Clinical predictive factors for endometriosis in a Portuguese infertile population

Transcription

1 Human Reproduction Page 1 of 6 Hum. Reprod. Advance Access published June 30, 2004 DOI: /humrep/deh374 Clinical predictive factors for endometriosis in a Portuguese infertile population C.Calhaz-Jorge 1,3, Ben W.Mol 2, J.Nunes 1 and A.P.Costa 1 1 Human Reproduction Unit, Department of Obstetrics and Gynaecology, Hospital de Santa Maria, Av. Prof. Egas Moniz, Lisboa, Portugal and 2 Department of Obstetrics and Gynaecology, Academic Medical Centre, Amsterdam, The Netherlands 3 To whom Correspondence should be addressed. calhazjorge@mail.telepac.pt BACKGROUND: Endometriosis is an important clinical situation associated with subfertility. It would be very useful to identify patients at increased risk for endometriosis prior to laparoscopy. In the present study, we evaluate the demographic and clinical characteristics in a cohort of Portuguese subfertile women in relation to the presence of endometriosis. METHODS: Consecutive subfertile patients scheduled for laparoscopy were interviewed prior to the procedure. At subsequent laparoscopy, the presence of endometriosis was scored according to the revised classification of the American Society for Reproductive Medicine (ASRM). Data available from the medical history were tabulated against the presence or absence of endometriosis. We used logistic regression analysis to evaluate whether data from the patient s medical history could predict the presence of endometriosis. RESULTS: Among the 1079 women that were studied, 358 had minimal/mild endometriosis and 130 had moderate/severe endometriosis. Primary subfertility, regularity of menstrual cycles, dysmenorrhoea, chronic pelvic pain, obesity, ever use of oral contraceptives and smoking were the most important predictors of endometriosis. The prediction model had an area under the receiver operating characteristic curve of CONCLUSIONS: Both the presence of endometriosis (all stages) and the presence of severe endometriosis per se can be predicted from the medical history. These data should be used in the decision to perform laparoscopy at an early stage or a later stage in the work-up for subfertility. Key words: endometriosis/epidemiology/infertility/predictive factors Introduction Endometriosis is characterized by the presence of tissue with epithelial and stromal characteristics of endometrium outside the uterus. It has been estimated to affect % of all women of childbearing age (Houston et al., 1987), but its reported frequency differs widely depending on the type of population studied (Mahmood and Templeton, 1991). Although the pathogenesis of endometriosis remains not well understood, most observations support the retrograde menstruation theory that was formulated originally almost a century ago (Sampson, 1927). However, because retrograde menstruation is a far more common phenomenon than the existence of endometriosis, some other characteristics are assumed to be necessary in the subset of women who develop the disorder. A genetic basis for endometriosis (Kennedy, 1999) is a possible explanation that has gained general acceptance. The contribution of environmental factors for the establishment of endometriosis has also been suggested (Koninckx et al., 1994; Pauwels et al., 2001; Cobellis et al., 2003). The relationship between endometriosis and some socio-demographic, reproductive, menstrual and lifestyle characteristics has been the subject of several previous studies. In general, they suggest that circumstances leading to an increased amount of menstrual flow, such as longer duration of bleeding (Cramer et al., 1986), intra-uterine device (IUD) use (Kirshon and Pointdexter, 1988), or a lifelong regular menstrual pattern of shorter cycles and heavy flows were associated with an increased risk of endometriosis (Cramer et