Random error and undercounting in birth defects surveillance data: implications for inference

Louis AM, Agopian AJ. Scientific impact of the National Birth Defects Prevention Network multistate collaborative publications

BACKGROUND

Given the lack of a national, population-based birth defects surveillance program in the United States, the National Birth Defects Prevention Network (NBDPN) has facilitated important studies on surveillance, research, and prevention of major birth defects. We sought to summarize NBDPN peer-reviewed publications and their impact.

METHODS

We obtained and reviewed a curated list of 49 NBDPN multistate collaborative publications during 2000-2022, as of December 31, 2022. Each publication was reviewed and classified by type (e.g., risk factor association analysis). Key characteristics of study populations and analytic approaches used, along with publication impact (e.g., number of citations), were tabulated.

RESULTS

NBDPN publications focused on prevalence estimates (N = 17), surveillance methods (N = 11), risk factor associations (N = 10), mortality and other outcomes among affected individuals (N = 6), and descriptive epidemiology of various birth defects (N = 5). The most cited publications were those that reported on prevalence estimates for a spectrum of defects and those that assessed changes in neural tube defects (NTD) prevalence following mandatory folic acid fortification in the United States.

CONCLUSIONS

Results from multistate NBDPN publications have provided critical information not available through other sources, including US prevalence estimates of major birth defects, folic acid fortification and NTD prevention, and improved understanding of defect trends and surveillance efforts. Until a national birth defects surveillance program is established in the United States, NBDPN collaborative publications remain an important resource for investigating birth defects and informing decisions related to health services planning of secondary disabilities prevention and care.

Impact evaluation of Zika epidemic on congenital anomalies registration in Brazil: An interrupted time series analysis

This study aimed to assess the impact of the Zika epidemic on the registration of birth defects in Brazil. We used an interrupted time series analysis design to identify changes in the trends in the registration of congenital anomalies. We obtained monthly data from Brazilian Live Birth Information System and used two outcome definitions: 1) rate of congenital malformation of the brain and eye (likely to be affected by Zika and its complications) 2) rate of congenital malformation not related to the brain or eye unlikely to be causally affected by Zika. The period between maternal infection with Zika and diagnosis of congenital abnormality attributable to the infection is around six months. We therefore used September 2015 as the interruption point in the time series, six months following March 2015 when cases of Zika started to increase. For the purposes of this analysis, we considered the period from January 2010 to September 2015 to be "pre-Zika event," and the period from just after September 2015 to December 2017 to be "post-Zika event." We found that immediately after the interruption point, there was a great increase in the notification rate of congenital anomalies of 14.9/10,000 live births in the brain and eye group and of 5.2/10,000 live births in the group not related with brain or eye malformations. This increase in reporting was in all regions of the country (except in the South) and especially in the Northeast. In the period "post-Zika event", unlike the brain and eye group which showed a monthly decrease, the group without brain or eye malformations showed a slow but significant increase (relative to the pre-Zika trend) of 0.2/10,000 live births. These findings suggest an overall improvement in the registration of birth malformations, including malformations that were not attributed to Zika, during and after the Zika epidemic.

A quality assessment of reporting sources for microcephaly in Utah, 2003 to 2013

BACKGROUND

Obtaining accurate microcephaly prevalence is important given the recent association between microcephaly and Zika virus. Assessing the quality of data sources can guide surveillance programs as they focus their data collection efforts. The Utah Birth Defect Network (UBDN) has monitored microcephaly by data sources since 2003. The objective of this study was to examine the impact of reporting sources for microcephaly surveillance.

METHODS

All reported cases of microcephaly among Utah mothers from 2003 to 2013 were clinically reviewed and confirmed. The UBDN database was linked to state vital records and hospital discharge data for analysis. Reporting sources were analyzed for positive predictive value and sensitivity.

RESULTS

Of the 477 reported cases of microcephaly, 251 (52.6%) were confirmed as true cases. The UBDN identified 94 additional cases that were reported to the surveillance system as another birth defect, but were ultimately determined to be true microcephaly cases. The prevalence for microcephaly based on the UBDN medical record abstraction and clinical review was 8.2 per 10,000 live births. Data sources varied in the number and accuracy of reporting, but a case was more likely to be a true case if identified from multiple sources than from a single source.

