Advertisement

Distress and reduced fertility: a follow-up study of first-pregnancy planners

      Abstract

      Objective: To assess the effect of psychological distress on time to first pregnancy.
      Design: A follow-up study of time to pregnancy with prospective data on distress, with controlling for potential confounding variables.
      Setting: Two university hospitals.
      Patient(s): Danish couples (n = 430) who were planning their first pregnancy and had no previous reproductive experience were followed for six menstrual cycles. Psychological distress was measured in each menstrual cycle by the General Health Questionnaire.
      Intervention(s): None.
      Main Outcome Measure(s): A clinically recognized pregnancy or a biochemical pregnancy detected in urine samples from each period of vaginal bleeding.
      Result(s): For cycles with the highest distress score (General Health Questionnaire score >80th percentile), the probability of conception per cycle was 12.8%, compared with 16.5% in other cycles (adjusted odds ratio [OR] 0.6; 95% confidence interval [CI] 0.4–1.0). The effect of distress was found almost exclusively among women with long menstrual cycles (OR 0.1; 95% CI 0.01–0.4 and OR 0.9; 95% CI 0.5–1.4 for women with cycles of ≥35 and <35 days, respectively). An increased incidence of early embryonal loss was also found among highly distressed women with long cycles, but was based on a small number of observations.
      Conclusion(s): Psychological distress may be a risk factor for reduced fertility in women with long menstrual cycles.

      Keywords

      The coexistence of infertility and psychosocial distress is supported by sound scientific evidence (
      • Wright J
      • Allard M
      • Lecours A
      • Sabourin S
      Psychosocial distress and infertility a review of controlled research.
      ,
      • Seibel M.M
      Emotional aspects of infertility.
      ). However, the causal direction may well be from infertility to psychological distress (
      • Boivin J
      • Takefman J.E
      Stress level across stages of in vitro fertilization in subsequently pregnant and nonpregnant women.
      ,
      • Lalos A
      • Lalos O
      • Jacobsson L
      • von Schoultz B
      The psychosocial impact of infertility two years after completed surgical treatment.
      ), and only a longitudinal study of distress can separate cause from effect. To describe the association between psychological distress and fertility, investigators must collect information on psychosocial distress before knowledge about the outcome and carefully determine possible confounding factors, including the timing and frequency of sexual intercourse. Furthermore, to avoid selection bias, exclusion of couples with known or suspected impairment of fertility is essential. None of the existing studies fulfill these criteria. We studied the association between distress and fertility in a follow-up study of couples with no a priori knowledge about their reproductive capability.

      Materials and methods

       Population

      From 1992 to 1994, a total of 430 couples were recruited after nationwide mailing of a personal letter to 52,255 trade union members (metal workers, office workers, retail clerks, nurses, and day-care workers) who were 20–35 years old, were living with a partner, and had no children. Couples were eligible only if they had no previous reproductive experience and intended to discontinue contraception to become pregnant. A detailed description of the cohort and design is given elsewhere (
      • Bonde J.P.E
      • Hjollund N.H.I
      • Jensen T.K
      • Ernst E
      • Kolstad H.A
      • Henriksen T.B
      • et al.
      A follow-up study of environmental and biologic determinants of fertility among 430 Danish first pregnancy planners design and methods.
      ). All subjects gave written informed consent, and the study was approved by the Danish Scientific Ethics Committee.
      Of the original cohort of 430 couples, 37 couples were part of a pilot study and were excluded because information on distress was not collected prospectively. Menstrual cycles with sexual abstinence from days 11 to 20 failed to result in a pregnancy and were excluded (n = 46 cycles). The analyses included 393 couples, contributing a total of 1,475 menstrual cycles (Table 1).
      TABLE 1Characteristics of menstrual cycles in a follow-up study of psychological distress and time to pregnancy (n = 393 couples).
      GHQ = General Health Questionnaire; QoL = Quality-of-Life scale.
      VariableCycleTotal
      123456
      Menstrual cycles in the analysis
      Cycles with sexual abstinence from day 11 to day 20 not included (n = 9, 6, 17, 7, 6, and 1 in cycles 1–6, respectively).
      3843242472111651441,475
      Termination of follow-up
      Pregnancy585840372218233
      Stop of attempts after completed cycle58610433
      GHQ/QoL score (mean values; range 0–36)
      Score from the current cycle not available in 165, 80, 46, 80, 57, and 66 cycles in cycles 1–6, respectively. Four couples became pregnant despite contraception and were assigned a waiting time of one menstrual cycle.
      11.211.811.912.112.212.111.8
      legend GHQ = General Health Questionnaire; QoL = Quality-of-Life scale.
      Cycles with sexual abstinence from day 11 to day 20 not included (n = 9, 6, 17, 7, 6, and 1 in cycles 1–6, respectively).
      Score from the current cycle not available in 165, 80, 46, 80, 57, and 66 cycles in cycles 1–6, respectively. Four couples became pregnant despite contraception and were assigned a waiting time of one menstrual cycle.

