.Gynecology-endocrinology

Transcription

1 --.Gynecology-endocrinology FERTILITY AND STERILITY Copyright 1] 1996 American Society for Reproductive Medicine VoL 65 No.6 June 1996 Printed on acid-free paper in U. S. A. Risk factors for ectopic pregnancy: a meta-analysis* Willem M. Ankum M.D. Ph.D.t:!: Ben W. J. Mol M.D.t Fulco Van der Veen M.D. Ph.D.t Patrick M. M. Bossuyt Ph.D. Academic Medical Centre University of Amsterdam Amsterdam The Netherlands Objective: To review current knowledge on the risk of ectopic pregnancy (EP) with the exception of contraceptive methods. Design: Meta-analysis. Setting: Case control and cohort studies published between 1978 and 1994 in English French German or Dutch retrieved by Medline search crossover search from the papers obtained and hand-search on recent medical journals. Patients: A total number of 6718 cases of EP in 27 case control studies and exposed women in 9 cohort studies. Main Outcome Measures: Detected studies were tested for homogeneity. If homogeneity was not rejected Mantel-Haenszel common odds ratios (OR) and 95% confidence intervals were calculated. Results: Previous EP previous tubal surgery documented tubal pathology and in utero diethylstilbestrol (DES) exposure were found to be associated strongly with the occurrence of EP. Previous genital infections (pelvic inflammatory disease [PID] chlamydia gonorrhoea) infertility and a lifetime number of sexual partners> 1 were associated with a mildly increased risk. For gonorrhoea PID previous EP previous tubal surgery and smoking a higher common OR was calculated when using pregnant controls compared with using nonpregnant controls. Conclusions: The strong risk in women with a previous EP previous tubal surgery documented tubal pathology or in utero DES exposure justifies the exploration of a screening policy for EP among these women. If a risk factor reduces fertility chances the OR detected when using pregnant controls is higher than the OR calculated using nonpregnant controls. Fertil Steril1996;65: Key Words: Ectopic pregnancy meta-analysis risk factors Current noninvasive diagnostic methods allow an early diagnosis of ectopic pregnancy (EP) even before the onset of any symptoms thereby improving prospects for the patient (1). Intervention before tubal Received September ; revised and accepted January * Supported in part by grant OG 93/007 from the Ziekenfonds Raad Amstelveen The Netherlands. t Department of Obstetrics and Gynecology. :j: Reprint requests: Willem M. Ankum M.D. Ph.D. Department of Obstetrics and Gynecology (H4-205) Academic Medical Centre University of Amsterdam P.O. Box DE Amsterdam The Netherlands (FAX: ). Department of Clinical Epidemiology and Biostatistics. integrity is lost and before the patient's condition has deteriorated improves clinical outcome. To enable early diagnosis a screening program for women in early pregnancy could be considered. Introduction of such a screening program only can be efficient if it is offered to those who are beyond a threshold risk for EP. Therefore a proper understanding of the risk factors associated with EP is a prerequisite. Although many risk factors for the development of EP have been identified none of the available review articles has tried to summarize and interpret these risk factors in a systematic quantitative way. The present study aims to do so by means of a meta-analysis of published case control and cohort studies. Vol. 65 No.6 June 1996 Ankum et al. Ectopic pregnancy and risk factors 1093

