Comparison of hysterosalpingography and laparoscopy in predicting fertility outcome

Ben W.J. Mol1,2,5, John A. Collins3,4, Elizabeth A. Burrows4, Fulco van der Veen2 and Patrick M.M. Bossuyt1

1 Departments of Clinical Epidemiology and Biostatistics and 2 Reproductive Medicine, Academic Medical Centre, University of Amsterdam, Amsterdam, the Netherlands, Departments of 3 Obstetrics and Gynaecology and 4 Clinical Epidemiology and Biostatistics, McMaster University, 1200 Main Street West, Room 3N52, Hamilton, Ontario, Canada L8N 3Z5


    Abstract
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
In this study, we compare the prognostic significance of hysterosalpingography (HSG) and laparoscopy for fertility outcome. In a prospective cohort study in 11 clinics participating in the Canadian Infertility Treatment Evaluation Study (CITES), consecutive couples who registered between 1 April 1984 and 31 March 1987 for the evaluation of subfertility and who underwent HSG and laparoscopy were included. Unilateral and bilateral tubal occlusion at HSG and laparoscopy were related to treatment-independent pregnancy. Cox regression was used to calculate fecundity rate ratios (FRR). Of the 794 patients who were included, 114 (14%) showed one-sided tubal occlusion and 194 (24%) showed two-sided tubal occlusion on HSG. At laparoscopy, 94 (12%) showed one-sided tubal occlusion and 96 (12%) showed two-sided tubal occlusion. Occlusion detected on HSG and laparoscopy showed a moderate agreement beyond chance (weighted {kappa}-statistic 0.42). The adjusted FRR of one-sided tubal occlusion at HSG was 0.80, whereas two-sided tubal occlusion showed an FRR of 0.49. For laparoscopy, the FRR were 0.51 and 0.15 respectively. After a normal or one-sided occluded HSG, laparoscopy showed two-sided occlusion in 5% of the patients, and fertility prospects in these patients were virtually zero. If two-sided tubal occlusion was detected on HSG but not during laparoscopy, fertility prospects were slightly impaired. Fertility prospects after a two-sided occluded HSG were strongly impaired in cases where laparoscopy showed one-sided and two-sided occlusion, with FRR of 0.38 and 0.19 respectively. Although laparoscopy performed better than HSG as a predictor of future fertility, it should not be considered as the perfect test in the diagnosis of tubal pathology. For clinical practice, laparoscopy can be delayed after normal HSG for at least 10 months, since the probability that laparoscopy will show tubal occlusion after a normal HSG is very low.

Key words: hysterosalpingography/laparoscopy/prognosis/subfertility/tubal occlusion


    Introduction
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Laparoscopy is currently regarded as the most reliable tool in the diagnosis of tubal causes of subfertility. Since laparoscopy visualizes morphological abnormalities of the Fallopian tubes directly, it is generally accepted as the reference standard for determination of the accuracy of other diagnostic tools for tubal pathology, such as hysterosalpingography (HSG) or Chlamydia antibody testing (Swart et al., 1995Go; Mol et al., 1997aGo; Land et al., 1998Go).

In the pre-in-vitro fertilization (IVF) era, surgical correction of tubal abnormalities was the only available treatment for tubal subfertility. Since then, IVF–embryo transfer has gradually replaced tubal surgery as the treatment of choice for tubal subfertility in many countries. It has become the treatment of last resort for persistent infertility due to any cause. In contrast to tubal surgery, the choice for IVF–embryo transfer does not depend on the detection of morphological abnormalities, but on fertility prognosis. Therefore, the central issue in the work-up of subfertility has changed from a diagnostic question into a prognostic question.

Previous studies have shown that bilateral tubal pathology diagnosed at HSG or laparoscopy did affect fertility prospects strongly, whereas unilateral pathology affected future fertility less severely (Norderskjöld and Ahlgren, 1983; Mol et al., 1997bGo). Although a substantial part of the population in these studies underwent HSG and laparoscopy, a direct comparison on the prognostic capacity between HSG and laparoscopy has never been made. In a previous study in 11 Canadian infertility clinics addressing the prognosis of live birth among untreated subfertile couples, presence of tubal pathology was shown to reduce prospects for treatment-independent pregnancy with 50% (Collins et al., 1995Go). In that particular study, tubal pathology was detected with HSG and/or laparoscopy. Comparison of HSG and laparoscopy was beyond the scope of that study. In the present study, the data of the Canadian study are reanalysed for this purpose.


