Spontaneous abortion occurs in 10–15% of clinically recognized pregnancies, usually in the first or early second trimester.1 It is generally defined as a pregnancy loss occurring before 20 weeks of gestation.2 Genetic anomalies are known to play a critical role in many early miscarriages, with 35–75% showing chromosomal abnormalities on karyotyping.3 Other factors known to be associated with miscarriage include advanced maternal age and a maternal history of recurrent losses (ie, two or more such outcomes in previous pregnancies), as well as maternal infections such as Listeria monocytogenes, Toxoplasma gondii, and rubella.1 Poorly controlled maternal insulin-dependent diabetes and thyroid autoimmunity are also linked to spontaneous abortion,4,5 as are antiphospholipid antibodies and factor V Leiden.1 Environmental toxins related to this outcome include benzene, gasoline, hydrogen sulfide, lead, and mercury,6–8 as well as tobacco,8–10 alcohol, and caffeine.11–13 Lifestyle factors, diet, deficiencies of folic acid or vitamin B12, and maternal obesity may also play a role.14–18
Although there is a voluminous literature on maternal factors in spontaneous abortion, less attention has been paid to the male partner's influence. Occupational exposures of men to pesticides, herbicides, dioxin, and sulfonamide have been investigated in relation to miscarriage,19–24 and there is a growing realization that paternal age influences diverse, complex diseases in adult offspring, ranging from cancer25,26 to schizophrenia27,28 and Alzheimer's disease.29,30 Our group has found paternal age to be related to preeclampsia,31 and it is known that advanced paternal age is associated with abnormalities in sperm, certain chromosome defects, and numerous birth defects associated with autosomal dominant mutations.25,26 The American Society for Reproductive Medicine has set an upper age limit of 40 years for semen donors because of the increased risk of genetic abnormalities in the offspring of older fathers.32
Several studies suggest an association of paternal age with spontaneous abortion33–38 but have failed to clarify whether there is a cutoff age or a progressive trend over the whole range of ages. Some were limited by low statistical power,33,35 inadequate adjustment for maternal age,35 an a priori treatment of paternal age that makes it difficult to interpret results,33 inclusion of late fetal deaths, or omission of some early losses.36,38 The current study tested the hypothesis that advancing paternal age is associated with an increased risk of spontaneous abortion, independent of maternal age as well as other factors. This study uses a large historical data set which contains information on many characteristics of mothers and fathers that might contribute to spontaneous abortion. This enabled us to control for multiple possible confounders of the relationship between paternal age and spontaneous abortion. Because delayed childbearing is increasingly common in industrialized countries, elucidation of this issue should provide important additional information for couples planning their families.
MATERIALS AND METHODS
This study draws on data from the Jerusalem Perinatal Study, a population-based cohort derived from 92,408 births in 1964–1976, whose design has been described in detail in a previous publication.39–42 Within this cohort, two subsets of mothers were interviewed regarding their obstetric history. The information they reported included a list of prior pregnancies and their outcomes along with month and year of each pregnancy. In 1965–1968 the mothers of 11,467 newborns were interviewed in antenatal clinics, usually at their first visit, and in November 1974 through December 1976 the mothers of 16,909 newborns were interviewed during their postpartum hospital stays. The first interview captured women attending free municipal antenatal clinics serving those at low risk for a poor obstetric outcome; these interviews were linked to 6.5% of births in Jerusalem in 1965 and 67.5, 64.7, and 50.5% of those in 1966, 1967, and 1968, respectively. The second interview captured 98% of the women delivering in the town's three largest obstetric units in the 26 months up to 1976 and was closely representative of the whole population, because 92% of all births occurred in these three hospitals.
The study was approved by the institutional review boards in New York and Jerusalem. We assembled two data sets from the computer-stored records of these interviews and constructed nested case–control studies for each data set. After comparing the data and results from the two data sets, we combined them. Cases were women whose previous pregnancy ended in a spontaneous abortion before the fifth month (20 weeks) of gestation. Controls were those whose previous pregnancy ended in a live birth. Cases and controls were restricted to women having their second or subsequent birth. This was necessary because women at their first birth with no prior pregnancies could not act as controls for women at their first birth after a prior spontaneous abortion. We excluded those reporting their previous pregnancy as ending in ectopic pregnancy, induced abortion, late fetal death (from the fifth month onwards) or intrapartum fetal death. The ages of the mother and her partner at the time of the case or control pregnancy were derived from their ages at interview, minus the time interval from the previous pregnancy until the interview.