al., 1986; Parazzini et al., 1995; Sangi-Haghpeykar and Poindexter, 1995). However, detailed analysis of data showed some conflicting results that are probably due to the great diversity of patients and controls used (Eskenazi and Warner, 1997). Another potential problem in the overall clinical valuation of data is the nature of endometriosis itself. In fact, endometriosis is not a homogeneous clinical entity and, at present, there is even a strong suggestion that peritoneal endometriosis may be a transitory finding in most women, whereas in a few of them a definitive organic disease develops endometriotic disease and evolves as a benign infiltrating tumour (Evers, 1994; Koninckx, 1994). The extension of endometriosis also seems a crucial point in the correct valuation of the clinical data of patients with the disorder. Human Reproduction q European Society of Human Reproduction and Embryology 2004; all rights reserved Page 1 of 6

2 C.Calhaz-Jorge et al. However, only a few studies (Matorras et al., 1995; Parazzini et al., 1995; Bérubé et al., 1998) reported this information in their population. Apart from these pathophysiological controversies, it is important to identify endometriosis at an early stage in the work-up for subfertility. Whereas women without endometriosis can be managed expectantly or profit from intra-uterine insemination, women with minimal or mild endometriosis may benefit from surgery (Marcoux et al., 1997), and women with moderate or severe endometriosis need surgery or even IVF. In view of these issues, we report on a subfertile Portuguese population in whom the presence of endometriosis was evaluated at laparoscopy. First, we investigated factors that may be related to either minimal/mild or moderate/severe endometriosis. Moreover, we evaluated whether data from the clinical history and symptomatology could predict the presence of endometriosis at laparoscopy. Materials and methods The study was performed between 1993 and 2000 at the Unit of Human Reproduction, Department of Obstetrics and Gynaecology, in the Hospital de Santa Maria in Lisbon, Portugal. The local Ethics Committee approved the study. We studied consecutive subfertile women who underwent either diagnostic or therapeutic laparoscopy. Subfertility was defined as a period of at least 12 months without conception despite unprotected intercourse. Patients with previous pelvic surgery were not excluded. None of the patients had medical treatment for endometriosis within 3 months prior to laparoscopy. Before surgery, all patients answered, by personal interview, a standard questionnaire regarding general characteristics (age at laparoscopy, weight and height, race, education), lifestyle habits (smoking habits), reproductive history (obstetric history, duration of subfertility and use of oral contraceptives), menstrual characteristics (age at menarche, average duration of bleeding and average cycle length), and presence and intensity of pelvic symptomatology (dysmenorrhoea, dyspareunia and pelvic pain). Dysmenorrhoea was categorized as mild (mild discomfort with no use of analgesic medication), moderate (significant pain with need of analgesic medication most of the time) and severe (intense pain with a need for medication every menstrual flow, with or without a need for bed rest and absence from work). A regular menstrual pattern was defined as a regular menstrual cycle shorter than 35 days and a variation in cycle length of,10 days. Laparoscopy could be performed at any day of the menstrual cycle except during menstruation. Endometriosis was confirmed by direct visualization or biopsy of lesions. The characteristics of endometriotic lesions were registered prospectively in a database. No blind biopsies of apparently normal peritoneum were taken. The stage of the disorder was determined according to the revised classification of the