Evidence from observational epidemiology
An observational study in an urban South African township found that extreme childhood poverty, particularly that involving frequent hunger, was associated with increased obesity risk in women but not in men (7). While >75% of the women in the 2004–2005 household survey were overweight or obese, only 30% of men were. Decomposition analysis indicated that childhood hunger accounted for 11% of the disparity in obesity prevalence between women and men (7). Extreme childhood hunger likely correlates with fetal and infant nutritional deficits.
A US observational study found that the female-male gap in black young adult obesity prevalence was largest in adults whose parents had low educational attainment (8). Low parental education is a proxy for low childhood socioeconomic status (SES).
In the United States, low-SES populations are at especially high risk for low-birth weight (9) and early nutritional deficiencies coupled with high caloric and fat intake in adolescence and adulthood (10). Other social factors, such as family structure and household size, only weakly modified the gender gap (8). The specificity of parental education as a modifier of the gender gap lends credibility to the hypothesis that in utero or infant exposures related to low-SES heighten long-term adult susceptibility to obesity in a sex-specific manner. Further, the finding was generalizable to other race groups. A gender disparity was also found among the small subgroup of Whites with low-SES childhood backgrounds.
Evidence from quasi-experimental studies
The previous evidence is from observational research and assumes that Americans who were low-SES as children and black South Africans who reported frequent hunger as children experienced an environmental mismatch between their early life and adult nutritional environments. Stronger causal evidence that in utero or perinatal nutritional deprivation influences gender differences in obesity susceptibility comes from quasi-experimental studies. In observational studies, the relationship between an exposure—such as fetal nutritional deprivation—and an outcome may be confounded by many other risk factors. Therefore, it is difficult to tease apart the effect of the exposure from the effects of other risk factors. In quasi-experimental studies, however, the exposure of interest is exogenous to, or statistically independent of, most other risk factors.
Perhaps the most well-known quasi-experimental study exploring the influence of fetal deprivation on later health is the Dutch famine study. The Dutch famine study is a quasi-experimental retrospective cohort of adults who were in utero in the Netherlands before, during, and after a World War II-era siege and ration cut (11). Exploiting the quasi-random nature of a 5-month ration cut, contemporary researchers examined 59-year-old adults who were exposed to the ration cut at different stages of gestation (160 males, 190 females) and nonexposed controls sampled from same-sex siblings (180 males, 180 females) and births at the same hospitals as the cases 2 years before or after the ration cut (137 males, 159 females). During the 5-month exposure period, rationed calories for adults ranged from 900 kcal/day down to 500 kcal/day, consisting almost entirely of bread and potatoes by the end. The nutritional deprivation ceased within weeks of liberation in spring 1945; residents were accorded abundant nutrition. Female—but not male—offspring who were in gestation during the ration cut had higher BMIs and waist circumferences in late middle age than controls, who did not gestate during the ration cut (11).
A Nigerian study had similar findings (12). From 1967–1970, during a civil war, Nigerian military forces pushed 7 million Biafrans, a minority group who had declared independence from Nigeria, into a small enclave and cut off food supplies to the people segregated there (12). An estimated 1–3 million people died, mostly of starvation. In 2009, researchers surveyed 1,339 adults at marketplaces in the former capital of the breakaway state of Biafra. Adults born between 1965 and 1967 were categorized as exposed to famine in early childhood (246 males, 142 females). Adults born between 1968 and January 1970 were categorized as exposed to famine in fetal life and infancy (189 males, 103 females). Those born between 1971 and 1973 were classified as unexposed (353 males, 133 females). In women, obesity (BMI ≥30 kg/m2) and overweight (25.0 kg/m2 ≤ BMI < 30.0 kg/m2) were more common in those exposed to the famine in the fetal and infant period compared to those born after the famine. No effects were observed in men.
A third investigation was designed to exploit deprivation caused by the Great Famine in China (1959–1961) (13). Using data from the nationally representative 2002 Chinese Nutrition and Health Survey (CNHS), the researchers divided rural respondents into those born during the 1959–1961 famine (1,997 men, 2,366 women) and a control group born in 1964, after the famine (1,220 men, 1,473 women). In men, there were no differences between the famine—exposed and control groups in mean BMI nor prevalence of study-defined overweight (24 ≤ BMI ≤ 27.9 kg/m2) or obesity (BMI ≥ 28 kg/m2). Women born in famine years, however, had higher BMI, greater prevalence of overweight, and greater prevalence of obesity than control women born in the same regions.
A final quasi-experimental study found no sex differences in effects of early life famine on later BMI, waist-to-hip ratio, or subscapular triceps skinfold ratio (14). A German blockade of Leningrad (present-day St Petersburg) between September 1941 and January 1944 included an especially severe period of starvation between November 1941 and May 1942. During this period, the average daily ration was around 300 cal, almost entirely of bread. Using a regularly updated record of all people living in or born in the city of Leningrad during the siege, the study assembled a group exposed to the blockade in utero (born November 1941 to June 1942; 37 males, 132 females) and a group exposed in infancy (born January 1941 to June 1941; 62 males, 130 females). A control group born in the same province but outside the besieged area between January 1941 and June 1942 were recruited as adults from hospital clinics and six local workplaces (50 males, 138 females). Diabetics were excluded from all analyses. The study found no differences among the intrauterine, infant, and control groups, in overall or sex-stratified analyses.