Introduction

Catastrophic health expenditure is a measure of financial risk protection and it is often incurred by households who have to pay out of pocket for health care services that are not affordable. Catastrophic health expenditure has been defined as a state when out-of-pocket payments cross estimated threshold share of household expenditure at which the household is forced to sacrifice their basic needs, sell their assets, incur unwarranted debt, or become impoverished [1,2,3]. There is no consensus in the existing literature and among health economists on the threshold proportion of household expenditure. However, there is an agreement that catastrophic health expenditure is medical spending or out-of-pocket expenditure that exceeds a defined threshold of a household’s total consumption or non-food consumption expenditure annually [1,2,3,4,5]. Different thresholds have been used by researchers to estimate catastrophic health expenditure in different countries [6]. Some studies have used 10% threshold of total household expenditure [3, 7] while others have used 40% of non-food consumption expenditure [5, 8]. Globally, many households incur catastrophic health expenditure and are pushed into poverty due to out-of-pocket payments [6]. According to the World Health Organization in 2010, it was estimated that 150 million people incur catastrophic health expenditure while 100 million people were pushed into poverty [9]. Populations in low- and middle-income countries (LMICs) including Nigeria experience high levels of catastrophic health expenditure. Studies conducted assessing the financial burden of out-of-pocket payment in Enugu State and Anambra State in Nigeria showed that the incidence of catastrophic health expenditure (at a 40% threshold of non-food expenditure) was 14.8 and 27%, respectively [10, 11]. Furthermore, another study conducted in Nigeria on the incidence and intensity of catastrophic health expenditure showed that 23% of households incurred catastrophic health expenditure at 10% threshold [12]. Evidence suggests that less than 5% of the Nigerian populations are under any health insurance coverage [13]. The Nigerian government’s expenditure on health is consistently below 30% (1995–2016) and this is far below what the population is contributing to health [14]. Out-of-pocket payment as a percentage of private health expenditure has also consistently been over 90% [14]. Furthermore, over 70% of the Nigerian population lives below the poverty line of $1.25 a day [15]. Nigeria is yet to implement the Abuja Declaration of 2001 in which African head of states pledge to set a target of earmarking at least 15% of their annual budget to improve the health sector [16]. In addition, Nigeria still lags far behind in moving closer to universal health coverage (UHC). Statistics show that more than half of the Nigerian population does not have equitable access to quality health care services [17]. All the metrics for measuring progress towards UHC recently proposed by Wagstaff and his colleagues showed that Nigeria may not achieve UHC until it stops the over-reliance on out-of-pocket payments as a major source of financing the health system [18]. It is therefore important to understand the factors that are associated with catastrophic health expenditure among households and individuals in Nigeria. The information generated will enable policy-makers and political actors take to action towards reducing the incidence and intensity of catastrophic health payments. The aim of this study is to determine the factors associated with catastrophic health expenditure.

Literature review

Many studies in both developed and developing countries have analyzed the determinants of catastrophic health expenditure [19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39] but there is limited evidence on the determinants of catastrophic health expenditure among households in Nigeria. Using the Nigeria General Household Survey data of 2010, Adisa [40] found that higher education and lack of health insurance are factors that are associated with the risk of incurring catastrophic health expenditure among urban elderly households while higher income, female-headed households, and larger household size are factors that are less likely to place urban elderly households at risk of incurring catastrophic health expenditure. Ukwaja et al. [41] analyzed the determinants of catastrophic payments for tuberculosis in Nigeria and found that age greater than 40 years, male gender, urban residence, formal education, receiving care at a private health facility, poor households, and households where the patient is the primary earner were significant determinants of catastrophic health expenditure for TB care. Ilesanmi et al. [42] found that a household in the lowest wealth quintile is a significant determinant of catastrophic health expenditure. However, household size and health insurance status were factors that decrease catastrophic health expenditure. A recent study by Omotosho and Ichoku [43] using the Harmonised Nigeria Living Standard Survey (HNLSS) 2009/10 dataset and the Xu methodology [44] to examine the determinants of catastrophic health expenditure in Nigeria found that factors such as household heads with age greater than 65 years, households with spouse greater than one, households in the north eastern zone, households in the north western zone, and households in the south eastern zone increases the risk of incurring catastrophic health expenditure.

