SciELO - Scientific Electronic Library Online

vol.21 issue1Chromosome 6q deletion: Report of a new case and review of the literatureBirth weight of twins: 2. Fetal genetic effect on birth weight author indexsubject indexarticles search
Home Pagealphabetic serial listing  

Services on Demand




Related links


Genetics and Molecular Biology

Print version ISSN 1415-4757On-line version ISSN 1678-4685

Genet. Mol. Biol. vol. 21 n. 1 São Paulo Mar. 1998 

Birth weight of twins:
1. The fetal growth patterns of twins and singletons


Bernardo Beiguelman1,2, Gloria M.D.D. Colletto 2, Carla Franchi-Pinto 3 and Henrique Krieger 2
1 Curso de Pós-Graduação em Genética, UNICAMP, Campinas, SP, Brasil.
2 Laboratório de Epidemiologia Genética, Departamento de Parasitologia, Instituto de Ciências Biomédicas, USP. Send correspondence to B.B. Rua Angelina Maffei Vita, 408, Apto. 41, 01455-070 São Paulo, SP, Brasil.
3 Serviço de Genética Médica da Santa Casa de Misericórdia de São Paulo, SP, Brasil.




Distributions of birth weights of twins and singletons born at three southeastern Brazilian hospitals were compared after adjustment for gestational age, its quadratic and cubic terms, sex, type of pregnancy, as well as their interactions. The pattern of twin fetal growth rate was retarded in comparison to that of singletons, regardless of the socioeconomic level of the examined population, but the gestational age at which this retardation started seems to be correlated to the income of the mothers. In all cases, after 28 weeks of gestation, female fetal growth was slightly but consistently lower than that of males.




On average, twins are born with a lower gestational age and lighter weight than singletons (McKeown and Record, 1952; Powers, 1973; Gedda et al., 1981; Leroy et al., 1982; Watson and Campbell, 1986; Alfieri et al., 1987; Bonnelykke et al., 1987; Buckler and Buckler, 1987; Bleker et al., 1988; Luke et al., 1991; Keith, 1994; Luke, 1996, inter allia). When twins and singletons were distributed into two classes of birth weight (less and more than 2500 g), or grouped into three classes of gestational age (preterm, term, and post-term), the proportions of newborns with low birth weight (less than 2500 g) or preterm deliveries (less than 37 weeks) were dramatically higher among twin births than among singletons.

Perinatal mortality is strongly influenced by low birth weight, due to its association with respiratory, metabolic, immunologic and neurologic disorders (Erhardt et al., 1964; Harper and Wiener, 1965; Fitzhardinge and Steven, 1972; Chandra, 1975; Laski et al., 1975, inter allia). Since the rate of perinatal deaths is disproportionately high among twins (Keith, 1994; Luke, 1996) one might suppose that this could merely be a consequence of their low birth weight. However, recent data of Luke (1996) confirmed Gedda et al.‘s (1981) claim that low birth weight in twins and in singletons are not comparable and have different implications for child growth and survival. Gedda et al. (1981) have also stressed that "when considering the relation between birth weight and perinatal mortality, the data should always be broken down according to whether they refer to twins or to singletons, rather than combined as it is usually done."

In fact, Luke (1996) demonstrated that the low birth weight of twins may be beneficial to them, since in the United States of America the lowest fetal death rate for twins was observed at an earlier gestational age (36-37 weeks) with a lower birth weight (2500-2800 g) compared to singletons, who showed the lowest fetal death rate at 40-41-weeks gestational age, with a birth weight of 3700-4000 g.

These recent analyses of gestational agespecific birth weights of twins compared to singletons have added a great deal to our understanding of this problem. Nevertheless, much work is still necessary to further understand this matter, particularly in the third world countries. The present study compares the birth weight distributions of twins and singletons born to southeastern Brazilian mothers of different socioeconomic backgrounds after adjustment of these weights by multiple regression analysis to some biological variables.



