Comment www.thelancet.com Vol 381 February 9, 2013 435 In The Lancet, Ai Koyanagi and colleagues 1 analyse data from WHO’s Global Survey on Maternal and Perinatal Health to describe the prevalence of macrosomia and its association with adverse maternal and perinatal outcomes in developing countries. The investigators analysed data for 276 436 births in 363 institutions in 23 developing countries in Asia, Africa, and Latin America. Inevitably, interpretation of this information has united the demands placed by the multinational nature of this study and data collection with the frustrations presented by interinstitutional variability in terms of standards of data collection. However, the greatest challenge was in the identification of a consensus definition of macrosomia across different nations and ethnicities. In developed countries, no absolute consensus exists about what defines macrosomia 2 or the principles that underlie diagnosis. Should macrosomia be diagnosed on the basis of birthweight alone? Should birthweight be combined with body composition? Should associated maternal and perinatal complications be incorporated when macrosomia is validated as a measure of true clinical significance? The American College of Obstetricians and Gynecologists 3 supports use of a 4500 g cutoff, irre- spective of gestational age, because babies weighing more than this value have an increased risk of adverse maternal and perinatal outcomes. Other groups 4,5 have noted increased validity with use of individualised cutoff points for neonatal weight calculated from population- specific growth curves. Agreement on a precise definition for diagnosis of macrosomia is difficult when its true clinical significance lies on a continuum. 6 In view of the multinational, multiethnic hetero- geneity of their study sample, Koyanagi and colleagues used a population-based approach for diagnosis, defin- ing macrosomia as birthweights greater than the 90th percentile of country-specific birthweights. When the investigators applied this definition to the entire study cohort, which notably includes institutional deliv- eries alone, they showed a substantially lower cutoff birthweight of 3750 g for the diagnosis of macrosomia than those of 4000 g or 4500 g often used in developed countries. The validity of a country-specific diagnosis of macrosomia is supported by the associated clinical risks and complications for both mother and baby. In line with findings from developed countries, maternal diabetes, increased gestational body-mass index (BMI), and higher parity were significantly associated with increased odds for macrosomia in all regions. 1 Furthermore, combined scores for maternal morbidity and mortality showed that babies with macrosomia (>90th percentile) had consistently higher odds for adverse maternal outcome in all regions than did those in the 10th–90th percentile, although the perinatal morbidity and mortality index was only significant for Africa. With interpretation of large cohort analyses, a key question is whether or not the study sample is representative of the population it aims to describe. 7 In the accompanying study, the WHO survey used for secondary data analysis used a stratified multistage cluster sampling design. Notably, the sample was selected from only institutions with more than 1000 deliveries per year and those where caesarean section was done. The results of Koyanagi and colleagues’ study should therefore be interpreted in the context of this unavoidable selection bias, because in most developing countries home births still account for the greatest proportion of deliveries. 8 When broad comparison is made with the Demographic and Health Surveys, 8 which selected births from both rural and urban cohorts across sub-Saharan Africa, Koyanagi and colleagues report consistently increased rates of obesity for countries included in both studies. Therefore, additional bias can be reasonably assumed for mater- nal and perinatal indices, conferred by the urban demographic of this study. However, in consideration of this limitation, the results presented by Koyanagi and colleagues point to a worrying trend of increased gestational obesity and its related complications 9 in urban centres in developing countries that are rapidly undergoing urbanisation and adoption of western culture. 10 Of equal concern is the associated high rate of maternal diabetes, which is probably under-represented in the present study. 11 In general, health facilities in many developing coun- tries are under-resourced and often rely on support from international organisations. 12 Hence, the risks and challenges presented by gestational obesity and diabetes might be perceived as a low priority compared with other population health concerns, such as HIV. Macrosomia: defining the problem worldwide Published Online January 4, 2013 http://dx.doi.org/10.1016/ S0140-6736(12)62090-X See Articles page 476 Getty Images