2 Search Strategy MATERIALS AND METHODS A computerized Medline search was undertaken to identify relevant literature. Search conditions for titles and abstracts were "ectopic pregnancy" together with "risk factors." The search was restricted to literature published in English French German and Dutch between January 1978 and January Furthermore a hand search was performed for 1993 and 1994 issues of the top 10 gynecological journals and the top 5 epidemiological journals of the Science Citation Index. From the reference lists of the articles obtained a further crossover search was carried out. Inclusion Criteria The available articles had to meet predefined inclusion criteria. They should either report on a case control or a cohort study. In case control studies cases should be women with EP confirmed either by an operation report or by histopathological examination. Controls should be defined as nonpregnant or pregnant women. Cohort studies should compare women exposed to a risk factor with nonexposed controls. A 2 X 2 table should be available from the articles either directly or in case the paper concentrated on adjusted odds should be retractable from the data supplied; obviously the latter was possible only when data on excluded patients were provided. All reported possible risk factors were analyzed with the exception of contraceptive methods which were reported separately (2). Statistical Analysis From each study 2 X 2 tables were constructed for pregnant and nonpregnant controls. Odds ratios (OR) and their 95% confidence intervals (95% CI) were calculated. Homogeneity was tested by means of the Breslow-Day test (3). P values <0.05 were considered to be statistically significant. If homogeneity was not rejected a common OR with 95% CI was calculated for each exposure by means of the Mantel-Haenszel method thus pooling the ORs of the evaluated studies (3). If homogeneity had to be rejected the range of point estimates is reported. Study Characteristics RESULTS A total of 211 papers meeting the search conditions were obtained from the Medline search. The crossover search revealed another 21 studies whereas 2 studies were detected by hand search. Among these 233 papers there were 43 case control and 23 cohort studies. The other 167 papers were inappropriate for our purpose; 17 papers reported on the incidence and mortality ofep 7 reported exclusively on contraceptive methods while there were 23 review articles 19 observational studies 29 studies on diagnosis 3 editorials 18 case reports and 23 letters. Another 28 papers concerned a variety of other subjects. Of the 43 case control studies thus obtained 4 papers were duplicates of other reports on the same study whereas another 12 did not meet the inclusion criteria leaving 27 papers available for analysis. Of the 23 cohort studies obtained 4 were duplicates of other studies and 10 did not meet the inclusion criteria leaving 9 studies available for analysis. The studies detected reported on one or more of the following possible risk factors: previous genital infections (gonococcal infection pelvic inflammatory disease [PID] chlamydia serology) previous surgical interventions (previous EP abdominal or pelvic surgery and tubal surgery) previous spontaneous or medical abortion infertility and/or tubal pathology (documented by hysterosalpingography or diagnostic laparoscopy) in utero diethylstilbestrol (DES) exposure and several lifestyle factors (smoking vaginal douching lifetime number of sexual partners and age at first sexual intercourse) (4-39). Pregnant Controls The results of the analysis of case control studies that used pregnant controls are shown in Figure 1. A common OR could be calculated for the exposures gonorrhoea chlamydia immunoglobulin G (IgG) antibodies > 1:32 and> 1:64 PID previous EP previous tubal surgery smoking lifetime number of sexual partners> 1 and age of first sexual intercourse < 18 years. Common ORs varied from 1.6 for age of first sexual intercourse < 18 years to 21 for previous tubal surgery. Homogeneity was rejected for chlamydia IgG antibodies> 1:16 previous pelvic and/or abdominal surgery medical abortion spontaneous abortion infertility documented tubal pathology and vaginal douching. For chlamydia IgG > 1:16 the range of ORs of four studies varied from 3.9 to 7.1 ( ) whereas one study reported an OR of 0.72 (26). For pelvic and/or abdominal surgery the range of ORs varied from 0.93 to 3.8. For medical abortion the range of seven studies varied between 0.95 and 2.4. Three of these studies reported a significantly 1094 Ankum et a1. Ectopic pregnancy and risk factors Fertility and Sterility

