Assessment of universal health coverage for adults aged 50 years or older with chronic illness in six middle-income countries

Abstract Objective To assess universal health coverage for adults aged 50 years or older with chronic illness in China, Ghana, India, Mexico, the Russian Federation and South Africa. Methods We obtained data on 16 631 participants aged 50 years or older who had at least one diagnosed chronic condition from the World Health Organization Study on Global Ageing and Adult Health. Access to basic chronic care and financial hardship were assessed and the influence of health insurance and rural or urban residence was determined by logistic regression analysis. Findings The weighted proportion of participants with access to basic chronic care ranged from 20.6% in Mexico to 47.6% in South Africa. Access rates were unequally distributed and disadvantaged poor people, except in South Africa where primary health care is free to all. Rural residence did not affect access. The proportion with catastrophic out-of-pocket expenditure for the last outpatient visit ranged from 14.5% in China to 54.8% in Ghana. Financial hardship was more common among the poor in most countries but affected all income groups. Health insurance generally increased access to care but gave insufficient protection against financial hardship. Conclusion No country provided access to basic chronic care for more than half of the participants with chronic illness. The poor were less likely to receive care and more likely to face financial hardship in most countries. However, inequity of access was not fully determined by the level of economic development or insurance coverage. Future health reforms should aim to improve service quality and increase democratic oversight of health care.


Introduction
The World Health Assembly in 2005 and the United Nations General Assembly in 2012 called for universal health coverage to reinforce the human right to health. All Member States were requested to guarantee affordable promotive, preventive, curative and rehabilitative health care of the highest attainable standard for everyone, without distinction. 1,2 However, over the past three decades, market deregulation and political crises have led to increased inequalities in income and opportunity in many countries. These inequalities are reflected in highly fragmented health and social security systems, which are increasingly differentiated by socioeconomic strata, and in setbacks for publicly funded health services. As a result, poorer social groups, including the historically marginalized and those more recently excluded from social protection systems, are forced either to forego care or to pay for access to increasingly costly health-care services. Consequently, access to health care in many countries has deteriorated to such an extent that health problems have become a threat to social development and cohesion. 3 At the same time, the rise in chronic noncommunicable diseases makes international development goals more difficult to achieve and complicates strategies for attaining universal health coverage. Moreover, the epidemiological transition, which is characterized by a sharp increase in population growth and a change in the leading causes of death, is rapidly accelerating in the poorest strata of society, where people are less likely to have access to appropriate services and are at risk of catastrophic health-care costs. 4,5 Many middle-income countries, including those we investigated, have scaled up ef-forts to achieve universal health coverage through substantial health reforms with a particular emphasis on the poor and vulnerable. These reforms have adopted a range of voluntary and social health insurance schemes in their attempt to increase service utilization while avoiding financial hardship and encouraging equity of access. 6 The aim of this study was to investigate gaps in universal health coverage for specific socioeconomic groups by focusing on older adults with chronic illness in China, Ghana, India, Mexico, the Russian Federation and South Africa. We examined five key issues: (i) access to basic chronic care; (ii) protection against financial hardship; (iii) the influence of health insurance schemes; (iv) the influence of place of residence; and (v) general satisfaction with the health-care system. We also examined progress towards universal health coverage in the six countries.

Methods
The World Health Organization's (WHO's) Study on Global Ageing and Adult Health (SAGE) provides comparable, publicly available data on adults aged 50 years and older based on nationally representative household surveys for six countries: China, Ghana, India, Mexico, the Russian Federation and South Africa. 7 These countries have some of the fastest growing economies globally and together contain more than 40% of the world's population spread over four regions. 8 We analysed cross-sectional data from wave 1 of the study carried out between 2007 and 2010. The response rate in individual surveys ranged from 52% in Mexico to 93% in China. 7 Research Health care gaps for chronically ill older adults Christine Goeppel et al.
Our study population consisted of all participants in the Study on Global Ageing and Adult Health who could be categorized as older adults with chronic illness: they were aged 50 years and older and reported being diagnosed with at least one chronic disease, such as arthritis, hypertension, stroke, angina, diabetes, chronic lung disease, asthma or depression. Universal health coverage was assessed on three dimensions: (i) access to basic chronic care; (ii) income-related equity of access; and (iii) protection against financial hardship. We also investigated the influence of health insurance on both access to basic chronic care and financial hardship and the influence of rural or urban habitation on access and we compared levels of satisfaction with the health system between people who did and did not use outpatient care.
Access to basic chronic care was assessed using a compound indicator with three components: (i) the provision of treatment, such as medications or advice on physical activity or diet, for each of the patient's conditions; (ii) visiting outpatient services for the chronic condition or conditions one or more times in the last reported year; and (iii) maintenance of a stable health state after the last outpatient visit. In the surveys, health-care providers were categorized as medical doctors, nurses, physiotherapists or traditional practitioners. Equity of access was assessed on the basis of equal treatment for equal health needs. 9 Since all study participants had chronic conditions, they all needed access to health care. Therefore, any income-related disparity in access to basic chronic care within a country indicated the existence of an inequity.
Financial hardship was defined in two ways: (i) catastrophic household spending on health in the last reported year of more than 30% of annual average household income, after the deduction of food expenditure -health expenditure included prepayments and out-of-pocket expenses; and (ii) catastrophic out-ofpocket expenditure for the last outpatient visit of more than 30% of annual household per capita income, after the deduction of food expenditure -expenditure on the outpatient visit included doctor fees and the cost of medications, diagnosis and transport. For our analysis, we based household income quintiles on annual household per capita income. 10 A person with any type of health insurance was classified as insured. Dissatisfaction with the health-care system was assessed using two indicators: (i) dissatisfaction with health-care services; and (ii) insufficient involvement in health-care decision-making.

