UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Improving hospital care for patients who use illicit drugs in Vancouver, Canada Ti, Lianping 2015

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

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata


24-ubc_2015_september_ti_lianping.pdf [ 1.2MB ]
JSON: 24-1.0166322.json
JSON-LD: 24-1.0166322-ld.json
RDF/XML (Pretty): 24-1.0166322-rdf.xml
RDF/JSON: 24-1.0166322-rdf.json
Turtle: 24-1.0166322-turtle.txt
N-Triples: 24-1.0166322-rdf-ntriples.txt
Original Record: 24-1.0166322-source.json
Full Text

Full Text

 IMPROVING  HOSPITAL  CARE  FOR  PATIENTS  WHO  USE  ILLICIT  DRUGS  IN  VANCOUVER,  CANADA    by  Lianping  Ti    B.Sc.,  McGill  University,  2009  M.P.H.,  Simon  Fraser  University,  2012    A  THESIS  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)    June  2015    ©  Lianping  Ti,  2015   ii Abstract  Background:   People   who   use   illicit   drugs   (PWUD)   experience   a   number   of   health-­‐‑related  harms   that   often   lead   to   frequent  hospitalizations.  However,   there   exists   little  scientific  evidence  that  has  explored  utilization  and  retention  in  hospital  care,  including  leaving  hospital  against  medical  advice  (AMA),  among  this  population.  The  objective  of  this   thesis   is   to   provide   evidence   to   improve   hospital   care   for   PWUD   by   first,  identifying  individual  and  contextual  factors  associated  with  leaving  hospital  AMA  and  other   hospital-­‐‑related   experiences;   and,   second,   to   use   these   findings   to   develop   and  evaluate  novel  approaches  to  minimizing  the  risks  and  harms  that  PWUD  experience  in  hospital  settings.    Methods:  This  dissertation  draws  on  data  collected  from  two  open  prospective  cohort  studies   of   HIV-­‐‑positive   and  HIV-­‐‑negative   PWUD   in   Vancouver,   Canada.   These   data  were   confidentially   linked   to   a   hospital   discharge  database   as  well   as   comprehensive  records   of  HIV   treatment   and   related   clinical   outcomes   through   a   clinical  monitoring  laboratory.  A  variety  of  longitudinal  and  cross-­‐‑sectional  analytic  techniques  were  used  to   examine   the   impact   of   intersecting   individual   and   contextual   factors   on   various  hospital  service  outcomes.      Results:  This  dissertation   identified  hospitals  as  a   risk  environment   for  PWUD,  given  the  high  prevalence  of  hospital  discharge  AMA  and  active  illicit  drug  use  in  hospitals.  The  study  findings  highlighted  various  risk  and  protective  factors  for   leaving  hospital  AMA,   and   discussed   the   negative   consequences   of   being   denied   pain  medication   on  illicit   drug   use   in   hospitals.   The   findings   from   this   dissertation   also   point   to   novel  strategies   that   may   address   these   issues,   including   the   implementation   of   an   adult  HIV/AIDS  integrated  health  program  operating  in  proximity  to  a  hospital  to  minimize  hospital  discharge  AMA  among  HIV-­‐‑positive  PWUD,  as  well  as  the  potential  for  an  in-­‐‑hospital   supervised   injection   facility   (SIF)   to   reduce   the   harms   associated   with   illicit  drug  use  in  hospital.    Conclusion:  Despite  this  setting  of  universal  access  to  healthcare,  there  are  individual-­‐‑  and  contextual-­‐‑level  factors  that  play  a  pivotal  role  in  shaping  utilization  and  retention  in  hospital  care  among  PWUD.  The  collective  findings  of  this  dissertation  offer  insights  into   how   integrated   harm   reduction-­‐‑based   interventions   may   mitigate   the   risks  associated  with  leaving  hospital  AMA  and  active  illicit  drug  use  in  hospitals.     iii Preface  This  statement  certifies  that  all  of  the  work  presented  henceforth  was  conceived,  undertaken   and   written   by   the   candidate,   Lianping   Ti   (LT).   All   empirical   research  conducted   for   this   dissertation   was   approved   by   the   University   of   British  Columbia/Providence   Health   Care   Research   Ethics   Board   (certificate   H05-­‐‑50233   and  H05-­‐‑50234).   The   co-­‐‑authors   of   the  manuscripts,   including  Dr.   Thomas  Kerr   (TK),  Dr.  Jane  Buxton  (JB),  Dr.  M-­‐‑J  Milloy  (M-­‐‑JM),  Dr.  Evan  Wood  (EW),  Dr.  Julio  Montaner  (JM),  Dr.  Ryan  McNeil  (RM),  Dr.  Kanna  Hayashi,  Ms.  Rosalind  Baltzer  Turje  (RBT),  Mr.  Scott  Harrison   (SH),  Ms.   Sabina   Dobrer   (SD),  Ms.   Pauline   Voon   (PV)   and  Ms.   Lianlian   Ti  (LLT)   made   contributions   only   as   is   commensurate   with   supervisory   committee,  collegial   or   co-­‐‑investigator  duties.   The  principal   investigators   of   the  VIDUS   (TK)   and  ACCESS  (EW)  studies,  from  which  all  empirical  analyses  were  derived,  has  access  to  all  of  the  data  and  as  corresponding  author  takes  full  responsibility  for  the  integrity  of  the  results   and   the   accuracy   of   the   analyses.   Relative   contributions   of   the   author,  collaborators,  and  co-­‐‑authors  are  described  in  detail  below.  Chapters   1   and   7   are   original,   unpublished   intellectual   products   of   the   author.  With  substantive  guidance  and  input  from  co-­‐‑supervisors  (TK  and  JB)  and  supervisory  committee  member   (M-­‐‑JM),  LT   searched  and   reviewed  all   of   the   literature  presented,  designed  the  research  and  synthesized  the  findings  of  these  chapters.  A  version  of  Chapter  2  is  currently  under  review  for  peer-­‐‑reviewed  publication:  Ti  L,  Ti  L.  “Leaving  hospital  against  medical  advice  among  people  who  use  illicit  drugs:  A  systematic  review.”  LT  designed  and  led  the  systematic  review  presented  in  Chapter  2   and,   in   collaboration   with   the   second   author,   conducted   the   search   strategy   and  selected  eligible  studies  for  final  inclusion.  A  version  of  Chapter  3  is  currently  under  review  for  peer-­‐‑reviewed  publication:  Ti  L,  Milloy  M-­‐‑J,  McNeil  R,  Buxton   J,  Dobrer  S,  Hayashi  K,  Wood  E,  Kerr  T.  “Factors  that  predict  leaving  hospital  against  medical  advice  among  people  who  use  illicit  drugs  in  Vancouver,   Canada.”   LT   and   TK  designed   the   study.   LT   took   the   primary   role   in  analyzing   the  data  and  prepared   the   first  draft  of   the  manuscript.  M-­‐‑JM,  RM,   JB,   SD,  KH,   EW   and   TK   provided   input   to   the   draft   and   contributed   to   the   revision   of   the  manuscript.  A   version   of   Chapter   4   has   been   published   and   is   reused   here   with   kind  permission  from  Pulsus  Group  Inc:  Ti  L,  Voon  P,  Dobrer  S,  Montaner  J,  Wood  E,  Kerr  T.  “Denial  of  pain  medication  by  health  care  providers  predicts  in-­‐‑hospital  illicit  drug  use  among  individuals  who  use  illicit  drugs.”  Pain  Research  and  Management,  2015;20(2):84-­‐‑88.  LT  and  TK  designed  the  study.  LT  conducted  the  statistical  analyses  and  prepared  the   first  draft  of   the  manuscript.  PV,  SD,   JM,  EW  and  TK  provided   input   to   the  draft  and  contributed  to  the  revision  of  the  manuscript.   iv A  version  of  Chapter  5  is  currently  under  review  for  peer-­‐‑reviewed  publication:  Ti   L,   Milloy   M-­‐‑J,   Baltzer   Turje   R,   Montaner   J,   Wood   E,   Kerr   T.   “The   impact   of   an  HIV/AIDS  adult  integrated  health  program  on  leaving  hospital  against  medical  advice  among  HIV-­‐‑positive  people  who  use  illicit  drugs.”  LT  and  TK  designed  the  study.  LT  conducted  the  statistical  analyses  and  prepared  the  first  draft  of  the  manuscript.  M-­‐‑JM,  RBT,  JM,  EW  and  TK  provided  input  to  the  draft  and  contributed  to  the  revision  of  the  manuscript.  A   version   of   Chapter   6   has   been   published   and   is   reused   here   with   kind  permission  from  John  Wiley  and  Sons:  Ti  L,  Buxton  J,  Harrison  S,  Dobrer  S,  Montaner  J,  Wood   E,   Kerr,   T.   “Willingness   to   access   an   in-­‐‑hospital   supervised   injection   facility  among   hospitalized   people   who   use   illicit   drugs.”   Journal   of   Hospital   Medicine,  2015;10(5):301-­‐‑306.  LT  and  TK  designed  the  study.  LT  conducted  the  statistical  analyses  and  prepared   the   first  draft   of   the  manuscript.   JB,   SH,   SD,   JM,  EW  and  TK  provided  input  to  the  draft  and  contributed  to  the  revision  of  the  manuscript.    v Table  of  Contents  Abstract  .........................................................................................................................................  ii  Preface  .........................................................................................................................................  iii  Table  of  Contents……………………………………………………………………………….v  List  of  Tables  ............................................................................................................................  viii  List  of  Figures  .............................................................................................................................  ix  List  of  Abbreviations  ..................................................................................................................  x  Acknowledgements  ...................................................................................................................  xi  Dedication  ................................................................................................................................  xiii  Chapter  1:  Introduction  ..............................................................................................................  1  1.1   Hospitalization  among  people  who  use  illicit  drugs  ..........................................................  1  1.2   Study  justification  ....................................................................................................................  3  1.3   Study  setting  .............................................................................................................................  5  1.4   Conceptual  framework  ............................................................................................................  6  1.5   Study  objectives  and  hypotheses  ...........................................................................................  8  1.6   Study  design  and  methods  ...................................................................................................  12  1.7   Summary  .................................................................................................................................  16  Chapter  2:  Leaving  hospital  against  medical  advice  among  people  who  use  illicit  drugs:  A  systematic  review  .....................................................................................................  19   vi 2.1   Introduction  ............................................................................................................................  19  2.2   Methods  ...................................................................................................................................  20  2.3   Results  ......................................................................................................................................  23  2.4   Discussion  ...............................................................................................................................  27  Chapter  3:  Factors  that  predict  leaving  hospital  against  medical  advice  among  people  who  use  illicit  drugs  in  Vancouver,  Canada  ........................................................................  36  3.1   Introduction  ............................................................................................................................  36  3.2   Methods  ...................................................................................................................................  37  3.3   Results  ......................................................................................................................................  41  3.4   Discussion  ...............................................................................................................................  42  Chapter  4:  Denial  of  pain  medication  by  healthcare  providers  predicts  in-­‐‑hospital  illicit  drug  use  among  people  who  use  illicit  drugs  ...........................................................  51  4.1   Introduction  ............................................................................................................................  51  4.2   Methods  ...................................................................................................................................  52  4.3   Results  ......................................................................................................................................  55  4.4   Discussion  ...............................................................................................................................  56  Chapter  5:  The  impact  of  an  HIV/AIDS  adult  integrated  health  program  on  leaving  hospital  against  medical  advice  among  HIV-­‐‑positive  people  who  use  illicit  drugs  ...  62  5.1   Introduction  ............................................................................................................................  62  5.2   Methods  ...................................................................................................................................  64  5.3   Results  ......................................................................................................................................  66   vii 5.4   Discussion  ...............................................................................................................................  67  Chapter  6:  Willingness  to  access  an  in-­‐‑hospital  supervised  injection  facility  among  people  who  use  illicit  drugs  ....................................................................................................  73  6.1   Introduction  ............................................................................................................................  73  6.2   Methods  ...................................................................................................................................  74  6.3   Results  ......................................................................................................................................  76  6.4   Discussion  ...............................................................................................................................  77  Chapter  7:  Conclusion  ..............................................................................................................  85  7.1   Summary  of  findings  .............................................................................................................  85  7.2   Study  strengths  and  unique  contributions  .........................................................................  88  7.3   Limitations  ..............................................................................................................................  91  7.4   Recommendations  ..................................................................................................................  93  7.5   Future  research  .......................................................................................................................  95  7.6   Conclusions  .............................................................................................................................  98  References  ...................................................................................................................................  99       viii List  of  Tables  Table  2.1  Summary  of  included  studies  in  systematic  review  (n  =  17)  ..............................  34  Table  3.1  Baseline  characteristics  of  people  who  use  illicit  drugs  stratified  by  leaving  hospital  against  medical  advice  (n  =  488)  ...............................................................................  47  Table  3.2  Crude  and  adjusted  longitudinal  estimates  of  the  odds  of  leaving  hospital  against  medical  advice  among  people  who  use  illicit  drugs  (n  =  488)  ...............................  49  Table  4.1  Bivariable  logistic  regression  analyses  of  factors  associated  with  having  ever  used  street  drugs  in  hospital  among  people  who  use  illicit  drugs  (n  =  1053)  ...................  60  Table  4.2  Multivariable  logistic  regression  of  factors  associated  with  ever  having  used  street  drugs  in  hospital  among  people  who  use  illicit  drugs  (n  =  1053)  ............................  61  Table  5.1  Baseline  characteristics  of  HIV-­‐‑positive  people  who  use  illicit  drugs  stratified  by  leaving  hospital  against  medical  advice  (n  =  181)............................................................  71  Table  5.2  Crude  and  adjusted  longitudinal  estimates  of  the  odds  of  leaving  hospital  against  medical  advice  among  HIV-­‐‑positive  people  who  use  illicit  drugs  (n  =  181)  .......  72  Table  6.1  Factors  associated  with  willingness  to  access  an  in-­‐‑hospital  supervised  injection  facility  among  people  who  use  illicit  drugs  in  Vancouver,  Canada  (n  =  732)  ..  82  Table  6.2  Multivariable  logistic  regression  analysis  of  factors  associated  with  willingness  to  access  an  in-­‐‑hospital  supervised  inhalation  facility  among  people  who  use  illicit  drugs  in  Vancouver,  Canada  (n  =  732)  ....................................................................................  84     ix List  of  Figures  Figure  1.1  Conceptual  framework  for  predictors  of  leaving  hospital  against  medical  advice  (objective  2) ..................................................................................................................... 18	  Figure  2.1  Flowchart  of  screening  and  article  selection  process ......................................... 33	       x List  of  Abbreviations  ACCESS   AIDS  Care  Cohort  to  evaluate  Exposure  to  Survival  Services  AIC      Akaike  information  criterion  AMA      Against  medical  advice  AOR      Adjusted  odds  ratio  BC      British  Columbia  CI      Confidence  interval  DPC      Dr.  Peter  Centre  DTES      Downtown  Eastside  of  Vancouver  DTP      Drug  Treatment  Programme  GEE      Generalized  estimating  equations  HCV      Hepatitis  C  virus  IQR      Interquartile  range  NSP      Needle  syringe  program  OR      Odds  ratio  PHN      Provincial  Health  Number    pVL      Plasma  viral  load  PWID     People  who  inject  drugs  PWUD   People  who  use  illicit  drugs  QIC      Quasi-­‐‑likelihood  under  the  independence  model  criterion  SIF      Supervised  injection  facility  VIDUS   Vancouver  Injection  Drug  Users  Study     xi Acknowledgements  I  am  sincerely  grateful  to  my  co-­‐‑supervisors,  Drs.  Thomas  Kerr  and  Jane  Buxton,  and   to   my   supervisory   committee   member,   Dr.   M-­‐‑J   Milloy,   for   their   unwavering  support  and  guidance  throughout  my  doctoral  training.  The  ongoing  mentorship  I  have  received  from  each  of  these  individuals  is  invaluable  and  I  am  very  appreciative  to  have  had  the  opportunity  to  be  guided  by  such  a  supportive  and  experienced  committee.  In  particular,  I  offer  my  extended  gratitude  to  Dr.  Thomas  Kerr,  who  has  been  involved  in  my  career  development   from   the  very  beginning.   I   could  not   thank  him  more   for   the  support  and  encouragement  he  has  shown  me  over   the  years.   I  also  benefited  greatly  from  the  support  and  guidance  that  Dr.  Evan  Wood  has  offered  to  me  as  an  invaluable  research  mentor.  None   of   the   work   contained   in   this   dissertation   would   have   been   possible  without   the   contributions  made   by   the   participants   in   the   Vancouver   Injection   Drug  Users  Study  and  the  AIDS  Care  Cohort  to  evaluate  Exposure  to  Survival  Services  study.  I  deeply  thank  all  who  were  courageous  enough  to  share  their  experiences.  I  hope  that  this  research  will  contribute  to  lowering  the  barriers  that  they  face  in  hospitals  so  that  they  can  benefit   from  life-­‐‑saving  care   they  may  receive   in   these  settings.   I  would  also  like  to  thank  all  past  and  present  study  interviewers,  field  staff,  study  nurses,  as  well  as  research  and  administrative  staff  at  the  BC  Centre  for  Excellence  in  HIV/AIDS  for  their  efforts,  as  well  as  the  study  co-­‐‑investigators  and  my  co-­‐‑authors.       I   feel   privileged   to   have   the   opportunity   to  work  with   a   remarkable   group   of  colleagues  and  fellow  graduate  students  at  the  University  of  British  Columbia  and  the   xii BC  Centre  for  Excellence  in  HIV/AIDS.  Specifically,  I  would  like  to  thank  Ms.  Michaela  Montaner,  Dr.  Kanna  Hayashi  and  Dr.  Lindsey  Richardson  for  providing  me  with  much  support,   advice   and   encouragement   through  my   graduate   studies.   I   look   forward   to  continuing  to  collaborate  with  and  learn  from  you  all.  Funding  to  support  my  doctoral  training  and  research  activities  was  generously  provided  by   the  Canadian   Institutes   for  Health  Research   through   a  Banting   and  Best  Canada   Graduate   Scholarship   and   the   University   of   British   Columbia   Doctoral  Fellowship.   I   also   thank   the   Urban   Health   Research   Initiative   of   the   BC   Centre   for  Excellence  in  HIV/AIDS  for  providing  salary  support  during  my  graduate  studies.  Lastly,  I  would  like  to  extend  my  utmost  gratitude  to  my  family  and  friends  for  their  support  and  patience  throughout  my  doctoral  studies.  I  am  deeply  grateful  to  my  sister  Lianlian  Ti  for  the  love  and  advice  she  has  provided  me  over  a  lifetime,  as  well  as  my  brother  Lianyi  Ti.  Finally  but  most   importantly,   I  would   like   to   thank  my  parents  Hataichanok  and  Teow  Choo  Ti   for   everything   they  have  done   for  me.  Without  your  endless  love  and  support,  I  would  not  have  had  all  the  opportunities  that  I  have  been  fortunate  to  have  in  my  life.  Thank  you.   xiii Dedication  To  my  parents,  Hataichanok  and  Teow  Choo  Ti   1 Chapter  1: Introduction  1.1 Hospitalization  among  people  who  use  illicit  drugs  People  who  use  illicit  drugs  (PWUD)  are  vulnerable  to  an  array  of  health-­‐‑related  harms   that   frequently   lead   to   an   overutilization   of   the   healthcare   system,   including  emergency   rooms   and   acute   hospital  wards.1,2   Previous   research   conducted   in  North  American  settings  have  indicated  that  while  injection-­‐‑related  soft  tissue  infections  (e.g.,  abscess,   cellulitis)   are   currently   the   most   common   causes   of   hospitalization   among  people  who  inject  drugs  (PWID),3,4  overdose  and  HIV/AIDS-­‐‑related  illnesses  continue  to  be  the  leading  causes  of  mortality  among  this  population.5,6  Of  note,  it  is  concerning  that  many   PWUD   are   known   to   delay   seeking   care   and   often   present   to   hospitals   at   late  stages  of  their  illnesses.7,8  As  a  consequence,  the  adverse  health  outcomes  (e.g.,  increase  in  severity  of  infectious  diseases,  progression  of  injection-­‐‑related  soft  tissue  infections  to  sepsis  or  a  more  advanced   form  of  bacterial   infection)   these   individuals  may  contend  with  will  likely  require  lengthy  and  costly  in-­‐‑patient  hospital  admissions.4,9,10  A   growing   body   of   literature   has   identified   that   while   hospitals   provide   much  needed  care  to  PWUD,  they  can  also  confer  risk  for  various  negative  health  outcomes  in  this   population.11   For   example,   past   research   has   shown   that   leaving   hospital   against  medical  advice  (AMA)  is  common  among  PWID,  with  prevalence  ranging  from  25%  to  30%.12–14  Studies  have  demonstrated  that  patients  who   leave  hospital  AMA  experience  elevated   rates   of   readmission   with   worsening   of   infections   or   other   comorbid   2 conditions.12,15–17   Population   level   data   also   indicate   the   high   risk   of  mortality   among  individuals  who  leave  hospital  AMA,17,18  with  one  study  showing  that  discharge  AMA  was   associated   with   over   twice   the   odds   of   30-­‐‑day   all-­‐‑cause   mortality   compared   to  those  who  had  planned  discharges.17  Moreover,  it  is  known  that  the  healthcare  system  is  burdened   with   higher   and   unnecessary   costs   associated   with   readmission,19   as   these  patients   are   likely   to   require   longer   length   of   hospital   stay   at   readmission.15   Other  research   has   also   identified   a   high   prevalence   of   high-­‐‑risk   illicit   drug   use   in   hospital  settings.  In  particular,  one  qualitative  study  indicated  that  some  PWID  resort  to  syringe  sharing  and  using  drugs  alone  in  locked  washrooms  as  a  way  to  conceal  their  drug  use  from   healthcare   providers,   thus   potentially   increasing   their   risk   of   infectious   disease  transmission  as  well  as  fatal  or  non-­‐‑fatal  overdose,  respectively.20  Penchansky  &  Thomas  (1981)  describe  five  dimensions  of  access  to  healthcare  that  encompasses   not   only   accessibility,   but   also   other   factors   that   touch   on   the   equity   of  healthcare,   and   includes:   the   availability,   accessibility,   accommodation,   affordability  and   acceptability   of   healthcare.21   While   many   developed   countries   operate   under   a  universal   healthcare   system   where   essential   medical   services   are   provided   free-­‐‑of-­‐‑charge   to   patients   (i.e.,   affordable   healthcare   services),   including   in   Canada,   PWUD  continue  to  face  numerous  barriers  to  hospital  care.22,23  While  these  barriers  are  thought  to  be  complex,  issues  around  accommodation  such  as  long  hospital  waiting  lists  and  the  suboptimal  management  of  active  addiction  and  pain  for  PWUD,24,25  as  well  as  concerns   3 regarding   acceptability   (e.g.,   negative   attitudes   of   healthcare   professionals   towards  PWUD),26,27   can   unfortunately   result   in   PWUD   having   negative   experiences   with   the  healthcare   system.22,28   Furthermore,   most   hospitals   operate   under   abstinence-­‐‑based  policies;   thus,   PWUD   have   minimal   access   to   drug-­‐‑using   paraphernalia   while  hospitalized,  making  it  difficult  for  these  individuals  to  safely  manage  their  active  drug  use.  