    Materials and methods
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Patients
We used data prospectively collected in the Canadian Infertility Treatment Evaluation Study (CITES). All couples who registered for the first time in 11 Canadian academic infertility clinics between April 1, 1984 through March 31, 1987, and who had both HSG and laparoscopy done were included. Patients with an abnormal HSG usually underwent laparoscopy without delay, whereas in patients with a normal HSG, laparoscopy was only performed in cases where subfertility persisted for a longer period of time.

Findings at HSG were classified as no tubal occlusion, one-sided tubal occlusion or two-sided tubal occlusion (partial or total occlusion). Findings at laparoscopy were classified as normal, one-sided tubal occlusion or two-sided tubal occlusion. Additional tubal pathology observed at laparoscopy, i.e. phimosis or adhesions, was scored separately. Furthermore, endometriosis detected at laparoscopy was classified according to the classification of the American Fertility Society (1985).

The following potential prognostic factors were also used in the analysis: female age (per year older than 30 years), duration of subfertility at the time of laparoscopy, type of subfertility (primary or secondary), and the presence of subfertility factors, i.e. cervical hostility, ovulation factor and sperm factor.

Follow-up
Follow-up ended when treatment-independent pregnancy occurred, or at the day on which fertility treatment started. Pregnancy was defined as an ongoing pregnancy at a gestational age of 12 weeks. When ectopic pregnancy or spontaneous abortion occurred, follow-up ended on the estimated day of conception. If a woman did not become pregnant and had no treatment, follow-up ended on the day of last contact. A woman was presumed to be `at risk' for treatment-independent conception for as long as the couple was in the study.

Analysis
Tubal occlusion detected at HSG was compared with occlusion detected at laparoscopy in a three by three table. Sensitivity, specificity and 95% confidence intervals (CI) of HSG in the diagnosis of tubal occlusion were calculated, regarding laparoscopy as the reference standard. Sensitivity and specificity were calculated twice, once when tubal occlusion was defined as one-sided or two-sided occlusion, and once when the definition of tubal pathology was limited to two-sided occlusion. Agreement beyond chance between HSG and laparoscopy was expressed as a weighted k-statistic. Whereas an unweighted {kappa}-statistic judges disagreement on one tube in one patient as severely as disagreement on two tubes in one patient, a weighted {kappa}-statistic judges cases in which HSG and laparoscopy disagree on the status of one tube, but agree on the status of the other, less severely than cases in which HSG and laparoscopy disagree on both tubes. The latter is more appropriate for clinical practice, since the difference between a bilateral patent HSG and a one-sided occluded HSG is clinically not as important as the difference between a bilateral patent HSG and a two-sided occluded HSG.

Three-year cumulative pregnancy rates were calculated for each category of HSG and laparoscopy findings, using Kaplan–Meier analysis (Collett, 1994Go). Subsequently, fecundity rate ratios (FRR) and 95% confidence intervals (CI) for the occurrence of treatment-independent ongoing pregnancy were calculated for findings at HSG as well as for findings at laparoscopy through Cox regression modelling (Cox, 1972Go). Furthermore, the tubal status as assessed by a combination of HSG and laparoscopy was related to treatment-independent pregnancy. In addition, we determined FRRs for other potential prognostic factors.

An FRR expresses the probability of spontaneous intrauterine pregnancy (IUP) per time unit for patients with a particular feature, relative to the probability in those without that feature (Olsen et al. 1998Go; Spira 1998Go). For example, if a feature is associated with an FRR of 2, subjects with this feature have a two-fold increased chance of becoming pregnant in a given time interval. To adjust the FRR of HSG and laparoscopy-findings for other potential prognostic factors, multivariable analysis was performed. Since the aim of the study was to compare the prognostic significance of HSG and laparoscopy, we performed three separate multivariable analyses, one in which the FRRs of HSG were corrected for all other prognostic factors but findings at laparoscopy, one in which the FRRs of laparoscopy were corrected for all other prognostic factors but findings at HSG, and one in which the FRRs of combined results of HSG and laparoscopy were corrected for all other prognostic factors.


