Pre-pregnancy body mass index and gestational weight gain and their effects on pregnancy and birth outcomes: a cohort study in West Sumatra, Indonesia

Study setting

The research was carried out in the West Sumatra province of Indonesia. West Sumatra has a total area of 42,013 km² and in 2010 had a population of 4,846,909 [18]. In 2010, West Sumatra had a total fertility rate of 2.9% [19], an infant mortality rate of 47 per 1000 and 89.5% of mothers were attended at their birth by a health professional [18]. Overall Indonesia has a maternal mortality rate of 240 per 100,000 maternities and female life expectancy is 73 years [18].

Sampling strategy

Stratified random sampling was used to select 3 urban and 3 rural districts to participate within this cohort study, from the 19 districts in West Sumatra – see Fig. 1.

Recruitment

At least one midwife is placed in every village in West Sumatra. The Midwives Organization within each of the 6 areas was contacted and through collaboration with them, all 537 midwives in the area were recruited to help with the study. This meant every sub-district in each area had a midwife representative. All 537 midwives were trained by the researchers, including explanations about the study aims and objectives and data collection procedures. Women in the first trimester of pregnancy within the identified districts attending their first antenatal visit between August and December 2010 were invited to take part in the study. Study recruitment occurred in government run public health centres, with the exception of Pariaman where some women were recruited from private clinics.

Standard measures

LBW was defined as <2500 g at delivery and macrosomia as >4000 g. Western definitions were used as no Asian standards for birthweight are currently available [23]. Birthweight for gestational age was compared to Alexander et al. (1996) [24] which was the currently accepted standard [25]. Birthweights below the 10th centile were classified as SGA and those above the 90th centile as LGA.

Data analysis

Logical checks and data cleaning were carried out by the investigators and inconsistencies were returned to the field for clarification. All survey data were double-entered and cleaned using SPSS 24.0. For binary outcomes, logistic regression analyses were used for comparison of groups. Outcomes on a continuous scale were compared using analysis of covariance. Multivariate logistic and linear regression were used to adjust comparisons for confounding factors. For binary outcomes, crude and adjusted odds ratios are reported (OR and aOR) and for continuous outcomes, crude and adjusted mean difference (MD and aMD) are reported, all along with their 95% confidence interval (CI). Confounders included in the adjusted analyses were maternal education, age, parity and district (urban/rural) for BMI comparisons. The same confounders as well as maternal BMI were included in comparisons of gestational weight gain. The OR for perineal suturing was calculated for women achieving a vaginal delivery only. Women with a normal BMI and women with a recommended weight gain were taken as the referent groups. The difference in proportions between the two BMI classification systems were calculated using the two-proportion z-test.

Ethical approval

This study was conducted according to the guidelines laid down in the Declaration of Helsinki and all procedures involving human subjects/patients were approved by the Faculty of Medicine of Andalas University Ethics Committee (045/KEP/FK/2010). Written informed consent was obtained from all participants.

Maternal characteristics

Maternal pre-pregnancy BMI and its relation to pregnancy and birth outcomes

Pregnancy and birth outcome proportions according to pre-pregnancy BMI are presented in Additional file 1. Within this cohort, mean maternal GWG ± SD was 10.2 ± 6.0 kg (n = 544) and mean birthweight was 3165 ± 402 g (n = 577). The results of the regression and multivariate analyses for maternal and neonatal outcomes are shown in Table 3 for both international [20] and Asian [21] BMI classifications. GWG was significantly higher in the underweight group, adjusted mean difference (aMD) (95% CI) 2.48 (1.24 to 3.73)kg for international and aMD 2.07 (0.80 to 3.33)kg for Asian BMI classification compared to women with a BMI in the normal range. Weight gain during pregnancy was reduced for women who were overweight and obese compared to women with a normal BMI within both BMI classifications. However when classifying GWG according to IOM recommendations, the odds of inadequate weight gain were lower in underweight, overweight and obese women than in women of normal weight, however this was not significant in the obese subgroup when using international BMI classifications. Mean birthweight was significantly increased in women who were obese within the international classification aMD 556 (234–878)g and in both the overweight and obese categories using the Asian BMI classification, aMD 117 (29–205)g and aMD 205 (46–365)g respectively.

