Indiana birth certificate

Birth certificates are completed for all live-born infants in the US. The US standard birth certificate collects data on many factors, including presence of defects at birth. Although limitations to the quality of birth defect data from birth certificates have been noted, birth certificates are a stable data source that have both some exposures of interest and birth defect outcome information for approximately four million births per year. (1-3) The validity of using birth certificate data to support or refute findings from earlier studies and to identify possible etiologic risk factors for birth defects has not been assessed.

To test the validity of birth certificate data, we examined the relationship between maternal smoking and birth defects in these data. In 1997 and 1998, maternal smoking data and birth defect data were collected on birth certificates by 45 states, New York City, and the District of Columbia; therefore, data on more than three million births per year were available for this analysis. Maternal smoking has diverse, well-recognized adverse effects on the fetus, but information about its association with major malformations is limited and controversial. (4,5) Because approximately 24% of 18- to 44-year-old US women are current smokers, further elucidating a role of maternal smoking in major malformations is important. (6)

METHODS

We used public-use natality data tapes (National Vital Statistics System, National Center for Health Statistics) for all live-born infants to residents of 45 states, New York City, and the District of Columbia born in the US during 1997 and 1998. We restricted the analysis to 13 defects/defect categories from the birth certificate that are likely to be observable at birth and that are relatively well-defined categories of defect: anencephaly, spina bifida, hydrocephaly, microcephaly, rectal atresia/stenosis, tracheo-esophageal fistula/ esophageal atresia, omphalocele/gastroschisis, cleft lip/cleft palate, clubfoot, polydactyly/syndactyly/ adactyly, congenital diaphragmatic hernia, renal agenesis, and Down syndrome. We excluded malformed genitalia, other central nervous system anomalies, other gastrointestinal anomalies, other urogenital anomalies, heart malformations, other circulatory/respiratory anomalies, and other musculoskeletal/integumental anomalies because the range of defects that may be included in these broad categories is not clear, making any meaningful comparison with the literature difficult. Because individual defects are rare events, any infant whose birth certificate did not indicate the presence of a specific defect was assumed, for analytic purposes, not to have that defect. No attempt was made to control for multiple defects. We excluded birth certificates for residents of New Mexico because New Mexico does not require reporting of congenital anomalies on the birth certificate.

Maternal smoking during pregnancy was ascertained from the yes/no question on the birth certificate. We excluded from this analysis infants with birth certificates on which the maternal smoking information was left blank or was coded as unknown. We excluded birth certificates for residents of California, New York State (except for New York City), Indiana, and South Dakota because those states do not require reporting of maternal smoking in the standard format on the birth certificate. Birth certificates record the number of cigarettes smoked per day, but they do not contain information about the timing of smoking during pregnancy. We analyzed smoking dose in four exposure categories: [greater than or equal to] 21 cigarettes per day, 11 to 20 cigarettes per day, six to 10 cigarettes per day, and one to five cigarettes per day. All exposure levels were compared to the referent (nonsmokers).

We calculated the prevalence ratio for the association between maternal smoking and selected defects. Because demographic variables may confound this relationship, we adjusted for maternal age (< 30 years, [greater than or equal to] 30 years), maternal race/ethnicity (Hispanic, non-Hispanic black, non-Hispanic white, other), and maternal education (0 to 11 years, [greater than or equal to] 12 years). The adjusted prevalence ratios and 95% confidence intervals were calculated using the test-based Mantel-Haenszel method with SAS software (version 6.09). We calculated prevalence ratios for all infants and by gender of the infant, because some evidence suggests that the male fetus may be more susceptible to the effects of tobacco. (7,8)

We compared the effect estimates using birth certificate data with published studies on smoking and birth defects. We identified previous studies by searching the MEDLINE database for publications from 1966 to early 2000 on smoking and each of the selected defects (by subject heading and keyword), and by tracing references in identified studies. We excluded studies that did not include either comparable measures of association (odds ratios, risk ratios, or prevalence ratios) between maternal smoking and the defect or adequate data to calculate an effect estimate. We excluded previous studies based solely on birth certificate data from this comparison. Both the direction and strength of the effect estimate were considered in this comparison.

RESULTS

There were 3,051,349 live births in 1997 and 3,110,157 live births in 1998 to residents of 45 US states, the District of Columbia, and New York City. Six of the 13 defects analyzed had statistically significant positive associations with maternal smoking in 1997-1998 birth certificate data: hydrocephaly (adjusted prevalence ratio [PR] = 1.24; 95% confidence interval [CI] = 1.08, 1.43), microcephaly (PR 1.47; 95% CI 1.15, 1.88), omphalocele/gastroschisis (PR 1.37; 95% CI 1.22, 1.53), cleft lip/palate (PR 1.35; 95% CI 1.25, 1.45), clubfoot (PR 1.62; 95% CI 1.49, 1.75), and polydactyly/syndactyly/adactyly (PR 1.33; 95% CI 1.23, 1.43) (Table 1). Most of these had consistent effect estimates for 1997 and 1998; microcephaly was an exception, with a strong association with maternal smoking in 1998 and no apparent association with maternal smoking in 1997.

Four of these defects (omphalocele/gastroschisis, oral clefts, polydactyly/syndactyly/adactyly, and clubfoot) were positively associated with maternal smoking among both male and female infants (Table 2). For infants with each of these four defects, the adjusted prevalence ratio was higher for male infants than for female infants. Microcephaly, hydrocephaly, and congenital diaphragmatic hernia showed a stronger association with maternal smoking for females than for males. The 95% confidence intervals for both microcephaly and hydrocephaly included the null value for males, and the 95% confidence intervals for congenital diaphragmatic hernia included the null value for both males and females.

Although there was no clear dose-response effect, there was an indication of a stronger association between heavy maternal smoking ([greater than or equal to] 20 cigarettes per day) and infants with clubfoot, hydrocephaly, or oral clefts (Table 3). No dose-response effect of maternal smoking was observed for infants with omphalocele/ gastroschisis, microcephaly, or polydactyly/syndactyly/ adactyly. Renal agenesis increased nearly twofold among infants whose mothers smoked more than 20 cigarettes per day, but the association was not statistically significant.

DISCUSSION

Using US birth certificate data for 1997-1998, we found a positive association between reported maternal smoking and hydrocephaly, microcephaly, omphalocele/ gastroschisis, oral clefts, polydactyly/syndactyly/ adactyly, and clubfoot. There was a stronger positive association observed among male infants than female infants for omphalocele/gastroschisis, oral clefts, polydactyly/syndactyly/adactyly, and clubfoot, and a stronger association for hydrocephaly and microcephaly among female infants than among male infants.

To assess the validity of birth certificate data for identifying birth defect risk factors, we compared the estimates from our analysis with previous studies of maternal smoking and birth defects (Table 4). We excluded literature based solely on birth certificate data from this table.