    Results
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Of the 2198 couples that were included in CITES, 1357 had HSG performed. Among the 563 patients who had HSG but no laparoscopy, 107 had a treatment-independent pregnancy. The 794 who had HSG and laparoscopy were included in the present study. Mean female age was 29.6 years (range 20–42) and mean duration of infertility was 41 months (range 12–153). The median time between HSG and laparoscopy was 10 months when HSG was normal, 8.5 months when HSG was one-sided abnormal and 4.5 months when HSG was bilateral abnormal. Among the 794 included couples, 86 had a treatment-independent pregnancy. Of these 86 couples, four had an ectopic pregnancy, whereas 12 pregnancies resulted in miscarriage. Thus, 70 women had a treatment-independent ongoing pregnancy, of which 50 resulted in a live birth and three in perinatal death. Among 17 couples pregnancy outcome was unknown.

Table IGo shows tubal status detected at HSG as compared to tubal status detected at laparoscopy. At HSG, 114 (14%) showed one-sided tubal occlusion and 194 (24%) showed two-sided tubal occlusion. At laparoscopy, 94 (12%) showed one-sided tubal occlusion and 96 (12%) showed two-sided tubal occlusion. Sensitivity of HSG was 0.81 (95% CI 0.76–0.81) and specificity 0.75 (95% CI 0.71–0.78) when disease was defined as any form of tubal occlusion detected at laparoscopy, be it one-sided or two-sided. Sensitivity and specificity of HSG were 0.72 (95% CI 0.62–0.81) and 0.82 (95% CI 0.79–0.85) respectively when the definition of disease was limited to double-sided tubal occlusion detected at laparoscopy. The weighted {kappa}-statistic expressing the agreement between findings at HSG and laparoscopy beyond chance was 0.42 (95% CI 0.37–0.48).


View this table:
[in this window]
[in a new window]
 
Table I. Tubal status detected at HSG as compared to the tubal status detected at laparoscopy
 
Table IIGo shows the distribution of findings at HSG and laparoscopy in relation to the number of subsequent pregnancies and the 3 year cumulative ongoing pregnancy rate. Three year cumulative ongoing pregnancy rates varied between 11% in cases where HSG and laparoscopy were both two-sided patent or one-sided occluded and 0% in several categories.


View this table:
[in this window]
[in a new window]
 
Table II. Tubal status detected at HSG and laparoscopy in relation to the number of treatment-independent intrauterine pregnancies and the 3-year cumulative intrauterine pregnancy rate
 
Table IIIGo shows the results of the univariable and multivariable Cox-regression analysis. In the multivariable analysis, one-sided occlusion detected at HSG was found to decrease fertility prospects slightly (FFR 0.80), whereas a two-sided occlusion had a stronger impact on fertility prospects (FRR 0.49). Occlusion detected at laparoscopy had a stronger impact on fertility prospects, with an FRR of 0.51 for a one-sided occlusion and 0.15 for a two-sided abnormality. Other tubal pathology at laparoscopy, i.e. phimosis and adhesions, was associated with an FRR of 0.6. Endometriosis grade I/II had an FRR of 0.52, whereas there were no pregnancies among couples with endometriosis grade III/IV.


View this table:
[in this window]
[in a new window]
 
Table III. comparison of HSG and laparoscopy
 
If one-sided occlusion at HSG was followed by normal laparoscopy fertility prospects were slightly impaired (FRR 0.81). If a normal HSG or a one-sided occluded HSG was followed by a laparoscopy that showed one-sided occlusion, which occurred in 8% of patients (50/600) with such HSGs, fertility prospects were moderately decreased (FRR 0.58). When a normal HSG or a one-sided occluded HSG was followed by a laparoscopy that showed two-sided abnormalities, which occurred in 5% of the patients (27/600) with such HSG, no spontaneous pregnancies occurred (FRR 0). When a two-sided occluded HSG was followed by a normal laparoscopy, which occurred in 42% of the patients (81/194) with such HSG, fertility prospects were slightly impaired (FRR 0.70). When a two-sided abnormal HSG was followed by a laparoscopy that showed one-sided or two-sided abnormalities, which occurred in 23% (44/194) and 36% (69/194) of patients with such HSG respectively, fertility prospects were strongly impaired, with FRR of 0.38 and 0.19 respectively.