3 Risk fadors (...) Previous genital infections Gonorrhoea (13lOZ4U) Chlamydia IgG > I: 16 ( ) Chlamydia IgG > 1:32 ( ) Chlamydia IgG > 1:64 (20242!30.32)... 0.~. No Studies COR/OR (95% Cl) t p * ( ) < ( ) ( ) ( ) 0.26 Previous surgery Ectopic pregnancy (I.. I.. I.. I... ~...) Pelvicl abdominal surgery (1S ') Tubal surgery ("l'~) Previous abortion Medical abortion(i~is.i. 24.2I.3~3') Spontaneous abortion ( ) 0: 0 qjo OaID ( ) (9.3-47) Figure 1 Meta-analysis of case control studies with pregnant controls. *Common OR-range of DRs; t95% CI; :j:breslow-day test statistic on homogeneity; " common OR; e significant OR ofa single study; 0 not significant OR of a single study; - 95% CI. Infertility (1J.1S.1' )~3') Tubal pathology (1'.19) Lifestyle factors Current smoking (12 10lIS23435) Ever smoking (162.1) Vaginal douching (24.3~3') Lifetime no sexual partners> 1 (2502) Age lirst sexual intercourse < 18 (25)... ) po 10. i!!!!!i!! ""! f!! II!!!! ""! I o Odds Ratio and 95% CI ( ) ( ) ( ) ( ) 0.41 increased risk ( ). For spontaneous abortion six of eight detected studies showed no significant association with ORs between 0.63 and 1.5 ( ). One study showed a mildly increased OR of 3.3 (36) whereas another study showed a mildly decreased OR of 0.33 (25). Point estimates of the OR in the eight studies reporting on infertility varied between 2.5 and 23 with seven of these eight studies reporting a significantly increased risk ( ). Two studies reporting on tubal pathology had significant ORs with point estimates of 3.8 and 21 (16 19). One ofthree studies reporting on vaginal douching showed a significant association with an OR of 3.1 (24). Nonpregnant Controls The results of the analysis of case control studies using nonpregnant controls are shown in Figure 2. A common OR could be calculated for the exposures PID previous EP medical abortion and current smoking. Common ORs varied between 1.2 for medical abortion and 6.6 for previous EP. Homogeneity was rejected for the two studies re- Vol. 65 No.6 June 1996 Ankum et al. Ectopic pregnancy and risk factors 1095

4 Risk factors (_) No Studies COR/OR ( 95% CI) t p * Previous genital infections Gonorrhoea (31) Chlamydia IgG > 1: 16 (11) PID ( ) ( ) 16 (6.6-38) 1.7 ( ) 0.3 Previous surgery Ectopic pregnancy f" ) Pelvic! abdominal s&gery (23) ~ ( ) 1.4 ( ) 0.2 Tubal surgery (23) 4.7 ( ) Previous Abortion Medical abortion (111731) Spontaneous abortion (11.31).. q ( ) <0.01 Infertility (38) 3.6 ( ) Tubal pathology 0 Lifestyle factors Current smoking (1123) Ever smoking (23) Vaginal douching (27) Lifetime no sexual partners> 1 (31) Age frrst sexual intercourse < 18 (22) 0.1!!!!d : -- :---- :--- 1!!! "I 10 Odds Ratio and 95% CI ( ) 1.7 ( ) 1.6 ( ) 2.5 ( ) 1.5 ( ) 0.25 Figure 2 Meta-analysis of case control studies with nonpregnant controls. *Common OR-range of ORs; t95% CI; :j:breslow-day test statistic on homogeneity; I common OR;. significant OR of a single study; 0 not significant OR of a single study; - 95% CI. porting on spontaneous abortion (18 31). Only one of these two studies showed a significant OR with a point estimate of2.0 (31). For each of the risk factors gonorrhoea chlamydia IgG antibodies> 1:16 previous pelvic and/or abdominal surgery previous tubal surgery infertility ever smoking vaginal douching lifetime number of sexual partners > 1 and age of first sexual intercourse < 18 years only one study was found. Cohort Studies Figure 3 shows the results of the analysis of cohort studies. One study reported on two cohorts of women who were delivered by a caesarean section (14). The common OR calculated from these cohorts was 1.5. In five studies reporting on DES exposure homogeneity was not rejected and a common OR of 5.6 was calculated (4-710). The ORs ofpid medical abortion and infertility each derived from a single study also are shown ( ). DISCUSSION This meta-analysis assesses current knowledge of the risk factors for EP. Previous EP tubal surgery documented tubal pathology and in utero DES expo Ankum et ai. Ectopic pregnancy and risk factors Fertility and Sterility

