UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

A program of research addressing exposure assessment in epidemiological studies of shift work Hall, Amy 2017

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2017_november_hall_amy.pdf [ 1.75MB ]
Metadata
JSON: 24-1.0355851.json
JSON-LD: 24-1.0355851-ld.json
RDF/XML (Pretty): 24-1.0355851-rdf.xml
RDF/JSON: 24-1.0355851-rdf.json
Turtle: 24-1.0355851-turtle.txt
N-Triples: 24-1.0355851-rdf-ntriples.txt
Original Record: 24-1.0355851-source.json
Full Text
24-1.0355851-fulltext.txt
Citation
24-1.0355851.ris

Full Text

	A	PROGRAM	OF	RESEARCH	ADDRESSING	EXPOSURE	ASSESSMENT	IN	EPIDEMIOLOGICAL	STUDIES	OF	SHIFT	WORK			by			 Amy	Hall	 		BScN	(Hons),	The	University	of	Toronto,	2004	MSc,	The	University	of	British	Columbia,	2009		A	DISSERTATION	SUBMITTED	IN	PARTIAL	FULFILLMENT	OF	THE	REQUIREMENTS	FOR	THE	DEGREE	OF		DOCTOR	OF	PHILOSOPHY	in	THE	FACULTY	OF	GRADUATE	AND	POSTDOCTORAL	STUDIES	(Population	and	Public	Health)		THE	UNIVERSITY	OF	BRITISH	COLUMBIA	(Vancouver)			September	2017			©	Amy	Hall,	2017	ii  Abstract	Background	Shift	work	is	common	with	wide-ranging	implications	for	worker	health.	It	is	also	complex,	presenting	challenges	for	exposure	assessment	in	epidemiological	studies	and	the	development	of	strong	evidence	to	inform	health	interventions	and	policies.	This	dissertation	generated	new	information	on	the	measurement,	assignment,	and	determinants	of	shift	work	exposure,	in	order	to	address	important	limitations	in	this	field	of	epidemiology.	Methods	In	Chapter	2,	152	full-shift	personal	light-at-night	measurements	were	collected	from	102	shift	workers	in	emergency	services	and	healthcare	to	investigate	exposure	variability	and	different	exposure	metrics.	In	Chapter	3,	multiple	exposure	indicators	were	constructed	for	a	national	survey	of	nurses	(n=11,450)	to	demonstrate	the	impacts	of	exposure	assignment	on	observed	relationships	between	shift	work	and	depression.	In	Chapter	4,	interviews	were	conducted	with	88	employers	in	one	Canadian	province	to	examine	determinants	of	workplace-level	shift	work	policies	and	practices.	Results	In	Chapter	2,	average	light-at-night	exposures	varied	across	occupations	and	settings;	between-group	variance	exceeded	between-worker	and	within-worker	variance,	and	all	exposure	metrics	were	moderately-to-highly	correlated.	In	Chapter	3,	the	strongest	relationships	between	shift	work	and	depression	were	observed	in	the	model	with	highest	exposure	precision,	defined	by	shift	timing	and	rotation	intensity,	whereas	weak	relationships	were	observed	in	models	with	lower	exposure	precision,	defined	by	shift	timing	or	presence/absence	of	shift	work.	In	Chapter	4,	long	duration	shifts	varied	by	industry	and	were	more	likely	in	large	workplaces;	shift	work	education/training	was	more	likely	in	large	workplaces	and	those	without	seasonal	shift	work;	and	nighttime	lighting	policies	were	more	likely	in	workplaces	reporting	that	maintenance,	client	service	needs,	or	prior	nighttime	incidents	affected	shift	work.	iii  Conclusions	This	dissertation	can	inform	future	epidemiological	studies	of	shift	work.	Chapter	2	identified	high-level	exposure	indicators	(e.g.,	occupation	groupings)	and	flexibility	in	the	choice	of	highly	correlated	metrics	for	light-at-night	exposure	studies.	Chapter	3	showed	that	increasing	the	precision	of	exposure	assignment	reduced	measurement	error	and	effect	attenuation	for	the	outcome	of	depression.	Chapter	4	identified	determinants	of	workplace-level	shift	work	policies	and	practices	(e.g.,	industry,	employer	size,	temporary	work,	and	employer	motivations)	to	consider	in	future	research	and	interventions.	 	iv  Lay	Summary		One	third	of	Canadians	perform	shift	work	(something	other	than	a	regular	daytime	schedule).	Shift	work	is	linked	to	various	physical	and	mental	health	outcomes,	however	strong	evidence	to	clarify	these	relationships	and	inform	health	interventions	has	been	slow	to	develop.	One	reason	is	that	shift	work	“exposure”	is	complex	and	difficult	to	assess	in	health	research,	where	risks	are	evaluated	by	comparing	groups	of	people	with	differing	amounts	of	exposure	to	a	hazard.	This	dissertation	addressed	three	evidence	gaps	concerning	the	assessment	of	shift	work	exposure	in	health	research.	Chapter	2	measured	and	characterized	light-at-night	exposure	levels	of	shift	workers;	Chapter	3	assessed	how	different	definitions	of	shift	work	exposure	affects	researchers’	ability	to	detect	health	effects;	and	Chapter	4	identified	workplace	characteristics	that	may	determine	certain	types	of	shift	work	policies	and	practices.	These	findings	should	be	considered	in	future	studies	of	shift	work	and	health.	 v  Preface	The	work	presented	here	is	that	of	the	candidate,	with	usual	guidance	from	the	supervisory	committee	(Dr.	Mieke	Koehoorn,	Dr.	Hugh	Davies,	and	Dr.	Renee-Louise	Franche)	and	manuscript	co-authors.	Additional	contributions	to	the	research	are	as	follows:	For	Chapter	2,	Dr.	George	Astrakianakis	at	the	University	of	British	Columbia,	Canada	provided	support	with	study	design	and	recruitment.	Dr.	Mariana	Figueiro,	Dr.	Mark	Rea,	and	Mr.	Geoffrey	Jones	at	the	Rensselaer	Polytechnic	Institute’s	Lighting	Research	Center,	USA,	provided	scientific	expertise	and	technical	support	with	the	study	sampling	equipment.	For	Chapter	3,	the	Statistics	Canada’s	Research	Data	Centres	Program	granted	the	candidate	access	to	microdata	from	The	National	Survey	of	the	Work	and	Health	of	Nurses,	conducted	by	Statistics	Canada	and	the	Canadian	Institute	for	Health	Information.		For	Chapter	4,	the	candidate’s	development	of	an	updated	survey	and	data	collection	protocols	built	upon	a	prior	employer	survey	conducted	in	2003	by	Dr.	Ralph	Mistlberger	and	Dr.	Glenn	Landry.	A	graduate	student	(Ms.	Andrea	Smit,	Simon	Fraser	University,	Canada)	assisted	study	management;	two	undergraduate	students	(Ms.	Shannon	Gahan	and	Ms.	Meagan	Ratz,	Simon	Fraser	University,	Canada)	conducted	the	majority	of	survey	interviews.	Identification,	design,	and	performance	of	the	research	program		The	candidate,	with	input	and	support	from	the	thesis	supervisory	committee	and	manuscript	co-authors,	developed	each	study’s	research	topic	and	research	design.		For	Chapter	2,	the	candidate	conceptualized	the	research	questions	on	light-at-night	exposure,	designed	the	study,	and	selected	the	exposure	measurement	tool.	The	candidate	then	carried	out	participant	recruitment	and	data	collection	and	conducted	all	data	preparation	and	statistical	analyses.	vi  For	Chapter	3,	the	candidate	conceptualized	the	research	questions	on	shift	work	exposure	assignment	and	depression	outcomes,	designed	the	study,	and	applied	for	access	to	Statistics	Canada	microdata	(The	National	Survey	of	the	Work	and	Health	of	Nurses)	via	Statistics	Canada’s	Research	Data	Centres	Program.	Once	access	was	granted,	the	candidate	conducted	all	data	preparation	and	statistical	analyses	on	site	at	the	University	of	British	Columbia	(Vancouver	campus)’s	Research	Data	Centre.	For	Chapter	4,	the	candidate	conceptualized	the	research	question	on	determinants	of	workplace-level	shift	work	policies	and	practices,	developed	an	updated	survey,	and	established	data	collection	protocols.	The	candidate	then	conducted	participant	recruitment	in	collaboration	with	the	study	manager,	oversaw	data	collection	by	research	staff,	and	conducted	all	data	preparation	and	statistical	analyses.	Publications		Three	research	chapters	(Chapters	2	to	4)	were	written	as	manuscripts	for	publication	in	peer-reviewed	journals.	All	outlines	and	drafts	were	prepared	by	the	candidate	and	circulated	for	review	by	thesis	committee	members	and	manuscript	co-authors	as	appropriate;	one	to	two	rounds	of	revisions	occurred	for	each	Chapter.		A	version	of	Chapter	2	has	been	published.	Hall	AL,	Davies	HW,	Koehoorn	M.	“Personal	light-at-night	exposures	and	components	of	variability	in	two	common	shift	work	industries:	uses	and	implications	for	future	research”.	Scand	J	Work	Environ	Health	(Online-first)	Sept	27,	2017;	doi:10.5271/sjweh.3673.	The	candidate	conceptualized	the	study,	carried	out	all	recruitment	and	data	collection,	conducted	data	cleaning	and	statistical	analyses,	interpreted	results,	and	prepared	the	manuscript.	The	candidate’s	total	contribution	to	the	work	as	a	whole	was	90%.	A	version	of	Chapter	3	is	under	peer	review	(September	2017).	Hall	AL,	Franche	RL,	Koehoorn	M.	“Examining	the	effect	of	exposure	assignment	in	epidemiological	studies	of	shift	work:	a	study	on	depression	among	nurses".	The	candidate	conceptualized	the	study,	vii  performed	all	data	cleaning	and	analyses,	interpreted	results,	and	prepared	the	manuscript.	The	candidate’s	total	contribution	to	the	work	as	a	whole	was	85%.	A	version	of	Chapter	4	has	been	published.	Hall	AL,	Smit	AN,	Landry	GJ,	Mistlberger	RE,	Koehoorn	M.	"Organizational	characteristics	associated	with	shift	work	practices	and	potential	opportunities	for	intervention:	findings	from	a	Canadian	study".	Occup	Environ	Med	2017;(74):6-13.	The	candidate	conceptualized	the	research	questions	for	this	study,	led	the	design	of	a	new	employer	survey,	co-supervised	data	collection,	performed	all	data	analyses,	and	wrote	the	manuscript.	The	candidate’s	total	contribution	to	the	work	as	a	whole	was	70%.		Ethics	Approval	Ethics	approval	for	Chapter	2	was	obtained	from	The	University	of	British	Columbia,	Canada’s	Behavioural	Research	Ethics	Board	(Study	Number	H15-01720).	Ethics	approval	for	Chapter	3	was	obtained	from	The	University	of	British	Columbia	Canada’s	Behavioural	Research	Ethics	Board	(Study	Number	H13-02137).	Ethics	approval	for	Chapter	4	was	obtained	from	Simon	Fraser	University,	Canada	(Study	Number	2014s0337)	and	The	University	of	British	Columbia,	Canada’s	Behavioural	Research	Ethics	Board	(Study	Number	H14-01588).		viii  Table	of	Contents	Abstract ......................................................................................................................................... ii	Lay	Summary .............................................................................................................................. iv	Preface ............................................................................................................................................v	Table	of	Contents ..................................................................................................................... viii	List	of	Tables ............................................................................................................................. xiii	List	of	Figures ............................................................................................................................ xiv	List	of	Abbreviations ................................................................................................................. xv	Glossary ...................................................................................................................................... xvi	Acknowledgements ................................................................................................................ xvii	Dedication ................................................................................................................................. xix	Chapter	1:	Introduction	and	background ...............................................................................1	1.1	 Shift	work:	biological,	physical,	and	social	concepts ..................................................... 3	1.1.1	 Circadian	biology	and	the	role	of	light .................................................................... 3	1.1.2	 Conceptualizing	factors	related	to	shift	work	and	its	relationships	with	health 4	1.1.2.1	 Upstream	factors	affecting	shift	work .............................................................. 5	1.1.2.2	 Primary	exposure:	Shift	work	schedule ........................................................... 6	1.1.2.3	 Secondary	exposures	or	modifiers:	light-at-night	and	workplace,	social,	and	individual	factors ............................................................................................................... 8	1.1.2.4	 Rhythm	disturbances	and	health	effects ........................................................ 12	1.1.2.5	 Summary ............................................................................................................ 13	1.2	 Exposure	assessment	challenges	in	epidemiological	studies	of	shift	work .............. 14	ix  1.3	 Rationale	and	objectives ................................................................................................ 15	1.3.1	 Measuring	light-at-night	exposure	levels	and	characterizing	variability	in	shift	workers ................................................................................................................................ 16	1.3.2	 Characterizing	the	impacts	of	exposure	assignment	in	epidemiological	studies	of	shift	work ......................................................................................................................... 17	1.3.3	 Identifying	determinants	of	workplace-level	shift	work	policies	and	practices 17	Chapter	2:	Personal	light-at-night	exposures	and	components	of	variability	in	two	industry	sectors	where	shift	work	is	common .....................................................................19	2.1	 Introduction .................................................................................................................... 19	2.1.1	 Light	concepts ......................................................................................................... 21	2.1.2	 Study	rationale	and	objectives ............................................................................... 21	2.2	 Methods .......................................................................................................................... 22	2.2.1	 Equipment ............................................................................................................... 23	2.2.2	 Recruitment	and	data	collection ............................................................................ 23	2.2.3	 Study	variables ........................................................................................................ 24	2.2.4	 Statistical	analyses .................................................................................................. 24	2.3	 Results ............................................................................................................................. 25	2.3.1	 Descriptive	summaries ........................................................................................... 26	2.3.2	 Components	of	variance ......................................................................................... 27	2.4	 Discussion ....................................................................................................................... 27	2.4.1	 Light-at-night	exposure	levels ............................................................................... 27	2.4.2	 Components	of	variance ......................................................................................... 30	x  2.4.3	 Exposure	metrics .................................................................................................... 32	2.4.4	 Strengths	and	limitations ....................................................................................... 33	2.5	 Conclusions ..................................................................................................................... 35	Chapter	3:	Examining	the	impacts	of	exposure	assignment	in	a	study	of	shift	work	and	depression	among	nurses .................................................................................................37	3.1	 Introduction .................................................................................................................... 37	3.1.1	 Hypothesized	pathways	between	shift	work	and	depression ............................ 38	3.1.2	 Prior	research	into	shift	work	and	depression ..................................................... 39	3.1.3	 Study	rationale	and	objectives ............................................................................... 41	3.2	 Methods .......................................................................................................................... 42	3.2.1	 Data	source .............................................................................................................. 42	3.2.2	 Study	sample ........................................................................................................... 43	3.2.3	 Study	variables ........................................................................................................ 43	3.2.4	 Statistical	analyses .................................................................................................. 46	3.3	 Results ............................................................................................................................. 46	3.3.1	 Descriptive	summaries ........................................................................................... 46	3.3.2	 Logistic	regression .................................................................................................. 49	3.4	 Discussion ....................................................................................................................... 51	3.4.1	 Strengths	and	limitations ....................................................................................... 54	3.5	 Conclusions ..................................................................................................................... 58	Chapter	4:	Assessing	determinants	of	workplace-level	shift	work	policies	and	practices:	An	employer	survey	in	British	Columbia,	Canada ............................................59	xi  4.1	 Introduction .................................................................................................................... 59	4.1.1	 Study	rationale	and	objective ................................................................................ 60	4.2	 Methods .......................................................................................................................... 61	4.2.1	 Survey	development ............................................................................................... 61	4.2.2	 Recruitment	and	data	collection ............................................................................ 62	4.2.3	 Study	variables ........................................................................................................ 63	4.2.4	 Statistical	analyses .................................................................................................. 64	4.3	 Results ............................................................................................................................. 64	4.3.1	 Descriptive	summaries ........................................................................................... 64	4.3.1	 Logistic	regression .................................................................................................. 67	4.4	 Discussion ....................................................................................................................... 69	4.4.1	 Long	duration	shifts ................................................................................................ 69	4.4.2	 Provision	of	shift	work	education	materials/training	to	employees ................. 70	4.4.3	 Nighttime	lighting	policies ..................................................................................... 73	4.4.4	 Strengths	and	limitations ....................................................................................... 74	4.5	 Conclusion ...................................................................................................................... 76	Chapter	5:	Discussion ................................................................................................................77	5.1	 Summary	of	studies,	methodological	considerations,	and	recommendations	for	future	research ........................................................................................................................ 79	5.2	 Conflicting	imperatives	in	translating	shift	work	and	health	research ..................... 85	5.3	 Summary ......................................................................................................................... 87	Bibliography ................................................................................................................................89	xii  Appendix	A ............................................................................................................................. 109	A.1	 Additional	detail	on	study	recruitment ................................................................. 109	Appendix	B ............................................................................................................................. 110	B.1	 Depression	scoring	in	the	National	Survey	of	the	Work	and	Health	of	Nurses .. 110	Appendix	C ............................................................................................................................. 113	C.1	 Copy	of	questionnaire	used ..................................................................................... 114	C.2	 Summary	of	study	recruitment ............................................................................... 125		xiii  List	of	Tables	Table	1.1:	Type	of	work	schedule	reported	at	end	of	year,	Canada…………………..........................	2	Table	2.1:	Summary	of	grouping	schemes	for	shift	workers’	personal	light-at-night	exposures……………………………………………………………………………………………………………………..	25	Table	2.2:	Personal	light-at-night	exposure	averages	(assessed	in	the	23:00	to	05:00	period).………………………………………………………………………………………………………………………...	26	Table	2.3:	Correlations	between	4	light-at-night	exposure	metrics	(n	=	152	observations)	assessed	in	the	23:00	to	05:00	period	…………..........…………..…………..……..…………..………………27	Table	2.4:	Proportions	of	variance	in	4	light-at-night	exposure	metrics	across	grouping	schemes,	accounted	for	by	between-group,	between-worker,	and	within-worker	components	(n	=	152	observations)	(assessed	in	the	23:00	to	05:00	period)..……..…………..	28	Table	3.1:	Baseline	study	sample	characteristics	and	bivariable	associations	with	depression	(within	previous	12	months):	National	Survey	of	the	Work	and	Health	of	Nurses	(NSWHN),	2005…………………………………………………………………………………………………………....	47	Table	3.2:	Unadjusted	and	adjusted	logistic	regression	odds	ratios	(ORs)	and	confidence	intervals	(CIs)	modeling	depression	=	yes,	National	Survey	of	the	Work	and	Health	of	Nurses,	2005…………………………………………………………………………………………………………..........	50	Table	4.1:	Characteristics	of	participating	organizations:	British	Columbia	employer	survey,	2014-2015..……………………………………………………………………………………………………………….....	66	Table	4.2:	Final	logistic	models	for	associations	between	determinant	variables	and	outcomes	of	(1)	long	duration	shift	use,	(2)	provision	of	shift	work	materials/training	to	employees,	and	(3)	nighttime	lighting	policies	in	the	workplace,	British	Columbia	employer	survey,	2014-2015………….………………………………………………………………………………………….....68	 xiv  List	of	Figures	Figure	1:	Summary	of	factors	related	to	shift	work	and	its	relationships	with	health.…..….....	5	 xv  List	of	Abbreviations	BC:	British	Columbia,	Canada	NAICS:	North	American	Industry	Classification	System	NSWHN:	National	Survey	of	the	Work	and	Health	of	Nurses	SI:	International	System	of	Units	xvi  Glossary	Chronotype:	The	behavioural	manifestation	of	underlying	circadian	rhythms	Circadian	rhythm:	A	24-hour	cycle	in	the	processes	of	living	beings	Circadian	disruption:	Perturbations	in	endogenous	circadian	rhythmicity	Entrainment:	The	alignment	of	a	circadian	rhythm's	period	and	phase	to	the	period	and	phase	of	an	external	rhythm	Exposure	assignment:	The	application	of	exposure	categories	or	levels	to	epidemiological	study	subjects		Exposure	indicator:	A	qualitative	proxy	for	exposure	(e.g.,	category	or	group)	applied	to	epidemiological	study	subjects	during	exposure	assignment	Exposure	metric:	A	quantitative	description	of	exposure	(e.g.,	mean,	median,	90th	percentile)	Lux:	A	measure	of	light	intensity	as	perceived	by	the	human	eye	Photopic	illuminance:	The	density	of	light	falling	on	a	surface,	described	in	units	of	lux	Zeitgeber:	Environmental	or	social	cue	that	synchronizes	the	processes	of	living	beings	to	a	24-hour	cycle		xvii  Acknowledgements	This	PhD	would	not	have	been	possible	without	the	support	and	guidance	that	I	received	from	many	people.	First,	I	cannot	thank	Mieke	Koehoorn	enough	for	her	tireless	mentorship	and	support	as	my	supervisor	over	the	past	five	years.	It	has	been	a	pleasure	to	learn	from	her	research	expertise,	leadership	style,	and	attention	to	detail.	I	would	also	like	to	thank	the	other	members	of	my	Committee:	Hugh	Davies	and	Renee-Louise	Franche,	for	their	time	and	insightful	feedback	that	pushed	me	to	become	a	better	researcher.	I	am	very	grateful	to	all	participants	who	volunteered	to	take	part	in	this	research;	none	of	it	would	have	been	possible	without	them.	For	the	light	exposure	study,	I	sincerely	thank	George	Astrakianakis	at	The	University	of	British	Columbia	for	his	assistance	during	conceptualization	and	recruitment	phases,	as	well	as	Karen	Bartlett	and	Matty	Jeronimo	for	their	help	with	accessing	the	Occupational	and	Environmental	Division	lab	and	equipment.	I	am	also	grateful	to	Mariana	Figueiro,	Mark	Rea,	and	Geoffrey	Jones	at	the	Rensselaer	Polytechnic	Institute’s	Lighting	Research	Center,	USA,	for	their	scientific	expertise	and	technical	support	with	the	Daysimeters.	For	the	analysis	of	depression	in	nurses,	I	thank	the	Statistics	Canada’s	Research	Data	Centres	Program	for	granting	me	access	to	microdata	from	The	National	Survey	of	the	Work	and	Health	of	Nurses.	On-site	technical	support	provided	by	Statistics	Canada	data	analysts	(Lee	Grenon,	Wendy	Kei,	and	Cheryl	Fu)	at	the	University	of	British	Columbia’s	Research	Data	Centre	was	much	appreciated.	For	the	survey	of	shift	work	employers,	I	thank	Ralph	Mistlberger	at	Simon	Fraser	University	for	his	generosity	and	trust	in	my	ability	to	run	this	study.	Glenn	Landry	provided	helpful	input	prior	to	the	study,	and	Andrea	Smit	was	an	invaluable	support	throughout	planning,	recruitment,	and	interview	stages.	Shannon	Gahan	and	Meagan	Ratz	conducted	the	interviews	with	insight	and	care.	Thank	you,	all.	I	gratefully	acknowledge	the	funding	received	to	support	my	PhD	research	from	WorkSafeBC,	The	Bridge	Program,	The	University	of	British	Columbia’s	Four	Year	Fellowship,	The	Partnership	for	Work	Safety	and	Health,	and	The	University	of	British	Columbia’s	Faculty	of	Medicine.	I	am	also	thankful	to	The	Canadian	Association	for	Research	on	Work	and	Health,	The	Canadian	Institutes	of	Health	Research’s	Gender	and	xviii  Health	Institute,	and	the	25th	&	26th	EPICOH	organizing	committees	for	supporting	my	attendance	at	conferences	and	training	institutes.	The	Bridge	Program	introduced	me	to	interdisciplinary	research	and	provided	valuable	training	that	I	would	not	have	obtained	elsewhere.	I	am	very	grateful	for	Linda	Bonamis’	administrative	support	and	knowledge,	and	for	faculty	mentorship;	particularly	that	of	Mike	Brauer,	Hugh	Davies,	Karen	Bartlett,	Sarah	Henderson,	and	Mieke	Koehoorn.	I	also	learned	a	great	deal	from	my	fellow	“Bridgies”,	and	am	indebted	to	Andrea	Jones,	Emily	Rugel,	Kaylee	Byers,	and	Leela	Steiner	for	their	friendship	and	collegial	support	throughout	our	respective	PhD	journeys.	I	am	grateful	to	Göran	Kecklund,	Constanze	Leineweber,	and	Phil	Tucker	at	Stockholm	University’s	Stress	Research	Institute,	for	welcoming	me	to	work	with	them	as	part	of	a	research	internship	in	2017.	It	was	enjoyable	and	informative	to	spend	time	in	Sweden	and	meet	a	variety	of	international	researchers	under	their	mentorship.	Within	UBC’s	School	of	Population	and	Public	Health,	I	thank	Gary	Poole	&	Charlyn	Black	for	introducing	me	to	the	value	and	enjoyment	of	teaching;	Beth	Hensler	for	her	guidance	through	degree	requirements,	applications,	and	other	life	hurdles;	and	Suhail	Marino	and	Jacqueline	Carpio	for	their	administrative	support.	I	am	also	grateful	to	my	colleagues	at	the	Partnership	for	Work,	Safety	and	Health	(especially	Lillian	Tamburic,	Esther	Maas,	Robert	Macpherson,	and	Niloufar	Saffari)	for	their	encouragement	through	long	days	of	data	collection	and	writing.		Thank	you	to	my	good	friends	near	and	far,	particularly	Alison	Palmer,	Barb	Karlen,	Kim	McLeod,	Cheryl	Peters,	Midori	Courtice,	Elaine	Fuertes,	Josh	Bates,	and	Caitlin	Gallichan-Lowe,	for	keeping	me	grounded	in	the	world	“outside”.	Also	thanks	to	my	favourite	(and	dearly	missed)	4-legged	friends,	Zoe	and	Justin,	for	providing	stress	therapy	in	trying	times.	To	my	family,	many	of	whom	were	coerced	into	proofreading	numerous	scholarships	and	funding	applications,	I	extend	both	gratitude	and	apologies.	I	am	particularly	grateful	to	Denise	Hall,	Judy	Hall,	and	my	sister	Marley	Wictorin;	a	lifelong	ally	who	helped	me	to	move	through	tough	patches	with	a	sense	of	humour.	Finally,	I	sincerely	thank	Hans	Kromhout,	whose	patience	always	prevailed	when	mine	was	thin.	His	tireless	support,	companionship,	and	fine	coffee	making	skills	made	this	journey	so	much	easier.	xix  Dedication		To	Joan	Budd,	in	memoriam.	Champion	of	nature,	education,	and	logic;	always	moving,	thinking,	questioning.		                1 Chapter	1: Introduction	and	background	Up	to	the	late	1800s,	human	activity	was	governed	primarily	by	natural	solar	cycles	of	light	and	darkness.	Modern	society	has	introduced	light	into	times	and	places	where	it	does	not	naturally	occur,	allowing	for	work	and	social	behaviours	to	extend	beyond	the	solar	day.	“Shift	work”	generally	aims	to	extend	an	organization’s	operational	time	beyond	a	regular	8-hour	day,	using	a	succession	of	worker	teams	(1).	This	has	become	a	common	form	of	working	time;	shift	work	is	reported	by	a	third	of	the	Canadian	workforce	(representing	over	4	million	individuals)	(2)	and	by	up	to	30%	of	other	workforces	globally	(3).	Shift	work	enables	extended	or	24-hour	activity	in	sectors	where	continuous	services	(e.g.,	law	enforcement,	healthcare,	transport,	telecommunications)	or	operations	(e.g.,	power	and	water	utilities)	are	essential	(1).	It	is	also	used	for	economic	reasons,	such	as	to	maximize	cost	efficiency	in	production	and	manufacturing	(4).	In	recent	decades,	shift	work	has	been	increasingly	relied	upon	to	extend	the	provision	of	non-essential	services	(e.g.,	entertainment,	restaurants,	fuel,	and	shops)	(1,4,5).	The	largest	proportions	of	Canadian	shift	workers	are	located	in	trade,	manufacturing,	accommodation	and	food	services,	and	healthcare	and	social	assistance	sectors	(6);	similar	to	the	United	States	(7)	and	European	countries	(8).	Women	represent	the	majority	of	night	workers	in	healthcare	and	social	assistance,	trade,	and	accommodation	and	food	services	sectors,	whereas	men	represent	the	majority	of	night	workers	in	manufacturing,	business,	building	and	other	support	services,	and	public	administration	(6).	The	most	prevalent	types	of	shift	work	schedules	are	irregular	shifts	(shift	changes	that	are	usually	prearranged	one	week	or	more	in	advance	-	for	example,	pilots)	and	rotating	shifts	(those	that	periodically	change	between	days,	evenings,	and/or	nights)	(See	Table	1.1).	This	is	followed	by	regular	evening	shifts,	regular	night	shifts,	split	shifts	(two	or	more	distinct	work	periods	each	day),	and	on	call/casual	shifts	(no	prearranged	schedule	–	for	example,	substitute	teachers).	Between	1996	and	2011,	the	proportions	of	Canadian	workers	in	regular	day	schedules	decreased,	while	proportions	of	workers	in	irregular	schedules	and	regular	night	shifts	increased	(9–11).	The	majority	of	Canadian	shift	workers	(82%)	work	full-time	(30	hours	or	more	per	week)	(2).		 2 Table	1.1:	Type	of	work	schedule	reported	at	end	of	year,	Canada1,2			 1996	%	 2006	%	 2011	%	Regular	daytime	schedule	 68.4	 65.6	 66.1	Irregular	schedule	 9.7	 11.0	 12.3	Rotating	shift	 10.1	 10.7	 9.4	Regular	evening	schedule	 5.3	 5.5	 4.7	On	call	 2.2	 2.2	 2.7	Regular	night	or	graveyard	shift	 1.7	 2.2	 2.0	Other/don't	know/refusal	 1.6	 1.9	 1.7	Split	shift	 1.0	 0.9	 1.0	1	Population:	Persons	aged	16-69	years	and	had	a	job	during	the	reference	year	and	paid	worker	2	Sources:	Survey	of	Labour	and	Income	Dynamics	(9–11)	Despite	its	many	economic	and	social	benefits,	shift	work	can	disrupt	daily	personal	rhythms	of	biology,	physiology	and	behaviour,	with	negative	implications	for	worker	safety	and	health	(3–5,12–17).	Shift	work-related	injury	and	disease	produce	significant	personal	and	economic	burdens,	such	as	workforce	absenteeism,	productivity	losses,	and	increased	use	of	health	services	(18).	Research	into	shift	work	and	its	impacts	on	health	is	therefore	an	important,	yet	evolving,	field.		A	recognized	challenge	in	this	area	of	epidemiological	and	intervention	research	is	that	“exposure”	to	shift	work	encompasses	a	variety	of	different	scheduling,	work	environment,	social,	and	individual	characteristics	(19).	This	complexity	tends	to	result	in	coarse	exposure	assessment	and	assignment	that	produces	misclassification,	weakens	exposure	contrast,	and	masks	differences	across	groups	(20,21).	It	is	therefore	not	surprising	that	calls	have	been	made	to	improve	the	quality	of	exposure	assessment	in	epidemiological	studies	of	shift	work	(21–24).	Such	activities	are	well	justified,	given	the	increasing	prevalence	of	shift	work	(particularly	irregular	work)	in	Canada	and	internationally	(25,26)	and	its	preponderance	in	disproportionately	susceptible	groups	such	as	young,	lower-educated,	and	lower-income	workers	(27–30).		To	address	current	evidence	gaps	and	inform	future	exposure	assessment	methods	in	epidemiological	studies	of	shift	work,	this	dissertation	includes	a	series	of	three	distinct	 3 studies	that	investigated	the	measurement	(Chapter	2),	assignment	(Chapter	3),	and	determinants	(Chapter	4)	of	shift	work	exposure.	Chapter	1	will	provide	a	review	of	basic	concepts,	summarize	challenges	related	to	exposure	assessment	in	epidemiological	studies	of	shift	work,	and	introduce	research	objectives	for	this	dissertation.	Following	a	summary	of	each	study	in	Chapters	2,	3,	and	4,	Chapter	5	will	summarize	findings,	discuss	methodological	considerations,	and	present	recommendations	for	future	research.	1.1 Shift	work:	biological,	physical,	and	social	concepts	This	dissertation	does	not	seek	to	systematically	review	the	fields	of	circadian	biology,	lighting	science,	and	work	organization.	