There were 13,865 women in this study. Table 1 shows how the numbers were derived from the data files based on offspring. From the 9,827 women interviewed at least once in 1965–1968 we excluded 600 (6.1%) who delivered in hospitals not covered by our surveillance of diabetes. From the 5,743 potentially eligible, having their second or subsequent pregnancy with a prior live birth or spontaneous abortion, we excluded 62 (1.1%) with missing information on maternal or paternal age and a further 40 (0.7%) with missing information on other variables. In comparison, we excluded 293 (3.0%) with missing data on ages and 55 (0.6%) with missing data on other variables in our model from the 9,703 potentially eligible cohort. Women missing data on education were not excluded.
We used SAS 9.0 (SAS Institute Inc, Cary, NC) to analyze the data. After comparing demographic and other characteristics of cases and controls using cross tabulations, we developed unconditional logistic regression models, employing Proc Logistic provided by SAS. The regression results are presented as odds ratios with 95% confidence intervals. All tests were two tailed and P<.05 was chosen as the level of statistical significance.
Data for paternal age are presented in 5-year age groups; the category of 25–29 year olds was designated as the reference group because it was the largest. Maternal age was the first factor we adjusted for in our analysis of the relationship between paternal age and miscarriage. The maternal age variable fit the data best as a continuous linear variable, expressed as deviations from its mean. We used orthogonal coding of maternal and paternal ages to mitigate the effects of correlations between paternal and maternal age.
Other potential confounders or effect modifiers considered for inclusion in the models were maternal diabetes (recorded in the medical record) and smoking (ever compared with never, self-reported at interview, there being very few who had quit), history of spontaneous abortions in pregnancies before the pregnancy of interest, parity (history of 2, 3–4, or 5 or more pregnancies at interview in the first data set, or before the pregnancy just delivered, in the second data set), interval from interview to previous pregnancy (less than 3 years compared with 3 or more years), socioeconomic status (low, medium, high), years of maternal and paternal education(less than ninth grade compared with ninth grade or greater), and history of induced abortions. These variables were modeled as dichotomies or sets of dummy variables. We confirmed that the reference group we chose for each category produced the best overall model using Wald scores.43 Those variables that changed the crude paternal age effect on spontaneous abortion by 10% or more were included as confounders in the final regression model. We tested for effect modification by comparing the –2log likelihoods of the main effects model with the model containing the interaction term; if the interaction term was significant we included the variable in the final model. Data on paternal smoking was only available for the second interview and consequently was not included in our merged data set.
Table 2 compares the two data sets, showing the numbers of cases and controls and the percent distribution of their characteristics. Overall, data from the two interviews were very similar. The proportions reporting spontaneous abortion were 10.3% and 10.7% in the first and second interview, respectively, with 10.6% in the combined data set. The distributions of the women's ages and their partners' ages were similar in the data sets, and in both, case women and their partners tended to be older than controls. The ages of the women at the time of their index pregnancies ranged from 13 to 50 years. Mean ages for cases and controls were 29.0 (standard deviation 5.0) years and 26.3 (standard deviation 4.6) years, respectively in the combined data set. Table 2 shows that in both interviews the cases were more likely to be diabetics and were more often smokers. In the second data set the odds ratio (OR) for paternal smoking was 1.1 (95% confidence interval [CI] 0.93–1.2), but we did not have data on paternal smoking in the first interview and did not include this covariate in our final models. In both interviews, women reporting a spontaneous abortion in their index pregnancy were more likely than controls to have experienced spontaneous and induced abortions in earlier pregnancies. As expected, cases in both interviews reported a shorter interval from index pregnancy to interview. This was expected, because women who desired pregnancy and then miscarried could attempt conception sooner than women who carried a pregnancy to term. In both interviews cases were slightly more likely to report 3 or 4 pregnancies at interview, whereas controls were more likely to report either fewer or more children. On the other hand, social class and education did not differ greatly between cases and controls. Both education and social class rose in the 8–10 years between the two interviews; this was true in both cases and controls. Because the characteristics of the subjects in the two data sets were so similar, as were the results of the subsequent analyses when separate, we merged the data from the two interviews into a single data set.
