GRADE evidence table: quality of evidence assessment (confidence in effect estimates) per end point
| Quality assessment | Patients (n) | Effect | Quality | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| No of studies | Design | Risk of bias | Inconsistency | Indirectness | Imprecision | Other considerations | GnRH antagonists | Standard androgen suppression therapy | Relative (95% CI) | Absolute | |
| Overall mortality (follow-up 84–364 days) | |||||||||||
| 9 | Randomised trials* | Serious† | No serious inconsistency | No serious indirectness | Serious‡ | See comment§ | 35/1923 (1.8%) | 16/1097 (1.5%) | RR 1.35 (0.63 to 2.93) | 5 more per 1000 (from 6 fewer to 30 more) | ⊕⊕OO LOW |
| Treatment failure (follow-up 84–364 days) | |||||||||||
| 7 | Randomised trials¶ | Serious** | No serious inconsistency | No serious indirectness | Serious‡ | None | 146/1450 (10.1%) | 81/750 (10.8%) | RR 0.92 (0.64 to 1.33) | 9 fewer per 1000 (from 39 fewer to 36 more) | ⊕⊕OO LOW |
| PSA progression (follow-up 84–364 days) | |||||||||||
| 7 | Randomised trials†† | Serious‡‡ | No serious inconsistency | No serious indirectness | Serious‡ | None | 115/1566 (7.3%) | 75/923 (8.1%) | RR 0.83 (0.64 to 1.06) | 14 fewer per 1000 (from 29 fewer to 5 more) | ⊕⊕OO LOW |
| Quality of life related to IPSS, follow-up 84 days; better indicated by lower values | |||||||||||
| 3 | Randomised trials§§ | Serious¶¶ | No serious inconsistency | Serious*** | No serious imprecision | None | 286 | 173 | – | MD 1.84 lower (3 to 0.69 lower) | ⊕⊕OO LOW |
| Quality of life related to urinary symptoms (follow-up 84 days; better indicated by lower values) | |||||||||||
| 3 | Randomised trials§§ | Serious¶¶ | No serious inconsistency | Serious*** | No serious imprecision | None | 288 | 173 | – | MD 0.4 lower (0.94 lower to 0.14 higher) | ⊕⊕OO LOW |
*The following studies were included: 149-98-02, 149-98-03, ABACS1, CS21, CS28, CS30, CS31, CS35, CS37.
†Downgraded for study limitations (−1): High or unclear risk of bias in included studies (for details see ‘risk of bias’ tables). Despite the methodological limitations, we do not feel that results are likely to be influenced by lack of blinding. However, there was insufficient reporting of attrition and exclusions to permit judgment on incomplete outcome data in studies CS28, CS31, CS35, CS37, and ABACS1. Studies CS35 and CS37 were reported as conference abstracts or data presentation within combined data analyses. Study ABACS1 was reported as conference abstract or the trial information was published within narrative reviews or FDA safety data publications. Studies CS35 and CS37 were terminated early. Studies CS35 and CS37 reported patient baseline characteristics incompletely.
‡Downgraded for imprecision (−1): Imprecision due to low number of events and wide CIs.
§Information on mortality was not provided by a single study as time-to-event data. Therefore, we could not, as initially planned, analyse these data with HRs, but have to report number of deaths during study duration. After screening the available entries of the study protocols in the registries, mortality was not predefined as primary/secondary outcome in any of the included studies, but was only assessed as an adverse event outcome.
¶The following studies were included: 149-98-02, 149-98-03, 149-99-03, CS21, CS28, CS30, CS31.
**Downgraded for study limitations (−1): High or unclear risk of bias in included studies (for details see ‘risk of bias’ tables). Study 149-99-03 was reported as conference abstract only. There was insufficient reporting of attrition and exclusions to permit judgment on incomplete outcome data in studies CS28, CS31, and 149-99-03. Study CS28 was terminated early.
††The following studies were included: CS21, CS28, CS30, CS31, CS35, CS37, ABACS1.
‡‡Downgraded for study limitations (−1): High or unclear risk of bias in included studies (for details see ‘risk of bias’ tables). Despite the methodological limitations, we do not feel that results are likely to be influenced by lack of blinding. However, there was insufficient reporting of attrition and exclusions to permit judgment on incomplete outcome data in studies CS28, CS31, CS35, CS37, and ABACS1. Studies CS35 and CS37 were reported as conference abstracts or data presentation within combined data analyses only. Study ABACS1 was reported as conference abstract or the trial information was published within narrative reviews or FDA safety data publications. Studies CS35 and CS37 were terminated early. Studies CS35 and CS37 reported patient baseline characteristics incompletely.
§§The following studies were included: CS28, CS30, CS31.
¶¶Downgraded for study limitations (−1): High or unclear risk of bias in included studies (for details see ‘risk of bias’ tables). There was insufficient reporting of attrition and exclusions to permit judgment on incomplete outcome data in studies CS28 and CS31. Studies CS35 and CS37 were identified to measure quality of life outcomes. However, we found no publications of these studies that reported this outcome.
***Downgraded for indirectness (−1): The question addressed by this systematic review was different from the results presented in the available evidence. We expected a measurement of quality of life related to general health but found only an evaluation of quality of life related to urinary symptoms or IPSS.
FDA, Food and Drug Administration; GnRH, gonadotropin-releasing hormone; IPSS, International Prostate Symptom Score; MD, mean difference.