Improving Allocative Efficiency In Zimbabwe’s Health Sector 1 Improving Allocative Efficiency in Zimbabwe’s Health Sector RESULTS FROM THE HEALTH INTERVENTIONS PRIORITIZATION TOOL Xiaohui Hou Laurence Lannes Gerard Abou Jaoude Thomas David Wilkinson Lara Goscé Cliff Kerr Shepherd Shamu Hassan Haghparast-Bidgoli Chenjerai N. Sisimayi Jolene Skordis © 2021 The World Bank Group, 1818 H Street NW, Washington DC 20433. This report was prepared by World Bank staff with external contributions. The find- ings, interpretations, and conclusions expressed in this work do not necessarily re- flect the views of The World Bank, its Board of Executive Directors, or the govern- ments they represent. This report was originally published in English by the World Bank (Reforming the Basic Benefits Package in Armenia: Modeling Insights from the Health Interventions Prioritization Tool) in 2021. Where there are discrepancies, the English version will prevail. 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 any territory’s legal status or the endorsement or acceptance of such boundaries. The material in this report is subject to copyright. The World Bank encourages the dissemination of its knowledge. Hence, this work may be reproduced, in whole or in part, for noncommercial purposes if full attribution to this work is given. Any queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@worldbank.org. Improving Allocative Efficiency in Zimbabwe’s Health Sector RESULTS FROM THE HEALTH INTERVENTIONS PRIORITIZATION TOOL Xiaohui Hou Laurence Lannes Gerard Abou Jaoude Thomas David Wilkinson Lara Goscé Cliff Kerr Shepherd Shamu Hassan Haghparast-Bidgoli Chenjerai N. Sisimayi Jolene Skordis Table of Contents ACKNOWLEDGEMENTS 9 EXECUTIVE SUMMARY 11 ABBREVIATIONS 15 1. Introduction 17 2. Summary of Previous Health Sector Analysis 18 2.1 The National Health Accounts for 2015 18 2.2 The Resource Mapping Study by MOHCC 18 2.3 The Fiscal Space Analysis for Health in Zimbabwe 18 2.4 The Health Financing Strategy (HFS) 19 3. Methods and Data 19 Introduction of the Health Interventions Prioritization Tool 19 3.1  Mapping of Interventions to DCP3 Interventions 20 3.2  Data Inputs 20 3.3   ntervention Spending 21 3.3.1 I  ntervention Cost-Effectiveness Data 22 3.3.2 I  ata Validation 22 3.3.3 D 3.4 Optimization Analysis 22 4. Results 23 4.1 The Disease Burden 23 4.2 Allocation of Health Spending in 2016 24 Validating Estimated Allocations of National Health Spending in 2016, by Broad 4.2.1  Disease Categories, Compared with Previous Resource-mapping Studies 24 Health Spending Allocations Across Intervention Platforms in 2016 25 4.2.2  Interventions with the Highest Expenditure in 2016 26 4.2.3   stimated Impact of 2016 Allocations on National Health Spending 28 4.3 E Estimated Impact of 2016 Expenditure Allocations by Intervention Platform 28 4.3.1  Interventions with the Highest Impact in 2016 29 4.3.2   ptimised Allocation of National Health Spending in 2016 31 4.4 O Optimised Allocation Across Intervention Platforms 31 4.4.1  Optimised Interventions with the Highest Expenditure 33 4.4.2   stimated Impact of an Optimised Allocation of 2016 4.5 E National Health Spending 36 Potential Impacts of Allocation Optimization by Intervention Platform 36 4.5.1  Optimised Interventions with the Highest Impact 38 4.5.2  Limitations of the Study 39 5.  5.1 Challenges and Gaps in Data and Intervention Mapping 39 5.2 Limitations of HIPtool 40 Policy Implications 41 6.  Conclusion 43 7.  Table of Contents (cont.) APPENDIXES 43 Appendix A HIPtool: Technical Specifications for the Health Interventions Prioritization Tool 43 Background 45 Applicability and Methodology 45 Aims and Scope 45 Data Input Requirements 46 Impact Model 47 Effective Coverage and Maximal Effective Coverage 47 Equity and Financial Risk Protection Modules 47 Optimization Module 48 Limitations 49 Appendix B: Key Documents Reviewed 49 Appendix C: Additional Graphs on Top Ten Interventions by Existing and Optimised Spending, Within Intervention Platforms 51 World Bank Group 8 Improving Allocative Efficiency In Zimbabwe’s Health Sector 9 Acknowledgements This analytical and advisory activity is one of the deliverables under the Zimbabwe Health Financing and HRH Reforms Technical Assistance. This report was prepared by Xiaohui Hou (Senior Economist and Task Team Leader), Lara Gosce (University College London), Gerard Abou Jaoude (University College London), Laurence Lannes (Senior Economist), Shepherd Shamu (Consultant), Chenjerai N. Sisimayi (Consultant), and Thomas David Wilkinson (Consultant). Marelize Gorgens (Senior Special- ist), Christine Lao Pena (Senior Economist) and Jolene Skordis-Worrall (University College London) provided immense guidance on the Health Interventions Prioritization Tool or the country specific context. The team thanks Edit Velenyi (Senior Economist) and Benoit Mathivet (Senior Economist) for their peer review comments and Melody Molinoff for her editing. The team also thanks Magnus Lindelow (Practice Manager) and Mukami Kariuki (Country Manager) for their overall guidance and support. The team would like to sincerely acknowledge the Ministry of Health and Child Care for facilitating data collection and organizing two technical workshops in Harare. The team also thanks the Clinton Health Access Initiative and other development partners for their feedback. Financial support for this work was received from the Bill and Melinda Gates Foundation, Japan Policy and Human Resources Development Fund, and the United Kingdom Department for International Development. Executive Summary 10 Improving Allocative Efficiency In Zimbabwe’s Health Sector 11 Executive Summary The country of Zimbabwe has seen some option to achieve better health outcomes important improvements in key health out- in the current country context is to improve comes since 2009. However, despite prog- health sector spending efficiency. Accurately ress in some areas of the health sector, the identifying areas or interventions that should country did not meet its Millennium Devel- be targeted is essential to increasing spend- opment Goals (MDGs) and current progress ing efficiency and improving population falls short of the Sustainable Development health outcomes in Zimbabwe. Goals (SDGs) milestones. Globally, the availability of disease burden As is often the case, the poor and rural data, cost-effectiveness of interventions, populations in Zimbabwe bear a dispropor- Disease Control Priorities (DCP3), and tionate burden of disease and health risks. improvements to optimization algorithms The situation is compounded by national used in allocative efficiency tools, as well economic challenges and health sector as the analytic process itself, have enabled spending inefficiencies that have resulted in the development of the Health Interventions households bearing an increasing share of Prioritization Tool (HIPtool). HIPtool can help health sector financing, mainly through out- decision makers by informing stakeholder of-pocket expenditures. Households provide dialogues around which services should be approximately 25 percent of health sector prioritized to maximize a given set of objec- financing in Zimbabwe. Again, the poor and tives within a fixed budget. Therefore, the rural populations are hardest hit by this objective of this study was to apply HIPtool economic reality. to the Zimbabwe context to improve re- source allocation across health services. Economic growth projections for Zimbabwe indicate a growth decrease in 2019, coupled Zimbabwe was one of the few countries in with rising inflation. In addition, the health which HIPtool was piloted at the proof of sector’s 2019 budget falls significantly short concept stage. HIPtool enables the mathe- of the high impact scenario. How can Zim- matical prioritization of interventions based babwe protect and improve the health of its on existing data and a set of criteria. It population in this constrained environment? provides a technical foundation to further develop an essential health benefits package. In recent years, several analyses on health However, HIPtool, at this stage in develop- financing in Zimbabwe were conducted. ment, still has strong limitations, which are The results revealed that the most promising outlined along with results in this report. Executive Summary 12 HIPtool falling into the non-communicable diseases and injury or crosscutting care categories. Three main types of data are required to The analysis adjusted DCP3 average unit run an allocation efficiency analysis using costs to the Zimbabwe context, but it is far HIPtool: (1) burden of disease, measured from being perfect. This exercise highlights in disability-adjusted life years (DALYs); the need for more investment in collecting (2) intervention cost-effectiveness; and (3) and analyzing such data for broader use. intervention spending. Once populated with data, HIPtool can inform which of the 218 interventions in the DCP3 Essential Universal HIP Analysis Results Health Coverage (EUHC) package should An estimated US$980 million was spent on be prioritized for a given context and what health services in 2016. The largest share was an optimised allocation of spending across spent on HIV-related interventions and, in prioritized interventions might look like. The particular, on ART care in health centers for three datasets are combined in HIPtool to people living with HIV (PLHIV). The 2016 to- conduct an optimization analysis with the tal expenditure on health services is estimat- option of user-assigned weights for up to ed to have averted 1.6 million DALYs. Interest- three criteria: (1) health maximization, (2) ingly, the primary health center interventions equity, and (3) financial risk protection. alone accounted for approximately 67.3 Data collection methods for the Zimbabwe percent (1,102.6 thousand DALYs) of the total exercise included a desk review of relevant number of DALYs averted. Consistent with national documents to provide the country global literature, the most impactful inter- context; details on existing health services; ventions were delivered at lower platforms of baseline and target coverage of existing ser- care, such as Primary Health Centers (PHC). vices as well as intervention unit costs, bud- gets, and/or expenditures. Interviews were The 2016 optimised scenario, created by conducted with key officials from the pertinent HIPtool, continued to place important fo- ministries and with development partners. cus on HIV-related and tuberculosis (TB) interventions. Maternal and child health and The analysis was based on the 2016 fiscal NCD-related interventions received increases year, which was used both for intervention in the optimised scenario. An emphasis on data collection and inflation adjustment. integrated care emerged as an important step Expenditure and budget data were extract- to improving spending efficiency, such as ed from the 2016 Resource Mapping Report, integrated community case management and the National Health Accounts Report, and basic emergency neonatal and obstetric care the 2016 Ministry of Health and Child Care (BEmNOC) at PHCs. In addition, interventions (MoHCC) expenditure and appropriation for TB and pneumonia were also prioritized, account. Expenditure and budget data were particularly at lower platforms of care. then used to compare and validate aggregat- ed intervention expenditure estimated for the HIPtool analysis. Policy Implications for Zimbabwe The analysis indicated that a shift in spend- The availability and quality of some underly- ing from hospitals to community and PHC ing unit costs and utilization data remain platforms of care could significantly increase a significant challenge in the low- and mid- the amount of DALYs averted by the NHS dle-income country context. This data chal- package in Zimbabwe. HIPtool does not lenge is particularly severe for interventions Improving Allocative Efficiency In Zimbabwe’s Health Sector 13 analyze human resources for health however; HIPtool is still in its nascent stage and has the roles of community health workers and several limitations, which are outlined later in PHC staff are widely accepted as central to this report. However, while HIPtool is being a more integrated approach at lower plat- further refined and improved, the current forms of care, which will facilitate optimised HIPtool outputs can inform high-level discus- resource allocation. sions on allocative efficiency and provide an entry point into more specific and compre- HIV/AIDS, TB, and malaria remain the highest hensive analyses in collaboration with other portion of the disease burden for Zimbabwe. existing tools. HIPtool’s power and value lies An integrated PHC and community approach in its ability to nimbly adjust recommenda- is essential to making additional progress tions as updated input data becomes avail- in reducing this part of the disease burden. able. The key is to build the MoHCC staff‘s Similarly, maternal and child health interven- capacity to understand and use HIPtool. The tions, which have been identified as areas of gloomy fiscal situation in Zimbabwe necessi- focus by the government, and non-communi- tates improved efficiency in the allocation of cable diseases interventions, could be more health funds. HIPtool, combined with other broadly delivered at the PHC and community technical analysis and with consideration for levels to improve cost-effectiveness. political and implementation realities, will provide the necessary evidence to help deci- The analysis provides evidence on high- sion makers improve health fund allocations. level shifts in resource allocation that could save money and