Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The effects of bone mineral density testing on health related behaviours in a randomly selected population… Kingwell, Elaine 2007

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

Download

Media
831-ubc_2007-318034.pdf [ 18.31MB ]
Metadata
JSON: 831-1.0100632.json
JSON-LD: 831-1.0100632-ld.json
RDF/XML (Pretty): 831-1.0100632-rdf.xml
RDF/JSON: 831-1.0100632-rdf.json
Turtle: 831-1.0100632-turtle.txt
N-Triples: 831-1.0100632-rdf-ntriples.txt
Original Record: 831-1.0100632-source.json
Full Text
831-1.0100632-fulltext.txt
Citation
831-1.0100632.ris

Full Text

T H E E F F E C T S O F B O N E M I N E R A L D E N S I T Y T E S T I N G O N H E A L T H R E L A T E D B E H A V I O U R S I N A R A N D O M L Y S E L E C T E D P O P U L A T I O N O F C A N A D I A N S b y E L A I N E K I N G W E L L B . S c , U n i v e r s i t y o f W a l e s , S w a n s e a , 1 9 8 3 M . S c . , U n i v e r s i t y o f C a l g a r y , 1 9 8 6 A T H E S I S S U B M I T T E D I N P A R T I A L F U L F I L M E N T O F T H E R E Q U I R E M E N T S F O R T H E D E G R E E O F D O C T O R O F P H I L O S O P H Y i n T H E F A C U L T Y O F G R A D U A T E S T U D I E S ( H e a l t h C a r e a n d E p i d e m i o l o g y ) T H E U N I V E R S I T Y O F B R I T I S H C O L U M B I A J u l y 2 0 0 7 © E l a i n e K i n g w e l l , 2 0 0 7 Abstract Osteoporosis is a major health problem that can be mitigated by accurate prediction of future risk for bone fractures. Bone mineral density (BMD) testing can be conducted to assess people's risk of fracture, but its value rests on the ability to provide risk information effectively to those individuals who have low B M D ; those individuals must then be willing to initiate and maintain recommended lifestyle behaviours and medications. This study assessed the influence of low B M D results and direct-to-participant feedback of results on awareness of diagnosis, information seeking about osteoporosis, bone-related medication use and health behaviours. Participants included 1,837 women and 869 men aged 40-60 years, from the population-based prospective Canadian Multicentre Osteoporosis Study, who had undergone baseline dual energy X-ray absorptiometry of the hip and spine. Bone-related health behaviour, medication use and risk factors for osteoporosis were measured by questionnaire at baseline and at 3-year follow-up; recall of B M D test results and subsequent information seeking were assessed at Year 3. The diagnosis reported to the participant and/or the family physician, was documented retrospectively by review of the feedback reports. Correct recall of osteoporosis or osteopenia diagnoses was poor, particularly in men. After adjustment for baseline health behaviour and other important covariates, low B M D results at baseline were associated with subsequent information seeking, and with higher calcium intake, Vitamin D supplement use and osteoporosis-related medication use at follow-up. A report of • low B M D had no influence on exercise participation, smoking cessation, high alcohol consumption, or high caffeine intake. Direct-to-participant feedback v. feedback only to the physician was associated with increased awareness of diagnosis in those with a borderline or normal diagnosis, greater information seeking, and calcium supplementation in women and men, and increased exercise participation and reduced caffeine consumption in women. Educational strategies that specifically target men and specific interventions to support health behaviour change in those at increased risk of fracture are likely to be necessary i f B M D testing in this age-group is to have a significant impact on fracture risk. Direct-to-participant feedback may offer an effective method of increasing awareness of B M D test results and influencing some health behaviours. Table of Contents A B S T R A C T U T A B L E O F C O N T E N T S ,. I H LIST O F T A B L E S V LIST O F F I G U R E S „ VII A C K N O W L E D G E M E N T S VTII C H A P T E R 1: I N T R O D U C T I O N 1 l.l INTRODUCTION 1 l .2 OUTLINE OF THE THESIS STRUCTURE 3 C H A P T E R 2: L I T E R A T U R E R E V I E W 5 2.1 T H E BURDEN OF OSTEOPOROSIS-RELATED FRACTURES 5 2.2 T H E DIAGNOSIS OF OSTEOPOROSIS: B M D AND D X A 8 2.3 NON-MODIFIABLE RISK FACTORS FOR OSTEOPOROSIS .17 2.4 MODIFIABLE RISK FACTORS AND MEDICAL THERAPY 24 2.5 AWARENESS OF TEST RESULTS FOLLOWING B M D TESTING 36 2.6 INFORMATION SEEKING AND KNOWLEDGE ABOUT OSTEOPOROSIS 40 2.7 BEHAVIOURAL CHANGE FOLLOWING B M D TESTING ,45 2.8 MEDICAL THERAPY FOLLOWING B M D ASSESSMENT 55 2.9 DIRKCT-TO-PARTICIPANT FEEDBACK OF TEST RESULTS 57 2.10 SUMMARY 59 C H A P T E R 3: R A T I O N A L E A N D O B J E C T I V E S 61 3.1 RATIONALE : 61 3.2 OBJECTIVES OF THE STUDY 63 3.3 OVERVIEW OF THE STUDY DESIGN 64 3.4 CONCEPTUAL FRAMEWORK 66 C H A P T E R 4: M E T H O D S 69 4.1 THECAMOS.STUDY . . . . 69 4.2 SAMPLE SELECTION FOR THIS STUDY '. 71 4.3 T H E N P H S SAMPLE 72 4.4 COLLECTION AND DERIVATION OF VARIABLES 73 4.5 DATA MANAGEMENT AND ANALYSES 87 C H A P T E R 5: R E S U L T S I. T H E S T U D Y S A M P L E A N D T H E E X P L A N A T O R Y V A R I A B L E S 92 5.1 T H E STUDY SAMPLE 92 5.2 CONTENT OF FEEDBACK TO STUDY PARTICIPANTS 102 5 3 DESTINATION OF THE FEEDBACK.. 105 5.4 DIAGNOSIS AND ITS RELATIONSHIP WITH THE OTHER EXPLANATORY VARIABLES 107 5.5 DESTINATION, CENTRE AND THEIR RELATIONSHIPS WITH THE OTHER EXPLANATORY VARIABLES 112 5.6 PREVALENT FRACTURES FROM THE SPINAL X-RAYS 114 5.7 ASSOCIATIONS AMONG THE REMAINING EXPLANATORY VARIABLES 114 C H A P T E R 6: R E S U L T S II. C O R R E C T A W A R E N E S S O F B O N E D E N S I T Y T E S T R E S U L T S 117 6.1 "CORRECT" AWARENESS OF FEEDBACK DIAGNOSIS: OUTCOME VARIABLE 117 6.2 FACTORS ASSOCIATED WITH CORRECT AWARENESS: UNIVARIATE ANALYSES 119 6.3 POTENTIAL EXPLANATORY FACTORS ASSOCIATED WITH CORRECT AWARENESS: MULTIVARIABLE MODELLING 121 6.4 SUBGROUP ANALYSES OF CORRECT AWARENESS 124 6.5 SENSITIVITY ANALYSES OF CORRECT AWARENESS 132 iii 6.6 SUMMARY OF THE RESULTS OF THE ANALYSES OF CORRECT AWARENESS 133 C H A P T E R 7: R E S U L T S III: B O N E D E N S I T Y T E S T R E S U L T S A N D H E A L T H B E H A V I O U R 135 7.1 INFORMATION SEEKING ABOUT OSTEOPOROSIS 136 7.2 CALCIUM INTAKE FROM DIET AND SUPPLEMENTS 141 7.3 VITAMIN D SUPPLEMENT USE 155 7.4 PARTICIPATION IN A REGULAR EXERCISE PROGRAM i 59 7.6 CIGARETTE SMOKING .- 169 7.7 HIGH ALCOHOL USE 172 7.8 CAFFEINE INTAKE .". 175 7.9 SENSITIVITY ANALYSES: THE EFFECTS OF VARIATION IN PROTOCOL FOR DESTINATION OF FEEDBACK 182 7.10 SUBGROUP ANALYSES: T H E EFFECTS OF A REPORT OF A PREVALENT SPINAL FRACTURE 182 7.11 SUMMARY OF THE MULTIVARIABLE MODELS FOR CORRECT AWARENESS, INFORMATION SEEKING AND OSTEOPOROSIS-RELATED BEHAVIOURAL CHANGE ...187 C H A P T E R 8: S U M M A R Y A N D DISCUSSION .. 194 8.1 INTRODUCTION 194 8.2 SUMMARY OF THE FINDINGS 195 8.3 POPULATION AND GENERALIZABILITY 199 8.4 CORRECT AWARENESS OF B M D TEST RESULTS 202 8.5 INFORMATION SEEKING ABOUT OSTEOPOROSIS FOLLOWING B M D TESTING ; 205 8.6 BONE RELATED LIFESTYLE BEHAVIOURS FOLLOWING B M D TESTING 208 8.7 DIRECT-TO-PARTICIPANT FEEDBACK AND BONE-RELATED HEALTH BEHAVIOUR 216 8.8 RECOMMENDATION FOR FOLLOW-UP OR TREATMENT 218 8.9 STRENGTHS AND LIMITATIONS 219 8.10 IMPLICATIONS AND FUTURE DIRECTIONS 224 8.11 CONCLUSION .229 B I B L I O G R A P H Y . . .. 231 A P P E N D I X A 264 A P P E N D I X B 268 A P P E N D I X C — 270 A P P E N D I X D. . : . , 271 A P P E N D I X E 271 A P P E N D I X F 281 A P P E N D I X G .. . 283 List of Tables Table 2.1 R i s k F a c t o r s tha t I n d i c a t e a N e e d f o r O s t e o p o r o s i s A s s e s s m e n t 15 Table 2.2 P r e v i o u s S t u d i e s o f t h e E f f e c t s o f L o w B M D T e s t R e s u l t s o n B o n e - S p e c i f i c H e a l t h B e h a v i o u r 4 6 Table 4.1 S o u r c e a n d D e r i v a t i o n o f a l l E x p l a n a t o r y a n d O u t c o m e V a r i a b l e s 7 6 Table 5.1a C o m p a r i s o n o f B a s e l i n e C h a r a c t e r i s t i c s b e t w e e n I n c l u d e d W o m e n , E x c l u d e d C a M O S W o m e n a n d a M a t c h e d S a m p l e o f W o m e n from the N P H S 9 3 Table 5.1b C o m p a r i s o n o f B a s e l i n e C h a r a c t e r i s t i c s b e t w e e n I n c l u d e d M e n , E x c l u d e d C a M O S M e n a n d a M a t c h e d S a m p l e o f M e n f r o m t h e N P H S 9 6 Table 5.2 C o m p a r i s o n o f R e l a t i v e F r e q u e n c i e s o f R e p o r t e d D i a g n o s i s a n d W H O D i a g n o s i s b y C e n t r e . 1 0 4 Table 53 F r e q u e n c y o f D e s t i n a t i o n o f F e e d b a c k b y C e n t r e 1 0 6 Table 5.4 F e e d b a c k D i a g n o s i s a n d i t s A s s o c i a t i o n w i t h t he O t h e r E x p l a n a t o r y V a r i a b l e s i n W o m e n 1 0 7 Table 5.5 F e e d b a c k D i a g n o s i s a n d i t s A s s o c i a t i o n w i t h t h e O t h e r E x p l a n a t o r y V a r i a b l e s i n M e n 1 0 9 Table 6.1 P a r t i c i p a n t s ' S e l f R e p o r t s o f t h e i r D i a g n o s i s b y F e e d b a c k D i a g n o s i s 118 Table 6.2 D i s t r i b u t i o n o f A w a r e n e s s o f B a s e l i n e B M D R e s u l t s w i t h i n e a c h E x p l a n a t o r y V a r i a b l e S t r a t i f i e d G e n d e r . . . 1 1 9 Table 6.3 M u l t i p l e L o g i s t i c R e g r e s s i o n A n a l y s i s o f t h e E f f e c t s o f D e s t i n a t i o n , D i a g n o s i s a n d O t h e r E x p l a n a t o r y V a r i a b l e s o n C o r r e c t A w a r e n e s s f o r W o m e n a n d M e n 122 Table 6.4 R e s u l t s o f S u b g r o u p A n a l y s i s o f t h e E f f e c t s o f D e s t i n a t i o n a n d t h e O t h e r E x p l a n a t o r y V a r i a b l e s o n C o r r e c t A w a r e n e s s i n W o m e n w i t h D i a g n o s e s o f L o w B M D 1 2 6 Table 6.5 R e s u l t s o f S u b g r o u p A n a l y s i s o f t h e E f f e c t s o f D e s t i n a t i o n a n d t h e O t h e r E x p l a n a t o r y V a r i a b l e s o n C o r r e c t A w a r e n e s s i n M e n w i t h D i a g n o s e s o f L o w B M D 128 Table 6.6 R e s u l t s o f M u l t i p l e L o g i s t i c R e g r e s s i o n A n a l y s i s o f t h e E f f e c t s o f a R e p o r t o f a S i g n i f i c a n t F r a c t u r e o n C o r r e c t A w a r e n e s s i n W o m e n a n d M e n A g e d 5 0 Y e a r s a n d O l d e r . . . 131 Table 7.1 F r e q u e n c i e s o f R e p o r t e d I n f o r m a t i o n S e e k i n g b y G e n d e r 1 3 6 Table 7.2 M u l t i p l e L o g i s t i c R e g r e s s i o n A n a l y s i s o f F a c t o r s A s s o c i a t e d w i t h I n f o r m a t i o n S e e k i n g a b o u t O s t e o p o r o s i s f o r W o m e n a n d M e n 1 3 9 Table 7.3 F r e q u e n c y o f U s e o f C a l c i u m S u p p l e m e n t s at Y e a r 3 b y U s e o f C a l c i u m S u p p l e m e n t s at B a s e l i n e 1 4 2 Table 7.4 T o t a l a n d D i e t a r y C a l c i u m I n t a k e f o r W o m e n a n d M e n at B a s e l i n e a n d Y e a r 3 1 4 4 Table 7.5 M u l t i v a r i a b l e L o g i s t i c R e g r e s s i o n A n a l y s i s o f F a c t o r s A s s o c i a t e d w i t h C a l c i u m S u p p l e m e n t U s e at Y e a r 3 f o r W o m e n a n d M e n 1 4 7 Table 7.6 M u l t i v a r i a b l e L i n e a r R e g r e s s i o n A n a l y s i s o f F a c t o r s A s s o c i a t e d w i t h T o t a l C a l c i u m I n t a k e at Y e a r 3 f o r W o m e n a n d M e n 1 5 0 Table 7.7 Q u a r t i l e s o f B a s e l i n e T o t a l C a l c i u m I n t a k e i n W o m e n a n d M e n 1 5 2 Table 7.8 F r e q u e n c y o f U s e o f V i t a m i n D S u p p l e m e n t s at Y e a r 3 b y U s e o f V i t a m i n D S u p p l e m e n t s at B a s e l i n e 155 Table 7.9 M u l t i v a r i a b l e L o g i s t i c R e g r e s s i o n A n a l y s i s o f F a c t o r s A s s o c i a t e d w i t h V i t a m i n D S u p p l e m e n t U s e at Y e a r 3 f o r W o m e n a n d M e n 1 5 7 Table 7.10 F r e q u e n c y o f S e l f - R e p o r t e d R e g u l a r E x e r c i s e P a r t i c i p a t i o n at Y e a r 3 b y B a s e l i n e S e l f - R e p o r t e d E x e r c i s e P a r t i c i p a t i o n 1 6 0 Table 7.11 M u l t i v a r i a b l e L o g i s t i c R e g r e s s i o n A n a l y s i s o f F a c t o r s A s s o c i a t e d w i t h E x e r c i s e P a r t i c i p a t i o n at Y e a r 3 f o r W o m e n a n d M e n 1 6 2 v Table 7.12 F r e q u e n c y o f O s t e o p o r o s i s R e l a t e d M e d i c a t i o n U s e at Y e a r 3 b y B a s e l i n e O s t e o p o r o s i s R e l a t e d M e d i c a t i o n U s e i n W o m e n . . 1 6 5 Table 7.13 O s t e o p o r o s i s R e l a t e d M e d i c a t i o n U s e at Y e a r 3 i n W o m e n S t r a t i f i e d b y F e e d b a c k D i a g n o s i s a n d B a s e l i n e U s e o f M e d i c a t i o n s 1 6 6 Table 7.14 O s t e o p o r o s i s R e l a t e d M e d i c a t i o n U p t a k e at Y e a r 3 i n M e n S t r a t i f i e d b y F e e d b a c k D i a g n o s i s . . 1 6 7 Table 7.15 M u l t i v a r i a b l e L o g i s t i c R e g r e s s i o n A n a l y s i s o f F a c t o r s A s s o c i a t e d w i t h C u r r e n t U s e o f M e d i c a t i o n at Y e a r 3 f o r W o m e n 1 6 8 Table 7.16 S m o k i n g S t a t u s at Y e a r 3 b y B a s e l i n e S m o k i n g S t a t u s . 1 7 0 Table 7.17 M u l t i v a r i a b l e L o g i s t i c R e g r e s s i o n A n a l y s i s o f F a c t o r s A s s o c i a t e d w i t h S m o k i n g S t a t u s at Y e a r 3 f o r W o m e n a n d M e n 171 Table 7.18 F r e q u e n c y o f H i g h A l c o h o l U s e at Y e a r 3 b y B a s e l i n e H i g h A l c o h o l U s e 173 Table 7.19 M u l t i v a r i a b l e L o g i s t i c R e g r e s s i o n A n a l y s i s o f F a c t o r s A s s o c i a t e d w i t h A l c o h o l I n t a k e o f T w o o r M o r e D r i n k s p e r D a y at Y e a r 3 f o r W o m e n a n d M e n 1 7 4 Table 7.20 F r e q u e n c y o f H i g h C a f f e i n e U s e at Y e a r 3 b y B a s e l i n e H i g h C a f f e i n e U s e 1 7 6 Table 7.21 M u l t i v a r i a b l e L o g i s t i c R e g r e s s i o n A n a l y s i s o f F a c t o r s A s s o c i a t e d w i t h D r i n k i n g F o u r o r M o r e C u p s o f C o f f e e p e r D a y at Y e a r 3 f o r W o m e n a n d M e n 1 7 8 Table 7.22 M u l t i v a r i a b l e L i n e a r R e g r e s s i o n A n a l y s i s o f F a c t o r s A s s o c i a t e d w i t h T o t a l C a f f e i n e I n t a k e ( S q u a r e R o o t ) at Y e a r 3 f o r W o m e n a n d M e n 1 8 0 Table 7.23 R e s u l t s o f M u l t i p l e L o g i s t i c R e g r e s s i o n A n a l y s i s o f t h e E f f e c t s o f D i a g n o s i s , D e s t i n a t i o n a n d R e p o r t o f a S i g n i f i c a n t F r a c t u r e o n C a l c i u m S u p p l e m e n t U s e at Y e a r 3 i n W o m e n a n d M e n A g e d 5 0 Y e a r s a n d O l d e r ; 1 8 3 Table 7.24 R e s u l t s o f M u l t i p l e L o g i s t i c R e g r e s s i o n A n a l y s i s o f t h e E f f e c t s o f D i a g n o s i s , D e s t i n a t i o n a n d R e p o r t o f a S i g n i f i c a n t F r a c t u r e o n V i t a m i n D S u p p l e m e n t U s e at Y e a r 3 i n W o m e n a n d M e n A g e d 5 0 Y e a r s a n d O l d e r 1 8 4 Table 7.25 S u m m a r y o f S i g n i f i c a n t A s s o c i a t i o n s i n A d j u s t e d A n a l y s e s w i t h C o r r e c t A w a r e n e s s o f T e s t R e s u l t s , I n f o r m a t i o n s e e k i n g a n d O s t e o p o r o s i s - R e l a t e d H e a l t h B e h a v i o u r C h a n g e i n W o m e n 1 9 0 Table 7.26 S u m m a r y o f S i g n i f i c a n t A s s o c i a t i o n s i n A d j u s t e d A n a l y s e s w i t h C o r r e c t A w a r e n e s s o f T e s t R e s u l t s , I n f o r m a t i o n s e e k i n g a n d O s t e o p o r o s i s - R e l a t e d H e a l t h B e h a v i o u r C h a n g e i n M e n 1 9 2 v i List of Figures Figure 3.1 A C o n c e p t u a l F r a m e w o r k o f F a c t o r s C o n t r i b u t i n g t o t h e I m p a c t o f D X A T e s t R e s u l t s o n I n f o r m a t i o n S e e k i n g a n d O s t e o p o r o s i s S p e c i f i c H e a l t h B e h a v i o u r 6 8 Figure 5.1 S e l e c t i o n o f t h e S t u d y S a m p l e from t h e 4 0 - t o 6 0 - y e a r - o l d C a M O S P a r t i c i p a n t s 9 2 Figure 7.1 D i s t r i b u t i o n o f T o t a l C a l c i u m I n t a k e at Y e a r 3 i n W o m e n 143 Figure 7.2 D i s t r i b u t i o n o f T o t a l C a l c i u m I n t a k e at Y e a r 3 i n M e n . . . 143 vii Acknowledgements I g r a t e f u l l y a c k n o w l e d g e t h e s u p p o r t , u n d e r s t a n d i n g , p a t i e n c e a n d h u m o u r o f m y w o n d e r f u l f a m i l y a n d f r i e n d s w h o h a v e h e l p e d m e t o c o m p l e t e t h i s p r o j e c t . I n p a r t i c u l a r , m y c h i l d r e n , C a l l u m a n d H a n n a h , h a v e g r o w n w i t h m e t h r o u g h t h i s p r o c e s s a n d h a v e i n s p i r e d m e t o o v e r c o m e t h e h u r d l e s a n d t o r e a c h m y g o a l s . I h a v e b e e n v e r y f o r t u n a t e t o b e n e f i t f r o m the g u i d a n c e o f f o u r h i g h l y e x p e r i e n c e d r e s e a r c h e r s a n d r o l e m o d e l s as m y s u p e r v i s o r y c o m m i t t e e , e a c h o f w h o m h a v e b e e n e x c e e d i n g l y g e n e r o u s w i t h t h e i r t i m e a n d e x p e r t i s e . I w a n t t o e x p r e s s m y d e e p a p p r e c i a t i o n t o D r . S u s a n K e n n e d y f o r h e r s u p p o r t a n d c o m m i t m e n t a n d f o r k e e p i n g m e o n t h e r i g h t t r a c k t h r o u g h o u t m y d o c t o r a l s t u d i e s . I a m v e r y g r a t e f u l t o D r J e r i l y n n P r i o r f o r h e r e n t h u s i a s m a n d p a s s i o n , as w e l l as h e r u n f a i l i n g c o n f i d e n c e i n m y p o t e n t i a l a s a r e s e a r c h e r . I a m i n d e b t e d t o D r P a m e l a R a t n e r f o r h e r c o n s t r u c t i v e c r i t i c i s m , m e t i c u l o u s a t t e n t i o n t o d e t a i l a n d c o n t i n u a l e n c o u r a g e m e n t . A s p e c i a l t h a n k y o u t o D r S u s a n B a r r f o r a l l o f h e r a d v i c e , t h o u g h t f u l f e e d b a c k a n d p o s i t i v e r e a s s u r a n c e . T h e s t a f f a t t h e D e p a r t m e n t o f H e a l t h C a r e a n d E p i d e m i o l o g y h a s p r o v i d e d a s s i s t a n c e w i t h a d m i n i s t r a t i v e i s s u e s t h r o u g h o u t m y d o c t o r a l s t u d i e s ; i n p a r t i c u l a r , I g r e a t l y a p p r e c i a t e t h e g o o d h u m o u r e d s u p p o r t p r o v i d e d b y L a u r e l S l a n e y a n d M o i r a T h e j o m a y e n . I w o u l d l i k e t o e x t e n d a s p e c i a l t h a n k y o u t o t he C a n a d i a n M u l t i c e n t r e O s t e o p o r o s i s S t u d y ( C a M O S ) S t e e r i n g C o m m i t t e e f o r s h a r i n g t h e C a M O S d a t a . I w o u l d a l s o l i k e t o a c k n o w l e d g e t he d i r e c t o r s a n d c o o r d i n a t o r s at e a c h o f t he C a M O S c e n t r e s f o r t h e i r g e n e r o u s h o s p i t a l i t y a n d s u p p o r t w h i l e I v i s i t e d a n d c o l l e c t e d d a t a i n S a i n t J o h n ' s , H a l i f a x , Q u e b e c C i t y , K i n g s t o n , H a m i l t o n , T o r o n t o , S a s k a t o o n , C a l g a r y a n d V a n c o u v e r . I n a d d i t i o n , S u z e t t e P o l i q u i n a n d C l a u d i e B e r g e r h a v e b e e n i n v a l u a b l e r e s o u r c e s f o r C a M O S i n f o r m a t i o n a n d I t h a n k t h e m f o r t h e i r w i l l i n g n e s s t o r e s p o n d t o m y e n q u i r i e s a n d r e q u e s t s f o r h e l p t h r o u g h o u t t h e c o u r s e o f t h i s p r o j e c t . F i n a l l y , I w o u l d l i k e t o t h a n k t h e C a n a d i a n w o m e n a n d m e n w h o a g r e e d t o p a r t i c i p a t e i n C a M O S ; w i t h o u t t h e i r w i l l i n g n e s s t o d o n a t e t h e i r t i m e a n d t o s h a r e t h e i r p e r s o n a l i n f o r m a t i o n t h i s r e s e a r c h w o u l d n o t h a v e b e e n p o s s i b l e . F i n a n c i a l s u p p o r t w a s g e n e r o u s l y p r o v i d e d t o m e d u r i n g m y d o c t o r a l s t u d i e s b y t h e C a n a d i a n Ins t i t u tes o f H e a l t h R e s e a r c h , H e a l t h C a n a d a , t h e M i c h a e l S m i t h F o u n d a t i o n f o r H e a l t h R e s e a r c h a n d t h e B C M e d i c a l S e r v i c e s F o u n d a t i o n . CHAPTER 1: Introduction 1.1 Introduction Osteoporosis is a major health problem that is expected to contribute an increasing burden on health services as the population grows and ages. The annual cost associated with treatment and care of hip fractures alone in Canada has been estimated to be $650,000,000 (1). Because age is the major risk factor for low bone density and osteoporosis-related fracture (2), the epidemic of fractures attributable to osteoporosis, together with the financial and societal burden that accompanies them, are expected to increase dramatically as life expectancy increases around the globe. Prevention offers the greatest potential for reducing the impact of osteoporosis on morbidity, mortality, quality of life and economic burden: Once fractures occur, bone loss has reached a stage that is often irreversible. If individuals who are identified as being at higher relative risk for osteoporotic fracture can be encouraged to modify their lifestyles appropriately, or to take effective medications or supplements, then the burden of osteoporosis would be expected to decrease. For such preventive interventions to be effective, however, individuals at risk for fracture must first be reliably identified, individual risk information and recommendations for intervention must effectively reach them, and those at high risk must be prepared to make recommended lifestyle changes and take medications as appropriate, in order to reduce their risk. The measurement of bone mineral density (BMD) plays a central role in the diagnosis of osteoporosis and in the assessment of fracture risk; the current gold standard method for identifying low bone density in clinical practice and population based settings is dual energy X -ray absorptiometry (DXA). A low test result by D X A should ideally prompt physicians or other health professionals to recommend appropriate interventions and to support women and men with low B M D (and hence an increased risk of fracture) to make lifestyle changes and, i f appropriate, to take medications to reduce their risk. The identification of women and men who have low bone density by D X A testing is of potential benefit i f the communication of an increased risk of fracture were to encourage those at greater risk to undertake preventive measures. The potential benefit of D X A testing is offset by its low positive and negative predictive values: many who are identified with a low bone density by D X A are not destined to experience a fracture and similarly, many with normal bone densities will later suffer fractures. A low test 1 result may lead to increased stress and worry about silent "risk" that may never result in a fracture, or may result in inappropriate precautions (3), and a normal test result may provide false reassurance and complacency. Furthermore, D X A testing is costly. When assessing the value of B M D testing in selected women in the perimenopausal and early menopausal years, the British Columbia Office of Health Technology Assessment cited a lack of data demonstrating that women, or their physicians, acted on B M D results in a way that could affect health outcomes (4). The positive potential of D X A testing can only begin to be realized i f it results in positive action, such as appropriate lifestyle change or medication use. The primary focus of this dissertation was to address whether a sample of mid-aged women and men, from the general Canadian population, who received B M D test results from a D X A test were aware of their test results, and whether low test results prompted information seeking about osteoporosis and the modification of osteoporosis-related health behaviour. Specific health behaviours that were addressed by this research were those that have been demonstrated to be effective, at least to some extent, in reducing the risk of fracture or the loss of B M D or that are recommended preventive or treatment interventions in established osteoporosis practice guidelines: increased calcium intake from diet, use of calcium or vitamin D supplements, participation in exercise, smoking abstinence, moderate alcohol and caffeine intake, and the use of bone-specific medications. Direct-to-participant feedback of test results may increase the awareness of test results and may further provide the opportunity for individuals to take responsibility for their own health. On the other hand, direct feedback may provide no improvement in understanding and may be less effective than communication through the primary health-care provider. Furthermore, direct feedback may even hinder behaviour change by interfering with the opportunity for physician-initiated discussion of potential lifestyle or therapeutic interventions. A second focus of this dissertation was to explore the effects on correct recall and awareness of D X A test results (or diagnosis), information seeking and health behaviour change (including the use of bone-specific medications) of direct-to-participant feedback of test results, in comparison with feedback to the FP only. Although knowledge of an increased risk may be necessary, it may not be sufficient to motivate positive change; the increased risk must also be relevant to the individual and that individual must be prepared to initiate and maintain changes to his or her lifestyle. This suggests that those at greater risk for fracture, such as older women, individuals with a family history of osteoporosis and those with a history of exposure to high-risk medications such as 2 corticosteroids, would be expected to take more notice of their B M D test results and may be more likely to make appropriate behavioural changes. In other words, different subgroups of the population may respond differently to the equivalent assessment of risk (BMD test results). A third focus of this dissertation was to assess the potential effects of demographic factors and other risk factors for osteoporosis on correct awareness of B M D test results, information seeking, health behaviour change and osteoporosis-related medical therapy. This study was made possible by the provision of prospectively collected population-based data from women and men who took part in the Canadian Multicentre Osteoporosis Study (CaMOS) (5). Further data to address the specific questions of this dissertation were collected retrospectively from CaMOS participant files. 1.2 Outline of the Thesis Structure Chapter 2 follows this introductory chapter and provides an overview of the literature on D X A testing, osteoporosis and its risk factors, and prevention and treatment recommendations. Prior studies that have investigated correct awareness of B M D test results, information seeking and lifestyle change following bone mineral density testing are reviewed in this chapter, as is the small amount of prior research related to direct-to-participant feedback of B M D test results and some theoretical grounding for its consideration as a potentially effective intervention. The rationale for the study in the context of a summary of the literature discussed in Chapter 2 is presented in Chapter 3, together with the overall and specific objectives of this dissertation research and the conceptual framework that guided the development of the study. The designs of this study and the Canadian Multicentre Osteoporosis Study (CaMOS), from which this study is derived, are described in Chapter 4. The sample selection, measurement and derivation of the explanatory and response variables, manipulation of the data and analysis methods are described here. Chapter 4 also includes a description of the sampling strategy for the National Population Health Survey (NPHS), from which a selection of variables was obtained and compared with a subset of this study's data to assess the findings' generalizability. Chapter 5 provides a description of the study sample and a comparison between the sample and the mid-aged CaMOS participants who were excluded due to loss to follow up or lack of a B M D test. The sample is also compared with the equivalent age group from the cross-sectional NPHS (Cycle 2) sample that was completed at the same time as the CaMOS participants were surveyed. 3 Chapters 6 and 7 present the results of the univariate and multivariable data analyses and Chapter 7 ends with summary tables of the associations that were found between the awareness, information seeking and health behaviour outcomes and all of the explanatory variables that were considered. Finally, an interpretation of the study findings and their implications, together with a discussion of the limitations of this research are presented in Chapter 8. 1 4 CHAPTER 2: Literature Review 2.1 The Burden of Osteoporosis-Related Fractures Osteoporosis is a chronic disorder in which a complex array of physiological mechanisms brings about a progressive reduction in bone strength and increased risk of bone fracture. The fractures that result are associated with a large burden of morbidity and mortality and a significant cost to the health-care system (1;6-11). Major fragility fractures, particularly of the hip, are associated with excess mortality in both women and men (12; 13). Mortality within one year following hip fracture varies considerably by age, but overall, is between 10% and 20% (14-16); between 17-23% of these deaths are believed to be directly due to the hip fracture event. It has been estimated that 1.5% of all deaths in a population aged 50 years and over can be attributed to hip fractures (17). Excess mortality is also associated with clinically diagnosed fractures of the vertebrae1 (12), although this is possibly due to their relationship with comorbid conditions. Although the incidence of hip fracture is much more frequent in women than in men, mortality has consistently been found to be higher in similarly aged men than women following hip fracture (16-21). The morbidity associated with osteoporosis-related fractures, particularly of the hip, is substantial. Approximately one half of all hip fracture patients have difficulty walking one year after fracture (14;22); 8% of independently living women who have an osteoporotic fracture of the hip, spine or wrist are expected to require long term care and a further 7% are expected to become dependent for the basic activities of daily living (23). A recent Canadian study found that 41% of community-dwelling women and men aged 50 years and over were no longer living independently one year after a hip fracture (1). Given the significant associated morbidity, it is not surprising that health-related quality of life (QOL) is significantly affected by osteoporotic fractures (24-27), although the relationship is complex and likely mediated or moderated to some extent by other factors such as comorbidity. Prevalent fractures were associated with poor QOL, as measured by the Health Utilities Index in the Canadian Multicentre Osteoporosis Study (CaMOS) cohort (24), while a prospective study of predictors of incident low trauma fractures amongst the menopausal women from the same cohort found that poor QOL, as measured by the SF-36 at baseline, was a risk factor for osteoporotic fractures of the hip, spine and other non-vertebral sites (28). This suggests that other factors that are often associated with QOL such as age, comorbidity and ' Approximately two thirds of vertebral fractures remain silent and are not identified clinically. 