Validation study of nonsurgical diagnosis of endometriosis

FERTILITY AND STERILITY VOL. 76, NO. 5, NOVEMBER 2001 Copyright 2001 American Society for Reproductive Medicine Published by Elsevier Science Inc. Printed on acid-free paper in U.S.A. Validation study of nonsurgical diagnosis of endometriosis Brenda Eskenazi, Ph.D., a Marcella Warner, Ph.D., a Luigi Bonsignore, M.D., b David Olive, M.D., a Steven Samuels, Ph.D., a,c and Paolo Vercellini, M.D. d Hospital of Desio, Desio, Italy, and University of California, Berkeley, California Received February 9, 2001; revised and accepted May 2, 2001. Supported by grant nos. R82471 from the US Environmental Protection Agency, R01 ES07171 and F06 TWO2075-01 from the National Institutes of Health, and EA-M1977 from the Endometriosis Association. Reprint requests: Brenda Eskenazi, Ph.D., School of Public Health, University of California at Berkeley, 140 Warren Hall, Berkeley, California 94720-7360 (FAX: 510-642-9083; E-mail: eskenazi@ uclink4.berkeley.edu). a School of Public Health, University of California. b Department of Gynecology, Hospital of Desio. c Division of Occupational/ Environmental Medicine & Epidemiology, University of California, Davis, California. d Department of Obstetrics and Gynecology, Mangiagalli Hospital, University of Milan, Milan, Italy. 0015-0282/01/$20.00 PII S0015-0282(01)02736-4 Objective: To determine whether the surgical diagnosis of endometriosis can be predicted using symptoms, signs, and ultrasound findings. Design: Prospective study (study sample); retrospective record review (test sample). Setting: Hospital of Desio (study sample) and Mangiagalli Hospital (test sample), Italy. Patient(s): Ninety women scheduled to undergo laparoscopy or laparotomy (study sample); 120 women who underwent laparoscopy (test sample). Intervention: The study sample group was interviewed before surgery about infertility and dysmenorrhea, dyspareunia, and noncyclic pelvic pain and each member had a pelvic examination and a transvaginal ultrasound. At surgery, endometriosis was noted. For the test sample, the same information was abstracted from medical records after laparoscopy. Main Outcome Measure(s): The ability of symptoms, signs, and ultrasound to predict endometriosis at surgery. A classification tree was developed with the study sample and evaluated with the test sample. Result(s): Ovarian endometriosis, but not nonovarian endometriosis, could be reliably predicted with noninvasive tools. Ultrasound and examination best predicted ovarian endometriosis, correctly classifying 100% of cases with no false positive diagnoses in the study sample. Similar results were found in the test sample. Conclusion(s): Noninvasive tools may be used to identify women with ovarian, but not nonovarian endometriosis, with excellent agreement with surgical diagnosis. (Fertil Steril 2001;76:929 35. 2001 by American Society for Reproductive Medicine.) Key Words: Endometriosis, endometriomas, epidemiology, laparoscopy, infertility, dyspareunia, dysmenorrhea, pelvic pain Endometriosis can only be definitively diagnosed by pelvic surgery. The lack of a noninvasive method of diagnosis has limited our knowledge about the epidemiology of endometriosis and has impacted the ability of the clinician to choose appropriate treatment or surgery. Because it is not possible or ethical to screen a population with surgery, studies of prevalence and of risk factors have been based only on clinically-diagnosed cases. Such studies may be biased because they miss a significant number of cases (1). Other than pelvic surgery, the methods currently available to diagnose endometriosis include measurements of serum proteins such as CA-125, pelvic ultrasound, and magnetic resonance imaging (MRI). The specificity of currently available biomarkers is suboptimal (2), and MRI, although potentially a valuable diagnostic tool, is expensive and requires a radiologist skilled in identifying endometriosis. Thus, most presurgical diagnoses rely on clinical judgment based on medical history, symptoms, and signs. The purpose of the current study is to determine whether the surgical diagnosis of endometriosis can be predicted using common noninvasive tools including medical history, symptom report, pelvic examination, and transvaginal ultrasound. We develop a predictive decision tree based on one sample of women who are about to undergo laparoscopy (study sample) and test the utility of this decision tree on a different sample of women who underwent laparoscopy (test sample). 929