Our study differs from these previous studies on the determinants of catastrophic health expenditure in Nigeria. The present study used HNLSS 2009/10 dataset, Wagstaff and van Doorslaer methodology [3], and three different thresholds for the concepts of total household expenditure and non-food expenditure in order to get insights into the sensitivity of results to the threshold levels. This study provides evidence and contributes to the literature on factors associated with catastrophic health expenditure.

Methods

Secondary data from the Harmonised Nigeria Living Standard Survey (HNLSS) 2009/10 was used for the study. HNLSS is a nationally representative cross-sectional study conducted by the National Bureau of Statistics (NBS) with funding by UK DFID and the World Bank as a follow up to the Nigeria Living Standard Survey (NLSS) of 2003/4. The HNLSS 2009/10 has an enlarged scope that includes health, household income, consumption and expenditure, demography, and education and skill/training compared with previous surveys. HNLSS is a combination of Nigeria Living Standard Survey (NLSS) household questionnaire and Core Welfare Indicator Questionnaire (CWIQ) jointly developed by National Bureau of Statistics (NBS) and the World Bank. Thirty-six states of the federation and the federal capital territory (FCT), Abuja were covered in the survey. The first part of the survey using the welfare approach was conducted among 77,400 households while the second part of the survey using the consumption approach covered 50 households in each local government area (LGA). In total, 38,700 households were interviewed. Sampling frame for the survey used Enumeration Areas specified by the National Population Commission (NPC). Two-stage sample design was employed. The first stage involved the selection of enumeration areas (EAs) while the second stage involved the selection of households. Data was collected through interviews conducted by NBS enumerators with household members on a quarterly basis from November 2009 to October 2010.

Variables measurement

Dependent variable

The dependent and/or outcome variable in this study is catastrophic health expenditure defined as a state when medical spending or out-of-pocket health expenditure exceed some fraction of a household’s total consumption or non-food consumption expenditure annually [1,2,3,4,5]. This was determined using the proportion of out-of-pocket expenditure in total household expenditure and non-food expenditure [3].

Independent variables

The independent and/or explanatory variables were guided by Grossman’s theory of demand for health [45], Andersen’s behavioral model [46], the multidimensional nature of the factors that indicate vulnerability to financial risk in the health sector [47], and literature review on the determinants of catastrophic health expenditure, which shows that household and individual characteristics are associated with catastrophic health expenditure [19,20,21,22,23,24,25,26,27,28,29,30,31,32,33]. According to Grossman, the demand for health care is influenced by factors such as health status, education, and age. In the Andersen’s Behavioral Model, the variables that determine the demand for health care are categorized into: predisposing factors (age, sex, ethnicity, and household head characteristics); enabling factors (location, geo-political zone, education, health insurance status, and household income); and need factors (perceived severity of illness, self-reported health status, presence of physician diagnosing chronic diseases, and overweight). Mukherjee et al. [47] stated that vulnerability to shocks due to health care payment consist of root causes (socio-economic characteristics of households, expenditure level and employment status), intervening causes (age, sex, location/place of residence, level of education, and supply-side factors) and intermediate causes (coping with financial cost of illness and coping in the absence of financial protection).

In our study, age was categorized into six groups since it may have a non-linear relationship with catastrophic health expenditure. The education level of the head of the household was defined according to the survey questionnaire and categorized into no education, primary education, secondary education, and post-secondary education. The gender of the head of the household was also defined according to the survey questionnaire and categorized into male and female. Household size was defined as the number of individuals in a household and was recoded as less than five members and more than five members. Location or place of residence was grouped into urban and rural areas. Urban residents are households living in towns and urban neighborhood of cities while rural residents are households living in villages and suburban areas of the cities. Nigeria is also grouped into six geo-political zones: north central, north east, north west, south east, south west, and south south. The southern region is the most developed while the northern region is least developed. Socio-economic status was constructed using an asset index based on variables such as type of dwelling place, number of rooms occupied, type of previous dwelling, occupancy status, material of wall of dwelling, floor of dwelling, roof of dwelling, source of fuel for cooking, source of lighting, refuse collection used, source of drinking water, source of other water, and type of toilet used. These variables were summed up for each household and their members. The final scores were used for grouping into wealth quintile as poorest, poorer, middle, richer, and richest. Work status was defined according to the survey questionnaire and recoded as unemployed and employed. Health insurance status was also defined according to those who pay out of pocket for health care services (lack health insurance) and those who did not (have health insurance). The type of facility visited was defined according to the survey questionnaire and recoded as public health facility and private health facility. The type of illness suffered was also defined according to the survey questionnaire and recoded as chronic and non-chronic illness.

