Table 4. Risk of fractures for patients with and without post-stroke rehabilitation stratified by sex and age.
| Rehabilitation | Non-rehabilitation | Model 1† | Model 2‡ | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| n§ | Fracture cases | Person-years | Incidence rate* | n§ | Fracture cases | Person-years | Incidence rate* | aHR (95% CI) |
p value | aHR (95% CI) |
p value | |
| Male | ||||||||||||
| Age 20–64 | 1039 | 25 | 993.2 | 2.5 | 1039 | 30 | 991.8 | 3.0 | 0.85 (0.50–1.45) |
0.549 | 0.87 (0.51–1.49) |
0.618 |
| Age ≥ 65 | 1440 | 64 | 1313.0 | 4.9 | 1440 | 56 | 1280.0 | 4.4 | 1.12 (0.78–1.61) |
0.541 | 1.24 (0.85–1.81) |
0.274 |
| Female | ||||||||||||
| Age 20–64 | 476 | 17 | 451.7 | 3.8 | 476 | 11 | 454.9 | 2.4 | 1.60 (0.75–3.43) |
0.228 | 1.70 (0.78–3.71) |
0.183 |
| Age ≥ 65 | 1172 | 119 | 1026.3 | 11.6 | 1172 | 73 | 1012.6 | 7.2 | 1.62 (1.21–2.17) |
0.001 | 1.82 (1.33–2.50) |
<0.001 |
The propensity score matching procedure was performed after initial stratification for sex and age of the total of “11,806 patients”§ included in this study.
* Per 100 person-years, calculated by correcting immortal time.
† Model 1 used a multivariate Cox proportional hazard regression model to adjust for all baseline characteristics listed in Table 1.
‡ Model 2 used a stratified Cox proportional hazard regression model stratified for propensity-score matched groups.
§ After propensity score matching, a total of 8,384 patients were enrolled. However, the values in the table do not add up to 8,384 patients. This is because in order to keep all paired relationships when stratifying sex and age, we stratified sex and age from the initial total 11,806 patients. After stratification, we separately performed propensity score matching in each stratum. Therefore, there are 2078, 2880, 952 and 2344 patients included in each stratum.