As evident in Table 3, women who were obese had higher odds of having a macrosomic infant compared to the reference group of normal weight women. Women who were overweight had lower adjusted odds for giving their baby an initial breastfeed after delivery in both classification systems; adjusted odds ratio (95% CI) aOR 0.25 (0.12 to 0.56) for the international BMI classification and 0.31 (0.14 to 0.70) for Asian BMI classification. When using international BMI classifications, women with a high BMI also had lower odds of exclusively breastfeeding at discharge from hospital aOR 0.52 (0.29 to 0.92) and had higher odds of having an infant admitted to a neonatal intensive care unit (NICU) aOR 12.90 (1.01 to 164.98). When using the Asian BMI classification, women who were overweight had lower odds of haemoglobin less than 11.0 g/dl in trimester 2 aOR 0.45 (0.26 to 0.79). It was not possible to analyse the incidence of shoulder dystocia, APGAR score < 7 at 5 min, postnatal depression, Sudden infant death syndrome, and maternal mortality across the different BMI or weight gain categories as there were too few cases in the whole sample.

Gestational weight gain based on IOM recommendations and its relationship with pregnancy and birth outcomes

Irrespective of BMI classification system, a large proportion of women who were underweight (47.6%) or had a BMI in the normal range prior to pregnancy (>60%) gained inadequate weight according to IOM guidance during pregnancy (Fig. 2). In contrast for overweight or obese women, more gained the recommended amount of weight than inadequate weight. Over all BMI categories, when IOM recommendations were adapted to Asian BMI classifications, 50.7% of women gained inadequate weight during pregnancy, compared to 56.1% using international classifications. The difference in the proportion of women gaining inadequate weight using the two different BMI classifications was not quite significant (p = 0.073).

National and international context

The prevalence of women with a BMI of 25 kg/m² or greater was higher than in the study by Winkvist et al. (2002) a decade before in Java; however the level of under nutrition (BMI <18.5 kg/m²) prior to pregnancy was also higher [17]. This may therefore indicate that societal transition is leading to an exacerbated situation in terms of diversity in nutritional health in Indonesia; so that while still struggling with under-nutrition, they are also facing the challenge of overweight. Within this study a high proportion of women did not achieve the recommended weight gain during pregnancy, particularly women with a low or normal BMI. However the mean GWG and the proportion of women gaining the recommended amount of weight in this cohort was larger than in previous Indonesian studies in the coastal region of West Java in 1995 [16] and in a mainly rural area of Central Java in 2002 [17]. Recommended weight gain in this study was in line with other recent studies conducted in middle income countries including Pakistan [28] and Iran [29], however adherence was markedly different from high income country studies such as the US, Canada and Sweden where the proportion of women with excessive GWG is far higher [30–34]. The high incidence of inadequate GWG is of particular importance for clinicians, researchers and policy makers to develop and adapt strategic interventions in order to address these modifiable nutritional deficiencies which are known to have a significant impact on infant weight and wellbeing [11].

Birthweight outcomes

Birthweight was significantly higher in women who were obese and was significantly lower in those gaining inadequate weight within both BMI classifications. The odds of macrosomia was higher in women who were obese. The crude odds of LGA were also higher in overweight women with international BMI classification and in women in the obese category using the Asian BMI classification, however once adjusting for confounders these were no longer significant. This is in line with previous research which has shown increased pre-pregnancy BMI to be associated with an higher incidence of LGA and macrosomia using both international [28, 31, 35–37] and Asian BMI classifications [38–40]. When comparing IOM weight gain groups (inadequate, recommended, excessive) (Table 4) applied to Asian BMI classifications the odds of macrosomia was significantly lower in women gaining inadequate weight and the crude odds of LGA were higher in those gaining excessive weight, however once controlling for other factors this was no longer significant. The odds of SGA were also higher in women gaining inadequate weight using both international and Asian BMI classifications. There is much evidence that weight gain below that recommended by the IOM is associated with LBW [11, 41] and SGA infants [11, 31, 33, 36, 41] and exceeding weight gain recommendations is associated with higher proportions of macrosomia [41] and LGA infants [11, 31, 36, 37, 41]. Given that macrosomia is associated with a higher risk of mortality and morbidity [42] and LBW and SGA with neonatal mortality and chronic diseases later in life such as glucose intolerance, coronary heart disease, obesity and disturbed blood clotting [43]; ensuring adequate maternal nutrition and weight gain in pregnancy in order to improve birth outcomes is of paramount importance.