We considered whether the missing data in either interview had the potential to bias our results. For each of the variables, however, the proportions of potential cases or controls with missing data were similar; so we concluded that the excluded missing data were unlikely to bias our results. Women with missing education formed a larger group in interview 1, and constituted the bulk of the missing information for that data set. Cases and controls were similarly distributed in those with missing education data as compared with those with the education data, and education was not a risk factor for spontaneous abortion in either interview. Because of this, and because the major reason for exclusion of women in the first interview was unknown education, those with unknown education were included in both data sets.
Table 3 shows effects of maternal age on the odds ratios (ORs) for spontaneous abortion, derived from analyses of the combined data set and using logistic regression models. We also modeled the data for each interview separately; but because the odds ratios for each of the variables were so similar, we present only the findings from the combined data set. Note that because spontaneous abortion is not a rare outcome, the odds ratios are not equivalent to relative risks. The table compares the crude odds ratios with the odds ratios adjusted for other variables. As expected, there was a strong effect of maternal age on miscarriage; the crude odds ratios increased more than 10-fold from 0.42 in the group of women aged younger than 20 years to 5.3 at age 35 years or older. Adjustment for other variables changed this a bit, ie, from 0.41 in the youngest women to 8.3 in the oldest. The table also shows the ORs for maternal age fitting a linear trend as a continuous variable; in this case the ORs represent the increase in reporting of spontaneous abortion with each additional year of maternal age.
Table 4 shows relationships of spontaneous abortions to other variables, using the combined data set. The table compares the age-adjusted ORs with those that are adjusted further for other variables. Maternal smoking, abortions before the index pregnancy (both spontaneous and induced), 3 or fewer years from index pregnancy to interview, five or more births, and fewer years of maternal and paternal education were all associated with spontaneous abortion. Effects of education were small. Maternal diabetes recorded at interview was associated with an increased risk of miscarriage in the index pregnancy (Table 3). However, with the numbers of diabetic women available for study, this relationship did not achieve statistical significance.
We included maternal age, history of spontaneous abortion before index pregnancy, time interval from index pregnancy to interview, parity, maternal diabetes, and maternal smoking as covariates in subsequent analyses of effects of paternal age. The first four of these variables fulfilled our formal criteria for effect modification or confounding (being related both to paternal age and spontaneous abortion and causing at least a 10% change in the OR for paternal age, or achieving significance in the –2 log likelihood test described above) or improved the fit of the final regression model (by comparing the final model fit with or without the specific covariate). Although there were too few smokers or diabetics to achieve statistical significance, both are generally acknowledged as causes of miscarriage in the literature.40,41 Both were related to paternal age with substantial effect sizes in our sample. Consequently, we included them in our final models. Other variables were not included, because they had weak effects on spontaneous abortion, did not fulfill our criteria for confounding or effect modification, or did not improve model fit or affect the ORs estimated for paternal age.
Paternal age ranged from 15 to 72 years, with a mean value of 32.9 (standard deviation 6.0) years for cases and 29.9 (standard deviation 5.7) years for controls. Table 5 shows effects of paternal age in 5-year groups, analyzed in the combined data set, with age 25–29 years as the reference group. The crude ORs showed a nearly six-fold change from the youngest group of husbands (age younger than 25 years) to the oldest (40 years or older). Adjustment for maternal age attenuated this to a 2-fold difference, although the trend remained significant. Further adjustment for other variables led to further, smaller changes in the effect of paternal age on spontaneous abortion. Effects of paternal age were similar in the two data sets. In the first, the ORs (and 95% CIs) were 0.62 (0.41–0.96), 1 (reference group), 1.2 (1.0–1.6), 1.8 (1.4–2.5), 1.4 (.93–2.0); in the second they were 0.58 (0.42–0.79), 1 (reference group), 1.5 (1.2–1.8), 1.9 (1.5–2.4), and 1.7 (1.2–2.3). Treating paternal age as a continuous variable, in the combined data set, the model estimated an adjusted OR for spontaneous abortion of 1.08 (1.07, 1.09) per year of paternal age.
