Impact Evaluation of Service Delivery Grants to Improve Quality of Health Care Delivery in Cambodia Baseline Study Report Somil Nagpal, Sebastian Bauhoff, Kayla Song, Theepakorn Jithitikulchai, Sreytouch Vong, Manveen Kohli April 2019 Acknowledgements This baseline report was a mammoth undertaking. The H-EQIP pooled fund partners and the study authors sincerely acknowledge the support, guidance, inputs and insights received from several institutions and individuals, without whom this study would not have been possible. Our gratitude commences with immense appreciation and kind thanks to the Royal Government of Cambodia team, particularly the project management of H-EQIP (H.E. Prof. Eng Huot, H.E. Dr. Yuok Sambath and Dr. Lo Veasnakiry) as well as all the heads and officials of provincial health departments, operational district offices, and health centers that supported this study. Mr. Khun Vibol provided very valuable coordination support which is deeply acknowledged. Continued support of Australian Aid, German Development Cooperation (through KfW) and KOICA as H- EQIP pooled fund partners with World Bank, jointly provided the financial resources to undertake this impact evaluation. Financial support from KOICA for the baseline survey costs helped commence the baseline effort even before the pooled TA funding mechanism for H-EQIP was set up. Priya Agarwal- Harding closely supported the pooled fund partners’ coordination efforts and her support is gratefully recognized. Sebastian Bauhoff (now at Harvard School of Public Health and then with the Centre for Global Development) provided technical leadership for the research team in the survey design, sampling methodology and data analysis. The development and refinement of baseline survey tools, in addition to the work done by the study authors, received very important contributions from the World Bank team including Sovanratnak Sao, Voleak Van, Tan Try, Ung Vanny and Karishma D'Souza. Our UN partners, particularly Sok Sokun (Reproductive Health Programme Specialist, UNFPA) and Lailou Arnaud (Nutrition Specialist, UNICEF) provided valuable contributions to their respective sections. Guidance received from H.E. Prak Sophorneary (Under Secretary of State, MOH) on the nutrition related questions is also gratefully acknowledged. Our peer reviewers for this study, Son Nam Nguyen (Lead Health Specialist, GHN01) and Jeremy Veillard (Senior Health Specialist, GHNGE) as well as additional review comments provided by Myint Kyaw (Operations Officer, EACMM) helped us make the storyline and content of this report much more reader-friendly and lucid, and we extend our heartfelt thanks for their invaluable contributions. This survey used the World Bank’s Survey Solutions platform for its computer-aided data collection and analysis efforts. We sincerely acknowledge the close coordination by Bruce Lachlan (then with Angkor Research) and Sergiy Radyakin and Michael Lokshin from the World Bank's Development Data Group for their close support and making this a very smooth process. We thank the entire team at Angkor Research led by Ian Ramage and including John Nicewinter, Kimhorth Keo, Lachlan Bruce and Benjamin Lamberet for their persistent and dedicated efforts throughout the baseline survey, especially in trying to keep to the timelines and ensuring effective coordination with the quality assurance efforts. Sreytouch Vong coordinated the field survey from the World Bank side, including the quality assurance efforts. Sovanratnak Sao and Ung Vanny were important contributors for enumerator training, and also for the quality assurance efforts. The quality 1 assurance team also included Voleak Van, Khun Seila, Thorn Rasmy, Em Phalnida and Loeurng Samoeurn, and their efforts are much appreciated and acknowledged. The authors would also like to place on record their gratitude to Nareth Ly, Anne Provo, Sovanratnak Sao and Voleak Van for providing their insights and contributions for the policy discussions section of this study. Karishma D’Souza and Dan Han contributed immensely to the analysis of Operational Districts’ data and the L2 data analysis included in this report. Manveen Kohli, Karishma D'Souza and Miguel San Joaquin Polo coordinated this baseline study during different phases and kept the wheels of the study in motion. Their efforts are very sincerely acknowledged. Excellent editorial support amidst tight timelines from Usha Tankha is gratefully acknowledged. Da Lin and Sophinith Sam Oeun sustained this mammoth task with excellent and consistent administrative support for a very large number of team members and an even larger number of field visits, which was instrumental for this study seeing the light of the day, and we extend our heartfelt thanks and appreciation to them. 2 Table of Contents Acknowledgements .................................................................................................................................. 1 Abbreviations ............................................................................................................................................ 5 Executive Summary .................................................................................................................................. 7 1. Introduction ........................................................................................................................................ 14 1.1 Background and Genesis............................................................................................................................ 14 1.2 Health Equity and Quality Improvement Project (H-EQIP) ................................................................. 14 1.3 Country and Sector Context ...................................................................................................................... 16 1.4 Current Knowledge on Health Financing Innovations in Cambodia .................................................. 