Data analysis

Data were analyzed using SPSS version 22 software.

Measuring determinants of catastrophic health expenditure

Chi-square analysis was done to determine the association between catastrophic health expenditure and selected variables. Discrete choice approach was thereafter used to estimate the determinants of catastrophic health expenditure as in previous studies [19,20,21,22,23,24,25,26,27,28,29,30,31,32,33] using the formula:

$$\ln \left( {\frac{{\hat{p}}}{{(1 - \hat{p})}}} \right) = b_{0} + b_{1} X_{1} + b_{2} X_{2} + \cdots + b_{p} X_{p} .$$

The dependent variable (P) is the occurrence of catastrophic health expenditure dichotomized and defined as 1 when the household faces catastrophic health payments and 0 otherwise. The independent and/or explanatory variables in the logistic regression equation above included the following variables: location: urban and rural; gender of household head: male and female; education of household head: no education, nursery education, primary education, secondary education, and post-secondary education; health insurance status: insured and uninsured; household socio-economic status (poorest, poorer, middle, richer, richest); household size: less than five members and more than five members; type of health facility visited: public health facility or private health facility; type of illness suffered: chronic illness or non-chronic illness; age: very young (member below 5 years old), 6–14, 15–24, 25–54, 55–64 years and elderly (member above 65 years); geo-political zones: north central, north east, north west, south east, south south and south west; work status: employed and unemployed.

Two concepts were used to define catastrophic health expenditure. Catastrophic health expenditure was determined using the proportion of out-of-pocket expenditure in total household expenditure and non-food expenditure [3]. An out-of-pocket payment for health care is considered catastrophic when the payment exceeds 10, 25, and 40% of both total household expenditure and non-food expenditure.

Results

Descriptive statistics

Table 1 presents the population characteristics of households and individuals used in the study. The distribution of age in the population is 0–5 (13.8%), 6–14 (24.8%), 15–24 (18.4%), 25–54 (33.3%), 55–64 (5.1%), and 65 and above (4.6%). About half (46.6%) of households have no education, 0.1% have nursery education, 31.6% have primary education, 16.7% have secondary education, and 5.1% have post-secondary education. Also, 50.9% of household heads are male while 49.1% of them are female. A high proportion of households (77.5%) have more than five members while only 22.5% have less than five members. The majority of households (74.1%) live in the rural area while only 25.9% live in an urban area. Among the sampled population, 16.9% are from the north central, 12.5% from the north east, 27.7% from the north west, 12.3% from the south east, 15.0% from the south south, and 15.5% from the south west. More than half of households belong to the poorer (64.1%) and middle (35.7%) socio-economic class. A total of 62.8% of the study sample are employed while 37.2% are unemployed. More than two-thirds of the population (77.9%) lack health insurance coverage while only 22.1% have health insurance coverage. Among the sampled population, 99.7% visit a public health facility while only 4.7% visit a private health facility. A total of 99.2% of the study population suffered non-chronic illness while only 0.8% suffered chronic illness.

