Birth weight patterns by gestational age in Brazil

Birth weight patterns by gestational age in Brazil

Carlos E. Pedreira^{I, II}; Francisco A. Pinto^III; Silvia P. Pereira^IV; Elaine S. Costa^IV

^ICOPPE-PEE/UFRJ, Av. Horácio Macedo, 2030, Prédio do CT, Bloco H, 3° andar, Ilha do Fundão, Cidade Universitária, 21941-914 Rio de Janeiro, RJ, Brasil

^IIFaculdade de Medicina/UFRJ, Av. Carlos Chagas Filho, Prédio do CCS, Bloco K, 2° andar, Ilha do Fundão, Cidade Universitária, 21941-914 Rio de Janeiro, RJ, Brasil

^IIIDepartamento de Estatística/UFF, Rua Mário Santos Braga s/n, 24020-140 Niterói, RJ, Brasil

^IVNúcleo Transdisciplinar de Investigação em Saúde da Criança, Instituto de Puericultura e Pediatria Martagão Gesteira/UFRJ, Rua Bruno Lobo, 50, Ilha do Fundão, Cidade Universitária, 21941-912 Rio de Janeiro, RJ, Brasil

^{Correspondence to}

ABSTRACT

BACKGROUND AND OBJECTIVES: We present an updated birth weight-for-gestational-age portrait, based on nearly 8 million observations of an ethnic-mixed population. It comprises the first comprehensive charts with Brazilian data. This contribution intends to assist clinicians in classifying fetal growth, to provide a reference for investigations of predictors and to show the consequences of small and large patterns for gestational age delivery. Most of the reference data for assessing birth weight for gestational age deal with insufficient sample size, especially at low gestational age. Population-based studies with considerably large sample size refer to data collected more than 15 years ago.

METHODS: We accomplished a population-based study on births in all the Brazilian states from 2003 to 2005. Results were based on 7,993,166 singletons. We constructed the 3rd, 5th, 10th, 25th, 50th, 90th, 95th and 97th smoothed percentiles curves and gender-specific tables from 22 to 43 completed weeks.

RESULTS: The resulting tables and graphical representation provide a gender-specific reference to access the birth weights distribution according to the gestational age in the Brazilian population.

CONCLUSIONS: This is the first population-based reference constructed on a developing country data. These charts could provide an important tool to improve clinical assessment of growth in newborns.

Key words: birth weight, newborn, gestational age, fetal growth, preterm, postterm.

RESUMO

BACKGROUND E OBJETIVOS: Apresentamos um retrato atualizado de peso-por-idade-gestacional, baseado em quase 8 milhões de observações em uma população etnicamente misturada. Estas constituem as primeiras tabelas com dados brasileiros. Esta contribuição pretende dar assistência aos clínicos na classificação do crescimento fetal, e prover uma referência para pesquisas de prognósticos e consequências em partos com padrões pequenos e grandes para a idade gestacional. A maior parte dos dados de referência para estimar peso-por-idade-gestacional sofre de tamanho de amostra insuficiente, especialmente em baixa idade gestacional. Os estudos populacionais com uma amostra de tamanho considerável se referem a dados coletados há mais de 15 anos.

MÉTODOS: Nós realizamos um estudo populacional baseado em nascimentos em todos os estados brasileiros de 2003 a 2005. Os resultados foram baseados em 7.993.166 nascimentos de gravidez única. Nós construímos curvas suavizadas e tabelas gênero-específicas de 22 a 43 semanas completas para os percentiles 3°, 5°, 10°, 25°, 50°, 90°, 95° e 97°.

RESULTADOS: As tabelas e representações gráficas resultantes proveem uma referência gênero-específica para acessar a distribuição de peso ao nascimento de acordo com a idade gestacional na população brasileira.

CONCLUSÕES: Esta é a primeira referência populacional construída com dados de um país em desenvolvimento. Estas tabelas podem prover uma importante ferramenta para melhorara avaliação clínica do crescimento em recém-natos.

Palavras-chave: peso ao nascimento, recém-nascido, idade gestacional, crescimento fetal, pré-termo, pós-termo.

