From conference notes:
Dr. Kay Dickersin, Hopkins, Baltimore: to minimize selection bias, analyze results including dropouts added to the groups to which they were assigned. Otherwise it becomes an observational study. 2/3 of studies end up studying more than 1 primary outcome by publication time, according to reports from Denmark and Canada.
In response to my query about the Bekelman et al 2003 JAMA study saying that industry sponsorship increases the odds of a drug trial’s success 3.6-fold, Dr. Dickersin said industry-sponsored studies are very good EXCEPT in publishing negative results and EXCEPT in providing adequate dosage etc. of the control interventions against which their proposed treatments are compared.
Www.cochrane.us helps people decide about the effectiveness of interventions.
Andrew Branca, Cambridge Health Assoc. said: Big pharma spends a great deal more on development than on research. Prozac is thought of as very safe, but has the shortcoming that works in only about half of patients.
Dr. Branca gave this example of insufficient knowledge about a well-known pharmaceutical: 1 in 500 men over 50 die as the result of taking 1/day aspirin. It is unknown whether this is more or less than the incremental number that would die of stroke or heart attack without that daily aspirin.