Table 1 Population characteristic
Full size table

Determinants of catastrophic health expenditure

Table 2 shows the determinants of catastrophic health expenditure using Chi-square analysis for the concept of total household expenditure. Irrespective of the threshold for this concept, there was a statistically significant association between catastrophic health expenditure and variables such as age of household members, education of household heads, gender of household heads, location, geo-political zone, work status, health insurance status, type of health facility visited, and type of illness suffered. There was no statistically significant association between catastrophic health expenditure and household size at 25% threshold of total household expenditure. Also, the association between catastrophic health expenditure and socio-economic status was not statistically significant at 10% threshold of total household expenditure. Similarly, Table 3 shows the determinants of catastrophic health expenditure using Chi-square analysis for the concept of non-food expenditure. Irrespective of the threshold for this concept, there was a statistically significant association between catastrophic health expenditure and variables such as age of household members, education of household heads, geo-political zone, socio-economic status, work status, health insurance status, type of health facility visited, and type of illness suffered. There was no statistically significant association between catastrophic health expenditure and gender of household heads at 40% threshold of non-food expenditure. Also, the association between catastrophic health expenditure and household size was not statistically significant at 25 and 40% thresholds of non-food expenditure. Furthermore, there was no statistically significant association between catastrophic health expenditure and location at 25% threshold of non-food expenditure. Table 4 presents results of the determinants of catastrophic health expenditure at 10, 25, and 40% threshold of total household expenditure using logistic regression model. The result shows that, irrespective of the threshold, age of household members (p = 0.000) was significantly associated with the risk of incurring catastrophic health expenditure. Having no education (p = 0.000), having primary education (p = 0.000), and having secondary education (p = 0.000) also increased the probability of incurring catastrophic health expenditure. However, there was no significant relationship between having nursery education (p > 0.05) and the risk of incurring catastrophic health expenditure at 10, 25, and 40% thresholds. Results also revealed that irrespective of the threshold, male household head (p = 0.000) increased the risk of incurring catastrophic health expenditure. At 10 and 40% threshold, having more than five members in the household (p < 0.05) increased the probability of incurring catastrophic health expenditure. However, there was no significant relationship between having more than five members in the household (p = 0.340) and the risk of incurring catastrophic health expenditure at 25% threshold. Results also revealed that irrespective of the threshold, households living in an urban area (p = 0.000) increased the probability of incurring catastrophic health expenditure. Irrespective of the threshold, results showed that households living in the north central zone (p = 0.000), households living in the north eastern zone (p = 0.000), and households living in north western zone (p = 0.000) increased the risk of catastrophic health expenditure. At 25% threshold, there was a significant relationship between households living in south eastern zone (p = 0.001) and the risk of incurring catastrophic health expenditure. At 40% threshold, households living in the south eastern zone (p = 0.000) and households living in the south southern zone (p = 0.026) increased the probability of experiencing catastrophic health expenditure. However, there was no significant relationship between households living in the south eastern zone and the risk of incurring catastrophic health expenditure at 10% thresholds. Also, there was no significant relationship between households living in the south southern zone and the risk of incurring catastrophic health expenditure at 10 and 25% thresholds. Irrespective of the threshold, being unemployed (p = 0.000) was significantly associated