Service readiness of health facilities in Bangladesh, Haiti, Kenya, Malawi, Namibia, Nepal, Rwanda, Senegal, Uganda and the United Republic of Tanzania

1In adopting the sustainable development goals (SDGs) in September 2015, global governments and multilateral organizations endorsed universal health coverage as both a critical element of sustainable development and a prerequisite for achieving equity in global health. If the goal of universal coverage is to be achieved, we will need health-system financing that is adequate and sustainable and health services that are accessible and effective. While research from multiple fields has provided insights on health-system design and the utilization of care, there has been relatively little investigation of the capacity of health facilities to provide essential services.

2Service readiness – a subset of the structural quality of care in the Donabedian triad of structure, process and outcome – is a prerequisite to the delivery of quality health care. During the era of the Millennium Development Goals, the capacity of health facilities to provide care and the quality of the care delivered received far less scrutiny than access to care alone. The resultant deficit must be addressed if the SDGs are to be achieved. Although there have been many area-specific studies on the capacity of health facilities to provide disease-specific services,– there have only been a few comparative and multicountry studies of general service readiness., Between 2003 and 2013, recognizing the need for a comparable metric of health facility capacity, the World Health Organization (WHO) developed a general service readiness index. In 2013, this index was updated to cover 50 items: seven basic amenities such as water and power, six types of basic equipment such as stethoscopes, nine infection prevention measures such as the availability of gloves, eight types of diagnostic test and 20 essential medications – including drugs for infectious as well as noncommunicable diseases (Table 1). In representing the percentage of these items that are readily available in health facilities, the service readiness index is intended to capture overall capacity to provide the general health services that should be available in all facilities – whether they be at primary, secondary or tertiary level. It does not include indicators for hospital-specific readiness.

3The general service readiness index is increasingly being used in subnational or national assessments, and has also been adapted for disease-specific studies. There appears to have been only one published analysis of the service readiness index across multiple countries: a comparison of the readiness of health facilities in six countries, which identified major gaps. Since 2013, when the latter comparison was published, the content of the service readiness index has been updated to incorporate basic capacity to address noncommunicable diseases and nationally representative assessments of health systems have been conducted in multiple countries.

4In analyses of facility readiness, the use of a generalizable metric such as the service readiness index is important for several reasons. For example, such a metric can indicate the capacity of facilities to provide essential care, including their ability to respond to traditional and emerging health challenges. Use of such a metric enables the identification of within-country differences that may suggest inequities in resource distribution and can provide a basis for comparing the efficiency of health systems in translating financial inputs into readiness to meet population health needs. Theorists interested in health-system reform have classified the causal determinants of health-system performance into five main areas: behaviour, financing, organization, payment and regulation. Changes in any of these areas should affect the accessibility, efficiency and quality of health care and, ultimately, population health. In low- and middle-income countries, positive – though variable – associations between total health expenditure and population health have been observed. In addition, external donor assistance for health has been linked with reduced mortality, whereas greater reliance on individual out-of-pocket expenditure has been associated with lower health coverage. Despite these observations, relatively little is known about the relationship between investigated determinants of health-system performance and the capacity of health systems to deliver care.

5The objective of the study was to evaluate the service readiness of health facilities in Bangladesh, Haiti, Kenya, Malawi, Namibia, Nepal, Rwanda, Senegal, Uganda and the United Republic of Tanzania.

6The study sample comprised the health facilities that had been included in service provision assessments by the Demographic and Health Survey (DHS) programme between 2007 and 2015. Assessment data were available in January 2017 for 10 low- and middle-income countries: Bangladesh, Haiti, Kenya, Malawi, Namibia, Nepal, Rwanda, Senegal, Uganda and the United Republic of Tanzania.

7The data from Kenya, Nepal, Senegal, Uganda and the United Republic of Tanzania that we analysed came from samples that were considered to be nationally representative of the countries’ health systems. The data from Haiti, Malawi, and Namibia came from service provision assessments that covered all of the health facilities in the countries; the data from Rwanda covered nearly all health facilities, with a partial sample of private facilities with fewer than five staff. The data from Bangladesh came from a sample that was considered nationally representative of all public facilities and all private hospitals, but excluded small private facilities. Although each assessment was based on a questionnaire for the facility manager, the manager’s responses were verified, whenever possible, by direct observation of the available infrastructure and supplies.