20 Health Equity Fund System ......................................................................................................................................... 21 Service Delivery Grants ................................................................................................................................................ 29 References .......................................................................................................................................................... 34 2. Impact Evaluation and Methodology ............................................................................................. 36 2.1 Intervention: Service Delivery Grants ..................................................................................................... 36 Provincial Health Department and Operational District Level ........................................................................ 36 Hospital Level: CPA-1, CPA-2 and CPA-3 ................................................................................................................ 37 Health Center Level: MPA ........................................................................................................................................... 37 2.2 Theory of Change ........................................................................................................................................ 37 Operational Districts ..................................................................................................................................................... 38 Health Centers ................................................................................................................................................................ 39 Households ...................................................................................................................................................................... 39 Measurable Impact ....................................................................................................................................................... 39 2.3 Impact Evaluation Design .......................................................................................................................... 40 Research Questions....................................................................................................................................................... 40 Service Delivery Grant Treatment Assignment in Practice............................................................................... 42 Sample Design................................................................................................................................................................. 44 2.4 Baseline Data Collection ............................................................................................................................ 44 Sample ............................................................................................................................................................................... 44 Survey Instruments ....................................................................................................................................................... 45 2.5 Ethics ............................................................................................................................................................. 45 3. Findings ................................................................................................................................................ 46 3.1 Health Facility .............................................................................................................................................. 47 Infrastructure .................................................................................................................................................................. 51 Autonomy ......................................................................................................................................................................... 52 3.2 Health Workers............................................................................................................................................ 54 Medical Training............................................................................................................................................................. 55 Health Workers’ Satisfactions.................................................................................................................................... 61 Health Workers’ Secondary Job ................................................................................................................................ 63 3.3 Patient Exit Survey ...................................................................................................................................... 65 ANC Treatment and Counseling ................................................................................................................................ 65 Treatment and Counseling for Children Under 5 ................................................................................................ 66 3 Patient Travel and Expenditure ................................................................................................................................. 67 Traditional Birth Attendants....................................................................................................................................... 68 3.4 Community ................................................................................................................................................... 70 Community Profile......................................................................................................................................................... 