Recent population-based studies from Denmark1 and Italy2 suggested a link between advancing paternal age and delivery before 32 weeks gestation among mothers 20 to 29 years old. This observation has biologic plausibility in that the placenta is largely dependent on the expression of genes of paternal origin,3 and potentially harmful mutations in genes involved in placentation may be more frequent among older men. This mechanism is also consistent with the finding that pregnancies fathered by older men may be at a higher risk of miscarriage.4
We explored whether the association between advanced paternal age and delivery before 32 weeks could be observed in U.S. births.
We used the U.S. birth certificates from 1995 to 2000. It is well-known that gestational age by date of last menstrual period (LMP) is especially vulnerable to errors at early weeks.5,6 We therefore examined the association using the clinical assessment of gestation.7 We excluded births from California residents from all analyses, because the clinical estimate was not reported.8
Figure 1 summarizes the steps to obtain the sample. We restricted the analysis to first births and, to obtain results comparable with those of previous reports,1,2 we performed this analysis only among white non-Hispanic women. Women with missing information on Hispanic origin (1.4% of white primiparas) were included. The proportion of missing paternal age was 39% among unmarried women and 0.4% among married ones; we therefore restricted the analysis to married women. We excluded all records with gestational age <20 weeks or missing (0.7%). Like in the previous studies,1,2 we restricted our analysis to parents aged 20 years or older. To avoid confounding by other conditions related to advanced maternal age, we excluded women older than 34 years. We thus had 2,510,509 eligible births, including 9,379 records with missing paternal age.
We excluded 1,277 preterm babies with a birth weight in excess of 4 standard deviations based on gestational age-specific external standards9 and using birth weight in 250-g categories. The applied standards started from 22 weeks.9 For weeks 20 and 21, babies were included only if the LMP-based gestational age was also below 32 weeks (686 of 772 babies). Preterm babies with missing birth weight were also included if the LMP estimate was in the same category as the clinical estimate (<32 or 32–36 weeks), thus recovering 168 of 302 births. No babies with a clinical gestation of 37 weeks or longer were excluded.
We contrasted early preterm delivery (<32 weeks gestation) with deliveries occurring from week 37 onward. We estimated odds ratios (ORs) and 95% confidence intervals (CIs) through logistic regression models, stratifying on maternal age and adjusting for mother’s education and smoking. Information on smoking was systematically missing in some states. For each year, we identified states with 25% or more missing data, flagging the missing values from these states differently than the missing values from states with fewer than 25% missing, thus adding 2 levels to the smoking variable.
According to our estimate, 0.9% of babies were born before 32 completed gestational weeks and 7.0% between weeks 32 and 36.
The proportion of births before 32 weeks by mother’s age was 0.9% among women 20 to 24 years old, 0.8% among women 25 to 29 years old, and 0.9% among women 30 to 34 years old. Table 1 reports the main characteristics of the births stratified on the timing of delivery. Low education (or missing information), smoking (or missing information in states with regular reporting), and missing father’s age were more frequent among women delivering before 32 weeks. Disagreement with the LMP estimate was more frequent among women with a clinical estimate of preterm, especially for the fraction 32 to 36 weeks. The mean birth weight (±standard deviation) of babies born before 32 weeks was 1156 g (±471) after excluding the implausible values.
Delivery before32 weeks was not consistently associated with increasing father’s age (Table 2). There was no association between father’s age and delivery between 32 and 36 weeks (not shown).
Figure 1 compares our results with the previously published studies.1,2 This shows the estimated crude risk of delivery before 32 weeks by father’s age among women aged 20 to 29 years. In both Denmark and Italy, the likelihood of early delivery increased with paternal age.
In contrast with previous findings,1,2 in a large sample of U.S. births, we saw no evidence of an association between increasing father’s age and delivery before 32 weeks. An earlier analysis of U.S. data that examined differences between mother’s and father’s ages (rather than age itself) reported no associations with delivery before 36 weeks except for young men having children with older women.10 In Denmark and Italy, where older father’s age was associated with preterm risk, prenatal care is largely free and there are fewer socioeconomic disparities than in the United States. Also, men born in the 1960s in Italy have a median age at first child exceeding 33 years,11 and 50% of men in Denmark are still childless at 30,12 whereas 54% of the men in our sample had had their first child by age 30. It is possible that, in the United States, selection by socioeconomic factors related to fathers’ age could produce confounding that disguises a true biologic effect of paternal aging. However, we found little evidence of such confounding when we adjusted our results by maternal education.
