Materials and Methods
We undertook a cross-sectional study based on data from computerized live birth certificates without personal identifiers for the state of Texas from 1994 through 1998. There were 1,675,024 registered births during that time and the following exclusions were made: 461,992 (27.6%) women aged 30 years and above, 308 (0.02%) missing information on maternal age; 24,782 (2.5%) twins and higher-order births; and 9335 (0.6%) lacking both last menstrual period (LMP) and clinical estimate of gestation information, leaving a total of 1,178,607 singleton births. The adult comparison group was limited to women aged 20–29 years, who account for more than 50% of all births in Texas.
Approximate dates of conception were computed using an algorithm combining LMP-based and clinical estimates of gestation. Gestational age based on LMP is an accurate measure of gestation for most pregnancies ending at term,16,17 and was therefore used as the primary measure to compute the date of conception for 972,464 (82.5%) of the singleton births. The estimated date of conception was computed by adding 14 days to the onset of the LMP. However, for most postterm births and a significant number of premature births, LMP is an imprecise measure of gestation presenting a discrepancy with gestation based on ultrasound.16 The clinical estimate of gestation, based on antenatal ultrasound, postnatal examination, or other clinical measures, is an alternative measure of gestation available in birth certificates.18,19 Therefore, for 75,607 (6.4%) of births for which there was more than a 4-week discrepancy between LMP and the clinical estimate of gestation, the latter was used as the more reliable measure. For an additional 130,536 (11.1%) of births for which the LMP was missing, the clinical estimate of gestation was substituted. For all births for which the clinical estimate of gestation was used, we subtracted the weeks of gestation less 14 days from the date of birth to estimate the date of conception. A total of 99.7% of the final gestational age estimates ranged from 20 to 44 weeks. Birth weight was correlated more strongly with this gestational age (r = .56) than with gestational age based on LMP only (r = .47).
Adolescent age was defined as 17 years or less at conception, consistent with the Healthy People 20104 objectives focusing on minors. The group of teenagers 18–19 year olds or older was analyzed as one of two comparison groups. Adult women aged 20–29 years comprised the other comparison group. Maternal and paternal ages at conception were computed using their respective dates of birth and the estimated date of conception. (Previous studies used mother's age at delivery.) Late initiation of prenatal care was defined as enrollment into prenatal care after 3 months of gestation or not at all.
Border counties were those within 100 miles of the Mexico border; urban counties were 58 of the 254 Texas counties classified as metropolitan by the US Census Bureau.20 Educational status was based on occupational category. Exclusive students were those who reported only “student.” If another occupation was reported alone or in addition to student then student status was nonexclusive student.
Total number of monthly conceptions were computed and adjusted to an equivalent for a 30-day month before computing monthly proportions. Because the monthly patterns we observed were similar from year to year, we present results for all years in aggregate. The χ2 goodness-of-fit test was used to test the null hypothesis of there being a constant proportion of 8.33% of births conceived per month. The χ2 test for homogeneity was also used to assess the deviation of peaks and troughs from a hypothetical baseline of 8.33% per month. Teenage births were stratified by sociodemographic variables to determine corresponding differences in monthly variations of conception. According to information obtained from the Houston Independent School District, the largest district in Texas, the summer vacation occurs approximately from mid-May to mid-August annually; Christmas vacation extends from mid-December to January, and the spring break occurs in March.
The relationship between season of conception and late initiation of prenatal care was assessed by comparing proportions of births receiving prenatal care late by month of conception initially using the χ 2 test for homogeneity. Subsequently, logistic regression was used to model seasonality of late initiation of prenatal care among teenagers, adjusting for potential sociodemographic confounders. SPSS 10.0 (SPSS Inc., Chicago, IL) and STATA 6.0 (Stata Corp., College Station, TX) programs were used for the data management and statistical computations.
Among 1,178,607 singleton births to women aged 29 years and less retained in the analyses, 124,005 (10.5%) were births to adolescents (aged 17 years or less), 162,131 (13.8%) to older teenagers, and 892,471 (75.7%) to women aged 20–29 years. The sociodemographic characteristics of these women are presented in Table 1.
Each of the age groups demonstrated a seasonal pattern in conception of births (P < .001) (Figure 1). The peak period of conception occurred from November to January. The month with the largest proportion of conceptions was December with 9.1, 8.9, and 9.0% of births to adolescents, older teenagers, and adults, respectively. Adolescents and older teenagers, in addition, presented a second peak in March each accounting for 8.9% of conceptions. The December and March peaks were higher than a hypothetical baseline of 8.33% (P < .05). The nadir of conceptions occurred during the period from June to September for all three age groups. The lowest monthly proportions of 7.5, 7.6, and 7.9% of births were conceived in August to adolescents, older teenagers, and adults, respectively. The August trough for each group was also statistically lower than the hypothetical baseline of 8.33% (P < .05).