American Society for Reproductive Medicine (American Fertility Society, 1985). In addition to this classification, subtle, atypical, lesions were also considered for the diagnosis. Data analysis The characteristics of women with and without endometriosis were tabulated. We classified women as having no endometriosis, minimal to mild endometriosis, or moderate to severe endometriosis. We compared the baseline characteristics in the three groups using a x 2 test or analysis of variance (ANOVA) where appropriate. Page 2 of 6 To evaluate whether clinical data could predict the presence of endometriosis at laparoscopy, we used logistic regression analysis. The presence of endometriosis was considered to be the dependent variable, whereas data from the medical history as well as clinical symptoms were potential predictors. Univariable analysis was performed, in which odds ratios (ORs) and 95% confidence intervals (95% CIs) were calculated. Subsequently, multivariate analysis was used to select independent predictors for endometriosis. The analysis was performed twice. First, we assessed the predictive capacity for the presence of any type of endometriosis. Afterwards, we repeated the analysis, looking at the prediction of the presence of moderate to severe endometriosis only. For the multivariable analysis, we used stepwise logistic regression analysis, in which a P-value of 0.5 was used as entry criterion, whereas a P-value of 0.10 was the threshold for a variable to stay in the model. The performance of the model was expressed as a c-index, which is comparable with the area under the receiver operating characteristic (ROC) curve (AUC). An AUC of 0.5 indicates no discriminative performance, whereas an AUC of 1.0 indicates perfect discrimination. Moreover, we assessed the calibration of the model by comparing the predicted probability in a category of patients and the observed percentage of endometriosis in that category. To do so, we categorized the predicted probabilities in 10 groups, based on percentile points with steps of 10% per step. In each category, we compared the mean predicted probability in that particular category with the observed probability, i.e. the number of women with endometriosis in that category divided by the total number of women in that category. The results were plotted graphically. Results We included 1079 patients. There were 488 women with endometriosis and 591 women without endometriosis, representing a prevalence of endometriosis of 45%. There were 358 patients (33%) who had minimal or mild endometriosis and 130 patients (12%) who had moderate or severe endometriosis. The distribution of the relevant results from the medical history in patients without endometriosis, patients with minimal or mild endometriosis and patients with moderate or severe endometriosis is shown in Table I. There were 257 patients (24%) who had their menarche before their 11th birthday. The prevalence of any type of endometriosis in these women was 44%, whereas the prevalence of moderate to severe endometriosis in these women was 13%. Menarche before the 11th birthday had an OR for any type of endometriosis of 0.95 (95% CI ) and an OR for moderate to severe endometriosis of 1.1 (95% CI ). There were 776 women (72%) with dysmenorrhoea. The prevalences of minimal to mild endometriosis and moderate to severe endometriosis in women with dysmenorrhoea were 35 and 14%, respectively. In women without dysmenorrhoea, these prevalences were slightly lower, 28 and 8%, respectively. The OR for any type of endometriosis in the presence of dysmenorrhoea was 1.7 (95% CI ), whereas the OR for moderate to severe endometriosis was 2.0 (95% CI ). Only in women with primary dysmenorrhoea that

