Birth defects, including neural tube defects (NTD), are one of the leading causes of infant and neonatal mortality in countries undergoing an epidemiological transition because of declines in infant mortality and improvements in the environment( Reference Black, Cousens and Johnson 1 ). Globally, NTD prevalence is estimated to be over 300 000 new cases per annum, with over 40 000 deaths and 2·3 million disability-adjusted life years( Reference Christianson, Howson and Modell 2 ). Further, in low-income countries, 17 % to 70 % of neonatal deaths from birth defects are attributed to NTD( Reference Blencowe, Cousens and Modell 3 ). However, scanty and fragmented surveillance information hinders the ability to adequately determine the prevalence of NTD in more than 11 million births per year in Latin America( 4 ). Birth defects surveillance information is vital for monitoring and evaluating the impact of prevention and intervention programmes.
Observational studies reinforce the evidence from clinical trials that have shown conclusively that consumption of staples fortified with folic acid and adequate periconceptional folic acid supplementation reduce the risk of NTD( Reference Blencowe, Cousens and Modell 3 , 5 – Reference Berry, Li and Erickson 7 ). Currently, all Latin American countries except Venezuela have mandatory fortification legislation and programmes aimed at decreasing conditions related to deficiencies of folic acid and other micronutrients. Few countries, however, have established monitoring and evaluation components to assess the impact of their NTD prevention programmes and fewer still have identified time trends pre- and post-fortification( 8 ).
The present review had two main objectives: (i) to summarize NTD prevalence and time trend data in Latin American countries; and (ii) to compare pre- and post-fortification periods to assess the impact of folic acid fortification on NTD prevalence in these countries.
Methods
We carried out a review of studies published between 1990 and 2010 to identify reports containing information on NTD prevalence and, when appropriate, the time periods in which fortification programmes were initiated. We searched CINAHL, Cochrane Collaboration, EMBASE, Global Health, Google Scholar, Ingenta, Medline, the Pan-American Health Organization search engine, PubMed, Red de Revistas Científicas de América Latina y el Caribe, España y Portugal, Revista Médica de Chile, the Latin American and Caribbean Health Sciences Literature (LILACS) and Web of Science for published information. The review was conducted between March 2007 and December 2010. The titles and abstracts were reviewed to determine if the content was related to NTD prevalence and/or folic acid fortification in the region. Studies identified for potential inclusion were assessed by two of the co-authors.
We considered for inclusion observational studies (cohort, case–control, cross-sectional and ecological studies) that included the following information: a clear description of the study population and methods (case definition and methods of case ascertainment, demographics); diagnosis of NTD in live-born infants in the first year of life and in stillbirths; population setting (clinic, hospital or population derived); number of type-specific cases and/or total cases; prevalence rates or ratios; limitations and biases; and any information regarding folic acid fortification interventions, when available. Studies were scored independently by two of the co-authors based on the following aspects of study quality: (i) clarity of case definition; (ii) methods of case ascertainment; (iii) reported prevalence rates or ratios; (iv) number of live births; (v) study limitations; and (vi) biases. Each category contributed one point. The scores of each independent reviewer were averaged, and the articles or reports were then classified based on their total score as ‘very good’ (score = 5–6), ‘good’ (score = 3–4), ‘satisfactory’ (score = 2) or ‘poor’ (score = 0–1). Only studies classified as good or very good were included in the review.
We excluded publications with fewer than 5000 live births per year; those that did not report the number of cases or reported the NTD prevalence without inclusion of the total number of births; those that reported graphs without point estimates; publications in which the information was based on mortality only; and/or publications that included only one type of NTD (i.e. anencephaly or spina bifida or encephalocele). Publications containing total NTD cases only were included when the methodology specifically defined at least two forms of NTD.
