Sex ratios at birth in African populations: A review of survey data

Human Biology, Dec 2002 by Garenne, Michel

Abstract This study analyzes the distribution of sex ratio at birth in African populations using data collected in birth histories in sample demographic surveys (Demographic and Health Surveys and World Fertility Surveys). The average sex ratio from 56 surveys, totaling 1.130 million births, was 1.033 (95% CI, 1.029-1.037), significantly different from the world average of 1.055. The distribution of sex ratios across surveys was found to be heterogeneous, and different from what could have been expected from random fluctuations due to sample size. Three subsets were identified: a subset with lower sex ratios, primarily in countries of eastern and southern Africa of Bantu populations (1.010), a subset with average sex ratios (1.035), and a subset of countries with higher sex ratios, in particular Nigeria and Ethiopia (1.070). Further analysis revealed that African populations are as diverse as other populations, with sex ratios ranging from low values (below 1.00) to high values (above 1.08). Results are discussed in light of independent data sources and in comparison with other human populations.

Sex ratios at birth, defined as the ratio of the number of male births to female births, vary from population to population and from period to period. Part of the yearly variations of sex ratios at birth in national populations is due to random fluctuations, part is due to differences between populations, and part is due to changing sex ratios over time. In the world as a whole, the sex ratio at birth averages about 1.055. For instance, among the national statistics available since 1950 and published by the World Health Organization (WHO) in 1997, the average was 1.0545 with a small confidence interval (1.0533-1.0557) due to a large number of births (1.07 billion). Several authors have noted the relatively large variations among populations, from 1.02 or 1.03 among African American and Native American populations in the United States to 1.08 to 1.09 among Asian populations, with a world record in Korea: 1.131 between 1921 and 1929 (Ciocco 1938; Visaria 1967). In the sample of births collected by WHO, selecting countries with at least 100,000 births, the sex ratio at birth varied from less than 1.030 in the West Indies (Jamaica, Bahamas, Barbados) to more than 1.080 (Thailand, Paraguay). The sex ratio at birth has also been shown to vary over time in the same population. For instance, during the 20th century in France, the sex ratio at birth went through various phases: it tended to increase from 1900 to 1913 (1.041 to 1.048), peaked at the end of the first world war (1.064 from 1918 to 1920), declined afterwards from 1.051 in 1921 to 1.038 in 1941, peaked again during and shortly after the second world war (1.059 from 1942 to 1948), declined again afterwards from 1.053 in 1949 to 1.046 in 1963, increased somewhat for about 10 years to reach 1.055 in 1972, and stabilized with a mild declining trend in the last period, tending towards 1.051 in 1999 (recomputed from data published by INSEE, the French National Institute of Statistics). Similarly, in Latin America, the sex ratio at birth was found to increase steadily between 1982 and 1986 (Feitosa and Kreiger 1993), whereas it was decreasing in Chile between 1900 and 1964 (Visaria 1967).

Factors accounting for variations in the sex ratio are poorly understood. First, genetic factors are well recognized, in particular from the father's side. For instance, in the United States, African American populations have lower sex ratios than Americans from European origin. In the case of mixed couples (black/white) in the United States, children with African American fathers are less likely to be boys than children with an African American mother, and likewise for Native American/white mixed couples (Khoury et al. 1984). Other factors have also been found significant in various instances: age of mother and father, birth order, socioeconomic status, maternal nutritional status (measured by the body mass index of weight / height-square), marital status (in particular polygyny), coital frequency, and a variety of environmental factors, in particular season and exposure to uranium in mines (Feitosa and Kreiger 1993; Andersson and Bergstom 1998; Whiting 1993; Ruder 1985; Rostron and James 1977). Some of these findings on risk factors could not be replicated in other populations, and the debate remains quite open, primarily because of large random fluctuations of the sex ratios and a lack of data with a combination of all risk factors.

Very little is known on sex ratios in sub-Saharan African populations, since published data of registered births are scarce. In their classic review, Brass et al. (1968) assume a sex ratio of 1.03 for tropical Africa, close to that of black populations in the United States and the United Kingdom. In Congo (former Zaire), Romaniuk (1967) found an even lower sex ratio of 0.978 (95% CI, 0.960-1.005). However, in Nigeria several authors have noticed higher sex ratios, such as 1.067 in Katsina province (Rehan 1982).