Most of the trials asked would not hand over their patient records
A team wanting to test glucosamine on patient records rather than published averages asked twenty-one trialists. Six shared, five were usable, and all five were independently funded.
What the study found
Twenty-one trials were asked for their raw data; six handed it over
A team wanting to test glucosamine on patient records rather than published averages went to the original trialists and asked. Six of twenty-one shared. Five were usable, and every one had been funded independently of the industry. The authors report the refusals as a finding of their own.
The caveat · Missing trials, not missing patients
The five contributed trials still cover 55% of everyone ever randomised in a published placebo-controlled glucosamine trial, so the gap is in which trials are represented, not in people. Because no manufacturer-sponsored dataset arrived, the reanalysis cannot answer those trials on their own data.