5 depression may confound its relationship with fractures. Nevertheless, it is clear that people with osteoporosis-related fractures experience a poorer QOL compared with those without fractures (24-28), During one year (1990), 740,000 deaths globally were estimated to be attributable to hip fractures and a loss of 1.75 million disability adjusted life-years; equivalent to 0.1% of the global burden of disease world-wide (29). This relative burden is more than 10 times greater (1.4%) i f only women in Western countries are considered (29). Evidently, the burden imposed by fractures associated with osteoporosis is significant in terms of both mortality and morbidity, and because a major risk factor for low trauma fracture is advancing age (30;31), the epidemic of fractures has been projected to increase exponentially as life expectancy and median age around the globe increases (32;33). The cost of treatment and care for these cases of osteoporosis is also expected to vastly increase in the absence of effective and acceptable preventive and treatment interventions; the direct and indirect cost associated with hip fracture alone in Canada has been estimated at $650 million, with a projected increase to $2.4 billion by 2041 (1). Fracture treatment is costly and rehabilitation has a low success rate, hence early preventive interventions offer the greatest potential benefit in terms of reducing morbidity, mortality and economic burden. Age-adjusted incidence rates for osteoporotic fractures in women have decreased since the 1950's in a U.S. cohort (34), while rates stabilized between 1965 and 1983 in a Swedish sample (35), during the 1980's in populations in Canada (36) and Finland (37), and in the early 1990's in a second Swedish population (38). Age-adjusted rates for men, on the other hand, have shown increases over time relative to those seen in women (34;35) leading to a decrease in the male : female ratio of incidence rates in some populations. The increase in incident fractures as the population ages is therefore expected to be greater in men than in women. The age adjusted incidence of fractures in Asia is still increasing (39;40). As fracture rates stabilize or decrease in parts of Europe and North America, the aging populations of Asia, as well as South America, are expected to contribute an increasing relative global burden of fracture cases over the next half century (41;42). 2 "Disability Adjusted Life Years" (DALYs) are a combined estimate of the years of life lost due to premature mortality and the number of healthy years of life lost because of disability. These calculations take into account the disability weight associated with hip fracture (0.272 for each year of disability), the age of onset of the disability (because the relative importance of healthy life is adjusted for different ages), the duration of time lost due to death attributable to hip fracture (25% of mortality following hip fracture was assumed to be attributable to the fracture for these estimates) and a time preference, or discounting function, that adjusts for the greater value of health gains now compared to health gains in the future. 6 The results of two recent studies have provided more evidence that age-adjusted fracture rates are declining in Western countries where greater attention has been paid to the disease. In a retrospective cohort study of Swedish women and men aged 50 years and over, age adjusted hip fracture rates in women decreased between 1982 and 1996, while those in men increased. The authors projected that age-adjusted hip fracture rates would decrease overall by 11% by the year 2010 (compared to 1996). Based on the data for women and men, a decrease in hip fracture rates in women of 19% was projected and an increase of 7% was projected for men. Given these projections, and the expected increased age of the population in 2010, no change in the actual number of hip fractures was projected (43). Jaglal et al. (44) investigated similar trends in both hip and wrist fractures in women and men aged 50 years and over in the population of Ontario, Canada. The age-adjusted incident fracture rates for both hip and wrist fractures in women remained stable between 1992 and 1996, but then declined from 1997 to 2000. Consequently, the authors projected that rates would continue to decline between 2001 and 2005 and that the actual number of hip and wrist fractures would not increase substantially even as the population aged. As observed in the Swedish study, the age-adjusted fracture rates in men did not exhibit the same decline. These results, together with earlier indications of relative increased rates in men, suggest that education and awareness, as well as lifestyle and therapeutic interventions, have made an impact on women's bone health but not on men's, and that there is a need for greater education, awareness, prevention and intervention directed at men. The Ontario study (44) correlated the decline in fracture rates to concurrent substantial increases in the use of B M D tests in women aged 50 years and over and to increasing prescription claims for bone-specific medications (antiresorptives) between 1996 and 2003 in women aged 65 years and over. While it is possible, as the authors suggest, that increased diagnosis by B M D testing and increased treatment with antiresorptives were responsible for the observed decline in age-adjusted hip and wrist fracture rates in women in Ontario, D X A testing and treatment could not be causally linked to the fracture outcome because interventions and fracture rates were not studied at the individual level. Furthermore, information was not available regarding either D X A test results or adherence to antiresorptive therapy; the authors only had information about the proportion of the population who filled a prescription. Other potential explanations for the decline in fracture rates include increased awareness of the risk factors and consequences of osteoporosis amongst physicians and the lay population, which may have led to initiation, maintenance or improvement in women's exercise participation or healthful diet choices, for example. The use of ovarian hormone therapy (OHT), whether taken 7 as a preventive intervention for osteoporosis or for other indications, may also have contributed to the change in fracture incidence rates (prescription of OHT was not investigated by the authors). In summary, it is not possible to tell from this study i f increased D X A testing, and in particular, the specific results of the tests, influenced fracture rates in Ontario in the late 1990's. Studies of individual level associations between B M D testing, the results of the tests, medication uptake, lifestyle changes and ultimately fracture rates are necessary to clarify the role that D X A testing has played, and may play in the future.. 2.2 The Diagnosis of Osteoporosis: B M D and D X A The term "osteoporosis" is used interchangeably to describe both the process that contributes to fractures, a decreased relative bone mineral density (BMD), and the clinical outcome (fractures). The following operational definitions, based on bone mass measurements in white menopausal women have been developed by the World Health Organization (WHO) (45): • Severe osteoporosis. B M D lower than or equal to 2.5 standard deviations (SD) below the peak adult mean (the mean of a healthy young adult reference population) in the presence of a low trauma fracture. • Osteoporosis. B M D lower than or equal to 2.5 SDs below the peak adult mean without fracture. • Osteopenia or low bone mass. B M D between 1 and 2.5 SDs below the peak adult mean. • Normal. B M D no lower than 1 SD below the peak adult mean. Measurements of bone density, most often from the proximal femur or lumbar spine, are interpreted by calculating a T-score, which is the number of standard deviations by which a result deviates from the mean bone density in young adults of the same sex. The threshold values, as defined by the WHO, are then used to categorize patients as having normal bone, osteopenia or osteoporosis. Z-scores are also used as an alternative comparison method; the Z-score is the number of standard deviations by which a result deviates from an age- and sex-matched population. As bone density naturally decreases with age, a Z-score for a person past the age of 40 years, when peak bone mass starts to decline, will be higher than the corresponding T-score. Sixteen per cent of young adults within the age range for "peak bone mass" will have a B M D that is 1 SD below peak bone mass (a T-score of-1.0), whereas a person with a Z-score of 8 -1.0 lies in the lowest 16% of the population of equivalent age. Hence, with a T-score cut-off, the chance of being diagnosed with osteoporosis or osteopenia gradually increases with advancing age. For Swedish women, for example, the prevalence of a T-score at or below -2.5 (i.e., a diagnosis of osteoporosis), using 20-39 year-old "peak bone mass" in women as the norm (from the National Health and Nutrition Examination Survey, or NHANES) , has been estimated to be approximately 3.9% in women at the age of 50 years, increasing to 36.1% at the age of 80 years. For men, using 20-39 year-old men as the reference population, the prevalence of osteoporosis is approximately 1.9% at the age of 50 years and 12.6% at age 80 years (46). From the CaMOS data, the prevalence rate of osteoporosis of the hip in Canadians aged over 50 years has been estimated at approximately 8% for women and 5% for men (47). Equivalent estimates for U.S. population from the NHANES data, have been similar for men (6%) but higher for women (18%) (48); this difference in prevalence may be explained by an actual difference in the B M D of women aged 50 years and over between the two populations or by somewhat different age distributions amongst women aged 50 to 80 years in Canada compared to the U.S. (47). Dual energy absorptiometry (DXA), in which the amount of mineralised tissue in a scanned area of bone (g/cm2) in the hip, spine or total body is measured by passing low-dose X -rays through the bone, is the current gold standard for the clinical measurement of B M D . D X A is relatively more accurate and precise than other currently available technologies for measuring B M D and can be performed quickly with little patient discomfort (49). The accuracy error for D X A (potential error in measurement of the actual B M D of a site) ranges from 5% to 8% (50) and the precision error (the potential error in re-measuring the same sample) is approximately 1% to 2% (49). With a potential error of estimating bone density at 8% above or below an individual's true B M D , there is the potential of a resulting error in the diagnosis and prediction of fracture risk. As was pointed out by the British Columbia Office of Health Technology Assessment (BCOHTA), in their systematic review of the use of selective or population screening in well women, however, the technological performance of D X A is of much less importance than its overall clinical usefulness as a prediction tool for the clinical outcome -fracture (4). The diagnosis of osteoporosis and the assessment of fracture risk are based on the comparison of a D X A measurement to peak bone mass and are therefore heavily dependent on the population that is used as the reference. Originally, the referent populations for D X A testing 9 were the D X A manufacturers' samples, which were small samples of relatively healthy young (aged 20 to 29 years) adults. The methods used to obtain these data and descriptions of the populations used for these manufacturers' normal samples were not published. The hip measurements of the reference populations for the two main D X A suppliers (Hologic Inc.™ and Lunar Corp.™ ) have been found to differ, even when the systematic differences between the values of absolute bone densities between the two systems were corrected (51). The authors of a study that compared the manufacturers' norms to a sample of young British women found large differences in the two sources of norms for both hip and spine values, and concluded that the use of the manufacturer's norms would lead to over diagnosis, excess treatment and unnecessary worry (52). Since 1997, the larger, population-based young adult sample from the third National Health and Nutrition Examination Survey (NHANES III) has been used as the reference population for the hip (53). Because the average B M D values of this large U.S. national representative sample are lower than those of the small samples available in the earlier days of D X A testing, the T-scores from measurements derived by comparison with the NHANES III reference database are higher than those using the older manufacturers' hip norms and there is a reduced probability of diagnosis of osteoporosis or osteopenia (47). Canadian population norms for women and men for both femoral neck and lumbar spine values are now available. Peak bone mass values of young adult women and men in the CaMOS study were found to be very similar to those in the NHANES database, and significantly lower than the manufacturers' norms for both hip and spine measurements (47). Until the publication of the CaMOS data, no reliable population-based values for peak bone mass values at the spine were available. Measurements for both the lumbar spine (the first four lumbar vertebrae) and the hip are typically generated by the D X A scan, but there have been varying opinions and only limited agreement about which sites should be used to make a diagnosis. Hip measurements are provided for three separate regions of the proximal femur: neck, trochanteric and inter-trochanteric, as well as a total hip measurement which includes all three areas. A measurement for the Ward's triangle, a small area of mostly cancellous bone3 below the femoral neck, is also provided, but this area is the least reliable of all hip sites and is not recommended for diagnosis (54). The hip region used for clinical and research purposes has varied, both in Canada and internationally, but most of the published research regarding the risk of fracture has used the 3 Cancellous bone (or trabecular or spongy bone) is osseous tissue with low density and strength and a large surface area; it fills the inner cavities of long bones and protects the bone marrow. 10 neck of the hip (55;56). Up until 1996,4 when the Scientific Advisory Board of the Osteoporosis Society of Canada5 recommended that the neck of the hip be used for clinical diagnostic purposes (57), there were no Canadian guidelines regarding the area of the hip that be used. Very shortly afterwards the International Committee for Standards in Bone Measurement endorsed the use of the total hip measurement as the preferred site (58). A recent position statement by the International Society for Clinical Densitometry (ISCD) which recommended that the lowest of the total hip, trochanter and neck of hip be used (59) is indicative of the confusion and inconsistency that have been associated with the clinical diagnosis of osteoporosis in practice. The ISCD's position has been challenged by the International Osteoporosis Foundation as being without scientific basis (60) and has since been updated by the ISCD to a recommendation of the selection of the lowest of the total hip or femoral neck (61). Values from the femoral neck or the total hip are reportedly the most reliable for diagnostic purposes and, because these measurements are highly correlated, they have similar predictive values for fracture (2;62). The T-scores that result from measurements taken at the spine and the hip from the same individual can be considerably discrepant; the two sites were discordant for a diagnosis of normal bone density, osteopenia and osteoporosis in 44% of the women in one study (63). The recommended sites for comparison to norms and for diagnostic purposes have changed over time and there is persistent disagreement over which site(s) to use for the prediction of fracture risk (59;60;64). Although there has been a tendency to use the lowest of the measurements at either the spine or the hip to assess fracture risk in patients clinically, and an endorsement of this approach by some groups (59;65), a recent analysis of the B M D values from six population-based cohorts (64) found that selection of the lower of the spine and hip values does not improve the estimate of risk, either for hip fracture, or for any fracture. The hip value alone is reportedly a better predictor of hip fracture risk, whilst the hip value and spine value each predict fractures at other sites equally. A recently published study of a large clinical cohort of women selected from the population-based database of the Manitoba Bone Density Program reported that measurements taken at the hip were superior to those taken at the spine for predicting overall osteoporotic fractures (66). Essentially, the use of the lower of either site increases the proportion of the population that receives a "below normal" diagnosis without improving the 4 The baseline CaMOS data were collected during 1996 to 1997 and feedback of the D X A results to participants and physicians occurred before, and after, the introduction of the first guidelines on the preferred hip site for clinical diagnosis. 5 The Osteoporosis Society of Canada is now known as "Osteoporosis Canada". 11 predictive value of the D X A test. Furthermore, extraneous calcification can erroneously inflate B M D measurements at the spine (67;68), as can vertebral compression fractures. Both calcifications (such as osteophytes) and prevalent vertebral fractures become more common with advancing age, which means that estimates of B M D at the spine can be subject to greater error, particularly in men (67;69) and in people over 65 years (49;70). While the use of the hip value alone for diagnostic, or risk assessment purposes, would appear to avoid the problem of erroneously elevated spinal values, the dialogue about the preferred sites(s) for clinical practice is continuing and there remains uncertainty and inconsistency in the preferred site for the interpretation of bone density tests (71). 2.2.1 Bone Mineral Density and Fracture Risk Prospective studies have demonstrated that for each standard deviation in B M D value (hip or spine) below the young adult mean the relative risk of any osteoporotic fracture increases by approximately 1.5 times in both women and men (55;72;73). Measurements taken at the hip provide a higher relative risk for fracture at the same site; a hip value that is 1 SD below the peak young adult mean has been shown to have an associated relative risk for hip fracture of 2.5-3.0 (55;56;74). Analyses provided by Marshall et al. (56), Johnell et al. (55), and Kanis et al. (64) included the combined results of several cohort studies to estimate the relative risk of fracture by B M D . 6 Interestingly, the risk estimates generated from these studies tended to be the same whether populations were drawn from Australia, Sweden, Japan, the U.K. , the U.S or Canada. Although a small proportion of the studies had relatively long periods of follow-up, the majority of the subjects were followed for less than five years; the risk estimates generated from the cohorts with longer term follow-ups were comparable to those with shorter follow-up periods (56). Johnell et al. (55) included sufficient subjects followed for 10 years to provide 10-year risk estimates and found that the predictive ability of B M D remained stable as time elapsed after measurement for up to 10 years. Risk estimates of the association between B M D and fracture therefore appear to be stable over time (at least for 10 years), between men and women and across different populations. Although these authors concluded that the relative risk estimate for a fracture 10 years after B M D testing was reliable and stable, there was a trend (not statistically significant) towards a gradual decrease in relative risk as the time between testing and fracture 6 Note that the latter two studies overlap; a subsample of the data utilised for the study by Johnell et al. (55) was used to answer different questions in the study by Kanis et al. (64). 12 increased, which suggests that prediction of fracture risk at time periods much longer than 10 years in advance (i.e., for younger men and premenopausal women) may become smaller and less meaningful. The results of longer-term follow up are not yet available from the large population-based studies to determine the risk estimates for fracture at a much later age for young women and men. Notwithstanding the above mentioned limitations, there is evidence that B M D may be a "purer" indicator of fracture risk in younger people. In their study of the results of 12 large, population-based cohorts (a total of 29,082 women and 9,891 men), Johnell and associates identified a decreasing gradient of hip fracture risk with age; the relative risk for hip fracture for each 1 SD decrease in T-score decreased from 3.68 at age 50 years to 1.93 at age 85 years in women and men combined (55). While the relative risk of a hip fracture associated with a lower B M D may decrease with advancing age, the absolute risk of fracture, however, increases due to the strong association between age and risk of hip (or any osteoporotic) fracture. Although B M D testing can detect a risk for fracture, it cannot identify which individuals will later fracture. The bone density distribution of women with a hip fracture overlaps significantly with that of women without a hip fracture (75); consequently, the number of false positives or false negatives will vary depending on the position of the "cut-off value on the normal curve. Use of a cut-off T-score of-1.0 to define those who are at higher risk of fracture, and a relative risk of hip fracture of 2.5 per SD below the mean, for example, results in a relatively high specificity (approximately 85% of women aged 50 years who will be identified as "low risk" will not fracture), but a low sensitivity (only approximately 45% of women aged 50 years who will be identified as "higher risk" will experience hip fracture). This results in a positive predictive value (the probability that a woman aged 50 years who has a T-score of-1.0 will fracture) of only 3.8% (76) . 7 This means that, for a 50-year-old woman without any other risk factors, neither a negative nor a positive test result is very informative. The predictive value of a low B M D test result increases when other risk factors for osteoporotic fracture are taken into account. In one prospective study, Cummings et al. identified several risk factors that, when combined, predicted the incidence of first hip fracture in women more reliably than did low B M D alone (77), and the predictive value increased when B M D was included together with other risk factors. Amongst the risk factors that were 7 The positive predictive value (PPV) of a test depends both on the relative risk and on the prevalence of the disease in the population. These estimates are based on a 15-year risk of fracture, which has a relatively low prevalence in women aged 50 years. For women aged 65 years, the PPV using the same cut-off for "high risk" is estimated at 20.1% while the sensitivity and specificity rates are similar to those for women aged 50 years. 13 identified, however, are those that are of little or no value in predicting fracture risk in younger women and men, such as age > 80 years, inability to rise from a chair without using one's arms, poor visual contrast sensitivity, and history of a non-hip fracture since the age of 50 years. It could be argued that once a fracture has occurred (even non-hip), or once a woman reaches the age of 80 years and has several other risk factors for fracture (some of which would have been potentially modifiable at a younger age), most available preventive interventions will be limited and any preventive efforts have already proven themselves fruitless. Furthermore, although a woman over the age of 80 years may have a high risk of fracture within 10 years based on a combination of risk factors, her lifetime risk of fracture may be low once the competing risk of death within 10 years is taken into account. 2.2.2 Screening by D X A The value of D X A testing as a screening tool for the identification of those at increased risk of fracture and the selection of appropriate people for B M D assessment have been extensively reviewed and debated; there has been particular controversy over whether perimenopausal women or younger menopausal women should be screened for low bone density (4;50;78-80). A WHO study group8 (50) determined that bone mass measurements meet many of the essential criteria of screening tests as outlined by Cadman (83), and that screening of women at or close to menopause, to help with decisions about uptake of ovarian hormone therapy (OHT), would be appropriate. It has been argued that if women were more likely to adhere to prescribed effective therapy after being identified with low bone mass, then general screening in perimenopausal women may be worthwhile (84). The BC Office of Health Technology concluded, on the other hand, that even selective screening of those menopausal women deemed to be at higher risk for osteoporosis was not warranted (4); a lack of information on the uptake and adherence to therapy and lifestyle change recommendations as well as the low positive predictive value of a B M D test were cited amongst the many reasons why such screening was not recommended. It has since been suggested that, even though screening for osteoporosis may not be justified based on current evidence, developments in therapies and interventions may lead to re-appraisal of the value of screening in the future (78). Note that the WHO study group's report was released before the results of the Women's Health Initiative (WHI) revealed that the risks of ovarian hormone therapy may outweigh its benefits (81 ;82) 14 Most reviewers and experts have concluded that the results of a B M D test alone, in the absence of other risk factors, are not sufficiently predictive of future fracture to justify a population screening approach in women or men under the age of 65 years. Hence, a case-finding strategy is generally recommended whereby those at higher risk of having a low B M D , but also a somewhat higher risk of fracture even when adjusted for B M D , are selected. Variation among different guidelines, however, leads to large differences in the populations that are recommended to undergo B M D assessment. Table 2.1: Risk Factors that Indicate a Need for Osteoporosis Assessment' Major Risk Factors Minor Risk Factors - Age > 65 years - Vertebral compression fracture - Fragility fracture after age 40 years - Family history of osteoporotic fracture (especially maternal hip fracture) - Systemic glucocorticoid therapy - Malabsorption syndrome - Primary hyperparathyroidism - Propensity to fall - Osteopenia apparent on X-ray film - Hypogonadism - Early menopause (before age 45 years) - Rheumatoid arthritis - Past history of clinical hyperthyroidism - Chronic anticonvulsant therapy - Low dietary calcium intake (<1,500 mg per day for menopausal women and men over 50 years) - Smoker - Excessive alcohol intakeb - Excessive caffeine intake (>4 cups coffee per day) - Weight < 57 kg - Weight loss > 10% of weight at age 25 years - Chronic heparin therapy a From Brown et al. (85). b A definition of "excess" alcohol was not provided by the panel. The Scientific Advisory Council for Osteoporosis Canada recommends that all women and men over the age of 50 years be assessed for risk factors for osteoporosis and that those with one major or two minor risk factors (see Table 2.1) be assessed for low B M D (85;86). Notably, according to these guidelines, all women and men over age 65 years should receive a D X A test and lifestyle factors such as being a smoker, or having a low calcium intake, or high caffeine or alcohol intake, are considered sufficient as minor risk factors to warrant a B M D test. The Canadian Task Force on Preventive Health Care recommends screening all women over age 65 years and menopausal women under 65 years in the case of a previous fragility fracture, weight less than 60 kilograms (kg), or certain other risk factors (a white woman aged 50 15 years who weighs 54 kg would meet the criteria for B M D testing by these recommendations) (87). A low calcium intake is considered sufficient indication by itself for a referral for a B M D test by the WHO Task Force for Osteoporosis, as are other commonly accepted risk factors such as a maternal history of risk fracture or prolonged corticosteroid therapy (88). In the U.S., the National Osteoporosis Foundation endorses a more preventive approach by recommending screening of all menopausal women with at least one major risk factor. Major risk factors include a history of a fragility fracture, a family history of fracture, low body weight, history of use of corticosteroids, and current smoking. Additional risk factors are also listed as potential indications for assessment, including poor health, low calcium intake and low physical activity (89). The European Foundation for Osteoporosis and Bone Disease9 guidelines for case finding include the requirement of the presence of at least one specific risk factor, such as loss of height, low body mass index and history of corticosteroid therapy, but include the acknowledgement that physicians may need to manage patients who have no known risk factors who would take treatment i f their B M D were low (i.e., that B M D testing may be recommended to help women with decisions about taking OHT) (90). There is evidence that a significant number of women who do not meet the screening guidelines are requesting and attending D X A tests. There are a variety of clinical practice guidelines that address criteria for referral for a D X A test in the U.S, for example, and adherence to these guidelines is reportedly low amongst health-care providers (91). In England, 30% of women who were referred by one clinic did not meet the guidelines for referral in that country (92). The most common reason for a referral for D X A testing amongst a different sample of 3,530 British women was to help in decision making about taking OHT use (93). A qualitative study of general practitioners (GPs) in the U.K. revealed that physicians have difficulty deciding for whom and when to order B M D tests and that factors other than existing guidelines, such as patients' requests and perceived availability of D X A machines, influence these decisions (94). A similar qualitative study of family physicians (FPs) in Ontario, Canada revealed that FPs lack a rationale for ordering B M D tests (95). Flexible referral patterns are likely to be occurring in Canada, where a significant proportion of mid-aged women (under 65 years) are receiving D X A tests (4;44;96). It is believed in fact, that "mass screening" methods have been used to some extent throughout the world (72). It should be noted that few guidelines exist for B M D assessment in men and premenopausal women. On behalf of the ISCD, Khan et al. (97) recommended that senior men 9 Now known as the International Osteoporosis Foundation. 16 should be assessed by B M D measurement and that younger men and premenopausal women should be assessed in the presence of secondary causes of low B M D in conjunction with other risk factors for fracture (i.e., secondary causes of osteoporosis or a fragility fracture must be present). 2.3 Non-Modifiable Risk Factors for Osteoporosis 2.3.1 Age and Gender There is no disputing that advancing age and female gender are the most significant independent risk factors for osteoporosis (98-101). Compared with women in their 50s, women in their 60s have twice the risk of osteoporosis and women over 70 years have 4 times the risk of osteoporosis (102). Results from longitudinal studies have revealed that men lose approximately 5-10% of their cortical bone mass each decade and that the incidence of hip fractures in men increases exponentially with age just as it does in women (103). The 10-year probability of hip fracture for a Canadian woman at age 50 years is approximately 0.4%, at age 60 years it is 1.5%, at age 70 years it is 4.7%, and at age 80 years it is 13.7%. For men, the pattern is similar, but with approximately half the risk relative to women of the same age; at age 50 years the risk of hip fracture for a Canadian man is 0.3%, at age 60 years it is 0.9%, at age 70 years it is 2.1%, and at age 80 years it is 6.2% (104). The rates in men are about five years behind those of women, so that a man at age 70 years has the same risk of hip fracture as an average woman at age 65 years (105). Rates for clinically diagnosed vertebral fractures in women and men also increase with age, but more linearly and with a consistent ratio; after the age of 50 there is a 2:1 incidence of fractures in women compared with age-matched men (99; 106; 107). Age-adjusted mortality and reduction in life expectancy following fracture are comparable in men and women, although the proportion of years of life lost attributable to hip fracture is even greater in men because of their shorter life expectancy (19). Although age is the strongest determinant of fracture risk, as a non-modifiable risk factor that applies equally to the whole population up until death, it is of limited value for the prediction of an individual's future fracture risk. Age is of much greater value when it can be combined with other factors associated with an increased or decreased risk of fracture. 