However,	basic	concepts	related	to	shift	work-related	exposures	and	health	(circadian	biology,	exposure	to	light-at-night,	work	schedules,	and	pathways	between	shift	work	and	health)	are	summarized	in	the	following	sections	to	provide	the	reader	with	a	foundation	to	understand	the	studies	herein.		Further,	this	dissertation	aims	to	present	general	exposure	assessment	concepts	for	use	in	future	epidemiological	studies	of	shift	work.	While	these	broadly	discussed	concepts	may	apply	to	a	number	of	health	outcomes,	only	the	outcome	of	depression	will	be	thoroughly	discussed	(as	the	focus	of	the	epidemiological	study	presented	in	Chapter	3).	1.1.1 Circadian	biology	and	the	role	of	light	Rhythms	that	follow	the	24-hour	cycle	of	light	and	dark	are	circadian.	The	biological	processes	of	humans	and	many	other	living	organisms	operate	on	a	circadian	cycle.	In	mammals,	a	master	clock	resides	in	the	suprachiasmatic	nucleus,	located	in	the	brain’s	hypothalamus	(31).	The	suprachiasmatic	nucleus	drives	the	circadian	rhythmicity	(clocks)	of	various	downstream	physiological	processes,	including	the	sleep-wake	cycle,	hormone	release,	cardiovascular	function,	and	metabolism	(32,33).		Humans	are	diurnal	mammals,	meaning	they	are	naturally	awake	and	active	during	daylight	hours	and	resting/sleeping	at	night.	The	suprachiasmatic	nucleus	is	synchronized	to	a	24-hour	(circadian)	cycle	by	a	number	of	environmental	and	social	cues	called	 4 zeitgebers	(33–35);	the	term	“entrainment”	describes	the	stable	synchronization	of	the	biological	clock	with	its	zeitgebers.	The	strongest	zeitgeber	for	the	human	circadian	system	is	the	sun’s	24-hour	cycle	of	light	and	dark	(36).	Details	of	the	mechanisms	by	which	external	light	cues	are	translated	into	internal	physiological	signals	have	only	recently	been	explained	(37–39).	In	mammals,	the	zeitgeber	light	is	detected	exclusively	by	the	eyes	(40),	where	the	retinas	transduce	light	signals	to	the	suprachiasmatic	nucleus	via	specialized	receptors	called	intrinsic	photosensitive	retinal	ganglion	cells	(38).	The	suprachiasmatic	nucleus	in	turn	translates	these	light	signals	to	coordinate	the	circadian	rhythmicity	of	physiological	clocks	throughout	the	body	(33,41),	such	as	the	sleep-wake	cycle	(42),	metabolism	(43,44),	cardiovascular	function	(45),	and	hormone	release	(46,47).	Melatonin	is	one	important	hormone	that	is	produced	and	released	by	the	pineal	gland	of	mammals	on	a	circadian	cycle	(48).	Commonly	described	as	the	“biological	marker	of	night”,	melatonin’s	circadian	rhythm	aligns	closely	with	the	natural	light-dark	period,	with	circulating	levels	lowest	in	daytime,	increasing	in	the	evening,	peaking	at	approximately	4	am,	and	returning	to	near	nil	levels	by	late	morning	(49).	The	nocturnal	release	of	melatonin	into	the	blood	and	cerebrospinal	fluid	signals	time	of	day	information	to	cells,	tissues	and	organs	throughout	the	body	(50);	and	circulating	melatonin	is	widely	regarded	as	the	best	biological	index	of	circadian	timing	(51).		Perturbations	in	endogenous	circadian	rhythmicity	(e.g.,	displacement	of	sleep	relative	to	the	circadian	clock),	particularly	by	exposure	to	artificial	light	during	natural	periods	of	dark,	are	defined	as	“circadian	disruption”	(52).	This	concept	will	be	described	further	in	section	1.1.2.4.	1.1.2 Conceptualizing	factors	related	to	shift	work	and	its	relationships	with	health	The	structure	and	distribution	of	shift	work	is	influenced	by	labour	market	and	social	conditions.	In	turn,	shift	work	is	associated	with	exposure	to	light-at-night	and	a	number	of	circadian,	sleep,	and	social	disturbances	that	negatively	impact	on	health.	Pathways	between	these	disturbances	and	health	are	strongly	interconnected	(34,42,52–54)	and	a	 5 number	of	exacerbating	loops	may	feed	back	into	an	individual’s	work	schedule	(4).	A	conceptual	model	of	the	relationships	between	shift	work	and	health,	including	upstream	factors,	intermediaries,	and	mechanisms,	is	presented	in	Figure	1	and	discussed	in	the	following	sections.		Figure	1:	Summary	of	factors	related	to	shift	work	and	its	relationships	with	health	 	Adapted	with	permission	(55)		1.1.2.1 Upstream	factors	affecting	shift	work	Shift	work	is	a	product	of	evolving	social	and	labour	conditions	that	have	changed	significantly	in	recent	decades.	In	the	UK	and	USA	for	example,	recessions	in	the	1980s	were	followed	by	labour	market	de-regulation	and	large-scale	replacement	of	traditional,	relatively	secure	industrial	jobs	(conferring	strong	unionization	and	moderately	sized	 6 salaries)	with	lower-paid	work	in	the	service	economy	(56).	Women	and	older	people	represent	increasing	proportions	of	the	workforce;	part-time,	flexible	(e.g.,	freelancers),	and	multiple	work	arrangements	are	also	on	the	rise	(57).	Increased	labour	market	“flexibility”	with	less	regulation	by	trade	union	agreements	and	employment	laws	has	given	employers	greater	leeway	to	offer	insecure	work	and	to	structure	working	time	as	they	see	fit	(56).	Meanwhile,	tightening	rules	for	unemployment	benefits	has	resulted	in	higher	take-up	of	low-wage,	insecure	employment	(56).	Social	inequalities	in	the	quality	of	work,	including	the	structure	of	working	time,	are	well	known.	For	example,	workers	with	at	least	a	postsecondary	degree	are	less	likely	to	report	shift	work,	and	workers	in	lower	income	households	are	more	likely	to	report	evening,	night	or	irregular	shift	work	(58).		1.1.2.2 Primary	exposure:	Shift	work	schedule		Organizations	utilize	a	range	of	work	schedules	that	vary	widely	by	timing,	duration,	and	frequency	of	work.	This	includes	shift	work,	compressed	work,	overtime,	part-time,	flexible	hours,	and	on-call	work,	among	others.	“Shift	work”	refers	to	a	work	system	where	one	group	of	workers	replaces	another	throughout	the	workday;	a	“shift	worker”	is	typically	defined	as	someone	who	regularly	begins	or	ends	work	outside	of	daytime	hours	(1).		Shift	work	schedules	vary	considerably	with	respect	to	a	number	of	features	(54,59)	that	may	influence	their	impacts	on	circadian	rhythms,	sleep,	and	social	effects	(60–62).	These	features	include	start	and	end	times	of	shifts,	rotation	between	shifts	(including	considerations	of	frequency	and	direction	of	rotation),	duration	of	individual	shifts,	number	of	successive	work	days,	number	of	successive	shifts	of	a	particular	type,	days	of	the	week	worked	(weekday	versus	weekend),	duration	of	rest	between	shifts,	frequency	and	duration	of	breaks	within	shifts,	regularity	(or	irregularity)	of	work	schedule,	amount	of	notice	given	before	a	shift,	and	an	individual’s	control	over	their	work	schedule,	among	others	(1).		Decisions	concerning	the	design	of	shift	schedules	incorporate	a	number	of	factors,	including	(59):	 7 • legislation	(working	time	in	Canada	is	mostly	governed	by	general	duty	clauses	laid	out	by	individual	provinces	and	in	federally	regulated	workplaces);	• collective	agreements	(if	the	workforce	is	unionized);		• economic	aims,	e.g.,	varying	consumer	demand,	maximizing	use	of	equipment/facilities;	• labour	market,	e.g.,	availability	of	qualified	workers	and	of	full-time,	part-time,	or	casual	workers;	• characteristics	of	the	workforce,	e.g.,	age,	caretaking	or	social	responsibilities;	• physiological,	psychological,	and	social	recommendations	based	on	current	science	about	worker	health,	performance,	safety,	and	well-being	outside	of	work.	There	are	thousands	of	shift	work	schedules	in	use	throughout	the	world,	with	no	one	optimal	choice	(59).	Any	schedule	(including	standard	daytime	work)	may	be	problematic	if	it	does	not	align	with	an	individual	worker’s	needs;	however,	some	work	schedules	have	a	greater	likelihood	of	being	problematic.	Research	consensus	on	“best”	scheduling	practices	in	relation	to	health	is	generally	lacking	(60),	particularly	concerning	long-term	health	outcomes	(62).	However,	there	is	some	agreement	concerning	scheduling	practices	to	mitigate	short-term	effects.	For	example,	night	and	early	morning	work	(both	requiring	wakefulness	during	the	natural	sleep	period)	are	generally	considered	to	be	the	most	disruptive	to	normal	circadian	functioning	and	sleep	(21,59,61).	Reductions	in	consecutive	night	shifts	by	increasing	shift	rotation	speed	are	associated	with	improvements	to	sleep,	reduced	fatigue,	and	possibly	improved	work-life	balance	(60,62).	Flexible	working	patterns	to	increase	worker	control,	such	as	self-scheduling,	are	associated	with	positive	health	outcomes	and	decreased	levels	of	sickness	and	disability	absence	(60,63,64),	although	little	is	known	about	the	longer-term	impacts	of	such	interventions	on	sleep,	alertness,	and	other	health	outcomes	(22).	Changing	the	direction	of	rotating	shift	patterns	from	backward	(nights	to	days)	to	forward	(days	to	nights)	is	associated	with	better	physical	and	psychological	well	being,	lower	levels	of	chronic	fatigue,	and	greater	alertness	at	work	(62,65,66),	although	this	evidence	base	is	not	entirely	conclusive	(62,65).	Further	discussion	of	factors	such	as	frequency	of	shift	rotations	and	shift	duration	will	be	discussed	in	Chapters	3,	4,	and	5.			 8 In	sum,	while	there	may	be	no	“ideal”	choice	of	a	work	schedule,	it	has	been	argued	that	shift	work	should	be	structured	to	minimize	disruption	of	circadian	rhythms,	avoid	the	accumulation	of	sleep	deficits,	and	maximize	regular	social	interaction	(59).		1.1.2.3 Secondary	exposures	or	modifiers:	light-at-night	and	workplace,	social,	and	individual	factors		Light-at-night	levels	can	be	expected	to	vary	across	different	work	environments.	Regulatory	requirements	for	photopic	illuminance	levels	in	buildings	typically	incorporate	issues	of	politics	and	economics	(e.g.,	price	of	energy,	source	and	cost	of	equipment)	in	addition	to	the	needs	associated	with	the	structural	environment	in	question	(67,68).	In	the	province	of	British	Columbia,	recommended	workplace	illumination	levels	are	also	based	on	safety	requirements	and	task	categories	(69).	This	means	that	occupational	light-at-night	exposure	levels	may	vary	considerably	across	workplaces,	occupations	and	job	tasks.	The	impacts	of	light-at-night	exposure	on	physiological	responses	differ	across	various	lighting	aspects	(e.g.,	intensity,	timing,	and	wavelength)	(32,70).	Light-at-night	may	also	impact	on	individuals	differently	as	a	function	of	one’s	pupillary	size	(affecting	the	amount	of	light	incident	on	the	retinas)	as	well	as	age	(increasing	age	is	associated	with	decreased	circadian	responsiveness	to	light)	(52,70).	Workplace	factors	are	characteristics	that	apply	equally	to	all	individuals	within	an	organization	or	site	(71),	such	as	lighting,	availability	of	healthy	food	during	both	day	and	night,	provision	of	rest	areas	for	breaks	(most	available	evidence	suggests	that	naps	on	the	night	shift	provide	positive	benefits	in	terms	of	subjective	sleepiness	and	performance	(72)),	and	company	carpooling	programs	(to	reduce	the	increased	risk	of	motor	vehicle	accidents	related	to	sleepiness	and	fatigue	following	night	or	long	work	hours	(73,74)).	Workplace	factors	may	also	refer	to	psychosocial	conditions	within	workplaces.	The	“psychosocial	environment”	has	been	defined	as	a	socio-structural	set	of	opportunities	that	is	available	to	meet	an	individual’s	needs	of	well	being,	productivity,	and	positive	self	experience	(75).	Two	important	aspects	of	positive	self	experience	are	self	efficacy	and	self	esteem.	Self	efficacy	is	defined	as	an	individual’s	belief	in	their	ability	to	accomplish	tasks;	 9 this	is	often	assessed	using	demand-control	models	to	evaluate	the	combination	of	work	demands	with	control	over	task	performance	(57).	Self	esteem	is	defined	as	the	consistent	positive	experience	of	a	person’s	self	worth;	this	is	often	assessed	using	the	effort-reward	imbalance	model	to	evaluate	social	reciprocity	in	the	workplace	(i.e.,	the	extent	to	which	employee	efforts	are	reciprocated	by	equitable	rewards)	(57).	Other	psychosocial	conditions	with	potential	impacts	on	health	include	staffing	levels,	safety	culture	and	conditions,	support	from	supervisors	and	colleagues,	and	work	time	control	(4,76).	In	the	field	of	shift	work	epidemiology,	psychosocial	conditions	have	received	the	most	attention	concerning	their	relationship	to	increased	risks	of	cardiovascular	disease	(77)	and	mental	health	outcomes	(78,79)	in	shift	workers.	Additional	workplace	factors	are	physical	job	characteristics	(e.g.,	noise,	heat,	dust,	ergonomic	tasks)	(76,77)	where	exposures,	or	an	individual’s	susceptibility	to	them,	may	vary	across	time	of	day	worked	(80).	This	was	demonstrated	in	a	nationwide	survey	of	workplace	exposures	in	New	Zealand,	where	workers	with	non-standard	hours	were	more	likely	to	report	exposure	to	all	hazards	assessed	(exposure	to	dust,	smoke	or	fume,	gas,	oils	or	solvents,	acids	or	alkalis,	fungicides,	or	other	chemical	products)	than	their	regular	daytime	working	counterparts	(81).		Social	factors	describe	potential	interactions	between	a	worker’s	personal	and	professional	life,	such	as	marital	status	and	presence	of	children	at	home	(19,82),	the	time	required	to	commute	to/from	work	(4),	leisure	time	demands/activities	(4),	and	home	environment	factors	(e.g.,	daytime	sleep	environment,	neighbourhood	safety)	(4,59).	Relative	to	regular	daytime	workers,	shift	workers	are	more	likely	to	be	single	or	previously	married	(2,58);	in	turn,	marital	status	is	strongly	associated	with	an	increased	risk	of	mental	health	outcomes	(the	prevalence	of	major	depressive	disorder	is	lowest	in	married	and	highest	in	single	and	previously	married	Canadians)	(83,84)	and	other	health	outcomes.	Literature	on	health	and	mortality	has	consistently	indicated	that	un-partnered	individuals	generally	report	poorer	health	and	have	a	higher	mortality	risk	than	those	that	are	partnered,	with	men	being	particularly	affected	(85).	 10 Children	add	to	domestic	obligations	and	therefore	contribute	to	difficulties	in	coping	with	shift	work;	this	issue	is	especially	problematic	for	women,	although	men	are	also	affected	(82).	One	large	Canadian	survey	noted	that	males	living	in	households	with	children	were	less	likely	to	work	shift	than	those	without	children,	while	there	was	no	reported	difference	for	female	workers	(58).	However,	female	workers	were	more	likely	than	men	to	report	caring	for	family	as	their	main	reason	for	shift	work	(58).		Work/family	conflicts	are	common	in	shift	workers,	since	most	family	and	social	activities	are	organized	according	to	daytime	routines	of	the	general	population.	This	may	negatively	affect	marital	relationships	and	parental	roles,	and	may	also	lead	to	increased	sleep	problems,	chronic	fatigue,	and	psychosomatic	symptoms	(80).	Leisure	time	to	allow	for	renewal	of	mental	or	physical	strength	away	from	work	is	of	less	benefit	when	experienced	in	poor	conditions	(56).	For	example,	commuting	time	to	and	from	work	can	have	a	major	impact	on	a	shift	worker’s	ability	to	rest	and	sleep	between	shifts	(4).	Shift	workers	who	must	sleep	during	the	day	may	also	be	impacted	by	the	nature	of	their	home	environment	and	non-work	responsibilities	(such	as	childcare).	Recommendations	to	improve	daytime	sleep	hygiene	include	the	use	heavy	curtains	to	block	out	light,	ensuring	sound	insulation	of	the	home	envelope,	use	of	air	conditioning,	and	sleeping	in	an	adequate	bed	(71).	Such	measures	are	not	feasible	for	all	workers,	particularly	those	who	are	economically	disadvantaged	and	may	face	additional	challenges	such	as	lack	of	access	to	childcare	or	a	safe	home	environment	(4).		Individual	factors	that	may	affect	shift	work	schedule’s	impacts	on	health	include	genetic,	sex,	and	age	differences	in	biological	timing	(24,82,86).		Humans	differ	from	one	another	as	to	when	their	biological	night	starts	and	ends.	In	the	absence	of	external	zeitgebers,	the	intrinsic	period	of	the	human	biological	clock	averages	24.18	hours	(87).	Inter-individual	genetic	differences	affect	the	period	length	of	this	daily	cycle,	as	well	as	the	alignment	of	an	individual’s	internal	clock	with	external	zeitgebers	(e.g.,	time	difference	between	sunrise	and	wake	up,	or	between	sunset	and	bedtime)	(33).	This	effect	describes	differences	in	chronotype,	with	late	chronotypes	commonly	referred	 11 to	as	“owls”	and	early	chronotypes	as	“larks”.	It	has	recently	been	argued	that	dichotomous	“work	at	the	civil	night	versus	work	at	the	civil	day”	comparisons	are	insufficient,	since	circadian	disruption	may	not	necessarily	be	caused	when	individuals	work	at	night	as	defined	by	civil	time	(88).	For	example,	owls	may	not	experience	circadian	disruption	when	working	during	parts	of	the	civil	night,	but	may	incur	circadian	disruption	when	working	early	shifts	during	the	civil	day	(89).		Shift	workers	are	younger	relative	to	regular	daytime	workers	(2,58).	To	some	extent	this	may	relate	to	age-related	differences	in	chronotype.	Chronotype	is	delayed	in	adolescence;	reaching	a	maximum	of	“lateness”	at	approximately	20	years	old;	then	becoming	earlier	again	with	advancing	age	(the	peak	of	“earliness”	occurs	after	60	years	of	age)	(33).	This	phenomenon,	in	combination	with	decreased	duration	and	quality	of	sleep	with	advancing	age	(90),	might	explain	older	workers’	decreased	tolerance	for	night	work	and	long-duration	shifts	(91).		Sex	and	gender	differences	are	also	seen	in	shift	work	schedules	and	tolerance	to	night	work	(2,58).	Women	represent	approximately	37%	of	all	full-time	shift	workers	and	nearly	70%	of	all	part-time	shift	workers	in	Canada	(2).	Women	are	more	likely	than	men	to	work	rotating	shifts	(41%	versus	34%)	and	evening	shifts	(14%	versus	10%),	while	men	are	more	likely	than	women	to	work	irregular	shifts	(35%	versus	25%)	(2).	Males	are	on	average	later	chronotypes	than	females	(33);	this	could	explain	males’	relatively	higher	tolerance	for	shift	work	as	demonstrated	by	better	sleep	and	less	fatigue	and	sleepiness	at	work,	and	a	healthier	lifestyle	(24).	Gender	differences	in	total	work	time	may	also	explain	these	differences	in	tolerance,	since	total	weekly	work	time	(paid	plus	unpaid)	is	consistently	higher	in	employed	women	compared	to	men	(92).	Pre-existing	health	conditions	can	also	increase	an	individual’s	susceptibility	to	negative	impacts	arising	from	shift	work	(93).	Such	conditions	may	relate	to	a	higher	risk	health	behaviour	profile	arising	from	or	related	to	shift	work,	such	as	night	shift	workers’	increased	odds	of	smoking,	obesity,	and	low	socioeconomic	status	(94,95).			 12 1.1.2.4 Rhythm	disturbances	and	health	effects		The	acute	effects	associated	with	shift	work	often	mimic	symptoms	of	jetlag	(4),	including	fatigue,	sleep	disturbances,	digestive	problems,	and	impaired	cognition	and	performance	(96).	The	risk	of	acute	work-related	injuries	also	increases	during	night,	evening,	and	early	morning	hours,	relative	to	day	work	(97).		The	chronic	effects	associated	with	shift	work	include	increased	risk	of	developing	breast	and	other	cancers	(20,98,99),	as	well	as	cardiovascular	(96),	metabolic	(100),	and	mood	disorders	(16,101).	A	discussion	of	the	evidence	for	shift	work’s	effects	on	mental	health	outcomes,	such	as	mood	disturbances	and	depression,	is	presented	in	Chapter	3.	Shift	work	and	exposure	to	light-at-night	can	lead	to	various	circadian,	sleep,	and	social	rhythm	disruptions	(21,102–104).	These	disruptions	are	thought	to	negatively	impact	health	through	a	variety	of	pathways	that	are	highly	interconnected	and	difficult	(if	not	impossible)	to	disentangle	(32).		Both	light-at-night	and	altered	sleep	can	disrupt	circadian	rhythmicity	in	hormones,	gene	expression,	metabolism	markers,	and	a	number	of	other	physiological	parameters	(32,105),	including	melatonin	production	and	release	(46,106,107).	Melatonin	is	involved	in	a	number	of	human	physiological	and	pathophysiological	processes,	such	as	cancer	inhibition	(50),	immune	function	and	vascular	regulation	(108).	Melatonin	also	increases	propensity	for	sleep	at	night	(109),	sleep	duration,	and	sleepiness	(48,110,111).		Sleep	disturbances	in	shift	workers	are	well	documented	(13,61,112–114).	Working	shifts	is	associated	with	future	sleep	disturbances,	while	termination	of	shift	work	is	associated	with	fewer	sleep	disturbances	(112).	Poor	sleep	quantity	and	quality	is	in	turn	associated	with	impairments	of	cognitive	and	motor	performance	(leading	to	increased	risk	of	workplace	incidents)	(115,116),	and	various	chronic	effects	such	as	altered	immune	function	(117),	increased	risk	of	cardiovascular	and	metabolic	disorders	(and	associated	mortality)	(15,118–120),	and	depression	(121).	Disturbed	sleep	may	also	increase	the	likelihood	of	light-at-night	exposure	(e.g.,	insomnia	leading	to	increased	activity	during	the	biological	night).		 13 Social	de-synchronization	arising	from	night	shifts	or	other	irregular	work	schedules	can	affect	shift	workers’	participation	in	family,	social,	and	cultural	activities,	resulting	in	both	biological	disturbances	(e.g.,	gastrointestinal,	cardiovascular,	and	sleep)	(19,34,104)	and	psychological	disturbances	(e.g.,	increased	work-family	conflict,	marital	problems,	and	social	isolation)	(96,122)	that	may	feed	back	into	circadian	and	sleep	disturbances	(123,124).	Changes	in	work	organization	over	time	have	also	altered	the	boundaries	between	home	and	work	(25).	Conflict	between	work	and	home	demands,	which	appears	to	be	particularly	problematic	for	women	(125,126),	can	lead	to	stress	and	negative	health	consequences	(127).	1.1.2.5 Summary	There	is	no	“one	size	fits	all”	solution	to	address	the	negative	consequences	of	shift	work;	a	highly	variable	exposure	with	a	number	of	workplace,	situational,	and	personal	factors	influencing	its	effects	on	health.	The	preceding	section	describes	a	number	of	pathways	or	mechanisms	for	how	shift	work	may	arise	or	vary	within	the	workforce,	and	how	shift	work	and	its	characteristics	may	impact	health	or	modulate	health	effects.	Since	many	of	these	pathways	overlap	and	are	difficult	to	disentangle,	common	upstream	variables	(the	focus	of	this	dissertation,	i.e.,	work	schedule	and	related	exposures	such	as	light-at-night),	are	reasonable	exposure	targets	in	epidemiological	studies	of	shift	work.	This	reflects	a	“stimulus-based”	model	approach,	which	describes	shift	work	as	a	stressor	that	is	a	direct	cause	of	disease,	compared	to	a	“transactional”	model	approach,	which	implies	that	variables	aside	from	shift	work	are	responsible	for	poor	health	(128).	While	transactional	models	tend	to	place	responsibility	for	worker	health	on	the	individual	as	a	target	of	change,	stimulus-based	models	focus	on	environmental	factors,	thus	shifting	responsibility	from	the	individual	to	the	employer	and	society.	In	this	way,	stimulus-based	research	can	be	used	to	advocate	for	broadly	applied	interventions;	this	has	been	promoted	as	a	more	pragmatic	and	effective	means	of	reducing	shift	work-related	disease	burden	at	a	population	level	(129).		 14 1.2 Exposure	assessment	challenges	in	epidemiological	studies	of	shift	work	Exposure	assessment	is	an	important	factor	in	all	epidemiological	research	seeking	to	identify,	evaluate,	and	control	health	risks	(130).	In	the	epidemiological	context,	exposure	assessment	involves	the	identification	of	the	hazard	to	be	evaluated,	collection	of	data,	assignment	of	exposure	indicators,	and	selection	of	appropriate	metrics	to	estimate	exposures	(131).	These	activities	are	crucial	to	epidemiological	studies,	which	evaluate	health	risks	by	comparing	outcomes	across	differently	exposed	groups.	Generally,	the	term	“exposure”	refers	to:	(1)	“proximity	and/or	contact	with	a	source	of	a	disease	agent”,	or	(2)	“the	amount	of	a	factor	to	which	a	group	or	individual	was	exposed”	(132).	In	the	first	definition,	exposure	is	simply	regarded	as	an	attribute	that	is	either	present	or	absent	(“exposed”	or	“unexposed”).	This	simple	dichotomy	can	(and	should)	be	extended	to	include	considerations	of	exposure	quantity	as	outlined	in	the	second	definition,	since	exposure	measures	must	be	precise,	accurate	and	appropriate	for	the	study	design,	and	biologically	and	temporally	relevant	(133)	in	order	to	avoid	biased	associations	in	the	exposure-response	relationship	(134).		A	common	problem	in	studies	of	shift	work,	where	exposure	features	are	numerous	and	variable	within	and	between	individuals,	is	weak	exposure	assessment	that	does	not	appropriately	capture	characteristics	relevant	to	risk	of	health	outcomes	(20–22).	This	leads	to	non-differential	exposure	misclassification;	a	type	of	information	bias	where	misclassification	is	independent	of	the	outcome	and	leads	to	a	predictable	underestimation	of	health	effects	when	exposures	are	dichotomous,	and	an	underestimation	or	overestimation	of	health	effects	when	exposures	variables	have	more	than	two	categories	(135).		While	exposure	assessment	issues	confront	all	areas	of	shift	work	and	health	research,	studies	investigating	the	carcinogenicity	of	shift	work	provide	a	good	example	of	current	limitations	and	the	need	for	refinement.	In	2007,	the	International	Agency	for	Research	on	Cancer	(IARC)	classified	shift	work	that	involves	circadian	disruption	as	“probably	carcinogenic	to	humans”	(Group	2A),	based	on	limited	evidence	in	humans	working	night	 15 shifts,	and	sufficient	evidence	in	animal	studies	assessing	the	carcinogenicity	of	light	exposure	during	the	biological	night	(3).	The	working	group	responsible	for	this	classification	noted	“an	important	limitation	of	the	available	epidemiological	studies	is	that	there	have	not	been	clear	and	uniform	definitions	of	‘shift	work’	used”	(21).	For	instance,	most	large	epidemiological	studies	published	to	date	on	the	topic	of	shift	work	and	breast	cancer	have	used	coarse	definitions	of	exposure	to	shift/night	work	that	are	not	sufficient	to	properly	assess	the	risk	of	shift	work-related	circadian	disruption	on	cancer	risk	(20,21).	Examples	include	assessment	of	exposure	to	shift	work	and	night	work	primarily	based	on	sporadic	self-reports,	or	on	membership	in	a	work	sector	where	shift	work	involves	a	high	percentage	of	workers	(e.g.,	using	national	registries).	Reports	of	other	quantitative	and	qualitative	information	describing	shift	schedules	(such	as	number	of	night	shifts	worked	per	month	or	year,	number	of	consecutive	night	shifts,	direction	and	speed	of	rotation,	and	shift	length)	have	often	been	missing	or	highly	varied	(20).	Since	the	IARC	classification,	shift	work’s	causal	role	in	the	development	of	breast	cancer	continues	to	be	vigorously	discussed,	particularly	in	cases	where	coarse	exposure	assessment	is	perceived	to	promote	conclusions	of	“no	effect”	(136–142).	For	example,	a	recent	paper	by	Travis	et	al.	(136)	described	results	from	three	prospective	studies	published	since	the	IARC	review,	and	provided	an	updated	meta-analysis	of	findings	from	all	prospective	studies	of	shift	work	and	breast	cancer.	Based	on	their	findings,	the	authors	concluded	that	“classification	of	night	shift	work	as	a	probable	human	(breast)	carcinogen	is	no	longer	justified”	(136).	While	this	conclusion	was	accepted	by	some	(143,144),	a	number	of	researchers	published	critical	responses	to	Travis	et	al.’s	conclusions	(137–139,145),	pointing	primarily	to	the	likelihood	of	“severe	exposure	misclassification”	(138)	due	to	short	follow	up	times,	poorly	characterized	duration	and	intensity	of	exposure,	and	selection	bias	(137,138,140).		1.3 Rationale	and	objectives		Shift	work	is	a	common	working	arrangement	with	wide-ranging	implications	for	worker	health.	While	strong	exposure	assessment	is	an	essential	component	of	high-quality	 16 epidemiological	studies,	the	complexity	of	this	task	is	a	recognized	challenge	in	the	field	of	shift	work	and	health	research.	Various	methodological	challenges,	particularly	issues	with	exposure	assessment,	have	limited	the	development	of	high-quality	epidemiological	evidence	to	inform	strategies	to	mitigate	shift	work’s	negative	impacts	on	health	(21,23,80,146).	A	number	of	reviews,	reports,	and	commentaries	point	to	the	need	for	more	research	in	this	area	(20–23,32,62,147,148);	many	explicitly	emphasize	the	need	for	more	precise	analyzes	of	exposure	to	shift	work	in	epidemiological	studies	(20–23).		This	PhD	dissertation	aimed	to	generate	new	information	on	the	assessment	of	shift	work	exposure	in	epidemiological	research,	in	order	to	address	a	number	of	limitations	in	this	field.	It	includes	a	series	of	three	distinct	studies	focused	on	issues	concerning	the	measurement,	assignment,	and	determinants	of	shift	work	exposure,	as	follows:	1.3.1 Measuring	light-at-night	exposure	levels	and	characterizing	variability	in	shift	workers	Relationships	between	light	and	physiologic	responses	in	humans	and	animals	have	been	investigated	in	controlled	laboratory	experiments	(106,149–152).	This	has	led	to	the	understanding	that	exposure	to	light-at-night	has	negative	effects	on	human	biology	(153),	and	is	likely	a	strong	contributor	to	shift	workers’	increased	risk	for	various	biological	and	social	disruptions	(32,70,154,155).		While	controlled	setting	studies	can	be	very	useful	to	identify	relationships	between	shift	work-related	exposures	and	circadian	disturbances,	workplace-based	research	is	required	to	make	connections	between	scientific	findings	and	“real-world”	practices	(156).	An	empirical	understanding	of	occupational	exposures	to	light-at-night	is	needed	to	plan	efficient	sampling	strategies,	to	reduce	misclassification	and	attenuation	of	exposure-response	relationships	in	epidemiological	studies	(134),	and	to	target	at-risk	worker	groups	for	research	and	intervention	purposes	(157).	However,	an	investigation	of	light-at-night	exposure	levels	and	variability	within	and	between	shift	workers	has	not	been	conducted	to	date,	and	quantitative	data	describing	occupational	exposure	to	light-at-night	are	limited	in	number	and	quality	(20,62,158).	Furthermore,	there	are	no	standard	(health- 17 based)	light-at-night	exposure	metrics	available	for	use	in	epidemiological	studies	of	shift	work.	The	objective	of	Chapter	2	is	to	measure	personal	exposures	to	light-at-night,	and	to	assess	variability	and	metrics	of	exposure,	in	a	sample	of	emergency	services	and	healthcare	shift	workers	in	the	province	of	British	Columbia,	Canada.	1.3.2 Characterizing	the	impacts	of	exposure	assignment	in	epidemiological	studies	of	shift	work		Exposure	assignment	(the	application	of	exposure	categories	or	levels	to	study	subjects)	is	a	fundamental	consideration	in	an	epidemiological	study,	since	this	is	the	basis	for	comparing	health	outcomes	across	groups	with	differing	amounts	(or	types)	of	exposure	to	a	hazard.	The	common	use	of	coarse	exposure	indicators	(e.g.,	“day	worker”	versus	“night	worker”)	in	epidemiological	studies	ignores	a	number	of	potentially	important	exposure	characteristics	of	shift	work	(e.g.,	shift	timing,	rotation	frequency)	that	may	impact	on	health	(21).	This	trend	is	problematic,	since	coarse	exposure	assessment	and	assignment	can	produce	measurement	error	and	exposure	misclassification	within	groups	(134)	that	can	attenuates	effects	and	mask	true	exposure-response	relationships.	Progress	in	the	use	of	detailed	and	consistent	exposure	assignment	in	shift	work	research	has	been	widely	called	for,	to	improve	the	quality	of	epidemiological	evidence	in	this	area	(20–24).		The	objective	of	Chapter	3	is	to	characterize	the	impacts	of	exposure	assignment	on	the	relationship	between	shift	work	and	depression	in	a	national	sample	of	nurses	in	Canada,	by	assigning	and	comparing	exposure	indicators	with	varying	degrees	of	precision.	1.3.3 Identifying	determinants	of	workplace-level	shift	work	policies	and	practices	Modifiable	workplace-level	policies	and	practices	that	determine	the	nature	of	shift	work	exposures	may	contribute	to	the	quality	of	a	working	environment	for	shift	working	employees	(159).	Information	on	determinants	(e.g.,	workplace	size,	temporary	work,	and	employer	motivations)	of	workplace-level	shift	work	policies	and	practices	(e.g.,	work	scheduling,	lighting	policies,	and	health	education)	could	provide	useful	targets	for	 18 epidemiological	research	and	interventions.	Such	information	could	also	promote	a	better	understanding	of	the	barriers	and	facilitators	to	conducting	research	and	translating	evidence	into	best	practices	within	workplaces.		Empirical	research	into	determinants	of	workplace-level	shift	work	policies	and	practices	has	not	been	conducted	to	date.	For	example,	while	there	is	evidence	to	suggest	that	controlling	exposure	to	light	at	night	could	help	to	mitigate	the	negative	health	effects	and	risks	associated	with	shift	work	(160,161),	the	characteristics	of	workplaces	that	apply	nighttime	lighting	policies	have	not	been	defined.		The	objective	of	Chapter	4	is	to	describe	and	assess	determinants	of	workplace-level	shift	work	policies	and	practices	thought	to	affect	health,	across	a	range	of	industry	sectors,	in	a	sample	of	shift	work	employers	in	the	province	of	British	Columbia,	Canada.	 19 Chapter	2: Personal	light-at-night	exposures	and	components	of	variability	in	two	industry	sectors	where	shift	work	is	common	2.1 Introduction	Workplace	exposures	can	vary	across	a	number	of	factors,	such	as	work	content,	tasks	performed,	production	characteristics,	and	time	(162).	Information	on	exposure	variability	therefore	provides	an	important	scientific	basis	for	exposure	assessment	in	the	planning,	analysis,	and	interpretation	of	epidemiological	studies	(134,163).	However,	few	empirical	assessments	of	occupational	light-at-night	exposure	levels	have	been	conducted.	Exposure	to	light-at-night	and	circadian	disruption	arising	from	shift	work	can	result	in	a	lack	of	entrainment	between	natural	lighting	cues	and	workers’	internal	circadian	rhythms	(21),	producing	a	number	of	physiological	disruptions	(164).	Dose	dependence	between	light-at-night	intensity	and	human	physiologic	response	has	been	noted	in	laboratory	settings	(106,149),	but	there	has	been	limited	measurement	of	light-at-night	exposure	in	“real-life”	settings	(1),	and	the	variability	of	such	exposures	has	not	been	examined.	The	few	occupational	studies	that	have	been	conducted	to	measure	light-at-night	exposures	have	typically	assessed	effects	on	biomarkers	in	shift	workers	(20,165);	focusing	primarily	on	individual	workplaces	or	occupations	in	healthcare,	manufacturing,	and	public	safety	sectors	(62).		This	is	a	noteworthy	gap,	since	quantitative	measurements	are	essential	to	informing	questions	about	exposure	variability,	designing	future	studies,	and	permitting	the	exploration	of	exposure-response	relationships	(163).		In	epidemiological	studies,	exposure	measurements	are	used	to	estimate	and	assign	exposures	to	workers	(166).	This	can	take	two	forms:	individual	assessment,	where	exposures	are	measured	and	assigned	on	a	worker-by-worker	basis;	or	grouped	assessment,	where	subgroups	of	workers	are	formed	based	on	factors	such	as	occupation	or	work	area/tasks	that	may	affect	exposure.	