Exemfoamkit.co.uk

Human Reproduction vol.12 no.7 pp.1582–1588, 1997
The spontaneous pregnancy prognosis in untreated
subfertile couples: the Walcheren primary care study
H.K.A.Snick1, T.S.Snick1, J.L.H.Evers2,4 and
new forms of treatment, or to wait further for a spontaneous J.A.Collins3
pregnancy to occur. Clinical prediction models have beendeveloped in order to assist the clinician in estimating a 1Department of Obstetrics and Gynaecology, Ziekenhuis Walcheren, couple’s spontaneous baseline pregnancy chance (Cramer et al., Koudekerkseweg 88, 4382 EE Vlissingen, 2Department ofObstetrics and Gynaecology, Academisch Ziekenhuis Maastricht, 1979; Leridon and Spira, 1984; Comhaire, 1987; Bostofte and The University of Maastricht, P.O. Box 5800, 6202 AZ et al., 1993; Eimers et al., 1994; Collins et al., 1995). Prediction Maastricht, The Netherlands and 3Department of Obstetrics and models based on empirical observations are more useful to Gynecology and Department of Clinical Epidemiology and clinicians if the inception cohort is similar to the patients in Biostatistics, McMaster University, 1200 Main Street West, Room their practice and if there is an unambiguous outcome such as live birth (Collins et al., 1995). Ideally therefore, one should 4To whom correspondence should be addressed derive an empirical prediction model from data collected in a The spontaneous pregnancy prognosis of couples in a
group of subfertility patients, living in a well-defined geograph- primary care situation has never been studied. Prognostic
ical area and visiting a primary care fertility centre. Walcheren, models have been developed for referral populations only.
a former island in the Dutch province of Zeeland, appears to We wished to develop a prognostic model to estimate the
be almost ideally suited for this kind of research. The composi- likelihood of live birth and the impact of prognostic factors
tion of its population of 110 000 is a genuine reflection of the among untreated subfertile couples in a primary care
population of The Netherlands as a whole, both in demographic situation. With this aim, we conducted a cohort follow-up
and socio-economic respects. It has only one hospital, where study of 726 couples in the peninsula of Walcheren, a
one and the same gynaecologist (H.K.A.S.) has been respons- geographically isolated, but demographically and socio-
ible for all subfertility care delivered between 1985 and today.
economically representative area of an industrialized West-
Only persistent factors with a proven fertility-impairing effect ern society, The Netherlands. Of the Walcheren population,
are treated, and only if a treatment is available which is based 9.9% exhibit subfertility complaints at least once during
on sound scientific evidence. The lines of communication their lifetime. There were 201 live birth conceptions during
between general physicians and specialist care providers are 9915 months of untreated observation. The cumulative rate
short. For all patients the care they receive at the hospital of conceptions leading to live births was 52.5% when all
constitutes the first contact with specialist fertility investigation of the untreated observations were considered, and 72.0%
and treatment. It may be considered primary fertility care, in the subgroup of 342 couples who remained untreated
since general physicians do not start fertility treatment in the throughout their follow-up. The relevant prognostic factors
Walcheren region. The social situation on this former island in this primary care subfertility population were: abnormal
allows for an almost complete long-term follow-up.
post-coital test, tubal defect, ovulation defect, and duration
of subfertility. A prediction score based on these factors
would be accurate in ~76–79% of cases. Live birth prognosis
Materials and methods
can be estimated with sufficient accuracy to be useful in
Setting
counselling subfertility patients, and in planning clinical
management.
The former island of Walcheren is situated in the south-west of TheNetherlands. To the south the river Schelde flows. A ferry boat Key words: infertility/prognostic factors/prognostic model/ service provides a connection to the mainland of The Netherlands spontaneous pregnancy prognosis/subfertility and Belgium. To the west is the North Sea, and to the north a damconnects Walcheren to another small former island, Noord-Beveland,which does not possess a hospital. Finally to the east a dam givesaccess to the former island of Zuid-Beveland, with a hospital at Introduction
Recent years have brought numerous new treatment modalities Walcheren has 110 000 inhabitants, 24 200 of whom are females for subfertile couples, many of which have subsequently between 15 and 44 years of age (mean of the figures for 1990–93).
disappeared, but some of which appear to offer realistic 1319 births take place per year (mean of the figures for 1989–93).