       Enrollment and follow-up

      The couples were enrolled when they discontinued birth control and were observed for six menstrual cycles or until pregnancy was recognized by their family physician. The couples were enrolled at two centers: the Department of Occupational Medicine in Aarhus (west center) and the Department of Growth and Reproduction in Copenhagen (east center). At enrollment, both partners filled out a questionnaire about psychological distress and demographic, medical, reproductive, occupational, and lifestyle factors.
      On the 21st day of each menstrual cycle during follow-up, the couples completed questionnaires on psychological distress and changes in occupational exposure and lifestyle factors. The women recorded vaginal bleeding and sexual intercourse in a structured diary developed for the study. Mean length of the menstrual cycle was calculated as the arithmetic mean of all observed complete cycles. Women who became pregnant in the first cycle were assigned a mean menstrual length based on information in the enrollment questionnaire. First-morning urine samples from days 1 to 10 in each menstrual cycle, starting on the first morning with menstrual bleeding, were stored for determination of hCG.

       Measurement of psychological distress

      Our primary psychometric tool was the General Health Questionnaire (
      • Goldberg D.P
      ). The General Health Questionnaire has been used extensively to define psychological distress (
      • Bech P
      Quality of life in psychosomatic research.
      ). The content validity and concurrent validity of the 30-item questionnaire (General Health Questionnaire–30) have been tested in a Danish population (
      • Hedegaard M
      • Henriksen T.B
      • Sabroe S
      • Secher N.J
      Psychological distress in pregnancy and preterm delivery.
      ). This version was completed by both partners at the start of the follow-up period. To optimize feasibility during follow-up, a shorter version of the General Health Questionnaire was included in the monthly questionnaires (see )). This short version, the General Health Questionnaire/Quality of Life, is a 12-item subscale of the General Health Questionnaire covering elements of fatigue, anxiety, and depression; it has been developed and validated by Bech (
      • Bech P
      ) and Loldrup et al. (
      • Loldrup D
      • Langemark M
      • Hansen H.J
      • Bonnevie O
      • Elsborg L
      • Olesen P
      • et al.
      The validity of the melancholia scale (MES) in predicting outcome of antidepressants in chronic idiopathic pain disorders.
      ).
      The questionnaire was completed in each cycle on day 21 from the first day of the previous menstrual bleeding, i.e., before the women knew whether the attempt to become pregnant had been successful. To encourage completion of the questionnaire, each couple received a mailed reminder on day 16 of each cycle. Each item on the General Health Questionnaire was scored from 0 to 3 (0 = low distress; 3 = high distress), and total scores were calculated as the sum of item scores.

       Early embryonal loss

      Urinary hCG was measured in a time-resolved immunofluorometric assay for serum hCG (Delfia, Wallac, Finland). The method was validated for measurement of intact hCG in the urine (
      • Bonde J.P.E
      • Hjollund N.H.I
      • Jensen T.K
      • Ernst E
      • Kolstad H.A
      • Henriksen T.B
      • et al.
      A follow-up study of environmental and biologic determinants of fertility among 430 Danish first pregnancy planners design and methods.
      ). When one of the values during a period was >1 IU, followed by a decline, the woman was regarded as having an early embryonal loss (
      • Bonde J.P.E
      • Hjollund N.H.I
      • Jensen T.K
      • Ernst E
      • Kolstad H.A
      • Henriksen T.B
      • et al.
      A follow-up study of environmental and biologic determinants of fertility among 430 Danish first pregnancy planners design and methods.
      ). Human chorionic gonadotropin–detected pregnancies were included in analyses of spontaneous abortions, but not in analyses of fecundability, which were based on clinically recognized pregnancies only.