In	the	latter	case,	a	relative	parameter	of	the	group’s	exposure	distribution	is	applied	to	all	individuals	within	the	group	(134).	Grouped	exposure	assessment	implicitly	assumes	that	all	individuals	within	groups	are	uniformly	exposed,	although	in	reality,	within-group	variability	is	common.	This	potential	source	of	 20 exposure	measurement	error	and	misclassification	is	frequently	ignored	in	classical	statistical	analyses	of	exposure-response	relationships	(134),	leading	to	the	possibility	of	attenuated	risk	estimates	(163).		An	exploratory	study	can	be	very	useful	for	gathering	the	information	needed	to	optimize	exposure	assessment	strategies	and	to	choose	an	exposure	indicator	that	reduces	the	potential	for	measurement	error	(134,163).	This	should	be	done	by	assessing	a	representative	sample	of	the	study	population,	with	participants	chosen	among	all	relevant	exposure	categories	and	repeated	random	sampling	of	individuals	(134).	An	analysis	of	variance	components	within	the	smaller	sample	can	then	be	conducted	prior	to	undertaking	a	larger	field	study.	This	involves	partitioning	the	total	variance	of	exposures	within	a	population	into	component	parts:	between-worker	variance,	a	measure	of	variation	in	average	exposure	levels	between	workers;	within-worker	variance,	a	measure	of	temporal	variation	within	individuals;	and	between-group	variance,	a	measure	of	variation	between	exposure	groups	established	by	the	investigator	(e.g.,	occupation,	workplace,	industry)	(167).	An	examination	of	variance	components	allows	for	an	assessment	of	the	degree	and	nature	of	exposure	variability	present,	and	consideration	of	the	best	exposure	indicator	(e.g.,	occupation,	workplace,	or	industry)	to	optimize	exposure	assessment	strategies	(163).	Such	data	can	thus	be	helpful	to	inform	decisions	about	who,	where,	and	how	many	people	to	measure	in	an	epidemiological	study	(168).	The	selection	of	an	appropriate	exposure	metric	(quantitative	description	of	an	exposure)	is	also	an	important	scientific	consideration	in	exposure	assessment,	since	observed	exposure-response	relationships	can	be	sensitive	to	the	metric	chosen	(134).	Currently	there	are	no	standard	(health-based)	metrics	available	for	use	in	epidemiological	studies	of	shift	work.	In	situations	where	the	most	biologically	relevant	exposure	metric	is	unknown,	the	assessment	of	multiple	measures	can	be	useful	(131).	This	approach	has	been	demonstrated	in	other	contexts,	such	as	the	assessment	of	various	exposure	metrics	for	peak	occupational	exposure	to	organic	solvents	(169).	An	evaluation	of	correlations	between	exposure	metrics	provides	a	better	understanding	of	the	limitations	and	strengths	of	data	to	evaluate	exposure-response	relationships	(163).	 21 2.1.1 Light	concepts		Light	is	most	frequently	defined	as	the	visible	portion	of	the	electromagnetic	spectrum	(i.e.,	wavelengths	between	400	and	780	nm)	(170).	Photoreceptors	(rods	and	cones)	in	human	eyes	collect,	interpret,	and	transpose	this	range	of	electromagnetic	waves	into	meaningful	visual	signals	in	the	brain.		Light	as	it	affects	the	visual	system	(photopic	illuminance)	is	the	basis	for	conventional	photometry	and	for	most	commercially	available	light	measurement	devices.	Photopic	illuminance	is	described	in	units	of	“lux”,	that	is,	the	density	of	light	falling	on	a	surface	(1	lux	=	1	lumen/m2,	where	“lumen”	is	the	unit	of	total	visible	light	emitted	by	a	source)	(171).	Simply	put,	lux	is	a	measure	of	light	intensity	as	perceived	by	the	human	eye.		Some	light	exposure	thresholds	have	been	proposed	for	human	biological	end	points.	Increasing	light	levels,	measured	in	terms	of	photopic	illuminance,	have	been	shown	to	increase	alertness	(172).	One	study	examining	the	dose-response	relationship	between	illuminance	and	subjective	alertness	found	that	room	light	of	~100	lux	elicited	half	of	the	maximum	alerting	response	(that	occurred	at	bright	light	of	9100	lux)	(173).	Increasing	light	levels	measured	in	terms	of	photopic	illuminance	also	affect	human	melatonin	levels,	with	endogenous	melatonin	suppression	beginning	at	light	levels	of	~30	lux	for	1	hour	(saturation	occurring	at	~1000	lux)	(174).	Therefore,	multiple	photopic	illuminance	metrics	of	biological	relevance	(median	lux,	90th	percentile	lux,	sum	of	minutes	≥	30	lux,	and	sum	of	minutes	≥	100	lux)	may	be	useful	to	characterize	shift	workers’	exposure	to	light-at-night	for	various	health	endpoints.		2.1.2 Study	rationale	and	objectives	Measurements	of	workplace	exposures	are	crucial	for	both	the	evaluation	of	possible	health	risks,	as	well	as	their	reduction	through	control	measures	(157).	Information	on	exposure	variability	provides	an	important	scientific	basis	for	exposure	assessment	in	the	planning,	analysis,	and	interpretation	of	epidemiological	studies	(134,163).		 22 For	example,	knowledge	about	exposures	is	useful	for	selecting	samples	with	maximal	exposure	variability,	thereby	increasing	sampling	efficiency	by	minimizing	the	number	of	participants	needed	to	achieve	a	certain	level	of	study	power	(175).	The	use	of	an	appropriate	grouping	strategy	for	assigning	exposure	indicators	can	also	reduce	the	likelihood	of	exposure	misclassification	and	distorted	exposure-response	relationships	(166).	Furthermore,	empirical	data	that	describe	light-at-night	exposures	can	be	used	to	identify	groups	of	shift	workers	at	greatest	risk	of	negative	health	outcomes,	for	use	in	targeted	research	and	workplace	prevention	efforts.		Quantitative	measurements	of	light-at-night	exposures	can	be	used	for	a	number	of	epidemiological	purposes.	However,	few	empirical	assessments	of	occupational	light-at-night	exposure	levels	and	variability	have	been	conducted.	The	current	study	measured	shift	workers’	exposure	levels	to	light-at-night	to	characterize	occupational	exposures	and	inform	future	exposure	assessment	methods.	It	aimed	to	answer	the	following	questions:	1) What	are	the	light-at-night	exposures	of	shift	workers	in	healthcare	and	emergency	services,	and	do	these	exposures	vary	by	industry	and	occupation?	2) What	are	the	components	of	variance	for	light-at-night	exposures	(within-worker,	between-worker,	and	between-group)	across	different	exposure	indicators	and	metrics?		3) What	are	the	correlations	between	various	light-at-night	exposure	metrics?	2.2 Methods	Data	on	light-at-night	exposures	were	collected	from	shift	workers	in	emergency	services	and	healthcare	industries	employed	by	the	Provincial	Health	Services	Authority,	within	the	province	of	British	Columbia,	Canada.	These	industries	employ	a	substantial	proportion	of	shift	workers	in	British	Columbia	(the	healthcare	sector	alone	represents	15%	of	all	night	and	rotating	night	workers	in	the	province	(10))	and	were	expected	to	represent	a	range	of	lighting	conditions	(68).	 23 2.2.1 Equipment	For	this	study,	personal	measurements	of	photopic	illuminance	were	collected	using	the	Daysimeter	(176),	a	small	light	monitoring	device	that	has	been	validated	(177)	and	used	in	prior	studies	to	examine	the	impacts	of	light	exposures	on	health	(178–180).	This	device	continuously	records	photopic	illuminance	levels	from	ambient	light	sources	using	an	integrated	circuit	light	sensor	array	(177,181).		2.2.2 Recruitment	and	data	collection	Sampling	was	conducted	between	October	2015	and	April	2016	in	Vancouver	area	ambulance	stations,	in	Vancouver	and	Victoria	area	emergency	dispatch	offices,	and	in	the	British	Columbia	Women's	Hospital	in	Vancouver,	Canada.	Participants	were	emergency	health	services	workers	(paramedics	and	dispatch	officers)	and	hospital	workers	(nurses,	security	guards,	patient	care	aides,	unit	clerks,	pharmacy	and	medical	laboratory	staff)	working	one	or	more	overnight	shifts	that	included	the	23:00	to	05:00	period.		Purposeful	sampling	was	conducted	to	capture	workers	in	a	range	of	occupations	and	environments	within	the	aforementioned	worksites.	To	permit	calculations	of	exposure	variance	between	and	within	workers,	the	recruitment	aim	was	to	sample	100	shift	workers,	with	repeated	measurements	for	up	to	2	shifts	per	worker.	At	minimum	5	shift	workers	were	recruited	at	each	worksite.	Additional	information	on	recruitment	is	available	in	Appendix	A.1.	The	study	coordinator	(A.	Hall)	was	present	on	site	at	the	beginning	of	each	participant’s	first	sampling	work	shift	to	deliver	the	light	sampling	device	and	administer	a	short	in-person	training	session.	Each	participant	was	instructed	to	wear	his	or	her	monitor	on	the	upper	chest	(suspended	from	a	lanyard)	for	the	entire	night	shift.	Photopic	illuminance	measurements	were	logged	at	a	rate	of	one	measurement	per	minute	and	downloaded	at	the	end	of	each	shift.	 24 2.2.3 Study	variables	This	study	examined	four	metrics	of	photopic	illuminance	between	23:00–05:00	hours	for	each	participant	shift.	These	were:	median	lux,	to	characterize	central	tendencies;	90th	percentile	lux,	to	characterize	peak	exposures;	sum	of	minutes	≥30	lux,	to	characterize	duration	of	exposure	that	may	suppress	melatonin	production;	and	sum	of	minutes	≥100	lux,	to	characterize	duration	of	exposure	that	may	elicit	half	of	the	maximal	alerting	response.			 2.2.4 Statistical	analyses	All	study	participants	worked	overnight	shifts	that	included	the	period	from	23:00	to	05:00,	with	the	exception	of	security	guards	who	started	their	shifts	at	midnight.	No	distinct	exposure	patterns	were	noted	across	security	guards’	shifts	by	hour;	therefore	measurements	from	each	security	guard’s	00:00-01:00	period	were	applied	to	the	23:00-00:00	period.	This	was	done	to	retain	security	guards	in	the	analytic	sample,	and	to	reduce	the	potential	for	underestimating	their	exposures	in	comparison	with	other	worker	groups	across	the	6-hour	exposure	window.	Summary	statistics	for	each	exposure	metric	were	calculated	by	industry	and	occupation.	Correlations	among	metrics	were	examined	using	Pearson	correlation	coefficients.		Data	distributions	for	each	exposure	metric	were	examined	using	frequency	distributions,	probability	plots,	and	skew	and	kurtosis	values.	The	median	and	sum	of	minutes	≥	100	lux	exposure	distributions	were	lognormal	and,	therefore,	log-transformed	for	the	components	of	variance	analyses.	The	90th	percentile	and	sum	of	minutes	≥	30	lux	exposure	distributions	were	approximately	normal	and	were	not	transformed.		Each	exposure	metric	was	grouped	by	industry	(2	groups),	by	workplace	(4	groups),	by	work	site	(9	groups)	and	by	occupation	(10	groups),	presented	in	Table	2.1.	To	identify	each	variance	component’s	relative	contribution	to	total	variance,	a	series	of	random	effects	(or	“null”)	models	was	generated	using	PROC	MIXED	in	SAS	(Version	9.4)	(182).	The	first	model	included	only	worker	as	a	random	effect	(Model	1);	four	subsequent	models	 25 examined	each	grouping:	worker	+	industry	(Model	2),	worker	+	workplace	(Model	3),	worker	+	worksite	(Model	4),	and	worker	+	occupation	(Model	5).	Table	2.1:	Summary	of	grouping	schemes	for	shift	workers’	personal	light-at-night	exposures	(n	participants,	n	measurements1)	Industry	 Workplace2	 Worksite3	 Occupation	Emergency		 Call	Centre	(19,	30)	 Call	Centre	-	Vancouver	(14,	22)	 Dispatch	Officer	(19,	30)	Services		(33,	47)				Various	(14,	17)	 Call	Centre	-	Victoria	(5,	8)	Various	-	Surrey	(9,	11)	Various	-	Delta	(5,	6)		Paramedic	(14,	17)	Healthcare		 Hospital	Unit	(58,	84)	 Hospital	Unit	–	LDR	(23,	31)	 Nurse	(38,	56)	(69,	105)	 Hospital	Other	(11,	21)	 Hospital	Unit	–	NICU	(27,	34)	 Lab	Technologist	(2,	4)		 	 Hospital	Unit	–	Ward	(12,	20)	 Lab	Assistant	(3,	6)			 	 Hospital	Other	–	Lab	(5,	10)		 Care	Aide	(11,	17)		 	 Hospital	Other	–	Circ	(6,	11)	 Security	Guard	(5,	9)		 	 	 Unit	Clerk	(4,	7)		 	 	 Pharmacist	(4,	4)			 		 		 Respiratory	Therapist	(2,	2)	2	Groups	 4	Groups	 9	Groups	 10	Groups	1	Maximum	number	of	measurements	per	participant	=	2	2	Various	=	ambulance	stations,	hospitals,	other	indoor	settings,	and	outdoor	settings;	Hospital	Unit	=	Labour	and	Delivery,	Neonatal	Intensive	Care,	or	General	Ward;	Hospital	Other	=	Laboratory	and	non-fixed	areas	throughout	hospital			3	Various	=	ambulance	stations,	hospitals,	other	indoor	settings,	and	outdoor	settings;	LDR	=	Labour	and	Delivery;	NICU	=	Neonatal	Intensive	Care	Unit;	Lab	=	Medical	Laboratory,	Circ	=	Circulating			2.3 Results	A	total	of	155	personal	full-shift	light-at-night	measurements	were	obtained	from	104	participants	over	45	nights.	Three	measurements	were	excluded	for	technical	(n=1)	or	compliance	(n=2)	reasons,	resulting	in	a	final	analytic	sample	of	102	participants	and	152	measurements.	Repeated	measurements	on	the	same	participant	(n	=	50,	or	49%)	were	performed	on	average	11	days	after	their	initial	shift	(range	of	1	to	42	days).		 26 Table	2.2:	Personal	light-at-night	exposure	averages	(assessed	in	the	23:00	to	05:00	period)			k	 n	 Median		(lux)	90th	percentile		(lux)	Sum	of	minutes		≥	30	lux	Sum	of	minutes		≥	100	lux		 	 	 	 	 	 	Shift	workers	(all)	 102	 152	 23	 73	 122	 28		 	 	 	 	 	 	Healthcare	(all)	 69	 105	 29	 88	 151	 36						Nurse	 38	 56	 22	 77	 120	 30						Care	Aide	 11	 17	 36	 143	 197	 58						Security	Guard	 5	 9	 28	 66	 153	 17						Unit	Clerk	 4	 7	 19	 43	 107	 3						Laboratory	Assistant	 3	 6	 67	 121	 262	 83						Laboratory	Technologist	 2	 4	 60	 112	 292	 69						Pharmacist	 4	 4	 30	 97	 178	 40						Respiratory	Therapist	 2	 2	 14	 38	 94	 6		 	 	 	 	 	 	Emergency	medical	services	(all)	 33	 47	 11	 40	 59	 11						Dispatch	Officer	 19	 30	 8	 12	 15	 1						Paramedic	 14	 17	 18	 88	 137	 28			 		 		 		 		 		 		k	=	number	of	workers	n	=	(number	of	workers)	x	(measurements	shifts	per	workers)	=	total	worker-shifts	2.3.1 Descriptive	summaries	Average	light-at-night	exposures	for	the	23:00-05:00	period	over	all	measurement	shifts	are	presented	in	Table	2.2	by	industry	and	occupation.	Laboratory	workers	(technologists	and	assistants)	and	care	aides	displayed	the	highest	levels	for	all	light	exposure	metrics	among	shift	worker	occupations	in	the	study.	Emergency	dispatch	officers	displayed	the	lowest	levels	for	all	light	exposure	metrics	among	shift	workers.	Correlations	between	light-at-night	exposure	metrics	are	presented	in	Table	2.3.	Correlations	were	generally	high	between	all	metrics	(>0.657),	but	the	highest	correlations	were	observed	between	the	median	lux	and	sum	of	minutes	≥	30	lux	metrics	(r	=	0.893),	and	between	the	90th	percentile	lux	and	sum	of	minutes	≥	100	lux	metrics	(r	=	0.810).	 27 Table	2.3:	Correlations	between	4	light-at-night	exposure	metrics	(n	=	152	observations)	assessed	in	the	23:00	to	05:00	period			Median		(lux)	90th		percentile		(lux)	Sum	of	minutes		≥	30	lux		Sum	of	minutes		≥	100	lux	Median	lux	 1.000	 	 	 	90th	percentile	lux	 0.657	 1.000	 	 	Sum	of	minutes	≥	30	lux	 0.893	 0.759	 1.000	 	Sum	of	minutes	≥	100	lux		 0.755	 0.810	 0.691	 1.000			 	 	 		2.3.2 Components	of	variance	Within-worker,	between-worker,	and	between-group	variances	proportions	for	each	grouping	scheme	across	all	exposure	metrics	are	presented	in	Table	2.4.	Between-group	variance	was	large	for	all	exposure	metrics	and	increased	as	grouping	precision	increased	(moving	from	industry	to	occupation).	Within-worker	variance	was	small	for	all	exposure	metrics	and	groupings,	with	the	lowest	values	observed	for	median	lux.	2.4 Discussion	This	study	measured	light-at-night	exposures	in	a	sample	of	emergency	services	and	healthcare	shift	workers.	Personal	light	measurements	were	collected	from	participants	across	industries,	workplaces,	work	units,	and	occupations,	and	repeated	measurements	were	obtained	on	two	separate	night	shifts	for	approximately	half	of	participants.	2.4.1 Light-at-night	exposure	levels		Average	light-at-night	exposures	for	the	23:00-05:00	period	varied	significantly	across	industry	and	occupation.	In	the	healthcare	industry,	the	highest	average	levels	for	all	exposure	metrics	were	observed	in	laboratory	workers	(technologists	and	assistants)	and	the	lowest	average	levels	for	all	exposure	metrics	were	observed	in	respiratory	therapists	and	unit	clerks.	These	results	are	consistent	with	recommended	light	levels	for	visual	tasks	 28 associated	with	laboratory	work	activities	(between	200	and	2000	lux)	and	recommended	light	levels	in	and	around	nursing	stations	(ranging	from	20	lux	in	nighttime	corridors	up	to	500	lux	in	general	areas),	where	unit	clerks	and	respiratory	therapists	in	this	study	spent	much	of	their	time	(68)	(note	that	photopic	illuminance	measured	at	the	cornea,	or	as	proxy	for	the	cornea,	is	substantially	lower	than	photopic	illuminance	measured	on	the	horizontal	plane	for	lighting	applications).	Table	2.4:	Proportions	of	variance	in	4	light-at-night	exposure	metrics	across	grouping	schemes,	accounted	for	by	between-group,	between-worker,	and	within-worker	components	(n	=	152	observations)	(assessed	in	the	23:00	to	05:00	period)			Median	(lux)1	90th	percentile	(lux)	Sum	of	minutes		≥	30	lux		Sum	of	minutes		≥	100	lux1	Model	1:	No	grouping	-	Worker	random	effect	only						Between-worker	variance	(%)	 97.3	 90.4	 92.7	 93.1						Within-worker	variance	(%)	 3.7	 10.6	 8.3	 7.9	Model	2:	Industry	(2	groups)						Between-industry	variance	(%)	 91.4	 62.8	 64.1	 62.7						Between-worker	variance	(%)	 5.6	 25.8	 27.6	 29.5						Within-worker	variance	(%)		 3.0	 11.4	 8.4	 7.8	Model	3:	Workplace	(4	groups)						Between-workplace	variance	(%)	 93.2	 70.7	 78.9	 75.3						Between-worker	variance	(%)	 4.1	 18.8	 14.6	 18.1						Within-worker	variance	(%)		 2.7	 10.5	 6.5	 6.6	Model	4:	Worksite	(9	groups)						Between-worksite	variance	(%)	 93.6	 71.7	 79.9	 76.5						Between-worker	variance	(%)	 3.6	 17.4	 13.7	 17.1						Within-worker	variance	(%)		 2.8	 10.8	 6.4	 6.4	Model	5:	Occupation	(10	groups)						Between-occupation	variance	(%)	 94.6	 80.1	 85.8	 81.5						Between-worker	variance	(%)	 3.1	 10.7	 9.1	 12.7						Within-worker	variance	(%)		 2.3	 9.2	 5.1	 5.9	1	Log-transformed	In	the	emergency	services	industry,	dispatch	officers	working	in	call	centres	displayed	the	lowest	average	lighting	exposures	across	all	light-at-night	metrics	for	shift	workers.	This	finding	was	expected,	since	study	participants	in	these	work	environments	expressed	a	 29 preference	for	low	general	lighting	levels	during	their	overnight	shifts.	However,	low	lighting	levels	may	not	occur	in	night-time	office	settings	beyond	those	studied	here,	given	a	range	of	recommended	lighting	levels	for	such	environments	(68)	and	personal	differences	in	preferred	lighting	levels	(183).	Depending	upon	the	goals	of	the	study	or	measurement	strategy,	light-at-night	exposure	measurements	could	be	used	to	identify	shift	workers	at	greatest	risk	of	fatigue	(i.e.,	where	light	exposures	are	consistently	low)	or	of	melatonin	suppression	(i.e.,	where	light	exposures	are	consistently	high	during	the	natural	darkness	period).	Previous	studies	provide	some	opportunities	for	comparison	with	this	study’s	light-at-night	exposure	level	results.	Median	light	levels	of	38	lux	between	the	hours	of	00:00	and	05:00	were	found	in	one	Spanish	study	of	workers	from	a	variety	of	industries	(including	hospitals)	(184).	In	Canada,	researchers	reported	a	mean	light	intensity	of	7.2	lux	(SD	6.73)	between	the	hours	of	00:00	and	05:00	for	night	shift	nurses	in	one	study	(21)	and	a	maximum	of	37.2	lux	for	the	same	time	period	among	nurses	in	a	separate	study	(185).	In	the	current	study,	the	average	median	exposure	for	nurses	(sampled	in	labour	and	delivery,	neonatal	intensive	care,	and	low-acuity	wards)	between	the	hours	of	23:00	–	05:00	was	22	lux,	and	for	hospital	workers	overall	the	median	was	29	lux.	These	results	compare	fairly	well	with	the	aforementioned	studies,	since	distributions	of	light-at-night	measurements,	similar	to	other	exposure	measurements,	are	log	normally	distributed	(186);	therefore	higher	median	lux	levels	are	expected	when	compared	to	mean	levels.	Regardless	of	work	schedule,	most	people	living	in	the	modern	age	are	exposed	to	light-at-night	to	some	extent.	Despite	the	unavoidable	problem	of	assessing	health	risks	in	“exposed”	shift	workers	against	“unexposed”	day	workers	(who	are	not	truly	unexposed)	(32),	it	has	been	reasoned	that	night	workers	would	have	substantially	higher	exposures	to	light-at-night	(21).	To	test	this	assumption	and	provide	an	exploratory	comparison	with	the	shift	workers’	light-at-night	exposures,	the	current	study	collected	light-at-night	measurements	from	a	small	convenience	sample	of	13	daytime	office	workers	employed	within	the	same	health	authority	as	the	participating	shift	workers.	Very	low	average	light	levels	were	observed	for	all	exposure	metrics	in	this	group	(median	lux	=	1;	90th	percentile	 30 lux	=	2;	Sum	of	minutes	≥	30	lux	=	2;	Sum	of	minutes	≥	100	lux	=	1).	This	compares	closely	with	other	findings	of	low	intensity	light-at-night	exposures	in	daytime	workers,	such	as	in	a	study	of	school	teachers	(99%	of	measurements	taken	between	the	hours	of	00:00-04:00	<	1	lux)	(179)	as	well	as	in	daytime	working	nurses	(between	the	hours	of	00:00-05:00,	mean	lighting	levels	=	2.48	lux)	(187).	These	results	support	the	notion	that	on	average,	shift	workers’	light-at-night	exposures	in	even	the	lowest	categories	(represented	in	this	study	by	dispatch	officers)	exceed	“background”	light-at-night	exposures	of	daytime	working	populations.	2.4.2 Components	of	variance		Exposure	assessment	strategies	based	on	individual	versus	grouped	measurement	strategies	each	have	benefits	and	disadvantages.	Individual-based	approaches	generally	increase	precision	of	exposure-response	relationships,	but	at	the	expense	of	introducing	bias	and	effect	attenuation	(175).	Grouped	exposure	assessment	is	commonly	used	in	occupational	epidemiological	studies	since	data	on	individual	exposures	are	often	missing	(in	the	case	of	historical	exposures)	or	can	be	impractical/costly	to	collect	(163).	In	addition	to	efficiency	benefits,	grouped	exposure	assessment	can	provide	reasonably	unbiased	estimates	of	exposure-response	relationships,	since	overestimation	or	underestimation	of	some	group	members’	exposures	produces	less	attenuation	of	an	exposure-response	relationship	than	would	be	produced	when	each	individual	is	assigned	the	mean	of	their	own	exposure	measurements	(a	Berkson	error	structure)	(175,188).	This	being	said,	the	validity	of	a	grouped	exposure	assessment	approach	relies	on	the	assumption	that	individuals	within	assigned	groups	are	similarly	exposed	(163),	and	that	assigned	groups	provide	sufficient	exposure	contrast	(189).	This	assumption	is	not	always	true	(163),	as	observed	for	a	variety	of	occupational	exposures	where	within-worker	and	between-worker	exposure	variability	has	been	high	relative	to	between-group	variability	for	some	exposure	groupings	(162,190–192).		Within-worker	exposure	differences	are	primarily	driven	by	temporal	variability	(134),	therefore	the	collection	of	repeated	measurements	for	individual	workers	is	needed	to	 31 examine	the	degree	and	nature	of	such	variability	(163).	The	current	study	collected	multiple	measurements	for	49%	of	participants,	and	observed	low	within-worker	variability	(ranging	from	2.3%	up	to	11.4%	of	total	variance)	for	all	light-at-night	exposure	metrics.	This	differs	from	other	studies	that	have	demonstrated	high	within-worker	variability	for	occupational	exposures	such	as	magnetic	fields	(190),	dust	(193),	and	mercury	(192).	The	low	within-worker	variability	in	light-at-night	exposure	observed	in	this	study	is	likely	explained	by	the	fact	that	many	of	the	nighttime	work	environments	assessed	demand	few	lighting	changes	over	time	(e.g.,	call	centres,	hospital	laboratories)	and	light-at-night	levels	are	likely	less	affected	by	specific	work	tasks	or	activities	compared	to	other	occupational	exposures.	Another	possible	reason	for	the	low	within-worker	variability	observed	is	that	comparisons	of	measurements	collected	over	longer	periods	(such	as	a	full	shift)	are	more	likely	to	be	similar,	compared	to	repeated	shorter	duration	measurements	(168,194).		Between-worker	differences	may	arise	from	differences	in	job	characteristics	(e.g.,	work	location,	tasks)	and	personal	characteristics	(e.g.,	sex,	age,	body	size,	techniques	used	to	carry	out	tasks).	As	workers	are	consolidated	into	smaller	classification	groups	with	fewer	subjects,	between-worker	variability	typically	decreases	and	between-group	variability	increases	(195).	This	study’s	light-at-night	exposure	variance	components	reflect	this,	with	between-worker	variance	decreasing	as	classification	groups	become	more	precise	(from	large	industry	groupings	to	smaller	occupation	groupings).	Between-group	variance	was	lowest	for	industry	(2	groups,	33-69	workers	per	group)	and	highest	for	occupation	(10	groups,	2-38	workers	per	group)	for	all	exposure	metrics.		In	this	study,	within-worker	variance	was	low	compared	to	both	between-worker	and	between-group	variance.	The	greatest	contrast	between	groups	was	observed	for	the	occupational	grouping	(across	all	exposure	metrics	assessed),	and	for	the	median	lux	metric	(across	all	exposure	groupings	assessed).	The	median	exposure	metric	for	photopic	illuminance	exhibited	the	greatest	between-group	contrast	for	all	exposure	groupings,	with	up	to	94.6%	of	variance	explained	by	differences	across	occupations.	The	90th	percentile,	sum	of	minutes	≥	30	lux,	and	sum	of	minutes	≥	100	lux	exposure	metrics	also	showed	good	 32 contrast	for	the	occupation	grouping,	with	between-group	differences	explaining	80.1%,	85.8%,	and	81.5%	of	variance,	respectively.		Since	residual	classical	error	and	attenuation	in	exposure-response	are	minimized	when	the	ratio	of	between-group	to	within-group	variability	is	large	(134,188),	the	high	between-group	variability	observed	for	all	exposure	groupings	assessed	in	this	study,	particularly	when	using	the	median	exposure	metric,	shows	promise	for	future	sampling	strategies	and	exposure	assignment	in	epidemiological	studies.	The	information	required	to	develop	such	groupings	(e.g.	occupation)	is	relatively	easy	to	access	and	may	prove	to	be	a	consistently	efficient	way	of	grouping	light-at-night	exposures	to	minimize	within-group	heterogeneity	and	therefore	reduce	exposure	measurement	error.	The	use	of	high-level	groupings	is	also	an	important	consideration	regarding	measurement	burden,	since	night	work	is	prevalent	and	individual-based	assessments	are	not	always	practical	due	to	financial	or	logistical	constraints	(166).	2.4.3 Exposure	metrics	The	selection	of	an	appropriate	exposure	indicator	for	use	in	epidemiological	studies	has	implications	for	observed	associations	between	exposure	and	disease	(166),	however	this	decision	is	often	not	straightforward	(131,166).	Since	toxicological	mechanisms	linking	exposure	to	health	outcome	are	not	always	fully	understood,	one	recommended	strategy	is	to	evaluate	multiple	metrics,	in	order	to	increase	the	chance	that	the	appropriate	one	will	be	used	(131).	In	this	study,	multiple	light-at-night	exposure	metrics	with	potential	biological	relevance	for	epidemiological	studies	were	constructed,	and	their	correlations	were	assessed.	The	finding	of	moderate	to	high	correlations	between	the	four	light-at-night	exposure	metrics	(ranging	from	r	=	0.66	to	r	=	0.89)	indicates	some	consistency	in	exposures	over	a	work	shift	and	suggests	that	a	number	of	metrics	may	be	appropriate	to	assess	similar	aspects	of	light-at-night	exposure,	particularly	for	highly	correlated	metrics	(e.g.,	use	of	median	lux	versus	sum	of	minutes	≥	30	lux,	or	use	of	90th	percentile	lux	versus	sum	of	minutes	≥	100	lux).	Additional	studies	should	be	conducted	to	assess	correlations	between	light-at-night	 33 exposure	metrics	and	compare	them	to	those	observed	in	the	current	study.	Although	multiple	metrics	of	exposure	may	be	used	in	situations	where	disease	mechanisms	are	unclear,	the	choice	of	“best”	exposure	metrics	should	be	based	as	much	as	possible	on	the	conceptual	nature	of	the	research	question,	and	on	biological	considerations	of	relationships	between	the	exposure	and	the	outcome	in	question	(21,196,197).	For	example,	cumulative	exposure	measures	are	commonly	used	in	chronic	disease	studies	whereas	short	duration	(peak)	exposure	measures	are	often	most	appropriate	to	assess	acute	effects	(169).	A	summary	measure,	such	as	median	level,	may	be	the	most	relevant	exposure	metric	for	studies	examining	the	long-term	effects	of	light-at-night	and	circadian	disruption	on	health	(21).	The	amount	of	time	spent	below	or	above	a	given	illumination	threshold	(172,197)	could	be	a	more	relevant	metric	for	studies	examining	light-at-night’s	effects	on	alertness	and	performance.	Despite	uncertainties	about	the	“best”	metrics	to	use	in	epidemiological	studies,	researchers	are	afforded	a	number	of	options	in	this	regard,	since	current	exposure	monitoring	equipment	is	relatively	simple	to	use	and	permits	the	collection	of	light	measurements	over	hours	or	even	days.	2.4.4 Strengths	and	limitations	In	this	study,	the	measurement	of	light-at-night	exposures	in	shift	workers	focused	only	on	nighttime	exposures	at	work.	This	could	be	viewed	as	a	limitation,	since	an	individual’s	exposure	to	light	throughout	the	24-hour	period	can	affect	their	circadian	entrainment	(and	disruption)	(198,199).	Other	research	examining	24-hour	patterns	of	light	exposure	in	shift	workers	has	been	conducted	(178,180)	to	examine	the	broader	impacts	of	light	on	health.	The	objective	of	this	study	was	to	provide	baseline	data	on	light-at-night	exposures	and	exposure	variability	at	work,	using	various	exposure	grouping	schemes	and	metrics.	This	information	serves	a	different	purpose,	such	as	to	inform	sampling	strategies	that	maximize	exposure	contrast	in	occupational	studies	of	nighttime	light	exposures,	or	to	develop	exposure	indices	that	take	light-at-night	exposure	variability	into	account	(e.g.,	assessing	injury	rates	of	shift	workers	employed	across	a	variety	of	lighting	environments).	An	additional	benefit	to	assessing	light-at-night	exposure	in	the	workplace	is	that	broadly	 34 applied	workplace	interventions,	such	as	lighting,	offer	the	opportunity	for	systematic	implementation	and	assessment	for	use	in	research.		The	recruitment	aim	of	this	study	was	to	capture	measurements	from	a	variety	of	occupations	and	work	environments	within	two	industries	where	shift	work	is	common.	Given	the	ubiquitous	nature	of	light,	it	is	unlikely	that	this	study’s	results	were	affected	by	non-response	bias	(workers	with	higher	or	lower	light-at-night	exposure	being	more	or	less	likely	to	participate).	However,	small	measurement	numbers	within	strata,	or	not	enough	strata,	may	have	limited	the	amount	of	light-at-night	variability	captured.	Future	studies	should	be	conducted	to	assess	exposures	in	other	industries,	work	sites,	and	occupations,	where	exposure	levels	and	characteristics	within	and	between	workers	may	differ.	Since	the	longest	period	between	repeated	measurements	in	this	study	was	42	days,	within-worker	variability	in	light-at-night	exposures	could	increase	over	longer	time	periods	(e.g.,	due	to	occupational,	architectural,	and	engineering	changes).	It	should	also	be	noted	that	only	49%	of	participants	were	repeatedly	measured,	with	a	maximum	of	two	full-shift	samples	obtained	per	participant.	Since	repeated	measurement	rates	varied	across	occupations	(0%	for	pharmacists	and	respiratory	therapists,	up	to	100%	for	laboratory	workers),	these	estimates	may	not	reflect	the	full	extent	of	within-worker	variability	for	all	occupations	reported.	Despite	this	limitation,	it	should	be	noted	that	the	nighttime	lighting	environments	for	occupations	that	lacked	repeated	measurements	(pharmacists	and	respiratory	therapists	in	neonatal	intensive	care)	appeared	to	be	equally	or	less	dynamic	compared	to	the	nighttime	lighting	environments	for	other	occupations	where	a	high	number	of	repeated	samples	were	obtained	(e.g.,	care	aides	and	nurses).	A	strength	of	this	study’s	sampling	strategy	is	that	data	were	collected	from	shift	workers	not	commonly	captured	in	epidemiological	research	(e.g.,	laboratory	staff,	unit	clerks,	security	guards)	in	addition	to	more	commonly	assessed	occupations	(e.g.,	nurses,	paramedics).	These	findings	do	not	necessarily	apply	to	other	shift	work	industries	and	workplaces	where	lighting	environments	may	be	different	or	more	temporally	varied.	If	a	wider	range	of	workplaces	were	included,	total	exposure	variability	would	likely	increase,	 35 and	the	relative	contribution	of	each	variability	component	could	change.	In	practical	terms	the	latter	issues	may	be	moot,	since	prior	to	any	epidemiological	study	a	small	pre-investigation	of	exposure	variability	is	recommended	to	guide	an	optimal	exposure	assessment	strategy	and	reduce	the	potential	for	measurement	error	(134,163).	The	mixed	modelling	methods	used	to	assess	components	of	variance	incorporate	several	assumptions.	It	was	assumed	that	light-at-night	measurements	were	normally	distributed	and	independent,	and	that	within-subject	variance	was	equal	for	each	worker.	Between-subject	variance	was	also	assumed	to	be	equal	for	all	groups	of	workers.	The	distributions	of	light-at-night	metrics	assessed	in	this	study	were	either	approximately	normal	or,	in	the	case	of	lognormal	distributions,	were	log-transformed	prior	to	analysis.	Participants	were	selected	using	a	stratified	approach	with	random	sampling	within	strata,	and	repeated	measurements	were	taken	11	days	apart	on	average,	so	the	majority	of	observations	may	be	assumed	to	be	independent.	Within-group	and	within-worker	variability	in	light-at-night	measurements	was	not	perfectly	equal	across	groups	or	workers,	however	the	grouping	variables	accounted	for	the	majority	of	variability	in	all	cases.	2.5 Conclusions	This	study	represents	a	first	step	to	evaluating	personal	light-at-night	levels	and	variability	in	industries	where	shift	work	is	common.	Results	demonstrate	the	feasibility	of	assessing	full-shift	light-at-night	exposures	in	a	variety	of	shift	workers,	and	indicate	that	light-at-night	exposures	in	shift	workers	exceed	those	of	day	workers.	An	examination	of	exposure	variability	components	suggests	that	high-level	light-at-night	exposure	indicators	(e.g.,	grouping	at	the	level	of	occupation)	could	provide	a	simple	yet	relatively	precise	way	of	characterizing	individual	light-at-night	exposures	in	future	epidemiological	studies	of	shift	work.	A	number	of	biologically-based	light-at-night	exposure	metrics	were	considered	and	may	be	useful	in	future	epidemiological	studies,	including	median	lux,	90th	percentile	lux,	sum	of	minutes	≥	30	lux,	sum	of	minutes	≥	100	lux.	