As  mentioned  previously,   this  may   result   in  many  PWUD   resorting   to   high-­‐‑risk  drug-­‐‑using   practices   (e.g.,   syringe   sharing,   injecting   alone)   in   the   hospital   or   leaving  hospital   AMA,   which   may   ultimately   lead   to   further   adverse   health   outcomes,  including  mortality.11    1.2 Study  justification  Despite  the  growing  body  of  research  in  the  area  of  hospital  care  among  PWUD,  there  continues  to  be  major  gaps  in  the  scientific  literature  on  hospital  service  outcomes,  including  leaving  hospital  AMA,  among  this  population.  While  substantial  harms  and  costs   associated   with   leaving   hospital   prematurely   are   well   known,   such   as   being  readmitted   to   the   hospital  with   a  worsened  health   condition   or  mortality,17,18   persons  with  drug  dependence  are  difficult   to  follow  for  research  purposes,  and  therefore  few  longitudinal   studies   have   explored   hospital   service   outcomes   and   its   consequences  among   this   population.   Despite   the   substantial   burden   of   disease   attributable   to   the   4 negative   outcomes   that   PWUD   experience   in   hospital   care,   little   attention   has   been  devoted  to  exploring  potential  solutions  to  this  problem.  As   indicated   in  section  1.4,   there   is  a  growing  body  of   research   that  has  utilized  Rhodes’  Risk  Environment  Framework   to  better  understand   the   social,   structural   and  physical  environments  that  shape  health-­‐‑related  harms  among  PWUD.29–31  However,  to  date,   there   appears   to   be   no   known   studies   that   have   focused   on   identifying   these  contextual-­‐‑level  factors  and  how  they  may  interact  with  individual-­‐‑level  behaviours  to  shape   PWUD’s   risk   of   leaving   hospital   AMA   and   other   hospital-­‐‑related   outcomes.  Given  the  gaps  in  information  and  the  significant  human  suffering  and  healthcare  costs  associated   with   leaving   hospital   AMA,   there   is   now   a   pressing   need   to   better  understand  the  full  complexity  of   this  phenomenon  and  to   identify  effective  solutions  for  patients  who  use  illicit  drugs.    Consequently,   the   primary   research   aim   of   this   thesis   is   to   provide   scientific  evidence   to   further   improve   hospital   care   for   PWUD   by   first   identifying   various  individual-­‐‑   and   contextual-­‐‑level   factors   associated   with   leaving   hospital   AMA   and  related-­‐‑hospital  outcomes,  particularly  in-­‐‑hospital  illicit  drug  use.  Then,  I  will  use  these  findings   to   develop   and   evaluate   effective   approaches   to   minimizing   these   negative  outcomes   among   PWUD.   Findings   from   this   research   have   the   potential   to   produce  important  scientific  evidence  and  practical  policy  recommendations  that  may  minimize  the   health-­‐‑related   harms   that   PWUD   contend   with   in   hospital   settings,   and   thereby   5 improve   the   overall   quality   of   life   among   this   population.   In   addition,   these   findings  may  lead  to  more  efficient  use  of  valuable  healthcare  resources.    1.3 Study  setting  The  downtown  eastside  (DTES)  neighbourhood  of  Vancouver   is  one  of  Canada’s  poorest  neighbourhoods  and  for  decades  has  been  characterized  as  the  country’s  largest  open   illicit   drug   scene.32,33   This   neighbourhood   is   particularly   known   for   a   high  prevalence  of  drug  use,  poverty,  mental  illness  and  crime,  and  disproportionately  high  prevalence  of  drug-­‐‑  and  health-­‐‑related  harms.34–36  Beginning  in  the  1990s,  the  DTES  area  experienced  explosive  outbreaks  of  HIV  and  hepatitis  C  virus   (HCV)   infection  among  the  PWID  population,37,38  which   can   largely   be   attributed   to   the   lack   of   available   and  accessible   public   health   interventions   such   as   needle   syringe   programs   (NSPs).32,39  Concurrently,   evidence   of   high   levels   of   poly-­‐‑substance   use,   including   heroin   and  cocaine  injection,  as  well  as  a  high  incidence  of  adverse  health  outcomes  resulting  from  the  dramatic   rise   in   injection  drug  use   (e.g.,  overdose  and   injection-­‐‑related   infections)  has   been   documented   in   the   literature.4,38   As   a   result,   past   studies   have   consistently  shown   a   remarkable   increase   in   the   use   of   emergency   departments   and   acute   care  hospitals  locally.1,2,9,40  In   response   to   these   epidemics,   a   public   health   emergency   in   Vancouver   was  declared  in  1996  and,  among  other  efforts,  a  large,  centralized  needle  exchange  program   6 was  implemented  in  an  attempt  to  reduce  injection-­‐‑driven  HIV  and  HCV  incidence  as  well   as   drug-­‐‑related   harms   in   this   population.   In   the   following   two   decades,   the  province   of   British   Columbia   (BC)   has   made   extraordinary   progress   in   expanding  access  to  a  comprehensive  set  of  public  health  and  harm  reduction  programs,41  such  as  the   decentralization   and   expansion   of   NSPs,39   the   expansion   of   opioid   substitution  therapy,42,43   and   the   implementation   of   Insite,   North   America’s   first   sanctioned  supervised   injection   facility   (SIF).44   These   advancements,   alongside   readily   available  HIV-­‐‑related  care  at  no  cost  to  patients  provided  since  1986,  which  include  the  provision  of   antiretroviral   medications   to   all   HIV-­‐‑positive   individuals   in   BC,45   have   been  associated  with  significant  success  in  reducing  the  incidence  of  infectious  diseases  and  drug-­‐‑related   harms   and   improving   treatment   and   care   for   the   PWUD   population.   It  should  be  noted  that  beginning  in  1992,  the  BC  Centre  for  Excellence  in  HIV/AIDS  Drug  Treatment   Programme   (DTP)   has   been   delivering   centralized   HIV-­‐‑related   care,  including  antiretroviral  dispensation  and  clinical  monitoring,  for  the  entire  province  at  no  cost  to  patients.46,47    1.4 Conceptual  framework  This   research   draws   on   Rhodes’   Risk   Environment   Framework,   with   specific  modifications   to   conceptualize   the   possible   relationships   between   various   individual-­‐‑  and   contextual-­‐‑level   variables   and   hospital   service-­‐‑related   outcomes.29,48   In   the   past,   7 research  conducted  on   infectious  disease   risk  and  health  services  among  PWUD  have  been   largely   driven   by   individually-­‐‑focused   theories,   including   the   Health   Belief  Model49  and   the  Theory  of  Reasoned  Action.50  However,   these   theories   fail   to  account  for  the  full  variability  of  health  risk  among  this  population.29  In  contrast,  Rhodes’  Risk  Environment  Framework  suggests  that  the  interactions  among  macro-­‐‑,  meso  and  micro-­‐‑level  dimensions   of   social,   structural   and  physical   environments  have  been   shown   to  exert   influence   on   the   distribution   of   disease   and   other   health-­‐‑   and   drug-­‐‑related  outcomes   among   PWUD   populations.   For   example,   past   studies   have   shown   that  structural   factors   such   as   heavy   policing   practices   have   deterred   vulnerable  populations,  including  PWUD,  from  accessing  healthcare  services,  thus  increasing  their  risk   of   HIV   infection.51,52   As   well,   laws   that   prohibit   opioid   substitution   therapies,   a  known   protective   factor   for   HIV   transmission,53   can   undermine   HIV   prevention  initiatives  and  ultimately  increase  PWUD’s  HIV  vulnerability.31  To   date,   the   Risk   Environment   Framework   has   not   been   extensively   applied   to  hospital   service   outcomes.  Most   studies   of   hospitalization   among   PWUD   continue   to  rely   on   hospital   administrative   databases.   This   commonly   used   approach   has  limitations,  including  a  lack  of  consideration  of  time-­‐‑updated  behavioural  information,  and   a   lack   of   focus   on   how   broader   contextual   factors   may   be   influencing   hospital-­‐‑related  outcomes  among  PWUD.  There   is  a  growing  body  of   research   that   recognizes  that   research   on   individual   and   behavioural   factors   alone   are   insufficient   for   8 understanding   how   PWUD   access   healthcare   services.   Given   past   studies   that  demonstrated   that   access   to   healthcare   and   harm   reduction   services   are   indeed  dependent  on  the  environmental  context,52  this  research  will  be  pursued  by  employing  the   Risk   Environment   Framework   to   study   the   influence   of   socio-­‐‑demographic,  behavioural   and   individual-­‐‑level   exposure   to   social,   structural   and   physical  environment  factors  on  leaving  hospital  AMA  among  PWUD  in  Vancouver.  Figure  1.1  presents  the  conceptual  framework  that  was  employed  for  the  second  study  objective  of  this  research  project  (see  section  1.5).  Various  contextual  factors  will  be  examined,  with  a  focus  on  those  that  have  been  shown  to  shape  drug-­‐‑related  harms  among  PWUD  (e.g.,  difficulty   accessing   harm   reduction   and   healthcare   services   and   exposure   to   the  criminal   justice   system).54,55   In   all   analyses,   hospitals   will   be   considered   as   a   distinct  physical   environment.   By   encompassing   various   contextual   influences   as   potential  explanatory   factors,   this   dissertation   may   be   able   to   identify   factors   to   minimize  premature  hospital  discharge  that  may  be  more  amenable  to  intervention.    1.5 Study  objectives  and  hypotheses  The  aims  of  this  research  project  are  two  fold:  first,  this  research  seeks  to  examine  how   intersecting   individual   and   contextual   forces   shape   discharge   AMA   and   other  healthcare   experiences   in   hospital   settings   among   PWUD.   The   second   part   of   the  research  aims   to   identify  potential  novel   interventions   that   intend   to  minimize  harms   9 associated   with   the   high-­‐‑risk   environment   that   PWUD   contend   with   in   hospital  settings.  Specifically,  the  research  study  has  the  following  five  objectives  and  associated  research   hypotheses,   and   will   draw   largely   upon   data   from   two   longstanding   and  ongoing   community-­‐‑recruited   cohorts   of   HIV-­‐‑positive   and   HIV-­‐‑negative   PWUD   in  Vancouver:  1. To   systematically   assess   the   literature   examining   hospital   discharge   AMA  among   this   population.   Chapter   2   provides   the   results   of   this   systematic  review  examining  the  current  literature  around  the  prevalence  and  predictors  of   leaving   hospital   AMA   among   this   population,   as   well   as   potential  interventions  to  minimize  this  phenomenon.  Specific  objectives  and  hypotheses  were  generated  for  analyses  in  Chapters  3  to  6  following  the  findings  from  this  review.  2. To   identify   the   prevalence   of   leaving   hospital   AMA   among   hospitalized  PWUD  and  to  characterize  factors  associated  with  leaving  hospital  AMA  in  this  population.  A  limited  number  of  studies  have  addressed   the  risk   factors  associated   with   leaving   hospital   AMA   specifically   among   the   PWUD  population.12,13  Using  quantitative  data  collected  through  two  parallel  ongoing  prospective  cohort  studies  of  PWUD,  which  as  part  of  this  thesis,  was  linked  to  a   hospital   administrative   database,   Chapter   3   analyzes   how   individual-­‐‑   and  select   contextual-­‐‑level   factors   shape   hospital   discharge   AMA.   It   is   10 hypothesized   that  high   frequency  drug  use,  a  history  of  negative  experiences  with  healthcare  professionals  and  exposure  to  the  criminal   justice  system  will  be   associated  with   an   increased   odds   of   leaving   hospital   AMA,  while   stable  employment  and  access  to  methadone  maintenance  therapy  will  be  negatively  associated  with  the  outcome.  3. To   examine   the   effect   of   being   denied   pain   medication   by   healthcare  providers  on  in-­‐‑hospital  illicit  drug  use  among  hospitalized  PWUD.  Studies  have  shown  that  undertreated  pain  is  common  among  PWUD,11  and  can  often  reflect   the   reluctance   of   healthcare   providers   to   provide   pain   medication   to  individuals  with   concurrent   substance   use   disorders.56,57   However,   there   is   a  limited   body   of   evidence   that   has   explored   being   denied   pain   medication  among  PWUD  in  acute  care  settings.  To  better  understand  the  experiences  of  PWUD   in   hospital   care   that   shape   their   hospital   utilization   and   retention  outcomes   in   this   setting,   the   study   presented   in   Chapter   4   will   test   the  hypothesis  that  being  denied  pain  medication  is  independently  associated  with  a  higher  odds  of  in-­‐‑hospital  illicit  drug  use.  Taken  together  with  the  findings  in  Chapter   3,   it   is   hoped   that   these   findings  will   identify   barriers   to   utilization  and   retention   in   hospital   care   and  may   serve   to   inform   the   development   of  interventions  that  aim  to  improve  care  among  PWUD  in  this  setting.   11 4. To   assess   the   role   of   an   HIV/AIDS   adult   integrated   health   program   on  leaving   hospital   AMA   among   HIV-­‐‑positive   PWUD.   Alternative   models   of  care  have  been  implemented  in  various  settings  that  aim  to  minimize  risks  and  improve  overall  health  outcomes  for  PWUD.  The  Dr.  Peter  Centre  (DPC)  is  an  established   specialty  HIV/AIDS-­‐‑focused  adult   integrated  health  program   that  provides  support  to  some  of  the  city’s  most  vulnerable  citizens  and  operates  in  proximity  to  a  hospital.58  As  part  of  their  program,  the  DPC  also  operates  a  SIF,  which   involves   the   supervision   of   illicit   drug   injection   by   nurses.59   Given  evidence   linking  HIV   to   an   elevated   risk   of   hospitalization   among   PWUD,60  and   studies   showing   that   severely   addicted   PWUD   account   for   a   large  proportion  of  all  discharges  AMA  from  an  HIV/AIDS  ward,12   the  objective  of  this  study  is  to  assess  the  impact  of  an  innovative  HIV/AIDS  adult   integrated  health   program   on   leaving   hospital   AMA   among   HIV-­‐‑positive   PWUD.   It   is  hypothesized   that   DPC   participants  will   be   associated  with   a   lower   odds   of  leaving  hospital  AMA  compared  to  non-­‐‑DPC  participants.  5. To  assess   the  prevalence  of  willingness   to   access   an   in-­‐‑hospital  SIF  and   to  examine   the   characteristics   of   PWUD   who   are   willing   to   access   such   a  facility.  An   extensive   body   of   literature   has   documented   the   effectiveness   of  SIFs   on   improving   social-­‐‑   and   health-­‐‑related   outcomes   among   PWID   in  Vancouver.61–64   However,   SIFs   that   are   available   in   the   community   are   12 currently  not  available   in  hospital   settings.  Using  a  hypothetical  scenario,   the  analyses  presented  in  Chapter  6  will  test  the  hypothesis  that  various  high-­‐‑risk  behaviours,   including   frequent   drug   use,   leaving   hospital   AMA,   and   using  illicit  drugs  in-­‐‑hospital  will  be  positively  associated  with  willingness  to  use  an  in-­‐‑hospital  SIF.    1.6 Study  design  and  methods  The  analyses  presented  in  Chapters  3  to  6  of  this  thesis  rely  on  data  derived  from  two   long-­‐‑running  prospective  cohort  studies  of  HIV-­‐‑negative  PWID  and  HIV-­‐‑positive  PWUD  in  Vancouver.  The  analytic  approaches  for  each  chapter  differ  from  each  other,  but   to  avoid  duplication,   elements  of   the   study  methodology   that   are   common   to   the  chapters  are  described  here.  The  Vancouver  Injection  Drug  Users  Study  (VIDUS)  and  the  AIDS  Care  Cohort  to  evaluate   Exposure   to   Survival   Services   (ACCESS)   studies   are   two   open   prospective  cohort  studies  of  PWUD  who  have  been  recruited  through  self-­‐‑referral,  street  outreach  and  word-­‐‑of-­‐‑mouth  since  May  1996.  To  be  eligible  for  enrollment,  participants  had  to  be  at  least  18  years  of  age  and  reside  in  the  greater  Vancouver  region.  The  recruitment  and  follow-­‐‑up  procedures   for   the   two  studies  are   largely   identical,   allowing   for   combined  analyses,  with   the   only   difference   being   that  HIV-­‐‑positive   individuals  who   use   illicit  drugs   other   than   cannabis   in   the  month  prior   to   enrollment   are   followed   in  ACCESS   13 whereas  HIV-­‐‑negative  individuals  who  injected  drugs  in  the  month  prior  to  enrollment  are  followed  in  VIDUS.  Moreover,   individuals  who  seroconvert   following  recruitment  are  transferred  from  the  VIDUS  study  into  the  ACCESS  study.    After   individuals  are  screened  for   inclusion  and  offered  enrollment   in   the  study,  they  complete  an  informed  consenting  process  in  a  private  and  secure  room  at  the  study  office.   During   this   process,   the   participant   and   study   representative   review   a  standardized  letter  explaining  the  study,  the  associated  protocols  and  possible  risks  and  benefits   of  participation.  Only   individuals  who  provide  written   informed   consent   are  included  in  the  study.  At  baseline  and  semi-­‐‑annually  thereafter,  participants  complete  a  harmonized   interviewer-­‐‑administered   questionnaire   and   provide   blood   samples   for  HIV   (for   HIV-­‐‑negative   participants   only)   and   HCV   testing,   as   well   as   HIV   disease  monitoring   (for   HIV-­‐‑positive   participants   only).   During   the   one-­‐‑on-­‐‑one   interview,  participants   provide   confidential   answers   to   questions   pertaining   to   various   socio-­‐‑demographic  characteristics,  drug  use  patterns  and  behaviours,  experiences  with  health  and   harm   reduction   services,   as  well   as   related   exposures.   Participants  were   given   a  stipend  ($20  CDN)  at  each  study  visit  for  their  time  and  transportation.  In  June  2013,  the  stipend  amount  increased  from  $20  CDN  to  $30  CDN.    For   the   analyses   presented   in  Chapter   3   and   5,   the  VIDUS   and  ACCESS   cohort  data  were  augmented  by  confidential  linkages  to  sources  containing  records  of  hospital  admissions   and   discharges   at   St.   Paul’s   Hospital   in   Vancouver.   A   Provincial   Health   14 Number   (PHN),   a   unique   and   persistent   identifier   issued   for   billing   and   tracking  purposes   for   all   residents   of   BC   by   the   medical   system,   enabled   these   confidential  linkages.  Specifically,  the  hospital  health  records  and  discharge  database  from  St.  Paul’s  Hospital   records   all   admissions,   diagnoses,   and   discharge   information,   and   are  routinely  linked  to  the  cohort  data.  To  enable  the  linkage,  study  participants’  PHNs  and  date  of  birth  (DOB)  were  given  to  a  data  analyst  at  St.  Paul’s  Hospital  in  an  encrypted  and  secure  file.  The  data  analyst  at  St.  Paul’s  Hospital  extracted  all  hospital  data  records  for  PHN-­‐‑  and  DOB-­‐‑matched  participants  who  were  hospitalized  at  St.  Paul’s  Hospital  between   2001   and   2011.   For   confidentiality   purposes,   full   names   of   patients   were  removed  from  the  received  file.  To  merge  the  cohort  data  to  the  hospital  data,  a  number  of  rules  were  created:  first,  the  candidate  calculated  the  absolute  difference  (in  days)  between  the  cohort  interview  date  and   the  hospital  discharge  date,  and  merged   the  data   from  the  closest   follow-­‐‑up  interview   date   to   the   hospital   discharge   record;   second,   if   the   absolute   difference  obtained  was  within  180  days,  then  the  data  from  that  cohort  observation  was  kept  in  the  dataset;  however,  if  the  absolute  difference  obtained  was  greater  than  180  days,  then  the   data   from   that   cohort   observation   was   set   to   missing.   A   six-­‐‑month   period   was  chosen   given   that   the   PWUD   population   is   dynamic   with   regard   to   their   drug   use  patterns  and  risk  behaviours,  and  therefore  the  candidate  wanted  to  match  the  hospital  discharge  database   to   the  variables   that  were  most   representative  of   their  exposure  at   15 that  time  to  account  for  this.  Moreover,  if  a  hospital  discharge  occurred  greater  than  180  days  before  participants  were  enrolled  in  the  study,  then  the  corresponding  cohort  data  was   set   to  missing   in   the  merged  dataset,   as   participants   could   not   have   answered   a  survey  questionnaire  prior  to  their  enrollment  in  the  cohort  studies.  In  total,  ten  years  of  data  (2001-­‐‑2011)  have  been  merged  for  this  thesis.     For  Chapter  5,  the  ACCESS  data  was  additionally  augmented  by  comprehensive  information   on   HIV   care   and   treatment   outcomes   from   the   DTP,   a   local   centralized  HIV/AIDS  registry,  at  the  BC  Centre  for  Excellence  in  HIV/AIDS.  As  mentioned  above,  the   BC   Centre   for   Excellence   in   HIV/AIDS   is   the   provincial   body   responsible   for  providing  HIV  treatment  and  care  for  all  HIV-­‐‑positive  individuals  in  BC.  Specifically,  a  complete  clinical  profile  of  all  CD4  T-­‐‑cell  counts,  HIV-­‐‑1  RNA  plasma  viral   load  (pVL)  observations   and   exposure   to   specific   antiretroviral   agents   for   each   participant   are  routinely  obtained.  Similarly  to  the  hospital  data,  participants’  HIV  clinical  monitoring  data  was   linked   to   the   cohort   data   using   each   participants’   PHN.   All   pVL   and   CD4  measurements   conducted   through   the   study   as   well   as   any   pVL   and   CD4   tests  conducted   outside   of   the   study   (e.g.,   in   community   and   correctional   settings)   were  obtained.   In   the   event   that   more   than   one   pVL   or   CD4   cell   count   observation   was  conducted   in   the  prior  six  months,   the  median  of  all  observations  were   included.  The  Roche   Amplicor   Monitor   assay   was   used   to   determine   plasma   viral   load   from  participant  blood  samples  (Roche  Molecular  Systems,  Pleasanton,  California).     16 The   University   of   British   Columbia/Providence   Health   Care’s   Research   Ethics  Board  has  approved  the  VIDUS  and  ACCESS  studies,  and  linkage  to  St.  Paul’s  Hospital  database   and   the   DTP   at   the   BC   Centre   for   Excellence   in   HIV/AIDS,   as   well   as   this  doctoral  research.    1.7 Summary  In   summary,   this  dissertation  contains   seven  chapters.  This   introductory  chapter  provides  an  overview  of  the  barriers  associated  with  hospital  utilization  and  retention  in   care   among   PWUD,   as   well   as   an   overview   of   this   dissertation’s   objectives,  hypotheses  and  methods.  Chapter  2  is  a  systematic  review  of  the  literature  focused  on  leaving   hospital   AMA   among   PWUD,   focusing   on   PWUD   both   as   an   explanatory  variable  of  interest  and  PWUD  as  a  population  of  interest.  Chapters  3  to  6  present  the  quantitative  empirical  analyses  of   this  dissertation.  The  empirical  studies  presented  in  Chapters   3   and   4   explore   healthcare   experiences   in   hospital   settings   among   PWUD  patients.   Specifically,   informed   by   the   Risk   Environment   Framework,   Chapter   3  explores   various   individual   and   select   contextual   factors   associated   with   leaving  hospital  AMA.  Chapter  4   tests   the  hypothesis   that  being  denied  pain  medication  by  a  healthcare  provider   is   independently  associated  with  an   increased  odds  of   illicit  drug  use  in  hospital.  Chapters  5  and  6  investigate  potential  programs  and  interventions  that  may   improve   hospital   care   for   PWUD.   Specifically   in  Chapter   5,   an   evaluation   of   an   17 integrated   HIV/AIDS   health   program   operating   in   proximity   to   a   hospital,   which  includes  a  SIF,  and  its  impact  on  leaving  that  hospital  AMA  was  conducted.  Chapter  6  explores  the  characteristics  of  PWUD  who  are  willing  to  use  an  in-­‐‑hospital  SIF.  Finally,  Chapter   7   summarizes   and   synthesizes   the   key   findings   of   this   dissertation.   The  implications   of   the   study   findings,   methodological   limitations,   and   future  recommendations  are  also  discussed  in  the  final  chapter.     