measuring health system efficiency in low- and middle
TRANSCRIPT
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE
Revisiting Health Financing Technical Initiative Efficiency Collaborative
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE
Revisiting Health Financing Technical Initiative Efficiency Collaborative
Resource Guide Disclaimer
This Resource Guide was produced by the Joint Learning Network for Universal Health Coverage (JLN), an innovative learning platform where practitioners and policy makers from around the globe co-develop global knowledge that focuses on the practical “how-to” of achieving universal health coverage. For questions or inquiries about this manual or other JLN activities, please contact JLN at [email protected] or Naina Ahluwalia at [email protected].
The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Nothing herein shall constitute or be considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are specifically reserved.
The findings, interpretations, and conclusions made by country contributors do not necessarily reflect the views of the organization, institutions or governments they represent.
Rights and PermissionsThis work is licensed under the Creative Commons Attribution-ShareAlike 4.0 International License (CC BY-SA 4.0). To view a copy of this license, visit https://creativecommons.org/licenses/by-sa/4.0/legalcode.
The content in this document may be freely used and adapted in accordance with this license, provided it is accompanied by the following attribution:
Measuring Health System Efficiency in Low- and Middle-Income Countries: A Resource Guide, Copyright © 2020, Joint Learning Network for Universal Health Coverage, International Decision Support Initiative (iDSI), The World Bank Group.
Product and company names mentioned herein may be the trademarks of their respective owners.
Suggested citationHafez, R., ed. Measuring Health System Efficiency in Low- and Middle-Income Countries: A Resource Guide. Joint Learning Network for Universal Health Coverage, 2020
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE iii
Acknowledgments
The Resource Guide was co-produced with the following country members of the JLN Efficiency Collaborative, and facilitated by the Joint Learning Network’s Efficiency Collaborative technical team.
TECHNICAL EDITORReem Hafez
COUNTRY CONTRIBUTORS:BangladeshAyesha Afroz Chowdhury, Ministry of Health & Family Welfare.Subrata Paul, Ministry of Health & Family Welfare.EthiopiaMideksa Adugna, Oromiya Regional Health Bureau.GhanaKwabena Boakye Boateng, Ghana Health Services.Kwakye Kontor, Ministry of Health.Titus Sorey, National Health Insurance Authority.Vivian Addo-Cobbiah, National Health Insurance Authority,.IndiaKavita Singh, Ministry of Health and Family Welfare.IndonesiaAhmad Ansyori, Council of National Social Security (DJSN).Eka Yoshida, Ministry of Health.Pandu Harimurti, The World Bank.MongoliaB. Munkhtsetseg, Ministry of Health.NigeriaBolaji Aduagba, Federal Ministry of Health.Kayode A. Obasa, Ministry of Budget and National Planning.Nneka Orji, Federal Ministry of Health.PhilippinesAndrea Margreth S. Ora-Corachea, Department of Health.Lavinia A. Oliveros, Department of Health.Martha de la Paz, Department of Health.Paulus Magnus Bacud, PhilHealth.VietnamNguyen Khanh Phuong, Health Strategy and Policy Institute.
Special thanks are owed to the contributors of the two country pilots:Kenya team: Ahmed Omar, Ministry of Health, Esther Wabuge, JLN CCG Coordinator, Isabel Maina, Ministry of Health, Kenneth Munge, KEMRI Wellcome Trust, Mercy Mwangangi, Ministry of Health
Malaysia team: Abdul Hakim Bin Abdul Rashid, Ministry of Health, Lee Kun Yun, Ministry of Health, Mohammad Najib Bin Baharuddin, Ministry of Health, Nor Izzah binti Hj. Ahmad Shauki, Ministry of Health, Yussni Binti Aris A. Haris, Ministry of Health
Other country contributors to the Guide include: Md. Abul Bashar Sarker, Md. Anwar Sadat and Md. Sarwar Bari from Bangladesh; Genet Mulugeta Hirpesa, Eyerusalem Animut and Tseganeh Guracha, from Ethiopia; Daniel Adin Darko, Emmanuel Ankrah Odame, and Ernest Asiedu Konadu from Ghana; Alok Saxena and B K Datta from India; Alice Wangui, Anne Musuva, Andrew Mulwa, Cyrus Matheka and Joseph Githinji from Kenya; Noraryana Bt Hassan from Malaysia; A. Khishigbayar, Bayanjargal Ganbol, L. Munkhzul and Ts. Tsolmongerel from Mongolia; Okechukwu Okwudili, Oritseweyimi Ogbe, Shamsudeen Saad and Uchenna Eugenes Ewelike from Nigeria; Arturo Alcantara and Israel Pargas from the Philippines; Nguyen Lan Huong from Vietnam
The Revisiting Health Financing Technical Initiative team would like to thank the Country Core Groups across JLN countries, as well as the JLN Steering Group for continued support for leveraging existing resources for health, as ‘efficiency’ is a very high technical priority for the JLN.
Valuable feedback was received by the Health Financing Global Solutions Area under the leadership of Christoph Kurowski and other members of the World Bank HNP Leadership team. The technical facilitation team wishes to express special thanks to David Wilson for chairing the peer review process and Owen Smith, David Evans and Laura Di Giorgio for their time and effort in reviewing the guide.
The Resource Guide could not have been co-produced without the financial cooperation of the Government of Japan, the Rockefeller Foundation, the Bill and Melinda Gates Foundation and Australian Aid. The International Decision Support Initiative (iDSI), represented through the Center for Global Development (CGD), were technical partners to the World Bank for facilitation of the Efficiency Collaborative. Management Sciences for Health (MSH) provided continued support and external perspective as the JLN Network Manager.
A special thanks also to the World Bank JLN focal points from Bangladesh, Ethiopia, Ghana, India, Indonesia, Kenya, Malaysia, Mongolia, Nigeria, Philippines and Vietnam in facilitating and encouraging this work.
The Guide technical facilitation team was guided by Reem Hafez from the World Bank and included Somil Nagpal, Danielle Bloom, Naina Ahluwalia and Lydia Ndebele. Melissa Ouellet provided research and data support. Dorothee Chen, Marvin Ploetz, Alessia Thiebaud, Moritz Piazzi, Federica Secci, Manuela Villar Uribe provided data and input to several fact sheets. Ruth Young and QUO Global provided editing, data visualization, layout and production support.
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDEiv
Genesis of the efficiency collaborative
The Joint Learning Network (JLN) is a global network connecting practitioners and policy makers from 31 countries around the globe to co-develop knowledge that focuses on the practical ‘how-to’ of universal health coverage reform in low- and middle-income country contexts.
Over a series of consultations, JLN members reported that one of the most common issues facing their health care systems was how to assess the efficient use of resources. As a result, the Efficiency Collaborative (EC) was launched in April 2017. Participants from 11 JLN member countries (Bangladesh, Ethiopia, Ghana, Indonesia, Kenya, Malaysia, Mongolia, Nigeria, Philippines, Sudan, Vietnam) decided on two strategic work streams:
i) the measurement and information stream (MIS) which aims to provide a framework for identifying and measuring efficiency, and
ii) the systematic priority setting stream (SPS) which aims to support countries in maximizing their stated health sector priorities within a given resource envelope.
There are three complementary products tied to these workstreams – Health Priority Setting: A Practitioner’s Handbook* and a Health Priority Setting and Resource Allocation Tool and Database** under the SPS stream – and this guide, which introduces readers to the concept of efficiency and gives guidance on how to assess efficiency in a practical setting. It includes a list of indicators most often used for tracking health system performance and several ‘fact sheets’ in Annex 2 meant to inform practitioners on how best to use, visualize, and interpret each indicator.
The development of this guide and the final list of proposed indicators was the result of an iterative process between technical facilitators and country participants over four in-person meetings, two webinars, and individual discussions with four countries testing out the validity and feasibility of the proposed approach.
Health Priority Setting: A Practitioner’s Handbook
Health Priority Setting and Resources Allocation Benchmarking Database
* https://www.jointlearningnetwork.org/resources/health-priority-setting-a-practitioners-handbook/.
** https://www.jointlearningnetwork.org/resources/health-priority-setting-and-resource-allocation-tool/
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE v
Acknowledgments iii
Genesis of the efficiency collaborative iv
Introduction 1
Concepts and principles of efficiency 7
A framework for measuring efficiency: a benchmarking plus approach 9
Why this approach? 13
From indicators to decision making 15 Conclusion 23
References 24
Annex 1: Indicator lists reviewed 25Annex 2: Indicator fact sheets 26
CONTENTS
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDEvi
List of boxes, figures and tables
BoxesBox A What are the main sources of inefficiency in health? 1
Box B What makes health spending particularly prone to inefficiency? 2
Box C A hypothetical application 11
Box D What potential sources are available to collect information for efficiency analysis? 16
Box E Kenya pilot 18
Exhibit E.1 Ratio of nurses per 100,000 population by county (latest year available) 19
Exhibit E.2 Maternal health coverage indicators, %, (2014–2018) 19
Exhibit E.3 Maternal service readiness index by county (2013) 19
Exhibit E.4 Maternal service readiness index by ownership and tier of care (2013) 19
Box F Malaysia pilot 20
Exhibit F.1 Diabetes prevalence (%), 2017 21
Exhibit F.2 Diabetes care cascade in Malaysia (absolute number and %), 2017 21
Exhibit F.3 Adherence to clinical practice guidelines for diabetes mellitus in Malaysia (%), 2017 21
FiguresFigure 1 Ways to benchmark 8
Figure 2 Life Expectancy vs. Government Health Expenditure per capita (2014) 9
Figure 3 The results chain or theory of change 10
Figure 4 Using data to inform decision making 15
TablesTable 1 Common indicators used to assess efficiency in hospital and pharmaceutical sub-systems 13
Table 2 Common indicators that can be used in an efficiency analysis 14
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE 1
Universal health coverage (UHC) will inevitably require governments to find additional resources for health; and efficiency is critical to achieving that objective. Health expenditure growth1 has exceeded gross domestic product (GDP) growth in nearly every country in the world over the past two decades. While there are several ways in which additional fiscal space for health can be met,2 the pressure to extract greater value for money from health spending is mounting as health systems increasingly absorb a larger share of government, employer, and household incomes. But perhaps more important than how much countries are spending on health is whether they are using their limited resources efficiently (World Bank, 2019). The 2010 World Health Report highlighted that between 20–40% of all resources spent on health are wasted. It identified human resources, hospitals, and pharmaceuticals as the biggest sources of inefficiency (Box A) which result in potential savings of ~USD 1.2 billion3 globally from these sources alone (Chisholm & Evans, 2010; Couffinhal & Socha-Dietrich, 2017). A more comprehensive description of sources of inefficiency can be found here.4
Box A: What are the main sources of inefficiency in health?
The main sources of inefficiency can be grouped into two types:
1. Inefficiencies arising from system-level resource allocation decisions – from poor planning or slow response to the changing health needs of the population:• Inappropriate or costly input/staff mix;• Inappropriate hospital size;• Sub-optimal deployment of health workers and facilities.
2. Inefficiencies that result from facility- or physician-level decisions linked to poor incentives or lack of accountability measures:• Inappropriate hospital admissions or length of stay;• Over-use of health care technology;• Sub-optimal quality of care and medical error;• Under-use of generic drugs;• Irrational use of drugs.
1 Health expenditure growth is driven by rapidly ageing populations, growing burdens of noncommunicable and chronic diseases, technological progress, and rising population expectations.
2 A parallel JLN Collaborative on Domestic Resource Mobilization for health covers four of these methods in greater detail. In short, they are: i) conducive macroeconomic conditions and increases in overall government revenue; ii) earmarking and specific commitment devices for health; iii) re-prioritization of health in the overall budget; and iv) access to health sector grants and donor aid.
3 Inflated to current 2019 USD from USD 965 billion in 2009.4 http://pubdocs.worldbank.org/en/616561494511379987/UHC-Efficiency-Background-Paper-V10-170419-1100.pdf0.
Introduction
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE2
Box B: What makes health spending particularly prone to inefficiency?
There is a common perception amongst Ministries of Finance and Budget and Planning that health is more inefficient than other sectors. However, there are many factors that are beyond the control of the Ministry of Health that impact health system efficiency:i. Uncertainty: The demand for health services is unknown. It is difficult to predict when and where people get
sick. This makes the planning and management of resources challenging. Public financial management (PFM) systems often restrict the ability to make mid-year adjustments to respond to the changing health needs of the population.
ii. Information asymmetry: The average patient generally has less information than providers about the need for and quality of health care. This leads to complicated principal-agent relationships in which a patient (the principal) delegates decision making authority to a better-informed doctor (the agent). The assumption is that the doctor will act in the best interests of the patient, but this can sometimes be at odds with the doctor’s own interest of wanting to maximize income. The end result is an increase in unnecessary (and total) health expenditures.
iii. Difficulty in linking inputs to outcomes: Health outcomes are more difficult to observe than outcomes in other sectors (e.g., a road being built or the number of students completing secondary education). In health, budgeting is often linked to inputs such as human resources or infrastructure (e.g., hospitals) and not to what is being purchased or delivered (e.g., health care services or treatment) and even less to health outcomes (e.g., health status or financial protection). Budget format and PFM rigidities also mean there is little flexibility to move funds around between line items.
iv. Fragmented financing: The health sector in low- and lower-middle income countries is marked by the presence of several specific programs (such as HIV/AIDS, malaria, tuberculosis, immunization, and family planning). Many of these programs are funded through external assistance. Often, this means that there are multiple donors present in the sector; health funds do not always flow through government budgets; and there is a lack of clarity on the full resource envelope. This not only poses a challenge for budgeting and planning but also leads to the duplication and waste of resources, the misalignment of resources with country priorities, and the use of parallel procurement, financial management, and monitoring and evaluation systems – placing a significant burden (financial and time) on the Ministry of Health.
This compromises Ministry of Health officials’ ability to argue for additional resources for the sector. The JLN’s Domestic Resource Mobilization Collaborative is producing a forthcoming messaging guide on how to make the case for health. It aims to facilitate discussion between Ministry of Health and Ministry of Finance officials highlighting the particularities of the health sector.
5 Stakeholders involved in the priority setting or benefit package selection process may include Ministries of Finance, Ministries of Health, health insurance agencies, market authorization and quality control agencies for drugs, national procurement agencies, professional medical associations, health technology agencies or academia, pharmaceutical lobbies, industry, the media, and the public.
Some inefficiencies will be easier to address than others. Several characteristics of the health sector make health spending particularly prone to inefficiency. Uncertainty in the demand for health, informational asymmetries between patients and providers, difficulties in linking inputs to outcomes, and fragmented sources of financing often lead to lower actual spending on health than the allocated budget (Box B). Improving allocative efficiency is generally more difficult, as many stakeholders outside the Ministry of Health are involved and decisions are often constrained by civil servant regulations and budget rigidities.5 The SPS produced a Handbook and a Resource Allocation Benchmarking Database to help tackle these issues. In comparison, addressing inefficiencies that result from facility- or physician-level decisions are more straightforward as accountable entities are easier to identify and fall directly under the supervision of the Ministry of Health (for example, regional hospital managers, hospital managers, hospital departments, or individual practitioners). It is the latter type of inefficiency that this guide focuses on.
While defining efficiency is relatively simple, measuring efficiency is challenging. The most basic definition of efficiency is maximizing outcomes relative to inputs. Therefore, the first key characteristic of efficiency analysis is that it requires information on both inputs and outputs. Efficiency analysis attempts to explain the unexplained variation across accountable entities – that is, why some individual providers, facilities, or health systems perform better than others. Therefore, a second key characteristic of efficiency analysis is that it involves making a comparison. The standard methods for economists to assess the efficiency of health care systems – stochastic frontier analysis (SFA) or data envelopment analysis (DEA) – empirically measure the productive efficiency of country health systems at the macro-level, or facilities at the micro-level. They take a production function approach, modeling the maximum attainable output achieved given a set of inputs. Yet, they struggle to produce actionable policy recommendations on how to improve efficiency. There are notable challenges to this approach. First, efficiency scores and rankings relative to the best performers are highly sensitive to model specification and the data used. Second, the results rarely help policy makers identify how they can improve efficiency, as efficiency scores – on their own – do little to tell a policy makers why the health system is inefficient.6 Finally, they are data intensive, time consuming, and require particular expertise. In low- and middle-income countries SFA and DEA are almost always carried out by external consultants as a one-off exercise, which does not help institutionalize routine monitoring of the efficiency of health care resources.
This guide proposes an alternative ‘benchmarking plus’ approach that can be readily used by practitioners and policy makers – the primary audience of the Efficiency Collaborative. The focus is less on empirical measurement and more on enabling the routine assessment of health system performance from an efficiency perspective. Benchmarking is simple to perform and easy to interpret; but on its own, it too will not tell policy makers how to improve efficiency. To be able to pinpoint the cause of variation, it will almost always be necessary to look at several indicators along the process of how inputs get transformed into particular outcomes and complement that with additional contextual knowledge in order to identify the appropriate policy action.
This guide gives a brief overview of concepts and principles of efficiency, and provides a framework for identifying and measuring efficiency in a practical way. It provides a list of indicators most often used for tracking health system performance and gives guidance on how they can be used to measure efficiency. It does not try to reiterate the work of the more comprehensive references from which it draws (World Bank, 2017) (Smith, 2012) (Cylus, et al., 2016) (Papanicolas & Smith, 2017) but rather attempts to operationalize the approach first suggested by (Smith, 2012)7 for low- and middle-income country settings.
6 Health systems have many goals – be it maximizing health status, financial protection, and/or patient satisfaction. There are also many sub-systems or sectors that contribute to the efficiency of the overall health system – such as preventive and promotive health, primary health care, the hospital sector, and the procurement, distribution, and use of pharmaceuticals and medical equipment. One part of the system may operate better than others.
7 Smith describes a “partial indicator” approach, which suggests benchmarking indicators across the entire results chain from inputs, to outputs, and outcomes.
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE 3
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE4
A BENCHMARKING PLUS APPROACH TO EFFICIENCY ANALYSIS
There are two key characteristics of efficiency analysis:1. Efficiency requires information on both inputs and outcomes
The most basic definition of efficiency is
MAXIMIZING
relative toOUTCOMES INPUTS
OUTCOMESINPUTS
If only inputs, the analysis is a resource mapping exercise that answers the question “What resources do I have and how are they distributed?”
If only outcomes, the analysis is an effectiveness analysis that answers the question “Does this intervention work?” or “Is it doing what it is supposed to do?”
Efficiency analysis
Looks at both inputs and outcomes and answers the question “Are we achieving maximum benefits at cheapest cost given our resources?”
2. Efficiency involves making a comparison
Benchmarking makes a comparison based on average performance, relative to the best performer, relative to a clinical norm or target, or relative to past performance.
Efficiency analysis attempts to explain the unexplained variation across accountable entities – that is why some individual providers, facilities, or health systems perform better than others.
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE 5
OUTPUTS OUTCOMESSERVICE
DELIVERYINPUTS$
Inefficiencies can occur at any point in the results chain which transforms inputs into outcomes
Benchmarking indicators along an entire results chain could help look at service-delivery bottlenecks through an efficiency lens.
Benchmarking is helpful at detecting emerging issues such as stagnating outcomes, outliers, or significant changes from one period to the next. However, on its own, it is not enough.
Policy makers have a range of policy levers to influence the behavior of accountable entities. Together, they affect how resources are allocated to different goods and services and how inputs are used to produce a given set of goods and services.
The ‘plus’ in benchmarking plus…
…allows more actionable policy recommendations
As specific issues emerge, additional indicators and key informant interviews may be necessary to better understand whether an issue is truly an inefficiency or not.
FINANCING PROVIDER INCENTIVES
REGULATIONORGANIZATION
A BENCHMARKING PLUS APPROACH TO EFFICIENCY ANALYSIS (CONTINUED)
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE6
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE 7
Concepts and principles of efficiency
The most basic definition of efficiency is maximizing outcomes relative to inputs.
• Inputs: Most often, people use inputs to refer to the costs, resources, or investments used to buy or produce health care inputs; and while financing for the health care system is one of the most fundamental inputs, others include health workforce (e.g., doctors, nurses, midwives, community extension workers); physical infrastructure (e.g., health care facilities, medical supply stores); drugs and medical products, equipment (e.g., MRI machines), and information – often the most overlooked input (e.g., data on civil registration and vital statistics, disease-specific registries, patient reported health outcomes, drug stocks).
• Outcomes can refer to the consequences, effectiveness, or benefits of service delivery interventions. In most health care systems around the world, the main outcomes of interest concern health status, financial risk protection, and public satisfaction. However, in practice, most efficiency metrics use intermediate outcomes or outputs. Outputs include information on the quantity (e.g., availability, access, coverage) and quality (e.g., diagnostic accuracy, treatment success rates) of the goods and services provided.
A key characteristic of efficiency analysis is that it requires information on both inputs and outcomes. If only inputs are examined, the assessment is essentially a resource mapping exercise. The question being asked is ‘what resources do I have and how are they distributed?’ If only outcomes are examined, the assessment is called an efficacy or effectiveness analysis.8 This asks the question ‘does this intervention work?’ or ‘is this intervention doing what it is supposed to do?’ Efficiency looks at both resources used (inputs) and effectiveness of interventions (outcomes), and assesses:
i) whether the maximum benefit to society is reached given the current use of inputs (‘doing the right things’), and
ii) whether the same benefits to society can be reached at cheaper cost (‘doing things in the right way’).
INPUTS
INTERMEDIATE OUTCOMES OR
OUTPUTS
Health care services
Fina
ncin
g
Quality
Financial protection
Health status
Public satisfaction
QuantityAvailability AccessCoverage
Equipment Physical infrastructure
Health workforce Drugs
Medical products Information
FINAL OUTCOMES
8 Efficacy refers to whether an intervention produces the expected result under ideal circumstances among a particular test group (such as a randomized control trial). Effectiveness refers to whether an intervention is beneficial to the wider public under ‘real world’ clinical settings.
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE8
Figure 1: Ways to benchmark
Relative to a clinical/ international norm or target
Before and after
Time
Effic
ienc
y m
easu
re
Out
puts
Inputs
Based on averageperformance
Out
puts
Relative to best performer
Inputs
Ideally, policy makers want to do the right things and do them in the right way. To improve efficiency, decision makers can either:
• Reorganize the current level of inputs (direct resources to different priorities), or adjust how services are delivered to increase benefits, which is sometimes referred to as allocative efficiency; or
• Find a way to produce the same amount of benefit but at a cheaper cost, which is sometimes referred to as technical efficiency.
A second key characteristic of efficiency analysis is that it involves making a comparison. To identify whether resources are being used efficiently, there needs to be some standard by which a comparison can be made. Benchmarking makes comparisons based on average performance, relative to the best performer, relative to a clinical norm or target, or relative to past performance.
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE 9
Figure 2: Life Expectancy vs Government Health Expenditure per capita (2014)
A framework for measuring efficiency: a benchmarking plus approach
Typically, the first step to benchmarking is to define the scope of the analysis. At the system level, the focus is on how health sector investments are contributing to overall health system objectives. At the sub-system or frontline-provider level, the focus might be on whether spending on broad health categories improves intermediate health outcomes, or on how facility resources are increasing service utilization.
Defining the scope of the analysis will help inform the choice of comparator. This could be countries of similar socioeconomic status or level of health system development; geographic regions or population groups within a country; or individual facilities or providers. Internal benchmarking is likely the most relevant for countries as understanding distribution across regions, hospitals, clinics, and population groups can help policy makers and program managers target resources more effectively.
While benchmarking can help identify outliers or raise flags, looking at discrete indicators does not tell policy makers how to improve efficiency. Figure 2 assesses efficiency at the health system level by plotting government health spending versus the average life expectancy of several countries. Countries can assess their efficiency relative to the average (red line). However, there are a number of drawbacks to this approach. First, it focuses on a single input (government health spending) and a single outcome (life expectancy) but many other inputs contribute to life expectancy, and the health system may have many other important outcomes of interest. Second, it does not tell us why for the same level of expenditure, one country is performing better than another given the same level of government spending (for example, Nigeria vs. Sudan). Information on the processes required to translate inputs into outcomes is missing; this is sometimes referred to as the ‘black box’ of service delivery.
Source: (World Bank, 2019)
100 250 500 1,000 2,500 10,000
GGHE Per capita, PPP
85
75
65
55
45
Life
exp
ecta
ncy
System level Frontline-provider levelSub-system level
INPUTS INPUTS INPUTSOUTCOME OUTPUT OUTPUT
Total health spending
Spending on primary health care
Number of specialists
Life expectancy
Immunization coverage
Number of admissions
Sudan
Philippines
Kenya
Malaysia
Vietnam
Indonesia Mongolia
Bangladesh
Ethiopia
Ghana
Nigeria
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE10
Inefficiencies can occur at any point in the results chain that transforms financing into outcomes. Instead of focusing on financing and final outcomes – the two ends of the results chain – it is more informative to benchmark indicators along an entire results chain to help pinpoint areas of inefficiency. Indicators that provide information on the management of resources, and on access to and quality of services are especially informative. Pinpointing areas of inefficiency will help narrow down the appropriate accountable entity for follow-up action – for example, regional hospital managers, hospital departments, or individual practitioners.