Finally, we completed a logistic regression analysis where we ran our complete model stratified by interview. The ORs (and 95% CIs) were 0.59 (.45–.76), 1 (reference group), 1.4 (1.2–1.6), 1.9 (1.6–2.3), and 1.6 (1.3–2.0) for fathers aged younger than 25 years, 25–29 years, 30–34 years, 35–39 years, and 40 years or older, respectively.
Our study found that women with partners aged 35–39 years or 40 years or older had a nearly three-fold increase in spontaneous abortions as compared with women conceiving with men aged younger than 25 years. The relationship with the man's age is highly significant, consistent in two data sets collected 8–10 years apart. The finding is independent of the woman's age and not explained by confounding by diabetes, smoking, parity, or previous spontaneous abortions. Other findings from this study are consistent with those from previous epidemiologic studies of spontaneous abortions; these include a strong effect of maternal age and weaker relationships with maternal smoking and diabetes.
In the prospective study by Ford et al,34 father's age was described as younger than 35 years or 35 years and older. Ford et al analyzed dozens of environmental and occupational exposures as well as four different outcomes in 585 couples. The outcomes were reproductive outcome, live birth, miscarriage, and infertility. The possible effects of having different numbers of factors were examined for miscarriage and 9 months infertility. Nine factors were examined for miscarriage, and couples were found to have from zero to seven of these. The observed rate of pregnancy loss ranged from 3.7% to 75% with increasing numbers of factors. The description of paternal age combined with the multiple exposures and outcomes curtailed the specific information generated on paternal age effects. Nevertheless, they reported an increase in spontaneous abortion with paternal age older than 35 years. Nybo Andersen et al36 conducted a study using data from 23,821 women in which they found that the risk of early fetal death was elevated slightly (hazard ratio 1.38, 95% CI: 0.66–2.88) for fathers aged 50 years or older but was similar for all other levels of paternal age. This study included 124 fathers aged 50 years or older for whom only eight pregnancies resulted in early fetal death. Their observation was not statistically significant, possibly because it was based on the observation of only eight early fetal deaths, and the authors themselves write that their results should be interpreted with caution. Slama et al38 published a study that found an increase in miscarriage associated with paternal age older than 35 years, although their study did not adjust for maternal diabetes or parity effects on the primary association. Using data from 5,121 women, they found that the risk of spontaneous abortion associated with a paternal age of 35 years or older was 1.27, with a 95% CI of 1.00–1.61. Research by De la Rochebrochard et al33 restricted its focus to last planned pregnancies and collected data on 3,174 women from Denmark, Germany, Italy, and Spain. The authors created a combined parental age variable for analysis and reported that spontaneous abortion rates were highest when both parents were more advanced in age. If the woman was aged 35 years or older and the man was aged 40 years or older, the risk of miscarriage was higher, with an OR of 1.97 (95% CI 1.03–3.77) when comparing the “highest” and the “high risk zones” (“high zone” defined by the authors as significant increase in miscarriage risk, and “highest zone” as major increase in miscarriage risk). The study by al-Ansary et al35 analyzed 452 pregnancies, with only 38 mothers aged 35 or older. Consequently, researchers had limited ability to control for maternal age when analyzing paternal age. They report no significant association between spontaneous abortion and maternal age in their study. However, the authors did observe that women reporting spontaneous abortion were more frequently married to husbands older than 50 years (relative risk 2.4).