70 Access to Facilities Providing Basic Health Services in the Community ....................................................... 70 Distance to the Nearest Facility ................................................................................................................................ 71 Time Taken to Travel to the Nearest Facility ........................................................................................................ 72 Social Capital and Community Empowerment ..................................................................................................... 73 3.5 Household..................................................................................................................................................... 74 Household Characteristics .......................................................................................................................................... 74 Health Utilization ........................................................................................................................................................... 77 Health-Related Expenditures ..................................................................................................................................... 83 3.6 Maternal and Child Health Survey ........................................................................................................... 85 Antenatal Care (ANC) and Postnatal Care (PNC) ................................................................................................. 85 Reproductive Health ..................................................................................................................................................... 87 Children Under 5 Years of Age................................................................................................................................... 89 Limitations........................................................................................................................................................................ 91 3.7 Operational Districts................................................................................................................................... 92 Capacity Building for In-service and Pre-service Training................................................................................. 92 Planning, Monitoring and Supervision .................................................................................................................... 93 Operational District Staff............................................................................................................................................. 94 Financial Planning and Budgeting............................................................................................................................. 95 3.8 Synthesis of Existing Analyses .................................................................................................................. 99 (A) L2 Assessment Score Analysis ............................................................................................................................. 99 (B) HEF Utilization Survey......................................................................................................................................... 107 4. Discussion and Policy Implications ............................................................................................... 110 Utilization of Health Equity Fund System.................................................................................................... 111 Quality of Care ................................................................................................................................................. 112 Managerial Practices, Autonomy and Empowerment .............................................................................. 113 Other Policy Implications ............................................................................................................................... 114 Changing Disease Burden: Noncommunicable Diseases and Chronic Malnutrition ............................. 114 Citizen Engagement, Gender, Outreach and Community Links ................................................................... 114 Equity and Sustainability .......................................................................................................................................... 114 References ........................................................................................................................................................ 115 Annex A: SDG Results Chain .............................................................................................................. 116 Annex B: Power Calculations ............................................................................................................. 118 Annex C: Additional Details for the Power Calculations for Health Centers ............................ 121 Annex D: List of Provinces, ODs, Health Facilities.......................................................................... 123 Annex E: Data Collection Overview .................................................................................................. 124 Annex F: Operational District Survey Questionnaire .................................................................... 