Table 3.
Description of PRO statistical inferential techniques captured, with associated trials separated by the treatment under investigation.
| Statistical inferential technique | Description | Drug trial (n = 8) | Radiotherapy trial (n = 12) |
|---|---|---|---|
| Wilcoxon signed-rank test | Change in PRO between baseline and another timepoint. | 30 | 28,50,56 |
| Wilcoxon rank-sum test | Association between PRO and dosimetric parameters at timepoints. | 28 | |
| Mixed-effect model | Fit model predicting PRO score across study from baseline accounting for intra-patient correlation. | 40 | 8,51 |
| Compare change in PRO score across each dose cohort when time is fitted as a random effect | 43 | ||
| Linear Regression | Fit model predicting average PRO score for specific symptoms across study from baseline. | 54 | |
| t-tests | Change in PRO score between baseline and another timepoint or difference in PRO score between dose cohorts. | 25,34,35,58 | 29,38,41,46,48 |
| Fisher's exact test | Association between MCID and each dose cohort | 45 | |
| Survival Analysis | Fit model predicting time from inclusion in study until deterioration or grade 3/4 toxicity using a Cox proportional hazards model. | 33 |
PRO: patient-reported outcome; MCID: minimal clinically important difference; HRQoL: health related quality of life.