Table 3.

Summary of methodological characteristics of nested case–control studies.

	[Blouin et al. 2008]	[Cotte et al. 2008]	[Meijer et al. 2008]	[Rabenda et al. 2008]	[Weycker et al. 2007]
Treatment	ALD, RSD	ALD, RSD, ETD, RAL, SR	ALD, RSD, ETD	ALD	ALD, RSD, CCT, HT, RAL
Evaluated drug class separately	NA	No	NA	NA	No
Baseline (lookback) period	2 years	NP	1 year	3 months	6 months
Period of covariate identificationa	1) Baseline: 1 year for all study covariates (prior fx 2 years) 2) Follow-up: comorbidities	1) Baseline 2) Follow-up: comorbidities, comedications and supplements in year prior to fx	1) Baseline 2) Follow-up: Use of non-OP comedications	Baseline only	1) Baseline 2) Follow-up: Age at fx date
New users (drug)	Yes (all OP drugs)	Yes (all OP drugs)	Yes (any BP)	Yes (Any BP or RAL)	Yes (all OP drugs)
Allowed switching	No censored switchers	Yes	NP	No excluded switchers	No excluded switchers
>1 Rx needed	No	No	1 + in two years before outcome	No	No
Cases –start of fracture identification	Fx at least 1 year after initiation	Fx 90 days after initiation	Fx 3 months after initiation	First fx after initiation	Fx 90 days after initiation
Matching of controls
Age	Yes (+/−1yr)	Yes	No	Yes	Yes (+/−3 yr)
Sex	Yes	Yes	Yes	Yes	Yes
Treatment	Yes	No	Yes	Yes	Yes
Index Date	Unclear	Yes	Unclear	Yes	Yes
Duration of Follow-up	Yes	Yes	Yes	Yes	Yes
Minimum confounders for adjustment in model
Age	No (matched)	No (matched)	Yes	No (matched)	Yes
Prior fracture	Yes	Yes	Yes	No	Yes
Comorbidity Score	Yesb	No	No	No	No
Immeasurable time	Assumed 100% compliance	NP	NP	NP	Assumed 100% compliance
Compliance measure	PDC	MPR	PDC	PDC	MPR
How compliance quantified (1 = Primary 2 = Secondary)	1) ≥80% 2) <50% vs. ≥ 50%, <90% vs. ≥ 90%	1) <20%, 20–<40%, 40–<60%, 60–<80%, ≥ 80% 2) Used ROCs to generate threshold	≥ 80%	1) 1 Year PDC ≥ 80% 2) Continuous 0–100%	<30%, 30–69%, 70–89%, ≥ 90%

ALD, alendronate; BP, bisphosphonates; CCT, calcitonin; ETD, etidronate; Fx, fracture; GCC, glucocorticoids; HT, hormone therapy; IBD, ibandronate; MPR, medication possession ratio; NA, not applicable; NP, data not provided; OP, osteoporosis; PDC, proportion of days covered; RAL, raloxifene, RSD, risedronate; Rx, prescription; SR, strontium ranelate; Unclear, not explicitly stated in the manuscript, however plausible based on the text.

^aCovariates identified based on 1) ‘baseline’ – data collected during pretreatment index (lookback) period; or 2) ‘follow-up’ – data collected after treatment index date or as time varying.

^bConfounder considered – however, it is unclear what was included in the final adjusted analysis.