Among adolescents, stratification by mother's occupation revealed a secondary early summer (May–June) peak of 8.8% per month, in births conceived to exclusive students (Figure 2). Stratification by father's occupation (not shown) also revealed a similar summer peak; however, information on paternal occupation was not available for about 40% of adolescent conceptions. There was a small secondary early summer peak in conceptions for black and for white but not for Hispanic adolescents (Figure 3). Monthly patterns did not differ by maternal marital status, rural or urban residence, border or nonborder residence, or parity. Stratification of adolescents into two age groups, up to 14 years and 15–17 years, resulted in random fluctuations in monthly proportions of conception among the younger group. The 15–17 year olds maintained the same overall pattern for adolescents. Only 6% of the adolescents were up to 14 years old.
Adolescents were much more likely to initiate prenatal care late or not at all compared with older teenagers and adult women for all months of conception. There was a bimodal pattern of late initiation of prenatal care with modes corresponding to births conceived in April–May and September–October for all age groups (Figure 4).
The final model of the relationship between late initiation of prenatal care and month of conception, adjusting for potential sociodemographic confounders among adolescents, is presented in Table 2. Adolescents conceiving during the months of April [odds ratio (OR) 1.18, 95% confidence interval (CI) 1.11, 1.25] and May (OR 1.14, 95% CI 1.08, 1.21) were most likely to initiate prenatal care late compared with those conceiving during other months. Compared with non-Hispanic white adolescents, Asian adolescents were twice more likely to have late prenatal care (OR 1.94, 95% CI 1.53, 2.46). There was a monotonic decrease in late initiation of prenatal care with age (not shown) and the very young adolescents (11–12 years) were twice more likely than those aged 17 years to have late prenatal care (OR 2.44, 95% CI 1.23, 4.86). Adolescents with previous births were more likely to initiate prenatal care late (OR 1.91, 95% CI 1.83, 2.00) and living in a border county was associated with a higher risk of late prenatal care (OR 1.51, 95% CI 1.45, 1.57).
In our study of more than 1.1 million singleton births to Texas women aged 29 years and less at conception, we found a modest but significant monthly variation in the proportion of conceptions. Seasonal patterns were similar for adolescents and older teenagers and not significantly different from adult women. Peaks in conception of live births occurred in March and December among adolescents and older teenagers. Texas vital data do not include information on abortions. To the best of our knowledge, there is no existing database with information on monthly frequency of abortions in the state. An estimated 24% of clinically recognized pregnancies (excluding miscarriages) end in abortions in Texas.2 We therefore focused on live births, which represent the most frequent clinical pregnancy outcome, with probably the greatest adverse consequences for teenagers. However, based on two studies, it is unlikely that the inclusion of abortions would induce any significant change in the patterns that we observed. One study reported the absence of any seasonality in induced abortions, miscarriages, and ectopic pregnancies in the United States beyond that due to seasonality in conception.10 This finding was also observed for induced abortions in the South Carolina study focusing on adolescents.15
The year-end peak found in all age groups studied is consistent with previous studies in the United States and elsewhere.10,13–15 This finding suggests that biologic processes or common behaviors may account for the seasonal variation. Biologic hypotheses include deterioration of sperm quality during summer, seasonal differences in anterior pituitary-ovarian function caused by changes in the daylight length, and variation in quality of the ovum or endometrial receptivity.12 Increased sexual activity associated with end-of-year holiday festivities has also been postulated13,15 as a possible behavioral explanation for the December peak in conceptions. The exact reasons remain unknown.10,14
In comparison with our findings of a summer nadir in conceptions to adolescents in Texas, the study of South Carolina adolescent conceptions from 1979 to 1986 found a summer peak.15 The South Carolina study included induced abortions, which were not included in the present study. These may explain some of the differences in the findings. Another possible explanation is a changing pattern in seasonality of conception between 1979–1986 and 1994–1998.
When stratified by race, a small secondary early summer peak in conception of live births was identified among black and non-Hispanic white adolescents but not for Hispanic adolescents. This difference may be an important explanation for some of the lack of congruity with the South Carolina study, because unlike South Carolina, most births in Texas are to Hispanic individuals.
The secondary early summer peak among exclusive students suggests that behavioral factors and increased opportunity for sexual activity at the end of the school year may still be important among unemployed students. Rodgers et al21 reported that teenagers were more likely to have their sexual debut in the summer than in other months. The presence of a mid-spring (March) peak common to adolescents and older teenagers, an unexpected finding, could be related to increased sexual activity during the spring break period among these age groups. This hypothesis is supported by the observation that exclusive students present a strong mid-spring peak that was absent in nonexclusive students.
We found an increase in late initiation of prenatal care among births conceived in April and May, consistent with the findings of Petersen and Alexander15 in South Carolina. They suggested that lack of access to school-based prenatal programs during the summer vacation might explain this finding. However, the fact that this pattern is consistent across all age groups in our study suggests that any barriers to accessing reproductive services during the summer are common to all women regardless of age. In addition, the small secondary peak we observed for late initiation of prenatal care for conceptions in September and October is suggestive of barriers to diagnosing pregnancy or accessing health services during the busy end-of-year holiday period, common to all age groups.