       Analysis and statistical methods

      The cycle-specific General Health Questionnaire/Quality of Life scores were missing, invalid, or completed after cycle day 27 in 33% of the cycles (n = 486). The scores in these cycles were assigned the values from the previous cycle. Our analyses were performed before and after exclusion of these cycles. Primarily, the outcomes of all cycles (pregnant/not pregnant) were compared across quintiles of General Health Questionnaire/Quality of Life scores (percentiles 0–20, 21–40, 41–60, 61–80, and 81–100).
      Secondarily, in a longitudinal approach, each woman served as her own reference, and each cycle was assigned the difference in General Health Questionnaire/Quality of Life score from the score in her initial cycle, as well as from the mean score of all her cycles. Women with only one observed cycle were excluded in these analyses because no differences could be calculated.
      The unit of observation was a menstrual cycle, and pregnancy was defined as a clinically recognized pregnancy. Time to pregnancy is a true discrete biologic phenomenon because ovulation takes place only once during a menstrual cycle. Therefore, a logistic regression is equivalent to discrete survival analysis (
      • Weinberg C.R
      The influence of occupational activity on the menstrual cycle and fecundability.
      ,
      SAS Institute Inc
      ), and the association between psychological distress and fertility was estimated by a logistic regression model while controlling for confounding factors (
      • Hosmer D.W.J
      • Lemeshow S
      ,
      SAS Institute Inc
      ). This model provides a fecundability odds ratio (OR), which is the estimated odds of conception in menstrual cycles with high distress divided by the corresponding odds for cycles with low distress.
      Menstrual cycles were numbered consecutively after discontinuation of contraception, and cycle numbers were entered as dummy variables to adjust for changes in baseline fecundability with increasing number of cycles. Potential confounding variables were selected by their biologic relevance, regardless of the association with exposure in the present data set.
      Mean menstrual cycle length and frequency of sexual intercourse may be determinants of fertility but can also be intermediate steps in a causal chain between psychosocial distress and fertility. In the absence of firm evidence on this issue, we performed the analyses with and without these factors. Recent users of oral contraceptives (OC) are less fertile in the first cycles after discontinuation (
      • Burkman Jr, R.T
      Noncontraceptive effects of hormonal contraceptives bone mass, sexually transmitted disease and pelvic inflammatory disease, cardiovascular disease, menstrual function, and future fertility.
      ). To account for an interaction between cycle number and OC use, we assigned a dummy variable indicating OC use for each of the six cycles. The interaction between distress and other variables was evaluated by stratified analyses (
      • Rothman K.J
      • Greenland S
      ).
      Urine samples for hCG analysis were obtained in 864 cycles. In an additional 159 cycles, the women became clinically pregnant, but were equipped with sampling devices that would have been able to collect urine had vaginal bleeding occurred. The proportions of pregnancies terminated by an early spontaneous abortion [early abortion/(early abortion + clinical pregnancy)] were compared for different categories of General Health Questionnaire/Quality of Life score in logistic regression analyses.