The other main difference between the previous reports1,2 and ours is that the proportion of deliveries before 32 weeks was about twice as high in the United States compared with Denmark and Italy. This higher background rate in the United States may dampen our ability to detect a trend with father’s age.
We assessed a large population with a wide range of variation for paternal age, which allowed us to examine this association with a substantial sample size within each category of maternal age. Misclassification of gestational age, if random, would bias the associations toward null values. If the most extreme values of paternal age reflect a higher likelihood of misclassification, we should still be able to see a trend in the less extreme categories, as observed in the European reports,1,2 but we do not.
The methods for making the “clinical assessments” of gestational age recorded on birth certificates are unspecified and presumably subject to error.7 However, we did not see an association between paternal age and delivery before 32 weeks even when we used the LMP assessment of gestation (data not shown). As a further check for errors in gestational age that might have biased the association, we used a birth weight of <1750 g as proxy for early delivery. This analysis also showed no evidence of paternal age effects (data not shown). We used Canadian standards to exclude babies too large for gestation using birth weight in 250-g categories. Although the adopted standard may not have been optimal for this population, the birth weight distributions at early weeks by clinical gestation showed much less evidence of misclassification than those by LMP, and shortening the right tail of the distribution virtually eliminated the problem. We doubt that different standards would have yielded qualitatively different results. Our sample was restricted to married women, because father’s age was missing for a large fraction of unmarried women. When we included unmarried women, however, the results were virtually the same (not shown).
Because paternal genes must be activated for successful placentation,3 it is biologically plausible that paternal age could play a role in increasing the risk of disorders associated with abnormal placentation. However, our findings from a large sample of U.S. births do not support previously reported effects of father’s age on very preterm delivery.
We thank Aimin Chen and Clarice Weinberg for reading an earlier draft of the manuscript.
1. Zhu JL, Madsen KM, Vestergaard M, et al. Paternal age and preterm birth. Epidemiology
2. Astolfi P, De Pasquale A, Zonta LA. Paternal age and preterm birth in Italy, 1990 to 1998. Epidemiology
3. Miozzo M, Simoni G. The role of imprinted genes in fetal growth. Biol Neonate
4. Slama R, Bouyer J, Windham G, et al. Influence of paternal age on the risk of spontaneous abortion. Am J Epidemiol
5. Gjessing HK, Skjaerven R, Wilcox AJ. Errors in gestational age: evidence of bleeding early in pregnancy. Am J Public Health
6. Parker JD, Schoendorf KC. Implications of cleaning gestational age data. Paediatr Perinat Epidemiol
7. Hospitals’ and Physicians’ Handbook on Birth Registration and Fetal Death Reporting
. DHHS Publication no (PHS) 87-1107. Hyattsville, MD: National Center for Health Statistics, Public Health Service; 1987.
8. Martin JA, Hamilton BE, Ventura SJ, et al. Births: Final Data for 2000. National Vital Statistics Reports, vol 50, no 5. Hyattsville, MD: National Center for Health Statistics; 2002. Available at: http://www.cdc.gov/nchs/data/nvsr/nvsr50/nvsr50_05.pdf
. Accessed 6 March 2006.
9. Kramer MS, Plat RW, Wen SW, et al. A new and improved population-based Canadian reference for birth weight for gestational age. Pediatrics
10. Kinzler WL, Ananth CV, Smulian JC, Vintzileos AM. Parental age difference and adverse perinatal outcomes in the United States. Paediatr Perinat Epidemiol
11. Rosina A, Fraboni R. Il primo figlio. In: Rosina A, Sabbadini LL, eds. Diventare padri in Italia. Fecondità e figli secondo un approccio di genere, Serie Argomenti, Istat, Roma; 2005:85–106. Available at: http://www.istat.it/salastampa/comunicati/non_calendario/20051020_00/
. Accessed 6 March 2006.
12. Kohler HP, Knudsen LB, Skytthe A, Christensen K. The fertility patterns of twins and general population compared: evidence from Danish cohorts 1954–64. Demographic research. 2002;6,article 14. Available at: http://www.demographic-research.org/volumes/vol6/14/6-14.pdf
. Accessed 6 March 2006.