18   Figure  1.1  Conceptual  framework  for  predictors  of  leaving  hospital  against  medical  advice  (objective  2)    Hospital-related Outcomes • Hospital discharge against medical advice Individual Factors • Age • Gender • Aboriginal ancestry • HIV serostatus • Drug use patterns  Social-structural Factors • Relationship status • Incarceration • Access to stable employment • Access to methadone maintenance therapy • Access to harm reduction services (i.e., access to drug-using paraphernalia) • Poor treatment by healthcare providers  Environmental Factors • Hospital setting • Housing status • Drug scene exposure • Sex work scene exposure Individual factors interact with select social, structural and environmental factors to facilitate pathways to hospital-related outcomes. Adapted from the Risk Environment Framework (Rhodes, 2002).   19 Chapter  2: Leaving  hospital  against  medical  advice  among  people  who  use  illicit  drugs:  A  systematic  review  2.1 Introduction  As   described   in   section   1.1,   leaving   hospital   AMA   is   an   increasing   problem   in  acute  care  settings  and  is  linked  to  an  array  of  negative  health  consequences,  including  increased  morbidity,  hospital  readmission  and  mortality.17,18,65  Studies  have  also  shown  that  leaving  hospital  AMA  introduces  a  huge  financial  burden  on  the  healthcare  system,  as   these   individuals   fail   to  make   a   full   recovery   the   first   time   they   are   treated  which  necessitates  readmission  to  a  hospital.19  Of  concern,  the  existing  literature  on  this  subject  has  consistently  shown  that  substance  misuse  is  related  to  an  increased  risk  of  leaving  hospital  AMA;17,19  however,  due  to  the  complex  issues  associated  with  treating  PWUD  for  various   comorbidities,   including  addiction,  mental  health,   infectious  diseases,   and  drug-­‐‑related  adverse  events,  the  reason  for  this  association  remains  unclear.  In  addition,  programs   and   interventions   for   reducing   the   rate   of   leaving   hospital   prematurely  among  PWUD  in  acute  care  settings  have  not  been  well  studied.  Therefore,  the  objective  of   this   study   was   to   systematically   assess   the   existing   literature   examining   the  prevalence  and  predictors  of   leaving  acute  care  hospital  AMA  among  this  population,  as  well  as  potential  interventions  that  aim  to  minimize  this  phenomenon.       20 2.2 Methods  The   candidate   referred   to   the   Preferred  Reporting   Items   for   Systematic  Reviews  and  Meta-­‐‑Analyses  (PRISMA)  guidelines  for  the  development  of  systematic  reviews  in  the  present  study.66    2.2.1 Search  strategy  The   candidate   searched   the   following   five   electronic   databases   to   identify  relevant  studies  published  in  peer-­‐‑reviewed  journals  from  database  inception  to  August  2014:   CINAHL,   EMBASE,   Medline,   PsychInfo   and   Web   of   Science.   Search   terms  included:   “patient   discharge”,   “hospital   discharge”,   “against   medical   advice”,   “drug  user”,   “substance-­‐‑related   disorders”   and   “intravenous   substance   abuse”.   The   terms  were   mapped   to   database-­‐‑specific   subject   headings   and   controlled   vocabulary   terms  when   available.   Where   possible,   methodological   filters   were   used   to   exclude   case  reports   and   case   series.   The   candidate   hand   searched   reference   lists   of   published  reviews  and  included  studies  which  were  relevant.  The  candidate  restricted  the  search  to   English-­‐‑language   publications   but   did   not   restrict   it   with   respect   to   year   of  publication.     21 2.2.2 Inclusion  and  exclusion  criteria  Studies   were   eligible   for   inclusion   if   they   were   published   in   a   peer-­‐‑reviewed  journal  as  an  original  research  article.  Grey  literature,  case  reports,  case  series,  reviews  and  editorials  were  excluded.  The  criteria  for  considering  studies  for  this  review  were  organized   using   the   PICO   (population,   intervention/exposure,   comparison,   outcome)  framework.  The   candidate   retained  original   studies   that   reported   illicit   drug  use   as   a  predictor  of  leaving  hospital  AMA  and  studies  of  discharge  AMA  that  included  PWUD  as  a  population  of  interest.  Studies  had  to  include  analyses  of  factors  associated  with  the  outcome  of   interest,  with   significance   assessed   through  appropriate   statistical   tests   or  the  estimation  of  effect  measures  and  confidence  intervals  (CIs),  in  order  to  be  included  in  the  analysis.  Given  the  candidate’s  interest  in  patients  who  were  hospitalized  in  acute  care   settings,   studies   that   reported   leaving   hospital   AMA   from   psychiatric   hospitals,  drug  treatment  centres  and  emergency  departments  were  excluded.  While  there  is  some  clinical   overlap   between   acute   medical   patients   and   patients   in   psychiatric   or   drug  treatment  centres,  the  data  are  generally  too  heterogeneous;  thus,  making  it  difficult  to  compare   these   populations.  Moreover,   the   candidate  was   interested   in   acute   diseases  and  patients  in  psychiatric  or  drug  treatment  centres  are  generally  admitted  for  mental  illness   and   drug   dependence,   respectively,  which  was   considered   chronic   conditions.  The  candidate  excluded  patients   in  emergency  departments  given  the  dynamic  nature  and  high  turnover  of  patients   in   this  setting,  which   is  vastly  different   from  acute  care   22 settings.  While  patients  are  able  to  stay  in  the  emergency  department  for  treatment  for  considerable  periods   of   time  without   transferring   to   acute   care,   length   of   stay   can   be  highly  variable  in  these  settings  and  therefore  may  be  subject  to  heterogeneity  as  well.  Studies  were  also  excluded  if  they  discussed  misuse  of  alcohol  but  not  of  illicit  drugs.    2.2.3 Study  selection  and  data  collection  process  The   candidate   conducted   the   database   search   and   entered   study   abstracts  matching  the  keywords  and  search  criteria  into  an  electronic  database.  After  removing  duplicates,   the   candidate   and   a   second   reviewer   independently   reviewed   potentially  eligible  articles  by  scanning  the  titles,  abstracts  and  full  texts  of  articles.  At  each  review  stage,   studies   clearly   not   meeting   the   inclusion   criteria   were   excluded   from   further  review.  The  candidate  and  reviewer  identified  studies  for  which  eligibility  was  unclear  and  a  final  decision  to  include  or  exclude  such  studies  were  made  after  thorough  review  by  both  investigators.  Data  were  extracted  for  each  eligible  record  using  a  standardized  form  which   was   developed   based   on   a   review   of   previous   published   peer-­‐‑reviewed  systematic  reviews  that  included  information  on  study  design,  setting,  sample  size  and  characteristics,   and   major   relevant   findings.   This   information   was   entered   by   the  candidate   and   reviewed   independently   for   accuracy   and   completeness   by   the   second  reviewer.     23 2.3 Results  2.3.1 Literature  search  Database  and  hand  searching  yielded  a  total  of  1,649  potentially  eligible  studies.  After  initial  title  and  abstract  screenings,  1,627  studies  were  removed  from  the  analysis.  Additionally,   five   studies   were   excluded   following   an   assessment   of   the   full   text  articles.  In  three  of  these  five  articles,  the  setting  criteria  was  not  met:  Beck  et  al.’s  study  was  conducted  in  community  mental  health  centres,  Green  et  al.’s  study  was  conducted  in   an   inpatient   detoxification/dual   diagnosis   unit,   and   Levine   et   al.’s   study   was  conducted  in  a  drug  treatment  hospital.  In  two  of  these  five  articles,67,68  the  studies  were  not  replicable  due  to  poor  description  in  the  methods  section  and  thus  were  excluded  from  the  final  analysis.  Consistency  was  high  between  investigator  searches,  with  only  two  differences  in  study  eligibility  initially  reviewed  by  both  investigators.  In  total,  17  studies  published  between  1977  and  2014  met  the  eligibility  criteria  and  were  included  in  the  final  synthesis.  Included  studies  are  presented  in  Table  2.1.    2.3.2 Summary  of  included  studies  Of  the  17  included  studies,  all  but  one  (in  Australia)  were  conducted  in  Canada  (five   in  Vancouver)  or   the  United  States.   In  total,   three  studies  were  conducted  before  2000.  Most  studies  employed  retrospective  study  designs  (n  =  13),  while  few  used  case  control  designs  (n  =  2),  cross-­‐‑sectional  designs  (n  =  1)  or  mixed  methods  designs  (n  =  1).   24 The  majority   of   studies  were   conducted   among  general  medical   inpatients   (n   =   9).   In  addition,  three  studies  were  conducted  specifically  among  PWID  (including  one  study  conducted   among   HIV-­‐‑positive   PWID   patients),   two   studies   were   conducted   among  patients   with   pneumonia   (including   one   study   conducted   among   HIV-­‐‑positive  pneumonia  patients),  one  study  was  conducted  among  HIV-­‐‑positive  patients,  one  study  was   conducted   among   patients  with   cirrhosis,   and   one   study  was   conducted   among  female   post-­‐‑partum  patients.  All   studies   relied   on  hospital   administrative   records   for  their  outcome  measure  of  leaving  hospital  AMA.  All  studies  relied  on  patient  medical  records  to  define  their  substance  misuse  variable.  Substance  misuse  was  inconsistently  defined  across   studies,   including   such  definitions   as   injection  drug  use   status   (n   =   6);  drug  abuse  or  dependence  (n  =  5);  alcohol  and/or  drug  abuse  or  dependence  (n  =  5);  and  mental  illness,  including  drug  abuse,  alcohol  dependence,  depression  and  psychosis  (n  =  1).      2.3.3 Substance  misuse  predicts  leaving  hospital  AMA  Of   the   five   studies   that   included   drug   abuse   or   dependence   as   a   predictor   of  leaving  hospital  AMA,   four   studies   reported  drug  abuse   as   a   significant   and  positive  correlate   of   leaving   hospital   AMA.   These   studies   were   conducted   among   general  hospital   inpatients,69   postpartum   female   patients,70   patients   with   pneumonia71   and  patients  with  cirrhosis.72  One  study  conducted  in  Boston,  Massachusetts  concluded  that   25 drug   addiction   was   a   significant   correlate   of   leaving   hospital   AMA;   however,   a   test  statistic  to  determine  the  strength  of  the  association  could  not  be  calculated  as  one  cell  contained  zero  counts.73    Similarly,   all   five   studies   that   examined   alcohol   and/or   drug   abuse   or  dependence  as  a  predictor  of  leaving  hospital  AMA  among  general  hospital  inpatients  found  it  to  be  a  significant  correlate  of  the  outcome  of  leaving  hospital  AMA.17,19,74–76  It  is  noteworthy  that  the  authors  of  these  five  studies  failed  to  account  for  clustering  among  patients   for  multiple   admissions   and   discharges   and   as   a   result,   these   analyses  may  have  falsely  concluded  a  significant  relationship  between  the  explanatory  variable  and  leaving   hospital   AMA.   In   a   United   States   national   sample   of   adult   patients  who   left  hospitals  between  1988  and  2006,  one  study  found  a  positive  and  statistically  significant  association   between   mental   illness   diagnosis—which   includes   drug   abuse,   alcohol  dependence,   depression   and   psychosis—and   leaving   hospital   AMA.   However,   when  patients  with  substance  abuse  diagnoses  were  removed  from  the  mental  illness  variable,  the  adjusted  odds  ratio  was  attenuated  from  5.1  to  3.3,  although  the  estimate  was  still  statistically  significant  in  multivariable  analyses.77  Three   studies   included   injection   drug   use   status   as   an   explanatory   variable   of  interest   in   their   analyses   examining   potential   correlates   of   leaving   hospital   AMA.   A  retrospective   matched   cohort   study   conducted   in   Vancouver   concluded   that   PWID  were   more   likely   to   leave   hospital   AMA   (54%)   compared   to   their   non-­‐‑drug-­‐‑using   26 counterparts  (23%).78  Similarly,  Anis  et  al.  (2002)  reported  that  injection  drug  use  was  a  significant  predictor  of   leaving  hospital  AMA  among  HIV-­‐‑positive  patients.15  Lastly,  a  study   conducted   by   Palepu   et   al.   (2003)   found   no   statistically   significant   association  between   injection   drug   use   and   leaving   hospital  AMA  on   early   hospital   readmission  among  HIV-­‐‑positive  pneumonia  patients.79    2.3.4 Prevalence  and  predictors  of  leaving  hospital  AMA  among  PWID  Three  studies  assessed  the  prevalence  and  various  predictors  of  leaving  hospital  AMA   among   PWID   in   Vancouver.   Leaving   hospital   AMA   was   commonly   reported  among   PWID   in   the   three   studies,   ranging   from   25%   to   30%.12–14   Analysis   of   a  retrospective  cohort  study  found  that  among  HIV-­‐‑positive  PWID,  recent  injection  drug  use,  Aboriginal  ancestry,  being  in  a  hospital  over  a  weekend  or  social  assistance  cheque  day  were  positive  correlates  of  leaving  hospital  AMA,  while  in-­‐‑hospital  methadone  use,  social  support  and  older  age  were  negatively  associated  with  this  outcome.12  In  BC,  the  government   issues   social   assistance   cheques   to   eligible   recipients  monthly   on   the   last  Wednesday  of  each  month.  Similarly,  Riddell  &  Riddell  (2006)  indicated  that  there  was  a  16%  increase  in  the  probability  of  leaving  hospital  AMA  on  Wednesdays  when  social  assistance   cheques  were   administered   in   comparison   to   any   other  Wednesday.14   In   a  retrospective   study  of   PWID,   Jafari   et   al.   (2014)  demonstrated   that   the   risk   of   leaving  hospital   AMA   was   significantly   lower   among   clients   admitted   to   the   Community   27 Transitional   Care   Team   (CTCT;   a   community   care   model   of   intravenous   antibiotic  therapy  for  PWID  with  deep  tissue  infection)  compared  to  those  who  were  admitted  to  the  hospital.13  While  this  particular  study  focuses  on  a  community  model  of  care,  it  was  included   in   the  analysis  given   that   the  comparison  group   to  patients  of   the  CTCT  are  patients  of  an  acute  care  hospital.    2.4 Discussion  The  present   systematic   review   found   that   there  was   consistency   in   the   evidence  presented   by   eligible   studies,   which   was   a   positive   association   between   substance  misuse  and  leaving  hospital  AMA  among  patients  in  acute  care.  A  few  studies  explored  predictors   of   leaving  hospital  AMA  among  PWID   specifically,   and   found   that   factors  such   as   recent   injection   drug   use   and   being   in   the   hospital   during   the   monthly  distribution   of   social   assistance   cheques   were   positively   associated   with   this  outcome.12,14   Moreover,   the   results   of   the   review   suggest   that   various   mitigating  factors—including  in-­‐‑hospital  methadone  use,  social  support  and  a  community  model  of  care  to  treat  PWID  with  deep  tissue  infections—have  the  potential  to  reduce  the  rate  of  leaving  hospital  prematurely.    As   mentioned   previously,   a   small   number   of   studies   investigated   various  predictors   of   leaving   hospital   AMA   specifically   among   PWID.   However,   one   of   the  major   drawbacks   of   this   data   is   that   it   fails   to   include   any   measurement   of   social,   28 structural   and   physical   environmental   factors   that   might   account   for   some   of   the  explained   variability   in   the   effect   of   leaving   hospital   AMA.   For   instance,   previous  studies  have  shown  that  having  negative  interactions  with  healthcare  providers,  a  social  environmental   factor,  was   associated  with   poor   utilization   of   healthcare   services   and  retention  in  care  among  PWUD.20,26  Indeed,  a  qualitative  study  that  was  not  included  in  the  present  systematic  review  documented  instances  of  both  voluntary  and  involuntary  discharges   AMA   due   to   negative   cultural   stereotypes   in   healthcare   settings.20  Furthermore,  a  growing  body  of  research  has  suggested  that  individuals  who  have  been  exposed   to   the   criminal   justice   system,   a   structural   environmental   factor   known   to  shape   numerous   health-­‐‑   and   drug-­‐‑related   outcomes,   were   less   likely   to   utilize  healthcare  services.80,81  Future  studies  should  address   the   limitations  of  administrative  data   by   linking   routinely   collected   behavioural   data   with   local   hospital   and   health  service  data  sources,  as  well  as   include  information  on  exposure  to  various  contextual  factors,  in  an  effort  to  provide  a  better  understanding  of  the  individual  and  contextual  factors  that  influence  leaving  hospital  AMA  among  PWUD.  The   paucity   of   evidence   regarding   potential   interventions   to   minimize   hospital  discharge  AMA  among  PWUD  warrants  some  attention.  Given  the  frequent  occurrence  of  discharge  AMA  among  PWUD,  it  is  surprising  that  only  two  studies  offer  potential  solutions   for   reducing   the   prevalence   of   premature   hospital   discharge   for   this  population.   The   present   systematic   review   is   consistent   with   a   narrative   review   that   29 concluded   a   lack   of   evidence   on   how   to   effectively   reduce   the   prevalence   of   leaving  hospital  AMA   among   PWUD  patients   in   acute   care.67  Nevertheless,   it   is   encouraging  that  one   study   included   in   the  present   review  observed  a   significant   reduction   in   the  rate  of   leaving  AMA  among  patients   in   the  CTCT  program  compared   to   the  hospital,  though  the  length  of  stay  in  the  hospital  ranged  from  16  to  22  days  compared  to  50  to  90  days   at   the   CTCT   for   the   same   clients.13   The   fact   that   the   CTCT   program   has   better  retention   outcomes   is   a   particularly   valuable   finding   given   that   PWID   are   most  commonly   hospitalized   for   cutaneous   injection-­‐‑related   infections   (e.g.,   cellulitis,  abscess,   osteomyelitis).4,9   Community-­‐‑based   programs   may   complement   traditional  acute  care  in  an  effort  to  minimize  premature  hospital  discharge.  Also  of  interest  is  the  finding   that   the   availability   and   use   of   in-­‐‑hospital   methadone   reduced   hospital  discharge  AMA  among  HIV-­‐‑positive  PWID.12  The  provision  of  methadone   in  hospital  settings  may  address   some  of   the   issues  around  underlying  addictive  behaviours  and  the  wish  to  acquire  more  drugs.71  However,  it  should  be  noted  that  opioid  substitution  therapy   is   only   appropriate   for   opioid-­‐‑dependent   PWUD   and   may   not   apply   to  individuals  who  are  dependent  on  other   illicit   substances,   in  particular   stimulants.   In  addition,   past   studies   have   shown   that   many   physicians   do   not   prescribe   adequate  doses  of  methadone  to  hospitalized  patients  with  opioid  dependence,82,83  and  as  a  result  of  opioid  withdrawal   from  poor  management  of  addiction  and  pain,24,82  many  PWUD  may  leave  hospital  AMA.   30 While  beyond  the  scope  of   the  present  systematic   review,   there   is  a  much   larger  body   of   scientific   evidence   around   AMA   discharges   from   psychiatric   hospitals   and  drug   treatment   centres.   Given   the   overlap   in   patient   characteristics   between   these  settings  and  acute   care  hospitals,   it  may  be  of   interest   to  draw  upon   this   literature   to  gain  insights  into  strategies  for  preventing  premature  hospital  discharge.  For  instance,  one   effective   strategy   consistently   reported   in   the   psychiatric   literature   is   the  involvement   of   a   patient   advocate   or   psychiatric   consultant   who   can   proactively  identify  and  address  issues  of  substance  misuse  during  hospital  admission.84,85  Another  strategy   for   preventing   a   discharge   AMA   that   was   noted   in   the   literature   included  various   motivational   interviewing   techniques   to   help   patients   make   positive   and  achievable   changes   to   overcome   their   drug   dependence.67   By   doing   so,   it   may   be  possible  for  physicians  to  understand  reasons  why  PWUD  leave  hospital  AMA  and  to  address  this  early  on  following  hospital  admission.  Limitations  common  to  many  of  the  included  studies  should  be  stated  in  order  to  better   contextualize   the   findings.   First,  most   of   the  published  data  on   this   subject   are  from   retrospective   analyses   and   case-­‐‑control   studies,   which   may   limit   the   ability   to  establish   temporality   or   infer   any   causal   relationships   between   the   explanatory  variables   and   outcome   variable   of   interest.   Future   research   of   higher  methodological  quality   is   required   to   better   understand   the   complex   nature   of   substance  misuse   and  leaving   hospital   AMA,   and   to   identify   potential   methods   for   minimizing   this   31 phenomenon.  However,  it  is  noteworthy  that  given  the  unethical  nature  of  randomizing  patients  to  leave  and  not  leave  hospital  AMA,  future  studies  will  likely  be  restricted  to  observational  designs  (e.g.,  prospective  cohort  studies)  and  thus,  will  retain  the  threat  of  residual  and  unmeasured  confounding.  Second,  many  of  these  studies  were  analyzed  at  the  hospital   admission/discharge   level,  yet   clustering  among   individual  patients  were  not  taken  into  account.  As  a  result,  these  studies  may  have  been  more  likely  to  falsely  detect  a  significant  difference  that  may  have  biased  their  findings.  Lastly,  as  a  limitation  of  the  review,  it  is  possible  that  some  eligible  studies  in  the  search  strategy  were  missed.  Specifically,   the   analysis   was   restricted   to   acute   care   hospitals.   While   the   candidate  chose   to   include   Janfari   et   al.’s   study   because   the   authors   used  patients   in   acute   care  hospitals   as   their   comparison   group   to   patients   in   the   CTCT   program,   the   present  systematic   review   may   have   missed   similar   non-­‐‑acute   hospital   care   initiatives.   The  candidate  also  recognizes  that  the  selection  and  qualitative  synthesis  of  eligible  studies  is   a   subjective   process.   However,   the   candidate   sought   to   minimize   this   bias   by  duplicating  the  search  and  using  two  investigators  to  conduct  the  screening  procedure  independently.  In   summary,   the   systematic   review   revealed   that   there   exists   little   scientific  evidence  on  risk  factors  associated  with  leaving  hospital  AMA  among  PWUD.  To  better  understand  this  phenomenon,  there  is  a  need  for  future  research  to  consider  the  effect  of  individual   and   contextual   characteristics   on   leaving   hospital   AMA   among   this   32 population.  Ascertaining   time-­‐‑updated   behavioural   information   on  drug  use   patterns  and  risk  behaviours,  as  well  as  various  social,  structural  and  environmental  exposures,  may  be  a  way  to  address  this  issue.  Furthermore,  given  the  limited  number  of  studies  exploring   interventions   to   reduce   the   rate   of   leaving   hospital   prematurely   among  PWUD,   the   development   and   evaluation   of   novel   methods   to   address   this   issue   is  needed.     33 Figure  2.1  Flowchart  of  screening  and  article  selection  process     34 Table  2.1  Summary  of  included  studies  in  systematic  review  (n  =  17)     35     36 Chapter  3: Factors  that  predict  leaving  hospital  against  medical  advice  among  people  who  use  illicit  drugs  in  Vancouver,  Canada  3.1 Introduction  As   described   in   section   1.1,   leaving   hospital  AMA   remains   a  major   healthcare  challenge   that   often   leads   to   an   array   of   negative   health   consequences   and   related  costs15,17–19,67,   particularly   among   patients   with   a   history   of   substance   use.15,19,67   For  example,  previous  research  conducted  in  Vancouver  indicated  that  almost  a  third  of  a  sample   of   PWID   admitted   to   an   inner   city   tertiary   care   hospital   were   discharged  AMA.12,15  Other  studies  have  documented  that  PWID  were  two  to  four  times  more  likely  to   be   discharged   AMA   compared   to   their   non-­‐‑PWID   counterparts.12,15   From   a  population   health   perspective,   this   is   concerning   given   the   known   health   risks  associated  with  illicit  drug  use  and  leaving  hospital  AMA.7,86,87  Despite   the   growing   body   of   research   dedicated   to   exploring   the   complex  phenomena   of   leaving   hospital   AMA,   the   systematic   review   presented   in   Chapter   2  found   that   there   are   few   studies   focused   specifically   on   PWUD   populations.  Additionally,   the   review   concluded   that   existing   research   has   relied   heavily   on  retrospective   data   collected   from   hospital   records   and   noted   that   this   approach   has  limitations,  including  a  lack  of  focus  on  how  various  more  proximal  individual,  as  well  as  broader  social,  structural  and  environmental   factors  may  influence   leaving  hospital  AMA  among   this   population.   