3 Predictive factors for endometriosis in infertile women Table I. Results of the univariate analysis No endometriosis AFS grade I/II AFS grade III/IV P-value AFS any type AFS grade III/IV Age (years) (SD) 30.9 (4.2) 30.9 (3.9) 30.7 (4.0) ( ) 0.99 ( ) Duration of subfertility 56 (34) 51 (30) 50 (36) ( ) 0.99 ( ) (per month) Caucasian 504 (53%) 333 (35%) 121 (13%),0.001 Negroid 78 (74%) 19 (18%) 8 (8%) 0.38 ( ) 0.57 ( ) Other 9 (56%) 6 (38%) 1 (6%) 0.86 ( ) 0.46 ( ) Age at menarche (SD) 12.6 (1.7) 12.6 (1.4) 12.3 (1.4) ( ) 0.89 ( ) No dysmenorrhoea 194 (65%) 86 (28%) 23 (7%),0.001 Secondary dysmenorrhoea 50 (56%) 29 (32%) 11 (12%) 1.4 ( ) 1.7 ( ) Primary dysmenorrhoea 315 (51%) 229 (37%) 73 (12%) 1.7 ( ) 1.6 ( ) Primary dysmenorrhoea; 30 (45%) 14 (21%) 23 (34%) 2.2 ( ) 6.4 ( ) recently intensified Missing value No dysmenorrhoea 194 (64%) 86 (28%) 23 (8%),0.001 Dysmenorrhoea day (51%) 207 (35%) 77 (13%) 1.4 ( ) 1.3 ( ) Dysmenorrhoea all days 97 (51%) 65 (34%) 30 (16%) 1.2 ( ) 1.5 ( ) No dysmenorrhoea 194 (64%) 86 (28%) 23 (8%),0.001 Mild dysmenorrhoea 219 (60%) 116 (32%) 29 (8%) 1.2 ( ) 1.0 ( ) Moderate dysmenorrhoea 142 (45%) 124 (39%) 51 (16%) 2.2 ( ) 2.3 ( ) Severe dysmenorrhoea 36 (38%) 32 (34%) 27 (28%) 2.9 ( ) 4.8 ( ) No dyspareunia 470 (56%) 278 (33%) 97 (11%) 0.39 Always dyspareunia 67 (50%) 52 (39%) 16 (12%) 1.1 ( ) 1.9 ( ) Recent dyspareunia 44 (53%) 25 (30%) 14 (17%) 1.3 ( ) 1.1 ( ) Missing value No dyspareunia 470 (56%) 278 (33%) 97 (11%) 0.53 Sometimes dyspareunia 100 (52%) 69 (36%) 24 (12%) 1.2 ( ) 1.5 ( ) Always dyspareunia 17 (49%) 11 (31%) 7 (20%) 1.3 ( ) 1.0 ( ) Missing value Chronic pelvic pain absent 525 (54%) 333 (35%) 105 (11%) Chronic pelvic pain present 66 (56%) 25 (22%) 25(22%) 0.91 ( ) 2.2 ( ) Always regular menstrual cycle 271 (52%) 172 (33%) 82 (16%) Generally regular menstrual cycle 159 (51%) 117 (38%) 35 (11%) 1.0 ( ) 0.69 ( ) Irregular cycle 161 (66%) 69 (28%) 13 (5%) 0.54 ( ) 0.31 ( ) Duration of menstrual flow (SD) 4.5 (1.7) 4.4 (1.3) 4.5 (1.4) ( ) 1.0 ( ) Mild menstrual flow 161 (66%) 70 (29%) 13 (5%) Moderate menstrual flow 338 (51%) 232 (35%) 91 (14%) 1.9 ( ) 3.4 ( ) Severe menstrual flow 92 (53%) 56 (32%) 26 (15%) 1.8 ( ) 2.8 ( ) Education High school 154 (65%) 65 (28%) 17 (7%) University 286 (53%) 181 (33%) 75 (14%) 0.54 ( ) 0.54 ( ) Other 126 (49%) 100 (39%) 32 (12%) 0.90 ( ) 1.1 ( ) Missing Smoking Never 311 (51%) 220 (36%) 80 (13%) 0.05 Previous smoker 73 (54%) 49 (36%) 14 (10%) 0.90 ( ) 0.76 ( ) Current 1 10 cigarettes/day 93 (62%) 37 (25%) 20 (13%) 0.64 ( ) 1.02 ( ) Current cigarettes/day 78 (63%) 37 (30%) 8 (7%) 0.60 ( ) 0.46 ( ) Current.20 cigarettes/day 36 (60%) 16 (27%) 8 (13%) 0.69 ( ) 1.02 ( ) BMI (missing) 5 0 0, BMI,20 kg/m 2 62 (38%) 74 (45%) 27 (17%) 1.7 ( ) 1.3 ( ) BMI kg/m (51%) 206 (36%) 76 (13%) BMI kg/m (64%) 64 (26%) 23 (9%) 0.58 ( ) 0.68 ( ) BMI.30 kg/m 2 72 (81%) 14 (16%) 3 (3%) 0.23 ( ) 0.21 ( ) Previous pregnancies None (primary subfertility) 350 (49%) 256 (36%) 101 (14%), (62%) 69 (29%) 22 (9%) 0.60 ( ) 0.60 ( ) 2 52 (67%) 22 (28%) 4 (5%) 0.49 ( ) 0.32 ( ) $3 43 (75%) 11 (19%) 3 (5%) 0.32 ( ) 0.33 ( ) Age at first pregnancy (SD) 22.4 (4.9) 24.1 (5.2) 24.3 (4.2) ( ) 1.06 ( ) Previous delivery Never pregnant 350 (50%) 256 (36%) 101 (14%), Pregnancy, no delivery 153 (61%) 79 (31%) 19 (8%) 0.62 ( ) 0.49 ( ) Previous delivery 88 (73%) 23 (19%) 10 (8%) 0.58 ( ) 1.1 ( ) OAC never 176 (64%) 76 (28%) 21 (8%), OAC ever 415 (51%) 282 (35%) 109 (14%) 1.7 ( ) 1.9 ( ) Duration of OAC use (per year) 3.5 (3.2) 3.9 (3.2) 4.6 (3.2), ( ) 1.09 ( ) OR ¼ odds ratio (95% CI); BMI ¼ body mass index; OAC ¼ oral contraceptives; SD ¼ Standard Deviation. Page 3 of 6