with catastrophic health expenditure. Results also revealed that irrespective of the threshold, lack of health insurance coverage (p = 0.000) increased the risk of incurring catastrophic health expenditure. Households visiting a private health facility (p = 0.000) was significantly associated with catastrophic health expenditure at 10, 25, and 40% threshold. Irrespective of the threshold, having household members with non-chronic illness (p = 0.000) also increased the probability of incurring catastrophic health expenditure. Results for Table 5 showed the determinant of catastrophic health expenditure at 10, 25, and 40% thresholds of non-food expenditure. Irrespective of the threshold, catastrophic health expenditure was associated with having a member aged between 6 and 14 years (p < 0.05), having a member aged between 15 and 24 years (p = 0.000), and having a member aged between 25 and 54 years (p = 0.000). At 25 and 40% threshold, having a member below 5 years (p < 0.05) increased the risk of incurring catastrophic health expenditure. Also, there was a significant relationship between having a member aged between 55 and 64 years (p < 0.05) and catastrophic health expenditure at 10 and 25% threshold. However, there was no significant relationship between having a member below 5 years (p = 0.117) and the risk of incurring catastrophic health expenditure at 10% thresholds. Also, results revealed that having a member aged between 55 and 64 years (p = 0.556) was not significantly associated with catastrophic health expenditure at the 40% threshold. Irrespective of the threshold, having no education (p < 0.05), having primary education (p = 0.000), and having secondary education (p = 0.000) increased the probability of incurring catastrophic health expenditure. However, there was no significant relationship between having nursery education (p > 0.05) and the risk of incurring catastrophic health expenditure at 10, 25, and 40% thresholds. Results also revealed that at 10 and 25% threshold, male household head (p < 0.05) increased the risk of incurring catastrophic health expenditure. However, there was no significant relationship between male household head (p = 0.223) and the risk of incurring catastrophic health expenditure at 40% threshold. At 10% threshold, having more than five members in the household (p = 0.025) increased the probability of incurring catastrophic health expenditure. However, there was no significant relationship between having more than five members in the household (p > 0.05) and the risk of incurring catastrophic health expenditure at 25 and 40% threshold. Results also revealed that at 10 and 40% threshold, households living in an urban area (p < 0.05) increased the probability of incurring catastrophic health expenditure. However, there was no significant relationship between households living in urban areas (p = 0.156) and the risk of incurring catastrophic health expenditure at 25% threshold. Irrespective of the threshold, results showed that households living in the north central zone (p = 0.000), households living in the north eastern zone (p = 0.000), and households living in the south southern zone (p = 0.000) increased the risk of catastrophic health expenditure. At 40% threshold, there was a significant relationship between households living in the south eastern zone (p = 0.004) and the risk of incurring catastrophic health expenditure. However, there was no significant relationship between households living in the north western zone (p > 0.05) and the risk of incurring catastrophic health expenditure irrespective of the threshold. Also, there was no significant relationship between households living in the south eastern zone (p > 0.05) and the risk of incurring catastrophic health expenditure at 10 and 25% threshold. Irrespective of the threshold, being unemployed (p = 0.000) was significantly associated with catastrophic health expenditure. Results also revealed that irrespective of the threshold, lack of health insurance coverage (p = 0.000) increased the risk of incurring catastrophic health expenditure. Households visiting a private health facility (p = 0.000) was significantly associated with catastrophic health expenditure at 10, 25, and 40% threshold. Irrespective of the threshold, having household members with non-chronic illness (p = 0.000) also increased the probability of incurring catastrophic health expenditure.