Other outcomes

Compared to women of a normal BMI, the odds of anaemia (haemoglobin level < 11.0 g/dl) were lower in overweight women when first tested in pregnancy in trimester 2. This may have been one of the mediating factors for higher birthweight with increasing BMI within this cohort, as maternal anaemia is known to influence birthweight outcomes [44]. When haemoglobin levels were measured in the 3rd trimester a lower incidence of anaemia was noted (Additional file 1) and there were no differences in odds between BMI categories. This may be due to the Indonesian policy of distributing iron supplementation tablets to all pregnant mothers. Given the known association between adverse outcomes and anaemia [44] and the large proportion of women within our study who were anaemic in early pregnancy, supplementation of Indonesian women prior to pregnancy seems appropriate. Any measures to optimise maternal BMI for pregnancy in Indonesia will need to address a potential resultant increased vulnerability to anaemia. Furthermore, our finding of a high incidence of anaemia and under-nutrition in this population may indicate insufficiency in other macro/micronutrients [45] which were outside the scope of this study.

There were lower odds of breastfeeding initiation at delivery for women who were overweight or obese compared to those of normal BMI, although once adjusting for other factors this only remained significant in the overweight group. The reduction in breastfeeding in higher BMIs in our study is of interest as it is consistent with global evidence [6, 11, 46]. Many varied reasons have been offered for these differences including physiological, psychological and socio-cultural factors, as well as the increase of co-existing medical pathologies alongside their associated higher risk of assisted birth that occurs with obesity [6]. The crude odds of CS were higher in obese women compared to women with a normal BMI, which is consistent with the literature [35, 36].

Comparison of Asian specific and international BMI classification systems

Although there is an ongoing debate about the appropriateness of Asian or International BMI classifications for specific communities, due to a high correlation between BMI, body fat percentage and health risks, WHO has suggested the use of specific BMI categories for Asian populations [21]. Evidence on adverse pregnancy outcomes and BMI cut off points for Asian populations, are limited. Assessing accuracy or comparing predictability or appropriateness of each BMI classification system for the pregnant women was not a primary objective of this study. As a side observation we have found similar results in observing significant associations between adverse pregnancy outcomes using international and Asian classifications in this population, however the Asian criteria seems to be more sensitive in identifying adverse outcomes. Encouraging women to gain weight in accordance with the IOM recommendations adapted for ethnicity could maximise the outcome of pregnancy in this group of women.

Strengths and limitations

The strengths of this research were that data collection was prospective and that it incorporated women from diverse regions of West Sumatra, including urban, rural, mainland and coastal areas, with a widespread engagement of midwives. A heterogeneous sample was obtained, as ascertained from maternal and partner occupation and education levels.

This is one of the few studies attempting to use both international and Asian specific BMI classifications in addressing the interrelationship among maternal nutritional status via anthropometric characteristics and weight gain during pregnancy. This large study, is of significance addressing global priority areas for women and their babies in a developing country with considerable societal and economic transitions.

A limitation of the study was that delivery data was only available for 60% of those initially recruited. However when we looked at demographic data for participants without delivery data compared to those with delivery data, there were no significant differences between the 2 groups except for the included women having significantly lower education levels and their partners occupation being more likely to be in agriculture or a labourer. Even when interpreting the results in the context of these differences, they still demonstrate high proportions of under nutrition and inadequate weight gain in these women. Delivery data only being available for 60% of the women recruited also led to small numbers within some categories, which may have reduced the power of calculating statistical differences and the number of outcomes that could be analysed. However the sample size has been adequate to demonstrate statistically significant differences in main outcomes such as birthweight, SGA and preterm birth.

It was difficult to ascertain the representativeness of our study sample to the actual population, due to the lack of available information on maternal age and parity in this region and due to numerous classification systems for occupation being in place within the official regional statistics. Our sample did however appear to be slightly more educated than the population in general and due to the stratified sampling technique our sample had more women living in urban and coastal areas than the average across the region.

A further limitation of the study is that pre-pregnancy BMI was calculated from women’s self-reported pre-pregnancy weight, however this is common within the literature in this area [4, 12, 23]. Furthermore within this study average self-reported pre-pregnancy weight and average actual weight in trimester 1 only differed by 0.9 kg and there was a high correlation between the two variables suggesting that biases from using estimated pre-pregnancy weight would be minimal.