The intial databases searches yielded 771 records in which 48 duplicates were removed. A further 571 papers were excluded after screening their titles and abstract (Fig. 1). The full-text of the remaining articles were screened based on the eligibility criteria. This led to 53 papers which were eglible for inclusion. Two studies were later excluded due to potentially erroneous results. One paper reported that 100% of the population was facing impoverishing spending while the other reported an incidence of 50% for catastrophic spending in another country. A total of 51 studies, all published in English were included in the final review (Fig.1).
3.1 Study characteristics
The studies were all published between 2003 and 2019, with the majority (31/50) published in 2015 or later. Over half of the publications (26/50) were authored by individuals affiliated with academic institutions, followed by the World Bank and WHO. Also, five individuals authored one-third of the papers.
Most of the publications concentrated on a single country, but more than half of the studies (27/50) were from four countries (Ghana, Kenya, Nigeria and Rwanda) as described in Table 1. The remainder of the papers were part of multi-country analyses at the global or regional levels (countries within the Africa region). The review contained articles on the majority of the countries (41/47) in the WHO African Region either in a single or multi-country analysis.
3.2 Data sources
Nationally representative household surveys were used for over three-quarters of the publications reviewed. Table 1 shows the full range of surveys used in the various articles, including the Living Standard Measurement Surveys (LSMS), Household Expenditure and Utilization Survey (HEUS), Household Budget Surveys and World Health Surveys. The median survey year was 2005. The remaining studies (12/50) are cross-sectional surveys at the state, district and community levels, which may not be generalizable to other contexts. Some of these papers focused on the impact of national health insurance initiatives in states or districts (n=6) or among specific communities (n=2)
3.3 Analytical methods used by included studies
The studies assessed different aspects of financial protection (Table 2). Most of the studies (40/51) determined the incidence of catastrophic payments using either the budget-share or capacity-to-pay approaches. Some of these publications (9/40) also determined the intensity (i.e. the extent to which households exceed a given threshold using the mean positive overshoot or gap measure). About half of the papers (25/50) assessed the incidence of impoverishment.
The studies which focused on single countries, also conducted an in-depth analysis of financial protection by examining incidence trends over time, equity strafiers and determinants of financial protection. Over half (28/50) of the papers identified the factors that are associated with financial protection indicators using logistic regression as the most common statistical model. Most (37/50) of the studies conducted equity analysis using characteristics such as income quintiles, geographical location (rural vs urban) and gender and used either regression analysis or concentration index for their analysis. A few of the studies (12/50) assessed the trends in financial protection with most of the studies (5/8) in a single country (i.e. Rwanda). All the global multi-country papers assessed trends over time.
3.4 Measuring catastrophic health expenditures
About half (n=24) of the publications focused only on catastrophic expenditures while only four papers exclusively assessed impoverishing out-of-pocket healthcare expenditures. The remaining papers (n=23) assessed both measures of financial protection. More than 50% of the papers used the budget-share approach. Twenty-one studies used the capacity-to-pay approach. Some of the studies used multiple methods to assess catastrophic expenditures due to the lack of consensus in the literature. However, all publications with at least one author from the WHO (n=6) used the capacity-to-pay approach that accounts for subsistence needs, while studies by authors from the World Bank used the budget-share method (n=10) either alone or in combination with the capacity-to-pay approach using food expenditure as a measure of basic subsistence.
The threshold at which out-of-pocket payments are defined as catastrophic varied (5-40%). The majority (n=44) of the studies used a uniform range of thresholds regardless of the method selected. Most of the studies that used the budget-share approach also used multiple thresholds; however, the most common threshold was 10% (24/27). Studies which used the capacity-to-pay approach mostly used a threshold of 40% (n=16/21). A few studies (n=3) used rank-dependent thresholds based on socio-economic status[12-14]. Only one study used both a uniform and variable thresholds[15].
Incidence and intensity of catastrophic health expenditures
Figure 2 shows the results using the share of out-of-pocket payments over total consumption or income (budget-share method) at a threshold of 10% for all the nationally representative household surveys. Wagstaff et al. (2017a) and the WHO and World Bank (2017) used the same methods, data source and years and have the same results. Therefore, only the results for the WHO and the World Bank report are included in this review.