CONCLUSION

While some reporting sources are more likely to identify possible and true microcephaly cases, maintaining a multiple source methodology allows for more complete case ascertainment. Surveillance programs should conduct periodic assessments of data sources to ensure their systems are capturing all possible birth defects cases. Birth Defects Research (Part A) 106:983-988, 2016. © 2016 Wiley Periodicals, Inc.

Retrospective Assessment of Cost Savings From Prevention: Folic Acid Fortification and Spina Bifida in the U.S

INTRODUCTION

Although fortification of food with folic acid has been calculated to be cost saving in the U.S., updated estimates are needed. This analysis calculates new estimates from the societal perspective of net cost savings per year associated with mandatory folic acid fortification of enriched cereal grain products in the U.S. that was implemented during 1997-1998.

METHODS

Estimates of annual numbers of live-born spina bifida cases in 1995-1996 relative to 1999-2011 based on birth defects surveillance data were combined during 2015 with published estimates of the present value of lifetime direct costs updated in 2014 U.S. dollars for a live-born infant with spina bifida to estimate avoided direct costs and net cost savings.

RESULTS

The fortification mandate is estimated to have reduced the annual number of U.S. live-born spina bifida cases by 767, with a lower-bound estimate of 614. The present value of mean direct lifetime cost per infant with spina bifida is estimated to be $791,900, or $577,000 excluding caregiving costs. Using a best estimate of numbers of avoided live-born spina bifida cases, fortification is estimated to reduce the present value of total direct costs for each year's birth cohort by $603 million more than the cost of fortification. A lower-bound estimate of cost savings using conservative assumptions, including the upper-bound estimate of fortification cost, is $299 million.

CONCLUSIONS

The estimates of cost savings are larger than previously reported, even using conservative assumptions. The analysis can also inform assessments of folic acid fortification in other countries.

Adjusting for bias due to incomplete case ascertainment in case-control studies of birth defects

Case-control studies of birth defects might be subject to selection bias when there is incomplete ascertainment of cases among pregnancies that are terminated after a prenatal diagnosis of the defect. We propose a simple method to estimate inverse probability of selection weights (IPSWs) for cases ascertained from both pregnancies that end in termination and those that do not end in termination using data directly available from the National Birth Defects Prevention Study and other published information. The IPSWs can then be used to adjust for selection bias analytically. We can also allow for uncertainty in the selection probabilities through probabilistic bias analysis. We provide an illustrative example using data from National Birth Defects Prevention Study (1997-2009) to examine the association between prepregnancy obesity (body mass index, measured as weight in kilograms divided by height in meters squared, of ≥30 vs. <30) and spina bifida. The unadjusted odds ratio for the association between prepregnancy obesity and spina bifida was 1.48 (95% confidence interval: 1.26, 1.73), and the simple selection bias-adjusted odds ratio was 1.26 (95% confidence interval: 1.04, 1.53). The probabilistic bias analysis resulted in a median adjusted odds ratio of 1.22 (95% simulation interval: 0.97, 1.47). The proposed method provides a quantitative estimate of the IPSWs and the bias introduced by incomplete ascertainment of cases among terminated pregnancies conditional on a set of assumptions.

The increasing incidence of foetal echogenic congenital lung malformations: an observational study

OBJECTIVES

The aim of this study was to investigate the incidence of congenital lung malformations over the past 19 years. Congenital lung malformations (CLM) are a heterogeneous group of lung abnormalities. The antenatal diagnosis is important for foetal and neonatal management but there have been no studies examining whether the reported incidence of this abnormality is constant.

METHODS

A retrospective cross-sectional study of cases identified from the Wessex Antenatally Detected Anomalies (WANDA) register 1994-2012.

RESULTS

One hundred and thirty-three cases of CLM in 524 372 live and stillbirths were identified. All but seven were identified on antenatal ultrasound. During the early registry (1994-1998) the average incidence of CLM was 1.27 per 10,000 births. By the last 4 years (2008-2012) this had risen to 4.15 per 10,000 births, with a progressive increase during the intervening years.

CONCLUSION

There was over a three-fold increase in the antenatally detected CLM in the Wessex region 1994-2012. Comparison with the antenatal detection of diaphragmatic hernia suggests that this is a true rise in incidence rather than an artefactual increase due to increased antenatal recognition secondary to improved ultrasound resolution and operator experience. These results have clinical and cost implications for practitioners of foetal medicine, neonatology and paediatric surgery services.

