Investing in health systems for universal health coverage in Africa

Objective 1: Prepare a synthesis on the situation of health systems components

We analysed data on the distribution of the key types of human resources for health, health facilities, health technologies, health financing, information, and leadership and governance. This entailed calculation of frequencies and descriptive statistics, as reported under Synthesis of health system components in the Results and discussion section.

Objective 2: Analyse the correlation between health MDG-related interventions and some health systems components

This objective was addressed through analysing frequency distributions on coverage of public health interventions related to MDG 5 on improving maternal health, MDG 4 on reducing child mortality, MDG 6 on combating HIV/AIDS, malaria and other diseases, and MDG 8 on creating a global partnership for development. For MDG 8 we dealt with only Target 8.E, which reads In cooperation with pharmaceutical companies, provide access to affordable essential drugs in developing countries. We used simple linear regression to estimate correlation coefficients (r) and the t-test of the null hypothesis (H₀) that there was no relationship between pairs of the variables concerned. Details on the analysis are shown under Progress towards universal coverage of MDG-related interventions in the Results and discussion section.

Objective 3: Provide overview of four major thrusts towards universal health coverage

To address the gaps in providing access to interventions related to health MDGs, we interrogated the current issues and challenges to draw up the four thrusts that we propose as essential actions for the countries to advance towards UHC. The argument for these thrusts is presented under Major thrusts towards universal health coverage in the Results and discussion section.

Human resources for health

In 2013, globally there were estimated 8 652 107 physicians, 16 689 250 nursing and midwifery personnel, 1 227 822 dentistry personnel, 2 114 282 pharmaceutical personnel, and 169 602 psychiatrists [35]. In spite of the fact that the African Region bears about 25% of the global burden of disease, it had a paltry 1.4% of the physicians, 2.8% of the nursing and midwifery personnel, 1.4% of the dentistry personnel, 1.6% of the pharmaceutical personnel, and 0.7% of the psychiatrists available globally in 2013.

It is estimated that the African Region has a shortage of at least 817 992 health workers [41]. According to the World Health Report 2006 [41], out of the 57 countries identified as facing a human resources for health (HRH) crisis with health workforce density ratios below 2.3 per 1000 population, 36 were in the African Region. Health workforce shortages in the Region have been attributed to inadequate institutional capacity for HRH management, low levels of national investment in HRH production, slow progress in educational reforms, skewed distribution of health workers, lack of incentives, and ineffective retention strategies.

Health facilities

The densities of primary, secondary and tertiary level health facilities in the African Region were lower than those recommended by WHO. For instance, 56% (15/27) of the countries had fewer than 10 health posts per 100 000 population, 85% (28/33) had fewer than 10 health centres per 100 000 population, 82% (28/34) had fewer than one district hospital per 100 000 population, and 100% (27) had fewer than one provincial hospital per 100 000 population. Worse still, 41% of the countries had fewer than 0.1 provincial hospitals and 66% (21/32) fewer than 0.1 tertiary hospitals per 100 000 population [36].

The low densities of primary and secondary level health facilities might partially explain why many countries had low coverage of essential public health interventions and curative services for case management of communicable and non-communicable diseases. Furthermore, the very low densities of tertiary hospitals may undermine African countries capacity to provide specialised referral services, produce physicians and other specialised health workforce cadres, conduct basic and applied research, and utilise available medical innovations.

Health technologies

Availability of essential medicines in public health facilities varied widely, ranging from 17.9% to 87.1%, and in private health facilities from 13.6% to 88% [35]. WHO estimates that in Africa about half of those in need of essential medicines do not have access to them [42], mainly because of cost reasons. The median consumer price ratio (MPR) is the ratio of the local price to an international reference price converted into the local currency. An MPR of 1 means that the local price is equivalent to the reference price, and an MPR of 2 that the local price is twice the reference price [35]. The MPR among public health facilities varied from 1.3 to 6.5, meaning that in some cases the local price was almost seven times the international reference price [35]. The MPR among private health facilities ranged from 2.2 to 15.1. In 53% of the countries, the local prices of selected generic medicines in the private health sector were more than four times the international reference prices. MPRs in the private health sector were remarkably higher than those in the public health sector [35]. The main challenges related to essential medicines are associated with their limited access; the circulation of fake, substandard and counterfeit medicines; weak regulatory capacity; non-rational use of medicines; and weak research and innovation capacity to address priority needs [43].