It is recommended to have discussions with key informants who can provide additional contextual knowledge and help get a sense of the relevance and extant of a particular source of inefficiency. They can help fill in the gaps when data is unavailable, incomplete, or unreliable, and be the starting conversation for thinking through appropriate policy actions and helping to determine which areas to tackle first. Addressing inefficiency is politically complicated, as there will be winners and losers in the process. This highlights the importance of having a transparent process around prioritization. For more information, see here,9 and here.10
Having identified the areas of inefficiency, policy makers have a range of policy levers to influence the behavior of accountable entities. We use the systems framework presented in listing financing, provider incentives, organization, and regulation as the key policy instruments to improve efficiency (Yip & Hafez, 2015).
i. Financing is one of the most powerful levers for influencing the efficient delivery of care, especially through the design of the benefits package. For example, a common practice for health insurance schemes in South and East Asian countries has been to focus on low probability, high-cost inpatient care over high probability, low-cost outpatient care, reasoning that catastrophic and impoverishing health expenditures are more likely to occur from expensive hospital services. But by covering only hospital care, the incentive is to admit patients even for simple conditions. Instead, offering free health promotion activities and primary health care can prevent the development of more serious and costly conditions. Another example of a financing lever is tiered cost-sharing arrangements that incentivize the use of lower-level facilities over tertiary hospitals (such as higher co-payments at hospitals versus primary health care facilities).
ii. Provider incentives – financial or otherwise – are another powerful policy lever that influences the efficiency (and quality) of care. For example, in prospective and bundled payments (such as capitation and case-based payments) greater risk is borne by the provider, incentivizing them to manage resources more efficiently. However, they may also incentivize other undesirable behaviors such as skimping on costs and avoiding chronic or sicker patients. For these reasons, most countries use a combination of payment methods to balance out the strengths and weaknesses of various methods. Non-financial provider incentives include relative social ranking, frequent interim feedback, and information availability. For more on provider payment methods please see here.11
iii. Organization covers a broad set of policies for managing and coordinating the delivery system. For example, for clinical interventions and diagnostic services, purchasing from both the private and public sector reinforces competition, and generally leads to lower prices and improved quality where effective regulatory and monitoring capacity are in place. But for goods and services with low-profit margins (for example, public health interventions), the public sector is recommended to ensure the socially efficient level of provision as these types of services are labor intensive, have low-profit margins, and are generally under-provided by the private sector. Also, policies that encourage disease prevention, gatekeeping, coordinated care, and a shift from hospital-based to ambulatory care help promote the cost-effective use of care.
iv. Regulation involves setting the rules, standards and operating guidelines within which the system is meant to operate. These may include medical and clinical protocols, national essential drug lists, procurement
Figure 3: The results chain or theory of change
OUTPUTS OUTCOMESSERVICE
DELIVERYINPUTS$
9 https://www.jointlearningnetwork.org/resources/health-priority-setting-a-practitioners-handbook10 https://www.jointlearningnetwork.org/resources/health-priority-setting-and-resource-allocation-tool11 https://www.jointlearningnetwork.org/resources/assessing-health-provider-payment-systems-a-practical-guide-for-countries-w/
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE 11
Box C: A hypothetical application
By benchmarking several indicators across the entire results chain, JLNotopia hopes to better understand why it underperforms on infant mortality compared to countries with similar government health spending.
To narrow down the cause of high infant mortality, JLNotopia needs to look at other indicators such as availability of providers and capacity to deliver immunization services.
The problem does not appear to be staff numbers.
Government health
expenditure per capita
Nurses and midwives per 1,000 people
Percent of fully vaccinated children
Infant mortality rate
Immunization specific readiness
index• Availability of guidelines• Number of trained staff• Equipment availability• Drug availability
OUTPUTS OUTCOMESSERVICE
DELIVERYINPUTS$
Nurses and midwives
100 250 500 1,000 2,500 10,000
GGHE Per capita, PPP
20
10
5
2.5
1.2
.6
Per 1
,000
birt
hs
100 250 500 1,000 2,500 10,000
GGHE Per capita, PPP
80
40
20
10
5
Rate
per
1,0
00 li
ve b
irths
Infant mortality
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE12
However, facilities have on average only 42% of all service-specific readiness requirements for immunization. Low availability of drug and vaccines and a lack of staff trained in child immunization are likely impacting service delivery.
As a result, the number of fully vaccinated children is low, and the infant mortality is high.
Primary health centers
By benchmarking several indicators across the entire results chain the analysis was able to pinpoint issues in the supply chain and in provider training. However, there is not enough information to know what specifically in the supply chain or training of providers is the issue. Taking supply chain as an example, the issue could be poor inventory management and capacity to forecast needed drugs, in which case the implementation of an inventory management or tracking system may be a solution. Or, the issue could be delays in the distribution of medicines, in which case outsourcing the distribution function to third parties might yield better outcomes. The ‘plus’ in the ‘benchmarking plus’ approach refers to the fact that it will almost always be likely to dig deeper. It might be necessary to look at additional indicators used to measure supply chain efficiency (for example, unit price of vaccines compared to international reference, time to process orders, number of days items were unavailable, etc.) and/or interview key informants at the national procurement agency, medical storage centers, and facilities to narrow down the right accountable entity for follow-up action. Only then can a conversation be had on how to reprioritize resources towards the right intervention based on a country’s prioritization process.
42%
42%
0% 20% 40% 60% 80% 100%
78%
31%
82%
77%
15%
13%
78%
78%
34%
34%
33%
32%
24%
5%
31%
22%
Overall
Guidelines for child immunizationStaff trained in child immunization
Cold box/vaccine carrier with ice packs
Refrigerator
Sharps container/safety box
Auto-disposable syringes
Temperature monitoring device in refrigerator
Adequate refrigerator temperature
Immunization cards
Immunization tally sheets
Staff and guidelines
Equipment availability
Measles vaccine
DPT-Hib+HepB vaccine
Oral polio vaccine
BCG vaccine
Pneumococcal vaccine
Rotavirus vaccine
IPV (Inactivated Poliovirus vaccine)
HPV (Human Papillomavirus)
Drug/vaccine availability
Children
% fully vaccinated* IMR per 1,000 live births
Residence
Urban 40.5 57
Rural 14.8 88
Wealth quintile
Lowest 4.4 90
Highest 56.1 51
Total 25.7 70.3
Note: *BCG, measles, and three doses each of DPT and polio vaccine
For a more in-depth discussion of various policy instruments to address inefficiencies, see here.12
guidelines for hospital equipment and drugs, quality control of drugs, manuals for claims management and audit (including fraud detection), certification and accreditation of facilities, licensing of the health workforce, price controls, and private sector service delivery.
12 http://pubdocs.worldbank.org/en/616561494511379987/UHC-Efficiency-Background-Paper-V10-170419-1100.pdf
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE 13
Why this approach?
The approach reflects the realities of many low- and middle-income countries attempting to measure health system efficiency. Every JLN deliverable is co-produced and driven by country participants over a series of face-to-face meetings. Technical experts help facilitate meetings to produce an end product that is based on topical expertise and country demand. The initial request was to produce a list of indicators to routinely assess the efficiency of health sector spending, especially in areas that are known to be major sources of inefficiency and consume the most resources, such as hospitals and pharmaceuticals (Table 1). However, many ‘true’ efficiency measures that combine inputs and outputs in a ratio, such as average consultations per doctor or unit cost per episode of care, are not readily available. Even if data is available, collection is difficult as datasets are often paper-based and housed in different units or administrative levels requiring a significant investment to standardize, aggregate, and clean data for analysis. In the absence of clear-cut efficiency metrics, benchmarking health system performance indicators along a results chain could help look at service delivery bottlenecks through an efficiency lens.
Table 1. Common indicators used to assess efficiency in pharmaceutical and hospital sub-systems
Pharmaceuticals Hospitals
• Pharmaceutical spending as % of total health expenditure (THE)
• Antibiotics spending as % of total pharmaceutical spending
• Unit price of drugs/medical consumables• Unit price compared to international reference
prices (especially for high-cost/use items)• Cost of freight/distribution to facilities• Order/use of high-cost items• High-use items• Number or % of expired items• Value of expired items • Stock-outs• Antibiotic prescription rates• Percent of encounters that end up in antibiotics
being prescribed• Time to process orders• Time to pay suppliers• Drug availability• Rate of anti-microbial resistance
• Spending by function (e.g., outpatient, inpatient, pharmaceutical, primary health care, public health or prevention, curative care) as % of government health expenditure (GGHE)
• Hospitals per 100,000 population, hospital bed density, bed occupancy rate
• General service readiness • Number of visits/admissions per day/month/year/per capita• Share of outpatient/inpatient• Diagnostic accuracy for tracer condition• Adherence to clinical guidelines• Number of incidents per 1,000 patient days
(e.g., center line-associated bloodstream infections, standardized infection ratio)
• Avoidable admissions for chronic obstructive pulmonary disease (COPD), asthma, hypertension, diabetes
• Referral rate• Average length of stay • Readmission rate• Caesarean section (C-section) rates
While many of the indicators considered are more commonly used to assess health system performance, any health system performance indicator can be viewed from an efficiency angle. For example, quality indicators such adherence to clinical guidelines, diagnostic accuracy, and treatment success rates would become the outputs, and a doctor’s visit or consultation the input. These could be benchmarked against a target (such as 100% diagnostic accuracy) relative to other providers, or compared to own past performance. In this way, indicators help fulfill the two key characteristics of efficiency analysis – first, it considers both inputs and outcomes and second, it involves making a comparison.
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE14
Table 2. Common indicators that can be used in an efficiency analysis
Financing (management)
Inputs (management)
Outputs (access)
Outputs (quality)
Outputs (risk factors)
Outputs (management)
Outcomes (health status)
Outcomes (financial protection)
• Total health expenditure (THE) as % of GDP or in per capita terms
• Government health expenditure (GGHE) as % of GDP, as % of budget, as % of THE, or in per capita terms
• Share of pre-paid/pooled spending as % of THE; out-of-pocket expenditure (OOP) as % of THE; external funds as % of THE
• THE, GGHE, OOP spending by function (e.g. outpatient, inpatient, pharmaceutical, primary health care, public health or prevention, curative care) as % of THE, as % of GGHE, as % of OOP expenditure, or in per capita terms
• Density of doctors, nurses, midwives
• Hospital density or hospital bed density
• Availability of basic facility infrastructure, essential medicines, equipment
• General services readiness or service-specific readiness for tracer condition(s)
• Service utilization
• Antenatal coverage
• Births attended by skilled health personnel
• Immunization coverage
• TB case detection
• TB treatment success rate
• Diagnostic accuracy for tracer condition(s)
• Adherence to clinical guidelines for tracer condition(s)
• Diabetes control
• % of incidents per 1,000 patient days
• Avoidable admissions
• Raised blood glucose/diabetes among adults
• Children under 5 who are stunted
• Tobacco use among persons aged 18+
• Caseload
• Absenteeism
• TB notification rate
• Average length of stay in hospital
• Bed occupancy rate
• Referral rate
• Hospital readmission rates
• C-section rates
• Expired drugs
• Stock-outs
• Claims ratio
• Budget execution rates
• Life expectancy at birth
• Under-five mortality rate
• Maternal mortality rate
• TB burden
• Mortality between 30-70 years age from cardiovascular diseases, cancer, diabetes or chronic respiratory diseases
• Catastrophic health expenditure
Note: Based on stated country priorities, indicators for immunization and maternal health were used as tracer conditions for assessing efficiency of primary health care, tuberculosis as a tracer condition for assessing efficiency of communicable diseases and a functioning surveillance system, and diabetes as a tracer condition for assessing efficiency of non-communicable diseases. Maternal health and diabetes can also be used as proxy conditions for assessing the efficiency of a functioning referral or integrated system.
The final list of indicators presented in this guide reflects a balance of current data availability in participating countries as well as some aspirational indicators. There are many health indicator lists that have been developed by international organizations, academics, advocacy groups, and others (including the JLN) grouped for different purposes – for tracking progress towards universal health coverage (UHC), assessing primary health care (PHC), ensuring hospital quality, and/or evaluating health provider payment systems (Annex 1). These existing lists were used as a starting point to compile a consolidated list of over 200 indicators. Indicators were organized according to the levels of the results chain (financing, inputs, outputs, and outcomes) and by domains of interest, such as access to care, quality, management, health status, risk factors, and financial protection. During the collaborative process, countries were asked to narrow down indicators based on:
i) countries’ health system priorities; ii) the availability of data in-country; and iii) the frequency of collection.
Indicators that were not frequently collected included those that covered the quality dimension, as these are more difficult to assess and time consuming to collect via facility surveys, vignettes, or patient record review. Hospital-centric indicators (e.g. average length of stay, bed occupancy rates, avoidable admissions, referrals) were also absent. As a result, the indicators that were initially chosen by participants focused only on financing, outputs related to the access dimension, and outcomes, as definitions were standardized and availability widespread. However, it was important to include indicators that may not be currently collected by health information systems in all participant countries. These are the indicators that will help countries move from assessing ‘coverage’ of key interventions towards assessing ‘effective coverage’. It may also help generate demand for collecting these indicators. Table 2 summarizes the final list of chosen indicators.
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE 15
Figure 4: Using data to inform decision making
From indicators to decision making
The intention is not to have countries track all indicators listed above. The choice of indicators should be driven by country priorities. Given the mix of country issues and varying level of data availability, countries should view Table 2 as a menu of indicators from which to choose. Figure 4 describes a process for using data to inform decision making. This section provides additional guidance on how to get started. It also includes case studies from two participant countries – Kenya and Malaysia – which have piloted the benchmarking plus approach.
Step 1 – Identifying strategic priorities: As mentioned in the introduction, the main sources of inefficiency have been widely documented in the literature; having discussions with key informants around the top ten can help get a sense of the relevance and extant of a particular source of inefficiency. This can be a starting point for countries thinking through which are the most important in their settings, and which they should tackle first. Addressing inefficiencies can be politically sensitive. Therefore, it is helpful to be transparent around the process. Health sector priorities listed in a National Health Strategic Plan, and routine reviews of health sector performance and budget implementation are natural entry points.
Step 2 – Selecting indicators: Having identified a health priority area, for example, maternal health, focus on choosing the right mix of inputs, outputs, and outcomes that clearly highlight the theory of change across the entire results chain – that is, indicators that are relevant for improving maternal health. For example, spending on maternal health, the number of nurses and midwives, the availability of minimum level of equipment and medicines for basic obstetric care, antenatal coverage, adherence to clinical guidelines for antenatal care activities, diagnostic accuracy for post-partum hemorrhage or hypertensive disorders such as eclampsia and preeclampsia, and maternal mortality.
Identify strategic priorities
Routinely report data
Assess options for
action
Select indicators
Collect data
Interpret the results
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE16
Step 3 – Collecting data: Policy makers tend to underestimate the availability of data and evidence in their countries. Routine administrative and health financing data are amongst the most valuable sources of information for conducting efficiency analysis. However, the lack of standardized reporting and accounting formats, the low prevalence of electronic health records and patient reported health outcomes, unreliable internet connectivity, and poor reporting compliance mean that significant effort and investment are needed to collect and process data. Until transactional data, administrative data, and/or health management and information systems are able to routinely generate dashboards or alerts for timely analysis and action by decision makers, countries can rely on several other less-frequent sources such as population and health facility surveys. In the meantime, it is helpful to document any data quality issues (e.g., completeness, validity, timeliness) so that an action plan to improve the quality and use of data for decision making can be implemented. Box D summarizes the main sources of data used in efficiency analysis.
Step 4 – Reporting data: Routine reporting provides the first indication that something may be wrong – reports are pushed to users who are then expected to ask the right follow-up questions to extract meaning from the report and take appropriate action. Ideally, the data reporting process is automated; however, standard reporting formats and regular review meetings to discuss areas of concern and action are also effective.
Step 5 – Interpreting the results: It is important to review and question the data to gain better insight. Fact sheets were developed to help guide practitioners in the use of indicators. For each indicator, the fact sheet provides a simple definition, identifies potential sources of data, suggests potential comparators, and gives guidance on benchmarking and interpretation. This includes suggestions on other indicators that might help provide additional insight. The fact sheets are presented in Annex 2.
Box D: What potential sources are available to collect information for efficiency analysis?
National health accounts, budget documents, and public expenditure reviews/public expenditure tracking surveys are used to assess how public resources – financial, human, in-kind – are allocated and managed and how they flow across administrative levels (e.g., from central government to frontline providers).
Routine or administrative data is generated as part of the operation of the health care system for purposes other than research – typically for administrative reasons or to support care. They include infectious disease surveillance data, medical records, hospital episode data, claims data, and supply chain data on medicines and commodities.
Health facility surveys are used to measure different dimensions of health facility performance supply-side readiness, provider training/knowledge, and provider effort.
Household or population surveys can be used to better understand health-seeking behavior and health status based on household characteristics (such as income, education, location, risk preferences, etc.).
Qualitative surveys or key informant interviews are also valuable instruments to better understand the political, organizational, and institutional environment in which decisions are made and interventions implemented.
The strengths and weaknesses of various data sources are summarized in Annex 1 of the SPS’s Health Priority Setting: A Practitioner’s Handbook.13
13 https://www.jointlearningnetwork.org/resources/health-priority-setting-a-practitioners-handbook/
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE 17
i. Indicators may not always have a recommended benchmark. Where there are clear recommendations, targets or clinical guidelines, benchmarking provides a nice signpost. For example, stock-out rates should ideally be zero, and C-section rates, of higher than 10% at the population level are not associated with reductions in maternal and newborn mortality. But where there are no set norms, benchmarks should be used with caution. For example, numerous health-spending targets have been recommended – 15% of government budgets, or 5% of GDP. While these targets can serve as global benchmarks, they are usually not helpful for determining appropriate levels of spending at the country level. Many countries spend more than these targets and have yet to provide a basic package of services to their population, while others spend less and achieve near universal levels of population coverage. This again highlights the importance of considering both inputs and outputs when interpreting indicators through an efficiency lens.
ii. It is important to make the right comparisons. Cross-country comparisons and national averages may mask regional variations or socioeconomic inequalities – especially in decentralized contexts. For most output indicators related to service delivery it is best to compare across facility types or providers within a country or region (internal benchmarking) as clinical protocols may vary from country to country. Looking at trends and past performance is also a helpful comparison – especially if there is no obvious comparator – as it informs policy makers on whether there is improvement, no change, or worsening performance.
iii. Inefficiency is difficult to attribute. As already mentioned, benchmarking is helpful for detecting emerging issues such as stagnating outcomes, outliers, or significant changes from one period to the next; however, on its own, it is not enough. Many other factors need to be considered such as time lags, data availability and quality, issues in costing, disease burden, and economies of scale. For example, the cost of service delivery can be much higher to achieve the same outcomes in remote areas that require significant outreach or in districts where the disease burden is more pronounced. As specific issues emerge, additional indicators and key informant interviews may be necessary to delve deeper and understand whether an issue is truly an inefficiency or not. The fact sheets in Annex 2 give additional guidance on benchmarking, choice of comparator, and interpretation for each indicator.
Step 6 – Assess options for action: Once inefficiencies have been identified, decision makers need an action plan to take recommendations forward. This requires having clear accountability processes in place and knowing the responsible unit or budget holder. Most countries reported that they typically collect hundreds of indicators as part of their regular monitoring and evaluation of the health sector – indicators that feed into annual budget reviews, annual program reports, key performance indicator dashboards, and statistical yearbooks. However, few were able to articulate whether and how this information was being used. These ‘reporting’ tools fall shy of the ‘analysis’ that might generate insight to guide policy and decision-making. While participant countries are already undertaking steps 1 through 4, only a few are assessing results, and even fewer are coming up with an action plan to address issues identified.
Ghana, Kenya, Malaysia, and the Philippines volunteered as pilot countries to test the validity and feasibility of the benchmarking plus approach. Boxes E and F summarize the Kenya and Malaysia case studies, respectively, applying the steps from identifying strategic priorities and choice of indicators to assessing options for action.
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE18
Box E: Kenya pilot
What was the main issue? Maternal mortality remains a challenge in Kenya, where the maternal mortality rate (MMR) has been stagnant since 1993, prompting the government to set reduction of MMR as a flagship program in the Kenya Health Sector Strategic Plan 2014–2018 (KHSSP). The aim was to reduce maternal mortality by half or decrease from 400 at baseline to 150 maternal deaths per 100,000 live births between 2014 and 2018. In the current KHSSP 2018–2022, the MMR baseline was 250 with a target reduction of 30% (Kenya Ministry of Health, 2018). Participants from Kenya’s JLN team applied the ‘benchmarking plus’ approach to identify inefficiencies in the service delivery chain.
What indicators were used and what were the findings? The team looked at the number of doctors and nurses, access to antenatal care (ANC) services and skilled birth attendance (SBA), and readiness to deliver basic obstetric care services as a proxy for quality of care. According to the 2018 Kenya Economic Report, Kenya has a significant shortage of health workers, exacerbated by poor distribution in the 47 counties in the country. There are 0.25 doctors, 1.82 registered nurses and 3.08 enrolled nurses per 1,000 population compared with the WHO recommendations of 3.0 doctors, 2.6 registered nurses and 5.4 enrolled nurses (Exhibit E.1) (KIPPRA, 2018). The percentage of women attending at least one ANC is high (82%), however, those attending four or more ANC visits was much lower at 49%. Data on SBA stood at 65% (Exhibit E.2). On average facilities only had 32% of items necessary to provide maternal health services with readiness highest at hospitals and private facilities; however, national averages mask wide variations (Exhibits E.3 and E.4).
What were the limitations? Was additional contextual information provided? Regular and up-to-date data for strategic planning, resource prioritization, and routine performance monitoring is not easily available. For example, administrative data does not routinely capture disaggregated data on the availability of personnel, medicines, and equipment, or utilization of services. However, population and facility surveys such as the Demographic and Health Surveys, the Service Availability and Readiness Assessment, and the Service Delivery Indicators were able to provide snapshots at a point in time. In addition, it was shared that only 36% of health facilities conducted audits for maternal deaths between July 2015–June 2016; only 43.7% of providers adhered to clinical guidelines; and only 44.6% of providers were able to manage maternal and neonatal complications based on clinical vignettes (World Bank, 2013). The team also shared that nurse strikes in December 2016 and June 2017 significantly affected the percentage of women receiving ANC and SBA services.
What recommendations are suggested? The main bottleneck happens early in the service delivery chain with workforce shortages. Therefore, next steps should focus on mapping out the distribution of providers, considering the labor market dynamics, and assessing the cost-effectiveness of various strategies, such as curriculum reform in medical schools and training, to increase the supply and retention of primary health care workers; introducing new and alternative cadres of health workers including options for telemedicine; monetary and non-monetary incentives for providers working in rural and primary health care practices; and regulatory measures that require workers to practice in remote areas for a certain number of years. In addition, increased supervision and oversight may enhance adherence to clinical protocols, but non-compliance may also be due to overworked personnel given shortages.
Source: Dr. Isabel Maina, Kenneth Munge, Dr. Mercy Mwangangi, Dr. Ahmed Omar, Esther Wabuge
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE 19
Exhibit E.1: Ratio of nurses per 100,000 population by county (latest year available)
Exhibit E.3: Maternal service readiness index by county (2013)
Exhibit E.2: Maternal health coverage indicators, %, (2014-2018)
Pregnant woman with at least 1 ANC visit
Delivery by skilled birth attendant
Pregnent women with 4 ANC visits
Postnatal care
2014 2015 2016 2017 2018
76.481.9
10.4–50.0
50.1–75.0
75.1–100.0
100.1–131.9
53.5
65.0
35.9
48.8
0
22.8
Source: (Kenya Ministry of Health, 2018)
32%
Source: (Government of Kenya, 2014)
Source: (Government of Kenya, 2014)
Exhibit E.4: Maternal service readiness index by ownership and tier of care (2013)
Public
Private
Hospitals
Health centers
Dispensaries
Medical clinics
0 10 20 30 40 50 60 70 80 90
6 33 47
27 43 58
17 28 42
28 62
7 30 69
20 56 84
70
60
50
40
30
20
10
0
Kir
inya
ga
Elge
yo-M
arak
wet
Mer
u
Bom
et
Bar
ingo
Kit
ui
Embu
Mac
hako
s
Ker
icho
Kili
fi
Ken
ya
Kaj
iado
Gar
issa
Mak
ueni
Man
dera
Mar
sabi
t
Bun
gom
a
Laik
ipia
Lam
u
Hom
a B
ay
Kis
ii
Kak
ameg
a
Mig
ori
Kia
mbu
Bus
ia
Kis
umu
Kw
ale
Isio
lo
Tran
s-N
zoia
Wes
t Po
kot
Nai
robi
Turk
ana
Mom
basa
Sam
buru
Nan
di
Waj
ir
Uas
in G
ishu
Tana
Riv
er
Nar
ok
Nye
ri
Nak
uru
Tha
raka
Nit
hi
Siay
a
Vih
iga
Nya
ndar
ua
Tait
a Ta
veta
Nya
mir
a
Mur
anga
Avg.
