Background: Hysterosalpingography (HSG) is a commonly used step in the diagnostic work-up of subfertile couples for the diagnosis of tubal pathology, whereas a definite diagnosis by diagnostic laparoscopy is deferred for six to twelve months in case of a normal HSG. Test accuracy of HSG, i.e. sensitivity and specificity, is considered stable and independent of other patient characteristics, but this assumption has never been tested. In this study, we address whether test accuracy of HSG is dependent on patient characteristics using individual patient data meta-analysis. Material and Methods: We approached authors of primary studies on the accuracy of HSG for their original data. In the data obtained, we stratified our analysis of sensitivity and specificity of HSG for female age ( < 32.5 or > =32.5 years), duration of subfertility ( < 1.5 or > =1.5 years), type of subfertility, history of pelvic inflammatory disease (PID), Body Mass Index (BMI) ( < 25 or > 25 kg/m2) and Chlamydia antibody test (CAT). Sensitivity and specificity of HSG will be estimated with the bivariate nonlinear regression model, and differences between subgroups tested by including these as covariates. Results: We collected data from 20 studies with a total of 4,803 women, of which data from 2,186 women could be analysed in this study. In patients suffering from subfertility ≥�1.5 years, the sensitivity of HSG was 0.13 versus 0.63 for patients subfertile < 1.5 years (p=0.004), whereas specificity did not significantly change. In patients with a negative CAT sensitivity was
0.50 versus 0.77 in patients with a positive CAT (p=0.006), whereas specificity remained comparable. Female age, primary or secondary subfertility, BMI and history of PID had no significant impact on the accuracy of HSG. Conclusion: The sensitivity of hysterosalpingogram is associated with duration of subfertility and the result of the Chlamydia antibody test.