MATERIALS AND METHODS Study Sample Women participating in the study (n 90) were scheduled to undergo laparoscopy or laparotomy for pelvic pain, infertility, tubal ligation, or masses of the adnexus or uterus at the Hospital of Desio, Department of Gynecology, between July 1998 and December 1999. We initially included women undergoing laparotomy; however, with the sharp increase in the number of laparoscopic surgeries in the initial phases of the study, we subsequently limited the study to those scheduled for laparoscopy. Indications for surgery remained the same. Women visiting the clinic for acute conditions such as ectopic pregnancy, for evaluation of endometrial or ovarian cancer, or for treatment of already diagnosed endometriosis were excluded from the study. This study was approved by the Institutional Review Boards of the participating institutions. The average age of the women was 35.5 years (SD, 7.2 years; range, 20 49). Thirty-one percent of the women were nulligravid, and 47% of the women were nulliparous. Twentysix percent had at least a high school education. Four women were currently using oral contraceptives. Data Collection At the preadmission appointment, women were invited to participate. They were informed that participation included an interview, pelvic examination, transvaginal ultrasound, and permission to access the current surgery medical record. The median time between the interview and the surgery was 7 days (range, 0 34 days). After an informed consent form was signed, a trained nurse-interviewer conducted a structured 1-hour interview. The interview asked about a woman s reproductive history including infertility, gynecologic diagnoses or treatments, and contraception. Women were asked to categorically rate the level of dysmenorrhea, noncyclic pelvic pain, and dyspareunia experienced in the past year. For dysmenorrhea, described as menstrual cramps, pain levels were defined as no pain, mild, some loss of work efficiency, moderate, in bed part of day, occasional loss of work, or severe, in bed 1 or more days, incapacitation. For noncyclic pelvic pain, described as persistent pain other than during menstruation, pain levels were defined as no pain, mild, occasional pelvic discomfort, moderate, noticeable discomfort for most of cycle, or severe, requires strong analgesics. For dyspareunia, defined as deep pain with intercourse, pain levels were defined as no pain, mild, tolerated discomfort, moderate, intercourse painful to the point of causing interruption, or severe, avoidance of intercourse because of pain. Pain in the past year was also assessed using a visual analog scale (3). For each type of pain, women were shown a 10-cm line and told the left end of the line represented no pain and the right end represented unbearable pain. Women were then asked to place a mark on the line indicating the degree of pain experienced in the past year. Positive current dysmenorrhea, pelvic pain, or dyspareunia was defined as either a verbal rating of moderate or severe pain or a visual analog scale rating of 5 cm. Positive current infertility was defined as trying to become pregnant for more than 12 months. Immediately after the interview, the nurse-interviewer abstracted information from the questionnaire and completed a form that summarized the medical, pregnancy, infertility, menstrual, infertility, and pain history. This form was then given to the gynecologist before the examination and ultrasound. A single gynecologist (L.B.) conducted all pelvic examinations and transvaginal ultrasounds and completed coded data forms for both of these procedures. The pelvic examination was defined as positive if there was notation of uterosacral ligament scarring, nodularity or pain; nodularity or pain at the Pouch of Douglas; vaginal/endometriotic lesions; painful or fixed adnexal masses; or fixed uterus or pain on movement of uterus. The transvaginal ultrasound examination was performed in the lithotomy position with the ATL Color Doppler HDI 3000 (Advanced Technology Laboratories, Bothwell, WA), 8-4 MHz transvaginal probe for imaging. The ultrasound was defined as positive if a cyst or mass with endometriosisspecific characteristics was noted, i.e., thick walls, regular margins, and homogeneous low