Technical note

Developing an index for the coverage of essential health services

Daniel Hogan, Ahmad Reza Hosseinpoor, and Ties Boerma Department of Evidence, Information and Research (WHO, Geneva) May 2016

Preamble

Monitoring coverage of essential health services is a key component of the WHO/World Bank framework for monitoring Universal Health Coverage (UHC). One approach to summarizing coverage of essential health services is to compute an index which combines tracer indicators of health service coverage into a single metric. This technical note documents the approach used to construct a service coverage index, focusing on national coverage levels, which was published in WHO’s World Health Statistics 2016: Monitoring Health for the SDGs, and was consistent with the wording of SDG indicator 3.8.1 as of May 2016. It is anticipated that the inputs and methods used to compute the index will evolve in the future, in particular to capture inequality in service coverage and to improve the relevance of the index to higher income countries, as well as to incorporate further feedback from various stakeholders.

Background

Universal health coverage (UHC) has recently been adopted as Target 3.8 of the Sustainable Development Goals, which will require regular reporting on progress. UHC is a process of progressive realization in which all people receive the quality, essential health services they need, without being exposed to financial hardship. Monitoring UHC therefore requires measuring health service coverage and financial protection within countries, including coverage among disadvantaged populations to track equity. In this technical note, we describe a methodological approach to monitoring the health service coverage component of the UHC target within the SDGs (indicator 3.8.1), which was published in WHO’s World Health Statistics 2016: Monitoring Health for the SDGs (1). The measurement approach will likely be refined in the future. Countries provide a wide range of services for health protection, promotion, prevention, treatment and care, but it is possible to define a set of tracer indicators that provide a good picture of overall service coverage (2,3). These tracer indicators can then be combined in a systematic way into an index that summarizes national service coverage with a single numerical value (4). Ideally, this index would be constructed so that it incorporates information on how equitably health service coverage is distributed across subpopulations. Tracer indicators were selected considering several criteria (3). First, an indicator should be relevant, reflecting epidemiological burden and the presence of (cost) effective interventions. Second, an indicator should be conceptually sound, with a measurable numerator and denominator, a clear target, and a definition that captures effective coverage (5). Third, it must also be feasible, with current, comparable data available for most countries, which can be disaggregated for equity analysis. Lastly, indicators should be usable, in the sense they are easy to communicate; indicators that are already reported across countries are appealing as they reduce reporting burden. Currently, no potential tracer indicators meet all of these criteria, and therefore proxy indicators, to approximate ideal quantities of interest, and statistical models, to bridge data gaps, are necessary in many cases. Ideal data sources are typically nationally representative, population-based surveys, which enable the measurement of those who need an intervention, in addition to counting those who receive it, and allow for disaggregation of coverage by different subpopulations for equity analysis. In other cases, administrative data collected from health facilities may be used, but often require additional analytic steps to estimate denominators or conduct equity analyses. A set of tracer indicators for service coverage were selected following an extensive review and discussion of potential indicators (2,6-8). These indicators are grouped into four main categories: (1) reproductive, maternal, newborn and child health, (2) infectious diseases, (3) noncommunicable diseases and (4) service capacity and access and health security. Coverage estimates for these tracer indicators are combined into a UHC service coverage index, which ultimately should be adjusted to reflect the degree of inequity in service coverage within a country. In the following sections, we summarize the tracer indicators for monitoring UHC service coverage (including detailed metadata), describe how the UHC service coverage index was computed for World Health Statistics 2016, and discuss areas for future work to further refine the measurement approach. Description of methods Tracer Indicators A total of 16 tracer indicators were selected, with four indicators chosen for each of the four service coverage categories (Table 1, with metadata in Annex 1). Effective coverage, which captures the extent to which those in need of health services receive quality care that obtains the desired result,

was the guiding principle for choosing tracer indicators. However, as can be seen in Table 1, in several cases the selected indicator is only a proxy for effective coverage, which is currently not measurable for many important health service areas. Perhaps surprisingly, given the proliferation of health surveys in developing countries over the past 20 years, for several coverage indicators we have more readily available comparable data from poorer countries than wealthier ones. This is particularly true for estimating sub-national coverage across key inequality dimensions, as surveys typically collect the variables necessary for disaggregated analysis, as compared to administrative data, which may play a greater role in monitoring service coverage in high income countries. The availability of data and comparable coverage estimates at country level, and our ability to disaggregate estimates by key inequality dimensions, are presented in Table 2. 1. Reproductive, maternal, newborn and child health In the area of reproductive, maternal, newborn and child health (RMNCH), data from a large number of population-based health surveys exist, for example from the Demographic and Health Surveys (DHS) and the Multiple Indicator Cluster Surveys (MICS). The selected indicators are therefore well-measured in many countries. These include indicators for family planning (demand satisfied with a modern method), pregnancy care (the average coverage of 4 or more antenatal care visits and skilled birth attendance/institutional delivery rates), full immunization for infants (for now measured with DTP3 coverage as a proxy), and child treatment (care seeking for suspected pneumonia). 2. Infectious diseases For infectious diseases, indicators were selected for treatment of TB and HIV, use of insecticide treated bed nets (ITN) among populations at risk of malaria, and household access to improved water and sanitation. National estimates of TB and HIV treatment coverage are derived from a mixture of facility data, for the numerator, and model-based estimates of the need for treatment, for the denominator (Annex 1). More work is required to improve the accuracy of ART coverage estimates in high income countries and other countries with low HIV burden. ITN coverage is estimated by combining household survey data with information on purchasing and distribution of bed nets for countries with a high burden of malaria, and data on coverage of improved water and sanitation sources are collected during household surveys. 3. Noncommunicable diseases A growing set of surveys, such as the WHO-supported STEPs surveys, include biomarkers that allow for important new possibilities for monitoring treatment coverage of some NCDs. However, currently no comparable national estimates of effective coverage of cardiovascular disease and diabetes treatment, nor treatment for elevated cardiovascular risk, are available. Until they become available, the prevalence of hypertension and raised blood glucose (diabetes) are used as proxy measures, which are meaningful in their own right as indicators of both the success of prevention efforts and screening and treatment programs. Non-use of tobacco is included as an effective coverage indicator of interventions to reduce tobacco use through the health sector and in other sectors, which has large implications for the control of cardiovascular disease, chronic obstructive lung diseases, and cancer. Finally, in addition to non-use of tobacco, cervical cancer screening was selected as an indicator of coverage of interventions against cancer. Currently, data for this indicator are very limited across countries. 4. Service capacity and access and health security The final indicator category, which attempts to represent more general features of capacity and access to care within a health system, includes indicators on hospital care, the density and distribution of core health professionals, psychiatrists and surgeons, access to essential medicines, and a measure of health security. These indicators depart from the effective coverage paradigm that has been used to motivate the selection of tracer indicators for monitoring UHC progress (3). This