A systematic evaluation of different methods for calculating adolescent vaccination levels using immunization information system data

OBJECTIVE

Immunization information systems (IISs) are valuable surveillance tools; however, population relocation may introduce bias when determining immunization coverage. We explored alternative methods for estimating the vaccine-eligible population when calculating adolescent immunization levels using a statewide IIS.

METHODS

We performed a retrospective analysis of the Michigan State Care Improvement Registry (MCIR) for all adolescents aged 11-18 years registered in the MCIR as of October 2010. We explored four methods for determining denominators: (1) including all adolescents with MCIR records, (2) excluding adolescents with out-of-state residence, (3) further excluding those without MCIR activity ≥ 10 years prior to the evaluation date, and (4) using a denominator based on U.S. Census data. We estimated state- and county-specific coverage levels for four adolescent vaccines.

RESULTS

We found a 20% difference in estimated vaccination coverage between the most inclusive and restrictive denominator populations. Although there was some variability among the four methods in vaccination at the state level (2%-11%), greater variation occurred at the county level (up to 21%). This variation was substantial enough to potentially impact public health assessments of immunization programs. Generally, vaccines with higher coverage levels had greater absolute variation, as did counties with smaller populations.

CONCLUSION

At the county level, using the four denominator calculation methods resulted in substantial differences in estimated adolescent immunization rates that were less apparent when aggregated at the state level. Further research is needed to ascertain the most appropriate method for estimating vaccine coverage levels using IIS data.

Inclusion of non-viable neonates in the birth record and its impact on infant mortality rates in Shelby County, Tennessee, USA

Rates of infant death are one of the most common indicators of a population's overall health status. Infant mortality rates (IMRs) are used to make broad inferences about the quality of health care, effects of health policies and even environmental quality. The purpose of our study was threefold: i) to examine the characteristics of births in the area in relation to gestational age and birthweight; ii) to estimate infant mortality using variable gestational age and/or birthweight criteria for live birth, and iii) to calculate proportional mortality ratios for each cause of death using variable gestational age and/or birthweight criteria for live birth. We conducted a retrospective analysis of all Shelby County resident-linked birth and infant death certificates during the years 1999 to 2004. Descriptive test statistics were used to examine infant mortality rates in relation to specific maternal and infant risk factors. Through careful examination of 1999-2004 resident-linked birth and infant death data sets, we observed a disproportionate number of non-viable live births (≤20 weeks gestation or ≤350 grams) in Shelby County. Issuance of birth certificates to these non-viable neonates is a factor that contributes to an inflated IMR. Our study demonstrates the complexity and the appropriateness of comparing infant mortality rates in smaller geographic units, given the unique characteristics of live births in Shelby County. The disproportionate number of pre-viable infants born in Shelby County greatly obfuscates neonatal mortality and de-emphasizes the importance of post-neonatal mortality.

Do the results of pregnancy registries contradict one another?

Malformation Risks of Antiepileptic Drugs in Pregnancy: A Prospective Study from the UK Epilepsy and Pregnancy Register

J Morrow, A Russell, E Guthrie, L Parsons, I Robertson, R Waddell, B Irwin, R C, McGivern, P J Morrison, J Craig

J Neurol Neurosurg Psychiatry 2006;77;193–198

Objective

To assess the relative risk of major congenital malformation (MCM) from in utero exposure to antiepileptic drug (AED).

Methods

Prospective data collected by the UK Epilepsy and Pregnancy Register were analyzed. The presence of MCMs recorded within the first 3 months of life was the main outcome measure.

Results

Full outcome data were collected on 3,607 cases. The overall MCM rate for all AED exposed cases was 4.2% (95% confidence interval (CI): 3.6–5.0%). The MCM rate was higher for polytherapy (6.0%) ( n = 770) than for monotherapy (3.7%) ( n = 2,598) (crude odds ratio (OR) = 1.63 ( p = 0.010), adjusted OR = 1.83 ( p = 0.002)). The MCM rate for women with epilepsy who had not taken AEDs during pregnancy ( n = 239) was 3.5% (1.8–6.8%). The MCM rate was greater for pregnancies exposed only to valproate (6.2% (95% CI: 4.6–8.2)) than only to carbamazepine (2.2% (1.4–3.4%)) (OR = 2.78 ( p < 0.001)); adjusted OR = 2.97 ( p < 0.001)). There were fewer MCMs for pregnancies exposed only to lamotrigine than only to valproate. A positive dose response for MCMs was found for lamotrigine ( p = 0.006). Polytherapy combinations containing valproate carried a higher risk of MCM than combinations not containing valproate (OR = 2.49 (1.31—4.70)).