Most of the publications on NTD used data from national or regional registries and surveillance/vital statistics systems. The registries cited most often were the Latin American Collaborative Study of Congenital Malformations (ECLAMC)( Reference Castilla and Orioli 9 ), the Costa Rica Congenital Malformations Registry (CARCM)( Reference Benavides-Lara, Barboza-Arguello and Umaña-Solis 10 ), the Cuban Congenital Malformations Registry (RECUMAC)( 11 ), the Mexican External Malformations Epidemiological Surveillance Registry (RYVEMCE)( Reference Mutchnick, Lisker and Babinsky 12 ) and the Puerto Rico Congenital Malformations Surveillance Systems( 13 ). In addition, reports from Argentina, Brazil and Mexico also used data sources based on national or local hospital discharge data( 14 – Reference Aguiar, Campos and Aguiar 16 ). Additionally, we included information from institutional reports published by the Universidad de San Carlos in Guatemala, the Costa Rica Ministry of Health, the Dominican Republic Ministry of Health and the Fundación de Niños Saludables in Honduras.
Analysis
To increase stability of the NTD prevalence estimates, we grouped years of reported cases and births when possible. Time trends were estimated from those publications that provided at least four time points. We present basic trends in prevalence by computing prevalences (with 95 % confidence intervals) of NTD over time within source using exact Poisson limits( Reference Daly 17 , Reference Price 18 ). We analysed basic time trends using Poisson and negative binomial distribution models for each source separately:
$${\rm{log}}({\rm{NTD}})\: \approx \:{\rm{intercept}}\: + \:\beta \:\times \:{\rm{year}}\: + \:{\rm{log}}({\rm{LB}}),\eqno\rm$$
where NTD is the case count and LB is the live birth count. Results of the model provided a summary of temporal trends as prevalence ratios, expressed as relative changes in prevalence per unit changes in time.
The impact of fortification was evaluated by comparing prevalences before and after the onset of fortification. The most basic comparison was the calculation of a single prevalence rate ratio (PRR)( Reference Greenland 19 , Reference Pearce 20 ), with approximate limits for 95 % confidence:
$$ {\rm{PR}}{{{\rm{R}}}_{{\rm{Trend}}}}\: = \:{\rm{Prevalenc}}{{{\rm{e}}}_{{\rm{Post}}}}\:/\:{\rm{Prevalenc}}{{{\rm{e}}}_{{\rm{Pre}}}}\,. \eqno\rm$$
For countries with sufficient data for both pre- and post-fortification periods, we used a more complex model:
$${{\rm{log}}({\rm{NTD}})\: \approx \ {\rm{intercept}}\: + \:{{\beta }_{{\rm{fort}}}}\:\times \:{\rm{fort}}\: + \:{{\beta }_{{\rm{year}}}}\:\times \:{\rm{year}}\: \cr + \:{{\beta }_{{\rm{fort\times year}}}}\:\times \:{\rm{fort}}\:\times \:{\rm{year}}\: + \:{\rm{log}}({\rm{LB}}),\eqno\rm$$
whose term
$$--><$>{{\beta }_{{\rm{fort}}\:\times \:{\rm{year}}}} $$$
gives a basis for a test of change in trend across fortification. Additionally, judicious use of the β
fort and β
year terms allowed estimation of trends for both pre- and post-fortification periods. SAS GENMOD version 9·2 software was used to produce all estimations and standard errors(
21
).
Results
Search results
The search identified a total of 2457 citations published from January 1990 to December 2010. Of these, 2295 were excluded because they duplicated data from original reports or were commentaries on previously published data. Of the remaining 162 citations, thirty-three were excluded because reports could not be located after extensive library and electronic searches as well as three or more unsuccessful attempts to contact the authors. Another sixty-five were excluded after reading the abstract and/or full text because the reports did not include specific NTD information, included only one type of NTD, or had a denominator of less than 5000 live births. This process identified a total of sixty-four reports. In addition, we included data from five institutional reports, two of which were published by the Universidad de San Carlos in Guatemala and one each by the Costa Rica Ministry of Health, the Dominican Republic Ministry of Health and the Proyecto Niños Saludables in Honduras. Of the sixty-nine reports that satisfied the inclusion criteria, fifty-one (73·9 %) were published in peer-reviewed journals. Table 1 summarizes the studies included in the present review by country, study design, time period covered, data sources, number of NTD cases by type and total, number of live births, and prevalence of NTD per 1000 live births by type and total.