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE20
Box F: Malaysia pilot
What was the main issue? In 2017, Malaysia had the highest diabetes prevalence (16.9%) in the East Asia Pacific region (Exhibit F.1) (International Diabetes Federation, 2019). Compared to other countries, Malaysia has high rates of hospital admission for long-term complications of diabetes (such as premature heart disease and stroke, blindness, amputations, and kidney failure), which are 10–12 times costlier to treat than regular follow-up diabetes care (Mustapha, et al., 2017). In Malaysia, diabetes is also the fifth highest cause of death and disability combined (IHME, 2019). For this reason, the JLN Malaysia team sought to explore potential efficiency gains along the results chain that could lead to improved outcomes for diabetes.
What indicators were used and what were the findings? The team looked at indicators along the diabetes care cascade – diabetes prevalence, percent of diabetes patients who are diagnosed, and among diagnosed the percent of patients who are on treatment and who have their diabetes under control (Exhibit F.2). Findings from an annual clinical audit of type 2 diabetes patients provided additional insight into service delivery by looking at adherence to clinical guidelines at public health care facilities and the rate of complications (Exhibit F.3). The breakdown happens early in the continuum of care, as more than half (59%) of diabetes patients remain undiagnosed. While 86.5% of diagnosed patients receive annual Ac1 screening, fundoscopy and renal and liver function tests are less prevalent. Among diabetes patients 58.2% had good glycaemic control Ac1<8% while 1 in 5 patients had an HbAc1>10%. In terms of complications, in 2017, 11.5% of diabetes patients developed nephropathy, 9.6% developed retinopathy, 5.1% developed ischaemic heart disease, 1.5% developed cerebrovascular disease, 1.3% developed a diabetic foot ulcer, and 0.7% had a limb amputation.
What were the limitations? Was additional contextual information provided? Population-level data on the percent of diagnosed diabetes patients that are on treatment, the percentage whose treatment adheres to clinical guidelines, and the rate of complications was not available. In the absence of such data, the percentages from the National Diabetes Registry (NDR) Clinical Audit sample-based study were used. While the data from the NDR does not include patients seen in the private sector, the 2015 National Health and Morbidity Survey reported that only 15.1% and 3.6% of diabetes patients sought care in private clinics and private hospitals respectively, making these reasonable estimates for the broader population. The Malaysia team complimented the analysis with additional insight on how an increasing number of patients were referred to tertiary hospitals due to certain limitations at primary health care facilities in diagnosing and managing diabetes care. However, there is now a push to have more fundus cameras available and more Assistant Medical Officers certified and trained to carry out fundus exams at district-level primary health care clinics. Multiple efforts have been undertaken by various sectors to overcome the inadequate adherence to clinical guidelines for diabetes in health care facilities as well as the mismatch of knowledge on diabetes, poor compliance of medications, and aversion to insulin delivery methods among patients.
What recommendations are suggested? Based on these findings, recommendations are divided into four perspectives – healthcare providers, patients, policy makers, and researchers. For healthcare providers, additional training is required to improve the knowledge and skills to detect diabetes early on. This will also help promote best practices in managing diabetes and its related complications for improved quality of care. For patients, continued effort by the health promotion team to improve awareness and increase health literacy among diabetic patients. Policy makers should consider alternative approaches to service delivery such as financial incentives to the case managers who oversee long-term care especially among high-risk patients, community empowerment programs that conduct screening, or a complete patient-centered integrated care model that incorporates community resources, patient self-management, integrated information systems, and innovative delivery systems such as clinical decision support applications, mobile health devices, and bundled payment options between primary health care and hospital settings. Researchers should focus on the effectiveness, efficiency, and cost-effectiveness of diabetes interventions to provide evidence and assist policy makers in the prioritization and decision-making process.
The authors would like to thank the Director General of Health Malaysia for his permission to publish this article.
Source: Dr Nor Izzah Hj Ahmad Shauki, Dr Lee Kun Yun, Dr Abdul Hakim b Abdul Rashid, Dr Mohd Najib b Baharuddin, Dr Yussni bt Aris
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE 21
Exhibit F.1: Diabetes prevalence (%), 2017
Malaysia
Fiji
Vanuatu
Singapore
India
China
New Zealand
Philippines
Thailand
Timor
Indonesia
Vietnam
Japan
Australia
Myanmar
Cambodia
Laos
16.7
14.5
12.0
11.0
10.4
9.7
8.1
7.1
7.0
6.9
6.3
6.0
5.7
5.1
4.6
4.0
4.0
0 5 10 15 20
Source: International Diabetes Federation (https://idf.org/our-network/regions-members/western-pacific/members/108-malaysia.html)
Note: *Based on sample-based clinical audit of diagnosed diabetes patients. http://www.iku.gov.my/images/IKU/Document/REPORT/nhmsreport2015vol2.pdf
Sources: International Diabetes Federation (https://idf.org/our-network/regions-members/western-pacific/members/108-malaysia.html) and *National Diabetes Registry of Malaysia
4,000,000
3,500,000
3,000,000
2,500,000
2,000,000
1,500,000
1,000,000
500,000
0
1009080706050403020100
%
Abs
olut
e nu
mbe
r
16.9
40.6
86.5
58.2
Prevalence Diagnosed Among diagnosed, on treatment*
Among diagnosed,
under control (HbA1c<8%)*
Exhibit F.2: Diabetes care cascade in Malaysia (absolute number and %), 2017
Exhibit F.3: Adherence to clinical practice guidelines for diabetes mellitus in Malaysia (%), 2017
Note: Sample size 160,075
Source: National Diabetes Registry of Malaysia
Had exam in past year %Patients who underwent HbA1c 86.5
Patients who underwent fundoscopy 51.8
Patients who underwent foot exam 76.3
Patients who underwent proteinuria 67.0
Patients who underwent urine microalbumin 52.9
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE22
The findings from the pilot exercise highlighted challenges to applying the proposed framework as well as lessons learned.
Challenges:• Availability of data: A key limiting factor was the availability of data at the right level of disaggregation to make meaningful comparisons (e.g., by geographical region, across population groups) – especially for decentralized countries. It was also difficult to get data across different parts of the health system as information is held in different departments or programs within the Ministry of Health or was spread out across several Ministries/Agencies or levels of government. However, when pressed, all pilot countries were able to bring in anecdotal evidence or refer to country studies to help explain their data.
• Defining the scope of the efficiency analysis: All four countries attempted to present country data for the entire list of priority indicators. This made it harder to focus on any one area and to delve deeper into identifying where along the results chain inefficiencies might lie.
• Interpreting the data: There was little interpretation of what the data might mean and/or packaging the information into a policy-relevant story line.
Lessons learned:• Focus the scope of the analysis: That is, looking at hospital efficiency, pharmaceuticals or a specific tracer
condition within primary health care, such as maternal health or tuberculosis, where relevant indicators were more likely to be available under responsible units or budget holders.
• Additional indicators should be analyzed when specific issues emerge in the course of routine monitoring: This includes going beyond indicators and drawing on practitioner’s valuable knowledge of the sector to help fill in the gaps in the storyline when data is not available. This will help with the interpretation of findings and the assessment of options for further action.
• Efficiency analysis needs to be embedded into the formal decision-making process, not a one-off exercise. Many health system performance indicators are already regularly collected and reported as part of the routine monitoring and evaluation of the health sector. Building on existing processes and applying an efficiency lens to reviewing indicators that question why some regions, hospitals, or providers perform better than others is the first step to institutionalizing efficiency analysis.
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE 23
Conclusion
There are inefficiencies in all health care systems, the scale of which can be significant. But while the main sources of inefficiency have been widely discussed in the literature, what is less available is a practical approach that policy makers can use to inform change. Stochastic frontier analysis (SFA) or data envelopment analysis (DEA) are the standard methods for economists to assess efficiency. However, these methods are often time consuming, data intensive, and require particular expertise. In addition, they do not necessarily identify the truly efficient behavior, nor do they tell policy makers how to improve efficiency.
This guide presents an alternative ‘benchmarking plus’ approach that can be readily used by practitioners and policy makers. In setting up routine monitoring systems, countries should ideally benchmark a few common efficiency metrics in areas that consume the most resources or are prone to inefficiency such as caseload, average length of stay, avoidable readmissions, unit cost per episode of care, stock-out rates, etc. However, if not available, benchmarking indicators along the process of how inputs are used to deliver services, which produce outputs that contribute to outcomes, can help pinpoint service delivery bottlenecks for further inspection. Discussion with key informants provides additional contextual information on the relevance and extant of identified problems. Only then can appropriate recommendations be considered based countries’ prioritization criteria and resource constraints.
However, even this simpler approach presented challenges for participating countries given data limitations. Almost all participating countries lacked indicators to measure those aspects of service delivery that are related to quality. Instead, available outputs mostly focused on availability, access, and coverage. But quality metrics are needed to help monitor countries’ progress from assessing ‘coverage’ of key interventions towards assessing ‘effective coverage’. In 2018, poor quality of health services contributed to more amenable deaths (5 million) than non-utilization of services (3.6 million) across lower-middle income countries. Hospital quality metrics are especially important given recent findings that 10% of hospitalized patients acquire infections during their stay; clinical guidelines are followed in less than 45% of cases; and harmful medical errors and preventable complications account for 15% of hospital costs (Kruk, et al., 2018). All of these represent sources of inefficiency because resources are spent on services that produce reduced or no added health benefit.
Countries cannot improve what they cannot measure; however, other resources are likely available to support efficiency analysis. In the absence of regular reporting on key efficiency metrics collected by health management and information systems, almost all countries will be able to fill in the gaps using past studies, ad hoc population and facility surveys, and interviews with key informants, as was demonstrated by the Kenya and Malaysia case studies. For this reason, a benchmarking plus approach is recommended while investments for foundational healthcare analytic systems are made.
Fortunately, countries are increasingly digitizing data which will help fill in these gaps and increase the potential for efficiency analysis. The demand for data-driven decision-making and the rising pressure to reduce healthcare spending while improving patient outcomes is driving the growth of healthcare analytics – a USD 14 billion industry globally that is projected to reach USD 50.5 billion by 2024. Currently, low- and lower-middle income countries still struggle to produce basic descriptive analytics – a precursor to more advanced predictive and prescriptive analytics. Technological advances in text and image storage, cloud computing, and big data are paving the way to extract more value from data by integrating electronic medical records, patient monitoring systems, claims data, and other core administrative data. However, the biggest limiting factor is the broad range of skills necessary to develop a healthcare analytics team that includes information technology specialists to deal with hardware, software, and connectivity issues; clinical coders and data management specialists to enable standardization and interoperability; and policy makers, health care providers, and data scientists to develop applications that help answer policy-relevant questions. All must work together.
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE24
References
Chisholm, D. & Evans, D. B., 2010. Improving health system efficiency as a means of moving towards universal coverage, Geneva: The World Health Organization.
Couffinhal, A. & Socha-Dietrich, K., 2017. Ineffective spending and waste in health care systems: frameworks and findings. In: Tackling wasteful spending on health. Paris: OECD Publishing, pp. 17–59.
Cylus, J., Papanicolas, I. & Smith, P. C., 2016. Health system efficiency: how to make measurement matter for policy and management. London: European Observatory on Health Systems and Policies.
Drummond, M. F. et al., 2015. Methods for economic evaluation of health care programmes, 4th Edition. Oxford: Oxford University Press.
Government of Kenya, 2014. Kenya Service Availability and Readiness Assessment Mapping (SARAM), Nairobi: Ministry of Health.
Hafez, R., Somanathan, A. & Wilson, D., 2016. Raising funds for health - Background paper for 1st universal health coverage financing forum. Washington DC: The World Bank.
IHME, 2019. Global Burden of Disease, s.l.: http://www.healthdata.org/malaysia.
International Diabetes Federation, 2019. Diabetes Atlas, s.l.: https://diabetesatlas.org/across-the-globe.html.
Kenya Institute for Public Policy Research and Analysis (KIPPRA), 2018. Kenya Economic Report 2018: Boosting Investments for Delivery of the Kenya Vision 2030, Nairobi.
Kenya Ministry of Health, 2018. Kenya Health Sector Strategic Plan (KHSSP), July 2018 - June 2022, Nairobi: Ministry of Health.
Kruk, M. E. et al., 2018. High quality health systems in the sustainable development goals era: time for a revolution. Lancet Glob Health , Volume 6, pp. e1196–252.
Morgan , D., 2010. Chapter 1: how much is too much? value for money in health spending. In: Value for money in health spending. Geneva: OECD, pp. 21–42.
Mustapha, F. I. et al., 2017. What are the direct medical costs of managing Type 2 Diabetes Mellitus in Malaysia. Med J Malaysia, 72(5), pp. 271–277.
Papanicolas, I. & Smith, P. C., 2017. Theory of system level efficiency in healh care. In: Encyclopedia of Health Economics. San Diego: Elsevier, pp. vol. 3: 386–394.
Smith, P. C., 2012. What is the scope for health system efficiency gains and how can they be achieved?. Eurohealth Observer, pp. 18(3)3–6.
Tandon, A. & Cashin, C., 2010. Assessing Public Expenditure on Health From a Fiscal Space Perspective, Washington, DC: The World Bank.
UNFPA, 2014. The state of the world’s midwifery, https://www.unfpa.org/data/sowmy/KE: UNFPA.
WHO, 2019. Global Health Observatory Database, Geneva: WHO.
World Bank, 2013. Kenya Service Delivery Indicators, Washington, DC: World Bank.
World Bank, 2017. Greater efficiency for better health and financial protection: Background paper for 2nd Annual UHC Financing Forum, April 20-21, 2017, Washington DC: The World Bank.
World Bank, 2019a. Health Equity and Financial Protection Database, Washington, DC: The World Bank.
World Bank, 2019b. World Development Indicators, Washington, DC: The World Bank.
Yip, W. & Hafez, R., 2015. Reforms for improving the efficiency of health systems: lessons from 10 country cases, Geneva: The World Health Organization.
Yip, W., Hafez, R. & Chi, Y.L., 2014. Strategic purchasing for universal health coverage: policy brief for the Eastern-Mediterranean region, Geneva: The World Health Organization.
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE 25
Annex 1 – Indicator lists reviewed
The World Health Organization-World Bank UHC Measurement FrameworkThe WHO-World Bank UHC Measurement Framework tracks eight core health service coverage indicators chosen because they involve interventions for which every individual should benefit – mainly maternal and child health, and infectious disease control. Not currently included are tracers for non-communicable diseases. In addition, the framework also tracks two measures of financial risk protection, namely out-of-pocket spending as a share of total consumption and health expenditures that drive households beneath the poverty line. These indicators are generally collected from frequent household surveys such as the demographic and health survey, the multiple indicator cluster survey, and/or household consumption surveys. The limited number of indicators and their relative straightforward measurability allow for comparability across countries making them a useful common framework for tracking progress towards UHC globally.For more information: https://www.who.int/healthinfo/universal_health_coverage/report/2015/en
The World Health Organization’s Global Reference List of 100 Core Health IndicatorsThis list is a more comprehensive (though not exhaustive) set of indicators organized by four domains i) health status, ii) risk factors, iii) service coverage, and iv) health systems. In addition to a broader range of maternal and child health, and infectious disease control indicators, the global reference list also includes a range of risk factors associated to non-communicable diseases, and outcomes that act as proxies for access and quality of care. Unlike the UHC monitoring framework that looks at a limited section of the results chain (i.e. outputs and outcomes), this expanded list allows policy makers to identify weaknesses along the entire results chain. In fact, they also provide the same list of indicators re-organized by sections of the results chain. However, the focus remains on areas associated with the Millennium Development Goals agenda.For more information: https://www.who.int/healthinfo/indicators/2018/en
Primary Health Care Performance InitiativeThe PHCPI focuses exclusively on primary health care. The 36 Vital Signs indicators align and complement the Global Reference List of 100 Core Health Indicators by focusing on important service delivery processes that are crucial for achieving UHC and other global priorities. It looks at indicators across the entire results chain and goes one step further in collecting information related to provider knowledge (e.g., diagnostic accuracy), effort (e.g., caseload), and quality of service delivery (e.g., continuity of care) – indicators that are more complex to produce requiring vignettes, and health facility surveys that are less commonly produced and standardized limit comparability across countries and over time. These might be more useful for a within country comparison (e.g., regional variation).For more information: https://improvingphc.org/phcpi-core-indicators
Organization for Economic Cooperation and Development Health at a Glance DashboardThe OECD dashboard of indicators is meant to provide an overview of health system performance – organized around similar domains of health care resources, health status, and risk factors. However, unlike the WHO and PHCPI lists, the focus of the access to and quality of care indicators is on non-communicable diseases – a reflection of these countries’ epidemiology and level of health system development. The dashboard is used for cross-country comparisons.For more information: https://www.oecd.org/health/health-systems/health-at-a-glance-19991312.htm
Commonwealth Fund Scorecard for State Health System PerformanceThe 44 indicators on the Commonwealth Fund Scorecard – used to track performance over time and rank US states relative to one another – fall into four domains: access and affordability, prevention and treatment, avoidable hospital use and cost, and healthy lives. Similar to the OECD dashboard, indicators are more focused on hospital use and care, non-communicable diseases and associated risk factors – again a reflection of the US’s disease burden and health care system.For more information: https://www.commonwealthfund.org/publications/fund-reports/2018/may/2018-scorecard-state-health-system-performance
Joint Learning Network’s Indicators for Monitoring Provider Payment Methods Finally, the JLN developed a list of indicators specifically geared towards monitoring provider payment methods. Given the diversity of health care systems and provider payment methods across countries, these would ideally be used for within-country comparison at the facility/provider level.For more information: https://www.jointlearningnetwork.org/resources/data-analytics-for-monitoring-provider-payment-toolkit
Annex 2 – Fact sheets
Financing (management) • Total health expenditure as % of GDP • Government health expenditure (GGHE) as % of GDP, as % of budget, as % of THE, or in
per capita terms •Healthexpenditurebysourceoffinancing • Current health expenditure (CHE) by function
Inputs (management) • Health worker density • Hospital bed density
Outputs (access) • Antenatal care • Skilled birth attendance
Outputs (quality) •TBcasedetection,TBnotificationrate,andTBtreatmentsuccessrate •Generalservicereadiness;Servicespecificreadiness • Diagnostic accuracy for a tracer condition • Adherence to clinical guidelines • Diabetes control • Avoidable hospital admissions for primary care sensitive conditions
Outputs (risk factors) • Children under 5 who are stunted • Tobacco use among persons aged 18+
Outputs (management) • Caseload • Absenteeism • Average length of stay (ALOS) vs. Bed occupancy rate (BOR) • Hospital re-admission rate • Caesarean section (C-section rates) • Expired drugs • Drug stockouts • Claims ratio • Budget execution rates (BERs)
Outcomes (health status) • Life expectancy at birth •Under-fivemortality • Maternal mortality rate • TB burden
Outcomes (financial protection) • Catastrophic health expenditure
Health Financing Technical Initiative: The Efficiency CollaborativeIndicator Fact Sheets
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE26
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE 27Pote
ntia
l sou
rce
of d
ata:
Abo
ut 9
5 co
untr
ies
have
pro
duce
d fu
ll na
tiona
l hea
lth
acco
unts
. The
W
HO
’s G
loba
l Hea
lth
Expe
nditu
re
Dat
abas
e al
so h
as in
tern
atio
nally
co
mpa
rabl
e da
ta o
n he
alth
spe
ndin
g fo
r clo
se t
o 19
0 co
untr
ies
from
200
0 to
201
6.
In-c
ount
ry s
ourc
es in
clud
e na
tiona
l he
alth
acc
ount
repo
rts,
pub
lic
expe
nditu
re re
port
s, s
tatis
tical
ye
arbo
oks
and
othe
r per
iodi
cals
, an
d re
port
s an
d da
ta p
rovi
ded
by c
entr
al s
tatis
tical
offi
ces
and
min
istr
ies.
ht
tp://
apps
.who
.int/
nha/
data
base
/Sel
ect/
Indi
cato
rs/e
n
Benc
hmar
king
gui
delin
esA
num
ber o
f hea
lth
spen
ding
ta
rget
s ha
ve b
een
set:
5%
of G
DP
(alle
ged
WH
O);
15%
of g
over
nmen
t sp
endi
ng (A
buja
dec
lara
tion)
. It
is
also
sug
gest
ed t
o be
nchm
ark
agai
nst
past
per
form
ance
and
rele
vant
re
gion
al/in
com
e co
untr
y co
mpa
rato
rs
and/
or a
vera
ges.
Whi
le t
hese
tar
gets
can
ser
ve
as g
loba
l ben
chm
arks
, the
y ar
e us
ually
not
hel
pful
for d
eter
min
ing
appr
opria
te le
vels
of s
pend
ing
at t
he
coun
try
leve
l - e
spec
ially
whe
re t
otal
he
alth
exp
endi
ture
is d
riven
by
OO
P sp
endi
ng. I
nste
ad, i
t is
mor
e us
eful
to
com
pare
aga
inst
wha
t is
fisc
ally
fe
asib
le, w
hat
the
coun
try
is t
ryin
g to
ac
hiev
e, a
nd h
ow m
uch
is n
eede
d to
co
ver a
n es
sent
ial b
enefi
t pa
ckag
e.
Sour
ce: W
HO
(201
9). G
loba
l Hea
lth E
xpen
ditu
re D
atab
ase
and
Wor
ld B
ank
(201
9). W
orld
Dev
elop
men
t Ind
icat
ors.
(h
ttps
://da
taba
nk.w
orld
bank
.org
/sou
rce/
wor
ld-d
evel
opm
ent-
indi
cato
rs)
Percentage (%)
-
0510152025
Bang
lade
sh
Indo
nesi
a
Nig
eria M
alay
sia
Ethi
opia
Phili
ppin
es
Gha
na
Mon
golia
Keny
a
Suda
n
Vie
tnam
Hig
h in
com
e
Upp
er m
iddl
e in
com
e
Low
er m
iddl
e in
com
e
Low
inco
me
Tota
l hea
lth e
xpen
ditu
re a
s %
of G
DP
Sim
ple
defin
ition
: All
spen
ding
on
heal
th c
are
good
s an
d se
rvic
es a
nd c
apita
l goo
ds. I
n th
e N
atio
nal H
ealt
h A
ccou
nt t
erm
inol
ogy
this
is re
ferr
ed t
o as
the
sum
of C
urre
nt H
ealt
h Ex
pend
iture
s (C
HE)
and
Cap
ital
Hea
lth
Expe
nditu
re (H
K) w
here
: CH
E is
the
sum
of a
ll go
vern
men
t sc
hem
es in
clud
ing
com
puls
ory
cont
ribut
ory
heal
th c
are
finan
cing
sch
emes
(HF.
1), a
ll vo
lunt
ary
heal
th c
are
paym
ent
sche
mes
(i.e
. vol
unta
ry o
r priv
ate
insu
ranc
e) (H
F.2)
, hou
seho
ld
out-
of-p
ocke
t (O
OP)
(HF.
3), r
est
of t
he w
orld
fina
ncin
g sc
hem
es (i
.e. e
xter
nal d
onor
fund
ing)
(HF.
4), a
nd o
ther
not
els
ewhe
re c
lass
ified
(HF.
nec)
and
HK
is d
efine
d at
the
acq
uisi
tion
of p
rodu
ced
asse
ts; t
hat
is
asse
ts in
tend
ed fo
r use
in t
he p
rodu
ctio
n of
oth
er g
oods
and
ser
vice
s.
Freq
uenc
y of
dat
a co
llect
ion:
Ann
ual.
Sugg
este
d co
mpa
rato
r: A
cros
s co
untr
ies;
if c
ount
ry is
dec
entr
aliz
ed s
ubna
tiona
l com
paris
ons
are
high
ly in
form
ativ
e; p
ast
perf
orm
ance
/tre
nd d
ata.
Mos
t re
cent
ly, t
he t
hird
edi
tion
of t
he D
isea
se C
ontr
ol P
riorit
ies
init
iati
ve (D
CP3)
est
imat
ed t
he t
otal
cos
t pe
r pe
rson
for
sust
aini
ng a
hig
h pr
iorit
y pa
ckag
e (H
PP) a
nd a
n es
sent
ial u
nive
rsal
hea
lth
cove
rage
pac
kage
(EU
HC)
at
80%
co
vera
ge w
ould
be
US
$42
and
$72
resp
ecti
vely
in lo
w in
com
e (L
IC) c
ount
ries
and
$58
and
$110
res
pect
ivel
y in
low
er-m
iddl
e in
com
e (L
MIC
) cou
ntrie
s.
Tota
l hea
lth
expe
ndit
ure
as a
sha
re o
f GD
P (%
), 20
16
Fina
ncin
g (m
anag
emen
t)
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE28 Gov
ernm
ent h
ealth
exp
endi
ture
(GG
HE)
as
% o
f GD
P,
as %
of b
udge
t, as
% o
f TH
E, o
r in
per c
apita
term
s
Gov
ernm
ent
heal
th s
pend
ing
per c
apit
a, 2
016,
in U
S$
Not
e: U
sing
the
WH
O’s
Glo
bal H
ealth
Exp
endi
ture
dat
abas
e th
e y-
axis
cor
resp
onds
to H
F.1
or G
GH
E-D
Sour
ce: S
ourc
e: W
HO
(201
9). G
loba
l Hea
lth E
xpen
ditu
re D
atab
ase
and
Wor
ld B
ank
(201
9). W
orld
Dev
elop
men
t Ind
icat
ors
( h
ttps
://da
taba
nk.w
orld
bank
.org
/sou
rce/
wor
ld-d
evel
opm
ent-
indi
cato
rs)
Government health expenditure per capita, US$
LOW
INCO
ME
LOW
ER M
IDD
LE
INCO
ME
UPP
ER M
IDD
LE
INCO
ME
HIG
H IN
COM
E
100
110100
1000
1000
0
1000
1000
0
EUH
C (L
MIC
) 110
$
HPP
(LIC
) 42$
Ethi
opia
Bang
lade
shN
iger
ia
Keny
aSu
danPh
ilipp
ines
Indo
nesi
aM
ongo
lia
Mal
aysi
a
Gha
naVie
tnam
Sim
ple
defin
ition
: All
spen
ding
on
heal
th b
y go
vern
men
t sc
hem
es a
nd c
ompu
lsor
y co
ntrib
utor
y he
alth
car
e fin
anci
ng s
chem
es.