echogenicity of fluid (4). During laparoscopy or laparotomy, the gynecologist surveyed the entire pelvis and upper abdomen. Biopsies were performed on all suspected lesions when possible, and specimens were fixed in formalin and prepared in a standard manner. All specimens were stained with hematoxylin and eosin and read by a pathologist experienced in histologic appearance of endometriosis. The slides were sent to a second pathologist if the visual diagnosis and histologic report differed and if the suspected lesion had been biopsied. However, if the suspected lesion had not been biopsied because of its location (such as on the bladder), then the visual diagnosis prevailed (n 1). All laparoscopic surgeries were videotaped, and the tapes were later reviewed by a second gynecologist (D.O.) who was blinded to diagnosis. The extent of endometriosis was graded according to the revised American Fertility Society classification (5). Statistical Analysis Statistical analyses were performed using STATA 6.0 (6). The positive predictive value (PPV) is the percent of women who were predicted to have disease and who were found to have the disease at surgery. Conversely, negative predictive value (NPV) is the percent of women predicted not to have disease and who were found not to have endometriosis at surgery. The kappa statistic measures agreement between the presurgical assignment and surgical diagnosis. 930 Eskenazi et al. Nonsurgical diagnosis of endometriosis Vol. 76, No. 5, November 2001

To find a rule to predict endometriosis disease status, we employed the Classification and Regression Trees (CART) program (7). We categorized women as having (1) no endometriosis; (2) nonovarian endometriosis; and (3) ovarian endometriosis. The predictor variables for CART were results of the ultrasound, physical examination, and report of current symptoms (pelvic pain, dyspareunia, dysmenorrhea, infertility), all coded as positive or negative. We accepted the CART default rules for node splitting (gini) and tree pruning (minimal standard error) and set the costs of misclassification of each outcome category to be equal. We report the percent correctly classified as estimated by 10-fold cross-validation. The tree developed on these data was tested on retrospective data from a different sample of women, the test sample. Test Sample The women in the test sample had undergone laparoscopy by a single gynecologist (P.V.) at Mangiagalli Hospital, University of Milan. These women had an average age of 31 years (SD, 5.1), and 65% had presented for infertility. We selected as endometriosis cases those women who had most recently undergone laparoscopy with a confirmed diagnosis of endometriosis. We selected 15 women from each AFS-r stage group (I IV), for a total of 60 cases. Control subjects were the last 60 women who had undergone laparoscopy, but who were diagnosed as not having endometriosis. The medical records were abstracted for results of ultrasound, pelvic examination, pain scales, infertility history, indication for laparoscopy, and laparoscopy findings. The same pain scales (categorical and visual analog) and methods of recording physical examination signs were used in the study and test samples. RESULTS Of the 90 participants in the study sample, 80% had laparoscopy; the remainder had laparotomy. The indications for surgery are shown in Table 1. Approximately 28% of the women had uterine masses and 30% had adnexal masses. For 21% of the cases, pelvic pain was an indication for surgery, and for 13%, infertility was an indication. Endometriosis was suspected in 14 (16%) women. Fifty of the 90 women (56%) reported at least one of the symptoms (see Table 2). Forty (44%) women reported dysmenorrhea; 20 (22%) reported pelvic pain, and 20 (22%) reported dyspareunia. Of the 90 women, 16 (18%) reported infertility. Forty-two (47%) of the 90 women had a positive examination. Specific signs included uterosacral pain (n 28), fixed adnexal mass (n 26), painful adnexal mass (n 25), pain in the Pouch of Douglas (n 17), nodularity in the Pouch of Douglas (n 7) or uterosacral ligaments (n 7), scarring in the uterosacral ligaments (n 4), pain on movement of uterus (n 2), and vaginal or endometriotic lesions TABLE 1 Indications for pelvic surgery, Desio, Italy, 1998 1999. Indication for surgery n % a Pelvic pain 19 21 Infertility 12 13 Ovarian cysts 27 30 Suspected endometriosis 