To   address   these   gaps   in   the   literature,   this   study  was   37 undertaken   to   examine  various   individual   and   select   contextual   factors   that   influence  leaving  hospital  AMA  among  PWUD  in  Vancouver,  with  a  focus  on  those  factors  that  have   been   shown   to   shape   drug-­‐‑related   harms   among   this   population   (e.g.,  incarceration).   This   setting,   with   its   universal   healthcare   system,   is   unique   in   that   it  allows   for   the   investigation   of   factors   associated   with   these   outcomes   without   the  confounding  effect  of  explicit  financial  barriers  to  essential  medical  care.  It  is  hoped  that  the  findings  from  this  study  may  serve  to  inform  the  development  of  interventions  that  aim  to  reduce  the  rate  of  PWUD  leaving  hospital  prematurely.   3.2 Methods  Data  for  these  analyses  were  ascertained  from  the  VIDUS  and  ACCESS  studies,  as  well  as  St.  Paul’s  hospital  records,  as  described  in  section  1.6.    3.2.1 Study  sample  The   present   study  was   restricted   to   participants  who   experienced   at   least   one  hospitalization  at  St.  Paul’s  Hospital  between  2005  and  2011.  Although  the  hospital  and  cohort  data  were  linked  and  merged  between  2001  and  2011  (described  in  section  1.6),  the   candidate   restricted   the   sample   from   2005   to   2011   given   that   the   measures   for  important   explanatory   variables   were   only   available   for   this   period   (e.g.,   negative  experiences  with  healthcare  professionals).  Despite  past   research   focusing  on  patients   38 with   a   history   of   injection   drug   use   and   leaving   hospital   AMA,   the   candidate  purposefully   included   injection   and   non-­‐‑injection   drug   users   in   the   study   given   that  PWUD  who  do  not   inject  are  also  vulnerable   to  an  array  of  health  harms   that   require  hospitalization.87,88  Moreover,  as   in   the  case  of  PWID,   the  candidate  hypothesized  that  PWUD  who   do   not   inject   are   also   at   risk   of   leaving   hospital   AMA.   For   the   present  study,  PWUD  will  henceforth  be  used  to  denote  both  injectors  and  non-­‐‑injectors.    3.2.2 Variable  selection  The   primary   outcome   of   interest   for   this   analysis   was   leaving   hospital   AMA,  which   was   obtained   from   the   St.   Paul’s   Hospital   administrative   admission   and  discharge   database.   Determined   by   hospital   policies,   participants   were   defined   as  having  left  hospital  AMA  if  they  left  the  hospital  without  physician  permission  and  did  not   return  within   six   hours   (e.g.,   if   a   participant   left   the   hospital   and   returned   seven  hours  later  they  were  considered  as  having  left  the  hospital  AMA  and  their  subsequent  admission  would  be  considered  a  new  admission  to  the  hospital).  Participants  were  also  designated   as   having   left   hospital  AMA   if   they   did   not   return   to   hospital  within   six  hours   after   obtaining   physician   permission   to   leave   on   a   pass   (unless   the   physicians  authorized   a   longer   period).   A   range   of   socio-­‐‑demographic,   behavioural   and  individual-­‐‑level   exposures   to   social   and   structural   factors   that   may   potentially   be  associated   with   leaving   hospital   AMA   were   explored,   including:   age   (per   year   39 younger);   gender   (male  vs.   female);   reported  Aboriginal   ancestry   (yes  vs.   no)   12,15;   lab  confirmed   HIV   serostatus   (positive   vs.   negative);   in   a   stable   relationship,   defined   as  either   legally   married/common   law   or   having   a   regular   partner   (yes   vs.   no);  homelessness,  such  as  living  on  the  street  or  having  no  fixed  address  (yes  vs.  no);  daily  heroin   injection   (yes   vs.   no);   daily   cocaine   injection   (yes   vs.   no);   daily   crystal  methamphetamine  injection  (yes  vs.  no);  daily  prescription  opioid  injection,  defined  as  injection   of   either   OxyNeo,   OxyContin,   Percocet,   Tylenol   3,   Morphine,   Dilaudid,  Demerol,   Methadone,   Fentanyl,   Hydrocodone,   or   Talwin   (yes   vs.   no);   daily   crack  cocaine   non-­‐‑injection   (yes   vs.   no);   binge   drug   use,   defined   as   having   injected   drugs  and/or   used   non-­‐‑injection   drugs   “more   than   usual”   (e.g.,   if   they   usually   inject   drugs  three  times  a  day  but  had  injected  five  times  on  a  particular  day;  yes  vs.  no);  need  help  injecting  (yes  vs.  no);  stable  employment,  defined  as  having  a  regular  job,  temporary  job  or   self   employed   (yes   vs.   no);   drug   dealing   (yes   vs.   no);   sex   work   (yes   vs.   no);  incarceration   (yes  vs.  no);  enrollment   in  methadone  maintenance   therapy  (yes  vs.  no);  difficulty  finding  drug  use  paraphernalia  in  any  setting,  including  syringes,  needles  and  crack   cocaine   pipes   (yes   vs.   no);   and   ever   had   negative   experiences   with   healthcare  professionals,   including  being  poorly  treated  by  a  healthcare  professional  (yes  vs.  no).  All  variables  refer  to  the  previous  six  months  unless  otherwise  indicated.     40 3.2.3 Statistical  analyses  Bivariable   and  multivariable   statistics   were   used   to   identify   factors   associated  with   leaving   hospital   AMA.   Since   analyses   of   factors   potentially   associated  with   the  outcome   of   interest   included   serial   measures   for   each   subject,   the   candidate   used  generalized   estimating   equations   (GEE)   for   binary   outcomes   with   logit   link   for   the  analysis  of  correlated  data  to  determine  factors  associated  with  leaving  hospital  AMA.  These  methods  provided  standard  errors  adjusted  by  multiple  observations  per  person  using   an   exchangeable   correlation   structure.   Therefore,   data   from   every   participant  follow-­‐‑up  visit  were  considered  in  this  analysis.  As  a  first  step,  GEE  bivariable  analyses  were   conducted   to  obtain  unadjusted  odds   ratios   and  95%CI   for  variables  of   interest.  Then,  a  multivariable  model  was  fit  using  an  a  priori-­‐‑defined  statistical  protocol  based  on  examination  of  the  Quasi-­‐‑likelihood  under  the  independence  model  criterion  (QIC)  for   GEE   and   p-­‐‑values.89   First,   a   preliminary   model   was   constructed   including   all  variables  significant  in  bivariable  analyses  at  p  <  0.10.  Following  this,  each  variable  with  the  highest  p-­‐‑value  was  removed  sequentially,  with  the  final  model  including  the  set  of  variables   associated   with   the   lowest   QIC.   The   severity   of   multicollinearity   was   also  quantified  using  the  variance  inflation  factor.  All  p-­‐‑values  are  two-­‐‑sided.    41 3.3 Results     Between  September  2005  and  July  2011,  a  total  of  488  PWUD  had  experienced  at  least  one  hospitalization  and  were  included  in  the  study:  211  (43.2%)  were  female  and  the  median   age   at   baseline  was   44   years   (interquartile   range:   38   –   50   years).   In   total,  there  were   1,176   hospital   admissions   among   these   individuals,   and   during   the   study  period,   participants   were   admitted   to   the   hospital   between   1   and   30   times   per  participant.  Among  these  participants,  212  (43.4%)  unique  individuals  left  the  hospital  AMA  at  least  once  during  the  study  period.  Table  3.1  shows  the  baseline  characteristics  of  the  study  sample  stratified  by  leaving  hospital  AMA  at  baseline.     As   indicated   in   Table   3.2,   in   bivariable   analyses,   factors   significantly   and  positively   associated   with   leaving   hospital   AMA   included:   younger   age   (odds   ratio  [OR]  =  1.04;  95%  confidence  interval  [CI]:  1.03  –  1.06);  homelessness  (OR  =  1.51;  95%CI:  1.10  –   2.08);  daily  heroin   injection   (OR  =  1.76;   95%CI:   1.24  –   2.48);  daily   crack   cocaine  non-­‐‑injection   (OR   =   1.43;   95%CI:   1.09   –   1.89);   drug   dealing   (OR   =   1.53;   95%CI:   1.10   –  2.12);  and  incarceration  (OR  =  1.90;  95%CI:  1.26  –  2.85).  In  addition,  stable  employment  (OR  =  0.56;  95%CI:  0.36  –  0.88)  and  enrollment  in  MMT  (OR  =  0.73;  95%CI:  0.55  –  0.98)  were  negatively  associated  with  the  outcome.    Table  3.2  also  presents  multivariable  analyses  of  factors  associated  with  leaving  hospital  AMA.  Variables  that  remained  significantly  and  positively  associated  with  the  outcome  included:  younger  age  (adjusted  odds  ratio  [AOR]  =  1.04;  95%CI:  1.02  –  1.06);   42 daily  heroin  injection  (AOR  =  1.49;  95%CI:  1.05  –  2.11);  and  incarceration  (AOR  =  1.63;  95%CI:   1.07   –   2.49),   while   stable   employment   (AOR   =   0.52;   95%CI:   0.33   =   0.82)   was  negatively   associated   with   the   outcome.   The   variance   inflation   factor   estimate  calculated  revealed  no  significant  multicollinearity  in  the  final  model.   3.4 Discussion  In  the  present  study,  it  was  found  that  a  substantial  proportion  of  PWUD  in  the  sample   had   left   the   hospital   AMA   at   some   point   during   the   study   period   and   that  various  markers  of  risk  and  vulnerability  were  associated  with  this  phenomenon.   It   is  concerning  that  hospital  discharge  AMA  occurs  frequently  among  this  population  given  the  known  health-­‐‑related  harms  and  costs  linked  to  such  events,  which  are  described  in  detail  in  section  1.1.17  The  development  and  implementation  of  novel  interventions  that  aim   to   minimize   leaving   hospital   prematurely   among   this   population   is   needed   to  better  understand  the  challenges  that  PWUD  contend  with  in  hospital  settings  that  lead  them  to  discharge  from  hospital  AMA.    Findings   from   the   present   study   revealed   that   frequent   heroin   injection   was  positively   associated   with   leaving   hospital   AMA.   Given   the   unavailability   of   illicit  opioids   in   hospital   settings,   frequent   heroin   injectors   may   face   severe   withdrawal  symptoms   and   as   a   result,   may   discharge   themselves   from   hospital   AMA   to   obtain  heroin   to   maintain   their   drug   addiction.   The   higher   odds   of   leaving   hospital   AMA   43 among   PWUD   who   are   frequent   heroin   injectors   could   also   be   explained   by   the  complexities   around   treating   opioid-­‐‑dependent   patients   for   pain.24,82,90   In   particular,  studies  have  documented   the  provision  of   inadequate  pain  management  among   these  individuals,24,57  which  may  ultimately  contribute  to  an  increased  risk  of  leaving  hospital  AMA  to  self-­‐‑manage  their  pain.91     Previous   studies   have   indicated   that   PWUD   who   have   been   exposed   to   the  criminal  justice  system  are  less  likely  to  access  and  utilize  healthcare  services.80,81,92  The  findings   from   this   study   add   to   the   existing   literature   by   revealing   a   positive  relationship  between  recent  incarceration  and  leaving  hospital  AMA.  This  may  likely  be  due  to  the  negative  experiences  PWUD  contend  with  in  correctional  settings,  including  poor  relations  between  inmates  and  prison  health  staff  that  may  create  distrust  towards  authority  among  those  previously  incarcerated.93  As  well,  PWUD  who  were  previously  incarcerated  may   have   experienced  withdrawal   during   their   time   in   prison   and   thus  may  be  more  likely  to  leave  hospital  AMA  to  maintain  their  drug  use.  Given  the  limited  evidence  concerning  this  dynamic,  future  research  should  seek  to  untangle  the  precise  causal   relationships   underlying   these   associations.   Regardless,   a   number   of  interventions  that  seek  to  build  and  improve  patient-­‐‑physician  relationships  should  be  considered,   including   educating   physicians   on   how   to   effectively   identify   and  appropriately   address   the   needs   of   PWUD   patients.   In   addition,   specialized   training  among  healthcare  providers  and  changes  in  institutional  attitudes  towards  PWUD  that   44 may   help   to   remove   the   stigma   and   discrimination   PWUD   sometimes   experience   in  hospitals  should  be  a  priority.    The  present  study  found  that  having  stable  employment  was  a  protective  factor  independently  associated  with  leaving  hospital  AMA.  One  explanation  for  this  finding  may   be   that   being   employed   is   a   marker   of   greater   stability   and   lower   intensity  addiction,94,95   though   further   research   is   needed   to   tease   out   these   relationships.  Nevertheless,   previous   studies   have   demonstrated   a   strong   link   between   unstable  employment   through  street-­‐‑based   income  generating  activities   (e.g.,  drug  dealing,   sex  work,  scavenging)  and  illicit  drug  use  and  dependence.94,96  Another  explanation  for  this  finding  may  be  that,  among  individuals  with  unstable  employment  and  who  engage  in  street-­‐‑based  income  generation  activities,  being  hospitalized  and  absent  from  the  drug  scene  could  mean  not  being  able   to  secure   funds  for  drugs  and  the   loss  of  a   job.  As  a  result,   people   with   unstable   employment   are   more   likely   to   leave   hospital   AMA   to  maintain  both  their  job  and  their  addiction.  However,  it  is  noteworthy  that  despite  the  fact   that   having   stable   employment   has   been   shown   to   be   associated   with   various  positive   outcomes   for   PWUD,97   there   are   a   number   of   barriers   towards   obtaining  employment  and  studies  have  consistently  shown  low  levels  of  employment  among  this  population.95,98,99    The  limitations  common  to  all  studies  are  presented  in  section  7.3;  however,  the  limitations  specific  to  this  chapter  are  presented  here.  First,  while  diagnostic  measures   45 at   discharge   were   collected   for   every   patient   admitted   to   acute   hospital   care,   it   was  decided  by  the  candidate  to  exclude  diagnosis  as  a  potential  explanatory  variable  given  that   PWUD   often   experience   an   array   of   health-­‐‑related   issues   that  may   lead   to   their  hospitalization.  As  a  result,  the  primary  diagnosis  recorded  at  discharge  may  not  have  accurately   captured   their  underlying   comorbidities,   and   these  may  have   instead  been  captured  as  their  secondary  or  tertiary  diagnoses.  Given  that  it  was  difficult  to  tease  out  the   primary   cause   for   being   admitted   into   the   hospital,   the   inclusion   of   this   type   of  measure   may   have   introduced   unpredictable   bias   into   the   study.   Therefore,   future  research  that  is  able  to  more  effectively  identify  a  primary  diagnosis  for  PWUD  patients  with   potentially   several   comorbidities   should   be   considered.   Second,   this   research  project   included   only   one   hospital.   This   limitation   would   have   resulted   in   an  underestimation  of  hospital  use,  as  some  participants  may  have  accessed  other  hospitals  during   the   study   period,   especially   those   who   have   had   negative   experiences   at   St.  Paul’s  Hospital.  However,  it  is  noteworthy  that  St.  Paul’s  Hospital  is  located  in  the  heart  of  downtown  Vancouver  and  is  known  to  service  the  majority  of  PWUD  in  this  setting.     In  sum,  the  study  found  that  a  substantial  proportion  of  PWUD  in  the  sample  left  hospital  AMA.  Younger   individuals,   those  who  inject  heroin  with  high  frequency  and  those  who  were  previously   incarcerated  were  more   likely   to  have   left   hospital  AMA.  Having   stable   employment   was   a   protective   factor   for   leaving   hospital   AMA.   These  findings  highlight  the  need  to  address  substance  abuse  issues  soon  following  admission   46 to  hospital.  These  findings  also  suggest  a  need  to  develop  novel  interventions  to  reduce  the  rate  of  PWUD  leaving  hospital  prematurely.        47 Table  3.1  Baseline  characteristics  of  people  who  use  illicit  drugs  stratified  by  leaving  hospital  against  medical  advice  (n  =  488)  Leaving hospital against medical advice   Characteristic Yes n (%) n = 156 No n (%) n = 332 Odds Ratio (95% CI) p - value Age     median 41 45 1.04 (1.02 – 1.07) <0.01 IQR (35 – 48) (39 – 51)   Gender     male  78 (50.0) 199 (59.9) 0.67 (0.46 – 0.98) 0.04 female 78 (50.0) 133 (40.1)   Aboriginal ancestry     yes 65 (41.7) 112 (33.7) 1.40 (0.95 – 2.07) 0.09 no 91 (58.3) 220 (66.3)   HIV serostatus     positive 93 (59.6) 191 (57.5) 1.09 (0.74 – 1.60) 0.66 negative 63 (40.4) 141 (42.5)   In a stable relationship*     yes 41 (26.3) 93 (28.0) 0.92 (0.60 – 1.41) 0.69 no 115 (73.7) 239 (72.0)   Homelessness*     yes 63 (40.4) 100 (30.1) 1.57 (1.06 – 2.34) 0.03 no 93 (59.6) 232 (69.9)   Daily heroin injection*     yes 42 (26.9) 60 (18.1) 1.67 (1.06 – 2.62) 0.02 no 114 (73.1) 272 (81.9)   Daily cocaine injection*     yes 18 (11.5) 26 (7.8) 1.54 (0.82 – 2.89) 0.18 no 138 (88.5) 306 (92.2)   Daily crystal methamphetamine injection*    yes 7 (4.5) 8 (2.4) 1.90 (0.68 – 5.34) 0.22 no 149 (95.5) 324 (97.6)   Daily prescription opioid injection*    yes 8 (5.1) 16 (4.8) 1.07 (0.45 – 2.55) 0.88 no 148 (94.9) 316 (95.2)   Daily crack cocaine non-injection*    yes 84 (53.8) 124 (37.3) 1.96 (1.33 – 2.88) <0.01 no 72 (46.2) 208 (62.7)   Binge drug use*     yes 63 (40.4) 144 (43.4) 0.88 (0.60 – 1.30) 0.53 no 93 (59.6) 188 (56.6)      48  Leaving hospital against medical advice   Characteristic Yes n (%) n = 156 No n (%) n = 332 Odds Ratio (95% CI) p - value Need help injecting*     yes 29 (18.6) 58 (17.5) 1.08 (0.66 – 1.77) 0.76 no 127 (81.4) 274 (82.5)   Stable employment*     yes 16 (10.3) 55 (16.6) 0.58 (0.32 – 1.04) 0.07 no 140 (89.7) 277 (83.4)   Drug dealing*     yes 61 (39.1) 89 (26.8) 1.75 (1.17 – 2.62) <0.01 no 95 (60.9) 243 (73.2)   Sex work*     yes 25 (16.0) 44 (13.3) 1.25 (0.73 – 2.13) 0.41 no 131 (84.0) 288 (86.7)   Incarceration*     yes 35 (22.4) 36 (10.8) 2.38 (1.43 – 3.96) <0.01 no 121 (77.6) 296 (89.2)   Enrollment in MMT*     yes 64 (41.0) 155 (46.7) 0.79 (0.54 – 1.17) 0.24 no 92 (59.0) 177 (53.3)   Difficulty finding drug use paraphernalia*   yes 35 (22.4) 78 (23.5) 0.94 (0.60 – 1.48) 0.80 no 121 (77.6) 254 (76.5)   Negative experiences with healthcare professionals   yes 25 (16.0) 51 (15.4) 1.05 (0.62 – 1.77) 0.85 no 131 (84.0) 281 (84.6)   CI: confidence interval; IQR: interquartile range; MMT: methadone maintenance therapy *Refers to the six-month period prior to follow-up     49 Table 3.2 Crude and adjusted longitudinal estimates of the odds of leaving hospital against medical advice among people who use illicit drugs (n = 488)  Unadjusted  Adjusted Characteristic Odds Ratio (95% CI) p - value  Odds Ratio (95% CI) p - value Age      (per year younger) 1.04 (1.03 – 1.06) <0.01  1.04 (1.02 – 1.06) <0.01 Gender      (male vs. female) 0.76 (0.56 – 1.04) 0.09    Aboriginal ancestry      (yes vs. no) 1.27 (0.92 – 1.75) 0.15    HIV serostatus      (positive vs. negative) 0.99 (0.72 – 1.37) 0.95    In a stable relationship*      (yes vs. no) 0.93 (0.69 – 1.26) 0.65    Homelessness*      (yes vs. no) 1.51 (1.10 – 2.08) 0.01    Daily heroin injection*      (yes vs. no) 1.76 (1.24 – 2.48) <0.01  1.49 (1.05 – 2.11) 0.03 Daily cocaine injection*      (yes vs. no) 1.07 (0.65 – 1.74) 0.80    Daily crystal methamphetamine injection*     (yes vs. no) 1.87 (0.85 – 4.15) 0.12    Daily prescription opioid injection*     (yes vs. no) 1.19 (0.63 – 2.26) 0.60    Daily crack cocaine non-injection*     (yes vs. no) 1.43 (1.09 – 1.89) 0.01    Binge drug use*      (yes vs. no) 0.87 (0.66 – 1.14) 0.31    Need help injecting*      (yes vs. no) 1.35 (0.94 – 1.93) 0.11    Stable employment*      (yes vs. no) 0.56 (0.36 – 0.88) 0.01  0.52 (0.33 – 0.82) <0.01 Drug dealing*      (yes vs. no) 1.53 (1.10 – 2.12) 0.01    Sex work*      (yes vs. no) 1.46 (0.98 – 2.18) 0.06    Incarceration*      (yes vs. no) 1.90 (1.26 – 2.85) <0.01  1.63 (1.07 – 2.49) 0.02 Enrollment in MMT*      (yes vs. no) 0.73 (0.55 – 0.98) 0.04       50  Unadjusted   Adjusted  Characteristic Odds Ratio (95% CI) p - value  Odds Ratio (95% CI) p - value Difficulty finding drug use paraphernalia*     (yes vs. no) 1.12 (0.82 – 1.54) 0.47    Negative experiences with healthcare professionals    (yes vs. no) 1.21 (0.84 – 1.75) 0.30    GEE: generalized estimating equations; CI: confidence interval; MMT: methadone maintenance therapy *Refers to the six-month period prior to follow-up          51 Chapter  4: Denial  of  pain  medication  by  healthcare  providers  predicts  in-­‐‑hospital  illicit  drug  use  among  people  who  use  illicit  drugs  4.1 Introduction  As   specified   in   section   1.1,   PWUD   are   vulnerable   to   an   array   of   health-­‐‑related  harms  that  often  lead  to  an  overreliance  on  emergency  rooms  and  acute  hospital  wards  as  a  regular  source  of  care.1,2  Despite  the  large  health  burden  associated  with  illicit  drug  use,   various  barriers  have  been  known   to   impede   access,   utilization,   and   retention   in  care  among  this  population.22,27,100  In  particular,  undertreated  pain  is  a  common  concern  among   PWUD,   which   may   reflect   the   challenges   that   healthcare   providers   face   in  providing  adequate  pain  medication  to  individuals  with  addictions.101,102  Pain  may  also  be   a   consequence   of   the   condition   that   PWUD  present   to   the   hospital  with.  As  well,  appropriate   treatment  of  pain  among  PWUD  is  often  complex  due  to   the  concomitant  use  of  opioid  substitution  therapies,  comorbidities,  and  the  lack  of  clear  guidelines  for  pain  management  among  this  population.103  A   growing   body   of   literature   has   shown   that   in   Vancouver,   a   substantial  proportion   of   the   illicit   drug-­‐‑using   population   are   hospitalized   annually.9,40   It   is  noteworthy  that  the  high  rates  of  hospitalization  among  active  PWUD  has  resulted  in  a  well  recognized  local  drug  market  where  patients  can  obtain  illicit  drugs  for  injection  or  inhalation   around   hospital   premises.104   Given   the   limited   body   of   evidence   that   has  explored   experiences   with   accessing   pain   medication   among   PWUD   in   acute   care   52 settings,  the  present  study  sought  to  quantitatively  examine  the  impact  of  being  denied  pain  medication  by  any  healthcare  provider  ever  on  the  use  of  illicit  drugs  in  hospitals  ever   while   admitted   to   hospital.   The   findings   from   this   study   may   be   useful   for  informing  public  health  and  clinical  practice  that  aim  to  minimize  the  harms  associated  with  high-­‐‑risk  illicit  drug-­‐‑use  in  hospital.    4.2 Methods  Data  for  these  analyses  were  ascertained  from  the  VIDUS  and  ACCESS  studies,  as  described  in  section  1.6.    4.2.1 Study  sample  The   present   study   was   restricted   to   participants   who   experienced   at   least   one  hospitalization  in  their  lifetime.      4.2.2 Variable  selection  The  primary  outcome  of  interest  for  this  analysis  was  having  ever  reported  using  illicit   drugs   in   hospital.   To   differentiate   between   illicit   drug   use   and   licit   drug   use,  participants   were   asked:   “Did   you   ever   use   street   drugs   while   in   hospital?”   The  primary  explanatory  variable  of  interest  was  having  ever  been  denied  pain  medication  by   a   healthcare   provider,   ascertained   by   asking   participants   the   following   questions:   53 “Have   you   ever   requested   a   prescription   for   pain  medication?   If   yes,  were   you   ever  refused  a  prescription?”  Secondary  variables  believed  to  be  confounders  included:  age  (per   year   increase);   gender   (male   vs.   female);   daily   injection  drug  use,   defined   as   the  cumulative   proportion   of   reported   daily   injection   drug   use   in   the   past   six   months  during   the   cohort   study  period   (≥   50%  of   the   time  vs.   <   50%  of   the   time);   daily   non-­‐‑injection  drug  use,  defined  as  the  cumulative  proportion  of  reported  daily  non-­‐‑injection  drug  use  in  the  past  six  months  during  the  cohort  study  period  (≥  50%  of  the  time  vs.  <  50%  of  the  time);  and  binge  drug  use,  defined  as  the  cumulative  proportion  of  reported  binge  drug  use  by   injection  or  non-­‐‑injection   in   the  past   six  months  during   the   cohort  study  period  (≥  50%  of  the  time  vs.  <  50%  of  the  time).  As  the  present  study  is  a  cross-­‐‑sectional  study  drawn  from  a  prospective  cohort  study,  the  latter  three  variables  were  derived   from   longitudinal   data   beginning   from   the   participant’s   initial   date   of  enrollment  to  the  current  study  period.  For  instance,  if  one  participant  completed  a  total  of   five   follow-­‐‑up   questionnaires   in   addition   to   their   baseline   questionnaire,   and  reported  at  least  daily  injection  drug  use  in  the  past  six  months  in  four  of  the  six  follow-­‐‑up  questionnaires,  then  their  cumulative  proportion  of  reported  daily  injection  drug  use  in  the  past  six  months  during  the  cohort  study  period  would  be  67%  (4/6  =  0.67);  thus,  they  would  fall   into  the   ‘>50%  of   the  time’  category  for   that  particular  variable.  These  measures  hoped  to  address   the   issue  of   temporality   in   the  present  analysis  given   that  the  outcome  variable  was  a  lifetime  measure  of  illicit  drug  use  in  hospital.   