choice is largely due to data limitations, which leave major gaps in the measurement of coverage for critical areas such as treatment for mental illnesses, emergency care, and surgical procedures. Data are limited because for many conditions of interest, such as treatment of major depression or emergency care for a road traffic injury, even if we knew the number of people who received care, it is difficult to define the denominator, as it requires diagnosis of need for treatment from survey questions, and severity can differ substantially. Second, even with effective survey instruments for determining need, the incidence of some conditions are likely to be too rare to be precisely estimated from population-based surveys. Thus, while indicators like effective coverage of trauma surgery are unlikely to be widely measured in the next several years, the capacity of the health system to provide life-saving acute care is clearly an essential feature of UHC.

An additional challenge of measuring service capacity and access is that we often do not know what the optimal level of coverage is, and high values may be disadvantageous. Indeed, in high income countries over use of health services may be a significant problem. We also do not know if those who receive a health service are those who need it, e.g. for Caesarean section rates. Despite this, in developing countries, low utilization rates or health worker density are indicative of poor access and use of essential health services. In the absence of better data, tracking utilization rates and health worker density up to some threshold is therefore considered useful for tracking UHC progress. These thresholds are set based on lower bounds observed in OECD countries; once a country exceeds the threshold, the value is fixed at 100%. Thus, these are only indicators of under-use and low capacity; no attempt is made to measure overuse.

As mentioned above, a final indicator in this category is for health security, namely the International Health Regulations core capacity index for responding to epidemics and other health threats. This is also an indicator of general capacity, which is presumably more meaningful than, for example, effective coverage against a pandemic, as it is unclear how to measure the latter on a regular basis. Data availability Ten of the 16 tracer indicators have recent, comparable estimates of national coverage (Table 2). For another four, including pregnancy care, care seeking for suspected pneumonia in children, hospital in-patient admission rates, and health worker density, well-maintained databases of country data points from either survey or administrative data are available. For the remaining two indicators, cervical cancer screening and access to essential medicines, there are currently no comprehensive data bases or comparable estimates available. These two indicators are left out of the calculation of the index presented in this technical note; in future years more information on these indicators is expected to be available. Missing country indicator values for year 2015 were imputed in one of three ways. The first option was to carry forward the most recently available estimate for a country from previous years. If this was not possible, in most cases a regional value, based on World Bank regions, was assigned as a placeholder. The use of regional averages, as opposed to regression-based imputation, was chosen to increase the accessibility of the index calculations for those interested in computing the index with different data inputs. Despite this choice to avoid modeling to impute missing values, there were three instances where there were not enough country data to allow for sensible regional estimates across all regions. These were inpatient admission rates and surgeons per capita, for which regressions fit to the log of GDP per capita (2011 PPP) were used to impute missing country values, and care-seeking for pneumonia, for which missing values were imputed from a regression fit to the log of the pneumonia mortality rate for children under five (9). Finally, given that ITN use is only relevant for a subset of countries, ITN coverage was only included in index calculations for countries in Africa with available estimates (Box 1).

UHC Health Service Coverage Index A diagram of the calcluations used to construct an index of health service coverage is presented in Box 1. While the ultimate goal is to construct an index that is adjusted for inequalities in coverage by socio-economic status or other key inequality dimensions, given current data limitations we focus on constructing an index based on average national levels of coverage in this Technical Note. All indicators are structured so they occur on a scale of 0 to 100%, with 100% the target. For example, non-use of tobacco, rather than use of tobacco, is used, and as described above, in-patient admission rates and health professional density are rescaled onto a scale of 0 to 100%. The index is constructed from geometric means of component indicators, first within each of the four categories, and then across those category-specific means to obtain the final summary index.1 Geometric means are used instead of arithmetic means because they are more sensitive to extreme values, thus implicitly giving more weight to health services with lower coverage, and because the country rankings implied by an index constructed from geometric means are less sensitive to the scale on which input variables are measured, as compared to when using arithmetic means. The Human Development Index was recently revised to use geometric means (10), and geometric means were used in previous work on an UHC index (4). It is noted that weights could be assigned to each indicator, each broad category, or both when computing the index. Potential approaches could involve weighting based on the population in need of a particular service, the burden of disease represented by a particular service area, the expected health impacts that each service is expected to impart on population health, or even the cost-effectiveness of interventions within each health service area. There is, however, no clear way to identify weighting values for any of these approaches, and so a simple equal weighting approach was selected. In World Health Statistics 2016, we present the distribution of country quintiles for the UHC index by WHO region based on provisional estimates using the above methods. These methods will be refined in the coming year, with baseline country estimates of UHC service coverage index scheduled for publication in December 2016. We also note that a single index cannot possibly characterize all of the necessary and critical health services that are required to achieve universal health coverage, however selecting a manageable set of tracer indicators and constructing an index provides a concrete summary of health service coverage within UHC, which can serve as a point of reference for policy discussions, such as for SDG target 3.8, and help highlight specific issues and patterns across countries. This monitoring framework is also flexible enough for countries to add additional indicators that they deem to be important. Ongoing methodological work Incorporating Equity Wagstaff et al (4) demonstrated how average national health service coverage estimates could be adjusted to account for the unequal distribution of coverage across economic status groups. This was operationalized by computing a concentration index, which summarizes the distribution of coverage of a particular health service in relation to household economic status. This approach is conceptually related to the Gini coefficient, but is expanded in the sense that it summarizes the distribution of one variable (coverage) in relation to a second one (economic status). The