Health financing

The data used for this section were for 2011. On average in the African Region general government expenditure on health constituted 48.7% of total expenditure on health, compared with the global level of 57.8%. G overnment expenditure on health as a percentage of total government expenditure was 9.7%, compared with 15.2% globally. By the end of 2011, only six of the 45 countries reporting on this factor had met the Abuja target of allocating at least 15% of total government expenditure to the health sector.

The private expenditure on health constituted 51.7% of total expenditure on health compared with 41.1% globally. About 56.5% of the private expenditure on health was from household out-of-pocket spending, compared with the global level of 48.8%, and 31.3% was from private prepaid plans, compared with 37.2% globally. The fact that out-of-pocket payments form over 50% of private spending on health in 38 (83%) of the countries is of great concern given that such a high level of spending on health exposes households to the risk of financial catastrophe and impoverishment. In fact, there is evidence that where out-of-pocket spending is less than 20% of total health expenditure, the risk of catastrophic health expenditure is negligible. Unfortunately, in 2010, out-of-pocket payments as a share of total health expenditure were above 20% in 35 countries, meaning that significant sizes of the population in those countries were exposed to the risk of financial catastrophe and impoverishment.

External resources for health made up less than 20% of total health expenditure in 21 (46%) of the countries, 20-40% in 15 countries, and over 40% in 10 countries. The average donor contribution to total health expenditure was 11.8%, compared with 0.4% globally. For 22 countries (45.7%) more than 25% of the expenditure on health was from external resources.

Per capita government expenditure on health was less than US$ 10 in seven countries, US$ 10 to US$ 30 in 21 countries and over US$ 30 in 18 countries. The range was from US$ 6 to US$ 570. The African Regions average per capita government expenditure on health was US$ 49, compared with US$ 613 globally. Some 46% of the countries had a per capita government expenditure on health of less than US$ 20.

Information

Mbondji and colleagues [20] questionnaire-based survey found that 36% (16/44) of the countries in sub-Saharan Africa had a functional national health research governance mechanism, 20% (9/44) had clear terms of reference and 20% had a functional national health research management forum. Kebede and colleagues [44] questionnaire survey of 847 health research institutions revealed that 49% of these institutions had computer laboratories, 50% had network computers, 38% had information technology support, and 67% had a library.

The descriptive analysis of data sources undertaken by Mbondji and colleagues [18] revealed that in almost all countries in the Region there was heavy reliance on household surveys for indicators. Few countries had civil registration systems that permitted adequate and regular tracking of mortality and cause of death data. The health management information systems generated considerable data but these were used rarely because of concerns about bias, quality and timeliness. About 93% (43/46) of the countries in the Region had introduced an integrated disease surveillance and response system.

In Mbondji and colleagues [20] questionnaire survey, 75% (33/44) of the countries reported to have functional ethical review committees. About 57% of the 847 health research institutions surveyed by Zielinski and colleagues [21] responded to the survey, with 51% of these reporting to have policies on research ethics, 58% to have written policies requiring researchers to obtain informed consent of research participants, 34% to have established ethics review committees, 42% to require that studies go through ethics review committees, 46% to have linkages with national or regional ethics organisations, 53% to have adopted standard operating procedures, and 36% to provide some type of ethics training for staff [36].

Leadership and governance

Although no readily available indicators exist for the leadership and governance component of health systems, it is a crucial element for ensuring that appropriate strategic policy frameworks exist for the pursuit of the UHC vision and are combined with effective and accountable building of partnerships. In this study we used three indicators for this component: existence of a valid national health policy and strategic plan. About 87% (41/47) of the countries had a valid national health policy and strategic plan [37].