improvements. All assisted reproduction treatments, apart from In 1994, 39 170 people were employed for Ͼ15 h a week. Per 1000 their medical advantages, also carry a medical, financial, inhabitants, the following numbers were active in the respective emotional and social burden. Therefore, clinicians are in need employment categories (numbers for The Netherlands in parentheses): of guidelines to counsel their patients whether to accept these agriculture and fishing 3.1 (6.5), industry, building, construction and European Society for Human Reproduction and Embryology Spontaneous pregnancy prognosis
transport 124 (118), trade and travel 65 (73), services 160 (170) activities until registration at the clinic; female and male partner’s (Source: Dutch Central Office of Demography Statistics, CBS, 1993).
age, age at the time of registration.
Only one hospital serves the whole population of Walcheren. The The diagnostic protocol was according to the recommendations of four consultant gynaecologists in the hospital see all gynaecological the ESHRE workshop on ‘unexplained infertility’ (Crosignani et al., and obstetric cases by themselves. There are no residents or house 1993; ESHRE, 1995), and included clinical examination of both officers employed. One gynaecologist (H.K.A.S.) coordinates all partners, one (i.e. the first) semen analysis according to WHO (1987) fertility investigations and sees all problem cases himself. He super- criteria, recording of a basal body temperature (BBT) chart, a mid- vises the fertility laboratory, and performs all morphology assessments luteal progesterone determination, a post-coital test (PCT), and in the semen analyses in person. The department of internal medicine/ hysterosalpingography (HSG). HSG was substituted by laparoscopy endocrinology and the department of urology refer all subfertility as the initial tubal factor investigation if the history was suggestive patients for evaluation to the senior author. The special interest in of a serious pelvic infection or if abnormal findings were encountered subfertility in the Walcheren hospital, and the availability of an at pelvic examination. If HSG was abnormal, a laparoscopy was done artificial insemination service with donor spermatozoa, cause an influx in one of the subsequent cycles. If HSG was normal, laparoscopy of patients from Noord- and Zuid-Beveland into the region. These patients, ~30 in number, were excluded from the present analysis as Semen analysis was judged normal if sperm concentration was their inclusion would have created an erroneously high incidence of Ͼ20ϫ106/ml, if grade A progressive motility was Ͼ25% and if the infertility complaints. General physicians in Walcheren do not refer percentage of normal morphology was ജ20% (WHO, 1987). A WHOsemen defect was diagnosed if one or more of these criteria were not subfertility patients to hospitals outside Walcheren. This enabled the met. Oligozoospermia was defined as Ͻ5ϫ106 motile sperm in the study of a primary population with fertility problems in a geographic- ally well-defined and demographically and socio-economically repres- The menstrual cycle was considered ovulatory if serum progesterone entative area of The Netherlands with an almost complete follow-up.
was Ͼ18 nmol/l and the duration of the menstrual cycle was Ͻ8weeks. An ovulation defect was diagnosed if one or both criteria Patients
Patients were admitted to the study if they wished to conceive and The PCT was considered positive if more than one forward-moving had a history of at least 1 year of unprotected intercourse without spermatozoon was found in the whole cervical mucus sample, between pregnancy, if no pregnancy existed at the moment of registration, and 8 and 16 h after intercourse. In patients with regular menstrual cycles if they were permanent inhabitants of Walcheren. Patients who had the PCT was scheduled on day 15 or 16 before the expected onset had a sterilization were excluded, as were patients from outside of the next menstrual period. In patients with irregular cycles the PCT was timed after tracking the appearance of the cervical mucus.
Between 1 January 1985 and 31 December 1993, 726 couples As soon as the mucus was considered to reflect the preovulatory fulfilled these criteria, for a mean of 81 couples per year. This means state, the couple was advised to have intercourse. Clear cervical that in a 30 year reproductive life span (ages 15–44), if this number mucus, with Ͼ6 cm spinnbarkeit, without leukocytes, was considered were to remain stable, 2430 couples would seek fertility care, or preovulatory. The PCT was repeated every 48 h until the BBT chart 9.9% of the population (Hull et al., 1985). Follow-up ended on 1 showed a persistent rise and/or (from 1988 onwards) the dominant September 1995. Of 724 couples (99.7%), the reproductive outcome follicle (ജ20 mm) had disappeared on ultrasound examination. A was available at that moment, either by direct contact with the couple, PCT defect (abnormal PCT) was diagnosed if the best PCT showed when they still frequented the clinic, or by their completing a written 0 or 1 forward-moving spermatozoon. Data analysis of PCT by χ2- questionnaire, or by the senior author calling them or their general test indicated that this test performs much better as a predictor of physician by telephone. Only two couples were lost to follow- infertility in a primary care population than in the patient groups from which prognostic models are more usually derived.