4 C.Calhaz-Jorge et al. was intensified recently was the risk of endometriosis slightly more increased than in women with endometriosis in whom dysmenorrhoea was not intensified recently. Among women with mild dysmenorrhoea, the prevalence of endometriosis was not increased as compared with women without dysmenorrhoea. However, both moderate and severe dysmenorrhoea were associated with a 2- to 5-fold increase of the risk of endometriosis. Dyspareunia was present in just over 20% of the women, and associated with a slightly but not statistically significant increased risk on endometriosis. In women with chronic pelvic pain, the risk of moderate to severe endometriosis was doubled. Among the 243 (23%) women who had irregular cycles, the probability of endometriosis was strongly decreased. We found no significant association between endometriosis and shorter menstrual cycles. A moderate to severe menstrual flow was present in 59% of the women, and these women had an increased risk of having endometriosis as compared with women with a mild menstrual flow. Current smokers had a decreased probability of having endometriosis as compared with non-smokers or past smokers, but there seemed to be no dose response association. Whereas in women with a body mass index (BMI),20 kg/m 2 the probability of endometriosis was increased, this risk decreased where BMI exceeded 25 kg/m 2. In women with a BMI.30 kg/m 2, the risk of endometriosis was even decreased almost 5-fold as compared with the risk in women with a BMI between 20 and 25 kg/m 2. The risk of endometriosis in women with a previous pregnancy was decreased as compared with women with primary subfertility. The impact of previous pregnancies appeared to be dose-dependent, i.e. the risk of endometriosis decreased where the number of previous pregnancies increased. In women with a previous pregnancy, the mean age of the women at their first pregnancy was lower in those without endometriosis as compared with those with endometriosis. Previous use of oral contraception was associated with an almost doubled risk of endometriosis. Moreover, in women who had ever used oral contraceptives, the duration of use was longer in women with endometriosis, as compared with women without endometriosis. The results of the multivariable analysis are shown in Table II. The risk of endometriosis of any type was decreased in Negroid women, in obese women, in current smokers and in women with irregular menstrual cycles or previous pregnancy. Women with a BMI,20 kg/m 2 and women who had previously used oral contraceptives showed an increased risk of endometriosis. The model had an AUC of 0.71, indicating an acceptable discriminative performance. Table II also shows the results of the multivariable analysis for the prediction of moderate to severe endometriosis. A previous pregnancy and an irregular cycle decreased the risk of severe endometriosis, but the impact of smoking habits, BMI and race was limited. In contrast to the prediction of any type endometriosis, dysmenorrhoea was an important risk indicator for the presence of moderate to severe endometriosis. The model had an AUC of 0.74, indicating an acceptable discriminative performance. Figure 1 shows the Page 4 of 6 Table II. Results of the multivariable analysis AFS any type AFS grade III/IV Negroid women 0.50 ( ) Dysmenorrhoea any type 2.5 ( ) Mild dysmenorrhoea 0.62 ( ) Moderate dysmenorrhoea 1.7 ( ) Severe dysmenorrhoea 2.8 ( ) Recently intensified 2.4 ( ) dysmenorrhoea Primary dysmenorrhoea 1.4 ( ) Dysmenorrhoea day ( ) Chronic pelvic pain 2.0 ( ) Generally regular menstrual 0.60 ( ) cycle Irregular cycle 0.60 ( ) 0.29 ( ) BMI,20 kg/m ( ) BMI kg/m ( ) BMI.30 kg/m ( ) Smoker 1 10 cigarettes/day 0.57 ( ) Smoker cigarettes/day 0.52 ( ) 0.47 ( ) Smoker.20 cigarettes/day 0.56 ( ) Previous pregnancy 0.65 ( ) 0.58 ( ) Ever use of oral contraceptives 1.6 ( ) 2.2 ( ) AUC 0.71 AUC 0.74 OR ¼ odds ratio (95% CI); BMI ¼ body mass index; AUC ¼ area under the receiver operating characteristic curve. calibration of the prediction model for endometriosis of any type (Figure 1A) and for moderate to severe endometriosis (Figure 1B). Since predicted probabilities were virtually similar to the observed rates of