8The core of each DHS service provision assessment was a facility audit – i.e. a standardized assessment of the readiness of each surveyed facility to provide essential health services. The audit that was employed varied slightly between countries and was substantially revised, in 2012, to reflect a broader health system focus and to include health services for noncommunicable diseases. We extracted variables from the DHS service provision assessments to match as many as possible of the 50 items used in the formal assessment of the WHO service readiness index. Depending on the country involved, we calculated an index using between 37 and 49 of the 50 items (Table 1). In calculating each value of the index, we followed the WHO definition – i.e. we determined a percentage score for the items assessed in each of five domains and then calculated the mean of the resultant five scores to give an overall service readiness index for each facility.

9We assessed levels of association between the index and modifiable inputs to health-system performance in the domains of financing, organization and payment. The available data were, however, insufficient for us to identify potential determinants in the areas of behaviour and regulation.

10The available comparable national data on health-system financing included indicators of total health expenditure per capita, resources for health provided from sources external to the country – as a percentage of total health expenditure – and out-of-pocket expenditures on health – again as a percentage of total health expenditure.

11The available comparable facility data on the organization of the health system were whether the facility was a hospital or a health centre, clinic or dispensary and whether it was publicly or privately managed.

12Data on payment included the methods used to transfer money to health-care providers, such as user fees and donor funding to a facility.

13In addition, we investigated the relationship between the service readiness index and the land area of each study country – as a non-modifiable factor that could partially explain health-system readiness. Continuous covariates were natural-log-transformed.

14Our analysis proceeded in three steps: (i) description of service readiness; (ii) comparison of readiness by facility type; and (iii) assessment of the potential determinants of service readiness.

15We conducted descriptive analyses separately for hospitals and all other facilities. Within each study country, we calculated a mean service readiness index for the surveyed hospitals and a corresponding mean value for the surveyed health centres/clinics. We calculated intraclass correlation coefficients to compare the between-country variance in the values of the index with the corresponding within-country variance.

16To compare readiness by facility type, we plotted service readiness for the surveyed private and public facilities – and the surveyed rural and urban facilities – within each country. To test for differences in the service readiness index between facility type, we used linear regression, with country as the stratifying variable.

17To assess potential determinants of service readiness we limited the sample to the nine study countries that were low- or lower-middle-income and excluded Namibia, which is an upper-middle-income country. We used linear regression. We regressed the service readiness index on each of the financing, organizational and payment factors we investigated – as well as on the logarithm of land area. We regressed the service readiness index on each of the four facility-level characteristics (hospital, private, donor funding and user fees) and the four national factors (log THE, log OOP, log external support for health and log land area) in separate models. All models were adjusted for survey year to account for temporal trends and the development of the methods used for the DHS service provision assessments. From these eight linear regression models, we selected all the factors that gave a P-value below 0.20 and used them to create a combined model. Each regression model took the general form as follows:where Y represents service readiness, B0 is the intercept (average service readiness for the reference group of facilities), α represents the survey year, X is the facility-level or national factor of interest and ε is an error term.

18By including interaction terms between hospital and each other factor in a single model, we assessed evidence in the data for differences between the service readiness index’s associations in hospitals and the corresponding associations in the health centres/clinics. We retained the interaction terms that gave P-values below 0.05 for the final model. To provide realistic and easily comparable estimates, coefficients for each continuous factor were transformed to express the estimated difference in the service readiness index for a 10% change in that factor. All regression models were limited to facilities with complete data. Each multicountry analysis was weighted using the sampling weights used in the DHS service provision assessments, but rescaled so that each country contributed equally to the analysis. We calculated confidence intervals (CI) based on standard errors that accounted for clustering by country. Unless indicated otherwise, a P-value of 0.05 or lower was considered indicative of a statistically significant difference.

19Our 10 study countries represent a range of low- and middle-income economies of varying geographical size and development status (Table 2). At the time of the DHS service provision assessments providing our data, mean life expectancies at birth ranged from 54 years in Uganda – in 2007 – to 71 years in Bangladesh – in 2014. In all of our study countries except Bangladesh and Nepal, these mean life expectancies were lower than the global mean values for low- and middle-income countries – which increased from 68 to 70 years between 2007 and 2015. At the time of the DHS service provision assessments, gross domestic products per capita varied from a low of 341 United States dollars (US$) in Malawi to a high of US$ 4134 in Namibia. At the same time, health-system funding varied. Median total annual health expenditure per capita was US$ 43. Some countries were characterized by high out-of-pocket expenditures on health – e.g. Bangladesh, where such expenditure represented 67% of total health expenditure – whereas others showed a heavy reliance on external funds – e.g. Malawi, where such funds represented 69% of total health expenditure.