The	most	closely	correlated	measures	were	median	lux	and	sum	of	minutes	≥	30	lux	(r	=	0.893),	and	90th	percentile	lux	and	sum	of	minutes	≥	100	lux	(r	=	0.810);	the	selection	of	“best”	metric	depends	upon	the	conceptual	nature	of	the	research	question	and	its	specific	biological	hypotheses.	Given	the	 36 early	stage	of	epidemiological	research	into	the	effects	of	light-at-night	exposure	on	health,	the	relative	ease	of	collecting	exposure	measurements,	and	the	need	to	clarify	mechanisms	linking	exposures	to	health	outcomes,	researchers	may	wish	to	investigate	multiple	exposure	indices	and	metrics	in	future	studies.			 37 Chapter	3: Examining	the	impacts	of	exposure	assignment	in	a	study	of	shift	work	and	depression	among	nurses	3.1 Introduction				“Exposure	misclassification”	describes	an	error	whereby	the	assignment	of	an	exposure	level	or	category	to	an	individual	does	not	appropriately	reflect	their	true	exposure	to	a	given	agent.	Many	studies	investigating	shift	work	have	used	crude	exposure	categories	when	assessing	health	risk,	such	as	the	binary	“shift	work-exposed”	or	“not	shift	work-exposed”	(20).	Such	categorizations	ignore	a	number	of	exposure	characteristics	(e.g.,	shift	timing,	intensity,	rotation	frequency)	that	may	have	a	bearing	on	health	outcomes	(21,23).		This	is	problematic,	since	crude	exposure	assessment	and	assignment	can	produce	measurement	error	and	exposure	misclassification	within	groups	(134),	often	leading	to	low	exposure	contrast	and	attenuation	of	true	effects.	The	impact	of	exposure	misclassification	is	illustrated	by	a	large	Swedish	cohort	study	that	examined	the	effects	of	shift	work	on	the	risk	of	myocardial	infarction	(200).	When	shift	work	was	defined	as	“not	day	work”,	the	Standardized	Mortality	Ratio	(SMR)	was	115%	(95%	CI:	104-126),	whereas	when	shift	work	was	defined	as	“night	shift”,	the	SMR	increased	to	148%	(95%	CI:	112-191)	(23).		Despite	this	evidence	and	other	calls	to	dissuade	the	practice	(21,22,24,80),	coarse	categorizations	are	still	commonly	used	to	describe	exposures	in	studies	of	shift	work	and	health.	In	an	effort	to	further	clarify	the	impact	of	exposure	misclassification	in	epidemiological	studies	of	shift	work,	the	current	investigation	sought	to	examine	associations	between	shift	work	and	depression.		Major	depressive	disorder	is	experienced	by	5%	of	the	Canadian	adult	population	annually	(201)	and	represents	a	significant	source	of	disability	and	economic	burden	(202).	While	shift	work	has	been	linked	to	an	increased	risk	of	mental	disorders	such	as	anxiety	and	depression	(16,96,101),	these	relationships	and	their	explanatory	mechanisms	have	not	been	well	characterized.	This	topic	requires	further	investigation,	since	there	is	increasing	 38 interest	in	the	prevention	of	depressive	disorders	globally	(203)	and	workplace	risk	factors	may	provide	a	useful	target	for	depression	strategies	(204).		3.1.1 Hypothesized	pathways	between	shift	work	and	depression	Mood	in	healthy	individuals	can	vary	across	the	24	hour	cycle	due	to	a	combination	of	circadian	and	sleep	influences;	typically	showing	a	deterioration	in	evening	compared	with	morning	hours	(31,123,205).	A	number	of	physiological	and	social	pathways	may	link	shift	work	exposure	with	pathological	mood	and	depression	outcomes,	including	exposure	to	light	at	night,	disrupted	circadian	rhythms	(including	sleep	and	melatonin)	and	altered	social	functioning	(206,207).	Exposure	to	light	at	“abnormal”	times	in	the	circadian	cycle	has	been	associated	with	depressive-like	behaviours	in	mice	(208,209)	and	mood	disorders	in	humans,	such	as	seasonal	affective	disorder	and	major	depression	(31).	While	it	is	established	that	exposure	to	light	plays	an	important	role	in	modulating	mood,	the	mechanisms	behind	such	relationships	are	not	well	understood	(210).	Proposed	pathways	between	light	exposure	and	mood	disorders	include	light’s	effects	on	the	timing	of	circadian	cycles	(for	example,	circadian	alterations	in	melatonin	release	are	implicated	in	a	number	of	physiological	processes	related	to	mood,	circadian	entrainment,	and	sleep	(50,53)),	as	well	as	light’s	more	direct	impacts	on	sleep	via	the	eyes’	retinal	receptors	(210).		Sleep	disturbances	are	the	most	widely	reported	circadian	disruptions	associated	with	depression	(31);	individuals	suffering	from	insomnia	are	more	likely	to	have	a	depressive	illness,	with	longitudinal	research	showing	that	persistent	insomnia	is	associated	with	new	depressive	episode	onset	(121).	Sleep	disturbances	in	shift	workers	are	well	documented	(13,61,112–114),	with	termination	of	shift	work	associated	with	decreased	sleep	disturbances	and	working	shifts	associated	with	increased	future	sleep	disturbances	(112).	Working	night	shifts	and	early	morning	shifts	appears	to	be	most	strongly	associated	with	acute	sleep	loss	(211);	further,	exposure	to	light-at-night	may	directly	suppress	sleep	(212)	and	disrupt	the	normal	sleep-wake	cycle	(153).	 39 The	social	zeitgeber	theory	postulates	that	stressful	life	events	may	trigger	depressive	episodes	by	causing	disruptions	in	social	routines,	leading	to	altered	biological	rhythms	(207).	Shift	work	has	been	linked	to	disruptions	in	social	and	family	life	patterns	(213),	and	dissatisfaction	with	work-life	balance	is	particularly	pronounced	in	shift	workers	reporting	split,	on	call	or	casual,	or	irregular	schedules	(2).	Since	a	lack	of	social	support	is	an	important	risk	factor	for	depressive	disorder	(214),	shift	work’s	effects	on	social	patterns	may	represent	an	important	pathway	to	depression.	3.1.2 Prior	research	into	shift	work	and	depression	Evidence	linking	shift	work	with	an	increased	risk	of	negative	mental	health	outcomes	is	mixed,	perhaps	owing	somewhat	to	the	positive	effects	of	work	in	general	for	mental	health	and	well-being	(215)	versus	the	negative	effects	of	sub-optimal	work	conditions	for	mental	health	(216).	However,	few	studies	have	sought	to	investigate	shift	work’s	effects	on	depression,	and	as	identified	herein,	even	fewer	have	used	refined	measures	of	shift	work	for	investigating	exposure-response	relationships.	A	British	household	panel	survey	found	that	working	varied	shifts	(with	no	usual	pattern)	was	related	to	increased	risks	of	anxiety/depression	and	minor	mental	disorders	in	females	over	time,	but	years’	duration	of	night	work	did	not	show	strong	effects	for	these	outcomes;	whereas	for	males,	years’	duration	of	night	shifts,	but	not	varied	shifts,	increased	risk	of	anxiety/depression	and	minor	mental	disorders	over	time	(16).	Another	prospective	cohort	study	found	an	increased	risk	of	depressive	disorder	in	current	or	former	female	shift	workers	versus	never	shift	workers	in	adjusted	analyses;	no	strong	relationships	were	noted	in	current	or	former	male	shift	workers	(101).	Finally,	a	short	prospective	cohort	study	of	nurses	found	no	strong	relationships	between	working	nights	and	working	rotating	shifts	and	independent	measures	of	anxiety	and	depression	(78).	In	a	cross-sectional	study	of	hotel	workers,	an	elevated	risk	of	depression	was	observed	in	all	shift	worker	categories	relative	to	regular	day	workers	(217).	In	another	small	cross-sectional	study	of	nursing	assistants,	regular	evening	and	rotating	shift	work	were	 40 associated	with	an	increased	risk	of	depressive	disorder	whereas	night	work	showed	a	protective	effect	(with	all	confidence	intervals	spanning	“1”)	(218).	It	should	be	noted	that	the	exposure	indicators	assigned	in	all	of	these	studies	were	either	dichotomous:	“night	work”	or	“varied	shifts”	versus	“none”	(16),	“shift	worker”	versus	“never	shift	worker”	(101),	or	incorporated	considerations	of	shift	timing	only:	“permanent	night	shift”,	“day	and	evening	rotating	shifts”,	or	“day,	evening,	and	night	rotating	shifts”	versus	“permanent	day	shift”	(78);	“rotating	day	shift”,	“rotating	night	shift”,	or	“fixed	night	shift”	versus	“fixed	day	shift”	(217),	“evening	shift”,	“night	shift”,	or	“rotating	shift”	versus	“day	shift”	(218).	Furthermore,	the	frequency	of	shift	rotations	(referring	to	the	number	of	shift	changes	in	a	given	time	period)	was	not	incorporated	into	the	exposure	assignment	of	any	of	these	studies.	This	is	an	interesting	gap,	since	shift	rotation	frequency	could	be	an	important	exposure	characteristic	to	assess	in	studies	of	shift	work	and	mental	health.	A	high	shift	rotation	frequency	increases	the	likelihood	of	“quick	returns”	between	shifts	(i.e.	changeovers	from	morning/day	to	night	shifts,	night	to	evening	shifts,	or	evening	to	morning/day	shifts	where	11	hours	or	less	of	free	time	is	scheduled	between	shifts	(219)).	Quick	returns,	identified	as	a	“big	problem	in	life”	by	some	shift	workers	(220),	have	been	associated	with	poor	sleep	quality,	increased	fatigue,	and	disrupted	social	relationships	(221).	High	shift	rotation	frequency	may	also	reflect	precarious	employment	situations	(222)	that	have	been	associated	with	increased	social	disruption,	stress,	and	depressive	symptoms	(223–225).	Nurses,	the	largest	occupational	group	in	Canada’s	health	sector,	experience	particularly	high	rates	of	depression.	In	a	recent	large	representative	survey,	9%	of	both	male	and	female	nurses	experienced	depression	in	the	previous	year	(226),	compared	to	6%	of	all	employed	females	and	3%	of	all	employed	males	in	the	general	working	population	(227).	Nurses	also	reported	higher	rates	of	shift	work	in	the	survey,	with	54%	of	respondents	working	something	other	than	a	regular	day	shift	(226)	compared	to	28%	of	the	general	working	population	(2).		 41 Depression	is	strongly	linked	to	work	absence	(202,228,229).	Healthcare	workers,	particularly	nurses,	experience	high	rates	of	work	absence	compared	to	other	Canadian	industry	sectors	(226,230).	Interestingly,	quick	returns	between	shifts	have	been	noted	to	be	more	common	in	healthcare	versus	other	industry	sectors	(220);	providing	one	possible	explanation	for	the	higher	rates	of	depression	and	sickness	absence	seen	in	this	worker	population.	3.1.3 Study	rationale	and	objectives	In	a	recent	meta-regression	examining	the	epidemiology	of	depression	in	Canada	between	1994	and	2013,	both	the	incidence	and	duration	of	major	depressive	disorder	were	found	to	be	steady	over	time	(231).	Since	many	work	related	variables	associated	with	psychological	ill	health	and	work	absence	are	potentially	amenable	to	change	(232,233),	the	identification	of	modifiable	workplace	factors	related	to	depressive	outcomes	(such	as	shift	scheduling)	could	play	an	important	role	in	reducing	the	burden	of	this	prevalent	disease.	The	National	Survey	of	the	Work	and	Health	of	Nurses	(NSWHN),	2005	(234)	is	a	nationally	representative	cross-sectional	survey	that	was	conducted	by	the	Canadian	Institute	for	Health	Information	and	Statistics	Canada	to	provide	a	picture	of	the	health	and	working	conditions	of	nurses	in	Canada.	The	NSWHN	collected	a	wealth	of	information	on	work	scheduling,	workplace	characteristics,	and	health	variables	compared	to	most	other	large-scale	surveys.	Importantly,	it	also	collected	information	on	frequency	of	shift	rotations;	an	exposure	characteristic	that	is	rarely	described	in	other	studies	of	shift	work	and	health.		The	NSWHN	offered	a	unique	opportunity	to	investigate	the	impacts	of	exposure	misclassification	on	relationships	between	shift	work	and	depression	among	nurses	(a	large	working	population	with	high	prevalence	of	both	shift	work	and	depression)	by	using	exposure	indicators	with	varying	degrees	of	precision.	This	study	was	particularly	interested	in	examining	relationships	between	shift	rotation	frequency	and	depression,	since	this	had	not	previously	been	done.	The	hypotheses	were:	 42 	(A)	Associations	between	shift	work	schedule	and	depression	will	strengthen	as	the	precision	of	exposure	indicators	increases,	defined	as:	1)	High	precision,	considers	shift	timing	and	rotation	frequency,	2)	Moderate	precision,	considers	shift	timing	only,	and	3)	Low	precision,	considers	the	absence/presence	of	shift	work	only;	and		(B)	Shift	workers	involved	in	regular	night	work	and	high	frequency	rotation	shift	work	(the	latter	being	a	uniquely	detailed	measure	of	rotating	shift	work)	will	show	higher	odds	of	depression	relative	to	regular	daytime	workers.		3.2 Methods	3.2.1 Data	source	Data	for	this	study	were	obtained	from	the	National	Survey	of	the	Work	and	Health	of	Nurses	(NSWHN),	2005,	via	Statistics	Canada’s	Research	Data	Centres	Program.	The	Research	Data	Centres	program	provides	access	to	the	microdata	or	master	files	of	a	select	number	of	Statistics	Canada	Surveys,	in	secure	settings	that	govern	all	aspects	of	data	access	from	analysis	through	to	publication	of	analytic	output	(235).			The	NSWHN	was	a	one-time	survey	dedicated	to	assessing	the	working	conditions	and	health	of	nurses	at	a	national	level.	The	target	population	was	regulated	nurses	(registered	nurses,	registered	psychiatric	nurses,	and	licensed	practical	nurses)	aged	21	years	or	older	who	were	registered	and	employed	in	Canada	at	the	time	of	survey.	To	construct	the	sampling	frame,	Statistics	Canada	received	membership	lists	from	all	nursing	organizations	and	regulatory	bodies	in	each	of	the	10	provincial	and	3	territorial	jurisdictions	in	Canada.	From	a	national	total	of	331,992	nurses,	24,443	were	selected	at	random	using	a	stratified	design	to	ensure	adequate	sample	sizes	for	each	jurisdiction	and	for	each	type	of	regulated	nurse	(registered	nurse,	licensed	practical	nurse,	and	registered	psychiatric	nurse).	Within	these	strata,	secondary	stratification	was	conducted	for	age	group,	place	of	work	(hospital,	long-term	care	facility,	community	health	setting,	and	other)	and	employment	status	(full-time,	part-time,	and	casual).	Computer	assisted	telephone	interviewing	was	carried	out	by	trained	interviewers	between	October	2005	and	January	2006	to	obtain	a	survey	sample	 43 size	of	18,676	(response	rate	=	79.7%).	Further	details	of	the	NSWHN	survey	measures,	methodology,	and	quality	control	methods	have	been	reported	by	Statistics	Canada	(234).	3.2.2 Study	sample			To	increase	the	precision	of	exposure	indicators	based	on	working	conditions	and	health	in	the	12	months	prior	to	survey	interview,	and	to	enhance	the	comparability	of	exposure-response	relationships	across	exposure	indicator	categories,	the	analyses	excluded	individuals	who	were	(1)	employed	in	more	than	one	nursing	job	(n	=	2,846),	(2)	not	exclusively	providing	direct	care	to	patients	or	residents	(n	=	3,222),	(3)	employed	in	their	current	position	for	less	than	12	months	(n	=	209),	(4)	temporarily	absent	from	their	nursing	position	for	12	months	or	more	at	time	of	interview	(n	=	104),	and	(5)	self-employed	(n	=	295).	The	sample	was	further	restricted	to	individuals	with	valid	responses	for	the	outcome	variable,	primary	explanatory	variable,	and	all	other	variables	included	in	the	analyses.	These	criteria	produced	an	un-weighted	sample	size	of	n	=	11,450	respondents.		3.2.3 Study	variables	The	outcome	variable	was	major	depressive	disorder	occurring	within	the	12	months	prior	to	interview.	To	identify	major	depressive	episodes,	the	NSWHN	interview	utilized	a	predictive	instrument	called	the	Composite	International	Diagnostic	Interview	Short	Form,	Major	Depression	section	(CIDI-SFMD),	developed	by	Kessler	and	colleagues	(236,237).	Depression	categories	for	this	study	were	dichotomized	into	“Yes”	(respondent	indicates	5	or	more	major	depression	symptoms	on	the	CIDI-SFMD)	and	“No”	(respondent	indicates	fewer	than	5	major	depression	symptoms	on	the	CIDI-SFMD).	Prior	validation	studies	have	indicated	that	75	to	90%	of	participants	reporting	5	or	more	major	depression	symptoms	on	the	CIDI-SFMD	(approximating	symptoms	listed	in	the	Diagnostic	and	Statistical	Manual	of	Mental	Disorders,	3rd	Edition	(DSM-III-R)	criteria	for	major	depression	(238))	have	had	a	major	depression	episode	in	the	preceding	12	months	(237,239).	The	full	list	of	symptoms	is	reported	in	Appendix	B.	The	primary	explanatory	variable	was	shift	work	schedule.	Three	groups	of	exposure	 44 indicators	for	shift	work	schedule	were	derived	from	questions	pertaining	to	work	hours	at	respondents’	main	job:	“Do	you	usually	work	days,	evenings,	or	nights?”	and	(for	the	high	precision	definition	only)	“In	the	past	2	weeks,	how	many	times	did	you	change	shifts	(for	example,	from	days	to	evenings,	or	evenings	to	nights)?”		Exposure	Grouping	1	-	High	Precision	Shift	Work	Schedule,	seven	categories	(considers	shift	timing	and	rotation	frequency):		A)	Regular	Days	B)	Regular	Evenings	C)	Regular	Nights	D)	Slow	Rotating	Shifts	(0-1	shift	changes	in	past	2	weeks);		E)	Medium	Rotating	Shifts	(2-3	shift	changes	in	past	2	weeks);	F)	Rapid	Rotating	Shifts	(4+	shift	changes	in	past	2	weeks);						G)	Undefined	Rotating	Shifts	(did	not	work	in	past	2	weeks;	could	not	be	classified).	Exposure	Grouping	2	-	Moderate	Precision	Shift	Work	Schedule,	four	categories	(considers	shift	timing	only):	A)	Regular	Days	B)	Regular	Evenings	C)	Regular	Nights	D)	Rotating	Shifts	Exposure	Grouping	3	-	Low	Precision	Shift	Work	Schedule,	two	categories	(considers	absence/presence	of	shift	work	only):	A)	Regular	Days		B)	Any	work	outside	of	regular	daytime	hours	(regular	evenings,	regular	nights,	or	rotating	shifts)	Additional	variables			 45 The	National	Survey	of	the	Work	and	Health	of	Nurses	asked	a	number	of	questions	about	socio-demographic,	health,	and	work	characteristics,	including	psychosocial	work	factors	that	have	been	linked	to	depression	(240).	Since	this	study’s	goal	was	to	measure	the	relationship	between	shift	work	schedule	and	depression,	only	variables	with	the	potential	to	confound	this	relationship	(but	unlikely	to	lie	on	the	causal	pathway)	were	sought	for	inclusion	in	the	models.	Determinants	or	risk	factors	for	depression,	that	are	not	associated	with	shift	work	and	therefore	not	confounders	of	the	relationship	under	investigation,	were	also	excluded	from	the	analyses.		This	was	done	using	a	priori	knowledge	and	causal	diagrams	(241)	to	conceptualize	relationships	between	variables.		Psychosocial	work	factors	(e.g.,	high	psychological	demands,	low	social	support)	may	be	implicated	in	the	aetiology	of	depression	(79,242,243).	All	psychosocial	work	variables	available	in	the	survey	(job	strain,	role	overload,	autonomy,	control,	psychological	demands,	social	support,	organizational	support,	and	scheduling	flexibility)	were	entered	individually	into	a	preliminary	logistic	model	of	depression	and	work	schedule	(high	precision	definition)	to	identify	the	strongest	potential	confounders	for	retention	in	the	final	model.	Only	autonomy,	organizational	support,	and	scheduling	flexibility	produced	a	substantial	(>10%)	shift	in	the	point	estimates	for	the	effect	of	work	schedule	on	depression.	Examination	of	cross-tabs	and	Pearson	chi-square	tests	showed	strong	interrelations	between	these	three	variables,	so	only	scheduling	flexibility	(conceptually	the	least	likely	variable	to	be	driven	by	shift	work	schedule,	thus	least	likely	to	be	on	the	causal	pathway)	was	retained.		The	following	additional	variables	were	included	in	the	models	as	potential	confounders	since	they	are	risk	factors	for	depression	and	other	mental	health	outcomes,	and	may	also	be	related	to	work	schedule:	age	(244,245),	sex	(246),	family/living	situation	(245),	socioeconomic	status	(245,246),	presence	of	chronic	health	conditions	(245,247,248),	scheduling	flexibility	(249),	workplace	type	(250),	overtime	(251),	and	typical	shift	durations	(252)	(see	Table	3.1	for	detailed	categories).	 46 3.2.4 Statistical	analyses	Analyses	were	conducted	through	the	Statistics	Canada’s	Research	Data	Centres	Program	(235)	at	the	University	of	British	Columbia,	using	SAS	Version	9.4	(182).	All	bivariable	cell	sizes	were	30	observations	or	more,	as	required	for	confidential	microdata	release	by	Statistics	Canada.	To	appropriately	account	for	the	NSWHN	sampling	procedures	(234),	probability	survey	weights	provided	by	Statistics	Canada	were	applied	in	all	analyses	to	produce	variance	estimates	that	adjusted	for	the	sampling	strategy	and	reduced	bias	in	the	estimates	obtained.	In	the	final	models,	estimates	were	bootstrapped	using	500	replicates	as	per	Statistics	Canada	guidelines.	Crude	odds	ratios	with	95%	confidence	intervals	were	calculated	for	the	relationship	between	shift	work	schedule	and	depression.	Logistic	regression	was	conducted	to	assess	the	presence	and	strength	of	a	relationship	between	shift	work	schedule	and	depression	while	adjusting	for	all	previously	stated	confounders.	Individual	models	were	run	for	each	shift	work	exposure	indicator	group	(high,	moderate,	and	low	precision).	3.3 Results	3.3.1 Descriptive	summaries	The	distribution	of	study	variables	within	the	study	sample	(n	=	11,450)	is	provided	in	Table	3.1.	Over	60%	of	respondents	reported	something	other	than	a	regular	daytime	schedule.	Outside	of	a	regular	day	schedule	(38.4%	of	respondents),	the	most	frequently	reported	schedules	were	slow	(20.1%)	and	medium	(15.4%)	rotating	shifts.	Depression	was	observed	in	9.1%	of	the	study	sample’s	respondents,	similar	to	the	overall	survey	finding	of	9.4%.	Depression	was	most	prevalent	in	respondents	working	rapid	rotating	shifts	(13.8%),	and	in	those	who	reported	rotating	shifts	but	who	had	not	worked	in	the	2	weeks	prior	to	the	survey	(12.7%).			 47 Table	3.1:	Baseline	study	sample	characteristics	and	bivariable	associations	with	depression	(within	previous	12	months):	National	Survey	of	the	Work	and	Health	of	Nurses	(NSWHN),	20051		 Frequency	(n	=	11,450)	 (%)	 Depression	%	No	 Depression%	Yes	Depression	 		 		 		 		No	 10,446	 90.9	 	 	Yes		 1004	 9.1	 	 	High	Precision	Shift	Work	Exposure	Grouping	 		 		Regular	Days	 4310	 38.4	 90.7	 9.3	Regular	Evenings	 874	 8.6	 89.9	 10.1	Regular	Nights	 899	 9.7	 91.4	 8.6	Slow	Rotating	Shifts	 2514	 20.1	 92.8	 7.2	Medium	Rotating	Shifts	 1919	 15.4	 91.3	 8.7	Rapid	Rotating	Shifts	 535	 4.1	 86.2	 13.8	Undefined	Rotating	Shifts	 399	 3.4	 87.3	 12.7	Moderate	Precision	Shift	Work	Exposure	Grouping		 		Regular	Days		 4310	 38.4	 90.7	 9.3	Regular	Evenings	 874	 8.6	 89.9	 10.1	Regular	Nights	 899	 9.7	 91.4	 8.6	Rotating	Shifts		 5367	 43.2	 91.2	 8.8	Low	Precision	Shift	Work	Exposure	Grouping	 		 		Regular	Days		 4310	 38.4	 90.7	 9.3	Shift	Work	(regular	evenings,	regular	nights,	or	rotating	shifts)	 7140	 61.6	 91.1	 8.9	Sex		 		 		 		 		Female	 10694	 94.7	 91.0	 9.0	Male	 756	 5.3	 89.4	 10.6	Age	(years)	 		 		 		 		Less	than	35	 2131	 20.5	 91.5	 8.5	35-44	 3063	 27.8	 89.8	 10.2	45-54	 3973	 34.3	 90.4	 9.6	55	and	over	 2283	 17.5	 93.0	 7.0	Family/Living	Situation	 		 		 		 		Living	with	spouse/partner	 3153	 24.9	 92.8	 7.2	Unattached	living	alone	 1514	 13.3	 88.4	 11.6	Living	with	spouse/partner	+	children	 4885	 44.8	 91.8	 8.2	Other	 1898	 17.1	 87.8	 12.2	Household	Income2	 		 		 		 		High	 10548	 94.5	 91.1	 8.9	Low	 902	 5.5	 87.4	 12.6		 	 	 	 	 48 	 Frequency	(n	=	11,450)	 (%)	 Depression	%	No	 Depression%	Yes	Chronic	Health	Conditions3	 		 		 		 		None		 3693	 33.9	 94.7	 5.3	1	or	more	 7757	 66.1	 89.0	 11.0	Employment	Type	 		 		 		 		Permanent,	Full-time	 6407	 56.9	 91.0	 9.0	Permanent,	Part-time	 3459	 31.0	 90.7	 9.3	Non-Permanent,	Full-time	 497	 3.6	 90.0	 10.0	Non-Permanent,	Part-time	 1087	 8.5	 91.5	 8.5	Scheduling	Flexibility4	 		 		 		 		Yes	 3154	 27.3	 92.0	 8.0	No	 8296	 72.7	 90.5	 9.5	Workplace	Type	 		 		 		 		Hospital		 5936	 66.1	 91.2	 8.8	Long-term	Care	Facility	 2879	 16.3	 90.1	 9.9	Community	 1669	 10.7	 90.5	 9.6	Other	 966	 6.9	 90.7	 9.3	Paid	Overtime5	 		 		 		Yes	 3481	 33.5	 89.8	 10.2	No	 7969	 66.5	 91.5	 8.5	Typical	Shift	Duration	 		 		 		 		8	hours	or	less	 7026	 62.8	 90.1	 9.9	12	hours	or	more	 3646	 31.3	 92.1	 7.9	Various	 778	 6.0	 92.9	 7.1	1	All	percentages	weighted	to	account	for	NSWHN	probability	sampling;	values	may	not	add	up	to					100%	due	to	rounding.	2	Household	Income:	categories	modelled	after	Statistics	Canada	methods	(58),	based	on	2	questions:					“What	is	your	best	estimate	of	the	total	income,	before	taxes	and	deductions,	of	all	household							members	from	all	sources	in	the	past	12	months?”	and	“Number	of	people	in	the	household”.	Low					household	income	=	(1-2	people;	income	<$30,000)	or	(3-4	people;	income	<$40,000)	or	(5+	people;					income	<$60,000);	High	household	income	=	(1-2	people;	income	≥$30,000)	or	(3-4	people;	income					≥$40,000)	or	(5+	people;	income	≥$60,000).	Note:	Donor	imputation	was	used	in	the	NSWHN	for	7%						of	respondents	who	did	not	state	their	household	income	(234).	3	Chronic	Health	Conditions:	Derived	variable	including	only	long-term	conditions	that	have	or	are	expected						to	last	6	months	or	more,	and	that	have	been	diagnosed	by	a	health	professional:	allergies,	asthma,					fibromyalgia,	arthritis	or	rheumatism,	back	problems,	migraine	headaches,	diabetes	(non-pregnancy					related),	heart	disease,	cancer,	stomach	or	intestinal	ulcers,	bowel	disorder	(such	as	Crohn's	disease					or	colitis),	thyroid	condition,	chronic	fatigue	syndrome,	multiple	chemical	sensitivities.	Respondents					indicating	yes	to	one	or	more	of	these	conditions	were	assigned	yes	for	this	variable.	4	Scheduling	Flexibility:	Based	on	the	question	“Does	your	employer	offer	flexibility	in	the	hours					nurses	can	choose	to	work?”	Respondents	indicating	“No”	or	“Don’t	know”	were	assigned	no	for	this					variable.	5	Paid	Overtime:	Based	on	the	question	“How	many	hours	of	paid	overtime	do	you	usually	work	per					week?”	Respondents	indicating	zero	hours	were	assigned	no	for	this	variable.		 	 49 In	terms	of	potential	confounding	variables,	the	majority	of	respondents	were	females	in	the	high	household	income	category,	with	one	or	more	chronic	health	conditions,	employed	in	permanent	full-time	work,	working	in	hospital	settings,	and	with	typical	shift	durations	of	8	hours	or	less.	Nearly	half	of	respondents	reported	living	with	a	spouse/partner	and	children.	Approximately	one	third	reported	some	degree	of	flexibility	in	their	work	hours,	and	working	overtime	every	week,	respectively.	3.3.2 Logistic	regression	Associations	between	the	three	shift	work	exposure	indicator	groups	and	depression	are	presented	in	Table	3.2.	Only	the	model	using	the	high	precision	group	showed	strong	associations	between	type	of	shift	work	and	depression.	The	adjusted	odds	ratio	for	depression	was	higher	in	the	rapid	rotating	shifts	category	(OR	=	1.51,	95%	CI	=	0.91-2.51)	and	in	the	undefined	rotating	shifts	category	(OR	=	1.67,	CI	=	0.92-3.02),	while	it	was	lower	in	the	slow	rotating	shifts	category	(OR	=	0.79,	95%	CI	=	0.57–1.08).	No	strong	relationships	emerged	between	shift	work	schedule	and	depression	(OR	range	0.95-0.99;	Table	3.2)	in	the	adjusted	models	using	moderate	precision	and	low	precision	shift	work	schedule	indicator	groups.		For	the	other	potentially	confounding	variables,	the	strongest	relationships	with	depression	were	observed	in	the	35	to	44	year	old	category	(relative	to	55	years	old	and	over);	in	unattached	individuals	living	alone	and	those	with	“other”	family/living	circumstances	(relative	to	those	living	with	a	spouse	or	partner);	and	in	those	reporting	one	or	more	diagnosed	chronic	health	conditions	(relative	to	those	reporting	no	diagnosed	chronic	health	conditions).	Higher	odds	ratios	for	depression	were	also	observed	for	other	younger	age	categories,	for	males,	for	those	with	low	household	incomes,	for	those	with	shorter	(8	hours	or	less)	shift	duration,	for	those	working	some	weekly	paid	overtime,	and	for	those	reporting	no	scheduling	flexibility;	the	95%	CIs	for	all	of	these	estimates	included	‘1’.	These	relationships	were	consistent	across	all	models	(low,	moderate,	and	high	precision).	 50 Table	3.2:	Unadjusted	and	adjusted	logistic	regression	odds	ratios	(ORs)	and	confidence	intervals	(CIs)	modeling	depression	=	yes,	National	Survey	of	the	Work	and	Health	of	Nurses,	2005		Unadjusted	OR	 95%	CI	 	Adjusted	OR1	 95%	CI	High	Precision	Shift	Work	Exposure	Grouping	 		 		 		 		 		 		Schedule	 	 	 	 	 	 	 	 	 	 				Regular	Days	 Ref	 	 	 	 Ref	 	 				Regular	Evenings	 1.09	 1.03	 1.15 	 1.00	 0.70	 1.43				Regular	Nights	 0.91	 0.86	 0.96	 	 0.96	 0.63	 1.45				Slow	Rotation	 0.75	 0.72	 0.79	 	 0.79	 0.57	 1.08				Medium	Rotation	 0.92	 0.88	 0.97	 	 1.00	 0.71	 1.40				Rapid	Rotation	 1.55	 1.45	 1.66	 	 1.51	 0.91	 2.51				Undefined	Rotation		 1.42	 1.31	 1.53	 	 1.67	 0.92	 3.02	Sex		 	 	 	 	 	 	 				Female	 	 	 	 	 Ref	 	 				Male	 	 	 	 1.28	 0.84	 1.96	Age	Group	 	 	 	 	 	 	 				55	and	over	 	 	 	 	 Ref	 	 				45-54	 	 	 	 	 1.50	 1.09	 2.06				35-44	 	 	 	 	 1.77	 1.27	 2.46				Less	than	35	 	 	 	 	 1.42	 0.98	 2.05	Family/Living	Situation	 	 	 	 	 	 	 				Living	with	spouse/partner	 	 	 	 Ref	 	 				Unattached	living	alone	 	 	 1.66	 1.16	 2.38				Living	with	spouse/partner	+	children	 	 	 1.00	 0.74	 1.33				Other	 	 	 	 	 1.62	 1.19	 2.20	Household	Income	 	 	 	 	 	 	 				High	 	 	 	 	 Ref	 	 				Low	 	 	 	 	 1.30	 0.95	 1.76	Chronic	Conditions	(diagnosed)	 	 	 	 	 	 	 				None		 	 	 	 	 Ref	 	 				1	or	more	 	 	 	 	 2.29	 1.74	 3.00	Employment	Type	 	 	 	 	 	 	 				Permanent,	Full-time	 	 	 	 	 Ref	 	 				Permanent,	Part-time	 	 	 	 	 1.10	 0.86	 1.41				Non-Permanent,	Full-time	 	 	 1.04	 0.56	 1.92				Non-Permanent,	Part-time	 	 	 1.06	 0.72	 1.55	Scheduling	Flexibility	 	 	 	 	 	 	 				Yes	 	 	 	 	 Ref	 	 				No	 	 	 	 	 1.19	 0.92	 1.53				 	 	 	 	 	 	 	 51 	Unadjusted	OR	 95%	CI	 	Adjusted	OR1	 95%	CI	Workplace	Type				Hospital		 	 	 	 	 Ref	 	 				Long-term	Care	Facility	 	 	 	 1.08	 0.84	 1.38				Community	 	 	 	 	 1.10	 0.79	 1.53				Other	 	 	 	 	 1.17	 0.76	 1.78	Paid	Overtime	 	 	 	 	 	 	 				No	 	 	 	 	 Ref	 	 				Yes	 	 	 	 	 1.20	 0.95	 1.52	Typical	Shift	Duration	 	 	 	 	 	 	 				12	hours	or	more	 	 	 	 	 Ref	 	 				8	hours	or	less	 	 	 	 	 1.29	 0.98	 1.69				Various	 	 	 	 	 0.87	 0.52	 1.44	Moderate	Precision	Shift	Work	Exposure	Grouping	 		 		 		 		 		Schedule		 	 	 	 	 	 	 	 	 	 				Regular	Days		 Ref	 	 	 	 Ref	 	 				Regular	Evenings	 1.09	 1.03	 1.15	 	 0.99	 0.69	 1.42				Regular	Nights	 0.91	 0.86	 0.96	 	 0.95	 0.63	 1.44				Rotating	Shifts	 0.93	 0.90	 0.96	 	 0.99	 0.75	 1.29	Low	Precision	Shift	Work	Exposure	Grouping	 		 		 		 		 		 		Schedule		 	 	 	 	 	 	 	 	 	 				Day	Worker	 Ref	 	 	 	 Ref	 	 				Shift	Worker	 0.95	 0.92	 0.98	 		 0.98	 0.77	 1.25	1	Bootstrap	weights	applied;	models	adjusted	for	sex,	age,	living/family	situation,	household	income,	chronic	health	conditions,	employment	type,	scheduling	flexibility,	workplace	type,	paid	overtime,	and	typical	shift	duration		3.4 Discussion		The	high	precision	shift	work	schedule	exposure	grouping	used	in	this	study	divided	the	sample	into	7	categories	of	shift	work:	regular	day	(38.4%),	regular	evening	(8.6%),	regular	night	(9.7%)	and	rotating	work	based	on	shift	changes	in	the	past	two	weeks:	0-1	(20.1%),	2-3	(15.4%),	4+	(4.1%),	and	undefined	(3.4%).	For	the	moderate	precision	shift	work	schedule	exposure	grouping,	the	latter	categories	were	aggregated	to	form	one	rotating	category	that	represented	43%	of	the	total	sample.	For	the	low	precision	shift	work	schedule	exposure	grouping,	all	categories	of	non-regular	daytime	work	were	further	aggregated	into	one	shift	work	category	that	represented	61.6%	of	the	total	sample.		 52 As	hypothesized,	the	strongest	relationships	between	shift	work	schedule	and	depression	are	observed	in	the	high	precision	shift	work	schedule	exposure	model	that	considered	elements	of	shift	timing	and	rotation	frequency.	The	odds	ratio	for	depression	was	higher	in	the	rapid	rotating	and	the	undefined	rotating	shifts	categories,	and	decreased	in	the	slow	rotating	shifts	category.	No	relationship	was	observed	with	depression	for	the	low	and	moderate	precision	shift	work	schedule	exposure	indicator	groups.	While	the	bootstrapped	high	precision	shift	work	schedule	model	results	are	not	statistically	significant	in	that	the	95%	confidence	intervals	include	the	OR	value	of	‘1’,	it	is	worth	noting	that	confidence	intervals	are	intended	to	serve	as	a	general	guide	to	the	amount	of	random	error	in	the	data	rather	than	as	a	literal	measure	(253).	The	point	estimates	presented	for	the	high	precision	shift	work	schedule	model	are	stronger	than	those	obtained	for	the	medium-	and	low-precision	models.	Odds	ratios	close	to	‘1’	are	noted	in	the	moderate	precision	shift	work	schedule	exposure	grouping	that	considered	elements	of	shift	timing	only,	and	in	the	low	precision	shift	work	schedule	exposure	grouping	that	dichotomized	shift	work	exposure	into	yes/no	categories.	The	observed	elevated	odds	for	depression	for	the	work	schedule	variable	in	the	high	precision	model	persisted	after	adjustment	for	confounders.	The	elevated	odds	ratio	for	depression	in	this	study’s	rapid	rotating	shifts	category	is	a	new	finding,	but	consistent	with	reports	from	other	emerging	research.	Rapidly	rotating	shifts	may	reflect	precarious	employment	situations	(work	involving	temporary,	contract,	or	casual	on-call	positions	that	often	lack	benefits	and	job	security;	also	referred	to	as	“casual”,	“seasonal”,	“temporary”,	“non-standard”	or	“contingent”	work)	(222).	Workers	in	precarious	employment	often	experience	unexpected	changes	in	work	schedule,	and	have	little	advance	notice	of	their	work	schedule	(254).	Short	notice	(<	1	month)	of	a	new	work	schedule	has	been	linked	to	negative	social	effects	in	shift	workers	(220).	Precarious	employment	and	job	insecurity	have	been	associated	with	a	number	of	related	detriments	(222),	including	increased	social	disruption,	stress,	and	depressive	symptoms	(223–225).		Rapidly	rotating	shifts	are	also	likely	to	increase	the	likelihood	of	quick	returns	(i.e.	where	11	hours	or	less	of	free	time	is	scheduled	between	shifts	(219)).	Some	shift	workers	view	 53 quick	returns	as	being	more	problematic	than	night	work	(220);	with	recent	evidence	showing	that	quick	returns	have	negative	effects	on	sleep	and	fatigue	(255–257)	and	risk	of	sick	leave	(258).	In	these	studies	(255–258),	the	negative	effects	of	quick	returns	were	more	severe	than	those	of	night	work.		The	preceding	evidence	on	quick	returns	could	explain	the	lack	of	strong	relationships	noted	between	regular	night	shift	work	and	depression	in	this	study.	A	body	of	strong	evidence	shows	that	night	work	is	linked	to	increased	risks	of	circadian	disruption	and	negative	health	outcomes	(3,259),	and	some	studies	have	noted	associations	between	regular	night	shifts	and	depressive	outcomes	(16,217).	However,	it	is	possible	that	social	issues	related	to	rapidly	rotating	shift	schedules	represent	an	equally	strong	(or	stronger)	link	to	negative	mental	health	outcomes	in	shift	workers,	as	compared	to	circadian	disruption	associated	with	night	work.	Rapidly	rotating	schedules	may	impose	a	greater	mental	strain	on	workers	than	regular	night	work,	particularly	in	cases	of	irregular	work	schedules.	