54   4.2.3 Statistical  analyses  Bivariable   analyses  were   conducted   to   determine   factors   associated  with   having  ever   reported   using   illicit   drugs   in   hospital   using   simple   logistic   regression.   To   fit   a  multivariable  logistic  regression  model,  the  candidate  first  employed  a  p-­‐‑value  cutoff  of  p  <  0.10  to  determine  which  variables  were  potentially  associated  with  having  ever  used  illicit  drugs  in  hospital  in  the  simple  logistic  regression  analyses  described  above.  Then  the  candidate  fit  a  full  model  including  these  explanatory  variables,  noting  the  value  of  the  coefficient  associated  with  having  ever  been  denied  pain  medication.  Given  the  lack  of  scientific  evidence  that  explored  being  denied  pain  medication  and  illicit  drug  use  in  hospital,  the  candidate  was  unable  to  draw  from  a  list  of  confounding  variables  known  to   be   associated   with   both   the   main   explanatory   and   outcome   variables   of   interest.  Therefore,   the   candidate   decided   that   a   stepwise   approach   to   including   confounders  was   the   most   appropriate   and   conservative   method   of   analysis.   Specifically,   the  secondary   explanatory   variable   corresponding   to   the   smallest   relative   change   in   the  effect  of  having  ever  been  denied  pain  medication  on  having  ever  used  illicit  drugs  in  hospital  was  removed  from  further  consideration.  This  iterative  process  continued  until  the  minimum  change  of   the  value  of   the  coefficient   for  having  ever  been  denied  pain  medication   from   the   full   model   exceeded   5%.   Remaining   variables   were   considered  confounders  in  multivariable  analysis.  All  p-­‐‑values  were  two-­‐‑sided.   55   4.3 Results  Among   all   the   participants  who  were   interviewed   between  December   2012   and  May  2013,  a  total  of  1053  (96%)  PWUD  had  experienced  at  least  one  hospitalization  and  were   included  in   the  study:  341  (32.4%)  were  female,  509  (48.3%)  had  completed  high  school   and   the   median   age   was   48   years   (IQR:   42   –   54   years).   In   terms   of   age  distribution  by  gender,  males  were  slightly  older  than  women,  with  a  median  age  of  50  years  (IQR:  43  –  55  years)  among  males  compared  with  a  median  age  of  46  years  (39  –  52  years)   among   females.  During   the   study  period,   132   (12.5%)   reported  being  homeless  and  229  (21.7%)  reported  having  stable  employment  (i.e.,  a  regular  job,  temporary  job,  or  self-­‐‑employed).  In  total,  465  (44.2%)  reported  having  ever  used  illicit  drugs  while  in  hospital   and   504   (47.9%)   reported   having   ever   been   denied   pain   medication   by   a  healthcare  provider.  As  indicated  in  Table  4.1,  in  bivariable  analyses,  having  ever  been  denied  pain  medication  was  positively  associated  with  having  ever  used  illicit  drugs  in  hospital   (OR   =   1.44;   95%CI:   1.13   –   1.83).   Secondary   factors   positively   associated  with  having   ever   used   drugs   in   hospital   included:   daily   injection   drug   use,   daily   non-­‐‑injection   drug   use   and   binge   drug   use,   while   older   age   and   male   gender   were  negatively  associated  with  the  outcome  (all  p  <  0.05).    Table   4.2   presents   the   results   of   the   multivariable   analysis   examining   the  relationship  between  having  ever  been  denied  pain  medication  and  having  ever  used   56 illicit   drugs   in   hospital.   In   the   multivariable   logistic   regression   model   adjusted   for  various   confounders,   having   been   denied   pain   medication   remained   positively   and  independently  associated  with  having  used  illicit  drugs  in  hospital  (AOR  =  1.46;  95%CI:  1.14  –  1.88).    4.4 Discussion  In  the  present  study  it  was  found  that  a  substantial  proportion  of  a  community-­‐‑recruited  sample  of  PWUD  in  Vancouver  had  reported  having  ever  used  illicit  drugs  in  hospital.  A   statistically   significant   association   between   having   ever   been   denied   pain  medication  and  in-­‐‑hospital  illicit  drug  use  was  also  observed,  after  adjusting  for  a  range  of  confounders.  The  findings  suggest  that  PWUD  may  resort  to  the  self-­‐‑management  of  pain  via  high-­‐‑risk  methods   after   being  denied  pain  medication   in   acute   care   settings.  There  are  several  possible  explanations  for  the  findings.  First,  it  may  be  that  stigma  and  discrimination  PWUD  sometimes  experience   in  hospitals  may  have  contributed  to   the  denial   of   pain  medication   in   the   setting,  which   are   previously   described   experiences  among   this   population.26,57,105   Second,   this   finding   is   also   consistent  with   past   studies  that   indicate   the   reluctance   of   healthcare   providers   to   provide   pain   medication   for  reasons  that  include  concerns  over  exacerbating  the  patient’s  drug  addiction,  relapse  or  ‘drug-­‐‑seeking’   behaviour.56,106   Third,   physicians   may   be   hesitant   to   prescribe   pain  medication   to   PWUD   for   fear   of   being   disciplined   by   their   professional   regulatory   57 bodies.   Indeed,   the   College   of   Physicians   and   Surgeons   of   British   Columbia   have  warned   physicians   against   prescribing   opioids   to   high-­‐‑risk   populations,   including  “patients   with   the   lifelong   disease   of   addiction”   and   “those   with   major   psychiatric  illness   or   personality   disorders”.107   Moreover,   the   American   Pain   Society   and   the  American   Academy   of   Pain   Medicine’s   clinical   guidelines   caution   that   the   potential  risks  of  opioid  therapy  may  outweigh  the  benefits  associated  with  drug  misuse,  abuse,  and  addiction  for  some  patients  with  a  history  of  substance  abuse.108  Lastly,  given  the  challenges   associated  with   procuring   and   using   drugs   in   hospitals,11   higher   intensity  drug  users  may  be  more  likely  to  request  and  be  denied  additional  pain  medication.    The  findings  have  important  public  health  implications,  particularly  considering  the   high   prevalence   of   in-­‐‑hospital   illicit   drug   use   and   denial   of   pain  medication   that  study  participants  reported.  Firstly,  appropriate  pain  management  for  PWUD  in  acute  care   settings   may   serve   to   minimize   preventable   morbidity   and  mortality   associated  with   the   risk   of   self-­‐‑managing  pain   via   illicit   drug  use.   Secondly,   consistent  with   the  recommendations   in   Chapter   3,   efforts   to   improve   cultural   competency   and   remove  negative   stereotypes   associated   with   addiction   through   education   and   training  programs   that   specialize   in   addiction  medicine   are   needed   in   this   setting.109   Thirdly,  there   may   be   a   potential   role   for   harm   reduction   programs   in   hospital   settings   to  mitigate  the  harmful  effects  of  in-­‐‑hospital  drug  use.104,110  Further  discussed  in  Chapters  5  and  6,  the  present  findings  lend  support  to  the  argument  for  a  structural  shift  in  policy   58 that  moves  away  from  abstinence-­‐‑based  drug  policies  and  towards  the  implementation  of   a   comprehensive  package  of   harm   reduction  programs,   including   supervised  drug  consumption  facilities  and  NSPs,  in  hospital  settings.  Indeed,  past  research  has  shown  that   integrating   harm   reduction   services   within   clinical   settings   has   had   a   positive  impact   on   the   health   of   PWUD.61,111   The   recommendations   are   also   consistent   with  studies  that  suggest  that  a  harm  reduction  approach  has  the  potential   to  reduce  drug-­‐‑related   risks   from   in-­‐‑hospital   drug   use   as   well   as   discharge   AMA   among   this  population.11   Future   research   that   seeks   to   explore   potential   harm   reduction-­‐‑based  interventions  may   be   useful   to   inform   hospital   policy   changes.   Related   to   this,   there  may  be   a   role   for   alternative  non-­‐‑hospital   services   available   in   the   community  where  PWUD   can   access   pain   medication   in   a   more   flexible   and   less   abstinence-­‐‑based  environment.  Lastly,   there   is  a  need   to   re-­‐‑evaluate  current   clinical  guidelines   for  pain  management,108   as   these  may   not   necessarily   be   appropriate   for   healthcare   providers  who   care   for   PWUD  patients,   particularly   among   those  who   contend  with   comorbid  addiction  and  mental  health  complications.  This   study  has   several   limitations.  While   the   limitations   common   to  all   studies  are  presented   in  section  7.3,   the   limitations  specific   to   this  chapter  are  presented  here.  First,   the   candidate  was   unable   to   determine  whether   having   ever   been   denied   pain  medication   resulted   in   illicit   self-­‐‑medication.  However,   conceptually,   the   direction   of  this   association   seems   more   plausible   than   the   opposite   direction,   which   states   that   59 PWUD  who  used   illicit   drugs   in  hospital  were  more   likely   to  have  been  denied  pain  medication   compared   to   those   who   did   not   use   illicit   drugs   in   hospital.   Future  longitudinal   research   should   seek   to  more   effectively   estimate   the   causal   relationship  between  having  been  denied  pain  medication  and   illicit  drug  use   in  hospital.  Second,  data   on   participants’   medical   condition,   diagnosis   or   other   information   on   their  admission   were   unavailable   given   that   these   variables   were   not   asked   in   the  questionnaire.    In   summary,   the   present   study   found   that   a   substantial   proportion   of   PWUD  experienced  being  denied  pain  medication  by  healthcare  providers  and  reportedly  used  illicit  drugs  within  hospital  settings.  The  findings  also  suggest  that  after  controlling  for  various   confounders,   denial   of   pain   medication   by   healthcare   providers   was  independently   associated  with   in-­‐‑hospital   illicit   drug  use.   These   findings   indicate   the  need  for  novel  efforts  to   improve  pain  management  among  this  population,   including  education   and   training   for   healthcare   providers,   implementation   of   harm   reduction  programs   within   hospitals,   and   appropriate   clinical   guidelines   for   managing   pain  among  PWUD.  Ultimately,   these   efforts  may   serve   to  minimize   the   severe  drug-­‐‑   and  health-­‐‑related   harms   associated  with   the   self-­‐‑management   of   pain   via   high-­‐‑risk   illicit  drug  use.      60 Table 4.1 Bivariable logistic regression analyses of factors associated with having ever used street drugs in hospital among people who use illicit drugs (n = 1053)  Ever used drugs in hospital   Characteristic Yes n (%) n = 452 No n (%) n = 575 Odds Ratio (95% CI) p - value Ever denied pain medication    yes 246 (52.9) 258 (43.9) 1.44 (1.13 – 1.83) 0.004 no 219 (47.1) 330 (56.1)   Age     median 47  50  0.99 (0.99 – 1.00) <0.001 IQR (41 – 53) (43 – 55)   Gender     male 282 (60.6) 430 (73.1) 0.57 (0.44 – 0.73) <0.001 female 183 (39.6) 158 (26.9)   Proportion of daily injection drug use over time   ≥ 50% of the time 146 (31.4) 103 (17.5) 2.16 (1.61 – 2.88) <0.001 < 50% of the time 319 (68.6) 485 (82.5)   Proportion of daily non-injection drug use over time   ≥ 50% of the time 249 (53.5) 229 (38.9) 1.81 (1.41 – 2.31) <0.001 < 50% of the time 216 (46.5) 359 (61.1)   Proportion of binge drug use over time    ≥ 50% of the time 204 (43.9) 209 (35.5) 1.42 (1.10 – 1.82) 0.006 < 50% of the time 261 (56.1) 379 (64.5)   CI: confidence interval    61 Table 4.2 Multivariable logistic regression of factors associated with ever having used street drugs in hospital among people who use illicit drugs (n = 1053) Variable Adjusted  Odds Ratio  95% CI  p - value Ever denied pain medication   (yes vs. no) 1.46 (1.14 – 1.88) 0.003 Age    (per year increase) 0.97 (0.96 – 0.99) <0.001 Proportion of daily injection drug use over time   (≥ 50% vs. < 50%) 1.93 (1.44 – 2.60) <0.001 CI: confidence interval          62 Chapter  5: The  impact  of  an  HIV/AIDS  adult  integrated  health  program  on  leaving  hospital  against  medical  advice  among  HIV-­‐‑positive  people  who  use  illicit  drugs  5.1 Introduction  HIV-­‐‑positive  PWUD  are  at  an  elevated  risk  of  leaving  hospital  AMA  compared  to  their   non   drug-­‐‑using   counterparts.12,15   This   has   been   documented   in   past   research  conducted   in   Vancouver   that   has   shown   that   hospital   discharge   AMA   is   common  among  HIV-­‐‑positive  PWID,  and  severely  addicted  PWID  account  for  a  large  proportion  of  all  discharges  AMA  from  an  HIV/AIDS  ward.12  These  findings  are  concerning  on  two  fronts:   first,  HIV-­‐‑positive  PWUD  are  vulnerable   to  a  wide  range  of  comorbidities  and  other  health-­‐‑related  harms,  including  opportunistic  infections  and  soft-­‐‑tissue  infections,  that  usually  require  lengthy  hospital  visits,4,12,15  and  second,  there  are  numerous  barriers  to   accessing  healthcare   services   (e.g.,   stigmatizing  attitudes)   that  HIV-­‐‑positive  PWUD  continuously   face   that   makes   retention   in   care   of   particular   importance   among   this  population.22,23    Alternative  models  of  care  have  been  implemented  in  various  settings  that  aim  to  minimize  risks  and  improve  overall  health  outcomes  for  PWUD,  including  those  living  with   HIV   disease.112,113   In   Vancouver,   the   Dr.   Peter   Centre   (DPC)   is   an   established  specialty  HIV/AIDS-­‐‑focused  adult   integrated  health  program  that  provides  support   to  some   of   the   city’s   most   vulnerable   citizens   who   face   poverty,   homelessness,   mental   63 health   and   addiction   issues   in   addition   to   HIV/AIDS.58   The   DPC   is   a   non-­‐‑profit  organization   and   is   situated   one   street   block   away   from  St.   Paul’s  Hospital,   an   acute  care  hospital   located  in  downtown  Vancouver,  and  provides  three  core  programs  that  include   a   day   health   program,   a   24-­‐‑hour   specialized   nursing   care   residence   and   an  enhanced  supported  housing  program.  Within  these  programs,  a  range  of  medical  and  harm  reduction  programs  are  provided  to  adults  at  the  DPC,  including:  addiction  and  other  counselling,  provision  of  food  and  nutrition  services,  art  and  music  therapy,  and  a  SIF.  The  DPC  started  the  SIF  in  2002  in  response  to  the  rise  in  injection-­‐‑related  adverse  events   (e.g.,   overdose)   in   Vancouver   and   this   was   supported   by   the   College   of  Registered  Nurses  of  BC  stating  that  supervising  injections  was  indeed  within  the  scope  of  nursing  practice.59,114  Despite   the   substantial   harms   and   costs   associated  with   the   problem   of   leaving  hospital   AMA,   there   is   limited   empiric   research   focused   on   ways   to   address   this  problem  among  PWUD  who  are   living  with  HIV  disease.  Given   the   evidence   linking  HIV   to  an  elevated   risk  of  hospitalization  among  PWUD,60   the  objective  of   this   study  was   to  assess   the   impact  of  an   innovative  HIV/AIDS  adult   integrated  health  program  operating  in  proximity  to  a  hospital  on  leaving  that  hospital  AMA  among  HIV-­‐‑positive  PWUD.     64 5.2 Methods  Quantitative  data   for   this   study  were  derived   from   the  ACCESS   study,   the  DTP  and  St.  Paul’s  Hospital  health  records  and  discharge  database,  as  described  in  detail  in  section  1.6.      5.2.1 Study  sample  The   present   study   was   restricted   to   participants   who   experienced   at   least   one  admission  to  in-­‐‑patient  care  at  St.  Paul’s  Hospital  between  July  2005  and  2011.      5.2.2 Variable  selection  Consistent   with   the   analyses   presented   in   Chapter   3,   the   primary   outcome   of  interest  for  this  analysis  was  leaving  hospital  AMA,  which  was  obtained  from  St.  Paul’s  Hospital   administrative   discharge   database.   The   primary   explanatory   variable   of  interest   was   being   a   participant   of   the   DPC,   defined   as   responding   “yes”   to   the  following  question:  “Are  you  a  participant  at  the  Dr.  Peter  Centre?”  To  be  a  participant  of   the   DPC,   eligible   individuals   must   be   diagnosed   with   HIV/AIDS   and   live   with  several  other  challenges,  including  mental  health  conditions,  unstable  housing,  limited  financial   support,   or   physical   disabilities.   An   admission   assessment   interview  conducted  by  the  DPC  staff  will  determine  if  the  DPC  meets  the  needs  of  the  individual  and  admission  is  prioritized  based  on  need.58     65 Secondary   socio-­‐‑demographic,   behavioural   and   clinical   variables   believed   to   be  confounders   included:   age   (per   year   increase);   gender   (male   vs.   female);   ancestry  (Caucasian  vs.  other);  homeless  (yes  vs.  no);  mental  illness  diagnosis  (yes  vs.  no);  illicit  drug  use   (≥  daily  vs.   <  daily);   enrollment   in  methadone  maintenance   therapy   (yes  vs.  no);   difficulty   finding   equipment   specific   to   injection   drug   use   (yes   vs.   no);   pVL  (copies/mL   plasma,   per   log10   increase);   CD4   cell   count   (per   100   cells/mL);   and   ART  initiation  ever  (yes  vs.  no).  In  contrast  to  Chapter  3  where  ‘Aboriginal  ancestry’  was  an  explanatory  variable  of  interest,  in  this  Chapter,  ‘Caucasian  ancestry’  was  used  instead  given  that  there  was  a  lack  of  supporting  literature  to  include  ‘Aboriginal  ancestry’  as  a  potential   confounding   factor   between   being   a   DPC   participant   and   leaving   hospital  AMA.   While   the   present   study   included   all   HIV-­‐‑positive   PWUD   who   had   been  previously   hospitalized   at   St.   Paul’s   Hospital   between   2005   and   2011,   the   variable  ‘difficulty   finding  equipment   specific   to   injection  drug  use’  was   restricted   to   injection  paraphernalia  given  the  hypothesis  that  PWUD  patients  may  be  leaving  the  hospital  to  use  the  DPC  SIF,  which  currently  only  provides  harm  reduction  services  related  to  drug  injection   and  does   not   have   a   supervised   inhalation   facility   for   people  who  use   non-­‐‑injection  drugs.  All  time-­‐‑varying  variables  are  time-­‐‑updated  and  refer  to  the  six-­‐‑month  period  prior  to  the  follow-­‐‑up  interview,  unless  otherwise  stated.       66 5.2.3 Statistical  analyses  As   a   first   step,   the   candidate   compared   selected   baseline   characteristics   among  participants  who  left  hospital  AMA  compared  to  those  who  did  not  leave  hospital  AMA  using   simple   logistic   regression.   Next,   consistent   with   longitudinal   analyses   of  correlated  data  presented  in  Chapter  3,  the  candidate  used  GEE  to  estimate  crude  ORs  for  the  effect  of  being  a  DPC  participant  and  all  other  secondary  explanatory  variables  on  leaving  hospital  AMA.  As   a   final   step,   a   multivariable   GEE   model   was   constructed   to   estimate   the  independent  effect  of  being  a  DPC  participant  on  leaving  hospital  AMA,  adjusting  for  various  confounders.  All  confounding  variables  were  entered  in  the  final  multivariable  model   regardless   of   statistical   significance   in   bivariable   analyses   as   the   candidate  decided  a-­‐‑priori  based  on  existing  literature  that  these  variables  should  be  considered  as  confounders  in  this  analysis.12,15,65,78,115,116  All  p-­‐‑values  are  two-­‐‑sided.    5.3 Results  Between   July   2005   and   July   2011,   a   total   of   181   HIV-­‐‑positive   PWUD   had  experienced  at  least  one  hospitalization  and  were  included  in  the  study:  81  (44.8%)  were  female  and  the  median  age  at  baseline  was  43  years  (interquartile  range:  38  –  50  years).  Of  the  406  hospital  admissions  among  these  individuals,  73  (39.9%)  participants  left  the  hospital  AMA  a   total   of   126   (31.0%)   times.   In   total,   44   (24.3%)   individuals  were  DPC   67 participants.  The  number  of  times  participants  were  hospitalized  ranged  between  1  and  10  hospitalizations  per  participant  over  the  six-­‐‑year  study  period.  Table  5.1  shows  the  baseline   characteristics   of   the   study   sample   stratified   by   leaving   hospital   AMA   at  baseline.  The   crude   longitudinal   estimates   of   the   odds   of   leaving   hospital   AMA   are  presented   in   Table   5.2.   DPC   participants   had   significantly   lower   odds   of   leaving  hospital   AMA   in   unadjusted   analysis   compared   to   those   who   were   not   DPC  participants  (OR:  0.43;  95%  CI:  0.20  –  0.94).  Also  presented  in  Table  5.2,  a  multivariable  model   adjusted   for   various   demographic,   socioeconomic   and   clinical   confounders  showed   that   being   a  DPC  participant   remained   independently   associated  with   lower  odds  of  leaving  hospital  AMA  (AOR  =  0.42;  95%CI:  0.19  –  0.89).    5.4 Discussion  The  present  study  observed   that   leaving  hospital  AMA  was  common  among  the  participants,  with  approximately  40%  leaving  hospital  AMA  at  some  point  during  the  six-­‐‑year  study  period.  After  adjusting  for  a  range  of  relevant  factors,  the  present  study  found  that  being  a  DPC  participant  was  associated  with  a  significantly  reduced  odds  of  leaving  AMA  compared   to  non-­‐‑DPC  participants.  To   the   candidate’s   knowledge,   this  study  is  the  first  to  demonstrate  the  beneficial  role  of  a  comprehensive  and  integrated   68 HIV/AIDS-­‐‑focused   health   program   operating   in   proximity   to   a   hospital   on   leaving  hospital  prematurely  among  HIV-­‐‑positive  PWUD.  Similar   to   the   findings   presented   in   Chapter   3   and   previous   research   on   HIV-­‐‑positive   PWID,12   the   current   findings   indicate   that   a   substantial   proportion   of   HIV-­‐‑positive  PWUD  in   the  sample   left  hospital  AMA.  This   is  concerning  given   the  known  risks  associated  with  leaving  hospital  prematurely.17,18,78  The  findings  also  indicate  that  an   HIV/AIDS   adult   integrated   health   program   operating   in   proximity   to   a   hospital,  which   implements   various   harm   reduction   strategies,   may   curb   the   rate   of   leaving  hospital  prematurely  among  PWUD  patients.  Given   that   there   is  a  paucity  of  data  on  novel  interventions  specific  to  AMA  among  this  population,67  drawing  from  the  current  literature   on   PWUD   in   acute   care   settings   more   broadly   may   be   useful.   A   body   of  research  has  demonstrated  that  stigma,  discrimination  and  drug  addiction  within  acute  care  settings  are  tightly  related.26,57  As  mentioned  earlier,  abstinence-­‐‑based  policies  and  lack  of  harm  reduction  services  within  hospital  settings  may  also  contribute  to  PWUD  leaving  hospital  before  completing  their  treatment.11  As  such,  the  DPC’s  harm  reduction  programs  incorporated  into  their  integrated  healthcare  program  may  be  responsible  for  driving  the  negative  association  between  being  a  DPC  participant  and  leaving  hospital  AMA.  For  instance,  it  may  be  that  DPC  participants  who  are  admitted  into  hospital  are  provided  with   a   safe   environment   in   a   nearby   location  where   they   can   access   sterile  drug   paraphernalia   and,   after   a   nursing   assessment,   use   their   illicit   drugs   under   the   69 supervision   of   a   nurse   without   increasing   their   risk   of   leaving   hospital   AMA.61  Moreover,   the   DPC   may   provide   respite   from   the   hospital,   particularly   around   the  stigma   and   abstinence-­‐‑based   policies   that   may   promote   AMA.59,117   Other   broader  integrated  health  strategies  offered  at  the  DPC,  including  education,  social  support  and  food  and  nutrition  services,  may  also  be  driving  the  aforementioned  association.  Given  that  little  is  known  about  this  topic,  future  in-­‐‑depth  qualitative  research  should  seek  to  explore  why  participants  of  an  HIV/AIDS  adult  integrated  health  program  in  proximity  to   a  hospital,  which  has  an  environment   that   recognizes   the  need   for  harm  reduction  strategies,  are  less  likely  to  leave  hospital  AMA.    While   the   limitations   common   to   all   studies   are   presented   in   section   7.3,   the  limitations   specific   to   this   chapter   are  presented  here.   In  particular,   it   is   important   to  note  that  there  may  be  selection  bias  due  to  the  fact  that  participants  were  not  randomly  assigned   to   be   a   DPC   participant.   Additionally   and   similarly   to   the   limitations  presented  in  Chapter  3,  this  research  project  included  only  one  hospital.  This  limitation  would  have   resulted   in  an  underestimation  of  hospital  use,  as   some  participants  may  have  accessed  other  hospitals  during   the  study  period.  However,  as  discussed  earlier,  St.  Paul’s  Hospital  services   the  majority  of   illicit  drug  users   in   this  setting.1,2,9  As  well,  the   present   study   aimed   to   examine   the   impact   of   a   HIV/AIDS   program   that   is  operating  in  proximity  to  a  hospital  and  therefore,  it  would  not  be  appropriate  to  assess   70 the  relationship  between  this  program  and  leaving  hospital  AMA  at  a  hospital  that  was  not  adjacent  to  the  DPC.    