20In the DHS service provision assessments that provided our data, 8606 (97%) of the 8881 facilities sampled were considered to have been successfully surveyed. Of these, 163 (2%) were outreach facilities assessed with a shortened questionnaire; our analysis was confined to data from the 8443 facilities that each completed the full survey questionnaire (Table 2). At national level, the mean service readiness index for health centres/clinics ranged from 41% in Bangladesh and Uganda to 68% in Namibia, whereas the corresponding values for hospitals ranged from 69% in Senegal to 82% in both Namibia and the United Republic of Tanzania (Table 2). The index varied much more within countries than between countries. The within-country differences were the cause of 71% and 91% of the variation seen in the values for health centres/clinics and hospitals, respectively. Overall, although they still fell short of complete readiness for basic services, hospitals demonstrated higher service readiness than health centres/clinics (mean service readiness index: 77% versus 52%).

21Of the 8443 facilities included in our analysis, 636 (8%) were hospitals. Compared with the health centres/clinics, hospitals were more likely to be privately managed (53% versus 30%), to be in urban areas (73% versus 26%) and to receive financial support from donor organizations (37% versus 20%) and user fees (84% versus 58%) (Table 3). Over 75% (2052 of 2673) of the private facilities charged user fees – in comparison with about half (2957 of 5699) of the public facilities.

22The number of facilities with complete readiness in any one of the five domains investigated was low (Table 3). About a third (191/636) of the hospitals, but only 2% (148/7807) of the health centres/clinics appeared to have full service readiness in terms of basic amenities. Only 199 (2%) and 151 (2%) of the facilities we investigated had all of the diagnostics and essential medications that could be assessed in our calculations of the service readiness index, respectively. Although service readiness appeared generally better in the hospitals than in the smaller facilities, most of the hospitals still appeared inadequate in terms of basic amenities, diagnostic capacity and essential medications. Of the 8443 facilities sampled, just one had a perfect overall service readiness index of 100%. The assessed items that were most likely to be unavailable, even in the higher-performing facilities, were medications and diagnostics for noncommunicable diseases – e.g. haemoglobin tests, inhalers and statins.

23Fig. 1 depicts the medians and interquartile ranges for the values of the service readiness index recorded in public and/or private facilities in each of the 10 study countries and in the full sample. Compared with the median values for the public health centres/clinics, those for private health centres/clinics were significantly lower in Namibia, similar in Rwanda and Senegal and significantly higher in Haiti, Kenya, Malawi, Nepal, Uganda and the United Republic of Tanzania; Bangladesh was excluded from this comparison because small private clinics were not surveyed in that country’s DHS service provision assessment. Compared with the median values for the public hospitals, those for private hospitals were significantly higher in Haiti, Kenya, Malawi, Uganda and the United Republic of Tanzania but significantly lower in Senegal. Fig. 2 illustrates – for the five countries for which the relevant information was available – the differences in the service readiness index between urban and rural facilities. Urban health centres scored significantly higher than rural health centres in Bangladesh, Haiti, Malawi and the United Republic of Tanzania but not in Senegal. Urban hospitals had a similar service readiness index to rural hospitals in Haiti, Malawi and the United Republic of Tanzania, but had a significantly lower index than rural hospitals in Bangladesh and Senegal.

24In the analysis of financial, payment and organizational factors associated with the service readiness index across the nine study countries that were low- or lower-middle-income, the charging of user fees, hospital as facility type, the percentage of total health expenditure represented by external contributions and private management each explained at least 10% of the variance seen in separate models adjusted only for survey year (Table 4). The fully adjusted model explained 34% of the total variance seen in the service readiness index. In the same model, hospitals had a significantly higher service readiness index than health centres/clinics (linear increase: 21.5%; 95% CI: 16.9 to 26.2). If other factors were held constant, each 10% increase in the percentage of total health expenditure represented by external contributions was associated with a difference of 0.7 of a percentage point in the service readiness index (95% CI: −0.3 to 1.7). In hospitals the association of facility funding sources – i.e. user fees and donor support – with the index differed significantly from that calculated for health centres/clinics. In the model with interaction terms for these factors, the estimated service readiness index was 53% for health centres/clinics with user fees and 46% for health centres/clinics without such fees. The corresponding values for hospitals were 73% and 72%, respectively. Donor funding was associated with estimated linear increases in the service readiness index of 3.0 percentage points in health centres/clinics and 4.4 percentage points in hospitals.

