CANADIAN NETWORK FOR ADVANCED INTERDISCIPLINARY METHODS FOR

COMPARATIVE EFFECTIVENESS RESEARCH

M. Abrahamowicz (McGill), S Bernatsky, (McGill), L. Pilote (McGIll) A. Levy, (Dalhousie) L. Levesque (Queen’s), Y.

Moride (U de M) et al.

Research scope To address drug safety & effectiveness from a

‘prospective cohort’ perspective

Over-arching goal To enhance the validity and accuracy of research on real-

world comparative drug effects

Ultimate aim To better understand which treatment works best for

which patients.

Cristiano Moura1, Sasha Bernatsky1, Elham Rahme1, Marie-Eve Beauchamp1, Louis Bessette2, Jonathan Adachi3, Alexandra

Papaioannou3, David Goltzman1, Jerilynn Prior4, Nancy Kreiger5, Tanveer Towheed6, William Leslie7, Stephanie Kaiser8, Laura

Pickard9, George Ioannidis3, Lisa-Ann Fraser9, Michal Abrahamowicz1

1McGill University, 2Université Laval, 3McMaster University, 4University of British Columbia, 5University of Toronto, 6Queen's

University, 7University of Manitoba, 8Dalhousie University, 9Western University

Antidepressants are suspected of increasing risk of fractures, especially selective serotonin & serotonin/ norepinephrine reuptake inhibitors, SSRIs, SNRIs

Objective: To evaluate whether SSRI/SNRI use is associated with fracture risk

Introduction

Sedation, daytime

drowsiness

Fracture

Increase risk of falls

Serotonin and bone

physiology

Decreased BMD

CaMos: prospective cohort

Nine regional centres across the country

Women and men ≥25 years

Recruitment: July 1995 to Sept. 1997

Follow-up: 10-years

Exposure and outcomes

Drug exposure assessed at 3 time points: year zero (baseline) and then at each 5 year follow-up

Annual questionnaires ascertained if subjects had experienced a fracture in the previous year; reports were confirmed by radiography. Fragility fractures were defined as those due to minimal trauma (eg, fall from standing height).

For this study, we examined only participants aged >50

Methods

Other covariates

Demographic and socioeconomic

Age, sex, employment status

Bone mineral density, BMD

Lumbar spine (L1–L4) and total hip

History of falls

Depressive symptoms

Mental component score (MCS) from SF-36

Clinical variables and lifestyle-habits

Smoking, alcohol, physical activity, Charlson Comorbidity Index

Other drug exposures

Anxiolytics, other antidepressants, glucocorticoids, bisphosphonates, calcium and vitamin D supplements

Methods

Data analysis

Multivariable Cox proportional hazard regression

Time to first fragility fracture

Failure to respond to the annual questionnaire was considered a loss to follow-up

SSRI/SNRIs and other drugs were modeled as time-dependent variables

Adjusted for previously described demographic and clinical variables

Methods

Descriptive characteristics

6,645 participants

192 (2.9%) SSRI/SNRI users at baseline 330 (5.9%) at year 5 and 333 (8.3%) at year 10

955 events 74 in the exposed group and 881 in the non-exposed group

Results

Results: Table 1 - Baseline selected characteristics - N (%) or mean ± SD

Variable SSRI+SNRI

nonusers

SSRI+SNRI

users

Sex (Female) 1873 (70.97) 32 (83.3)

Age (years ± SD) 65.7 ± 8.9 65.2 ± 8.7

Education level (High school or higher) 3896 (60.4) 125 (65.1)

Modified Charlson Index (mean ± SD) 0.37 ± 0.74 0.59 ± 0.93

Bone mineral density (mean ± SD)

Total hip

Lumbar spine

0.88 ± 0.16

0.95 ± 0.18

0.86 ± 0.14

0.95 ± 0.18

Depressive symptoms (MCS<42) 885 (13.8) 75 (39.7)

Falls in previous month 398 (6.2) 31 (16.2)

Medication use

Other antidepressants

Anxiolytics

Biphosphonates

Corticosteroids (oral+IV)

251 (0.04)

294 (0.05)

112 (0.02)

89 (0.01)

19 (9.9)

28 (14.6)

4 (0.02)

2 (0.01)

Results

Incidence & unadjusted/adjusted* hazard ratios (HR) with confidence intervals (CI): SSRI/SNRIs & fragility fractures

User Non-User

Events 74 881 -

Person-years 2257 50216 -

Incidence/100 person-years 3.3 1.7 -

HR (unadjusted) and 95% CI - - 1.87 (1.48-2.39)

HR (adjusted) and 95% CI - - 1.68 (1.32-2.14)

*adjusted for age, sex, Charlson, depressive symptoms, falls at baseline, BMD

Time-to-event analysis: SSRI/SNRI use was associated with fragility fracture

Sensitivity analyses Using separate time-dependent indicators of recent

and former use, the adjusted HR for recent use was 1.62 (95%CI: 1.20-2.19), and for former use HR: 1.42; 95%CI: 0.88- 2.31)