Freq
uenc
y of
dat
a co
llect
ion:
Ann
ual.
Sugg
este
d co
mpa
rato
r: A
cros
s co
untr
ies;
if c
ount
ry is
dec
entr
aliz
ed s
ubna
tiona
l com
paris
ons
are
high
ly in
form
ativ
e; p
ast
perf
orm
ance
/tre
nd d
ata.
Pote
ntia
l sou
rce
of d
ata:
Abo
ut 9
5 co
untr
ies
have
pro
duce
d fu
ll na
tiona
l hea
lth
acco
unts
. The
W
HO
’s G
loba
l Hea
lth
Expe
nditu
re
Dat
abas
e al
so h
as in
tern
atio
nally
co
mpa
rabl
e da
ta o
n he
alth
spe
ndin
g fo
r clo
se t
o 19
0 co
untr
ies
from
200
0 to
201
6. In
-cou
ntry
sou
rces
incl
ude
natio
nal h
ealt
h ac
coun
t re
port
s,
publ
ic e
xpen
ditu
re re
port
s, s
tatis
tical
ye
arbo
oks
and
othe
r per
iodi
cals
, and
re
port
s an
d da
ta p
rovi
ded
by c
entr
al
stat
istic
al o
ffice
s an
d m
inis
trie
s.ht
tp://
apps
.who
.int/
nha/
data
base
/Sel
ect/
Indi
cato
rs/e
n
Benc
hmar
king
gui
delin
esA
num
ber o
f hea
lth
spen
ding
tar
gets
ha
ve b
een
set:
5%
of G
DP
(alle
ged
WH
O);
15%
of g
over
nmen
t sp
endi
ng
(Abu
ja d
ecla
ratio
n). I
t is
als
o su
gges
ted
to b
ench
mar
k ag
ains
t pa
st
perf
orm
ance
and
rele
vant
regi
onal
/in
com
e co
untr
y co
mpa
rato
rs a
nd/o
r av
erag
es.
Whi
le t
hese
tar
gets
can
ser
ve
as g
loba
l ben
chm
arks
, the
y ar
e us
ually
not
hel
pful
for d
eter
min
ing
appr
opria
te le
vels
of s
pend
ing
at t
he
coun
try
leve
l - e
spec
ially
whe
re t
otal
he
alth
exp
endi
ture
is d
riven
by
OO
P sp
endi
ng. I
nste
ad, i
t is
mor
e us
eful
to
com
pare
aga
inst
wha
t is
fisc
ally
fe
asib
le, w
hat
the
coun
try
is t
ryin
g to
ac
hiev
e, a
nd h
ow m
uch
is n
eede
d to
co
ver a
n es
sent
ial b
enefi
t pa
ckag
e.
Mos
t re
cent
ly, t
he t
hird
edi
tion
of t
he D
isea
se C
ontr
ol P
riorit
ies
initi
ativ
e (D
CP3)
est
imat
ed t
he t
otal
cos
t pe
r per
son
for s
usta
inin
g a
high
prio
rity
pack
age
(HPP
) and
an
esse
ntia
l uni
vers
al h
ealt
h co
vera
ge p
acka
ge (E
UH
C) a
t 80
% c
over
age
wou
ld b
e $U
S42
and
$72
resp
ectiv
ely
in lo
w in
com
e (L
IC) c
ount
ries
and
$58
and
$110
resp
ectiv
ely
in lo
wer
-mid
dle
inco
me
(LM
IC) c
ount
ries.
The
se e
stim
ates
sug
gest
tha
t cu
rren
t G
GH
E w
ill n
eed
to d
oubl
e or
trip
le t
o fin
ance
HPP
or E
UH
C pa
ckag
es in
LIC
/LM
IC.
Fina
ncin
g (m
anag
emen
t)
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE 29
Hea
lth
expe
ndit
ure
by s
ourc
e (2
016)
Out
-of-
Pock
etO
ther
Gov
ernm
ent
Exte
rnal
Perc
enta
ge
71.2
21.7
31.1
40.8
39.9
25.9
43.1
57.8
Hea
lth e
xpen
ditu
re b
y so
urce
of fi
nanc
ing
Sim
ple
defin
itio
n: S
hare
of
gove
rnm
ent
spen
ding
, vol
unta
ry a
nd/o
r pr
ivat
e H
I, O
OP,
res
t of
the
wor
ld a
s %
of T
HE.
Th
e ov
eral
l cur
rent
spe
ndin
g on
hea
lth in
clud
e ex
pend
iture
s of
gov
ernm
ent s
chem
e (H
F.1)
, vol
unta
ry s
chem
es (H
F.2)
, out
of p
ocke
t spe
ndin
g (H
F.3)
, res
t of t
he w
orld
(HF.
4) a
nd o
ther
not
els
ewhe
re c
lass
ified
ex
pend
iture
s (H
F.ne
c).
Freq
uenc
y of
dat
a co
llect
ion:
Ann
ual.
Sugg
este
d co
mpa
rato
r: A
cros
s co
untr
ies;
if c
ount
ry is
dec
entr
aliz
ed s
ubna
tion
al c
ompa
rison
s ar
e hi
ghly
info
rmat
ive;
pas
t pe
rfor
man
ce/t
rend
dat
a.
By c
onst
ruct
ion,
the
hig
her t
he s
hare
of g
over
nmen
t an
d/or
pre
paid
spe
ndin
g th
e lo
wer
the
sha
re o
f OO
P; a
nd t
he lo
wer
the
sha
re o
f OO
P sp
endi
ng, t
he b
ette
r. G
loba
lly, c
ount
ries
that
are
clo
sest
to
atta
inin
g un
iver
sal h
ealt
h co
vera
ge (U
HC)
ha
ve O
OP
spen
ding
leve
ls t
hat
are
less
tha
n 15
-20%
of t
otal
hea
lth
spen
ding
.
Pote
ntia
l sou
rce
of d
ata:
Abo
ut 9
5 co
untr
ies
have
pro
duce
d fu
ll na
tiona
l hea
lth
acco
unts
. The
W
HO
’s G
loba
l Hea
lth
Expe
nditu
re
Dat
abas
e al
so h
as in
tern
atio
nally
co
mpa
rabl
e da
ta o
n he
alth
spe
ndin
g fo
r clo
se t
o 19
0 co
untr
ies
from
200
0 to
201
6. In
-cou
ntry
sou
rces
incl
ude
natio
nal h
ealt
h ac
coun
t re
port
s,
publ
ic e
xpen
ditu
re re
port
s, s
tatis
tical
ye
arbo
oks
and
othe
r per
iodi
cals
, and
re
port
s an
d da
ta p
rovi
ded
by c
entr
al
stat
istic
al o
ffice
s an
d m
inis
trie
s.
http
://ap
ps.w
ho.in
t/nh
a/da
taba
se/S
elec
t/In
dica
tors
/en
Benc
hmar
king
gui
delin
esN
o es
tabl
ishe
d be
nchm
ark.
Sug
gest
be
nchm
arki
ng a
gain
st p
ast
or o
wn
coun
try
perf
orm
ance
.
Not
e: U
sing
the
WH
O’s
Glo
bal H
ealth
Exp
endi
ture
dat
abas
e G
GH
E-D
+OO
PS+E
XT+V
HI+
OTH
ER=1
00%
or H
F.1+
HF.
2+H
F.3+
HF.
4+H
F.ne
c=CH
ESo
urce
: WH
O (2
019)
. Glo
bal H
ealth
Exp
endi
ture
Dat
abas
e
Fina
ncin
g (m
anag
emen
t)
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE30 Curr
ent h
ealth
exp
endi
ture
(CH
E) b
y fu
nctio
nSi
mpl
e de
finiti
on: T
he s
hare
of c
urre
nt h
ealth
exp
endi
ture
by
func
tion
(e.g
. cur
ativ
e ca
re, p
harm
aceu
tical
s, p
reve
ntiv
e ca
re, a
dmin
istr
ativ
e). U
sing
the
WH
O’s
Glo
bal H
ealth
Exp
endi
ture
dat
abas
e de
finiti
ons:
· CH
E= C
urat
ive
care
(HC.
1) +
Reh
abili
tativ
e ca
re (H
C.2)
+ L
ong-
term
car
e (H
C.3)
+ A
ncill
ary
serv
ices
(HC.
4) +
Med
ical
goo
ds (H
C.5)
+ P
reve
ntiv
e ca
re (H
C.6)
+ G
over
nanc
e an
d he
alth
sys
tem
and
fina
ncin
g ad
min
istr
atio
n (H
C.7)
+ O
ther
hea
lth c
are
serv
ices
not
els
ewhe
re c
lass
ified
(HC.
9)· C
urat
ive
care
(HC.
1)=I
npat
ient
cur
ativ
e ca
re (H
C.1.
1) +
Day
cur
ativ
e ca
re (H
C.1.
2) +
Out
patie
nt c
urat
ive
care
(HC.
1.3)
+ H
ome-
base
d cu
rativ
e ca
re (H
C.1.
4) +
Uns
peci
fied
cura
tive
care
(HC.
1.ne
c)· P
rimar
y he
alth
car
e ex
pend
iture
=Out
patie
nt C
urat
ive
Care
(HC.
1.3)
+ H
ome-
base
d cu
rativ
e ca
re (H
C.1.
4) +
Out
patie
nt lo
ng-t
erm
car
e (H
C.3.
3) +
Hom
e-ba
sed
long
-ter
m c
are
(HC.
3.4)
+ 8
0% o
f med
ical
goo
ds (H
C.5)
+
Prev
entiv
e ca
re (H
C.6)
+ 8
0% o
f Gov
erna
nce,
and
hea
lth s
yste
m a
nd fi
nanc
ing
adm
inis
trat
ion
(HC.
7). H
owev
er, i
t is
reco
mm
ende
d to
adj
ust t
his
defin
ition
to b
ette
r refl
ect c
ount
ry c
onte
xt.
Freq
uenc
y of
dat
a co
llect
ion:
Ann
ual.
Sugg
este
d co
mpa
rato
r: A
cros
s co
untr
ies;
Pas
t per
form
ance
/tre
nd d
ata.
Pote
ntia
l sou
rce
of d
ata:
Beyo
nd O
ECD
cou
ntrie
s, o
nly
a se
lect
nu
mbe
r of c
ount
ries
have
sta
rted
re
port
ing
expe
nditu
res
by fu
nctio
n to
th
e W
HO
’s G
loba
l Hea
lth E
xpen
ditu
re
Dat
abas
e (h
ttp:
//app
s.w
ho.in
t/nh
a/da
taba
se/S
elec
t/In
dica
tors
/en)
. Th
e W
HO
has
pub
lishe
d th
e fir
st e
ver
com
para
ble
mea
sure
s of
prim
ary
heal
th c
are
spen
ding
usi
ng d
ata
from
20
16 fo
r 46
low
- an
d m
iddl
e-in
com
e co
untr
ies
(LIN
K: h
ttps
://ap
ps.w
ho.in
t/iri
s/bi
tstr
eam
/han
dle/
1066
5/27
6728
/W
HO
-HIS
-HG
F-H
F-W
orki
ngPa
per-
18.3
-eng
?ua=
1).
In-c
ount
ry s
ourc
es m
ay in
clud
e na
tiona
l hea
lth a
ccou
nt re
port
s an
d pu
blic
exp
endi
ture
revi
ews
(htt
p://
sear
ch.w
orld
bank
.org
/per
).
Benc
hmar
king
gui
delin
esN
o es
tabl
ishe
d be
nchm
ark.
Sug
gest
be
nchm
arki
ng a
gain
st c
ount
ries
of s
imila
r inc
ome/
heal
th s
yste
m
deve
lopm
ent.
This
indi
cato
r is
diffi
cult
to c
ompa
re a
cros
s co
untr
ies.
Spe
ndin
g by
func
tion
is g
ener
ally
driv
en b
y fa
ctor
s su
ch a
s co
untr
y an
d he
alth
sys
tem
dev
elop
men
t, d
emog
raph
ics
and
epid
emio
logy
, tre
atm
ent p
roto
cols
, and
pat
ient
pre
fere
nces
, etc
. For
ex
ampl
e, in
cou
ntrie
s w
here
the
qual
ity o
f prim
ary
care
is p
oor o
r lon
g te
rm c
are
optio
ns a
re li
mite
d, s
pend
ing
on c
urat
ive
care
/hos
pita
ls w
ill b
e hi
gher
. It i
s al
so im
port
ant t
o no
te th
at m
ore
than
20%
of c
urre
nt h
ealth
exp
endi
ture
rem
ains
unc
lass
ified
in
som
e co
untr
ies
sugg
estin
g a
lack
of a
vaila
bilit
y of
mor
e gr
anul
ar d
ata
for p
rodu
cing
hea
lth a
ccou
nts.
Spec
ial n
ote
on p
harm
aceu
tical
spe
ndin
g: F
our p
ossi
ble
fact
ors
influ
ence
a c
ount
ry’s
phar
mac
eutic
al s
pend
ing:
i) c
ount
ry p
opul
atio
n an
d vo
lum
e of
dru
gs c
onsu
med
, ii)
drug
util
izat
ion
per p
erso
n, ii
i) ty
pe a
nd m
ix o
f dru
gs c
onsu
med
(e.g
., ge
neric
s ve
rsus
bra
nd-n
ame
drug
s), a
nd iv
) pric
es a
t whi
ch d
rugs
are
sol
d. C
ompa
ring
drug
pric
es a
cros
s co
untr
ies
is a
com
plic
ated
and
impe
rfec
t pro
cess
, prim
arily
bec
ause
of t
he p
ropr
ieta
ry n
atur
e of
the
reba
tes
drug
man
ufac
ture
rs o
ffer
diff
eren
t pay
ers.
Not
es: B
oxpl
ots
show
the
inte
rqua
rtile
rang
e of
val
ues
with
the
med
ian
mar
ked
by a
line
insi
de th
e ba
r and
the
mea
n m
arke
d by
an
X.
The
lines
from
the
bar e
xten
d to
the
max
imum
and
min
imum
val
ues
with
out
liers
exc
lude
d. In
patie
nt c
urat
ive
=HC1
.1+H
C1.2
; Out
patie
nt
cura
tive=
HC1
.3+H
C1.4
; Med
ical
goo
ds=H
C5; P
reve
ntiv
e ca
re=H
C6; G
over
nanc
e an
d ad
min
istr
atio
n=H
C7. D
ata
is fo
r 34
low
- an
d lo
wer
-m
iddl
e in
com
e co
untr
ies
for w
hich
dat
a is
ava
ilabl
e: A
fgha
nist
an; A
rmen
ia; B
urun
di; B
urki
na F
aso;
Bhu
tan;
Cot
e d’
Ivoi
re; C
ongo
, Dem
. Rep
.; Co
ngo,
Rep
; Cab
o Ve
rde;
Djib
outi;
Eth
iopi
a; G
eorg
ia; G
hana
; Gui
nea;
Gua
tem
ala;
Hai
ti; In
dia;
Ken
ya; K
yrgy
z Re
publ
ic; C
ambo
dia;
Lib
eria
; Sri
Lank
a; M
oldo
va; M
ali;
Mau
ritan
ia; N
epal
; Phi
lippi
nes;
Tog
o; T
ajik
ista
n; T
imor
-Les
te; T
unis
ia; T
anza
nia;
Uga
nda;
Zam
bia
Sour
ce: W
HO
(201
9). G
loba
l Hea
lth E
xpen
ditu
re D
atab
ase
(htt
p://a
pps.
who
.int/
nha/
data
base
/Sel
ect/
Indi
cato
rs/e
n)
55.0
50.0
45.0
40.0
35.0
30.0
25.0
20.0
15.0
10.0 5.0
0.0
Curr
ent
heal
th e
xpen
ditu
re b
y fu
ncti
on (2
016)
Inpa
tient
cu
rativ
eO
utpa
tient
cu
rativ
eM
edic
al
good
sPr
even
tive
care
Gov
erna
nce,
an
d ad
min
istr
atio
n
Percent of CHE
Fina
ncin
g (m
anag
emen
t)
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE 31
Sour
ce: W
orld
Ban
k (2
019)
. Wor
ld D
evel
opm
ent I
ndic
ator
s. (h
ttps
://da
taba
nk.w
orld
bank
.org
/sou
rce/
wor
ld-d
evel
opm
ent-
indi
cato
rs)
Phys
icia
ns, n
urse
s, a
nd m
idw
ives
, by
coun
try
(per
1,0
00 p
eopl
e)
(Mos
t rec
ent v
alue
in 2
010-
2016
)
05
1015
2025
High
inco
me
Uppe
r mid
dle
inco
me
Low
er m
iddl
e in
com
e
Low
inco
me
Hea
lth w
orke
r den
sity
Sim
ple
defin
itio
n: N
umbe
r of
phy
sici
ans,
nur
ses,
and
mid
wiv
es a
vaila
ble
(in a
reg
ion)
per
1,0
00 h
abit
ants
.
Freq
uenc
y of
dat
a co
llect
ion:
Ann
ual.
Sugg
este
d co
mpa
rato
r:
• In
the
ory,
the
rel
ativ
e st
raig
htfo
rwar
d an
d st
anda
rdiz
ed m
easu
rabi
lity
of t
hese
indi
cato
rs s
houl
d al
low
for
com
para
bilit
y ac
ross
cou
ntrie
s •
It m
ay a
lso
be o
f int
eres
t to
pol
icy
mak
ers
to c
ompa
re b
y re
gion
(e.g
. pro
vinc
e, u
rban
/rur
al)
Glo
bal d
ata
is le
ss fr
eque
nt t
han
natio
nal s
ourc
es a
nd m
ost
rece
nt v
alue
in a
5 o
r 10
year
per
iod
may
be
need
ed fo
r cro
ss-c
ount
ry c
ompa
rison
s.
It is
impo
rtan
t to
che
ck h
ow c
ount
ry d
efini
tions
diff
er fr
om g
loba
l pra
ctic
e if
com
parin
g to
oth
er c
ount
ries.
In p
ract
ice,
dat
a co
mpa
rabi
lity
is li
mite
d by
diff
eren
ces
in d
efini
tions
and
tra
inin
g of
med
ical
per
sonn
el. I
n pa
rtic
ular
, com
mun
ity
heal
th w
orke
rs (C
HW
s) a
re g
ener
ally
not
incl
uded
whi
ch m
ay b
e pr
oble
mat
ic in
cou
ntrie
s th
at a
re in
crea
sing
ly s
hift
ing
task
s to
CH
Ws
and
rely
ing
on t
hem
to
prov
ide
basi
c he
alth
ser
vice
s.
Her
e to
o it
may
be
mor
e us
eful
to c
onsi
der c
ount
ry c
onte
xt -
wha
t is
need
ed to
reac
h he
alth
sys
tem
goa
ls a
nd p
rovi
de e
quita
ble
acce
ss to
hea
lth c
are
serv
ices
. Thi
s is
esp
ecia
lly re
leva
nt in
spa
rsel
y po
pula
ted
and
hard
to re
ach
area
s w
ithin
a c
ount
ry.
It w
ould
be
good
to
exam
ine
this
indi
cato
r at
the
subn
atio
nal l
evel
alo
ngsi
de t
he c
asel
oad
indi
cato
r to
asse
ss a
lloca
tive
and
tech
nica
l (w
orke
r pro
duct
ivity
) effi
cien
cy.
Pote
ntia
l sou
rce
of d
ata:
The
WH
O c
ompi
les
data
from
ho
useh
old
and
labo
r for
ce s
urve
ys,
cens
uses
, and
adm
inis
trat
ive
reco
rds
( htt
p://a
pps.
who
.int/
gho/
data
/nod
e.m
ain?
show
only
=HW
F ). D
ata
can
also
co
me
dire
ctly
from
nat
iona
l sou
rces
in
clud
ing
prof
essi
onal
regi
strie
s an
d H
R sy
stem
s.
Benc
hmar
king
gui
delin
esTh
e W
HO
est
imat
es t
hat
at le
ast
2.5
med
ical
sta
ff (p
hysi
cian
s, n
urse
s an
d m
idw
ives
) per
1,0
00 p
eopl
e ar
e ne
eded
to
prov
ide
adeq
uate
co
vera
ge w
ith p
rimar
y ca
re
inte
rven
tions
. Alt
erna
tivel
y, t
he S
DG
in
dex
thre
shol
d is
4.4
5 ph
ysic
ians
, m
idw
ives
and
nur
ses
per 1
,000
for
UH
C an
d SD
G fu
lfilm
ent.
Inpu
ts (m
anag
emen
t)
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE32
Glo
bal d
ata
is le
ss fr
eque
nt t
han
natio
nal s
ourc
es a
nd m
ost
rece
nt v
alue
in a
5 o
r 10
year
per
iod
may
be
need
ed fo
r cro
ss-c
ount
ry c
ompa
rison
s.A
maj
or li
mita
tion
of h
ospi
tal d
ensi
ty is
tha
t it
does
not
tak
e in
to a
ccou
nt t
he s
ize
of fa
cilit
ies
and
it m
ay b
e m
ore
info
rmat
ive
to lo
ok a
t in
patie
nt b
ed d
ensi
ty. H
owev
er, d
iffer
ence
s in
defi
nitio
ns a
nd w
heth
er m
ater
nal b
eds
and/
or fa
ith-
base
d be
ds a
re in
clud
ed m
ay u
nder
/ove
rest
imat
e nu
mbe
rs. T
here
fore
, it
is im
port
ant
to c
heck
how
cou
ntry
defi
nitio
ns d
iffer
from
glo
bal p
ract
ice
if co
mpa
ring
to o
ther
cou
ntrie
s.
Sim
ilar t
o hu
man
reso
urce
s, it
may
be
mor
e us
eful
to
cons
ider
cou
ntry
con
text
- w
hat
is n
eede
d to
reac
h he
alth
sys
tem
goa
ls a
nd p
rovi
de e
quita
ble
acce
ss t
o he
alth
car
e se
rvic
es. H
ere,
it w
ould
be
good
to
exam
ine
bed
dens
ity a
t th
e su
bnat
iona
l lev
el a
long
side
the
bed
occ
upan
cy ra
te in
dica
tor t
o as
sess
allo
cativ
e an
d te
chni
cal e
ffici
ency
.
Sour
ce: W
HO
, var
ious
yea
rs
Hos
pita
l Bed
s pe
r 1,0
00 P
eopl
e, L
ates
t Ye
ar A
vaila
ble
Keny
a
Hos
pita
l bed
den
sity
Sim
ple
defin
itio
n: N
umbe
r of
hos
pita
ls o
r ho
spit
al b
eds
per
1,00
0 po
pula
tion
.
Freq
uenc
y of
dat
a co
llect
ion:
Ann
ual.
Sugg
este
d co
mpa
rato
r:
• In
the
ory,
the
rel
ativ
e st
raig
htfo
rwar
d an
d st
anda
rdiz
ed m
easu
rabi
lity
of t
hese
indi
cato
rs s
houl
d al
low
for
com
para
bilit
y ac
ross
cou
ntrie
s.
• It
may
als
o be
of i
nter
est
to p
olic
y m
aker
s to
com
pare
by
regi
on (e
.g. p
rovi
nce,
urb
an/r
ural
).
Pote
ntia
l sou
rce
of d
ata:
The
WH
O c
ompi
les
data
from
cou
ntry
fo
cal p
oint
s fr
om m
inis
trie
s of
hea
lth
thro
ugh
base
line
faci
lity
surv
eys.
Dat
a ca
n al
so c
ome
dire
ctly
from
na
tiona
l adm
inis
trat
ive
reco
rds.
Serv
ice
Prov
isio
n A
sses
smen
ts (S
PA),
a ty
pe o
f Dem
ogra
phic
and
Hea
lth
Surv
ey, m
ay a
lso
be h
elpf
ul t
o es
tabl
ish
estim
ates
.So
urce
sW
HO
: htt
p://a
pps.
who
.int/
gho/
data
/nod
e.m
ain.
HS0
7?la
ng=e
nSP
A: h
ttps
://dh
spro
gram
.com
/Wha
t-W
e-D
o/Su
rvey
-Typ
es/S
PA.c
fmD
HS:
htt
ps://
dhsp
rogr
am.c
om/
Benc
hmar
king
gui
delin
esU
sual
ly t
here
is a
cou
ntry
tar
get
for
faci
lity
dens
ity.
WH
O s
elec
ts a
n ar
bitr
ary
benc
hmar
k of
2.5
inpa
tient
bed
s pe
r 1,0
00 b
ased
on
glo
bal,
low
er a
nd u
pper
-mid
dle
inco
me
coun
try
aver
ages
of 2
.7, 1
.8,
and
3.9
hosp
ital b
eds
per 1
,000
re
spec
tivel
y.
