Results from the sensitivity analyses.

Background
This paper presents results from one of the few comparative effectiveness evaluations of novel antiandrogen medications (NHT) against standard of care (SoC) for patients suffering from metastatic castrate-resistant prostate cancer (mCRPC).
Methods
The design and the analysis are published in a protocol before accessing outcome data. Two groups of patients are balanced on hundreds of important covariates measured before the prostate cancer diagnosis and up to the date of the prescription. While the design yields balance on the observed covariates, one cannot discard the possibility that unobserved confounders are not balanced. The unconfoundedness assumption is assessed by estimating placebo regressions on two health measures, not included in the design but added together with the outcome data after protocol publication.
Results
We find a substantial (64 percent) increase in mortality for patients prescribed with NHT rather than SoC. However, based on the results from one of the two placebo regressions, we cannot rule out that the difference in mortality may be due to confounding. Using a bounding strategy of the effect, we can, however, rule out that NHT reduces mortality compared to SoC. Under an empirical valid assumption that most mCRPC patients who die suffer from bone metastases, we have a strong indication of increased skeleton-related events in patients if prescribed NHT against SoC.
Conclusions
Generally, the SoC for this group of patients is docetaxel. Given the substantially higher costs of many of the NHT, the finding of no positive effects from NHT on both mortality and SRE is important. More comparative studies, including studies analysing quality of life outcomes, are thus needed.