17 2.3.2 Attainment and Maintenance of Peak Bone Mass Osteoporosis is brought about by a gradual loss, or resorption, of bone over time after peak bone mass has been attained. The exact age at which peak bone mass is reached appears to vary by skeletal site, but the results of cross-sectional studies of bone mineral content suggest that bone accrual slows considerably after the age of 15 to 16 years in girls and 18 years in boys (108;109). Bone mass then begins a noticeable natural decline as the rate of bone resorption starts to exceed the rate of bone formation; the age at which this decline first becomes evident varies considerably between sites and has been inconsistent between studies. Cross sectional data from the CaMOS women and men suggest that a decline in femoral neck bone density may be evident as early as the end of the third decade but lumbar spine bone density may remain constant until the end of the fourth decade (47). Regardless of the uncertainty about the age at which peak bone mass starts to decline, it is presumed that the higher the peak bone mass, the longer it will take for this decline to take place with age. Low bone density in mid-aged women and men reflects both the maximum amount of bone mass achieved and the loss of bone after attainment of peak bone mass. In addition to genetics, there are many factors that are known, or believed, to influence the attainment and maintenance of peak bone mass (or "full genetic potential" for bone strength as defined by Heaney et al. (110)). Many of the factors that play a role in attainment and maintenance of peak bone mass can be categorized as those that influence reproductive and other hormone (such as Cortisol) concentrations. Late age at menarche, early age at pregnancy, irregular menstrual cycles and amenorrhea secondary to excessive leanness, intense physical activity, or anorexia nervosa, for example, have all been associated with lower peak bone mass (110; 111). Sub clinical ovulatory disturbances as a result of lower progesterone production (i.e., anovulatory cycles or cycles with a short luteal phase) have also been linked to bone loss from the spine in premenopausal women (112; 113). Other experiences that are reported to disrupt the hormonal milieu and potentially lead to reduced bone mass include dietary restraint (114-116) and intense exercise (112; 117). Low bone density may also result from the use of oral contraceptive agents (OCAs); a recent analysis of the premenopausal women in the CaMOS cohort revealed that those who had ever used OCAs had lower B M D measurements at the spine and trochanter than did women who had never used OCAs (118). The effects of OCAs on premenopausal bone is not conclusive however, as both increases and decreases in B M D have been reported (119). Although there have been reports of an association between a history of 18 O C A and increased risk of fracture (120; 121), these studies have included only limited adjustment for potential confounders. Apart from hormonal influences, strength, mobility or bone-loading, and alterations in nutrition (particularly sufficient calcium and vitamin D, but also phosphorus, protein and other nutrients) and lifestyle (such as tobacco use and possibly alcohol use) play a role in the attainment and maintenance of peak bone mass (111). Gains and losses in weight (or "weight cycling") also put premenopausal women at risk for reduced B M D (122) and have been associated with increased risk of hip fracture in older women and men (123). After the achievement of peak bone mass both men and women lose approximately 0.5-1.0% of bone per year, although there is considerable variation between individuals. In addition, there is a phase of more rapid loss during and soon after the menopausal transition (see below) in women. Both men and women may therefore become osteoporotic, with age being a prominent risk factor and women losing bone earlier and more extensively than men. Other risk factors combine with this decline in peak bone mass to bring about a certain level of risk for fractures at an older age. The changes are much more evident in trabecular bone than in cortical bone due to the higher metabolic activity in the former consequently the sites where trabecular bone predominate (the vertebrae and at the ends of long bones) are the most frequently affected by fragility fracture. By the age of 40 years, women and men have achieved their peak bone mass and, in fact their bone mass is likely to already be on the decline, particularly at vulnerable sites such as the hip. Apart from age and gender, there are several other non-modifiable risk factors for osteoporosis that mid-aged people may possess. 2.3.3 The Menopause and Women's Reproductive Hormones An accelerated rate of bone loss is associated with changes that occur during the first five to ten years of menopause in women (2; 124); this is believed to be associated with low estrogen levels (125). Many studies have demonstrated that this amplified rate of bone loss actually begins before menopause, in the perimenopausal years, as hormone levels begin to fluctuate and ovulatory disturbances become more prevalent, and that bone loss during the perimenopause may even exceed that in the early menopause years (126-130). Menopause, or perimenopause, has also been directly associated with an increased risk of fracture in mid-aged women: In the Kuopio Osteoporosis Risk Factor and Prevention Study, a population-based retrospective cohort study of 14,220 women aged between 47 and 56 years, a comparison was made between the 49% 19 of women who were menopausal (defined, rather liberally, as at least six months since last menstrual period) by the start of the study period and the remaining women who were grouped together and inconsistently defined as premenopausal or perimenopausal (although this group also included women who were taking hormone therapy). Increased fracture risk in the five year window was associated with being menopausal when adjusted for age, weight, height and other gynaecological factors (131). Due to the young age of this cohort, however, many of these women who were classified as menopausal (i.e., between the ages of 42 and 51 years) would have been very recently perimenopausal and the results seem to implicate perimenopause, rather than menopause, as a risk factor for fracture in the fifth or sixth decade of life. An increased duration of exposure to gonadal hormones in women tends to provide protection against low B M D and fracture. Exposure is increased by earlier menarche and by later menopause; although the former may have a greater relative influence on risk of fracture in later life (132). Increasing age at menopause has been found to be associated with reduced risk of fracture by some authors (133-135), while others have found that it is not predictive once other important confounders are taken into account (136-140). Surgical menopause represents a variation on early menopause, but also involves the abrupt cessation of estrogen release from the ovaries, which confers an even greater risk for osteoporosis than the gradual change in hormone levels that accompanies natural menopause (141-145). Hysterectomy (without ovariectomy) has also been tentatively linked to an increased risk of osteoporosis (144;146); it has been suggested that indications for hysterectomy, such as ovulation disturbances and high estrogen levels that in turn are related to menstrual bleeding problems, may be instrumental in an increased risk in hysterectomized women (127). Given that there are serious disparities in the rates of hysterectomy, for various indications, across educational and socioeconomic strata (147; 148), hysterectomy may well be linked to an increased osteoporosis risk through both biological and sociocultural pathways. 2.3.4 Race or Ethnicity There are significant differences in the bone mass and the fracture risk of women and men of different racial or ethnic groups. African American women aged 50 and over, for example, have a lower risk of osteoporosis than non-Hispanic white women (102); African American women appear to attain a higher peak bone mass than white women of an equivalent weight (149;150), and also lose bone less rapidly during early menopause (150). Associated age-adjusted hip fracture rates are approximately 50% lower in African American women than in 20 white women (151). Similarly, African American men have a lower rate of prevalent vertebral fractures than do white men (152). Overall, white women and men are at the highest risk of osteoporotic fracture (153; 154); hip fracture rates are highest in whites living in North America and Europe, intermediate in Asian populations and lowest in black populations. Although hip fracture rates have been found to be somewhat lower in Asian, compared with white, women and men (155; 156), bone densities have typically been found to be equivalent in the two populations (102;157), which implicates a role for factors independent of bone density. 2.3.5 Body Mass Index Even though body mass index (BMI) (or specifically the weight component of BMI) is potentially modifiable, a low B M I is typically treated as a "marker" or a non-modifiable risk factor for osteoporosis in mid-aged or older women and men; a larger body size is believed to have a protective influence. There are well known serious risks 1 0 of other diseases that are associated with excess weight or obesity and the associations between BMI , loss or gain of weight, weight cycling, and B M D and fracture risk are complex. Thus, it is not surprising that a recommendation for an increase in B M I is not prevalent in clinical practice or in the literature as a potential osteoporosis prevention or treatment intervention for those considered at risk for osteoporosis due to low BMI. The evidence for the association between low body mass index (BMI), or weight, and increased risk of fracture, independently of age and sex, is consistent (102;154;158-160). De Laet et al. (161) recently analysed the combined international data from 12 prospective population-based cohorts and confirmed that low B M I is a reliable predictor of fracture risk at the hip and other sites in both women and men and that this association is independent of age. Low B M I is also associated with low B M D (162;163) but at low values of B M I (<20 kg/m2), B M I predicts fracture risk independently of B M D ; the association between B M I and fracture risk does not appear to be independent of B M D at greater values of B M I (161). Although weight gain during adulthood appears to be protective for osteoporosis and fracture (77; 158; 164), a history of variability in weight (or weight cycling), irrespective of BMI , is associated with a low B M D or higher risk of hip fracture (122; 123). Weight loss (>10% of maximum body weight) is associated with an increased likelihood of low B M D and fracture risk (123;165;166). 1 0 Health problems that are associated with excess weight and obesity include hypertension, cardiovascular disease, type 2 diabetes and some cancers. 21 2.3.6 Family History The risk of osteoporosis and fracture is greater for people with a family history of osteoporotic fracture. Perimenopausal and menopausal women with a family history of osteoporosis are at increased risk of osteoporosis and fracture (77; 102; 146; 163) and premenopausal women with a family history of osteoporosis have been found to have lower B M D and may experience a greater loss of bone density over time (162;167;168). Men with a maternal history of fracture also have lower B M D (154; 169). A meta-analysis of seven international prospective cohort studies confirmed that family history of fracture, particularly hip fracture, is associated with increased risk of osteoporotic fracture in both women and men and that the associated risk is independent of B M D (170). 2.3.7 Chronic Illness A history of certain chronic illnesses, including rheumatoid arthritis, anorexia nervosa or eating disorders, malabsorption syndromes (such as inflammatory bowel disease, Crohn's disease, chronic liver disease), renal disease, hyperparathyroidism, hyperthyroidism and Cushing's syndrome has been included in some position papers, consensus statements and guidelines as a potential risk factor for fracture or low bone density (85;87;89;171;172). Based on a systematic review of the available literature up to 1997, Espallargues et al. identified anorexia, type 1 diabetes, primary hyperparathyroidism and pernicious anaemia as "high" risk factors for osteoporosis, and hyperthyroidism, type 2 diabetes and rheumatoid arthritis as "moderate" risk factors (160). There is reasonable evidence of an increased risk of fracture in both women and men with a history of rheumatoid arthritis; this association persists even when prior corticosteroid use is taken into account (173;174). A history of anorexia nervosa is a well established risk factor for decreased bone density and fracture (175-177). While the evidence for a relationship between bulimia and low bone density has been suggested (178) it seems likely that associations observed in early studies can be explained by comorbidity with anorexia (177; 179). Osteoporosis guidelines have not generally included diabetes as a risk factor, although type 1 diabetes has been identified as a risk factor for decreased B M D and associated fracture and for low B M D (180-182). An increased B M D in women and men with type 2 diabetes was identified, while controlling for BMI , in a cross-sectional analysis of the CaMOS cohort (183). A n association between increased B M D and type 2 diabetes has been similarly reported, while 22 taking B M I or weight into account, by others (184-186) although type 2 diabetes has also been found to be associated with increased risk of fracture (185). A prospective cohort study of risk factors for incident low trauma fracture in middle age (44 to 50 years) identified diabetes (type not specified) as a major risk factor for fractures of the hip, vertebrae, and forearm in men and women; the other chronic conditions that were considered by these authors, and not found to be associated with increased risk, were angina, cancer and hypertension (187; 188). Even though risk ratios varied somewhat, the overall risk factors for men and women were found to be very similar (187; 188). The greatest bulk of the evidence for the role of a history of chronic conditions has been derived from studies of women and there is relatively little information on the association between a history of chronic conditions and fracture in men. Still, risk factors for osteoporosis in men are often assumed to be the same as those in women for practical purposes (86; 189) and there is no reason to expect the impact of chronic illnesses, or medications, on bone to be qualitatively or quantitatively different in men compared with women (103). 2.3.8 Exposure to Medications Certain medications, particularly anticonvulsants and oral or inhaled corticosteroids, have been associated with increased risk of osteoporosis and fractures. Oral corticosteroids are considered to be an established risk factor for osteoporosis and fracture (85; 190); this robust association is independent of age, gender and underlying disease (174; 191). Use of inhaled corticosteroids has also been associated with decreased bone mineral density (192;193), although the association between inhaled corticosteroids and risk of fracture is less consistent (192;194). Anticonvulsant use has been associated with an increased risk of fracture in women and men (77;195;196); this association may be partly explained by seizures (197) but there is also reasonable evidence of a direct, deleterious effect of these medications on bone density (198-202). In their systematic review of studies of risk factors for fracture, Espallargues et al. classified the history of use of corticosteroids as a "high" risk factor (i.e., previous studies had found an associated relative risk or odds ratio of 2 or more); a history of use of anticonvulsants carried a similarly high risk for fracture (160). Evidence for the effects of other medications on fracture risk, including diuretics, thyroid hormone therapy, and anti-ulcer agents has been less consistent (160). 23 2 . 3 . 9 G e n e r a l H e a l t h a n d S o c i o e c o n o m i c S t a t u s Although general health, education and socioeconomic status are not typically considered to be risk factors for osteoporosis or fracture, these factors have been observed to be associated with this risk, just as they are with other chronic conditions, and can therefore serve as indicators of those at greater risk. Poor self-rated general health is associated with low B M D (203;204) and is predictive of future fracture in both men and women (28;77; 187; 188). A case-control study of menopausal women in Sweden revealed that, even when other relevant risk factors were taken into account, women who had experienced a hip fracture were more likely to have a lower household income, to have lived in more crowded housing or to have lived without a partner in the three decades preceding the fracture; education level, on the other hand, was not associated with hip fracture risk (in unadjusted or adjusted analyses) (205). In contrast, an earlier case-control study of menopausal women with hip fracture in Southern Europe found a protective effect of more education (206). Two separate population-based studies in the U.S. that relied on estimated neighbourhood incomes (generated from postal codes) as a proxy for individual socioeconomic status identified a link between higher rates of hip fracture in women and men and lower neighbourhood socioeconomic status (207;208). Further, these disparities in risk were apparent even when other potential confounders such as self-reported race or ethnicity and estimated fluency in English were taken into account (208). 2 . 3 . 1 0 P r e v i o u s F r a c t u r e A history of a low trauma fracture is associated with an increased risk of another fracture (146;209;210). A review of studies of the risk associated with previous fracture (211) and a recent international meta-analysis of 11 cohorts that included the data from 15,259 men and 44,902 women (212) both reported that the risk of fracture is consistently about twice as high in people with a prior fracture compared with those with no prior fracture. Furthermore, this association is evident in both women and men, it is independent of B M D (211;212) and the risk of subsequent fracture increases with the number of prior fractures (i.e., the relationship is dose dependent) (211). 2.4 Modifiable Risk Factors and Medical Therapy Certain risk factors for osteoporosis, including older age, female gender and "white" race are not modifiable. Several modifiable or lifestyle factors are, however, believed to play a role 24 in bone loss after peak bone mass has been achieved. Information about the role of these risk factors in minimizing the rate of bone loss is typically given by family physicians or other specialized sources, such as Osteoporosis Canada, to individuals who are considered to be at risk for fracture or who have been diagnosed with osteoporosis. A sufficient intake of calcium and vitamin D, by diet or by supplements is advised as well as regular participation in weight bearing exercise (such as walking, running, aerobics), abstinence from tobacco smoking and a restriction of alcohol consumption to a maximum of two drinks a day. Some sources also suggest restricting caffeine intake to less than four cups of coffee per day. Variations in lifestyle appear to be associated with significant differences in risk for fracture and may account for a large component of the risk (132), which suggests that successful modification of unhealthful behaviours would contribute to a reduction in fracture risk. Modifiable risk factors, or health behaviours, may be particularly targeted amongst those who receive low B M D test results and have a greater relative risk of fracture. OVarian hormone therapy (OHT) acts by both slowing bone resorption and stimulating new bone formation and has been recommended, and has been offered in the past, to menopausal women as a common method of preventing and treating osteoporosis (213). OHT is no longer endorsed as a preventive option however (85;214) since the results of the Women's Health Initiative (WHI) revealed that the risks of OHT may outweigh the benefits (81;82). Currently available treatments for osteoporosis include bisphosphonates, raloxifene, calcitonin and teriperatide. Three bisphosphonates are approved for the treatment or prevention of osteoporosis in Canada; etidronate, alendronate and risedronate. These medications act by binding to the surface of bone and slowing the activity of the (bone resorbing) osteoclasts. Bisphosphonates are indicated for the prevention and treatment of osteoporosis in men, menopausal women and individuals who are taking steroid medications. Calcitonin, a thyroid hormone, is indicated for the treatment (but not prevention) of osteoporosis and also acts by slowing down osteoclast activity. Raloxifene's effects on bone are similar to those of estrogen and it is indicated for the prevention and the treatment of osteoporosis in menopausal women. Teriperatide, a parathyroid hormone analog, activates the (bone building) osteoblasts and is approved for treatment of menopausal women and men with severe osteoporosis who do not respond to or cannot tolerate other osteoporosis therapies. 25 2.4.1 Calcium There is a vast literature on the role of calcium on bone health and osteoporosis, and there is substantial evidence that sufficient dietary calcium is necessary to maintain bone mass and to protect against osteoporosis. The benefits of increased dietary calcium or the addition of calcium supplements to fracture risk in mid aged women or men are less certain however. A meta-analysis of controlled trials published in 1990 by Cumming showed that the rate of bone loss was reduced by about 50% in menopausal women taking calcium supplements and that calcium supplementation appears to be most effective when there are low baseline intakes of calcium (i.e., there is a threshold effect), at an older age and when B M D is lower or there is clinical evidence of osteoporosis (215). In a more recent summary of randomised controlled trials (RCTs) and observational studies of children, adolescents, adult women and a few studies that included men, Heaney concluded that increased calcium intake was clearly effective at improving bone health. More than 90% of the RCTs reviewed and three quarters of the observational studies reported reduced or arrested bone loss at some skeletal site or decreased fracture risk (216). Studies of women in the early menopausal years generally found smaller effects on bone mass than those of premenopausal or later menopausal women (216). Even though it is unlikely that a woman in her perimenopausal or early menopausal years can take enough calcium (either through diet or supplements) to restore the accelerated bone loss that occurs during this time, the goal is to achieve sufficient calcium intake that bone loss is minimized and to ensure that insufficient calcium is not one of the factors that contributes to a decline in B M D . Although the relationship between calcium intake and fracture risk has been elusive in many observational studies (217-219), randomized clinical trial data have tended to demonstrate an association between increased calcium intake and decreased risk of fracture in menopausal women. A Cochrane review of clinical trials in menopausal women, published up to 2001, concluded that calcium supplementation had a clear positive effect on bone density, and there was a trend towards a reduction in risk for vertebral fractures, while the effect on non-vertebral fractures was uncertain (220). Results from the WHI randomized double blind placebo controlled trial of calcium and vitamin D on fracture risk in menopausal women were recently reported (221). Whereas B M D was significantly higher in the treated group, there were no significant differences in hip or non-hip fracture rates in the intention to treat analysis. Comparison between only those who adhered to therapy did, however, reveal a significant protective effect of supplementation on hip fracture. 26 As seen in other studies, effects were greater in older women, in those who were not taking additional calcium supplements and in those who had low baseline intakes of calcium (<800 mg/day) from all sources. Very few of the women in this study had a low intake of calcium and further supplementation was allowed in both arms of the study; the average intake of calcium at baseline was over 1,000 mg per day, which means that women in the treatment group were taking over 2,000 mg per day on average. An increased risk of kidney stones was observed in the calcium and vitamin D treated group, which also reflects the particularly high calcium intake levels. Thus, while these results indicate that calcium supplementation has no impact on fractures in healthy women who already have adequate intakes of calcium, the study results also demonstrate that there are potential benefits for women with low dietary calcium intake (221). A low dietary calcium intake has been documented as a significant risk factor for low B M D in observational studies of middle-aged men (222;223), but there is a distinct lack of clinical trials that have assessed the impact of calcium supplementation on men. The results from one prospective study of men aged 30 to 87 years indicated that calcium supplementation for three years in those with a high baseline calcium intake (mean = 1,159 mg/day) has no effect on B M D change (224) and a randomized controlled trial that included just over 800 elderly men with previous fractures who were followed for up to five years found that calcium (combined with vitamin D) was not effective at preventing a further fracture (225). The potential long-term effects of calcium supplementation in mid-aged men on fracture risk are unknown. There is evidence that calcium supplementation may interact with other interventions such as OHT and other medications (226;227), even during the period of rapid bone loss in women. A retrospective cohort study of early menopausal women taking OHT, for example, found that those who took calcium supplements in addition to OHT lost less bone mass over the two year study period than women on OHT alone (226). A review of clinical trials of OHT that compared studies that had administered OHT plus calcium supplements or diet modification to those that administered OHT only concluded that significantly greater increases in bone mass were seen in menopausal women with higher calcium intakes. There was also some evidence of a similar potentiation by calcium of the effect of calcitonin (227). Despite the variable evidence of its influence in some subgroups, calcium is recommended as a necessary adjunct to therapy for osteoporosis in both women and men, and essential levels of calcium intake are advised for the prevention of osteoporosis in pre-, peri- and menopausal women, and in men (85;228). A statement prepared at the 1993 Consensus Conference of the Osteoporosis Society of Canada concluded that dietary calcium was 27 unambiguously important for bone health and recommended an optimum minimum intake as a preventive measure against osteoporosis of 1,000 mg/day of calcium for women and men aged 19 to 49 years, and between 1,000-1,500 mg/day for women and men aged 50 years and over (229). Clinical practice guidelines available in Canada in 1996 (at the time of the CaMOS baseline) recommended that adults obtain between 1,000 and 1,500 mg of calcium per day for "optimal bone health" (230). Osteoporosis Canada recommendations for daily intake of calcium for the prevention of osteoporosis are now 1,000 mg/day for premenopausal women and for men between the ages of 18 and 50 years, and 1,500 mg/day for menopausal women and men over 50 years (228). Health Canada and the Institute of Medicine recommend 1,000 mg/day for those aged between 19 and 50 years and 1,200 mg/day for those aged over 50 years (231). 2.4.2 Vitamin D A deficiency of vitamin D is known to lead to below optimal calcium absorption and to increased parathyroid hormone excretion, which in turn stimulates osteoclast activity and increased bone loss (232); adequate amounts of vitamin D are necessary for bone health. Synthesis in the skin by U V light is a major source of vitamin D, but a significant portion of vitamin D is derived from diet, particularly during the winter or in parts of the world where sun exposure is more limited (233). Other factors that are associated with decreased synthesis of vitamin D in the skin are use of sunscreen, greater melatonin pigmentation and advancing age (234;235). It is believed that suboptimal dietary intake of vitamin D contributes .to accelerated loss of bone mass and may also be associated with increased fracture risk (49;234): Vitamin D inadequacy (or low levels of vitamin D intake) have been linked to increased fracture risk in cross-sectional (236) and prospective (237) studies of menopausal women. Studies in Europe and North America, including Canada, have demonstrated that vitamin D insufficiency is prevalent in the middle-aged and in the elderly (233-235;238;239), which suggests that supplementation of vitamin D may be required for optimal bone health for many people, even in middle age. Results from intervention studies have indicated a reasonably clear benefit of vitamin D supplementation on bone mass and fracture risk in older menopausal women, but the evidence for an effect in perimenopausal or early menopausal women, or in younger men is less consistent. 28 In a placebo controlled study of healthy menopausal women (mean age 62 years), treatment with 400 IU/day of vitamin D 3 1 1 slowed the normal seasonal bone loss and increased bone remodelling that occurs in the winter months (242). A later study by the same authors compared vitamin D 3 supplements of 100 IU per day to 700 IU per day and found a dose response effect; the higher dose was more effective at slowing bone loss in menopausal women (243) . A trial that compared three groups of women and men randomized to either vitamin D 3 supplements, calcium supplements or placebo reported that bone loss at the hip in the vitamin D 3 treated group over four years was half way between the calcium and placebo treated groups (244) , but not significantly different from either. The baseline level of vitamin D in these women and men (who were all aged 60 years or over) was relatively high, which suggests that supplementation in people with higher levels of vitamin D may have less potential benefit. Fracture risk appears to be reduced with vitamin D supplementation in older women and possibly in men, although many of the trials of vitamin D supplementation have included calcium supplementation so that the effects of each cannot be separated. Trials in non-institutionalized women and men aged 65 years and over (245) and in healthy ambulatory elderly women (246;247), for example, have shown that supplementation of vitamin D 3 together with calcium supplementation reduces the incidence of hip (246;247) or other non-vertebral fractures (245;246). On the other hand, vitamin D 3 supplements, alone or combined with calcium, were of no measurable benefit in preventing further fracture in a large trial of vitamin D 3 supplementation for elderly women and men with a previous fracture (225). Overall, the balance of the evidence suggests that vitamin D supplementation is beneficial for older women and men: A recent meta-analysis of vitamin D 3 supplementation in ambulatory or institutionalized people aged 60 years or over concluded that supplements of 700 to 800 IU per day were effective at reducing hip and non-vertebral fracture risk in this population, but 400 IU per day was not effective (248). An earlier review and meta-analysis of placebo controlled trials of various forms of vitamin D 3 therapy in menopausal women concluded that vitamin D supplementation is effective at reducing the risk of vertebral fracture (249). It is plausible that supplementation is of more benefit when baseline intake of vitamin D is low or suboptimal. As pointed out by the authors of one meta-analysis (249), very few authors have documented baseline levels of vitamin D intake which limits the assessment of this potential explanation. ' 1 Vitamin D supplements are available in two separate forms, as Vitamin D 2 (ergocalciferol) and as Vitamin D 3 (cholecalciferol). Vitamin D 3 is now known to be a more potent and effective form of Vitamin D in the human body (240;241). The majority of intervention trials of the effects of Vitamin D on bone have used Vitamin D 3 , even though Vitamin D 2 has been the more readily available form in prescriptions in North America. 