Inpu
ts (m
anag
emen
t)
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE 33
36.8
51.3
51.7
56.3
57.6
58.7
62.3
63.4
73.1
01020304050607080
Pote
ntia
l sou
rce
of d
ata:
Usu
ally
der
ived
from
hou
seho
ld
surv
eys
with
repr
esen
tativ
e sa
mpl
es
of w
omen
of r
epro
duct
ive
age
such
as
the
Dem
ogra
phic
and
Hea
lth
Surv
ey (D
HS)
or M
ultip
le In
dica
tor
Clus
ter S
urve
ys (M
ICS)
.ht
tps:
//dhs
prog
ram
.com
/dat
a/av
aila
ble-
data
sets
.cfm
http
s://m
ics.
unic
ef.o
rg/s
urve
ys
Benc
hmar
king
gui
delin
esU
sed
to b
e 4
visi
ts, m
ost
rece
nt W
HO
re
com
men
datio
n is
8 v
isits
. How
ever
, w
hat
is m
ore
impo
rtan
t th
an t
he
num
ber o
f vis
its i
s th
e qu
ality
of
AN
C vi
sits
and
the
inte
rven
tions
de
liver
ed d
urin
g an
AN
C vi
sit.
Perc
ent
of p
regn
ant
wom
en w
ho h
ad a
ll 4
AN
C vi
sits
, Ken
ya (2
014)
Sour
ce: K
enya
Dem
ogra
phic
and
Hea
lth S
urve
y 20
14
Nor
th
East
ern
Rift
Val
ley
Tota
lCo
ast
Nai
robi
Wes
tern
East
ern
Nya
nza
Cent
ral
Ant
enat
al c
are
Sim
ple
defin
itio
n: P
erce
ntag
e of
wom
en a
ged
15-4
9 w
ith
a liv
e bi
rth
wit
hin
a gi
ven
tim
e pe
riod
that
rec
eive
d an
tena
tal c
are
prov
ided
by
a sk
illed
hea
lth
prof
essi
onal
(e.g
. phy
sici
ans,
mid
wiv
es, n
urse
s) d
urin
g th
eir
preg
nanc
y. A
nten
atal
car
e in
clud
es s
cree
ning
or
prov
idin
g tr
eatm
ent
to p
reve
nt o
r re
duce
the
ris
k of
poo
r pr
egna
ncy
outc
omes
.
Freq
uenc
y of
dat
a co
llect
ion:
Usu
ally
eve
ry 5
yea
rs.
Sugg
este
d co
mpa
rato
r: A
cros
s co
untr
ies.
It m
ay a
lso
be o
f in
tere
st t
o po
licy
mak
ers
to c
ompa
re b
y re
gion
(e.g
. pro
vinc
e, u
rban
/rur
al) o
r so
cioe
cono
mic
sta
tus
(e.g
. wea
lth
quin
tile
, lev
el o
f ed
ucat
ion)
. Pas
t pe
rfor
man
ce/t
rend
dat
a.
If a
vaila
ble,
it w
ould
be
impo
rtan
t to
look
for i
nfor
mat
ion
on A
NC
serv
ice-
spec
ific
avai
labi
lity
and
read
ines
s an
d ad
here
nce
to q
ualit
y as
sura
nce
prot
ocol
s/gu
idel
ines
from
faci
lity
surv
eys
such
as
WH
O’s
Serv
ice
Avai
labi
lity
and
Read
ines
s A
sses
smen
t (S
ARA
) (ht
tps:
//ww
w.w
ho.in
t/he
alth
info
/sys
tem
s/sa
ra_i
ntro
duct
ion/
en/)
or t
he W
orld
Ban
k’s
Serv
ice
Del
iver
y In
dica
tors
(SD
I) (h
ttps
://w
ww
.sdi
ndic
ator
s.or
g). D
HS
surv
eys
also
usu
ally
hav
e a
deta
iled
brea
kdow
n of
the
com
pone
nts
of a
nten
nal c
are.
Out
puts
(acc
ess)
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE34 Skill
ed b
irth
atte
ndan
ce
Pote
ntia
l sou
rce
of d
ata:
Usu
ally
der
ived
from
hou
seho
ld
surv
eys
with
repr
esen
tativ
e sa
mpl
es
of w
omen
of r
epro
duct
ive
age
such
as
the
Dem
ogra
phic
and
Hea
lth
Surv
ey (D
HS)
or M
ultip
le In
dica
tor
Clus
ter S
urve
ys (M
ICS)
. ht
tps:
//dhs
prog
ram
.com
/dat
a/av
aila
ble-
data
sets
.cfm
http
s://m
ics.
unic
ef.o
rg/s
urve
ys
Benc
hmar
king
gui
delin
esN
o es
tabl
ishe
d be
nchm
ark.
Her
e to
o,
wha
t is
mor
e im
port
ant
than
the
pe
rcen
tage
of b
irths
att
ende
d by
a
skill
ed h
ealt
h pe
rson
nel i
s th
e qu
ality
of
mat
erna
l and
new
born
car
e.
Birt
hs a
tten
ded
by s
kille
d he
alth
sta
ff b
y in
com
e qu
inti
le in
Sub
-Sah
aran
Afr
ica
(L
ates
t yea
r ava
ilabl
e)
Sour
ce: W
B (2
019)
. HEF
PI d
atab
ase
(htt
p://d
atat
opic
s.w
orld
bank
.org
/hea
lth-e
quity
-and
-fina
ncia
l-pr
otec
tion)Q4:
Ric
her
Q5:
Ric
hest
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Ethi
opia
Ghan
aKe
nya
Libe
riaM
adag
asca
rM
alawi
Mali
Mau
ritan
iaM
ozam
biqu
eN
iger
Nig
eria
Rwan
daSo
mal
iaSu
dan
Tanz
ania
Ugan
daZa
mbi
a
Q1:
Poo
rest
Q2:
Poo
rer
Q3:
Mid
dle
Sim
ple
defin
itio
n: P
erce
ntag
e of
live
birt
hs a
tten
ded
by s
kille
d he
alth
per
sonn
el d
urin
g a
spec
ified
tim
e pe
riod.
Freq
uenc
y of
dat
a co
llect
ion:
Usu
ally
eve
ry 5
yea
rs.
Sugg
este
d co
mpa
rato
r: A
cros
s co
untr
ies.
It m
ay a
lso
be o
f in
tere
st t
o po
licy
mak
ers
to c
ompa
re b
y re
gion
(e.g
. pro
vinc
e, u
rban
/rur
al) o
r so
cioe
cono
mic
sta
tus
(e.g
. wea
lth
quin
tile
, lev
el o
f ed
ucat
ion)
. Pa
st p
erfo
rman
ce/t
rend
dat
a.
Ther
e ar
e so
me
issu
es in
defi
ning
who
is in
clud
ed a
s a
skill
ed h
ealt
h pr
ofes
sion
al. I
n pa
rtic
ular
, com
mun
ity h
ealt
h w
orke
rs (C
HW
s) a
re g
ener
ally
not
incl
uded
whi
ch m
ay b
e pr
oble
mat
ic in
cou
ntrie
s w
here
CH
Ws
are
used
as
mai
n se
rvic
e pr
ovid
er t
o de
liver
bab
ies.
Whe
n co
nsid
erin
g w
heth
er t
o in
clud
e CH
Ws
or n
ot, i
t is
enc
oura
ged
to lo
ok a
t w
hat
trai
ning
eac
h ca
dre
has
and
wha
t se
rvic
es t
hey
are
mea
nt t
o pr
ovid
e.If
ava
ilabl
e, it
wou
ld b
e im
port
ant
to lo
ok fo
r inf
orm
atio
n on
ser
vice
-spe
cific
ava
ilabi
lity
and
read
ines
s fo
r bas
ic o
bste
tric
and
new
born
car
e, a
dher
ence
to
qual
ity a
ssur
ance
pro
toco
ls/g
uide
lines
, and
pro
vide
r kno
wle
dge
(e.g
. abi
lity
to
diag
nose
, tre
at, a
nd m
anag
e lif
e th
reat
enin
g co
mpl
icat
ions
suc
h as
pos
t-pa
rtum
hem
orrh
age
and
birt
h as
phyx
ia) f
rom
faci
lity
surv
eys
such
as
WH
O’s
Serv
ice
Avai
labi
lity
and
Read
ines
s A
sses
smen
t (S
ARA
) (ht
tps:
//ww
w.w
ho.in
t/he
alth
info
/sy
stem
s/sa
ra_i
ntro
duct
ion/
en/)
or t
he W
orld
Ban
k’s
Serv
ice
Del
iver
y In
dica
tors
(SD
I) (h
ttps
://w
ww
.sdi
ndic
ator
s.or
g)
Out
puts
(acc
ess)
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE 35
Sim
ple
defin
itio
n:
TB c
ase
dete
ctio
n: P
erce
ntag
e of
est
imat
ed n
ew a
nd re
laps
e TB
cas
es d
etec
ted
in a
giv
en y
ear.
TB n
otifi
catio
n ra
te: N
umbe
r of n
ew a
nd re
laps
ed T
B ca
ses
notifi
ed in
a g
iven
yea
r. A
not
ifica
tion
mea
ns t
hat
the
case
was
repo
rted
on
the
natio
nal s
urve
illan
ce s
yste
m.
TB t
reat
men
t su
cces
s ra
te: P
erce
nt o
f TB
case
s th
at w
ere
succ
essf
ully
tre
ated
out
of a
ll no
tified
TB
case
s du
ring
a sp
ecifi
ed p
erio
d of
tim
e.
Freq
uenc
y of
dat
a co
llect
ion:
Adm
inis
trat
ive
data
cou
ld b
e co
llect
ed m
onth
ly, y
earl
y. F
acili
ty s
urve
ys a
re u
sual
ly a
d ho
c w
hen
fund
ing
is a
vaila
ble.
Inve
ntor
y or
logi
stic
s m
anag
emen
t sy
stem
s ca
n pr
oduc
e da
ily/r
eal-
tim
e re
port
s.
Sugg
este
d co
mpa
rato
r: A
cros
s co
untr
ies
or c
ompa
red
to p
ast
perf
orm
ance
/tre
nd d
ata.
It m
ay a
lso
be o
f int
eres
t to
pol
icy
mak
ers
to c
ompa
re b
y re
gion
(e.g
. pro
vinc
e, u
rban
/rur
al),
acro
ss f
acili
ty t
ype
(prim
ary
heal
th c
linic
vs
hosp
ital
) or
prov
ider
typ
e (p
ublic
/priv
ate)
.
TB c
ase
dete
ctio
n, T
B no
tifica
tion
rate
, and
TB
trea
tmen
t suc
cess
rate
Pote
ntia
l sou
rce
of d
ata:
Adm
inis
trat
ive
data
from
faci
litie
s,
surv
eilla
nce
syst
ems,
TB
regi
ster
s.
Ann
ual c
ase
notifi
catio
ns a
re
repo
rted
by
coun
trie
s to
the
WH
O
and
then
repo
rted
in T
B an
nual
re
port
s an
d in
the
cou
ntry
pro
files
.ht
tps:
//ww
w.w
ho.in
t/tb
/cou
ntry
/dat
a/profi
les/en/
Benc
hmar
king
gui
delin
esFo
r TB
case
det
ectio
n an
d no
tifica
tion
rate
the
re is
no
esta
blis
hed
benc
hmar
k, b
ut t
he h
ighe
r the
bet
ter.
Sugg
est
benc
hmar
king
aga
inst
pas
t pe
rfor
man
ce. F
or T
B tr
eatm
ent
succ
ess
rate
– a
t le
ast
90%
.
TB c
asca
de in
Indo
nesi
a, 2
017
Sour
ce: G
loba
l Tub
ercu
losi
s Re
port
201
8 (h
ttps
://w
ww
.who
.int/
tb/p
ublic
atio
ns/g
loba
l_re
port
/en/
)
It w
ould
be
help
ful t
o lo
ok a
t in
dica
tors
alo
ng t
he e
ntire
TB
casc
ade
to e
nsur
e co
ntin
uity
of c
are.
TB
trea
tmen
t su
cces
s ra
tes
will
onl
y be
a g
ood
mea
sure
of a
per
form
ance
if d
etec
tion
of T
B is
als
o hi
gh. T
reat
men
t su
cces
s ra
tes
also
rely
on
the
outc
ome
of t
he t
reat
men
t be
ing
reco
rded
in ro
utin
e in
form
atio
n sy
stem
s an
d so
tre
atm
ent
succ
ess
rate
s w
ill a
lso
rely
on
havi
ng ro
bust
sys
tem
s. R
outin
e ca
se n
otifi
catio
ns m
ay b
e bi
ased
if t
he p
rivat
e se
ctor
doe
s no
t/un
der-
repo
rts
new
ca
ses.
Sur
veill
ance
sys
tem
may
als
o on
ly re
gist
er b
acte
riolo
gica
lly c
onfir
med
cas
es -
esp
ecia
lly if
tes
ting
and
diag
nost
ic c
apac
ity is
wea
k.If
ava
ilabl
e, it
wou
ld b
e im
port
ant
to lo
ok fo
r inf
orm
atio
n on
TB
serv
ice-
spec
ific
avai
labi
lity
and
read
ines
s an
d pr
ovid
er k
now
ledg
e su
ch a
s di
agno
stic
acc
urac
y of
TB,
and
adh
eren
ce t
o qu
ality
ass
uran
ce p
roto
cols
/gui
delin
es fr
om fa
cilit
y su
rvey
s su
ch a
s W
HO
’s Se
rvic
e Av
aila
bilit
y an
d Re
adin
ess
Ass
essm
ent
(SA
RA) (
http
s://w
ww
.who
.int/
heal
thin
fo/s
yste
ms/
sara
_int
rodu
ctio
n/en
/) o
r the
Wor
ld B
ank’
s Se
rvic
e D
eliv
ery
Indi
cato
rs (S
DI)
(htt
ps://
ww
w.s
dind
icat
ors.
org)
A h
igh
TB n
otifi
catio
n ra
te m
ay a
lso
act
as g
ood
prox
y fo
r the
eff
ectiv
enes
s of
hea
lth
surv
eilla
nce
syst
ems
over
all.
8420
00
4462
6044
2172
3837
83.6
TB in
cide
nce
TB d
etec
tion
TB n
oti�
cati
onTB
trea
tmen
t and
suc
cess
53%
99%
87%
Number of patients
Out
puts
(qua
lity)
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE36 Gen
eral
ser
vice
read
ines
s; S
ervi
ce s
peci
fic re
adin
ess
Benc
hmar
king
gui
delin
esN
o es
tabl
ishe
d be
nchm
ark,
th
e hi
gher
the
bet
ter.
Sugg
est
benc
hmar
king
ag
ains
t pa
st p
erfo
rman
ce,
aver
age
faci
lity/
prov
ider
, or
bes
t pe
rfor
mer
.
Gen
eral
ser
vice
read
ines
s an
d se
rvic
e sp
ecifi
c re
adin
ess
desc
ribes
the
ava
ilabi
lity
of t
he c
ompo
nent
s ne
eded
to
deliv
er s
ervi
ces
such
as
basi
c am
eniti
es, d
rugs
, and
equ
ipm
ent.
SA
RA, S
DI a
nd S
PA p
roto
cols
are
incr
easi
ngly
bei
ng s
tand
ardi
zed
to a
llow
gre
ater
com
para
bilit
y be
twee
n co
untr
ies.
Nev
erth
eles
s, s
tand
ard
defin
ition
s m
ay n
ot m
ake
sens
e fo
r the
ser
vice
del
iver
y co
ntex
t an
d as
sess
men
t ag
ains
t a
coun
try’
s ow
n st
anda
rds
and
prot
ocol
s is
als
o ad
vise
d.W
hile
oft
en u
sed
as a
pro
xy fo
r qua
lity
– se
rvic
e re
adin
ess
only
mea
sure
s av
aila
bilit
y of
tra
cer i
tem
s. It
is s
ugge
sted
to
com
plem
ent
this
mea
sure
with
oth
ers
such
as
diag
nost
ic a
ccur
acy,
adh
eren
ce t
o cl
inic
al g
uide
lines
, etc
.
All
faci
lity
type
s (4
9%)
0%20
%40
%60
%80
%10
0%0%
20%
40%
60%
80%
100%
0%20
%40
%60
%80
%10
0%
Priv
ate
Clin
ics
(67%
)PH
Cs (3
6%)
35%
40%
68%
36%
67%
50%
20%
36%
58%
79%
81%
74%
71%
77%
44%
52%
31%
34%
66%
31%
67%
45%
16%
32%
Basi
c A
men
ities
Infr
astr
uctu
re a
vaila
bilit
y (S
DI)
Basi
c eq
uipm
ent
Equi
pmen
t ava
ilabi
lity
(SD
I)
Infe
ctio
n pr
even
tion
Dia
gnos
tic c
apac
ity
Esse
ntia
l med
icin
e
Esse
ntia
l dru
g av
aila
bilit
y
Not
e: S
ARA
and
SD
I sur
veys
hav
e sl
ight
ly d
iffer
ent d
efini
tions
/che
cklis
t ite
ms
for b
asic
am
eniti
es/in
fras
truc
ture
ava
ilabi
lity,
equ
ipm
ent,
and
ess
entia
l med
icin
esSo
urce
: Nig
eria
Hea
lth F
acili
ty S
urve
y 20
15-2
016Gen
eral
ser
vice
read
ines
s by
faci
lity
type
, Nig
eria
.
Pote
ntia
l sou
rce
of d
ata:
Faci
lity
surv
eys
such
as
WH
O’s
Serv
ice
Avai
labi
lity
and
Read
ines
s A
sses
smen
t (S
ARA
) or W
orld
Ban
k’s
Serv
ice
Del
iver
y In
dica
tors
(S
DIs
); a
dmin
istr
ativ
e da
ta.
http
s://d
hspr
ogra
m.c
om/d
ata/
avai
labl
e-da
tase
ts.c
fmht
tps:
//mic
s.un
icef
.org
/sur
veys
Sim
ple
defin
itio
n: M
easu
res
the
capa
city
of h
ealt
h fa
cilit
ies
to p
rovi
de b
asic
ser
vice
s ba
sed
on t
he a
vaila
bilit
y of
sel
ecte
d tr
acer
item
s fo
r ba
sic
amen
itie
s, e
quip
men
t, e
ssen
tial
med
icin
es, d
iagn
osti
c ca
paci
ty,
and
stan
dard
pre
caut
ions
for
infe
ctio
n pr
even
tion
.
Freq
uenc
y of
dat
a co
llect
ion:
Sur
vey
data
is u
sual
ly d
one
ever
y 5
year
s. F
acili
ty s
urve
ys a
re u
sual
ly a
d ho
c w
hen
fund
ing
is a
vaila
ble.
Sugg
este
d co
mpa
rato
r: A
cros
s fa
cilit
ies
and/
or p
rovi
ders
. Acr
oss
diff
eren
t ty
pes
of s
ervi
ces.
Out
puts
(qua
lity)
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE 37
Sim
ple
defin
itio
n: P
rovi
ders
cor
rect
ly d
iagn
ose
cond
itio
n X
base
d on
vig
nett
e su
rvey
dat
a.
Freq
uenc
y of
dat
a co
llect
ion:
Fac
ility
sur
veys
are
usu
ally
ad
hoc
whe
n fu
ndin
g is
ava
ilabl
e.
Sugg
este
d co
mpa
rato
r: A
cros
s fa
cilit
ies,
pro
vide
rs, c
ondi
tion
s.
Birt
h as
phyx
iaDi
abet
esDi
arrh
eaM
alar
iaPn
eum
onia
Post
-par
tum
hem
orrh
age
Tube
rcul
osis
PHCs
NC
NE
NW
SE SS SWPr
ivat
eNC
NE
NW
SE SS SW
0.0
100.
0
Dia
gnos
tic a
ccur
acy
for a
trac
er c
ondi
tion
Pote
ntia
l sou
rce
of d
ata:
Faci
lity
surv
eys
such
as
WH
O’s
Serv
ice
Avai
labi
lity
and
Read
ines
s A
sses
smen
t (S
ARA
) htt
ps://
ww
w.w
ho.
int/
heal
thin
fo/s
yste
ms/
sara
_rep
orts
/en
/ or W
orld
Ban
k’s
Serv
ice
Del
iver
y In
dica
tors
(SD
Is) h
ttps
://w
ww
.sd
indi
cato
rs.o
rg/in
dica
tors
Benc
hmar
king
gui
delin
esN
o es
tabl
ishe
d be
nchm
ark,
th
e hi
gher
the
bet
ter.
Sugg
est
benc
hmar
king
aga
inst
pas
t pe
rfor
man
ce a
vera
ge p
rovi
der,
or
bes
t pe
rfor
mer
.
Dia
gnos
tic
accu
racy
by
cond
itio
n, p
rovi
der t
ype,
and
regi
on (%
), N
iger
ia, 2
016
Prot
ocol
s ar
e in
crea
sing
ly b
eing
sta
ndar
dize
d to
allo
w g
reat
er c
ompa
rabi
lity
betw
een
coun
trie
s. H
owev
er, g
iven
pro
vide
r tra
inin
g an
d tr
eatm
ent
prot
ocol
s m
ay v
ary,
cou
ntrie
s ha
ve a
lso
conv
ened
focu
s gr
oups
to
disc
uss
vign
ette
in
terp
reta
tion
and
mak
e fin
ding
s m
ore
rele
vant
to
loca
l ser
vice
del
iver
y co
ntex
t. F
or t
his
reas
on it
may
be
bett
er t
o co
mpa
re d
iagn
ostic
acc
urac
y w
ithin
cou
ntry
acr
oss
regi
ons,
faci
litie
s, p
rovi
ders
.To
bet
ter a
sses
s pr
ovid
er k
now
ledg
e, it
is a
lso
reco
mm
ende
d to
look
at
addi
tiona
l ind
icat
ors
such
as
adhe
renc
e to
clin
ical
gui
delin
es, t
reat
men
t ac
cura
cy, a
nd a
bilit
y to
man
age
cert
ain
trac
er c
ondi
tions
whi
ch a
re g
ener
ally
als
o co
llect
ed a
s pa
rt o
f the
faci
lity
surv
ey v
igne
ttes
.
Not
e: N
iger
ia h
ealth
faci
lity
surv
ey, 2
015–
2016
Sour
ce: I
nven
tory
man
agem
ent s
yste
m, C
ount
ry X
. 201
7
Out
puts
(qua
lity)
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE38
Sim
ple
defin
itio
n: W
heth
er p
rovi
ders
ask
or
adhe
re t
o th
e qu
esti
ons
and
exam
s th
ey s
houl
d be
ask
ing
or d
oing
whe
n ch
ecki
ng a
pat
ient
for
a p
arti
cula
r co
ndit
ion.
Freq
uenc
y of
dat
a co
llect
ion:
Sur
vey
data
is u
sual
ly d
one
ever
y 5
year
s. F
acili
ty s
urve
ys a
re u
sual
ly a
d ho
c w
hen
fund
ing
is a
vaila
ble.
Sugg
este
d co
mpa
rato
r: A
cros
s fa
cilit
ies
and/
or p
rovi
ders
. Acr
oss
diff
eren
t ty
pes
of s
ervi
ces.
Adh
eren
ce to
clin
ical
gui
delin
es
Pote
ntia
l sou
rce
of d
ata:
• Fac
ility
sur
veys
suc
h as
WH
O’s
Serv
ice
Avai
labi
lity
and
Read
ines
s A
sses
smen
t (S
ARA
) htt
ps://
ww
w.
who
.int/
heal
thin
fo/s
yste
ms/
sara
_re
port
s/ o
r Wor
ld B
ank’
s Se
rvic
e D
eliv
ery
Indi
cato
rs (S
DIs
) htt
ps://
ww
w.s
dind
icat
ors.
org/
indi
cato
rs• H
ouse
hold
sur
veys
suc
h as
the
D
emog
raph
ic a
nd H
ealt
h Su
rvey
(D
HS)
htt
ps://
dhsp
rogr
am.c
om/d
ata/
avai
labl
e-da
tase
ts.c
fm. T
he U
NIC
EF
Mul
tiple
Indi
cato
r Clu
ster
Sur
vey
(MIC
S) h
ttps
://m
ics.
unic
ef.o
rg/s
urve
ys
can
also
be
used
if g
uide
lines
are
ex
plic
it as
to
wha
t se
rvic
es s
houl
d be
pro
vide
d fo
r cer
tain
typ
es o
f se
rvic
es e
.g. a
nten
atal
car
e vi
sits
.
Benc
hmar
king
gui
delin
esN
o es
tabl
ishe
d be
nchm
ark,
th
e hi
gher
the
bet
ter.
Sugg
est
benc
hmar
king
aga
inst
pas
t pe
rfor
man
ce, o
r bes
t pe
rfor
mer
av
erag
e fa
cilit
y/pr
ovid
er.
Sour
ce: N
iger
ia H
ealth
Fac
ility
Sur
vey
2015
-201
6
Impo
rtan
t pr
ecur
sors
to
bein
g ab
le t
o m
easu
re t
his
indi
cato
r inc
lude
cle
arly
defi
ned
trea
tmen
t pr
otoc
ols,
tra
ined
ser
vice
pro
vide
rs, a
nd g
ood
docu
men
tatio
n or
info
rmat
ion
syst
ems
to b
e ab
le t
o m
onito
r adh
eren
ce.