5 Pelvic and/or abdominal surgery (14) In utero DES-exposure ('10) PIO(3') Medical Abortion (9) Infertility (21) ~ - No Studies COR/OR (95% CI) t p 1: NA NA NA Figure 3 Meta-analysis of cohort studies. *Common ORrange of ORs; t95% CI; :j:breslow-day test statistic on homogeneity; _ common OR;. significant OR of a single study; 0 not significant OR of a single study; - 95% CI. Common Odds Ratio/ Range of Odds Ratios. 95 % Confidence Intervals t Breslow-Day test-statistic on homogeneity Odds Ratio and 95% CI _ COR Significant OR of a single study Not Significant OR of a single study -- 95%CI sure are associated strongly with the occurrence of EP. Previous genital infections (PID chlamydia gonorrhoea) infertility and a lifetime number of sexual partners exceeding one are associated with a mildly increased risk. Previous pelvic and/or abdominal surgery smoking vaginal douching and an early age of first sexual intercourse are associated with a slightly increased risk. The impact of medical and spontaneous abortion on the occurrence of EP seems to be very limited. It should be noted that the traditional meta-analysis used in the present study does not provide an insight in the interplay between risk factors some of which may be highly correlated. Although a multivariate approach has been described for meta-analysis unfortunately the individual papers do not report sufficient data to allow such a strategy (40). The majority of papers available for the metaanalysis were case control studies. The fundamental problem in the interpretation of case control studies is control definition (41). If a risk factor interferes with fertility the OR for the occurrence of EP is dependent on the selection of the control group. This is explained by the cumulative effect oftwo probabilities involved in the occurrence of EP. First there is the probability of achieving conception. Conditional on the probability of achieving conception there is the second probability of subsequent ectopic nidation. Most risk factors in this meta-analysis interfere with both. For example a history of tubal surgery leads to a decreased probability of conception but increases the probability of ectopic nidation once a pregnancy occurs. Studies using pregnant controls describe the risk of EP only for those who are currently pregnant. Therefore ORs from these studies describe only the second probability i.e. ectopic nidation conditional on the presence of a pregnancy. Studies using nonpregnant controls however incorporate both probabilities mentioned earlier i.e. the probability of conception and the probability of subsequent ectopic nidation. Consequently ORs in these studies express the joint occurrence of two events. This implies that common ORs from studies using pregnant controls are higher than those from studies using nonpregnant controls whenever risk factors reduce the probability of achieving a pregnancy. For instance for tubal surgery the OR of EP compared with nonpregnant controls is 4.7 whereas it is 21 compared with pregnant controls. The fact that the latter common OR is significantly higher reflects the decreased fertility prospects after tubal surgery. This mechanism also explains the difference in common ORs for gonorrhoea PID previous EP and smoking. The findings from the cohort studies are in concordance with those from case control studies. In PID however there seems to be a discrepancy: the cohort study reports an OR of 5.7 whereas the common ORs from case control studies using nonpregnant and pregnant controls were 1. 7 and 2.6 respectively. The difference might result from insufficient data as the 95% CIs are overlapping. It also could be explained by a different definition ofpid: in the cohort study the diagnosis was made laparoscopically whereas in case control studies it was obtained from interviews or medical records. The common denominator in the etiology of EP seems to be the fallopian tube. Previous EP tubal surgery documented tubal pathology PID gonorrhoea and in utero DES exposure which show the Vol. 65 No.6 June 1996 Ankum et al. Ectopic pregnancy and risk factors 1097