Dose effect analysis: using reference categories based on ‘defined daily dose*’, the effect of these drugs in the highest dose group was greater (2.02; 95% CI=1.24-3.17) than the reference group

*average maintenance dose per day

Our results regarding antidepressants use & fracture risk are consistent with previous investigations

Case-control study in Denmark (Vestergaard et al. 2013)

Meta-analysis including 34 studies (Rabenda et al. 2013)

CaMos report of 5 years follow-up (Richards et al. 2007)

Confounding by indication

Depression itself has been associated with fracture risk

We attempted to control for depressive symptoms

Discussion

Strengths

Large longitudinal cohort, 10-year period

Fractures were confirmed by radiology

Limitations

Impossible to know if individuals started or stopped these therapies between assessment points

Sensitivity analysis showed consistent results under differents assumptions for exposure

SNRI and fracture risk

Limited number of participants under this drugs

Discussion

Conclusions SSRI/SNRI use was associated with an important

increased risk of fragility fractures, after controlling for important confounders and/or effect modifiers.

Risks and benefits should be carefully considered when these drugs are prescribed for older people.

Risk stratification: History of falls, concurrent use of anxiolytics

Consider monitoring/prophylaxis

• Moura C, Bernatsky S, Rahme E, Beauchamp M-E, Bessette L, Adachi J, Papaioannou A, Goltzman D, Prior J, Kreiger N, Towheed T, Leslie W, Kaiser, Pickard L, Ioannidis, Fraser L-A, Abrahamowicz M. Psychotropic medication use and 10-year incident fracture risk in men and women ages 50 and older in the population-based Canadian Multicentre Osteoporosis Study Canadian Association for Population Therapeutics Conference Toronto, November 17-19, 2013.

• Moura C, Bernatsky S, Rahme E, et al. SSRIs and fracture risk in a population based population: Preliminary Results CaMOS Investigators meeting, Montreal, April 2013

• Upcoming teleconference (April 10th) with stakeholders

• Webinar later this year (CANRAD network)

• CAN-AIM team

• CIHR-DSEN

• CaMos

Baseline Year 5 Year 10

SSRI/SNRI non-users

SSRI/SNRI users

SSRI/SNRI non-users

SSRI/SNRI users

SSRI/SNRI non-users

SSRI/SNRI users

N 6453 192 5226 330 3678 333 Female (%) 4580 (71) 160 (83.3) 3752 (71.8) 290 (87.9) 2677 (72.8) 285 (85.6)

Age (years±SD) 65.71 (8.9) 65.25 (8.7) 64.96 (8.5) 65 (8.2) 63.43 (7.9) 63.65 (8.0) High school +(%) 3896 (60.4) 125 (65.1) 3238 (62) 201 (60.9) 2408 (65.5) 198 (59.5) Employed (%) 1082 (16.8) 24 (12.5) 943 (18) 44 (13.3) 782 (21.3) 53 (15.9) Regular activity 3699 (57.3) 106 (55.2) 3082 (59) 184 (55.8) 2225 (60.5) 182 (54.7) Smoker (%) 930 (14.4) 39 (20.3) 711 (13.6) 48 (14.5) 458 (12.5) 53 (15.9) Alcohol mean±SD) 0.462 (3) 0.231 (2) 0.462 (4) 0.231 (2) 0.692 (4) 0.231 (2) Comorbidities 1583 (24.5) 70 (36.5) 1170 (22.4) 74 (22.4) 730 (19.8) 74 (22.2) Depressive (%) 885 (13.8) 75 (39.7) 649 (12.5) 121 (36.9) 426 (11.7) 101 (30.3) Falls 398 (6.2) 31 (16.1) 313 (6) 30 (9.1) 220 (6) 24 (7.2) Other antidepressants Anxiolytics Antihypertensives Antipsychotics Biphosphonates Corticosteroids

251 (3.9) 294 (4.6) 1276 (19.8) 28 (0.4) 112 (1.7) 89 (1.4)

19 (9.9) 28 (14.6) 34 (17.7) 0 (0) 4 (2.1) 2 (1)

183 (3.5) 198 (3.8) 955 (18.3) 21 (0.4) 86 (1.6) 69 (1.3)

47 (14.2) 49 (14.8) 66 (20) 2 (0.6) 9 (2.7) 6 (1.8)

111 (3) 113 (3.1) 586 (15.9) 13 (0.4) 50 (1.4) 36 (1)

44 (13.2) 39 (11.7) 66 (19.8) 2 (0.6) 11 (3.3) 3 (0.9)

Calcium Vitamin D

902.9 (754.0) 3.214 (8.9)

969.4 (810.9) 4.7 (9.4)

907.9 (751.9) 3.393 (8.9)

1028 (867.0) 3.75 (9.7)

914.4 (743.1) 3.393 (8.9)

974.9 (925.5) 3.661 (9.5)

*19% loss to follow up by year 5, 40% by year 10