During the period of investigation, 94 patients qualified for in- Endometriosis was scored according to the revised American vitro fertilization (IVF) treatment and were referred to an IVF centre Fertility Society classification (1985). Only stages 3 and 4, or cysts (outside Walcheren). Another 19 couples were referred for further Ͼ6 cm diameter were taken into account in defining endometriosis analysis or specialized treatment to referral centres outside the region: in the present investigation. Laparoscopy was performed according for microsurgical repair of epididymal (n ϭ 8) or tubal (n ϭ 8) to the ESHRE (1995) guidelines for the investigation of the infertile obstruction, or for specialized diagnosis of andrological problems couple. This implies that an unknown proportion of patients with (n ϭ 2). One couple was referred to a centre outside Walcheren for endometriosis (undefined by stage, but unlikely to harbour stage 3 or gonadotrophin treatment of difficult ovulatory disturbances.
4 disease) may exist within the unexplained infertility group. This is During the period of investigation, seven couples sought a second typical of most studies based on similar investigative protocols.
opinion in a hospital outside Walcheren without referral by the A normal tubal status was diagnosed as no abnormalities on HSG gynaecologist concerned. None of them became pregnant. No patient and/or laparoscopy: no obstruction of either tube, no fertility-impairing was seen for obstetric care at the Walcheren hospital after infertility peritubal adhesions. A tubal defect was diagnosed if one or more of investigation or treatment elsewhere without referral. Notwithstanding the Dutch system of home deliveries, of all pregnant patients in Pregnancy was defined as amenorrhoea of ജ6 weeks in combination Walcheren 80% are seen at some time during their pregnancy by one with a positive urine pregnancy test, or positive ultrasound findings.
For a diagnosis of abortion or ectopic pregnancy to be made, surgicalpathology confirmation was required. The product of conception was Definitions and management
defined as a child from a pregnancy duration of 28 weeks onwards.
The following definitions were used: infertility, no conception after For the purpose of these analyses, success was defined as live birth; 1 year of unprotected intercourse; primary infertility, no prior preg- cases in which unsuccessful pregnancy occurred were considered nancy in this partnership; secondary infertility, a prior pregnancy in failures and observations were censored at the time of the last this partnership, regardless of the pregnancy outcome; duration of menstrual period. A live birth was defined further as a living child 1 infertility, the interval in months from discontinuation of contraceptive H.K.A.Snick et al.
for minimal endometriosis by a locum tenens gynaecologist, onceclomiphene citrate was prescribed to a woman with a normal ovulatorymenstrual cycle.
To concentrate on untreated conceptions, all calculated rates excluded conceptions that occurred if either partner was treated afterregistration. For live birth calculations, all couples were included inthe denominator, and the numerator was the number of untreatedconceptions. In the life-table and approximate fecundity calculations,the observation months among untreated couples were combined withthe observation months before the first treatment among treatedcouples. Cumulative live birth rate was the life-table estimate of timeto conception in cases associated with live births. Approximatefecundity was the number of live births per 100 months of untreated Figure 1. Cumulative rate of live births (and 95% CI) among 342
untreated couples (upper curve) and 726 couples (lower curve). Theupper curve excludes observations prior to treatment among the 384 Statistical analysis
Prognostic variables were evaluated by means of Pearson’s χ2(categorical variables) or analysis of variance (continuous variables).
The strategy for selecting variables to include in the proportionalhazards analysis was based on these univariate analyses (Cox, 1972),and on life-table analysis of live birth during follow-up. Termsrepresenting pregnancy history, duration of infertility, female age,and diagnosis were considered. Binary variables for duration (cut-offpoint at 24 months) and female age (cut-off point at 30 years) wereconstructed in order to estimate relative risks. Semen analysis wasentered as normal or abnormal depending on published criteria forsperm concentration, progressive motility and normal morphology(WHO, 1987). Semen variables also were used to calculate totalmotile sperm count per ejaculate. The significance level for enteringcovariates into the proportional hazards analysis models was 0.05,and for removing covariates 0.051.