endometriosis in each group, the calibration of the model was good. Discussion In the present study, we evaluated socio-demographic, clinical and behavioural characteristics in a cohort of infertile women with the purpose of investigating the association of those characteristics with the prevalence of endometriosis. Moreover, we evaluated whether a combination of such characteristics could be used in the risk assessment of endometriosis in the same population. Endometriosis was present in 45% of our patients. Although in agreement with previous reports (Mahmood and Templeton, 1991; Koninckx et al., 1991), the prevalence of endometriosis might be somewhat inflated due to the fact that the women undergoing laparoscopy might be a nonrandom selection of the total subfertile population in our clinic, in which women with endometriosis are slightly overrepresented. Indeed, the mean duration of subfertility was 4 years. If one were to use our model to identify patients at a duration of subfertility of 1 year, both prevalence of endometriosis and the magnitude of predictive factors might be different. Overall, our results indicate that the prevalence of endometriosis in subfertile women is related to race, BMI, irregular menstrual cycles, intensity of menstrual flow, dysmenorrhoea, chronic pelvic pain, obstetric history, oral contraceptive use and smoking habits. Some of our findings support the retrograde menstruation theory because increased exposure to menstruation was shown to be associated with an

5 Predictive factors for endometriosis in infertile women Figure 1. The relationship between predicted and observed probability of patients with endometriosis any type (A) and moderate to severe endometriosis (B). The predictive performance of both models is acceptable, since the 95% confidence intervals of the observed probabilities overlap with the predicted probabilities. increased risk of developing endometriosis. Heavier menstrual flow and earlier age of menarche were associated with a higher risk of endometriosis, although the latter association was with moderate to severe disease only. Moreover, a decreased risk was related to irregular, long menstrual cycles and to a history of previous gravidity, and in this case the decrease of the risk was directly related to the number of pregnancies. In our population, we did not find an increased risk of endometriosis for shorter cycles or longer duration of menstrual flow as reported by others (Cramer et al., 1986; Parazzini et al., 1995; Arumugam and Lim, 1997). Like us, Darrow et al. (1993) found no association between endometriosis and cycle length, and Sangi-Haghpeykar and Poindexter (1995) reported that their patients were more likely to have longer cycles than controls. BMI showed an inverse relationship to the presence of endometriosis, as was reported previously (Cramer et al., 1986; Signorello et al., 1997; Bérubé et al., 1998). Obesity frequently is associated with irregular menstrual cycles, a factor repeatedly related to a reduction of the risk of endometriosis (Candiani et al., 1991; Gruppo Italiano per lo Studio dell Endometriosi, 1999). It is conceivable that the reduction of the frequency of menstrual episodes counterbalances the relative hyperestrogenism of most of these women. Another explanation might be that obesity is an independent risk factor for endometriosis. Thus, obese women undergoing a laparoscopy for subfertility might be at decreased risk for endometriosis. In accordance with an earlier report, we found an increased risk for endometriosis in ever users of oral contraceptives compared with never users (Parazzini et al., 1994). The association of oral contraceptive use and endometriosis is difficult to interpret and we cannot provide a clear pathophysiological explanation for our finding. Risk indicators for endometriosis have been the subject of many