This article is a US Government work and, as such, is in the public domain in the United States of America.Presented as a poster at the 2006 Congress of Epidemiology, a joint meeting of the American College of Epidemiology, Epidemiology Section of the American Public Health Association, and the Society for Epidemiologic Research, June 21–24, 2006, in Seattle, Washington.Presented as a poster at the 9th annual meeting of the National Birth Defects Prevention Network, January 30–February 1, 2006, in Arlington, Virginia.The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention/the Agency for Toxic Substances and Disease Registry.
Folic acid fortification in the United States became mandatory January 1, 1998, to reduce the occurrence of neural tube defects (NTDs). We evaluated the impact of folic acid fortification on orofacial clefts using United States birth certificate data for 45 states and the District of Columbia.
Prevalence ratios (PRs) were calculated comparing orofacial cleft prevalence among births prefortification (1/1990–12/1996) and postfortification (10/1998–12/2002), based on fortification status at conception. The JoinPoint Regression Program and exponentially weighted moving average charts (EWMA) were used to assess the timing of any statistically significant changes in prevalence. Data were stratified by maternal race/ethnicity, age, smoking, and timing of prenatal care.
Orofacial clefts declined following folic acid fortification (PR = 0.94; 95% CI: 0.92–0.96). The EWMA chart flagged a significant decrease in the fourth quarter of 1998. The JoinPoint graph had one change in slope, with a significant quarterly percent change (−0.34) between 1996 and 2002. The decline in orofacial clefts occurred in non-Hispanic Whites but not other racial/ethnic groups, nonsmokers but not women who reported smoking during pregnancy, and women who received prenatal care in the first trimester but not women who began receiving care later in pregnancy.
In 1992, the U.S. Public Health Service recommended that women of reproductive age receive 400 μg of folic acid daily to reduce the risk of neural tube defects (NTDs). They proposed that this could be achieved by fortifying the U.S. food supply (Centers for Disease Control and Prevention, 1992). While the initial motivation behind folic acid fortification was to reduce the number of NTDs, a number of observational studies have suggested that folic acid taken periconceptionally could also reduce the occurrence of cleft lip with or without cleft palate and cleft palate alone (Czeizel et al., 1996, 1999; Itikala et al., 2001; Loffredo et al., 2001; Shaw et al., 1995; Tolarova and Harris, 1995; van Rooij et al., 2004) A few studies have examined the effects of folic acid fortification on the occurrence of orofacial clefts. Ray et al. (2003) assessed the occurrence of orofacial clefts in Ontario, Canada among women who had undergone a maternal serum screening following folic acid fortification of grains and found no statistically significant reduction. Simmons et al. (2004) also found no decrease in prevalence of clefts in Arkansas and folic acid fortification. Using data from multiple birth defects surveillance programs in the United States, Canfield et al. (2005) found a 12% reduction in cleft palate alone and a smaller nonsignificant decline in cleft lip with or without cleft palate following folic acid fortification.
The objective of this study was to evaluate the impact of folic acid fortification in the United States on orofacial clefts using data from U.S. birth certificates. Although birth certificates underestimate the number of orofacial clefts occurring each year, they are very useful for monitoring national trends.
MATERIALS AND METHODS
The Food and Drug Administration authorized optional folic acid fortification of grains in the United States in March 1996, with mandatory fortification beginning January 1, 1998. Enriched grains are required to be fortified with 140 μg of folic acid per 100 g of product (U.S. Food and Drug Administration, 1996). This level was chosen to increase the number of reproductive-age women obtaining 400 μg from fortification and dietary sources combined, while allowing the amount of folic acid being consumed daily to remain <1 mg for most Americans.
