Post hoc ergo propter hoc. Arvind Veiraiah, Lalndough Hospital 22 November 2005 I believe that this study is fatally flawed due to the following problems:1) It merely chronicles subjective improvement in a group of chronically ill patients without using a comparator. While the authors accept that this study cannot rule out effects due to additional conventional medications, they ignore the fact that these patients might simply have improved due to the placebo effect of participating in such a study (lots of attention, health education, recognition of concurrent problems, placebo effect of the drugs themselves...). The results are meaningless in the absence of placebo control.2) The authors assume that the diseases are indeed chronically of the baseline severity and attribute improvement in symptoms and quality of life to treatment. But severity varies quite a bit during the course of a chronic illness, although recall bias ensures that patients always remember the worst experiences (and sometimes build upon them). Why did the patients present themselves to the homeopathic practitioner (or any practitioner) in the first instance? Probably because the disease was worse than usual. Experience dictates that symptoms will usually improve from their worst state - so why are these improvements not just the effect of time?3) The authors state that an assumption that the quality of life of these patients is similar to that of the general population is conservative. But why? Surely the quality of life at baseline for these chronically ill patients should be worse than that of the general population? Why could that not have been ascertained in the beginning of the study? Would not an assumption that the baseline quality of life should be low in this cohort mean a greater expected change due to regression to mean than that allowed for in this study?This important research question needs studies that are not just very complex and carefully monitored. The initial question has to be well thought out; and the effects of time and active monitoring have to be considered. Unfortunately this study does not fulfil these expectations. Competing interests None.