Table 4.
Summary of methodological considerations by study design and overall.
| Cohort (n = 9) n (%) | Case—control (n = 5) n (%) | All studies (n = 14) n (%) | |
|---|---|---|---|
| Drug exposure | |||
| Bisphosphonate only | 7 (78) | 3 (60) | 10 (71) |
| Bisphosphonate or other osteoporosis drug | 2 (22) | 2 (40) | 4 (29) |
| Baseline prior to index date to define covariates and prior use | |||
| ≤ 6 months | 3 (33) | 2 (40) | 5 (36) |
| 1 year | 3 (33) | 1 (20) | 4 (29) |
| 2 years | 1 (11) | 1 (20) | 1 (7) |
| variable/not provided | 2 (22) | 1 (20) | 3 (21) |
| New user design | 7 (78) | 5 (100) | 12 (86) |
| Bisphosphonates | 6 (67) | 1 (20) | 7 (50) |
| All osteoporosis drugs | 1 (11) | 4 (80) | 5 (36) |
| Need > 1 prescription for inclusion | 2 (22) | 1 (20) | 3 (21) |
| Accounted for switching | 4 (44) | 4 (80) | 8 (57) |
| Allowed between bisphosphonates | 3 (33) | 0 | 3 (21) |
| Allowed any switching | 1 (11) | 1 (20) | 2 (14) |
| Excluded switchers | 0 | 2 (40) | 2 (14) |
| Censored switchers | 0 | 1 (20) | 1 (7) |
| Accounted for immeasurable time Assumed 100% compliance | 2 (22) | 2 (40) | 4 (29) |
| Follow-up time for fracture identification began | |||
| 90 days | 1 (11) | 3 (60) | 4 (29) |
| 6 months | 3 (33) | 0 | 3 (21) |
| 1 year | 1 (11) | 1 (20) | 2 (14) |
| Any | 4 (44) | 1 (20) | 5 (36) |
| Compliance measurement method | |||
| Ascertainment period | 0 | ||
| Entire follow-up period | 6 (67) | ||
| Time varying compliance | 3 (33) | ||
| Primary method of quantifying compliance | |||
| ≥ 80% | 7 (78) | 3 (43) | 10 (71) |
| > 2 Categorical groups | 1 (11) | 2 (40) | 3 (21) |
| Continuous (linear regression) | 1 (11) | 0 | 1 (7) |
| Minimum confounders for adjustment in model | |||
| Age | 9 (100) | 2 (40)a | 11 (79) |
| Prior fracture | 9 (100) | 4 (80) | 13 (93) |
| Sex | 9 (100) | 5 (100)b | 14 (100) |
| Comorbidity score | 2 (22) | 0 | 2 (14) |
| Case—control matching | |||
| Age | 4 (80) | ||
| +/-1 year | 1 (20) | ||
| +/-3 years | 1 (20) | ||
| No specification | 2 (40) | ||
| Drug class | 4 (80) | ||
| Sex | 5 (100)b | ||
| Date | 3 (60)c | ||
| Follow up | 5 (100) | ||
| All five criteria | 2 (40) |
Index date defined as date of treatment initiation.
a
Does not consider matching.
b
Studies only evaluated females.
c
Three clearly matched on date: it is unclear whether the other two case—control studies matched on date.