131 4 Abbreviations ANC Antenatal Care ART Antiretroviral Therapy CBHI Community-Based Health Insurance CMS Central Medical Store CPA Complementary Package of Activities CSES Cambodian Socio-Economic Survey DFAT Australian Department of Foreign Affairs and Trade DLI Disbursement Linked Indicator DPHI Department of Planning and Health Information FM Finance Management GNI Gross National Income HC Health Center HEF Health Equity Fund HEFO Health Equity Fund Operator H-EQIP Health Equity and Quality Improvement Project HMIS Health Management Information System HSSP Health Sector Support Project HSSP2 Second Health Sector Support Project HW Health Worker IDPoor Identification of Poor Households Program IMCI Integrated Management of Childhood Illness IPD Inpatient Departments iSHPS Integrated Social Health Protection Scheme ISM Implementation Support Mission KfW Kreditanstalt für Wiederaufbau KHR Khmer Riel KOICA Korea International Cooperation Agency L2 Level 2 Quality Assessment MDG Millennium Development Goal MEF Ministry of Economy and Finance MOH Ministry of Health MPA Minimum Package of Activities NCD Noncommunicable Disease NGO Nongovernmental Organization NSPPF National Social Policy Protection Framework NSSS National Social Security System NQEMP National Quality Enhancement Monitoring Program NQEMT National Quality Enhancement Monitoring Tool OD Operational District OOP Out-of-Pocket OPD Outpatient Department PAC Priority Access Card PBG Performance-Based Service Delivery Grant PCA Payment Certification Agency PHD Provincial Health Department 5 PHRD Japan Policy and Human Resources Development PMRS Patient Management and Registration System PMTCT Prevention of Mother to Child Transmission of HIV PNC Postnatal Care PRH Provincial Referral Hospital QA Quality Assurance QOC Quality of Care RGC Royal Government of Cambodia RH Referral Hospital SCD Systematic Country Diagnostic SDG Service Delivery Grant SOA Special Operating Agency STI Sexually Transmitted Infection TBA Traditional Birth Attendant UHC Universal Health Coverage VHSG Village Health Support Group 6 Executive Summary Though structured as a ‘baseline report’ for the impact evaluation, the purpose of this report is two - fold: to provide an overview of the health system innovation in Cambodia that constituted the backdrop for the Health Equity and Quality Improvement Project (H-EQIP) and for the specific interventions which are being evaluated here; and to present the baseline findings of the Service Delivery Grants (SDG) impact evaluation. In the light of these two objectives, the report also includes a discussion and outlines some policy implications on these themes that are embedded in the closing chapter of the report. A. Health System Innovation in Cambodia The Cambodian economy had an average growth rate of 7.6 percent over the period 1994-2015, ranking sixth in the world. With strong economic growth, and a gross national income (GNI) per capita that more than tripled from US$300 in 1994 to an estimated US$1,070 in 2015, Cambodia entered the ranks of a lower middle-income economy. Poverty incidence under the national poverty line h decreased from 47.8 percent in 2007 to 13.5 percent in 2017. Regardless, there is high vulnerability to financial and weather shocks, including health episodes, with Cambodia ranking among the world’s top ten countries in terms of the share of out-of-pocket (OOP) health expenditure. Meanwhile, social protection schemes are in nascent stages, with Cambodia spending less than 0.1 percent of GDP on social assistance compared to the world average of 1.6 percent. Health Equity and Quality Improvement Project (H-EQIP) H-EQIP (2016-2021) is the current flagship project of the Cambodian Ministry of Health (MoH, #50) with cofinancing from the Governments of Australia, Germany and Korea, and the World Bank. Key shifts in H-EQIP design from its immediate predecessor (HSSP2), with an underlying focus on sustainability of financing and implementation for major health system initiatives, include: (i) mainstreaming implementation of project activities through Royal Government of Cambodia (RGC) systems; (ii) increasing funding flows to the subnational implementation level; (iii) building domestic capacity to take over project implementation support, information systems and monitoring roles; and (Wiseman, #34) enhancing the use of output-based payments through Health Equity Funds (HEF), performance-based financing through Service Delivery Grants (SDG), and Disbursement Linked Indicators (DLIs). Operationally, the framework for quality improvement initiatives under H-EQIP includes the following: • To provide additional, flexible financial resources for health facilities to improve the use and functionality of available infrastructure, and maintain availability of necessary supplies and consumables, through fixed lump sum SDGs for all levels of the health system. • To measure and reward improvements in performance of health facilities, health workers’ knowledge and clinical skills, hygiene and infection control, availability of medicines and consumables through performance-based SDGs for health facilities. • To improve performance of health facilities through performance-based SDGs for supervisory levels, to improve monitoring and supervision, introduce standardized assessments of performance, and reward coaching and other measures taken by supervisors. • To improve the competencies and skills of health workers through pre-service as well as in-service training opportunities through disbursement linked indicators. • To promote access and availability of services, especially in remote areas through necessary augmentations to health infrastructure. 