Table 1 Prevalence of neural tube defects (NTD) in Latin America
An, anencephaly; SB, spina bifida; En, encephalocele; LB, live births; ICBDSR, International Clearinghouse for Birth Defects Surveillance and Research; ECLAMC, Latin American Collaborative Study of Congenital Malformations; UFMG, Universidade Federal de Minas Gerais; SINASC, Surveillance System on Live Births; RECUMAC, Cuban Congenital Malformations Registry; CGPM, Provincial Center of Medical Genetics; REDOMALCO, Dominican Republic Congenital Malformations Registry; RYVEMCE, Mexican External Malformations Epidemiological Surveillance Registry.
*January 2003 to June 2005.
†July 2005 to December 2007.
Reports were available from fifteen countries and one sub-region. Information was not available for Bolivia, El Salvador, Nicaragua and Panama, and although Paraguay reported NTD prevalence, its birth cohort was too small to satisfy our inclusion criteria.
Within each country, the data, which covered single or multiple locations and different time periods and hospitals, varied by methodology used and geographic areas covered. Data for most countries covered regional and/or local areas; however, data for Argentina, Costa Rica, Cuba and Puerto Rico were also collected at the national level. ECLAMC reported NTD prevalence for South America and several locations in the subcontinent. In a majority of reports, spina bifida cases were the largest contributor to total NTD cases.
Among the sixty-nine reports, forty-two (60·9 %) were based on hospital registries (structured case definition and inclusion criteria), fourteen (20·3 %) on review of medical records, seven (10·1 %) on hospital discharge data and the remaining six (8·7 %) were based on population-based registries, review of hospital delivery logs or reports from live birth surveillance systems. Costa Rica, Cuba and Puerto Rico also included reports from specialty clinics to capture post-discharge diagnoses. In addition, data from Cuba included pregnancy terminations.
Prevalence of neural tube defects
Reported NTD prevalence by country or location (Table 1) showed wide geographic variation within and between countries, ranging from 0·2 to 9·6 per 1000 live births.
National prevalence estimates
National NTD prevalence estimates were available for seven countries: Argentina, Costa Rica, Cuba, Ecuador, Guatemala, Mexico and Puerto Rico. These estimates varied by country and methodology. National registry data showed that NTD prevalence per 1000 live births was 0·45 in Costa Rica (2007)( 22 ), 1·10 in Cuba (2005–2006)( 23 ), 0·82 in Ecuador (2001–2005)( 24 ) and 0·90 in Puerto Rico (2008)( 25 ). National hospital discharge data showed an NTD prevalence of 1·62 in Argentina (2005)( Reference Calvo and Biglieri 15 ) and 0·47 in Mexico (2004)( Reference Mancebo-Hernández, González-Rivera and Díaz-Omaña 26 ). NTD prevalence based on national hospital delivery logs was 2·82 in Guatemala (2001–2003)( 27 ).