In 1989, the U.S. standard birth certificate was revised to include checkboxes for congenital anomalies (Freedman et al., 1988). The revised certificate includes only one checkbox for “cleft lip/palate” to record births affected by cleft lip with or without cleft palate and cleft palate alone. Data for 45 states and the District of Columbia for 1990 to 2002 were included in this analysis. Five states were excluded because of inconsistent reporting of congenital anomaly data. Connecticut and Maryland were excluded because, for multiple years early in the time period assessed, more than 25% of congenital anomalies were reported as “not stated.” New Mexico was excluded because congenital anomalies were not reported for any of the years during the study period. Oklahoma and New York were excluded because congenital anomalies were not reported for 1 year and 4 years, respectively, during this time period. When the data were stratified by maternal smoking status, California, Indiana, and South Dakota were excluded because they did not record data on maternal tobacco use during pregnancy using the standard format. Of the states that were included in the analysis, only New Hampshire did not ascertain Hispanic ethnicity for the entire time period; reporting of Hispanic ethnicity in this state began in 1993. However, New Hampshire was included in the race analysis because the states that did report Hispanic ethnicity accounted for ∼99% of the Hispanic population (Matthews et al., 1998).
To determine the consistency of the birth certificate to detect birth defects after the addition of checkboxes, we looked at the prevalence for other congenital anomalies listed on the birth certificate by year for 1990 through 2002. Birth certificates that mentioned orofacial clefts, NTDs, or were marked only as “other” were excluded; those that mentioned at least 1 of the 18 other specified congenital anomalies were included. NTDs were excluded due to the existing evidence that they have declined following folic acid fortification (Honein et al., 2001). The “other” checkbox was excluded because it likely included many infants with perinatal conditions that are not structural birth defects.
For this study, time periods of interest were defined on the basis of folic acid fortification status at the time of conception. Because folic acid fortification of cereal grains became mandatory in January 1998, the first cohort of births exposed to fortified grains periconceptionally were those born in the fourth quarter of 1998. Prefortification was defined as the years 1990 through 1996. Births between January 1997 and September 1998 (conceived during the period of optional fortification) were excluded from the prevalence ratio calculations comparing pre- and postfortification because it is unknown when the majority of manufacturers began fortifying foods.
Prevalence ratios (PRs) and 95% CIs were calculated to compare the prevalence of orofacial clefts among births before and after fortification. Using information recorded on the birth certificate, the data were stratified by maternal race/ethnicity, maternal age, maternal smoking during pregnancy, and timing of initiation of prenatal care. PRs were calculated for each stratum.
To assess the timing of any statistically significant changes in prevalence, the data were analyzed by quarter of birth using two statistical approaches: exponentially weighted moving average (EWMA) and JoinPoint regression. The EWMA method compares postfortification data to a predefined baseline (prefortification) and detects statistically significant changes from the weighted average (Lucas and Saccucci, 1990). Upper and lower bound limits were created based on the mean and standard deviation, which were calculated using prefortification prevalence (1990–1996). The EWMA statistic crosses either the upper or lower boundary when a significant change from the baseline mean occurs. We reset the statistic to the 1990–1996 baseline mean each time a statistically significant change occurred. Based on the proportion of time pre- versus postfortification, the weight for the weighted average statistic was set to 0.075, which gave 92.5% of the weight to the prefortification years and 7.5% of the weight to the current data (Crowder, 1987).
The second method used was the JoinPoint Regression Program, which detects deviations from a linear slope over time by examining breaks in the slope; the breaks are referred to as joinpoints (Kim et al., 2000; National Cancer Institute, 2003). The program first assumes a zero joinpoint null hypothesis model by fitting a straight line to the data. Next, it permutes the residuals from the null model and adds them back to the null modeled means to obtain permutation data sets. Then it fits the alternative model (with number of joinpoints) to the permuted data sets using the F statistic as a goodness-of-fit measure. Finally, the Monte Carlo method for P value calculation is used to find the significant joinpoints, as well as a quarterly percentage change (QPC) with a 95% CI for each line segment.