INTRODUCTION

After Lubchenco's article (Lubchenco et al. 1963) in the sixties, a number of reference data for assessing birth weight for gestational age have been proposed in the literature (Kramer et al. 2001, Zhang and Bowes 1995, Alshimmiri et al. 2004, Bonellie et al. 2008, Alexander et al. 1996, Shin et al. 2005, Skjæ rven et al.2000). Most of them refer to developed countries and none of the underdeveloped or developing country studies are population-based. Despite the obvious importance of these contributions, some deal with methodological troubles (Kramer et al. 2001), e.g. non-population-based studies, unisex references and small sample sizes, which is especially critical for low gestational age. In fact, some of these problems derive from intrinsic challenges in constructing birth weight for gestational age charts, e.g. the requirement of population-based studies and the need of reasonable sized samples even for unfrequent events like extreme preterm. Moreover, it should be taken into account that some hardships directly follow from the birth registration documents, which can not be controlled.

In this paper we conducted a study resulting in a gender-specific reference of birth weight for gestational age. It was based on a dataset including all newborns in all the Brazilian states between 2003 and 2005. These are the first comprehensive charts with Brazilian data. Currently, most clinical evaluations in Brazil are based on international data. Graphical representations as well as tables for the 3rd, 5th, 10th, 25th, 50th, 90th, 95th and 97th percentiles are displayed. Brazil has a large population and a quite high birth rate, which allows the results to be based on a large sample size. This is particularly relevant for low gestational ages and on the extreme (3rd, 5th, 95th and 97th) percentiles calculation. These percentiles are frequently used as cutoffs to define if newborns are small or large for gestational age. Although formed by a considerably large number of observations, our dataset span for just 3 years, avoiding possible undesirable effects due to birth weight patterns changes in the considered period (Bonellie and Raab 1997, Chike-Obi et al. 1996).

METHODS

For this study, we used data from births in all the Brazilian states from 2003 to 2005 provided by live birth certificates supplied by the Brazilian General Health System (SUS). The issue of live birth certificates is mandatory in Brazil, and unregistered births are almost inexistent and may accordingly be disregarded. Our results were based on 7,993,166 singleton births (4,093,316 male and 3,899,832 female newborns) after exclusions. Neonates from multiple gestations (n = 169,373) and with major congenital anomalies (n = 53,891) were excluded from the dataset. Registrations with unrecorded major congenital anomalies (817,867), gestational age (79,137), birth weight (52,967) and multiple gestations (15,467) were also eliminated. We also eliminated a few (less than 0.1%) obviously erroneous measurements. Apart from these variables, the dataset also provided information on ethnicity and parity. We decided not to use this information since the physiologic-pathologic nature of ethnic differences in fetal growth is still unclear (Kierans et al. 2008). In the Brazilian birth registration documents, gestational weeks are presented in ranges: less than 22, 22 to 27, 28 to 31, 32 to 36, 37 to 41, and more than 41 weeks.

Gestational age refers to the interval, in completed weeks, between the first day of the mother's last menstrual period (LMP) and the day of delivery. It can alsobe any estimate of this interval based on ultrasound, a physical examination or other method. Brazilian birth registration manual recommend the use of LMP. Other methods, such as ultrasound estimation and obstetric measures, may have been also used in some cases.

The Brazilian Information System on Live Births (SINASC), implemented in 1990, covers 90% of all births in Brazil. This birth registration system includes all babies born alive, independently of the gestation age. Babies with very low gestational age are not considered to be stillborn. It is worth mentioning that babies with extremely low gestational ages have a survival ratearound 50%, reflecting a considerably large net of Neonatal Intensive Care Units to assist extremely premature newborns in Brazil.

We constructed separate curves and tables formale and female newborns for the 3rd, 5th, 10th, 25th, 50th, 90th, 95th and 97th percentiles from 22 to 42 completed weeks based on smoothed estimated curves. The curves and tables were smoothed by a shape-preserving piecewise cubic interpolation method (Fritsch and Carlson 1980, Kahaner et al. 1989) using MATLAB software program (Mathworks, Natick, MA).