In   sum,   the   present   study   found   that   a   substantial   proportion   of   PWUD   in   this  study   reported   leaving   hospital   AMA.   It   was   also   demonstrated   that   an   HIV/AIDS  integrated   health   program   in   proximity   to   hospitals   may   curb   the   rate   of   leaving  hospital   prematurely.   The   findings   are   supportive   of   the   development   of   similar  programs  in  similar  settings,  which  may  minimize  high  AMA-­‐‑related  human  and  fiscal  costs.  Additionally,  employing  a  comprehensive  harm  reduction  program  within  acute  care   settings   and   improving   cultural   competency   among   healthcare   workers  may   be  valuable  for  reducing  the  negative  consequences  associated  with  leaving  hospital  AMA.    71 Table 5.1 Baseline characteristics of HIV-positive people who use illicit drugs stratified by leaving hospital against medical advice (n = 181)  Left hospital against medical advice  Characteristic Yes n (%) n = 52 No n (%) n = 129 Odds Ratio (95% CI) p - value Dr. Peter Centre participant    yes 6 (11.5) 31 (24.0) 0.41 (0.16 – 1.06) 0.06 no 46 (88.5) 98 (76.0)   Age     median 41 44  0.96 (0.92 – 1.00) 0.05 IQR (35 – 48)  (39 – 51)   Gender     male 27 (51.9) 73 (56.6) 0.83 (0.43 – 1.58) 0.57 female 25 (48.1) 56 (43.4)   Ancestry     Caucasian 23 (44.2) 75 (58.1) 0.57 (0.30 – 1.09) 0.09 other 29 (55.8) 54 (41.9)   Homeless*     yes 15 (28.8) 38 (29.5) 0.97 (0.48 – 1.97) 0.94 no 37 (71.2) 91 (70.5)   Mental illness diagnosis*     yes 24 (46.2) 69 (53.5) 0.75 (0.39 – 1.42) 0.37 no 28 (53.8) 60 (46.5)   Illicit drug use*     ≥ daily 38 (73.1) 71 (55.0) 2.22 (1.10 – 4.48) 0.02 < daily 14 (26.9) 58 (45.0)   Enrollment in methadone maintenance therapy*   yes 22 (42.3) 61 (47.3) 0.82 (0.43 – 1.57) 0.54 no 30 (57.7) 68 (52.7)   Difficulty finding injection equipment*    yes 13 (25.0) 31 (24.0) 1.05 (0.50 – 2.22) 0.89 no 39 (75.0) 98 (76.0)   Plasma HIV-1 RNA viral load* (per log10 cells/mL)   median 3.74 2.85  1.09 (0.87 – 1.36) 0.47 IQR (1.65 – 4.54)  (1.65 – 4.66)   CD4 cell count* (per 100 cells/mL)    median 3.12 2.85  1.10 (0.95 – 1.28) 0.21 IQR (2.10 – 4.61)  (1.50 – 4.35)   Antiretroviral therapy initiation ever   yes 48 (92.3) 121 (93.8) 0.79 (0.23 – 2.76) 0.75 no   4 (7.7)     8 (6.2)   *Refers to the six month period prior to follow-up   72 Table 5.2 Crude and adjusted longitudinal estimates of the odds of leaving hospital against medical advice among HIV-positive people who use illicit drugs (n = 181)  Crude  Adjusted Characteristic Odds Ratio (95% CI) p - value  Odds Ratio (95% CI) p - value Dr. Peter Centre participant     (yes vs. no) 0.43 (0.20 – 0.94) 0.03  0.42 (0.19 – 0.89) 0.02 Age      (per year older) 0.95 (0.92 – 0.98) <0.01  0.96 (0.92 – 0.99) 0.02 Gender      (male vs. female) 0.77 (0.45 – 1.31) 0.33  1.01 (0.59 – 1.75) 0.96 Ancestry      (Caucasian vs. other) 0.73 (0.43 – 1.24) 0.24  0.83 (0.48 – 1.45) 0.52 Homeless*      (yes vs. no) 1.04 (0.56 – 1.93) 0.90  0.83 (0.45 – 1.54) 0.55 Mental illness diagnosis*      (yes vs. no) 1.02 (0.59 – 1.77) 0.95  0.98 (0.58 – 1.65) 0.94 Illicit drug use*      (≥ daily vs. < daily) 1.43 (0.87 – 2.37) 0.16  1.21 (0.71 – 2.06) 0.49 Enrollment in methadone maintenance therapy*    (yes vs. no) 0.81 (0.47 – 1.38) 0.43  0.76 (0.42 – 1.35) 0.34 Difficulty finding equipment*     (yes vs. no) 1.52 (0.90 – 2.58) 0.12  1.59 (0.92 – 2.75) 0.10 Plasma HIV-1 RNA viral load*     (per log10 increase) 1.16 (0.98 – 1.38) 0.08  1.08 (0.91 – 1.29) 0.39 CD4 cell count*      (per 100 cells) 1.01 (0.90 – 1.13) 0.88  1.01 (0.93 – 1.20) 0.41 Antiretroviral therapy initiation ever     (yes vs. no) 0.63 (0.18 – 2.21) 0.47  0.71 (0.19 – 2.63) 0.61 CI: confidence interval *Refers to the six month period prior to follow-up      73 Chapter  6: Willingness  to  access  an  in-­‐‑hospital  supervised  injection  facility  among  people  who  use  illicit  drugs  6.1 Introduction  SIFs   are   sanctioned   environments   where   PWUD   can   inject   pre-­‐‑obtained   illicit  drugs  under  the  supervision  of  healthcare  staff.  Internationally,  SIFs  have  been  shown  to  improve  public  health  and  public  order  within  surrounding  communities.64,118–120  For  example,   a   dramatic   decline   in   fatal   overdoses   in   Vancouver’s   DTES   neighbourhood  was   attributed   to   the   implementation   of   a   SIF   in   the   area.64   Changes   in   risk   injecting  behaviour   have   also   been   observed   among   individuals   who   access   SIFs.63   It   is   also  noteworthy   that   one   Vancouver-­‐‑based   study   conducted   a   cost-­‐‑effective   analysis   and  found  that  the  implementation  of  a  SIF  was  associated  with  improved  health  and  cost  savings.121  While   a   large   body   of   evidence   supports   SIFs   as   an   effective   approach   for  minimizing   the   drug-­‐‑   and   health-­‐‑related   harms   within   street-­‐‑based   drug   scenes,119,122  little  is  known  about  whether  there  is  a  role  for  SIFs  within  hospital  settings.  Currently  in  Vancouver,  harm  reduction  services  that  are  available  in  the  community  (e.g.,  NSPs,  SIFs)  are  generally  not  being  provided  within  hospital  settings.  Therefore,  the  candidate  sought   to   conduct   a   needs   assessment   to   identify   the   prevalence   and   correlates   of  willingness  to  access  an  in-­‐‑hospital  SIF  among  PWUD  in  Vancouver.  These  data  may  be  crucial  for  planning  appropriate  programs  and  services  to  reduce  health-­‐‑related  harms  and  leaving  hospital  AMA  among  PWUD  in  hospital  settings.   74   6.2 Methods  Data  for  these  analyses  were  ascertained  from  the  VIDUS  and  ACCESS  studies,  as  described  in  section  1.6.    6.2.1 Study  sample  Given   that   the   variable   measure   was   based   on   a   hypothetical   scenario,   all  participants  who   completed   the   survey   between   June   2013   and  November   2013  were  eligible  for  inclusion  regardless  of  their  current  injection  drug  use  behaviour.    6.2.2 Variable  selection  The  primary  outcome  of  interest  for  this  analysis  was  willingness  to  access  an  in-­‐‑hospital   SIF   (yes   vs.   no   or   unsure),   ascertained   by   asking   participants   the   following  hypothetical  question:  “If  you  were  admitted  into  a  hospital,  and  if  a  supervised  safer  injection  site  was  available   in   that  hospital,  would  you  use   it?”  Given  the  existence  of  two   SIFs   in   the   local   environment,   PWUD   in  Vancouver   are   familiar  with   the  design  and  operation  of  such  programs.  Variables  considered  included:  age  (per  year  increase),  gender  (male  vs.  female),  HIV  serostatus  (positive  vs.  negative),  heroin  injection  (≥  daily  vs.  <  daily),  cocaine  injection  (≥  daily  vs.  <  daily),  crystal  methamphetamine  injection  (≥  daily  vs.   <  daily),   prescription  opioid   injection,  defined   as   in  Chapter   3   (≥  daily  vs.   <   75 daily),   binge   injection   drug   use   (yes   vs.   no),   ever   left   hospital   AMA   because   they  wanted  or  needed  to  use  illicit  drugs  (yes  vs.  no),  ever  used  illicit  drugs  in  hospital  (yes  vs.   no),   previously   used   a   SIF   (yes   vs.   no),   ever   had   negative   experiences   with  healthcare  providers  (yes  vs.  no)  and  ever  had  negative  experiences  with  police,  defined  by  having   ever  been   confronted  and/or   assaulted  by  police   (yes  vs.   no).  All   variables  refer  to  the  previous  six  months  unless  otherwise  indicated.      6.2.3 Statistical  analyses  First,  the  candidate  compared  PWUD  who  were  and  were  not  willing  to  access  an  in-­‐‑hospital   SIF   using   simple   logistic   regression   analyses.   To   identify   factors  independently  associated  with  willingness  to  access  an  in-­‐‑hospital  SIF,  a  multivariable  logistic  regression  model  was  constructed  using  an  a  priori-­‐‑defined  statistical  protocol  based  on  examination  of   the  AIC  and  p   values.  First,   the   candidate   constructed  a   full  model   that   included   all   variables   significant   at   p   <   0.10   in   bivariable   analyses.   After  noting  the  AIC  of  the  model,  the  variable  with  the  largest  p  value  was  removed  and  a  reduced  model  was  built.  This  iterative  process  continued  until  no  variables  remained  for  inclusion.  The  candidate  selected  the  multivariable  model  with  the  lowest  AIC  score.  All  p  values  were  two  sided.  As  a  sub-­‐‑analysis,  participants  who  were  willing  to  access  and  in-­‐‑hospital  SIF  were  asked  to  indicate  reasons  why  they  were  willing  to  access  such  a  facility.   76   6.3 Results  In  total,  732  PWUD  participated  in  the  study;  250  (34.2%)  were  female,  the  median  age  was  48  years  (interquartile  range:  41  –  53  years),  and  307  (41.5%)  were  HIV-­‐‑positive.  Among  the  study  sample,  499  (68.2%)  participants  were  willing  to  access  an  in-­‐‑hospital  SIF   if   it   were   available.   As   indicated   in   Table   6.1,   in   bivariable   analyses,   factors  significantly   and   positively   associated   with   willingness   to   access   an   in-­‐‑hospital   SIF  included:  daily  heroin  injection  (OR  =  2.15;  95%CI:  1.35  –  3.40);  ever  left  hospital  AMA  because  they  wanted  or  needed  to  use  illicit  drugs  (OR  =  5.07;  95%CI:  1.18  –  21.83);  ever  used  illicit  drugs  in  hospital  (OR  =  1.65;  95%CI:  1.20  –  2.27);  and  previously  used  a  SIF  (OR  =  1.70;  95%CI:  1.23  –  2.36).  As   indicated   in   Table   6.2,   in   multivariable   analyses,   factors   that   remained  significantly   and   positively   associated   with   willingness   to   access   an   in-­‐‑hospital   SIF  included:  daily  heroin  injection  (AOR  =  1.90;  95%CI:  1.20  –  3.11);  ever  used  illicit  drugs  in  hospital   (AOR  =   1.63;   95%CI:   1.18   –   2.26);   and  previously  used   a   SIF   (AOR  =   1.53;  95%CI:  1.10  –  2.15).  Among  participants  who  were  willing  to  access  an  in-­‐‑hospital  SIF,  the  most   common   reasons   included:   to   be   able   to   stay   in   hospital   (45.9%);   to   reduce  their  drug-­‐‑related  risks  (37.9%);  and  to  reduce  stress  associated  with  being  kicked  out  of  the  hospital  because  they  were  using  drugs  (19.4%).     77 6.4 Discussion  The  present  study  found  that  over  two-­‐‑thirds  of  PWUD  participating  in  the  study  were  willing  to  access  an  in-­‐‑hospital  SIF  if  such  a  service  was  available.  These  findings  also  revealed  positive  and  independent  relationships  between  willingness  to  access  an  in-­‐‑hospital   SIF   and   various   risk   factors,   including   frequent   heroin   injection   and   ever  having   used   illicit   drugs   in-­‐‑hospital.   Having   previously   used   a   SIF   also   predicted  willingness  to  access  an  in-­‐‑hospital  SIF.  Among  those  who  were  willing  to  access  an  in-­‐‑hospital  SIF,   the  main  reasons  given  by  participants   for  wanting  to  access   this   type  of  service  was  so  that  they  could  stay  in  the  hospital  and  reduce  their  drug-­‐‑related  risks.  The   limitations   of   this   study   are   related   to   the   study   design   and,   along   with   the  limitations   common   among   all   the   analyses   presented   in   each   of   the   chapters,   are  included  in  section  7.3.  The   finding   that   many   PWUD   were   willing   to   access   an   in-­‐‑hospital   SIF   is  encouraging  given  that  a  large  proportion  of  PWUD  are  hospitalized  annually  for  acute  and   chronic   diseases.4,9   Previous   studies   have   documented   the   positive   impact   of  incorporating   a   harm   reduction   model   within   hospital   settings,   resulting   in   more  comprehensive   care   for   PWUD.111,123   For   example,   the   previously   mentioned   DPC  program  provides   a   SIF   for  HIV-­‐‑positive   PWUD   to   safely   use   illicit   drugs   under   the  supervision  of  trained  nurses  and  was  at  one  time  located  at  St.  Paul’s  Hospital.58  While  the  DPC   currently   operates   outside   of   St.   Paul’s  Hospital,   it  may  be   advantageous   to   78 model   an   in-­‐‑hospital   SIF   after   the  DPC’s  harm   reduction   room  given   their   success   in  facilitating  access  and  delivery  of  comprehensive  care  for  PWUD.123    Previous   studies   have   identified   various   high-­‐‑risk   locations   where   individuals  with  drug  dependence  use  illicit  drugs  to  maintain  their  established  drug-­‐‑using  habits,  including   using   in   locked   washrooms   in   hospitals.11   The   study   found   a   positive  association  among  PWUD  who  had  used  illicit  drugs  in  the  hospital  and  willingness  to  use   an   in-­‐‑hospital   SIF.   The   finding   is   reassuring  given   that   these   individuals   are   at   a  higher   risk  of  negative  health   consequences   (e.g.,   fatal  overdose)   from  using  drugs   in  hospital.11   These   findings   support   recent   calls   for   an   integrated   approach   to   care   by  implementing  harm  reduction  services  within  hospital  settings  in  an  effort  to  minimize  the  harms  associated  with  illicit  drug  use.104,110  The   study   also   found   that   high   frequency   heroin   injection   was   associated   with  willingness  to  access  an  in-­‐‑hospital  SIF.  As  one  of  the  many  possible  explanations  that  have   been   put   forward,   this   relationship   may   be   a   result   of   the   complex   nature   of  treating   opioid-­‐‑dependent   patients   for   pain   and   appropriately,   is   in   line   with   the  findings  discussed  in  Chapters  3  and  4.  Challenges  that  make  it  difficult  to  adequately  prescribe   pain   medication   to   these   individuals   have   been   previously   discussed   and  among   other   things,   include   issues   around   the   already   established   high   tolerance   for  opioids  due  to  the  concomitant  use  of  opioid  substitution  therapies  and  ongoing  drug  use,   and   withdrawal   symptoms.56,124–127   Given   the   complexities   arising   from   high   79 intensity   heroin   use,   pain   management,   and   healthcare   professionals’   perceptions  regarding   PWUD,   further   research   should   seek   to   untangle   the   causal   relationships  underlying  these  associations.  The   study   found   an   association   between   recent   use   of   a   SIF   and  willingness   to  access  an  in-­‐‑hospital  SIF.  As  mentioned  previously,  a  large  body  of  research  has  shown  improvements   in   various   health   outcomes   and   reductions   in   related   harms   in  surrounding   communities   where   SIFs   were   implemented.64,119   It   is   unfortunate   that  while  progress  in  reducing  the  harms  of  injection  drug  use  has  been  seen  in  community  settings   globally,   the   same   cannot   be   said   about   hospitals.   Given   that   many   PWUD  often  present  to  emergency  care  late  in  the  course  of  illness  and  require  admission  to  a  hospital  bed,1  it  is  important  to  ensure  that  harm  reduction  services  that  are  available  in  the  community  are  also  made  available  in  hospitals.    It  is  noteworthy  that  the  interpretation  of  these  results  should  not  be  generalized  to  other  PWUD  populations  outside  of  Vancouver.  However,  it  is  important  to  consider  that  over  the  past  few  decades,  community-­‐‑based  SIFs  have  been  successfully  operating  in   international  settings  such  as  Europe  and  Australia;120,128   thus,   the  concept  of  an   in-­‐‑hospital   SIF  may   not   be   far   from   actual   inpatient   practice   in   these   settings.   It   is   also  important   to   acknowledge   the   progress   made   towards   the   implementation   of  community-­‐‑based   SIFs   in   other   settings,   including   the   United   States.   For   example,   80 feasibility   studies   have   been   conducted   in   San   Francisco   and   New   York   and   have  shown  increasing  support  for  the  implementation  of  SIFs  in  these  areas.129,130  As   indicated   in   section  1.1,   a   large  body  of   evidence  has  documented   the  health  harms   associated  with   leaving   hospital   AMA,   including   readmission   for   a   worsened  illness  and  mortality.17,18  However,  when  faced  with  abstinence-­‐‑based  policies  that  exist  in   most   hospital   settings,   it   is   not   uncommon   for   PWUD   to   leave   the   hospital   to  maintain  their  active  addiction  or  to  address  their  drug  withdrawal.11  While  the  study  failed  to  find  a  statistically  significant  association  between  being  discharged  AMA  and  willingness  to  access  an  in-­‐‑hospital  SIF,  it  is  noteworthy  that  in  the  sub-­‐‑analysis  it  was  found  that  PWUD  who  were  more  likely  to  access  an  in-­‐‑hospital  SIF  reported  doing  so  because  they  wanted  to  stay  in  the  hospital  and  reduce  their  drug-­‐‑related  risks.  Given  that  low  counts  of  reported  AMA  discharge  events  were  observed  resulting  in  wide  CIs,  further  exploration  of  this  topic  is  warranted.    The   present   study   found   that   a   substantial   proportion   of   PWUD   in   the   sample  were   willing   to   access   an   in-­‐‑hospital   SIF   if   this   service   was   available.   PWUD   who  expressed  a  willingness  to  use  an  in-­‐‑hospital  SIF  were  more  likely  to  be  high-­‐‑intensity  heroin  users,   to  have  previously  used  illicit  drugs   in  hospital  and  were  more   likely  to  have  previously  used  a  SIF.  The   findings  highlight   the  potential  of   in-­‐‑hospital  SIFs   to  complement   existing   harm   reduction   programs   that   serve   PWUD.   Moreover,   an   in-­‐‑ 81 hospital   SIF   may   minimize   the   harms   associated   with   high-­‐‑risk   illicit   drug   use   in  hospital.          82 Table 6.1 Factors associated with willingness to access an in-hospital supervised injection facility among people who use illicit drugs in Vancouver, Canada (n = 732)  Willingness to access an in-hospital SIF  Characteristic Yes n (%) n = 499 No n (%) n = 233 Odds Ratio (95% CI) p - value Age     median 48 48 0.98 (0.97 – 1.00) 0.085 IQR (41 – 53) (42 – 54)   Gender     male 331 (66.3) 151 (64.8) 1.07 (0.77 – 1.48) 0.685 female 168 (33.7)   82 (35.2)   HIV serostatus     positive 203 (40.7) 104 (44.6) 0.85 (0.62 – 1.16) 0.313 negative 296 (59.3) 129 (55.4)   Heroin injection*     ≥ daily 106 (21.2)   26 (11.2) 2.15 (1.35 – 3.40) <0.001 < daily 393 (78.8) 207 (88.8)   Cocaine injection*     ≥ daily   46 (9.2)   19 (8.2) 1.14 (0.65 – 2.00) 0.637 < daily 453 (90.8) 214 (91.8)   Crystal methamphetamine injection*    ≥ daily   46 (9.2)   16 (6.9) 1.38 (0.76 – 2.49) 0.287 < daily 453 (90.8) 217 (93.1)   Prescription opioid injection*    ≥ daily   34 (6.8)     9 (3.9) 1.82 (0.86 – 3.86) 0.114 < daily 465 (93.2) 224 (96.1)   Binge drug use*     yes 141 (28.3)   61 (26.2) 1.11 (0.78 – 1.58) 0.558 no 358 (71.7) 172 (73.8)   Ever left hospital against medical advice    yes   21 (4.2)     2 (0.9) 5.07 (1.18 – 21.83) 0.012 no 478 (95.8) 231 (99.1)   Ever used illicit drugs in hospital    yes 238 (47.7)   83 (35.6) 1.65 (1.20 – 2.27) 0.002 no 261 (52.3) 150 (64.4)   Ever had negative experiences with healthcare providers   yes 131 (26.3)   64 (27.5) 0.94 (0.66 – 1.33) 0.729 no 368 (73.7) 169 (72.5)   Ever had negative experiences with police   yes 383  (76.8) 169  (72.5) 1.25 (0.88 – 1.78) 0.217 no 116  (23.2)     64  (27.5)      83 Willingness to access an in-hospital SIF  Characteristic Yes n (%) n = 499 No n (%) n = 233 Odds Ratio (95% CI) p - value Used a SIF*   yes 228 (45.7)   77 (33.0) 1.70 (1.23 – 2.36) 0.001 no 271 (54.3) 156 (67.0)   SIF: supervised injection facility; CI: confidence interval; IQR: interquartile range * Activities reported in the six months prior to interview     84 Table 6.2 Multivariable logistic regression analysis of factors associated with willingness to access an in-hospital supervised inhalation facility among people who use illicit drugs in Vancouver, Canada (n = 732) Variable Adjusted  Odds Ratio  (AOR) 95% Confidence Interval (CI) p - value Heroin injection*    (≥ daily vs. < daily) 1.90 (1.20 – 3.11) 0.008 Ever left hospital against medical advice   (yes vs. no) 3.74 (1.06 – 23.72) 0.079 Ever used illicit drugs in hospital   (yes vs. no) 1.63 (1.18 – 2.26) 0.003 Used a SIF*    (yes vs. no) 1.53 (1.10 – 2.15) 0.013 * Activities reported in the six months prior to interview    85 Chapter  7: Conclusion  7.1 Summary  of  findings  The  objective  of   this   thesis  was   to  provide  scientific  evidence   to   further   improve  hospital   care   for  PWUD  by   first   identifying  various   individual   and   contextual   factors  associated  with  leaving  hospital  AMA  and  illicit  drug  use   in  hospitals  among  PWUD,  and  second,  by  developing  and  evaluating  potential  approaches  to  minimizing  the  risks  and  harms  that  PWUD  experience   in  hospitals.  A  summary  of  each  specific  chapter   is  listed  below.  Chapter  2  summarizes   the   findings   from  a  systematic  review  focused  on   leaving  hospital  AMA  and  illicit  drug  use.  The  review  described  how  few  studies  have  explored  various   risk   factors   associated   with   leaving   hospital   AMA   among   PWUD   patients  despite  increasing  evidence  that  these  individuals  are  most  vulnerable  to  experiencing  this   phenomenon.   Moreover,   none   of   the   studies   included   in   the   systematic   review  incorporated  information  on  the  social,  structural  and  environmental  factors  associated  with   leaving   hospital   AMA   and   relied   only   on   information   from   hospital   records.  Finally,  the  review  highlighted  that  only  a  small  number  of  studies  proposed  potential  solutions  to  the  problem  of  leaving  hospital  AMA  among  PWUD.  The   empirical   studies   presented   in   Chapters   3   and   4   explored   healthcare  experiences   in   hospital   settings   among   PWUD   patients.   Specifically,   employing   a  modified   version   of   the   Risk   Environment   Framework,   Chapter   3   sought   to   identify   86 various   intersecting   socio-­‐‑demographic,   behavioural   and   contextual   influences   that  shape   discharge   AMA   experiences   among   PWUD.   The   study   found   that   43%   of   a  sample   of   hospitalized   PWUD   left   hospital   AMA   over   a   six-­‐‑year   period.   In  multivariable   analyses,   leaving   hospital   AMA   was   positively   and   independently  associated  with  various  risks  for  drug-­‐‑  and  health-­‐‑related  harm,  including  younger  age,  frequent  heroin  injection  and  incarceration,  while  stable  employment,  as  hypothesized,  was  negatively  associated  with  the  outcome.    Chapter   4   examined   the   effect   of   being   denied   pain   medication   by   healthcare  providers   on   illicit   drug   use   in   hospital.  A   large   proportion   of   PWUD   reported   both  having  been  denied  pain  medication  by  a  healthcare  provider  (48%)  and  illicit  drug  use  in   hospital   (44%).   In   adjusted   analyses,   having   been   denied   pain   medication   was  positively   and   independently   associated   with   using   illicit   drug   in   hospital.   Taken  together   with   Chapter   3,   these   findings   are   consistent   with   previous   research   that  described  hospitals  as  a  risk  environment  for  PWUD.20  As  well,  the  findings  highlighted  the   importance   of   addressing   the   contextual   circumstances   surrounding   PWUD   as   a  means  of  preventing  the  risks  and  harm  that  these  individuals  sometimes  experience  in  hospital  settings.  The   quantitative   analyses   in   Chapters   5   and   6   explored   possible   programs   and  interventions  with  potential   to   reduce   the  burden  of   leaving  hospital  AMA  and  high-­‐‑risk  illicit  drug  use  in  hospital  among  PWUD  patients.  Specifically,  Chapter  5  examined   87 the   impact   of   the  DPC,   an  HIV/AIDS   integrated   health   program   operating   close   to   a  hospital,   on   leaving   that   hospital   AMA   among   HIV-­‐‑positive   PWUD.   It   was  hypothesized  that  being  a  client  of  the  DPC  would  reduce  the  odds  of  leaving  hospital  AMA.  A  multivariable  model   concluded   that   indeed,   DPC   clients  were   less   likely   to  have   left  hospital  AMA  compared   to   those  who  were  not  DPC  clients.  These   findings  suggest   that   the   provision   of   a   broad   range   of   clinical,   harm   reduction   and   support  services  through  an  innovative  HIV/AIDS-­‐‑focused  adult  integrated  health  program  may  curb   the   rate   at   which   PWUD   leave   hospital   prematurely   and   thereby   reduce   the  incidence  of  preventable  morbidity  and  mortality  resulting  from  leaving  care  AMA.    Chapter  6   identified   the  prevalence  and  correlates  of  willingness   to  access  an   in-­‐‑hospital   SIF.   In   this   study,   68%   of   a   sample   of   hospitalized   PWUD   reported   their  willingness   to  access   such  a   facility   if   it  were  available.  The   results  of  a  multivariable  analysis   indicated   that   various   factors   were   positively   and   independently   associated  with  willingness  to  access  an  in-­‐‑hospital  SIF,  including  frequent  heroin  injection,  having  used   illicit   drugs   in   hospital   and   having   recently   used   a   community-­‐‑based   SIF.   The  findings   highlight   the   potential   of   in-­‐‑hospital   SIFs   to   complement   existing   harm  reduction  programs  that  serve  IDU  and  may  minimize  the  harms  associated  with  high-­‐‑risk  illicit  drug  use  in  hospital.  Along  with  the  findings  from  Chapter  5,  these  findings  demonstrate  that  an  integrated  harm  reduction  approach  to  the  provision  of  healthcare   88 services   for   PWUD   may   be   most   effective   for   reducing   the   risks   these   individuals  contend  with  in  hospital  settings.  