In	Canadian	workers	for	example,	dissatisfaction	with	work-life	balance	is	most	pronounced	in	shift	workers	reporting	split,	on	call	or	casual,	or	irregular	schedules	(2).	This	could	also	explain	the	decreased	odds	ratio	for	depression	noted	in	the	slow	rotating	shift	schedule	category;	rare	or	occasional	changes	in	shift	timing	(0-2	times	per	month)	are	likely	to	be	more	predictable	and	may	even	confer	greater	work-life	balance	relative	to	regular	daytime	workers.		The	increased	odds	ratio	for	depression	in	the	undefined	rotating	shifts	category	(individuals	not	working	in	the	2	weeks	prior	to	survey)	is	not	straightforward	to	interpret.	Depression	is	associated	with	temporary	work	leave	and	work	impairment	(202,228,229,245),	therefore	one	explanation	of	the	increased	odds	ratio	for	depression	in	this	group	is	that	depression	status	(unrelated	to	shift	work	schedule)	resulted	in	the	reported	work	absence.	Another	possibility	is	that	the	characteristics	of	workers’	rotating	schedules	(e.g.,	high	frequency	rotations	and	associated	stressors)	resulted	in	depression	and	sick	leave,	as	has	been	observed	elsewhere	(101).	A	third	(though	least	likely)	possibility	is	that	respondents	not	working	in	the	past	two	weeks	were	off	work	for	reasons	unrelated	to	depression,	but	were	coincidentally	more	depressed	than	other	 54 respondents.	Since	the	NSWHN’s	cross-sectional	design	lacks	the	temporal	detail	needed	to	assess	which	scenario	is	most	likely,	this	study’s	finding	of	high	depression	outcomes	in	the	workers	with	undefined	rotating	shifts	should	be	interpreted	with	caution.	3.4.1 Strengths	and	limitations	In	addition	to	precise	and	hypothesis-driven	exposure	assessment	and	assignment,	the	accurate	estimation	of	exposure-response	relationships	relies	on	the	control	of	variables	that	may	confound	the	relationship	between	primary	exposure	and	outcome.	In	this	study’s	multivariable	models,	the	observed	relationships	between	shift	schedule	and	depression	persisted	after	adjustment	for	a	number	of	known	and	potential	confounders	available	in	the	NSWHN.	The	relationships	observed	between	these	confounders	and	depression	are	consistent	with	a	body	of	research	that	has	found	higher	rates	of	depression	among	younger	adults	(244,245),	single	or	previously-married	individuals	(245),	low	income	earners	(245),	and	those	with	chronic	health	conditions	(245,247,248).	Low	control	over	work	time	(relating	to	low	scheduling	flexibility)	(249),	long	working	hours	(relating	to	overtime)	(251),	and	shorter	shift	durations	(252)	also	have	been	respectively	linked	to	negative	mental	health	outcomes,	providing	additional	face	validity	to	the	models	in	the	current	study.	It	is	interesting	to	note	that	depression	prevalence	was	higher	in	males	than	females	in	this	study’s	sample,	unlike	general	population	surveys	where	depression	prevalence	is	consistently	higher	in	females	(84,244).	Higher	depression	prevalence	in	males	in	nursing	populations	has	been	observed	in	other	studies	(260,261)	and	may	be	explained	by	greater	physical	demands	placed	on	male	nurses	(261),	or	by	the	disproportionate	emotional	effects	of	lower-prestige	“female-type”	occupations	on	male	nurses	relative	to	their	female	counterparts	(262).		To	reduce	the	likelihood	of	exposure	misclassification,	this	study’s	analyses	were	restricted	to	individuals	working	one	nursing	job,	since	shift	work	schedule	information	was	collected	for	respondents’	“main	nursing	job”	only.	However,	this	study’s	shift	work	categories	do	retain	some	degree	of	misclassification,	since	the	NSWHN	does	not	differentiate	between	 55 rotating	shift	work	involving	nights	versus	rotating	shift	work	that	involves	only	days	and	evenings.	Furthermore,	the	NSWHN	does	not	capture	other	exposure	characteristics	such	as	direction	of	shift	rotations	(60,65),	history	of	shift	work,	and	individual	morning/evening	preference	(89)	that	may	be	important	to	consider	when	assessing	the	impacts	of	shift	work	schedule	on	depression.	The	current	study’s	findings	support	the	importance	of	assigning	detailed	shift	work	exposure	indices	in	order	to	identify	effects	where	they	exist,	underlining	the	need	to	collect	and	utilize	such	variables	in	future	studies.	The	relatively	homogenous	nature	of	NSWHN	respondents	reduced	sources	of	residual	confounding	that	are	often	present	in	general	population	surveys,	such	as	differences	in	work	environments	and	tasks	that	may	impact	on	depressive	outcomes	(263).	The	NSWHN’s	size	also	permitted	the	restriction	of	analyses	to	nurses	working	in	direct	care	areas	and	not	self-employed,	further	reducing	unmeasured	differences	in	exposures	between	comparison	groups	and	the	potential	for	a	biased	assessment	of	the	relationship	between	shift	work	schedule	and	depression.	Although	useful	for	the	purpose	of	these	analyses,	the	sample’s	homogeneity	does	introduce	the	possibility	that	the	findings	do	not	apply	to	other	occupations	and	work	environments,	where	the	type	and	timing	of	work	demands	may	have	different	impacts	on	depression.	A	widely	acknowledged	challenge	in	shift	work	research	is	the	self-selection	of	individuals	in	to	and	out	of	shift	work	(the	“healthy	worker	effect”)	leading	to	a	workforce	of	shift	workers	that	is	healthier	than	day	workers	(23)	and	biasing	results	toward	underestimated	effects.	Workers	that	have	remained	employed	(“survived”)	in	shift	work	for	long	periods	are	likely	less	sensitive	to	its	negative	effects,	which	may	also	bias	results	towards	the	null.	Self-selection	into	shift	work	(primary	selection)	may	be	less	of	a	problem	in	nurses,	since	most	direct	care	areas	require	some	degree	of	night	work,	particularly	for	new	graduates.	Self-selection	out	of	shift	work	(secondary	selection)	is	likely	a	bigger	issue	affecting	this	study	population.	Recent	longitudinal	studies	have	shown	that	the	presence	of	depressive	symptoms	(101)	and	other	depression-related	outcomes	(264)	at	baseline	is	associated	with	a	change	in	work	schedule	(leaving	night	work).	In	the	present	study,	depressed	workers	who	had	moved	from	night	work	into	a	regular	day	schedule	could	have	diluted	 56 the	reference	category	and	produced	attenuated	associations.	Despite	this,	a	relationship	between	work	schedule	and	depression	was	observed	when	the	high	precision	exposure	grouping	was	used.	It	seems	less	likely	that	depressed	individuals	would	differentially	move	into	more	disruptive	schedules	(i.e.	rapidly	rotating	work)	that	would	be	required	to	overestimate	the	strength	of	association	between	shift	work	schedule	and	depression	in	this	category	of	workers.	The	NSWHN’s	cross-sectional	design	and	lack	of	information	on	history	of	shift	work	did	not	allow	for	a	temporal	assessment	of	self-selection	effects.	Therefore,	it	is	certain	that	some	amount	of	misclassification	was	present	in	the	“day	worker”	category	(i.e.,	some	former	shift	workers	would	be	classified	as	day	workers).	Despite	this	important	limitation,	the	NSWHN’s	stratified	random	sample	sampling	ensured	representation	across	age	categories	which	is	an	improvement	over	other	studies	of	shift	work	that	have	assessed	disproportionate	numbers	of	middle-	or	advanced-age	participants	(136,146).	The	assessment	of	a	short-latency	health	outcome	may	have	also	minimized	the	effects	of	self-selection	out	of	shift	work	(i.e.,	shift	workers	recently	affected	by	depression	may	not	have	moved	out	of	a	night	work	schedule	at	the	time	of	survey).	Although	this	study	could	not	conduct	a	thorough	investigation	of	healthy	worker	bias,	this	is	an	important	methodological	issue	to	consider	in	future	research	into	shift	work	(23,265).		In	an	attempt	to	limit	the	effect	of	self-selection	out	of	shift	work	in	the	current	study,	the	sample	included	respondents	who	had	not	worked	in	the	past	two	weeks	prior	to	the	survey,	as	well	as	those	who	were	off	work	for	less	than	one	year	at	the	time	of	survey.	Individuals	who	were	off	work	for	more	than	one	year	at	the	time	of	study	were	excluded	to	permit	the	assessment	of	both	shift	work	schedule	and	depression	occurring	in	the	past	12	months.	This	exclusion	criterion	could	have	resulted	in	an	underestimation	of	depression	prevalence	and	biased	associations	towards	the	null,	although	this	effect	may	be	minimal	since	the	prevalence	of	depression	in	this	sample	(9.1%)	is	essentially	the	same	as	that	reported	in	the	full	survey	(226).	The	NSWHN’s	high	response	rate	provides	further	assurance	that	error	due	to	differences	between	respondents	and	non-respondents	was	relatively	low.	 57 This	study’s	exposure	and	outcome	measures	are	based	on	self-reports	that	were	not	validated	against	objective	data,	or	by	clinical	assessment.	Concerning	the	exposure	measure,	the	validity	of	self-reported	shift	work	exposure	was	recently	examined	in	a	Finnish	study	that	compared	self-reports	to	payroll	registry	data	(266).	Self-reported	work	involving	night	shifts	(regular	and	rotating)	showed	greater	sensitivity	and	precision	than	shift	work	not	involving	nights;	imprecise	reports	of	shift	work	not	involving	nights	(e.g.,	shift	workers	that	rotated	between	day	and	evening	shifts	identifying	themselves	as	day	workers)	resulted	in	a	bias	towards	the	null	when	assessing	work	schedule’s	effects	on	fatigue	(266).	In	the	NSWHN,	the	type	of	rotating	shift	work	(involving	nights	versus	no	nights)	was	not	differentiated.	Since	the	majority	of	respondents	(62.8%)	in	the	current	study	indicated	8-hour	shift	duration,	rotations	between	day	and	evening	shifts	without	any	night	shift	work	likely	occurred	in	a	(unknown)	number	of	workers,	and	may	have	produced	a	bias	towards	the	null	when	assessing	relationships	between	shift	work	schedule	and	depression.	Concerning	the	validity	of	the	outcome	measure,	the	CIDI-SF	assessment	tool	used	to	evaluate	the	likelihood	of	Major	Depressive	Disorder	in	the	NSWHN	was	derived	from	a	subset	of	items	from	the	Composite	International	Diagnostic	Interview	(CIDI),	a	well-validated	(237,239)	WHO-endorsed	tool	that	has	been	used	successfully	in	other	studies.	Social	desirability	bias	(arising	from	the	stigma	attached	to	mental	illness)	may	have	produced	an	underreporting	of	depressive	symptoms	as	noted	in	another	large	general	population	health	survey	conducted	in	Canada	(267),	although	this	would	likely	be	non-differential	across	shift	type	and	would	therefore	exert	a	conservative	bias	on	the	exposure-response	relationship.		Prior	to	the	NSWHN,	information	available	on	Canadian	nurses	was	deemed	insufficient	to	conduct	comprehensive,	reliable,	and	valid	evaluations	of	work	and	health	(268).	Although	data	collection	for	the	NSWHN	was	conducted	over	a	decade	ago,	this	survey	remains	the	largest	and	most	detailed	source	of	information	on	work	and	health	characteristics	at	the	occupational	level,	in	one	of	Canada’s	largest	working	populations.	The	use	of	this	data	source	to	investigate	relationships	between	shift	work	and	depression	still	holds	relevance	 58 today,	since	the	hypothesized	underlying	relationships	are	temporally	consistent.	Furthermore,	the	annual	prevalence	of	major	depression	in	Canadians	remained	steady	between	2002	and	2012	(4.8%	and	4.7%,	respectively)	(201)	and	an	increasing	proportion	of	employed	Canadians	(approximately	33%)	worked	some	form	of	shift	work	in	2011	(11),	compared	to	28%	in	2005	(2).	This	study’s	findings	are	also	relevant	elsewhere,	given	the	elevated	prevalence	of	depression	and	other	measures	of	psychological	morbidity	observed	in	nursing	populations	outside	of	Canada,	such	as	the	USA	and	the	UK	(269,270).	3.5 Conclusions	As	with	other	areas	of	shift	work	epidemiology,	the	quality	of	evidence	linking	shift	work	to	depression	is	challenged	by	the	use	of	coarse	exposure	assignment	that	does	not	sufficiently	consider	characteristics	of	shift	work	exposure	with	impacts	on	the	outcome.	The	high	precision	shift	work	schedule	exposure	indicator	assigned	in	this	study	(an	improvement	over	many	others	used	in	studies	of	shift	workers)	incorporated	considerations	of	both	shift	timing	and	intensity	of	shift	rotations.	This	definition	reduced	within-group	heterogeneity	compared	to	the	low-	and	moderate-precision	indicators,	and	produced	the	strongest	associations	with	depression	in	this	sample	of	nurses.	This	study’s	findings	support	the	need	to	use	precise	and	conceptually	driven	exposure	assignment	in	future	studies	of	shift	work	and	health,	to	correctly	assess	and	identify	exposure-response	relationships	within	individual	studies,	and	to	appropriately	target	health	interventions	to	reduce	the	personal	and	economic	burden	associated	with	shift	work.	Further	research	into	the	effects	of	shift	rotation	frequency	on	depression	is	also	recommended.	 59 Chapter	4: Assessing	determinants	of	workplace-level	shift	work	policies	and	practices:	An	employer	survey	in	British	Columbia,	Canada	4.1 Introduction	Shift	worker	health	is	influenced	by	a	variety	of	workplace,	social,	and	personal	factors	(4,19).	The	complexities	of	shift	work	exposure	and	its	various	pathways	to	health	outcomes	point	to	the	need	for	a	comprehensive	systems	approach	(e.g.,	targeting	workplace,	social,	and	personal	factors)	to	mitigate	its	impacts	(271).		Interventions	to	reduce	the	health	and	social	burden	associated	with	shift	work	should	include	multiple	leverage	points,	such	as	workplace,	social,	and	individual	factors	(129).	Although	the	scheduling	preferences	and	needs	of	shift	workers	may	vary	considerably	across	a	number	of	personal	and	social	factors	that	should	be	considered	for	prevention	purposes	(71),	workplace-level	interventions	(defined	as	“planned,	behavioural,	theory-based	actions	that	aim	to	improve	employee	health	and	well-being	through	changing	the	way	work	is	designed,	organized,	and	managed”	(272))	are	a	vital	means	for	health	promotion	at	a	population	level	(273).	For	instance,	workplace-level	interventions	can	be	broadly	applied,	and	do	not	rely	on	the	participation	of	a	small	number	of	highly	motivated	individuals	(129).	A	number	of	modifiable	workplace	policies	and	practices	-	concerning	shift	work	scheduling	(60,61),	light-at-night	levels	(155),	and	health	promotion	to	target	individual-level	factors	(62)	-	have	been	linked	to	positive	health	outcomes	in	shift	workers,	and	are	potential	targets	for	further	research	and	intervention.		One	aspect	of	shift	work	scheduling	policy	and	practice	that	has	received	considerable	attention	is	the	optimal	length	of	a	shift	to	maintain	health	(4),	with	some	viewing	long	work	hours	as	being	most	problematic	to	shift	workers	(1).	While	there	is	no	consensus	definition	for	this	concept	within	a	work	shift	(i.e.,	extended	or	“long”	hours)	(1),	the	effects	of	altering	shift	duration	as	a	health	intervention	are	frequently	assessed	by	comparing	outcomes	in	8-hour	versus	10-	or	12-hour	shifts.	Evidence	to	support	the	benefits	and	 60 drawbacks	of	typical	8-hour	shifts	versus	long	duration	shifts	is	mixed	(71)	and	partly	depends	on	the	outcome	in	question.	For	example,	long	duration	shifts	appear	to	confer	greater	worker	satisfaction	(274),	while	short	duration	shifts	tend	to	optimize	workplace	safety	outcomes	and	provide	more	time	for	work	recovery	between	shifts	(4,275).		As	described	in	section	1.1.2.3	of	this	dissertation,	exposure	to	light-at-night	has	been	linked	to	disrupted	circadian	rhythmicity	in	various	physiological	processes	(e.g.,	melatonin	suppression	and	sleep	disruption)	(32,46)	that	may	impact	on	shift	worker	performance	and	health	outcomes.	Lighting	in	workspaces	serves	a	variety	of	purposes,	such	as	a	sense	of	personal	security,	detection	of	hazards,	and	support	of	visual	performance	(67,68),	and	lighting	levels	can	be	expected	to	vary	across	different	work	environments	and	tasks	(68,69).	However,	despite	light-at-night’s	ubiquity	and	its	potential	effects	on	performance	and	health,	workplace-level	policies	and	practices	that	determine	worker	exposure	levels	are	not	well	described	in	the	literature.	Individual-level	workplace	policies	and	practices	include	employer	provision	of	shift	work	education	materials	and	training	to	employees,	such	as	training	sessions	for	shift	workers	and	their	partners,	individual	counselling,	or	written	materials	(71).	Such	resources	may	cover	a	number	of	individual-level	behavioural	strategies	to	cope	with	shift	work,	including	topics	such	as	alertness	at	work,	safe	driving,	and	healthy	sleep,	eating,	physical	activity,	and	family/social	relationships	(62,71).	Though	not	a	substitute	for	scheduling	and	work	environment	interventions,	shift	work	education	may	be	incorporated	into	a	comprehensive	systems	approach	to	mitigating	the	impacts	of	shift	work	(271).				4.1.1 Study	rationale	and	objective	Although	a	singular	“silver	bullet”	solution	is	unlikely,	interventions	that	target	modifiable	workplace	policies	and	practices	(such	as	scheduling,	lighting,	and	health	promotion)	are	important	aspects	of	comprehensive	approaches	to	promote	workplace	health	and	safety	(276).	However,	little	empirical	information	exists	to	describe	the	determinants	and	prevalence	of	modifiable	shift	work	policies	and	practices	in	workplaces.		 61 Such	knowledge	could	be	used	to	identify	research	targets	and	inform	interventions	to	mitigate	negative	health	outcomes	in	shift	workers.	This	is	particularly	relevant	for	future	workplace-based	research	into	shift	work	and	health,	where	random	assignment	of	worker	clusters	(e.g.,	units	within	a	hospital)	to	an	intervention	has	been	recommended	as	a	more	practical	alternative	to	the	randomization	of	individual	workers	(62).	The	objective	of	this	exploratory	study	was	to	identify	determinants	of	workplace-level	shift	work	policies	and	practices	with	potential	impacts	on	health	across	a	range	of	industry	sectors,	in	a	survey	of	employers	reliant	on	shift	work.	4.2 Methods	This	interview-administered	study	focused	on	organizations	that	employ	shift	workers	across	a	range	of	industries	within	the	province	of	British	Columbia,	Canada.	In	2014,	British	Columbia	had	a	workforce	of	2.3	million	(277),	approximately	one	third	of	whom	were	shift	workers	(278).	4.2.1 Survey	development	Survey	items	concerning	shift	work	schedule	characteristics	and	the	provision	of	shift	work	education	materials/training	to	assist	workers	with	shift	work	adaptation,	performance,	and	safety	were	drawn	from	a	previous	study	of	shift	work	practices	conducted	in	British	Columbia	(279).	Workplace-level	factors	hypothesized	to	affect	the	use	of	shift	work	schedules,	nighttime	lighting	policies,	and	educational	resources	were	identified	following	a	review	of	literature	that	described	various	aspects	of	work	organization	and	shift	system	design	(e.g.,	(25,59))	and	consultation	with	colleagues	in	relevant	research	fields	(e.g.	epidemiology,	circadian	science,	and	human	resources).		The	survey	also	contained	questions	about	every	shift	schedule	(defined	as	a	generally	consistent	pattern	of	shifts	worked	by	an	employee	that	differed	from	another	employee’s	pattern	of	shifts)	used	within	a	given	workplace.	Shift	schedules	were	characterized	by:	shift	coverage	period	(e.g.,	day,	evening,	or	night),	shift	duration	in	hours,	specific	start	and	end	times	of	shifts,	regularity	(e.g.,	always	work	nights	versus	rotating	between	days	and	nights),	type	of	shift	rotation	if	present	(e.g.,	switching	from	days	to	nights	versus	nights	to	days),	and	typical	number	of	days’	rest	between	shifts.		 62 An	initial	version	of	the	survey	was	piloted	with	human	resources	and	labour	relations	managers	employed	in	the	entertainment	and	manufacturing	sectors	to	ensure	relevance	and	clarity	of	wording;	feedback	from	this	process	informed	improvements	for	the	final	survey	version	(Appendix	C.1).	4.2.2 Recruitment	and	data	collection	An	existing	database	of	organizations	employing	shift	workers	in	British	Columbia	was	used	for	recruitment.	This	database,	developed	in	2003	for	the	purpose	of	other	shift	work	research	(279),	contained	178	organizations	that	represented	a	range	of	industry	sectors	and	employer	sizes	in	the	province	Between	September	2014	and	July	2015,	contact	by	phone	and/or	email	was	attempted	with	all	178	organizations	in	the	database.	An	interview	was	conducted	with	a	representative	employed	at	every	consenting	organization	(each	organization	was	counted	as	one	study	participant).	In	situations	where	the	listed	contact	individual	had	retired,	changed	positions,	or	left	the	organization,	an	alternate	representative	with	appropriate	knowledge	of	work	scheduling	within	the	organization	was	sought	for	interview.	This	was	determined	by	asking	“who	at	your	organization	is	most	knowledgeable	about	work	scheduling?”	The	majority	of	interviews	were	conducted	with	representatives	in	management,	n	=	55;	followed	by	human	resources,	n	=	25;	and	company	leadership	(e.g.,	president,	vice-president,	or	owner),	n	=	8.	In	addition	to	the	shift	work	employer	database,	additional	participants	in	healthcare,	construction,	and	retail	were	recruited	to	ensure	adequate	representation	of	all	British	Columbia	shift	work	sectors	where	shift	work	is	common,	as	identified	in	the	Canadian	Census	(278).	This	recruitment	was	conducted	via	contacts	known	to	the	study	team.		Interviews	were	conducted	by	four	study	team	members	who	had	received	survey	administration	training	by	A.	Hall	to	increase	consistency	in	recording	and	reporting	for	this	study.	Other	measures	to	promote	inter-interviewer	and	temporal	consistency	included	interviewer	prompts	embedded	within	the	survey,	an	interviewer	guide	with	 63 survey	terminology	and	procedural	reminders,	and	group	meetings	conducted	throughout	the	study	to	discuss	survey	progress	and	interview	techniques.	4.2.3 Study	variables		Hypothesized	determinants	of	workplace-level	shift	work	policies	and	practices	were	organized	into	two	groups.	Group	1	represented	more	fixed	characteristics	(less	variable	over	time):	industry	group,	organization	size,	workplace	size,	and	workforce	unionization.	Group	2	represented	more	flexible	characteristics	(more	variable	over	time)	that	were	reported	to	influence	shift	work	scheduling	within	the	organization:	economic	factors	(e.g.,	production	or	business	demand),	client	service	or	care	needs,	site	maintenance	needs	(e.g.,	stocking,	cleaning,	or	equipment	maintenance),	temporal	factors	(e.g.,	changes	across	seasons,	weeks,	or	days),	previous	accidents	or	incidents	that	occurred	during	non-daytime	hours,	concern	for	employee	health	(mental,	physical,	injuries,	or	other),	and	employee	preference	concerning	their	schedule.	Three	discrete	workplace-level	shift	work	policies	or	practices	with	potential	impacts	on	health	were	investigated:		A. Use	of	long	(12	hours	or	more)	versus	short	(less	than	12	hours)	duration	shifts,	derived	from	the	shift	start	and	end	times	reported	by	each	organization	B. Provision	of	shift	work	education	materials/training	to	employees,	obtained	from	the	question:	“Does	your	organization	provide	employees	with	any	training	or	educational	materials	designed	to	help	them	adapt	to	working	shifts?	(e.g.,	safety	meeting,	workshop,	manual,	or	information	pamphlet)”		C. Existence	of	nighttime	lighting	policies	in	the	workplace,	obtained	from	the	question:	“Do(es)	you(r	organization)	have	workplace	policies	regarding	lighting	levels	at	night	(that	is,	official	requirements	to	lower	or	brighten	lighting	at	certain	times?)”		 64 4.2.4 Statistical	analyses	Analyses	were	performed	using	SAS	software,	version	9.4.	(182)	There	were	no	missing	values	for	any	of	the	variables	assessed.	To	assess	the	influence	of	selected	determinants	on	workplace-level	shift	work	practices,	logistic	models	were	constructed	for	each	outcome:	A. Organization	uses	“long”	duration	shifts	of	12	hours	or	more	(Yes/No)	B. Organization	provides	shift	work	education	materials/training	to	employees	(Yes/No)	C. Organization	has	nighttime	lighting	policies	in	the	workplace	(Yes/No)	The	objective	of	this	exploratory	study	was	to	generate	evidence	to	expand	a	limited	body	of	evidence	on	shift	work	practices	and	policies	and	their	determinants.	Therefore,	the	modeling	approach	focused	on	choosing	the	strongest	determinants	of	shift	work	practices	from	the	survey	data.	For	each	outcome,	automatic	stepwise	modeling	was	used	to	select	determinants	from	Variable	Group	1	first	and	then	determinants	from	Variable	Group	2,	using	a	significance	level	cut-off	of	p	=	0.10	for	both	entry	and	retention	in	the	model.	Retained	determinant	variables	from	Group	1	and	Group	2	models	were	then	combined	and	manual	stepwise	regression	was	conducted	to	develop	a	final	logistic	model	for	each	outcome,	using	a	significance	level	cut-off	of	p	=	0.10.	This	level	was	chosen	to	ensure	that	other	potentially	informative	variables	(that	did	not	reach	“statistical	significance”	at	the	traditional	p	=	0.05	level	cut-off)	were	reported.	4.3 Results	4.3.1 Descriptive	summaries	Participating	organizations	were	recruited	from	all	major	industry	sectors	in	British	Columbia	where	shift	work	was	used.	For	industries	employing	larger	proportions	of	shift	workers,	multiple	organizations	were	recruited	to	promote	adequate	representation	of	each	sector.		 65 Out	of	the	original	shift	work	database	(n	=	178	organizations),	13	had	merged	into	6;	18	had	gone	out	of	business;	37	could	not	be	reached	by	the	study	team;	and	33	were	unwilling	to	participate.	This	produced	a	database	response	rate	of	54%	(83	out	of	153	organizations	still	in	business).	Two	participating	organizations	no	longer	had	hours	of	operation	meeting	the	study	definition	of	“shift	work”,	and	were	therefore	excluded.	Targeted	recruitment	(response	rate	of	100%)	produced	an	additional	7	participating	organizations	in	the	healthcare,	construction,	and	retail	sectors.	This	resulted	in	a	final	study	sample	of	88	participating	organizations,	representing	50,000	workers	(out	of	which	30,700	were	shift	workers)	in	the	province	of	British	Columbia.	A	summary	of	study	recruitment	is	provided	in	Appendix	C.	Grouping	by	the	North	American	Industry	Classification	System	(NAICS)	2012	two-digit	level	industry	sectors,	the	largest	numbers	of	participating	organizations	were	located	in	Manufacturing	(n	=	16),	Transportation	and	Warehousing	(n	=	14),	Public	Administration	(n	=	9),	Wholesale	and	Retail	Trade	(n	=	8),	Healthcare	(n	=	8),	and	Accommodation	and	Food	Services	(n	=	7).	To	ensure	adequate	cell	sizes	for	analytic	purposes,	participants	were	grouped	into	three	industry	categories	(shown	in	Table	4.1).		Additional	participating	organization	characteristics	are	presented	in	Table	4.1.	Nearly	half	of	all	organizations	(49%,	n	=	43)	reported	that	concern	for	employee	health	influenced	shift	work	scheduling	at	their	organization;	however	only	one	third	of	organizations	(33%,	n	=	29)	provided	employees	with	education	materials/training	to	assist	with	adaptation	to	shift	work.	One	third	of	organizations	(34%,	n	=	30)	reported	official	workplace	policies	regarding	nighttime	lighting	levels,	and	one	third	of	organizations	(35%,	n	=	31)	used	long	duration	shifts	of	12	hours	or	more.	Responses	to	these	questions	were	not	mutually	exclusive;	6	participating	organizations	reported	“Yes”	for	all	three	outcomes,	and	29	reported	“No”	for	all	three	outcomes.	The	most	common	overlap	was	observed	in	participating	organizations	that	reported	“Yes”	for	both	long	duration	shifts	and	provision	of	shift	work	education	materials/training	(14%,	n	=	12).	 66 Table	4.1:	Characteristics	of	participating	organizations:	British	Columbia	employer	survey,	2014-2015		 	 Long	duration	shifts	(12	hours	or	more)	used1:		Yes	Shift	work	education	materials/training	provided1:	Yes	Nighttime	lighting	policies	in	workplace1:	Yes		 	n	(%	of	all	participants)								 	n	(%	of	all	participants)								 	n	(%	of	all	participants)								 	n	(%	of	all	participants)									 88	(100)	 31	(35)	 29	(33)	 30	(34)	Variable	Group	1:	“Fixed”	characteristics		 	n	(%	of	all	participants)								 	n	(%	of	variable	category)								 	n	(%	of	variable	category)								 	n	(%	of	variable	category)								Industry	group	Government	(municipal,	provincial,	federal)	 25	(28)	 11	(44)	 10	(40)	 10	(40)	Private		(primary	industry	&	manufacturing)	 23	(26)	 12	(52)	 9	(39)	 5	(22)	Private		(service	&	accommodation)	 40	(46)	 8	(20)	 10	(25)	 15	(38)	Organization	size	<1000	employees	 43	(49)	 9	(21)	 11	(26)	 12	(28)	1000+	employees	 45	(51)	 22	(49)	 18	(40)	 18	(40)	Workplace	size	<99	employees	 42	(48)	 5	(12)	 8	(19)	 13	(31)	100+	employees	 46	(52)	 26	(57)	 21	(46)	 17	(37)	Workforce	unionized	Yes	 56	(64)	 23	(41)	 23	(41)	 20	(36)	No	 32	(36)	 8	(25)	 6	(19)	 10	(31)		 	 	 	 	Variable	Group	2:	“Flexible”	characteristics	(reported	to	influence	shift	work	scheduling)		 	n	(%	of	all	participants)								 	n	(%	of	variable	category)								 	n	(%	of	variable	category)								 	n	(%	of	variable	category)								Economic	factors	(e.g.,	production	or	business	demand)	Yes	 64	(73)	 20	(31)	 22	(34)	 22	(34)	No	 24	(27)	 11	(46)	 7	(29)	 8	(33)	Client	service	or	care	needs	Yes	 56	(64)	 18	(32)	 18	(32)	 24	(43)	No		 32	(36)	 13	(41)	 11	(34)	 6	(19)	Site	maintenance	needs	(e.g.,	stocking,	cleaning,	or	equipment	maintenance)		 67 	 n	(%	of	all	participants)	 n	(%	of	variable	category)	 n	(%	of	variable	category)	 n	(%	of	variable	category)	Yes	 29	(33)	 11	(38)	 10	(34)	 14	(48)	No		 59	(67)	 20	(34)	 19	(32)	 16	(27)	Temporal	factors	(seasonal	work)	Yes	 49	(56)	 11	(22)	 11	(22)	 21	(43)	No		 39	(44)	 20	(51)	 18	(46)	 			9	(23)	Temporal	factors	(days	of	the	work	week)	Yes	 39	(44)	 12	(31)	 14	(36)	 15	(38)	No		 49	(56)	 19	(39)	 15	(31)	 15	(31)	Temporal	factors	(hours	of	the	work	day)	Yes	 42	(48)	 11	(26)	 17	(40)	 16	(38)	No		 46	(52)	 20	(43)	 12	(26)	 14	(30)	Collective	agreements	Yes	 44	(50)	 19	(43)	 20	(45)	 18	(41)	No		 44	(50)	 12	(27)	 		9	(20)	 12	(27)	Previous	accidents	or	incidents	that	occurred	during	non-daytime	hours	Yes	 16	(18)	 		8	(50)	 		8	(50)	 10	(63)	No	 72	(82)	 23	(32)	 21	(29)	 20	(28)	Concern	for	employee	health	(mental,	physical,	injuries,	or	other)		Yes	 43	(49)	 19	(44)	 19	(44)	 17	(40)	No	 45	(51)	 12	(27)	 10	(22)	 13	(29)	Employee	preference	Yes	 49	(56)	 18	(37)	 16	(33)	 17	(35)	No		 39	(44)	 13	(33)	 13	(33)	 13	(33)	1	Outcomes	are	not	mutually	exclusive	and	therefore	do	not	add	to	100%	4.3.1 Logistic	regression	Associations	between	determinant	variables	and	shift	work	outcomes	are	presented	in	Table	4.2	for	the	final	multivariable	logistic	regression	models.	The	odds	ratio	for	long	duration	shifts	was	higher	in	organizations	with	larger	workplaces,	and	in	government	organizations	and	private	organizations	in	primary	industry	and	manufacturing	(relative	to	private	organizations	in	service	and	accommodation	industries).	The	odds	ratio	for	providing	shift	work	education	materials/training	was	higher	in	organizations	with	larger	workplaces,	in	organizations	that	reported	no	seasonal	changes	in	shift	work,	and	in	organizations	that	reported	shift	work	scheduling	related	to	concerns	about	employee	health.	The	odds	ratio	for	having	nighttime	lighting	policies	in	the	workplace	was	higher	in		 68 Table	4.2:	Final	logistic	models	for	associations	between	determinant	variables	and	outcomes	of	(1)	long	duration	shift	use,	(2)	provision	of	shift	work	materials/training	to	employees,	and	(3)	nighttime	lighting	policies	in	the	workplace,	British	Columbia	employer	survey,	2014-2015	Organization	uses	long	duration	shifts	(12	hours	or	more)	(yes	versus	no)	Workplace	size	OR	 95%	CI			<99	employees	 1.0		 -			100+	employees	 11.63		 3.56	-	38.02	Industry	group	 				Private	(service	&	accommodation)	 1.0		 -			Private	(primary	industry	&	manufacturing)	 5.73		 1.53	-	21.47			Government	(municipal,	provincial,	federal)	 4.14		 1.15	-	14.93	Organization	provides	shift	work	education	materials/training	to	employees	(yes	versus	no)	Workplace	size	OR	 95%	CI			<99	employees	 1.0		 -			100+	employees	 2.81		 1.01	-	7.78	Shift	work	related	to	seasonal	needs	 				Yes	 1.0		 -			No		 3.37		 1.20	-	9.46	Shift	work	related	to	organizational	concern	for	employee	health	(mental,	physical,	injuries,	or	other)					Yes	 3.33		 1.18	-	9.38			No	 1.0		 -	Organization	has	nighttime	lighting	policies	in	the	workplace	(yes	versus	no)	Shift	work	related	to	previous	accidents	or	incidents	that	occurred	during	non-daytime	hours		OR	 95%	CI			Yes	 3.80		 1.16	-	12.49			No	 1.0	 -	Shift	work	related	to	site	maintenance	needs		 				Yes	 2.48		 0.92	-	6.68			No		 	1.0	 -	Shift	work	related	to	client	service	or	care	needs						Yes	 2.74		 0.93	-	8.08			No		 1.0	 -	OR	=	Odds	Ratio;	CI	=	Confidence	Interval	 69 organizations	that	reported	shift	work	scheduling	related	to	previous	non-daytime	workplace	accidents	or	incidents,	in	organizations	that	reported	shift	work	scheduling	related	to	maintenance	activities,	and	in	organizations	that	reported	shift	work	scheduling	related	to	client	service	or	care	needs.	4.4 	Discussion	This	study	identified	determinants	of	workplace-level	shift	work	policies	and	practices	among	a	sample	of	88	employers	in	the	Canadian	province	of	British	Columbia,	representing	over	30,000	shift	workers.	Over	400	distinct	shift	work	schedules	were	reported	(results	not	shown),	providing	yet	another	example	of	shift	work’s	complexity. Larger	workplace	size	was	strongly	associated	with	the	outcomes	of	long	duration	shifts	and	the	provision	of	education	resources	to	shift	workers,	but	not	to	the	presence	of	nighttime	lighting	policies.	Industry	group	was	also	strongly	associated	with	long	duration	shifts,	with	government	and	private	organizations	in	primary	industry	&	manufacturing	sectors	more	likely	to	use	long	duration	shifts	than	private	organizations	in	service	&	accommodation	sectors.	Factors	related	to	shift	work	scheduling	within	an	organization,	including	seasonal	(temporary)	work	and	concern	for	worker	health,	were	strongly	associated	with	employer	provision	of	shift	work	education	materials/training	to	employees.	Other	factors	related	to	shift	work	scheduling,	including	previous	workplace	accidents	and	incidents	occurring	during	non-day	shifts,	site	maintenance	needs,	and	client-	or	service-based	needs,	were	associated	with	the	presence	of	nighttime	lighting	policies	in	the	workplace.		4.4.1 Long	duration	shifts	Approximately	one	third	(35%,	n	=	31)	of	participating	organizations	in	this	study	reported	the	use	of	long	duration	shifts	(12	hours	or	more)	in	their	workplaces,	of	which	the	majority	(n	=	29)	were	12	hours.	The	odds	ratio	for	long	duration	shifts	was	higher	as	a	function	of	larger	workplace	size	and	type	of	industry	sector,	with	government	organizations	and	private	organizations	in	primary	industry	and	manufacturing	more	likely	 70 to	use	shifts	with	extended	durations	relative	to	private	organizations	in	service	and	accommodation	industries.	These	findings	agree	with	previously	noted	differences	in	shift	work	characteristics	across	industries.	