improve health outcomes. Zimbabwe recently identified strengthening budget formulation processes as a key and urgent reform. This is a first step toward actively utilizing HIPtool to appraise the cost and impact of the NHS package and improve health outcomes in Zimbabwe. Abbreviations ART Assessment of antiretroviral treatment BEmNoC Basic Emergency Neonatal and Obstetric Care CHAI Clinton Health Access Initiative DHS Demographic health surveys EUHC Essential Universal Health Coverage GBD Global burden of disease GDP Gross Domestic Product HFS Health financing strategy HIP Health investment prioritization IHME Institute for Health Metrics and Evaluation ICCM Integrated Community Case Management MCH Maternal and child health MNCH Maternal, neonatal and child health MICS Multiple Indicator Cluster Surveys MDG Millennium Development Goal MoFED Ministry of Finance and Economic Development MoHCC Ministry of Health and Child Care M&E Monitoring and evaluation MDR-TB Multi-drug resistant tuberculosis NCD Non-communicable disease NHA National Health Accounts NHS National Health Service PFM Public financial management SADC South African Development Community SRH Sexual reproductive health STI Sexually transmitted infection SDG Sustainable Development Goal TB Tuberculosis UNAIDS Joint United Nations Programme on HIV/AIDS UNICEF United Nations Children’s Fund WHO World Health Organization World Bank Group 16 Photo: Jason Zhao Improving Allocative Efficiency In Zimbabwe’s Health Sector 17 1. Introduction further heightening the risk of limited access to quality health services in the immediate Despite improvements in key health out- to short term. Economic growth was pro- comes since 2009, Zimbabwe’s health sector jected to decrease from four percent in 2018, did not meet its Millennium Development to 3.1 percent in 2019, but increase to above Goals (MDGs), and current progress falls seven percent in 2020; whilst annual aver- short of the Sustainable Development Goals age inflation was projected to increase from (SDGs) milestones. Zimbabwe’s human cap- 8.3 percent in 2018, to 22.4 percent in 2019. ital index1 is 0.44, which is on par with the Inflation was expected to increase signifi- Southern African Development Community cantly especially during the first half of 2019, (SADC) average.2 In 2016, life expectancy with medical inflation being a major driver reached 61 years of age. Maternal and infant in recent months. Despite improvements mortality has decreased, as has HIV and in nominal allocation from previous years, tuberculosis (TB) prevalence. However, Zim- the 2019 health sector budget allocation of babwe’s disease burden remains high and its US$694 million fell significantly short of the maternal and child health (MCH) outcomes US$1.39 billion required for the NHS High are among the worst in the SADC region. Impact Scenario for 2019.4 Sixty-five percent of annual deaths are at- In recent years, several analyses on health tributed to communicable, maternal, perina- financing in Zimbabwe were conducted. The tal, and nutritional illness, although the share results strengthened the information base of deaths attributed to non-communicable and revealed that the most promising option diseases has been increasing.3 The poor and to achieve better results in the current coun- rural populations shoulder a disproportionate try context is to improve spending efficiency. burden of disease and health risks. The analysis presented in this report was Cycles of fragility and macroeconomic chal- triggered by the conclusions and recommen- lenges coupled with health sector spending dations of these analyses and aim to provide inefficiencies have increasingly shifted the tangible solutions to increase spending effi- burden of health care financing to house- ciency. The main conclusions of these reports holds, which has affected service utilization. are listed in section 2. Among all the possible Zimbabwe’s total health spending per capita options to improve the health sector’s fiscal compares favorably with the sub-Saharan space in Zimbabwe, the most urgent and average. However, due to limited fiscal plausible option is to increase efficiency gains space/public financing in health, absence of by maximizing the current level of resources a prepayment mechanism offering financial allocated to the health sector. protection to the population, and ineffi- Globally, the availability of disease burden ciencies in the sector, households provide a data, information on the cost-effectiveness significant share of health sector financing of health interventions, the outputs of the (25 percent), mainly through out-of- Disease Control Priorities (DCP3 ) initiative, pocket expenditures. and progress in allocative efficiency research The continued deterioration of the economic have enabled the development of the Health landscape has compounded existing supply- Investment Prioritization Tool (HIPtool) and demand-side constraints in the sector,  he index measures the amount of human capital that a child born today can expect to attain by age 18, given the 1 T risks of poor health and poor education that prevail in the country where s/he lives. 2  SADC is a regional organization comprised of 14 member countries. 3  World Health Organization. 4  Zimbabwe National Health Strategy 2016-2020 Costing Report. World Bank Group 18 to help inform the prioritization of health 2.2 The Resource Mapping resources. Using a mathematical algorithm, Study by MOHCC an optimised allocation of spending across health interventions is generated to maxi- The resource mapping showed that the Gov- mize one or more of the following objectives: ernment of Zimbabwe is the major funder of (a) health maximization (i.e., maximization of the health sector and that its contribution DALYs averted), (b) equity, and (3) financial is heavily skewed towards health worker risk protection. salaries. There is a significant cost-sharing imbalance between the government and The objective of the study is to apply HIPtool partners as government funding goes mostly in Zimbabwe to improve resource allocation toward health systems costs while partner across health services. funding goes toward disease-specific ac- tivities (e.g., commodities). The reliance on external funding for key cost categories rep- 2. Summary of Previous resents a challenge in terms of sustainability Health Sector Analysis and predictability of health system funding. Funding from external partners is critical for This section summarizes the main findings drugs, and some important items (research from the National Health Accounts, the annu- and M&E or infrastructure and equipment-re- al resource mapping, the fiscal space analysis lated expenses) are only paid for by external for health, and the health financing strategy.5 assistance. HIV, vaccines, malaria, reproduc- tive and maternal, neonatal and child health (MNCH), and TB programs are highly 2.1 The National Health donor dependent. Accounts for 2015 The NHA revealed important inefficiencies in 2.3 The Fiscal Space Analysis for current public and private health spending Health in Zimbabwe and recommended measures such as allocat- ing more resources for preventive care since The 2017 fiscal space analysis6 concluded current health financing allocations favor that the current financial crisis and macro- curative services. The increasing burden from economic situation in Zimbabwe did not con- non-communicable diseases (NCDs) requires stitute an enabling environment for gener- prioritization of preventive care to avert the ating fiscal space for health. In this context, high costs of NCD treatment; developing using the current level of resources allocated strategic purchasing mechanisms to en- to the health sector in the most effective and hance the efficiency of available funding; and efficient way is the most urgent and plausible strengthening the integration of vertical and option for increasing fiscal space for health. disease-specific programs, which are concen- Looking at the disease profile and health trated on HIV/AIDS, malaria, and tuberculosis, seeking behaviours of the poorest, the fiscal to leverage disease-specific donor funding. space analysis recommended reallocating more resources to the lowest levels of care, where most of the vulnerable go and where most cases can be treated at a lower cost. Reassigning some resources from curative  ttp://www.mohcc.gov.zw/index.php?option=com_phocadownload&view=category&download=53: 5 h zimbabwe-health-financing-strategy&id=6:acts-policies&Itemid=552  he fiscal space analysis looked at opportunities for making additional resources available for government spend- 6 T ing on health through: (i) establishing conducive macroeconomic and fiscal conditions; (ii) prioritizing health within the government budget; (iii) allocating health sector-specific financing from other sources; (iv) negotiating higher development assistance for health; and (v) improving efficiency of outlays for health. Improving Allocative Efficiency In Zimbabwe’s Health Sector 19 to preventive services, focused on NCD pre- countries to use locally-generated cost, vention, could also alleviate the significant effectiveness, and coverage data to deter- burden of NCD curative treatments on health mine a mathematically optimised resource systems and health financing. The analysis allocation that maximizes one or more of the also suggested improvements in budget following objectives: (1) health maximization, processes, from planning to execution and (2) equity, and (3) financial risk protection. implementing PFM reforms to better turn In this analysis, only health maximization, allocated funds into inputs. It also acknowl- measured by averted Disability Adjusted Life edged that the large wage bill represents Years (DALYs) was used due to lack to finan- a major constraint for the country and the cial risk protection data in the HIPtool health sector, and substantial efficiency gains at the time. could be achieved within the sector without implementing public sector wage reform. The mainstay of HIPtool is the findings from Disease Control Priorities (DCP3), which looks at 21 essential packages of interven- 2.4 The Health Financing tions at five different platforms of care: (1) Strategy (HFS) population, (2) community, (3) health center, (4) first level hospitals, and (5) referral or The HFS recommends a significant focus on central hospitals. Combined, the packages efficiency gains, which triggered the analy- form a single Essential Universal Health sis presented in this paper. The HFS places Coverage (EUHC) health benefits package emphasis on implementing reforms that pri- developed by DCP3, which contains the 218 oritize low-cost, high-impact interventions; interventions included in HIPtool. All 218 in- on improving allocative and implementation terventions are not relevant to each country, efficiencies; and on improving the integration and even if they were, low- and middle- of services at all levels of the health system. income countries would not have sufficient The HFS short- to medium-term interven- resources to fully implement each. Therefore, tions focus on efficiency; notably, the im- tools such as HIPtool are important to iden- portance of increasing efficiency gains from tify which of the 218 interventions should be existing resources and improving efficiency prioritized with sufficient levels of coverage of external assistance on health. in different contexts. HIPtool is able to determine the services 3. Methods and Data that should be prioritized to maximize a given set of objectives within a fixed budget. HIPtool has the potential to consider health 3.1 Introduction of the Health outcomes, financial protection, and equity as Interventions Prioritization Tool a comprehensive objective function mirroring The Health Interventions Prioritization Tool the UHC goals. Although HIPtool is still in its (HIPtool) is a cloud-based, open-access, nascent stages, this analytical work comple- user-friendly, high-impact resource to assist ments prior analyses and tools that have been with the design of national health bene- developed and demonstrates the potential fits packages. It combines context-specific ability to inform national discussions on im- data on disease burden and intervention proving the efficiency of resource allocation effectiveness to help stakeholders identify in the health sector. funding priorities and targets. HIPtool allows Introduction 20 “Zimbabwe’s communicable, maternal, neonatal, and nutritional disease burden remains high, yet, the country faces a double burden of communicable and non-communicable diseases (NCDs).” 