Giving it two years did not change the answer
662 participants stayed on their assigned capsules for a further 18 months. At two years none of glucosamine, chondroitin, the pair or celecoxib had moved knee pain meaningfully.
What the study found
Two years on, no treatment reached a pain difference that would matter
Waiting longer is the usual defence of a slow-acting treatment, so 662 participants stayed on their assigned capsules for a further eighteen months. At two years, none of glucosamine, chondroitin, the pair or celecoxib had produced a clinically important change in knee pain or function.
The caveat · A continuation, not a new trial
This is a continuation in the participants who met stricter X-ray criteria, not a fresh randomisation, so it carries the parent trial's high placebo response with it. Two years of dropout leaves less power, not more. The authors call the trends beneficial but not significant, which describes a null.