29 Intervention studies of vitamin D therapy on B M D in early menopausal women and younger men have not shown significant effects. Loss of B M D at the lumbar spine or femoral neck was no different over five years in early menopausal non-osteoporotic women treated with low dose vitamin D 3 compared with placebo (250), although the dose of vitamin D 3 was likely insufficient to confer a benefit and it is not known whether selection of high risk women would have shown different results. Likewise, a twin study of young menopausal women (with a mean age of 59 years) to determine the effects of larger supplements of vitamin D 3 (800 IU per day) found no effect on B M D after two years of treatment (251). A randomized controlled trial of vitamin D 3 supplementation combined with calcium in normal healthy men aged 30 to 87 years also found no impact of the combined therapy compared with placebo after three years (224). The effects of vitamin D supplements on subgroups of younger women or men with either low baseline B M D or low baseline vitamin D levels are unknown. Although there is only limited evidence that vitamin D supplementation alone is sufficient as a therapy for osteoporosis or to prevent fracture, adequate vitamin D is still recommended as an essential adjunct to preventive therapy and treatment for osteoporosis (85;228), and a dietary intake of between 400 and 800 IU of vitamin D per day was recommended for people with osteoporosis by guidelines and statements published by the Scientific Advisory Board of the Osteoporosis Society of Canada during the time that the CaMOS study began (229;230). Dietary sources of vitamin D are considered necessary in Canada because sun exposure is not sufficient to provide the full vitamin D requirements. Osteoporosis Canada's recommended intake level for vitamin D (from diet and supplements combined) for all men and women under 50 years of age is now 400 IU per day; this recommended requirement increases to 800 IU per day for men and women over 50 years of age (85;228). Joint guidelines provided for the Canadian and U.S. public from the Institute of Medicine and Health Canada (231) list somewhat lower dietary reference intake values for Vitamin D; 200 IU per day for men and women aged 19 to 50 and 400 IU per day for those aged 51 to 70 years.12 2.4.3 Exercise Exercise plays a role in reducing bone loss or enhancing bone gain and, in people who have been diagnosed with osteoporosis, the benefits of exercise are believed to include an 1 2 Health Canada's website now notes that these 1997 values may underestimate Vitamin D requirements and recommends that all women and men aged 50 years and over should take a daily Vitamin D supplement of 400 IU per day in addition to following Canada's Food Guide (252) 30 indirect protection against fractures by improving mobility and muscle strength and by reducing the risk of falls (253). A statement paper prepared at the 1995 consensus conference of the Osteoporosis Society of Canada concluded from a review of the available evidence that regular, moderate physical activity as a therapy for people with osteoporosis was warranted (254). In the same year that CaMOS began, Canadian guidelines for the management of osteoporosis from the Scientific Advisory Board of the Osteoporosis Society of Canada were published that included the recommendation of regular participation in an exercise program as part of a treatment regimen for osteoporosis (230). The weight of evidence from cross-sectional and prospective non-randomised studies indicates that current or recent exercise has a protective effect on bone mass as well as on fracture risk in women and men (102;107;154;159;255-258). Even though this protective effect has not been consistently found, either for B M D (162;259) or for fracture risk (139;260), physical inactivity is clearly considered a "high" risk factor for osteoporotic fracture (160). A Cochrane review of randomized controlled trials to assess the effects of exercise in preventing and treating osteoporosis in menopausal women was published in 2002 (261). This review included 18 studies (only one of which included fractures as an outcome) with a total of 1,423 participants. The authors concluded that aerobic exercise, resistance exercises and walking were all more effective than no prescribed program of exercise for reducing loss of B M D at the spine, while walking was effective at the hip. A similar meta-analysis that was published two years earlier and included all but two of the studies in the Cochrane review, as well as an additional eight RCT studies in premenopausal women, reached a similar conclusion about the effects of exercise in premenopausal women; the relative effectiveness of exercise on spinal B M D was comparable in premenopausal and menopausal women (262). The potential impact of exercise therapy on fracture risk is less definitive. A recent review of trials that included exercise therapy as an intervention and fracture as an outcome identified only a handful of trials that met the eligibility criteria: subjects who would typically be considered at high risk for fractures (including all elderly people and menopausal women), fracture as an outcome and follow-up longer than eight weeks. Three trials of exercise in mid-aged populations were identified; all three included menopausal women. A meta-analysis of these three trials, which incorporated only 322 women when combined, showed a non-significant reduction in spinal fractures with the intervention (RR = 0.52, 95% CI = 0.17-1.60). The authors concluded that it is still unclear whether exercise interventions reduce the risk of fracture and that more RCTs are required to evaluate their effectiveness (263). RCTs of exercise 31 interventions have tended to be small, of short duration and are challenged by high rates of non-compliance or drop-out. Despite the dearth of evidence from prospective studies for the effectiveness of exercise interventions on fracture risk, there is no doubt that the potential physical and mental health benefits of regular exercise in mid-age reach well beyond osteoporosis and fracture prevention. Current Canadian guidelines recommend that women and men should be encouraged to participate in exercise as a preventive measure against osteoporosis (85;87). 2.4.4 Smoking The majority of studies that have investigated the effects of smoking on bone have found a positive association between smoking and reduced B M D (154;158;258;264) as well as increased fracture risk (107;159;219;255); but other studies have found no association between smoking and fracture risk (139;265). The effects of smoking on bone may be modified by other factors; there is evidence that smoking has a greater effect on B M D or fracture risk in those with a lower B M I (266), advancing age (267) and male gender (158) and that there is an interaction between smoking and OHT use in menopausal women (264;268). A recent meta-analysis of smoking and risk of fracture in men and women from 10 prospective cohorts, including CaMOS, showed that current smokers were at greater risk of low trauma fracture compared with non-smokers and that this association was partly independent of B M D and BMI. Further, the relative risk of smoking was greater in men than in women for osteoporotic fracture at sites other than the hip, but the relative risk for hip fracture was comparable between men and women (269). A n earlier meta-analysis of 19 cohort and case control studies and 29 cross-sectional studies reached a similar conclusion; smoking was associated with lower B M D and with fracture in menopausal women and this association was independent of low B M I and less exercise participation (267). These authors estimated that 13% of hip fractures in women are attributable to smoking and that the relative risk of hip fracture in smokers compared with non-smokers increases with advancing age. Similar patterns were observed in men, although the data were limited (267). The effects of smoking on B M D and on fracture risk are greater in current smokers than past smokers (264;267;269;270), which suggests that smoking cessation has the potential to improve risk status. Smoking is included as a risk factor for osteoporosis by guidelines in Canada and the U.S. (85;89); European guidelines have included discussion of the potential positive influences of smoking cessation on bone health but have not gone as far as to list smoking as a risk factor (90; 172). 32 2.4.5 Alcohol Although moderate alcohol intake may have a protective effect and has been observed to be associated with increased B M D in women and men (144;162;271), excessive alcohol intakes have consistently been found to be predictive of low B M D (272;273), loss of B M D over time (158) and fracture risk (136;274-276). A heavier intake of alcohol (more than two drinks per day) has been implicated as a risk factor for osteoporosis and fracture in both women and men. A recent international analysis of 5,939 men and 11,032 women from three cohorts, including CaMOS, determined that alcohol intake had a nonlinear effect on the risk of fracture; an increased risk of hip or any low trauma fracture was evident in women and men who consumed more than two alcoholic drinks per day. This increased risk appears to be at least partly independent of B M D and smoking (277). Excessive alcohol use is currently listed as a risk factor for osteoporosis by Osteoporosis Canada (86). 2.4.6 Caffeine Excessive caffeine intake is believed to increase the risk of osteoporosis; the effect may be mediated by its interference with calcium retention and its association with low calcium intake. There is evidence from retrospective and prospective cohort studies for a moderate association between a high caffeine intake and fracture risk or low B M D (77;184;275;278-280), although the vast majority of the subjects studied have been women. Other prospective studies have found no association between caffeine intake and either bone loss (158) or fracture risk (219). It has further been suggested that the effects of high caffeine intake are only evident in those with low calcium intakes (278;279;281). High caffeine intake (> 4 cups of coffee a day) is regarded as a minor risk factor for osteoporosis (85;282) or as a lifestyle factor that should be modified in preventive interventions (87) in some published guidelines. Other guidelines do not mention caffeine at all (88-90; 172). 2.4.7 Medical Therapy OHT (estrogen with or without progesterone) was the recommended first choice of treatment or prevention of osteoporosis in menopausal women during the time of the CaMOS baseline and Year 3 follow-up data collection (230). OHT has consistently been found to be effective in the prevention or slowing of bone loss and in the reduction of the risk of osteoporotic fracture (107;163;184;213;283). The Women's Health Initiative (WHI) randomised controlled primary 33 prevention trial of estrogen plus progestin vs. placebo was stopped in 2002 when an increased risk for invasive breast cancer in the treatment arm was detected. Other significant risks of OHT that were identified by this trial were increased risks of coronary heart disease, stroke and pulmonary embolism. On the other hand, the risk of hip fracture, as well as colorectal cancer, was significantly reduced in those treated with combined OHT (81). It is now generally agreed that the risks of OHT are likely to outweigh its benefits and it is no longer recommended as the first choice of therapy for osteoporosis (85;214). Two bisphosphonates, alendronate and etidronate, were approved in Canada for the prevention or treatment of osteoporosis when the CaMOS baseline was conducted. The selective estrogen receptor modulator (SERM), raloxifene was approved for the prevention and treatment of postmenopausal osteoporosis, in 1998, before the three-year follow up of the CaMOS cohort began. Calcitonin was also available and reportedly being used as an alternative treatment, although it was not specifically approved as a medical intervention and therapy for osteoporosis (230). Bisphosphonates, raloxifene and calcitonin have been found to effectively increase and prevent loss of spinal bone density and fracture risk in menopausal women (284-293). There is evidence that alendronate is effective in increasing B M D and reducing vertebral fracture risk in men (294;295), while cyclical etidronate has been found to prevent bone loss in men with corticosteroid induced osteoporosis (296;297). There is limited information about the effects of calcitonin on men (297). Other agents have been approved for the treatment or prevention of osteoporosis since the completion of the three-year follow-up, such as risedronate and parathyroid hormone. Androgen therapy has been found to improve B M D in men with osteoporosis that is secondary to hypogonadism (low levels of testosterone); however, the majority of men with osteoporosis have normal gonadal function (297). None of these medications is without potential drawbacks; all have associated side effects that contribute to low rates of adherence. 2.4.8 Summary of Prevention and Treatment Recommendations Appropriate lifestyle choices, such as adequate calcium intake and exercise, and smoking cessation are indicated for everyone as a means of maintaining or achieving optimal health, including bone health; in theory, these lifestyle decisions should not be influenced by measurements of bone density (57;85;87;89;172). Guidelines from Osteoporosis Canada (86) emphasize adequate calcium and exercise as preventive interventions and list excessive caffeine and alcohol consumption and smoking amongst the risk factors for osteoporosis. The Canadian 34 Task Force on Preventive Health Care (87) recommends that menopausal women, irrespective of their diagnosis from a B M D test, should have their lifestyle assessed. Physicians and other health professionals in Canada should be expected to stress the importance of these health behaviours to women and men with low B M D test results in particular, either as the only intervention or as an adjunct to pharmacological therapy. In turn, receipt of a low bone density measurement may be expected to motivate people to change their lifestyle in order to reduce their fracture risk. In addition to its general preventive role, calcium is specifically recommended in many countries as a treatment (together with other interventions) for osteoporosis (90), and both calcium and vitamin D are recommended as an adjunct to treatment for osteoporosis in Canada (85;228). The Canadian Task Force on Preventive Health Care recommended in 2004 that menopausal women who are found to have osteoporosis should be treated with a bisphosphonate or raloxifene; if these cannot be tolerated then OHT or calcitonin should be considered. Treatment, other than lifestyle evaluation and modification i f indicated, is not recommended for women with osteopenia i f they are under age 65. The lifestyle recommendations include adequate calcium intake (1,000-1,500 mg/d), adequate vitamin D intake (400-800 IU/d), sufficient exercise (three times per week for at least 20-30 minutes each time), moderate caffeine intake (fewer than four cups of coffee per day) and smoking abstinence (87). Bisphosphonates are recommended as the first-line treatment for men with osteoporosis (or low bone mass) (85;297), but are not recommended for treatment of premenopausal women except in the presence of identified secondary causes of osteoporosis (85). At the time of the CaMOS baseline between 1996 and 1997, there were limited recommendations and guidelines for the management of women with osteoporosis, and osteoporosis in men was barely addressed. The Osteoporosis Society of Canada recommended in 1996 that sufficient intake of calcium and vitamin D should be attained and that people should be counselled to exercise. In addition, high alcohol intake and smoking were recognized as high-risk behaviours for osteoporosis. OHT was recommended as both the preventive and treatment therapy of choice for menopausal women and bisphosphonates (alendronate and etidronate were approved at the time) were cited as a reasonable alternative for women with established osteoporosis, and for men. Other agents that have been used to prevent or treat osteoporosis, such as raloxifene, calcitonin, sodium fluoride, calcitriol and cyclical clodronate were not specifically approved for these purposes at the time of the CaMOS baseline but were available by prescription (230). 35 2.5 Awareness of Test Results Following B M D Testing It is generally accepted that patients who are informed and knowledgeable about their health and about specific diseases that may affect them are more likely to be proactive in seeking information, to make positive changes to their lifestyles, to initiate medications when appropriate, and to adhere to physicians' recommendations or prescriptions. Hence we would expect individuals who are aware that they are at greater risk for fracture and those who are more knowledgeable about osteoporosis in general to be more likely to make positive changes than those who are less knowledgeable or are uncertain about their risk. Although an increase in knowledge (or awareness) does not always lead to a change in behaviour, it is likely that some kind of knowledge is necessary before a conscious decision to make a change in health behaviour will be made. Knowledge is therefore typically considered to be a necessary, but not a sufficient factor for health behaviour change and many intervening factors are presumed to be involved in the process of behavioural change as explained in the following summary by Green and Rreuter: "Behaviour may not change immediately in response to new awareness or knowledge, but the cumulative effects of heightened awareness, increased understanding, and a greater command (recognition and recall) of facts seeps into the system of beliefs, values, attitudes, intentions and self-efficacy, and eventually into behaviour." 1 3 There is a vast body of literature concerned with the role of factors such as beliefs, values, attitudes, intentions and self-efficacy in the complex pathway between knowledge and health behaviour. The scope of the following review is limited to studies that have focused on the specific assessment of correct knowledge of B M D test results and its association with information seeking and health behaviour or uptake of, and adherence to, medical therapy. Estimates of correct knowledge or awareness of results (or diagnosis) after bone density testing have mostly been derived from retrospective studies of women referred for testing by their physicians (3;299;300). A limitation of such retrospective studies is that the women who took more notice of their results, and were therefore more likely to be correct, may also have been more likely to respond to the invitation to take part in the study, which would tend to give an inflated estimate of correct awareness. Furthermore, the definition of "correct" awareness of 1 3 From Green, L.W. and Kreuter, M.W. (1991) Health Promotion Planning. An Educational Approach. 2nd Edition., p. 156. (298) 36 test results has varied and this has a potentially large influence on the apparent awareness that screened women have of their test results. In Rubin and Cummings' widely cited retrospective study of 327 women (aged between 23 and 96 years) who had been referred for a B M D test seven months previously, 85% of the participants who reported that they had received a "low" test result (defined as a B M D value that fell below the age-, sex- and race-adjusted mean) were correct (3). The proportions of women that incorrectly reported that their test result was "normal" or that reported that they did not know their results were not documented; the assessment of correct awareness of test results was therefore incomplete. The women in the study were predominantly white and menopausal with a high level of education and high prevalence of risk factors'for osteoporosis, which limited the generalizability of the results; these characteristics of the sample may explain the reported high rate of correct awareness of test results. A more recent study reported that one third of 956 women aged 46 - 90 years who had received a diagnosis of osteopenia or osteoporosis after referral for B M D testing and then had initiated OHT, raloxifene or a bisphosphonate, were unclear about their test results (whether they had osteopenia or osteoporosis) between four and twelve months after the initiation of therapy (299). Correct awareness was low, even though this sample would be expected to be biased towards women who were more likely to be correct about their test results due to the selection of only those who had initiated therapy, and therefore must have discussed their results with a physician. Medication adherence was found to be associated with self report of a diagnosis of osteoporosis; those who were uncertain about their diagnosis, or who believed that it was not osteoporosis were 1.6 times more likely to discontinue their medication. This latter subgroup, however, included the women who were correct about a diagnosis of osteopenia, who may have appropriately had less reason to adhere to therapy. Despite limits to the interpretation of the results due to the grouping together of those women who were correct about their osteopenia results with women who were incorrect about their osteoporosis results, the findings demonstrate that perceived osteoporosis disease status is a factor in the discontinuation of therapy, which implies that a correct awareness of test results could also be a factor in the initiation and maintenance of bone-specific health behaviour change. In another study, only 48% of 977 women and 37 men (aged 34 - 93 years) were able to provide their correct diagnosis 18 months after they had been referred for D X A testing (300). Responses in this case were counted as correct if: (a) the participants who were informed of a diagnosis of either osteopenia or osteoporosis reported a low test result and (b) the participants 37 recalled a normal test result as such. This estimate of correct awareness in the referred population is likely to be more representative than those derived by the two studies above (3;299) because of its more complete definition of correct reporting as well as inclusion of referred patients with both normal and low test results. Just less than 50% correct awareness of test results is a disappointing rate for a referred sample. Furthermore, the authors did not include the 20% of participants who returned a questionnaire but declined to answer this question. If these selective non-responders are assumed to have been unaware or unsure of their test results, correct awareness of test results in the sample drops to as low as 38%. As has been observed in other studies (299;301), the participants who received low B M D test results were less likely to be correct than those with normal results; in particular those with a diagnosis of "osteopenia" were the least likely of all to be correct about their diagnosis (31% were correct or 26% i f the non-responders are included). Despite the low rate of correct responses, 80% of the participants in this study reported that their physician had passed on the results from the test.- This suggests that either the communication between the physician and patient (or the "patient education") was at fault of that the participants were not sufficiently interested to take notice of their results or to retain the information. Thus, although it is assumed that test results must first be communicated to patients before they can be aware of their risk, even when results are communicated, patients may not always understand or commit them to memory. The participants who were correct about their low test results, compared with those who were incorrect about low test results, were more likely to have been prescribed and to have adhered to medication. This analysis was not stratified by diagnosis, however, so there is potential confounding by a diagnosis of osteopenia, which was associated with more incorrect responses and also carries a lower indication for therapy and arguably greater justification for non-adherence. Although this study is unique in that the sample included men, no information specific to the men was provided; the results were not stratified by sex (presumably because of the small number of men in the study) (300). A Canadian study has similarly reported that women are less likely to be correct about low B M D test results than about normal results; "osteopenia" was the least frequently correctly reported diagnosis reported by a prospective cohort of Canadian women aged over 50 years who had been referred for B M D testing (301). Fifty one percent of the women with an osteopenia diagnosis thought that their results were normal when interviewed by telephone three months after testing and a further 25% were unclear about their results. In contrast, only 6.4% of women with a diagnosis of osteoporosis thought that their results were normal, while 50% were unclear. Women who had a normal result were the most likely to be correct; none of them reported a low 38 test result and 30% were unclear of their result. Women who were correct about their osteoporosis diagnosis were more likely to have initiated OHT or bisphosphonates than those who were incorrect. The authors interpreted their results as indicating that communication between the physician and patient is crucial to therapeutic decision-making (301). It is not, however, possible to determine the cause: Patients may have taken prescribed medications because they had a better understanding of their low test results and associated risk for fracture or they may have paid more attention to their results because their physicians recommended therapy. Pickney and Arnason (300) commented that older age of the patient, specialty of the physician and time since the D X A were not associated with correctly reporting test results in unadjusted analyses. Except for the diagnosis from B M D testing and whether the results were discussed with a physician, other potential predictors or confounders of correct awareness of test results (such as education level and risk factors for osteoporosis) have not been addressed by previous researchers (3;299-301). The results from these observational studies of referred samples (299-301) suggest that the perception of low test results in women is associated with uptake or adherence to therapy, and that improvement in communication between physicians and patients or patient education may lead to improved appropriate medication uptake. Such results further imply that improved communication of B M D test results or diagnosis could be associated with positive lifestyle changes. The only previous study of correct awareness of B M D test results in a non-referred sample included 515 mid-aged women (aged 45-54 years) who were randomly selected from a population health register in Aberdeen, Scotland to participate in a randomized controlled trial to assess the effect of direct-to-participant feedback on knowledge of test results (302). The cut-off for a report of a low bone density for this study was the lowest quartile of the sample; these women were informed that their B M D was below average (in the lowest 25%) 1 4 and that they may be at risk of osteoporosis in later life. Two years after testing only 43% of the women were able to correctly recall whether they were at increased risk; this is comparable to the rates of correct awareness reported in referred samples (300;301). Also in line with previous studies, the 1 4 The study by Campbell et al. was conducted before the release of the WHO diagnostic guidelines for osteoporosis (50). Other investigations of the effects of B M D feedback in premenopausal women, conducted since the release of the W H O diagnostic criteria, have also utilized alternative, less stringent, methods of defining higher risk for fracture in their feedback (303-306). Few premenopausal women meet the W H O criteria for high risk ("osteoporosis"), furthermore, the criteria were derived for use with menopausal women, although they are now often applied to premenopausal women and men in clinical practice or research. 39 25% who were told that they were at greater risk were less likely to be correct; only 26% of the women in the lowest quartile were able to recall their low result. 2.6 Information seeking and Knowledge about Osteoporosis Seeking out or accessing information would be expected to lead to an increased level of knowledge about osteoporosis, particularly i f the baseline level of knowledge in the population is low. In general, studies that have assessed knowledge about osteoporosis have demonstrated only poor to moderate levels of knowledge about the consequences of osteoporosis, its risk factors and preventive interventions (307). Women (308-310), people with a higher education (308;311-313), and those with a family history of osteoporosis (314;315) are the most likely to be knowledgeable about the disease. People who have previously received information about osteoporosis are also more knowledgeable (312;316), which indicates that information seeking about osteoporosis is likely to improve knowledge about osteoporosis. A recent presentation at the 2006 American Society for Bone and Mineral Research Conference reported that men aged 40 to 79 years who had been referred for D X A evaluation and diagnosed with osteoporosis or osteopenia were no more knowledgeable about the risk factors for osteoporosis than were matched controls who had not received a D X A test or an equivalent diagnosis. Knowledge about risk factors was low overall in these men who were from predominantly rural communities, but was particularly poor regarding the role of smoking and excessive alcohol consumption. Diagnosis was not associated with knowledge, even though men who had been diagnosed with osteoporosis or osteopenia were less likely to report a low calcium intake or excessive alcohol use. Although no causal relationships between diagnosis and changes in knowledge and health behaviour can be inferred from this case-control study, because baseline levels of these variables were unavailable, it is the only available report of levels of knowledge about risk factors in a population of men who have been deemed to be at high risk (317). As pointed out by Werner (307) in her review of studies of people's knowledge of osteoporosis, the assessment of knowledge is based on the assumption that increased knowledge will be associated with increased participation in preventive behaviours. A logical extension of this assumption is that information seeking about osteoporosis will lead to increased knowledge and may have an indirect impact on preventive behaviours. It has been proposed that health information seeking behaviour could be regarded as a process that may lead to specific health behaviours, or that it may define people who are considering health behaviour change (318). 