Split	shifts,	which	are	typically	of	shorter	duration,	are	most	common	in	retail	trade,	hospitality,	and	entertainment/recreation	industries	(4,280);	all	of	which	were	categorized	into	this	study’s	industry	group	of	private	organizations	in	service	and	accommodation.	Employees	in	primary	industry	(such	as	mining	operations)	in	British	Columbia	may	perform	many	consecutive	long	duration	shifts	while	working	in	remote	areas.	The	use	of	long	duration	shifts	may	also	arise	from	historical	traditions	of	working	patterns	in	some	industries,	such	as	in	municipal	protection	services	(included	in	this	study),	or	in	large	employers	that	have	been	established	for	long	periods.	Evidence	and	implications	concerning	the	health	effects	of	long	shift	durations	is	discussed	further	in	Section	5.2.	4.4.2 Provision	of	shift	work	education	materials/training	to	employees	In	this	study,	only	one	third	(33%,	n	=	29)	of	participating	organizations	reported	the	provision	of	education	materials	or	training	designed	to	assist	with	adaptation	to	shift	work.	This	low	proportion	may	indicate	a	lack	of	recognition	regarding	shift	work’s	potential	health	impacts,	or	there	may	be	insufficient	expertise	at	the	workplace	level	to	obtain,	evaluate,	and	distribute	shift	work	education	materials	and	training	to	employees	in	some	instances.		Concerning	the	impacts	of	workplace	size	on	shift	work	education	or	training,	a	body	of	research	indicates	that	small	workplaces	have	relatively	greater	difficulties	controlling	occupational	health	and	safety	risks	(281–283)	and	may	not	always	use	information	and	supports	when	they	are	available,	due	to	economic	and	cultural	factors	(282).	Conversely,	larger	organizations	tend	to	have	greater	access	to	worker	health	expertise	(e.g.,	industrial	hygienist	on	staff,	health	and	safety	committees)	and	economic	resources	to	distribute	educational	resources.	In	the	Canadian	context,	it	has	been	suggested	that	occupational	health	and	safety	practices	in	small	workplaces	could	be	improved	through	the	 71 development	of	information	systems	that	gather	occupational	health	and	safety	data	by	workplace	size;	such	information	could	then	be	used	to	inform	the	case	for	legislative	and	policy	changes	that	incorporate	small	business	considerations	(283).	Another	study	conducted	in	the	Canadian	province	of	Quebec	found	that	the	most	common	prevention	activities	in	small	firms	were	linked	to	production	efficiency	(284).	This	suggests	that	the	integration	of	prevention	into	production	activities	(e.g.,	championing	the	business	incentives	of	shift	worker	education	and	training)	could	be	an	effective	strategy	for	shift	workers	in	small	workplaces.	Temporary	shift	work	(defined	in	this	study	as	differences	in	shift	work	scheduling	across	months	or	seasons)	was	reported	by	over	50%	of	organizations	in	industries	such	as	Construction,	Agriculture,	and	Accommodation	and	Food	Services.	This	study’s	finding	that	organizations	with	seasonal	(precariously	employed)	shift	workers	are	less	likely	to	provide	shift	work	education	is	consistent	with	other	research	that	suggests	the	transient	nature	of	temporary	workforces	challenges	the	maintenance	of	workplace	health	programs,	risk	assessment,	and	injury/health	surveillance	(222).	Temporary	workers	have	also	been	noted	to	receive	less	training	during	paid	working	hours	relative	to	permanent	workers	(254,285).	The	issue	of	how	to	support	precariously	employed	workers	has	recently	received	greater	attention	in	Canada	(254).	Recommendations	focus	on	taking	a	broader	view	to	workforce	development,	by	creating	government	policies	that	are	responsive	to	labour	market	needs	yet	also	ensure	access	to	training	opportunities	that	support	precariously	employed	workers	and	enable	them	to	find	more	secure	employment	(254).		The	finding	that	employer	concern	for	health	drives	the	provision	of	shift	work	education	and	training	is	somewhat	intuitive.	Out	of	the	29	participating	organizations	that	provided	shift	work	education	and	training	to	their	employees,	13	reported	concern	about	mental	health,	12	about	physical	health,	9	about	injuries,	and	7	about	“other”	health	concerns	(6	of	the	7	were	fatigue-related).	While	this	is	interesting,	the	interview	structure	of	this	study	cannot	identify	which	specific	health	concerns	led	to	the	introduction	of	shift	work	education	materials/training	in	participating	organizations.	However,	taken	together	with	 72 findings	on	prior	history	of	work-place	injuries,	the	results	suggest	that	general	concern	about	shift	worker	health	and	safety	may	be	a	good	lever	for	the	implementation	and	provision	of	workplace-level	shift	work	policies	and	resources.	There	has	been	surprisingly	little	published	research	on	the	short-	and	long-term	efficacy	of	shift	work	education	programs.	Where	provided,	the	type	and	quality	of	shift	work	education	materials	and	programs	varies	widely	across	organizations	(279),	and	such	information	does	not	always	reach	its	intended	audience	(159).	Further	research	is	needed	to	assess	which	types	of	training	programs	and	materials	are	effective	for	enhancing	tolerance	to	shift	work,	and	to	ensure	that	such	interventions	provide	improvements	to	health	and	well-being	without	inducing	further	harm.	This	is	particularly	important	given	the	complexities	of	individual	factors	(e.g.,	age,	gender,	genetics)	and	other	mechanisms	that	may	drive	shift	work’s	effects	on	health,	such	as	sleep,	social	stress,	lifestyle,	and	environmental	factors	(17,19,146).	The	need	for	input	from	shift	workers	themselves	has	also	been	identified	as	a	future	priority	in	this	area,	to	ensure	that	working	environment	interventions	to	improve	shift	worker	health	actually	reach	their	intended	targets	(159).		Government	incentives	to	support	the	provision	of	education	and	training	to	shift	workers	could	also	serve	as	a	lever	to	promote	the	broad	reach	of	such	individual-level	interventions.	A	specific	recommendation	that	could	be	applied	to	shift	work	education	practices	refers	to	an	already	established	(and	purportedly	successful)	model	in	the	province	of	Quebec,	where	employee	training	is	encouraged	through	a	government	tax	incentive	for	employers,	and	a	1%	payroll	tax	applied	to	employers	who	do	not	provide	training	(254).	In	turn,	the	planning	and	implementation	of	other	workplace-level	health	interventions,	such	as	a	shift	system	change,	would	be	supported	by	worker	education	and	training,	since	information,	communication,	and	worker	participation	have	been	identified	as	important	elements	in	this	process	(71).		 73 4.4.3 Nighttime	lighting	policies	Official	nighttime	lighting	policies	in	the	workplace	were	reported	by	34%	(n	=	30)	of	participating	organizations	in	this	study.	This	suggests	that	considerations	of	shift	worker	health	do	not	typically	drive	decisions	concerning	workplace	lighting,	although	interpretations	are	limited	by	a	lack	of	explicit	detail	on	the	policies	themselves	(for	example,	dimming	versus	brightening	of	lights	at	certain	times).	The	odds	ratio	for	having	nighttime	lighting	policies	in	the	workplace	was	higher	in	organizations	that	reported	shift	work	scheduling	related	to	previous	accidents	or	incidents	during	non-daytime	hours,	client	service	or	care	needs,	and	site	maintenance	activities.	Specific	motivating	factors	and	circumstances	for	nighttime	lighting	policies	were	not	assessed	in	this	study,	although	determinants	showing	the	strongest	associations	with	light	policies	have	face	value.	For	example,	client	service/care	needs	and	site	maintenance	activities	are	logical	drivers	of	lighting	use.	Previous	accidents/incidents	in	the	workplace	were	also	associated	with	nighttime	lighting	policies.	Of	the	30	participating	organizations	that	reported	nighttime	light	policies,	33%	(n=10)	reported	that	a	previous	nighttime	incident	had	affected	shift	work	scheduling	practices	in	the	organization.	Of	the	58	participating	organizations	that	did	not	report	nighttime	light	policies,	10%	(n=6)	reported	that	a	previous	nighttime	incident	had	affected	shift	work	scheduling	practices	in	the	organization.	This	suggests	that	previous	nighttime	incidents	or	accidents	might	serve	as	a	lever	or	instigator	for	the	implementation	of	best	practices.	For	example,	workplace	incidents	may	prompt	an	occupational	health	and	safety	investigation	or	regulatory	inspection	activity	that	leads	to	change,	particularly	when	penalties	are	applied	(286).		Workplace-level	interventions	may	be	driven	by	a	range	of	motivations	other	than	concern	for	employee	health,	such	as	structural	changes	(downsizing	or	upsizing),	or	other	business	concerns	(e.g.,	efficiency,	productivity,	cost)	(287).	This	study	cannot	verify	that	concern	about	employee	health	was	a	major	driver	for	nighttime	lighting	policies	in	the	organizations	that	reported	them.	Of	the	30	participating	organizations	that	reported	nighttime	light	policies,	57%	(n	=	17)	reported	that	concern	about	employee	health	affected	shift	work	scheduling	within	their	organization,	versus	45%	(n	=	26)	of	the	58	 74 participating	organizations	that	did	not	report	nighttime	light	policies.	In	another	survey	question	(results	not	shown),	14%	(n	=	12)	of	participating	organizations	did	report	providing	bright	lighting	to	enhance	shift	worker	performance,	alertness,	safety,	and	well-being.	These	findings	suggest	that	lighting	policies	may	be	driven	somewhat	by	concern	for	worker	safety	and	health,	however	“energy	efficiency”	and	“cost	savings”	were	mentioned	by	multiple	participating	organizations	as	the	driving	factors	for	nighttime	lighting	policies	(results	not	shown).	Therefore,	future	research	that	makes	both	health	and	economic	arguments	for	workplace	interventions	could	provide	a	good	incentive	for	organizations	to	implement	nighttime	lighting	policies	with	multiple	benefits.	4.4.4 Strengths	and	limitations	This	exploratory	study	of	determinants	of	workplace-level	shift	work	policies	and	practices	used	stepwise	modeling	to	select	the	best	combination	of	variables	for	each	shift	work	outcome,	before	offering	these	variables	jointly	to	build	the	final	models.	While	the	majority	of	determinant	variables	in	the	final	models	met	the	more	traditional	cut-off	value	of	p	=	0.05,	a	less	restrictive	cut-off	value	of	p	=	0.10	provided	information	on	additional	determinants	of	shift	work	practices	(site	maintenance	needs	and	client	service/care	needs)	that	may	be	useful	to	consider	in	future	research	and	intervention	strategies.	Although	Type	I	error	(the	incorrect	identification	of	an	effect	that	is	not	truly	present)	is	possible	given	the	number	of	variables	and	models	that	were	analyzed,	the	final	workplace	determinants	have	face	validity	to	support	their	associations	with	assessed	shift	work	outcomes.	The	study	results	also	included	the	95%	confidence	intervals	around	the	odds	ratios	to	provide	evidence	of	the	precision	around	these	estimates.		The	widths	of	these	intervals	reflect	the	study’s	small	sample	size	and	variability	in	reported	shift	work	practices.	Despite	imprecision	around	the	point	estimates,	the	strength	of	associations	and	their	consistency	with	other	literature	supports	their	validity.	The	study’s	small	sample	size	(n	=	88)	required	the	aggregation	of	some	explanatory	variables	to	increase	cell	sizes	and	stabilize	regression	estimates;	this	resulted	in	the	 75 masking	of	potentially	valuable	information	at	a	finer	level	(e.g.,	industry	type,	workplace	size).	The	small	sample	size	also	restricted	the	investigation	of	additional	shift	work	practices	with	potential	benefit	to	shift	workers,	such	as	flexible	working	time	arrangements	(61,63,288)	and	type	of	rotational	shift	work	(60,62).	For	example,	few	participating	organizations	in	this	study	reported	worker	self-scheduling	(16%;	n	=	14)	or	backward	rotations	(2%;	n	=	2)	(results	not	shown)	precluding	analyses	of	relationships	between	these	variables	and	workplace-level	determinants	in	this	sample.	The	small	numbers	observed	also	suggest	that	employers	in	British	Columbia	may	not	commonly	use	such	scheduling	policies	and	practices.	The	individuals	interviewed	on	behalf	of	participating	organizations	were	selected	based	on	their	familiarity	with	shift	work	scheduling	at	each	organization.	Due	to	differences	in	workplace	size	and	employment	position	(scheduling	versus	management)	interviewees	may	not	have	been	equally	familiar	with	“on	the	ground”	practices	within	workplaces,	and	some	misclassification	may	have	occurred	for	both	the	determinant	variables	and	outcome	variables.	Although	not	ideal,	such	misclassification	would	likely	be	non-differential,	and	therefore	more	likely	to	bias	effect	measures	towards	the	null	(not	produce	spurious	effects).	If	social	desirability	bias	affected	the	reporting	of	workplace	concern	about	employee	health,	the	association	between	this	variable	and	the	provision	of	shift	work	education	materials/training	to	employees	may	have	been	overestimated.	However,	less	than	half	(49%)	of	participating	organizations	indicated	that	concern	for	employee	health	influenced	shift	work	practices	within	their	organization,	suggesting	that	this	effect	may	be	small.		This	study	included	organizations	from	a	range	of	industry	sectors	that	were	hypothesized	to	be	a	potential	source	of	variability	in	shift	work	practices.	Although	a	range	of	workplace	sizes	was	also	sought,	the	final	study	sample	over-represents	large	workplaces	(16%	of	participating	organizations	represented	workplaces	with	more	than	500	employees,	versus	the	8%	provincial	average)	and	under-represents	small	workplaces	(9%	of	participating	organizations	represented	workplaces	with	fewer	than	20	employees,	versus	the	38%	provincial	average)	(289).	Since	results	may	vary	due	to	the	resources	available	within	 76 workplaces	of	differing	size,	more	research	is	needed	to	clarify	determinants	of	shift	work	practices	in	smaller	organizations	that	employ	shift	workers.	4.5 Conclusion	This	study	represents	the	first	known	analytic	examination	of	determinants	of	workplace-level	shift	work	policies	and	practices.	Results	point	to	high-level	characteristics	that	may	be	used	to	guide	interventions	and	research	where	they	are	most	needed	(or	most	likely	to	be	effective)	concerning	shift	work	scheduling,	light-at-night,	and	health	promotion	for	shift	workers.	For	example,	industry	sector	and	workplace	size	may	be	important	considerations	in	research	or	interventions	focused	on	shift	duration.	Workplace	size,	seasonal	changes	in	shift	work,	and	employer	concern	for	worker	health	may	be	important	considerations	in	research	or	interventions	focused	on	individual-level	health	promotion.	Previous	workplace	incidents	occurring	during	non-day	shifts,	site	maintenance	needs,	and	client-	or	service-based	needs	may	be	important	considerations	in	studies	or	interventions	focused	on	nighttime	lighting	policies	and	practices	in	workplaces.								 77 Chapter	5: Discussion	Shift	work	is	a	prevalent	work	arrangement	in	modern	society	that	can	disrupt	workers’	natural	rhythms	of	biology	and	socialization.	Although	it	may	not	fit	into	the	traditional	occupational	hygiene	paradigm,	shift	work	appears	to	be	an	important	causal	factor	for	a	number	of	health	outcomes,	and	should	assessed	with	the	same	rigour	as	other	occupational	hazards.	In	fact,	greater	rigour	in	exposure	assessment	may	be	warranted,	since	the	composition	and	intensity	of	shift	work	exposures	extend	beyond	traditional	factors	such	as	job	title	and	workplace	characteristics	(290).	High	quality	workplace-based	research	is	needed	to	assess	shift	work’s	effects	on	health,	including	intervention	studies	aimed	at	minimizing	its	impacts.	However,	a	number	of	methodological	issues	in	epidemiological	studies	of	shift	work	have	limited	the	development	of	evidence-based	interventions	and	policies	to	support	shift	worker	health.	One	major	methodological	challenge	is	that	shift	work	represents	a	highly	variable	mixture	of	exposures	that	is	difficult	to	study.	This	reality	will	endure	for	as	long	as	shift	work	exists;	in	other	words,	indefinitely.	Therefore,	an	important	and	relevant	question	for	health	researchers	is:	How	can	we	do	a	better	job	of	studying	shift	work,	in	order	to	develop	stronger	evidence	and	thus	clearer	guidance	to	reduce	its	negative	impacts?		It	has	been	proposed	that	the	ability	to	cope	with	shift	work	relies	on	the	mutually	interactive	domains	of	the	circadian,	sleep,	and	social/domestic	systems	(291).	It	is	therefore	interesting	that	much	shift	work	research	has	assessed	simple	linear	relationships,	often	focusing	on	biomedical	models	of	disrupted	sleep	and	circadian	rhythms	(i.e.,	the	time-of-day	effects	of	work	schedule),	while	ignoring	other	pathways	that	may	link	shift	work	exposures	to	health	(as	described	in	section	1.1.2).	Theoretical	development	and	testing	has	not	been	a	primary	focus	of	shift	work	research	(1).	Attempts	to	develop	more	complex	models	that	incorporate	general	stress	concepts	have	been	lauded	for	increasing	diversity	in	shift	work	research,	yet	have	also	been	criticized	for	their	lack	of	clarity	and	utility	as	heuristic	frameworks	rather	than	descriptions	of	data	(128).	Therefore,	as	was	recently	emphasized	elsewhere	(292),	opportunities	exist	to	develop	 78 comprehensive	shift	work	models	that	thoroughly	and	accurately	characterize	the	strongest	pathways	(biological,	social,	and	psychological)	linking	shift	work	with	health.	Such	work	is	needed	in	order	to	identify	the	strongest	pathways	(biological,	social,	and	psychological)	linking	shift	work	with	health.	Various	methods	have	been	suggested	for	such	refinements,	including:	1)	investigating	research	questions	that	arise	from	existing	shift	work	models	and	using	findings	to	clarify/diversify	theory	(e.g.,	incorporate	stress	research	on	individual	coping	mechanisms);	2)	developing	narrower	theories	that	specify	the	shift	work	features	that	are	related	to	health	outcomes,	and	in	which	organizational	circumstances;	3)	developing	symptom-focused	theories	to	explain	the	etiology	of	individual	health	problems	among	shift	workers,	by	describing	specific	physiological	and/or	psychological	mechanisms	(128).		Regardless	of	the	method	pursued,	it	is	clear	that	future	research	focused	on	the	effects	of	shift	work	on	health	should	take	a	comprehensive	approach	to	assessing	the	individual	in	addition	to	the	combined	effects	of	shift	scheduling,	work	demands,	sleep,	and	the	psychosocial	context	at	work	and	home.	In	these	studies,	multiple	outcome	measures	should	be	included	on	circadian	adjustment,	sleep,	and	recovery,	corresponding	to	the	hypothesized	underlying	health	mechanisms.	The	contribution	of	non-work	activities	and	stressors	should	also	be	assessed	to	understand	how	time	off	and	demands	outside	of	work	may	contribute	to	circadian	entrainment,	sleep,	and	recovery	(292).	Researchers	interested	in	a	wider	conceptualization	of	upstream	factors	relating	to	work	organization	and	shift	work	may	choose	to	integrate	broader	social	determinants	of	health	into	such	models.		The	insufficiency	of	simple	approaches	to	exposure	assessment	is	not	a	new	concept	in	occupational	epidemiology	(293),	and	the	need	for	more	sophisticated	and	standardized	exposure	assessment	in	epidemiological	studies	of	shift	work	has	increasingly	been	realized	in	recent	years	(21,23,80).	To	move	this	science	forward,	we	must	progress	beyond	the	status	quo	use	of	crude	exposure	assessment.	Relevant	shift	work	characteristics	and	their	variability	should	be	considered	at	every	step	of	the	epidemiological	process	(i.e.,	during	planning,	data	collection,	analysis,	and	interpretation)	to	meaningfully	describe	exposures	and	reduce	the	potential	for	exposure	misclassification,	 79 measurement	error,	and	bias.	These	considerations	require	the	collection	of	detailed	exposure	measurements	(e.g.,	light-at-night	exposure	data,	details	of	a	work	schedule,	history	of	exposure)	to	examine	which	fundamental	aspects	of	exposure	are	important	to	assess	underlying	relationships	(290).	Such	data	is	needed	to	inform	the	assignment	of	exposure	indices	that	are	specific	and	relevant	to	the	health	outcome(s)	of	interest	(or,	in	cases	where	a	priori	mechanisms	are	not	well	understood,	to	compare	different	exposure	indices	and	metrics	in	the	analysis)	(293).	Data	that	describe	determinants	of	workplace-level	shift	work	policies	and	practices	are	also	needed	to	inform	targeted	research	and	interventions	aimed	at	mitigating	population-level	health	risks,	based	on	strong	evidence	of	associations	where	they	exist.	5.1 Summary	of	studies,	methodological	considerations,	and	recommendations	for	future	research	This	dissertation	presents	a	series	of	three	exposure	assessment	studies	focused	on	the	measurement,	assignment,	and	determinants	of	shift	work	exposure	in	epidemiological	research.	These	studies	provide	new	information	that	can	be	applied	to	strengthen	future	studies,	including:	quantitative	data	that	describe	light-at-night	exposure	levels	and	their	variability	in	shift	workers,	and	an	examination	of	exposure	metrics	(in	Chapter	2),	evidence	to	support	the	assignment	of	specific	exposure	indices	that	are	based	upon	clearly	formulated	hypotheses	of	exposure-response	relationships	(in	Chapter	3),	and	new	insights	into	determinants	of	workplace-level	shift	work	policies	and	practices	(in	Chapter	4).	In	Chapter	2,	light-at-night	exposure	levels	in	shift	workers	were	documented	as	an	initial	step	in	the	measurement	of	exposure	levels	and	variability.	Personal	full-shift	light-at-night	exposure	measurements	(n	=	152)	were	collected	from	102	shift	workers	in	emergency	health	services	(paramedics,	dispatchers)	and	healthcare	industries	(nurses,	unit	clerks,	security	guards,	and	support,	pharmacy,	and	laboratory	staff).	Descriptive	and	variance	component	analyses	were	conducted	for	the	23:00-05:00	period	to	characterize	exposures	using	multiple	metrics	of	potential	biological	relevance.	The	highest	exposures	were	observed	in	medical	laboratory	workers	in	healthcare	while	the	lowest	exposures	were	observed	in	dispatch	officers	in	emergency	services.	Between-group	variance	was	large	 80 relative	to	between-worker	and	within-worker	variance	for	all	exposure	groupings	and	metrics,	with	variance	increasing	as	grouping	precision	increased.	All	exposure	metrics	were	moderately	to	highly	correlated.		This	study	provides	new	information	about	light-at-night	exposure	levels	and	variability	in	emergency	services	and	healthcare	workers,	using	multiple	exposure	indices	and	exposure	metrics	that	were	based	on	a	current	understanding	of	human	physiologic	responses	to	light.	Findings	suggest	that	light-at-night	exposures	may	be	more	temporally	consistent	than	other	traditional	workplace	exposures,	and	that	a	number	of	exposure	metrics	and	high-level	grouping	schemes	may	be	useful	to	characterize	and	describe	these	exposures.	While	the	methods	used	to	assess	variability	in	this	study	have	previously	been	used	to	assess	occupational	hazards	such	as	chemical	agents	(162),	dusts	(294),	and	musculoskeletal	strain	(168),	they	are	new	to	the	field	of	shift	work.	Measurement	collection	and	characterization	of	light-at-night	exposures	in	shift	workers	could	provide	significant	improvements	over	the	standard	use	of	“shift	work”	or	“shift	work	involving	nights”	as	a	proxy	for	light-at-night	exposure	in	studies	of	shift	work	and	health.	New	findings	from	this	study	are	informative,	but	they	should	not	be	regarded	as	prescriptive	for	all	shift	workers	in	healthcare	or	emergency	services.	Light-at-night	exposures	occurring	outside	of	the	assessed	sample	could	be	different	with	respect	to	their	intensity	or	temporal	variability.	As	described	elsewhere	(134),	the	collection	of	quantitative	exposure	measurements	prior	to	an	epidemiological	study	can	be	a	useful	way	to	maximize	it	impacts.	Such	data	can	be	used	to	inform	considerations	of	exposure	variability	in	a	target	research	population,	such	as	sampling	strategies	that	increase	study	efficiency,	as	well	as	the	assignment	of	grouping	schemes	that	maximize	accuracy	and	exposure	contrast	for	analyses.	For	these	reasons,	light-at-night	exposure	should	be	measured	and	assessed	prior	to	any	large	epidemiological	study	seeking	to	investigate	its	effects	on	health.			Research	into	light-at-night	exposure	and	health	is	a	new	and	evolving	scientific	area,	with	many	questions	still	to	be	answered	about	which	exposure	characteristics	(e.g.,	duration,	timing,	illumination	levels,	spectral	characteristics)	carry	the	strongest	impacts	on	a	variety	 81 of	health	outcomes.	The	assessment	of	photopic	illuminance	levels	and	variability	in	Chapter	2	allows	for	comparisons	with	most	prior	epidemiological	studies	of	nighttime	light	exposure	on	health.	Photopic	illuminance	is	an	appropriate	exposure	measure	for	studies	of	short-term	alertness	(172,196,197)	in	shift	workers,	and	is	also	a	relevant	measure	to	assess	melatonin	suppression	in	humans	(295,296).	However,	collecting	data	on	light-at-night	as	it	affects	the	visual	system	may	not	be	sufficient	to	fully	characterize	circadian	disruption	in	shift	workers	(181),	since	light	exposure	on	the	retina	affects	the	human	visual	and	circadian	systems	differently	(199).	For	example,	visual	reaction	time	is	in	the	order	of	milliseconds	and	can	be	stimulated	by	light	ranging	from	starlight	(~0.0001	lux)	to	midday	sun	on	a	clear	day	(~100,000	lux)	(199),	whereas	the	circadian	system	requires	sustained	light	stimulation	with	a	threshold	of	~30	lux	(saturation	occurring	at	~1000	lux)	(199).	Light-at-night’s	suppressive	effect	on	melatonin	also	depends	on	the	type	and	timing	of	light	exposure,	with	greater	impacts	observed	with	longer	exposure	to	high	intensity	light	in	the	short	wavelength	(“blue”)	range	during	periods	of	biological	darkness	(70).	Therefore,	while	Chapter	2	provides	useful	new	information	on	shift	workers’	light-at-night	exposure	levels	and	variability	during	the	biological	period	of	darkness,	there	are	opportunities	to	expand	upon	this	assessment	in	future	research	by	using	measures	of	the	circadian	system’s	sensitivity,	threshold,	and	saturation.		An	accumulation	of	strong	evidence	in	this	area	could	be	used	in	future	to	inform	lighting	standards	that	consider	the	health	impacts	of	lighting.	To	date,	authorities	such	as	the	Illuminating	Engineering	Society	and	the	American	Society	of	Heating,	Refrigerating	and	Air-Conditioning	Engineers	(ASHRAE)	have	only	incorporated	considerations	of	the	human	visual	system	(e.g.,	needs	for	safety	and	task	performance),	but	not	considerations	of	neuroscience	(e.g.,	nocturnal	melatonin	response,	circadian	stimuli)	into	regulated	lighting	standards	(67).	It	has	been	suggested	that	various	lighting	benefits	be	incorporated	into	a	broader	definition	for	use	in	lighting	standards	(67);	to	do	so,	collaboration	between	specialists	in	neuroscience	research	and	lighting	applications	is	necessary	to	develop	practical,	evidence-based	lighting	changes	that	both	improve	human	health	and	performance,	and	increase	energy	efficiency.	 82 In	Chapter	3,	the	implications	of	exposure	assignment	precision	on	the	nature	and	strength	of	shift	work’s	relationship	with	depression	was	investigated	using	a	large,	nationally	representative	survey	of	Canadian	nurses.	High,	moderate,	and	low	precision	shift	work	exposure	indicator	groups	were	used.	The	high	precision	exposure	indicator	group	considered	both	shift	timing	and	frequency	of	rotation,	while	the	moderate	precision	group	considered	only	shift	timing,	and	the	low	precision	group	considered	only	the	presence	or	absence	of	shift	work.	In	the	analyses,	the	point	estimates	of	associations	observed	in	the	high	precision	shift	work	schedule	model	(incorporating	considerations	of	rotation	frequency)	were	stronger	than	those	of	the	medium	and	low	precision	models,	where	weak	associations	were	observed	for	all	schedule	categories.		The	assignment	of	exposure	indicators	is	an	important	scientific	consideration	in	epidemiological	studies.	A	poorly	chosen	indicator	of	exposure	can	result	in	misclassification	of	subjects’	exposures,	thereby	introducing	measurement	error	and	distorting	relationships	(166).	Unfortunately,	these	important	concepts	are	frequently	ignored	in	epidemiological	studies	of	shift	work,	where	the	assignment	of	coarse	exposure	indicators	has	been	the	norm.	In	Chapter	3,	neither	the	low	nor	the	moderate	precision	exposure	definition	produced	strong	associations	between	shift	schedule	and	depression.	However,	the	high	precision	exposure	definition	(that	included	considerations	of	shift	timing	and	rotation	frequency)	found	positive	associations	between	high	frequency	shift	rotations	and	depression,	and	negative	associations	between	low	frequency	shift	rotations	and	depression,	relative	to	regular	day	workers.	These	results	demonstrate	that	in	this	study	population,	exposure	assignment	considering	only	shift	work	absence/presence,	or	considering	only	shift	timing	and	not	rotation	frequency,	would	have	produced	exposure	misclassification	and	a	conclusion	of	“no	effect”	for	depression	related	to	shift	work.	Findings	from	this	study	therefore	reinforce	the	need	to:	1)	collect	sufficiently	detailed	exposure	information	from	research	participants,	and	2)	apply	this	information	to	assign	specific	exposure	indices,	in	order	to	minimize	the	potential	for	measurement	error	due	to	exposure	misclassification.			 83 This	task	of	detailed	exposure	data	collection	and	assignment	should	be	informed	by	considerations	of	causal	mechanisms	that	are	outcome	specific,	since	different	types	of	shift	work	may	produce	varying	levels	of	risk	across	outcomes	(23).	For	instance,	an	appropriate	shift	work	exposure	definition	for	long-latency	diseases	such	as	cancer	(e.g.,	focused	on	history	of	exposure,	number	of	nights	worked,	etc.	(21))	may	not	be	appropriate	for	assessing	mental	health	effects	in	the	short-term,	where	other	shift	work	characteristics	(such	as	quick	returns	between	shifts,	or	short	notice	of	work	schedule)	may	constitute	equal	or	greater	problems	for	shift	workers	(220).		Future	research	that	collects	and	incorporates	detailed,	hypothesis-driven	characteristics	into	the	assignment	of	exposures	would	help	to	identify	the	most	accurate	and	precise	exposure	indices	to	assess	health	risks	within	studies.	The	consistent	assignment	of	specific	and	biologically	based	exposure	indicators	for	health	outcomes	would	also	permit	better	comparisons	across	studies	of	shift	work.	This	point	was	demonstrated	in	a	recent	study	that	used	administrative	data	on	individual	working	hours	to	examine	how	eight	different	exposure	definitions	impacted	calculations	of	the	number	of	night	shifts	worked	(297).	The	investigators	found	marked	differences	in	the	number	of	nights	worked	when	comparing	night	shift	defined	by	specified	periods	of	night	time	(e.g.,	working	at	least	3	hours	between	24:00	and	05:00)	versus	night	shift	defined	by	start	and	end	time	(e.g.,	beginning	after	19:00	and	ending	before	09:00).	Such	differences	would	affect	the	proportions	of	workers	categorized	as	exposed	or	unexposed	to	shift	work,	with	potential	impacts	on	both	the	estimation	of	risk	within	studies,	as	well	as	the	comparability	of	results	between	studies.	In	Chapter	4,	potential	determinants	of	workplace-level	shift	work	policies	and	practices	were	described.	Data	on	shift	work	scheduling,	provision	of	shift	work	education	materials/training	to	employees,	and	nighttime	lighting	policies	in	the	workplace	were	collected	during	phone	interviews	with	employers	across	the	Canadian	province	of	British	Columbia.	The	sample	included	88	participating	organizations,	representing	30,700	shift	workers	across	industries	common	to	most	western	countries,	such	as	manufacturing,	service,	retail,	and	healthcare.	Long	duration	(12+	hour)	shifts,	provision	of	shift	work	education	materials/training	to	employees,	and	nighttime	lighting	policies	were	each	 84 reported	by	approximately	one	third	of	participating	organizations.	The	use	of	long	duration	shifts	was	more	likely	in	larger	workplaces,	and	varied	by	industry.	Employer	provision	of	shift	work	education	materials/training	was	more	likely	in	larger	workplaces,	in	organizations	where	shift	work	scheduling	was	related	to	concern	for	shift	worker	health,	and	in	organizations	without	seasonal	changes	in	shift	work.	Workplace	nighttime	lighting	policies	were	more	likely	in	organizations	where	shift	work	scheduling	was	related	to	previous	workplace	accidents	or	incidents	occurring	during	non-daytime	hours,	site	maintenance	needs,	and	client	service	or	care	needs.		Shift	work	is	a	highly	variable	exposure	with	a	number	of	workplace,	social,	and	individual	factors	that	may	influence	its	effects	on	health,	and	there	is	no	blanket	solution	to	address	its	negative	consequences.	As	described	in	Chapter	4	however,	workplace-level	interventions	are	an	important	means	to	promote	population	health.	Reviews	on	the	topic	of	shift	work	(62)	and	organizational	and	stress	intervention	(273)	have	emphasized	the	need	for	more	intervention	research	conducted	at	the	workplace	level.	Despite	their	perceived	importance	to	research	and	public	health,	workplace	interventions	have	been	characterized	as	time	consuming,	expensive,	and	difficult	to	describe,	control,	and	evaluate	(273).	This	could	explain	why	few	intervention	studies	that	target	modifiable	risk	factors	for	health	outcomes	in	shift	workers,	such	as	workplace	policies	and	practices,	have	been	conducted	outside	of	experimental	or	simulated	settings	(62,156).	The	resulting	dearth	of	information	is	generally	problematic,	but	especially	so	for	long-term	health	promotion	in	shift	workers,	since	interventions	to	prevent	chronic	diseases	are	even	more	complex	to	conduct	(148).		Workplace-based	intervention	research	is	important	to	identify	what	works	for	whom,	and	in	what	circumstances,	by	studying	the	context	of	an	intervention	(e.g.,	in	what	conditions	are	an	intervention	effective?)	as	well	its	mechanisms		(what	makes	an	intervention	work?)	(272).	As	one	author	aptly	states:	“the	alert	researcher	should	always	look	for	windows	of	opportunity	to	perform	intervention	research”	(273).	A	better	understanding	of	the	links	between	high-level	workplace	characteristics	and	shift	work	policies	and	practices	is	therefore	useful	to	support	progress	in	this	area.	 85 Results	from	Chapter	4	point	to	determinants	of	workplace-level	shift	work	policies	and	practices	that	could	serve	as	targets	for	future	epidemiological	research	and	interventions.	These	include	industry	sector	and	workplace	size	(for	shift	duration),	workplace	size,	seasonal	changes	in	shift	work,	and	employer	concern	for	worker	health	(for	individual-level	health	promotion),	and	previous	workplace	accidents	or	incidents	occurring	during	non-daytime	hours,	site	maintenance	needs,	and	client	service	or	care	needs	(for	nighttime	lighting	policies).		