Regional and local prevalence estimates
Hospital-based registry data have been used to estimate NTD prevalence in specific locations in different countries. For example, Argentina hospital registry data from fifty-nine hospitals in seven regions showed an NTD prevalence of 1·99 per 1000 live births for the period 1994–2007( Reference Campaña, Pawluk and Lopez-Camelo 28 ). This prevalence is consistent with another hospital registry study in forty-one Argentinean hospitals, which showed that the prevalence of NTD for 1982–2007 was 2·01 per 1000( Reference Lopez-Camelo, Castilla and Orioli 29 ). Available hospital registry data from Chile in 1998–2000 showed a similar prevalence. However, NTD prevalence estimates based on hospital registry data varied within and between locations in Brazil, Colombia, Cuba, Mexico, Uruguay and Venezuela. For example, Brazilian hospital registry data from nineteen ECLAMC-participating hospitals for the periods 2003–2005 and 2005–2007 showed that NTD prevalence was 4·51 and 3·80 per 1000, respectively( Reference Lopez-Camelo, Castilla and Orioli 29 ). This NTD prevalence was almost half the 9·60/1000 prevalence reported from Porto Alegre for the time period 2000–2005( Reference Guardiola, Koltermann and Aguilar 30 ). In contrast, NTD prevalence in hospital registry data from Minais Gerais and Sao Paulo for comparable time periods ranged from 1·13 to 4·87 per 1000 live births( Reference Aguiar, Campos and Aguiar 16 , Reference Nascimiento 31 ).
Variations in NTD prevalence estimates were also observed between hospital delivery logs and hospital records data in several locations. For example, hospital delivery log data from two Guatemala City hospitals in 2004–2008 showed an NTD prevalence of 2·00 per 1000 live births( Reference Salguero-García, Barrios-Ruiz and Cardona de León 32 ), compared with 3·47 per 1000 in 2004–2005 identified in data derived from hospital records at the same hospitals( Reference Ortiz and Kestler 33 ).
Time trends in prevalence of neural tube defects
Information to assess NTD prevalence time trends was available for Bogota (Colombia), Costa Rica, Cuba, Dominican Republic, Guatemala City (Guatemala), Mexico (RYVEMCE) and Puerto Rico. Overall time trends of NTD prevalence exhibited average annual declines ranging from 2·5 % to 21·8 % (Table 2), with the exception of Cuba, which showed an increase; however, it was not possible to model the trend prevalence because the Cuban surveillance system changed its inclusion criteria for the period 2000–2004.
Table 2 Model trend prevalence rate ratios (PRR) for neural tube defects for Bogota (Colombia), Costa Rica, Cuba, Dominican Republic, Guatemala City (Guatemala), Mexico and Puerto Rico
RYVEMCE, Mexican External Malformations Epidemiological Surveillance Registry; N/A, not applicable.
*Time trend modelling is not possible because the Cuban surveillance system changed its inclusion criteria in middle of the period 2000–2004.
Fortification
All Latin American countries have mandatory folic acid fortification of wheat flour except Venezuela (Table 3). Fortification levels range from 0·35 mg/kg to 3·3 mg/kg. While most countries started fortification in the late 1990s, Argentina, Brazil, Peru and Uruguay introduced fortification regulation in 2002, 2002, 2004 and 2006, respectively( Reference David 34 – 37 ). In addition, Mexico also fortifies corn flour and Costa Rica also fortifies corn flour, rice and milk( Reference David 34 , 38 ).
Table 3 Fortification status in Latin American countries
NA, not available.
Argentina, Brazil, Chile, Costa Rica, Cuba, Puerto Rico and Mexico were the only countries for which information was available to perform a meaningful comparison of changes in pre- and post-fortification NTD prevalence. We directly compared NTD prevalence between these periods by computing NTD prevalence for years pooled pre- and post-fortification, summarizing effects as a prevalence ratio with 95 % confidence limits. Table 4 depicts the post-fortification changes in NTD prevalence and their corresponding PRR and 95 % CI for Argentina, Brazil, Chile, Costa Rica, Cuba, Mexico (RYVEMCE) and Puerto Rico. All sites showed significant declines in NTD prevalence ranging from 33·0 % to 59·0 %. In the case of Costa Rica we were able to fit a more complex model including an interaction term to assess whether the NTD secular trend changed after fortification. Our model yielded a statistically significant interaction term (P < 0·03) suggesting that NTD prevalence changed significantly following fortification. The pre-fortification prevalence was estimated at 1 % decline (95 % CI −4 %, +3 %) per year, and the post-fortification trend was estimated at 6 % decline (95 % CI −10·2 %, 2·1 %) per year.
