Analysis based on the main outcome (ie, the frequency of athletes with at least one injury complaint impacting on athletics participation during the study period)
| Risk/RD, % | 95% CI | P value | |
| Intention-to-treat analysis | |||
| Risk in control group (ref) | 69.9 | 62.2% to 77.6% | |
| Risk in intervention group | 66.8 | 60.8% to 72.8% | |
| Cumulative risk difference | −3.1 | −12.9% to 6.6% | 0.530 |
| As treated analysis | |||
| Risk amongst athletes not performing the AIPP-intervention (ref) | 67.8 | 63.0% to 72.7% | |
| Risk amongst athletes performing the AIPP-intervention | 80.7 | 53.8% to 107.6% | |
| Cumulative risk difference | 12.9 | −14.4% to 40.3% | 0.352 |
| Instrumental variable analysis | |||
| Counterfactual control risk (ref) | 70.2 | 61.7% to 78.6% | |
| Counterfactual intervention risk | 17.7 | −141.9% to 185.4% | |
| Cumulative risk difference | −52.5 | −218.7% to 113.7% | 0.536 |
RD=measure of association is cumulative risk difference measured by time-to-event (generalised linear model with pseudo observations).3 Risk is the cumulative risk in each of the exposure groups. Analysis performed after 40 weeks. Note that the 95% CI (−218.7% to 113.7%) for the cumulative risk difference as well as the 95% CI for the risk in the counterfactual intervention group ranging from -141.9% to 185.4% in the instrumentedvariable analysis is flawed because the assumption regarding a strong correlationbetween the instrument and the exposure is violated. Also note that the upper limit of the CI 107.6% is unrealistic as it is unable to exceed 100%. Few injuries affect the time-to-event model making it less robust.
AIPP, Athletics Injury Prevention Programme; Ref, reference.