Figure 2. Receiver operating characteristics curve for the
The prediction score was computed from the β coefficients of the prediction of live birth based on the proportional hazards model in terms in the model. In order to assess the accuracy of the prediction Table III. Closed squares are the observed operating points and the model, receiver-operating characteristics (ROC) curves were estimated smoothed line is the estimated binormal receiver-operating for the prediction of live births in the data, and the area under the curve (AUC) with 95% confidence intervals (95% CI) was calculated.
The reproducibility of the prediction score was evaluated by applyingthe model to a previously published independent set of data (Collins Table I. Live birth and other outcomes among 726 infertile couples by
Observation and outcome data
The mean duration of infertility was 20.7 (SD 13.9) months and the mean age was 29.1 (SD 4.5) years for 726 female partners and 31.2 (SD 5.3) for 511 male partners (the latter being collected only from December 1988 onward). Twenty- nine per cent of the couples had secondary infertility defined as infertility after a previous pregnancy in the partnership andin a further 7% the female partner had a pregnancy in a All treatment was ‘evidence-based’. As a rule of practice, only previous partnership. The status of all patients at 1 September persistent factors with a proven fertility-impairing effect were treated, 1995 is shown in Table I. The mean time under observation and only if a treatment was available which was based on sound or treatment was 24.5 months (SD 27.6, median 15.0, range scientific evidence. The only exception to this rule was embolization 1–124.9). Of the 495 pregnancies, 411 were live births; these of a varicocele grade III, which was performed in couples with live birth conceptions occurred after a mean of 13.2 months severe sperm abnormalities after ജ2 years of otherwise unexplained of observation before and during treatment (SD 15.5, median Patients’ preferences or requests for treatments other than those There were 342 (47%) untreated couples, and they contrib- based on scientific evidence were not honoured. The patient was uted 5918 months of observation (mean 17.3, SD 22.5, median counselled extensively on the findings instead. As a consequence, onan individual basis, apart from embolization of a varicocele, unproven 9.0, range 1–120 months). There were also 3997 months of treatment was used only twice: once danazol was prescribed observation before treatment for 384 treated couples (mean Spontaneous pregnancy prognosis
Table II. Live birth rate, cumulative live birth rate and approximate fecundity by diagnosis group
10.4, SD 15.2, median 5.0, range 1–124 months). The resulting Table III. Relative likelihood of live birth among 726 infertile couples
9915 months of untreated observations accounted for 56% of the total of 17 797 months of observations. The untreatedcouples had 239 conceptions and 201 live births; the mean time to conception for untreated couples who had a live birth was 8.1 months (SD 10.0, median 4.0, range 1–73 months).
The 201 live births among the 342 untreated couples accounted The cumulative rate of conceptions leading to live birth was 52.5% (95% CI 44.7–60.2) when all of the untreated observations were considered and 72.0% (95% CI 65.0–79.0) in the subgroup of couples who remained untreated throughout Effect of prognostic factors on live birth
Live birth rate, cumulative live birth rate at 36 months Data from follow-up in the Walcheren region. Model statistics were: and approximate fecundity were highest in couples with selected model, χ2 ϭ 88.129, 4 degrees of freedom, P Ͻ 0.0001; alternate unexplained infertility (Table II). These rates also were high model, χ2 ϭ 70.106, 6 degrees of freedom, P Ͻ 0.0001. The endometriosis in the less severe subgroup of the male defect diagnostic variable did not enter either model.
PCT ϭ post-coital test.
category. Live birth rates also were high in couples withshorter duration of infertility and secondary infertility and lowin couples with an abnormal PCT. Female age was not Table IV. Area under the curve for receiver-operating curve (ROC) analyses
significantly associated with live birth rates among these for prediction scores and observed live births in untreated infertile couples A proportional hazards analysis to evaluate the independent contribution of prognostic factors included four covariates inthe final model (Table III). The relevant variables are PCT result, tubal status, ovulation status and duration of infertility respectively, in order of importance. This model is referred to as the ‘selected model’. The PCT result variable entered the model, while variables defining semen defect by World Health aWalcheren: the present data; CITES: untreated prognosis among 2198 Organization standards (WHO, 1987) or by the presence of a couples in the Canadian Infertility Therapy Evaluation Study (Collins et al., total motile count Ͻ5ϫ106 per ejaculate did not enter the model. An ‘alternate model’ was developed in which the PCTresult variable was not offered to the analysis. The alternate model would be accurate in 79% of cases. AUC values Ͻ50% model included three additional variables, one representing imply an inoperative model, and 100% would constitute a male function (the WHO semen variable), one representing perfect prediction model. Confidence limits for areas under female function (female partner’s age Ͻ30 years) and one the ROC curves indicate that the small observed difference representing undefined couple factors (secondary infertility).