Statistical analysis
We adjusted study data for differences between countries in survey design and data collection in Stata version 13.1 (StataCorp. LP, College Station, United States of America) using person-level analysis weights based on selection probabilities in the survey sampling design and a post-stratification factor. All percentage estimates are weighted. Differences between countries in access to basic chronic care and financial hardship were described using weighted population means. Within countries, we assessed the effect of poverty by stratifying data by household income quintile and looked particularly at differences between the poorest quintile and the population mean. Differences in equity of access between countries were compared using concentration curves and indices. We employed logistic regression modelling to estimate: (i) the effect of health insurance on access to basic chronic care and financial hardship; (ii) the effect of rural or urban residence on access; and (iii) the effect of using outpatient services on dissatisfaction with the health system. Models were adjusted for sex, age, place of residence, educational level, income quintile, comorbidity and insurance. Finally, we examined the effect of macroeconomic and social factors on universal health coverage in different countries by determining whether gross national income per capita, public health expenditure per capita or the Gini coefficient was   Fig. 1). Access rates were highest for the richest household income quintile in all countries. In Ghana, there was a continuous gradient from poor to rich. In India, the access rate decreased from the poorest to the second poorest quintile and then increased continuously to the richest quintile. There was a sharp increase in the access rate for the fourth and fifth income quintiles in Mexico and, for the richest quintile, in the Russian Federation. In China and South Africa, only small changes in the proportion with access were observed across the quintiles (Fig. 2). Moreover, the concentration curves for all countries except South Africa lay below the equity line and tested dominant (Fig. 3, available at: http://www.who.int/bulletin/vol/umes/94/4/15-163832), which indicates that the rich had disproportionate access to chronic care. Inequity of access was most pronounced for Mexico. The related concentration index varied substantially from 0.003 (95% CI: −0.045 to 0.050) for South Africa to 0.249 (95% CI: 0.087 to 0.403) for Mexico -higher values indicate greater inequity between rich and poor (Fig. 4).
The proportion of households that faced catastrophic health spending in the last reported year varied between 23.5% (95% CI: 19.3 to 28. 3) in South Africa and 65.5% (95% CI: 60.6-69.8) in Ghana. Financial hardship was present in all socioeconomic strata, though the proportion affected was generally highest in the poorest household income quintile, except in Mexico, where the proportion affected was highest in the third quintile (