25The aim of our study was to improve our understanding of the extent to which service readiness of health facilities varies within and across low- and middle-income countries. This work updates and expands upon the prior comparative assessment of service readinessand highlights opportunities for improving health infrastructure at the start of the era of the SDGs. Our cross-country comparison of structural readiness in over 8000 health facilities revealed substantial and pervasive gaps in the basic capacity to provide health-care services. In general, health centres/clinics achieved barely over half – and hospitals just over three quarters – of the maximum possible score for service readiness. Although the service readiness index defines the resources that WHO hopes to see in any facility providing health-care services, only one assessed facility had all of those resources. In all of the study countries, gaps were particularly notable in single-use items such as medications and diagnostics. Small, multi-use items, such as basic equipment, were more prevalent. These patterns are similar to those seen in a previous, but smaller cross-country comparison of service readiness.

26While mean readiness differed among the countries in the study, the variation between facilities in the same country was much larger. Although all of the study countries appeared to be able to equip some facilities well, all of them were also failing to ensure consistent readiness throughout their health systems. Public health centres/clinics and rural health centres/clinics were particularly low-performing in many countries – but not all. Previous research in individual countries has suggested that, compared with the private sector, the public sector provides care of higher quality in some countries and care of lower quality elsewhere.– Our analyses, using a standard measure and data that are nationally representative, not only led to a similar observation but also indicate that, within any given country, the trend for the public sector to appear better –or worse – than the private can depend on facility type. In identifying the delivery of adequate care in rural areas as being a major challenge – because of major deficiencies in equipment and supplies – our findings support the results of earlier, single-country studies.,

27Compared with health centres/clinics, we found that hospitals appeared to have much better service readiness, especially in terms of basic amenities such as water and electricity supplies. This finding and the similar deficits identified in a study of maternal care in facilities without surgical capacity raise concerns about the general readiness of primary care facilities. Primary care is an important source of essential health care – including for underserved populations. Such care could be a powerful platform for responding to a range of health challenges in lower-income countries, – but only if we give increased attention to the infrastructure and supplies at the health centres and clinics that serve as the first line of care for populations, particularly in rural areas.

28Our analysis of some potential determinants of service readiness yielded unexpected results. While external support for the health system was positively associated with the mean values of the service readiness index, none of the elements of national financing tested in our analysis demonstrated a strong relationship with health-facility readiness. Variation within the nine countries included in this particular analysis may have overwhelmed any underlying causal relationships. Such variation in health-system capacity may explain why countries with similar incomes and epidemiological burdens have achieved very different health outcomes.,– The associations we observed between facility payment factors and service readiness reinforce the idea that, in any given country, health-system performance may be more greatly improved by reducing the variability between facilities than by simply increasing health funding at the national level. Research on within-country changes in health-system inputs and on the resulting capacity is needed. The results of our analyses indicate specific areas for follow-up research, such as the relatively strong performance of public health facilities – compared with private ones – in Namibia and Senegal.

29This research has several limitations. For example, depending on the country, no information on one to 13 items, of the 50 used in the formal assessment of the service readiness index, was available. The older DHS service provision assessments we considered did not assess medications for noncommunicable diseases. Lack of information on some items made it more difficult to gauge the full service readiness in some countries. However, our calculations of the service readiness index excluded these items and also controlled for survey year – to account for the temporal differences in the types of data recorded in the DHS service provision assessments. Although the assessment in Bangladesh excluded small private facilities, all of the other assessments were based on nationally representative samples or were complete – or nearly complete – censuses of all health facilities. To maximize sample size and comparability, we selected all the countries in which DHS service provision assessments had been conducted in the decade preceding our analyses. The inclusion of additional countries could provide greater capacity to test the contribution of national factors to facility readiness.

30The main aims of the present study were to describe service readiness, to compare it across key strata within countries and to assess potential determinants across countries. While our findings are not generalizable to all low- and middle-income countries – particularly those in regions not covered in the sample, such as Latin America and East Asia – they cover seven of 31 low-income countries in the world. Taken together with prior research on service readiness and process quality, in other lower-income countries, the results highlight the depth and breadth of the quality deficits in health care that must be addressed on the road to universal health coverage. Efforts to ensure access to health care will fail to improve population outcomes if health facilities lack the basic capacity to provide care. Although the service readiness index may be used to assess the quality of infrastructure and equipment, it cannot indicate the provision of competent care. The provision of such care is likely to be even rarer than good service readiness.,, The health systems in all 10 of our study countries experience shortages in basic resources for essential services. For these countries, efforts at quality improvement must be national in scope and, in most cases, designed with a specific focus on addressing the inequities for those accessing rural and/or public health centres/clinics. Countries may benefit from identifying the best-performing facilities as case studies for better practices and from working to standardize support for service readiness nationwide. More broadly, reducing variability and improving efficiency in the translation of health-system inputs to facility readiness – and, ultimately, the translation of readiness to the quality of care delivered – are critical steps in the pathway to health care that is both available and effective.