Can we rely on blind endometrial biopsy for detection of focal intrauterine pathology?

OBJECTIVE: To compare the diagnostic power of random endometrial biopsy with hysteroscopy for intrauterine lesions. STUDY DESIGN: A retrospective cohort study of 639 women evaluated by diagnostic office hysteroscopy and endometrial biopsy (Novak curette) was carried out between 10/1997-6/2000. Reasons for evaluation were postmenopausal bleeding, abnormal uterine bleeding, ultrasound or hystero-salpingography findings, intrauterine device removal, suspected retained products of conception, infertility, late abortions and recurrent abortions. RESULTS: The women's mean age was 43.4 +/- 13.3 years (range, 1888). The most prevalent indication for investigation was abnormal uterine bleeding (n = 218, 34.1%), followed by sonographic or hysterosalpingographic findings (n = 167, 26.1%). Hysteroscopy revealed a normal uterine cavity in 367 (57.4%) women. Endometrial polyps and submucosal fibroids were the most common hysteroscopic findings (in 151 [23.6%] and 72 [11.3%], respectively). The hysteroscopic findings were compared with the pathology results in 558 cases. The sensitivity of the Novak curette for detection of endometrial polyps and submucosal fibroids was only 8.4% and 1.4%, respectively. The positive predictive value (30.9%) and the negative predictive value (57.9%) for both lesions were likewise low. On the other hand, hysteroscopy was not effective in diagnosing the 27 cases of hyperplasia (26 simple and one complex) all without atypia. CONCLUSION: Random endometrial sampling alone is not effective for diagnosing focal lesions of the uterine cavity and should be combined with other modalities, preferably diagnostic hysteroscopy.