Adh
eren
ce t
o cl
inic
al g
uide
lines
may
als
o be
influ
ence
d by
mul
tiple
sys
tem
fact
ors.
It is
sug
gest
ed t
o co
mpl
emen
t th
is m
easu
re w
ith s
ervi
ce re
adin
ess
mea
sure
s su
ch a
s a
vaila
bilit
y of
dru
gs; d
iagn
ostic
cap
acity
and
pro
vide
r kno
wle
dge
mea
sure
s su
ch a
s di
agno
stic
acc
urac
y.
Adh
eren
ce t
o cl
inic
al g
uide
lines
by
cond
itio
n %
, Nig
eria
.
Dia
rrhe
a, 4
3
Pneu
mon
ia, 4
0D
iabe
tes,
41
Mal
aria
, 50
Birt
h A
sphy
xia,
17
Post
-par
tum
he
mor
rhag
e, 3
2
Tube
rcul
osis
, 44
Dia
rrhe
a, 3
1
Pneu
mon
ia, 2
7
Dia
bete
s, 2
3
Mal
aria
, 39
Birt
h A
sphy
xia,
9
Post
-par
tum
he
mor
rhag
e, 1
7
Tube
rcul
osis
, 30
Publ
icPr
ivat
e
Out
puts
(qua
lity)
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE 39Dia
bete
s con
trol
Sim
ple
defin
ition
: Per
cent
age
of p
atie
nts
who
hav
e th
eir d
iabe
tes
unde
r con
trol
. Peo
ple
with
dia
bete
s ge
nera
lly h
ave
high
leve
ls o
f glu
cose
in th
e bl
ood
mea
sure
d by
a h
emog
lobi
n A
1c>6
.5%
and
/or b
lood
glu
cose
>12
6mg/
dL.
Freq
uenc
y of
dat
a co
llect
ion:
Ana
lyze
d m
onth
ly, q
uart
erly
, ann
ually
. Sur
vey
data
usu
ally
eve
ry 5
yea
rs.
Sugg
este
d co
mpa
rato
r: A
cros
s fa
cilit
ies/
hosp
itals
; aga
inst
pas
t per
form
ance
; acr
oss
regi
ons
with
in c
ount
ry.
Pote
ntia
l sou
rce
of d
ata:
Med
ical
reco
rds,
adm
inis
trat
ive
reco
rds
or s
urve
illan
ce s
yste
ms
like
natio
nal d
iabe
tes
patie
nt re
gist
ries.
H
ouse
hold
sur
veys
suc
h as
the
Dem
ogra
phic
and
Hea
lth S
urve
y (D
HS)
, the
UN
ICEF
Mul
tiple
Indi
cato
r Cl
uste
r Sur
vey
(MIC
S).
Benc
hmar
king
gui
delin
esTh
ere
are
clin
ical
gui
delin
es th
at
act a
s be
nchm
arks
. Whi
le ta
rget
s m
ay b
e sp
ecifi
c to
cou
ntrie
s’ c
linic
al
prot
ocol
s, in
gen
eral
dia
bete
s co
ntro
l re
quire
s an
HbA
1c te
st b
etw
een
7-8%
fo
r pat
ient
s w
ith ty
pe II
dia
bete
s or
Fa
stin
g bl
ood
gluc
ose
betw
een
80-
130
mg/
dL o
r 4.4
to 7
.2 m
mol
/L.
If a
vaila
ble,
it w
ould
be
impo
rtan
t to
look
for i
nfor
mat
ion
on i)
dia
bete
s se
rvic
e-sp
ecifi
c av
aila
bilit
y an
d re
adin
ess
(i.e.
bas
ic it
ems
need
ed t
o te
st a
nd t
reat
dia
bete
s), i
i) ad
here
nce
to q
ualit
y as
sura
nce
prot
ocol
s/gu
idel
ines
, and
iii)
prov
ider
kn
owle
dge
mea
sure
s su
ch a
s di
agno
stic
acc
urac
y fr
om fa
cilit
y su
rvey
s su
ch a
s W
HO
’s Se
rvic
e Av
aila
bilit
y an
d Re
adin
ess
Ass
essm
ent
(SA
RA) o
r the
Wor
ld B
ank’
s Se
rvic
e D
eliv
ery
Indi
cato
rs (S
DI).
It w
ould
als
o be
hel
pful
to
look
at
a ra
nge
of
indi
cato
rs a
long
the
ent
ire d
iabe
tes
casc
ade
(e.g
. dia
bete
s pr
eval
ence
; per
cent
dia
gnos
ed; p
erce
nt o
n tr
eatm
ent;
per
cent
und
er c
ontr
ol; d
iabe
tes-
rela
ted
hosp
ital a
dmis
sion
s; a
nd m
orta
lity
due
to d
iabe
tes)
Mal
aysi
a di
abet
es c
asca
de
Not
e: *
Bas
ed o
n sa
mpl
e-ba
sed
clin
ical
aud
it of
dia
gnos
ed d
iabe
tes
patie
nts.
htt
p://w
ww
.iku.
gov.
my/
imag
es/IK
U/D
ocum
ent/
REPO
RT/
nhm
srep
ort2
015v
ol2.
Sour
ces:
Inte
rnat
iona
l Dia
bete
s Fe
dera
tion
(htt
ps://
idf.o
rg/o
ur-n
etw
ork/
regi
ons-
mem
bers
/wes
tern
-pac
ific/
mem
bers
/108
-mal
aysi
a.ht
ml)
and
*Nat
iona
l Dia
bete
s Re
gist
ry o
f Mal
aysi
a
Prev
alen
ceD
iagn
osed
Am
ong
diag
nose
d*A
mon
g di
agno
sed,
un
der c
ontr
ol
(HbA
1c <
8%)*
16.9
40.6
86.5
58.2
4,00
0,00
0
3,50
0,00
0
3,00
0,00
0
2,50
0,00
0
2,00
0,00
0
1,50
0,00
0
1,00
0,00
0
500,
000 0
100
90 80 70 60 50 40 30 20 10 0
Absolute number
%
Out
puts
(qua
lity)
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE40
Pot
enti
ally
avo
idab
le a
dmis
sion
s fo
r ch
roni
c ill
ness
es t
hat
coul
d be
ave
rted
if t
he p
atie
nts
had
qual
ity
outp
atie
nt c
are.
Cond
itio
ns f
or w
hich
hos
pita
l adm
issi
ons
coul
d be
pre
vent
ed b
y in
terv
enti
ons
in c
omm
unit
y se
ttin
gs.
Freq
uenc
y of
dat
a co
llect
ion:
Usu
ally
ana
lyze
d m
onth
ly, q
uart
erly
or
annu
ally
. Hou
seho
lds
surv
eys
are
less
freq
uent
ly a
vaila
ble
and
not
sugg
este
d fo
r ro
utin
e m
onit
orin
g.
Sugg
este
d co
mpa
rato
r: A
cros
s fa
cilit
ies/
hosp
ital
s. It
may
be
usef
ul f
or p
olic
y m
aker
s to
hav
e co
mpa
rison
s ag
greg
ated
at
subn
atio
nal l
evel
(e.g
. pro
vinc
es, d
istr
icts
).
Avoi
dabl
e ho
spita
l adm
issi
ons
for p
rimar
y ca
re s
ensi
tive
cond
ition
s
Pote
ntia
l sou
rce
of d
ata:
Idea
lly c
laim
s da
ta a
nd/o
r med
ical
re
cord
s. S
ome
hous
ehol
d su
rvey
s w
ith a
det
aile
d he
alth
mod
ule.
Benc
hmar
king
gui
delin
esN
o be
nchm
ark,
the
low
er is
bet
ter.
Sugg
est
benc
hmar
king
aga
inst
pas
t pe
rfor
man
ce o
r bes
t pe
rfor
mer
av
erag
e fa
cilit
y/pr
ovid
er.
Hyp
erte
nsio
n an
d di
abet
es a
dmis
sion
by
faci
lity,
Nor
th M
aced
onia
, 201
6
Sour
ce: H
ealth
Insu
ranc
e Fu
nd o
f Nor
th M
aced
onia
vary
. Con
ditio
ns g
ener
ally
con
side
red
are
diab
etes
, hyp
erte
nsio
n, a
nd a
cute
res
pira
tory
con
ditio
ns s
uch
as b
ronc
hitis
, ast
hma,
and
chr
onic
obs
truc
tive
pulm
onar
y di
seas
e as
mos
t ar
e no
t as
soci
ated
with
com
plic
atio
ns a
nd s
houl
d no
t re
quire
ho
spita
l adm
issi
ons.
With
in c
ount
ry c
ompa
rison
is a
lso
reco
mm
ende
d as
adm
issi
on a
nd t
reat
men
t pr
otoc
ols
and
the
read
ines
s an
d qu
ality
of
prim
ary
heal
th c
are
to a
ppro
pria
tely
man
age
thes
e co
nditi
ons
may
var
y. A
s th
e qu
ality
of
clai
ms
data
impr
oves
, cas
e-m
ix s
ever
ity s
houl
d al
so b
e ta
ken
into
acc
ount
.
Out
puts
(qua
lity)
050100
150
200
250
300
350
General municip
al hospita
l in Sko
pjeBito
la Clinica
l Hospita
l
Endocrinology U
niversi
ty hospita
l in Sko
pje
Gos�va
r Gen
eral Hosp
ital
Struga General H
ospita
lSh
�p Clinica
l Hosp
ital
Tetovo Clin
ical h
ospita
lPrile
p General Hosp
ital
Kavadarts
i General H
ospita
l
Kicevo General H
ospita
lVeles G
eneral Hosp
ital
Kocani G
eneral Hosp
ital
Debar Gen
eral Hosp
ital
Kumanovo General Hosp
ital
Strumica
General Hosp
ital
Ohrid General H
ospita
l
Gevgelija
General Hospita
l
Number of admissionsH
yper
tens
ion
and
diab
etes
adm
issi
ons b
y fa
cilit
y, M
aced
onia
, 201
6
Hyp
erte
nsio
n w
ithou
t com
plic
a�on
sD
iabe
tes w
ithou
t sev
ere
com
plic
a�on
s
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE 41
Inco
me
Coun
try
Mal
e
Fem
ale
Low
inco
me
Low
er m
iddl
e in
com
eU
pper
mid
dle
inco
me
Hig
h in
com
e
0204002040
Congo, Dem. RepNiger
EthiopiaChad
Sierra LeoneGuineaLiberia
CambodiaComoros
ZimbabweKorea, Dem. People’s Rep.
TogoTajikistan
Gambia, TheHaiti
Kyrgyz RepublicTimor-Leste
PakistanYemen, Rep.
GuatemalaZambia
IndiaIndonesia
BangladeshNigeria
DjiboutiCote d’Ivoire
PhilippinesMauritania
HondurasSenegal
GhanaSri LankaParaguayMoldovaEcuadorNamibia
GabonAzerbaijan
ThailandPeru
MexicoAlgeriaTunisiaTurkey
MontenegroJordan
Bosnia and HerzegovinaDominican Republic
JamaicaBarbados
KuwaitUnited States
Chile
Pote
ntia
l sou
rce
of d
ata:
Pote
ntia
l Sou
rce
Hou
seho
ld s
urve
ys
such
as
the
Dem
ogra
phic
and
Hea
lth
Surv
ey (D
HS)
(htt
ps://
dhsp
rogr
am.c
om),
the
UN
ICEF
Mul
tiple
Indi
cato
r Clu
ster
Su
rvey
(MIC
S) (h
ttps
://m
ics.
unic
ef.o
rg).
Nat
iona
l cen
sus
surv
eys.
Glo
bal
data
base
s fo
r cou
ntry
com
paris
ons:
th
e H
uman
Cap
ital P
roje
ct u
ses
data
on
stu
ntin
g fo
r con
stru
ctio
n of
its
HCI
from
the
UN
ICEF
-WH
O-T
he
Wor
ld B
ank:
Joi
nt C
hild
Mal
nutr
ition
Es
timat
es. (
http
s://w
ww
.wor
ldba
nk.o
rg/
en/p
ublic
atio
n/hu
man
-cap
ital) .
Benc
hmar
king
gui
delin
esN
o es
tabl
ishe
d be
nchm
ark,
the
lo
wer
the
bet
ter.
As
part
of t
he W
HO
G
loba
l Nut
ritio
n Ta
rget
s, a
40%
re
duct
ion
in t
he n
umbe
r of c
hild
ren
unde
r 5 w
ho a
re s
tunt
ed b
y 20
25
(thi
s is
roug
hly
10%
per
yea
r) u
sing
20
12 a
s ba
selin
e.
Prev
alen
ce o
f stu
ntin
g, c
hild
ren
unde
r age
5 (%
)
Stun
ting
refle
cts
poor
nut
ritio
n an
d in
fect
ion
expo
sure
dur
ing
preg
nanc
y an
d af
ter b
irth.
The
refo
re, i
f ava
ilabl
e, it
wou
ld b
e im
port
ant
to lo
ok fo
r inf
orm
atio
n on
AN
C an
d ch
ild h
ealt
h se
rvic
e-sp
ecifi
c av
aila
bilit
y an
d re
adin
ess,
dia
gnos
tic
accu
racy
of c
omm
on c
hild
hood
con
ditio
ns li
ke d
iarr
hea,
and
adh
eren
ce t
o qu
ality
ass
uran
ce p
roto
cols
/gui
delin
es fr
om fa
cilit
y su
rvey
s su
ch a
s W
HO
’s Se
rvic
e Av
aila
bilit
y an
d Re
adin
ess
Ass
essm
ent
(SA
RA) (
http
s://w
ww
.who
.int/
heal
thin
fo/
syst
ems/
sara
_int
rodu
ctio
n/en
/) o
r the
Wor
ld B
ank’
s Se
rvic
e D
eliv
ery
Indi
cato
rs (S
DI)
(htt
ps://
ww
w.s
dind
icat
ors.
org)
.
Not
e: M
ost r
ecen
t val
ue in
201
2-20
15So
urce
: Wor
ld B
ank
(201
9). W
orld
Dev
elop
men
t Ind
icat
ors
Child
ren
unde
r 5 w
ho a
re s
tunt
edSi
mpl
e de
finit
ion:
Per
cent
age
of s
tunt
ed (m
oder
ate
and
seve
re) c
hild
ren
aged
0–5
9 m
onth
s (m
oder
ate
= he
ight
-for
-age
bel
ow -
2 st
anda
rd d
evia
tion
s fr
om t
he W
HO
Chi
ld G
row
th S
tand
ards
med
ian;
se
vere
= h
eigh
t-fo
r-ag
e be
low
-3
stan
dard
dev
iati
ons
from
the
WH
O C
hild
Gro
wth
Sta
ndar
ds m
edia
n).
Freq
uenc
y of
dat
a co
llect
ion:
Sur
vey
data
usu
ally
eve
ry 5
yea
rs. U
NIC
EF-W
HO
-The
Wor
ld B
ank:
Joi
nt M
alnu
trit
ion
Esti
mat
es u
pdat
ed a
nnua
lly.
Sugg
este
d co
mpa
rato
r: A
cros
s co
untr
ies.
It m
ay a
lso
be o
f in
tere
st t
o po
licy
mak
ers
to c
ompa
re b
y re
gion
(e.g
. pro
vinc
e, u
rban
/rur
al) o
r so
cioe
cono
mic
sta
tus
(e.g
. wea
lth
quin
tile
, lev
el o
f ed
ucat
ion)
. Pa
st p
erfo
rman
ce/t
rend
dat
a.
Out
puts
(ris
k fa
ctor
s)
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE42
KiribatiNauru
GreeceSerbia
Russian FederationBosnia and Herzegovi..
ChileIndonesia
JordanLebanonBulgaria
CubaCroatiaLatvia
Sierra LeoneLao PDR
Czech RepublicEstonia
AndorraSamoa
UkraineGermanyRomania
IsraelSpain
BahrainTonga
LithuaniaGeorgiaAlbania
HungaryPoland
Slovak RepublicFrance
ArmeniaBelarus
Korea, Rep.Turkey
MongoliaSeychellesPhilippines
FijiKyrgyz Republic
KazakhstanMaltaChina
NetherlandsLesotho
ArgentinaUruguay
60 50 40 30 20
Pote
ntia
l sou
rce
of d
ata:
Hou
seho
ld s
urve
ys s
uch
as t
he
Dem
ogra
phic
and
Hea
lth
Surv
ey
(DH
S),
the
UN
ICEF
Mul
tiple
Indi
cato
r Cl
uste
r Sur
vey
(MIC
S), W
orld
Hea
lth
Org
aniz
atio
n ST
EPw
ise
(STE
PS)
surv
ey a
nd n
atio
nal s
urve
ys.
http
s://d
hspr
ogra
m.c
omht
tps:
//mic
s.un
icef
.org
http
s://w
ww
.who
.int/
ncds
/sur
veill
ance
/st
eps/
repo
rts/
en/
Benc
hmar
king
gui
delin
esN
o es
tabl
ishe
d be
nchm
ark,
the
low
er
the
bett
er. G
loba
l vol
unta
ry t
arge
t of
30
% re
lativ
e re
duct
ion
in t
he u
se o
f to
bacc
o in
per
sons
ove
r 15
year
s by
20
20 (r
efer
ence
yea
r: 2
013)
.
Prev
alen
ce o
f sm
okin
g of
any
tob
acco
pro
duct
, 201
3
Surv
eys
may
hav
e di
ffer
ent
incl
usio
n cr
iteria
bas
ed o
n ag
e. F
or in
stan
ce, S
TEPS
sur
vey
will
gen
eral
ly a
sk t
hose
18
year
s or
old
er w
here
as D
HS
asks
tho
se w
ho a
re 1
5 ye
ars
or o
lder
. Cur
rent
use
defi
nitio
ns m
ay a
lso
vary
dep
endi
ng o
n th
e su
rvey
. Tob
acco
use
can
be
both
sm
oke
and
non-
smok
ed b
ased
use
d.
Toba
cco
use
amon
g pe
rson
s ag
ed 18
+Si
mpl
e de
finit
ion:
Pre
vale
nce
of c
urre
nt t
obac
co u
se a
mon
g pe
rson
s ag
ed 1
8+ y
ears
.
Freq
uenc
y of
dat
a co
llect
ion:
Sur
vey
data
usu
ally
eve
ry 5
yea
rs.
Sugg
este
d co
mpa
rato
r: A
cros
s co
untr
ies.
It m
ay a
lso
be o
f in
tere
st t
o po
licy
mak
ers
to c
ompa
re b
y so
cioe
cono
mic
sta
tus
(e.g
. wea
lth
quin
tile
, lev
el o
f ed
ucat
ion,
gen
der)
. Pas
t pe
rfor
man
ce/t
rend
dat
a.
Not
e: T
op 5
0 hi
ghes
t cou
ntrie
sSo
urce
: Wor
ld B
ank
(201
9). W
orld
Dev
elop
men
t Ind
icat
ors
(htt
p://w
di.w
orld
bank
.org
)
Out
puts
(ris
k fa
ctor
s)
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE 43Pote
ntia
l sou
rce
of d
ata:
Faci
lity
surv
eys
such
as
WH
O’s
Serv
ice
Avai
labi
lity
and
Read
ines
s A
sses
smen
t (S
ARA
: htt
ps://
ww
w.w
ho.
int/
heal
thin
fo/s
yste
ms/
sara
_rep
orts
/en
/ ) or
Wor
ld B
ank’
s Se
rvic
e D
eliv
ery
Indi
cato
rs (S
DIs
: htt
ps://
ww
w.s
dind
icat
ors.
org/
indi
cato
rs);
ad
min
istr
ativ
e/fa
cilit
y re
cord
s.
Benc
hmar
king
gui
delin
esN
o es
tabl
ishe
d be
nchm
ark.
Su
gges
t be
nchm
arki
ng a
gain
st p
ast
perf
orm
ance
or a
vera
ge fa
cilit
y.
In O
ECD
cou
ntrie
s ge
nera
l pr
actit
ione
rs s
ee o
n av
erag
e be
twee
n 11
to
30 p
atie
nts
a da
y w
hile
in S
ub-
Saha
ran
Afr
ica
whe
re m
ost
of t
he S
DI
surv
eys
have
tak
en p
lace
cas
eloa
d ra
nges
bet
wee
n 4
to 1
7.
Case
load
Sour
ce: W
orld
Ban
k (2
016)
. Nig
eria
Nat
iona
l Hea
lth F
acili
ty S
urve
y
Case
load
is a
com
plex
indi
cato
r to
inte
rpre
t. It
cou
ld b
e aff
ecte
d by
man
y su
pply
sid
e is
sues
suc
h as
the
ove
r- o
r und
er-s
uppl
y of
hum
an re
sour
ces
the
sala
ry a
nd t
imel
ines
s of
pay
men
t to
pro
vide
rs, a
bsen
teei
sm ra
tes,
the
num
ber o
f pa
tient
s re
gist
ered
to
a pr
imar
y he
alth
car
e fa
cilit
y, t
he a
vaila
bilit
y of
dru
gs a
nd e
quip
men
t, t
he fu
nctio
nalit
y of
faci
litie
s, t
he s
ervi
ce p
rovi
sion
resp
onsi
bilit
ies
of d
iffer
ent
cadr
es o
r by
dem
and
side
issu
es s
uch
as a
pre
fere
nce
for d
iffer
ent
prov
ider
typ
es o
r set
tings
, the
dis
ease
bur
den,
the
aff
orda
bilit
y of
car
e, t
he d
ista
nce
to fa
cilit
ies,
or t
he p
erce
ived
qua
lity
of c
are.
Thi
s m
akes
it d
ifficu
lt t
o co
mpa
re a
cros
s co
untr
ies
as t
he h
ealt
h ca
re c
onte
xt m
ay b
e di
ffer
ent.
For
exa
mpl
e,
com
mun
ity h
ealt
h w
orke
rs (C
HW
s) a
re n
ot in
clud
ed in
the
WB’
s SD
I defi
nitio
n w
hich
may
be
prob
lem
atic
in c
ount
ries
that
are
incr
easi
ngly
shi
ftin
g ta
sks
to C
HW
s an
d re
lyin
g on
the
m t
o pr
ovid
e ba
sic
heal
th s
ervi
ces.
Eve
n w
ithin
cou
ntrie
s,
it is
reco
mm
ende
d to
pai
r thi
s in
dica
tor w
ith o
ther
s (e
.g. a
bsen
teei
sm, d
elay
of p
aym
ent,
num
ber o
f pat
ient
s re
gist
ered
to
a fa
cilit
y or
pro
vide
r, av
erag
e nu
mbe
r of p
atie
nt v
isits
with
in a
yea
r etc
.) to
iden
tity
the
appr
opria
te p
olic
y ac
tion.
W
hile
cas
eloa
d is
gen
eral
ly a
n ou
tpat
ient
/prim
ary
heal
th c
are
mea
sure
, oth
er m
easu
res
of p
rodu
ctiv
ity s
uch
as t
ime
per p
atie
nt p
er v
isit
(tho
ugh
less
com
mon
) hav
e al
so b
een
used
.
©20
19 M
apbo
x ©
Ope
nStr
eetM
ap
0-1
1-1.
51.
5-2
2-2.
52.
5-3
3-4
>4
Case
load
Sim
ple
defin
itio
n: N
umbe
r of
out
pati
ent
visi
ts p
er c
linic
ian
per
day.
Freq
uenc
y of
dat
a co
llect
ion:
Fac
ility
sur
veys
are
usu
ally
ad
hoc
whe
n fu
ndin
g is
ava
ilabl
e. A
dmin
istr
ativ
e/fa
cilit
y re
cord
s ar
e ge
nera
lly a
naly
zed
mon
thly
, qua
rter
ly, o
r an
nual
ly.
Sugg
este
d co
mpa
rato
r: A
cros
s fa
cilit
ies,
typ
es o
f pro
vide
rs.
Out
puts
(man
agem
ent)
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE44 Abs
ente
eism
ty fr
om th
e fa
cilit
y on
an
unan
noun
ced
visi
t.
Freq
uenc
y of
dat
a co
llect
ion:
Adm
inis
trat
ive
data
is u
sual
ly a
naly
zed
mon
thly
, qua
rter
ly, a
nd/o
r ann
ually
. Fac
ility
sur
veys
are
usu
ally
ad
hoc
whe
n fu
ndin
g is
ava
ilabl
e.
Sugg
este
d co
mpa
rato
r: A
cros
s fa
cilit
ies,
pro
vide
rs.
Pote
ntia
l sou
rce
of d
ata:
Adm
inis
trat
ive
data
. Fac
ility
sur
veys
su
ch a
s W
HO
’s Se
rvic
e Av
aila
bilit
y an
d Re
adin
ess
Ass
essm
ent (
SARA
) (h
ttps
://w
ww
.who
.int/
heal
thin
fo/
syst
ems/
sara
_int
rodu
ctio
n/en
/).
Or W
orld
Ban
k’s
Serv
ice
Del
iver
y In
dica
tors
(SD
Is) (
http
s://w
ww
.sd
indi
cato
rs.o
rg/)
.