      Results

      The mean (±SD) General Health Questionnaire–30 score for the women at the start of follow-up was 19.9 ± 7.6. The mean scores of the General Health Questionnaire/Quality of Life changed only slightly during follow-up (Table 1), and the scores did not increase during the waiting time among women who did not become pregnant during six cycles. Characteristics of the participants according to the General Health Questionnaire–30 scores are listed in Table 2. No consistent differences were found in the proportions of women with a high distress score at enrollment with respect to lifestyle factors, age, education, or previous contraception method.
      TABLE 2Characteristics of the population at the time of enrollment dichotomized at the 80th percentile of scores on the 30-item version of the General Health Questionnaire (n = 393 couples).
      CharacteristicTotalPercentage of women with high score (>24) (n = 79)
      Center
      Copenhagen19720
      Aarhus19620
      Trade union of index person
      Metal workers15719
      Office and commercial workers12325
      Nurses9415
      Day-care workers1922
      Education
      ≤10th grade1421
      High school1011
      Vocational education19116
      Further education17824
      Female age (y)
      ≤2413615
      25–2922923
      ≥302819
      Female body mass index (kg/m2)
      ≤198024
      20–2422619
      ≥258420
      Unknown30
      Smoking (no. of cigarettes per day)
      027222
      1–94613
      10–196113
      ≥201436
      Caffeine intake (mg/d)
      0–14912118
      150–2999723
      300–69914220
      ≥7003318
      Alcohol consumption (no. of drinks per week)
      010321
      1–415521
      5–98322
      ≥105212
      Mean menstrual cycle length (d)
      20–24729
      25–2919117
      30–3411423
      35–394123
      40–441421
      ≥451218
      Unknown1421
      Safe contraceptive method
      Regular use of intrauterine device or OCs for the last 12 months.
      Yes25621
      No13718
      OCs as the last method
      Yes13219
      No26121
      Frequency of sexual intercourse (no. of times per week)
      Mean midcycle value in the first cycle.
      0.1–0.93719
      1.0–1.99420
      2.0–2.910220
      3.0–3.94721
      ≥4.03816
      Unknown7520
      Male or female urogenital disorder
      Yes
      Female: salpingitis, ovarian cyst, gonorrhea, perforated appendicitis, Crohn’s disease, hormonal disorder, amenorrhea, or hormonal treatment. Male: mumps in adult age, epididymitis, gonorrhea, only one testicle, testicular cancer, varicocele, or sperm cyst.
      7129
      No32218
      Partner’s sperm density (×106/mL)
      0–104126
      11–203129
      21–6014417
      61+17019
      Unknown729
      Regular use of intrauterine device or OCs for the last 12 months.
      Mean midcycle value in the first cycle.
      Female: salpingitis, ovarian cyst, gonorrhea, perforated appendicitis, Crohn’s disease, hormonal disorder, amenorrhea, or hormonal treatment. Male: mumps in adult age, epididymitis, gonorrhea, only one testicle, testicular cancer, varicocele, or sperm cyst.
      A total of 233 of the 393 women (59%) had a clinically recognized pregnancy within the 6 cycles of follow-up (Table 1). When women had cycles with a General Health Questionnaire/Quality of Life score above the 80th percentile, the fecundability based on cycle-specific scores was 12.8%, compared with 16.5% in all other cycles (crude OR 0.73, 95% confidence interval [CI] 0.50–1.17; adjusted OR 0.6, 95% CI 0.4–1.0; P<.05) (Table 3). The fecundability of the five General Health Questionnaire/Quality of Life categories decreased slightly with increasing distress score.
      TABLE 3Crude and adjusted ORs for pregnancy per menstrual cycle by scores on the General Health Questionnaire/Quality of Life, categorized in quintiles.
      In 33 cycles, the mean menstrual cycle length was unknown.
      PercentileScoresNo. of pregnanciesAll cycles (n = 1,475)Menstrual cycle length <35 days (n = 1,171 cycles)Menstrual cycle length ≥35 days (n = 271 cycles)
      OR
      Adjusted for cycle number.
      (95% CI)
      OR
      Adjusted for cycle number, center, trade union, education, age, body mass index, contraceptive method in the year before enrollment, self-reported male or female reproduction-related disease, partner’s sperm count, smoking, and caffeine and alcohol consumption. All variables coded as in Table 2 and entered in the logistic regression model as a number of dummy variables equal to the number of categories minus one. The interaction between recent OC use and cycle number was adjusted by entering five dummy variables (OC × cycle).
      (95% CI)
      No. of cyclesOR
      Adjusted for cycle number, center, trade union, education, age, body mass index, contraceptive method in the year before enrollment, self-reported male or female reproduction-related disease, partner’s sperm count, smoking, and caffeine and alcohol consumption. All variables coded as in Table 2 and entered in the logistic regression model as a number of dummy variables equal to the number of categories minus one. The interaction between recent OC use and cycle number was adjusted by entering five dummy variables (OC × cycle).
      (95% CI)
      No. of cyclesOR
      Adjusted for cycle number, center, trade union, education, age, body mass index, contraceptive method in the year before enrollment, self-reported male or female reproduction-related disease, partner’s sperm count, smoking, and caffeine and alcohol consumption. All variables coded as in Table 2 and entered in the logistic regression model as a number of dummy variables equal to the number of categories minus one. The interaction between recent OC use and cycle number was adjusted by entering five dummy variables (OC × cycle).
      (95% CI)
      0–203–1073ReferenceReference364Reference610.9 (0.4–2.0)