3.2 Mapping of Interventions 3.3 Data Inputs to DCP3 Interventions Three main types of data are required to The NHS package was first mapped to EUHC run an allocative efficiency analysis using interventions to determine the extent to HIPtool: (1) burden of disease, measured in which it is aligned with global evidence pub- disability-adjusted life years (DALYs); (2) lished by DCP3. Mapping the existing NHS to intervention cost-effectiveness; and (3) inter- DCP3 interventions also enabled the use of vention spending. Country-specific disease relevant secondary data collated by DCP3, burden data is available from the Global pre-loaded in HIPtool, and in turn a timely Burden of Disease Study conducted by the analysis to inform the ongoing discussions on Institute for Health Metrics and Evaluation revisions to the NHS package. The mapping (IHME) and is pre-loaded in HIPtool. Data process included one-on-one intervention on the cost-effectiveness of interventions, mapping consultations with experts (health available from the EUHC data published by workforce and laypersons) and intervention DCP3, is also included in HIPtool. However, mapping workshops with clinicians, program given the lack of secondary data on interven- planners, and policy makers. Published NHS tion spending, this often has to be calculat- protocols were the primary source of data ed. Once populated with data, HIPtool can for mapping NHS services to EUHC interven- identify which of the 218 EUHC interventions tions. However, in the absence of a precise should be prioritized for a given context and and common definition the primary cause present an optimised allocation snapshot of disease was used to indicate the scope across prioritized interventions. The three of an intervention. datasets are combined in HIPtool to inves- tigate the latter by conducting an optimi- After mapping NHS services to EUHC inter- zation analysis with the option to assign ventions, necessary intervention datasets weights for up to three criteria: (1) health were collated by determining population in maximization, (2) equity, and (3) financial need, baseline and target coverage, and unit risk protection. cost per person served. Improving Allocative Efficiency In Zimbabwe’s Health Sector 21 The HIPtool framework, how data are com- request). Local unit cost data was mostly bined, and the optimization process are available for maternal and child health inter- described in detail in Appendix A. The ventions, and the resource mapping analyses Zimbabwe study assigned full weight provided spending data for certain HIV- and to health maximization. malaria-related interventions. In cases where unit costs were not available from local Data collection methods for the exercise sources, EUHC intervention unit costs pub- included a desk review of relevant national lished by DCP3 were adjusted to the Zimba- documents to provide the country context. bwe context. The review examined details on existing health services; baseline and target coverage Costing and utilization data for non-com- of existing services; and intervention unit municable diseases, injury and cross-cutting costs, budgets and/or expenditures (see Ap- care is very poor. For most interventions pendix B for the list of documents reviewed). in these categories, we had to rely on the Interviews were conducted with key clinical adjustment of EUHC unit costs, which are officials from the Ministry of Health and Child derived from averages across low-income Care (MoHCC) to gather information on the countries in 2012 US dollars.8 To generate types of interventions that are being imple- average EUHC intervention unit costs for mented in Zimbabwe. Interviews were also low-income settings, DCP3 assumed that 70 conducted with the Clinton Health Access percent of each EUHC unit cost is allocated Initiative (CHAI), the World Health Organiza- to health worker salaries.9 To further refine tion (WHO), and the Ministry of Finance and the estimate for a given EUHC intervention Economic Development (MoFED) to discuss unit cost, the average health worker salary resource-mapping data, data use and One- was adjusted to the country context and ap- Health country-specific data on unit costs plied to the assumed 70 percent of unit cost. and intervention coverage, and evidence on The ratio used to adjust the health worker fiscal space, respectively. salary components of EUHC unit costs was based on the health worker salary dataset 3.3.1 Intervention Spending used by DCP3 to generate the EUHC unit costs, and on the GDP multipliers generated Intervention spending was estimated by by the WHO-CHOICE econometric analysis combining unit cost and annual utilization to inform national averages for health worker estimates for each of the interventions.7 salaries.10 Once adjusted, the EUHC interven- Unit costs were sourced from the National tion unit costs were inflated to 2016 based Health Strategy 2015-2020 and expenditure on the World Bank consumer price index data from the resource mapping study of for Zimbabwe.11 interventions where local data was available (the full list of data sources is available upon  i=UC x Un. Where intervention spending is Si, unit cost is UC, and the annual number of people utilising an inter- 7 S vention is Un. In turn, the number of people utilising a service Un, is based on: Un=PN x Ap. Where PN denotes the number of people in need of an intervention, and Ap is the estimated access to an intervention as a percentage. 8  DCP3 unit costs published for the EUHC package of interventions reflect long-run average costs  or more details, see the DCP3 working paper: Watkins, D.A., Qi, J., Horton, S.E. 2017. Working Paper # 20: Costing 9 F Universal Health Coverage: The DCP3 Model. Disease Control Priorities in Developing Countries, 3rd Edition.  erje J, Bertram MY, Brindley C, Lauer JA. Global health worker salary estimates: an econometric analysis of 10 S global earnings data. Cost Effectiveness and Resource Allocation. 2018 Dec;16(1):10.  UHC unit costs were inflated using the average of World Bank data on “Inflation, consumer prices (annual %) - 11 E low-income group” for 2012-6. The average inflation rate between 2012-6 used was 5.35%. World Bank Group 22 To estimate utilization for a given interven- through a rigorous review of the evidence, tion, secondary data sources on population it was assumed that EUHC interventions fall need and the percentage of population with within the upper-bound of estimated country- access to an intervention were combined. specific, cost-effectiveness thresholds.13 Last, Baseline data from the 2015 NHS report, as before undertaking the optimization analysis, well as data from the WHO Global Health a quality reduction factor of 30 percent was Observatory, UNICEF, UNAIDS, Demograph- applied across all intervention cost-effective- ic Health Surveys (DHS), Multiple Indicator ness estimates. This reduction adjusted the Cluster Surveys (MICS), and other pub- trial-based data to reflect the loss of effec- lished peer-reviewed and grey literature tiveness that occurs outside of controlled were compiled to inform levels of access to settings and when implementing at scale.14 each intervention. In the absence of data, DCP3 assumptions on baseline coverage 3.3.3. Data Validation were used.12 To estimate population in need, disease-specific modeled prevalence for HIV The analysis was based on the 2016 fiscal was sourced from UNAIDS, and the GBD year, which was used both for intervention prevalence data was used for all other inter- data collection and inflation adjustment. ventions directly linked to a cause of disease. Expenditure and budget data were extracted Populations at risk, or eligible, for interven- from the 2016 Resource Mapping Report, the tions that are not directly linked to a GBD National Health Accounts Report, and the cause were estimated based on data from 2016 Ministry of Health and Child Care expen- the DHS, UNICEF, UN Populations Division, diture and appropriation account. Expendi- and from peer-reviewed studies conducted ture and budget data were then used to com- either in Zimbabwe or in Southern Africa. pare and validate aggregated intervention expenditure estimated for the HIPtool analysis (see section 4.2.1 below for key results from 3.3.2 Intervention Cost- the expenditure validation process). Effectiveness Data Data published by DCP3 on the cost-effec- 3.4 Optimization Analysis tiveness of EUHC interventions, linked to NHS services through the mapping process, The optimization module of HIPtool is de- were used in the HIPtool analysis. Howev- scribed in detail in Appendix A. Following er, due to the absence of some cost-effec- the steps outlined in sections 3.1 and 3.2, the tiveness estimates in the published EUHC three datasets necessary to run HIPtool were package, additional data was sourced from combined to calculate the maximum effec- other DCP3 volumes and annexes. For certain tive coverage for a given intervention. The EUHC interventions, data on cost-effective- maximum effective coverage for an interven- ness was not available in any of the DCP3 vol- tion informs the upper spending constraint umes or supplementary materials. For these for the optimization process, based on the interventions, given that DCP3 developed the estimated maximum impact an intervention EUHC package based on cost-effectiveness could have on the causes of the disease burden that it is linked to.15  ssumed baseline coverage of 40 percent of estimated prevalence for Tier I causes from the GBD study (commu- 12 A nicable, maternal, perinatal, and nutritional disorders). For Tier II and III causes (noncommunicable diseases and injuries), assumed baseline coverage of 10 percent. For cross-cutting packages (e.g. palliative care or rehabilita- tion), assumed baseline coverage of five percent.  oods B, Revill P, Sculpher M, Claxton K. Country-level cost-effectiveness thresholds: initial estimates and the 13 W need for further research. Value in Health. 2016 Dec 1;19(8):929-35.  rost A, Colbourn T, Seward N, Azad K, Coomarasamy A, Copas A, Houweling TA, Fottrell E, Kuddus A, Lewy- 14 P cka S, MacArthur C. (2013) Women’s groups practising participatory learning and action to improve mater- nal and newborn health in low-resource settings: a systematic review and meta-analysis. The Lancet. May 18;381(9879):1736-46.  ee Appendix A for details on how maximum effective coverage is calculated. 15 S Improving Allocative Efficiency In Zimbabwe’s Health Sector 23 It was possible to estimate the maximum ef- maternal, neonatal, and nutritional diseases fective coverage for 149 of the 168 interven- account for 57.5 percent of the total number tions that were costed. The 19 interventions of DALYs, while non-communicable diseases for which a maximum effective coverage (NCDs) and injuries account for 33.4 percent could not be estimated were not optimised; and 9.1 percent respectively. The burden of associated intervention spending was fixed disease doubled from 5.3 million DALYs in and therefore remained unchanged after the 1990, to 11.8 million DALYs in 2008 (figure optimization analysis. The ‘non-optimised 1), followed by a significant reduction to the interventions’ do not have a direct link to a seven million DALYs estimated for 2017. The GBD cause of disease, except for the ‘Fixed rapid increase in the disease burden from Injury and Rehabilitation Package,’ which was 1990–2008, and subsequent reduction, is excluded due to insufficient data to estimate driven primarily by HIV/AIDS and improve- the maximum effective coverage of related ments in the national HIV response. Also interventions. An optimization analysis was driving the increase in the disease burden then conducted with the objective to maxi- between 1990 and 2017 is a steady rise in mize health impact. the number of NCD-related DALYs, from 1.3 million to 2.3 million (figure 1). 4. Results Disaggregating the disease burden in Zim- babwe by age and specific causes of disease highlights that a quarter (23.9 percent) of all 4.1 The Disease Burden DALYs are experienced by the first year of life (figure 2), and an additional 572,000 A total of seven million disability-adjusted DALYs (eight percent) occur in the one- to life years (DALYs) are estimated to have oc- four-year age range. Maternal and neonatal curred in Zimbabwe in 2017. Communicable, FIGURE 1 Estimated Number of DALYs by Broad Disease Category, Zimbabwe, 1990-2017 Zimbabwe, Both Sexes, 11M All Ages, All Causes 10M Zimbabwe, Both Sexes, All Ages, Communicable, 9M Maternal, Neonatal, and Nutritional Diseases 8M Zimbabwe, Both Sexes, 7M All Ages, Injuries 6M Zimbabwe, Both Sexes, All Ages, Non- 5M communicable Diseases 4M 3M 2M 1M 0 1990 1995 2000 2005 2010 2015 Source: IHME Global Burden of Disease Study, April 16, 2019 World Bank Group 24 health conditions account for 80 percent of 4.2 Allocation of Health DALYs in the zero to six-day age range. Spending in 2016 Respiratory infections and TB, enteric in- fections, and nutritional deficiencies are the leading causes of DALYs in the seven-day to 4.2.1 Validating Estimated Allocations four-year age range. Across all age groups, of National Health Spending in 2016, by HIV/AIDS (13.9 percent), maternal and neona- Broad Disease Categories, Compared tal disorders (12.3 percent), lower-respiratory with Previous Resource-mapping Studies infections (9.9 percent), and TB (7.3 percent) An estimated US$980 million was spent on are the leading causes of DALYs in Zimbabwe. health services in 2016. This expenditure in-  stimated Number of DALYs by Age and Condition, Zimbabwe, 2017 FIGURE 2 E DALYs HIV/AIDS & STIs Respiratory Infections & TB Enteric Infections NTDs & Malaria Other Infectious Maternal & Neonatal Nutritional Deficiencies Neoplasms Cardiovascular Diseases 600k Chronic Respiratory Digestive Diseases Neurological Disorders Mental Disorders Substance Use Diabetes & CKD Skin Diseases Sense Organ Diseases 400k Musculoskeletal Disorders Other Non-Communicable Transport Injuries Unintentional Injury Self-harm & Violence 200k 0 28 27 ys s s 10 Ye s 15 Ye s 20 19 Y rs 25 4 Y rs 30 9 Y rs 35 4 Y rs 40 9 Y ars 45 4 rs 50 9 ars 55 4 Y rs 60 9 rs 65 4 Y rs 70 9 Y rs 75 4 Y rs 80 9 ars 85 4 Y rs 90 9 rs 4 rs + ars s 64 ay 1- Day 9 r r ar 5- ea -14 a a -2 ea -2 ea -3 ea -4 ea -5 Yea -5 ea -6 Yea -6 ea -7 ea -8 Yea -8 ea -9 Yea 7- Da -3 e -4 Ye -7 e 95 Ye Ye -3 D Y 6 4 0- - Source: IHME Global Burden of Disease Study, April 16, 2019. Improving Allocative Efficiency In Zimbabwe’s Health Sector 25  stimated Total Expenditure Across NHS Services by Disease Program, FIGURE 3 E in USD and as a % of Total Spending, 2016 HIV/AIDS 371,240,750 (37.8%) Health Systems Strengthening 97,572,109 (9.9%) Maternal & Newborn Health 95,156,993 (9.7%) NCDs Behaviour Charge Communication/Awareness 72,685,967 (7.4%) Non Optimised Injury & Rehabilitation Spending 69,374,389 (7.1%) CVD and Related Illness 64,505,636 (6.6%) Tuberculosis 51,752,290 (5.3%) Febrile Illness & NTDs 38,592,456 (3.9%) Environmental Health & Epidemic Prone Diseases 34,466,978 (3.5%) Surgery 26,091,430 (2.7%) Child & Adolescent Health 16,508,123 (1.7%) Other Cross Cutting Activities 16,310,211 (1.7%) Reproductive Health 8,800,523 (0.9%) Injury & Rehabilitation 5,693,124 (0.6%) Mental Health 4,757,236 (0.5%) Cancer 3,276,906 (0.3%) Palliative Care 1,953,890 (0.2%) Congenital Disorders 1,482,308 (0.2%) Musculoskeletal 837,997 (0.1%) Environmental Health 336,695 (0.0%) 50 10 15 20 25 40 30 35 0, 0, 0, ,0 0, 0, 0, 0, 0 0 0 0 0 0 0 0 0 0 0 0 0 0, 0 0 0, 0, 0, 0, 0, 0, 0, 0 0 0 Source: HIPtool analysis. 