1 In rare cases where coverage was 0%, it was replaced by a value of 1% to allow for the calculation of a geometric mean.

concentration index is a useful concept, but simpler stratification approaches may be more accessible and provide qualitatively similar information, for example comparing coverage among the poorest quintile (or two poorest quintiles) to the national average. Depending on the indicator, it may also be desirable to define a “disadvantaged” population for summarizing inequality in coverage in terms of other inequality dimensions, e.g., sex, age, education, or place of residence. In some countries, race, ethnicity or caste may be particularly important dimensions for monitoring equity. However defined, once coverage among the disadvantage population is computed, it can be used to adjust the national coverage value. One way to do this is to compute the service coverage index separately for the national population and the disadvantaged population, and then average them to obtain an equity-adjusted UHC coverage index (or, relatedly, to compute equity adjusted averages for each indicator separately, and then combine those values with an index). While this approach is appealing, we currently do not have the data necessary to compute coverage levels in the disadvantaged population for all tracer indicators. Until more data are available, an alternative approach could focus on computing an index based on the national coverage rates for all 16 tracer indicators, and then adjusting that index value by a single equity adjustment factor. This equity adjustment factor would be computed for each country from the subset of indicators that have sufficient data available to compute coverage across disadvantaged populations. Currently, this is largely confined to RMNCH coverage indicators in low- and middle-income countries. Ensuring relevance for all countries The process of selecting tracer indicators highlighted the massive data gaps that currently exist for monitoring health service coverage across countries. Some important health areas are left wanting, for example, there are currently no strong indicators for coverage of interventions against cancer, mental health and injuries. In general, there are few indicators of national health service coverage that are estimated consistently across most countries, and even fewer indicators that actually measure the effective coverage of health interventions. A limitation of the current UHC index is that it is arguably more relevant for low and middle income countries. High income countries have coverage levels approaching 100% for most of the RMNCH and service capacity and access indicators, and the infectious disease indicators may be poorly measured (ART) or not relevant (ITN and WASH). Perhaps surprisingly, however, the lack of standardized surveys across high income countries inhibits our ability to monitor equity in coverage. We arguably have more internationally comparable information on the equity of service coverage in developing countries as compared to developed ones, yet many of the gains required to achieve UHC in developed countries must come through improving coverage in disadvantaged populations. Going forward, one way to make a single index more relevant across all countries could be to employ a two-pronged approach for some indicators. For example, once hospital admission rates reach the threshold value, a second indicator of hospital care could be used, such as waiting times for hip-fracture surgery. These indicators could potentially be combined into one scale, with hospital admissions making up the scale from 0 to 50%, and waiting times contributing to the scale from 51 to 100%.

References (1) World Health Organization. World Health Statistics 2016: Monitoring Health for the SDGs.

Geneva: World Health Organization, 2016. (available from: http://www.who.int/gho/publications/world_health_statistics/2016/en/).

(2) World Health Organization, World Bank. Tracking Universal Health Coverage: First Global Monitoring Report. Geneva: World Health Organization, 2015. (available from: http://apps.who.int/iris/bitstream/10665/174536/1/9789241564977_eng.pdf?ua=1).

(3) Boerma T, AbouZahr C, Evans D, Evans T. Monitoring intervention coverage in the context of universal health coverage. PLoS Med. 2014;11(9):e1001728.

(4) Wagstaff A, Dmytraczenko T, Almeida G, Buisman L, Hoang-Vu Eozenou P, Bredenkamp C, et al. Assessing Latin America's Progress Toward Achieving Universal Health Coverage. Health Aff (Millwood). 2015;34(10):1704-12.

(5) Ng M, Fullman N, Dieleman JL, Flaxman AD, Murray CJ, Lim SS. Effective coverage: a metric for monitoring Universal Health Coverage. PLoS Med. 2014;11(9):e1001730.

(6) Boerma T, Eozenou P, Evans D, Evans T, Kieny MP, Wagstaff A. Monitoring progress towards universal health coverage at country and global levels. PLoS Med. 2014;11(9):e1001731.

(7) World Health Organization, World Bank. Monitoring Progress towards Universal Health Coverage at Country and Global Levels: A Framework. 2013. (available from: http://www.who.int/healthinfo/country_monitoring_evaluation/UHC_WBG_DiscussionPaper_Dec2013.pdf).

(8) World Health Organization, World Bank. Monitoring progress towards universal health coverage at country and global levels: Framework, measures and targets. Geneva: World Health Organization, 2014. (available from: http://apps.who.int/iris/bitstream/10665/112824/1/WHO_HIS_HIA_14.1_eng.pdf).

(9) World Health Organization, Maternal Child Epidemiology Estimation Group. Child causes of death 2000-2015. World Health Organization, 2016. (available from: http://www.who.int/healthinfo/global_burden_disease/estimates_child_cod_2015/en/.)

(10) Klugman J, Rodriguez F, Choi H-J. The HDI 2010: New Controversies, Old Critiques. United Nations Development Programme, 2011.