      21–4011551.1 (0.8–1.6)1.0 (0.7–1.6)2271.1 (0.7–1.7)640.7 (0.3–1.6)
      41–6012440.8 (0.6–1.2)0.8 (0.5–1.3)2420.8 (0.5–1.3)540.7 (0.3–1.6)
      61–8013250.9 (0.5–1.5)0.9 (0.5–1.4)1370.9 (0.5–1.5)220.5 (0.1–2.4)
      81–10014–20360.7 (0.5–1.1)0.6 (0.4–1.0)2010.9 (0.5–1.4)700.1 (0.01–0.4)
      legend In 33 cycles, the mean menstrual cycle length was unknown.
      Adjusted for cycle number.
      Adjusted for cycle number, center, trade union, education, age, body mass index, contraceptive method in the year before enrollment, self-reported male or female reproduction-related disease, partner’s sperm count, smoking, and caffeine and alcohol consumption. All variables coded as in Table 2 and entered in the logistic regression model as a number of dummy variables equal to the number of categories minus one. The interaction between recent OC use and cycle number was adjusted by entering five dummy variables (OC × cycle).
      The length of a menstrual cycle is measurable only when pregnancy fails to occur. There was a significantly increased risk of cycle lengths ≥35 days in cycles with a General Health Questionnaire/Quality of Life score above the 80th percentile (OR 2.1; 95% CI 1.4–3.2).
      The mean menstrual cycle length, calculated from all observed cycles for each woman, modified the relation between scores on the General Health Questionnaire/Quality of Life and fecundability (Table 3). Although no effect was found among women with a mean cycle length of <35 days, there was a strong relation between distress and fecundability in women with longer menstrual cycles. Fecundability ORs based on other cutoffs showed that the decline in the odds of fecundability was present already for cycle lengths of >30 days (Fig. 1). When mean menstrual cycle length was based on data from the enrollment questionnaire, similar results were found (data not shown).
      Figure thumbnail GR1
      FIGURE 1OR for pregnancy for cycles with high distress in women with long mean menstrual cycles according to the cutoff value for long cycle length. High distress is defined as a score above the 80th percentile in the General Health Questionnaire/Quality of Life. Solid line = OR; dashed lines = 95% CI.
      In the longitudinal approach, in which each woman was assigned her personal reference value of General Health Questionnaire score, a similar relation was found. Among women with long mean menstrual cycles, the probability of conception was increased in cycles in which the woman had a lower score on the General Health Questionnaire/Quality of Life compared with her individual mean value (Table 4). Using the primary cycle score or the mean score of all cycles provided essentially identical results (data not shown).
      TABLE 4Odds ratios for pregnancy per cycle and 95% CIs for individual changes in scores on the General Health Questionnaire/Quality of Life stratified by mean menstrual cycle length.
      Data are OR (95% CI). The analysis includes cycles from women who contributed with a current cycle General Health Questionnaire score in at least two cycles (n = 307 women). Odds ratios are adjusted for cycle number.
      VariableMenstrual cycle length <35 days (n = 770 cycles)Menstrual cycle length ≥35 days (n = 176 cycles)
      Difference from individual mean score
      For each woman, the individual mean score of all cycles was calculated, and for each cycle, the difference between the cycle score and the individual mean score was calculated. Differences were categorized according to tertiles in the distribution of all cycle differences.
      Higher score (upper tertile)1.5 (0.9–2.4)1.7 (0.3–8.9)
      Same score (middle tertile)ReferenceReference
      Lower score (lower tertile)1.1 (0.6–1.9)8.4 (1.6–45.3)
      Difference, continuous (OR/point)
      OR for pregnancy per point increase in score difference. The SD was 2.7 points.
      1.02 (0.95–1.11)0.73 (0.59–0.91)
      legend Data are OR (95% CI). The analysis includes cycles from women who contributed with a current cycle General Health Questionnaire score in at least two cycles (n = 307 women). Odds ratios are adjusted for cycle number.
      For each woman, the individual mean score of all cycles was calculated, and for each cycle, the difference between the cycle score and the individual mean score was calculated. Differences were categorized according to tertiles in the distribution of all cycle differences.
      OR for pregnancy per point increase in score difference. The SD was 2.7 points.
      Analyses based on another 12-item version of the General Health Questionnaire (
      • Goldberg D.P
      ) showed similar results (fecundability OR 0.1; 95% CI 0.01–0.7 for women with scores above the 80th percentile and with long menstrual cycles). Analyses stratified by cycle number demonstrated homogeneity across the cycles of waiting time, with the same direction in each of the six cycles. Stratification by frequency of sexual intercourse (5 levels) and the other potential confounding variables revealed essentially identical results. Restriction to couples without reported male or female urogenital disorders or sperm density of <20 × 106/mL (294 couples), couples who used an intrauterine device or OCs (256 couples), or cycles with cycle-specific General Health Questionnaire scores (981 cycles) did not change the estimates.
      Early embryonal loss was detected in 32 cycles. The proportion of pregnancies terminated by early embryonal loss was 16.8%. Compared with women with a menstrual cycle length of <35 days and a low score on the General Health Questionnaire/Quality of Life (below the median value of 12), the ORs for early embryonal loss were 1.1 (95% CI 0.4–3.0), 1.5 (95% CI 0.5–4.7), and 12.3 (95% CI 2.1–72.5) for women with cycles of <35 days with a high score, women with cycles of ≥35 days with a low score, and women with cycles of ≥35 days with a high score, respectively.