To assess the impact of changes in the maternal age distribution of women delivering in the United States, age-adjusted prevalence was calculated by direct standardization and was stratified by race/ethnicity. The standard distribution was based on the age distribution in the year 2000. The standardized age-adjusted prevalence for each quarter was entered into the EWMA program and the JoinPoint Program for comparison with the unadjusted analyses.
The 45 states included in this analysis had a total of 45,926,598 births in the period 1990–2002, ranging from a yearly low of 3,435,192 births in 1997 to a high of 3,655,217 births in 1990. The prevalence of congenital anomalies reported on the birth certificate, excluding NTDs and orofacial clefts, remained stable from 1990 through 2002 (Table 1). Cleft lip/palate was recorded on the birth certificate of 38,232 infants during this time period, with a lower prevalence in the period from 1998 to 2002 (Fig. 1). Orofacial clefts were reported on the birth certificates of 85.2 per 100,000 births during the period before folic acid fortification (1990–1996). This prevalence dropped to 80.2 per 100,000 births after folic acid fortification became mandatory (October 1998–December 2002) in the United States (PR = 0.94; 95% CI: 0.92–0.96) (Table 2). When the years 1991–1996 were used to calculate the prevalence for the period before fortification, the same prevalence ratio was obtained. The first statistically significant point with the EWMA analysis was a decrease in the fourth quarter of 1998, corresponding to the first birth cohort exposed to mandatory folic acid fortification at the time of conception. In total, seven decreases were observed by the EWMA method in the period following folic acid fortification. The JoinPoint Regression Program found similar results, with the best fitting model having one joinpoint at the first quarter of 1996 (95% CI: third quarter 1993–third quarter 1997); a significant QPC of −0.34 (95% CI: −0.51 to −0.17) was observed between 1996 and 2002 (Fig. 2).
Table 1. Prevalence of a Reported Congenital Anomaly on the Birth Certificate, Excluding NTDs and Orofacial Clefts, 45 U.S. States and Washington, D.C. from 1990 through 2002
Table 2. Prevalence and Prevalence Ratios of Orofacial Clefts after Folic Acid Fortification (Oct. 1998–Dec. 2002) Compared with the Prevalence before Fortification (Jan. 1990–Dec. 1996) by Selected Maternal Characteristics and Timing and Direction of Changes in Prevalence, 45 U.S. States and Washington, D.C.
Timing and direction of significant changes by EWMA analysis†
Stratifying by maternal race/ethnicity, only non-Hispanic White mothers had a statistically significant decline in cleft prevalence following folic acid fortification. Both before and after fortification, non-Hispanic Whites had the highest prevalence of orofacial clefts, and non-Hispanic Blacks had the lowest prevalence. The EWMA analysis for non-Hispanic Whites showed results similar to those for all births, with the first decrease in the fourth quarter of 1998. A statistically significant increase was seen in the prevalence of clefts in the first quarter of 1999 in the EWMA analysis for non-Hispanic Blacks. There were no increases or decreases noted in the EWMA chart for the Hispanic or other race/ethnicity category. Using the age-adjusted prevalence the same EWMA results were obtained as the unadjusted prevalence for non-Hispanic Whites and Hispanics; the significant increase for non-Hispanic Blacks changed to the fourth quarter of 2000 in the age-adjusted analysis.
Women who reported smoking during pregnancy had a higher prevalence of infants with orofacial clefts than did women who did not report smoking during pregnancy both before fortification (119.0 vs. 82.2 per 100,000 live births) and after fortification (125.3 vs. 78.3 per 100,000 live births). Comparing data before and after folic acid fortification, the prevalence of orofacial clefts increased among infants whose mothers smoked during pregnancy (PR = 1.05; 95% CI: 0.99–1.11), and declined significantly among infants whose mothers reported no use of tobacco during pregnancy (PR = 0.94; 95% CI: 0.91–0.96). The EWMA analysis stratified by maternal smoking status showed three statistically significant decreases among nonsmoking mothers following folic acid fortification, and one statistically significant increase among smoking mothers.
