Table 4.
Estimated treatment effects for the randomised trial of shoulder pain.
| Analysis | Estimate | SE | 95% CI | p-value | |
|---|---|---|---|---|---|
| Continuous outcomes | |||||
| Total shoulder flexion | |||||
| Unadjusted | 6.87, | 3.88 | (-0.80, | 14.54) | 0.079 |
| Covariate adjustment (linear regression) | 7.62, | 3.37 | (0.95, | 14.29) | 0.026 |
| IPTW (uncorrected SE) | 7.62, | 3.88 | (0.02, | 15.22) | 0.049 |
| IPTW (corrected SE) | 7.62, | 3.34 | (1.07, | 14.17) | 0.023 |
| SPADI | |||||
| Unadjusted | -2.05, | 3.55 | (-9.08, | 4.98) | 0.565 |
| Covariate adjustment (linear regression) | -1.91, | 3.16 | (-8.16, | 4.35) | 0.547 |
| IPTW (uncorrected SE) | -1.91, | 3.56 | (-8.88, | 5.06) | 0.592 |
| IPTW (corrected SE) | -1.91, | 3.14 | (-8.06, | 4.24) | 0.544 |
| Binary outcome | |||||
| Improvement (logarithm of risk ratio) | |||||
| Unadjusted | 0.30, | 0.139 | (0.03, | 0.57) | 0.031 |
| Covariate adjustment (binomial regression) | Convergence not achieved | ||||
| IPTW (uncorrected SE) | 0.30, | 0.139 | (0.02, | 0.57) | 0.033 |
| IPTW (corrected SE) | 0.30, | 0.137 | (0.03, | 0.57) | 0.030 |
| Improvement (risk difference) | |||||
| Unadjusted | 0.19, | 0.085 | (0.03, | 0.36) | 0.024 |
| Covariate adjustment (binomial regression) | 0.20, | 0.080 | (0.04, | 0.35) | 0.015 |
| IPTW (uncorrected SE) | 0.19, | 0.085 | (0.02, | 0.36) | 0.026 |
| IPTW (corrected SE) | 0.19, | 0.084 | (0.03, | 0.35) | 0.023 |
| Improvement (logarithm of odds ratio) | |||||
| Unadjusted | 0.86, | 0.389 | (0.09, | 1.62) | 0.028 |
| Covariate adjustment (binomial regression) | 0.89, | 0.400 | (0.11, | 1.66) | 0.026 |
| IPTW (uncorrected SE) | 0.85, | 0.391 | (0.08, | 1.61) | 0.030 |
| IPTW (corrected SE) | 0.85, | 0.383 | (0.10, | 1.60) | 0.027 |
For continuous outcomes, adjusted estimates are adjusted for the baseline value of the outcome measure
For binary outcomes adjusted estimates are adjusted for baseline SPADI.
IPTW = inverse probability-of-treatment weighting; SPADI = Shoulder and Pain Disability Index.