between the selected (four-variable) model and the alternate(six-variable) model is not significant (Table IV). The perform- Evaluation of the prediction score
ance of the prediction model was validated in an independent The prediction score based on the proportional hazards model data set which has been published previously (Collins et al., in the sample of 726 couples was evaluated by ROC analysis 1995). Because the comparison data did not include PCT (Figure 2). The AUC indicates that the selected prediction results, the alternate model was used for this validation test.
H.K.A.Snick et al.
The AUC was 67% when the alternate model was evaluatedagainst observed live birth rates in the previously publisheddata set. The AUC with the previously published score waslower when it was applied either in the original group ofCanadian couples or in the present data set.
Discussion
Only large prospective studies of subfertile couples visitingprimary care fertility centres with a long-term (and complete)follow-up may shed light on the baseline pregnancy prognosisin untreated subfertility. Studies from tertiary care facilities Figure 3. Cumulative live birth rate with 95% CI for infertile
may present less reliable data in this respect since their patient couples in Walcheren (n ϭ 726) and in the Canadian Infertility populations may be selected, and their composition skewed Therapy Evaluation Study (CITES, n ϭ 2198), truncated after 2years of observation for this figure only.
towards those with as yet undiscovered — but persistent —fertility-impairing factors. Many new treatment options forsubfertile couples have been introduced in recent years. Only one calculated from secondary and tertiary care populations some of them have been subjected to controlled trials in which (e.g. 21.2% in the Canadian Infertility Therapy Evaluation one group was randomly allocated to receive a placebo or no Study (CITES; Figure 3), reflecting a different composition of treatment. On the other hand, only in the minority of cases, the respective study populations, notwithstanding their identical i.e. those associated with azoospermia, prolonged amenorrhoea inclusion criteria of subfertility. When applying models, this or bilateral tubal obstruction, can the cause of infertility be difference should impact on the decision about when to resort identified easily. For the remaining couples, the interpretation of diagnostic test results is not straightforward, and the choice Proportional hazards analysis showed the likelihood of live of treatment often lacks a rationale. Growing concern exists birth during untreated observation to be influenced predomi- regarding the potential disadvantages, and even risks, of some nantly by three diagnostic variables (abnormal PCT, tubal of the newer treatment modalities, especially since they appear defect and ovulation defect) and by one clinical variable to be applied ever more liberally in patients whose fertility (duration of infertility Ͻ24 months). These four variables may sometimes be only marginally impaired. Under such were included in the so-called ‘selected model’ (Table III). If circumstances (often counterproductive) overtreatment is abnormal PCT was not offered to the model, its place could be taken by abnormal semen (WHO, 1987) or oligozoospermia The Walcheren follow-up study presents data collected in (total motile count Ͻ5ϫ106), but if all three variables were 726 couples in the course of a 9 year review period of primary available, only abnormal PCT entered. Even when we used a fertility care in the only hospital of a geographically isolated, cut-off point for total motile count at Ͻ 1ϫ106 per ejaculate but demographically and socio-economically representative [identifying 91 (12.5%) of cases as abnormal] its contribution area of an industrialized Western country, The Netherlands.