40 Efforts to obtain information about health in general have been found to be associated with positive health behaviours and fewer risky behaviours (318;319). Both gender (being female) (318;319) and older age (319) are predictive of more information seeking about health. People who participate in (or who are seriously considering participating in) osteoporosis-related health behaviours such as increased calcium intake or calcium supplement use (316;320-324), regular physical activity (320;323-325), and use of OHT (316;320), as well as those who demonstrate a greater likelihood of stopping steroid use (316), have been shown to have increased knowledge about osteoporosis in cross-sectional and intervention studies. It is evident however that knowledge alone does not necessarily influence health behaviour (326). Thus, the association between increased osteoporosis knowledge and positive health behaviours, or even the intention to change behaviour, has not been consistently identified. In one well-designed intervention study involving a large sample of randomly selected premenopausal women from the Tasmanian population, a group education intervention led to increased knowledge about osteoporosis, but this greater knowledge was not associated with increased self-efficacy regarding calcium intake and physical activity either six months, or two years, after B M D testing and education (303). A low B M D test result (defined as a mean T-score < 0) was similarly not associated with increased self-efficacy (303). Interestingly, women in the low B M D group showed a gradually increasing level of knowledge between six months and two years after testing compared with the normal group, which suggests that they may well have sought information over time as a result of their B M D test results (303). No association was found between knowledge and dietary calcium intake or Weight bearing exercise participation in two other studies involving young college aged women (327;328) and in a small intervention study of three distinct groups of women (31 young college students, a community sample of 35 women aged 22 to 83 years, and a group of 18 mid-aged nurses) in which each group received a different intervention (329). It is notable that studies that have found no association between increased knowledge and health behaviour change have most often involved younger women. A possible explanation for this discrepancy with studies of older women is that younger women may be less motivated to make behavioural changes, even when they are aware of the risks, because the threat of osteoporosis is more distant and may seem less relevant to them (330). Other authors have examined the association between whether subjects have discussed their B M D test results with a physician and their behaviour change or medication uptake and adherence. Discussion of B M D test results with a physician may represent a form of information 41 seeking (if consultation were initiated by patients or if patients sought information during the consultation). It is also possible that a physician would have initiated the consultation and that the transfer of information was also initiated by the physician. The latter scenario may be more likely in cases where the subjects had originally been referred for a B M D test by the physician as opposed to self-referred. A large proportion of women who have been referred for B M D testing report that they have discussed their test results with their physician. Ninety-five percent of women aged between 23 and 96 years in a retrospective sample that was randomly selected from those referred for B M D testing in the San Francisco area recalled that they had discussed their results with a physician or health-care provider (3), similarly 80% of retrospectively sampled women and men who had been referred for B M D testing at a Wisconsin health centre reported having received their test results from their physician (300). Likewise, a prospective study of women referred for D X A testing near Boston found that 79% of the participants had discussed their results with a physician (331). Because these were referred populations, the rates of a low diagnosis (either osteopenia or osteoporosis) were relatively high; ranging from 53% to 83% of the sample (3;300;331), suggesting that other risk factors may have been present and which may have influenced these women's sense of susceptibility. Non-referred subjects who have been invited to attend B M D testing as part of a study would be expected to have lower rates of low test results and hence potentially lower reported rates of discussion of test results with a physician. Only 33% of a convenience sample of premenopausal women who underwent testing said that they had discussed their test results with a physician. Twenty percent of the sample had been informed that they had low test results (Z-score < -1.0), however, and 76% of this subgroup had discussed their results with a physician, which indicates that, appropriately, diagnosis is a major factor (306). Although no association was reportedly found between discussion of the results with a physician and positive lifestyle change, it is not clear how the authors defined a change in health behaviour, and the number of women who changed their behaviour was not reported. Discussion of B M D test results with a physician or other health-care provider has been shown to be associated with the initiation of therapy (OHT or bisphosphonates) in prospective studies of Canadian women aged 50 years and over who had been referred for testing (301), and mid-aged women (aged 54 to 65 years) who volunteered to participate in a study on osteoporosis prevention (332). These findings suggest that those who discuss their results with a physician are more likely to start medication, although cause and effect cannot be established. It is not 42 possible to determine if the consultation influenced the participant's decision to start medication, or whether the physician initiated the consultation because of low test results and indication for therapy, or indeed if the participant initiated the consultation because of an interest in initiating therapy or lifestyle change. Furthermore, these studies (301 ;332) did not address whether the initiation of medication was appropriate; consultation or discussion with a physician is only potentially valuable i f it leads to appropriate treatment or behavioural change. Appropriate treatment following B M D testing was taken into account by Solomon et al. (331) in their study of a referred sample of mostly menopausal women. Bisphosphonates or calcitonin were much more likely to be taken by the women who received a diagnosis of osteoporosis and somewhat more likely to be taken by the women with a diagnosis of osteopenia. No women in the normal B M D group initiated bisphosphonate or calcitonin therapy, which was considered appropriate. In this homogeneous group of white women, many of whom had other risk factors for osteoporosis, OHT use did not appear to be influenced by the B M D test results; the women who had received a normal result were just as likely to be taking OHT as were the women who received a diagnosis of osteoporosis.15 Consultation with a physician was not associated with initiation of appropriate therapy. The women who reported that they understood their test results, however, were more likely to have initiated appropriate therapy than were the women who stated that they did not understand their results (331). This suggests that simply consulting with a physician is not sufficient; the quality of the discussion or consultation is likely to play a role in patient understanding, initiation of health behaviours and medication uptake. The sources of osteoporosis information that have been accessed by particular populations have been described by three previous studies. The findings have indicated that information sources that men and women use to learn about osteoporosis are not limited to the doctor's office. In fact, other sources of information about osteoporosis have been cited at least as frequently as have health professionals by Canadian seniors (310) and more frequently than health professionals by young college women (328) and women who were referred for (but not yet attended) D X A testing (333). Television, newspapers, books, friends, and family physicians were each cited by approximately 30% of seniors as common sources of osteoporosis information in a community-based convenience sample of women and men (310). Other sources, such as family members 1 5 This study was completed before the results of the Women's Health Initiative were available, when OHT was amongst the first line of therapy for the prevention of osteoporosis and heart disease, and for the relief of menopausal symptoms. 43 and the internet, were much less frequently mentioned. Women in this sample had a significantly higher awareness and knowledge about osteoporosis than had the men, even though the same number of women and men cited each of these top five sources. The likelihood of having accessed information from any source (or the number of people who had not sought information at all) was not presented (310). A study of college women with a mean age of 19 years reported media as the main source of information about osteoporosis. Seventy-nine percent of these young women who had received information about osteoporosis reported that it had been acquired from television programs, 68% had gained information from magazines and 43% reported receiving information from a doctor or other health-care provider (328). Fifty-seven percent of the women who had been referred for D X A testing at two Canadian centres had initially "heard about osteoporosis" through newspapers and magazines, but only 11 % had acquired such initial information from a health-care practitioner (333). The descriptive results from these three very different populations suggest that media sources such as television, newspapers and magazines are heavily relied upon as sources of information about osteoporosis and are more likely to be used as sources than are family physicians. The potential impact of B M D testing on information seeking about osteoporosis was assessed in a study of a convenience sample of women who were recruited through advertisements (334). This descriptive study found that, of the women whose test results fell in the lower third for their age, 72% sought information from either a health-care provider or another source while only 42% of those whose results fell in the upper third for their age sought information. Although this sample is likely to represent more highly motivated women than the general population, the results suggest that the reports received following B M D tests can play a role in prompting information seeking (334). In summary, the bulk of previous studies have indicated that increased knowledge about osteoporosis, or receipt of information about osteoporosis, is associated with an increased likelihood of positive health behaviour or medication uptake. So far, there has been no information about what predicts information seeking following B M D testing, or the preferred sources of sought information, in a population-based sample of women or men. People who receive low B M D test results and understand that they are at increased risk for fracture may be expected to seek information about osteoporosis as a means of learning about lifestyle modification options and the availability of medications that may decrease their risk; furthermore, such information seeking behaviour is likely to be a step in the process towards a 44 change in health behaviour. On the other hand, it is plausible that receipt of a low test result can lead to avoidance of more information when it causes anxiety or mental discomfort. Such avoidance of information has been documented, particularly with regard to genetic testing for cancer risk (335). In the case of osteoporosis, a lack of information seeking behaviour may indicate either avoidance of information, disinterest or lack of need (i.e., the level of knowledge is already high). 2.7 Behavioural Change Following B M D Testing The majority of studies that have attempted to determine whether bone-specific changes in behaviour follow the receipt of low BMD test results have included women who have been referred by their physicians for BMD testing. Many have used convenience samples, often recruited retrospectively, which limits the generalizability of the findings to these specific populations. Table 2.2 summarizes the published research that has specifically investigated the impact of low BMD test results (or awareness of low test results) on bone-specific health behaviour. The publications in Table 2.2 are limited to those that have specifically examined BMD test results as a predictor and health-behaviour change as an outcome. Studies of the effects of other interventions, such as educational materials on lifestyle changes, have not been included unless BMD test results were also considered as an explanatory variable in the study. Studies that have measured change after receipt of low BMD results, without comparison with changes made following normal or higher test results (i.e., no comparison group) have not been included. 45 T a b l e 2.2: P r e v i o u s S t u d i e s o f t he E f f e c t s o f L o w B M D T e s t R e s u l t s o n B o n e - S p e c i f i c H e a l t h B e h a v i o u r Authors / Study Population Baseline Measures / Definition of Low Time: Outcomes Findings: Sample Size / Age Adjustment for Test Results in B M D Association between Low Test at Follow-up Confounders / Feedback/ Test to Results and Health Behaviour Analysis Independent Follow-Variable (IV) up Cook et al., - Women No behavioural Above average, 1 year Self reported change: +ve association (below average v. 1991 (336) - Prospective baseline measures. average or below -Diet average/above average): - Volunteer sample average for age. - Exercise - Change in diet n = 380 from 4 medical Other predictors - Alcohol - Change in exercise practices considered: Age, IV: Feedback of - Smoking Income, Education. average, above - Other habits No association with: Age 30+ years average or below - Smoking Bivariate (Chi- average results. - Alcohol consumption squared). (No - Other habits adjusted analyses) Rubin and - Women No behavioural Above ("normal") 4-24 Self reported change: +ve association (low v. nonnal): Cummings, - Retrospective baseline measures. or below ("low") months - Ca+ supplements - Started Ca+ supplements 1992 (3) - Randomly mean for age. - Vitamin D supplements - Started Vit. D supplements sampled from all Covariates: (mean = - Milk/Ca+ rich foods - Increased milk/Ca+ rich foods n = 261 those referred for Education; Family 7 mths). - Exercise - Started or increased exercise B M D test during History; Previous IV: Self report of - Coffee, tea or cola - Reduced coffee, tea or cola 14 mth period. fracture; normal or low test - Alcohol - Started estrogen therapy comorbidity; results - Smoking - Started other therapy - Age 23-96 years perception of risk. - Increased worry about (mean age = 59) - Estrogen therapy osteoporosis Bivariate (Chi- - Other therapy (sodium fluoride or - Increased fear of falling • squared) and some calcitonin) multivariable No association with: regression - Worry or fear of falling - Decreased alcohol intake - Smoking cessation OS Authors / Sample Size at Follow-up Study Population / Age Baseline Measures / Adjustment for Confounders / Analysis Definition of Low Test Results in Feedback/ Independent Variable (TV) Time: B M D Test to Follow-up Outcomes Findings: Association between Low Test Results and Health Behaviour Holt et a l , 1997 (337) n = 312 - Women - Retrospective -Convenience sample from those referred for B M D testing - Aged 39 - 88 (mean age = 62) No baseline measures No adjustment for confounders Bivariate tests of association Feedback not described IV: Self report of normal or low test results Up to 6 months Self reported change: - Calcium - Exercise -Diet - Estrogen - Bisphosphonates - Calcitonin -i-ve association (low v. normal1): - Increased calcium intake - Increased exercise - Changes in diet - Started medications (estrogen, bisphosphonates or calcitonin). Jones and Scott, 1999 (305) n = 271 - Premenopausal women - Prospective - Convenience sample from another study (young mothers, more likely to be smokers, less likely to have breastfed) - Mean age = 33 Baseline data not comparable to follow-up (collected during pregnancy and 6 years before D X A ) Bivariate tests Normal (T-score > -1.0) or Low (T-score<-1.0) B M D . IV: Feedback of low or normal results 1 year Self reported change: - Use of Ca+ supplements - Calcium intake - Physical activity - Smoking Continuous measures at follow-up (unadjusted for baseline): - Calcium intake by food frequency questionnaire - Exercise questionnaire +ve association (low v. normal): - Increased Ca+ supplement use - Increased Ca+ intake - Increased physical activity - Higher calcium intake by food frequency questionnaire - Lower overall sports participation by questionnaire No association with: - Smoking cessation Authors / Study Population Baseline Measures / Definition of Low Time: Outcomes Findings: Sample Size / Age Adjustment for Test Results in B M D Association between Low Test at Follow-up Confounders / Feedback / Test to Results and Health Behaviour Analysis Independent Follow-Variable (TV) up Jamal et al., - Premenopausal Baseline measures Normal (Z-Score 1 year Change in behaviour relative to +ve association (low v. normal):! 1999 (306) women taken, using different > -1.0) or Low (Z- baseline (derivation of outcomes - Use of calcium supplements methods to follow- Score<-1.0) used for the analysis of association - Use of Vit. D supplements n = 669 - Prospective up. B M D . with diagnosis was not defined by - Convenience authors). sample recruited Covariates: Age, No association with: via adverts (79% weight, FH, IV: Feedback of - Use of Ca+ supplements - > 3 km walking / week university or education low or normal - Use of Vit. D supplements - > 1 cup milk / day college educated, results - > 3 km walking / week - Smoking 41% had F H of Unadjusted - > 1 cup milk / day - > 1 alcoholic drink / day osteoporosis) regression - Smoking - > 3 cups caffeinated drinks / day (presented) and -_> 1 alcoholic drink / day - Age 18-35 years adjusted regression - >_ 3 cups caffeinated drinks / day (mean age = 28) (data not presented) Rimes et al., - Women Baseline measures "Below fracture 1 week Behaviours undertaken "to prevent +ve association (low v. high): 1999 (334) taken into account. threshold" or not. and 3 or slow down bone loss" (open - Greater number of preventive -Prospective Cut-off for months ended question). Change behaviours n = 298 - Convenience No adjustment for feedback not determined by comparison with - At least one preventive sample recruited covariates. described baseline: behaviour via adverts - Calcium supplements - Uptake of calcium supplements Descriptive analyses, IV: Lowest V 3 - Other mineral or vitamin - Increased anxiety - Age 32-73 years Chi-squared, t-test compared with supplements - Increased perception of risk (mean age = 54) and A N O V A highest V 3 based -Diet on Z score - Exercise No association with: (analysis limited - Drug prescribed by doctor - Exercise ton= 180) -Other - Other mineral or vitamin supplements Changes in: - O H T uptake Worry, anxiety, perceived susceptibility 00 Authors / Study Population Baseline Measures / Definition of Low Time: Outcomes Findings: Sample Size / Age Adjustment for Test Results in B M D Association between Low Test at Follow-up Confounders / Feedback/ Test to Results and Health Behaviour Analysis Independent Follow-Variable (TV) up Marci et al., - Menopausal Baseline measures Normal (T-Score 15-55 Self report of changes to prevent +ve association flow v. normal): 2000 (338) women available; not taken > -2.0), moderate months. osteoporosis: - Started Ca+ supplement use - Retrospective into account in low bone mass (T- - Calcium supplements - Increased dietary calcium - Convenience defining change. Score < -2.0 and > (Mean = - Dietary consumption of milk / - Started exercise n = 701 sample of those -3.0), severe low 2.9 years) calcium-rich foods - Smoking cessation referred to an Covariates: Ed, age, bone mass (T- - Exercise habits - Decreased caffeine osteoporosis clinic history of OP, Score < -3.0). - Smoking habits - Initiated O H T previous fracture, - Caffeine intake - Initiated precautions to prevent - Age 50+ years comorbidity, F H of IV: Feedback of falls OP, worry about OP. normal or below - OHT - Increased worry about OP normal B M D Multivariable logistic - Limited activities to avoid falling regression to adjust - Safety precautions for confounders Rolnick et - Menopausal Baseline measured W H O definitions 6 months Self report of changes initiated in +ve association flow v. normal): al., 2001 Women but not taken into of normal, last 6 months since D X A test. - Increased vitamin D intake (332) - Prospective account in definition . osteopenia and - Calcium intake - Started OHT - Convenience of change. osteoporosis. - Vitamin D intake - Started other bone-specific sample invited - Diet or eating patterns medication' n = 508 from a care Covariates were not IV: Feedback of - Exercise organization taken into account in normal, osteopenia analyses of and osteoporosis Prescription data and self report: No association with: - Menopausal associations between - Started OHT - Increased calcium intake behaviour change - Started other medications - Modification of diet - Age 54-65 years and B M D test results. - Increased exercise Bivariate analysis (Chi-squared) 4^ Authors / Study Population Baseline Measures / Definition of Low Time: Outcomes Findings: Sample Size / Age Adjustment for Test Results in BMD Association between Low Test at Follow-up Confounders / Analysis Feedback/ Independent Variable (IV) Test to Follow-up Results and Health Behaviour Wallace et - Perimenopausal Baseline measures of WHO definitions 4 months - FFQ for total calcium +ve association Clow v. normal): al., 2002 and menopausal all.behaviours of normal, - Intake of total calcium (339) women osteopenia and - Ca+ intake meets recommended - Retrospective Covariates: Age, osteoporosis. daily intake; (Ca+ intake derived No association: n=129 - Convenience osteoporosis from FFQ dichotomised to more or -Intake > 1,500mgCa+ sample recruited knowledge, health IV: Feedback of less than 1,500 mg/day). - > 60 minutes exercise / week from women beliefs. normal, osteopenia referred for DXA and osteoporosis - Exercise scale; dichotomised to testing. Multivariable linear and logistic more or less than 60 minutes/ week. - Aged up to 65 regression, adjusted years for baseline (mean age = 54) behaviours and confounders Patel et al., - Men Baseline not taken WHO definitions Mean = 1 Self report of changes in: +ve association flow v. normal): 2003 (340) - Prospective into account in of normal, year - Started calcium supplements - Convenience definition of change osteopenia and - Calcium supplements - Started vitamin D supplements sample of patients osteoporosis. - Vitamin D supplements - Started bisphosphonates n=102 with prostate - Diet - Increased fear of falling cancer recruited Bivariate analyses IV: Feedback of - Exercise from urology only (no adjustment normal or - Smoking No association with: practice (24) and for covariates)- osteopenia v. - Caffeine intake - Change in dietary calcium healthy men feedback of - Use of bisphosphonates - Change in exercise patterns recruited from osteoporosis - Fall prevention habits - Change in alcohol intake adverts - Fear of falling - Change in smoking - Change in caffeine intake - Age 48-84 years - Limiting of activities related to falling © Authors / Sample Size at Follow-up Study Population / Age Baseline Measures / Adjustment for Confounders / Analysis Definition of Low Test Results in Feedback/ Independent Variable (TV) Time: B M D Test to Follow-up Outcomes Findings: Association between Low Test Results and Health Behaviour Hamel et al., 2005 (333) n = 1057 - Women - Prospective - Convenience sample recruited from women referred for D X A testing. - Mean age = 58 years Baseline measures of behaviours: Baseline calcium intake was taken into account in estimation of change in calcium in diet. No other adjustments for baseline measures. Other covariates: History of previous fracture, age > 45. Logistic regression WHO definitions of normal, osteopenia and osteoporosis. IV: Feedback of normal v. feedback of osteopenia or osteoporosis 3 months Self report of changes in: - Calcium intake - Exercise -Dietary calcium intake estimated from FFQ of calcium rich foods Self report of current use of: - calcium supplements - Vitamin D supplements - Bisphosphonates - Estrogen - Natural remedies +ve association (osteoporosis v. normal or osteopenia v. normal): - Self report increased calcium - Use of Ca supplements - Use of Vit.D supplements - Use of natural remedies - Use of bisphosphonates - Non-use of estrogen No association with: -Change in dietary calcium - Self report of increased exercise Winzenberg et al., 2006 (304) n = 415 - Premenopausal women - Prospective - Randomly selected population-based sample (64% response rate and 88% retention to follow-up) - Premenopausal -Age 2 4 - 4 4 years (mean age = 38) Baseline measures of all behaviours and covariates; not clearly taken into account in analysis of association between diagnosis and behavioural change Bivariate analyses Normal (T-Score >0)orLow(T-Score < 0) B M D . IV: Feedback of low or normal results. 2 years Self report of changes in: - Use of calcium supplements - Dietary Calcium intake - Physical activity - Smoking cessation Calcium intake and use of calcium supplements also derived from FFQ; physical activity also derived from questionnaires. Muscle strength and endurance fitness (objective measures) +ve association (low v. normal): - Started calcium supplements - Increased physical activity No association with: - Dietary calcium intake - Smoking cessation - Strenuous activity - Change in leg strength - Change in work capacity As can be seen from Table 2.2, the majority of previous investigations of the effects of B M D test results on health behaviour have assessed women; the single report that measured behavioural change in men recruited two thirds of its sample from men who had a history of prostate cancer (and two thirds of these men were hypogonadal and thus at high risk for osteoporosis) and the remaining third (men in the community) were recruited through advertisements (340). Virtually all of the studies either recruited women from a referred population or they used a convenience sample. Some of the referred samples were recruited retrospectively, after their test results were known (3;337;338), which may limit the generalizability of the results to more motivated individuals, who were more likely to have made changes because of their test results. There are no population-based studies of the association between B M D test results and bone-specific health behaviour change in perimenopausal or menopausal women, or in men. The association between B M D test results and behavioural change was investigated in a randomly selected population-based sample of premenopausal women in Southern Tasmania (304). Follow-up has been short term for the most part; only two studies included a mean follow-up period of more than one year after B M D testing (304;338). Although some of the authors listed in Table 2.2 reportedly measured pre-test levels of behaviour (such as the amount of calcium intake, exercise frequency, use of supplements, alcohol or caffeine intake) only three of these studies clearly documented that these values were taken into account when assessing at least some of the changes following B M D testing (333;334;339). Most often, the participants were simply asked to report i f they had changed these behaviours at the follow-up, whether or not the baseline behaviours had been measured (3;305;332;336-338;340). Holt et al. (337), for example, asked specifically, "What did you do differently because of your D X A test results?" This latter approach to questioning is subject to potential bias, not only because women with low test results may be more likely to report relatively more changes than women with normal results, just because they believe that the behavioural change is expected of them (recall bias), but also because behavioural changes that were not perceived by the women as due to B M D test results would not be documented (observation bias). In the large Canadian study reported by Hamel et al., a combination of approaches was used to assess behavioural change (333). A change in calcium intake was derived from the difference between the self report of current intake of calcium-rich foods at follow-up and the current intake reported at baseline, supplement and medication use were represented by reported use at follow-up with no adjustment for baseline, and subjects were asked simply i f they had 52 changed their calcium intake and exercise. Interestingly, diagnosis of osteoporosis was found to be associated with a self reported change in calcium intake in this sample, but not with the difference in calcium intake based on the food frequency questionnaire (FFQ). The assessment of change in calcium intake with the FFQ may be a more objective and less potentially biased method in this situation. On the other hand, the measurement of calcium intake based on a few select foods may have been too crude a measure to accurately reflect a change, as the authors suggested (333). Estrogen use was found to be significantly more frequent in the women who had been diagnosed with a normal B M D compared with those who had osteoporosis (333); the authors did not adjust their analysis for baseline differences and it is apparent that this difference is explained by more frequent use of estrogen in the "normal" diagnostic group, even before the D X A test. The much smaller study by Wallace et al. (339) included adjustment for baseline levels of health behaviours (calcium and exercise) in the analyses and found that neither dietary calcium intake nor exercise behaviour were associated with a low B M D test result: The only predictors of post scan calcium and exercise were pre scan calcium and exercise. It is feasible that these more objective measures of behaviour change, taking baseline levels into account provide a more accurate reflection of true behavioural change. There are other explanations for the observed lack of impact of B M D test results in this study, however, because the time between D X A testing and follow up was very short (only up to three months) and possibly too soon for behavioural changes to have taken place, and there was a particularly high attrition rate (50%), which limits the interpretation of the results. Rimes et al. (334) compared follow-up measures of specific behaviours that were reported in response to the question, "Do you do or take anything to try to prevent or slow down bone loss?" with responses to the same question at baseline to identify behavioural changes. Due to the wording of the question, information about participation in any of the behaviours for other (health or non-health) reasons was not available. Because subjects cannot be randomized to receive low or normal test results, baseline health behaviours would not be expected to be randomly distributed between B M D diagnostic groups and should be taken into account in any analysis of the association between test results and behavioural change. If women with low B M D results already participated in little exercise, or had a low baseline calcium intake, for example (as may be predicted given that lack of exercise and low calcium are risk factors for low bone density), they also have greater potential to make positive changes. Women who are already exercising several times a week on the other hand, have less room for improvement and ceiling effects may be observed. 