MDG 5: Maternal health

We tested the null hypothesis that no relationship existed between the percentage of births attended by skilled health personnel and number of health facilities, personnel numbers in various health workforce cadres, health spending, and availability of medical devices. We found that the correlation coefficients were significantly different from zero at the 97.5% confidence level for physicians per 10 000 population, nursing and midwifery personnel per 10 000, general government expenditure on health as a percentage of total health expenditure, private expenditure on health as a percentage of total health expenditure, per capita total expenditure on health, and radiotherapy units per a million population. Figure2 shows a positive relationship between percentage of births attended by skilled health personnel and nursing and midwifery personnel per 10 000 population. The correlation coefficient (r) was 0.509946 (r² = 0.260045), computed t-statistic with 31 degrees of freedom was 3.300676, two-sided P was 0.0024 and critical t-statistic from the statistical tables was 1.960 at the 97.5% confidence level. Since the computed t-statistic was higher than the critical t value, we rejected the null hypothesis (H₀) that there is no relationship between the two variables. We concluded that the correlation between the percentage of births attended by skilled health personnel and nursing and midwifery personnel per 10 000 population was significantly different from zero at the 97.5% confidence level.

Figure3 depicts the relationship between percentage of births attended by skilled health personnel and per capita total expenditure on health. We tested the null hypothesis that there was no relationship between these two variables. The estimated correlation coefficient (r) was 0.51424 (r² = 0.264442), computed t-statistic at 42 degrees of freedom was 3.885813, two-sided P was 0.0004 and critical t-statistic was 1.960 at the 97.5% confidence level. Since the computed t-statistic was greater than the critical t-statistic, we rejected the null hypothesis and concluded that the correlation coefficient was significantly different from zero. There was a positive relationship between the percentage of births attended by skilled health personnel and per capita total expenditure on health.

The correlation coefficient for the relationship between the percentage of births attended by skilled health personnel and general government expenditure on health as a percentage of total health expenditure was positive and significantly different from zero. On the other hand, the correlation coefficient for private expenditure on health as a percentage of total health expenditure was significantly different from zero but negatively related to the percentage of births attended by skilled health personnel.

Unmet need for family planning services in the African Region stands at 25%, compared with the global average of 12%. The levels varied widely among the countries, ranging from 6% to 38% [35]. Figure4 depicts the relationship between unmet need for family planning services and the number of district hospitals per 100 000 population. A test was conducted on the null hypothesis that there was no relationship between these two variables, i.e. the correlation coefficient was not significantly different from zero. The estimated correlation coefficient (r) was −0.539349 (r² = 0.290898), computed t-statistic with 21 degrees of freedom was −2.935117, two-sided P was 0.0079 and critical t-statistic from statistical tables was 1.721 at the 95% confidence level. Since the computed t-statistic was greater than the critical t-statistic, we rejected the null hypothesis and concluded that the correlation coefficient was significantly different from zero; that is, there was a negative relationship between unmet need for family planning services and the number of district hospitals per 100 000 population. The coefficients for other levels of health facilities, various categories of human resources, per capita total expenditure on health, and medical devices were not significantly different from zero.

The Regions contraceptive prevalence rate of 27% was much lower than the global average of 63%. Contraceptive prevalence among the countries varied from 4% to 65% [35]. Figure5 shows the relationship between contraceptive prevalence rate and nursing and midwifery personnel per 10 000 population. We tested the null hypothesis that there was no relationship between these two variables and, thus, that the correlation coefficient was not significantly different from zero. The estimated correlation coefficient (r) was 0.589924 (r² = 0.34801), computed t-statistic at 30 degrees of freedom was 4.001619, two-sided P was 0.0004 and critical t-statistic was 1.960 at the 97.5% confidence level. Since the computed t-statistic was greater than the critical t-statistic, we rejected the null hypothesis and concluded that the correlation coefficient was significantly different from zero. There was a positive relationship between contraceptive prevalence rate and the number of nursing and midwifery personnel per 10 000 population. The correlation coefficient for the number of physicians per 10 000 population was positive and significantly different from zero, signifying the existence of a relationship between this variable and contraceptive prevalence rate.