Similar declines in the prevalence of orofacial clefts following folic acid fortification were observed for mothers <35 years of age (PR = 0.94; 95% CI: 0.92–0.97) and mothers ≥35 years (PR = 0.92; 95% CI: 0.87–0.99). Using the EWMA analysis for each strata of maternal age, some statistically significant decreases were observed following folic acid fortification; however, other increases and decreases were also observed. We repeated the analyses after age adjustment of the prevalence to the maternal age distribution in 2000. Using the age-adjusted prevalence, the first statistically significant decrease on the EWMA chart occurred in the first quarter of 1999. The JoinPoint program selected the model with zero joinpoints as the best fitting model; however, fitting a joinpoint model with one break resulted in a joinpoint at the first quarter of 1996 (95% CI: second quarter 1993–first quarter 1998). The QPC from 1996 to 2002 for the age-adjusted rates was −0.26 (95% CI: −0.44 to −0.08).
Orofacial clefts were more common among infants of mothers who received no prenatal care or who received care starting in the third trimester than among infants whose mothers received prenatal care in the first trimester. A decline in the occurrence of orofacial clefts following folic acid fortification was observed only for infants of mothers who received prenatal care in the first trimester (PR = 0.93; 95% CI: 0.90–0.95). This was also the only group for whom a significant decline was seen in the EWMA analysis.
Analysis of the data from birth certificates suggests that folic acid fortification in the United States was associated with an ∼6% decline in the prevalence of orofacial clefts. This decrease was seen even though there was no noticeable decline in the sensitivity of birth certificate records over the same period and is consistent with results from the National Birth Defects Prevention Network using data from multiple state registries, which found a 12% decline in cleft palate alone and a 5% decline in cleft lip with or without cleft palate (Canfield et al., 2005). The decrease in prevalence in our analysis was limited to infants of non-Hispanic White mothers, nonsmoking mothers, and mothers who received prenatal care in the first trimester of their pregnancy.
A recent review on the use of supplements containing folic acid and dietary intake of folic acid indicated this nutrient may play a role in the prevention of orofacial clefts (Bailey and Berry, 2005). Earlier studies found that high doses of folic acid (6–10 mg) may decrease the occurrence of orofacial clefts, though later studies found similar weak effects with lower doses of folic acid (0.4–1.0 mg) (Czeizel et al., 1996; Itikala et al., 2001; Loffredo et al., 2001; Tolarova and Harris, 1995). However, other studies have not found any association between folic acid intake and orofacial clefts (Czeizel, 1993; Czeizel et al., 1999; Hayes et al., 1996; Hill et al., 1988). The findings of these studies, both positive and negative, come from various design types (e.g., intervention trials, case-control, observational) and data sources (e.g., surveillance systems, orofacial cleft registries).
A few studies have looked at the effects of dietary folic acid alone. Shaw et al. (1995) assessed the effects of fortified cereal consumption without the use of a multivitamin and found a statistically significant decrease in cleft lip with or without cleft palate, indicating that dietary folic acid alone may have an effect. Van Rooij et al. (2004) also found a similar significant decrease in mothers who reported high levels of dietary folate intake and did not use a supplement. They also found a dose-response effect when looking at dietary folate intakes and cleft lip with or without cleft palate: the largest reduction was seen in mothers who had a high dietary folate intake and reported use of a folic acid supplement.
The association between maternal smoking and the occurrence of orofacial clefts has been consistently reported, including a recent meta-analysis of 24 studies (Little et al., 2004). It is thought that smoking decreases blood folate levels and reduces the uptake of antioxidants and micronutrients (Mannino et al., 2003; Mansoor et al., 1997; Piyathilake et al., 1994). And, if smoking increases the risk of clefting due to its impact on blood folate levels, smoking mothers might not receive the full benefit of folic acid fortification because their blood folate levels might not have increased to a level that could reduce the risk of clefting. Alternatively, the association between smoking and orofacial clefts might be due to maternal hypoxia, vasoconstriction of fetal and maternal blood vessels, or a direct effect of a component of tobacco smoke (Bailey et al., 1995; Bronsky et al., 1986; Millicovsky and Johnston, 1981; Nelson, 2001; Ross et al., 1973; van Rooij et al., 2001). And, mothers who use tobacco during pregnancy might differ from nonsmoking mothers in a number of characteristics, including their likelihood of taking multivitamin supplements before and during early pregnancy, their exposure to other substances including alcohol and illicit drugs, and the adequacy of their routine nutritional intake.