6 highest ORs all affect the fallopian tube directly. The same pathophysiologic mechanism seems involved in infertility and previous pelvic/abdominal surgery although apparently these risk factors not by definition affect the tube. Smoking is thought to affect tubal motility thus increasing the risk of ectopic nidation but might also represent a certain lifestyle associated with an increased risk. The observed influence of lifestyle factors in the occurrence of EP most likely is explained by confounding. Sexually transmitted diseases probably play an important role in this respect. In conclusion this meta-analysis shows previous EP previous tubal surgery documented tubal pathology and in utero DES exposure to be the strongest risk factors associated with the occurrence ofep. Current intrauterine contraceptive device use (range of ORs between 4.2 and 45) and sterilization (common OR 9.3) were detected previously as strong risk factors when compared with pregnant controls (2). Screening women with these risk factors for EP seems promising (1). However the success of such a screening program depends not only on the OR of the risk factor but also on the pretest probability of EP. A low pretest probability reduces the benefits from screening whereas a high pretest probability may justify the screening of women with mildly increased risk factors. Currently we are evaluating the cost effectiveness of such a screening policy and hope to report our results in the near future. Acknowledgment. The authors thank Miss Heleen Dyserinck for assistance in performing the search and Mr. Hans Groenendijk for collecting literature. REFERENCES 1. Ankum WM Van der Veen F Hamerlynck JVTH Lammes FB. Laparoscopy: a dispensable tool in the diagnosis of ectopic pregnancy? Hum Reprod 1993;8: Mol BWJ Ankum WM Bossuyt PMM Van der Veen F. Contraception and the risk of ectopic pregnancy: a meta-analysis. Contraception. 1995; 52: Kuritz SJ Landis JR Kocg GG. A general overview of Mantel-Haenszel methods: applications and recent developments. Annu Rev Public Health 1988;9: Barnes AB Colton T Gundersen J Noller KL Tilley BC Strama T et al. Fertility and outcome of pregnancy in women exposed in utero to diethylstilbestrol. N Engl J Med 1980;302: Cousins L Karp W Lacey C Lucas WE. Reproductive outcome of women exposed to diethylstilbestrol in utero. Obstet GynecoI1980;56: Herbst AL Hubby MM Blough RR Azizi F. A comparison of pregnancy experience in DES-exposed and DES-unexposed daughters. J Reprod Med 1980;24: Kaufman RH Adam E Binder GL Gerthoffer RN. Upper genital tract changes and pregnancy outcome in offspring exposed in utero to diethylstilbestrol. Am J Obstet Gynecol 1980; 137: Ory HW Women's health study. Ectopic pregnancy and intrauterine contraceptive devices: new perspectives. Obstet Gynecol 1981; 57: Chung CS Smith RG SteinhoffPG Mi M. Induced abortion and ectopic pregnancy in subsequent pregnancies. Am J Epidemiol 1982; 115: Mangan CE Borow L Burtnett-Rubin MM Egan V Giuntoli RL Mikuta JJ. Pregnancy outcome in 98 women exposed to diethylstilbestrol in utero their mothers and unexposed siblings. Obstet Gynecol 1982;59: Svensson L Mardh P-A Ahlgren M Nordenskjiild F. Ectopic pregnancy and antibodies to Chlamydia trachomatis. Fertil Steril 1985;44: WHO Task Force on Intrauterine Devices for Fertility Regulation. A multinational case-control study of ectopic pregnancy. Clin Reprod Fertil 1985;3: Brunham RC Binns B McDowell J Paraskevas M. Chlamydia trachomatis infection in women with ectopic pregnancy. Obstet GynecoI1986;62: Hemminki E. Longterm maternal health effects of caesarian section. J Epidemiol Community Health 1991;45: Thorburn J Berntsson C Philipson M Lindblom B. Background factors of ectopic pregnancy. I. Frequency distribution in a case-control study. Eur J Obstet Gynecol