studies in the past, but several methodological aspects of our study support its additional value. First, in the aetiology of endometriosis, both genetic and environmental factors are important and the risk of endometriosis has never been assessed before in a Portuguese population. A second strength of our study is its power, as well as the fact that it is a cohort study. We are not aware of any study that has assessed the presence of endometriosis in such a large cohort of subfertile women. Three case control studies reported on infertility subjects; the largest one included 591 women (Bérubé et al., 1998). Finally, all women attended the same public University hospital. The study period was relatively long, and one might question whether during this period the assessment of the pelvis might have been changed. Pelvic assessment was done in a standardized way. During the study period, the staffing of our unit was very stable, leading to few changes in the management of subfertile women with respect to laparoscopy and no changes in criteria of evaluation of endometriotic lesions. Moreover, at least one of the authors (C.C.-J. or A.P.C.) was present in the operating theatre in.90% of the laparoscopies. The incidence of endometriosis showed no significant change throughout the study period. The inter-observer variability of the diagnosis of minimal and mild endometriosis using the rasrm classification is considerable because of the subjectivity. Biopsies, that might help to make the diagnosis independent of the observer, were only used in cases where the laparoscopist was in doubt about the nature of a lesion in the absence of other obvious lesions. Endometriosis can only be diagnosed definitively at laparocopy. In view of the absence of reliable tools that can establish the diagnosis in a non-invasive way, there is a strong need for clinical prediction. We constructed a predictive model based on medical history that showed an acceptable predictive performance. Its future testing should also take into account the therapeutic consequences of a diagnosis of minimal to mild endometriosis and moderate to severe Page 5 of 6

6 C.Calhaz-Jorge et al. endometriosis. Eskenazi et al. (2001) developed a model incorporating medical history, physical examination and transvaginal sonography. Their model appeared to be reliable in the prediction of ovarian endometriosis, but it was shown to be of limited value for non-ovarian endometriosis. A difference from their study is that we evaluated a prediction model for absence or presence of endometriosis, whereas they classified endometriosis into three categories: absent, non-ovarian or ovarian. Their algorithm also had an almost perfect specificity. One explanation for the difference from our findings might be that a drop in specificity might have increased sensitivity to an acceptable level. Moreover, we want to stress that in the work-up for subfertility, the question is not whether a test should be ordered now or never, but that the timing of test scheduling is of importance. In cases where the probability of endometriosis is low, then it might be acceptable that the diagnosis is delayed for 6 months; however, the risk of a delayed diagnosis should be communicated to the patient. In conclusion, in the present study, we have demonstrated that data from the medical history are useful in the risk assessment for endometriosis in a cohort of subfertile women, both for any type of endometriosis and for moderate to severe endometriosis. Further analysis of the data, resulting in prediction scores for minimal to mild endometriosis and moderate to severe endometriosis, is warranted. Moreover, a predictive model based on clinical data is proposed. References American Fertility Society (1985) Revised American Fertility Society classification of endometriosis: Fertil Steril 43, Arumugam K and