The duller report.
The Study: If you’re new to QuadShot, you should know we loathe population database studies that try to provocatively compare disease outcomes for different types of cancer treatment. That’s why we love this study, which we’d be willing to bet has something to do with this. In it, survival outcomes of population-based observational studies comparing cancer treatments (e.g., using SEER and NCDB) were weighed against actual randomized trials comparing those same exact treatments. Not surprisingly, there was a ratio of nearly three comparative population studies for every one randomized trial. The hazard ratios for survival had essentially no correlation between study types. In fact, only 40% of population studies agreed with the treatment effect demonstrated in randomized trials. Perhaps there is some way of predicting the reliability of a population-based treatment comparison study? Nope. No measure of study quality was significantly associated with randomized trial agreement.
Bottom Line: Differences in survival outcomes reported by population-based observational studies appear to have virtually no correlation with those reported by randomized trials. | Soni, J Clin Oncol 2019