was only marginally significant. The presence of this variable The 726 couples represent 9.9% of the Walcheren population, reduced the significance of the abnormal PCT variable, so that seeking medical care for subfertility problems at least once it was associated with P ϭ 0.0089 and a χ2 to remove equal during their reproductive life span. Strict diagnosis and treat- to 6.83. When abnormal PCT was not offered to the model, ment protocols were adhered to, and only evidence-based the information that it contributed seems to have been replaced treatment was instituted. If no such treatment was available by one male variable (WHO semen defect), one female variable for the couple under consideration, they were counselled (age Ͻ30 years), and one couple variable (secondary infertility) extensively on the findings instead. Management was expectant (‘alternate model’; Table III). In the Walcheren data, median in such cases. This allows for the calculation of the baseline duration of infertility is 16 months, compared with the Canadian pregnancy prognosis in untreated subfertility couples from a data where the median was 36 months. In our group, diagnostic representative population in a developed country. The couples assessments have much more powerful effects on the prognosis, described in this study had a shorter duration of infertility and indefinite variables are less important (female age, second- (mean 20.7 Ϯ 13.9 months) than in most published studies, ary infertility and duration of infertility simply indicating the which may be explained by the fact that most other studies presence of unknown factors). Apparently, as the duration of considered patients from referral institutions, whereas our infertility increases, factors which are as yet undiscovered patients visited the Walcheren hospital for their initial fertility become more important than known factors in estimating work-up. Given the short lines of communication between general physicians and specialist care providers in Walcheren, Only a handful of prognostic models have yet accrued the regional organization of fertility care, and the demographic adequate evidence of accuracy, generality, and effectiveness characteristics of the population in the present investigation, (Anonymous, 1995). The prediction model based on the we propose our study to reliably reflect baseline fertility observations in the Walcheren population was tested by means prognosis in untreated couples. This baseline prognosis is of ROC curves and AUC. It was then validated in an independ- much better (2 year cumulative live birth rate 41.9%) than the ent previously published population, the CITES group of Spontaneous pregnancy prognosis
subfertility patients (Collins et al., 1995). To allow for evaluat- Table Va. Average baseline prognosis of live birth
ing our prediction model in the CITES group, which did not (ABPLB) after 3, 6, 12, 24 and 36 months of require PCT results to be collected, the PCT parameter was made unavailable, and this resulted in our alternate (six- variable) model (Table III), which included the WHO semendefect parameter. The predicted likelihood of live birth derived from our alternate (six-variable) prediction model was com- pared with the observed births in the Walcheren and the Canadian patient groups respectively. These live birth event rates were submitted to an ROC analysis program generatingROC curves and AUCs plus their 95% CI (Table IV): theAUC was 76% when the Walcheren alternate model was Table Vb. Effect of prognostic factors, expressed in
applied to the Walcheren data, and, significantly worse, 67% multiplication factors (MF) of the baseline prognosis when our model was applied to the Canadian data. It should be remembered that the Canadian model applied to the Canadian data rendered an AUC of 62% in a split training/ validation sample set-up (Collins et al., 1995). The Canadian model applied to the Walcheren data reached 65%, indicating that the probability of the Canadian model correctly predicting which of two Walcheren couples will conceive first is 65%, as compared to 67% when our model is used in two Canadian couples. There was no significant difference between the AUC of our selected (four-variable) model, 79%, and of our alternate (six-variable) model, 76%, when applied to our own data.
In the 1995 CITES publication, ovulation defect was not Live birth rate prognosis after 3 to 36 months ϭABPLBϫMF(s).
selected into the model, whereas ovulation defect was asignificant component of both the selected and alternateWalcheren models. As noted above, the duration of infertility of a pregnancy, would be a frequent reason for couples in the Canadian couples was longer than in the Walcheren selectively dropping out of prediction studies, this might reduce couples. Ovulation defects are easily treated in primary care, the accuracy of the models derived from these studies to and only the most difficult cases would be referred. Therefore predict spontaneous live births. In contrast to many studies a lower proportion of the Canadian ovulation defects would reporting prediction models, in the present study loss to follow- be likely to resolve spontaneously and lead to untreated up was virtually non-existent (0.3%).
The present study offers a model for the prediction of live It has already been mentioned that the couples described in birth among untreated subfertile couples, seeking reproductive this study had a shorter duration of infertility than those in health care for the first time, in a developed western country.
the Canadian study. With longer duration of infertility, the In order to allow prediction for an individual couple, we most fertile couples will have conceived and the most easily constructed a simplified model for the clinical prediction of treatable disorders will have been successfully treated. These average cumulative live birth rate for various future time couples therefore do not come to the secondary or tertiary periods and prognostic factors (Tables Va and Vb). This model care clinics that usually publish prediction models and do not allows estimation of the fertility prognosis of a couple by contribute to the data set published from these clinics. This multiplying their baseline prognosis by the relative hazard for will have profound effects on the composition of the study each clinical predictor present in that particular couple. For population, on the conception figures in this group and on the example, a couple with primary infertility of 1 year duration, prediction models derived from data obtained in this group.