    Discussion
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Laparoscopy is currently regarded as the best available method to assess tubal occlusion. In this study, tubal occlusion was detected in 24% of the laparoscopies. Two-sided occlusion appeared to affect fertility prospects considerably, whereas one-sided occlusion affected fertility prospects less strongly. In the same sample of patients, HSG detected tubal occlusion in 38%. The weighted {kappa}-statistic was 0.42 (95% CI 0.37–0.48), indicating moderate agreement beyond chance between HSG and laparoscopy. Findings at laparoscopy had a stronger impact on spontaneous fertility course than HSG results. After a completely normal HSG or a HSG with a one-sided abnormality, a one-sided occluded laparoscopy affected fertility prospects slightly, whereas no spontaneous pregnancies occurred after double-sided occlusion detected at laparoscopy. After a HSG with two-sided abnormalities fertility prospects were only slightly decreased in cases where laparoscopy showed patent tubes. However, in cases where laparoscopy showed one-sided or two-sided abnormalities in these patients, fertility prospects were strongly decreased, with FRR of 0.38 and 0.19 respectively.

Laparoscopy was performed at the end of the work-up for subfertility, after the performance of HSG. Patients with an abnormal HSG usually underwent laparoscopy with a short delay, whereas in patients with a normal HSG laparoscopy was withheld for a longer time. This selection bias possibly hampers the interpretation of our findings in two ways. Firstly, unexplained subfertility, that might be associated with an unknown fertility reducing factor, could be over-represented in patients with a normal HSG. If such a selection bias were to play a role in our study, this bias might cause an underestimation of the prognostic capacity of HSG as detected in this study. Secondly, the delay of laparoscopy among patients with normal HSG results in an overestimation of patients with abnormal laparoscopy among all patients undergoing HSG.

Although the analysis incorporated a cohort of almost 800 women, only 70 women had a treatment-independent pregnancy. As a consequence, the 95% confidence intervals of the FRR were wide and many included 1, thereby indicating that the results were not statistically significant. Consequently, the results of our study must be interpreted with caution, and there is still a need for confirmation of the results by other studies. On the other hand, the point estimates that we calculated for the FRR of tubal pathology at HSG and laparoscopy were comparable to estimates provided in other studies (Nordenskjöld and Ahlgren, 1983Go; Mol et al., 1997bGo). A second consequence of the limited power was that the impact of a more subtle classification of HSG findings, for example a distinction between partial and total occlusion or a distinction between proximal and distal occlusion could not be evaluated.

Many studies have associated morphological abnormalities with fertility outcome in patients who underwent microsurgical correction for tubal occlusion (Caspi et al., 1979Go; Te Velde et al., 1989Go; Strandell et al., 1995Go). So far, only one study has assessed the significance of findings at laparoscopy in patients evaluated for subfertility. Norderskjöld and Ahlgren (1983) reported on 433 subfertile women who had laparoscopy. The presence of adhesions reduced fertility prospects on the same order as unilateral tubal occlusion, with relative risks of 0.74 (95% CI 0.57–0.98) and 0.73 (95% CI 0.39–1.4) respectively. None of 10 patients with a double-sided occlusion of the tube became pregnant. The fact that 101 (23%) of the patients had microsurgery some time after laparoscopy was not addressed in that study, thereby hampering interpretation of the results. Furthermore, dividing subfertile couples into couples who conceived and couples who did not conceive does not address the true nature of (sub)fertility. Instead, analysis of time to pregnancy, as was done in the present study, is more appropriate (Olsen et al., 1998Go; Spira, 1998Go).

In a meta-analysis comparing results of HSG and laparoscopy for the diagnosis of tubal pathology, HSG had a sensitivity of 65% for the diagnosis of tubal occlusion, in cases where laparoscopy was presumed to be the `gold' standard (Swart et al., 1995Go). This finding indicates that 35% of the tubes that were found to be occluded at laparoscopy showed patency at HSG. That particular finding made the choice of laparoscopy as the `gold' standard questionable, since patency at HSG in our opinion proves that laparoscopy was incorrect in diagnosing tubal occlusion in these patients. The results of the present study seem to indicate that laparoscopy is a better predictor for infertility than HSG, albeit not a perfect one. However, this conclusion is hampered by the fact that the median interval to laparoscopy after normal HSG was 10 months, compared to 4.5 months in cases where HSG was two-sided abnormal. Thus, the difference in prognostic capacity between HSG and laparoscopy is likely to be overestimated. Only prognostic studies in which HSG and laparoscopy are performed at the same moment can overcome this issue.