Research
Health care gaps for chronically ill older adults Christine Goeppel et al.
Participants with health insurance were significantly more likely to have access to basic chronic care than those without in China, Ghana, India and Mexico but not South Africa (Table 2). In India, insurance increased the odds more than threefold. Nevertheless, insurance did not necessarily protect against financial hardship. In China, Ghana, India and South Africa, the risk of catastrophic health spending in the last reported year was the same or even higher for the insured as the uninsured. In India, health insurance was also associated with a higher risk of catastrophic out-of-pocket expenditure for the last outpatient visit. However, in Mexico insurance coverage was associated with a significantly lower risk of catastrophic out-of-pocket expenditure for the last outpatient visit (odds ratio, OR: 0.35; 95% CI: 0.14 to 0.84) and a nonsignificantly lower risk of catastrophic health spending in the last reported year (OR: 0.49; 95% CI: 0.22 to 1.07). In Ghana, the insured had a significantly lower risk of catastrophic out-of-pocket expenditure (OR: 0.38; 95% CI: 0.23 to 0.62) but a nonsignificantly higher risk of catastrophic health spending in the last year (OR: 1.22; 95% CI: 0.86 to 1.73). Living in a rural rather than an urban area was not associated with a lower likelihood of access to basic chronic care in any country except Ghana, where people living in rural areas were significantly less likely to have access (Table 3).
Only 4.5% (95% CI: 3.0 to 6.7) of the participants in Ghana were dissatisfied or very dissatisfied with health-care services, as were only 6.3% (95% CI: 5.2 to 7.5) in China (Table 4, available at: http://www.who.int/bulletin/ volumes/94/4/15-163832). The highest proportion who were dissatisfied or very dissatisfied was in Mexico (20.8%; 95% CI: 16.0 to 26.7), where, in addition, 19.1% (95% CI: 14.3 to 25.1) rated their involvement in health-care decisionmaking as "bad" or "very bad". In China, Ghana, Mexico, the Russian Federation and South Africa, people who did not use outpatient care tended to be less satisfied with the health system than those who did ( Table 4).
The six countries in our study differed markedly in their level of economic and social development: for example, gross national income per capita in the Russian Federation was approximately six times that in Ghana (   Notes: Chronic illness was defined as being diagnosed with at least one chronic disease. The concentration index varies from -1 to +1 and is 0 for equity of access to basic chronic care. A value greater than 0 indicates that the rich have better access to care than the poor. The Gini coefficient measures the extent to which the distribution of income or consumption expenditure among individuals or households within an economy deviates from a perfectly equal distribution: a coefficient of 0 represents perfect equality, whereas a coefficient of 1 implies perfect inequality.  (Fig. 4). In fact, South Africa performed best in terms of achieving universal health coverage: 45.5% (95% CI: 41.1 to 50.0) of all participants had access to basic chronic care without incurring catastrophic out-of-pocket expenditure for the last outpatient visit (Fig. 7). The Russian Federation had the highest gross national income per capita and the highest public health expenditure per capita. Generally, access to basic chronic care without financial hardship in the country was similar to that in South Africa, but the poorest quintile was disadvantaged. Although China's gross national income per capita was comparable to South Africa's, both public health expenditure and the proportion of participants with access to basic chronic care without financial hardship were lower. Mexico had the second highest public health expenditure but performed poorly in terms of providing universal health coverage: a ORs and 95% CI were calculated using logistic regression models that controlled for sex, age, urban or rural residence, educational level, household income quintile and comorbidity. b Health insurance included voluntary and mandatory insurance. c Participants had reported being diagnosed with at least one chronic disease d As insurance coverage was almost universal in the Russian Federation no ORs could be calculated. e Basic chronic care included: (i) the provision of treatment, such as medications or advice on physical activity or diet, for each of the patient's conditions; (ii) visiting outpatient services for the chronic condition or conditions one or more times in the last reported year; and (iii) maintenance of a stable health state after outpatient care. f Catastrophic health spending in the last year was defined as the household spending more on health in the last reported year than 30% of annual household income, after deduction of food expenditure. g Catastrophic out-of-pocket expenditure was defined as spending more on the last outpatient visit than 30% of annual household per capita income, after deduction of food expenditure.

