The diagnostic accuracy of outpatient miniature hysteroscopy in predicting premalignant and malignant endometrial lesions

Objective To determine the accuracy of outpatient miniature hysteroscopy in the diagnosis of premalignant and malignant endometrial lesions. Design A prospective cohort study (1996-1997). Setting Minimal access surgical training centre in a large teaching hospital. Methods Outpatient hysteroscopy and endometrial sampling was carried out on 248 patients with abnormal uterine bleeding. Any patients unable to undergo these outpatient procedures were admitted for a formal inpatient hysteroscopy and curettage. Test performance characteristics were computed for hysteroscopy, comparing its findings with the histological diagnosis which served as a 'gold standard'. Main outcome measures The estimate of the accuracy of the hysteroscopic findings was based, for binary results, on sensitivity specificity and predictive values. For multilevel results, the diagnostic accuracy was computed using likelihood ratios (LRs). With a normal hysteroscopy result, an LR of <1 indicated a decreased probability that a premalignant/malignant lesion was present, whereas with an abnormal hysteroscopy result, an LR of >1 indicated an increased probability that such a lesion was present. Results Hysteroscopy revealed features of normal endometrium in 228 women, of whom 12 were found histologically to have premalignant hyperplasia/cancer. Hysteroscopic features of premalignant/malignant endometrial lesions were suspected in 20 patients but the diagnosis was confirmed histologically in six cases only This amounted to a sensitivity of 33.3% (95% confidence intervals (CI) 14.4-58.8), a specificity of 93.9% (95% CI 89.8-96.5), a positive predictive value of 30.3%. (95% CI 12.8-54.3), and a negative predictive value of 94.7%. (95% CI 90,8-97.1). Analysis using likelihood ratios revealed that for a normal hysteroscopy the LR was 0.7 (95% CI 0.5-0.9). The LR was 1.9 (95% CI 0.5-6.6) when there was intermediate abnormality (increased endometrial thickness, abnormal vascularization, polypoid formations, mamillations and cerebroid irregularities). The LR was 51.1 (95% CI 7.9-326.9) when there was definite hysteroscopic abnormality (intermediate abnormality features associated with irregular polylobular friable excrescences with necrosis or bleeding). Conclusion The probability of the presence of a premalignant/malignant lesion is not particularly altered by the finding of normal hysteroscopic appearances or of intermediately abnormal features. However; if the hysteroscopic findings are definitely abnormal, the probability of there being a premalignant/malignant lesion is substantially increased. Normal findings at hysteroscopy are not conclusive and do not eliminate the need for endometrial sampling as they are not a substitute for benign findings on histological investigation.