Benc
hmar
king
gui
delin
esN
o es
tabl
ishe
d be
nchm
ark,
the
low
er
the
bett
er. S
ugge
st b
ench
mar
king
ag
ains
t pas
t per
form
ance
or a
vera
ge
faci
lity/
prov
ider
.
Onl
y un
-exc
used
abs
ence
s sh
ould
be
acco
unte
d fo
r - fo
r exa
mpl
e if
prov
ider
is o
n a
trai
ning
, sic
k, o
r on
leav
e th
ey d
o no
t cou
nt a
s ab
sent
. It i
s al
so im
port
ant t
o ch
eck
whi
ch p
rovi
ders
are
cou
nted
whe
n do
ing
cros
s co
untr
y co
mpa
rato
rs.
Sour
ce: A
dmin
istr
ativ
e ho
spita
l dat
a, C
ount
ry X
, 201
7
Abs
ente
eism
am
ong
prim
ary
heal
th c
are
prov
ider
s, c
ount
ry a
vera
ge (%
)
Uga
nda
(201
3)To
go
(201
3)G
uine
a Bi
ssau
(2
018)
Nig
er
(201
5)N
iger
ia
(201
3)Si
erra
Le
one
(201
8)
Keny
a (2
013)
Mad
agas
car
(201
6)M
ozam
biqu
e (2
014)
Sene
gal
(201
0)Ta
nzan
ia
(201
4)
46.7
37.6
34.2
33.1
31.7
29.9
27.5
27.4
23.9
2014
.3
50 45 40 35 30 25 20 15 10 5 0
Absenteeism rate (%)
Out
puts
(man
agem
ent)
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE 45Pote
ntia
l sou
rce
of d
ata:
Adm
inis
trat
ive
data
from
fa
cilit
ies,
bill
ing
or c
laim
s da
ta.
Benc
hmar
king
gui
delin
esA
LOS:
No
esta
blis
hed
benc
hmar
k.
Sugg
est
benc
hmar
king
aga
inst
pa
st p
erfo
rman
ce o
r to
coun
try
hosp
ital a
vera
geBO
R: A
n op
timal
bed
occ
upan
cy
rate
of 8
5% h
as o
ften
bee
n ci
ted
in
the
liter
atur
e al
thou
gh it
is u
ncle
ar
whe
ther
thi
s is
bas
ed o
n cl
inic
al
indi
catio
n. H
owev
er, b
ed o
ccup
ancy
ra
tes
abov
e 85
% h
ave
been
obs
erve
d to
resu
lt in
regu
lar b
ed s
hort
ages
an
d hi
gher
hea
lthc
are-
acqu
ired
infe
ctio
ns.
Sour
ce: M
inis
try
of H
ealth
, Mal
aysi
a (2
017)
Leng
th o
f sta
y an
d by
ext
ensi
on b
ed o
ccup
ancy
rate
s ca
n be
a re
flect
ion
of t
he e
ffici
ency
of h
ospi
tal s
ervi
ces
as a
sho
rter
sta
y w
ill re
duce
cos
ts a
nd s
hift
the
car
e fr
om in
patie
nt t
o ou
tpat
ient
set
tings
. How
ever
, it
is d
ifficu
lt t
o co
mpa
re t
his
mea
sure
acr
oss
coun
trie
s as
tre
atm
ent
prot
ocol
s m
ay v
ary.
ALO
S m
ay a
lso
refle
ct t
he re
adin
ess
and/
or q
ualit
y of
out
patie
nt/p
rimar
y he
alth
car
e. T
here
fore
com
paris
on b
etw
een
hosp
itals
with
in-c
ount
ry is
reco
mm
ende
d. A
s th
e qu
ality
of
clai
ms
data
impr
oves
it is
reco
mm
ende
d to
com
pare
ALO
S by
cas
e-m
ix.
Ave
rage
leng
th o
f sta
y (d
ays)
vs
bed
occu
panc
y ra
te (%
) by
stat
e, M
alay
sia,
201
7
BOR
(%)
ALOS (days)
PERL
IS
PULA
U P
INA
NG
PERA
K
SELA
NG
OR
WP
KUA
LA L
AM
PUR
WP
PUTR
AJA
YA
WP
LABU
AN
NEG
ERI S
EMBI
LAN
MEL
AKA
JOH
OR PA
HA
NG
TERE
NG
GA
NU
KELA
NTA
N
SABA
H
SARA
WA
K
3
3.54
4.55
3540
4550
5560
6570
7580
8590
9510
0
Ave
rage
ALO
S: 3
.81
days
Ave
rage
BO
R: 6
4.2
%
KEH
DA
H
Aver
age
leng
th o
f sta
y (A
LOS)
vs.
Bed
occ
upan
cy ra
te (B
OR)
Sim
ple
defin
itio
n: A
LOS:
Tot
al a
vera
ge n
umbe
r of
bed
-day
s fo
r a
hosp
ital
adm
issi
on o
r di
scha
rge.
Bed
-day
s re
fer
to t
he n
umbe
r of
bed
s in
the
fac
ility
mul
tipl
ied
by t
he n
umbe
r of
day
s ob
serv
ed.
BOR:
The
per
cent
age
of b
ed-d
ays
that
are
occ
upie
d ov
er a
spe
cifie
d pe
riod
of t
ime.
Occ
upie
d be
d-da
ys r
efer
s to
the
num
ber
of d
ays
the
beds
wer
e oc
cupi
ed.
Freq
uenc
y of
dat
a co
llect
ion:
Adm
inis
trat
ive
data
is u
sual
ly c
olle
cted
dai
ly a
nd a
naly
zed
mon
thly
, qua
rter
ly, a
nd/o
r an
nual
ly.
Sugg
este
d co
mpa
rato
r: A
cros
s ho
spit
als,
typ
e of
hos
pita
ls.
WP
KUA
LA L
UM
PUR
Out
puts
(man
agem
ent)
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE46 Hos
pita
l re-
adm
issi
on ra
teSi
mpl
e de
finiti
on: P
erce
ntag
e of
unp
lann
ed a
nd u
nexp
ecte
d ho
spita
l rea
dmis
sion
s fo
r con
ditio
ns th
at a
re tr
acke
d th
roug
h th
e he
alth
sys
tem
like
acu
te m
yoca
rdia
l inf
arct
ion,
pne
umon
ia, a
sthm
a an
d di
abet
es.
Freq
uenc
y of
dat
a co
llect
ion:
Usu
ally
ana
lyze
d m
onth
ly o
r ann
ually
.
Sugg
este
d co
mpa
rato
r: A
cros
s fa
cilit
ies/
hosp
itals
.
Pote
ntia
l sou
rce
of d
ata:
Adm
inis
trat
ive
data
, cla
ims
data
, m
edic
al re
cord
s.
Benc
hmar
king
gui
delin
esN
o es
tabl
ishe
d be
nchm
ark.
Su
gges
t ben
chm
arki
ng a
gain
st p
ast
perf
orm
ance
or a
vera
ge fa
cilit
y.
Sour
ce: A
dmin
istr
ativ
e ho
spita
l dat
a, C
ount
ry X
, 201
7
This
indi
cato
r is
tric
ky to
mea
sure
acr
oss
coun
trie
s as
it is
influ
ence
d by
fact
ors
such
as
trea
tmen
t pro
toco
ls, p
rovi
der i
ncen
tives
, qua
lity
of c
are,
and
the
avai
labi
lity
of o
utpa
tient
and
com
mun
ity c
are.
For
thes
e re
ason
s, i
t is
sugg
este
d to
look
for o
utlie
r fac
ilitie
s or
pro
vide
rs w
ithin
cou
ntrie
s or
regi
ons.
If d
oing
cro
ss-c
ount
ry c
ompa
rison
, it i
s im
port
ant t
o ch
eck
how
read
mis
sion
s ar
e de
fined
– fo
r wha
t con
ditio
ns a
nd w
ithin
w
hat t
ime
perio
d. C
omm
on ti
me
perio
ds a
re 3
0-da
y an
d 90
-day
read
mis
sion
s an
d co
mm
on c
ondi
tions
incl
ude
chro
nic
obst
ruct
ive
pulm
onar
y di
seas
e, a
cute
myo
card
ial i
nfar
ctio
n, p
neum
onia
, ast
hma,
dia
bete
s, h
ip a
nd
knee
repl
acem
ents
. Ana
lysi
s is
als
o de
pend
ent o
n th
e qu
ality
of d
ata,
goo
d m
edic
al re
cord
kee
ping
, cod
ing
prac
tices
, and
repo
rtin
g co
mpl
ianc
e. T
his
indi
cato
r als
o re
quire
s un
ique
pat
ient
iden
tifier
s to
be
able
to tr
ack
patie
nts
in th
e he
alth
care
sys
tem
.
30-d
ay H
ospi
tal R
eadm
issi
on ra
te (%
), M
ay 2
017
Nat
iona
l A
vera
geH
ospi
tal 1
Hos
pita
l 2H
ospi
tal 3
Hos
pita
l 4H
ospi
tal 5
Hos
pita
l 6H
ospi
tal 7
22.3
25.1
21.2
16.6
22.4
22.1
22.3
25.4
7.3
8.2
5.1
6.6
12.3
6.9
7.3
4.4
30.0
25.0
20.0
15.0
10.0 5.0
0.0
Readmission rate (%)
Hip
repl
acem
ent
COPD
Out
puts
(man
agem
ent)
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE 47Cesa
rean
sect
ion
(C-s
ectio
n) ra
tes
Sim
ple
defin
ition
: The
per
cent
of p
regn
ant w
omen
who
hav
e a
C-se
ctio
n in
a s
peci
fic g
eogr
aphi
cal a
rea
and
refe
renc
e pe
riod.
Freq
uenc
y of
dat
a co
llect
ion:
Adm
inis
trat
ive
data
is u
sual
ly a
naly
zed
mon
thly
, qua
rter
ly, a
nd/o
r ann
ually
. Sur
vey
data
is u
sual
ly e
very
5 y
ears
.
Sugg
este
d co
mpa
rato
r: A
cros
s co
untr
ies;
acr
oss
faci
litie
s/pr
ovid
ers;
pas
t per
form
ance
/tre
nd d
ata;
by
urba
n/ru
ral;
by p
ublic
/priv
ate
faci
lity.
Pote
ntia
l sou
rce
of d
ata:
Num
erat
or: c
linic
al re
gist
ries
for d
ata
in a
giv
en g
eogr
aphi
cal a
rea
on th
e nu
mbe
r of C
-sec
tions
per
form
ed;
estim
ates
of t
he n
umbe
r of b
irths
in
that
are
a; p
opul
atio
n- b
ased
sur
veys
fo
r sel
f-re
port
ed C
-sec
tions
onl
y.
Den
omin
ator
: all
live
birt
hs d
urin
g th
e re
fere
nce
perio
d. W
here
dat
a on
the
num
ber o
f liv
e bi
rths
are
ab
sent
, eva
luat
ors
can
calc
ulat
e to
tal
estim
ated
live
birt
hs u
sing
cen
sus
data
fo
r the
tota
l pop
ulat
ion
and
crud
e bi
rth
rate
s in
a s
peci
fied
area
.
Benc
hmar
king
gui
delin
esU
NIC
EF/W
HO
/UN
FPA
reco
mm
end
a C-
sect
ion
rate
bet
wee
n 5
and
15 p
erce
nt o
f all
birt
hs, b
ased
on
estim
ates
from
a v
arie
ty o
f sou
rces
. Ra
tes
less
than
5%
can
indi
cate
in
adeq
uate
cov
erag
e or
acc
ess
to
emer
genc
y ob
stet
ric c
are.
Rat
es o
ver
15%
can
indi
cate
exc
essi
ve u
se o
f c-
sect
ions
resu
lting
in u
nnec
essa
ryex
posu
re to
sur
gery
and
ane
sthe
sia
risks
for w
omen
and
a w
aste
of
reso
urce
s. It
is im
port
ant t
o no
te th
at
the
refe
renc
e be
nchm
ark
is a
t the
po
pula
tion
leve
l whi
ch u
ses
all l
ive
birt
hs d
urin
g th
e re
fere
nce
perio
d as
th
e de
nom
inat
or.
As
trea
tmen
t pro
toco
ls m
ay v
ary
acro
ss c
ount
ries,
com
paris
on w
ithin
cou
ntry
is a
lso
reco
mm
ende
d. F
or w
ithin
cou
ntry
com
paris
on, a
n al
tern
ativ
e in
dica
tor i
s th
e pr
opor
tion
of fa
cilit
y de
liver
ies
that
are
C-s
ectio
ns.
Her
e, th
e so
urce
of d
ata
can
be fr
om fa
cilit
y or
cla
ims
data
. The
den
omin
ator
is li
ve b
irths
at t
he fa
cilit
y no
t all
live
birt
hs w
ithin
a c
atch
men
t are
a. T
his
indi
cato
r may
be
sign
ifica
ntly
hig
her t
han
c-se
ctio
n ra
tes
at th
e po
pula
tion
leve
l bec
ause
it d
oes
not a
djus
t for
hig
her r
isk
preg
nanc
ies
and
refe
rral
s. W
hile
spe
cify
ing
an a
ppro
pria
te ra
nge
of ta
rget
per
cent
ages
with
in a
faci
lity
is im
prac
tical
, it m
ight
stil
l be
inte
rest
ing
to
benc
hmar
k ag
ains
t ave
rage
faci
litie
s/pr
ovid
ers
of s
imila
r typ
e to
look
for o
utlie
rs. T
he R
obso
n Sc
ale
has
bee
n ap
plie
d in
som
e se
ttin
gs to
hel
p m
onito
r the
rate
s of
c-s
ectio
n am
ongs
t sub
grou
ps o
f pre
gnan
t wom
en
(e.g
. hig
h ris
k ve
rsus
low
risk
) how
ever
this
will
like
ly re
quire
acc
ess
to p
atie
nt h
isto
ry a
nd m
edic
al re
cord
s.
(Lin
k to
Rob
son
Scal
e:ht
tps:
//ap
ps.w
ho.in
t/iri
s/bi
tstr
eam
/han
dle/
1066
5/25
9512
/978
9241
5131
97-e
ng.p
df;js
essi
onid
=A04
8ECA
988C
32D
049A
D25
B2E2
D07
E68A
?seq
uenc
e=1)
Caes
area
n se
ctio
n ra
tes,
201
5 (o
r nea
rest
yea
r)
Turkey
Portugal
United Kingdom
Estonia
Korea
Slovak Rep
ublic
Spain
Norway
Italy
OECD33
France
Finland
Mexico
United States
Canada
Sweden
Hungary
Ireland
Denmark
Iceland
Australia
Luxembourg
Belgium
Chile
Germany
Czech
Republic
Israel
Poland
Austria
Latvia
Netherl
ands
Switzerl
and
New Zea
land
Slovenia
53.1
46.8
46.0
38.0
37.2
36.2
35.3
34.0
32.9
32.3
32.2
30.2
30.2
30.1
28.7
27.927.8
26.0
26.3
25.4
24.5
21.1
21.0
20.8
20.8
20.8
18.7
17.3
16.216.1
16.0
15.9
15.5
26.2
60 50 40 30 20 10 0
Per 1
00 li
ve b
irths
Out
puts
(man
agem
ent)
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE48
Yello
w fe
ver
(5 d
ose)
vac
cine
9900
7119
4835
4364
3848
2975
Vic
ryl H
S-37
mm
1
Hal
oper
idol
tab
1mg
Stri
p (b
lood
ch
oles
tero
l)
Mag
nesi
um s
ulfa
te
past
e 0%
(100
mg)
Chol
era
vacc
ine
(2 d
ose
vial
)
Pote
ntia
l Sou
rce
Faci
lity
base
d su
rvey
s, a
dmin
istr
ativ
e da
ta, S
ervi
ce P
rovi
sion
Ass
essm
ents
. If
ava
ilabl
e, L
ogis
tics
Man
agem
ent
and
Info
rmat
ion
Syst
em (L
MIS
) or
inve
ntor
y m
anag
emen
t sy
stem
Benc
hmar
king
gui
delin
esId
eally
zer
o. S
ugge
st b
ench
mar
king
ag
ains
t pa
st p
erfo
rman
ce o
r av
erag
e fa
cilit
y
Loss
due
to
expi
red
drug
s in
Hos
pita
l A, T
op 6
item
s in
USD
, 201
7
Sour
ce: I
nven
tory
man
agem
ent s
yste
m, C
ount
ry X
. 201
7
mig
ht in
clud
e th
e or
der t
urna
roun
d tim
e (i.
e. t
he n
umbe
r of d
ays
betw
een
whe
n th
e or
der w
as re
ceiv
ed a
nd d
eliv
ered
), th
e fr
eque
ncy
of o
rder
s an
d th
e pr
escr
iptio
n ra
te o
r util
izat
ion
of c
omm
only
exp
ired
prod
ucts
.
Expi
red
drug
s P
ropo
rtio
n of
pro
duct
s pa
st t
heir
use-
by d
ate
in s
tock
ove
r th
e to
tal a
mou
nt o
f pro
duct
in s
tock
. Can
als
o be
mea
sure
d in
mon
etar
y va
lue
if th
e qu
anti
ty o
f di
scar
ded
expi
red
drug
is r
ecor
ded
alon
g w
ith
the
mar
ket
pric
e of
the
dru
g.
Freq
uenc
y of
dat
a co
llect
ion:
Adm
inis
trat
ive
data
cou
ld b
e co
llect
ed m
onth
ly, y
earl
y. F
acili
ty s
urve
ys a
re u
sual
ly a
d ho
c w
hen
fund
ing
is a
vaila
ble.
Inve
ntor
y or
logi
stic
s m
anag
emen
t sy
stem
s ca
n pr
oduc
e da
ily/r
eal-
tim
e re
port
s.
Sugg
este
d co
mpa
rato
r: A
cros
s fa
cilit
ies,
pha
rmac
ies,
med
ical
sto
rage
cen
ters
.
Pote
ntia
l sou
rce
of d
ata:
Faci
lity
base
d su
rvey
s, a
dmin
istr
ativ
eda
ta, S
ervi
ce P
rovi
sion
Ass
essm
ents
.If
avai
labl
e, L
ogis
tics
Man
agem
ent
and
Info
rmat
ion
Syst
em (L
MIS
) or
inve
ntor
y m
anag
emen
t sys
tem
.
Be
nchm
arki
ng g
uide
lines
Idea
lly z
ero.
Sug
gest
ben
chm
arki
ngag
ains
t pa
st p
erfo
rman
ce o
rav
erag
e fa
cilit
y.
Out
puts
(man
agem
ent)
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE 49
180
days
120
days
10-6
9 da
ys
6 da
ys60-8
3 da
ys 40 d
ays80
day
s55
day
s
051015202530354045
Faci
lity
1 Faci
lity
2Faci
lity
3
Faci
lity
4Faci
lity
5 Faci
lity
6Faci
lity
7 Faci
lity
8Faci
lity
9 Faci
lity
10Faci
lity
11 Faci
lity
12Faci
lity
13 Faci
lity
14Faci
lity
15 Faci
lity
16
6-55
day
s 4-40
day
s
Pote
ntia
l sou
rce
of d
ata:
Faci
lity
base
d su
rvey
s, a
dmin
istr
ativ
e da
ta, S
ervi
ce P
rovi
sion
Ass
essm
ents
. If
ava
ilabl
e, L
ogis
tics
Man
agem
ent
and
Info
rmat
ion
Syst
em (L
MIS
) or
inve
ntor
y m
anag
emen
t sy
stem
.
Benc
hmar
king
gui
delin
esId
eally
zer
o. S
ugge
st b
ench
mar
king
ag
ains
t pa
st p
erfo
rman
ce o
r av
erag
e fa
cilit
y.
Stoc
k-ou
t ra
te (%
) and
num
ber o
f day
s it
ems
wer
e no
t av
aila
ble,
ac
ross
faci
litie
s, 2
017
Sour
ce: I
nven
tory
man
agem
ent s
yste
m, C
ount
ry X
. 201
7
Just
like
exp
ired
prod
ucts
, dru
g st
ocko
uts
coul
d al
so b
e a
refle
ctio
n of
wea
knes
ses
thro
ugho
ut t
he s
uppl
y ch
ain.
Lac
k of
fund
s an
d/or
com
plic
ated
and
leng
thy
proc
edur
es t
o ac
cess
fund
s an
d or
der d
rugs
cou
ld b
e an
issu
e. H
ere
too,
in
dica
tors
suc
h as
the
pro
port
ion
of e
xpire
d or
ruin
ed d
rugs
, the
ord
er t
urna
roun
d tim
e, t
he fr
eque
ncy
of o
rder
s, a
nd t
he p
resc
riptio
n ra
te o
r util
izat
ion
of s
tock
ed-o
ut it
ems
mig
ht h
elp
faci
litie
s/ph
arm
acie
s/ce
ntra
l med
ical
sto
res
bett
er
fore
cast
and
man
age
drug
inve
ntor
ies.
Dru
g st
ocko
uts
Sim
ple
defin
itio
n: N
umbe
r or
per
cent
of
serv
ice
deliv
ery
unit
s th
at e
xper
ienc
e a
stoc
kout
of a
par
ticu
lar
prod
uct
that
sho
uld
be in
sto
ck o
ver
a de
fined
per
iod
of t
ime.
It c
ould
als
o be
defi
ned
as n
umbe
r, or
pe
rcen
t of
pro
duct
s th
at a
re u
nava
ilabl
e at
a p
arti
cula
r se
rvic
e de
liver
y un
it.
Freq
uenc
y of
dat
a co
llect
ion:
Adm
inis
trat
ive
data
cou
ld b
e co
llect
ed m
onth
ly, y
earl
y. F
acili
ty s
urve
ys a
re u
sual
ly a
d ho
c w
hen
fund
ing
is a
vaila
ble.
Inve
ntor
y or
logi
stic
s m
anag
emen
t sy
stem
s ca
n pr
oduc
e da
ily/r
eal-
tim
e re
port
s.
Sugg
este
d co
mpa
rato
r: A
cros
s fa
cilit
ies,
pha
rmac
ies,
med
ical
sto
rage
cen
ters
.
Out
puts
(man
agem
ent)
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE50 Pote
ntia
l sou
rce
of d
ata:
Clai
ms
and
prem
ium
col
lect
ion
data
.
Benc
hmar
king
gui
delin
esTh
e “e
xpec
ted”
cla
ims
ratio
sho
uld
be 1
00%
min
us a
mar
gin
for
adm
inis
trat
ive
expe
nses
and
a
reas
onab
le le
vel o
f cla
im re
serv
es.
For a
priv
ate
insu
ranc
e co
mpa
ny,
a m
argi
n fo
r pro
fit w
ould
als
o be
su
btra
cted
. Fo
r mat
ure
natio
nal h
ealt
h in
sura
nce
syst
ems,
one
wou
ld e
xpec
t th
e ra
tio
to b
e ab
out
90%
.
Clai
ms
rati
o (%
) by
mem
bers
hip
grou
p, 2
014-
2017
Sour
ce: N
atio
nal h
ealth
insu
ranc
e ag
ency
cla
ims
and
prem
ium
dat
abas
e (2
019)
If t
he c
laim
s ra
tio is
regu
larly
ove
r 100
% o
ver a
n ex
tend
ed n
umbe
r of y
ears
it m
eans
eith
er t
he p
rem
ium
s ar
e no
t be
ing
prop
erly
cal
cula
ted
usin
g so
und
actu
aria
l prin
cipa
ls a
nd/o
r the
re a
re is
sues
on
the
clai
ms
paym
ent
and
adm
inis
trat
ive
expe
nse
side
. If c
ompa
ring
acro
ss in
sura
nce
sche
mes
, it
is im
port
ant
to a
ccou
nt fo
r diff
eren
ces
in p
rem
ium
s, m
embe
rshi
p ty
pes,
cov
erag
e ra
tes,
and
cos
t co
ntai
nmen
t fe
atur
es.
Percentage (%)
Poor
and
nea
r po
orPr
ivat
e fo
rmal
Info
rmal
sec
tor
Reti
red/
pens
ione
rs
2014
2015
2016
2017
Clai
ms
ratio
Sim
ple
defin
itio
n: A
rat
io c
alcu
late
d as
acc
rued
cla
ims
divi
ded
by a
ccru
ed p
rem
ium
s. T
his
is s
omet
imes
als
o re
ferr
ed t
o as
the
loss
rat
io.
Freq
uenc
y of
dat
a co
llect
ion:
Mon
thly
, qua
rter
ly, a
nnua
lly.
Sugg
este
d co
mpa
rato
r: B
y m
embe
rshi
p ty
pe, a
gain
st p
ast
perf
orm
ance
. If
mor
e th
an o
ne s
chem
e by
insu
ranc
e sc
hem
e.