Conclusions

Only 4.2% of live births to women with epilepsy had an MCM. The MCM rate for polytherapy exposure was greater than for monotherapy exposure. Polytherapy regimens containing valproate had significantly more MCMs than those not containing valproate. For monotherapy exposures, carbamazepine was associated with the lowest risk of MCM. Maternal Valproate Dosage and Foetal Malformations Vajda FJ, Eadie MJ Acta Neurol Scand 2005;112(3):137–143

Objective

To study the possible dose dependence of the foetal malformation rate after exposure to sodium valproate in pregnancy.

Methods

Analysis of records of all foetuses in the Australian Registry of Antiepileptic Drugs in Pregnancy exposed to valproate, to carbamazepine, lamotrigine or phenytoin in the absence of valproate, and to no antiepileptic drugs.

Results

The foetal malformation rate was higher ( p < 0.05) in the 110 foetuses exposed to valproate alone (17.1%), and in the 165 exposed to valproate, whether alone or together with the other antiepileptic drugs (15.2%), than in the 297 exposed to the other drugs without valproate (2.4%). It was also higher ( p < 0.10) than in the 40 not exposed to antiepileptic drugs (2.5%). Unlike the situation for the other drugs, the malformation rate in those exposed to valproate increased with increasing maternal drug dosage ( p < 0.05). The rate was not altered by simultaneous exposure to the other drugs. Valproate doses exceeding 1,400 mg per day seemed to be associated with a more steeply increasing malformation rate than at lower doses and with a different pattern of foetal malformations.

Conclusions

Foetal exposure to valproate during pregnancy is associated with particularly high, and dose-dependent risks of malformation compared with other antiepileptic drugs, and may possibly involve different teratogenetic mechanisms.

Feasibility of using the national hospital discharge survey to estimate the prevalence of selected birth defects

BACKGROUND

Nationally representative data on the prevalence of certain birth defects are largely unavailable. We evaluated the feasibility of using data from the National Hospital Discharge Survey (NHDS) to describe the prevalence of selected birth defects.

METHODS

All live births recorded in the NHDS during 1999-2001 were included. The prevalence for selected birth defects was calculated using weighted ratio estimators. Prevalence ratios comparing the NHDS estimates to published national estimates from the National Birth Defects Prevention Network (NBDPN) were calculated.

RESULTS

With the exception of common truncus, the NHDS prevalence for the selected defects was consistently lower than the NBDPN estimates. The prevalence ratios ranged from 0.38 for trisomy 18 and anopthalmia/micropthalmia to 1.16 for common truncus. The NHDS prevalence estimates for spina bifida without anencephaly (PR 0.89, 95% CI: 0.57-1.22) and gastroschisis/omphalocele (PR 0.94, 95% CF: 0.48-1.40) most closely approximated the NBDPN estimates.

CONCLUSIONS

NHDS data underestimate the prevalence of most birth defects. Additional research is needed to determine whether NHDS estimates may be useful for evaluating trends in certain conditions. Surveillance systems employing active case-finding continue to provide more accurate estimates of birth defects prevalence.

Evaluating the accuracy of Malformations Surveillance Program in detecting virilization due to congenital adrenal hyperplasia

Malformations surveillance programs of newborn infants have been developed as a method for identifying serious and relatively common birth defects. The virilization of newborn infants with the classic 21-hydroxylase form of congenital adrenal hyperplasia must be identified early if the associated metabolic crisis in the perinatal period is to be prevented. We compared the detection of virilization associated with 21-hydroxylase congenital adrenal hyperplasia in infants by three methods: an 'active' malformations surveillance of medical records at a large urban hospital; routine medical care by examining physicians; and newborn biochemical screening of blood samples. The experience at a large maternity center in Boston, since 1972, showed that pediatricians often recognized affected females (6/6), but not males (0/2); the state newborn screening program, begun in 1990, identified correctly all affected males and females. The Active Malformations Surveillance Program was the least effective screening method, identifying four of six affected females and neither of the affected males. The low rate of detecting affected females by the Surveillance Program was attributed to a failure to sensitize the research assistants to the importance of physicians' notations regarding the signs and symptoms of virilization. The failure of examining physicians, and thereby, the malformations surveillance program, to detect virilized newborn males was due to the lack of consistent associated physical features. These comparisons between these three methods of detection can be used to design and improve malformations surveillance programs.