Table 1. UHC tracer indicators for monitoring progress on health service coverage. Tracer area Tracer indicator

Reproductive, maternal, newborn and child health a. Family planning Demand satisfied with modern method among women 15-

49 who are married or in a union (%) b. Pregnancy care Average coverage of 4 or more antenatal visits and skilled

birth attendance (%) c. Full child immunization One year old children who have received 3 doses of a

vaccine containing diphtheria, tetanus and pertussis (%) d. Child treatment Care seeking behaviour for children with suspected

pneumonia (%) Infectious diseases

a. Tuberculosis treatment TB cases detected and cured (%) b. HIV treatment People living with HIV receiving ART (%) c. Malaria prevention Population at risk sleeping under insecticide treated bed

nets (%) d. Improved water and sanitation Average coverage of households with access to improved

water and sanitation (%) Noncommunicable diseases

a. Treatment of cardiovascular disease Prevalence of raised blood pressure (%)1 b. Management of diabetes Prevalence of raised blood glucose (%)1 c. Cervical cancer screening Cervical cancer screening among women 30-49 (%) d. Tobacco control Adults age >=15 years not smoking tobacco in last 30 days

(%) Service capacity and access

a. Hospital access2 In-patient admissions per capita (w/ threshold) b. Health worker density2 Health professionals per capita (w/ threshold): physicians,

psychiatrists, and surgeons c. Access to essential medicines Average proportion of WHO-recommended core list of

essential medicines present in health facilities d. Health security International Health Regulations core capacity index

1. As more data become available, these will likely be replaced by either (i) proportion of population with hypertension or diabetes receiving effective treatment (ii) proportion of eligible persons (defined as aged 40 years and older with a 10-year cardiovascular risk ≥30%, including those with existing cardiovascular disease) receiving drug therapy and counselling (including glycaemic control) to prevent heart attacks and strokes. 2. There is no clear optimal level of service capacity and access. Therefore, these indicators are measured against a threshold defined by minimum rates observed in OECD countries. Thus, countries with low rates of service capacity and access will score below 100% on these indicators, and countries that are above the threshold will be set at 100%. It is currently unclear how to estimate the “over-use” of health services.

Table 2. Tracer indicators for UHC service coverage, with data availability

Indicator Data sources Number of countries with national data since 2010

Number of countries with comparable estimates

Measurability of key dimensions of inequalitya,b

Reproductive, maternal, newborn and child health Family planning coverage Surveys 98 184 W,E,R,(A) Antenatal and delivery care Surveys 121 194 W,E,R,(A) Full child immunization Surveys, Admin 193 194 W,E,R,S Health-seeking behaviour for child pneumonia

Surveys 72 None W,E,R,S

Infectious diseases Tuberculosis effective treatment

Admin 190 190 (R)

HIV antiretroviral treatment

Admin, Surveys, Surveillance

118 118 (R)

ITN coverage for malaria prevention

Surveys, Admin 40c 40c W,E,R,S

Improved water source and adequate sanitation

Surveys 156 192 W,R

Noncommunicable diseases Prevalence of raised blood pressure

Surveys 86 192 (E),(R),S,A

Prevalence of raised blood glucose

Surveys 76 192 (E),(R),S,A

Cervical cancer screening Surveys <30 None - Tobacco (non-use) Surveys 146 123 (W),(E),(R),S,(A) Service capacity and access Basic hospital access Facility data 105 None (R) Health-worker density Administrative

data 166 None (R)

Access to essential medicines

Facility surveys <30 None (R)

Health security: IHR compliance

Country reported

191 None -

a W = household wealth quintile; E = educational attainment; R = place of residence (typically urban vs. rural); S = sex; and A = age. Letters in parentheses indicate that data sources exist to estimate coverage by the indicated dimension but that more analytical work is needed to prepare disaggregated estimates. b Information to estimate coverage across key inequality dimensions typically comes from population-based surveys. Standardized population-based surveys are typically only conducted in developing countries, and therefore there is currently a lack consistent data sources to characterize equity for service coverage in many high-income countries. c Only pertains to countries with highly endemic malaria.

Box 1. Calculation of UHC service coverage index based on national levels for coverage

Notes 1. Cervical cancer screening and access to essential medicines are excluded due to low data availability. 2. The percentage of the adult population with non-elevated blood pressure and non-elevated blood glucose are based on age-standardized estimates. These distributions are rescaled to provide finer resolution for the index, based on looking at observed minima across countries. For non-elevated blood pressure, the rescaled indicator = 100-(100-X)/(100-66.67), and for non-elevated blood pressure, the rescaled indicator = 100-(100-X)/(100-75). 3. Similar to (2), non-use of tobacco is also based on age-standard estimates, and rescaled to provide a finer resolution using a minimum bound of 50%, so that the rescaled indicator = 100-(100-X)/(100-50). 4. In-patient hospital admission rates are rescaled and capped based on a threshold of 8 per 100, based on minimum rates observed in OECD countries. Values below 0.08 are rescaled as X/0.08*100, and values above 0.08 are set to 100%. 5. Similar to (4), health worker density is rescaled and capped based on threshold values. Physician density has a threshold of 0.9 per 1000, psychiatrists have a threshold of 1 per 100 000, and surgeons have a threshold of 14 per 100 000. After rescaling these values (i.e., max(100, X/threshold*100), they are combined into a HWD composite variable for entry into the above index calculations, computed as = (physicians * psychiatrists * surgeons)1/3. 6. Pregnancy care is computed as (ANC4 * SAB)1/2 and WASH is computed as (water * sanitation)1/2. In the future these could be replaced by co-coverage estimates, e.g., fraction of women receiving both ANC4 and SAB.