      Discussion

      The main finding of this study was substantially reduced fecundability among highly distressed women with long menstrual cycles. Distress did not affect women with a normal duration of menstrual cycles.

       Sources of bias

      There is no consensus on how best to measure distress, although methodologic issues have been discussed extensively (
      • Paarlberg K.M
      • Vingerhoets A.J
      • Passchier J
      • Dekker G.A
      • Van Geijn H
      Psychosocial factors and pregnancy outcome a review with emphasis on methodological issues.
      ,
      • Kristensen T.S
      Job stress and cardiovascular disease a theoretical critical review.
      ). The original General Health Questionnaire inventory consists of 60 items, and several subscales exist. We used the General Health Questionnaire/Quality of Life subscale because these items cover relevant components with a higher degree of inclusiveness (
      • Bech P
      ). Another 12-item version of the General Health Questionnaire has a higher specificity for more severe psychological distress and psychiatric disease (
      • Goldberg D.P
      ). Analyses based on this scale revealed similar results.
      The main challenge in this study was to control the direction of causality. Because the failure to become pregnant may be a stressor per se (
      • Wright J
      • Allard M
      • Lecours A
      • Sabourin S
      Psychosocial distress and infertility a review of controlled research.
      ,
      • Seibel M.M
      Emotional aspects of infertility.
      ,
      • Boivin J
      • Takefman J.E
      Stress level across stages of in vitro fertilization in subsequently pregnant and nonpregnant women.
      ,
      • Lalos A
      • Lalos O
      • Jacobsson L
      • von Schoultz B
      The psychosocial impact of infertility two years after completed surgical treatment.
      ), an association could be due to bias if the woman knew whether the actual attempt to become pregnant was or was not successful when she provided the information on distress. We attempted to account for this in two ways. First, we enrolled only couples who had no knowledge about their fertility. Thus, we excluded couples who had a potentially raised level of distress caused by unsuccessful attempts to conceive before entry. Second, we instructed the couples to complete the cycle-specific questionnaire on the 21st day in each menstrual cycle. Most questionnaires (90%) were completed before day 23, and the results remained unchanged when questionnaires completed after day 24 were excluded. Thus, our data on distress and fertility are genuinely prospective.
      Planning bias may arise if the use of unsafe birth-control methods is associated with the exposure of interest (
      • Weinberg C.R
      • Baird D.D
      • Wilcox A.J
      Sources of bias in studies of time to pregnancy.
      ). If highly distressed couples use less efficient birth control, the most fertile of these couples may conceive unintentionally and thus never become eligible for this study. Planning bias cannot be ruled out completely without knowledge of exposure among nonplanners, but if birth-control methods vary considerably, this should probably be reflected among the enrolled couples as well.
      Among couples who had used safe contraceptive methods for the last 12 months, 21% of the women had a high score on the General Health Questionnaire–30, compared with 18% in the group who had used less safe contraceptives (Table 2). This difference is not statistically significant and is in fact in the opposite direction of what would be suspected if a substantial planning bias were present. It is therefore not likely that planning bias led to an underrepresentation of highly fertile, highly distressed women. Furthermore, similar associations were found in intraindividual analyses, which are less vulnerable to this bias.

