Summary estimates of sensitivity and specificity, and positive and negative likelihood ratios (with 95% CIs) were calculated using the Der Simonian and Laird random-effects model. Post-test probabilities were calculated based on pooled likelihood ratios and overall pretest probability of a diagnosis of adenomyosis. Area under the summary receiver operating characteristic (SROC) curve was estimated, but the associated SROC plot was not presented.
Possible sources of heterogeneity were investigated using meta-regression analysis. Explanatory variables were study design, number of participants, test description, QUADAS score, country of publication, language of publication, type of ultrasound transducer and whether the study included participants with symptoms consistent with adenomyosis. Subgroup analyses were performed for patients who had a hysterectomy for symptoms consistent with adenomyosis and for patients who had a hysterectomy for any reason.
Publication bias was assessed by visual inspection of funnel plots and using the Egger test.