7 • To improve health workers’ morale, motivation and remuneration in a manner closely linked to their performance on quality, improved productivity and higher patient satisfaction through an overall focus on results, performance and performance-linked payments. Health Equity Fund (HEF) The HEF is a notable health financing system that purchases services from public health facilities on a pay-for-performance (output) basis (through a reimbursement of user fees on behalf of the poor) that has improved access to health services for the poor, leveraged quality improvements and provided a major source of flexible revenue within the health system. As of May 2015, the HEF system had reached nation-wide coverage to reach over 1,200 health facilities, including all Health Centers, all former District Hospitals, all Referral Hospitals and one of the eight National Hospitals. A further expansion to cover five more national hospitals was completed in mid-2018. The HEF system has grown from a series of small NGO-run pilots in the early 2000s to a government-owned nation-wide social health protection and health financing mechanism providing comprehensive coverage to about 3 million poor people in Cambodia. In light of the overall increase in health services utilization, the recent body of evidence indicates a growth in public health service utilization among HEF beneficiaries compared to those without the entitlement. Several challenges still exist, often arising from systematic factors that contribute to under-utilization among the poor, such as distance and low awareness of the HEF entitlement. Regardless, several studies suggest that the HEF system has made access to health care more affordable, and improved financial protection, especially for higher cost services. Service Delivery Grant (SDG) In 2008, the Cambodian MOH established special operating agencies (SOAs), either based in a provincial referral hospital (PRH) or in an operational district (OD) with the aim to increase utilization and quality of care in identified locations. This was part of an initiative to pay performance-related grants, or SDGs under an internal contract between MOH/Provincial Health Department (PHD) and the SOA. SOAs are designated organizational units with service delivery functions, which are granted some additional delegation of managerial authority and flexibility, jointly by the MOH and Ministry of Economy and Finance (MEF), under a sub-decree and other policy issuances, in return for stronger accountability for performance. Under H-EQIP, the SOA and SDG system underwent a major revision. In the redesigned system, the payment of SDGs to health centers and hospitals is more closely linked to performance in the delivery of basic and comprehensive packages of services, such as critical reproductive, maternal, neonatal, child, and adolescent health services. The MOH SDG manual outlines the intended use for the fixed lump sum grant and performance-based grants (MoH, 2016 #50): • Fixed Lump sum Grants are intended as a complement to the facility operational budget, to manage and implement direct spending for the purpose of promoting quality and equity in patient care. • Performance-based Grants are intended to reward health facilities for quality performance and to reward OD and PHD offices and in particular certified assessors for conducting quality ex-ante assessment. Up to 80 percent of Performance-based Grants can be spent for staff incentives. At least 20 percent of Performance-based Grants are eligible for any other SDG-eligible expenditures. B. SDG Impact Evaluation: Baseline Study The purpose of the SDG Impact Evaluation is to measure the impact – outputs and short-term outcomes – of the implementation of performance-based SDGs, which is in effect at three levels: PHD and OD 8 Offices, Referral Hospitals and Health Centers. Performance of these entities is measured by National Quality Enhancement Monitoring Tools (NQEMT), which are applied quarterly by certified assessors from the OD and PHD offices. These payments supplement the revised fixed lump sum SDG that complement health facilities’ existing operating budget. The quality of health service delivery is expected to be positively reinforced by the quarterly quality assessment using NQEMT, leading to the regular disbursement and use of performance-based SDG. The baseline survey was conducted at the level of OD Offices, Health Centers and Households to measure the existing quality of health service delivery prior to the implementation of the intervention. The schematic in Figure E.1 outlines the theory of change by the SDG implementation, including the expected short-term outcomes1. Figure E.1 Schematic representation of the theory of change from SDG implementation in Cambodia Expected Outcomes Supply Demand A. Improved monitoring and evaluation A. Improved perceptions of quality of B. Improved managerial capacity and financial autonomy care by citizens C. Improved perceptions of quality of care by health worker B. Improved financial protection D. Increased motivation and morale of health worker C. Increased utilization of public health E. Improved health worker clinical competency facilities F. Increased availability of health worker at health facilities G. Improved use and functionality of available infrastructure H. Increased availability of necessary supplies and consumables OD office | Health Center | Health Worker | Exit Exit | Household SDG Impact Evaluation Survey Instruments Methods By focusing on the expected areas of impact, as outlined above, the impact evaluation aims to answer the following research questions: 1. Does the National Quality Enhancement Monitoring Program (NQEMP) at the levels of the operational district and health centers have an average impact? 2. Are the impacts of the intervention heterogeneous with regard to differences in the local context? 