The birth weights of 1,158 twin pairs and 12,609 singletons, whose registered records did not fail to indicate gestational age in weeks, were investigated. Records were obtained from three maternity hospitals in the State of São Paulo, Brazil: Maternidade de Campinas (MC) in the city of Campinas, SP (631 twin pairs and 1,112 singletons); Hospital Santa Catarina (HSC) in the city of São Paulo, SP (339 twin pairs and 300 singletons) and Hospital e Maternidade Escola Vila Nova Cachoeirinha (VNC) in the city of São Paulo, SP (188 twin pairs and 11,197 singletons). MC is the largest obstetric center in Campinas. All social segments of the population are found among the women assisted there (40% private patients, 59% welfare program and 1% indigents). In contrast, HSC is mostly dedicated to private patients, while VNC is a public hospital where neither hospital nor medical care is charged, since it provides assistance to the poorest layers of the population. VNC is also a reference center that gives obstetrical care for high-risk poor pregnant women. Stillbirths were not omitted, but newborns weighing 500 g or less were excluded, as they are classified as abortions (Belitzki et al., 1978).

Since birth weight has a large variance, differences among the three hospitals were assessed using Kruskal-Wallis test. The values of birth weight in grams (dependent variable) were transformed into natural logarithms, and adjusted by multiple stepwise regression analysis for gestational age in weeks, its quadratic and cubic terms, sex, and type of pregnancy (twin or singleton), as well as their interactions, using the SPSS® package.



The results of the Kruskal-Wallis test presented in Table I clearly show that neither the twins nor the singletons born at the three hospitals can be pooled together, since they represent different birth weight populations. On average, both twins and singletons were the heaviest at HSC, followed by MC, and VNC. Therefore, a multiple regression analysis was performed separately for each hospital. The natural logarithm of birth weight was considered as the dependent variable, while the independent variables included gestational age, (gestational age)2, (gestational age)3, sex, gestational age ´ sex, (gestational age)2 ´ sex, (gestational age)3 ´ sex, type of pregnancy, gestational age ´ type of pregnancy, (gestational age)2 ´ type of pregnancy, (gestational age)3 ´ type of pregnancy, sex ´ type of pregnancy, gestational age ´ sex ´ type of pregnancy, (gestational age)2 ´ sex ´ type of pregnancy, and (gestational age)3 ´ sex ´ type of pregnancy.


Table I - Mean birth weights and standard deviations in grams of twins and singletons born at three Brazilian hospitals, and Kruskal-Wallis test results.





Mean ± SD

Mean rank


Mean ± SD

Mean rank



2444 ± 540



3222 ± 479




2295 ± 570



3115 ± 413




2245 ± 631



3041 ± 578


K-W test
(2 d.f.)

H = 104.6; P < 0.001

H = 72.52; P < 0.001

HSC, Hospital Santa Catarina; MC, Maternidade de Campinas; VNC, Hospital e Maternidade Escola Vila Nova Cachoeirinha.


Table II shows the influence of the selected independent variables on birth weight. For newborns at MC these variables were gestational age, (gestational age)3, gestational age ´ type of pregnancy, and sex. For newborns at HSC they were gestational age, (gestational age)2, sex, and (gestational age)3 ´ type of pregnancy. Concerning newborns at VNC, the selected independent variables were gestational age, (gestational age)3, gestational age ´ sex, gestational age ´ type of pregnancy, (gestational age)3 ´ type of pregnancy. Figure 1 presents the adjusted curves derived from the analyses shown in Table II.


ms1973f1.gif (23501 bytes)

Figure 1 - Distribution of the birth weight of twins and singletons born at three Brazilian hospitals (HSC, MC, VNC) after adjustment according to the regression model shown in Table II. Twins are symbolized by squares, and singletons by circles. Males are represented by closed symbols, and open symbols designate females.



Table II - Significant regression coefficients obtained by multiple stepwise regression analysis of the logarithm of birth weight in grams(Y) on gestational age, its quadratic and cubic terms, sex, and type of pregnancy, as well as their interactions.

