The aim of the present study was to compare gender differentials in health in Havana (Cuba) with those in Mexico City (Mexico) and in the US Hispanic population. Our findings indicate that both the absolute and, in particular, the relative gender gaps in four major health domains were consistently more pronounced in the Havana sample than in the samples for Mexico City and foreign-born Hispanics in the USA. The gender gap observed among the Havana respondents persisted even after controlling for SES, family characteristics, and smoking. Although the larger relative differentials in Cuba can be attributed to the much lower overall (national) and gender-specific prevalence of poor health, the present study shows that there was a strikingly large female disadvantage in health in Havana with respect to both absolute and relative gender gaps.
Our findings suggest that although Cuba has been maintaining high levels of gender and social equity (including universal access to health care) and can be regarded as one of the longevity vanguards in Latin America and Caribbean region, women in Havana (aged 60+) bear a disproportionate burden of ill health, as tends to be the case for women in much less equitable societies. This pattern may also be a consequence of the major causes of death shifting towards the stage when chronic age-related diseases are dominant.
The Cuban experience suggests that ensuring universal access to basic medical care, which is very efficient, for example, in the prevention of infectious diseases (Cooper 2006; Macdonald et al. 2006b), may be not sufficient to address recently emerging health threats that require modern and costly medical technologies for early diagnosis and treatment. Importantly, older adults can be considered a vulnerable group in the Cuban health system due to the scarcity of health care resources, which are predominantly allocated to provide care for children and women of childbearing age (Da Silva Coqueiro et al. 2010). Our study provides indirect evidence that Cuba has limited health care resources for the older population, as the treatments and procedures needed by older patients tend to be expensive and sophisticated and often involve more risk due to the presence of comorbid conditions. Furthermore, Verstraeten et al. (2016) showed that women in the Caribbean were generally more disadvantaged than men because of their poorer working conditions, higher risk of experiencing sexual and physical violence, and disproportionately large care burdens as heads of single-parent families. Our study has highlighted the disadvantages of Cuban women in several key health dimensions and points towards a contradiction between the formally declared high levels of gender and social equality under communism (Sarmiento 2010) and the actual gender gaps observed.
The gender gaps in health in Mexico City and among the foreign-born Hispanics in the USA were less consistent. The finding of no gender gaps in poor SRH and ADL disability among foreign-born Hispanics in the USA was particularly surprising.
Male-dominated migration of Latin Americans to the USA continues to occur despite the increase in female migrants in recent years (Riosmena and Massey 2012). Because our study population included only foreign-born Hispanics in the USA aged 60+, most of them are likely to have migrated years ago. Thus, the initial health advantage of Hispanic immigrants to the USA may have decreased over time because of adverse effects of hazardous jobs following immigration to the USA and poorer lifestyles, which are more prevalent among men than among women (Antecol and Bedard 2006). The existing literature has also shown that the health of US Hispanics tends to differ substantially by their generational status (foreign-born versus US-born), place of residence, and how quickly they assimilate the American lifestyle (Escarce et al. 2006). While we were not able to include some of these measurements, our results provided some indications that lower SES among women might be an important contributor to the disadvantage in mobility limitations observed among the foreign-born Hispanic women.
Our study also explored the question of whether gender differentials in health can be explained by differences in the socio-economic and behavioural characteristics of men and women. We found that controlling for differences in socio-economic status, family characteristics, and smoking behaviour did not produce substantial changes in the originally observed age-adjusted relative gender differences.
Our findings that there were no female disadvantages in SRH and ADL disability partially contradict previous results that reported gender differences among Mexicans aged 60+ in the prevalence of diabetes, depression, anaemia, and malnutrition (Wheaton and Crimmins 2015). A possible explanation for these inconsistent findings is related to where the respondents live, as the place of residence may be an important component of the impact of socio-economic characteristics on gender differences in health. As almost all of the population of Mexico City live in urban areas, gender differences may be less evident there.