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 cludes both public health spending and most spending was allocated to MNCH-related external financing.16 The allocations estimat- interventions, 5.3 percent (US$51.8 million) ed by disease are broadly in line with the on TB-related interventions, and an addition- results from the resource mapping studies al 3.25 percent (US$38.6 million) on malaria (figures 3 and 4). Approximately US$371.24 and neglected tropical diseases (figure 3). million was allocated to HIV-related inter- ventions, equivalent to 37.8 percent of total 4.2.2 Health Spending Allocations spending (figure 3), and US$262.5 million Across Intervention Platforms in 2016 (26.7 percent) was spent on providing ART care in health centers for people living with Approximately 41 percent of national health HIV (PLHIV) as shown in figure 5. Overall, expenditure was spent on interventions deliv- around 11.4 percent (US$111.8 million) of total ered mainly through primary health centers.  he estimation of national health expenditure includes both domestic public financing and external financing. For 16 T external financing estimation, the method is described in the resource mapping study as follows “The MOHCC DPP provided the Resource Mapping team with a list of key health stakeholders to include in data collection. The Resource Mapping team distributed the data entry tool to these stakeholders and requested that they com- plete the tool and return it within 8 weeks. The result included a response rate of 90%, with submission from the MoHCC, 70 local authorities, 4 parastatals (National AIDS Council, Zimbabwe National Family Planning Council, Medicines Control Authority of Zimbabwe and National Pharmaceutical Authority of Zimbabwe) and 18 donors and 35 NGOs.” It was estimated about 5 percent of financing were not included due to nonresponse from exter- nal stakeholders. The national health spending does not include private health spending. World Bank Group 26  otal Expenditure Across NHS Services and Mapped to Disease Program Areas, 2016-2017 FIGURE 4 T Allocation to Disease Programs * = 2017 % Allocation $19 *(1%) Other Diseases (mental health, nutrition, NTDs) Other Cross-Cutting Activities $37 *(2%) Vaccines $7 *(3%) Eye, Ear, and Skin Conditions $31 *(3%) Environmental Health & Epidemic Prone Diseases $32 *(4%) Tuberculosis, Excluding HIV/TB $34 *(4%) Malaria $33 *(4%) Non-Communicable Diseases $73 *(7%) Health System Strengthening $61 *(8%) Respiratory Infections $96 *(10%) Reproductive, Maternal, and Newborn Health $104 *(11%) HIV Including STIs $404 *(43%) 2017 2016 500 Millions Source: CHAI resource mapping study 2016–2017. Using the platform allocation for DCP3 in- lation-wide health interventions. Overall, 31 terventions, the HIPtool analysis was able to percent (US$308.3 million) of spending was predict expenditure by platform. As shown not optimised due to the reasons outlined in in figure 5, approximately US$405.3 million section 3.3. was allocated to interventions delivered through primary health centers in 2016. 4.2.3 Interventions with the Around 14 percent (US$136.8 million) of total Highest Expenditure in 2016 health spending was allocated to interven- tions delivered in first-level hospitals, nine The 15 interventions with the highest ex- percent (US$92.7 million) to interventions penditure (figure 6) accounted for around delivered at the community level, three 80 percent (US$542.2 million) of total percent (US$27.8 million) to interventions health spending in 2016. After removing the provided at referral and specialized hospitals, non-optimised interventions the amount is and one percent (US$9.1 million) to popu- equivalent to 55 percent of health spend- Improving Allocative Efficiency In Zimbabwe’s Health Sector 27 ing. Spending on six of the 15 interventions, care in health centers for people living with delivered at primary health centers, amount- HIV (PLHIV) – equivalent to 65 percent of ed to US$366.1 million. Another six (US$101.1 total spending on interventions provided in million), provided at first-level hospitals, ac- health centers. In 2016, four of the 15 interven- counted for US$101.1 million. Of the remain- tions with the largest expenditures addressed ing three interventions, two (US$16.3 million) maternal and child health conditions. Overall, were referral and specialty hospital-based US$83.5 million was spent on the latter; of interventions, and one (US$58.7 million) which 45 percent was spent in primary health was a community-based intervention. centers and the rest in first-level (44 per- cent) and referral-level hospitals (11 percent). Five of the 15 interventions with the highest Aside from two TB-related interventions, the amounts of spending (figure 6) were HIV-re- remaining four interventions address NCDs at lated, and US$262.5 million (27 percent of first-level and referral hospitals.17 total spending) was spent on providing ART  imbabwe Health Expenditure Across Intervention Platforms, 2016 FIGURE 5 Z Spending in Millions (2016 $, USD) 92.7 Community Health Centre 308.3 First-level Hospital Referral and Specialty Hospital Population-based Health Interventions Non-optimised Expenditure 9.1 405.3 27.8 136.8 Source: HIPtool analysis. nsufficient local data was available to estimate and validate spending on NCD-related interventions. There is 17 I therefore uncertainty around the amounts stated for NCD interventions in figure 6. World Bank Group 28  he 15 Highest Expenditure Interventions, Zimbabwe, US$ millions, 2016 FIGURE 6 T Spending in Millions (2016 USD, $) (Health Center) ART Care for PLHIV 262.5 (Community) Community-based HIV Services 58.7 (Health Center) BEmNOC 37.8 (First-level Hospital) Labor and Delivery In High Risk Women 21.7 (First-level Hospital) Acute Asthma and COPD Management 20.9 (Health Center) PMTCT of HIV (Option B+) and Syphilis 20.1 (First-level Hospital) Referral for DST and MDR-TB Treatment 19.6 (Health Center) Isoniazid Preventative Therapy for TB 18.2 (First-level Hospital) Acute Critical Limb Ischemia Management 17.6 (First-level Hospital) Surgical Termination of Pregnancy 15.1 (Health Center) Medical Male Circumcision 14.7 (Health Center) Testing and Counseling for HIV, STIs, Hepatitis 13.0 (Referral and Specialty Hospital) Full Supportive Care for Preterm Newborns 8.9 (Referral and Specialty Hospital) Retinopathy Screening & Treatment 7.4 (First-level Hospital) Osteomyelitis Management 6.2 Source: HIPtool analysis. 4.3 Estimated Impact of 2016 the total number of DALYs averted (figure Allocations on National 7). A similar number of DALYs, 233.4 and Health Spending 235.7 thousand, were averted through in- terventions delivered at the community and first-level hospital platforms respectively. 4.3.1 Estimated Impact of 2016 Expendi- Combined, community and first-level hospital ture Allocations by Intervention Platform interventions accounted almost 30 percent The existing amount and allocation of na- of all DALYs averted. Referral and specialty tional health spending is estimated to have hospitals and population-wide interventions averted 1.6 million DALYs in 2016. Interven- yielded three percent (49.2 thousand DA- tions provided at the primary health center LYs) and one percent (16.3 thousand DALYs), platform alone accounted for approximately respectively, of the total 1.6 million DALYs 67.3 percent (1,102.6 thousand DALYs) of averted in 2016. Improving Allocative Efficiency In Zimbabwe’s Health Sector 29 “Approximately 41 percent of national health expenditure was spent on interventions delivered mainly through primary health centers.” 4.3.2 Interventions with the in 2016. Seven of the 15 interventions with Highest Impact in 2016 the greatest impact were delivered through primary health centers, and combined they Interventions with the greatest impacts averted around 957 thousand DALYs – equiv- were delivered at the community and health alent to 60 percent of all DALYs averted in center levels. The 15 interventions with the 2016. The remaining eight interventions with greatest impact on the disease burden are the greatest impact were delivered by com- shown in figure 8 and accounted for 80 munity (four interventions) and first-level percent the total number of DALYs averted (four interventions) hospitals, which accounted  stimated Impact of 2016 Health Spending Allocations Across Intervention Platforms FIGURE 7 E Impact of 2016 NHS Spending Allocations (DALYs, thousands) 233.4 49.2 Community 235.7 16.3 Health Centre First-level Hospital Referral and Specialty Hospital Population-based Health Interventions 1,102.6 Source: HIPtool analysis. World Bank Group 30  016 NHS 15 Most Impactful Interventions, Zimbabwe FIGURE 8 2 Intervention Impact, in DALYs Averted (Thousands) (Health Center) ART Care for PLHIV 585.0 (Health Center) BEmNOC 179.8 (Health Center) Diagnosis of TB and First-line Treatment 61.5 (Health Center) PMTCT of HIV (Option B+) and Syphilis 55.9 (Health Center) Medical Male Circumcision 53.2 (First-level Hospital) Referral for DST and MDR-TB Treatment 51.9 (First-level Hospital) Relief of Urinary Obstruction 43.4 (Health Center) Testing and Counseling for HIV, STIs, Hepatitis 38.0 (Health Center) Integrated Management of Childhood Illness 36.5 (Community) Pneumococcus Vaccination 35.2 (Community) Indoor Residual Spraying 33.2 (Community) Diagnosis of and Treatment of Malaria 29.1 (First-level Hospital) Care for Fetal Growth Restriction 27.0 (First-level Hospital) Severe Acute Malnutrition Management 25.3 (Community) Acute Severe Malnutrition Management 20.6 Source: HIPtool analysis. for 7.4 percent (118 thousand DALYs) and 9.2 of the total number of DALYs averted. Six percent (148 thousand DALYs) of the total of the 15 highest-impact interventions were number of DALYs averted. None of the 15 maternal, child health and nutrition interven- interventions with the greatest impact were tions, which combined account for a fifth of delivered at referral and specialty hospitals or all DALYs averted (324.5 thousand DALYs). were a part of population-wide interventions. Out of the six maternal, child, and nutrition interventions, Basic Emergency Neonatal and The most impactful interventions are HIV, Obstetric Care (BEmNoC) in health centers maternal, child health and nutrition related. alone accounts for 11 percent of all DALYs Four of the 15 interventions with the great- averted. The remaining five highest-impact est impact were HIV-related; ART for PLHIV interventions were two TB interventions, in health centers alone was responsible for two malaria interventions, and the relief of approximately 585 thousand (36 percent) urinary obstruction. Improving Allocative Efficiency In Zimbabwe’s Health Sector 31 4.4 Optimised Allocation of spending allocation, but decreasing invest- National Health Spending in 2016 ment in both first-level and referral hospital spending was recommended so that funds can be reallocated to lower platforms of care 4.4.1 Optimised Allocation Across (figure 10). First-level and referral hospitals Intervention Platforms would retain 74.9 percent (-US$34.34 mil- lion) and 36.3 percent (-US$17.7 million) of The mathematical optimization suggests spending under an optimised allocation increasing investment to primary health of spending. centers and community-based interventions. As shown in figure 9, an optimised alloca- Population-wide health interventions re- tion of 2016 national health spending would mained largely unchanged under the opti- increase investment in primary health cen- mised national health spending allocation, ters (+US$33.08 million) to US$438 million, but decreasing investment in both first-level equivalent to 44.7 percent of total spending. and referral hospital spending was recom- Community-based interventions were also mended so that funds can be reallocated to prioritized under an optimised allocation of lower platforms of care (figure 10). First-lev- spending (+US$19.47 million), accounting el and referral hospitals would retain 74.9 for US$112.2 million or 11.4 percent of total percent (-US$34.34 million) and 36.3 percent national health spending. Population-wide (-US$17.7 million) of spending under an opti- health interventions remained largely un- mised allocation of spending. changed under the optimised national health TABLE 1 2016 NHS Spending Comparison, Actual v. Optimised, Zimbabwe, US$ Optimised Platform 2016 Spending Difference Spending Community 92,703,361 112,174,729 + 19,471,367 Health Centre 405,315,540 438,392,777 + 33,077,237 First-level Hospital 136,843,381 102,498,635 - 34,344,746 Referral and Specialty Hospital 27,809,711 10,098,716 - 17,710,995 Population-based 9,091,860 8,598,997 - 492,863 Health Interventions Fixed Programme 308,266,683 308,266,683 N/A (interventions not optimised) Source: HIPtool analysis. World Bank Group 32 FIGURE 9 Optimised Allocations of 2016 National Health Spending Across Intervention Platforms Optimized Spending In Millions (2016 $, USD) 112.2 Community Health Centre 308.3 First-level Hospital Referral and Specialty Hospital Population-based Health Interventions Non-optimised Expenditure 438.4 8.6 10.1 102.5 Source: HIPtool analysis. FIGURE 10 Variations in 2016 National Health Spending by Intervention Platform Difference in Spending by Platform (USD, millions) 40 Community 30 Health Centre 20 First-level Hospital 10 Referral and Specialty Hospital - Population-based Health Interventions (10) (20) (30) (40) Source: HIPtool analysis. Improving Allocative Efficiency In Zimbabwe’s Health Sector 33 4.4.2 Optimised Interventions with hospital or population-wide platforms. The the Highest Expenditure optimised allocation of 2016 national health spending also prioritized integrated interven- The resource allocation on health inter- tions that address multiple conditions (figure ventions can be much more focused. The 10). Integrated community case management 15 optimised interventions with the highest (+US$63.4 million); testing for HIV, STIs, and amount of spending were equal to 58.3 per- hepatitis (+US$23.2 million); and integrated cent of total 2016 national health spending management of childhood illnesses (+US$8.7 (figure 10).18 Nine of the 15 interventions were million) all receive additional funding un- provided through primary health centers der an optimised allocation of spending. (US$418.4 million), four through first-level Though the table provides specific numbers, hospitals (US$71.2 million), and two were it should be used as a directional tool, both community-based interventions (US$81.8 due to limitations in the analysis and because million). The nine health center interven- amounts are contingent on the total national tions shown in figure 10 accounted for 42.7 health-spending envelope specified in the percent of total 2016 national health spend- analysis. HIPtool is able to provide analysis ing. None of the 15 optimised interventions of different scenarios with varying levels of with the highest amount of spending were national health spending. delivered through the referral and specialty t accounts for around 85 percent (US$571.4) of total optimised spending, excluding non-optimised interventions. 18 I World Bank Group 34 TABLE 2 Comparison of 15 Optimised Interventions with the Highest Expenditure, US$, 2016 Optimised Intervention 2016 Spending Difference Spending (Health Center) ART care 262,471,347 238,100,095 - 24,371,252 for PLHIV (Community) Integrated 1,485,542 64,842,867 + 63,357,325 community case management (Health Center) BEmNOC 37,765,604 54,740,233 + 16,974,628 (Health Center) Testing and 12,956,836 36,162,784 + 23,205,948 counseling for HIV, STIs, hepatitis (Health Center) Medical 14,658,943 27,734,623 + 13,075,680 male circumcision (First-level Hospital) Labor and 21,742,752 21,742,752 - delivery in high risk women (Health Center) PMTCT of HIV 20,053,179 21,503,299 + 1,450,120 (Option B+) and syphilis (First-level Hospital) Referral for 19,568,107 21,057,756 + 1,489,649 DST and MDR-TB treatment (Community) Cotrimoxazole 538,496 16,937,770 + 16,399,274 for children (First-level Hospital) 6,170,745 14,206,573 + 8,035,828 Osteomyelitis management (First-level Hospital) Septic 6,170,745 14,206,573 + 8,035,828 arthritis management (Health Center) Medical 1,474,547 11,858,629 + 10,384,081 management of heart failure (Health Center) Integrated 3,026,984 11,691,625 + 8,664,641 management of childhood illness (Health Center) Diagnosis of 4,306,782 8,628,375 + 4,321,593 TB and first-line treatment (Health Center) Psychological 1,587,945 7,961,859 + 6,373,913 and antidepressant therapy Source: HIPtool analysis. Improving Allocative Efficiency In Zimbabwe’s Health Sector 35 FIGURE 11 Optimised Spending for 15 Highest Expenditure Interventions, US$ millions, 2016 Optimized Spending in Millions (2016 USD, $) (Health Center) ART Care for PLHIV 238.1 (Community) Integrated Community Case Management 64.8 (Health Center) BEmNOC 54.7 (Health Center) Testing and Counseling for HIV, STIs, Hepatitis 36.2 (Health Center) Medical Male Circumcision 27.7 (First-level Hospital) Labor and Delivery in High Risk Women 21.7 (Health Center) PMTCT of HIV (Option B+) and Syphilis 21.5 (First-level Hospital) Referral for DST and MDR-TB Treatment 21.1 (Community) Cotrimoxazole for Children 16.9 (First-level Hospital) Septic Arthritis Management 14.2 (First-level Hospital) Osteomyelitis Management 14.2 (Health Center) Medical Management of Heart Failure 11.9 (Health Center) Integrated Management of Childhood Illness 11.7 (Health Center) Diagnosis of TB and First-line Treatment 8.6 (Health Center) Psychological and Anti-Depressant Therapy 8.0 Source: HIPtool analysis. HIV-related interventions continued to Maternal and child health interventions account for five of the 15 optimised inter- remained critical in the optimised allocation ventions, compared with the 2016 allocations and most interventions were delivered at of national health spending. The highest the community and primary health center expenditures by category are shown in figure level. Except for high-risk labor and delivery, 11. The provision of cotrimoxazole to children the maternal and child health interventions at the community level was prioritized over shown in figure 11 were all delivered at com- community-based HIV treatment. Similarly, munity or primary health centers. Under the under the optimised 2016 national health optimised spending allocation, an additional spending allocation, investment for male US$16.9 million was allocated to BEmNOC, circumcision increased by US$13.0 million, equivalent to a 44.9 percent increase. while the amount of spending on ART provi- sion in health centers decreased by US TB interventions remained a strong area $24.4 million. of focus. An additional US$5.8 million was designated to the two TB interventions. The interventions involved diagnosis and first-line treatment at the primary health center level and MDR-TB diagnosis and treatment at a first-level hospital. World Bank Group 36 Four of the 15 optimised interventions with described in Appendix A. As shown in figure the greatest amount of spending addressed 12, under an optimised spending allocation, NCDs. The management of musculoskeletal primary health center interventions would be disorders at first-level hospitals (+US$16.1 responsible for approximately 60 percent of million), heart failure management (+US$10.4 all DALYs averted, which marks an additional million), and psychological therapy (+US$6.4 457 thousand DALYs averted compared with million) at the health center level were all pri- the 2016 spending allocations. Similarly, an oritized ahead of the more specialized NCD additional 349 thousand and 130 thousand treatments delivered at higher platforms of DALYs were averted through community and care under 2016 allocations of national first-level hospital interventions, respectively. health spending. In the optimised spending scenario, the num- ber of DALYs averted by population-wide interventions remained largely unchanged; 4.5 Estimated Impact of an the overall impact of referral and specialty Optimised Allocation of 2016 hospital interventions decreased by nine National Health Spending thousand DALYs. Therefore, an optimised allocation of 2016 national health spending 4.5.1 Potential Impacts of Allocation increased the impact of all but one platform Optimization by Intervention Platform of care, and generated cost-savings across first-level hospital interventions that yielded An additional million DALYs (2.6 million) greater impact with a 25 percent spending could be averted with optimised reallo- reduction (see section 4.4.1). cations of 2016 national health spending, according to the optimization algorithm TABLE 3 National Health Spending Impact Comparison Actual v. Optimised Scenarios, DALYs, 2016 2016 DALYs Optimised Platform Difference Averted DALYs Averted Community 233,380 582,517 + 349,136 Health Centre 1,102,562 1,559,066 + 456,504 First-level Hospital 235,691 365,892 + 130,201 Referral and Specialty Hospital 49,153 39,881 - 9,272 Population-based 16,263 17,864 + 1,601 Health Interventions Source: HIPtool analysis. Improving Allocative Efficiency In Zimbabwe’s Health Sector 37  stimated Impact of an Optimised Allocation of 2016 National Health FIGURE 12 E Spending Across Interventions, DALYs, thousands Impact Impact an ofof an Optimized Allocation Optimized Allocation ofof 2016 2016 NHS NHS Spending Spending (DALYs, (DALYs, thousands) thousands) 365.9 365.9 39.9 39.9 Community Community 17.9 17.9 Health Health Centre Centre 582.5 582.5 First-level First-level Hospital Hospital Referral Referral and and Specialty Specialty Hospital Hospital Population-based Population-based Health Health Interventions Interventions 1,559.1 1,559.1 Source: HIPtool analysis.  ifference in Impact of National Spending by Intervention FIGURE 13 D Platform, DALYs, thousands, 2016 Difference in Spending by Platform (DALYs, thousands) 500 Community 400 Health Centre 300 First-level Hospital 200 Referral and Specialty Hospital 100 Population-based Health Interventions 0 -100 Source: HIPtool analysis. World Bank Group 38 4.5.2 Optimised Interventions with thousand DALYs) of the total number of the Highest Impact DALYs averted, respectively. None of these 15 optimised impactful interventions were Nine of the 15 interventions with the great- delivered at referral and specialty hospitals est impact were delivered through prima- or were population-wide interventions. ry health centers, and combined averted around 1,462 thousand DALYs – equivalent Five of the 15 optimised interventions with to 56.2 percent of all DALYs averted. The 15 the greatest impact were HIV-related, and optimised interventions with the greatest combined they accounted for 39.9 percent impact on the disease burden are shown in of all DALYs averted under an optimised figure 14, and represent 76 percent of the allocation of 2016 NHS spending (figure 14). total number of DALYs averted under an op- The HIV-related intervention that yielded the timised allocation of 2016 NHS spending. The greatest increase in the number of DALYs remaining six interventions with the greatest averted was cotrimoxazole for children. The impact were comprised of four first-level additional US$16.4 million invested in cotri- hospitals and two community-based inter- moxazole under the optimised scenario was ventions, which accounted for 8.5 percent estimated to avert 179 thousand more DALYs. (222 thousand DALYs) and 11.0 percent (285  ntervention Impacts in DALYs Averted of the 15 Most Impactful FIGURE 14 I Interventions Under the Optimised Spending Scenario Intervention Impact, in DALYs Averted (Thousands) (Health Center) ART Care for PLHIV 585.0 (Health Center) BEmNOC 260.7 (Community) Cotrimoxazole for Children 185.3 (Health Center) Integrated Management of Childhood Illness 141.1 (Health Center) Diagnosis of TB and First-line Treatment 123.3 (Health Center) Testing and Counseling for HIV, STIs, Hepatitis 106.1 (Health Center) Medical Male Circumcision 100.6 (Community) Integrated Community Case Management 99.5 (First-level Hospital) Relief of Urinary Obstruction 71.3 (Health Center) PMTCT of HIV (Option B+) and Syphilis 60.0 (First-level Hospital) Referral for DST and MDR-TB Treatment 55.8 (First-level Hospital) Care for Fetal Growth Restriction 51.5 (Health Center) Medical Management of Heart Failure 48.3 (First-level Hospital) Repair of Perforations 43.0 (Health Center) Provision of Insecticide Nets 36.6 Source: HIPtool analysis. Improving Allocative Efficiency In Zimbabwe’s Health Sector 39 BEmNOC remained the intervention that demonstrated how HIPtool can assist coun- yielded the second highest number of DALYs tries with their resource allocation decisions. averted under an optimised allocation. BEm- However, a large number of interventions NOC, under the optimal spending increase of have been considered, and the analysis is US$17 million, generated an additional 80.8 therefore affected more than usual by a scar- thousand averted DALYs. city of data – particularly for NCDs, injuries, and cross-cutting services. The other two critical interventions were integrated management of childhood The quality and robustness of some coun- illnesses and integrated community case try-level unit costs was a major area of con- management, which yielded a significant in- cern, particularly intervention costs offered crease in impact under an optimised spend- at the hospital level. Public hospital service ing allocation. The additional US$72 million charges were used as proxy for the under- invested in both interventions would avert lying costs. However, it proved a very poor 202.8 thousand DALYs more than the 2016 representation of unit costs. The NHS costing spending allocations. had no clear reference to the origins of the unit costs that were used for each interven- Diagnoses and treatment of TB, provision tion among the service categories, making it of treatment for drug-resistant TB, and difficult to ascertain whether the costs were NCD-related interventions comprised the locally generated or were from the interna- rest of top 15 high impact interventions in tional unit cost databases. The country lacks Zimbabwe. Under an optimised allocation of a defined framework that standardizes unit spending (a US$5.8 million increase), diag- costs. Even for the few small studies that nosis and treatment of drug-susceptible and have been done on costing, there is clearly drug-resistant TB (figure 14) would avert an no reference to the use of any standardized additional 65.7 thousand DALYs. Aside from costing protocols. Some unit costs sourced the provision of insecticide nets, the remain- from the NHS were significantly higher than ing three optimised interventions with the the DCP3, but there were also DCP3 unit greatest impact on the disease burden were costs that were significantly higher than NCD interventions. In contrast, the 2016 ac- locally-generated costs. This made it diffi- tual spending allocations highest-impact in- cult to decide which costs were appropriate terventions did not include interventions that for use in the model. In those cases, health addressed NCDs. Combined, relief of urinary expenditure figures in resource mapping obstruction, medical management of heart analysis were crosschecked to make the failure, and the repair of stomach perfora- best selection. tions accounted for 162.5 thousand DALYs averted with an US$19.6 million allocation. Despite extensive consultations with direc- tors, managers, clinicians, national reports, and guidelines there have been challenges 5. Limitations of the Study when mapping local interventions to glob- ally recommended packages and published effectiveness data – particularly for injuries 5.1 Challenges and Gaps in Data and surgical interventions. In some cas- and Intervention Mapping es there was a thin line separating health activities from being misconstrued as health Better quality local data on unit costs and interventions, blurring the actual understand- health coverage is needed to improve ing of the DCP3 interventions. In some cases, HIPtool analysis outputs. This pilot exercise the DCP3 interventions were aggregated. For World Bank Group 40 example, diagnosis, treatment, and manage- Third, HIPtool is not a costing or budgeting ment of a condition were included in one tool, and is intended only to contribute one intervention making it difficult to map to component in the overall process of deter- local interventions. Data on coverage proved mining an effective HBP. Political, logisti- another challenge as some higher-level inter- cal, and other considerations need to be ventions were missing data, particularly for considered outside the context of HIPtool surgical, injury and rehabilitation, congenital in determining what HBPs are feasible. In disorders, and palliative care. order for an explicit benefits package to be sustainably and consistently defined in the context of changing disease trend, shifting 5.2 Limitations of HIPtool cost structures and ways of delivering care, HIPtool has potential to be a transformative and evolving evidence base, localised priority enabler in the difficult country-level process setting mechanisms should be established. of defining what health interventions will be These considerations are outside the scope offered under a UHC benefits package in re- of HIPtool, but are critical for the successful source constrained settings. However, at the adoption of a HBP. current point in development, there are some Fourth, all EUHC interventions are assigned critical limitations of HIPtool. a platform that allows useful indication of First, since there is a trade-off between platform spend. However, the design of usability and flexibility, there are many highly the interventions is not platform specific as complex interactions between diseases and many interventions requiring cross-platform health interventions that cannot be captured care. In addition, the impact and efficiency in full, as these would require more data at which health interventions can be offered than are available in most country contexts. is highly dependent on the availability of As a consequence, overlaps and synergies other interventions, particularly in co-morbid between the interventions included in HIP- conditions. The current algorithm does not tool are not considered; currently, only the provide for intervention interdependency. first-order impacts are incorporated into Last but not least, since HIPtool adopts the analyses. a health system perspective, it is unable Second, the current HIPtool is not a disease to capture cross-sectoral benefits that lie modelling tool and therefore does not ac- outside of the health budget. For exam- count for disease progression and infectious- ple, effects such as gains in productivity or ness. Instead, it builds on the best existing school attendance would not be captured, projections of disease burden and studies of which means that the positive impact of a intervention effects in terms of DALYs avert- HBP estimated by HIPtool is likely to be an ed, which limits outputs of HIPtool to DALYs underestimate compared to its full cross- averted. Analyses using HIPtool are only as sectoral impact. valid as the data entered into them; and in HIPtool will be continuously updated and the absence of high-quality, context-specific some of these limitations will be addressed data, its analyses are more likely to replicate in the later versions. the findings of global studies such as DCP3. Improving Allocative Efficiency In Zimbabwe’s Health Sector 41 6. Policy Recommendations also provides an opportunity to refine interventions proposed in the country’s Zimbabwe was one of the first few countries Community Health Strategy.19 in which HIPtool was piloted at the proof of Similarly, results from the HIP analysis concept stage. The power of HIPtool extends suggests investing more resources in MCH beyond a static and polished report. Rather, interventions that can be offered at the HIPtool can provide instant results based on community and primary care levels, par- various budget scenarios and can be con- ticularly BEmNOC interventions. Maternal stantly updated with the improvement of and child-related interventions are among the underlying unit cost and utilization data. the top 15 in terms of both spending and As discussed earlier, at its nascent stage, impacts. This is consistent with the govern- HIPtool still has strong limitations. Therefore, ment focus on promoting maternal and child the policy discussion below is more indica- health interventions. The country developed tive of potential high-level recommendations its first Maternal and Neonatal Health Road- that could be withdrawn from the results. map (2007–2015) in response to the African The targeted audience are government and Union’s call for African countries to develop development partners. their own MNH Roadmaps. The roadmap A shift of resource allocation from hospi- emphasized the development of a clear tals to community and health centers can referral system for maternal services – from result in a significant increase of DALYs the community level right up to the highly averted. This is consistent with government specialized tertiary and central/quaternary and development partner’s recent discus- hospitals. Most of Zimbabwe’s primary health sion on strengthening community health care facilities are able to offer basic obstet- benefits. One such community initiative is ric care, while most secondary level facilities the Integrated Community Case Manage- and upwards are able to offer comprehen- ment (ICCM). In Zimbabwe, ICCM involves sive emergency obstetric care. The roadmap the training and deployment of community emphasizes effective utilization of scarce health workers to provide diagnostics and resources for cost effective and high impact treatment for communities. The diseases MNH interventions. It promotes the four that they can diagnose and treat include pillars of Safe Motherhood: family planning, pneumonia, diarrhea, malaria, and neonatal antenatal care, clean and safe delivery for conditions for children of families that have mother and newborn, and essential obstetric difficulties accessing treatment at health care. Further, the services have integrated facilities. This intervention is likely to be HIV and STI services. The HIPtool analysis highly cost effective, and investment in its show that the country can achieve better expansion is expected to provide good value health outcomes by investing more resources for money. The extent to which the approach in these interventions that can be offered at is comprehensively implemented varies the community and primary care levels. This across districts and disease areas. The results certainly has implications for health workers from the HIP analysis show that more than a training at the frontline and the delivery of 100,000 DALYs can be averted by reallocat- necessary medical equipment and supplies. ing more resources to this intervention. This  Community Health Strategy is under development, a draft is ready and still to be finalized. 19 A World Bank Group 42 The HIPtool analysis provides evidence on tic financing and external financing. From the the integration of care, such as the prioriti- domestic financing budgeting and execution zation of testing and counseling for HIV, STIs perspectives, implementing PFM reforms and Hepatitis at health centers, that can be to better turn allocated funds into inputs at harmonized by the government with other various service platforms is essential. Under initiatives to holistically enhance efficiencies. external financing, close development part- MoHCC recently undertook some studies on ner coordination with a stronger government implementation efficiencies with support leadership will help implement a coherent from the World Bank and other partners. A health sector strategy. A more transparent three-year study conducted from 2015–2018 budget calendar, broad consultation, and to assess the efficiency gains from integrat- incorporation of evidence-based prioritiza- ing HIV and Sexual Reproductive Health tion mechanisms are some of the keys (SRH) services provided evidence and in- for implementation. sights on the opportunities to improve tech- nical efficiencies through integrated service delivery. The study showed that between 7. Conclusion 2013 and 2016, Zimbabwe’s HIV and SRH response became more integrated at a time Zimbabwe was one of the first few countries that there was also task shifting to primary in which HIPtool was piloted at the proof health care sites. The evaluation showed of concept stage. HIPtool demonstrates its that Zimbabwe could deliver – for the same potential to mathematically estimate an funding – more SRH services. Integration re- optimised spending allocation across a set of sulted in a nine percent drop in the average health interventions to maximize three UHC cost of delivering HIV and SRH services at objectives. It provides a technical foundation district hospitals in Zimbabwe and more than to further develop an essential health benefit a 20 percent drop in the average cost of de- package that can be rapidly and regularly livering services at primary health care sites. adjusted. Admittedly, at this stage, HIPtool Clearly, in the context of UHC with its focus still has several strong limitations. In addi- on people-centered and integrated care, the tion, the tool is constrained by the availability efforts by Zimbabwe to integrate not just and quality of underlying unit cost and uti- within the HIV and SRH program, but across lization data in the low- and middle-income different programs and to put the patient country context. This limitation is particularly first, is essential. severe for interventions falling into non-com- municable diseases, injury or cross-cutting Overall, the analysis provides some concrete care categories, and underscores the global evidence in terms of resource allocation. need to invest in collecting and analyzing However, more in-depth analyses are re- such data for broader use. HIPtool will be quired and implementation remains a chal- continuously updated and applied in several lenge. The government’s efforts in expanding more countries. fiscal space for health need to be comple- mented with the reprioritization of health in order to improve efficiencies in spending the available funding. This includes both domes- Improving Allocative Efficiency In Zimbabwe’s Health Sector 43 Appendixes 44 Photo: Arpit Rastogi Improving Allocative Efficiency In Zimbabwe’s Health Sector 45 Appendixes Appendix A HIPtool: Applicability and Methodology Technical Specifications HIPtool, the Health Interventions Prioriti- zation Tool, leverages the disease burden for the Health Interventions framework of the Institute for Health Metrics Prioritization Tool Evaluation’s (IHME) Global Burden of Disease (GBD) estimates, as well as the interventions framework of DCP (specifically the EUHC Background package), and allows tailoring to specif- This document briefly outlines the Health ic country needs and data. These studies Interventions Prioritization Tool (HIPtool), represent a synthesis of the best available which combines all available country-specific international evidence on the priorities for evidence on intervention cost, coverage, and disease burden and disease control in differ- impact with demographics, disease burden, ent sectors. The following sections provide resource availability, and other data. HIPtool technical details on how HIPtool can be used can help inform policymakers via the three as a preliminary step in exploring the impli- components used to determine interven- cations of different HBP choices. tions: (1) maximize disability-adjusted life years (DALYs) averted, (2) maximize equity, Aims and Scope and/or (3) maximize financial risk protection. Specifically, HIPtool is designed to address HIPtool is designed for countries at vari- the following questions: (1) What is the cost ous stages of progress toward UHC to help and impact of an optimised national package achieve their strategic goals. For countries of health services or interventions based on defining a HBP for the first time, as well as global and local evidence? (2) What health for countries that are reviewing their HBPs, services or interventions outside of the HIPtool facilitates a multi-variate approach to optimised HBP would be cost-effective and decision-making by incorporating available important to deliver? (3) How do changes evidence on costing, impact, and disease in available funding affect the interventions burden within a single analytical framework. included in an optimised HBP? The results HIPtool allows countries to estimate a HBP’s from these analyses can then be linked to potential impact and facilitate preliminary delivery platforms to provide an initial step discussions on priority setting and how to towards developing an optimal HBP. Appendixes 46 improve a package by balancing its project- is by default based on UN Population Divi- ed health impacts with equity and financial sion estimates, although national or other risk protection for certain populations. In estimates can also be used. Default disease addition, HIPtool can be useful for Ministries burden data for each primary cause is based of Health seeking to draft an economic and on the IHME GBD database, although users social case to justify the need and poten- are able to add, remove, or edit causes. Caus- tial returns from a national health insurance es are defined by the following properties: scheme or an increase in funding allocated in a certain way. 1. Primary cause name 2. Health category Questions HIPtool is able to address 3. Population prevalence by year* are as follows: 4. Number of people affected by year* 5. Total DALYs by year  hat packages of health services or inter- •W 6. Total mortality by year ventions should be prioritized for consider- ation for inclusion in an optimised HBP? * If one of these quantities is entered or updated, the other will be automatically calculated.  