53 Winzenberg et al. (304) comprehensively measured baseline behaviours and determined change in calcium intake and exercise both objectively, by comparing follow up behaviours to baseline, as well as by self reported change. These randomly selected premenopausal women were more likely to begin taking calcium supplements (as measured by the FFQ) and to increase their physical activity (by self report) i f they received a report of a low B M D test result, which suggests that young, low risk women would be prepared to make adjustments to their lifestyle in response to low B M D results. Furthermore, the authors found that after two years, bone mineral density had increased in the women who had commenced calcium supplements and in the women who reported persistent increases in physical activity relative to the women who did not report such lifestyle change. In contrast to the impact of a low B M D test result, an education intervention (small group education v. leaflets) did not have any effect on health behaviour or change in B M D . A l l of the studies listed in Table 2.2 that investigated change in the use of calcium or vitamin D supplements found that those who had been told, or believed, that they had low bone mineral density were more likely to start taking such supplements after they received low test results (3;304-306;333;334;337;338;340). On the other hand, changes to dietary calcium intake were observed to occur following low B M D test results in some studies (3;305;337;338), but not in others (304;306;332;333;339;340). Changes in exercise behaviour and decreased caffeine intake in response to low test results were inconsistently reported. The methods for defining the diagnosis, length of follow-up and definition of "change" likely influenced these outcomes to some extent. Two of the publications that found positive change in many of the behaviours, for example, used women's self reported diagnosis, rather than the actual diagnosis received, as the predictor variable (3;337). The women who perceived that they were at greater risk for osteoporosis were more likely to participate in bone-specific health behaviours (341) or adhere to bone-specific medications (299) than were the women who perceived themselves at lower risk or as "normal", irrespective of their actual B M D test results. Marci et al. (338) reported that women with a low B M D were more likely to stop smoking than were women with normal results. These findings, while encouraging, are surprising because smoking behaviour is notoriously difficult to modify. It is not clear, however, whether women who were diagnosed with a normal B M D were just as likely to quit smoking: The women were asked specifically if they had stopped smoking "to prevent osteoporosis as a result of having their first bone density test". These women may have stopped smoking for several reasons (in keeping with the secular trend), but those with a diagnosis of osteoporosis 54 were more likely to cite the diagnosis as the reason for quitting and therefore more likely to respond positively to this question. No other studies that have assessed smoking behaviour have identified any impact of low B M D test results (3;304-306;336;340), and alcohol intake has been similarly resistant to change (3;306;336;340). Factors that are associated with a low B M D test result, such as age, a family history of osteoporosis or a history of corticosteroid use may also be expected to be associated with behavioural change. People with other risk factors may have greater motivation to make changes in an attempt to lower their own risk of fragility fractures once they learn that they have a low B M D . On the other hand, those with other risk factors may have already modified their behaviours and would then be less likely to respond to a low B M D test result with a lifestyle change. In either case, such other risk factors for osteoporosis could confound the relationship between a report of a low B M D and health behaviour change. Marci et al. specifically reported the effects of factors other than diagnosis on behavioural change in multivariable adjusted models (338). They found that, in addition to a low B M D test result, a higher education level was associated with decreased caffeine intake, increased exercise and smoking cessation after testing and a history of a previous fracture was associated with decreased caffeine and increased dietary calcium intake (338). Pre-test knowledge and health beliefs about osteoporosis and age were considered as potential confounders by Wallace et al. (339), but none of these variables was found to be associated with behaviour change after B M D testing. A history of a previous fracture was considered as the other explanatory variable of interest (together with diagnosis by D X A ) by Hamel et al. (333); previous fracture was not associated with any lifestyle change, but may have influenced uptake of etidronate following D X A testing in women with low B M D test results. Subgroup analysis of only those women aged 45 years and over was carried out, but no further covariates were considered. Other authors who have reportedly adjusted their analyses for other risk factors have not reported the effects of these covariates (3;306). The role of the presence of other risk factors in the response to low B M D test results requires further investigation; there is limited information about whether lifestyle changes are more or less likely to occur in high-risk subgroups, such as those with a history of corticosteroid or anticonvulsant use, low body mass index or smoking. 2.8 Medical Therapy Following B M D Assessment Although receipt of low B M D test results has consistently been found to be associated with greater uptake of bone-specific medications, such as bisphosphonates 55 (3;332;333;337;340;342) and OHT (3;332;338;343-345), several authors have observed that uptake of medication has been lower than would be expected: It is not uncommon for individuals who have been diagnosed with osteoporosis, even if they have been referred by their physician for B M D testing, to fail to initiate medical therapy. Only 58% of women aged 45 years and over who had been referred for testing to one of two Canadian centres and then received a diagnosis of osteoporosis were taking a bisphosphonate or OHT six months after D X A testing (333). A survey of physicians who had referred women for B M D testing to a community hospital in Boston revealed that, although physicians had recommended calcium or vitamin D to a large proportion of the women whose results indicated osteopenia or osteoporosis, only 54% were prescribed medication (38% OHT and 16% bisphosphonates) (346). A similar proportion of women in California who were diagnosed with osteoporosis filled a prescription for OHT or a bone-specific medication; 50% of the women who had been referred by a physician and found to have osteoporosis apparently did not receive medical therapy (342). Given that adherence to both OHT (347-350) and bisphosphonates (347;351) has been shown to be poor, far fewer of these women would be expected to remain on these therapies. Other risk factors may play a role in decisions about therapy; women with a history of corticosteroid use or a previous fracture for example, were more likely to initiate therapy even when diagnosis was taken into account (342). Only 41%) of a sample of men (67% of whom had prostate cancer with the majority being hypogonadal and on androgen deprivation therapy) who received a diagnosis of osteoporosis had initiated bisphosphonate therapy within a year of receiving B M D test results; although the prevalence of other medical conditions in this atypical sample may have influenced the rate of treatment with medication (340). Even lower rates of medication uptake may be expected in people who have not been referred by their physicians for B M D testing. Indeed, an observational study of the effects of D X A testing on hip fracture reported that a sub-sample of the population-based cohort of community-dwelling women and men aged 65 years and older who participated in the Community Heart Health Study and were invited for D X A screening showed only minimal uptake of medications within a year of B M D testing after being informed that their results were below average for their age (352). Only 1% of the women and men with below average results started taking bisphosphonates, only 7% took up calcium supplements and, amongst the women, there was no difference in estrogen use between those who received average and those who received below average results (352). Likewise, in another study, only 18% of women (mean age 59 years) and men (mean age 63 years) who were reported to have T-scores < -2.0 after being invited for a heel ultrasound, while attending a health fair, had initiated treatment of estrogens, bisphosphonates or calcitonin six months later, even though the test results had been reported directly to their physicians (353). This latter study compared therapy following low B M D test results in women and men; no difference was evident between the number of men and women who initiated therapy following a low B M D test result, although generalizability is limited because of the sampling method (353). Differences have been observed in the treatment of men and women following moderate or low trauma fractures; men are less likely than women to undergo appropriate further investigation (such as a D X A test) or to be treated for osteoporosis following such a fracture (354-357). The proportion of women and men who have received medical therapy for osteoporosis, or even been referred for B M D testing, after experiencing a low trauma fracture in middle age or later has consistently been found to be well below expectations (163;355-357). Clinical guidelines emphasize that low trauma fracture is a major risk marker for further fracture in women and one that requires appropriate evaluation for osteoporosis and treatment (85;89;230); but the association between previous fracture and osteoporosis risk in men, as well as guidelines for assessment of risk in men are less well established (97). 2.9 Direct-to-Participant Feedback of Test Results It is feasible that directly informing women, as well as men, of their test results may prompt changes in bone-specific lifestyle behaviours. Individuals who have already received information about their test results and their personal risk estimate prior to a consultation with their physician would be expected to have an increased level of awareness that could lead to active enquiry and exploration about the risks of osteoporosis and the options for prevention, possibly due to an increased level of engagement. When patients have more active participation in their interactions with physicians (i.e., as "activated" patients) their clinical outcomes, as well as their satisfaction with the interaction with the medical system, are generally more positive (358). As discussed by Roter (359), increased patient participation in their health care serves to facilitate dialogue and ultimately the full involvement of patients in the negotiation and evaluation of health services that apply to them. A recent discussion of potential system changes to improve health related behaviour in patients accessing the health-care system has suggested that "pre-activation" would serve to assist patients to make more enquiries and to take more 57 action in their own care (360). Pre-activated patients have also been found to have better health outcomes, possibly because of closer adherence to clinical recommendations (361). Pre-activation can be accomplished by the provision of written materials, as well as face-to-face discussions or computer interactions, prior to a medical consultation: Patients who have received adequate information about their condition and have participated in decision making are more satisfied, have greater self-efficacy and healthier behaviour overall (360). Although the activated patient concept goes beyond simply providing patients with test results, informing people directly about their relative risk of fracture based on their B M D test results may facilitate patient empowerment and participation in decisions about their treatment or health promotion interventions. Receipt of test results directly may provide both the opportunity and the incentive for women and men to initiate a consultation with their physician and to seek further information about osteoporosis from health providers and from other sources. Ultimately, direct-to-participant feedback allows people to maintain control of their own test results as well as their bone health, which may in turn lead to increased motivation to make positive lifestyle changes. Direct-to-participant feedback of test results may also serve to prime participants by giving them a chance to formulate questions and to approach a physician consultation prepared to seek the information that they require to make decisions about lifestyle change or therapy. Physician-patient interviews that involve more information-giving compared with question-asking by the physician are associated with adherence to recommended therapies (362;363). On the other hand, direct-to-participant feedback may provide more immediate and more accessible information about test results to those who either would not choose to follow-up with their physicians to learn about their results, or whose physicians may not initiate a consultation; the latter situation may be more likely to arise in the case of "normal" test results. Wallace et al. (339) found that four months after testing, referred women with normal test results who were randomized to receive their results from a consultant directly after their scan reported lower susceptibility to osteoporosis than did women whose normal results were sent to their physician. There were no differences in perceived susceptibility by women with low test results, or in changes in dietary calcium or exercise behaviour between the two interventions. Although the authors concluded that direct feedback to the patient by the consultant did not add anything for those women with low test results, the enhanced perception of susceptibility in the women with a normal test result whose feedback went directly to their physician supports the value of ensuring that feedback reaches all participants or patients. Direct feedback presumably provided 58 reassurance and had no apparent negative impact on the health behaviour of the women with normal test results. Campbell et al. evaluated the impact of direct-to participant feedback on correct awareness of D X A test results in a randomly selected population-based sample of mid-aged women (302). The women who were randomized to receive their feedback directly (in addition to their physician) were significantly more likely to be correct about their results compared with those whose results were sent only to their physician (58% v. 36%). The authors did not find, however, that direct-to-participant feedback had a significant effect on exercise level, smoking status or uptake of OHT in these predominantly perimenopausal women. The results suggest that while direct-to-participant feedback may offer an improved means of communicating B M D test results as an alternative to the traditional method of sending D X A test results to the physician, it may not be sufficient to facilitate behavioural change. Interpretation of the study by Campbell et al. (302) is limited, however, because the follow-up questionnaires were mailed to the participants, which provided women with the opportunity to look up their reports at home: Women who had received their results directly therefore had a potential advantage because their results would be expected to be more easily accessible. In addition, the WHO diagnostic guidelines were not used to assess risk of fracture in this group; women in the lowest quartile were informed that their B M D was below average (in the lowest 25%) and that they may be at risk of osteoporosis in later life, which limits generalizability to the current standard practice that uses the WHO criteria to determine risk of fracture (50). Further clarification of the effectiveness of direct-to-participant feedback on correct awareness of B M D test results and behaviour change is required: The method for assessment of correct awareness should provide no advantage to the group that receives feedback directly and, as is now common in clinical practice, the standard WHO criteria for diagnosis (50) should be used to define risk. Furthermore, the potential impact of direct-to-participant feedback on awareness and behavioural change in men must be evaluated. 2.10 Summary This literature review began with an overview of the burden of osteoporosis and the role of D X A testing in identifying those at increased risk of fracture. A background was provided regarding the role of non-modifiable risk factors relevant to women and men in middle age, and the potential contributions of modifiable lifestyle factors and medical therapy to the risk of 59 osteoporosis and fracture were discussed in the context of the published literature. A selection of studies that have explored relationships between knowledge, awareness, information seeking and osteoporosis related health behaviour was briefly discussed and the literature specific to the impact of B M D test results on awareness of risk, information seeking and health behaviour change was reviewed in greater detail. The review included the only study that investigated the effects of direct-to-participant feedback on correct awareness of test results. This review has indicated that information regarding awareness of B M D test results and the impact of B M D test results on information seeking and health behaviour modification in mid-aged women (particularly women who are perimenopausal) is limited and there is a paucity of information regarding the impact of B M D testing in men. The potential impact of direct-to-participant feedback of D X A test results has not been sufficiently explored. If findings about these potential effects of D X A testing are to be generalized to the more broadly screened population, these questions must be investigated in a large randomly selected sample of both women and men, followed prospectively, for a long enough period to allow new behaviours and treatment interventions to become established. A relatively large study with longer follow-up, in a general population that includes men, has so far been lacking. 60 CHAPTER 3: Rationale and Objectives 3.1 Rationale The burden of morbidity and mortality arising from osteoporosis is substantial and is increasing with the advancing age of the population, and these increases are expected to become more evident in men relative to women. Because the treatment and rehabilitation of osteoporosis and its consequences, namely bone fractures, are costly and often not successful, preventive interventions offer the greatest potential in alleviating some of the burden. Although there is evidence of decreasing trends in age-adjusted fracture rates amongst women in Canada and other developed countries, the role that D X A testing may have played in this downturn is unknown. Several non-modifiable risk factors for osteoporosis have been recorded in the literature, and these can and have been used to identify mid-aged women and men who are at increased risk of fracture. Modifiable behavioural risk factors, such as low calcium intake, low vitamin D intake, low exercise participation, high alcohol use, smoking, and high caffeine intake can be targeted and potentially modified in people who are at relatively increased risk of fracture as a preventive intervention, or as an adjunct to medical therapy for the treatment of osteoporosis. Reliable medical therapies are available for the treatment of people who are identified as having a low B M D by D X A testing. Although D X A testing has a low positive predictive value for fracture, it still offers the most reliable means of predicting whether a woman or man of a specific age has a relative increased risk of osteoporotic fracture. D X A testing is only recommended for men and for women under 65 years of age i f other risk factors are present: There is evidence in the literature, however, that these guidelines are not always followed, or well understood, and that perimenopausal and early menopausal women receive a significant proportion, i f not the majority, of D X A tests. Furthermore, it has been suggested that a more generalized approach to screening may be warranted in the event that more reliable treatments and interventions become available or i f adherence rates are to be improved. Given that B M D testing is likely utilised for more than diagnostic purposes, and that a more general population-screening approach remains open to debate, a better understanding of the potential effectiveness of this intervention in terms of awareness of risk, health behaviour change and medication uptake or adherence in a randomly selected population is required. B M D testing can only be expected to have an impact on the morbidity and mortality associated with osteoporosis i f the provision of test results has an influence on the health 61 behaviours that affect osteoporosis risk and on effective medical interventions. Previous authors have cited a lack of information about such potential benefits of B M D testing on preventive interventions (4;56). As pointed out by one author in the field: "Bone densitometry, like any other diagnostic test, has no role whatever i f it is not to influence either physician or patient in their management or behaviour" (364). In order for B M D testing to have an impact, risk information and recommendations for interventions must effectively reach individuals who have been identified as being at greater risk for fracture by D X A , and who will implement the recommended or warranted changes to their lifestyles and medical therapy. An understanding of the test results, and an increased risk status, is likely to be a necessary step in bringing about change in behaviour or uptake of therapy. Not surprisingly, previous studies of the effects of D X A testing have demonstrated a link between correct awareness of low test results and uptake of therapy. The majority of the literature reporting on awareness and understanding of B M D test results, however, has involved volunteer or convenience samples of women who have been referred for D X A testing by their physician. In addition, other potential predictors or confounders of correct awareness of B M D results, such as risk factors other than B M D , have not been adequately addressed. Correct awareness of test results in men has not been described at all. B M D testing may be expected to prompt information seeking about osteoporosis, and those with low test results may be the most likely to seek further information. Information seeking may be a process that is indicative of a readiness to make a behavioural change, even i f a change is not evident or measurable. An association between information seeking about osteoporosis and health behaviour change has been reported by some authors, but not others. The impact of D X A testing on information seeking about osteoporosis and the sources of information that are used after D X A testing have not been examined in a population-based sample of women, and have not been addressed at all in men. As with the literature about correct awareness of test results, most previous studies of lifestyle modification following B M D testing have utilised volunteer or convenience samples of women who have been referred for testing, and the studies have tended to have short follow-up periods. Very few previous studies have taken into account baseline levels of the behaviours of interest in their analyses of behavioural change and in the interpretation of their results. Furthermore, the investigation of the potential role of other subject-related factors, such as age dr other risk factors for osteoporosis, has been limited to the assessment of a select few, i f any, of these factors. There is, therefore, virtually no information about whether lifestyle changes are 62 more or less likely after D X A testing in high-risk subgroups, such as those with a family history of osteoporosis, low body mass index or a history of corticosteroid or anticonvulsant use. It is assumed that in order for D X A test results to motivate lifestyle change or uptake of medication in those at higher risk for fracture, the increased risk must first be conveyed to the individual, whether directly or through a physician, and the individual must take notice of their "diagnosis" or risk status. An awareness of increased fracture risk is likely to be a prerequisite for lifestyle modification; an efficacious method of communicating B M D results and risk information would therefore be expected to be critical to the effective management of osteoporosis and fracture risk. Direct-to-participant feedback of test results may provide a relatively effective means of conveying risk about osteoporosis. The literature suggests that patients who receive adequate information and have had more active participation in their interactions with physicians (activated patients) demonstrate greater self-efficacy and healthier behaviour. Direct feedback to participants may also serve to relay test results in the most effective way, particularly for people whose results do not indicate an increased risk of fracture. One previous study investigated the effects of direct-to-participant feedback on test result awareness and health behaviour in mid-aged women (302). This research was conducted in Scotland, before the availability of the WHO diagnostic criteria; the impact of direct-to-participant feedback of D X A test results has not been studied with women in Canada, using the WHO diagnostic criteria for feedback, or not in men at all. 3.2 Objectives of the Study The overall aim of this study was to examine the potential effects of bone mineral density testing in a population-based sample of mid-aged Canadian women and men by describing its association with correct awareness of bone density test results, information seeking about osteoporosis, and osteoporosis related health behaviour change. The specific objectives were: (1) To describe the relationships between correct knowledge of test results three years following testing and both a diagnosis of osteoporosis or osteopenia by bone density measurement and the method of test result feedback. (2) To describe the information sources accessed by women and by men following D X A testing. 63 (3) To examine the associations between self reported information seeking about osteoporosis during the three years following testing and both bone density test results and the method of test result feedback. (4) To explore the associations between changes in osteoporosis related health behaviour three years after D X A testing and both bone density test results and the method of test result feedback. (5) To assess the influence of other known risk factors on test result awareness, information seeking and health behaviour change while taking diagnosis by bone density measurement into account. 3.3 Overview of the Study Design This research was designed to assess the impact of an "at risk" diagnosis of osteopenia or osteoporosis, compared with a normal test result or "no increased risk" as a result of B M D testing on awareness of test results, information seeking about osteoporosis, osteoporosis related health behaviour change and the use of medical therapy three years after D X A . In addition, the effectiveness of a simple intervention in the form of direct-to-participant feedback of test results was explored. These objectives were addressed using prospectively collected baseline (1996-1997) and three-year follow-up (1999-2000) data and retrospective data collected from participants' files, and by means of a natural experiment (or quasi-experiment) that occurred within the ongoing Canadian Multicentre Osteoporosis Study (CaMOS). A proportion of the data utilised for this study was collected as part of CaMOS; a prospective cohort study designed to estimate the incidence and prevalence of osteoporosis and fractures, as well as the effect of potential risk factors, in a random sample of women and men from nine centres across Canada (5;365). Further data were collected from the CaMOS participant files, with the specific purpose of addressing the unique objectives of this dissertation. It would be neither ethical nor practical to compare the effects of feedback of low B M D results with feedback of normal B M D results to study participants by randomised intervention. For this study, the potential effects of a low B M D test result compared with a normal test result were examined by the observational method, with consideration for the potential effects and influences of a broad range of uncontrolled factors. The selection of these latter variables was informed by the review of the published literature, but also determined by the availability of the 64 relevant information from data collected during the CaMOS baseline interview or retrospectively from the participant files at the CaMOS centres. A l l participants were invited to attend a D X A test at baseline as part of the CaMOS study protocol. Because the method for interpretation and feedback of these test results to the study participants was not part of the CaMOS protocol, each participating centre adopted its own method of relaying test results. The results of the D X A test were relayed to the study participant, to the physician, or to both by the CaMOS centres using standard reports that resembled those that were typically generated following B M D tests of clinically referred patients by the testing centres involved. The method of feedback of test results varied between centres, and also within some centres. There were no further study interventions; all health behaviour interventions and actions were left to the physicians, the existing structure and the community programs that were available. The study is presumed to provide information about the effectiveness of a report of low B M D test results in the context of the programs, facilities and resources that were available at the time of testing. Hence the current study examined outcomes following a "real world" scenario of feedback of normal or "at risk" test results, by a quasi-experimental study of direct-to-participant feedback of simple test results in a standard format. The study population was composed of a randomly selected sample of Canadian women and men aged between 40 and 60 years. Study participants were not selected for their risk status, although they naturally included a proportion of women and men who would meet the criteria for selection for D X A testing (85;87;228). In this respect, the study population is considered to model a community screening program for mid-aged women and men across Canada. The results of this dissertation research provide insight into test result awareness, information seeking, behavioural change and use of bone-specific medications in women and men who are selectively screened for osteoporosis, as well as the impact that B M D testing may have i f general D X A screening were provided to the mid-aged population. The findings also provide information about those subgroups of the population that are relatively resistant to health behaviour change specific to osteoporosis so that these groups may be targeted by osteoporosis prevention and treatment programs. Finally, this study provides an indication of whether direct-to-participant feedback of test results leads to improved correct awareness of test results, or increased information seeking about osteoporosis, lifestyle change or bone-specific medication use three years after D X A testing compared with feedback only to the physician. 65 3.4 Conceptual Framework The conceptual framework for this thesis research is outlined in Figure 3.1. An organizational structure for the characterization of the factors that may be expected to influence the effects of D X A test results on awareness of results, information seeking and health behaviour change has been borrowed from parts of the PRECEDE model of health promotion (366). The variables that have been measured and taken into account are emphasised in the outlined model in bold, and are described in greater detail in the following chapter. The PRECEDE model, which was originally designed to guide the systematic development and evaluation of health education programs, groups the various influences of health behaviour into three categories: predisposing, enabling and reinforcing factors (366). Predisposing factors provide the rationale and motivation for change and are typically, although not always, psychological in nature. Specific psychological predisposing factors, such as beliefs or attitudes have not been measured as part of this study. Factors that have been measured, however, include those that may be expected to influence such psychological predisposing factors (or "motivation for change"). These include the presence of risk factors for osteoporosis that study participants and their physicians may have been aware of, such as a family history of osteoporosis or a history of corticosteroid use, as well as demographic characteristics (many of which are also related to an increased or decreased risk of osteoporosis) such as age, sex and level of education. Enabling factors are antecedents that enable motivation for behaviour change to be realized. Typically these factors are related to systemic conditions such as programs and resources that are available in the community. Within the context of this study, enabling factors may have been influenced by, or have been associated with, the nine CaMOS study centres; services and resources such as osteoporosis and health promotion programs may have been available to a varying extent in the different communities.16 In addition, inclusion in the participants' or physicians' feedback of a recommendation for further action (such as the 1 6 Environmental, system and cultural variables that may be associated with the location of the testing centres have not been measured for this research. However, the potential influence of the "Centre" level variable on the outcomes is considered in the analyses and interpretation. The CaMOS centres are highly correlated to the method of feedback of the test results, which presents limitations to the interpretation of the influence of these two enabling factors. This is discussed in more detail in Chapter 4. 