Figure 3 Weighted random effects meta‐regression analyses: effect sizes (ES) of the individual studies according to pain (A,B) and self‐reported disability (C,D), following different rates of weight reductions (A,C) at different intensities (B,D); a decrease in ES would indicate clinical improvement. The area of each circle is inversely proportional to the random effects variance²⁰ of the standardised mean difference; the fitted random‐effects regression line is shown with 95% prediction intervals—indicating the degree of variance predicted by the weight loss magnitude and intensity.