hat health services or interventions out- •W side of the optimal HBP would be cost- All cause data is visualizable and editable, effective and important to deliver? with import/export options to Excel avail- able. Users are able to enter data for multiple  ow do changes in available funding •H years if available (e.g., 2015, 2020, 2025, affect the interventions included in an and 2030). optimal HBP? Each intervention is defined by the HIPtool is primarily aimed as a gateway to following properties: designing a HBP, by comparing different possibilities of optimal HBPs depending on 1. Intervention name policy objectives and available budgets. 2. Targeted disease By synthesizing and linking best available 3. Delivery platform evidence, HIPtool seeks to provide a starting 4. Unit cost per person covered point for different options for HBPs and their 5. DALYs averted per person covered* potential impacts. Therefore, HIPtool pro- 6. Cost per DALY averted* vides an accessible starting point, preceding 7. Default coverage of intervention† analysis provided by more detailed and spe- 8. Maximum coverage of intervention† cific costing and implementation tools such 9. Equity score as OneHealth. 10. Financial risk protection score † Default data not present. Percentage coverage Data Input Requirements estimates are usually collated from secondary HIPtool is based on country-specific disease sources, and where unavailable DCP3 coverage burden data. Users select their country from assumptions are used. a drop-down selector, with the full list of available countries listed in Appendix A. Where possible, country-specific estimates This is used to pre-populate demographic for each of these measures are utilized; a and disease burden data. Demographic data simple tabular graphical interface is provid- Improving Allocative Efficiency In Zimbabwe’s Health Sector 47 ed to easily update estimates, with full data ratio between target nominal coverage TC import/export available via Excel. Default and current nominal coverage NC, multiplied values for indicators are based on interna- by the effective coverage EC: tional estimates, including DCP3, and in a for- mat that could be populated using data from ECM = (TC /NC)*EC sources such as Tufts Cost-Effectiveness The effective coverage EC is dependent on Analysis Registry (for cost per DALY averted) the disease burden b (in terms of DALYs) and WHO CHOICE (for unit costs). All default and the outcome of the single intervention values are visible to the user, editable, and o expressed as burden averted (in terms of fully documented. In addition, the user has DALYs). This burden is given by the ratio the option to add, edit, or delete interven- of the spending needed to implement the tions, providing a fully customizable list of intervention S and the incremental cost- interventions for a given country context. effectiveness ratio (ICER) relative to the implementation of the intervention. This Impact Model ICER can be reduced by a rate Q (as men- Each intervention in HIPtool is linked to one tioned above) to account for a loss of or more causes of disease burden classified effectiveness during implementation. in the GBD conducted by IHME. The linking EC = O/b of interventions to GBD causes of disease was carried out with guidance from WHO experts. In turn, the burden of disease data Equity and Financial Risk Protection Modules (prevalence, mortality and DALYs) associ- As noted above, individual interventions in- ated with EUHC interventions in HIPtool is cluded in HBPs have health equity and finan- based on this linking exercise. The impact or cial risk protection scores assigned to them outcome of a given set of interventions on by default, sourced from DCP3, which can burden of disease is defined as: be modified by the user based on in- country needs. O = S/(ICER/Q) Health equity is defined in terms of the where O is the outcome expressed as burden heath-adjusted age at death (HAAD). Three averted in DALYs, S is the total amount of general HAAD cut-offs are used to assign a spending on an intervention, and Q is the high- or low-ranking equity score to a health quality factor that reflects realistic imple- intervention. For example, if an intervention mentation of interventions (assumed to be addresses a cause for which individuals have 70 percent reduction in cost-effectiveness). a HAAD of less than 40 years, the interven- tion receives a score of 3; interventions ad- Effective Coverage and Maximal Effective Coverage dressing a cause with a HAAD of more than Each intervention is defined by a maximal 40-50 years receives a score of 2 while caus- effective coverage ECM to reflect real con- es with a HAAD more than 50 years receives straints of scaling up an intervention by pa- a score of 1. HIPtool provides the option to rameterising the upper-bounds of interven- include the current life expectancy, allowing tion spending for the optimization process. for the HAAD cut-offs to be automatically Maximal effective coverage is defined as the scaled up or down, tailoring the health equi- ty scores to a given context. Where data are A key aim of HIPtool is to generate an HBP within a given budget and to meet three defin objectives, which are Optimization Module to (1) maximize DALYS averted, (2) maximize equity, and/or maximize financial risk protection.Module Optimization A key aim of HIPtool is to generate an HBP within a given budget and to meet three defin Optimization Module which A key aim of HIPtool is to generate an HBP within a given budget and to me objectives, Optimizations can be arerun to in (1) two maximize different DALYS averted, (2) maximize equity, and/or to (1)modes. Constrained mode is used to optimize Appendixes 48 objectives, which are maximize DALYS averted, (2) maximize eq A key aim of HIPtool maximize is to generate health impactfinancial an risk (for HBP which protection. within the user a given budget chooses to and maximize to meetDALYs three defined averted), with constrai maximize financial risk protection. objectives, which are imposedto (1) equity andDALYS onmaximize financial averted, risk (by (2) default, maximize the constraint equity, and/or is that equity (3) and financial r available, additional factors can be included constraints may be implemented by the user: maximize Optimizations financial risk protection. protection must canstay be run the same in twoor different improve modes. withintervention the optimized Constrained mode package compared is used to optimize to baselin in the calculation of the equity score, Optimizations in- (1) canfundingbe run for ain given two different modes. must Constrained mode is used health Weighted impactmode (for which instead the user performs a chooses to maximize user-specified weighted DALYs optimization averted), over with health constrai impa cluding socio-economic status, geographic health impact remain (for constant which the (i.e.,user be excluded chooses from to the maximize DALYs averted), Optimizations can beimposed run in equity, andon two equity different financial and risk financial modes. protection; risk (by Constrained default, default mode the weights constraint is used are to 60 is optimize that percent, equity for 20 is and financial percent, andr location, or gender. imposed on equity optimization);and financial (2) funding risk (by cannot default, scale the up orconstraint that equity a health impact (for protectionwhich the percent, user stay must respectively, the chooses normalizedsame or to maximize down improve fasterwith DALYs respect than a withgiven the averted), to maximum rate optimized (e.g., with30 and package per- constraints minimum compared possible to outcom baselin protection must stay the same or improve with the optimized package compa The financial risk protection imposed on equity andWeighted financial mode module risk instead of (by default, performs a user-specified weighted optimization riskover health impa for each measure. Two modethe additional constraint types is that equity and financial cent per year). Ifof we constraints define the funding may be for implemented by the user: HIPtool is based on equity, three dimensions: Weighted (1) risk instead performs a user-specified weighted optimization ove protection must stay the same funding andfor financial orequity, improve a givenand with protection; intervention eachthe optimized intervention default must package as remaincompared a weights budget constant vector are 60 B to , (i.e., the percent, baseline). be 20 excluded percent, fromand t likelihood of impoverishment (LOI) in the financial risk protection; default weights are 60 percent, 20 Weighted percent, mode instead optimization); respectively, performs a user-specified (2) fundingnormalized health cannot with weighted outcome scale respect up optimization (DALYs or down to maximum averted) faster over correspond- health than and a minimum impact, given rate possible (e.g., 30outcom perc absence of public financing; (2) urgency percent, of respectively, normalized with respect to maximum and minimum po equity, and financialfor per each risk protection; year). measure. If we Two default define additional the ing to weights fundingthis types budget are for of 60 each as constraints O percent, ( B ) intervention(as above), 20 may percent, as bethe a implemented budget and 20 vector by the , theuser: hea need of the intervention; and (3) average for each measure. age total equity Two as E(B), andtypes additional the total of constraints financial may be implemented percent, of deathrespectively, and level offunding normalized outcome disability, for (DALYs with awith a givenrespect averted) favour- interventionto maximum corresponding must and toremain minimum this budget constant possible as ()(i.e., outcomes(as be above), excluded the total from equt funding for risk given intervention a protection as F(B), then must we have: remain constant (i.e., be excl for each measure. optimization); Two additional (2) types funding of cannot constraints scale may up beor down implemented faster than a have: given rate (e.g., 30 perc able weighting as () for interventions , and the that total address optimization); financial risk (2) funding protection cannot asscale () up orby , then down the we user: faster(1) than a given rate funding for acausing high disability perdiseases given year). If intervention andwe define must improve the remain funding for each constant (i.e., intervention be excluded as a from budget the vector , the hea per year). If we define the funding for each intervention as a budget vect outcome optimization); (2) funding cannot (DALYs averted) scale up(DALYs or down correspondingfaster = than to ∫ a∑ this given budget rate as ()30 percent the total equ (as above), to(e.g., the health of wage-earners. The LOI in the outcome () averted) corresponding () , budget as () (as above), this absence of public health per year). If we define the, funding as () and the for financing is total espe - each financial risk protection intervention as a budget as () vector , then we , the have: health cially likely to vary between countries be- , and the total financial as () = 0 risk =1 protection as () , then we have: outcome (DALYs averted) corresponding to this budget as () (as above), the total equity cause it is based on unit cost data. Following as (), and the total financial risk protection as () , then () () = = ∫ ∫ we ∑ have: ∑ () (),, any updates made to intervention unit costs () = ∫ ∑ () , by users, HIPtool adjusts the weighting of =0 =1 = 0 =1 =0 =1 () = ∫ ∑ () the LOI accordingly, in order to reflect more , closely the level of financial risk protection= =1 () = ∫ ∑ () , where and are the 0equity where e