a female partner aged 28 years, a normal PCT, and oligozoo- Those not conceiving by a median of 36 months are less likely spermia has an estimated 12 month cumulative live birth to conceive. This group with persistent infertility is more likely rate equivalent to 41.1% (27.4%ϫ1.5) in the selected (four- to harbour unknown factors which are as yet undiscovered.
variable) model, and to 34.5% (27.4%ϫ1.5ϫ1.4ϫ0.6) in the These factors are relatively more important in the untreated alternate (six-variable) model. In this way, the prognostic prognosis than the known factors, but they are unknown and model presented here combines important items of patient data therefore are not included in the prediction score. Thus the to predict untreated clinical outcome in patients with impaired prediction score developed in the Canadian long-duration fertility. The model is of potential value in counselling patients, group explains less of the observed live birth prognosis in the in deciding on their referral for secondary care, in estimating short-duration Walcheren population seeking primary care the effect of treatment, and in selecting uniform comparison groups of patients for clinical trials. Also, if no results from Another factor which may affect reliability of prediction randomized clinical trials are available, the outcome of a models is the proportion of patients lost to follow-up. If the particular treatment modality can be compared to the approxi- occurrence of a pregnancy, or for that matter the non-occurrence mated spontaneous pregnancy rate in the group of subfertility H.K.A.Snick et al.
patients studied. In our opinion, the most important asset ofthe present study is that it offers the possibility to estimatelive birth rates among untreated subfertile couples from aprimary care environment with sufficient accuracy to be usefulin the clinical management of subfertility patients by primaryfertility care providers. The difference between the resultsfrom the present study and those from academic fertility centrespublished before, underlines the importance of assessing thesimilarity of the inception cohort of the study to the patientsin one’s own practice.
Acknowledgement
F.H.Comhaire MD, Department of Internal Medicine, State UniversityHospital, Ghent, Belgium, is acknowledged for his help in thepreparatory phase of this project.
References
American Fertility Society (1985) Revised classification of endometriosis.
Fertil. Steril., 43, 351–352.
Anonymous (1995) Commentary: Prognostic models: clinically useful or quickly forgotten? Br. Med. J., 311, 98–100.
Bostofte, E., Bagger, P., Michael, A. et al. (1993) Fertility prognosis for infertile couples. Fertil. Steril., 59, 102–107.
Collins, J.A., Burrows, E.A. and Willan, A.R. (1995) The prognosis of live birth among untreated infertile couples. Fertil. Steril., 64, 22–28.
Comhaire, F.H. (1987) Simple model and empirical method for the estimation of spontaneous pregnancies in couples consulting for infertility. Int. J.
Androl., 10, 671–680.
Cox, D.R. (1972) Regression models and life tables. J. R. Statist. Soc., 34,
Cramer, D.W., Walker, A.M. and Schiff, I. (1979) Statistical methods in evaluating the outcome of infertility therapy. Fertil. Steril., 32, 80–86.
Crosignani, P.G., Collins, J., Cooke, I.D. et al. (1993) Recommendations of the ESHRE workshop on ‘unexplained infertility’. Hum. Reprod., 8,
977–000.
Eimers, J.M., Te Velde, E.R., Gerritse, R. et al. (1994) The prediction of the chance to conceive in subfertile couples. Fertil. Steril., 61, 44–52.
ESHRE (European Society for Human Reproduction and Embryology) (1995) Investigation and treatment of infertile couples: ESHRE guidelines for good
clinical and laboratory practice. Hum. Reprod., 10, 1246–1271.
Hull, M.G.R., Glazener, C.M.A., Kelly, N.J. et al. (1985) Population study of causes, treatment, and outcome of infertility. Br. Med. J., 291, 1693–1697.
Leridon, H. and Spira, A. (1984) Problems in measuring the effectiveness of infertility therapy. Fertil. Steril., 41, 580–586.
World Health Organization (1987) Laboratory Manual for the Examination of Human Semen and Semen–Cervical Mucus Interaction. CambridgeUniversity Press, Cambridge.
Received on January 6, 1997; accepted on May 6, 1997

Source: http://www.exemfoamkit.co.uk/pdf/Encl%207_BC%20ref2_Snick%20et%20al_Human%20Reprod.pdf