The analysis in Figure 2 covers many countries (n=36) with data ranging from 1993-2014. The incidence ranged from 0.29% in Zambia in 2010 to 16.4% in Nigeria in 2009. The mean incidence across the countries was 8.1% with the median incidence being 7.0% (IQR: 3.4-11.0). For countries with more than one data point, the results showed that the incidence of catastrophic expenditures had increased over time except for Ghana, and Zambia. Of the four upper-middle-income countries which were included in the analysis, only one had an incidence of catastrophic spending that was significantly below the regional incidence average. Low-income countries experienced various levels of catastrophic spending.
A few of the studies assessed the intensity of catastrophic expenditures. Results from five countries which used the budget-share method at a 10% threshold estimate the mean intensity at 16.0%, meaning that, on average, households spend 16% over the 10% catastrophic threshold.
Figure 3 shows the incidence of catastrophic expenditures at the 25% threshold (budget-share), which is lower than the incidence at the 10% threshold. The mean incidence of catastrophic expenditure at the 25% threshold was 2.3%. Generally, the incidence at the 25% threshold has also increased over time in countries with some exceptions. Only two studies in Kenya determine the intensity of catastrophic expenditures at the 25% threshold.
A mean incidence of catastrophic payments was estimated at 3.3% using the 40% threshold for the capacity-to-pay approach based on non-food spending (Additional File 2). Although not directly comparable, the results in Figure 3 generally show that the incidence of catastrophic payment at the 10% threshold using the budget-share approach is higher than the incidence using the capacity-to-pay approach at the 40% threshold.
3.5 Measuring impoverishing health expenditures
About half (25/51) of the publications assessed impoverishing expenditures using various approaches, as shown in Table 2. Majority of the studies assessed impoverishing out-of-pocket payments (13/25) used an international poverty line (mostly the $1.25 per person per day line) to determine the incidence and depth of poverty. However, the recent global multi-country analysis used the $1.90 and $3.10[3, 16] international absolute poverty lines. Nearly all of these papers (12/13) used multiple poverty lines. Two studies used both the national and international poverty lines[12, 14].
Incidence of impoverishing out-of-pocket health payments
The incidence of impoverishment at the $1.90 poverty line in 32 countries in the African Region with all the countries experiencing a certain level of impoverishment with the lowest incidence of impoverishment in Cabo Verde at 0.09% in 2007 and the highest incidence at 9.37% in Guinea in 2012. The mean incidence of impoverishment is 1.81% at the $1.90 and 3.2% at the $3.10 poverty lines.
Trend of financial protection
The global multi-country analyses show that generally, the incidence of catastrophic expenditure has increased in the African Region. For single-country papers, it was found that the incidence of catastrophic spending has decreased over time in Uganda, Rwanda and Ghana and has increased in Mauritius. Studies in Rwanda assessed inequality in catastrophic spending over time and showed that that the gap between the poorest and the least poor has significantly decreased over time[17, 18].
For impoverishing health expenditures, the trend of its incidence depends on the poverty line used. According to the global-multi-country analyses, the incidence of impoverishing health expenditures decreased for the $1.25 poverty line while at the $3.10 poverty line, this incidence increased. Only one study in the single-country showed the trend of impoverishing health expenditure with the incidence of impoverishment decreasing.
Analysis of financial protection by equity stratfiers
For the studies (n=21) which use nationally representative surveys and accounted for income quintiles in catastrophic expenditures, majority of them found that poorer households experienced higher incidences of catastrophic health expenditures. However, exceptions are found in Nigeria and Mauritius where the better-off households had a higher incidence of catastrophic expenditures.[13, 19, 20]. The studies also found that catastrophic health expenditure was concentrated in rural households compared with those in urban areas. Only four papers reported income quintiles for impoverishment expenditures, and there was no clear pattern of concentration of impoverishment among the income quintiles.[20-22] [23]
Due to the few studies, which used concentration indexes to assess whether the poor are more likely than the rich to incur financial catastrophe, it is not possible to determine whether catastrophic spending among the different income groups is sensitive to certain thresholds and methods in the African region.
Determinants of financial hardship
Among the 13 studies reporting the determinants of catastrophic expenditure using national household surveys, most found that household size, use of health facility especially inpatient services, age of household members, and education level of household heads are significantly associated with catastrophic health expenditures. In Nigeria and Rwanda, health insurance was a significant factor associated with catastrophic health spending. While in Kenya there were mixed findings on the significance of health insurance on catastrophic health spending.[24-26]