Our study's strengths include its large sample size and its extensive data, which permit consideration of important potential confounders not included together in other analyses. These include variables such as maternal diabetes, parity, history of prior spontaneous and induced abortions, socioeconomic status, and for retrospective studies, interval from index pregnancy to interview. This study focused exclusively on spontaneous abortion as the outcome. The nested case–control design optimizes this study because most spontaneous abortions are embryonic or preembryonic losses that occur by the 10th or 11th week of gestation.1 This study sample captured a history of miscarriage reported at interview, even if that miscarriage occurred before a woman sought prenatal care. Research of spontaneous abortion that recruits women when they begin prenatal care introduces two challenges to the study of miscarriage. First, many women miscarry at an early gestational age, and if they miscarry before they begin prenatal care, they will be missing from any study sample that recruits women seeking prenatal care (such as the Nybo Andersen and Slama studies). For example, in the study by Slama et al,38 women were recruited when they called their health plan to make their first prenatal appointment, but were excluded from the study if their pregnancy ended before their interview. Second, women who recognize their pregnancies early in the first trimester and seek prenatal care at that time, so that their early pregnancy losses are captured by a prospective study, may be different from the general population of women.
An additional strength of this study is that it explores the trend of paternal age influence on spontaneous abortion over several categories of paternal age. The finding of a step-wise increase in the likelihood of spontaneous abortion with increasing paternal age categories generates considerable support for an association between increasing paternal age and increasing spontaneous abortion. Moreover, the mothers in this study's cohort come from varied backgrounds; although 98% are Jewish, many are first- or second-generation immigrants from Islamic countries of Western Asia and North Africa, whereas others originated from Central and Eastern Europe. This broad mix of backgrounds makes our study findings more generalizable.
This study's cohort's extreme religious conservatism, low rate of out-of-wedlock births, and low divorce rates make paternity more certain than in some other cultures. Also, this study included only fertile women. Although this may make the results less generalizable, it also may remove potential distortions of the association between paternal age and spontaneous abortion. For example, if women with uterine anatomic anomalies precluding viable pregnancies were differentially distributed among paternal ages this could obscure the primary relationship being examined.
The limitations of retrospective studies of spontaneous abortion have been enumerated by Weinberg et al.44 They include inaccurate recall of prior pregnancies and a potential bias resulting from shorter interval from previous pregnancy to interview in those with a previous spontaneous abortion outcome as compared with a live birth outcome. This is because many women who experience a spontaneous abortion tend to conceive again until they have a live birth, whereas women who have a live birth may not get pregnant again for a longer time.44 Thus, recall for cases may be less accurate than recall for controls, because of the more likely longer interval from prior pregnancy to data collection in our study. We adjusted for interval from index pregnancy to interview. For a spurious effect of paternal age, recall would need to be biased by paternal age, and there is no indication that this should be the case in our study.
Another limitation to our study design was the possible misreporting of induced abortions as spontaneous abortions. Both data subsets showed that case women were more likely to report induced abortions (Table 2). If control women reported more induced abortions than cases, then one would consider whether cases were more frequently misreporting some of their induced abortions as spontaneous abortions. Because this was not the case, we were less concerned about this source of bias in our study. Additionally, in Israel at the time of the first interview, induced abortions were widely tolerated and easily available from gynecologists, many of whom had trained in Eastern Europe, where abortion was legal in the decades before our study. In our second data set most of the previous pregnancies occurred at a time when abortion was legal in Israel. There was no decrease in the proportion of pregnancies reported as spontaneous abortions between the first and second data sets; therefore, we do not believe that the first data set was biased for induced abortion reported as spontaneous abortion.
Another potential drawback of this study's design is that the two interviews were conducted at different times; the first interview was during pregnancy, usually at the first antenatal visit, while the second interview was during the first few postpartum days. Our results were concordant in the two interviews, however, despite the theoretical concern.
Our study generates strong support for the association of increasing paternal age with increasing spontaneous abortion, and our findings are corroborated by other published studies. We describe a step-wise increase in this association, and we show that this relationship persists after controlling for maternal age, maternal smoking, maternal diabetes, parity, history of spontaneous abortion before the index pregnancy, and interval from index pregnancy to interview. Advanced paternal age may only result in a slight increase in the chance of spontaneous abortion for a specific couple. Nevertheless, as childbearing is increasingly delayed in Western societies, this study provides important information for people who are planning their families.
1. American College of Obstetricians and Gynecologists. Clinical management guidelines of obstetrician–gynecologists. ACOG Practice Bulletin 24. Washington, DC: ACOG; 2001.
2. Cunningham FG, MacDonald PC, Gant NF, Leveno KJ, Gilstrap LC, Hankins GD, et al. Williams obstetrics. 20th ed. Stamford (CT): Appleton & Lange; 1997. p. 582.