Guided  by   the  Risk  Environment  Framework,   this  dissertation   investigated  how  intersecting   individual   and   select   contextual   factors   shape   PWUD   patients’   broader  health   service-­‐‑related   outcomes,   including   patient   discharge   AMA   and   high-­‐‑risk   in-­‐‑hospital   illicit  drug  use.  Potential   interventions  that   incorporate  a  harm  reduction  and  integrated   approach   to   care   for   PWUD   patients   admitted   to   acute   care   were   also  identified.  The  current  dissertation  presented   four  separate  empirical  analyses  of  data  gathered   through   the   VIDUS   and   ACCESS   studies,   two   longstanding   prospective  cohort   studies   of   HIV-­‐‑negative   and   HIV-­‐‑positive   PWUD.   Findings   from   this  dissertation   add   to   the   growing   body   of   literature   investigating   the   experiences   of  PWUD  patients  in  hospitals  as  well  as  possible  solutions  to  address  the  issues  they  face  in  these  settings.    7.2 Study  strengths  and  unique  contributions  This   dissertation   has   several   strengths   and   makes   unique   contributions   to   the  scientific   literature   focused  on   the  delivery  and  efficacy  of  hospital   services  and   illicit  drug   use.   First,   this   dissertation   advances   the   field   of   health   services   research   by  employing  a  conceptual  framing  that  is  relatively  novel  to  the  study  of  hospital  use  by  PWUD.   The   approach   used   diverges   from   previous   research   focused   mainly   on   89 individual   and   behavioural   factors   by   accounting   for   the   dynamic   interplay   between  these   factors   and   select   social,   structural,   and   physical   environments.   The   current  dissertation   adopts   a  modified   version   of   the   Risk   Environment   Framework   that   has  largely  been  absent  from  the  literature  on  hospital  service  utilization  and  demonstrates  how   certain   contextual   factors   contribute   to   leaving   hospital   AMA   among   PWUD.  Specifically,   in   Chapter   3,   various   socio-­‐‑demographic,   behavioural   and   individual  exposures   to   social   and   structural   factors   such   as   younger   age,   high   severity   of   drug  dependence,   unstable   employment   and   incarceration   emerged   as   key   themes  characterizing  individuals  who  are  highly  vulnerable  to  leaving  hospital  AMA.  Indeed,  the   capacity   to   link   hospital   administrative   data   to   time-­‐‑updated   cohort   studies  afforded   a   more   effective   determination   of   which   risk   or   protective   factors   were  associated  with  this  phenomenon.    Second,   Chapters   3   and   5   in   this   dissertation   made   use   of   two   comprehensive  datasets:   a   comprehensive   hospital   admission   and   discharge   database;   and   a  comprehensive  HIV  care  and   treatment  database   from  the  province’s   sole  provider  of  HIV/AIDS   treatment   and   care   (Chapter   5   only).   The   integration   of   hard  measures   of  discharge   status   as   well   as   complete   prospective   clinical   profiles   including   CD4   cell  counts   and   HIV-­‐‑1   plasma   viral   loads   with   information   on   socio-­‐‑demographic,  behavioural  and  contextual  factors  produced  a  dataset  with  high  validity  for  modeling  outcomes  related  to  hospital  care  among  PWUD.  Moreover,  given  the  universal  no-­‐‑cost   90 access   to   all   essential   medical   care   in   this   setting,   including   HIV   care,   the   resulting  analyses   are   largely   free   of   the   possible   confounding   influences   of   personal   financial  ability.  Third,   the   quantitative   analyses   in   this   dissertation   are   driven   by   data   derived  from   a   long-­‐‑running   cohort   of   community-­‐‑recruited   PWUD.   Analyses   presented   in  Chapters   3   and   5   use   longitudinal   statistical  modeling   techniques  with   time-­‐‑updated  measures   (e.g.,  GEE   regression  methods)   and   stand   in   contrast   to  many  other   studies  which   have   relied   on   cross-­‐‑sectional   modeling   procedures   using   data   solely   from  retrospective  medical   chart   reviews.  Although  no  claims  are  made  with   regard   to   the  generalizability  of  the  findings  to  settings  outside  of  Vancouver,   it   is  believed  that  the  analyses   presented   in   this   dissertation   are   analytically   strong   as   the   candidate  accounted   for   clustering  within  and  between  participants.   It   should  be  noted   that   the  analyses  in  Chapters  4  and  6  did  not  employ  longitudinal  methods;  specifically,  for  the  analyses   presented   in   Chapter   4,   the   candidate   was   only   able   to   conduct   a   cross-­‐‑sectional  study  since  the  main  explanatory  and  outcome  variable  of  interest  were  only  introduced  and  asked  in  a  recent  follow-­‐‑up  questionnaire.  As  well,  in  Chapter  6,  it  was  decided  that  a  cross-­‐‑sectional  study  was  sufficient  for  conducting  a  preliminary  needs  assessment   to   examine   the   prevalence   of   and   factors   associated   with   willingness   to  access  an  in-­‐‑hospital  SIF.     91 Lastly,   the   analyses   in   this   dissertation   strongly   support   two   novel   integrated  harm  reduction  programs  that  have  the  potential  to  minimize  the  negative  experiences  in  hospital  care  that  PWUD  often  contend  with.  As  described  in  Chapters  4  and  6,  the  candidate  found  that  an  integrated  health  program  that  incorporates  a  harm  reduction  component  may  be  an  important  tool  to  consider  in  hospital  settings.  These  findings  are  important  contributions  to  the  existing  literature  given  that  only  a  few  scientific  studies  to  date  have  explored  potential  interventions  to  reduce  the  rate  at  which  PWUD  leave  hospital  prematurely.12,13   Empirical   evidence  of   this   kind   is  useful   for   informing  drug  and  hospital  policies,   especially  given  a  growing   recognition  among  key   stakeholders  (e.g.,   scholars,   policy   makers,   administrators,   healthcare   providers,   members   of   the  affected  community)   that  harm  reduction  principles  are  essential  components  of   these  policies.    7.3 Limitations  As   with   all   research,   the   research   presented   in   this   dissertation   has   several  limitations.  While   the   limitations   of   each   study   have   been   described   in   detail   in   the  individual  chapters,  those  common  to  all  studies  are  presented  here.  First,  participants  in   the   VIDUS   and   ACCESS   studies   were   not   randomly   sampled,   and   therefore   the  findings  may  not  be  representative  of  or  generalizable  to  PWUD  populations  outside  of  Vancouver.  It  is  noteworthy  that  randomization  methods  cannot  be  used  to  generate  a   92 representative   sample  given   that   comprehensive   lists   of  PWUD   in  Vancouver   are  not  available.   Second,   several   variables   of   interest   were   measured   via   self-­‐‑report,   which  could   introduce   reporting   biases   such   as   recall   bias   and   social-­‐‑desirability   reporting.  However,  study  interviewers  employed  several  measures  to  minimize  the  likelihood  of  bias,   including   asking   sensitive   questions   towards   the   end   of   the   interview   to   allow  rapport   to   be   established   between   the   interviewer   and   participant.   As   indicated  previously,   Chapters   3   and   5   of   this   dissertation   also   included   hard   administrative  measures   of   hospital   discharge   AMA   and   HIV   clinical   measures   (Chapter   5)   to  minimize  biases  that  may  occur  from  self-­‐‑reporting.  Nevertheless,  a  number  of  studies  have   demonstrated   that   self-­‐‑reports   from   PWUD   are   reliable.131–133   Third,   due   to   the  nature  of  observational  studies,  the  possible  influence  of  confounding  variables  on  the  outcomes  cannot  be  ignored.  In  Chapters  4  and  5,  the  candidate  conducted  confounding  models   and   aimed   to   minimize   the   effect   of   residual   or   unmeasured   confounding  analytically  by  using  a  multivariable  modeling  procedure  that  systematically  included  or  excluded  secondary  explanatory  variables  based  on  their  statistical  effect  on  both  the  primary   explanatory   and   outcome   variable   of   interest.   As   well,   the   candidate  constructed   directed   acyclic   graphs   to   visualize   and   confirm   that   the   potential  confounding   variables   were   not   in   the   causal   pathway.   In   addition,   the   candidate  included   variables   that   are   known   confounders   from   the   literature   despite   their   93 statistical  significance  in  the  study.  Lastly,  temporality  could  not  be  truly  assessed  and  relatedly,  the  candidate  was  unable  to  determine  causation  from  this  research.      7.4 Recommendations  Specific   recommendations  resulting   from  each  empirical  analysis  are   included   in  Chapters   3-­‐‑6.   This   section   considers   all   study   findings   as   a   whole   and   includes  recommendations  for  public  health-­‐‑  or  policy-­‐‑based  interventions  in  Vancouver.      The   findings   presented   herein   clearly   indicate   that   hospitals   may   be   an  environment  within  which  PWUD  patients  are  at  risk  of  experiencing  various  negative  health  outcomes  despite  the  fact  that  essential  care  for  a  variety  of  illnesses  and  injuries  are  provided  within   such   settings.  These   findings  underscore   the  value  of   employing  the  Risk  Environment  Framework  to  conceptualize  the  effect  of  various  individual-­‐‑  and  contextual-­‐‑level  exposures  on  hospital  service  outcomes.  Specifically,   the  findings  add  to   the  existing   literature   that  utilizes   the  Risk  Environment  Framework   to  understand  barriers   to   accessing   community-­‐‑based   healthcare   and   harm   reduction   services   by  demonstrating  that  this  particular  framework  can  also  be  applied  to  hospital  settings.51,52  Therefore,   it   is   important   to   continue   to   apply   the   Risk   Environment   Framework   to  outcomes   related   to   health   service   and   hospital   use,   and   to   investigate   how   specific  aspects   of   this   environment   could   be   modified   to   improve   these   outcomes   among  PWUD.     94 Consistent   with   previous   research   that   demonstrated   the   presence   of   stigma,  discrimination  and  negative  stereotypes  against  PWUD  in  some  healthcare  settings,26,27  the   findings   from   the   study   suggest   potential   value   in   educating   physicians   and  healthcare   workers   about   addiction   broadly,   as   well   as   recent   developments   in  addiction   medicine.   An   improved   understanding   of   addictive   behaviours   and   its  neurobiology,   as   well   as   the   contextual   circumstances   surrounding   addiction,   will  enable  physicians  to  more  effectively  identify  and  respond  to  issues  relating  to  ongoing  substance  use  in  patients.  The  ability  to  do  so  will  likely  improve  the  medical  outcomes  associated   with   the   many   comorbid   issues   that   PWUD   patients   contend   with   (e.g.,  chronic   pain,   HIV,   HCV).109   Encouragingly,   there   is   growing   interest   in   the  development  and  implementation  of  addiction  medicine  training  programs  for  medical  residents  and  physicians  to  gain  expertise  in  treating  and  preventing  substance  misuse  and   related   harms.134,135   Efforts   that   continue   to   prioritize   specialized   education   in  addiction  medicine  may  serve  to  better  address  the  needs  of  PWUD  and  minimize  the  health  risks  and  harms  that  PWUD  face  in  hospital  and  other  healthcare  settings.  The   findings   are   consistent  with   previous   calls  made   by   researchers   and   public  health   advocates   that   point   to   the   need   for   an   integrated   harm   reduction-­‐‑based  approach  to  care  for  PWUD  in  hospital  settings.104,110  As  discussed  in  Chapters  5  and  6,  the   implementation   of   a   harm   reduction  model  within   hospital   settings  may   provide  opportunities   for   PWUD   to   obtain   a   variety   of   support   and   care   in   one   convenient   95 location,   and   thus   improve   access,   utilization   and   retention   in   these   settings.   In  particular,   efforts   to   provide   sterile   drug-­‐‑using   paraphernalia   and   establish   an   in-­‐‑hospital  SIF  may  be  useful  to  address  the  harms  associated  with  leaving  hospital  AMA  and   illicit   drug   use   in   hospital.   In   addition,   this   type   of   approach  may   provide   cost-­‐‑effective   treatment   and   care,   and  may   improve   patient-­‐‑physician   interactions   and   by  consequence  better  health  outcomes  for  PWUD.111,123  Given  the  potential  benefits  of  an  integrated  harm  reduction  approach   to  hospital  care   for  PWUD,   initiating  discussions  around   this   topic   and   the   development   of   these   types   of   programs   among   key  stakeholders,   including   healthcare  workers,   patients,   and   administrative   staff,   should  be   prioritized.   Indeed,   this   would   ensure   that   hospitals   are   providing   care   that   is  consistent   with   international   guidelines,   including   those   from   the   World   Health  Organization  and  the  Joint  United  Nations  Programme  on  HIV/AIDS  which  call  for  the  implementation   of   a   range   of   harm   reduction   and   prevention   programs,   including  needle  and  syringe  distribution  and  opioid  substitution  treatments.136      7.5 Future  research  The   studies   presented   herein   provide   a   body   of   evidence   that   demonstrate   the  risks   that   PWUD   experience   in   hospital   settings   and   offer   potential   solutions   to  minimize  this  concern.  While  the  findings  suggest  new  directions  for  hospital  and  drug   96 policy,  more  research  is  needed  to  strengthen  evidence-­‐‑informed  recommendations  for  these  policies.  This   dissertation   focused   on   experiences   of   PWUD   in   hospital   care   but   did   not  explore   the   perspectives   of   healthcare   providers   and   other   patients   who   do   not   use  illicit  drugs.  It   is  important  to  examine  the  views  and  experiences  of  individuals  other  than   PWUD   patients   to   help   inform   the   development   of   effective   policy   and  programming   to   reduce   the   harms   that   PWUD   contend   with   in   these   settings.  Accordingly,   future   research   should   include   physicians,   nurses   and   other   healthcare  providers  who  are  involved  in  the  management  and  day-­‐‑to-­‐‑day  operation  of  hospitals.  As   described   earlier,   a   number   of   studies   have   identified   negative   attitudes   around  illicit  drug  use  among  some  healthcare  providers,  as  well  as  challenges  associated  with  finding   a   balance   between   prescribing   adequate   pain   management   and   exacerbating  drug   use   habits.26,102,125   Exploring   the   experiences   of   key   stakeholders   who   have   an  influence  on  hospital  policies  and  clinical  guidelines  using  in-­‐‑depth  qualitative  research  methods  may  be  useful   for   identifying  opportunities   and   challenges   to   reshaping   the  way  PWUD  are  cared  for  in  hospitals.  The   findings   from   this   dissertation   touch   on   the   importance   of   addressing   the  issue   of   pain   among   PWUD   patients.   Consistent   with   previous   research   suggesting  inadequate   pain   management   among   hospitalized   PWUD,   the   finding   that   a   large  proportion   of   PWUD   are   denied   pain   medication   and   that   pain   may   play   a   role   in   97 increasing   the   risk   of   PWUD  patients   leaving   hospital   AMA   is   a   significant   concern.  While   the   issue   of   pain   management   is   complex   and   beyond   the   scope   of   this  dissertation,  it  should  be  a  topic  for  future  research  and  further  discussion.  Specifically,  future  studies  should  seek  to  investigate  prescribing  patterns  of  pain  medication  among  physicians  and  to  understand  how  to  more  effectively  prescribe  pain  to  PWUD  patients,  considering   a   variety   of   factors   such   as   tolerance   and   withdrawal   symptoms.   Along  these   lines,   intervention   studies   that   aim   to   monitor   and   optimize   access   to   pain  medication  may  be  an  important  avenue  of  research  to  pursue.  Indeed,  there  have  been  recent   calls   for   opioid   stewardship   programs,   modeled   after   successful   antibiotic  stewardship  programs  for  controlling  the  overprescribing  of  antibiotic  medications,137–139  for   addressing   inadequate   and   possibly   inappropriate   prescribing   of   pain  medication  for  patients.140–143  It  was   found   that   the   integration   of   a   harm   reduction   approach  within   hospital  care  may  minimize   the   risks  and  harms   that  PWUD  experience   in   these  settings.  This  finding   is  particularly   timely  given  that  St.  Paul’s  Hospital’s  AIDS  ward   is  now  being  repurposed   to   focus   on   a   broader   range   of   treatments   related   to   co-­‐‑morbidities   of  HIV/AIDS,   including  addiction,  due   to   the   rapidly  declining  number  of  people   living  with  AIDS  and   severity  of  AIDS   cases.144,145  Public  health   advocates  may  benefit   from  feasibility  studies   to  estimate   the  potential  use  of   such   integrated  programs  and  what  these  programs  would  look  like  if  they  were  to  be  implemented  in  hospital  settings.     98   7.6 Conclusions  This  dissertation  examined  PWUD’s  experiences  in  hospital  care,  highlighting  the  important   role   that   contextual   factors   play   in   enabling   or   impeding   utilization   and  retention   in   hospital   care   among   this   population.   Specifically,   it   makes   a   significant  contribution  to  this  area  by  demonstrating  that,  in  addition  to  its  current  use  as  a  model  to  measure  vulnerability  to  HIV  infection  and  related  outcomes,  the  Risk  Environment  Framework   can   also   serve   as   a   valuable   conceptual   model   to   study   hospital-­‐‑related  outcomes   among   PWUD.   Various   contextual   factors   such   as   access   to   stable  employment,   exposure   to   the   criminal   justice   system   and   access   to   appropriate   pain  management  are  challenges  that  PWUD  contend  with  when  seeking  the  care  they  need  in   hospital   settings.   Furthermore,   this   dissertation   aimed   to   minimize   the   gap   that  currently  exists  in  the  literature  around  specific  interventions  that  address  the  obstacles  that  PWUD  patients  face  in  hospitals.  Specifically,  it  pointed  to  the  potential  value  of  an  integrated   harm   reduction   approach   to   hospital   care,   including   an   in-­‐‑hospital   SIF.  Taken   together,   the   findings   not   only   highlight   the   need   to   recognize   and   address  substance  abuse  issues  soon  following  hospital  admission,  but  also  offer  critical  insight  into   how   incorporating   harm   reduction   models   in   hospital   settings   can   promote  positive  health  outcomes  by  retaining  PWUD  in  care.        99 References  1.     Kerr  T,  Wood  E,  Grafstein  E,  Ishida  T,  Shannon  K,  Lai  C,  et  al.  High  rates  of  primary  care  and  emergency  department  use  among  injection  drug  users  in  Vancouver.  J  Public  Health.  2005;27(1):62–6.    2.     Palepu  A,  Strathdee  SA,  Hogg  RS,  Anis  AH,  Rae  S,  Cornelisse  PGA,  et  al.  The  social  determinants  of  emergency  department  and  hospital  use  by  injection  drug  users  in  Canada.  J  Urban  Health  Bull  N  Y  Acad  Med.  1999;76(4):409–18.    3.     Gibson  A,  Randall  D,  Degenhardt  L.  The  increasing  mortality  burden  of  liver  disease  among  opioid-­‐‑dependent  people:  cohort  study.  Addiction.  2011;106(12):2186–92.    4.     Lloyd-­‐‑Smith  E,  Wood  E,  Zhang  R,  Tyndall  MW,  Sheps  S,  Montaner  JSG,  et  al.  Determinants  of  hospitalization  for  a  cutaneous  injection-­‐‑related  infection  among  injection  drug  users:  a  cohort  study.  BMC  Public  Health.  2010;10:327.    5.     Davidson  PJ,  McLean  RL,  Kral  AH,  Gleghorn  AA,  Edlin  BR,  Moss  AR.  Fatal  heroin-­‐‑related  overdose  in  San  Francisco,  1997-­‐‑2000:  a  case  for  targeted  intervention.  J  Urban  Health  Bull  N  Y  Acad  Med.  2003;80(2):261–73.    6.     Degenhardt  L,  Singleton  J,  Calabria  B,  McLaren  J,  Kerr  T,  Mehta  S,  et  al.  Mortality  among  cocaine  users:  A  systematic  review  of  cohort  studies.  Drug  Alcohol  Depend.  2011;113(2–3):88–95.    7.     Binswanger  IA,  Kral  AH,  Bluthenthal  RN,  Rybold  DJ,  Edlin  BR.  High  prevalence  of  abscesses  and  cellulitis  among  community-­‐‑recruited  injection  drug  users  in  San  Francisco.  Clin  Infect  Dis  Off  Publ  Infect  Dis  Soc  Am.  2000;30(3):579–81.    8.     Ndiaye  B,  Salleron  J,  Vincent  A,  Bataille  P,  Bonnevie  F,  Choisy  P,  et  al.  Factors  associated  with  presentation  to  care  with  advanced  HIV  disease  in  Brussels  and  Northern  France:  1997-­‐‑2007.  BMC  Infect  Dis.  2011;11(1):11.    9.     Palepu  A,  Tyndall  MW,  Leon  H,  Muller  J,  O’Shaughnessy  MV,  Schechter  MT,  et  al.  Hospital  Utilization  and  Costs  in  a  Cohort  of  Injection  Drug  Users.  Can  Med  Assoc  J.  2001;165(4):415–20.    10.     Wood  E,  Kerr  T,  Spittal  P,  Tyndall  M,  O’Shaughnessy  M,  Schechter  M.  The  health  care  and  fiscal  costs  of  the  illicit  drug  use  epidemic:  The  impact  of  conventional  drug  control  strategies.  BC  Med  J.  2003;45(3):128–34.     100 11.     McNeil  R,  Small  W,  Wood  E,  Kerr  T.  Hospitals  as  a  “risk  environment”:  An  ethno-­‐‑epidemiological  study  of  voluntary  and  involuntary  discharge  from  hospital  against  medical  advice  among  people  who  inject  drugs.  Soc  Sci  Med.  2014  Mar;105C:59–66.    12.     Chan  AC,  Palepu  A,  Guh  DP,  Sun  H,  Schechter  MT,  O’Shaughnessy  MV,  et  al.  HIV-­‐‑positive  injection  drug  users  who  leave  the  hospital  against  medical  advice:  the  mitigating  role  of  methadone  and  social  support.  J  Acquir  Immune  Defic  Syndr.  2004;35(1):56–9.    13.     Jafari  S,  Joe  R,  Elliot  D,  Nagji  A,  Hayden  S,  Marsh  DC.  A  community  care  model  of  intravenous  antibiotic  therapy  for  injection  drug  users  with  deep  tissue  infection  for  “reduce  leaving  against  medical  advice”.  Int  J  Ment  Health  Addict.  2014;13:49–58.    14.     Riddell  C,  Riddell  R.  Welfare  Checks,  Drug  Consumption,  and  Health:  Evidence  from  Vancouver  Injection  Drug  Users.  J  Hum  Resour.  2006;41(1):138–61.    15.     Anis  AH,  Sun  H,  Guh  DP,  Palepu  A,  Schechter  MT,  O’Shaughnessy  MV.  Leaving  hospital  against  medical  advice  among  HIV-­‐‑positive  patients.  CMAJ  Can  Med  Assoc  J.  2002;167(6):633–7.    16.     Hwang  SW,  Li  J,  Gupta  R,  Chien  V,  Martin  RE.  What  happens  to  patients  who  leave  hospital  against  medical  advice?  Can  Med  Assoc  J.  2003;168(4):417–20.    17.     Southern  WN,  Nahvi  S,  Arnsten  JH.  Increased  risk  of  mortality  and  readmission  among  patients  discharged  against  medical  advice.  Am  J  Med.  2012;125(6):594–602.    18.     Glasgow  JM,  Vaughn-­‐‑Sarrazin  M,  Kaboli  PJ.  Leaving  Against  Medical  Advice  (AMA):  Risk  of  30-­‐‑Day  Mortality  and  Hospital  Readmission.  J  Gen  Intern  Med.  2010;25(9):926–9.    19.     Aliyu  ZY.  Discharge  against  medical  advice:  sociodemographic,  clinical  and  financial  perspectives.  Int  J  Clin  Pract.  2002;56(5):325–7.    20.     McNeil  R,  Small  W,  Wood  E,  Kerr  T.  Hospitals  as  a  “risk  environment”:  an  ethno-­‐‑epidemiological  study  of  voluntary  and  involuntary  discharge  from  hospital  against  medical  advice  among  people  who  inject  drugs.  Soc  Sci  Med.  2014;105:59–66.    21.     Penchansky  R,  Thomas  JW.  The  concept  of  access:  definition  and  relationship  to  consumer  satisfaction.  Med  Care.  1981;19(2):127–40.     101 22.     McCoy  CB,  Metsch  LR,  Chitwood  DD,  Miles  C.  Drug  use  and  barriers  to  use  of  health  care  services.  Subst  Use  Misuse.  2001;36(6-­‐‑7):789–806.    23.     Arbelaez  C,  Wright  EA,  Losina  E,  Millen  JC,  Kimmel  S,  Dooley  M,  et  al.  Emergency  provider  attitudes  and  barriers  to  universal  HIV  testing  in  the  emergency  department.  J  Emerg  Med.  2012;42(1):7–14.    24.     Huxtable  CA,  Roberts  LJ,  Somogyi  AA,  MacIntyre  PE.  Acute  pain  management  in  opioid-­‐‑tolerant  patients:  a  growing  challenge.  Anaesth  Intensive  Care.  2011;39(5):804–23.    25.     McCreaddie  M,  Lyons  I,  Watt  D,  Ewing  E,  Croft  J,  Smith  M,  et  al.  Routines  and  rituals:  a  grounded  theory  of  the  pain  management  of  drug  users  in  acute  care  settings.  J  Clin  Nurs.  2010;19(19-­‐‑20):2730–40.    26.     Van  Boekel  LC,  Brouwers  EPM,  van  Weeghel  J,  Garretsen  HFL.  Stigma  among  health  professionals  towards  patients  with  substance  use  disorders  and  its  consequences  for  healthcare  delivery:  Systematic  review.  Drug  Alcohol  Depend.  2013;131(1–2):23–35.    27.     Sayles  JN,  Wong  MD,  Kinsler  JJ,  Martins  D,  Cunningham  WE.  The  Association  of  Stigma  with  Self-­‐‑Reported  Access  to  Medical  Care  and  Antiretroviral  Therapy  Adherence  in  Persons  Living  with  HIV/AIDS.  J  Gen  Intern  Med.  2009;24(10):1101–8.    28.     Doab  A,  Treloar  C,  Dore  GJ.  Knowledge  and  attitudes  about  treatment  for  hepatitis  C  virus  infection  and  barriers  to  treatment  among  current  injection  drug  users  in  Australia.  