Discussion
The prevalence of diagnosed chronic conditions in people aged 50 years or more varied widely among the six study countries and was higher in the more developed and more urbanized areas. This may be due to the negative health impact of the lifestyle changes accompanying modernization. 18 None of the six countries provided access to basic chronic care for more than half the participants, which is in line with evidence of gaps in essential services for noncommunicable diseases in low-and middle-income countries. 19 Although it is often assumed that wealthier countries may be better at providing health services for noncommunicable diseases, we found no evidence that a higher level of development was associated with greater universal health coverage. 20 The inequities in health coverage we observed in middle-income countries and that have been reported in other studies persist despite substantial health reforms aimed at improving universal health coverage, especially for poor and vulnerable groups. [21][22][23][24] The poor chronically ill were less likely to receive basic chronic care and more likely to face financial hardship than the better off in all countries in our study, except in South Africa, where primary health care is provided free to all citizens. 25 Although previous evidence suggests that rural residents have more limited access to primary care than urban residents and are less likely to have health insurance, 26 we found that participants in rural and urban areas had similar access to basic chronic care in all study countries except Ghana. More detailed country-specific data are required to explore this potential difference in greater depth.
In Mexico, we found that almost 80% of the participants did not have access to basic chronic care, which is consistent with the Mexican President Peña Nieto's statement in 2013 that much of the population cannot exercise their right to health. 27 In 2014, the Mexican National Council for the Evaluation of Social Development Policy reported that only 21.4% of the population received medical care for their health problems but that between 84% and 97% of people with health problems did receive medical treatment. 28 The inconsistency may be due to the difference between perceived and medically defined health needs. Since individuals' experiences and expectations of the probable outcome of medical care can shape perceptions of their health status, many older people may not think their health problem requires medical care despite being diagnosed with a chronic condition. 29 We found that people who were dissatisfied with the health system were less likely to seek care despite medical need. The same appears to be true for people who perceive it as ineffective. 30 In India, the common belief that the private sector offers better quality care coupled with inadequate public provision has led many people to use private facilities and bear high out-of-pocket costs despite increased investment in public health and the exemption of vulnerable groups from user fees. 4 We found that both the insured and uninsured could experience financial hardship in all study countries. Although health insurance improved access to health care, it also increased the risk of catastrophic health spending in most countries. In the Russian Federation, universal health insurance became mandatory in 1993 and health services are provided free at the point of care. However, the cost of pharmaceuticals excluded from guaranteed packages and informal payments can result in catastrophic expenditure. 23 In Ghana, where a national health insurance scheme was established in 2003, we found that insurance increased access to basic chronic care and protected against catastrophic out-of-pocket expenditure for the last outpatient visit, in agreement with the previous findings. 31 However, insured households were more likely to incur catastrophic spending during the last year, perhaps due to more frequent service utilization by the insured. 31 In China, health insurance did not significantly influence the likelihood of catastrophic out-of-pocket expenditure for the last outpatient visit but the insured were more likely to experience catastrophic spending during the last year, as reported previously. 32,33 This suggests that recent social health insurance programmes in China have neither reduced the risk of catastrophic spending nor relieved the financial burden on older people with chronic conditions. In India and South Africa, a small minority of insured people had an increased risk of catastrophic spending -they were mainly covered by private insurance schemes that may have encouraged the use of specialist providers with higher co-payments. 21,34 In Mexico, insurance generally increased protection against financial hardship. However, our findings provide only partial evidence that the voluntary Seguro Popular insurance scheme introduced in 2003 protected against financial hardship because health service utilization was extremely low, especially among the poor, and little information was available on insurance schemes in Mexico's highly fragmented health insurance system. 35,36 Our study had several limitations. The WHO Study on Global Ageing and Adult Health provides the best, available, comparable data on older adults with chronic illness in middle-income countries because it uses a unified method but the self-reported prevalence of noncommunicable disease is less than the actual prevalence in older people. In particular, in some countries the poor are less likely to be given a diagnosis. 20 Accordingly, the level of access to basic chronic care may have been overestimated and, consequently, inequities may have been underestimated. Although the national representativeness of the household survey could have been weakened by the variation in response rate between countries, the results of our sensitivity analysis confirmed the validity of the samples (data available from the corresponding author).
Universal health coverage remains a distant hope for many older adults with chronic illness in middle-income countries. Although allocating a higher share of a country's gross national income to health might improve services and subsidize health care for the poor, economic development does not in itself guarantee universal health coverage or greater equity. Nevertheless, lower socioeconomic inequality generally leads to more equal distribution of health services. Yet, as evident in South Africa, the provision of free primary health care can help achieve equitable universal health coverage despite high socioeconomic inequality, which suggests that universal social protection may guard better against catastrophic expenditure than insurance schemes. If gaps in universal health coverage are to be closed, it is essential that the care provided is acceptable to the population. Consequently, health reforms should aim to improve service quality and promote democratic oversight of health care through increased social participation in addition to expanding insurance schemes. The provision of universal health coverage for older people with chronic conditions is particularly challenging for low-and middle-income countries, especially given the ongoing epidemiological transition. It is crucial, therefore, that future health policies are tailored to the specific needs of older people. ■
Health care gaps for chronically ill older adults Christine Goeppel et al.

Household income quintile
Notes: Chronic illness was defined as being diagnosed with at least one chronic disease. Catastrophic health spending in the last year was defined as the household spending more on health in the last reported year than 30% of annual household income, after deduction of food expenditure. Household income quintiles were based on annual household per capita income. The greater the distance between quintiles, the greater the effect of household income on the likelihood of catastrophic health spending. Poorest 20%

Household income quintile
Notes: Chronic illness was defined as being diagnosed with at least one chronic disease. Catastrophic out-of-pocket expenditure was defined as expenditure more than 30% of annual household per capita income, after the deduction of food expenditure at the last outpatient visit. Household income quintiles were based on annual household per capita income. The greater the distance between quintiles, the greater the effect of household income on the likelihood of catastrophic out-of-pocket expenditure. The Gini coefficient measures the extent to which the distribution of income or consumption expenditure among individuals or households within an economy deviates from a perfectly equal distribution: a coefficient of 0 represents perfect equality, whereas a coefficient of 1 implies perfect inequality.