Annex 1 Tracer area Family planning Indicator name Demand for family planning satisfied with modern methods Indicator definition Percentage of women of reproductive age (15−49 years) who are married or in-

union who have their need for family planning satisfied with modern methods. Numerator Number of women aged 15-49 who are married or in-union who use modern

methods Denominator Total number of women aged 15-49 who are married or in-union in need of family

planning Main data sources Population-based health surveys Method of measurement Household surveys include a series of questions to measure modern contraceptive

prevalence rate and demand for family planning. Total demand for family planning is defined as the sum of the number of women of reproductive age (15–49 years) who are married or in a union and who are currently using, or whose sexual partner is currently using, at least one contraceptive method, and the unmet need for family planning. Unmet need for family planning is the proportion of women of reproductive age (15–49 years) either married or in a consensual union, who are fecund and sexually active but who are not using any method of contraception (modern or traditional), and report not wanting any more children or wanting to delay the birth of their next child for at least two years. Included are: 1. all pregnant women (married or in a consensual union) whose pregnancies

were unwanted or mistimed at the time of conception; 2. all postpartum amenorrhoeic women (married or in consensual union) who

are not using family planning and whose last birth was unwanted or mistimed; 3. all fecund women (married or in consensual union) who are neither pregnant

nor postpartum amenorrhoeic, and who either do not want any more children (want to limit family size), or who wish to postpone the birth of a child for at least two years or do not know when or if they want another child (want to space births), but are not using any contraceptive method.

Method of estimation The United Nations Population Division produces a systematic and comprehensive series of annual estimates and projections of the percentage of demand for family planning that is satisfied among married or in-union women. A Bayesian hierarchical model combined with country-specific data are used to generate the estimates, projections and uncertainty assessments from survey data. The model accounts for differences by data source, sample population, and contraceptive methods. See here for details: http://www.un.org/en/development/desa/population/theme/family-planning/cp_model.shtml

UHC-related notes

Tracer area Pregnancy care Indicator definition Percentage of women aged 15-49 years with a live birth in a given time period who

received antenatal care four or more times and who had skilled health personnel attending the birth

Numerator Number of women aged 15−49 years with a live birth in a given time period who (1) received antenatal care four or more times as well as (2) had skilled health personnel attending the birth

Denominator Total number of women aged 15−49 years with a live birth in the same period. Main data sources Household surveys and routine facility information systems. Method of measurement Data on four or more antenatal care visits is based on questions that ask if and

how many times the health of the woman was checked during pregnancy. Data on skilled birth attendance come from questions that ask respondents about who helped during delivery. Note that the definition of skilled birth attendant varies between countries, but should include doctors, nurses or midwives, who are trained in providing live-saving obstetric care giving the necessary supervision, care and advice for women during pregnancy, childbirth and postpartum, to conduct deliveries on their own, and to care for newborns. Household surveys that can generate this indicator include DHS, MICS, RHS and other surveys based on similar methodologies. Service/facility reporting systems can be used where the coverage is high, usually in industrialized countries.

Method of estimation WHO and UNICEF maintain data bases on coverage of antenatal care and births attended by skilled health personnel. Considerable effort is spent on verifying skilled birth attendant definitions from survey data; in many cases survey reports will present coverage of “skilled birth attendance” but use cadres that are not considered skilled. These figures must be adjusted to ensure comparability for global monitoring purposes. During 2016, WHO and UNICEF are collaborating to conduct a full review and country consultation on this issue to obtain a final set of data sources and estimates.

UHC-related notes This indicator is intended to provide a more comprehensive measure of pregnancy care as compared to monitoring antenatal care and skill birth attendance coverage separately. To calculate it, original survey data must be re-analyzed to determine the joint distribution of ANC4 and SAB coverage across respondents. To date, this exercise has not been completed; as a proxy, the geometric mean of national estimates of ANC4 and SAB coverage is computed, which is an overestimate. Note that this indicator could later be expanded to incorporate a measure of post-delivery care if desired.

Tracer area Full child immunization Indicator definition Percentage of infants receiving three doses of diphtheria-tetanus-pertussis

containing vaccine Numerator Children 1 year of age who have received three doses of diphtheria-tetanus-

pertussis containing vaccine Denominator All children 1 year of age Main data sources Household surveys and facility information systems. Method of measurement For survey data, the vaccination status of children aged 12–23 months is

collected from child health cards or, if there is no card, from recall by the care-taker. For administrative data, the total number of doses administered to the target population is extracted.

Method of estimation Together, WHO and UNICEF derive estimates of DTP3 coverage based on data officially reported to WHO and UNICEF by Member States, as well as data reported in the published and grey literature. They also consult with local experts - primarily national EPI managers and WHO regional office staff - for additional information regarding the performance of specific local immunization services. Based on the available data, consideration of potential biases, and contributions from local experts, WHO/UNICEF determine the most likely true level of immunization coverage. For details, see here: http://www.who.int/bulletin/volumes/87/7/08-053819/en/ http://www.who.int/immunization/monitoring_surveillance/routine/coverage/en/index4.html

UHC-related notes There is variability in national vaccine schedules across countries. Given this, one option for monitoring full child immunization is to monitor the fraction of children receiving vaccines included in their country’s national schedule. A second option, which may be more comparable across countries and time, is to monitor DTP3 coverage as a proxy for full child immunization. Diphtheria-tetanus-pertussis containing vaccine often includes other vaccines, e.g., against Hepatitis B and Haemophilus influenza type B, and is a reasonable measure of the extent to which there is a robust vaccine delivery platform within a country.

Tracer area Child treatment (care-seeking for symptoms of pneumonia) Indicator definition Percentage of children under 5 years of age with suspected pneumonia (cough and

difficult breathing NOT due to a problem in the chest and a blocked nose) in the two weeks preceding the survey taken to an appropriate health facility or provider.

Numerator Number of children with suspected pneumonia in the two weeks preceding the survey taken to an appropriate health provider.

Denominator Number of children with suspected pneumonia in the two weeks preceding the survey.

Main data sources Household surveys Method of measurement During the UNICEF/WHO Meeting on Child Survival Survey-based Indicators, held

in New York, 17–18 June 2004, it was recommended that acute respiratory infections (ARI) be described as “presumed pneumonia” to better reflect probable cause and the recommended interventions. The definition of presumed pneumonia used in the Demographic and Health Surveys (DHS) and in the Multiple Indicator Cluster Surveys (MICS) was chosen by the group and is based on mothers’ perceptions of a child who has a cough, is breathing faster than usual with short, quick breaths or is having difficulty breathing, excluding children that had only a blocked nose. The definition of “appropriate” care provider varies between countries. WHO maintains a data base of country-level observations from household surveys that can be accessed here: http://www.who.int/gho/child_health/prevention/pneumonia/en/

Method of estimation There are currently no internationally comparable estimates for this indicator. UHC-related notes This indicator is not typically measured in higher income countries with well-

established health systems. For countries without observed data, coverage was estimated from a regression that predicts coverage of care-seeking for symptoms of pneumonia (on the logit scale), obtained from the WHO data base described above, as a function of the log of the under-five pneumonia mortality rate, which can be found here: http://www.who.int/healthinfo/global_burden_disease/estimates_child_cod_2015/en/.