14 16 Fibroids 25 28 Tubal ligation 6 6 Other b 8 9 a Percentages total to more than 100% because many women had more than one indication. b Bleeding (n 3), urinary incontinence (n 2), prophylactic bilateral ovariectomy (n 1), adenomyosis (n 1), pelvic mass (n 1). (n 1). Twenty-two (24%) of the women had a positive ultrasound. Based on the surgery and pathology reports, 37 women had endometriosis (86% histologically confirmed) and 53 women did not. Of the 37 women with endometriosis, 16 had nonovarian and 21 had ovarian endometriosis. Fourteen (38%) women had stage I II disease and 23 (52%) women had stage III IV disease. Twenty-one of the 22 women with a positive ultrasound were found to have ovarian endometriosis at surgery (20 had stage III or IV disease, one had stage I), and one woman was found not to have endometriosis (Table 2). As shown in Table 3, positive ultrasound correctly classified 81% of the women, with 95% PPV and 76% NPV. The kappa statistic indicated moderate agreement between prediction and surgical diagnosis (kappa 0.58). Twenty-eight of 42 women with a positive examination were found to have endometriosis (74% correctly classified, 67% PPV, 81% NPV); 21 (100%) with ovarian endometriosis and 7 (44%) with nonovarian endometriosis had a positive examination; 14 (26%) of the 53 women without endometriosis had signs on pelvic examination (see Table 2). As presented in Table 2, 28 of 50 women who reported having dysmenorrhea, pelvic pain, dyspareunia, or infertility were found to have endometriosis; 18 (86%) had ovarian endometriosis and 10 (63%) had nonovarian endometriosis. As presented in Table 3, reporting any symptoms correctly classified 66% of women, with 56% PPV and 78% NPV. The percent correctly classified was highest (68%) for dysmenorrhea, followed by pelvic pain (63%), dyspareunia (54%), and infertility (52%). Approximately twice the proportion of those with endometriosis versus those without endometriosis reported moderate to severe dysmenorrhea (65% vs. 30%) and noncyclic pelvic pain (32% vs. 15%), whereas an equal proportion of the two groups reported dyspareunia (22% vs. 23%) (see FERTILITY & STERILITY 931

TABLE 2 Number and proportion of women undergoing pelvic surgery who were found to have positive ultrasound, exam, and symptomatology before surgery by surgical diagnosis, Desio, Italy, 1998 1999. Total (%) n 90 Endometriosis diagnosis No (%) n 53 Yes (%) n 37 Nonovarian endometriosis (%) n 16 Ovarian endometriosis (%) n 21 Positive ultrasound 22 (24) 1 (2) 21 (57) 0 (0) 21 (100) Positive exam 42 (47) 14 (26) 28 (76) 7 (44) 21 (100) Any symptoms a 50 (56) 22 (42) 28 (76) 10 (63) 18 (86) Dysmenorrhea 40 (44) 16 (30) 24 (65) 9 (56) 15 (71) Pelvic pain 20 (22) 8 (15) 12 (32) 5 (31) 7 (33) Dyspareunia 20 (22) 12 (23) 8 (22) 1 (6) 7 (33) Infertility 16 (18) 11 (21) 5 (14) 1 (6) 4 (19) a Symptoms include dysmenorrhea, pelvic pain, dyspareunia, and infertility. Table 2). Almost all of the patients with endometriosis who reported dyspareunia had ovarian endometriosis. The level of dysmenorrhea as indicated by the number of centimeters on the visual analog scale was significantly correlated with severity of disease as indicated by the AFS-r score (r 0.32, P.05); however, there was no correlation between level of dyspareunia (r 0.06, P.5) or pelvic pain (r 0.02, P.8) and severity of disease. Figure 1 presents the results of the CART decision tree analysis using the data from the study sample and tested with data from the test sample. The CART analysis indicated that ovarian endometriosis, but not nonovarian endometriosis, could be reliably predicted from the noninvasive methods. Positive ultrasound and examination were the best predictors of ovarian endometriosis; all of the women with ovarian endometriosis were correctly classified, and none of the other women were classified into the ovarian endometriosis group. However, nonovarian endometriosis was poorly predicted, even after symptomatology was added to ultrasound and examination results; only six (38%) of the 16 diagnosed with nonovarian endometriosis at surgery were correctly classified. Note that the tree makes a single classification by a plurality rule. These findings were duplicated when the test sample data were applied to the CART model in Figure 1, which was created with the study data. In the test sample, 85% of women with ovarian endometriosis diagnosed