There seems to be no straightforward explanation for the difference in predictive capacities of HSG and laparoscopy that were found in the present study. Possibly, the direct visualization of methylene blue in relation to the Fallopian tubes, as compared to the indirect visualization of contrast at HSG might be responsible for this difference. Despite the fact that laparoscopy seems to be a better predictor for infertility than HSG, we postulate that HSG should keep its place in the diagnostic work-up for subfertility. Normal HSG reduces the probability that a tubal factor plays a role in future fertility prospects. Only in 5% of the patients with normal HSG was double-sided tubal occlusion detected at laparoscopy. Although fertility prospects in these couples were virtually zero, we think that a probability of 1 in 20 to detect severe abnormalities at laparoscopy – observed after a median time between HSG and laparoscopy of 10 months – is so low that earlier laparoscopy is not justified. It should be kept in mind that the median delay of laparoscopy of 10 months implicates that the fraction of abnormal laparoscopy among all patients with a normal HSG is even lower than 5%. When deciding about a delay of laparoscopy, one should also take into account other aspects of the prognostic profile of the couple, of which female age is the most important. In case female age is >36 years, the success rates of IVF–embryo transfer are expected to decline and delay of diagnosis and treatment is not justified (Templeton et al., 1996Go; Van Kooij et al., 1996Go).

In contrast, laparoscopy performed after a two-sided abnormal HSG showed no abnormalities in 42% of the patients. Since fertility prospects in these patients were only slightly impaired, whereas patients with two-sided occluded HSG and laparoscopy showing bilateral occlusion had strongly impaired fertility prospects, laparoscopy performed after a two-sided abnormal HSG could be very useful, since it divides patients with a two-sided occluded HSG into a large group in whom fertility prospects are slightly impaired and a large group in which fertility prospects are strongly impaired.

Apart from HSG and laparoscopy, other tests are available for the assessment of the status of the Fallopian tubes. Atalas et al. (1997) recently demonstrated that sonohysterography performed better than HSG in the detection of intrauterine abnormalities, using laparoscopy as the reference strategy. The functional approach of the Fallopian tubes might be evaluable with radionuclide HSG, although a recent report on this method indicates that at present it does not seem to be a reliable method in the infertility work-up (Lundberg et al., 1997Go). Transvaginal hydrolaparoscopy allows direct visualization of the ovaries and Fallopian tubes in an outpatient setting (Gordts et al., 1998Go). However, all these techniques have been evaluated in relatively small groups of patients, and the prognostic capacity for fertility has not been established yet. Before their implementation in clinical practice, their capacity to predict future fertility should be assessed, preferably in a direct comparison with HSG and/or laparoscopy, as was done in the present study.

When comparing HSG and laparoscopy, we should keep in mind that both procedures provide more information than the condition of the Fallopian tubes alone. Whereas HSG provides information on the status of the intrauterine cavity, laparoscopy allows inspection of the intra-abdominal cavity, for instance to see if endometriosis is present. The latter has become especially important, since it was recently shown that laparoscopic treatment of endometriosis improves fertility prospects by 13% (Marcoux et al., 1997Go). Thus, in the final decision on the clinical value of HSG and laparoscopy, one should consider issues other than solely tubal pathology. However, such an analysis is beyond the scope of this study. When focusing on tubal pathology, we conclude that laparoscopy should not be considered as perfect in the diagnosis of tubal pathology. For clinical practice, we recommend that laparoscopy can be postponed until at least 10 months after a normal or one-sided abnormal HSG, whereas laparoscopy provides useful information immediately after a two-sided abnormal HSG.


    Acknowledgments
 
The authors acknowledge M.Y.Bongers, from the Department of Obstetrics and Gynaecology, St Joseph Hospital, Veldhoven, The Netherlands for critically reading the manuscript.