       Evaluation of the results

      Animal studies and physiologic considerations have suggested that psychological distress may influence the female reproductive system at various levels: through the autonomic nervous system, the endocrine system, and the immune system (
      • Seibel M.M
      Emotional aspects of infertility.
      ,
      • Johnson E.O
      • Kamilaris T.C
      • Chrousos G.P
      • Gold P.W
      Mechanisms of stress a dynamic overview of hormonal and behavioral homeostasis.
      ,
      • Meyerhoff J.L
      • Oleshansky M.A
      • Kalogeras K.T
      • Mougey E.H
      • Chrousos G.P
      • Granger L.G
      Neuroendocrine responses to emotional stress possible interactions between circulating factors and anterior pituitary hormone release.
      ). Stress-induced changes in sexual behavior constitute a more obvious possible mechanism of action. The frequency of sexual intercourse was only weakly associated with distress (Table 2) and had no impact on the relation between distress and fecundability. Results were also unchanged after exclusion of couples who reported sexual problems such as loss of libido or vaginal dryness. Thus, our results cannot be explained by behavioral changes alone.
      The pronounced interaction between menstrual cycle length and distress was not part of our initial hypothesis and must be corroborated by others. It is well established that exposure to physical stressors can lead to changes in characteristics of the menstrual cycle (
      • Bullen B.A
      • Skrinar G.S
      • Beitins I.Z
      • von Mering G
      • Turnbell B.A
      • McArthur J.W
      Induction of menstrual disorders by strenuous exercise in untrained women.
      ,
      • Rebar R.W
      • Cumming D.C
      Reproductive function in women athletes.
      ), and it appears that vigorous exercise alters the pulsatile release of GnRH (
      • Bonen A
      Exercise-induced menstrual cycle changes.
      ). These results often have been extended to include mental stress, although human scientific evidence is limited.
      Berga et al. (
      • Berga S.L
      • Daniels T.L
      • Giles D.E
      Women with functional hypothalamic amenorrhea but not other forms of anovulation display amplified cortisol concentrations.
      ) reported increased activity of the hypothalamic-pituitary-adrenal axis in women with functional hypothalamic amenorrhea and reduced GnRH drive, compared with women with other causes of amenorrhea. This was interpreted as a stress effect, and the investigators suggested that if stress can be ameliorated or resolved, then GnRH drive should recover and ovulation should resume. Harlow and Matanoski (
      • Harlow S.D
      • Matanoski G.M
      The association between weight, physical activity, and stress and variation in the length of the menstrual cycle.
      ) found an increased risk of long menstrual cycles during stressful periods in the first year at college. They suggested that moderate stress levels may increase the probability of long cycles in susceptible individuals.
      The primary outcome of interest in our study was pregnancy. The length of cycles that result in a pregnancy is not observable, but in previous cycles, an association was found between distress and menstrual cycle length. Thus, the mean menstrual cycle length may be a marker of susceptibility to distress, and the demonstrated effect of distress on fecundability may be mediated through a disturbance of the normal menstrual cycle.
      The reduced fertility in women with high General Health Questionnaire scores and long menstrual cycles might be explained in part by an increased incidence of early embryonal loss. The latter finding, however, is based on relatively few observations and should be interpreted with caution. Animal research has indicated that exposure to stressors is associated with increased rates of resorption (
      • Pratt N.C
      • Lisk R.D
      Effects of social stress during early pregnancy on litter size and sex ratio in the golden hamster (mesocricetus auratus).
      ,
      • deCatanzaro D
      • Macniven E
      Psychogenic pregnancy disruptions in mammals.
      ), although little is known about the mechanisms involved.
      In conclusion, our findings indicate that psychological distress may be a risk factor for reduced fertility among women with long menstrual cycles.

      Acknowledgements

      The authors are indebted to several trade union officials, especially Ernst Bliesmann, Peter Olesen, Rigmor Laulund, and Niels Nedergaard.

      Appendix 1.

       The general health questionnaire/quality of life

      legend
      Have you during the last month:
      (1) - been feeling perfectly well and in good health?
      Numbers refer to items in the 60-item version of the General Health Questionnaire (6).
      (Better than usual / Same as usual / Worse than usual / Much worse than usual)
      (7) - been able to concentrate on whatever you’re doing?(Better than usual / Same as usual / Less than usual / Much less than usual)
      (16) - been feeling full of energy?(Better than usual / Same as usual / Less energy than usual / Much less energetic)
      (21) - been managing to keep yourself busy and occupied?(More than usual / Same as usual / Rather less than usual / Much less than usual)
      (23) - tended to lose interest in your ordinary activities?(Not at all / No more than usual / Rather more than usual / Much more than usual)
      (28) - felt on the whole you were doing things well?(Better than usual / About the same / Less well than usual / Much less well)
      (30) - been satisfied with the way you’ve carried out your task?(More satisfied / About same as usual / Less satisfied than usual / Much less satisfied)
      (31) - been able to feel warmth and affection for those near to you?(Better than usual / About same as usual / Less well than usual / Much less well)
      (35) - felt that you are playing a useful part in things?(More so than usual / Same as usual / Less useful than usual / Much less useful)
      (36) - felt capable of making decisions about things?(More so than usual / Same as usual / Less so than usual / Much less capable)
      (42) - been able to enjoy your normal day-to-day activities?(More so than usual / Same as usual / Less so than usual / Much less than usual)
      (54) - been feeling reasonably happy, all things considered?(More so than usual / About same as usual / Less so than usual / Much less than usual)
      legend Numbers refer to items in the 60-item version of the General Health Questionnaire (6).