Tracer area Tuberculosis detection and treatment Indicator definition Percentage of incidence TB cases that are detected and successfully treated in a

given year Numerator Number of new and relapse cases detected in a given year and successfully treated Denominator Number of new and relapse cases in the same year Main data sources Facility information systems, surveillance systems, population-based health

surveys with TB diagnostic testing, TB register and related quarterly reporting system (or electronic TB registers)

Method of measurement This indicator requires three main inputs: (1) The number of new and relapse TB cases diagnosed and treated in national TB control programmes and notified to WHO in a given year. (2) The number of incident TB cases for the same year, typically estimated by WHO. (3) Percentage of TB cases successfully treated (cured plus treatment completed) among TB cases notified to the national health authorities. The final indicator = (1)/(2) x (3)

Method of estimation Estimates of TB incidence are produced through a consultative and analytical process led by WHO and are published annually. These estimates are based on annual case notifications, assessments of the quality and coverage of TB notification data, national surveys of the prevalence of TB disease and information from death (vital) registration systems. Estimates of incidence for each country are derived, using one or more of the following approaches depending on available data: 1. incidence = case notifications/estimated proportion of cases detected; 2. incidence = prevalence/duration of condition; 3. incidence = deaths/proportion of incident cases that die. These estimates of TB incidence are combined with country-reported data on the number of cases detected and treated, and the percentage of cases successfully treated, as described above.

UHC-related notes To compute the indicator using WHO estimates, one can access necessary files here: http://www.who.int/tb/country/data/download/en/, and compute the indicator as = c_cdr x c_new_tsr

Tracer area HIV treatment Indicator definition Percentage of people living with HIV currently receiving antiretroviral therapy

(ART) Numerator Number of adults and children who are currently receiving ART at the end of the

reporting period Denominator Number of adults and children living with HIV during the same period Main data sources Facility reporting systems, sentinel surveillance sites, population-based surveys Method of measurement Numerator: The numerator can be generated by counting the number of adults

and children who received antiretroviral combination therapy at the end of the reporting period. Data can be collected from facility-based ART registers or drug supply management systems. These are then tallied and transferred to cross sectional monthly or quarterly reports which can then be aggregated for national totals. Patients receiving ART in the private sector and public sector should be included in the numerator. Denominator: Data on the number of people with HIV infection may come from population-based surveys or, as is common in sub-Saharan Africa, surveillance systems based on antenatal care clinics.

Method of estimation The numerator is calculated using the above methods by WHO/UNAIDS. To estimate the number of people living with HIV across time, UNAIDS in collaboration with countries uses an epidemic model (Spectrum) that combines surveillance data on prevalence with the current number of patients receiving ART and assumptions about the natural history of HIV disease progression. Since ART is now recommended for all individuals living with HIV, monitoring ART coverage is less complicated than before, when only those with a certain level of disease severity were eligible to receive ART. Estimates of ART coverage can be found here: http://apps.who.int/gho/data/node.main.626

UHC-related notes There are currently no comparable estimates of ART coverage in high income countries, but estimates are expected within the next year.

Tracer area ITN coverage for malaria prevention Indicator definition Percentage of population in malaria-endemic areas who slept under an ITN the

previous night. Numerator Number of people in malaria-endemic areas who slept under an ITN. Denominator Total number of people in malaria endemic areas. Main data sources Data on household access and use of ITNs come from nationally representative

household surveys such as Demographic and Health Surveys, Multiple Indicator Cluster Surveys, and Malaria Indicator Surveys. Data on the number of ITNs delivered by manufacturers to countries are compiled by Milliner Global Associates, and data on the number of ITNs distributed within countries are reported by National Malaria Control Programs.

Method of measurement Many recent national surveys report the number of ITNs observed in each respondent household. Ownership rates can be converted to the proportion of people sleeping under an ITN using a linear relationship between access and use that has been derived from 62 surveys that collect information on both indicators.

Method of estimation Mathematical models can be used to combine data from household surveys on access and use with information on ITN deliveries from manufacturers and ITN distribution by national malaria programmes to produce annual estimates of ITN coverage. WHO uses this approach in collaboration with the Malaria Atlas Project. Methodological details can be found in the Annex of the World Malaria Report 2015: http://www.who.int/malaria/publications/world-malaria-report-2015/report/en/.

UHC-related notes WHO produces comparable ITN coverage estimates for 40 high burden countries. For other countries, ITN coverage is not included in the UHC service coverage index.

Tracer area Improved water and adequate sanitation source Indicator definition Percentage of households using improved water and improved sanitation facilities Numerator Population living in a household with drinking water from: piped water into

dwelling, plot or yard; public tap/stand pipe; tube well/borehole; protected dug well; protected spring; or rainwater collection AND living in a household with: flush or pour-flush to piped sewer system, septic tank or pit latrine; ventilated improved pit latrine; pit latrine with slab; or composting toilet.