600
500
400
300
200
100 0
Out
puts
(man
agem
ent)
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE 51
0%20%
40%
60%
80%
100%
120%
Gen
eral
Pub
licSe
rvic
esDe
fenc
ePu
blic
Ord
eran
d Sa
fety
Econ
omic
Affai
rsEn
viro
nmen
tal
Prot
ec�o
nHo
usin
g an
dCo
mm
unity
Amen
i�es
Heal
thTo
urism
and
Crea�v
eEc
onom
y
Relig
ion
Educ
a�on
Socia
lPr
otec�o
n
Adju
sted
bud
get
Orig
inal
bud
get
Pote
ntia
l sou
rce
of d
ata:
No
know
n gl
obal
com
para
tor
data
base
. Fin
anci
al M
anag
emen
t In
form
atio
n Sy
stem
s (F
MIS
) or
Publ
ic E
xpen
ditu
re a
nd F
inan
cial
A
ccou
ntab
ility
Fra
mew
orks
(PEF
A)
give
exe
cutio
n ra
tes
for g
over
nmen
t as
a w
hole
and
by
sect
or. P
ublic
Ex
pend
iture
Rev
iew
s (P
ERs)
may
als
o ha
ve th
is in
form
atio
n al
thou
gh th
e m
etho
dolo
gy a
nd fr
eque
ncy
are
less
co
nsis
tent
.Th
e W
orld
Ban
k’s
BOO
ST in
itiat
ive
(dep
loye
d in
~40
coun
trie
s gl
obal
ly)
prov
ides
bud
get a
nd e
xpen
ditu
re
data
at a
mor
e gr
anul
ar le
vel b
y go
vern
men
t lev
el (c
entr
al o
r loc
al),
adm
inis
trat
ive
unit
(min
istr
y,
depa
rtm
ent,
agen
cy, u
nive
rsity
, ho
spita
l or s
choo
l), e
cono
mic
serv
ices
, cap
ital e
xpen
ses)
, fun
ctio
nal
sour
ce (b
udge
t, do
mes
tic o
r for
eign
bo
rrow
ing)
.ht
tps:
//pef
a.or
g/;
http
://gl
obal
prac
tice
s.w
orld
bank
.org
/go
vern
ance
/Pag
es/S
iteP
ages
/PER
%20
aspx
;ht
tp://
sear
ch.w
orld
bank
.org
/per
Man
y of
the
reas
ons
for p
oor B
ERs
are
gene
rally
out
side
the
cont
rol o
f the
MO
H (e
.g. w
eak
capa
city
to fo
reca
st re
venu
es, d
elay
s in
the
rele
ase
of fu
nds,
in-y
ear b
udge
t adj
ustm
ents
, or r
igid
ities
on
how
fund
s ca
n be
spe
nt a
cros
s pr
ogra
ms)
. For
asse
ssm
ent o
f exe
cutio
n ra
tes
is a
lso
extr
emel
y he
lpfu
l. Fo
r ins
tanc
e, M
inis
trie
s th
at a
re v
ery
hum
an re
sour
ce in
tens
e (li
ke e
duca
tion
and
heal
th) a
re li
kely
to h
ave
a be
tter
exe
cutio
n ra
te th
an o
ther
s be
caus
e w
ages
are
qua
si s
tatu
tory
. For
this
re
ason
, it m
ight
mak
e m
ore
sens
e to
com
pare
the
BER
of w
ages
and
sal
arie
s, n
on-w
age
recu
rren
t exp
endi
ture
s, a
nd in
vest
men
t exp
endi
ture
s in
the
sect
or s
epar
atel
y.
A=a
ggre
gate
exp
endi
ture
out
turn
w
as b
etw
een
95%
and
105
% o
f th
e ap
prov
ed a
ggre
gate
bud
gete
d ex
pend
iture
in a
t le
ast
two
of t
he
last
thr
ee y
ears
.B=
aggr
egat
e ex
pend
iture
out
turn
w
as b
etw
een
90%
and
110
% o
f th
e ap
prov
ed a
ggre
gate
bud
gete
d ex
pend
iture
in a
t le
ast
two
of t
he
last
thr
ee y
ears
.C=
aggr
egat
e ex
pend
iture
out
turn
w
as b
etw
een
85%
and
115
% o
f th
e ap
prov
ed a
ggre
gate
bud
gete
d ex
pend
iture
in a
t le
ast
two
of t
he
last
thr
ee y
ears
.D
=per
form
ance
is le
ss t
han
requ
ired
for a
C s
core
.
Sour
ce: P
EFA
Var
ious
Yea
rsBudg
et E
xecu
tion
Rat
es, b
y se
ctor
, Ind
ones
ia, 2
016,
%Be
nchm
arki
ng g
uide
lines
Whi
le t
here
are
no
esta
blis
hed
benc
hmar
ks t
o as
sess
hea
lth
sect
or
budg
et e
xecu
tion
rate
s, t
here
ar
e PE
FA s
corin
g gu
idel
ines
for
aggr
egat
e ex
pend
iture
out
turn
whi
ch
coul
d be
app
lied
to a
sses
s th
e he
alth
se
ctor
bud
get
relia
bilit
y.
Budg
et e
xecu
tion
rate
s (B
ERs)
by
agg
rega
te e
xpen
ditu
re o
uttu
rn).
Freq
uenc
y of
dat
a co
llect
ion:
revi
ews
are
less
freq
uent
.
Sugg
este
d co
mpa
rato
r: A
cros
s co
untr
ies;
Oth
er s
ecto
rs/l
ine
min
istr
ies;
by
expe
ndit
ure
cate
gory
(e.g
. cap
ital
vs
recu
rren
t); t
rend
dat
a.
Out
puts
(man
agem
ent)
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE52 Life
exp
ecta
ncy
at b
irth
Pote
ntia
l sou
rce
of d
ata:
Vita
l reg
istr
y da
ta. H
ouse
hold
su
rvey
s su
ch a
s th
e D
emog
raph
ic
and
Hea
lth
Surv
ey (D
HS)
, th
e U
NIC
EF M
ultip
le In
dica
tor
Clus
ter S
urve
y (M
ICS)
.ht
tps:
//dhs
prog
ram
.com
http
s://m
ics.
unic
ef.o
rg
Benc
hmar
king
gui
delin
esN
o es
tabl
ishe
d be
nchm
ark
the
high
er t
he b
ette
r. Su
gges
t al
so b
ench
mar
king
aga
inst
pa
st p
erfo
rman
ce o
r rel
evan
t re
gion
al/in
com
e co
untr
y co
mpa
rato
rs/a
vera
ges.
Life
exp
ecta
ncy
(in y
ears
), 20
17
Sour
ce: W
orld
Ban
k (2
019)
. Wor
ld D
evel
opm
ent I
ndic
ator
s. (h
ttps
://da
taba
nk.w
orld
bank
.org
/sou
rce/
wor
ld-d
evel
opm
ent-
indi
cato
rs)
Ave
rage
num
ber
of y
ears
tha
t a
new
born
is e
xpec
ted
to li
ve if
cur
rent
mor
talit
y ra
tes
cont
inue
to
appl
y.
Freq
uenc
y of
dat
a co
llect
ion:
Vit
al r
egis
try
data
sho
uld
be a
nnua
l. Su
rvey
dat
a us
ually
eve
ry 5
yea
rs.
Sugg
este
d co
mpa
rato
r:
The
rela
tive
str
aigh
tfor
war
d an
d st
anda
rdiz
ed m
easu
rabi
lity
of t
hese
indi
cato
rs a
llow
s fo
r co
mpa
rabi
lity
acro
ss c
ount
ries.
It
may
als
o be
of
inte
rest
to
polic
y m
aker
s to
com
pare
by
soci
oeco
nom
ic s
tatu
s (e
.g. w
ealt
h qu
inti
le, l
evel
of e
duca
tion
, gen
der)
.
As
a hi
gh le
vel o
utco
me
indi
cato
r, it
will
be
impo
rtan
t to
choo
se re
leva
nt h
ealth
sys
tem
inpu
t, o
utpu
t, a
nd p
roce
ss m
easu
res
in d
evel
opin
g a
resu
lts c
hain
nar
rativ
e.
Out
com
es (h
ealt
h st
atus
)
77
75
72
7070
68
66
63
Mal
dive
s
Sri L
anka
Bang
lade
sh
Nep
al
Bhut
an
Indi
a
Paki
stan
Afgh
anis
tan
Life
exp
ecta
ncy
in S
outh
Asia
, 201
7
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE 53Pote
ntia
l sou
rce
of d
ata:
Vita
l reg
istr
y da
ta. H
ouse
hold
su
rvey
s su
ch a
s t
he D
emog
raph
ic
and
Hea
lth
Surv
ey (D
HS)
, th
e U
NIC
EF M
ultip
le In
dica
tor C
lust
er
Surv
ey (M
ICS)
.ht
tps:
//dhs
prog
ram
.com
http
s://m
ics.
unic
ef.o
rg
Benc
hmar
king
gui
delin
esN
o es
tabl
ishe
d be
nchm
ark,
the
low
er
the
bett
er. S
ugge
st b
ench
mar
king
ag
ains
t pa
st p
erfo
rman
ce o
r re
leva
nt re
gion
al/in
com
e co
untr
y co
mpa
rato
rs/a
vera
ges.
Sus
tain
able
D
evel
opm
ent
Goa
l tar
gets
aim
to
redu
ce u
nder
5 m
orta
lity
to le
ss t
han
25 p
er 1
000
live
birt
hs b
y 20
30.
Und
er 5
-mor
talit
y ra
te (p
er 1
,000
), 20
17
010
2030
4050
6070
8090
100
110
120
130
Sub-
Saha
ran
Afr
ica
Mau
ritiu
s
Mal
dive
s
Cuba
Japa
n
Isra
el
Slov
enia
Cana
da Uni
ted
Stat
es
Turk
men
ista
n
Djib
outi
Lao
PDR
Hai
ti
Paki
stan
Som
alia
East
Asi
a &
Pac
i�c
Nor
th A
mer
ica
Sout
h A
sia
Lati
n A
mer
ica
&
Cari
bbea
n
Mid
dle
East
&
Nor
th A
fric
a
Euro
pe &
Ce
ntra
l Asi
a
Und
er-fi
ve m
orta
lity
Sim
ple
defin
itio
n: U
nder
-five
mor
talit
y (U
5M) r
ate
is t
he p
roba
bilit
y th
at a
chi
ld w
ill d
ie b
efor
e re
achi
ng a
ge fi
ve.
Freq
uenc
y of
dat
a co
llect
ion:
Vit
al r
egis
try
data
sho
uld
be a
nnua
l. Su
rvey
dat
a us
ually
eve
ry 5
yea
rs.
Sugg
este
d co
mpa
rato
r: T
he r
elat
ive
stra
ight
forw
ard
and
stan
dard
ized
mea
sura
bilit
y of
the
se in
dica
tors
allo
ws
for
com
para
bilit
y ac
ross
cou
ntrie
s. It
may
als
o be
of
inte
rest
to
polic
y m
aker
s to
com
pare
by
regi
on (e
.g. p
rovi
nce,
urb
an/r
ural
), or
soc
ioec
onom
ic s
tatu
s (e
.g. w
ealt
h qu
inti
le, l
evel
of e
duca
tion
, gen
der)
.
Not
e: B
est a
nd w
orst
cou
ntrie
s by
regi
onSo
urce
: Wor
ld B
ank
(201
9). W
orld
Dev
elop
men
t Ind
icat
ors
http
s://d
atab
ank.
wor
ldba
nk.o
rg/s
ourc
e/w
orld
-dev
elop
men
t-in
dica
tors
Neo
nata
l (le
ss th
an 2
8 da
ys) a
nd in
fant
(les
s th
an 1
yea
r) m
orta
lity
rate
s ar
e im
port
ant p
arts
of t
he o
vera
ll un
der-
5 m
orta
lity
rate
. Dis
aggr
egat
ed a
naly
ses
can
help
iden
tify
vuln
erab
le p
opul
atio
ns fo
r bet
ter t
arge
ting.
As
mos
t U5M
dea
ths
are
prev
enta
ble,
it w
ill b
e im
port
ant t
o ch
oose
rele
vant
inpu
t, o
utpu
t and
pro
cess
indi
cato
rs in
dev
elop
ing
resu
lts c
hain
nar
rativ
e. F
or e
xam
ple,
num
ber o
f nur
ses,
pos
tnat
al c
over
age,
fully
-im
mun
ized
chi
ldre
n, re
leva
nt c
hild
-hea
lth s
peci
fic re
adin
ess
mea
sure
s (S
ALA
(htt
ps://
ww
w.w
ho.in
t/he
alth
info
/sys
tem
s/sa
ra_i
ntro
duct
ion/
en/)
or S
DI (
http
s://w
ww
.sdi
ndic
ator
s.or
g), a
nd o
ther
s su
ch a
s di
agno
stic
acc
urac
y of
com
mon
chi
ldho
od il
lnes
ses
like
diar
rhea
or p
neum
onia
, and
adh
eren
ce to
clin
ical
gu
idel
ines
.
Out
com
es (h
ealt
h st
atus
)
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE54 Pote
ntia
l sou
rce
of d
ata:
Vita
l reg
istr
y da
ta, h
ealt
h se
rvic
es
reco
rds,
cen
sus.
Hou
seho
ld s
urve
ys
such
as
the
Dem
ogra
phic
and
Hea
lth
Surv
ey (D
HS)
, the
UN
ICEF
Mul
tiple
In
dica
tor C
lust
er S
urve
y (M
ICS)
.ht
tps:
//dhs
prog
ram
.com
http
s://m
ics.
unic
ef.o
rg
Benc
hmar
king
gui
delin
esN
o es
tabl
ishe
d be
nchm
ark,
th
e lo
wer
the
bet
ter.
Sugg
est
benc
hmar
king
aga
inst
pas
t pe
rfor
man
ce o
r rel
evan
t re
gion
al/
inco
me
coun
try
com
para
tors
/av
erag
es. S
usta
inab
le D
evel
opm
ent
Goa
l tar
gets
aim
to
redu
ce m
ater
nal
mor
talit
y to
less
tha
n 70
per
10
0,00
0 liv
e bi
rths
by
2030
.
Mat
erna
l mor
talit
y pe
r 100
,000
live
birt
hs (1
990-
2016
)
Sour
ce: W
orld
Ban
k (2
019)
. Wor
ld D
evel
opm
ent I
ndic
ator
s. (h
ttps
://da
taba
nk.w
orld
bank
.org
/sou
rce/
wor
ld-d
evel
opm
ent-
indi
cato
rs)
As
mos
t mat
erna
l dea
ths
are
prev
enta
ble,
it w
ill b
e im
port
ant t
o ch
oose
rele
vant
inpu
t, o
utpu
t and
pro
cess
indi
cato
rs in
dev
elop
ing
the
resu
lts c
hain
nar
rativ
e. F
or e
xam
ple,
num
ber o
f mid
wiv
es, a
nten
atal
car
e co
vera
ge, a
cces
s to
ski
lled
care
dur
ing
http
s://w
ww
.who
.int/
heal
thin
fo/s
yste
ms/
sara
_int
rodu
ctio
n/en
/) or
SD
I (ht
tps:
//ww
w.s
dind
icat
ors.
org)
, and
oth
ers
such
as
diag
nost
ic a
ccur
acy,
adh
eren
ce to
clin
ical
gu
idel
ines
. Mat
erna
l dea
th a
udits
are
als
o an
impo
rtan
t add
ition
al s
ourc
e of
info
rmat
ion
to e
xpla
in c
ause
s of
mat
erna
l mor
talit
y.
Rate per 100,000 live births
The
annu
al n
umbe
r of
mat
erna
l dea
ths
from
any
cau
se r
elat
ed t
o or
agg
rava
ted
by p
regn
ancy
or
its
man
agem
ent,
rel
ativ
e to
the
tot
al n
umbe
r of
live
birt
hs. T
he in
dica
tor
is r
epor
ted
as d
eath
s pe
r 10
0,00
0 liv
e bi
rths
.
Freq
uenc
y of
dat
a co
llect
ion:
Vit
al r
egis
try
data
sho
uld
be a
nnua
l. Su
rvey
dat
a oc
curs
usu
ally
eve
ry 5
yea
rs.
Sugg
este
d co
mpa
rato
r: T
he s
tand
ardi
zed
mea
sura
bilit
y of
the
se in
dica
tors
allo
ws
for c
ompa
rabi
lity
acro
ss c
ount
ries.
It m
ay a
lso
be o
f int
eres
t to
pol
icy
mak
ers
to c
ompa
re b
y re
gion
(e.g
. pro
vinc
e, u
rban
/rur
al),
or s
ocio
econ
omic
sta
tus
(e.g
. wea
lth
quin
tile
, lev
el o
f edu
cati
on, g
ende
r).
Mat
erna
l mor
talit
y ra
teO
utco
mes
(hea
lth
stat
us)
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE 55
Sim
ple
defin
itio
n: T
he b
urde
n of
TB
dise
ase
can
be m
easu
red
in t
erm
s of
: i) I
ncid
ence
– t
he n
umbe
r of
new
and
rel
apse
cas
es o
f TB
aris
ing
in a
giv
en t
ime
perio
d, u
sual
ly 1
yea
r; ii
) Pre
vale
nce
– th
e nu
mbe
r of
ca
ses
of T
B at
a g
iven
poi
nt in
tim
e; a
nd ii
i) M
orta
lity
– th
e nu
mbe
r of
dea
ths
caus
ed b
y TB
in a
giv
en t
ime
perio
d, u
sual
ly 1
yea
r.
Freq
uenc
y of
dat
a co
llect
ion:
TB
noti
ficat
ions
/sur
veill
ance
dat
a ar
e id
eally
cap
ture
d in
rea
l-ti
me
and
rev
iew
ed w
eekl
y or
mon
thly
. Pre
vale
nce
surv
eys
are
at le
ast
5 ye
ars
apar
t.
Sugg
este
d co
mpa
rato
r: A
cros
s co
untr
ies,
by
regi
on, h
igh
risk
popu
lati
ons
(e.g
. peo
ple
livin
g w
ith
HIV
, inj
ecti
ng d
rug
user
s, p
rison
s, n
ursi
ng h
omes
, hea
lthc
are
wor
kers
).
Indi
a
Phili
ppin
esPa
kist
an
Indo
nesi
a
Chin
aSo
uth
Afric
aN
iger
ia
DR C
ongo
Bang
lade
sh
UR T
anza
Mya
nmar
Moz
ambi
que
Keny
a
Ethi
opia
Viet
Na
Thai
land
DPR
Kor
Ango
la
Russ
ian
Fede
rati
onBr
azil
TB b
urde
n
Pote
ntia
l sou
rce
of d
ata:
The
WH
O p
rodu
ces
a ye
arly
Glo
bal
TB re
port
. The
re a
re a
lso
i) TB
pr
eval
ence
sur
veys
– a
met
hod
used
if
notifi
catio
n da
ta fr
om ro
utin
e su
rvei
llanc
e ar
e t
houg
ht t
o be
in
accu
rate
or i
ncom
plet
e an
d w
hen
ther
e is
an
estim
ated
rela
tivel
y hi
gh p
reva
lenc
e of
TB
(mor
e th
an
100
per 1
00,0
00 p
opul
atio
n) a
s th
ey a
re e
xpen
sive
to
cond
uct;
ii)
TB n
otifi
catio
ns –
a m
etho
d us
ed
by 1
44 c
ount
ries
with
low
leve
ls
of u
nder
repo
rtin
g (g
ener
ally
hig
h in
com
e an
d so
me
sele
cted
mid
dle-
inco
me
coun
trie
s); a
nd ii
i) N
atio
nal
inve
ntor
y st
udie
s –
thes
e m
easu
re
the
leve
l of u
nder
repo
rtin
g of
de
tect
ed T
B ca
ses
(met
hod
used
in 6
co
untr
ies)
; and
iv) C
ase
notifi
catio
n da
ta c
ombi
ned
with
exp
ert
opin
ion
abou
t ca
se-d
etec
tion
gaps
– t
his
met
hod
is t
he le
ast
pref
erre
d an
d it
is re
lied
upon
onl
y if
one
of t
he
othe
r thr
ee m
etho
ds c
anno
t be
use
d (c
urre
ntly
use
d in
43
coun
trie
s).
Benc
hmar
king
gui
delin
esSD
G 3
incl
udes
a t
arge
t to
end
th
e gl
obal
TB
epid
emic
by
2030
, w
ith T
B in
cide
nce
(new
and
rela
pse
case
s pe
r 100
,000
pop
ulat
ion
per
year
) defi
ned
as t
he in
dica
tor f
or
mea
sure
men
t of
pro
gres
s. T
he 2
030
targ
ets
set
in t
he E
nd T
B St
rate
gy a
re
a 90
% re
duct
ion
in T
B de
aths
and
an
80%
redu
ctio
n in
TB
inci
denc
e,
com
pare
d w
ith le
vels
in 2
015.
TB b
urde
n: 2
0 hi
ghes
t co
untr
ies,
Inci
denc
e nu
mbe
r in
thou
sand
s, 2
017
Sour
ce: W
HO
(201
8). G
loba
l Tub
ercu
losi
s Re
port
201
8. h
ttps
://w
ww
.who
.int/
tb/p
ublic
atio
ns/g
loba
l_re
port
/en
The
ultim
ate
goal
is t
o di
rect
ly m
easu
re T
B bu
rden
from
TB
notifi
catio
ns in
all
coun
trie
s. N
otifi
catio
ns a
re c
onsi
dere
d a
good
pro
xy o
f inc
iden
ce in
cou
ntrie
s th
at h
ave
high
-per
form
ance
sur
veill
ance
sys
tem
s an
d go
od a
cces
s to
qua
lity
care
th
at re
sult
in fe
w u
ndia
gnos
ed c
ases
. As
the
surv
eilla
nce
syst
em im
prov
es a
nd t
he n
umbe
r of c
ases
dec
reas
es, c
ount
ries
no lo
nger
und
erta
ke e
xpen
sive
TB
surv
eys
that
wou
ld re
quire
larg
er a
nd la
rger
sam
ple
size
s. T
o as
sess
pro
gres
s m
ade
due
to a
pro
gram
or i
nter
vent
ion
it m
ay b
e m
ore
info
rmat
ive
to a
lso
look
at
indi
cato
rs a
long
the
ent
ire T
B ca
scad
e in
clud
ing
othe
r TB
indi
cato
rs s
uch
as T
B ca
se d
etec
tion,
TB
notifi
catio
n, T
B tr
eatm
ent
succ
ess
rate
s, T
B-se
rvic
e sp
ecifi
c re
adin
ess
mea
sure
s (S
ARA
htt
ps://
ww
w.w
ho.in
t/he
alth
info
/sys
tem
s/sa
ra_r
epor
ts/e
n/ a
nd S
DI h
ttps
://w
ww
.sdi
ndic
ator
s.or
g/in
dica
tors
) and
dia
gnos
tic a
ccur
acy
and
adhe
renc
e to
clin
ical
gui
delin
es.
51-1
00
101-
150
151-
200
201-
500
501-
1000
>100
0
Out
com
es (h
ealt
h st
atus
)
Per
cent
age
of h
ouse
hold
s w
ho s
pend
mor
e th
an 1
0% o
f the
ir to
tal h
ouse
hold
con
sum
ptio
n on
hea
lth
expe
ndit
ures
.
Freq
uenc
y of
dat
a co
llect
ion:
Usu
ally
eve
ry 5
yea
rs h
owev
er s
ome
regi
ons/
coun
trie
s co
llect
dat
a on
an
annu
al b
asis
. The
Hea
lth
Equi
ty a
nd F
inan
cial
Pro
tect
ion
Indi
cato
rs d
atab
ase
will
be
upda
ted
annu
ally
.
Sugg
este
d co
mpa
rato
r: A
cros
s co
untr
ies.
Som
e su
rvey
s al
so p
rodu
ce r
epre
sent
ativ
e da
ta a
t th
e su
bnat
iona
l lev
el a
llow
ing
wit
hin
coun
try
com
paris
ons.
It m
ay a
lso
be o
f in
tere
st t
o po
licy
mak
ers
to c
ompa
re b
y so
cioe
cono
mic
sta
tus
(e.g
. wea
lth
quin
tile
or
inco
me
grou
ps).
Cata
stro
phic
hea
lth e
xpen
ditu
re
Pote
ntia
l sou
rce
of d
ata:
Hou
seho
ld s
urve
ys. M
ore
rece
ntly
the
W
orld
Ban
k ha
s st
arte
d to
pro
duce
re
gula
r est
imat
es t
hrou
gh it
s H
ealt
h Eq
uity
and
Fin
anci
al P
rote
ctio
n da
taba
se (H
EFPI
).ht
tp://
data
topi
cs.w
orld
bank
.org
/hea
lth-
Benc
hmar
king
gui
delin
esN
o es
tabl
ishe
d be
nchm
ark,
the
low
er
the
bett
er. S
ugge
st b
ench
mar
king
ag
ains
t pa
st p
erfo
rman
ce o
r re
leva
nt re
gion
al/in
com
e co
untr
y co
mpa
rato
rs/a
vera
ges.
Inci
denc
e of
cat
astr
ophi
c he
alth
exp
endi
ture
, % o
f pop
ulat
ion
(Lat
est
year
, 10
Perc
ent
thre
shol
d)
prot
ectio
n it
mus
t be
inte
rpre
ted
with
cau
tion
as it
can
not
dis
tran
spor
tatio
n, t
ime
cost
s, in
form
al p
aym
ents
, etc
.
Not
e: G
rey
= no
dat
aSo
urce
: WB
(201
9) H
EFPI
dat
abas
e
0-3
3-6
6-10
10-1
515
-45
The World Bank’s support to the Joint Learning Network for UHC is made possible with financial contributions from the following partners:
MEASURING HEALTH SYSTEM EFFICIENCY IN LOW- AND MIDDLE-INCOME COUNTRIES: A RESOURCE GUIDE 59
www.jointlearningnetwork.org