Table 4 Comparison of pre- and post-fortification prevalence rate ratios (PRR) for neural tube defects for Argentina, Brazil, Chile, Costa Rica, Cuba, Mexico and Puerto Rico
RYVEMCE, Mexican External Malformations Epidemiological Surveillance Registry; N/A, not applicable.
*Pre- and post-fortification comparison is not possible because the Cuban surveillance system changed its inclusion criteria. Post-fortification period included pregnancy terminations and additional hospitals.
†January 2003 to June 2005.
‡July 2005 to December 2007.
Discussion
To our knowledge, the present publication is the first one that reviews data on NTD prevalence in Latin America, including time trends and the impact of folic acid fortification. The majority of countries showed a generalized decrease in the NTD prevalence in time, similar to time trend declines previously reported elsewhere( Reference Olney and Mulinare 39 – Reference Williams, Rasmussen and Flores 41 ).
Our review showed that fifteen countries reported local and/or regional NTD prevalence and seven of them reported, in addition, national prevalence of NTD. Comparisons of NTD prevalence between and within countries showed regional and/or local differences, most probably due to variations in data collection methods. In our review, the main data collection methods used to estimate NTD prevalence were hospital-based registries, clinical examination at birth and review of hospital records, hospital discharge data, hospital delivery logs and live birth statistics. Hospital-based registries have defined inclusion and exclusion criteria and clear diagnostic criteria including specific definitions for case ascertainment and information recording. Hospital clinical examinations at birth can be as effective as hospital registries in recording numbers and types of NTD, if there are in place specific protocols for diagnostic criteria, case inclusion and exclusion, and case ascertainment. However, hospital clinical examinations without specific protocols are more susceptible to biases than registries or structured surveillance systems because these examinations are not standardized and clinicians have differences in case definition, differences in how newborns are examined and differences in how results are recorded. Similarly, hospital discharge data are more susceptible to biases due to differences in criteria within and across hospitals and physicians related to case ascertainment, case recording and ICD (International Classification of Diseases) code assignment in the discharge diagnosis fields. Also, hospital discharge data are susceptible to including multiple records of the same individual and hospital transfers or readmissions, and might contain records of patients who do not belong to the hospital catchment area. In addition to being susceptible to differences in hospitals and medical practices, hospital delivery logs are limited because they do not include cause-specific morbidity or mortality. In addition, NTD are serious defects with a very high mortality and it is essential to count stillbirths when estimating the prevalence of NTD to avoid an underestimation of prevalence.
In summary, hospital-based data, although readily available in many countries, also reflect variations in access to and utilization of health services. The impact of different data collection methods and sources of information on NTD prevalence estimates has been reported previously( Reference Lix, Yogendran and Shaw 42 – Reference Hexter and Harris 46 ).
Differences in NTD prevalence by country can also be explained by geographic variation. For example, higher NTD prevalence in some areas of Brazil, Guatemala, Honduras and Mexico is consistent with higher levels of poverty, higher conception rates for younger mothers, and less access to health services and fortified staples( 47 , Reference Imhoff-Kunsch, Flores and Dary 48 ). In contrast, the observed increase in Cuba's NTD prevalence trend was most likely due to differential ascertainment: changes in inclusion criteria, increase in number of participating hospitals and inclusion of pregnancy terminations( Reference Piloto Morejón, Choconta Sanabria and Menendez Garcia 49 ).
Nevertheless, despite data limitations and geographic variation, these data are important because they can show changes over time.