66 suggestion that the participant be referred to a specialty clinic) may also be expected to influence enabling factors. Reinforcing factors provide reinforcement, or incentives, for the health behaviour to persist once it has been initiated and include the effects of support from significant others, health-care providers and peers as well as vicarious reinforcement. Reinforcing factors, although important, have not been addressed in this study. The framework acknowledges that both correct awareness of test results and information seeking about osteoporosis are likely to be important process variables that precede a positive change in health behaviour. These outcomes, however, were all measured at the same time (at the Year 3 follow-up); the assessment of potential causal relationships between these outcomes is not addressed. Instead, each of these outcomes is considered as a desirable endpoint of an effective screening procedure. Osteoporosis specific health behaviour change is the most desirable of these outcomes in people who are identified at higher risk for fracture by D X A testing, and where a health behaviour change is warranted, because correct awareness of test results and information seeking do not have direct effects on B M D or fracture risk. Correct awareness of risk status (or test results) and information seeking about osteoporosis do indicate, however, that people have received and understood their test results, and that they are in a position to make an informed decision about their own health behaviour. The conceptual framework presented in Figure 3.1, was derived with the aid of borrowed concepts from the PRECEDE model in order to guide the selection of important variables, as well as to inform the design of the study. No behavioural model was included in the conceptual design of the original CaMOS study; the CaMOS study was not designed to assess behavioural outcomes. The research for this dissertation was designed as an observational study to determine the effectiveness of D X A test results and direct-to-participant feedback. While the important role of intervening psychological and cognitive processes as well as environmental factors between these interventions and the behavioural outcomes is acknowledged, these intervening variables were not measured or addressed by this research. Thus, this study was not designed to test any particular model of health behaviour, health promotion or health-care utilisation; the conceptual model outlined in Figure 3.1 has been generated only as a guide for study design. 67 Figure 3.1: A Conceptual Framework of Factors Contributing to the Impact of DXA Test Results on Information Seeking and Osteoporosis Specific Health Behaviour Demographics and risk ^ factors for osteoporosis - Age - Sex - Race - Education - Menopausal status - Family history - Income quintile - Previous diagnosis - B M I - Prevalent fracture - Comorbidity - Medication history - Self-rated health - Smoking status - D X A T e s t R e s u l t s - D i r e c t - t o - P a r t i c i p a n t F e e d b a c k Other Influences -Centre - Recommended Action Predisposing Factors "Motivation " - Attitudes -Beliefs - Values -Needs Reinforcing. Factors - Social support - Peer influences - Feedback and advice by health care providers Enabling Factors - Accessibility of programs and resources - Referrals A w a r e n e s s o f R i s k - Correct knowledge of test results I n f o r m a t i o n - S e e k i n g H e a l t h B e h a v i o u r - Ca+ in diet - Ca+ supplements - Vit. D supplements - Exercise - Smoking status - Alcohol intake - Caffeine intake - Medical Therapy Bone Mineral Density Fracture Risk CHAPTER 4: Methods This was a retrospective analysis of a sub-sample of the prospective Canadian Multicentre Osteoporosis Study (CaMOS) (365). Women and men aged 40 to 60 years at baseline who completed the full CaMOS interview, had participated in bone density testing at baseline, and then took part in a CaMOS follow-up interview three years later were eligible for this study. A full description of the participants is included in Chapter 5. 4.1 The C a M O S Study The larger CaMOS cohort is an age-, sex-, and region-stratified random sample of community-dwelling women and men aged 25 years and over from nine study centres across Canada (Vancouver, Calgary, Saskatoon, Hamilton, Toronto, Kingston, Quebec City, Halifax, and St. John's). At baseline, all of the participants resided within 50 kilometres of one of the study centres. Individuals were eligible for participation i f they matched the age- and sex-stratification criteria and if they spoke and understood English (all centres), French (Quebec City) or Cantonese or Mandarin (Vancouver and Toronto). Households were originally selected via a random sample of listed residential telephone numbers. These randomly selected households were first contacted by mail and the eligible residents were invited to participate. The households were then contacted by telephone and one individual was randomly selected from those household members that satisfied the stratification and selection criteria. Complete details of the recruitment methods and criteria are available (5;365). To account for the greater risk of fracture with age and in women, the CaMOS strata increased in size with age and twice as many women as men were selected within each age group. The population-based random sample is believed to represent approximately 40% of the population of Canada, but it does not include nor represent either institutionalized Canadians or the Aboriginal populations of northern Canada (5;365). In addition, the sample represents English-speaking Canadians, as well as a proportion of the population who speak French, Cantonese or Mandarin. CaMOS, which is now in its 12 th year of follow-up, involves researchers affiliated with universities across Canada.17 The main objectives of the CaMOS study are to 1 7 Dr Nancy Krieger and Dr David Goltzman are the national principal investigators for CaMOS; the founding national principal investigator was Dr Alan Tenenhouse. Centre directors for CaMOS are Dr. 69 determine the incidence and prevalence of osteoporosis in the Canadian population by region, to describe and determine the influence of certain risk factors on bone density and fracture risk, to assess the impact of osteoporosis on health status and to estimate the cost of osteoporosis and associated fractures in Canada (5;365). Baseline data collection for CaMOS involved bone mineral density (BMD) measurements by dual energy X-ray absorptiometry (DXA) of the lumbar vertebrae (LI to L4) and three sites at the hip (neck, trochanter and Ward's area), and height and weight measurements. An extensive questionnaire regarding demographics, illnesses and medications, family history, diet, exercise, reproductive history, smoking, alcohol consumption and health status was administered, one on one, by trained interviewers. In addition, all participants aged 50 years and older were asked to undergo lateral X-rays of the lumbar and thoracic spine for the diagnosis of prevalent vertebral fractures. The recruitment and baseline data collection phases of the CaMOS study took place during 1996 and 1997 (5;365). Following baseline data collection, an interpretation of the results of the D X A tests and spinal X-rays was sent to every participant and/or his or her family physician (FP). The procedure for interpretation and feedback of these test results was left to the discretion of the CaMOS centre directors; each CaMOS centre therefore adopted its own method for interpretation and feedback of results. A l l CaMOS participants aged 40 to 60 years at baseline were eligible for follow up at Year 3, at which time the B M D , height and weight measurements were repeated and information was collected by individual interview regarding current diet and other lifestyle variables as well as changes in reproductive status, family history, medications, illnesses and health status. The participants also were queried about their knowledge of their baseline B M D and whether they had sought information about osteoporosis. The CaMOS mid-aged sub-sample (i.e., those aged 40 to 60 years at baseline) of "full" participants included 3,139 men and women with a mean age at baseline of 52 years, 89% of whom were followed-up at Year 3 (889 men and 1,904 women). Although they were considered "full" participants by the CaMOS study (they completed the full Jerilynn Prior (Vancouver), Dr. David Hanley (Calgary), Dr. Wojciech Olszynski and Dr. K . Shawn Davison (Saskatoon), Dr. Rick Adachi and Dr. Alexandra Papaioannou (Hamilton), Dr. Robert Josse and Dr. Sophie Jamal (Toronto), Dr. Tassos Anastassiades and Dr. Tanveer Towhead (Kingston), Dr. Jacques Brown and Dr. Louis Bessette (Quebec City), Dr. Susan Kirkland and Dr. Stephanie Kaiser (Halifax) and Dr. Carol Joyce and Dr Christopher Kovacs (St John's). National co-ordination of the study, since its inception, is provided by Ms. Suzette Poliquin. 70 questionnaire), a small number of these participants did not take part in the baseline D X A test (see below). As part of the CaMOS protocol, participants were invited to complete a short "refusal" questionnaire at baseline i f they declined to participate in the full study. Thirty percent of the women and 33% of the men aged 25 to 80+ years who were contacted completed this short questionnaire, which included questions related to major risk factors for osteoporosis including, sex, race, family history of osteoporosis and current cigarette smoking status. A comparison of the CaMOS participants who completed the full questionnaire and those who completed the refusal questionnaire indicated that full responders were more likely to have risk factors for osteoporosis (a family history of osteoporosis or a history of a previous fracture) and were less likely to be current smokers (367). The CaMOS sample, in general, may be over representative of those at higher risk of developing osteoporosis. No information was available, however, on risk factors for 25%) of the women and 30% of the men who were contacted for the study; these participants did not agree to complete the full questionnaire or the refusal questionnaire. Participation rates within subgroups of the men and women who were invited to take part in CaMOS have shown that, although the full participation rate for the CaMOS sample (all ages) was only 43%, the participation rate was higher in the sample aged 40 to 59 years (this age range includes most of the participants who were eligible for the present study). The full participation rate for this mid-aged group was 67% for women and 50% for men. The highest baseline response rate by gender, and of any age group in CaMOS, was that of the 40- to 49-year-old women (68% of women in this age group who were invited to participate agreed to full participation) (367). Further generalizability of the sample selected for this study to the Canadian population was assessed and is presented in the following chapter (Chapter 5) by comparison of the study sample with cross sectional data from the 1996 Canadian National Population Health Survey (described below). 4.2 Sample Selection for this Study A l l 3,139 mid-aged (40 to 60 years old at baseline) CaMOS participants were initially eligible for the study. Because the study objectives were focused on the effects of reported baseline B M D test results, those subjects who did not complete a baseline B M D test were excluded (n =117). The outcome variables for this study were derived from 71 responses to questions on the Year 3 questionnaire; subjects who were lost to follow up at Year 3 were therefore excluded from the study sample (n = 316 of those who had completed B M D testing at baseline). Information regarding baseline diagnosis and feedback destination (for those who completed a B M D test), as well as demographic, medical and health behaviour information from the baseline questionnaire, was collected for all 3,139 CaMOS mid-aged participants. The information collected from those who did not undergo a B M D test or were not followed up at Year 3 was used for comparison with the study sample (i.e., the included participants). 4.3 The NPHS Sample Responses to selected demographic and health measurements that were equivalent and available on both the CaMOS questionnaire and the household component of the 1996-97 National Population Health Survey (NPHS) questionnaire (368) were compared in an attempt to evaluate how the participants from the current CaMOS sub-sample compared with a much larger randomly selected survey population that is considered to be representative of a significant proportion of the Canadian population. The NPHS (cycle 2) was conducted by Statistics Canada during the same time period that the CaMOS baseline sample was collected, between 1996 and 1997. The purpose of the NPHS was to collect and measure information related to health and the health status of Canadians. Although a substantial proportion of the respondents to the NPHS were part of a longitudinal component that has been surveyed every two years, the data used for the current comparison are cross-sectional. The household component population of the NPHS was made up of household residents of both rural and urban areas in all Canadian provinces (the Yukon, Northwest Territories and Nunavut were excluded). Certain populations were not part of the sample: people living on native reserves, on Canadian Forces bases, and in institutions such as hospitals, residential care facilities and correctional centres. In addition, some remote areas of Quebec and Ontario were not covered by the survey. The NPHS sample was created using a two-stage stratified sampling design (dwellings were selected from within clusters). In all provinces except Quebec, the Labour Force Survey design was used to select the sample. In Quebec, the Enquete sociale et de sante conducted by Sante Quebec in 1992-1993, with a two-stage design similar to the 72 Labour Force Survey design, was used. After an initial random selection of households, one member of the household was chosen to be the respondent. Methods were incorporated to eliminate potential bias of over-sampling occupants of smaller households. The complex sampling scheme was a carefully designed process aimed at minimizing sampling bias. Interviewers were well trained and supervised. Quality assurance measures were implemented at all steps of data collection and processing in an attempt to minimize non-sampling error. The overall response rate for the NPHS survey was approximately 18 79% (368). Sampling weights to adjust for differing opportunity to respond and for non-response were included in the dataset provided by Statistics Canada and were employed in the current generation of frequencies from the NPHS sample. Comparisons were made between the study sample from CaMOS and the equivalent age-19 and gender-matched sub-sample of the NPHS. 4.4 Collection and Derivation of Variables 4.4.1 Collection of the Feedback Data Two of the main exposures of interest for this study were the diagnosis that the participant received (based on the lower of the spine or hip D X A measurements and labelled here as "Normal", "Osteopenia" or "Osteoporosis") and the destination of the feedback ("to the FP", "to the participant" or "to both the FP and the participant"). Early investigation of selected feedback letters from the nine study centres revealed that the diagnostic criteria for osteoporosis or osteopenia, as well as the recommendations accompanying the diagnostic information (such as whether treatment or referral to an osteoporosis clinic should be initiated), were not consistent between centres and within some centres. The form and destination of the feedback were not part of the original CaMOS study protocol and each of the nine study centres developed its own feedback content and method of delivery. The recipients of the feedback (the participant and/or the FP) were therefore determined within each centre. The NPHS household response rate was 82.6% and the response rate of selected persons within the households was 95.6%. 1 9 The NPHS public use data files provide "age" only as a grouped variable, in 5-year intervals (i.e., ages 40-44, 45-49, 50-54, 55-59 and 60-64 years). The NPHS sample aged 40 to 64 years was selected for comparison with the CaMOS sample aged 40 to 60 years. The NPHS sample included approximately equal distributions of participants in each age range, whereas the CaMOS sample included a greater number of participants in each increasing age range because of sampling design. Because the age distributions within each of the 2 populations were not directly comparable, any differences that were found between the 2 populations were investigated further by analyses stratified by 5-year age group. 73 To ensure that the diagnosis received by each participant was accurately coded, copies of standard "form" letters as well as random samples of feedback letters pertaining to B M D results and spinal X-ray results (stratified by date, B M D measurements, and gender) were requested from each centre. These letters were carefully reviewed for consistency. It was determined that to accurately code the information that was sent to the participant and/or his or her physician, and to determine the destination of the feedback in each case, all available individual feedback letters should be reviewed. It was also determined that the consistency of the feedback protocol and diagnostic criteria at each centre would best be determined by a review of the files and other information on site at each centre. Consequently, the data related to feedback diagnosis (and criteria for the diagnosis), any recommendation included in the feedback to the participant or FP, the destination of the feedback and reasons for declined B M D tests were collected from the original participant files on site at each of the nine study centres (Vancouver, Calgary, Saskatoon, Hamilton, Toronto, Kingston, Quebec City, Halifax and St. John's) by the author. The data were collected retrospectively after the Year 3 follow up of the CaMOS participants had been completed. For each of the 3,022 mid-aged baseline CaMOS participants who completed the B M D test, copies of the baseline D X A reports, all available copies of feedback letters addressed to the participants and/or their family physicians, radiology reports and letters regarding the X-ray and all notes, log books and other documentation dating to the time of the original feedback were reviewed, in every case, to determine where the results were sent and the exact content of the feedback. The files of the 117 CaMOS mid-aged participants who did not have a baseline B M D test were reviewed for information regarding reasons for declining the B M D test. In addition to an examination of all available information in the individual files, the study coordinator was interviewed at every centre to determine the general protocol followed at baseline and whether there were any deviations from or flexibility in the planned feedback protocol. In centres where other individuals played a significant role in the feedback process, those individuals were interviewed: a technician (Saskatoon), interviewers (Saint John's) and a secretary (Kingston). In three centres (Saskatoon, Quebec City and Halifax), there was a change in study coordinator between the beginning of the CaMOS study and baseline feedback, and the time of this data collection. The original study coordinator was contacted by telephone and interviewed in each of these 74 three centres in addition to the current coordinator. The directors of the centres were also consulted directly regarding the feedback rationale and protocol in seven of the nine centres (the director was unavailable at two centres). The data collected at the centres were used to generate the "Feedback Destination", "Diagnosis by Feedback", "Report of a Significant Vertebral Fracture" (from X-ray) and "Recommendation for Further Action" explanatory variables. Information regarding reasons for no B M D test, together with information regarding the adopted criteria for diagnosis of osteopenia or osteoporosis was summarized descriptively. The Vancouver data were collected from the participant files twice; once at the beginning of the data collection period and again at the end, after all other centres had been visited. This step was performed as a check to ensure that the coding criteria had remained unchanged over the course of data collection. It was not feasible to re-visit the other eight centres. It was found, however, that the criteria for the feedback variables were reliable; when the Vancouver data were entered a second time the coding had not changed. A l l information regarding feedback was entered and coded directly into an Excel database by the researcher. 4.4.2 Derivation of Variables from the CaMOS Questionnaire The remaining variables, including all outcome variables, were derived directly from responses to one or more questions on the CaMOS questionnaires at baseline and Year 3. Table 4.1 summarizes the sources of each of the explanatory and response variables and summarizes how each variable was derived and classified. Further description of the derivation of some of the more complicated variables follows in the text below. Copies of the relevant questions from the CaMOS questionnaires at baseline and Year 3 are listed in Appendices A and B, respectively. A l l questions included in the CaMOS questionnaire that were used for this study were developed or selected by individuals in the original CaMOS research group (5;365). In some cases, the questions were modified from existing questionnaires, but the modified versions did not undergo reliability or validity testing. In other cases, the questions were written by members of the CaMOS research group without reference to existing questionnaires. The exceptions amongst the questions used for this study were the question regarding race or ethnicity and the general health question from the SF-36 (see Appendix A). The former was taken directly from the questionnaire used by Statistics Canada for the National Population 75 Health Survey (368). The reliability and validity of the SF-36 (369) has been tested extensively on diverse populations. Table 4.1: Source and Derivation of all Explanatory and Outcome Variables Variable Source Notes/Comments Main Explanatory Variables Feedback Diagnosis Copies of feedback letters, logbooks and interviews with CaMOS personnel Potential Explanatory Variable and used to derive "Correct" outcome (see below). 3 Categories based on lowest of hip and spine values: - Normal, Osteopenia or Osteoporosis Recalled Diagnosis Year 3 questionnaire: Question 1.5 Used to derive "Correct" outcome variables (see below) and as an alternative potential explanatory variable for the health behaviour outcomes. - "Normal / High", "Osteopenia / Borderline / Low without Osteoporosis", "Osteoporosis / Low" and "Don't Know" WHO Diagnosis Raw B M D test data from feedback reports and CaMOS database Alternative explanatory variable to feedback diagnosis, used to derive "Correct by W H O " alternative outcome variable. Categories based on raw (unstandardised) B M D results and WHO guidelines: ' - Normal (T-scores > -1.0) - Osteopenia (T-scores < -1.0 and > -2.5) - Osteoporosis (T-scores < -2.5) Destination of Feedback Copies of feedback letters, logbooks and interviews with CaMOS study personnel Potential Explanatory Variable in 3 Categories: - To family physician only - To participant only - To family physician and participant Potential Effect Modifiers or Confounders Recommendation for Action Copies of feedback letters Dichotomous Variable (Yes/No): - Recommendation for treatment or for referral to an osteoporosis clinic in the feedback 76 Variable Source Notes/Comments Report of a Copies of Radiology reports Only applicable to those aged 50 to 60 years who Significant Fracture0 and feedback letters attended the X-ray examination. Dichotomous Variable: - A noted significant prevalent fracture in the spinal X-ray report to the FP and/or participant - No significant prevalent fracture reported to the FP and/or participant Age Group Baseline questionnaire: 2 categories derived from continuous data: Age in years at baseline Aged 40-49 or 50-60 years. Body Mass Index (BMI) Baseline measurements of height and weight taken at D X A appointment or during interview Divided into quartiles of B M I within each gender - Lowest quartile - Second lowest quartile - Second highest quartile B M I = Weieht (kel - Highest quartile Height (metres)2 Continuous variable for comparisons among excluded/included/NPHS participants Race/Ethnicity" Baseline questionnaire: Dichotomous: Question 1.7. "White" or "Non-white" e Reproductive Status (Women only) Baseline questionnaire: Questions 5.2, 5.3, 5.4. 4 categories: - Premenopausal - Naturally Menopausal - Surgically Menopausal - Premenopausal Hysterectomy Education Baseline questionnaire: 4 categories Question 1.8 - Incomplete High School (HS): < Grade 12 (< Grade 13 for those who attended HS in Ontario) with no high school certificate or diploma. - Complete HS: Graduated HS with certificate or diploma. - Postsecondary Ed.: Trade/professional certificate or some university without certificate or diploma - University degree. Estimated Neighbourhood Income Quintile Baseline questionnaire: First 3 digits of the postal code. Quintiles of estimated income level: - Lowest quintile - Second lowest Income levels were - Middle generated using the PCCF+ - Second highest Version 3G program from - Highest quintile Statistics Canada. 77 Variable Source Notes/Comments Centre Baseline questionnaire: 9 Centres: Vancouver, Calgary, Saskatoon, Hamilton, Centre (location) at baseline Toronto, Quebec City, Kingston, Halifax or Saint John's; all participants lived within 50 km of the centre at baseline. Awareness of a Family History of Osteoporosis Baseline questionnaire: Question 7.2 Dichotomous: - Awareness of at least one family member with a diagnosis of osteoporosis ("yes") - No awareness of a family member with a diagnosis of osteoporosis ("no", "don't know" or "not applicable" to each type of family member) Awareness of a Previous Diagnosis of Osteoporosis Baseline questionnaire: Question 2.1 (Osteoporosis) Dichotomous: - Awareness of a previous diagnosis ("yes") - No awareness of a previous diagnosis ("no" or "don't know") History of Comorbidity Baseline questionnaire: Question 2.1 Dichotomous: - Awareness of a previous diagnosis of Rheumatoid (Rheumatoid Arthritis, Arthritis and/or an Eating Disorder ("yes") Eating Disorder) - No awareness of a previous diagnosis of either disorder ("no" or "don't know") History of "High Risk" Medication Exposure Baseline questionnaire: Question 3.1 Dichotomous: - Awareness of ever having taken cortisone (oral, (anticonvulsants and/or inhaled or injection) and/or anticonvulsants corticosteroids)) (phenytoin, phenobarbital, seizure pills); "yes" - No awareness of ever having taken these medication; "no" or "don't know" Self-Reported Health Statusf Baseline questionnaire: Question 14.1 Collapsed to Dichotomous Variable: - "Excellent" or "Very Good" - "Good", "Fair" or "Poor" "Correct" Awareness: Outcome Variables llilSllli^tSc •{• 'TtSM^^^^^^^^^^^ Correct Year 3 questionnaire: Dichotomous. Question 1.5 and Feedback - "Correct"; exact match between recalled diagnosis Diagnosis and feedback diagnosis - "Not Correct" (includes incorrect responses and "don't know") 78 Variable Source Notes/Comments Closely Correct Year 3 questionnaire: Question 1.5 and Feedback Diagnosis Alternative Outcome. Dichotomous. - "Closely Correct": Correct recall of a normal diagnosis or recall of "osteopenia" or "osteoporosis" if the feedback diagnosis was either osteoporosis or osteopenia - "Not Closely Correct": Incorrect by above definition or "don't know" Correct by WHO Year 3 questionnaire: Question 1.5 and WHO Diagnosis (see above) Alternative Outcome Variable. Dichotomous: - Exact match between recalled diagnosis and WHO diagnosis. - Incorrect report of WHO diagnosis, or "don't know" Health Behaviour: Potential Effect Modifiers (baseline) and Outcome Variables (Year 3). Information seeking Year 3 questionnaire: Question 1.6 Dichulomous (Yes. No;: - Information sought about osteoporosis from at least one source. Current Cigarette Smoker Baseline questionnaire: Question 9.1 and 9.2 Year 3 questionnaire: Question 9.1 and 9.3 Dichotomous (yes/no) - Has ever smoked cigarettes daily for at least 6 months and is currently smoking cigarettes. Calcium (diet alone); mg/day Calcium (supplements and diet); mg/day Baseline and Year 3 questionnaires: Q. 3.2 and 10.1. Continuous Variables: - Estimated elemental calcium from dietary sources derived by CaMOS from report of average portion size and frequency of eating calcium-rich foods over previous 12 months. - Estimated total elemental calcium derived by combining dietary calcium and amount of calcium intake from supplements (derived by CaMOS, from dose and frequency of use of any reported current supplements taken on a regular basis). Current Calcium Supplements Baseline and Year 3 questionnaires: Q. 3.2 Dichotomous (yes/no): - Calcium supplements currently taken regularly Current Vitamin D Supplements Baseline and Year 3 questionnaires : Q. 3.2 Dichotomous (yes/no): - Vitamin D supplements currently taken regularly Current Regular Exercise Baseline and Year 3 questionnaires: Q. 11.3 Dichotomous (yes/no): - Currently participates in any regular activity or program for exercise 79 Variable Source Notes/Comments Current Specific Osteoporosis Medication Use Baseline and Year 3 questionnaires: Q. 3.2 Dichotomous (yes/no) - Currently taking any medication commonly or occasionally used to treat or prevent osteoporosis (other than hormones): bisphosphonates, raloxifene8, calcitonin or fluoride Current Ovarian Hormone Therapy (OHT) Use (Women only) Baseline Questionnaire: Q. 5.5,5.6 Year 3 questionnaire: Q. 5.10, 5.11 and 5.12 Dichotomous (yes/no): - Currently taking estrogen and/or progestin/progesterone. Current Osteoporosis Related Medication Use Baseline Questionnaire: Q. 3.2,5.5,5.6 Year 3 questionnaire: Q. 3.2, 5.10., 5.11,5.12. Dichotomous (yes/no): -Current use of OHT and/or any specific osteoporosis medication Derived from a combination of above 2 variables High Alcohol Intake Baseline and Year 3 questionnaires: Q. 10.2 Dichotomous: - Less than 2 alcoholic drinks per day - 2 or more alcoholic drinks per day. Derived from reported frequency of servings of alcoholic drinks on average over past 12 months High Coffee Intake Baseline and Year 3 questionnaires: Q. 10.2 Dichotomous: - Less than 4 cups of coffee per day - 4 or more cups of coffee per day. Derived from reported frequency of servings of caffeinated coffee on average over past 12 months Total Caffeine Baseline and Year 3 questionnaires: Q. 10.2 Continuous variables: Total caffeine derived by CaMOS from report of the average frequency of servings of caffeinated and decaffeinated coffee, tea and colas over the past 12 months. c Prevalent fractures were diagnosed by radiologists at each of the 9 CaMOS centres. Reports of fractures were considered to be "significant" i f there was a report of a fracture in the feedback and no indication that it was "minor" or "insignificant". Compression fractures that specifically stated that there was no more than 20% compression on any vertebra were considered to be not significant. d The term "race" is used here as a descriptor of participants' self identified race or ethnicity when asked to describe their race or colour and is assumed to reflect participants' perception of their personal ethnic or cultural background. e The "Non-white" group includes all participants who indicated a race other than "White", even i f they also described themselves as "White"; participants were not limited to one response to this question. f From the SF-36 (369), a widely used generic measure of health status. The complete SF-36 was administered as part of the CaMOS protocol; only the response to the first question (the "in general..." health question) was used for this study. 8 Raloxifene was not approved by Health Canada for the prevention of osteoporosis in women until after completion of the baseline data collections, but was available at Year 3. 