The relative percentual differences between previous published charts and the present paper are computed for the 10th, 50th and 90th percentiles using the Brazilian data as reference. Relative percentual differences were calculated as:

Here, Brasil_perc represents the Brazilian percentiles,while Other_perc are the percentiles published in (Kramer et al. 2001, Zhang and Bowes 1995, Alshimmiri et al. 2004) and (Bonellie et al. 2008).

The institutional ethical research board considered that this study is exempt of approval since the data is publicly available in the Brazilian government site.

RESULTS

In Table I one can find the 3rd, 5th, 10th, 25th, 50th, 90th, 95th and 97th percentiles of birth weights for gestational age for male and female newborns, respectively. Figure 1 shows the graphical representation of these percentiles.

In Table II we present a comparative view concerning some design aspects, of the present paper and other different studies. It is worth noting that most of the previous charts was based on developed countries data and that the Brazilian charts were built up with nearly 8 million newborns, more than double the sample size of the largest early studies.

In Table III one may find a comparative tabulation of the relative percentual differences for the 10th, 50th and 90th percentiles, for males, between four previous published charts (Kramer et al. 2001, Alshimmiri et al. 2004, Bonellie et al. 2008, Alexander et al. 1996) and the present paper, used as reference. Negative andpositive quantities in Table III reflect that the Brazilian percentiles were, respectively, smaller or larger, in comparison with the other four references. Negative numbers indicate that the percentiles are overestimated in comparison to the Brazilian ones, while positive values means underestimation in relation to present study percentiles. Note that all values are percentual. For instance, the male percentile 90 from Table III for 40 weeks would be overestimated by 4.1% if (Bonellie et al. 2008) reference was used.

Following the way the references are published by different authors, for Bonellie we used the average between nulliparous and multiparous for comparison purpose. Although (Alexander et al. 1996) provide the 10th percentile values for both males and females, the data is not gender-specific for most percentiles, and so the same values were used in Table III.

DISCUSSION

Birth weight for gestational age charts is an essential tool providing relevant information to clinicians regarding which newborns may be at higher risk of neonatal morbidity and subsequent mortality or developmental delay. In this paper we present the first comprehensive charts with Brazilian data.

Table II allows a comparative view of the Brazilian study presented in this paper and some of the published charts. It can be noticed that the only published population-based studies with considerably large sample size refer to data collected more than 15 years ago.

From the ethnic standards point of view, the Brazilian population is a truly melting pot. Accordingly, the present study constitutes a portrait of the birth weightfor gestational age of a mixed population based on nearly 8 million observations.

Table III should be viewed with caution since the studies over ethnic and socioeconomic diversities embrace substantial methodological differences among each other. Nevertheless, it is interesting to note that the Brazilian chart produced lower percentiles (negativevalues in Table III) for the majority of the gestational weeks if compared to (Alexander et al. 1996). Furthermore, the 50th percentile is generally higher for the Brazilian data at very low gestational ages (up to 26th weeks) and lower at term. This may be an indication of poorer survival rates at lower gestations in Brazil. Overall, despite the methodological discrepancies and the expected consequent diversion in percentile values, the Brazilian chart is quite consistent with previous studies. The greatest divergences appeared in relation to (Alshimmiri et al. 2004) charts, which are maybe explained by its quite small sample size. It should be noticed that the differences among these studies maybe in part attributed to differences in the statistical procedures applied in different papers.

Several approaches are concerned with misestimation of the gestational age for a proportion of newborns (Kramer et al. 2001, Bonellie et al. 2008, Plattet al.2001, Oja et al. 1991, Hutcheon and Platt 2008). This misclassification may produce curves that arenot smooth or biologically plausible. Undesirable effects like bumps in extreme percentiles, specially around weeks 28 to 30, have been reported (Kramer et al. 2001, Bonellie et al. 2008). We did not observe these distortions in our curves maybe because of the huge size of the sample.

We hope the presented charts will be useful to clinicians in classifying fetal growth. They may also contribute as a reference for investigations of predictors and to show the consequences of small and large patterns for gestational age delivery. At last, some limitations should be acknowledged. Our study is cross-sectional, as all population-based gestational age references. Possible bias due to missing data caused by the absence of the weights of the fetuses still in utero (Hutcheon and Platt 2008) is also common to all charts, including ours. Finally, the accurate determination of the gestational age in population-based studies is an open challenge for all charts and the practical adequacy of different measures is a stirring investigation problem (Wingate et al. 2007).