Accounting for family characteristics did not modify the magnitude of gender differences in most health outcomes in all three study populations, but they may have played an important role in the relationship between gender and depression in the Havana sample. Controlling for current partnership and number of children slightly reduced the gender gap in depression, which is in line with the social support literature suggesting that emotional support has a positive effect on mental health (Kawachi and Berkman 2001).
Finally, we explored the potential role of smoking in explaining the gender gap in the selected health characteristics. In the Havana sample, gender differences in poor SRH, ADL disability, and mobility limitation slightly increased after adjusting for smoking behaviour, while in the Mexico City and the foreign-born US Hispanic samples, gender differences remained almost unchanged. The differences in effects of smoking may be explained by the relatively high prevalence of women who smoke in Havana in comparison with Mexico City and the foreign-born US Hispanics (See Supplementary Table 1).
Strengths and limitations
The major strength of our study is the extension of research on gender inequalities in health to Latin American populations and the consideration of various health outcomes. The study also includes three settings with very different gender equities (Cuba and Mexico), selection into old age (US Hispanics), and health care access.
However, this study comes with some limitations. SABE was conducted only in the very big cities of Havana and Mexico City. Therefore, the data do not represent the entire Cuban or Mexican populations. It is noteworthy that the health care infrastructure, particularly for the 60+ population, appears to be more advanced in urban areas than in rural areas. In Cuba, Havana has the highest concentration of the population aged 60+ (19.5% for 2009), and the living circumstances of older people are particularly challenging due to a lack of services and poor housing conditions. Thus, the (older) population in Havana may not differ from the Cuban population as a whole with respect to health or quality of life (Coyula 2010).
In Mexico City, the situation is not much different. In a newspaper article, Cruz-Flores (2018) reported that the capital is one of the richest areas of the country but suffers from an uneven distribution of health care providers (i.e. doctors and nurses) and infrastructure within the city itself. The hospital infrastructure in particular is lagging behind, as most of Mexico City’s hospitals were built over 60 years ago and are not equipped to deal with the chronic conditions that affect the population of the capital, or the infectious diseases that are common in the peripheral neighbourhoods of the city.
Furthermore, our results obtained from the HRS sample (2000) may not be generalisable to the whole population of US Hispanics, as the characteristics of this population, like place of residence and country of origin, have become more diverse in recent years (Ennis et al. 2011).
Another key concern in this cross-national comparison study is the comparability of the SABE and the HRS data. Although the data on the Havana and Mexico City populations were collected using the same data collection format, the HRS applied a different methodology, which could interfere with the comparison of health disparities across populations within each gender. Moreover, the comparability of mobility limitations across settings can be hindered by differences in public or housing environments, for example, the presence of ramps, elevators, and adequately designed stairs. However, the aim of the present study was not to provide a cross-country comparison of health, but to examine gender differences in health. The latter comparison is to be likely less sensitive to cross-country differences in data collection instruments, response patterns, and public or housing environments. Supporting this proposition, recent research found apparent cross-country differences but no clear gender differences in reporting of health (Jürges 2007; Oksuzyan et al. 2019; Spitzer and Weber 2019). These findings suggest that the comparison of self-reported health measures between genders across countries should be credible.
Moreover, SABE and HRS included only non-institutionalised populations. Since widowhood is more prevalent among women than men (Dupre et al. 2009), a man in need of care is more likely to be living in the community with his female spouse, while a woman in need of care is more likely to be living in a nursing home or other long-term care facility (Hurd et al. 2014). Thus, gender differences in more serious disabilities can be underestimated in community-based samples (Kelfve et al. 2013). However, we do not expect that underestimation occurred in our study, given that for cultural and social reasons, institutionalisation is a much less common practice among Cubans, Mexicans, and US Hispanics than it is among the populations of high-income countries (Angel et al. 2014).
Finally, we cannot exclude the possibility that some respondents could misinterpret or misunderstand the questions posed to them or misinform, voluntarily or not, by incorrectly assessing their condition (Rosenman et al. 2011).