Figure6 shows the relationship between contraceptive prevalence rate and per capita total expenditure on health. We tested the null hypothesis that there was no relationship between these two variables and, thus, the correlation coefficient was not significantly different from zero. The estimated correlation coefficient (r) was 0.578638 (r² = 0.334822), computed t-statistic at 37 degrees of freedom was 4.315581, two-sided P was 0.0001 and critical t-statistic was 1.960 at the 97.5% confidence level. Since the computed t-statistic was greater than the critical t-statistic, we rejected the null hypothesis and concluded that the correlation coefficient was significantly different from zero. There was a positive relationship between contraceptive prevalence rate and per capita total expenditure on health.

MDG 4: Routine immunisation coverage

As portrayed in Table5, African Regions average measles immunisation coverage among 1-year-olds was 73%, compared with the global average of 84% [35]. Measles immunisation coverage varied among the countries, ranging from 42% to 99%. Figure7 shows the relationship between measles immunisation coverage among 1-year-olds and general government expenditure on health as a percentage of total health expenditure. We tested the null hypothesis that there was no relationship between these two variables. The estimated correlation coefficient (r) was 0.376021 (r² = 0.141392), computed t-statistic at 44 degrees of freedom was 2.691791, two-sided P was 0.01 and critical t-statistic was 1.960 at the 97.5% confidence level. Since the computed t-statistic was greater than the critical t-statistic from the statistical tables, we rejected the null hypothesis at the 95% confidence level. We concluded that the correlation coefficient was significantly different from zero, meaning that there was a positive relationship between measles immunisation coverage among 1-year-olds and general government expenditure on health as a percentage of total health expenditure. This makes even more sense when we recall that in national health accounts estimates general government expenditure on health also includes funding from partners [45].

Interventions	African region average (A)	Global average (B)	Gap [C = B/A)]
Measles (MCV) immunization coverage among 1-year-olds (%)	73	84	1.2-fold
BCG immunization coverage among 1-year-olds (%)	80	89	1.1-fold
DPT3 immunization coverage among 1-year-olds (%)	72	83	1.2-fold
Hib3 immunization coverage among 1-year-olds (%)	65	45	0.7-fold
HepB3 immunization coverage among 1-year-olds (%)	72	79	1.1-fold

Sources: WHO [35].

We also tested the null hypothesis that there was no relationship between measles immunisation coverage among 1-year-olds and private expenditure on health as a percentage of total health expenditure. The estimated correlation coefficient (r) was 0.376021 (r² = 0.141392), computed t-statistic at 44 degrees of freedom was 2.691791 and two-sided P equalled 0.01. As the computed t-statistic of 2.691791 was greater than the critical t value of 1.960 from the statistical tables, we rejected the null hypothesis at the 97.5% confidence level. We concluded that the correlation coefficient was significantly different from zero, meaning that there was an inverse relationship between measles immunisation coverage among 1-year-olds and private expenditure on health as a percentage of total health expenditure. This implies that private health spending, which largely constitutes of direct household out-of-pocket payments in the Region, adversely affects access to immunization services, and hence, immunization coverage.

Diphtheria tetanus toxoid and pertussis (DTP3) vaccine coverage among 1-year-olds for the African Region was 72%, compared with the global average of 83%. Coverage among the countries varied from 33% to 99% [35]. HepB3 immunisation coverage among 1-year-olds was 72%, compared with the global average of 79%. Coverage among the countries ranged from 41% to 99% [35]. The regional average of 65% for haemophilus influenzae type B vaccine (Hib3) coverage among 1-year-olds was higher than the global average of 45%. Coverage varied widely among the countries, ranging from 10% to 99% [35].