Table 2 Determinants of catastrophic health expenditure using Chi-square test
Full size table
Table 3 Determinants of catastrophic health expenditure using Chi-square test
Full size table
Table 4 Determinants of catastrophic health expenditure using logistic regression model
Full size table
Table 5 Determinants of catastrophic health expenditure using logistic regression model
Full size table

Discussion

The results from our study indicate that some factors are associated with catastrophic health expenditure among households in Nigeria. At 10% threshold of total household expenditure, household socio-economic status was not strongly associated with catastrophic health expenditure. This is in contrast to findings from studies in India [28] and Kenya [34] where socio-economic status increases the risk of catastrophic health payment. The level of education of household heads, such as having no education, having primary education, and having secondary education, was significantly associated with catastrophic health expenditure in the study. A similar study also in India [28] supports this finding. Type of health facility visited was positively associated with catastrophic health expenditure. This is in line with findings from a study in Kenya [34] where type of health facility visited increases the likelihood of experiencing catastrophic health expenditure. At 40% threshold of non-food expenditure, our study revealed that gender of household heads was not associated with catastrophic health expenditure. This is in line with a study in Burkina Faso [7] where gender of household heads was not a significant determinant of catastrophic health expenditure but in contrast to similar studies in China [35] and Iran [31] where gender of household heads was a significant determinant of catastrophic health expenditure. Household socio-economic status such as being poorer, being in the middle, and being richer was also an important determinant of catastrophic health expenditure. Similar studies in India [28, 36], Mexico [37], China [35], Iran [31], Thailand [32], Turkey [27], Georgia [24], Vietnam [38], and Burkina Faso [7] support this finding. Having a member above 65 years was not significantly associated with catastrophic health expenditure in Nigeria. This is in contrast to studies in Mexico [37], China [35], Iran [31], Thailand [32], Turkey [27], and Vietnam [38]. Our study also revealed that type of illness suffered such as suffering non-chronic illness was strongly associated with the risk of facing catastrophic health expenditure. This could be due to the high burden of communicable diseases in Nigeria. Studies in Thailand [32], Kenya [19], and India [36] support this finding. Type of health facility visited such as visiting private health facility was also a significant determinant of catastrophic health expenditure. This could be due to the high cost of health care in a private health facility compared to a public health facility. However, this is in line with studies in Thailand [32], Kenya [19], and India [36], where type of health facility visited increases the risk of experiencing catastrophic health expenditure. Geo-political zone, for example households living in the north central zone, households living in the north eastern zone, households living in the south eastern zone, and households living in the south southern zone was associated with catastrophic health expenditure. This could be due to the variation in the financing, delivery, and provision of health care services across the six geo-political zones. However, similar studies in India [28], Georgia [24], Kenya [19], and Nigeria [43] support this finding. Level of education of the household head such as having no education, having primary education, and having secondary education was an important determinant of catastrophic health expenditure. This is in line with studies in India [28] and China [35] where level of education of household heads increases the risk of catastrophic health expenditure but in contrast to studies in Kenya [19], Turkey [27], and Burkina Faso [7], where the level of education of household heads was statistically insignificant. Location, such as living in an urban area, increased the probability of incurring catastrophic health expenditure. This could be due to the high cost of health care services in urban areas compared to rural areas. Also, urban areas have many private hospitals that charge high out-of-pocket payments. Households in the rural areas may not seek health care due to the inability to pay for health care services. Similar studies in India [28, 36], Mexico [37], Botswana [21], China [35], and Turkey [27] support this finding. In contrast, a study in Burkina Faso [7] found that location was not a significant determinant of catastrophic health expenditure. Our study revealed that health insurance status such as lack of any health insurance was a significant determinant of catastrophic health expenditure. This could be due to the poor coverage of health insurance in Nigeria, which is less than 10%. However, this is in line with studies in China [35], Iran [31], and Turkey [27], where lack of health insurance increases the probability of incurring catastrophic health expenditure. Our study also showed that household size, such as having more than five members, was not associated with catastrophic health expenditure. This is in contrast to studies in Mexico [37], Lesotho [21], China [35], Iran [31], India [36], Vietnam [38], and Burkina Faso [7], where household size was significantly associated with catastrophic health expenditure.

There is an agreement that UHC will only be achieved by public investment and in particular health insurance. Policy-makers need to expand social health insurance, scale-up community-based health insurance (CBHI), and provide social health protection as a priority for the poor and most vulnerable population. Our findings have implications for health system financing reforms in all resource-poor countries. Health insurance systems should be tailored for the benefit of the poor and most vulnerable population through the adoption of a tax financed non-contributory UHC scheme and a government-run CBHI program for rural dwellers.

Sensitivity of results

For the concept of total household expenditure, variables such as having more than five members, households living in the south eastern zone, households living in the south southern zone, and household socio-economic status were sensitive to the three thresholds (details are given in Table 4). For the concept of non-food expenditure, variables such as having household members below 5 years, household members aged between 55 and 64 years, having no education, having male headed households, having more than five members, households living in an urban area, households living in a south eastern zone, and household socio-economic status were sensitive to the three thresholds (details are given in Table 5). This implies that the different concepts and thresholds have a significant effect on the determinants of catastrophic health expenditure.

Limitations of the study

As is common with all national household surveys around the world, the estimates of catastrophic health expenditure in our study is affected by the structure of the questionnaire, mode of data collection, recall bias, as well as issues of validity, reliability, and comparability. However, these limitations do not invalidate our work, as the estimates are important in research on health care financing.

Conclusions

Findings from our study show that some household and individual characteristics are associated with catastrophic health expenditure in Nigeria. Many households experience catastrophic health payments due to factors such as age, education of household head, health insurance status, geo-political zone, type of health facilities visited, and type of illness suffered. Governments are yet to find fair and innovative ways of financing the health system so as to reduce the financial burden of out-of-pocket payments on households and individuals in Nigeria. This implies that many households and individuals still experience inequitable access to quality health care services and face financial hardship as a consequence. Policy-makers and political actors need to design equitable health financing policies that will increase financial risk protection for people in both the formal and informal sectors the economy.