at surgery were correctly classified. Again, there was poor prediction for nonovarian endometriosis; only four (12%) cases found at surgery were correctly classified by the classification tree. DISCUSSION We found that noninvasive procedures (history, pain reports, physical examination, ultrasound) have moderate success in predicting a surgical diagnosis of endometriosis. TABLE 3 Sensitivity, specificity, and predictive value of ultrasound, exam, and symptomatology before surgery in classifying women according to presence or absence of endometriosis at surgery, Desio, Italy, 1998 1999. Presurgical findings Sensitivity Specificity Positive predictive value Negative predictive value % Correctly classified Kappa statistic Positive ultrasound 0.57 0.98 0.95 0.76 81 0.58 Positive exam 0.76 0.74 0.67 0.81 74 0.48 Any symptoms a 0.76 0.58 0.56 0.78 66 0.32 Dysmenorrhea 0.65 0.70 0.60 0.74 68 0.36 Pelvic pain 0.32 0.85 0.60 0.64 63 0.19 Dyspareunia 0.65 0.70 0.40 0.59 54 0.01 Infertility 0.14 0.79 0.31 0.57 52 0.08 a Symptoms include dysmenorrhea, pelvic pain, dyspareunia, and infertility. 932 Eskenazi et al. Nonsurgical diagnosis of endometriosis Vol. 76, No. 5, November 2001

FIGURE 1 Decision tree for classifying women as negative for endometriosis, nonovarian endometriosis, and ovarian endometriosis based on noninvasive criteria including ultrasound, pelvic exam, and symptomatology, Desio, Italy, 1998 1999. When ovarian and nonovarian endometriosis are considered together, ultrasound diagnosis alone correctly classified 81% of the women, signs on examination correctly classified 74% of the women, and symptomatology including dysmenorrhea, pelvic pain, dyspareunia or infertility correctly classified 66% of the women. However, the decision tree analysis indicated that these results are driven by the reliable prediction of ovarian endometriosis only. All ovarian endometriosis cases were correctly classified by ultrasound and examination, but only 38% of nonovarian endometriosis diagnosed at surgery was accurately predicted by a combination of indicators. Similar results were found with the test sample data. Previous studies have reported that ultrasound can be used to evaluate pelvic masses and to differentiate between ovarian endometriosis and other ovarian masses, especially when using a vaginal probe (8). In a study by Mais et al. (9), transvaginal ultrasound had a sensitivity of 75% and a specificity of 99% in the screening for ovarian endometriosis, and a sensitivity of 84% and specificity of 90% in screening for other ovarian masses. The utility of ultrasound in distinguishing ovarian endometriosis from other ovarian masses FERTILITY & STERILITY 933

was confirmed by Kurjak and Kupesic (4). Our specificity was similar (98%) to Mais et al. (9), however, our sensitivity was considerably lower (57%), because of the inclusion in the current study of women who had no ovarian masses. A few studies have examined the predictive ability of the pelvic examination to identify women with endometriosis. In a study of women presenting for laparoscopy because of infertility or pelvic pain, the sensitivity, specificity, and PPV of focal pelvic tenderness to predict endometriosis was 79%, 32%, and 64%, respectively (10). Focal tenderness was most highly specific at the uterosacral ligaments and cul de sac pelvic zones, with 83% PPV when these locations were considered alone. To some extent, focal tenderness may depend on the location, depth, and appearance of the lesion (11). Koninckx et al. (12) reported that painful nodularities found at pelvic examination during menstruation were highly predictive of deep endometriosis, ovarian endometriosis, or severe cul de sac disease with a 79% sensitivity and 92% specificity. Although our pelvic examinations were not conducted during menses and included women presenting for laparoscopy for a variety of indications, our findings of 76% sensitivity and 74% specificity are in line with the above studies. Few previous studies have attempted to predict endometriosis at surgery from symptoms of pain or infertility. Instead, a number of investigations have been restricted to women who present with pain (13) or infertility (14, 15). Studies have reported that a wide range (14% 67%) of infertile women who undergo laparoscopy have endometriosis (14, 16, 17). A few studies have compared the frequency of dysmenorrhea in