It	is	surprising	that	little	other	information	is	available	to	describe	high-level	determinants	of	shift	work	practices	and	policies.	There	are	many	opportunities	for	future	studies	to	assess	the	validity	of	Chapter	4’s	findings	and	to	expand	this	knowledge	base,	such	as	through	similar	investigations	that	assess	larger	samples	or	different	jurisdictions.	Industry-based	studies	would	provide	a	means	to	probe	reasons	behind	the	relationships	observed	in	Chapter	4,	such	as	motivating	factors	for	the	use	of	certain	types	of	shift	work	schedules,	the	existence	of	light-at-night	policies,	and	the	provision	of	worker	education.	In	addition	to	informing	targets	for	future	research	and	prevention,	such	information	could	also	provide	a	clearer	picture	of	the	barriers	and	facilitators	to	conducting	shift	work-related	research	and	translating	the	evidence	obtained	into	best	practices	within	workplaces	(272).	Time	requirements	for	recruiting	and	interviewing	participants	in	future	studies	is	an	important	consideration;	for	example,	recruitment,	interview,	and	follow	up	activities	for	each	participant	(organization)	in	Chapter	4	required	approximately	6	hours.	5.2 Conflicting	imperatives	in	translating	shift	work	and	health	research	While	eliminating	shift	work	might	be	the	best	way	to	minimize	circadian	disruption	and	negative	health	impacts,	this	would	clearly	ignore	the	requirements	of	work	and	life	in	modern	society	(32).	Instead,	practical	questions	for	worker	health	relate	to	questions	such	as:	How	can	nighttime	lighting	be	applied	for	maximal	benefit?	Which	work	schedules	cause	minimal	impact?	How	can	information	about	health	maintenance	and	self-protection	be	meaningfully	communicated	to	workplaces	and	individuals?	High	quality	epidemiological	studies	are	needed	to	answer	these	questions,	and	to	inform	policies	and	workplace	interventions	that	will	effectively	support	the	health	of	shift	workers.	Such	 86 evidence	might	also	help	to	address	the	challenge	of	conflicting	imperatives	(23)	that	emerge	due	to	the	range	of	health	effects	arising	from	a	myriad	of	shift	work	exposures.	One	clear	example	of	conflicting	imperatives	arising	in	this	dissertation	relates	to	the	development	of	recommendations	for	“healthy”	nighttime	lighting	in	workplaces.	Exposure	to	light-at-night	is	a	strongly	hypothesized	mechanism	linking	shift	work	to	cancer	(3)	and	as	such,	measures	to	minimize	its	effects	on	natural	sleep-wake	rhythms	are	recommended	(161),	particularly	since	total	adaptation	to	night	work	is	unlikely	in	most	circumstances	(1).	Conversely,	exposure	to	bright	light	has	a	direct	alerting	effect	(172,296,298)	that	has	been	found	to	confer	acute	safety	benefits	by	increasing	alertness	and	enhancing	performance	during	work	at	night	(196,197,299,300).	At	first	glance,	this	conflict	suggests	an	impossible	situation	concerning	the	short-term	protection	of	worker	safety	versus	the	long-term	maintenance	of	health.	However,	recent	scientific	findings	have	indicated	that	longer	wavelength	“red”	light	might	provide	alerting	benefits	without	suppressing	melatonin	in	shift	workers	(161,301).	This	points	to	future	research	opportunities	into	“healthy”	nighttime	lighting	technologies	that	allow	individuals	to	see	and	remain	alert	while	concurrently	minimizing	melatonin	suppression	and	disruption	to	circadian	rhythms	and	sleep.		A	second	example	of	conflicting	imperatives	arising	in	this	dissertation	relates	to	developing	practical	advice	for	“optimal”	shift	duration,	due	to	conflicting	evidence	for	the	benefits	of	using	short	versus	long	duration	shifts	(62,71).	A	number	of	studies	have	demonstrated	positive	effects	of	12-hour	shifts	in	nurses	(302)	and	other	shift	workers	(61),	perhaps	due	to	related	benefits	such	as	longer	periods	of	time	off	from	work,	more	time	for	family	and	friends,	greater	satisfaction	with	working	hours	(303)	and	work-life	balance	(274),	and	improved	sleep	compared	to	an	8-hour	shift	schedule	(62).	Beneficial	effects	of	longer	shifts	(12+	hours	versus	8	hours	or	less)	on	depressive	outcomes	are	also	identified	in	Chapter	3	of	this	dissertation.	Other	research	indicates	that	8-hour	shifts	are	associated	with	less	fatigue	and	sleepiness	during	the	work	period	(303)	and	decreased	risk	of	workplace	incidents	and	errors	compared	to	12-hour	shifts	(275,304),	with	a	cumulative	increasing	risk	of	incidents	in	shifts	exceeding	8	hours	(305).	In	industrial	 87 environments,	shorter	shifts	can	be	advantageous	to	reduce	the	potential	for	prolonged	toxic	exposures	and	increase	time	off	between	shifts	to	clear	chemicals	with	longer	half-lives	(303).	Shorter	workdays	(of	6	hours)	have	also	been	associated	with	decreased	risk	for	musculoskeletal	pain	relative	to	work	shifts	of	7	hours	or	more	(306).		Inconsistent	findings	regarding	the	impact	of	long	duration	shifts	on	health	may	be	due	to	differences	in	the	occupations	or	job	tasks	studied	(e.g.,	workload,	monotonous	tasks),	workplace	factors	(e.g.,	rest	break	policies,	staff	resources,	distribution	of	work	shifts	and	rest	days),	or	other	more	general	factors	(e.g.,	commuting	time)	(4).	While	these	inconsistencies	challenge	the	specification	of	universally	applicable	recommendations	for	optimal	shift	duration,	some	of	the	conceptual	conflicts	could	be	resolved	with	more	methodologically	rigorous	intervention	studies	in	the	future	(61).	Furthermore,	future	evidence-based	recommendations	should	consider	the	reasons	for	long	duration	shifts	being	favoured	in	large	workplaces	and	in	certain	types	of	industries,	as	described	in	Chapter	4.				These	examples	demonstrate	some	of	the	challenges	facing	the	translation	of	research	findings	into	practice.	It	is	interesting	to	note	that	very	few	shift	worker	intervention	designs	have	assessed	a	combination	of	approaches	to	promote	health	(62),	when	this	could	be	a	more	effective	strategy	to	address	the	complexities	of	shift	work	exposure	and	its	various	health	effect	modifiers.	Future	health	studies	and	interventions,	particularly	those	that	examine	a	broad	spectrum	of	actions	(to	address	multiple	exposures	and	modifiers)	and	take	a	long-term	perspective	(to	demonstrate	long-lasting	effects)	(148),	require	strong	research	methods	in	order	to	produce	accurate	results	that	can	be	effectively	interpreted	and	applied.	Therefore,	it	is	certain	that	the	need	for	strong	exposure	assessment	in	intervention-focused	epidemiological	studies	will	persist.			5.3 Summary			This	PhD	dissertation	aimed	to	provide	new	information	on	the	measurement,	assignment,	and	determinants	of	shift	work	exposure	in	epidemiological	studies	of	shift	work,	to	address	current	evidence	gaps	and	inform	future	methods	in	this	area.	Chapter	2	was	 88 conducted	to	measure	personal	light-at-night	exposure	levels	in	emergency	services	and	healthcare	workers	in	British	Columbia,	as	an	initial	step	to	characterizing	occupational	exposure	levels,	variability,	and	metrics.	It	was	found	that	exposures	varied	across	occupations	and	work	environments	yet	were	generally	stable	over	time	in	this	population,	suggesting	that	high-level	grouping	schemes	may	be	appropriate	to	identify	and	describe	light-at-night	exposures	in	future	research.	The	exposure	metrics	assessed	were	moderately	to	highly	correlated;	therefore	multiple	metrics	may	be	useful	in	future	epidemiological	studies,	depending	upon	the	conceptual	nature	of	the	research	question.	Chapter	3	was	conducted	to	examine	how	exposure	assignment	precision	would	affect	observed	relationships	between	shift	work	schedule	and	depression	in	a	large	sample	of	Canadian	nurses.	The	strongest	associations	with	depression	were	observed	with	the	high-precision	exposure	indicator	group	that	measured	both	timing	of	work	and	frequency	of	shift	changes.	This	finding	supports	the	need	for	detailed	and	hypothesis-based	exposure	assignment	in	future	studies,	and	provides	new	epidemiological	evidence	of	shift	work’s	effects	on	mental	health.	Finally,	Chapter	4	was	conducted	to	describe	and	assess	determinants	of	workplace-level	shift	work	policies	and	practices	thought	to	affect	health	across	a	range	of	industry	sectors	in	British	Columbia,	Canada.	Results	indicate	that	industry	sector,	workplace	size,	seasonal	work,	past	workplace	incidents	occurring	during	non-daytime	hours,	the	nature	of	work	tasks,	and	concern	for	worker	safety	and	health	are	associated	with	shift	work	policies	and	practices;	such	factors	should	be	considered	in	future	epidemiological	research	and	interventions.			 89 Bibliography			1.		 Totterdell	P.	Work	Schedules.	In:	Barling	J,	Kelloway	EK,	Frone	MR,	editors.	Handbook	of	work	stress.	Thousand	Oaks,	California:	SAGE	Publications;	2005.	p.	710.		2.		 Williams	C.	Work-life	balance	of	shift	workers.	Statistics	Canada	Perspectives.	Ottawa,	Canada:	Statistics	Canada;	2008.		3.		 International	Agency	for	Research	on	Cancer.	IARC	Monographs	on	the	Evaluation	of	Carcinogenic	Risks	to	Humans	Volume	98:	Painting,	Firefighting	and	Shiftwork.	Lyon,	France;	2010.		4.		 Tucker	P,	Folkard	S.	Conditions	of	Work	and	Employment	Series	No.	31:	Working	Time,	Health	and	Safety:	a	Research	Synthesis	Paper.	Geneva,	Switzerland:	International	Labour	Office;	2012.		5.		 Rajaratnam	SM,	Arendt	J.	Health	in	a	24-h	society.	Lancet.	2001;358(9286):999–1005.		6.		 CAREX	Canada.	Shiftwork	[Internet].	2017	[cited	2017	Mar	20].	Available	from:	http://www.carexcanada.ca/en/shiftwork/occupational_estimate/	7.		 McMenamin	TM.	A	time	to	work:	recent	trends	in	shift	work	and	flexible	schedules.	Mon	Labor	Rev.	2007;130(12):3–15.		8.		 Boisard	P,	Cartron	D,	Gollac	M,	Valeyre	A.	Time	and	work:	Duration	of	work	[Internet].	Dublin:	European	Foundation	for	the	Improvement	of	Living	and	Working	Conditions;	2003.	Available	from:	http://edz.bib.uni-mannheim.de/www-edz/pdf/ef/02/ef0211en.pdf	9.		 Survey	of	Labour	and	Income	Dynamics,	1996.	Ottawa,	Canada:	Statistics	Canada;	2009.		10.		 Survey	of	Labour	and	Income	Dynamics,	2006.	Ottawa,	Canada:	Statistics	Canada;	2009.		11.		 Survey	of	Labour	and	Income	Dynamics,	2011.	Ottawa,	Canada:	Statistics	Canada;	2014.		12.		 Harrington	JM.	Health	effects	of	shift	work	and	extended	hours	of	work.	Occup	Environ	Med.	2001;58(1):68–72.		13.		 Åkerstedt	T,	Wright	KP.	Sleep	loss	and	fatigue	in	shift	work	and	shift	work	disorder.	Sleep	Med	Clin.	2009;4(2):257–71.		14.		 Lowden	A,	Moreno	C,	Holmbäck	U,	Lennernäs	M,	Tucker	P.	Eating	and	shift	work	-	effects	on	habits,	metabolism	and	performance.	Scand	J	Work	Environ	Health.	2010;36(2):150–62.		15.		 Frost	P,	Kolstad	HA,	Bonde	JP.	Shift	work	and	the	risk	of	ischemic	heart	disease	-	a	systematic	review	of	the	epidemiologic	evidence.	Scand	J	Work	Environ	Health.	 90 2009;35(3):163–79.		16.		 Bara	A-C,	Arber	S.	Working	shifts	and	mental	health	-	findings	from	the	British	Household	Panel	Survey	(1995-2005).	Scand	J	Work	Environ	Health.	2009;35(5):361–7.		17.		 McLaughlin	C,	Bowman	ML,	Bradley	CL,	Mistlberger	RE.	A	prospective	study	of	seasonal	variation	in	shift-work	tolerance.	Chronobiol	Int.	2008;25(2):455–70.		18.		 Culpepper	L.	The	social	and	economic	burden	of	shift-work	disorder.	J	Fam	Pract.	2010;59(1	Suppl):S3–11.		19.		 Costa	G.	Factors	influencing	health	of	workers	and	tolerance	to	shift	work.	Theor	Issues	Ergon	Sci.	2003;4(3–4):263–88.		20.		 Costa	G,	Haus	E,	Stevens	R.	Shift	work	and	cancer	-	considerations	on	rationale,	mechanisms,	and	epidemiology.	Scand	J	Work	Environ	Health.	2010;36(2):163–79.		21.		 Stevens	RG,	Hansen	J,	Costa	G,	Haus	E,	Kauppinen	T,	Aronson	KJ,	et	al.	Considerations	of	circadian	impact	for	defining	“shift	work”	in	cancer	studies:	IARC	Working	Group	Report.	Occup	Environ	Med.	2011;68(2):154–62.		22.		 Härmä	M,	Kecklund	G.	Shift	work	and	health	-	how	to	proceed?	Scand	J	Work	Environ	Health.	2010;36(2):81–4.		23.		 Knutsson	A.	Methodological	aspects	of	shift-work	research.	Chronobiol	Int.	2004;21(6):1037–47.		24.		 Saksvik	IB,	Bjorvatn	B,	Hetland	H,	Sandal	GM,	Pallesen	S.	Individual	differences	in	tolerance	to	shift	work	–	A	systematic	review.	Sleep	Med	Rev.	2011;15(4):221–35.		25.		 Kompier	MAJ.	New	systems	of	work	organization	and	workers’	health.	Scand	J	Work	Environ	Health.	2006;32(6):421–30.		26.		 De	Witte	H.	Job	insecurity:	Review	of	the	international	literature	on	definitions,	prevalence,	antecedents	and	consequences.	SA	J	Ind	Psychol.	2005;31(4):1–6.		27.		 Mechanic	D,	Tanner	J.	Vulnerable	people,	groups,	and	populations:	societal	view.	Health	Aff.	2007;26(5):1220–30.		28.		 Salminen	S.	Have	young	workers	more	injuries	than	older	ones?	An	international	literature	review.	J	Safety	Res.	2004;35(5):513–21.		29.		 Lynch	JW,	Smith	GD,	Kaplan	GA,	House	JS.	Income	inequality	and	mortality:	importance	to	health	of	individual	income,	psychosocial	environment,	or	material	conditions.	BMJ.	2000;320(7243):1200–4.		30.		 Cutler	D,	Lleras-Muney	A.	Education	and	Health:	Evaluating	Theories	and	Evidence	(NBER	Working	Paper	No.	12352).	Cambridge,	MA:	National	Bureau	of	Economic	Research;	2006.		31.		 Monteleone	P,	Martiadis	V,	Maj	M.	Circadian	rhythms	and	treatment	implications	in	depression.	Prog	Neuro	Psychopharmacol	Biol	Psychiatry.	2011;35(7):1569–74.		32.		 Stevens	RG,	Zhu	Y.	Electric	light,	particularly	at	night,	disrupts	human	circadian	 91 rhythmicity:	is	that	a	problem?	Philos	Trans	R	Soc	Lond	B	Biol	Sci.	2015;370(1667):20140120.		33.		 Roenneberg	T,	Kuehnle	T,	Juda	M,	Kantermann	T,	Allebrandt	K,	Gordijn	M,	et	al.	Epidemiology	of	the	human	circadian	clock.	Sleep	Med	Rev.	2007;11(6):429–38.		34.		 Mistlberger	RE,	Skene	DJ.	Social	influences	on	mammalian	circadian	rhythms:	animal	and	human	studies.	Biol	Rev	Camb	Philos	Soc.	2004;79(3):533–56.		35.		 Stephan	FK.	The	“other”	circadian	system:	food	as	a	zeitgeber.	J	Biol	Rhythms.	2002;17(4):284–92.		36.		 Roenneberg	T,	Kantermann	T,	Juda	M,	Vetter	C,	Allebrandt	K	V.	Light	and	the	human	circadian	clock.	Handb	Exp	Pharmacol.	2013;(217):311–31.		37.		 Provencio	I,	Jiang	G,	De	Grip	WJ,	Hayes	WP,	Rollag	MD.	Melanopsin:	An	opsin	in	melanophores,	brain,	and	eye.	Proc	Natl	Acad	Sci	U	S	A.	1998;95(1):340–5.		38.		 Berson	DM,	Dunn	FA,	Takao	M.	Phototransduction	by	retinal	ganglion	cells	that	set	the	circadian	clock.	Science.	2002;295(5557):1070–3.		39.		 Hattar	S,	Liao	H-W,	Takao	M,	Berson	DM,	Yau	K-W.	Melanopsin-containing	retinal	ganglion	cells:	architecture,	projections,	and	intrinsic	photosensitivity.	Science.	2002;295(5557):1065–70.		40.		 LeGates	TA,	Fernandez	DC,	Hattar	S.	Light	as	a	central	modulator	of	circadian	rhythms,	sleep	and	affect.	Nat	Rev	Neurosci.	2014;15(7):443–54.		41.		 Golombek	DA,	Rosenstein	RE.	Physiology	of	circadian	entrainment.	Physiol	Rev.	2010;90(3):1063–102.		42.		 Arendt	J.	Melatonin,	circadian	rhythms,	and	sleep.	N	Engl	J	Med.	2000;343(15):1114–6.		43.		 Cagampang	FR,	Bruce	KD.	The	role	of	the	circadian	clock	system	in	nutrition	and	metabolism.	Br	J	Nutr.	2012;108(3):381–92.		44.		 Bellet	MM,	Sassone-Corsi	P.	Mammalian	circadian	clock	and	metabolism	–	the	epigenetic	link.	J	Cell	Sci.	2010;123(22):3837–48.		45.		 Portaluppi	F,	Tiseo	R,	Smolensky	MH,	Hermida	RC,	Ayala	DE,	Fabbian	F.	Circadian	rhythms	and	cardiovascular	health.	Sleep	Med	Rev.	2012;16(2):151–66.		46.		 Arendt	J.	Melatonin	and	human	rhythms.	Chronobiol	Int.	2006;23(1–2):21–37.		47.		 Ota	T,	Fustin	J-M,	Yamada	H,	Doi	M,	Okamura	H.	Circadian	clock	signals	in	the	adrenal	cortex.	Mol	Cell	Endocrinol.	2012;349(1):30–7.		48.		 Arendt	J.	Melatonin	and	the	Mammalian	Pineal	Gland.	London,	UK:	Chapman	&	Hall;	1994.		49.		 Arendt	J,	Skene	DJ.	Melatonin	as	a	chronobiotic.	Sleep	Med	Rev.	2005;9(1):25–39.		50.		 Blask	DE.	Melatonin,	sleep	disturbance	and	cancer	risk.	Sleep	Med	Rev.	2009;13(4):257–64.		 92 51.		 Klerman	EB,	Gershengorn	HB,	Duffy	JF,	Kronauer	RE.	Comparisons	of	the	variability	of	three	markers	of	the	human	circadian	pacemaker.	J	Biol	Rhythms.	2002;17(2):181–93.		52.		 Stevens	RG,	Brainard	GC,	Blask	DE,	Lockley	SW,	Motta	ME.	Adverse	health	effects	of	nighttime	lighting:	comments	on	american	medical	association	policy	statement.	Am	J	Prev	Med.	2013;45(3):343–6.		53.		 Srinivasan	V,	Smits	M,	Spence	W,	Lowe	AD,	Kayumov	L,	Pandi-Perumal	SR,	et	al.	Melatonin	in	mood	disorders.	World	J	Biol	Psychiatry.	2006;7(3):138–51.		54.		 Foster	R,	Wulff	K.	The	rhythm	of	rest	and	excess.	Neuroscience.	2005;6(5):407–14.		55.		 Folkard	S.	Editorial:	Special	issue	on	night	and	shiftwork.	Ergonomics.	1993;36(1–3):1–2.		56.		 Bartley	M,	Ferrie	J,	Montgomery	SM.	Health	and	labour	market	disadvantage:	unemployment,	non-employment,	and	job	insecurity.	In:	Marmot	M,	Wilkinson	RG,	editors.	Social	Determinants	of	Health,	Second	Edition.	Oxford,	UK:	Oxford	University	Press;	2006.	p.	78–96.		57.		 Marmot	M,	Siegrist	J,	Theorell	T.	Health	and	the	psychosocial	environment	at	work.	In:	Marmot	M,	Wilkinson	R,	editors.	Social	Determinants	of	Health,	Second	Edition.	Oxford,	UK:	Oxford	University	Press;	2006.	p.	97–130.		58.		 Shields	M.	Shift	work	and	health.	Heal	Reports.	2002;13(4):11–33.		59.		 Knauth	P.	The	design	of	shift	systems.	Ergonomics.	1993;36(1–3):15–28.		60.		 Bambra	CL,	Whitehead	MM,	Sowden	AJ,	Akers	J,	Petticrew	MP.	Shifting	schedules:	the	health	effects	of	reorganizing	shift	work.	Am	J	Prev	Med.	2008;34(5):427–34.		61.		 Sallinen	M,	Kecklund	G.	Shift	work,	sleep,	and	sleepiness	-	differences	between	shift	schedules	and	systems.	Scand	J	Work	Environ	Health.	2010;36(2):121–33.		62.		 Neil-Sztramko	SE,	Pahwa	M,	Demers	PA,	Gotay	CC.	Health-related	interventions	among	night	shift	workers:	a	critical	review	of	the	literature.	Scand	J	Work	Environ	Health.	2014;40(6):543–56.		63.		 Joyce	K,	Pabayo	R,	Critchley	JA,	Bambra	C.	Flexible	working	conditions	and	their	effects	on	employee	health	and	wellbeing.	Cochrane	Database	Syst	Rev.	2010;17	Feb(2):CD008009.		64.		 Vahtera	J,	Laine	S,	Virtanen	M,	Oksanen	T,	Koskinen	A,	Pentti	J,	et	al.	Employee	control	over	working	times	and	risk	of	cause-specific	disability	pension:	the	Finnish	Public	Sector	Study.	Occup	Environ	Med.	2010;67(7):479–85.		65.		 Barton	J,	Folkard	S.	Advancing	versus	delaying	shift	systems.	Ergonomics.	1993;36(1–3):59–64.		66.		 Tucker	P,	Smith	L,	Macdonald	I,	Folkard	S.	Effects	of	direction	of	rotation	in	continuous	and	discontinuous	8	hour	shift	systems.	Occup	Environ	Med.	2000;57(10):678–84.		67.		 Rea	M.	The	lumen	seen	in	a	new	light:	Making	distinctions	between	light,	lighting	and	 93 neuroscience.	Light	Res	Technol.	2015;47(3):259–80.		68.		 DiLaura,	David;	Houser,	Kevin;	Mistrick,	Richard;	Steff	G,	editor.	The	Lighting	Handbook	10th	Edition.	New	York,	USA:	The	Illuminating	Engineering	Society;	2011.		69.		 WorkSafeBC.	Regulation	Part	4	General	Conditions	-	Illumination	[Internet].	2017	[cited	2017	Mar	24].	Available	from:	https://www.worksafebc.com/en/law-policy/occupational-health-safety/searchable-ohs-regulation/ohs-regulation/part-04-general-conditions	70.		 Rea	MS,	Figueiro	MG.	What	is	“Healthy	Lighting?”	Int	J	High	Speed	Electron	Syst.	2011;20(2):321–42.		71.		 Knauth	P,	Hornberger	S.	Preventive	and	compensatory	measures	for	shift	workers.	Occup	Med	(Chic	Ill).	2003;53(2):109–16.		72.		 Pallesen	S,	Bjorvatn	B,	Magerøy	N,	Saksvik	IB,	Waage	S,	Moen	BE.	Measures	to	counteract	the	negative	effects	of	night	work.	Scand	J	Work	Environ	Health.	2010;36(2):109–20.		73.		 Rogers	A,	Holmes	S,	Spencer	M.	The	effect	of	shiftwork	on	driving	to	and	from	work.	J	Hum	Ergol	(Tokyo).	2001	Dec;30(1–2):131–6.		74.		 Barger	LK,	Cade	BE,	Ayas	NT,	Cronin	JW,	Rosner	B,	Speizer	FE,	et	al.	Extended	Work	Shifts	and	the	Risk	of	Motor	Vehicle	Crashes	among	Interns.	N	Engl	J	Med.	2005	Jan	13;352(2):125–34.		75.		 Siegrist	J,	Marmot	M.	Health	inequalities	and	the	psychosocial	environment—two	scientific	challenges.	Soc	Sci	Med.	2004	Apr	1;58(8):1463–73.		76.		 Bøggild	H,	Burr	H,	Tüchsen	F,	Jeppesen	HJ.	Work	environment	of	Danish	shift	and	day	workers.	Scand	J	Work	Environ	Health.	2001;27(2):97–105.		77.		 Puttonen	S,	Härmä	M,	Hublin	C.	Shift	work	and	cardiovascular	disease	-	pathways	from	circadian	stress	to	morbidity.	Scand	J	Work	Environ	Health.	2010;36(2):96–108.		78.		 Berthelsen	M,	Pallesen	S,	Magerøy	N,	Tyssen	R,	Bjorvatn	B,	Moen	BE,	et	al.	Effects	of	psychological	and	social	factors	in	shiftwork	on	symptoms	of	anxiety	and	depression	in	nurses:	A	1-year	follow-up.	J	Occup	Environ	Med.	2015;57(10):1127–37.		79.		 Netterstrom	B,	Conrad	N,	Bech	P,	Fink	P,	Olsen	O,	Rugulies	R,	et	al.	The	relation	between	work-related	psychosocial	factors	and	the	development	of	depression.	Epidemiol	Rev.	2008;30(1):118–32.		80.		 Costa	G.	Shift	work	and	health:	current	problems	and	preventive	actions.	Saf	Health	Work.	2010;1(2):112–23.		81.		 Jay	SM,	Gander	PH,	Eng	A,	Cheng	S,	Douwes	J,	Ellison-Loschmann	L,	et	al.	New	Zealanders	working	non-standard	hours	also	have	greater	exposure	to	other	workplace	hazards.	Chronobiol	Int.	2017;34(4):519–26.		82.		 Nachreiner	F.	Individual	and	social	determinants	of	shiftwork	tolerance.	Scand	J	Work	Environ	Health.	1998;24	Suppl	3:35–42.		 94 83.		 Patten	SB,	Wang	JL,	Williams	JV,	Currie	S,	Beck	CA,	Maxwell	CJ,	et	al.	Descriptive	epidemiology	of	major	depression	in	Canada.	Can	J	Psychiatry.	2006;51(2):84–90.		84.		 Patten	SB,	Williams	JVA,	Lavorato	DH,	Wang	JL,	McDonald	K,	Bulloch	AGM.	Descriptive	epidemiology	of	major	depressive	disorder	in	Canada	in	2012.	Can	J	Psychiatry.	2015;60(1):23–30.		85.		 Robards	R,	Evandrou	M,	Falkingham	J,	Vlachantoni	A.	Marital	status,	health	and	mortality.	Maturitas.	2012	Dec	1;73(4):295–9.		86.		 Härmä	M.	Individual	differences	in	tolerance	to	shiftwork:	a	review.	Ergonomics.	1993;36(1–3):101–9.		87.		 Czeisler	CA,	Duffy	JF,	Shanahan	TL,	Brown	EN,	Mitchell	JF,	Rimmer	DW,	et	al.	Stability,	precision,	and	near-24-hour	period	of	the	human	circadian	pacemaker.	Science.	1999;284(5423):2177–81.		88.		 Erren	TC,	Groß	J	V,	Fritschi	L.	Focusing	on	the	biological	night:	towards	an	epidemiological	measure	of	circadian	disruption.	Occup	Environ	Med.	2017;74(3):159–60.		89.		 Erren	TC,	Morfeld	P.	Computing	chronodisruption:	how	to	avoid	potential	chronobiological	errors	in	epidemiological	studies	of	shift	work	and	cancer.	Chronobiol	Int.	2014;31(4).		90.		 Mukherjee	S,	Patel	SR,	Kales	SN,	Ayas	NT,	Strohl	KP,	Gozal	D,	et	al.	An	Official	American	Thoracic	Society	Statement:	The	Importance	of	Healthy	Sleep.	Recommendations	and	Future	Priorities.	Am	J	Respir	Crit	Care	Med.	2015;191(12):1450–8.		91.		 Costa	G,	Di	Milia	L.	Aging	and	Shift	Work:	A	Complex	Problem	to	Face.	Chronobiol	Int.	2008;25(2–3):165–81.		92.		 Sayer	LC,	England	P,	Bittman	M,	Bianchi	SM.	How	Long	Is	the	Second	(Plus	First)	Shift?	Gender	Differences	in	Paid,	Unpaid,	and	Total	Work	Time	in	Australia	and	the	United	States.	J	Comp	Fam	Stud.	2009;40(4):523–45.		93.		 Nicholson	PJ,	D’Auria	DAP.	Shift	work,	health,	the	working	time	regulations	and	health	assessments.	Occup	Med	(Chic	Ill).	1999;49(3):127–37.		94.		 Ramin	C,	Devore	EE,	Wang	W,	Pierre-Paul	J,	Wegrzyn	LR,	Schernhammer	ES.	Night	shift	work	at	specific	age	ranges	and	chronic	disease	risk	factors.	Occup	Environ	Med.	2014;72(2):100–7.		95.		 Wang	X-S,	Travis	RC,	Reeves	G,	Green	J,	Allen	NE,	Key	TJ,	et	al.	Characteristics	of	the	Million	Women	Study	participants	who	have	and	have	not	worked	at	night.	Scand	J	Work	Environ	Health.	2012;38(6):590–9.		96.		 Vogel	M,	Braungardt	T,	Meyer	W,	Schneider	W.	The	effects	of	shift	work	on	physical	and	mental	health.	J	Neural	Transm.	2012;119(10):1121–32.		97.		 Mustard	CA,	Chambers	A,	McLeod	C,	Bielecky	A,	Smith	PM.	Work	injury	risk	by	time	of	day	in	two	population-based	data	sources.	Occup	Environ	Med.	2013;70(1):49–56.		 95 98.		 Jia	Y,	Lu	Y,	Wu	K,	Lin	Q,	Shen	W,	Zhu	M,	et	al.	Does	night	work	increase	the	risk	of	breast	cancer?	A	systematic	review	and	meta-analysis	of	epidemiological	studies.	Cancer	Epidemiol.	2013;37(3):197–206.		99.		 Parent	M-E,	El-Zein	M,	Rousseau	M-C,	Pintos	J,	Siemiatycki	J.	Night	Work	and	the	Risk	of	Cancer	Among	Men.	Am	J	Epidemiol.	2012;176(9):751–9.		100.		 Tucker	P,	Marquié	J-C,	Folkard	S,	Ansiau	D,	Esquirol	Y.	Shiftwork	and	metabolic	dysfunction.	Chronobiol	Int.	2012;29(5):549–55.		101.		 Driesen	K,	Jansen	N,	van	Amelsvoort	L,	Kant	I.	The	mutual	relationship	between	shift	work	and	depressive	complaints	-	a	prospective	cohort	study.	Scand	J	Work	Environ	Health.	2011;37(5):402–10.		102.		 Åkerstedt	T,	Ingre	M,	Broman	J,	Kecklund	G.	Disturbed	sleep	in	shift	workers,	day	workers,	and	insomniacs.	Chronobiol	Int.	2009;29(2):333–48.		103.		 Fritschi	L,	Glass	DC,	Heyworth	JS,	Aronson	K,	Girschik	J,	Boyle	T,	et	al.	Hypotheses	for	mechanisms	linking	shiftwork	and	cancer.	Med	Hypotheses.	2011;77(3):430–6.		104.		 Haines	III	VY,	Marchand	A,	Rousseau	V,	Demers	A.	The	mediating	role	of	work-to-family	conflict	in	the	relationship	between	shiftwork	and	depression.	Work	Stress.	2008;22(4):341–56.		105.		 Spiegel	K,	Leproult	R,	Van	Cauter	E.	Impact	of	sleep	debt	on	metabolic	and	endocrine	function.	Lancet.	1999;354(9188):1435–9.		106.		 Lewy	A,	Wehr	T,	Goodwin	F,	Newsome	D,	Markey	S.	Light	suppresses	melatonin	secretion	in	humans.	Science.	1980;210(4475):1267–9.		107.		 Mirick	DK,	Davis	S.	Melatonin	as	a	biomarker	of	circadian	dysregulation.	Cancer	Epidemiol	Biomarkers	Prev.	2008;17(12):3306–13.		108.		 Macchi	MM,	Bruce	JN.	Human	pineal	physiology	and	functional	significance	of	melatonin.	Front	Neuroendocrinol.	2004;25(3):177–95.		109.		 Skocbat	T,	Haimov	I,	Lavie	P.	Melatonin	-	the	key	to	the	gate	of	sleep.	Ann	Med.	1998;30(1):109–14.		110.		 Lockley	SW,	Skene	DJ,	Tabandeh	H,	Bird	AC,	Defrance	R,	Arendt	J.	Relationship	between	napping	and	melatonin	in	the	blind.	J	Biol	Rhythms.	1997;12(1):16–25.		111.		 Brzezinski	A,	Vangel	MG,	Wurtman	RJ,	Norrie	G,	Zhdanova	I,	Ben-Shushan	A,	et	al.	Effects	of	exogenous	melatonin	on	sleep:	a	meta-analysis.	Sleep	Med	Rev.	2005;9(1):41–50.		112.		 Linton	SJ,	Kecklund	G,	Franklin	KA,	Leissner	LC,	Sivertsen	B,	Lindberg	E,	et	al.	The	effect	of	the	work	environment	on	future	sleep	disturbances:	a	systematic	review.	Sleep	Med	Rev.	2015;23:10–9.		113.		 Pilcher	JJ,	Lambert	BJ,	Huffcutt	AI.	Differential	effects	of	permanent	and	rotating	shifts	on	self-report	sleep	length:	a	meta-analytic	review.	Sleep.	2000;23(2):155–63.		114.		 Øyane	NMF,	Pallesen	S,	Moen	BE,	Akerstedt	T,	Bjorvatn	B.	Associations	between	night	work	and	anxiety,	depression,	insomnia,	sleepiness	and	fatigue	in	a	sample	of	 96 Norwegian	nurses.	PLoS	One.	2013;8(8):e70228.		115.		 Luyster	FS,	Strollo	PJ,	Zee	PC,	Walsh	JK,	Boards	of	Directors	of	the	American	Academy	of	Sleep	Medicine	and	the	Sleep	Research	Society.	Sleep:	a	health	imperative.	Sleep.	2012;35(6):727–34.		116.		 Williamson	A,	Lombardi	DA,	Folkard	S,	Stutts	J,	Courtney	TK,	Connor	JL.	The	link	between	fatigue	and	safety.	Accid	Anal	Prev.	2011;43(2):498–515.		117.		 Bryant	PA,	Trinder	J,	Curtis	N.	Sick	and	tired:	does	sleep	have	a	vital	role	in	the	immune	system?	Nat	Rev	Immunol.	2004;4(6):457–67.		118.		 Cappuccio	FP,	D’Elia	L,	Strazzullo	P,	Miller	MA.	Quantity	and	quality	of	sleep	and	incidence	of	Type	2	Diabetes.	Diabetes	Care.	2010;33(2):414–20.		119.		 Gangwisch	JE,	Malaspina	D,	Boden-Albala	B,	Heymsfield	SB.	Inadequate	sleep	as	a	risk	factor	for	obesity:	analyses	of	the	NHANES	I.	Sleep.	2005;28(10):1289–96.		120.		 Rod	NH,	Kumari	M,	Lange	T,	Kivimäki	M,	Shipley	M,	Ferrie	J.	The	joint	effect	of	sleep	duration	and	disturbed	sleep	on	cause-specific	mortality:	results	from	the	Whitehall	II	cohort	study.	PLoS	One.	2014;9(4):e91965.		121.		 Ohayon	MM.	Epidemiology	of	insomnia:	what	we	know	and	what	we	still	need	to	learn.	Sleep	Med	Rev.	2002;6(2):97–111.		122.		 Barnes-Farrell	JL,	Davies-Schrils	K,	McGonagle	A,	Walsh	B,	Milia	L	Di,	Fischer	FM,	et	al.	What	aspects	of	shiftwork	influence	off-shift	well-being	of	healthcare	workers?	Appl	Ergon.	2008;39(5):589–96.		123.		 Germain	A,	Kupfer	DJ.	Circadian	rhythm	disturbances	in	depression.	Hum	Psychopharmacol	Clin	Exp.	2008;23(7):571–85.		124.		 Vetter	C,	Fischer	D,	Matera	JL,	Roenneberg	T,	Wright	KP,	Bogan	RK,	et	al.	Aligning	work	and	circadian	time	in	shift	workers	improves	sleep	and	reduces	circadian	disruption.	Curr	Biol.	2015;25(7):907–11.		125.		 Leineweber	C,	Baltzer	M,	Magnusson	Hanson	LL,	Westerlund	H.	Work-family	conflict	and	health	in	Swedish	working	women	and	men:	a	2-year	prospective	analysis	(the	SLOSH	study).	Eur	J	Public	Health.	2013;23(4):710–6.		126.		 Strandh	M,	Nordenmark	M.	The	interference	of	paid	work	with	household	demands	in	different	social	policy	contexts:	perceived	work–household	conflict	in	Sweden,	the	UK,	the	Netherlands,	Hungary,	and	the	Czech	Republic.	Br	J	Sociol.	2006;57(4):597–617.		127.		 Allen	TD,	Herst	DEL,	Bruck	CS,	Sutton	M.	Consequences	associated	with	work-to-family	conflict:	A	review	and	agenda	for	future	research.	J	Occup	Health	Psychol.	2000;5(2):278–308.		128.		 Taylor	E,	Briner	RB,	Folkard	S.	Models	of	shiftwork	and	health:	An	examination	of	the	influence	of	stress	on	shiftwork	theory.	Hum	Factors	J	Hum	Factors	Ergon	Soc.	1997;39(1):67–82.		129.		 Smedley	B,	Syme	S.	Promoting	health:	intervention	strategies	from	social	and	 97 behavioral	research.	Am	J	Health	Promot.	2001;15(3):149–66.		130.		 Berglund	M,	Elinder	C-G,	Jarup	L.	Human	Exposure	Assessment:	An	Introduction.	Geneva,	Switzerland:	World	Health	Organization;	2001.		131.		 Stewart	P,	Stenzel	M.	Exposure	Assessment	in	the	Occupational	Setting.	Appl	Occup	Environ	Hyg.	2000;15(5):435–44.		132.		 Last	J.	A	dictionary	of	epidemiology.	Oxford,	UK:	Oxford	University	Press;	2001.	141	p.		133.		 Nieuwenhuijsen	M,	editor.	Exposure	assessment	in	occupational	and	environmental	epidemiology.	New	York,	USA:	Oxford	University	Press;	2003.		134.		 Loomis	D,	Kromhout	H.	Exposure	variability:	concepts	and	applications	in	occupational	epidemiology.	Am	J	Ind	Med.	2004;45(1):113–22.		135.		 Rothman	K.	Biases	in	Study	Design.	In:	Epidemiology:	An	Introduction.	New	York,	USA:	Oxford	University	Press;	2002.	p.	94–112.		136.		 Travis	RC,	Balkwill	A,	Fensom	GK,	Appleby	PN,	Reeves	GK,	Wang	X-S,	et	al.	Night	shift	work	and	breast	cancer	incidence:	Three	prospective	studies	and	meta-analysis	of	published	studies.	J	Natl	Cancer	Inst.	2016;108(12):djw169.		137.		 Schernhammer	ES.	RE:	Night	Shift	Work	and	Breast	Cancer	Incidence:	Three	Prospective	Studies	and	Meta-analysis	of	Published	Studies.	J	Natl	Cancer	Inst.	2017;109(4):135–40.		138.		 Stevens	RG.	RE:	Night	Shift	Work	and	Breast	Cancer	Incidence:	Three	Prospective	Studies	and	Meta-analysis	of	Published	Studies.	J	Natl	Cancer	Inst.	2017;109(4):853–60.		139.		 Erren	TC,	Morfeld	P,	Groß	JV.	RE:	Night	Shift	Work	and	Breast	Cancer	Incidence:	Three	Prospective	Studies	and	Meta-analysis	of	Published	Studies.	J	Natl	Cancer	Inst.	2017;109(4):282–6.		140.		 Hansen	J.	RE:	Night	Shift	Work	and	Breast	Cancer	Incidence:	Three	Prospective	Studies	and	Meta-analysis	of	Published	Studies.	J	Natl	Cancer	Inst.	2017;109(4):551–6.		141.		 Vistisen	HT,	Garde	AH,	Frydenberg	M,	Christiansen	P,	Hansen	ÅM,	Andersen	J,	et	al.	Short-term	effects	of	night	shift	work	on	breast	cancer	risk:	a	cohort	study	of	payroll	data.	Scand	J	Work	Environ	Health.	2017	Jan	1;43(1):59–67.		142.		 Stevens	RG.	Letter	in	reference	to:	“Short-term	effects	of	night	shift	work	on	breast	cancer	risk:	a	cohort	study	of	payroll	data.”	Scand	J	Work	Environ	Health.	2017;43(1):95.		143.		 Chan	N,	Choy	C.	No	link	between	breast	cancer	and	working	night	shifts.	Nurs	Stand.	2016;31(9):16–16.		144.		 Gulland	A.	Sixty	seconds	on	.	.	.	night	shifts.	BMJ.	2016;355:i5476.		145.		 Hansen	J,	Lassen	CF.	Nested	case-control	study	of	night	shift	work	and	breast	cancer	risk	among	women	in	the	Danish	military.	Occup	Environ	Med.	2012;69(8):551–6.		 98 146.		 Knutsson	A.	Mortality	of	shift	workers.	Scand	J	Work	Environ	Heal.	2017;43(2):97–8.		147.		 Council	on	Science	and	Public	Health	Report	4.	Light	Pollution:	Adverse	Health	Effects	of	Nighttime	Lighting.	Chicago,	IL:	American	Medical	Association	House	of	Delegates	Annual	Meeting;	2012.		148.		 Lowden	A,	Moreno	C.	Workplace	interventions:	a	challenge	for	promoting	long-term	health	among	shift	workers.	Scand	J	Work	Environ	Health.	2014;40(6):539–41.		149.		 Nathan	PJ,	Wyndham	EL,	Burrows	GD,	Norman	TR.	The	effect	of	gender	on	the	melatonin	suppression	by	light:	a	dose	response	relationship.	J	Neural	Transm.	2000;107(3):271–9.		150.		 Brainard	GC,	Richardson	BA,	Hurlbut	EC,	Steinlechner	S,	Matthews	SA,	Reiter	RJ.	The	influence	of	various	irradiances	of	artificial	light,	twilight,	and	moonlight	on	the	suppression	of	pineal	melatonin	content	in	the	Syrian	hamster.	J	Pineal	Res.	1984;1(2):105–19.		151.		 Bedrosian	TA,	Fonken	LK,	Walton	JC,	Nelson	RJ.	Chronic	exposure	to	dim	light	at	night	suppresses	immune	responses	in	Siberian	hamsters.	Biol	Lett.	2011;7(3):468–71.		152.		 Figueiro	MG,	Plitnick	B,	Rea	MS.	Pulsing	blue	light	through	closed	eyelids:	effects	on	acute	melatonin	suppression	and	phase	shifting	of	dim	light	melatonin	onset.	Nat	Sci	Sleep.	2014;(6):149–56.		153.		 Smolensky	MH,	Sackett-Lundeen	LL,	Portaluppi	F.	Nocturnal	light	pollution	and	underexposure	to	daytime	sunlight:	Complementary	mechanisms	of	circadian	disruption	and	related	diseases.	Chronobiol	Int.	2015;32(8):1029–48.		154.		 Straif	K,	Baan	R,	Grosse	Y,	Secretan	B,	Ghissassi	F	El,	Bouvard	V,	et	al.	Carcinogenicity	of	shift-work,	painting,	and	fire-fighting.	Lancet	Oncol.	2007;8(12):1065–6.		155.		 Figueiro	MG,	White	RD.	Health	consequences	of	shift	work	and	implications	for	structural	design.	J	Perinatol.	2013;33	Suppl	1(S1):S17-23.		156.		 Kantermann	T,	Juda	M,	Vetter	C,	Roenneberg	T.	Shift-work	research:	Where	do	we	stand,	where	should	we	go?	Sleep	Biol	Rhythms.	2010;8(2):95–105.		157.		 Kromhout	H.	Design	of	measurement	strategies	for	workplace	exposures.	Occup	Environ	Med.	2002;59(5):349–54.		158.		 Stevens	R.	Light-at-night,	circadian	disruption	and	breast	cancer:	assessment	of	existing	evidence.	Int	J	Epidemiol.	2009;38(4):963–70.		159.		 Nabe-Nielsen	K,	Jørgensen	MB,	Garde	AH,	Clausen	T.	Do	working	environment	interventions	reach	shift	workers?	Int	Arch	Occup	Environ	Health.	2016;89(1):163–70.		160.		 Boivin	DB,	James	FO.	Light	treatment	and	circadian	adaptation	to	shift	work.	Ind	Health.	2005;43(1):34–48.		161.		 Figueiro	MG,	Sahin	L,	Wood	B,	Plitnick	B.	Light	at	night	and	measures	of	alertness	and	performance:	implications	for	shift	workers.	Biol	Res	Nurs.	2015;18(1):90–100.		 99 162.		 Kromhout	H,	Symanski	E,	Rappaport	SM.	A	comprehensive	evaluation	of	within-	and	between-worker	components	of	occupational	exposure	to	chemical	agents.	Ann	Occup	Hyg.	1993;37(3):253–70.		163.		 Nieuwenhuijsen	MJ.	Exposure	assessment	in	occupational	epidemiology:	measuring	present	exposures	with	an	example	of	a	study	of	occupational	asthma.	Int	Arch	Occup	Environ	Health.	