3. Ohno M, Maeda T, Matsunobu, A. A cytogenetic study of spontaneous abortions with direct analysis of chorionic villi. Obstet Gynecol 1991;77:394–8.
4. Dorman JS, Burke JP, McCarthy BJ, Norris JM, Steenkiste AR, Aarons JH, et al. Temporal trends in spontaneous abortions associated with Type 1 diabetes. Diabetes Res Clin Pract 1999;43:41–7.
5. Prummel MF, Wiersinga WM. Thyroid autoimmunity and miscarriage. Eur J Endocrinol 2004;150:751–5.
6. Xu X, Cho S, Sammel M, You L, Cui S, Huang Y, et al. Association of petrochemical exposure with spontaneous abortion. Occup Environ Med 1998;55:31–6.
7. Hertz-Picciotto I. The evidence that lead increases the risk for spontaneous abortion. Am J Ind Med 2000;38:300–9.
8. Sharara F, Seifer DB, Flaws JA. Environmental toxicants and female reproduction. Fertil Steril 1998;70:613–22.
9. Garcia-Enguidanos A, Calle ME, Valero J, Luna S, Dominguez-Rojas V. Risk factors in miscarriage: a review. Eur J Obstet Gynecol Reprod Biol 2002;102:111–9.
10. Pattinson HA, Taylor PJ, Pattinson MH. The effect of cigarette smoking on ovarian function and early pregnancy outcome of in vitro fertilization treatment. Fertil Steril 1991;55:780–3.
11. Kesmodel U, Wisborg K, Olsen SF, Henriksen TB, Secher NJ. Moderate alcohol intake in pregnancy and the risk of spontaneous abortion. Alcohol Alcohol 2002;37:87–92.
12. Rasch V. Cigarette, alcohol, and caffeine consumption: risk factors for spontaneous abortion. Acta Obstet Gynecol Scand 2003;82:182–8.
13. Tolstrup JS, Kjaer SK, Munk C, Madsen LB, Ottesen B, Bergholt T, et al. Does caffeine and alcohol intake before pregnancy predict the occurrence of spontaneous abortion? Human Reproduction 2003;18:2704–10.
14. Di Cintio E, Parazzini F, Chatenoud L, Surace M, Benzi G, Zanconato G, et al. Dietary risk factors and risk of spontaneous abortion. Eur J Obstet Gynecol Reprod Biol 2001;95:132–6.
15. Gindler J, Li Z, Berry RJ, Zheng J, Correa A, Sun X, et al. Folic acid supplements during pregnancy and risk of miscarriage. Lancet 2001;358:796–800.
16. George L, Mills J, Johansson A, Nordmark A, Olander B, Granath F, et al. Plasma folate levels and risk of spontaneous abortion. JAMA 2002;288:1867–73.
17. Reznikoff-Etievant M, Zittoun J, Vaylet C, Pernet P, Milliez J. Low Vitamin B(12) level as a risk factor for very early recurrent abortion. Eur J Obstet Gynecol Reprod Biol 2002;104:156–9.
18. Wang J, Davies M, Norman R. Obesity increases the risk of spontaneous abortions during infertility treatment. Obes Res;2002:10:551–4.
19. Logman JF, de Vries LE, Hemels ME, Khattak S, Einarson TR. Paternal organic solvent exposure and adverse pregnancy outcomes: a meta-analysis. Am J Ind Med 2005;47:37–44.
20. Schnorr TM, Lawson CC, Whelan EA, Dankovic DA, Deddens JA, Piacitelli LA, et al. Spontaneous abortion, sex ration and paternal occupational exposure to 2,3,7,8-tetrachlorodibenzo-p-dioxin. Environ Health Perspect 2001;109:1127–32.
21. Petrelli G, Figa-Talamanca I, Tropeano R, Tangucci M, Cini C, Aquilani S, et al. Reproductive male-mediated risk: spontaneous abortion among wives of pesticide applicators. Eur J Epidemiol 2000;16:391–3.
22. Arbuckle TE, Savitz DA, Mery LS, Curtis KM. Exposure to phenoxy herbicides and the risk of spontaneous abortion. Epidemiology 1999;10:752–60.