      References

        • Wright J
        • Allard M
        • Lecours A
        • Sabourin S
        Psychosocial distress and infertility.
        Int J Fertil. 1989; 34: 126-142
        • Seibel M.M
        Emotional aspects of infertility.
        Fertil Steril. 1982; 37: 137-145
        • Boivin J
        • Takefman J.E
        Stress level across stages of in vitro fertilization in subsequently pregnant and nonpregnant women.
        Fertil Steril. 1995; 64: 802-810
        • Lalos A
        • Lalos O
        • Jacobsson L
        • von Schoultz B
        The psychosocial impact of infertility two years after completed surgical treatment.
        Acta Obstet Gynecol Scand. 1985; 64: 599-604
        • Bonde J.P.E
        • Hjollund N.H.I
        • Jensen T.K
        • Ernst E
        • Kolstad H.A
        • Henriksen T.B
        • et al.
        A follow-up study of environmental and biologic determinants of fertility among 430 Danish first pregnancy planners.
        Reprod Toxicol. 1998; 12: 19-27
        • Goldberg D.P
        The detection of psychiatric illness by questionnaire. Oxford University Press, London1972
        • Bech P
        Quality of life in psychosomatic research.
        Psychopathology. 1987; 20: 169-179
        • Hedegaard M
        • Henriksen T.B
        • Sabroe S
        • Secher N.J
        Psychological distress in pregnancy and preterm delivery.
        Br Med J. 1993; 307: 234-239
        • Bech P
        Rating scales for psychopathology, health status and quality of life. Springer, Berlin1993
        • Loldrup D
        • Langemark M
        • Hansen H.J
        • Bonnevie O
        • Elsborg L
        • Olesen P
        • et al.
        The validity of the melancholia scale (MES) in predicting outcome of antidepressants in chronic idiopathic pain disorders.
        Eur Psychiatr. 1991; 6: 119-125
        • Weinberg C.R
        The influence of occupational activity on the menstrual cycle and fecundability.
        Epidemiology. 1994; 5: 563-564
        • SAS Institute Inc
        SAS/STAT software: changes and enhancements through release 6.11. SAS Institute, Inc, Cary (NC)1996
        • Hosmer D.W.J
        • Lemeshow S
        Applied logistic regression. John Wiley & Sons, New York1989
        • SAS Institute Inc
        SAS/STAT user’s guide, version 6, volume 2. 4th ed. SAS Institute, Inc, Cary (NC)1989
        • Burkman Jr, R.T
        Noncontraceptive effects of hormonal contraceptives.
        Am J Obstet Gynecol. 1994; 170: 1569-1575
        • Rothman K.J
        • Greenland S
        Modern epidemiology. Lippincott-Raven, Philadelphia1998
        • Paarlberg K.M
        • Vingerhoets A.J
        • Passchier J
        • Dekker G.A
        • Van Geijn H
        Psychosocial factors and pregnancy outcome.
        J Psychosom Res. 1995; 39: 563-595
        • Kristensen T.S
        Job stress and cardiovascular disease.
        J Occup Health Psych. 1996; 1: 246-260
        • Weinberg C.R
        • Baird D.D
        • Wilcox A.J
        Sources of bias in studies of time to pregnancy.
        Stat Med. 1994; 13: 671-681
        • Johnson E.O
        • Kamilaris T.C
        • Chrousos G.P
        • Gold P.W
        Mechanisms of stress.
        Neurosci Biobehav Rev. 1992; 16: 115-130
        • Meyerhoff J.L
        • Oleshansky M.A
        • Kalogeras K.T
        • Mougey E.H
        • Chrousos G.P
        • Granger L.G
        Neuroendocrine responses to emotional stress.
        Adv Exp Med Biol. 1990; 274: 91-111
        • Bullen B.A
        • Skrinar G.S
        • Beitins I.Z
        • von Mering G
        • Turnbell B.A
        • McArthur J.W
        Induction of menstrual disorders by strenuous exercise in untrained women.
        N Engl J Med. 1985; 312: 1349-1353
        • Rebar R.W
        • Cumming D.C
        Reproductive function in women athletes.
        J Am Med Assoc. 1981; 246: 1590
        • Bonen A
        Exercise-induced menstrual cycle changes.
        A functional, temporary adaptation to metabolic stress. Sports Med. 1994; 17: 373-392
        • Berga S.L
        • Daniels T.L
        • Giles D.E
        Women with functional hypothalamic amenorrhea but not other forms of anovulation display amplified cortisol concentrations.
        Fertil Steril. 1997; 67: 1024-1030
        • Harlow S.D
        • Matanoski G.M
        The association between weight, physical activity, and stress and variation in the length of the menstrual cycle.
        Am J Epidemiol. 1991; 133: 38-49
        • Pratt N.C
        • Lisk R.D
        Effects of social stress during early pregnancy on litter size and sex ratio in the golden hamster (mesocricetus auratus).
        J Reprod Fertil. 1989; 87: 763-769
        • deCatanzaro D
        • Macniven E
        Psychogenic pregnancy disruptions in mammals.
        Neurosci Biobehav Rev. 1992; 16: 43-53