Radical prostatectomy v radiation: only a randomised trial can provide the answerBMJ 2014; 348 doi: https://doi.org/10.1136/bmj.g2266 (Published 25 March 2014) Cite this as: BMJ 2014;348:g2266
- Mack Roach 3rd, professor of radiation oncology and urology and chair of the Department of Radiation Oncology1
Sooriakumaran and colleagues’ study is one of several that conclude survival is better after prostatectomy than radiotherapy.1 2 These studies have many features in common. They assume a cause and effect association between survival and treatment. They claim that they corrected for confounders. They acknowledge suboptimal radiotherapy (dose and use of androgen deprivation therapy).3 They acknowledge limitations, including potential unaccounted for differences between groups. Yet they still imply that their findings are fact. Some studies ignore prostate specific antigen data and level I evidence on the use of androgen deprivation therapy and radiotherapy.3 This brings to mind the saying: “You can fool all the people some of the time, and some of the people all of the time, but you can’t fool all the people all the time” (Abraham Lincoln).
“Propensity scores” can’t correct for large effects of unknown factors when populations are fundamentally different. Such adjustments can make things worse by allowing us to fool ourselves. If it takes 34 000 patients to show “statistically significant” differences, they may not be clinically significant. When baseline variables show large differences, there are probably other uncorrected differences, such as differences in the percentage of positive biopsies or Gleason 4+3 versus 3+4 tumours. Patients chosen for surgery are very different, as shown by their baseline characteristics. The profound impact of these differences is well documented.4 5 Men treated with radiotherapy have better survival than men without cancer and a lower risk of all causes of death. I wish that researchers would stop trying to prove what only a phase III trial—which adjusts for unknown variables by randomisation—can prove.2
Cite this as: BMJ 2014;348:g2266
Competing interests: None declared.
Full response at: www.bmj.com/content/348/bmj.g1502/rr/689904.