MDG 6: Coverage of health interventions for combating HIV/AIDS, malaria and tuberculosis

In 2010, in the African Region 35% of the male and 29% of the female population aged 1524 years had comprehensive, correct knowledge on HIV/AIDS; that is, they could correctly identify the ways of preventing sexual transmission of HIV and reject misconceptions about HIV transmission. The ranges among the countries for this variable were 16% to 55% for the male population and 8% to 59% for the female population [35].

Figure8 depicts the relationship between the percentage of female population aged 1524 years without comprehensive, correct knowledge on HIV/AIDS and general government expenditure on health as a percentage of total health expenditure. We tested the null hypothesis that there was no relationship between these two variables. The estimated correlation coefficient (r) was −0.380545 (r² = 0.144815), computed t-statistic at 30 degrees of freedom equalled −2.253912, two-sided P was 0.0317 and critical t-statistic was 1.960 at the 97.5% level of significance. Since the computed t-statistic was greater than the critical t-statistic from the statistical tables, we concluded that the correlation coefficient was significantly different from zero. That means that there was a relationship between the percentage of the female population aged 1524 years without comprehensive, correct knowledge on HIV/AIDS and general government expenditure on health as a percentage of total health expenditure. This means that as government spending on health increases, health promotion (including health education) programmes would be expected to receive more funding, and hence health illiteracy would be expected to diminish. However, the correlation coefficients for various densities of health facilities, numbers in various health workforce cadres and medical devices were not significantly different from zero.

The African Regions 2012 average antiretroviral therapy coverage among people eligible for that treatment was 63%, compared with the global average of 61% [35]. For the 41 countries reporting on this, antiretroviral therapy coverage varied widely, ranging from 1% to 95% [35]. We tested the null hypothesis that there was no relationship between antiretroviral therapy coverage and densities of the various levels of health facilities and health workforce cadres, health spending level and medical devices. In all cases the correlation coefficient was not significantly different from zero. Figure9 serves as an example and shows the relationship between antiretroviral therapy coverage among people eligible for that treatment and the level of external resources for health as a percentage of total expenditure on health. The null hypothesis was that there was no relationship between the two variables. The estimated correlation coefficient (r) was 0.29535 (r² = 0.087231), computed t-statistic at 40 degrees of freedom was 1.431704, two-sided P was 0.16 and critical t-statistic from the statistical tables was 1.645 at the 95% level of confidence. Since the computed t-statistic was lower than the critical t-statistic we concluded that the correlation coefficient was not significantly different from zero.

In the African Region in 2012, 64% of pregnant women with HIV received antiretrovirals for prevention of mother-to-child transmission of HIV (PMTCT), compared with the global average of 62%. Coverage of antiretrovirals for PMTCT varied widely among the countries, ranging from 13% to over 95% [35].

Coverage of health interventions for reversing malaria burden

As depicted in Table6, on average 25% of children aged under 5years were sleeping under insecticide treated nets in the African Region. Among the 29 countries reporting on this item, coverage varied from 1% to 91% [35]. Figure10 depicts the relationship between percentage of children under 5years sleeping under insecticide treated nets and general government expenditure on health as a percentage of total government expenditure. We tested the null hypothesis that the correlation coefficient was not significantly different from zero. The estimated correlation coefficient (r) was 0.509 (r² = 0.259081), computed t-statistic at 27 degrees of freedom was 3.07266, two-sided P was 0.0048 and critical t-statistic was 2.052 at the 97.5% confidence level. Since the computed t-statistic was greater than the critical t-statistic, we rejected the null hypothesis and concluded that the correlation coefficient was significantly different from zero. This means that there was a positive relationship between the percentage of children under 5years sleeping under insecticide treated nets and general government expenditure on health as a percentage of total government expenditure.

