Repeated measurements

Repeated measurements refer to measurements made at different points in time. Thus, a dietary trial may report weight loss at 4 weeks, at 8 weeks and at 6 months. We can't include all of these in the same meta-analysis since again we'd be counting the same person more than once and we'd have a unit-of-analysis error. The problem of repeated measurements can partly be overcome by specifying in the protocol which time-points are of interest, and discarding the rest. It may be helpful to classify outcomes as 'short-term', 'medium-term' and 'long-term', and to perform separate meta-analyses for these different outcomes including only one time-point from each trial in each analysis. Alternatively, you may only be interested in a single time-point, or the longest available follow-up. The disadvantages of opting for longest available follow-up are that more patients may have been lost to follow-up, and it may vary considerably between studies introducing heterogeneity.