Table 6.
Effect of physical activity on mobility outcomes within levels of hospital exposure
| Exposurea | Operational definition | HR (95% CI) | P value* |
|---|---|---|---|
| Onset of MMD | |||
| Any hospitalization | Not hospitalized | 0.77 (0.62–0.95) | 0.903 |
| Hospitalized | 0.79 (0.58–1.1) | ||
| Number of hospitalizations | 0 | 0.77 (0.62–0.95) | 0.742 |
| 1 | 0.85 (0.59–1.2) | ||
| 2 or more | 0.66 (0.37–1.2) | ||
| Number of days hospitalizedb | 0 | 0.79 (0.66–0.95) | 0.804 |
| 3 | 0.80 (0.67–0.95) | ||
| 10 | 0.81 (0.65–1.0) | ||
| Recovery from MMD d | |||
| Any hospitalizationc | |||
| Model 1 | Not hospitalized | 1.2 (0.88–1.7) | 0.670 |
| Hospitalized | 1.5 (0.71–3.0) | ||
| Model 2 | Not hospitalized | 1.2 (0.94–1.6) | 0.392 |
| Hospitalized | 1.6 (0.90–2.9) | ||
| Model 3 | Not hospitalized | 1.3 (0.88–2.1) | 0.477 |
| Hospitalized | 2.0 (0.71–5.4) | ||
Abbreviations: CI confidence interval, HR hazard ratio, MMD major mobility disability
* Values represent statistical interaction between exposure and study group on mobility outcome
a Assessed during the interval preceding the outcome
b Results are provided for a range of fixed values. Three days was the median length of hospital stay, and 10 days allowed for long lengths of stay and more than one hospital admission
c Results are not available for the number of hospitalizations or number of days hospitalized because the number of participants with more than one hospitalization during the at-risk period was small
d All models include clinical site, age, and gender as covariates; Model 2 uses inverse probability weighting based on MMD, whereas Model 3 uses inverse probability weighting based on withdrawal/missed follow-up, as described in the Methods