Interventions	African region average (A)	Global average (B)	Gap [C = B/A)]
Children aged <5years sleeping under insecticide-treated nets (%)	25	30 (in low income countries)	1.2-fold
Children aged <5years with fever who received treatment with any antimalarial (%)	-	-	-
Antiretroviral therapy coverage among people eligible for treatment (%)	63	61	0.96-fold
Pregnant women with HIV receiving antiretrovirals to prevent mother-to-child-transmission (MTCT)	64	62	0.97-fold
Case-detection rate for all forms of tuberculosis (%)	59	67	1.1-fold
Treatment-success rate for smear-positive tuberculosis in 2011 (%)	82	87	1.1-fold

Source: WHO [35].

Among the 36 countries reporting on coverage of antimalarial treatment for fever, the levels ranged from 2% to 65% [35]. Figure11 shows the relationship between percentage of children aged under 5years with fever receiving antimalarial treatment and number of health posts per 100 000 population. We tested the null hypothesis that there was no relationship between the two variables. The estimated correlation coefficient (r) was 0.458264 (r² = 0.210006), computed t-statistic at 22 degrees of freedom was 2.418329 and two-sided P equalled 0.0243. Since the computed t-statistic was greater than the critical t-statistic value of 2.074 from the statistical tables, we rejected the null hypothesis at the 97.5% confidence level. We concluded that the correlation coefficient was significantly different from zero, implying that there was a positive relationship between the percentage of children aged under 5years with fever receiving antimalarial treatment and number of health posts per 100 000 population. The correlation coefficients for the densities of health facilities in the other levels of the system, i.e. district, provincial and tertiary hospitals, were not significantly different from zero.

Coverage of health interventions for tuberculosis control

In 2012, the African Region had the lowest average case detection rate for all forms of tuberculosis, at 59%, compared with the global average of 67%. This rate varied widely among the countries, ranging from 20% to 83% [35]. We found no relationship between case detection rate for all forms of tuberculosis and densities of various levels of health facilities, cadres of the health workforce, medical devices or spending on health. Figure12 shows the relationship between case detection rate for all forms of tuberculosis and per capita total expenditure on health. The estimated correlation coefficient (r) was 0.188845 (r² = 0.035663), computed t-statistic at 43 degrees of freedom was 1.261032, two-sided P was 0.2141 and critical t-statistic was 1.645 at the 95% confidence level. Since the computed t-statistic was lower than the critical t-statistic, we accepted the null hypothesis that the correlation coefficient was not significantly different from zero.

In 2011, the African Region had a success rate of 82% for smear-positive tuberculosis treatment compared with the global average of 87%. There was wide variation in this rate among the countries, ranging from 25% to 92% [35]. Figure13 portrays the relationship between the smear positive tuberculosis treatment success rate and general government expenditure on health as a percentage of total government expenditure. We tested the null hypothesis that the correlation coefficient was not significantly different from zero. The estimated correlation coefficient (r) was 0.338786 (r² = 0.114776), computed t-statistic at 43 degrees of freedom equalled 2.361204, two-sided P was 0.0228 and critical t-static from the statistical tables was 1.960 at the 97.5% confidence level. The computed t-statistic was greater than the critical t-statistic, so we rejected the null hypothesis and concluded that the correlation coefficient was significantly different from zero.

Strengthening public health infrastructure capacity

We define health infrastructure as the essential resources that are required for the delivery of health care. Infrastructure encompasses human resources, technologies and facilities [47].

The road map for expanding human resources for health for improved health service delivery in the African Region during 20122025 [48],[49] proposes six thrusts to countries: strengthening leadership and governance; strengthening regulatory capacity; scaling up education and training; optimising the deployment, retention and performance of available personnel; improving the generation of information to support evidence-based decision-making; and strengthening partnership and dialogue. In order for health workers to apply their knowledge and skills to solve specific health problems they need essential health technologies developed for that purpose. Acquisition of health technology should take cognisance of the available related infrastructure components. Each country ought to have a national health technology policy as an integral part of the overall national health policy, legislation or plan [50]-[52]. Effective implementation of that policy presupposes existence of skilled staff involved in selection, planning, procurement, management, maintenance and utilisation of health technologies.