Clin  Infect  Dis.  2005;40  Suppl  5:S313–320.    29.     Rhodes  T.  The  “risk  environment”:  a  framework  for  understanding  and  reducing  drug-­‐‑related  harm.  Int  J  Drug  Policy.  2002;13:85–94.    30.     Rhodes  T.  Risk  environments  and  drug  harms:  A  social  science  for  harm  reduction  approach.  Int  J  Drug  Policy.  2009;20:193–201.    31.     Strathdee  SA,  Hallett  TB,  Bobrova  N,  Rhodes  T,  Booth  R,  Abdool  R,  et  al.  HIV  and  risk  environment  for  injecting  drug  users:  the  past,  present,  and  future.  Lancet.  2010;376(9737):268–84.    32.     Strathdee  SA,  Patrick  DM,  Currie  SL,  Cornelisse  PG,  Rekart  ML,  Montaner  JS,  et  al.  Needle  exchange  is  not  enough:  lessons  from  the  Vancouver  injecting  drug  use  study.  AIDS.  1997;11(8):F59–65.     102 33.     Wood  E,  Kerr  T.  What  do  you  do  when  you  hit  rock  bottom?  Responding  to  drugs  in  the  city  of  Vancouver.  Int  J  Drug  Policy.  2006;17:55–60.    34.     Small  W,  Maher  L,  Lawlor  J,  Wood  E,  Shannon  K,  Kerr  T.  Injection  drug  users’  involvement  in  drug  dealing  in  the  downtown  eastside  of  Vancouver:  Social  organization  and  systemic  violence.  Int  J  Drug  Policy.  2013;24(5):479–87.    35.     McNeil  R,  Shannon  K,  Shaver  L,  Kerr  T,  Small  W.  Negotiating  place  and  gendered  violence  in  Canada’s  largest  open  drug  scene.  Int  J  Drug  Policy.  2014;25(3):608–15.    36.     Shannon  K  CJ.  VIolence,  condom  negotiation,  and  hiv/sti  risk  among  sex  workers.  JAMA.  2010;304(5):573–4.    37.     Patrick  DM,  Strathdee  SA,  Archibald  CP,  Ofner  M,  Craib  KJ,  Cornelisse  PG,  et  al.  Determinants  of  HIV  seroconversion  in  injection  drug  users  during  a  period  of  rising  prevalence  in  Vancouver.  Int  J  STD  AIDS.  1997;8(7):437–45.    38.     Urban  Health  Research  Initiative  of  the  British  Columbia  Centre  for  Excellence  in  HIV/AIDS.  Drug  Situation  in  Vancouver.  Vancouver:  British  Columbia  Centre  for  Excellence  in  HIV/AIDS;  2013.    39.     Kerr  T,  Small  W,  Buchner  C,  Zhang  R,  Li  K,  Montaner  J,  et  al.  Syringe  Sharing  and  HIV  Incidence  Among  Injection  Drug  Users  and  Increased  Access  to  Sterile  Syringes.  Am  J  Public  Health.  2010;100(8):1449–53.    40.     Fairbairn  N,  Milloy  M-­‐‑J,  Zhang  R,  Lai  C,  Grafstein  E,  Kerr  T,  et  al.  Emergency  Department  Utilization  among  a  Cohort  of  HIV-­‐‑positive  Injecting  Drug  Users  in  a  Canadian  Setting.  J  Emerg  Med.  2012;43(2):236–43.    41.     BC  Harm  Reduction  Strategies  and  Services.  The  History  of  Harm  Reduction  in  British  Columbia  [Internet].  British  Columbia  Centre  for  Disease  Control;  2012  [cited  2014  Dec  12].  Available  from:  http://www.bccdc.ca/NR/rdonlyres/7B9F63E8-­‐‑7D50-­‐‑4985-­‐‑9BC2-­‐‑38D78C4E77AD/0/UpdatedBCHarmReductionDocumentAug2012JAB_final.pdf  42.     Nosyk  B,  Marsh  DC,  Sun  H,  Schechter  MT,  Anis  AH.  Trends  in  methadone  maintenance  treatment  participation,  retention,  and  compliance  to  dosing  guidelines  in  British  Columbia,  Canada:  1996-­‐‑2006.  J  Subst  Abuse  Treat.  2010;39(1):22–31.    43.     Woo  A.  Report  says  B.C.  a  model  for  treating  opioid  addiction.  The  Globe  and  Mail  [Internet].  [cited  2015  Feb  17];  Available  from:   103 http://www.theglobeandmail.com/news/british-­‐‑columbia/report-­‐‑says-­‐‑bc-­‐‑a-­‐‑model-­‐‑for-­‐‑treating-­‐‑heroin-­‐‑addiction/article13628835/  44.     Wood  E,  Tyndall  MW,  Montaner  JS,  Kerr  T.  Summary  of  findings  from  the  evaluation  of  a  pilot  medically  supervised  safer  injecting  facility.  CMAJ  Can  Med  Assoc  J  J  Assoc  Medicale  Can.  2006;175(11):1399–404.    45.     Hogg  RS,  Heath  KV,  Yip  B,  Craib  KJ,  O’Shaughnessy  MV,  Schechter  MT,  et  al.  Improved  survival  among  HIV-­‐‑infected  individuals  following  initiation  of  antiretroviral  therapy.  JAMA.  1998;279(6):450–4.    46.     Hogg  RS,  Yip  B,  Chan  KJ,  Wood  E,  Craib  KJ,  O’Shaughnessy  MV,  et  al.  Rates  of  disease  progression  by  baseline  CD4  cell  count  and  viral  load  after  initiating  triple-­‐‑drug  therapy.  J  Am  Med  Assoc.  2001;286(20):2568–77.    47.     Wood  E,  Schechter  MT,  Tyndall  MW,  Montaner  JS,  O’Shaughnessy  MV,  Hogg  RS.  Antiretroviral  medication  use  among  injection  drug  users:  two  potential  futures.  AIDS  Lond  Engl.  2000;14(9):1229–35.    48.     Rhodes  T,  Singer  M,  Bourgois  P,  Friedman  S,  Strathdee  S.  The  social  structural  production  of  HIV  risk  among  injecting  drug  users.  Soc  Sci  Med.  2005;61:1026–44.    49.     Koniak-­‐‑Griffin  D,  Stein  JA.  Predictors  of  sexual  risk  behaviors  among  adolescent  mothers  in  a  human  immunodeficiency  virus  prevention  program.  J  Adolesc  Health  Off  Publ  Soc  Adolesc  Med.  2006;38(3):297.e1–11.    50.     Celentano  DD,  Cohn  S,  Davis  RO,  Vlahov  D.  Self-­‐‑efficacy  estimates  for  drug  use  practices  predict  risk  reduction  among  injection  drug  users.  J  Urban  Health  Bull  N  Y  Acad  Med.  2002;79(2):245–56.    51.     Small  W,  Kerr  T,  Charette  J,  Schechter  M,  Spittal  P.  Impacts  of  intensified  police  activity  on  injection  drug  users:  Evidence  from  an  ethnographic  investigation.  Int  J  Drug  Policy.  2006;17:85–95.    52.     Shannon  K,  Rusch  M,  Shoveller  J,  Alexson  D,  Gibson  K,  Tyndall  M.  Mapping  violence  and  policing  as  an  environmental–structural  barrier  to  health  service  and  syringe  availability  among  substance-­‐‑using  women  in  street-­‐‑level  sex  work.  Int  J  Drug  Policy.  2008;19:140–7.    53.     MacArthur  GJ,  Minozzi  S,  Martin  N,  Vickerman  P,  Deren  S,  Bruneau  J,  et  al.  Opiate  substitution  treatment  and  HIV  transmission  in  people  who  inject  drugs:  systematic  review  and  meta-­‐‑analysis.  BMJ.  2012;345:e5945.     104 54.     Mathers  B,  Degenhardt  L,  Ali  H,  Wiessing  L,  Hickman  M,  Mattick  R,  et  al.  HIV  prevention,  treatment,  and  care  services  for  people  who  inject  drugs:  a  systematic  review  of  global,  regional,  and  national  coverage.  Lancet.  2010;375:1014–28.    55.     Palepu  A,  Milloy  M-­‐‑J,  Kerr  T,  Zhang  R,  Wood  E.  Homelessness  and  adherence  to  antiretroviral  therapy  among  a  cohort  of  HIV-­‐‑infected  injection  drug  users.  J  Urban  Health  Bull  N  Y  Acad  Med.  2011;88(3):545–55.    56.     Haber  PS,  Demirkol  A,  Lange  K,  Murnion  B.  Management  of  injecting  drug  users  admitted  to  hospital.  Lancet.  2009;374(9697):1284–93.    57.     McCreaddie  M,  Lyons  I,  Watt  D,  Ewing  E,  Croft  J,  Smith  M,  et  al.  Routines  and  rituals:  a  grounded  theory  of  the  pain  management  of  drug  users  in  acute  care  settings.  J  Clin  Nurs.  2010;19(19-­‐‑20):2730–40.    58.     Dr.  Peter  AIDS  Foundation.  Dr.  Peter  AIDS  Foundation  [Internet].  2011  [cited  2014  Jun  19].  Available  from:  http://www.drpeter.org/  59.     Wood  R,  Zettel  P,  Stewart  W.  Dr.  Peter  Centre:  Harm  Reduction  Nursing.  Canadian  Nurse  [Internet].  2003;99(5).  Available  from:  http://www.drpeter.org/media/canadiannurse-­‐‑may03.pdf  60.     Himelhoch  S,  Chander  G,  Fleishman  JA,  Hellinger  J,  Gaist  P,  Gebo  KA.  Access  to  HAART  and  utilization  of  inpatient  medical  hospital  services  among  HIV-­‐‑infected  patients  with  co-­‐‑occurring  serious  mental  illness  and  injection  drug  use.  Gen  Hosp  Psychiatry.  2007;29(6):518–25.    61.     Krüsi  A,  Small  W,  Wood  E,  Kerr  T.  An  integrated  supervised  injecting  program  within  a  care  facility  for  HIV-­‐‑positive  individuals:  a  qualitative  evaluation.  AIDS  Care.  2009;21(5):638–44.    62.     DeBeck  K,  Kerr  T,  Bird  L,  Zhang  R,  Marsh  D,  Tyndall  M,  et  al.  Injection  drug  use  cessation  and  use  of  North  America’s  first  medically  supervised  safer  injecting  facility.  Drug  Alcohol  Depend.  2011;113(2-­‐‑3):172–6.    63.     Stoltz  J-­‐‑A,  Wood  E,  Small  W,  Li  K,  Tyndall  M,  Montaner  J,  et  al.  Changes  in  injecting  practices  associated  with  the  use  of  a  medically  supervised  safer  injection  facility.  J  Public  Health.  2007;29(1):35–9.    64.     Marshall  BD,  Milloy  M-­‐‑J,  Wood  E,  Montaner  JS,  Kerr  T.  Reduction  in  overdose  mortality  after  the  opening  of  North  America’s  first  medically  supervised  safer   105 injecting  facility:  a  retrospective  population-­‐‑based  study.  Lancet.  2011;377(9775):1429–37.    65.     Weingart  SN,  Davis  RB,  Phillips  RS.  Patients  discharged  against  medical  advice  from  a  general  medicine  service.  J  Gen  Intern  Med.  1998;13(8):568–71.    66.     Moher  D,  Liberati  A,  Tetzlaff  J,  Altman  DG,  PRISMA  Group.  Preferred  reporting  items  for  systematic  reviews  and  meta-­‐‑analyses:  the  PRISMA  statement.  BMJ.  2009;339:b2535.    67.     Alfandre  DJ.  “I’m  Going  Home”:  Discharges  Against  Medical  Advice.  Mayo  Clin  Proc.  2009;84(3):255–60.    68.     Endicott  P,  Watson  B.  Interventions  to  improve  the  AMA-­‐‑discharge  rate  for  opiate-­‐‑addicted  patients.  J  Psychosoc  Nurs  Ment  Health  Serv.  1994;32(8):36–40.    69.     Kraut  A,  Fransoo  R,  Olafson  K,  Ramsey  CD,  Yogendran  M,  Garland  A.  A  population-­‐‑based  analysis  of  leaving  the  hospital  against  medical  advice:  incidence  and  associated  variables.  BMC  Health  Serv  Res.  2013;13:415.    70.     Fiscella  K,  Meldrum  S,  Franks  P.  Post  partum  discharge  against  medical  advice:  who  leaves  and  does  it  matter?  Matern  Child  Health  J.  2007;11(5):431–6.    71.     Saitz  R.  Discharges  against  medical  advice:  time  to  address  the  causes.  CMAJ  Can  Med  Assoc  J.  2002;167(6):647–8.    72.     Myers  RP,  Shaheen  AAM,  Hubbard  JN,  Kaplan  GG.  Characteristics  of  patients  with  cirrhosis  who  are  discharged  from  the  hospital  against  medical  advice.  Clin  Gastroenterol  Hepatol  Off  Clin  Pract  J  Am  Gastroenterol  Assoc.  2009;7(7):786–92.    73.     Jankowski  CB,  Drum  DE.  Diagnostic  correlates  of  discharge  against  medical  advice.  Arch  Gen  Psychiatry.  1977;34(2):153–5.    74.     Seaborn  Moyse  H,  Osmun  WE.  Discharges  against  medical  advice:  a  community  hospital’s  experience.  Can  J  Rural  Med  Off  J  Soc  Rural  Physicians  Can  J  Can  Médecine  Rurale  J  Off  Société  Médecine  Rurale  Can.  2004;9(3):148–53.    75.     Yong  TY,  Fok  JS,  Hakendorf  P,  Ben-­‐‑Tovim  D,  Thompson  CH,  Li  JY.  Characteristics  and  Outcomes  of  Discharges  Against  Medical  Advice  Among  Hospitalized  Patients.  Intern  Med  J  [Internet].  2013  [cited  2013  May  22];  Available  from:  http://onlinelibrary.wiley.com/doi/10.1111/imj.12109/abstract   106 76.     Smith  DB,  Telles  JL.  Discharges  against  medical  advice  at  regional  acute  care  hospitals.  Am  J  Public  Health.  1991;81(2):212–5.    77.     Tawk  R,  Freels  S,  Mullner  R.  Associations  of  mental,  and  medical  illnesses  with  against  medical  advice  discharges:  the  National  Hospital  Discharge  Survey,  1988-­‐‑2006.  Adm  Policy  Ment  Health.  2013;40(2):124–32.    78.     Choi  M,  Kim  H,  Qian  H,  Palepu  A.  Readmission  Rates  of  Patients  Discharged  against  Medical  Advice:  A  Matched  Cohort  Study.  PLoS  ONE.  2011;6(9):e24459.    79.     Palepu  A,  Sun  H,  Kuyper  L,  Schechter  MT,  O’Shaughnessy  MV,  Anis  AH.  Predictors  of  Early  Hospital  Readmission  in  HIV-­‐‑infected  Patients  with  Pneumonia.  J  Gen  Intern  Med.  2003;18(4):242–7.    80.     Van  Olphen  J,  Freudenberg  N,  Fortin  P,  Galea  S.  Community  Reentry:  Perceptions  of  People  with  Substance  Use  Problems  Returning  Home  from  New  York  City  Jails.  J  Urban  Health  Bull  N  Y  Acad  Med.  2006;83(3):372–81.    81.     Olphen  J  van,  Eliason  MJ,  Freudenberg  N,  Barnes  M.  Nowhere  to  go:  How  stigma  limits  the  options  of  female  drug  users  after  release  from  jail.  Subst  Abuse  Treat  Prev  Policy.  2009;4(1):10.    82.     Hines  S,  Theodorou  S,  Williamson  A,  Fong  D,  Curry  K.  Management  of  acute  pain  in  methadone  maintenance  therapy  in-­‐‑patients.  Drug  Alcohol  Rev.  2008;27(5):519–23.    83.     Woody  GE,  McLellan  AT,  O’Brien  CP,  Luborsky  L.  Addressing  psychiatric  comorbidity.  NIDA  Res  Monogr.  1991;106:152–66.    84.     Targum  SD,  Capodanno  AE,  Hoffman  HA,  Foudraine  C.  An  intervention  to  reduce  the  rate  of  hospital  discharges  against  medical  advice.  Am  J  Psychiatry.  1982;139(5):657–9.    85.     Holden  P,  Vogtsberger  KN,  Mohl  PC,  Fuller  DS.  Patients  who  leave  the  hospital  against  medical  advice:  the  role  of  the  psychiatric  consultant.  Psychosomatics.  1989;30(4):396–404.    86.     Lloyd-­‐‑Smith  E,  Kerr  T,  Hogg  RS,  Li  K,  Montaner  JS,  Wood  E.  Prevalence  and  correlates  of  abscesses  among  a  cohort  of  injection  drug  users.  Harm  Reduct  J.  2005;2:24.     107 87.     Sá  LC  de,  Araújo  TME  de,  Griep  RH,  Campelo  V,  Monteiro  CF  de  S.  Seroprevalence  of  hepatitis  C  and  factors  associated  with  this  in  crack  users.  Rev  Lat  Am  Enfermagem.  2013;21(6):1195–202.    88.     Sordo  L,  Indave  BI,  Barrio  G,  Degenhardt  L,  de  la  Fuente  L,  Bravo  MJ.  Cocaine  use  and  risk  of  stroke:  a  systematic  review.  Drug  Alcohol  Depend.  2014;142:1–13.    89.     Pan  W.  Akaike’s  Information  Criterion  in  Generalized  Estimating  Equations.  Biometrics.  2001;57(1):120–5.    90.     Carroll  IR,  Angst  MS,  Clark  JD.  Management  of  perioperative  pain  in  patients  chronically  consuming  opioids.  Reg  Anesth  Pain  Med.  2004;29:576–91.    91.     Voon  P,  Callon  C,  Nguyen  P,  Dobrer  S,  Montaner  J,  Wood  E,  et  al.  Self-­‐‑management  of  pain  among  people  who  inject  drugs  in  Vancouver.  Pain  Manag.  2014;4(1):27–35.    92.     Freudenberg  N,  Daniels  J,  Crum  M,  Perkins  T,  Richie  BE.  Coming  Home  From  Jail:  The  Social  and  Health  Consequences  of  Community  Reentry  for  Women,  Male  Adolescents,  and  Their  Families  and  Communities.  Am  J  Public  Health.  2005;95(10):1725–36.    93.     Tompkins  C,  Neale  J,  Sheard  L,  Wright  N.  Experiences  of  prison  among  injecting  drug  users  in  England:  A  qualitative  study.  Int  J  Prison  Health.  2007;3(3):189–203.    94.     DeBeck  K,  Shannon  K,  Wood  E,  Li  K,  Montaner  J,  Kerr  T.  Income  generating  activities  of  people  who  inject  drugs.  Drug  Alcohol  Depend.  2007;91(1):50–6.    95.     Richardson  L,  Wood  E,  Li  K,  Kerr  T.  Factors  associated  with  employment  among  a  cohort  of  injection  drug  users.  Drug  Alcohol  Rev.  2010;29(3):293–300.    96.     Ti  L,  Richardson  L,  DeBeck  K,  Nguyen  P,  Montaner  J,  Wood  E,  et  al.  The  impact  of  engagement  in  street-­‐‑based  income  generation  activities  on  stimulant  drug  use  cessation  among  people  who  inject  drugs.  Drug  Alcohol  Depend.  2014;141:58-­‐‑64.    97.     Steensma  C,  Boivin  J,  Blais  L,  Roy  E.  Cessation  of  Injecting  Drug  Use  Among  Street-­‐‑Based  Youth.  J  Urban  Health.  2005;82(4):622–37.    98.     Alexandre  PK,  French  MT.  Further  evidence  on  the  labor  market  effects  of  addiction:  chronic  drug  use  and  employment  in  metropolitan  Miami.  Contemp  Econ  Policy.  2004;22(3):382–93.    99.     DeSimone  J.  Illegal  Drug  Use  and  Employment.  J  Labor  Econ.  2002;20(4):952–77.     108 100.    Wen  CK,  Hudak  PL,  Hwang  SW.  Homeless  People’s  Perceptions  of  Welcomeness  and  Unwelcomeness  in  Healthcare  Encounters.  J  Gen  Intern  Med.  2007;22(7):1011–7.    101.    Breitbart  W,  Rosenfeld  B,  Passik  S,  Kaim  M,  Funesti-­‐‑Esch  J,  Stein  K.  A  comparison  of  pain  report  and  adequacy  of  analgesic  therapy  in  ambulatory  AIDS  patients  with  and  without  a  history  of  substance  abuse.  1997;72(1-­‐‑2):235-­‐‑243.    102.    Lebovits  AH,  Florence  I,  Bathina  R,  Hunko  V,  Fox  MT,  Bramble  CY.  Pain  knowledge  and  attitudes  of  healthcare  providers:  practice  characteristic  differences.  Clin  J  Pain.  1997;13(3):237–43.    103.    Voon  P,  Kerr  T.  “Nonmedical”  prescription  opioid  use  in  North  America:  a  call  for  priority  action.  Subst  Abuse  Treat  Prev  Policy.  2013;8:39.    104.    Rachlis  BS,  Kerr  T,  Montaner  JS,  Wood  E.  Harm  reduction  in  hospitals:  is  it  time?  Harm  Reduct  J.  2009;6(1):19.    105.    Chan  KY,  Stoove  M,  Sringernyuang  L,  Reidpath  D.  Stigmatization  of  AIDS  patients:  Disentangling  Thai  nursing  students’  attitudes  towards  HIV/AIDS,  drug  use,  and  commercial  sex.  AIDS  Behav.  2008;12:146–57.    106.    Merrill  JO,  Rhodes  LA,  Deyo  RA,  Marlatt  GA,  Bradley  KA.  Mutual  mistrust  in  the  medical  care  of  drug  users:  the  keys  to  the  “narc”  cabinet.  J  Gen  Intern  Med.  2002;17(5):327–33.    107.    College  of  Physicians  and  Surgeons  of  British  Columbia.  Prescription  Review  Committee  Prescribing  Principles  for  Chronic  Non-­‐‑Cancer  Pain  [Internet].  College  of  Physicians  and  Surgeons  of  British  Columbia;  2014  [cited  2014  Mar  11].  Available  from:  https://www.cpsbc.ca/files/pdf/PRC-­‐‑Prescribing-­‐‑Principles.pdf  108.    Chou  R,  Fanciullo  GJ,  Fine  PG,  Adler  JA,  Ballantyne  JC,  Davies  P,  et  al.  Clinical  Guidelines  for  the  Use  of  Chronic  Opioid  Therapy  in  Chronic  Noncancer  Pain.  J  Pain.  2009;10(2):113–130.e22.    109.    Wood  E,  Samet  JH,  Volkow  ND.  Physician  education  in  addiction  medicine.  JAMA  J  Am  Med  Assoc.  2013;310(16):1673–4.    110.    Kerr  T,  Ti  L.  Drug  use  in  hospitals:  Is  there  a  role  for  harm  reduction?  Hospital  News  [Internet].  2013  Oct  17  [cited  2014  Mar  11];  Available  from:  http://hospitalnews.com/drug-­‐‑use-­‐‑in-­‐‑hospitals-­‐‑is-­‐‑there-­‐‑a-­‐‑role-­‐‑for-­‐‑harm-­‐‑reduction/   109 111.    Grau  LE,  Arevalo  S,  Catchpool  C,  Heimer  R.  Expanding  Harm  Reduction  Services  Through  a  Wound  and  Abscess  Clinic.  Am  J  Public  Health.  2002;92(12):1915–7.    112.    Ministry  of  Housing  and  Social  Development,  Canada  Mortgage  and  Housing  Corporation,  City  of  Vancouver.  $14M  Reopens  Supporting  Housing  in  Downtown  Eastside  [Internet].  City  of  Vancouver;  2009.  Available  from:  http://www2.news.gov.bc.ca/news_releases_2005-­‐‑2009/2009HSD0002-­‐‑000006.htm  113.    Solai  S,  Dubois-­‐‑Arber  F,  Benninghoff  F,  Benaroyo  L.  Ethical  reflections  emerging  during  the  activity  of  a  low  threshold  facility  with  supervised  drug  consumption  room  in  Geneva,  Switzerland.  Int  J  Drug  Policy.  2006;17(1):17–22.    114.    Davis  M.  Integrating  a  supervised  injection  site  into  health  care  -­‐‑  and  community.  The  Globe  and  Mail  [Internet].  Vancouver;  2012;  Available  from:  http://www.theglobeandmail.com/globe-­‐‑debate/integrating-­‐‑a-­‐‑supervised-­‐‑injection-­‐‑site-­‐‑into-­‐‑health-­‐‑care-­‐‑-­‐‑-­‐‑and-­‐‑community/article4099569/  115.    Brook  M,  Hilty  DM,  Liu  W,  Hu  R,  Frye  MA.  Discharge  against  medical  advice  from  inpatient  psychiatric  treatment:  a  literature  review.  Psychiatr  Serv  Wash  DC.  2006;57(8):1192–8.    116.    Kerr  T,  Craib  KJ,  Gataric  N,  Hogg  RS.  Assessing  the  impact  of  an  adult  day  program  on  hospital  utilization  by  persons  living  with  HIV/AIDS.  J  Acquir  Immune  Defic  Syndr.  2002;31(1):117–9.    117.    Griffiths  H,  Mackinnon  D.  Dr.  Peter  Centre:  Removing  Barriers  to  Health  Care  Services.  Nursing  BC  [Internet].  2002  [cited  2014  Aug  6];  Available  from:  http://www.drpeter.org/media/nursingbc_dec02.pdf  118.    Salmon  AM,  Van  Beek  I,  Amin  J,  Kaldor  J,  Maher  L.  The  impact  of  a  supervised  injecting  facility  on  ambulance  call-­‐‑outs  in  Sydney,  Australia.  Addiction.  2010;105(4):676–83.    119.    Wood  E,  Kerr  T,  Small  W,  Li  K,  Marsh  DC,  Montaner  JSG,  et  al.  Changes  in  public  order  after  the  opening  of  a  medically  supervised  safer  injecting  facility  for  illicit  injection  drug  users.  Can  Med  Assoc  J.  2004;171(7):731–4.    120.    Dolan  K,  Kimber  J,  Fry  C,  Fitzgerald  J,  McDonald  D,  Trautmann  F.  Drug  consumption  facilities  in  Europe  and  the  establishment  of  supervised  injecting  centres  in  Australia.  Drug  Alcohol  Rev.  2000;19(3):337–46.     110 121.    Bayoumi  AM,  Zaric  GS.  The  cost-­‐‑effectiveness  of  Vancouver’s  supervised  injection  facility.  CMAJ  Can  Med  Assoc  J  J  Assoc  Medicale  Can.  2008;179(11):1143–51.    122.    DeBeck  K,  Wood  E,  Zhang  R,  Tyndall  M,  Montaner  J,  Kerr  T.  Police  and  public  health  partnerships:  Evidence  from  the  evaluation  of  Vancouver’s  supervised  injection  facility.  Subst  Abuse  Treat  Prev  Policy.  2008;3(11).    123.    Krüsi  A,  Small  W,  Wood  E,  Kerr  T.  An  integrated  supervised  injecting  program  within  a  care  facility  for  HIV-­‐‑positive  individuals:  a  qualitative  evaluation.  AIDS  Care.  2009;21(5):638–44.    124.    Compton  P,  Charuvastra  VC,  Ling  W.  Pain  intolerance  in  opioid-­‐‑maintained  former  opiate  addicts:  effect  of  long-­‐‑acting  maintenance  agent.  Drug  Alcohol  Depend.  2001;63(2):139–46.    125.    Baldacchino  A,  Gilchrist  G,  Fleming  R,  Bannister  J.  Guilty  until  proven  innocent:  a  qualitative  study  of  the  management  of  chronic  non-­‐‑cancer  pain  among  patients  with  a  history  of  substance  abuse.  Addict  Behav.  2010;35(3):270–2.    126.    Berg  KM,  Arnsten  JH,  Sacajiu  G,  Karasz  A.  Providers’  experiences  treating  chronic  pain  among  opioid-­‐‑dependent  drug  users.  J  Gen  Intern  Med.  2009;24(4):482–8.    127.    Bahl  V,  McCreadie  S,  Stevenson  J.  Developing  dashboards  to  measure  and  manage  inpatient  pharmacy  costs.  Am  J  Health  Syst  Pharm.  2007;64(17):1859–66.    128.    Potier  C,  Laprévote  V,  Dubois-­‐‑Arber  F,  Cottencin  O,  Rolland  B.  Supervised  injection  services:  What  has  been  demonstrated?  A  systematic  literature  review.  Drug  Alcohol  Depend.  2014;145:48–68.    129.    Kral  AH,  Wenger  L,  Carpenter  L,  Wood  E,  Kerr  T,  Bourgois  P.  Acceptability  of  a  safer  injection  facility  among  injection  drug  users  in  San  Francisco.  Drug  Alcohol  Depend.  2010;110(1-­‐‑2):160–3.    130.    Broadhead  R,  Borch  C,  van  Hulst  Y,  Farrell  J,  Villemez  W,  Altice  F.  Safer  Injection  Sites  in  New  York  City:  A  Utilization  Survey  of  Injection  Drug  Users.  J  Drug  Issues.  2003;33(3):733–50.    131.    Zanis  DA,  McLellan  AT,  Randall  M.  Can  you  trust  patient  self-­‐‑reports  of  drug  use  during  treatment?  Drug  Alcohol  Depend.  1994;35(2):127–32.     111 132.    McElrath  K,  Chitwood  DD,  Griffin  DK,  Comerford  M.  The  consistency  of  self-­‐‑reported  HIV  risk  behavior  among  injection  drug  users.  Am  J  Public  Health.  1994;84(12):1965–70.    133.    Darke  S.  Self-­‐‑report  among  injecting  drug  users:  a  review.  Drug  Alcohol  Depend.  1998;51(3):253–263;  discussion  267–268.    134.    Ryan  D.  Health  expert  to  tackle  Vancouver’s  addiction  puzzle.  www.vancouversun.com  [Internet].  2014  [cited  2015  Feb  16];  Available  from:  http://www.vancouversun.com/health/Health+expert+tackle+Vancouver+addiction+puzzle/9820669/story.html  135.    Wood  E.  Canada  should  train  doctors  to  specialize  in  addiction.  The  Globe  and  Mail  [Internet].  2012  [cited  2015  Feb  16];  Available  from:  http://www.theglobeandmail.com/globe-­‐‑debate/columnists/canada-­‐‑should-­‐‑train-­‐‑doctors-­‐‑to-­‐‑specialize-­‐‑in-­‐‑addiction/article4487570/  136.    WHO,  UNODC,  UNAIDS.  WHO,  UNODC,  UNAIDS  technical  guide  for  countries  to  set  targets  for  universal  access  to  HIV  prevention,  treatment  and  care  for  injecting  drug  users  [Internet].  Geneva:  WHO;  2012.  Available  from:  http://apps.who.int/iris/bitstream/10665/77969/1/9789241504379_eng.pdf  137.    Hyun  DY,  Hersh  AL,  Namtu  K,  Palazzi  DL,  Maples  HD,  Newland  JG,  et  al.  Antimicrobial  stewardship  in  pediatrics:  how  every  pediatrician  can  be  a  steward.  JAMA  Pediatr.  2013;167(9):859–66.    138.    Mondain  V,  Lieutier  F,  Dumas  S,  Gaudart  A,  Fosse  T,  Roger  P-­‐‑M,  et  al.  An  antibiotic  stewardship  program  in  a  French  teaching  hospital.  Médecine  Mal  Infect.  2013;43(1):17–21.    139.    Lesprit  P,  de  Pontfarcy  A,  Esposito-­‐‑Farese  M,  Ferrand  H,  Mainardi  JL,  Lafaurie  M,  et  al.  Postprescription  review  improves  in-­‐‑hospital  antibiotic  use:  A  multicenter  randomized  controlled  trial.  Clin  Microbiol  Infect.  2015;21(2):180.e1–180.e7.    140.    Fredholm  L,  Bekanich  S.  A  call  for  opioid  stewardship.  Hospitalist  News  [Internet].  2013;  Available  from:  http://www.ehospitalistnews.com/index.php?id=779&tx_ttnews[tt_news]=142379&cHash=d3cb10f902c56ed416085990952981ce  141.    Andrews  LB,  Bridgeman  MB,  Dalal  KS,  Abazia  D,  Lau  C,  Goldsmith  DF,  et  al.  Implementation  of  a  pharmacist-­‐‑driven  pain  management  consultation  service  for   112 hospitalised  adults  with  a  history  of  substance  abuse.  Int  J  Clin  Pract.  2013;67(12):1342–9.    142.    Ghafoor  VL,  Phelps  P,  Pastor  J.  Implementation  of  a  pain  medication  stewardship  program.  Am  J  Health-­‐‑Syst  Pharm  AJHP  Off  J  Am  Soc  Health-­‐‑Syst  Pharm.  2013;70(23):2070,  2074–5.    143.    Expert  Working  Group  on  Narcotic  Addiction.  The  Way  Forward:  Stewardship  for  Prescription  Narcotics  in  Ontario  [Internet].  Ottawa;  2012.  Available  from:  http://www.health.gov.on.ca/en/public/publications/mental/docs/way_forward_2012.pdf  144.    CBC  News.  AIDS  ward  closes  in  B.C.  as  death  rate  plummets.  CBC  News  [Internet].  2014  May  27  [cited  2014  Dec  12];  Available  from:  http://www.cbc.ca/1.2655795  145.    Giovannetti  J,  Burgmann  T.  Closing  of  Vancouver  hospital’s  AIDS  ward  is  symbolic  of  progress.  The  Globe  and  Mail  [Internet].  2014  [cited  2015  Feb  16];  Available  from:  http://www.theglobeandmail.com/news/british-­‐‑columbia/citing-­‐‑decrease-­‐‑in-­‐‑aids-­‐‑cases-­‐‑vancouver-­‐‑hospital-­‐‑to-­‐‑close-­‐‑dedicated-­‐‑ward/article18868664/    


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items