Y ± SD

7.85 ± 0.30
b ± SE

7.86 ± 0.28
b ± SE

7.99 ± 0.24
b ± SE


1.598 ± 0.163

-2.034 ± 0.513

2.204 ± 0.093

gestational age

0.2309 ± 0.0068

0.4815 ± 0.0296

0.2119 ± 0.0038

(gestational age)2


-0.0057 ± 0.0004


(gestational age)3

-4.3 x 10-5 ± 1.7 x 10-6


-4.1 x 10-5 ± 9.8 x 10-7


0.0247 ± 0.0065

0.0259 ± 0.0098


gestational age x sex



0.0010 ± 8.2 ± 10-5

gestational age x type pregnancy

-0.0052 ± 0.0002


-0.0150 ± 0.0016

(gestational age)3 x type pregnancy


-3.2 x 10-6 ± 2.1 x 10-7

6.3 x 10-6 ± 1.2 x 10-6

SQ regression








SQ residual
















For abbreviations see Table I.


The adjusted curves illustrate that, after about 28 weeks of gestation, female fetal growth was slightly but consistently lower than that of males. They also demonstrate clearly that the pattern of twin fetal growth rate was delayed in comparison to that of singletons, independent of the socioeconomic level of the assisted inpatients. Moreover, these curves disclose that the lighter mean birth weight of twins cannot be attributed solely to their shorter mean gestational age, since this  variable  was included for the adjustment of all curves in Figure 1.

The adjusted curves derived from the data provided by HSC, that assists high income people, indicate that the fetal growth rates of twins and singletons were similar up to about 28 weeks of gestation. From this age on a slower intrauterine growth of twins compared to singletons began and the difference was progressively emphasized. The period of undifferentiated intrauterine growth seen in the HSC curves is similar to those reported by McKeown and Record (1952), Naeye et al. (1966) and Williams et al. (1982) for first world populations. However, Luke et al. (1991), based on unadjusted data of birth weights obtained at the John Hopkins Hospital, concluded that the difference between fetal growth of twins and singletons starts later, at 36 weeks of gestation.

The adjusted curves yielded by the data obtained from MC, where the assisted women have, on average, a lower income compared to those from HSC, are similar to HSC’s curves. Nevertheless, in MC, a slower rate of twin fetal growth is noted soon after 22 weeks of gestation in the curves representing MC. This differentiation is clearer in the adjusted curves yielded by the data obtained from the public hospital VNC, that provides care only to poor women and is a reference hospital for high-risk pregnancies. This pattern may be an indication that twin fetal growth is delayed earlier when they are generated by undernourished pregnant women. However, taking into account the fact that infants born to mothers with chronic hypertension are smaller and lighter than those born to normal mothers, and considering that the most common cause for high-risk transfers to VNC is chronic hypertension, one cannot exclude the possibility that the precocious delay of intrauterine growth of the twins born in this hospital might be exaggerated by an excess of these high-risk mothers.



This work was supported by CNPq. Publication supported by FAPESP.




As distribuições dos pesos de recém-nascidos de partos gemelares e únicos em três maternidades do sudeste brasileiro foram comparadas, depois do ajustamento desses pesos à idade gestacional, seus termos quadrático e cúbico, sexo e tipo de gestação e às interações dessas variáveis. O padrão da taxa de crescimento fetal dos gêmeos em comparação ao dos recém-nascidos de parto único é retardado, independentemente do nível sócio-econômico da população examinada, mas a idade gestacional em que começa esse atraso parece estar correlacionada ao nível econômico das mães. Em todos os casos, depois de 28 semanas de gestação, o crescimento dos fetos femininos mostrou-se levemente, mas consistentemente, inferior ao dos fetos masculinos.