    Notes
 
5 To whom correspondence should be addressed at: Academic Medical Centre, Department of Clinical Epidemiology and Biostatistics, University of Amsterdam, PO Box 22700, 1100 DE Amsterdam, The Netherlands Back


    References
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
American Fertility Society (1985) Revised American Fertility Society classification of endometriosis: 1985. Fertil. Steril., 43, 351–352.[Medline]

Alatas, C., Aksoy, E., Akasu, C., et al. (1997) Evaluation of intrauterine abnormalities in infertile patients by sonohysterosonography. Hum. Reprod., 12, 487–490.[ISI][Medline]

Caspi, E., Halperin, Y. and Bukovsky, I. (1979) The importance of periadnexal adhesions in tubal reconstructive surgery for infertility. Fertil. Steril., 31, 296–300.[ISI][Medline]

Collett, D. (1994) Modelling Survival Data in Medical Research. Chapman & Hall, London.

Collins, J.A., Burrows, E.A. and Willan, A.R. (1995) The prognosis for live birth among untreated infertile couples. Fertil. Steril., 64, 22–28.[ISI][Medline]

Cox, D.R. (1972) Regression models and life tables. J. R. Stat. Soc. 34, 187–220.[ISI]

Gordts, S., Campo, R., Rombauts, L. and Brosens, I. (1998) Transvaginal hydrolaparoscopy as an outpatient procedure for infertility investigation. Hum. Reprod., 13, 99–103.[Abstract]

Land J.A., Evers, J.L.H. and Goossens, V.J. (1998) How to use Chlamydia antibody testing in subfertility patients. Hum. Reprod., 13, 1094–1098.[Abstract]

Lundberg, S., Wramsby, H., Bremmer, S. et al. (1997) Radionuclide hysterosalpingography does not distinguish between fertile women, before sterilization, and infertile women. Hum. Reprod., 12, 275–278.[Abstract]

Marcoux, S., Maheux, R. and Bérubé, S. (1997) Laparoscopic surgery in infertile women with minimal or mild endometriosis. N. Engl. J. Med., 337, 217–222.[Abstract/Free Full Text]

Mol, B.W.J., Dijkman, B., Wertheim, P. et al. (1997a) Chlamydial antibody titers in the diagnosis of tubal pathology; a meta-analysis. Fertil. Steril., 67, 1031–1037.[ISI][Medline]

Mol, B.W.J., Swart, P., Bossuyt, P.M.M. and Van der Veen, F. (1997b) Is hysterosalpingography an important tool in predicting fertility outcome? Fertil. Steril., 67, 663–669.[ISI][Medline]

Nordenskjöld, F. and Ahlgren, M. (1983) Laparoscopy in female infertility. Acta Obstet. Gynecol. Scand. 62, 609–615.[ISI][Medline]

Olsen, J., Juul, S. and Basso, O. (1998) Measuring time to pregnancy: methodological issues to consider. Hum. Reprod., 13, 1751–1753.[Free Full Text]

Spira, A. (1998) Measuring time to pregnancy: the use of fecundability in epidemiological surveys. Hum. Reprod., 13, 1753–1756.[ISI][Medline]

Strandell, A., Bryman, I., Janson, P.O. and Thorburn, J. (1995) Background factors and scoring systems in relation to pregnancy outcome after fertility surgery. Acta Obstet. Gynecol. Scand., 74, 281-287.[ISI][Medline]

Swart, P., Mol, B.W.J, Van der Veen, F. et al. (1995) The accuracy of hysterosalpingography in the diagnosis of tubal pathology, a meta-analysis. Fertil. Steril., 64, 486–491.[ISI][Medline]

Templeton, A., Morris, J.K. and Parslow, W. (1996) Factors that affect outcome of in-vitro fertilisation treatment. Lancet, 348, 1402–1406.[ISI][Medline]

Te Velde, E.R., Boer-Meisel, M.E., Meisner, J. et al. (1989) The significance of preoperative hysterosalpingography and laparoscopy for predicting the pregnancy outcome in patients with a bilateral hydrosalpinx. Eur. J. Obst. Gyn. Repr. Biol. 31, 33–45.

Van Kooij, R.J., Looman, C.W.N., Habbema, J.D.F. et al. (1996) Age-dependent decrease in embryo implantation rate after in vitro fertilization. Fertil. Steril., 66, 769–775.[ISI][Medline]

Submitted on June 15, 1998; accepted on January 14, 1999.