When assessing timing of prenatal care in this study, a significant decline in orofacial clefts was seen only among infants of mothers who began prenatal care in the first trimester. Women who receive prenatal care after the first trimester or receive no care at all are less likely to be informed about the potential benefits of folic acid and are also more likely to be of lower socioeconomic status (Ahluwalia and Daniel, 2001; Institute of Medicine, 1988; Pagnini and Reichman, 2000). Women who plan their pregnancies are more likely to take prenatal or multivitamins prior to conception and are more likely to enter prenatal care early (Green-Raleigh et al., 2005; Sable and Wilkinson, 1998; Than et al., 2005). No data are available from the birth certificates on periconceptional intake of vitamins containing folic acid; however, it is likely that a higher proportion of women who began prenatal care in the first trimester also consumed a supplement containing folic acid in the periconceptional period than did women who began their prenatal care in the third trimester because previous studies have shown women who receive early prenatal care are more aware of the benefits (Ahluwalia and Daniel, 2001). Thus, more of the “first trimester prenatal care” mothers might have received a higher dose of folic acid (fortification plus supplements), and this higher dose might be more effective at preventing the occurrence of orofacial clefts.
We used three statistical methods to assess any change in the occurrence of orofacial clefts following folic acid fortification: (1) PRs and 95% CIs comparing the period before folic acid fortification was authorized to that time period after folic acid fortification was mandatory; (2) EWMA analysis of timing of any statistically significant changes in the quarterly prevalence; and (3) a JoinPoint analysis of the timing of a change in the slope of the line for quarterly prevalence. Although each method tested a somewhat different null hypothesis, all three methods suggested a statistically significant reduction in orofacial clefts near the time of the introduction of folic acid fortification. The first statistically significant decline by the EWMA analysis occurred in the fourth quarter of 1998, corresponding to the first births exposed to folic acid fortification at conception. The EWMA takes into account our prior knowledge of when folic acid fortification occurred. We chose the baseline for this analysis as the period prior to any folic acid fortification (1990–1996), and all subsequent time periods were compared to this baseline. The change in slope by the JoinPoint analysis occurred in the first quarter of 1996, which was before folic acid fortification; however, the JoinPoint analysis does not consider any prior knowledge, and thus the prevalence in the quarters after fortification was not being compared with any predefined baseline. Instead the program looked for changes in the slope at any point in the 13 year period from 1990 through 2002. An additional explanation for the joinpoint occurring before fortification is that it is not clear when manufacturers began fortifying grains. Two studies have found that blood folate levels began increasing as early as 1997, suggesting that some portion of grains were being fortified before mandatory fortification was in effect (Jacques et al., 1999; Lawrence et al., 1999).
The percentage of orofacial clefts that are prenatally diagnosed is relatively low, with 0% of cleft palate cases and 14% of cleft lip or palate cases being prenatally diagnosed (Forrester et al., 1998). Due to treatment and corrective surgery, termination rates for an affected pregnancy are generally low in the United States for isolated orofacial clefts. From 1987 to 1996 there was a slight increase in the percent of cases that were prenatally diagnosed, which may be attributable to the improved quality of prenatally diagnostic tests and the increased availability of the tests (Forrester et al., 1998). Even with this increase in prenatal detection, the proportion of elective terminations among orofacial cleft cases has stayed relatively the same (Forrester et al., 1998). Therefore, it is unlikely that termination rates would also increase with improved detection over the period of the study and account for the decline we found.