and andfinancial f are the equity protection risk = () and ∫ financial ∑score per () person , covered awarded by interventions in a given country. defined above or user-defined), and where risk protection score =0 coverage is shown =1 per person covered here as a function of budget (as =0 =1 () = ∫ defined ()or ∑ above , user-defined), and where Optimization Module where and are the equity coverage and financial is shown here risk as a function of score per person covered protection Constrained optimization where =0 is and defined =1 are B the as equity and financial risk protection score per per A key aim of HIPtool defined above an is to generate user-defined), and where coverage is shown here as a function of budget or HBP budget . a given within defined above or user-defined), and where coverage is shown here as a funct where and budget are the and to meet equity andthree financialConstrained risk protection score per optimization ∑person is defined as covered (as = . defined objectives, which defined above or user-defined), Constrained optimization are to (1) maximize and where coverage is defined is shownas ubject hereto as{ a function of budget . DALYS averted, (2) maximize equity, Constrained and/or max(()) optimization is defined as() ≥ , (3) maximize financial risk protection. () ≥ Constrained optimization is defined as ∑ = . ∑ = . Optimizations can be where run in and are the user-specified two different max(()) ubjectminimum ≥ to { ()values for , equity and financial r max(()) ubject to { () ≥ , modes. Constrained mode is used to opti- protection, respectively. ∑ = . () ≥ mize for health impact (for which the user () ≥ max(()) ubject to { () ≥ , chooses to maximize DALYs averted), with where Weighted and are optimization is definedthe user-specified () asare ≥ minimum values for equity and financial r constraints imposed on equity and where financial and where Emin and F the are user-specified the user-specified minimum values for equity a protection, respectively. min risk (by default, the constraint is that protection, respectively. equity minimum values for equity and financial risk where and and financial are the risk protection must user-specified max stay the( () minimum protection, values + () + respectively. for equity ()) and financial ubject to ∑ = risk . , protection, respectively. same or improve with Weighted the optimised optimization pack- is defined as Weighted optimization is defined as Weighted optimization is defined as age compared to baseline). Weighted mode where , , and weighted are user-chosen weights for diease outcome, equity, and financial r Weighted instead performsoptimization a user-specifiedis defined as ( max () + () + ()) ubject to ∑ = . , optimization over health protection, impact,respectively. equity, and max ( () + () + ()) ubject to ∑ = financial risk protection; default weights are 60 percent, max ( where 20 percent, () , and + 20 , () and percent, +are ()) user-chosen ubject to ∑for weights = . diease outcome, , equity, and financial r where , , and are user-chosen weights for diease outcome, equity, a respectively, normalized protection, with respect respectively. to maximum and minimum possible outcomes protection, respectively. where wo , we , and wf are user-chosen weights where , , and are user-chosen weights for diease outcome, equity, and financial risk for each measure. Two additional types of for disease outcome, equity, and financial protection, respectively. 38 risk protection, respectively. 38 38 38 Improving Allocative Efficiency In Zimbabwe’s Health Sector 49 Limitations text-specific data, its analyses are likely to replicate the findings of global studies such First, since HIPtool adopts a health system as DCP3 . perspective, it is unable to capture cross-sec- toral benefits that lie outside of the health Finally, HIPtool is not a costing or budgeting budget. For example, effects such as gains in tool, and is intended only to contribute one productivity or school attendance would not component in the overall process of deter- be captured, which means that the positive mining an effective HBP. Political, logistical, impact of a HBP estimated by HIPtool is like- and other considerations need to be consid- ly to be an underestimate compared to ered outside the context of HIPtool in deter- its full cross-sectoral impact. mining what HBPs are feasible. In addition, a given HBP needs to be carefully costed, Second, since there is a trade-off between with implications for implementation fully usability and flexibility, there are many highly considered. These considerations are outside complex interactions between diseases and the scope of HIPtool, but are critical for the health interventions that cannot be captured successful adoption of a HBP. in full, as these would require more data than are available in most country contexts. As a consequence, overlaps and synergies be- tween the interventions included in HIPtool Appendix B: are not considered, and currently, only the Key Documents Reviewed first-order impacts are incorporated into the analyses. Key documents reviewed are listed below: Third, the current HIPtool is not a disease (i) district core health services package, modeling tool and therefore does not ac- which defines the services that the popu- count for disease progression and infectious- lation is entitled to at the district level or ness. Instead, it builds on the best existing secondary level and below; projections of disease burden and studies of (ii) costed essential health services package; intervention effects in terms of mortality and DALYs averted. The analyses performed by (iii) essential drugs list for Zimbabwe (EDLIZ), HIPtool are only as valid as the data entered which has information on diseases, essential into it; in the absence of high-quality, con- medicines and treatment guidelines; Appendixes 50 (iv) the costed National Health Strategy (vii) resources mapping reports (2015 – (NHS, 2015 - 2020) which had the base- 2018) which mapped government and line and target coverage data and unit cost partner financing by disease and by prioritized packages such as maternal and major expenditure types; newborn health, child and adolescent health, reproductive health interventions, mental (viii) national health accounts (NHA – 2015); health interventions and selected cross-cut- (ix) the National Health Profiles ting packages such as surgical, rehabilitation (2014 - 2017); and and palliative care packages; (x) District Health Information System 2 (v) the health financing policy and health (DHIS2) summary indicator reports. financing strategy; Table B.1 shows an example of selected NHS (vi) fiscal space for health analyses; maternal and child-care interventions and how they were mapped to DCP3. TABLE B.1 Zimbabwe Maternal and Child Care Interventions DCP3 Intervention Intervention Name NHS Definition/Classification Code Antenatal and postpartum Defined as maternal and C1 education on family planning neonatal disorders Counseling of mothers on pro- viding thermal care for preterm C2 Defined as kangaroo mother care newborns (delayed bath and skin-to-skin contact) Management of labor and delivery Defined as labor and delivery man- in low risk women by skilled at- agement and Neonatal resuscitation, C3 tendants, including basic neonatal which is listed as separate interven- resuscitation following delivery tion. Defined as Pre-referral management of labour complications would be Detection and management of FLH1 linked to this, but could also be fetal growth restriction potentially included in “labor and delivery management” FLH2 Induction of labor post-term Defined as beyond 41 weeks Jaundice management FLH3 Defined as localised infection with phototherapy 0 A 50 150 100 250 200 300 RT C ar Millions e PM fo rP TC LH FIGURE C 1 T IV of H 262.5 IV Is (O BE 238.1 pt m on io N ia n O zi C d B+ ) US$ millions Pr an 37.8 ev d en Sy 54.7 ta ph tiv Te e ili s st Th in M e 20.1 g ra an ed ic py 21.5 d al fo C M Intervention Platforms rT Appendix C: Additional ou al B D n e ventions by Existing and ia se C 18.2 Graphs on Top Ten Inter- be lin irc te g um 0.00 s fo ci Sc rH si Optimised Spending, Within re IV on en ,S in TI 14.7 g s, & H Improving Allocative Efficiency In Zimbabwe’s Health Sector D C ep 27.7 ia ar e at gn f os o r iti s is A of t- R 13.0 TB is 36.2 C k on & A do Fi du m rst lts s -li O & ne 5.6 pp H Tr or or m ea 0.00 tu on tm ni st a en ic lC t on Sc tr 4.3 re a en ce 8.6 in pt g iv fo es rH yp 4.1 er 7.1 te  A Comparison of Top 10 Health Center Interventions by Existing Spending, ns io n 3.4 Current Spending 0.00 Optimised Spending 51 A 5 cu 0 0 15 A 10 25 20 La 50 te 150 100 250 A 200 300 RT H bo Te st ig r & C hm h st ar Millions Millions a R De in g e Re is l fo an k iv Appendixes an fe d W er d rP rr C C LH FIGURE C 2 FIGURE C 1 om y al O i ou IV fo PD en n ns A rD M 21.7 el 262.5 cu ST an in te ag 21.7 g BE 238.1 C an d em fo m M rH N rit D en IV O ic R t , C al -T ST US$ millions US$ millions Li B 20.9 M Is 37.8 m T b re 0.0 PM ed ,H 54.7 Is at TC ic ep ch m T al at Su em en M iti rg ia t of al e s ic M H C 13.0 al an 19.6 IV irc Te (O rm ag 21.1 M um 36.2 ed pt in em io ci at en ic n si on io t al B+ n In M ) O o fP 17.6 te a n an 14.7 st eo re 3.1 gr at ag d 27.7 em Sy m gn ed ye an M e nt ph ili lit is cy an of s Se M 15.1 ag H 20.1 pt ana em ea ic rt 21.5 A ge 0.0 D en Fa rt m ia to ilu hr en gn fC re iti t os h s is ild Re M 6.2 Ps yc of ho 1.5 lie an fo ag 14.2 ho TB od 11.9 In fU em lo gi & Ill to en ca Fi ne xi rin rs ss ca ar t la t- tio y 6.2 nd lin e 3.0 n/ O A Tr po bs tr 14.2 C nt ea 11.7 is uc on i-D tm on tio do ep en in n m re t g s ss Es M 2.9 & a 4.3 ch an H nt ar ag 4.8 or m Th 8.6 ot em on er om en ap al y y t C or on Fa 2.8 tr 1.6 ac sc io 0.0 ep 8.0 to tiv m  B Comparison of Top 10 Health Center Interventions by Optimised spending, y es 2.6 4.1  A Comparison of Top 10 First-level Hospital Interventions by Existing Spending, Current Spending Current Spending 0.0 7.1 Optimised Spending Optimised Spending 52 1 7 2 3 5 8 6 9 4 fo Fu 0 Re 10 Re fe 5 0 r P ll 15 10 tin 25 rr 20 re Sup La op te p al fo H bo Millions at rm or ig r & hy rD h Millions N tive ST R De Sc ew is l re bo Car an k iv FIGURE C 3 FIGURE C 2 en rn e d W er in s M om y i Tr g D ea & 8.9 R en n -T 21.7 tm Sp Tr ea 4.9 O B st Tr en ec tm eo ea 21.7 to fE ial en m tm ar iz ed t ye lit en t US$ millions ly 7.4 is -s ta TB Se M 19.6 Se 0.0 pt an ge rv ic ag 21.1 C ic A em hi es rt ld hr en ho 5.0 iti s t od Re M 6.2 Spending, US$ millions C 0.0 lie an an fo ag 14.2 Re C ce fU em pa at ir ar rs C ri en of act 2.8 ar e na t ry A Ex fo O 6.2 no tr 0.0 rF bs re ac In et tr 14.2 tic tio se al uc al n rt G tio M io ro n 2.2 n w al fo & th Re A 2.9 Improving Allocative Efficiency In Zimbabwe’s Health Sector Tr pa rm 0.7 Re R ea ir at cu m es 4.8 tm io te ov tr of C a ic en O ns rit lo tio to bs ic f n Tr f t 0.7 al C ea Ea et Li on 1.9 tm ric 0.0 m tr en rly Fi b ac 3.7 to St st Is ep ag ul ch tiv fE e a em es ar ia ly Br 0.2 M St ea 1.8 st 0.5 an ag C ag 3.7 e an Re C ce pa em ol r ir en or t ec 0.2 of ta Pe 17.6 lC 1.7 rf an or 3.1 Re ce at io pa ir r ns of 0.2 Tu C ba 0.7 lu b 1.5 lL 2.9 Fo ig at ot io n 0.1 2.22 Current Spending Current Spending 0.1  B Comparison of Top 10 First-level Hospital Interventions by Optimsied Spending, 2.22 Optimised Spending  A Comparison of Top 10 Referral and Specialized Hospital Interventions by Existing Optimised Spending 53 1 7 2 3 5 8 6 9 4 fo Fu 0 0 C 10 10 70 20 30 50 60 40 om r P ll m re Sup te p u Tr Millions Millions H nit ea rm or In IV y tm N tive Appendixes do Se -ba ew or rv se en S Tr bo Car FIGURE C 4 FIGURE C 3 Re ic d to ta ea rn e si es fE ge tm s P du 58.7 ar ly Br en 8.9 Ex ne al ea t er um Sp 0.0 St st f o 4.9 ci ra ag C Ea se oc yi e an r -b oc ng C ce ly as cu ol or r s US$ millions ed Va 5.3 ec 0.2 Pu 5.5 ta 1.7 Iro lm cc A lC n in cu & on at te C an a io a ce Fo ry n Ve ta ra r lic nt Re 5.2 ila c 0.2 A ha to tE ci d bi 5.2 ry xt 1.5 fo lit ra at Fa ct rP io ilu io re n re n gn an 4.8 M 2.2 Re an En tW 0.0 pa ag 0.7 vi Ro ir em Optimised Spending, US$ millions ro ta om of nm vi en O en en ru bs t ta s 4.1 te 0.1 Va tr lM ic an cc in 4.3 Fi 0.5 ag at Re st io Re ul V em n pa pa a ita en ir iro In m tf 2.5 of fC 0.2 te in C A or 0.0 le lu 0.5 gr at M In ft b an al se Fo ed d ar rt Li p C Z i ia io a ot om nc n nd m fo 2.0 of C 0.1 D Sh le ia un ity rC 4.3 Re un ft 0.1 hi tin Pa gn C ld tf la os as re op or is e n at H te of M 1.6 hy yd 0.0 an ana Sc ro d ge 0.1 re ce 0.1 Tr m en ph ea en in al tm t g us en an to 1.5  A Comparison of Top 10 Community Interventions by Existing Spending, d 0.0 Tr fM 64.8 ea 0.0  B Comparison of Top 10 Referral and Specialized Hospital Interventions by al tm ar en ia t 1.3 7.4 Current Spending Current Spending 1.3 0.0 Optimised Spending Optimised Spending 54 In 0 te 10 g 70 20 30 50 60 40 Ca rate se d C Millions ot M Co an m rim ag mu ox em ni FIGURE C 4 az en ty ol e t fo 1.5 In rC 64.8 do hi or ld Re re si n Pn d ua 0.5 En eu lS 16.9 vi m pr ro oc ay nm oc in en c g us ta Va 5.3 lM cc 5.5 Iro Optimised, US$ millions an in n ag at & em io Fo n lic en A t 5.2 f ci d or 5.2 fo M M r al ar as Pr ia s eg M ar na 2.0 ke nt Improving Allocative Efficiency In Zimbabwe’s Health Sector W tin W 4.3 A g om SH of en Be In se ha ct 4.1 A vi or ic id 4.3 cu te C e Se h an N et ve ge s re In 1.0 M te D al nu rv 2.3 ia en gn tr tio iti os is o n ns of M 0.1 an a na d ge 2.2 Tr ea m en tm t en to 0.9 fM 1.5 al ar ia  B Comparison of Top 10 Community Interventions by Expenditure, Current v. 1.3 Current Spending 1.3 Optimised Spending 55 Appendixes 56 WWW.WORLDBANK.ORG