Health posts and centres and district hospitals provide health services in a health district with a population of 100 000 and 300 000. Walley and colleagues [32] posit that for district health systems to work effectively, countries need to adopt four strategies: definition of the roles of health facilities, integration of health care, building and strengthening of the referral system, and decentralisation of management. District health systems form the operational level of primary care, and that role should be articulated. The relationship of district health systems with health facilities at secondary and tertiary levels, as well as the role of hospitals and public health programmes within the national health services, needs to be clearly defined.

Raising sufficient resources to strengthen health systems

We believe it is possible to increase domestic resources through improvement of the efficiency of tax revenue collection; prioritisation of government budgets to meet the 15% requirement defined by the Abuja commitment [46]; innovative financing, such as taxing tobacco, alcohol, air tickets, foreign exchange transactions, mobile phone use and diaspora remittances [53]; and leveraging the private sector [54]. Predictable, aligned, harmonised and increased development assistance for health is critically necessary, especially for low income countries [55],[56].

Promoting efficiency in national health services to optimise resource use and maximise results

Efficiency is essential in allocation of health resources and provision of national health services to optimise resource use and maximise results. The World Health Report 2010 on health systems financing estimated that about 20-40% of the resources spent on health are wasted [46]. Such resources could have been directed towards the pursuit of universal health care coverage. A number of tools exist for addressing health resource gaps in the African Region, including the Ouagadougou Declaration on Primary Health Care and Health Systems in Africa [10] and its implementation framework [11], the Algiers Declaration on Research for Health [57] and its implementation framework [58], the Addis Ababa Declaration on Community Health [59] and the Libreville Declaration on Health and Environment [60].

Removing financial risks and barriers to care and service access

Compulsory prepayment into a fund for health services before their need arises is recommended as a way of removing financial risks and barriers to access of health services. Where multiple funds exist, there will be need for cross-subsidisation. Prepayment arrangements can be organised through general taxation or compulsory contributions for health insurance or both. Since approximately 51.5% of the African Regions population lives on less than Int$ 1 per day [35], the poor will need assistance through either direct access to government-financed health services or subsidies for their health insurance premiums.

Since 2005, we have seen an increase in the number of African countries pursuing the goal of providing universal health coverage, including Benin, Burkina Faso, Congo, Democratic Republic of Congo, Ghana, Gabon, Lesotho, Kenya, Nigeria, Rwanda, South Africa, Swaziland, Uganda and Zambia. It is critically important that Africa draw lessons from countries around the world that have made significant strides in providing universal health coverage through taxation or social health insurance or a mix of the two.

The literature points to a number of factors that have facilitated countries to progress towards universal health coverage: a high per capita income that strengthened the capacity of businesses and citizens to prepay for health care; a large formal sector that enhanced the capacity to contribute and boosted collection of contributions; urbanisation and communication that facilitated the delivery of services; a skilled labour force, including in health services, to deliver the health care benefit packages and manage health financing schemes; solidarity of the society that facilitated cross-subsidisation from the rich to the poor and from the healthy to the ill; strong government stewardship capacity to launch, guide and sustain a process of compulsory prepayment; and availability of public and private health services of acceptable quality [6].

The majority of African countries lack an enabling environment and are characterised by relatively low per capita gross domestic product; large informal agricultural and small commerce/business sectors that are difficult to tax; limited volumes of taxable imports; insufficiently skilled labour force, including a crisis in human resources for health; and structural and institutional weaknesses that might compromise the efficiency of compulsory prepayment policies and mechanisms.

In spite of these challenges, several countries have introduced social health insurance schemes for their populations. So far, the two well-documented success stories are Ghana and Rwanda, which are implementing a mixed financing health insurance system, with Ghanas scheme being fundamentally tax based [61] while Rwandas is essentially community based [62].