Reprod BioI 1986;23: Yang C-P Chow WH Daling JR Weiss NS Moore DE. Does prior infertility increase the risk of tubal pregnancy? Fertil Steril 1987;48: Burkman RT Mason KJ Gold EB. Ectopic pregnancy and prior induced abortion. Contraception 1988; 37 : Parazzini F La Vecchia C Fasoli M Cecchetti G Mezzanotte G. Trends in ectopic pregnancies and use of intrauterine devices in Lombardy Italy Contraception 1988: 37: Tuomivaara L Kauppila A. Ectopic pregnancy: a case-control study of aetiological risk factors. Arch Gynecol Obstet 1988;243: Robertson IN Hogston P Ward ME. Gonococcal and chlamydial antibodies in ectopic and intrauterine pregnancy. Br J Obstet Gynaecol 1988;95: Varma TR Patel RH Bhathenia RK. Outcome of pregnancy after infertility. Acta Obstet Gynecol Scand 1988;67: Holt VL Daling JR Voight LF McKnight B Stergachis A Chu J et al. Induced abortion and the risk of subsequent ectopic pregnancy. Am J Public Health 1989;79: Ni H Daling JR Chu J Stergachis A Voight LF Weiss NS. Previous abdominal surgery and tubal pregnancy. Obstet GynecoI1990;75: Chow JM Yonekura ML Richwald GA Greenland S Sweet RL Schachter J. The association between Chlamydia trachomatis and ectopic pregnancy. JAMA 1990;263: Walters MD Eddy CA Gibbs RS Schachter J Holden AEC Pauerstein CJ. Antibodies to Chlamydia trachomatis and risk for tubal pregnancy. Am J Obstet Gynecol1988; 159: Tuomivaara LM. Ectopic pregnancy and genital infections: a case-control study. Ann Med 1990;22: Daling JR Weiss NS Schwarz SM Stergachis A Wang SP Foy HM et al. Vaginal douching and the risk of tubal pregnancy. Epidemiology 1991;2: Kalandidi A Doulgerakis M Tzonou A Hsieh CC Aravandinos D Trichopoulos D. Induced abortions contraceptive 1098 Ankum et al. Ectopic pregnancy and risk factors Fertility and Sterility

7 practices and tobacco smoking as risk factors for ectopic pregnancy in Athens Greece. Br J Obstet Gynaecol1991; 98: Nordenskjold F Ahlgren M. Risk factors in ectopic pregnancy. Results from a population-based case-control study. Acta Obstet Gynecol Scand 1991; 70: Picaud A Berthonneau JP Nlome-Nze AR Ogowet-Igumu N Engongah-Beka T Faye A. Serologie des Chlamydiae et grosse sse extra-uterine. Frequence du syndrome de Fitz Hugh-Curtis. J Gynecol Obstet BioI Reprod (Paris) 1991; 20: Stergachis A Scholes D Daling JR Weiss NS Chu J. Maternal cigarette smoking and the risk of tubal pregnancy. Am J Epidemiol 1991; 133: Chrysostomou M Karafyllidi P Papadimitriou V Bassiotou V Mayakos G. Serum antibodies to Chlamydia trachomatis in women with ectopic pregnancy normal pregnancy or salpingitis. Eur J Obstet Gynecol Reprod BioI 1992:44: Osser S Persson K. Chlamydial antibodies and deoxyribonucleic acid in patients with ectopic pregnancy. Fertil Steril 1992;57: Phillips RS Tuomala RE Feldblum PJ Schachter J Rosenberg MJ Aronson MD. The effect of cigarette smoking Chla- mydia trachoma tis infection and vaginal douching on ectopic pregnancy. Obstet Gynecol 1992; 79: Westrom L JoesoefR Reynolds G Hagdu A Thompson SE. Pelvic inflammatory disease and fertility. Sex Transm Dis 1992; 19: Michalas S Minaretzis D Tsionou Ch Maos G Kioses E Aravantinos D. Pelvic surgery reproductive factors and risk of ectopic pregnancy: a case-control study. Int J Obstet 1992;38: Basuki B Rossing MA Daling JR. Intrauterine device use and risk of tubal pregnancy: an Indonesian case-control study. Int J Epidemiol 1994;23: Rossing MA Daling JR Weiss NS Voigt LF Stergachis AS Wang SP et al. Past use of an intrauterine device and risk of tubal pregnancy. Epidemiology 1993;4: Kihlstrom E Lindgren R Ryden G. Antibodies to Chlamydia trachomatis in women with infertility pelvic inflammatory disease and ectopic pregnancy. Eur J Obstet Gynecol Reprod BioI 1990;35: Greenland S. Quantitative methods in the review of epidemiologicalliterature. Epidemiol Rev 1987;9: Weiss NS Daling JR Chow WHo Control definition in casecontrol studies of ectopic pregnancy. Am J Public Health 1985; 75:67-8. Vol. 65 No.6 June 1996 Ankum et a1. Ectopic pregnancy and risk factors 1099