80 4.4.3 Potential Effect Modifiers Factors that may be associated with correct awareness of test results or osteoporosis-related behavioural change, and that may also be associated with the potential explanatory variables were considered as potential confounders and as potential effect modifiers, where appropriate. Categories of certain variables (race, education and self-rated health status) that were derived directly from the CaMOS questionnaire(s) were combined (see Table 4.1) because of sparse numbers in some categories, to distribute the sample more evenly across subgroups and to facilitate the interpretation of results. Two continuous variables, age and body mass index, were converted into categorical variables. Age was dichotomized into a 40 to 49 year group and a 50 to 60 year group. Both women and men over the age of 50 may be considered at more serious risk for osteoporosis and may be more likely to be treated, or advised differently than are younger women and men: Previous guidelines concerned with the screening and treatment of osteoporosis have been restricted to those aged 50 years and over (72;85), in whom the potential benefits of assessment and intervention are considered to be greater. There is relatively little known about the potential benefits of any form of intervention (lifestyle or medication) on the long term bone density or fracture risk of women or men under the age of 50 years. The age of 50 years, which is close to the average age of menopause in women, is the cut-off age for increases in the recommended daily intakes of calcium and Vitamin D. Intakes recommended by Osteoporosis Canada increase from 1,000 mg/day of calcium and 400 IU/day of Vitamin D for women and men aged 31 to 50 years, to 1,500 mg/day of calcium and 800 IU/day of vitamin D for those aged 51 to 70 years (85). Furthermore, personal communications with physicians who managed patients with osteoporosis led to the conclusion that there is a "cut-off at age 50 years whereby patients are treated differently and test results may be taken more seriously. Finally, participants in the CaMOS study were only eligible for an X-ray test i f they were 50 years of age or over. Subgroup analysis within this age group allowed for the inclusion of a report of a significant prevalent fracture as a potential explanatory variable. The participants' body mass index (BMI) distribution was divided into quartiles within each gender to facilitate interpretation. A difference of one "unit" of B M I was not considered to be a meaningful measure for comparing behavioural outcomes; it is, however, common to describe individuals as "larger" or "smaller". The division of the sample into four equal groups created a "large" and a "small" group, with the two middle groups representing an average size. Significant associations could then be interpreted more meaningfully with reference to the 81 relatively "larger" or "smaller" group. In addition, preliminary analysis suggested that there were non-linear associations between the continuous form of B M I and some of the other variables, and that the distributions of B M I were skewed to the right (particularly for women); a categorical form of B M I was considered to be more appropriate. For comparison between the CaMOS participants that were included in this study and those that were excluded, and with the NPHS sample, however, the continuous measure of B M I was used. Since 1998, standardised categories of B M I have been available and widely used in clinical practice and in research to classify individuals as underweight (BMI < 18.5), normal (BMI of 18.5 - 24.9), overweight (BMI of 25 - 29.9), or obese (BMI of 30 or more) (370;371). Canadian guidelines for defining "health risk" categories of B M I (with different cut-off values to those that are now commonly used) were available before 1998 (372), but the use of these categories was much less prevalent in practice in 1996/1997 (at the time of the CaMOS data collection and feedback of B M D test results) than they are now. Canadian osteoporosis guidelines that were available at the time of the CaMOS baseline data collection simply stated that "low body weight" was an important predictor of osteoporosis (230). The B M I variable for this dissertation was therefore divided into quartiles to reflect the participants' and their physicians' expected general perception of their body size and potential risk for osteoporosis at the time. Derivation of the variables for Reproductive Status in women and for the Estimated Neighbourhood Income Quintile involved relatively more complex derivations. The methods for obtaining these particular variables are explained below. Reproductive Status: Women were not specifically asked to report their current reproductive status on the baseline questionnaire, although they were asked i f their periods had stopped for at least 12 months. To derive the closest measure possible of the women's perceived reproductive status at baseline, the answers to several baseline reproductive questions were used: Whether periods had stopped for at least 12 months; the age at which periods had stopped; whether there was a history of ovariectomy; the number of ovaries removed and the age at which they were removed; whether there was a history of hysterectomy and the age at the time of hysterectomy. Four categories of baseline reproductive status were created from the responses: • Premenopausal. Women who had not had a double ovariectomy and whose periods had not stopped for more than 12 months. Women with ovariectomy more than 12 months previously who were unaware of the number of ovaries removed, but whose periods had 82 not yet stopped for more than 12 months were included. Four women who had a hysterectomy (without a double ovariectomy) within the last year while still premenopausal were included. • Naturally Menopausal. Women whose periods had stopped for more than 12 months. Women who had an ovariectomy and/or hysterectomy after their periods had already stopped were included. • Surgically Menopausal. Women who had a double ovariectomy before their periods stopped naturally. Women with ovariectomy (without hysterectomy) who were unaware of the number of ovaries removed but whose periods stopped at the time of ovariectomy were included. • Premenopausal hysterectomy. Women who had a hysterectomy (without double ovariectomy) before their periods had stopped and their "menopausal" status cannot therefore be determined based on whether their periods had stopped. Three women who had a premenopausal hysterectomy with ovariectomy but were unsure how many ovaries were removed were included. Neighbourhood Income Quintile: The first three digits, or forward sortation area (FSA), of each participant's postal code were used to generate an estimate of their neighbourhood income quintile using the SAS program, PCCF+ Version 3G (373). This program automatically assigns an estimated income quintile to each valid postal code by matching it to a series of files derived from the Postal Code Conversion Files that have been prepared by Statistics Canada using summary statistics of the 1996 Canadian census. Although there are more recent postal code conversion files available, the 1996 census version was used to give the closest possible approximation to neighbourhood income quintiles at the time of the CaMOS baseline data collection. It should be noted that the size of FSAs vary considerably across Canada, and the precision of this estimate of neighbourhood income would therefore be expected to vary widely as well. In 1996, the population of Canada was 28,846,761 and there were 1,696 FSAs. Consequently, although the average population in each FSA would have approximated 17,000, the size of the FSAs included in the Canadian census ranges from populations as small as 45 (located in metropolitan Toronto) to populations as large as 129,420 (in rural Quebec) (374). 83 4.4.4 "Correct" Knowledge of B M D Test Results: Outcome Variables The outcome variable of interest for the analysis of correct awareness of test results was whether, when asked at Year 3, participants were able to correctly state their baseline diagnosis from their initial B M D feedback. Derivation of the exactly "correct" outcome variable is summarized in Table 4.1 and demonstrated in Chapter 6. Although the above definition of "correct" was used for the main analysis, other variations of the "correct" variable were derived for comparison because there were alternative ways of defining whether a participant was correct in his or her self-diagnosis. Because the diagnoses of "osteoporosis" and "osteopenia" could both be considered an "abnormal" diagnosis (or an increased risk of fracture) by participants, the categories for osteoporosis and osteopenia were collapsed to derive a "Closely Correct" variable that coded participants as correct based on a dichotomous division between normal and abnormal results. Coding of the "closely correct" variable is demonstrated in Chapter 6. This variable was used as an alternative outcome variable in the analysis of correct knowledge of test results for sensitivity analysis. The criteria used within the centres for a diagnosis of osteoporosis or of osteopenia varied. Although there were guidelines available at the time from the World Health Organization (WHO) regarding definitions of osteoporosis (particularly for menopausal women), as defined by T-scores (50), these guidelines were not always followed in the participant feedback for various reasons (see Chapter 5). In some centres, the WHO guidelines were followed strictly for the diagnosis of osteoporosis or osteopenia. Because the actual T-scores from the B M D test were sometimes sent to participants or to their family physicians, there is a possibility that participants or their physicians may have revised their diagnosis to the equivalent "WHO" diagnosis: In cases where the WHO diagnosis and the diagnosis received in the feedback were incongruent, a family physician may have re-interpreted the B M D scores to a participant and provided the WHO diagnosis, or a participant may have revised it themselves by discovering the WHO criteria on the Web, for example. This scenario presents a possibility that, for some cases, participants would have had a greater chance of being "incorrect" because the participants, either alone or through their physicians, may have modified their diagnoses to the WHO equivalent. For those cases where the diagnosis received in the feedback and the WHO diagnosis were incongruent, the participants may have been correct based on the WHO criteria, but not correct based on the diagnosis provided in their feedback. A version of the correct knowledge of test results was therefore derived whereby participants were correct i f they correctly provided their "WHO diagnosis". In addition, there was a definition of correct 84 whereby the participants were considered correct i f they provided either the correct feedback diagnosis or the correct WHO diagnosis (Either Correct). The WHO diagnosis was determined from the raw T-scores generated by the D X A machines at each centre; this was the measurement that was provided to participants or family physicians with the feedback information. The coding of this variable is identical to that for the "Correct" variable, except that the WHO criteria for osteoporosis and osteopenia were used to derive the diagnosis from the reported T-scores (see Table 4.1). These variables were used as alternatives to the exactly correct variables in sensitivity analyses. In summary, the main outcome variable, "Correct" and three variations of this variable were designed to derive binary outcomes of correct or incorrect responses at Year 3 regarding the B M D diagnosis at baseline. Main outcome variable • Exactly correct ("Correct") based on the diagnosis received in the feedback. Alternative Outcome Variables for sensitivity analyses: • Closely correct regarding a normal v. abnormal (or higher risk) result as reported in the feedback ("Closely Correct"). • Exactly correct based on the diagnosis derived using the WHO criteria ("Correct by WHO"). • Exactly correct based on the diagnosis derived from WHO criteria or the feedback diagnosis ("Either Correct"). 4.4.5 Health Behaviour: Baseline Measures and Outcome Variables The osteoporosis related health-behaviour outcomes were all derived from answers to questions on the CaMOS questionnaire at Year 3 (see Table 4.1); participants were asked about their current behaviour or their behaviour, on average, over the preceding 12 months. The same behavioural measures, covering comparable time frames were recorded at baseline for all of the behaviours (except for information seeking). The average mg per day of dietary calcium intake was derived by CaMOS from the responses to the food intake questions at baseline and Year 3 (see Appendices A and B). 2 0 As per WHO guidelines, T-scores > -1.0 were coded as normal bone density measurements, T-scores < -1.0 and > -2.5 were coded as osteopenia and scores <^  -2.5 were coded as osteoporosis. 85 Participants were asked to estimate their average intake of specific calcium-rich foods over the past 12-month period. These questions were formulated by the CaMOS research group and were not tested for validity or reliability. The calcium content of each of the foods in the questionnaire was estimated by CaMOS with reference to the Canadian Nutrient File (375); the amounts used by CaMOS to calculate calcium intakes are listed in Appendix C. The measurement of calcium in the diet is limited somewhat by shortcomings of the dietary calcium portion of the CaMOS questionnaire. During the period between baseline and the Year 3 follow up, Health Canada approved calcium fortification of orange juice and soy beverage products and these fortified beverages became widely available in Canada.21 These sources of dietary calcium were not, however, included with the questions regarding calcium-rich foods on the Year 3 questionnaire. The baseline measure of calcium in the diet may therefore be a relatively more accurate measure of calcium intake and the estimated intake of dietary calcium at Year 3 may be an underestimate because it does not take into account the potential contribution of fortified 22 beverages. The amount of elemental calcium acquired from supplements was derived by CaMOS from a database of supplements and multivitamins established for the specific purpose of estimating the amounts of calcium in various multivitamin and mineral supplement products. In contrast to the dietary questions that were referenced to the last 12 months, the questions about use of calcium supplements (as well as Vitamin D supplements, OHT and osteoporosis-specific medications) referred specifically to current use at the time of the interview. Total caffeine intake23 was calculated by CaMOS using the values listed in Appendix D. The sources, derivation, and classification of all of the behavioural variables are listed in the last section of Table 4.1. 2 1 Calcium-fortified soy and other plant-based products first became available on the Canadian market when Health Canada issued an Interim Marketing Authorization to allow their sale on November 11, 1997 (376). Fortified orange and tangerine juices became available to Canadians when the first of a series of Temporary Marketing Authorizations to permit the sale of fortified juice was issued on August 1, 1999 (Source: Dr. J. Johnson, Health Canada Food Director, Nutrition Evaluation Division, Health Products and Food Branch, Health Canada; Personal Communication, 2006). A n Interim Marketing Authorization was recently issued, in April 2006, for the sale of fortified orange and tangerine juices (377). 2 2 Results from unpublished marketing data (A.C. Nielson; by personal communication to Dr. Susan Barr, 2007) show that sales of fortified soy and rice beverages made up less than 1% of the total fluid dairy sales in Canada during 2003-2005, which suggests that the exclusion of these beverages from the questionnaire may not have had a substantial impact on dietary calcium intake estimates during 1999-2000. 2 3 The caffeine content of beverages was estimated by CaMOS with the aid of various sources including the Web and popular literature. The amounts used by CaMOS are comparable, however, to estimates from the Canadian Nutrient File (375). 86 4.5 Data Management and Analyses The data management, manipulations, statistical analyses and modelling were performed using SPSS, version 13.0 (SPSS Inc., Chicago IL.) and SAS, version 8.0 (SAS Institute Inc., Cary, NC.) 4.5.1 Data Management Where responses were missing to questions on the CaMOS questionnaires the data were coded as missing. Extensive efforts were made to minimize missing data from the information that was collected at the centres regarding content and destination of the feedback; i f essential24 original copies of letters were missing from the files every attempt was made to obtain alternative copies through the hospital information systems or from the central CaMOS data storage facility in Quebec. If the feedback information was still not retrievable, the information was coded as missing. Data obtained for every variable was checked for unusual or out of range values by means of histograms, bar charts or descriptive statistics. If unusual values were identified, they were verified by referral to the original questionnaires (CaMOS personnel at the data centre in Montreal, Quebec were contacted for this purpose), and corrected i f necessary. 4.5.2 Data Analyses A l l descriptive statistics and association and regression modelling procedures were stratified by gender. Comparisons between the eligible mid-aged participants and the ineligible CaMOS mid-aged participants (those who did not undergo a B M D at baseline, were not followed up at Year 3, or had missing data for their feedback diagnosis), as well as comparisons to an equivalent age- and gender-matched sample from the 1996 NPHS, were made using the Chi-squared test for homogeneity (for the categorical data) or the T-test (for continuous data). The agreement between the "feedback" diagnosis and the WHO diagnosis based on the raw T-scores from the B M D was investigated using the Kappa statistic. The analyses of correct knowledge of test results and of osteoporosis-related health behaviour began with a descriptive analysis of the main explanatory variables, the potential confounders or effect modifiers and the outcome variable(s). This was followed by univariate Copies of individual letters were considered essential in all centres where the diagnostic criteria or hip site for hip diagnosis varied between participants, where the diagnostic criteria were unclear or the content of the feedback was not identical for all participants or family physicians. 87 analyses of the relationships between the potential explanatory variables and the relationships between each explanatory variable and the outcome variable(s) to assess the potential for confounding and collinearity. Correlations were determined for the pairs of continuous variables. Associations between the pairs of categorical or discrete variables were estimated with the Chi-square test. Comparison of the means of the continuous variables between subgroups was carried out with the t-test (for variables with two subgroups) or one-way analysis of variance (for variables with three or more subgroups). Multiple regression modelling was used to assess the associations between the explanatory variables and each of the knowledge and health behaviour outcomes, while taking into account other variables that were associated with both the explanatory variables and the outcome. Logistic regression models were used for the binary outcomes; knowledge of test results, information seeking, calcium supplement use, Vitamin D supplement use, exercise participation, osteoporosis-related medication use, smoking status, high alcohol intake, and high caffeine intake. Linear regression models were used for the continuous outcomes (dietary and total calcium intake and caffeine intake) with transformations of the data, where necessary, as discussed below. The main focus of this study was the relationships among the diagnosis as a result of the B M D tests, the destination of the feedback, and the outcome variables: knowledge of test results and osteoporosis-related health behaviour after three years. For this reason, these main explanatory variables of interest were included in all models, whether or not they made a significant contribution to the model. In addition, age group was included in all models because it was seen as an important potential effect modifier; physicians and participants may have taken the risk of osteoporosis more seriously, or acted on it, i f the participant was older (in the 50- to 60-year-old age group). The inclusion of any form of recommendation for treatment or for further referral, in the feedback sent to the FP or the participant, was considered to be an important potential confounder of the association between the diagnosis, the destination of the feedback and the knowledge and behavioural outcomes; a recommendation for action was included in all models. Finally, to control for baseline health behaviour (i.e., before feedback of B M D test results), all models of osteoporosis-related health behaviour included the baseline measure of the behaviour as an explanatory variable when available.25 The exception was the information seeking variable where no baseline assessment was available. Participants were not asked about their previous information seeking behaviour at baseline. 88 To determine which other explanatory variables should be entered into the multivariable regression analysis (in addition to those listed in the paragraph above), a cut-off of p < 0.2 in the univariate analysis was used as a criterion for consideration of variables for the multivariable model (378). Manual backwards stepwise multivariate regression was performed using those variables found to be significantly associated with the outcome in the univariate analyses. Starting with all of the variables that met the above criteria, the variable with the weakest, non-significant association with the outcome variable was removed and the impact of removal of the variable was evaluated by calculating the difference between the Chi-squared value for the model with the variable and the model with the variable removed. If this difference was not significant (p > 0.05) the variable's contribution to the model was considered to be non-significant and the variable was removed. In addition, the effect of removal of the variable on the coefficients remaining in the model and their standard errors was considered. Finally, the multivariable model was estimated with only those variables included that contributed significantly to the model so that all cases with data for these variables were taken into consideration. The aim was to achieve the most parsimonious models possible while including all important and significant variables as well as the explanatory variable of interest. The analyses of the association between the feedback diagnosis and correct knowledge of the B M D test results are presented in Chapter 6. Descriptive information about correct awareness of test results is presented first, and is followed by the results of univariate analyses between the explanatory variables as well as the potential confounders and correct awareness. The results of the multivariable modelling are then presented followed by an examination of potential effect modification by diagnosis and then by age group. For the assessment of potential effect modification, the analyses were repeated in each of the three diagnostic subgroups (normal, osteopenia and osteoporosis) and in each of the two age groups (40 to 50 years and 51 to 60 years) for women and men, separately. A l l explanatory variables that were included in the main models were included as covariates except for those that were not relevant to the subgroup analyses (i.e., "diagnosis" in the diagnostic subgroup analyses and "age group" in the age group subgroup analyses). The separate analysis of the older age group includes the assessment of the potential effects of a report of a significant fracture from the X-ray on correct awareness of test results (see below). Chapter 6 ends with the sensitivity analyses (also described below). The associations among the feedback diagnosis and each of the behavioural outcomes are presented in Chapter 7. Both the feedback diagnosis and the recalled diagnosis, as reported at Year 3 by the participants (see Table 4.1), were considered, in separate models, as potential 89 explanatory variables for these behavioural analyses. The associations between the feedback diagnosis and the behavioural outcomes are presented as the main analyses so that the effectiveness of a diagnosis reported from a B M D test by D X A can be estimated. Analyses were repeated with the recalled diagnosis (which represents the diagnosis that participants believed that they had received three years earlier) as the main explanatory variable. The results of these subsequent analyses of "perceived" diagnosis and behavioural change pertains to the question of what the participants who had a correct "awareness" or "perception" of their test results did with this knowledge. These results are compared with those from the main analysis of the feedback diagnosis. A description of the health behaviour at Year 3 is presented first for each set of analyses of the association between diagnosis and health behaviour in Chapter 7. This is followed by the results of univariate analyses between the explanatory variables as well as the potential confounders and the behavioural outcomes. The results of the multivariable modelling are then presented. Subsequently, examination of the potential effects of the results of the X-ray test on those aged 50 to 60 years (as described below) are presented by subgroup analyses of the older age group of women and men who attended the X-ray test. The results of the sensitivity analyses of these models are also included in Chapter 7 (also described below). 4.5.3 Further Analyses, Sensitivity Analyses and Diagnostics As previously mentioned, only two thirds of the participants in this study were eligible to attend the spinal X-ray and receive the X-ray results because spinal X-ray was only offered to those CaMOS participants who were aged 50 year or older at baseline. A l l participants who attended the X-ray (or their physicians) received a report indicating whether a significant prevalent fracture was evident on the X-ray. The effect of a report of a significant prevalent fracture from the X-ray was examined by multivariable regression in the subgroups of women and men who were aged 50 to 60 years at baseline and attended the X-ray test. The question of interest was whether the report of a significant fracture increased the participants' knowledge of a diagnosis of osteopenia or osteoporosis, or contributed to osteoporosis related health-behaviour change. Because some of the centres followed the same protocol for their feedback of B M D results for all participants within their centre, the destination variable was partially nested within the centre. For this reason, analyses with "Centre" as an explanatory variable were conducted separately from those with the "Destination" variable and the models were compared. 90 For the knowledge outcome, sensitivity analyses were performed using the three alternative versions of correct awareness, as described above. Further, the analyses were repeated for all models (knowledge and behaviour outcomes) while excluding those cases in any centre where the feedback was not sent according to the protocol. The purpose of this sensitivity analysis was to examine the potential effects of a bias in the destination of the test results in those centres where the intended protocol deviated for some cases. Model diagnostics were performed on the final logistic and linear multivariate models to identify potentially influential cases and to determine whether the model's assumptions had been met. Predicted values were plotted against standardised residuals and the regression analyses were repeated after influential cases had either been removed, or their values modified by applying a maximum ceiling value or by. transformation. 91 CHAPTER 5: Results I. The Study Sample and the Explanatory Variables 5.1 The Study Sample The selection of participants for this study is demonstrated in Figure 5.1. There were 3,139 full participants aged 40 to 60 years at baseline in the CaMOS study. Of these, 1,837 women and 869 men (86% of the original CaMOS mid-aged sample) met the selection criteria of completion of a baseline B M D test in addition to the questionnaire and follow-up at Year 3. Figure 5.1: Selection of the Study Sample from the 40- to 60-year-old CaMOS Participants 2,105 women and 1,034 men enrolled in the CaMOS study and completed the full questionnaire at baseline (100%> of the initial sample) 83 women and 34 men declined the BMD test 3,022 completed the B M D test at baseline (96%o of the initial sample) 185 women and 131 men lost to follow up 2,706 participants followed up at Year 3 (86%o of the initial sample) S t u d y S a m p l e 1 ,837 w o m e n ( 8 7 % o f t h e i n i t i a l s a m p l e o f w o m e n ) 8 6 9 m e n ( 8 4 % o f t h e i n i t i a l s a m p l e o f m e n ) 2 6 A "full" participant is classified as an eligible person who was invited to participate and who completed the full version of the CaMOS questionnaire. 92 5.1.1 Missing Data In a few cases, copies of the reports that were sent to the participants and/or their family physicians were not available at the centres arid the feedback destination or diagnosis could not be derived from other information. Consequently, there were 18 women and 4 men in the study sample for whom the feedback diagnosis was missing and 5 women and 4 men With missing information on the destination of their feedback. Responses were missing to some questions on the CaMOS questionnaire, but the proportion did not exceed 5% for any one variable, and was typically much less than 1%. These responses, or the variables that were derived from them, were left as missing because they could not be inferred. A l l selected participants were included in all analyses except when there was a missing value for an included variable. Missing information on specific variables is noted below (see Tables 5.1a and 5.1b). Missing data on the outcome variables, and the self-recalled bone density test results (which was part of the outcome variable "correct" as well as an explanatory variable in the behavioural analyses), are addressed in the following chapters. 5.1.2 Baseline Characteristics and Generalisability of the Study Sample The baseline characteristics of the women and men who participated in this study are listed in Tables 5.1a and Table 5.1b, respectively. For comparison purposes, the distributions of the same characteristics are shown in the second column of the table for the 40-60 year-old participants from the CaMOS study that were "excluded" because of a missing B M D test or loss to follow-up at Year 3. The last column in the table compares the distributions of the gender- and age-matched sample of subjects from the 1996 NPHS for those variables that were comparable between the two studies. Table 5.1a: Comparison of Baseline Characteristics between Included Women, Excluded CaMOS Women and a Matched Sample of Women from the NPHS STUDY SAMPLE OF W O M E N N = 1837a E X C L U D E D W O M E N N = 268b NPHS W O M E N N = 8383 Body Mass Index0 Mean (SD) 27.1 (5.4) Mean (SD) 27.3 (5.2) Mean (SD) 25.3 (4.7)*** 93 STUDY SAMPLE OF W O M E N N = 1837A EXCLUDED WOMEN N = 268b NPHS WOMEN N = 8383 N (%) N (%) N (%) Diagnosis Normal in Feedback Osteopenia Osteoporosis 943 (52%) 563 (31%) 313 (17%) 92 (50%) 60 (32%) 33 (18%) Significant Prevalent Fractured 92 (8%) 14 (11%) Destination of Feedback To Family Physician To Participant To Both 382 (21%) 935 (51%) 515 (28%) 52 (28%)* 79 (43%) 53 (29%) Recommendation for Referral or Treatment 135 (7%) 19 (10%) Age Group 40-49 years 50-60 years 562 (31%) 1275 (69%) 60 (22%)** 208 (78%) "Non-white" Race/Ethnicity 107 (6%) 21 (8%) 701 (8%)*** Reproductive Status Premenopausal Naturally Menopausal Surgically Menopausal Premenopausal hysterectomy 638 (35%) 704 (38%) 140 (8%) 354 (19%) 75 (28%) 104 (39%) 28 (10%) 61 (23%) Education Incomplete High School Complete High School Postsecondary Education University Degree 458 (25%) 338 (18%) 708 (39%) 333 (18%) 98 (37%)*** 45 (17%) 92 (34%) 33 (12%) 1999 (24%)*** 2982 (36%) 1987 (24%) 1344(16%) Estimated Neighbourhood Income Quintile Lowest Second Lowest Middle Second Highest Highest 303 (17%) 331 (18%) 323 (18%) 389 (22%) 456 (25%) 62 (23%) 48 (18%) 49 (19%) 52 (20%) 54 (20%) Centre Calgary Hamilton Halifax Kingston Quebec City Saskatoon St. John's Toronto Vancouver 212 (12%) 190 (10%) 192 (11%) 190 (10%) 239 (13%) 205 (11%) 196 (11%) 194 (11%) 219 (12%) 19 (7%)*** 26 (10%) 45 (17%) 35 (13%) 37 (14%) 12 (5%) 22 (8%) 39 (15%) 33 (12%) Awareness of Family History 312 (17%) 44 (17%) 94 STUDY SAMPLE OF W O M E N N = 1 8 3 7 a E X C L U D E D W O M E N N = 268b NPHS W O M E N N = 8383 P r e v i o u s D i a g n o s i s o f O s t e o p o r o s i s 8 3 ( 5 % ) 21 (8%)* H i s t o r y o f C o m o r b i d i t y 9 5 ( 5 % ) 21 (8%) H i s t o r y o f " H i g h R i s k " M e d i c a t i o n E x p o s u r e 2 2 7 ( 1 2 % ) 42 (16%) G e n e r a l Poor/ Fair/ Good H e a l t h Very Good/ Excellent 6 3 9 ( 3 5 % ) 1 1 9 7 ( 6 5 % ) 118 (44%)** 150 (56%) 3450 (41%)*** 4933 (59%) C u r r e n t C i g a r e t t e S m o k e r 3 0 4 ( 1 7 % ) 66 (25%)** 1834 (22%)*** C u r r e n t C a S u p p l