Denominator Total population Main data sources Population-based household surveys and censuses Method of measurement Household-level responses, weighted by household size, are used to compute

population coverage. Method of estimation The WHO/UNICEF Joint Monitoring Programme has produced regular estimates of

coverage of improved water and improved sanitation for MDG monitoring. After compiling a database of available data sources, for each country, simple linear regressions are fitted to the country’s data series to obtain an in-sample estimate, as well as to produce a 2-year extrapolation beyond the last available data point, after which coverage is held constant for 4 years and then assumed missing. This is done separately for urban and rural regions, and then combined to obtain national coverage estimates. Details of the methodology and most recent estimates can be found here: http://www.wssinfo.org/

UHC-related notes This indicator is intended to provide a more comprehensive measure of water and sanitation coverage than estimating use of improved water and sanitation separately as two indicators. To calculate it, original survey data must be re-analyzed to determine the joint distribution of improved water and sanitation coverage across households. To date, this exercise has not been completed; as a proxy, the geometric mean of national estimates of improved water and improved sanitation coverage is computed, which is an overestimate. The SDG indicators for drinking water and sanitation are expanded versions of the MDG indicators, incorporating the safety/quality of drinking water and sanitation facilities. Once country data and estimates are available for these new indicators, they could be used for UHC monitoring in lieu of those based on the MDG indicator definitions described above.

Tracer area Treatment of cardiovascular disease Indicator definition Age-standardized prevalence of raised blood pressure among adults aged 18+ Numerator Number of adults aged 18 or older with systolic blood pressure >=140 mm Hg or

diastolic blood pressure >=90 mm Hg Denominator Number of adults aged 18 or older Main data sources Population-based surveys and surveillance systems Method of measurement Data sources recording measured blood pressure are used (self-reported data are

excluded). If multiple blood pressure readings are taken per participant, the first reading is dropped and the remaining readings are averaged.

Method of estimation For producing comparable national estimates, data observations of prevalence defined in terms of alternate SBP and/or DBP cutoffs are converted into prevalence of systolic blood pressure >=140 mm Hg or diastolic blood pressure >=90 mm Hg using regression equations. A Bayesian hierarchical model is then fitted to these data to calculate age-sex-year-country specific prevalences, which accounts for national vs. subnational data sources, urban vs. rural data sources, and allows for variation in prevalence across age and sex. Age-standardized estimates are then produced by applying the crude estimates to the WHO Standard Population. Details on the statistical methods are here: http://www.thelancet.com/journals/lancet/article/PIIS0140-6736(10)62036-3/abstract WHO and the Global Burden of Metabolic Risk Factors of Chronic Diseases Collaborating Group has produced comparable estimates for this indicator for year 2014, which are available here: http://apps.who.int/gho/data/node.main.A875?lang=en

UHC-related notes As more data become available, this indicator will likely be replaced by the fraction of population with hypertension receiving effective treatment. For now, prevalence of raised blood pressure is used as a proxy.

Tracer area Management of diabetes Indicator definition Age-standardized prevalence of raised blood glucose among adults 18+ Numerator Number of adults aged 18 or older with fasting plasma glucose >= 7.0 mmol/l or on

medication for raised blood glucose Denominator Number of adults aged 18 or older Main data sources Population-based surveys and surveillance systems Method of measurement Fasting plasma glucose (FPG) levels are determined by taking a blood sample from

participants who have fasted for at least 8 hours. Other related bio-indicators can also be incorporated into estimates (see below), in particular the two-hour oral glucose tolerance test (OGTT) or hemoglobin A1c (HbA1c) percentage. Self-reported data on diabetes diagnosis are excluded, but self-reported data are used for identifying whether or not a respondant is currently on medication for raised blood glucose.

Method of estimation For producing comparable national estimates, data observations of diabetes prevalence defined in terms of FPG, OGTT, HbA1c, or combinations therein, are all converted into diabetes prevalence in terms of FPG >= 7.0mmol/l. A Bayesian hierarchical model is then fitted to these data to calculate age-sex-year-country specific prevalences, which accounts for national vs. subnational data sources, urban vs. rural data sources, and allows for variation in prevalence across age and sex. Age-standardized estimates are then produced by applying the crude estimates to the WHO Standard Population. Methodological details can be found here: http://www.thelancet.com/journals/lancet/article/PIIS0140-6736(16)00618-8/fulltext WHO and the Global Burden of Metabolic Risk Factors of Chronic Diseases Collaborating Group has produced comparable estimates for this indicator for year 2014, which are available here: http://apps.who.int/gho/data/node.main.A869?lang=en

UHC-related notes As more data become available, this indicator will likely be replaced by the fraction of population with diabetes receiving effective treatment. For now, prevalence of raised blood glucose is used as a proxy.

Tracer area Cervical cancer screening Indicator definition Percentage of women aged 30−49 years who report ever having been screened for

cervical cancer Numerator Number of women aged 30−49 years who report ever having had a screening test

for cervical cancer using any of these methods: VIA, pap smear and HPV test. Denominator All women aged 30-49 years Main data sources Population-based surveys Method of measurement Self-reported data on respondents’ cervical cancer screening history are collected

through surveys. Method of estimation There are currently no comparable estimates of cervical cancer screening

coverage. UHC-related notes There are currently few countries with recent data for this indicator and it is

therefore excluded from the 2016 UHC service coverage index calculations. An additional challenge for international comparability is that data sources may use different time periods (ever screened vs. screened in past 5 years) and different age groups.

Tracer area Tobacco control Indicator definition Age-standardized prevalence of adults >=15 years not smoking tobacco in last 30

days Numerator Adults 15 years and older who have not smoked tobacco in the last 30 days Denominator Adults 15 years and older Main data sources Household surveys Method of measurement “Current tobacco smoking" includes cigarettes, cigars, pipes or any other smoked

tobacco products used in the past 30 days. Data are collected via self-report in surveys.

Method of estimation WHO estimates prevalence of current tobacco (non) smoking with a negative binomial meta-regression model, which generates comparable estimates by adjusting for differences in age groups and indicator definition across national surveys included in the analysis. These estimates are done separately for men and women. Methodological details can be found here: http://www.thelancet.com/journals/lancet/article/PIIS0140-6736(15)60264-1/supplemental. WHO estimates of the prevalence of tobacco smoking can be accessed here (see “current smoking of any tobacco product”): http://apps.who.int/gho/data/node.main.1250?lang=en

UHC-related notes Prevalence of not smoking tobacco is computed as 1 minus the prevalence of tobacco smoking.