Recent studies have shown that influenza and in particular a fever (hyperthermia) associated with the influenza may play an etiologic role in the origin of orofacial clefts (Acs et al., 2006; Metneki et al., 2005). One study found that there was a decrease in the prevalence of orofacial clefts among women with influenza who took an antifever medication; when folic acid was taken in addition to the antifever medication, a preventative effect was seen (Acs et al., 2005). However, influenza surveillance in the United States indicates that the percentage of patients presenting with influenza-like illnesses has remained fairly constant over the period of the study (Brammer et al., 2000, 2002). Therefore, while influenza may play a role in the formation of orofacial clefts, it is unlikely that it is responsible for the decrease we found.
Birth certificates represent a consistent source of data on selected variables, and the format has been consistent from 1989 through 2002 (Freedman et al., 1988). A birth certificate is completed for ∼4 million births per year, and the 45 states and District of Columbia included in this analysis represent ∼3.5 million births per year. Orofacial clefts are usually recognized at birth and result in a relatively prompt diagnosis. And, to improve the strength of our findings, we used multiple approaches to assess the timing of any change in prevalence. In this manner, we can be sure that our findings are not dependent only on our selection of statistical methods.
Despite the consistency of birth certificate data, it is important to acknowledge the limitations of using these data. Studies have shown that the quality of birth certificate data can vary significantly, depending on the variable of interest, although several studies have shown that demographic variables are more reliable than other variables ascertained from the birth certificate (Adams, 2001; Ananth, 2005; DiGiuseppe et al., 2002; Dobie et al., 1998; Piper et al., 1993). The relatively low sensitivity of using birth certificates in ascertaining most birth defects, including orofacial clefts, which were found to have a sensitivity of 38% and a positive predictive value of 98%, represents one of several limitations of this study (Watkins et al., 1996). In comparing our postfortification prevalence on the birth certificate to the postfortification prevalence found in the National Birth Defects Prevention Network, we were able to ascertain 56% of the prevalence the Network data were able to achieve (Canfield et al., 2005). Looking at the prevalence of 11 active surveillance systems combined and comparing it with our prevalence, we detected 48% of the prevalence from those systems (Centers for Disease Control and Prevention, 2006). Due to the low sensitivity of birth certificate use in detecting orofacial clefts we were not able to ascertain all the cases and therefore the decline that we observed may be an underestimate of the actual decrease.
A second limitation of using birth certificate data is that we may be achieving statistical significance due to large numbers when there may in fact be no effect. Third, we were very limited in that birth certificates include data only on all orofacial clefts combined and do not differentiate between isolated and syndromic cases, and although several studies have suggested different effects for cleft palate alone compared to cleft lip with or without cleft palate, we were not able to evaluate them separately (Canfield et al., 2005; Itikala et al., 2001; Loffredo et al., 2001; Werler et al., 1999). Fourth, as mentioned previously, we do not have any information on the trimesters of maternal smoking or any information on folic acid consumption in the periconceptional period. With no information on folic acid consumption, we were unable to separate the quantity and source of folic acid (dietary folate, vitamin supplementation, fortified grains, and/or other fortified foods) consumed. Finally, genetic factors are extremely important in the etiology of orofacial clefts, and we did not have any information on genetic factors or even family history of orofacial clefts (Lettieri, 2000; Saxen, 1974; Tolarova, 1987).
The magnitude of decline observed for orofacial clefts in this study is much smaller than that found for NTDs (Honein et al., 2001; Williams et al., 2002). Additional studies are necessary to better understand what contribution folic acid fortification has made to orofacial cleft prevention and whether various sources, as well as higher doses of folic acid, could further prevent cases of orofacial clefts. Nevertheless, the results of this study suggest an additional benefit of this public health intervention, as well as the potential effects of smoking on blood folate levels and orofacial clefts.
We would like to thank Owen Devine and Matthew Strickland, at the National Center for Birth Defects and Developmental Disabilities, for all their input and assistance.