23. Savitz DA, Arbuckle T, Kaczor D, Curtis KM. Male pesticide exposure and pregnancy outcome. Am J Epidemiol 1997;146:1025–36.
24. Prasad MH, Pushpavathi K, Devi GS, Reddy PP. Reproductive epidemiology in sulfonamide factory workers. J Toxicol Environ Health 1996;47:109–14.
25. Aitken RJ, Baker MA, Sawyer D. Oxidative stress in the male germ line and its role in the aetiology of male infertility and genetic disease. Reprod Biomed Online 2003;7:65–70.
26. Kuhnert B, Nieschlag E. Reproductive functions of the ageing male. Hum Reprod Update 2004;10:327–39.
27. Malaspina D, Corcoran D, Fahim C, Berman A, Harkavy-Friedman J, Yale S, et al. Paternal age and sporadic schizophrenia: evidence for de novomutations. Am J Med Genet 2002;114:299–303.
28. Dalman C, Allebeck P. Paternal age and schizophrenia: further support for an association. Am J Psychiatry 2002;159:1591–2.
29. Bertram L, Busch R, Spiegl M, Lautenschlager NT, Muller U, Kurz A. Paternal age is a risk factor for Alzheimer disease in the absence of a major gene. Neurogenetics 1998;1:277–80.
30. Whalley LJ, Thomas BM, Starr JM. Epidemiology of presenile Alzheimer's disease in Scotland (1974-1988) II. Exposures to possible risk factors. Br J Psychiatry 1995;167:732–8.
31. Harlap S, Paltiel O, Deutsch L, Knaanie A, Masalha S, Tiram E, et al. Paternal age and preeclampsia. Epidemiology 2002;13:660–7.
32. Bordson BL, Leonardo VS. The appropriate upper age limit for semen donors: a review of genetic effects of paternal age. Fertil Steril 1991;56:397–401.
33. De la Rochebrochard E, Thonneau P. Paternal age and maternal age are risk factors for miscarriage: results of a multicentre European study. Hum Reprod 2002;17:1649–56.
34. Ford JH, MacCormac L, Hiller J. PALS (pregnancy and lifestyle study): association between occupational and environmental exposure to chemicals and reproductive outcome. Mutat Res 1994;313:153–64.
35. al-Ansary LA, Babay ZA. Risk factors for spontaneous abortion: a preliminary study on Saudi women. J R Soc Health 1994;114:188–93.
36. Nybo Andersen AM, Hansen KD, Andersen PK, Davey Smith G. Advanced paternal age and risk of fetal death: a cohort study. Am J Epidemiol 2004;160:1214–22.
37. De La Rochebrochard E, McElreavey K, Thonneau P. Paternal age over 40 years: the “amber light” in the reproductive life of men? J Androl 2003;24:459–65.
38. Slama R, Bouyer J, Windham G, Fenster L, Werwatz A, Swan S. Influence of paternal age on the risk of spontaneous abortion. Am J Epidemiol 2005;161:816–23.
39. Davies AM, Prywes R, Tzur B, Weiskopf P, Sterk VV. The Jerusalem Perinatal Study. 1. Design and organization of a continuing, community-based, record-linked survey. Isr J Med Sci 1969;5:1095–106.
40. Windham GC, Von Behren J, Waller K, Fenster L. Exposure to environmental and mainstream tobacco smoke and risk of spontaneous abortion. Am J Epidemiol 1999;149:243–7.
41. Katz VL, Kuller JA. Recurrent miscarriage. Am J Perinatol 1994; 11:386–97.
42. Rush D, Cassano P, Harlap S. Perinatal outcome, maternal weight gain, cigarette smoking and social status in Jerusalem. Rev Epidemiol Sante Publique 1988;36:186–95.
43. Stokes ME, Davis CS, Koch GG. Categorical data analysis using the SAS system, 2nd ed. Cary (NC): SAS Institute; 2000.
44. Weinberg CR, Baird DD, Wilcox AJ. Bias in retrospective studies of spontaneous abortion based on the outcome of the most recent pregnancy. Ann N Y Acad Sci 1994;709:280–6.