The association of late initiation of prenatal care with having a previous live birth and younger age was reported by Petersen and Alexander.15 Interestingly, Asian individuals, though constituting only 0.4% of the adolescent birth population, presented the highest risk among ethnicities for initiating prenatal care late. This finding may suggest less tolerance of adolescent pregnancy among Asian populations. The higher risk of late or no initiation of prenatal care by adolescents living in border counties may be symptomatic of poor access to health care in border counties. Unknown age of fathers was also a risk factor for late initiation and may indicate that adolescents no longer had a relationship with the infant's father, a previously reported risk factor.22
Descriptive demographic data from birth certificates used in our analyses are generally reliable, but some overreporting of early initiation of prenatal care has been reported.23 In addition, there may have been some overreporting of exclusive student status. As the misclassification of these variables is unlikely to differ by month of conception (nondifferential), the associations we have reported may actually be stronger.
Overall, there was no summer peak in the conception of births to adolescents in Texas. Although statistically significant, seasonal differences in conception of births and initiation of prenatal care were modest. These findings suggest that rather than intensifying efforts during specific periods of the year, preventive and reproductive health services for teenagers and adults should be equally available throughout the year.
1. Ventura SJ, Freedman MA. Teenage childbearing in the United States, 1960–1997. Am J Prev Med 2000;19(Suppl):18–25.
2. The Alan Guttmacher Institute. Teenage pregnancy: Overall trends and state-by-state information. New York: Alan Guttmacher Institute, 1999.
3. Singh S, Darroch JE. Adolescent pregnancy and childbearing: Levels and trends in developed countries. Fam Plann Perspect 2000;32:14–23.
4. US Department of Health and Human Services. Healthy People 2010 Conference Edition. Vols. 1 and 2. Washington, DC: US Department of Health and Human Services, 2000.
5. Fraser AM, Brockert JE, Ward RH. Association of young maternal age with adverse reproductive outcomes. N Engl J Med 1995;332:1113–7.
6. Texas Medical Association (Special Task Force on Teen Pregnancy Prevention). Position paper on teen pregnancy prevention. Austin, Texas: Texas Medical Association, March 1999.
7. Kogan MD, Martin JA, Alexander GR, Kotelchuck M, Ventura SJ, Frigoletto FD. The changing pattern of prenatal care utilization in the United States, 1981–1995, using different prenatal care indices. JAMA 1998.279:1623–8.
8. Parnell AM, Rodgers JL. Seasonality of induced abortion in North Carolina. J Biosoc Sci 1998;30:321–32.
9. Kallan JE, Udry JR. Demographic components of seasonality of pregnancy. J Biosoc Sci 1989;21:101–8.
10. Warren CW, Gwinn ML, Rubin GL. Seasonal variation in conception and various pregnancy outcomes. Soc Biol 1986;33:116–26.
11. Warren CW, Gold J, Tyler CW, Jr, Smith JC, Paris AL. Seasonal variation in spontaneous abortions. Am J Public Health 1980;70:1297–9.
12. Rojansky N, Brzezinski A, Schenker JG. Seasonality in human reproduction: An update. Hum Reprod 1992;7:735–45.
13. Tembon AC. Seasonality of births in the North West Province, Cameroon: Implications for family planning programme. Cent Afr J Med 1990;36:90–3.
14. Erhardt CL, Nelson FG, Pakter J. Seasonal patterns of conception in New York City. Am J Public Health 1971;61:2246–58.
15. Petersen DJ, Alexander GR. Seasonal variation in adolescent conceptions, induces abortions, and late initiation of prenatal care. Public Health Rep 1992;107:701–6.
16. David RJ. The quality and completeness of birthweight and gestational age data in computerized birth files. Am J Public Health 1980;70:964–73.
17. Kramer MS, McLean FH, Boyd ME, Usher RH. The validity of gestational age estimation by menstrual dating in term, preterm, and postterm gestations. JAMA 1988;260:3306–8.
18. Zhang J, Bowes WA, Jr. Birth-weight-for-gestational-age patterns by race, sex, and parity in the United States population. Obstet Gynecol 1995;86:200–8.
19. Alexander GR, Tompkins ME, Petersen DJ, Hulsey TC, Mor J. Discordance between LMP-based and clinically estimated gestational age: Implications for research, programs, and policy. Public Health Rep 1995;110:395–402.
20. U.S. Census Bureau. Metropolitan counties in alphabetical order, by state, with metropolitan area title. Washington, DC: U.S. Census Bureau, 1999.
21. Rodgers JL, Harris DF, Vickers KB. Seasonality of first coitus in the United States. Soc Biol 1992;39:1–14.
22. Wiemann CM, Berenson AB, Pino LG, McCombs SL. Factors associated with adolescents' risk for late entry into prenatal care. Fam Plann Perspect 1997;29:273–6.
23. Piper JM, Mitchel EF, Snowden M, Hall C, Adams M, Taylor P. Validation of 1989 Tennessee birth certificate using maternal and newborn hospital records. Am J Epidemiol 1993;137:758–68.