Lim JMH (1997) Menstrual characteristics associated with endometriosis. Br J Obstet Gynaecol 104, Bérubé S, Marcoux S, Maheux R and the Canadian Collaborative Group on Endometriosis (1998) Characteristics related to the prevalence of minimal or mild endometriosis in infertile women. Epidemiology 9, Candiani GB, Danesino V, Gastaldi A, Parazzini F and Ferraroni M (1991) Reproductive and menstrual factors and risk of peritoneal and ovarian endometriosis. Fertil Steril 56, Cobellis L, Latini G, DeFelice C, Razzi S, Paris I, Ruggieri F, Mazzeo P and Petraglia F (2003) High plasma concentrations of di-(2-ethylhexyl)- phthalate in women with endometriosis. Hum Reprod 18, Cramer DW, Wilson E, Stillman RJ, Berger MJ, Belisle S, Schiff I, Albrecht B, Gibson M, Stadel BV and Schoenbaum SC (1986) The relation of endometriosis to menstrual characteristics, smoking, and exercise. J Am Med Assoc 255, Darrow SL, Vena JE, Batt RE, Zielezny MA, Michalek AM and Selman S (1993) Menstrual cycle characteristics and the risk of endometriosis. Epidemiology 4, Eskenazi B and Warner ML (1997) Epidemiology of endometriosis. Obstet Gynecol Clin North Am 24, Eskenazi B, Warner M, Bonsignore L, Olive D, Samuels S and Vercellini P (2001) Validation study of nonsurgical diagnosis of endometriosis. Fertil Steril 76, Evers JLH (1994) Endometriosis does not exist; all women have endometriosis. Hum Reprod 9, Gruppo Italiano per lo Studio dell Endometriosi (1999) Risk factors for pelvic endometriosis in women with pelvic pain or infertility. Eur J Obstet Gynecol Reprod Biol 83, Houston DE, Noller KL, Melton J, Selwyn BJ and Hardy RJ (1987) Incidence of pelvic endometriosis in Rochester, Minnesota, Am J Epidemiol 125, Kennedy S (1999) The genetics of endometriosis. Eur J Obstet Gynecol Reprod Biol 82, Kirshon B and Pointdexter AN (1988) Contraception: a risk factor for endometriosis. Obstet Gynecol 71, Koninckx PR (1994) Is mild endometriosis a condition occurring intermittently in all women? Hum Reprod 9, Koninckx PR, Meuleman C, Demeyere S, Lesaffre E and Cornillie FJ (1991) Suggestive evidence that pelvic endometriosis is a progressive disease, whereas deeply infiltrating endometriosis is associated with pelvic pain. Fertil Steril 55, Koninckx PR, Braet P, Kennedy SH and Barlow DH (1994) Dioxin pollution and endometriosis in Belgium. Hum Reprod 9, Mahmood TA and Templeton A (1991) Prevalence and genesis of endometriosis. Hum Reprod 6, Marcoux S, Maheux R, Bérubé S and the Canadian Collaborative Group on Endometriosis (1997) Laparoscopic surgery in infertile women with minimal or mild endometriosis. N Engl J Med 337, Matorras R, Rodíquez F, Pijoan JI, Ramón O, Terán GG and Rodríguez- Escudero F (1995) Epidemiology of endometriosis in infertile women. Fertil Steril 63, Parazzini F, Ferraroni M, Bocciolone L, Tozzi L, Rubessa S and La Vecchia C (1994) Contraceptive methods and risk of pelvic endometriosis. Contraception 49, Parazzini F, Ferraroni M, Fedele L, Bocciolone L, Rubessa S and Riccardi A (1995) Pelvic endometriosis: reproductive and menstrual risk factors at different stages in Lombardy, northern Italy. J Epidemiol Community Health 49, Pauwels A, Schepens PJC, D Hooghe T, Delbeke L, Dhont M, Brouwer A and Weyler J (2001) The risk of endometriosis and exposure to dioxins and polychlorinated biphenyls: a case control study of infertile women. Hum Reprod 16, Sampson JA (1927) Peritoneal endometriosis due to the menstrual dissemination of endometrial tissue into the peritoneal cavity. Am J Obstet Gynecol 14, Sangi-Haghpeykar H and Poindexter AN (1995) Epidemiology of endometriosis among parous women. Obstet Gynecol 85, Signorello LB, Harlow BL, Cramer DW, Spiegelma D and Hill JA (1997) Epidemiologic determinants of endometriosis: a hospital-based case control study. Ann Epidemiol 7, Submitted on March 10, 2004; accepted on May 25, 2004 Page 6 of 6