Fortification
Comparisons between pre- and post-fortification prevalence of NTD showed that fortification efforts were effective in reducing NTD prevalence in Argentina, Bogota (Colombia), Chile, Costa Rica, Guatemala City (Guatemala), Mexico (RYVEMCE) and Puerto Rico. This confirms the reduction in NTD prevalence reported elsewhere after fortification with folic acid( Reference Williams, Rasmussen and Flores 41 , Reference De Wals, Tairou and Van Allen 50 – Reference Ray, Meier and Vermeulen 55 ) and the previous declines in NTD prevalence reported in the region( Reference Lopez-Camelo, Castilla and Orioli 29 , Reference Tacsan and Ascencio Rivera 52 , Reference Hertrampf and Cortés 53 ).
A declining secular trend that started before the implementation of fortification programmes may obscure assessment of the NTD prevention effect of these programmes. Previous publications have reported techniques that include methodological approaches to assess the potential effects attributable to fortification when a previous declining trend has been identified( Reference Chen, Carmichael and Selvin 56 ). Using a methodology similar to that reported by Chen et al.( Reference Chen, Carmichael and Selvin 56 ) we attempted to determine the pre- and post-fortification slopes that represented summaries of the annual NTD prevalence before and after implementation of fortification. However, the only data set in which we could evaluate such changes was from Costa Rica because we had enough data to assess the pre- and post-fortification trend. For the pre-fortification period the slope of NTD prevalence was not different from zero; however, the post-fortification period showed a significant decline in NTD prevalence. This result re-confirms that the decline in NTD prevalence can be accelerated when countries select staples that are highly consumed by the population and monitor and evaluate the levels of folic acid in fortified staples and the impact of their fortification programmes. The present review re-asserts that fortification of staples with folic acid results in up to a 59 % decrease in NTD cases that could result in reductions in mortality, morbidity and financial burden associated with these conditions( Reference Llanos, Hertrampf and Cortes 57 , Reference Bentley, Weinstein and Willett 58 ).
Limitations
There are several limitations that could have a bearing on our findings regarding NTD prevalence in Latin America. The overall quality of the review and its results is dependent on the quality of information of the individual studies. The heterogeneity of case ascertainment and years of study across countries and across surveillance programmes affected our ability to pool estimates, make direct comparisons or quantitatively evaluate trends across time or countries. The use of voluntary hospital-based surveillance systems that capture only a proportion of the population at risk is also a potential limitation of the study. The under-representation of rural populations in the reported data from some countries can affect estimates.
Conclusion
The present publication is the first to review and report data on NTD prevalence in Latin America including time trends and the impact of folic acid fortification. The surveillance of NTD in Latin America is currently limited because few countries have established systems to report national and local NTD prevalence. However, when data are available, reported NTD prevalence, which varies by geographic region from 0·2 to 9·6 per 1000 live births, is in some areas of Latin America among the lowest in the world while in others is among the highest. Observed declines in NTD prevalence were largest in countries where folic acid fortification of staples reached the majority of the population at risk, such as Chile and Costa Rica. NTD prevalence among countries in which fortification had been implemented showed declines ranging from 33·0 % to 59·0 %. It was possible to show that fortification has an impact and was consistent for most countries. Selected registries in the region have become proxies for national surveillance systems, and even though they have limited coverage, they constitute the major source of information regarding NTD prevalence and time trends that allow for the monitoring of disease burden and impact of fortification programmes. The need for adequate data is central to a better understanding of the magnitude of the public health impact of NTD in the Latin American region and the assessment of the effectiveness of prevention programmes. The implementation of national NTD surveillance programmes could help to close this information gap.
Acknowledgements
Sources of funding: This research received no grant from any funding agency in the public, commercial or not-for-profit sectors. The Hispanic Association of Colleges and Universities provided funding to support J.C. Conflicts of interest: The authors declare to have no conflicts of interest in relation to the present manuscript. Ethics: Ethical approval was not required for this study. Authors’ contributions: J.R. and J.C. performed the literature search, data analysis and drafted the manuscript. C.J.A. contributed to the data analysis. The draft manuscript was critically revised by D.T., C.J.A., A.F. and J.F. All authors approved the final version of the manuscript.