Tracer area Hospital access Indicator definition Hospital (inpatient) admissions per population per year, relative to a maximum

threshold of 8 per 100 population per year Numerator Number of inpatient admissions (or discharges) per year Denominator Total population Main data sources Administrative systems / Health facility reporting system Method of measurement Country administrative systems are used to total the number of inpatient

admissions (or discharges) per year, and divided by the total estimated population for the same year. Various groups collate information on inpatient admission rates, including: OECD: http://www.oecd.org/els/health-systems/oecd-health-statistics-2014-frequently-requested-data.htm and doi: 10.1787/health_glance_ap-2014-graph74-en PAHO: http://ais.paho.org/phip/viz/basicindicatorbrowaser.asp WHO: http://www.euro.who.int/en/data-and-evidence/core-health-indicators-in-the-who-european-region/core-health-indicators-in-the-who-european-region-2015.-special-focus-human-resources-for-health

Method of estimation Using available data, the indicator is computed relative to the minimum inpatient admission rate observed across OECD countries since 2000, which is 8 per 100 per year. This indicator is designed to capture low rates of hospital access; the maximum threshold is used because high rates of inpatient admissions are not necessarily optimal. The indicator is computed as follows, using country data on inpatient admission rates (x), which results in values ranging from 0 to 100: • Country with an admission rate x < 8 per 100 per year, the indicator =

x /0.08*100. • Country with an admission rate x >= 8 per 100 per year, the indicator = 100. There are currently no comprehensive and comparable estimates of national hospital admission rates.

UHC-related notes There is currently sparse information available on inpatient admission rates for countries from three WHO regions: Africa, Eastern Mediterranean and South-East Asia

Tracer area Health workforce Indicator definition Health professionals (physicians, psychiatrists, and surgeons) per capita, relative to

maximum thresholds for each cadre Numerator Number of physicians, psychiatrists and surgeons Denominator Total population Main data sources National database or registry of health workers, ideally coupled with regular

assessment of completeness using census data, professional association registers, or facility censuses.

Method of measurement The classification of health workers is based on criteria for vocational education and training, regulation of health professions, and activities and tasks of jobs, i.e. a framework for categorizing key workforce variables according to shared characteristics. The WHO framework largely draws on the latest revisions to the internationally standardized classification systems of the International Labour Organization (International Standard Classification of Occupations), United Nations Educational, Scientific and Cultural Organization (International Standard Classification of Education), and the United Nations Statistics Division (International Standard Industrial Classification of All Economic Activities). Methodological details and data can be found here: http://www.who.int/hrh/statistics/hwfstats/en/

Method of estimation There are currently no comprehensive and comparable estimates health worker density across countries. Using available data, the indicator is computed by first rescaling, separately, health worker density ratios for each of the three cadres (physicians, psychiatrists and surgeons) relative to the minimum observed values across OECD countries since 2000, which are as follows: physicians = 0.9 per 1000, psychiatrists = 1 per 100,000, and surgeons = 14 per 100,000. This rescaling is done in the same way as that for the hospital inpatient admission rates indicator described above, resulting in indicator values that range from 0 to 100 for each of the three cadres. For example, using country data on physicians per 1000 population (x), the cadre-specific indicator would be computed as: • Country with x < 0.9 per 1000 per year, the cadre-specific indicator = x /0.9*100. • Country with x >= 0.9 per 1000 per year, the cadre-specific indicator = 100. As a final step, the geometric mean of the three cadre-specific indicator values is computed to obtain the final indicator of health workforce density.

UHC-related notes The “physicians” category would ideally be expanded to include all “core health professionals”, such as nurses and midwives. However, no internationally comparable data base exists that uses consistent definitions of non-physician core health professionals to allow for meaningful cross-country comparisons.

Tracer area Access to essential medicines Indicator definition Percentage of health facilities with essential medicines Numerator Number of facilities with essential medicines in stock Denominator Total number of health facilities Main data sources Special facility surveys or, potentially, routine facility information systems Method of measurement Data on the availability of a specific list of medicines are collected from a survey of

a sample of facilities. Availability is reported as the percentage of medicine outlets where a particular medicine was found on the day of the survey. If routine facility reporting on stocks is accurate and complete, it may also be possible to use data from the routine system. Regular independent verification will be required.

Method of estimation This indicator is still under development, both in terms of the core list of medicines to be monitored and data collection strategies. The Service Availability and Readiness Assessment (SARA) surveys have collected data for a limited number of countries, see here: http://www.who.int/healthinfo/systems/sara_methods/en/

UHC-related notes There are currently about 30 countries with recent data for this indicator and it is therefore excluded from the 2016 UHC service coverage index calculations.

Tracer area Health security Indicator definition International Health Regulations (IHR) core capacity index, which is the average

percentage of attributes of 13 core capacities that have been attained at a specific point in time. The 13 core capacities are: (1) National legislation, policy and financing; (2) Coordination and National Focal Point communications; (3) Surveillance; (4) Response; (5) Preparedness; (6) Risk communication; (7) Human resources; (8) Laboratory; (9) Points of entry; (10) Zoonotic events; (11) Food safety; (12) Chemical events; (13) Radionuclear emergencies.

Numerator Number of attributes attained Denominator Total number of attributes Main data sources Key informant survey Method of measurement Key informants report on attainment of a set of attributes for each of 13 core

capacities using a standard WHO instrument, as described here: http://apps.who.int/iris/bitstream/10665/84933/1/WHO_HSE_GCR_2013.2_eng.pdf Capacity-level indicator values can be found here: http://www.who.int/gho/ihr/monitoring/legislation/en/index1.html

Method of estimation The indicator is computed by averaging, across the 13 core capacities, the percentage of attributes for each capacity that have been attained.

UHC-related notes