Alfieri, A., Gatti, I. and Alfieri, A.C. (1987). Weight and height growth in twins and children born in the last decade. Acta Genet. Med. Gemellol. 36: 209-211.         [ Links ]

Belitzki, R., Fescina, R.E. and Ucieda, F. (1978). Definiciones y terminologias aplicables al periodo perrinatal. Recomendaciones de la Organización Mundial de la Salud y modificaciones de la F.I.G.O. Publicación Científica del CLAP, No. 757: 136-147.         [ Links ]

Bleker, O.P., Oosting, J. and Hemrika, D.J. (1988). On the cause of the retardation of fetal growth in multiple gestations. Acta Genet. Med. Gemellol. 37: 41-46.         [ Links ]

Bonnelykke, B., Sogaard, J. and Nielsen, J. (1987). Seasonality in twin birth rates, Denmark, 1936-1984. J. Epidemiol. Comm. Health 41: 338-343.        [ Links ]

Buckler,  J.M.H.  and  Buckler,  J.B. (1987).   Growth characteristics in twins  and higher order  multiple births. Acta Genet. Med. Gemellol. 36: 197-208.

Chandra, R.K. (1975). Fetal malnutrition and postnatal immunocompetence. Am. J. Dis. Child. 129: 450-454.         [ Links ]

Erhardt, C.L., Joshi, G.B., Nelson, F.G., Kroll, B.H. and Weiner, L. (1964). Influence of weight and gestational age on perinatal and neonatal mortality by ethnic group. Am. J. Public. Health. 54: 1841-1855.         [ Links ]

Fitzhardinge, P.M. and Steven, E.M. (1972). The small-for-date infant. II - Neurological and intellectual sequelae. Pediatrics 50: 50-57.         [ Links ]

Gedda, L., Brenci, G. and Gatti, I. (1981). Low birth weight in twins versus singletons: separate entities and different implications for child growth and survival. Acta Genet. Med. Gemellol. 30: 1-8.         [ Links ]

Harper, P.A. and Wiener, G. (1965). Sequelae of low birth weight. Ann. Rev. Med. 16: 405-420.         [ Links ]

Keith, L. (1994). Mortality and morbidity among twins: recent observations from the United States. Acta Genet. Med. Gemellol. 43: 25-31.         [ Links ]

Laski, R.E., Lechtig, A., Delgado, H., Klein, R.E., Engle, P., Yarbrough, C. and Martorell, R. (1975). Birth weight and psychomotor perfomance in rural Guatemala. Am. J. Dis. Child. 129: 566-570.         [ Links ]

Leroy, B., Lefort, F., Neveu, P., Risse, R.J., Trévise, P. and Jeny, R. (1982). Intrauterine growth charts for twin fetuses. Acta Genet. Med. Gemellol. 31: 199-206.         [ Links ]

Luke, B. (1996). Reducing fetal deaths in multiple births: optimal birth weight and gestational ages for infants of twin and triple births. Acta Genet. Med. Gemellol. 45: 333-348.         [ Links ]

Luke, B., Witter, F.R., Abbey, H., Feng, T., Namnoum, A.B., Paige, D.M. and Johnson, T.R.B. (1991). Gestational age-specific birthweights of twins versus singletons. Acta Genet. Med. Gemellol. 40: 69-76.         [ Links ]

McKeown, T. and Record, R.G. (1952). Observations on fetal growth in multiple pregnancy in man. J. Endocrinol. 5: 387-401.         [ Links ]

Naeye, R.L., Benirschke, K., Hagstrom, J.W.C. and Marcus, C.C. (1966). Intrauterine growth of twins as estimated from live-born birth-weight data. Pediatrics 37: 409-416.         [ Links ]

Powers, W.F. (1973). Twin pregnancy, complications and treatment. Obstet. Gynecol. 42: 795-808.         [ Links ]

Watson, P. and Campbell, D.M. (1986). Preterm deliveries in twin pregnancies in Oxford. Acta Genet. Med. Gemellol. 35: 193-199.         [ Links ]

Williams, R.L., Creasy R.K., Cunningham, G.C., Hawes, W.E., Norris, F.D. and Tashiro, M. (1982). Fetal growth and perinatal viability in California. Obstet. Gynecol. 59: 624-632.         [ Links ]


(Received October 8, 1997)

Creative Commons License All the contents of this journal, except where otherwise noted, is licensed under a Creative Commons Attribution License