endometriosis cases and controls (18 20) and have reported conflicting results. For example, the frequency of dysmenorrhea was found to be similar in endometriosis cases and infertile controls (20) but more prevalent in endometriosis cases than other surgical controls (18). Cramer et al. (19) found that 62% of endometriosis case patients but only 36% of pregnant control subjects reported having moderate to severe dysmenorrhea. These findings are similar to ours (65% of case patients vs. 30% of women without endometriosis) and suggest that although women with and without endometriosis have a similar prevalence of dysmenorrhea, patients may be more likely to have more severe symptoms. Naish et al. (21) reported that severe dysmenorrhea was better than dyspareunia and pelvic pain in predicting endometriosis, with 63% PPV and 95% NPV. In fact, in a study of 99 subfertile women, Forman et al. (14) found that severe dysmenorrhea was the only symptom predictive for endometriosis. We also report a significant positive correlation between the severity of dysmenorrhea using the visual analog scale and the extent of disease (AFS-r score). This finding confirms previous reports (22) and corroborates a report of women with stage III or IV disease having more severe dysmenorrhea than controls (20). The association of dyspareunia and endometriosis has been studied in a few prior investigations. In contrast to our finding of an equal proportion of women with and without endometriosis who reported dyspareunia, Fedele et al. (20) observed that dyspareunia was more frequent in infertile endometriosis case patients compared with other infertile women. Our sample, however, included fertile as well as infertile women. We also report no association of severity of disease (AFS-r stage) and level of dyspareunia (visual analog scale). This finding is discordant with one report which found an inverse relationship (23), but similar to others (3, 17, 24, 25). We found that women who were diagnosed with endometriosis compared with those who were not were about twice as likely to report moderate to severe noncyclic pelvic pain (32% vs. 15%). This finding is similar to that of Fedele et al. (20); pelvic pain was more frequently reported in infertile endometriosis case patients compared with other infertile women. However, Williams and Pratt (18) reported in almost 1,000 women undergoing celiotomies that noncyclic pelvic pain was less common in those with endometriosis. We also found no association between extent of disease and severity of symptoms. This finding is similar to those of Fedele et al. (24), but somewhat dissimilar to their findings in another study restricted to fertile women in which they reported more frequent pelvic pain in women with stage III and IV disease (20). In this latter study, only ovarian endometriosis was associated with severe pelvic pain; whereas, we report no difference in the proportion reporting pelvic pain among those with ovarian and nonovarian endometriosis. The differences across studies in associations between disease and signs and symptoms may be related to study selection criteria and referral biases inherent to specific clinics. Also, early-stage endometriosis was not well-recognized before 1983. Yet it is certain that some women experience infertility, dysmenorrhea, dyspareunia, and pelvic pain due to endometriosis. In fact, in a recent double-blind clinical trial of depot leuprolide, the level of each of these pain types decreased in the treated group (26). Given the large proportion of women without endometriosis who also experience infertility and pain, their presence may not accurately predict endometriosis in populations presenting for pelvic surgery. Moreover, in population-based epidemiological studies, where diagnosis may need to be made on non clinic-based populations, the sensitivity of these markers is likely to be extremely low. Nevertheless, this study does suggest that with high-resolution ultrasound read by experienced clinicians in combination with signs noted on physical examination, ovarian endometriosis can be accurately diagnosed in clinic-based populations. It appears that we may have to wait for the advent of a biomarker of 934 Eskenazi et al. Nonsurgical diagnosis of endometriosis Vol. 76, No. 5, November 2001

