Characteristics of trials accounting for postprogression therapy (PPT) bias on overall survival estimates
| Characteristic | PPT analysis | |||||
| All trials, N | Yes, N (%) | No, N (%) | OR | 95% CI | P value | |
| Total trials | 334 | 41 (12) | 293 (88) | |||
| Disease stage | ||||||
| Solid M0 | 65 | 1 (2) | 64 (98) | 0.10 | 0.01 to 0.49 | 0.03 |
| Solid M1 | 218 | 29 (13) | 189 (87) | Ref | ||
| Haematologic | 51 | 11 (22) | 40 (78) | 1.79 | 0.80 to 3.80 | 0.14 |
| Disease site | ||||||
| Breast | 50 | 8 (16) | 42 (84) | 1.14 | 0.39 to 3.37 | 0.81 |
| Gastrointestinal | 62 | 5 (8) | 57 (92) | 0.53 | 0.15 to 1.68 | 0.29 |
| Genitourinary | 43 | 5 (12) | 38 (88) | 0.79 | 0.22 to 2.56 | 0.70 |
| Haematologic | 51 | 11 (22) | 40 (78) | 1.65 | 0.61 to 4.64 | 0.33 |
| Thoracic | 72 | 4 (6) | 68 (94) | 0.35 | 0.09 to 1.19 | 0.10 |
| Other* | 56 | 8 (14) | 49 (86) | Ref | ||
| Treatment type | ||||||
| Systemic therapy | 322 | 41 (13) | 281 (87) | |||
| Local therapy | 12 | 0 (0) | 12 (100) | Ref | ||
| Cooperative group study | ||||||
| Yes | 86 | 6 (7) | 80 (93) | 0.46 | 0.17 to 1.05 | 0.09 |
| No | 248 | 35 (14) | 213 (86) | Ref | ||
| Industry sponsored | ||||||
| Yes | 284 | 38 (13) | 246 (87) | 2.42 | 0.83 to 10.30 | 0.15 |
| No | 50 | 3 (6) | 47 (94) | |||
| Enrolled patients† | NA | NA | NA | 0.999 | 0.998 to 1.00 | 0.04 |
| Publication date‡ | NA | NA | NA | 0.996 | 0.90 to 1.11 | 0.95 |
| Cross-over (allowed or required) | ||||||
| Yes | 49 | 16 (33) | 33 (67) | 5.04 | 2.42 to 1.38 | <0.0001 |
| No | 285 | 25 (9) | 260 (91) | Ref | ||
| Blinding | ||||||
| Double-blind | 128 | 15 (12) | 113 (88) | 0.92 | 0.46 to 1.79 | 0.81 |
| Open-label | 206 | 26 (13) | 180 (87) | Ref | ||
| Overall survival primary or coprimary endpoint | ||||||
| Yes | 168 | 18 (11) | 150 (89) | 0.75 | 0.38 to 1.44 | 0.38 |
| No | 166 | 23 (14) | 143 (86) | Ref | ||
| Surrogate survival and overall survival discordance | ||||||
| Yes | 107 | 20 (19) | 87 (81) | 2.26 | 1.16 to 4.38 | 0.02 |
| No | 227 | 21 (9) | 206 (91) | Ref | ||
Associations are evaluated by unadjusted binary logistic regression.
*Other disease sites included central nervous system, endocrine, gynaecological, head and neck, paediatric, sarcoma, and skin.
†The median trial enrolment was 585 patients (IQR 362–896).
‡Trends over time are reported in online supplemental figure 2.
§No local therapy trial accounted for PPT, so an odds ratio was not computed.