1997;70(5):295–308.		164.		 Cho	Y,	Ryu	S-H,	Lee	BR,	Kim	KH,	Lee	E,	Choi	J.	Effects	of	artificial	light	at	night	on	human	health:	A	literature	review	of	observational	and	experimental	studies	applied	to	exposure	assessment.	Chronobiol	Int.	2015;32(9):1294–310.		165.		 Folkard	S.	Do	permanent	night	workers	show	circadian	adjustment?	A	review	based	on	the	endogenous	melatonin	rhythm.	Chronobiol	Int.	2008;25(2–3):215–24.		166.		 Loomis	D,	Salvan	A,	Kromhout	H,	Kriebel	D.	Selecting	indices	of	occupational	exposure	for	epidemiologic	studies.	Occup	Hyg.	1999;5(1):73–91.		167.		 Tielemans	E,	Kupper	LL,	Kromhout	H,	Heederick	D,	Houba	R.	Individual-based	and	group-based	occupational	exposure	assessment:	Some	equations	to	evaluate	different	strategies.	Ann	Occup	Hyg.	1998;42(2):115–9.		168.		 Trask	CM,	Teschke	K,	Morrison	J,	Johnson	P,	Koehoorn	M.	Optimising	sampling	strategies:	components	of	low-back	EMG	variability	in	five	heavy	industries.	Occup	Environ	Med.	2010;67(12):853–60.		169.		 Preller	L,	Burstyn	I,	Pater	N	DE,	Kromhout	H.	Characteristics	of	Peaks	of	Inhalation	Exposure	to	Organic	Solvents.	Ann	Occup	Hyg.	2004;48(7):643–52.		170.		 European	Commission	Scientific	Committee	on	Emerging	and	Newly	Identified	Health	Risks.	Health	Effects	of	Artificial	Light.	Brussels,	Belgium;	2012.		171.		 Wulfinghoff	D.	Reference	Note	50	Measuring	Light	Intensity.	In:	Energy	Efficiency	Manual:	For	Everyone	who	Uses	Energy,	Pays	for	Utilities,	Controls	Energy	Usage,	Designs	and	Builds,	is	Interested	in	Energy	and	Environmental	Preservation.	Wheaton,	Maryland,	USA:	Energy	Institute	Press;	1999.		172.		 Cajochen	C.	Alerting	effects	of	light.	Sleep	Med	Rev.	2007;11(6):453–64.		173.		 Cajochen	C,	Zeitzer	JM,	Czeisler	CA,	Dijk	D-J.	Dose-response	relationship	for	light	intensity	and	ocular	and	electroencephalographic	correlates	of	human	alertness.	Behav	Brain	Res.	2000;115(1):75–83.		174.		 Rea	M,	Figueiro	M,	Bierman	A,	Hamner	R.	Modelling	the	spectral	sensitivity	of	the	human	circadian	system.	Light	Res	Technol.	2011;44(4):386–96.		175.		 Armstrong	BG.	Effect	of	measurement	error	on	epidemiological	studies	of	environmental	and	occupational	exposures.	Occup	Environ	Med.	1998;55(10):651–6.		176.		 Bierman	A,	Klein	TR,	Rea	MS.	The	Daysimeter:	a	device	for	measuring	optical	radiation	as	a	stimulus	for	the	human	circadian	system.	Meas	Sci	Technol.	2005;16(11):2292–9.		177.		 Figueiro	MG,	Hamner	R,	Bierman	A,	Rea	MS.	Comparisons	of	three	practical	field	 100 devices	used	to	measure	personal	light	exposures	and	activity	levels.	Light	Res	Technol.	2012;45(4):421–34.		178.		 Miller	D,	Bierman	A,	Figueiro	M,	Schernhammer	E,	Rea	M.	Ecological	measurements	of	light	exposure,	activity,	and	circadian	disruption.	Light	Res	Technol.	2010;42(3):271–84.		179.		 Rea	MS,	Brons	JA,	Figueiro	MG.	Measurements	of	light	at	night	(LAN)	for	a	sample	of	female	school	teachers.	Chronobiol	Int.	2011;28(8):673–80.		180.		 Young	CR,	Jones	GE,	Figueiro	MG,	Soutière	SE,	Keller	MW,	Richardson	AM,	et	al.	At-sea	trial	of	24-h-based	submarine	watchstanding	schedules	with	high	and	low	correlated	color	temperature	light	sources.	J	Biol	Rhythms.	2015;30(2):144–54.		181.		 Rea	MS,	Figueiro	MG,	Bierman	A,	Bullough	JD.	Circadian	light.	J	Circadian	Rhythms.	2010;8(1):2–11.		182.		 SAS	Institute	Inc.	SAS	9.4	Product	Documentation.	2013.		183.		 Butler	DL,	Biner	PM.	Preferred	lighting	levels:	variability	among	settings,	behaviors,	and	individuals.	Environ	Behav.	1987;19(6):695–721.		184.		 Papantoniou	K,	Pozo	OJ,	Espinosa	A,	Marcos	J,	Castaño-Vinyals	G,	Basagaña	X,	et	al.	Circadian	variation	of	melatonin,	light	exposure,	and	diurnal	preference	in	day	and	night	shift	workers	of	both	sexes.	Cancer	Epidemiol	Biomarkers	Prev.	2014;23(7):1176–86.		185.		 Grundy	A,	Tranmer	J,	Richardson	H,	Graham	CH,	Aronson	KJ.	The	influence	of	light	at	night	exposure	on	melatonin	levels	among	Canadian	rotating	shift	nurses.	Cancer	Epidemiol	Biomarkers	Prev.	2011;20(11):2404–12.		186.		 Esmen	NA,	Hammad	YY.	Log-normality	of	environmental	sampling	data.	J	Environ	Sci	Heal		Part	A	Environ	Sci	Eng.	1977;12(1–2):29–41.		187.		 Grundy	A,	Sanchez	M,	Richardson	H,	Tranmer	J,	Borugian	M,	Graham	CH,	et	al.	Light	intensity	exposure,	sleep	duration,	physical	activity,	and	biomarkers	of	melatonin	among	rotating	shift	nurses.	Chronobiol	Int.	2009;26(7):1443–61.		188.		 Seixas	NS,	Sheppard	L.	Maximizing	accuracy	and	precision	using	individual	and	grouped	exposure	assessments.	Scand	J	Work	Environ	Health.	1996;22(2):94–101.		189.		 Kromhout	H,	Heederik	D.	Occupational	epidemiology	in	the	rubber	industry:	Implications	of	exposure	variability.	Am	J	Ind	Med.	1995;27(2):171–85.		190.		 Kromhout	H,	Loomis	DP,	Mihlan	GJ,	Peipins	LA,	Kleckner	RC,	Iriye	R,	et	al.	Assessment	and	grouping	of	occupational	magnetic	field	exposure	in	five	electric	utility	companies.	Scand	J	Work	Environ	Health.	1995;21(1):43–50.		191.		 Carere	A,	Iacovella	N,	Turrio	Baldassarri	L,	Fuselli	S,	Iavarone	I,	Lagorio	S,	et	al.	Variability	of	benzene	exposure	among	filling	station	attendants	[Internet].	Rome,	Italy:	Istituto	Superiore	Di	Sanita;	1996	[cited	2017	Mar	21].	Available	from:	https://inis.iaea.org/search/search.aspx?orig_q=RN:28074192	192.		 Symanski	E,	Sällsten	G,	Barregård	L.	Variability	in	airborne	and	biological	measures	 101 of	exposure	to	mercury	in	the	chloralkali	industry:	implications	for	epidemiologic	studies.	Environ	Health	Perspect.	2000;108(6):569–73.		193.		 Heederik	D,	Attfield	M.	Characterization	of	dust	exposure	for	the	study	of	chronic	occupational	lung	disease:	a	comparison	of	different	exposure	assessment	strategies.	Am	J	Epidemiol.	2000	May	15;151(10):982–90.		194.		 Trask	CM,	Teschke	K,	Morrison	J,	Johnson	PW,	Village	J,	Koehoorn	M.	How	long	is	long	enough?	Evaluating	sampling	durations	for	low	back	EMG	assessment.	J	Occup	Environ	Hyg.	2008;5(10):664–70.		195.		 Symanski	E,	Maberti	S,	Chan	W.	A	meta-analytic	approach	for	characterizing	the	within-worker	and	between-worker	sources	of	variation	in	occupational	exposure.	Ann	Occup	Hyg.	2006;50(4):343–57.		196.		 Costa	G,	Ghirlanda	G,	Minors	DS,	Waterhouse	JM.	Effect	of	bright	light	on	tolerance	to	night	work.	Scand	J	Work	Environ	Health.	1993;19(6):414–20.		197.		 Daurat	A,	Foret	J,	Benoit	O,	Mauco	G.	Bright	light	during	nighttime:	effects	on	the	circadian	regulation	of	alertness	and	performance.	Biol	Signals	Recept.	2000;9(6):309–18.		198.		 Jewett	ME,	Rimmer	DW,	Duffy	JF,	Klerman	EB,	Kronauer	RE,	Czeisler	CA.	Human	circadian	pacemaker	is	sensitive	to	light	throughout	subjective	day	without	evidence	of	transients.	Am	J	Physiol	-	Regul	Integr	Comp	Physiol.	1997;273(5):R1800-9.		199.		 Rea	MS.	Human	health	and	well-being:	promises	for	a	bright	future	from	solid-state	lighting.	In:	Streubel	KP,	Jeon	H,	Tu	L-W,	Linder	N,	editors.	SPIE	7954,	Light-Emitting	Diodes:	Materials,	Devices,	and	Applications	for	Solid	State	Lighting	XV.	San	Francisco,	California:	International	Society	for	Optics	and	Photonics;	2011.		200.		 Alfredsson	L,	Spetz	C-L,	Theorell	T.	Type	of	occupation	and	near-future	hospitalization	for	myocardial	infarction	and	some	other	diagnoses.	Int	J	Epidemiol.	1985;14(3):378–88.		201.		 Patten	SB,	Williams	JVA,	Lavorato	DH,	Wang	JL,	McDonald	K,	Bulloch	AGM.	Major	depression	in	Canada:	what	has	changed	over	the	past	10	years?	Can	J	Psychiatry.	2016;61(2):80–5.		202.		 Lim	K,	Jacobs	P,	Ohinmaa	A,	Schopflocher	D,	Dewa	C.	A	new	population-based	measure	of	the	economic	burden	of	mental	illness	in	Canada.	Chronic	Dis	Can.	2008;28(3):92–8.		203.		 Cuijpers	P,	Beekman	ATF,	Reynolds	CF,	III.	Preventing	depression:	a	global	priority.	J	Am	Med	Assoc.	2012;307(10):1033–4.		204.		 Mykletun	A,	Harvey	SB.	Prevention	of	mental	disorders:	a	new	era	for	workplace	mental	health.	Occup	Environ	Med.	2012;69(12):868–9.		205.		 Boivin	DB,	Czeisler	CA,	Dijk	D-J,	Duffy	JF,	Folkard	S,	Minors	DS,	et	al.	Complex	interaction	of	the	sleep-wake	cycle	and	circadian	phase	modulates	mood	in	healthy	subjects.	Arch	Gen	Psychiatry.	1997;54(2):145–52.		206.		 McClung	CA.	How	might	circadian	rhythms	control	mood?	Let	me	count	the	ways...	 102 Biol	Psychiatry.	2013;74(4):242–9.		207.		 Grandin	LD,	Alloy	LB,	Abramson	LY.	The	social	zeitgeber	theory,	circadian	rhythms,	and	mood	disorders:	Review	and	evaluation.	Clin	Psychol	Rev.	2006;26(6):679–94.		208.		 LeGates	T,	Altimus	C,	Wang	H,	Lee	H,	Yang	S,	Zhao	H,	et	al.	Aberrant	light	directly	impairs	mood	and	learning	through	melanopsin-expressing	neurons.	Nature.	2012;491(7425):594–8.		209.		 Fonken	L,	Nelson	R.	Dim	light	at	night	increases	depressive-like	responses	in	male	C3H/HeNHsd	mice.	Behav	Brain	Res.	2013;243C:74–8.		210.		 Stephenson	KM,	Schroder	CM,	Bertschy	G,	Bourgin	P.	Complex	interaction	of	circadian	and	non-circadian	effects	of	light	on	mood:	shedding	new	light	on	an	old	story.	Sleep	Med	Rev.	2012;16(5):445–54.		211.		 Kecklund	G,	Axelsson	J.	Health	consequences	of	shift	work	and	insufficient	sleep.	BMJ.	2016;(355):i5210.		212.		 Lowden	A,	Akerstedt	T,	Wibom	R.	Suppression	of	sleepiness	and	melatonin	by	bright	light	exposure	during	breaks	in	night	work.	J	Sleep	Res.	2004;13(1):37–43.		213.		 Jansen	NWH,	Kant	I,	Nijhuis	FJN,	Swaen	GMH,	Kristensen	TS.	Impact	of	worktime	arrangements	on	work-home	interference	among	Dutch	employees.	Scand	J	Work	Environ	Health.	2004;30(2):139–48.		214.		 Paykel	ES.	Life	events,	social	support	and	depression.	Acta	Psychiatr	Scand	Suppl.	1994;(377):50–8.		215.		 Waddell	G,	Burton	AK.	Is	work	good	for	your	health	and	well-being?	Norwich,	UK:	The	Stationery	Office;	2006.		216.		 Harvey	SB,	Modini	M,	Joyce	S,	Milligan-Saville	JS,	Tan	L,	Mykletun	A,	et	al.	Can	work	make	you	mentally	ill?	A	systematic	meta-review	of	work-related	risk	factors	for	common	mental	health	problems.	Occup	Environ	Med.	2017;60(11):1105–15.		217.		 Moon	HJ,	Lee	SH,	Lee	HS,	Lee	K-J,	Kim	JJ.	The	association	between	shift	work	and	depression	in	hotel	workers.	Ann	Occup	Environ	Med.	2015;27(1):29–39.		218.		 Geiger-brown	J,	Muntaner	C,	Lipscomb	J,	Trinkoff	A.	Demanding	work	schedules	and	mental	health	in	nursing	assistants	working	in	nursing	homes.	Work	Stress.	2004;18(4):292–304.		219.		 Council	of	the	European	Union	EP.	Directive	2003/88/EC	of	the	European	Parliament	and	of	the	Council	of	4	November	2003	Concerning	Certain	Aspects	of	the	Organisation	of	Working	Time	[Internet].	2003	[cited	2017	Mar	23].	Available	from:	http://eur-lex.europa.eu/legal-content/EN/TXT/?uri=celex%3A32003L0088	220.		 Åkerstedt	T,	Kecklund	G.	What	work	schedule	characteristics	constitute	a	problem	to	the	individual?	A	representative	study	of	Swedish	shift	workers.	Appl	Ergon.	2017;59(Pt	A):320–5.		221.		 Vedaa	Ø,	Harris	A,	Bjorvatn	B,	Waage	S,	Sivertsen	B,	Tucker	P,	et	al.	Systematic	review	of	the	relationship	between	quick	returns	in	rotating	shift	work	and	health- 103 related	outcomes.	Ergonomics.	2016;59(1):1–14.		222.		 Quinlan	M,	Mayhew	C,	Bohle	P.	The	global	expansion	of	precarious	employment,	work	disorganization,	and	consequences	for	occupational	health:	a	review	of	recent	research.	Int	J	Heal	Serv.	2001;31(2):335–414.		223.		 Kim	W,	Park	E-C,	Lee	T-H,	Kim	TH.	Effect	of	working	hours	and	precarious	employment	on	depressive	symptoms	in	South	Korean	employees:	a	longitudinal	study.	Occup	Environ	Med.	2016;73(12):816–22.		224.		 Burgard	SA,	Brand	JE,	House	JS.	Perceived	job	insecurity	and	worker	health	in	the	United	States.	Soc	Sci	Med.	2009;69(5):777–85.		225.		 Lewchuk	W,	Lafleche	M.	Precarious	employment:	what	it	means	for	workers	and	their	families.	In:	Burke	RJ,	Page	KM,	editors.	Research	Handbook	on	Work	and	Well-Being.	Cheltenham,	UK:	Edward	Elgar	Publishing;	2017.	p.	150–69.		226.		 Shields	M,	Wilkins	K.	Findings	from	the	2005	National	Survey	of	the	Work	and	Health	of	Nurses.	Ottawa,	Canada:	Statistics	Canada	and	Canadian	Institute	for	Health	Information;	2006.		227.		 Shields	M.	Stress	and	depression	in	the	employed	population.	Ottawa,	Canada:	Statistics	Canada;	2006.		228.		 Kessler	RC,	Greenberg	PE,	Mickelson	KD,	Meneades	LM,	Wang	PS.	The	effects	of	chronic	medical	conditions	on	work	loss	and	work	cutback.	J	Occup	Environ	Med.	2001;43(3):218–25.		229.		 Dewa	CS,	Goering	P,	Lin	E,	Paterson	M.	Depression-related	short-term	disability	in	an	employed	population.	J	Occup	Environ	Med.	2002;44(7):628–33.		230.		 Gorman	E,	Yu	S,	Alamgir	H.	When	healthcare	workers	get	sick:	exploring	sickness	absenteeism	in	British	Columbia,	Canada.	Work.	2010;35(2):117–23.		231.		 Patten	SB,	Williams	JVA,	Lavorato	DH,	Bulloch	AGM,	Wiens	K,	Wang	J.	Why	is	major	depression	prevalence	not	changing?	J	Affect	Disord.	2016;190:93–7.		232.		 Michie	S,	Williams	S.	Reducing	work	related	psychological	ill	health	and	sickness	absence:	a	systematic	literature	review.	Occup	Environ	Med.	2003;60(1):3–9.		233.		 Franche	R-L,	Murray	E,	Ibrahim	S,	Smith	P,	Carnide	N,	Cote	P,	et	al.	Examining	the	Impact	of	Worker	and	Workplace	Factors	on	Prolonged	Work	Absences	Among	Canadian	Nurses.	J	Occup	Environ	Med.	Lippincott	Williams	&	Wilkins;	2011;53(8):919–27.		234.		 Statistics	Canada.	National	Survey	of	the	Work	and	Health	of	Nurses,	2005	-	Microdata	User	Guide.	Ottawa,	Canada;	2006.		235.		 Statistics	Canada.	The	Research	Data	Centres	Program	[Internet].	Government	of	Canada.	2017	[cited	2017	Mar	25].	Available	from:	http://www.statcan.gc.ca/eng/rdc/index	236.		 Kessler	RC,	McGonagle	KA,	Zhao	S,	Nelson	CB,	Hughes	M,	Eshleman	S,	et	al.	Lifetime	and	12-month	prevalence	of	DSM-III-R	psychiatric	disorders	in	the	United	States.	 104 Results	from	the	National	Comorbidity	Survey.	Arch	Gen	Psychiatry.	1994;51(1):8–19.		237.		 Kessler	RC,	Andrews	G,	Mroczek	D,	Ustun	B,	Wittchen	H-U.	The	World	Health	Organization	Composite	International	Diagnostic	Interview	short-form	(CIDI-SF).	Int	J	Methods	Psychiatr	Res.	1998;7(4):171–85.		238.		 Diagnostic	and	Statistical	Manual	of	Mental	Disorders,	Third	Edition,	Text	Revision	(DSM-III-R).	Washington	DC,	USA:	American	Psychiatric	Association;	1987.		239.		 Patten	SB,	Brandon-Christie	J,	Devji	J,	Sedmak	B.	Performance	of	the	composite	international	diagnostic	interview	short	form	for	major	depression	in	a	community	sample.	Chronic	Dis	Can.	2000;21(2):68–72.		240.		 Niedhammer	I,	Lesuffleur	T,	Algava	E,	Chastang	JF.	Classic	and	emergent	psychosocial	work	factors	and	mental	health.	Occup	Med	(Chic	Ill).	2015;65(2):126–34.		241.		 Greenland	S,	Pearl	J,	Robins	JM.	Causal	diagrams	for	epidemiologic	research.	Epidemiology.	1999;10(1):37–48.		242.		 Bonde	JP.	Psychosocial	factors	at	work	and	risk	of	depression:	a	systematic	review	of	the	epidemiological	evidence.	Occup	Environ	Med.	2008;65(7):438–45.		243.		 Siegrist	J.	Chronic	psychosocial	stress	at	work	and	risk	of	depression:	evidence	from	prospective	studies.	Eur	Arch	Psychiatry	Clin	Neurosci.	2008;258(S5):115–9.		244.		 Ferrari	AJ,	Charlson	FJ,	Norman	RE,	Patten	SB,	Freedman	G,	Murray	CJL,	et	al.	Burden	of	Depressive	Disorders	by	Country,	Sex,	Age,	and	Year:	Findings	from	the	Global	Burden	of	Disease	Study	2010.	PLoS	Med.	2013;10(11):e1001547.		245.		 Gilmour	H,	Patten	SB.	Depression	and	work	impairment.	Heal	Reports.	2007;18(1):9–22.		246.		 Alonso	J,	Angermeyer	MC,	Bernert	S,	Bruffaerts	R,	Brugha	TS,	Bryson	H,	et	al.	Prevalence	of	mental	disorders	in	Europe:	results	from	the	European	Study	of	the	Epidemiology	of	Mental	Disorders	(ESEMeD)	project.	Acta	Psychiatr	Scand.	2004;109(s420):21–7.		247.		 Patten	SB,	Beck	CA,	Kassam	A,	Williams	JVA,	Barbui	C,	Metz	LM.	Long-term	medical	conditions	and	major	depression:	strength	of	association	for	specific	conditions	in	the	general	population.	Can	J	Psychiatry.	2005;50(4):195–202.		248.		 Egede	LE.	Major	depression	in	individuals	with	chronic	medical	disorders:	prevalence,	correlates	and	association	with	health	resource	utilization,	lost	productivity	and	functional	disability.	Gen	Hosp	Psychiatry.	2007;29(5):409–16.		249.		 Nijp	HH,	Beckers	DG,	Geurts	SA,	Tucker	P,	Kompier	MA.	Systematic	review	on	the	association	between	employee	worktime	control	and	work–non-work	balance,	health	and	well-being,	and	job-related	outcomes.	Scand	J	Work	Environ	Health.	2012;38(4):299–313.		250.		 Cocco	E,	Gatti	M,	de	Mendonça	Lima	CA,	Camus	V.	A	comparative	study	of	stress	and	burnout	among	staff	caregivers	in	nursing	homes	and	acute	geriatric	wards.	Int	J	 105 Geriatr	Psychiatry.	2003;18(1):78–85.		251.		 Virtanen	M,	Ferrie	JE,	Singh-Manoux	A,	Shipley	MJ,	Stansfeld	SA,	Marmot	MG,	et	al.	Long	working	hours	and	symptoms	of	anxiety	and	depression:	a	5-year	follow-up	of	the	Whitehall	II	study.	Psychol	Med.	2011;41(12):2485–94.		252.		 Lowden	A,	Kecklund	G,	Axelsson	J,	Akerstedt	T.	Change	from	an	8-hour	shift	to	a	12-hour	shift,	attitudes,	sleep,	sleepiness	and	performance.	Scand	J	Work	Environ	Health.	1998;24(Suppl	3):69–75.		253.		 Rothman	K.	Random	Error	and	the	Role	of	Statistics.	In:	Epidemiology,	An	Introduction.	New	York,	USA:	Oxford	University	Press;	2002.	p.	113–29.		254.		 Lewchuk	W,	Lafleche	M,	Procyk	S,	Cook	C,	Dyson	D,	Goldring	L,	et	al.	The	Precarity	Penalty.	Toronto,	Canada:	Poverty	and	Employment	Precarity	in	Southern	Ontario	(PEPSO)	Research	Group;	2015.		255.		 Flo	E,	Pallesen	S,	Moen	BE,	Waage	S,	Bjorvatn	B.	Short	rest	periods	between	work	shifts	predict	sleep	and	health	problems	in	nurses	at	1-year	follow-up.	Occup	Environ	Med.	2014;71(8):555–61.		256.		 Eldevik	MF,	Flo	E,	Moen	BE,	Pallesen	S,	Bjorvatn	B,	Härmä	M,	et	al.	Insomnia,	excessive	sleepiness,	excessive	fatigue,	anxiety,	depression	and	shift	work	disorder	in	nurses	having	less	than	11	hours	in-between	shifts.	PLoS	One.	2013;8(8):e70882.		257.		 Dahlgren	A,	Tucker	P,	Gustavsson	P,	Rudman	A.	Quick	returns	and	night	work	as	predictors	of	sleep	quality,	fatigue,	work–family	balance	and	satisfaction	with	work	hours.	Chronobiol	Int.	2016;33(6):759–67.		258.		 Vedaa	Ø,	Pallesen	S,	Waage	S,	Bjorvatn	B,	Sivertsen	B,	Erevik	E,	et	al.	Short	rest	between	shift	intervals	increases	the	risk	of	sick	leave:	a	prospective	registry	study.	Occup	Environ	Med.	2016;Published	Online	First:	08	November	2016.		259.		 Haus	E,	Smolensky	M.	Biological	clocks	and	shift	work:	circadian	dysregulation	and	potential	long-term	effects.	Cancer	Causes	Control.	2006;17(4):489–500.		260.		 Martin	F,	Poyen	D,	Bouderlique	E,	Gouvernet	J,	Rivet	B,	Disdier	P,	et	al.	Depression	and	burnout	in	hospital	health	care	professionals.	Int	J	Occup	Environ	Health.	1997;3(3):204–9.		261.		 Cushway	D,	Tyler	PA,	Nolan	P.	Development	of	a	stress	scale	for	mental	health	professionals.	Br	J	Clin	Psychol.	1996;35(2):279–95.		262.		 Purvanova	RK,	Muros	JP.	Gender	differences	in	burnout:	A	meta-analysis.	J	Vocat	Behav.	2010;77(2):168–85.		263.		 Theorell	T,	Hammarström	A,	Gustafsson	PE,	Magnusson	Hanson	L,	Janlert	U,	Westerlund	H.	Job	strain	and	depressive	symptoms	in	men	and	women:	a	prospective	study	of	the	working	population	in	Sweden.	J	Epidemiol	Community	Health.	2014;68(1):78–82.		264.		 Waage	S,	Pallesen	S,	Moen	BE,	Magerøy	N,	Flo	E,	Di	Milia	L,	et	al.	Predictors	of	shift	work	disorder	among	nurses:	a	longitudinal	study.	Sleep	Med.	2014;15(12):1449–55.		 106 265.		 Picciotto	S,	Hertz-Picciotto	I.	Commentary:	Healthy	Worker	Survivor	Bias:	A	Still-Evolving	Concept.	Epidemiology.	2015;26(2):213–5.		266.		 Härmä	M,	Koskinen	A,	Ropponen	A,	Puttonen	S,	Karhula	K,	Vahtera	J,	et	al.	Validity	of	self-reported	exposure	to	shift	work.	Occup	Environ	Med.	2016;74(3):228–30.		267.		 O’Donnell	S,	Vanderloo	S,	McRae	L,	Onysko	J,	Patten	SB,	Pelletier	L.	Comparison	of	the	estimated	prevalence	of	mood	and/or	anxiety	disorders	in	Canada	between	self-report	and	administrative	data.	Epidemiol	Psychiatr	Sci.	2016;25(4):360–9.		268.		 Koehoorn	M,	Ratner	PA,	Shamian	J.	Feasibility	of	using	existing	Statistics	Canada	surveys	to	describe	the	health	and	work	of	nurses.	Nurs	Leadersh.	2003;16(2):94–106.		269.		 Letvak	S,	Ruhm	CJ,	McCoy	T.	Depression	in	hospital-employed	nurses.	Clin	Nurse	Spec.	2012;26(3):177–82.		270.		 Stansfeld	SA,	Fuhrer	R,	Shipley	MJ,	Marmot	MG.	Work	characteristics	predict	psychiatric	disorder:	prospective	results	from	the	Whitehall	II	Study.	Occup	Environ	Med.	1999	May;56(5):302–7.		271.		 Tepas	D.	Educational	programmes	for	shiftworkers,	their	families,	and	prospective	shiftworkers.	Ergonomics.	1993;36(1–3):199–209.		272.		 Nielsen	K,	Miraglia	M.	What	works	for	whom	in	which	circumstances?	On	the	need	to	move	beyond	the	“what	works?”	question	in	organizational	intervention	research.	Hum	Relations.	2017	Jan;70(1):40–62.		273.		 Kristensen	T.	Workplace	intervention	studies.	Occup	Med	(Chic	Ill).	2000;15(1):293–305.		274.		 Brough	P,	O’Driscoll	MP.	Organizational	interventions	for	balancing	work	and	home	demands:	An	overview.	Work	Stress.	2010;24(3):280–97.		275.		 Folkard	S.	Shift	work,	safety	and	productivity.	Occup	Med	(Chic	Ill).	2003;53(2):95–101.		276.		 DeJoy	DM.	Behavior	change	versus	culture	change:	Divergent	approaches	to	managing	workplace	safety.	Saf	Sci.	2005;43(2):105–29.		277.		 Statistics	Canada.	Labour	Force	Information	–	December	7	to	13,	2014	(Catalogue	no.	71-001-X).	Ottawa,	Canada;	2015.		278.		 Statistics	Canada.	2006	Census	of	Population.	Ottawa,	Canada;	2006.		279.		 Mistlberger	R.	Shiftwork	practise	in	British	Columbia	(Report	to	WorkSafeBC)	[Internet].	2004	[cited	2017	Mar	23].	Available	from:	https://www.worksafebc.com/en/resources/about-us/research/shiftwork-practice-in-british-columbia?lang=en&origin=s&returnurl=https%3A%2F%2Fwww.worksafebc.com%2Fen%2Fsearch%23q%3Dmistlberger%26sort%3Drelevancy%26f%3Alanguage-facet%3D%5BEnglish%5D&highlight=	280.		 Economic	Policy	Institute.	Briefing	Paper#394:	Irregular	Work	Scheduling	and	its	 107 Consequences.	Washington,	DC;	2015.		281.		 Hasle	P,	Limborg	HJ.	A	review	of	the	literature	on	preventive	occupational	health	and	safety	activities	in	small	enterprises.	Ind	Health.	2006;44(1):6–12.		282.		 MacEachen	E,	Kosny	A,	Scott-Dixon	K,	Facey	M,	Chambers	L,	Breslin	C,	et	al.	Workplace	health	understandings	and	processes	in	small	businesses:	a	systematic	review	of	the	qualitative	literature.	J	Occup	Rehabil.	2010;20(2):180–98.		283.		 Eakin	JM,	Champoux	D,	MacEachen	E.	Health	and	safety	in	small	workplaces:	refocusing	upstream.	Can	J	public	Heal.	2010;101(Suppl	1):S29-33.		284.		 Champoux	D,	Brun	J-P.	Occupational	health	and	safety	management	in	small	size	enterprises:	an	overview	of	the	situation	and	avenues	for	intervention	and	research.	Saf	Sci.	2003;41(4):301–18.		285.		 Aronsson	G,	Gustafsson	K,	Dallner	M.	Work	environment	and	health	in	different	types	of	temporary	jobs.	Eur	J	Work	Organ	Psychol.	2002;11(2):151–75.		286.		 Tompa	E,	Kalcevich	C,	Foley	M,	McLeod	C,	Hogg-Johnson	S,	Cullen	K,	et	al.	A	systematic	literature	review	of	the	effectiveness	of	occupational	health	and	safety	regulatory	enforcement.	Am	J	Ind	Med.	2016;59(11):919–33.		287.		 Egan	M,	Bambra	C,	Petticrew	M,	Whitehead	M.	Reviewing	evidence	on	complex	social	interventions:	appraising	implementation	in	systematic	reviews	of	the	health	effects	of	organisational-level	workplace	interventions.	J	Epidemiol	Community	Health.	2009;63(1):4–11.		288.		 Garde	AH,	Albertsen	K,	Nabe-Nielsen	K,	Carneiro	IG,	Skotte	J,	Hansen	SM,	et	al.	Implementation	of	self-rostering	(the	PRIO-project):	effects	on	working	hours,	recovery,	and	health.	Scand	J	Work	Environ	Health.	2012;38(4):314–26.		289.		 Statistics	Canada.	Labour	Force	Survey.	Ottawa,	Canada:	Government	of	Canada;	2014.		290.		 Smith	TJ,	Hammond	SK,	Hallock	M,	Woskie	SR.	Exposure	Assessment	for	Epidemiology:	Characteristics	of	Exposure.	Appl	Occup	Environ	Hyg.	1991	Jun	24;6(6):441–7.		291.		 Monk	TH.	Coping	with	the	stress	of	shift	work.	Work	Stress.	1988	Apr;2(2):169–72.		292.		 Merkus	SL,	Holte	KA,	Huysmans	MA,	van	Mechelen	W,	van	der	Beek	AJ.	Nonstandard	working	schedules	and	health:	the	systematic	search	for	a	comprehensive	model.	BMC	Public	Health.	2015;15(1):1084.		293.		 Heederik	D,	Boleij	JSM,	Kromhout	H,	Smid	T.	Use	and	Analysis	of	Exposure	Monitoring	Data	in	Occupational	Epidemiology:	An	Example	of	an	Epidemiological	Study	in	the	Dutch	Animal	Food	Industry.	Appl	Occup	Environ	Hyg.	1991;6(6):458–64.		294.		 Burdorf	A,	Lillienberg	L,	Brisman	J.	Characterization	of	exposure	to	inhalable	flour	dust	in	Swedish	bakeries.	Ann	Occup	Hyg.	1994;38(1):67–78.		295.		 Mclntyre	IM,	Norman	TR,	Burrows	GD,	Armstrong	SM.	Human	melatonin	 108 suppression	by	light	is	intensity	dependent.	J	Pineal	Res.	1989;6(2):149–56.		296.		 Lockley	SW,	Evans	EE,	Scheer	FAJL,	Brainard	GC,	Czeisler	CA,	Aeschbach	D.	Short-wavelength	sensitivity	for	the	direct	effects	of	light	on	alertness,	vigilance,	and	the	waking	electroencephalogram	in	humans.	Sleep.	2006;29(2):161–8.		297.		 Garde	AH,	Hansen	J,	Kolstad	HA,	Larsen	AD,	Hansen	ÅM.	How	do	different	definitions	of	night	shift	affect	the	exposure	assessment	of	night	work?	Chronobiol	Int.	2016;33(6):595–8.		298.		 Perrin	F,	Peigneux	P,	Fuchs	S,	Verhaeghe	S,	Laureys	S,	Middleton	B,	et	al.	Nonvisual	Responses	to	Light	Exposure	in	the	Human	Brain	during	the	Circadian	Night.	Curr	Biol.	2004;14(20):1842–6.		299.		 Lowden	A,	Åkerstedt	T,	Wibom	R.	Suppression	of	sleepiness	and	melatonin	by	bright	light	exposure	during	breaks	in	night	work.	J	Sleep	Res.	2004;13(1):37–43.		300.		 Campbell	SS,	Dawson	D.	Enhancement	of	nighttime	alertness	and	performance	with	bright	ambient	light.	Physiol	Behav.	1990;48(2):317–20.		301.		 Figueiro	MG,	Bierman	A,	Plitnick	B,	Rea	MS.	Preliminary	evidence	that	both	blue	and	red	light	can	induce	alertness	at	night.	BMC	Neurosci.	2009;10(1):105.		302.		 Stone	PW,	Du	Y,	Cowell	R,	Amsterdam	N,	Helfrich	TA,	Linn	RW,	et	al.	Comparison	of	nurse,	system	and	quality	patient	care	outcomes	in	8-hour	and	12-hour	shifts.	Med	Care.	2006;44(12):1099–106.		303.		 Knauth	P.	Extended	work	periods.	Ind	Health.	2007;45(1):125–36.		304.		 Geiger-Brown	J,	Trinkoff	AM.	Is	it	time	to	pull	the	plug	on	12-hour	shifts?	J	Nurs	Adm.	2010;40(3):100–2.		305.		 Wagstaff	AS,	Sigstad	Lie	J-A.	Shift	and	night	work	and	long	working	hours--a	systematic	review	of	safety	implications.	Scand	J	Work	Environ	Health.	2011;37(3):173–85.		306.		 Wergeland	EL,	Veiersted	B,	Ingre	M,	Olsson	B,	Akerstedt	T,	Bjørnskau	T,	et	al.	A	shorter	workday	as	a	means	of	reducing	the	occurrence	of	musculoskeletal	disorders.	Scand	J	Work	Environ	Health.	2003;29(1):27–34.			        109 Appendix	A			Supporting	documentation	for	Chapter	2	(Personal	light-at-night	exposures	and	components	of	variability	in	two	industry	sectors	where	shift	work	is	common)	A.1 Additional	detail	on	study	recruitment		Study	approval	to	contact	workers	and	conduct	on-site	measurements	was	obtained	from	the	Workplace	Health	Division	of	British	Columbia’s	Provincial	Health	Services	Authority,	and	from	upper	management	representing	both	the	British	Columbia	provincial	emergency	medical	services	and	the	British	Columbia	Women’s	Hospital.	Study	approvals	were	then	obtained	from	subsequent	levels	of	management	for	each	occupational	group	within	emergency	services	(paramedics	and	dispatch	workers)	and	the	Women’s	Hospital	(nursing,	security,	patient	care	aides,	unit	clerks,	and	laboratory	staff).	Worker	recruitment	was	conducted	in	a	manner	suitable	to	each	worksite.	For	instance,	paramedics	at	participating	ambulance	stations	were	informed	in	advance	about	the	study	via	emailed	study	introduction	letters	and	notices	posted	at	their	worksite;	workers	that	were	present	on	four	pre-determined	sampling	dates	were	then	invited	to	participate	ad	hoc.	For	nurses,	a	study	notice	was	sent	to	all	Women’s	Hospital	nursing	staff	in	a	weekly	emailed	newsletter,	and	notices	were	posted	in	staff	rest	areas	to	invite	nurses	to	contact	the	research	team	to	participate. 110 Appendix	B			Supporting	documentation	for	Chapter	3	(Examining	the	impacts	of	exposure	assignment	in	a	study	of	shift	work	and	depression	among	nurses)	B.1 Depression	scoring	in	the	National	Survey	of	the	Work	and	Health	of	Nurses		The	following	text	in	Section	B.1	was	taken	directly	from	Shields	and	Wilkins’	summary	of	findings	from	the	2005	National	Survey	of	the	Work	and	Health	of	Nurses	(1),	p.	107-109:	Using	the	methodology	of	Kessler	et	al.	(2),	history	of	a	major	depressive	episode	(MDE)	was	measured	using	a	subset	of	questions	from	the	Composite	International	Diagnostic	Interview.	These	questions	cover	a	cluster	of	symptoms	for	a	depressive	disorder,	which	are	listed	in	the	Diagnostic	and	Statistical	Manual	of	Mental	Disorders	(DSM-III-R)	(3).		Two	screening	questions	were	used	for	the	depression	module.	The	first	was,		• During	the	past	12	months,	was	there	ever	a	time	when	you	felt	sad,	blue,	or	depressed	for	two	weeks	or	more	in	a	row?	Respondents	who	replied	“yes”	to	this	question	were	instructed	to	think	about	the	2-week	period	during	the	past	12	months	when	these	feeling	were	the	worst,	in	reference	to	the	following	questions:		1. “During	that	time	how	long	did	these	feelings	usually	last	.	.	.all	day	long,	most	of	the	day,	about	half	of	the	day,	or	less	than	half	the	day?”	(Respondents	who	replied	“about	half	of	the	day”	or	“less	than	half	the	day”	were	not	asked	questions	2	to	11).			2. “How	often	did	you	feel	this	way	during	those	two	weeks	.	.	.	every	day,	almost	every	day,	or	less	often?”	(Respondents	who	replied	“less	often”	were	not	asked	questions	3	to	11.)			3. “During	those	two	weeks	did	you	lose	interest	in	most	things?”	(yes	/	no)			4. “Did	you	feel	tired	out	or	low	on	energy	all	of	the	time?”	(yes	/	no)			5. “Did	you	gain	weight,	lose	weight,	or	stay	about	the	same?”	(“gained	weight,”	“lost	weight,”	“stayed	about	the	same,”	“was	on	a	diet.”)			 111 6. “About	how	much	did	you	gain/lose?”	(not	asked	if	the	respondent	was	on	a	diet)			7. “Did	you	have	more	trouble	falling	asleep	than	you	usually	do?”	(yes	/	no)			8. “How	often	did	that	happen	.	.	.	every	night,	nearly	every	night,	or	less	often?”			9. “Did	you	have	a	lot	more	trouble	concentrating	than	usual?”	(yes	/	no)			10. “At	these	times,	people	sometimes	feel	down	on	themselves,	no	good,	or	worthless.	Did	you	feel	this	way?”	(yes	/	no)			11. “Did	you	think	a	lot	about	death;	either	your	own,	someone	else’s,	or	death	in	general?”	(yes	/	no).			The	second	screening	question	for	the	depression	module,	“During	the	past	12	months,	was	there	ever	a	time	lasting	two	weeks	or	more	when	you	lost	interest	in	most	things	like	hobbies,	work,	or	activities	that	usually	give	you	pleasure?”	was	asked	of	respondents	who	replied	“no”	to	the	first	screening	question	or	were	“skipped	out”	of	the	subsequent	questions	based	on	their	response	to	item	1	or	2.		Respondents	who	replied	“yes”	to	the	second	screening	question	were	asked	the	same	follow-up	questions	(1	to	11)	but	with	the	difference	that	they	were	instructed	to	“think	about	the	2-week	period	during	the	past	12	months	when	you	had	the	most	complete	loss	of	interest	in	things.”	As	well,	the	wording	for	item	1	was	revised	to	“During	that	two-week	period,	how	long	did	the	loss	of	interest	usually	last?”,	and	item	3	was	not	asked.		To	derive	a	depression	score,	all	respondents	were	assigned	an	initial	value	of	0.	For	each	of	the	eight	criteria	listed	below	that	was	met,	a	value	of	1	was	added	to	the	initial	value;	thus	the	total	score	could	range	from	0	to	8.		• a	response	of	“yes”	to	either	screening	question			• a	response	of	“yes”	to	item	3			• a	response	of	“yes”	to	item	4			• a	change	of	weight	of	at	least	10	pounds	(4.5	kilograms),	indicated	in	item	6			• a	response	of	“every	night”	or	“nearly	every	night”	to	item	8			• a	response	of	“yes”	to	item	9			• a	response	of	“yes”	to	item	10			 112 • a	response	of	“yes”	to	item	11			The	scores	were	transformed	into	a	probability	estimate	of	the	occurrence	of	an	MDE.	For	the	analysis,	if	the	estimate	was	0.9	or	more,	that	is,	90%	likelihood	of	an	MDE,	the	respondent	was	considered	to	have	experienced	a	depressive	episode	in	the	previous	12	months.	To	obtain	a	probability	of	0.9,	respondents	had	to	score	5	or	more.			References	1. Shields	M,	Wilkins	K.	Findings	from	the	2005	National	Survey	of	the	Work	and	Health	of	Nurses.	Ottawa,	Canada:	Statistics	Canada	and	Canadian	Institute	for	Health	Information;	2006.	p.	107-109.	2. Kessler	RC,	McGonagle	KA,	Zhao	S,	Nelson	CB,	Hughes	M,	Eshleman	S,	et	al.	Lifetime	and	12-month	prevalence	of	DSM-III-R	psychiatric	disorders	in	the	United	States.	Results	from	the	National	Comorbidity	Survey.	Arch	Gen	Psychiatry.	1994;51(1):8–19.	3. Diagnostic	and	Statistical	Manual	of	Mental	Disorders,	Third	Edition,	Text	Revision	(DSM-III-R).	Washington	DC,	USA:	American	Psychiatric	Association;	1987.										 113 Appendix	C			Supporting	documentation	for	Chapter	4	(Assessing	determinants	of	workplace-level	shift	work	policies	and	practices:	An	employer	survey	in	British	Columbia,	Canada)	 114 C.1 Copy	of	questionnaire	used	  115   116   117   118   119   120   121   122   123   124   125 C.2 Summary	of	study	recruitment		Industry	Description		(2-digit	code,	NAICS	2012)	%	Shift	workers	by	BC	industry1		Recruitment	of	organizations	via	shift	work	database		Targeted	recruit-ment	(n)	Study	total		Total	in	database		(n)	Consent	(n)		Refusal	(n)	Out	of	business	(n)		Non-response	(n)		Response	rate	(%)		Agriculture,	forestry,	fishing,	hunting	(11)		 3.4	 11	 5	 0	 1	 5	 50	 0	 5	Mining,	quarrying,	oil	and	gas	extraction	(21)		 5	 2	 0	 1	 2	 50	 0	 2	Utilities	(22)	 0.1	 2	 2	 0	 0	 0	 100	 0	 2	Construction	(23)	 2.0	 3	 1	 2	 0	 0	 33	 1	 2	Manufacturing	(31-33)	 13.8	 38	 16	 13	 4	 5	 47	 0	 16	Wholesale	trade	(41)	 23.8	 5	 4	 0	 1	 0	 100	 0	 4	Retail	trade	(44-45)	 5	 3	 2	 0	 0	 60	 1	 4	Transportation	and	warehousing	(48-49)	 7.7	 28	 15	 7	 0	 6	 54	 0	 142	Information	and	cultural	industries	(51)	 3.0	 9	 2	 1	 2	 4	 29	 0	 2	Arts,	entertainment	and	recreation	(71)	 10	 5	 0	 2	 3	 63	 0	 5	Finance	and	insurance	(52)	 1.2	 1	 0	 0	 1	 0	 0	 0	 0	Real	estate	and	rental	and	leasing	(53)	 0	 0	 0	 0	 0	 N/A	 0	 0	Professional,	scientific	and	technical	services	(54)	 0.3	 4	 0	 0	 0	 4	 0	 0	 0	Management	of	companies	and	enterprises	(55)	 4.6	 0	 0	 0	 0	 0	 N/A	 0	 0	Administrative	and	support,	waste	management	and	remediation	services	(56)	 8	 5	 1	 1	 1	 71	 0	 42	Educational	services	(61)	 1.0	 2	 2	 0	 0	 0	 100	 0	 2	Health	care	and	social	assistance	 15.5	 5	 5	 0	 0	 0	 100	 3	 8	 126 (62)	Accommodation	and	food	services	(72)	 17.8	 11	 6	 2	 2	 1	 67	 1	 7	Other	services	(except	public	administration)	(81)	 2.8	 3	 2	 1	 0	 0	 67	 0	 2	Public	administration	(91)	 3.2	 21	 8	 4	 3	 6	 44	 1	 9	TOTAL	 100	 171	 83	 33	 18	 37	 54	 7	 88	NAICS	=	North	American	Industry	Classification	System	BC	=	British	Columbia,	Canada	1	Reference:	Statistics	Canada.	2006	Census	of	Population.	Ottawa,	Canada;	2006.		2	One	organization	did	not	meet	the	study	definition	of	“shift	work”	and	was	excluded	  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0355851/manifest

Comment

Related Items