Table 3.
Fracture incidence among users of any insulin (N = 2,980) vs matched non-users (N = 24,900), stratified by quartiles of insulin medication possession ratio (MPR) within first year of first insulin use.
| MPR | Non-users of insulin (14,900) | Users of insulin (2,980) | SHR | adj SHR | |||||
|---|---|---|---|---|---|---|---|---|---|
| Quartile | Median (IQR) | N | Median follow-up (years) | Rate (per 1000 PYs) | N | Median follow-up (years) | Rate (per 1000 PYs) | ||
| 1 | 21 (21–31) | 235 | 4.5 | 8.8 | 11 | 0.8 | 7.8 | 0.94 (0.51–1.73) | 0.92 (0.50–1.71) |
| 2 | 41 (41–51) | 147 | 4.6 | 7.1 | 17 | 1.3 | 12.6 | 1.85 (1.12–3.04) | 1.85 (1.12–3.05) |
| 3 | 62 (62–72) | 97 | 4.6 | 7.2 | 18 | 1.7 | 17.9 | 2.58 (1.56–4.26) | 2.58 (1.56–4.26) |
| 4 | 103 (84–123) | 151 | 4.7 | 9.5 | 10 | 1.9 | 7.5 | 0.82 (0.43–1.57) | 0.84 (0.45–1.60) |
Only the first insulin use considered (i.e. switching types was not accounted for), then censored. All analyses include incident insulin use as a time dependent covariate. Analyses are PS matched, on a 5:1 ratio of non-users to users in a separate model from the main analysis. Adjusted analyses control for outstanding confounders (P < 0.1) and for propensity score.
Abbreviations: N (number of fractures), Interquartile range (IQR), person-years (PYs), subhazard ratio (SHR), adjusted (adj).