ACKNOWLEDGMENTS

This work has been partially supported by grants from Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq), and Fundação Carlos Chagas Filho de Amparo à Pesquisa do Estado do Rio de Janeiro (FAPERJ).

Manuscript received on July 27, 2009

accepted for publication on November 19, 2010

• ALEXANDER GR, HIMES JH, KAUFMAN RB, MOR J AND KOGAN M. 1996. A United States national reference for fetal growth. Obstetrics & Gynecology 87: 163-168.
• ALSHIMMIRI MM, AL-SALEH EA, ALSAEID K, HAMMOUD MS AND AL-HARMI JA. 2004. Birth weight percentiles by gestational age in Kuwait. Arch Gynecol Obstet 269: 111-116.
• BONELLIE S, CHALMERS J, GRAY R, GREER I, JARVIS S AND WILLIAMS C. 2008. Centile charts for birth weight for gestational age for Scottish singleton births. BMC Pregnancy Childbirth 25: 8-15.
• BONELLIE SR AND RAAB GM. 1997. Why are babies getting heavier? Comparison of Scottish births from 1980 to 1992. BMJ 315: 1205.
• CHIKE-OBI U, DAVID RJ, COUTINHO R AND WU SY. 1996. Birth weight Has Increased Over a Generation. Am J Epidemiol 164: 563-569.
• FRITSCH FN AND CARLSON RE. 1980. Monotone Piecewise Cubic Interpolation, SIAM. J Num Anal 17: 238-246.
• HUTCHEON JA AND PLATT RW. 2008. The missing data problem in birth weight percentiles and thresholds for small-for-gestational-age. Am J Epidemiol 167: 793-796.
• KAHANER D, MOLER C AND NASH S. 1989. Numerical Methods and Software. London: Prentice Hall, 495 p.
• KIERANS WJ, JOSEPH KS, LUO ZC, PLATT R, WILKINS R AND KRAMER MS. 2008. Does one size fit all? The case for ethnic-specific standards of fetal growth. BMC Pregnancy Childbirth 8: 1.
• Kramer M, Platt RW, Wen SW, Joseph KS, Allen A, Abrahamowicz M, Blondel B, Bréart G and Fetal/Infant Health Study Group of the Canadian Perinatal Surveillance System. 2001. A new and improved population-based reference for birthweight for gestational age. Pediatrics 108: E35.
• LUBCHENCO L, HANSMAN C, DRESSLER M AND BOYD E. 1963. Intrauterine growth as estimated from liveborn birth weight data at 24 to 42 weeks of gestation. Pediatrics 32: 793-800.
• OJA H, KORAINEN M AND RANTAKALLIO P. 1991. Fitting mixture models to birth weight data: a case study. Biometrics 47: 883-897.
• PLATT RW, ABRAHAMOWICZ M, KRAMER MS, JOSEPH KS, MERY L, BLONDEL B, BRÉART G AND WEN SW. 2001. Detecting and eliminating erroneous gestational ages: a normal mixture model. Stat Med 20: 3491-3503.
• SHIN SM, CHANG YP, LEE ES, LEE YA, SON DW,KIM MH AND CHOI YR. 2005. Birth Weight, Very Low Birthweight Rates and Gestational Age-Specific Birth weight Distribution of Korean Newborn Infants. J Korean Med Sci 20: 182-187.
• SKJÆ RVEN R, GJESSING HK AND BAKKETEIG LS. 2000. Birth weight by gestational age in Norway. Acta Obst et Gynec Scand 796: 440-449.
• WINGATE MS, ALEXANDER GR AND BUEKENS P. 2007. Comparison of Gestational Age Classifications: Date of Last Menstrual Period vs. Clinical Estimate. Ann Epidemiol 17: 425-430.
• ZHANG J AND BOWES W JR. 1995. Birth-weight-for-gestational-age patterns by race, sex, and parity in the United States population. Obstet Gynecol 86: 200-208.

Publication Dates

• Publication in this collection
  27 May 2011
• Date of issue
  June 2011

History

• Received
  27 July 2009
• Accepted
  19 Nov 2010