endometriosis to accurately and cheaply diagnose nonovarian disease. Acknowledgments: The authors thank Drs. Paolo Moccarelli and Silvio Arienti for their assistance in the clinical protocols; Stefania Casalini, Giuseppina Bonacina, and Andrea Castelnuovo for their assistance in data collection; Wan-Ying Chee for her assistance in data analysis. References 1. Eskenazi B, Warner M. Epidemiology of endometriosis. Obstet Gynecol Clin North Am 1997;24:235 58. 2. Olive SL, Schwartz LB. Endometriosis. N Engl J Med 1993;328:1759 69. 3. Vercellini P, Bocciolone L, Vendola N, Colombo A, Rognoni MT, Fedele L. Peritoneal endometriosis: morphopologic appearance in women with chronic pelvic pain. J Reprod Med 1991;36:533 6. 4. Kurjak A, Kupesic S. Scoring system for prediction of ovarian endometriosis based in transvaginal color and pulsed Doppler sonography. Fertil Steril 1994;62:81 8. 5. Revised American Fertility Society classification of endometriosis: 1985. Fertil Steril 1985;43:351 2. 6. Stata C. Stata Statistical Software: Release 6.0. College Station, TX: Stata Press, 1999. 7. Steinberg D. Classification and regression trees. San Diego: Salford Systems, 1997. 8. Fleischer AC, Entman SS. Differential diagnosis of pelvic masses. In: Ultrasound in obstetrics and gynecology. Chervenak FA, Isaacson GC, Campbell S, eds. Boston: Little, Brown, 1993: 1643 53. 9. Mais V, Guerriero S, Ajossa S, Angiolucci M, Paoletti AM, Melis GB. The efficiency of transvaginal ultrasonagraphy in the diagnosis of endometrioma. Fertil Steril 1993;60:776 80. 10. Ripps BA, Martin DC. Correlation of focal pelvic tenderness with implant dimension and stage of endometriosis. J Reprod Med 1992;37: 620 4. 11. Martin DC, Ling FW. Endometriosis and pain. Clin Obstet Gynecol 1999;42:664 86. 12. Koninckx PR, Meuleman C, Oosterlynck D, Cornillie FJ. Diagnosis of deep endometriosis by clinical examination during menstruation and plasma CA-125 concentration. Fertil Steril 1996;65:280 7. 13. Vercellini P, Fedele L, Molteni P, Arcaini L, Bianchi S, Candiani GB. Laparoscopy in the diagnosis of gynecologic chronic pelvic pain. Int J Gynecol Obstet 1990;32:261 5. 14. Forman RG, Robinson JN, Mehta Z, Barlow DH. Patient history as a simple predictor of pelvic pathology in subfertile women. Hum Reprod 1993;8:53 5. 15. Thornton JG, Morley S, Lilleyman J, Onwude JL, Currie I, Crompton AC. The relationship between laparoscopic disease, pelvic pain and infertility: an unbiased assessment. Eur J Obstet Gynecol Reprod Biol 1997;74:57 62. 16. Balasch J, Creus M, Fábregues F, Carmona F, Ordi J, Martinez-Román S, et al. Visible and non-visible endometriosis at laparoscopy in fertile and infertile women and in patients with chronic pelvic pain: a prospective study. Hum Reprod 1996;11:387 91. 17. Buttram VC, Jr. Conservative surgery for endometriosis in the infertile female: a study of 206 patients with implications for both medical and surgical therapy. Fertil Steril 1979;31:117 23. 18. Williams TJ, Pratt JH. Endometriosis in 1,000 consecutive celiotomies: incidence and management. Am J Obstet Gynecol 1977;129:245 50. 19. Cramer DW, Wilson E, Stillman RJ, Berger MJ, Belisle S, Schiff I, et al. The relation of endometriosis to menstrual characteristics, smoking, and exercise. JAMA 1986;255:1904 8. 20. Fedele L, Bianchi S, Bocciolone L, Di Nola G, Parazzini F. Pain symptoms associated with endometriosis. Obstet Gynecol 1992;79: 767 9. 21. Naish CE, Kennedy BH, Barlow DH. Correlation between pain symptoms and laparoscopic findings. Brosens IA, Donnez J, eds. In: Program and abstracts of the Third World Congress on Endometriosis 1993 June 1 3, Brussels, 1993. 22. Muzii L, Catalano GF, Marana R. Endometriosis externa and interna: endoscopic diagnosis. Rays 1998;23:683 92. 23. Vercellini P, Tresidi L, Giorgi OD, Cortesi I, Parazzini F, Crosignani PG. Endometriosis and pelvic pain: relation to disease stage and localization. Fertil Steril 1996;65:299 304. 24. Fedele L, Parazzini F, Bianchi S, Arcaini L, Candiani GB. Stage and localization of pelvic endometriosis and pain. Fertil Steril 1990;53: 155 8. 25. Marana R, Muzli L, Caruana S, Dell Acqua S, Mancuso S. Evaluation of the correlation between endometriosis extent, age of the patients and associated symptomatology. Acta Eur Fertil 1991;22:209 12. 26. Ling FW. Randomized controlled trial of depot leuprolide in patients with chronic pelvic pain and clinically suspected endometriosis. Pelvic Pain Study Group. Obstet Gynecol 1999;93:51 8. FERTILITY & STERILITY 935