1 The SDG results chain is outlined in Annex A 9 3. Does the intervention affect equity? The SDG baseline quantitative survey was conducted from June to July 2016 (supply side) and September to November 2016 (demand side). The planned and completed samples are outlined in Table E.1. Table E.1 Planned and completed samples across different units of observation Unit of observation Original plan Completed Operational district offices (across 23 provinces) 70 69 Health facilities (across 23 provinces) 140 140 Health workers (4 per facility) 560 546 Patient exit interviews (including four antenatal patients and four under-5-child patient 1,120 1,053 caregivers in each health center) Medical record review (4 records per facility) 560 560 Households (20 per facility catchment area, split HEF vs non-HEF) 2,800 2,474 Community (2 per operational district) 140 140 Synthesis of Findings from Existing Studies In order to complement the baseline findings and to provide a more comprehensive view of the health system prior to the H-EQIP, two additional analyses were incorporated using the L2 quality assessment scores and HEF utilization survey results, both of which pertain to a period just prior to the launch of H- EQIP. The L2 quality assessment scores were analyzed to measure any existing differences based on the SOA and HEF status, in the health center performance on clinical vignettes. In theory, being designated as an SOA does entail a different degree of investment and incentives, which would render these facilities different in terms of infrastructure, process and outcomes. The key finding of this analysis was that SOA and HEF Health Centers consistently scored better than non-SOA and non-HEF ones on average across all vignettes. The key findings of the HEF utilization survey revealed that HEF beneficiaries and the near poor were very similar in most assets and wealth indicators, including monthly income, which was around US$150 on average in both groups. But the near poor spent almost three times more than HEF beneficiaries during their last hospitalization. In addition, patients with a chronic illness were 43 percent more likely to use an HEF card for OPD care. Baseline Study Findings The report discusses the analysis of survey responses from seven different data collection instruments capturing both the supply and demand dimensions of the health system. The schematic in Figure E.2 outlines the contextual findings according to the three levels of impact analysis: OD office, health center and household. The results in the OD office and Health Center columns represent the supply-side context, outlining the key findings from the OD, health facility and health worker surveys, and partly 10 from the patient exit survey (focused on content of care); the results in the Household column represent the demand-side with the key findings from the community, household and maternal and child health surveys, and partly from the patient exit surveys (with a focus on patient opinion). Figure E.2 Schematic representation of the key findings of the baseline study across three levels of impact analysis OD Office Health Center Household •47% of the ODs conduct •50% of facilities owned a •Lower than 60% of HEF monthly M&E of health functioning computer; 44% household heads centers; 34% quarterly had access to internet understood their benefits •63% assessed the training •27% disagreed or strongly towards indirect cost needs, 63% organized in- disagreed on having exemption service training, and 63% enough authority to obtain •34% of HEF cardholders at helped with career the resources for the health centers reported planning for staff at health facility being reimbursed some of centers •Low satisfaction on the transport costs •64% ODs set and awarded quantity of medicine (54%) •The most accessible health financial & non-financial and equipment in facility center was on an average incentives for OD staff (44%) about 3 km away from the •93% of the ODs prepared •Satisfaction on salary was villages; the average an annual budget for the 37% distance to drug seller was 2015 fiscal year; 87% of the •45% of health workers 1km from the villages ODs monitored OD currently engaged in other •83% of HEF holders visited expenditures against the economic activities as their OPD from a private health agreed budget secondary job provider; 45% HEF holders •36% of the ODs reported •86% and 95% of total ANC visited IPD from a private having complete freedom patients received the health provider on resource allocation; 10% measurement of weight •49.4% of HEF households no freedom at all and blood pressure, and 49.8% of non-HEF •Expenditure on drugs respectively households sought care at accounted for the highest •HEF patients received more pharmacy and drug seller spending category at 26% laboratory services than at the most recent visit of all budget expenditures non-HEF patients: blood •94% of HEF households and •59% of OD directors think sample (66% HEF, 27% non- 91% of non-HEF that the OD staff is highly HEF), urine sample (43% households expressed out- motivated; 32% think the HEF, 8% non-HEF) of-pocket expenditures on staff is motivated •Almost all children received medicine (largely driven by weight measurement; 13% expenditure incurred at HEF/9% non-HEF children private providers) received height measurement; 3% HEF/2% non-HEF children received upper arm circumference measurement Discussion and Policy Implications The impact evaluation serves the important objective of providing evidence of the effectiveness of the SDG program to the key stakeholders, and helps facilitate the achievement of H-EQIP objectives in conjunction with the broader health system elements by guiding policy dialogue. • Utilization of HEF system 11 There are several persisting challenges that have been identified by the existing body of published evidence, and the emerging findings from the baseline survey, which suggest where future efforts at improving the program need to be directed. In particular, the competency and in some cases, also the attitude of providers at public facilities has been below expectations, which has led to underutilization of public health facilities. It is also increasingly clear that indirect costs are important barriers to access among the poorest members of the population, which have deterred HEF beneficiaries from seeking care at public health facilities. It appears that patients with chronic diseases have not been able to benefit from their HEF entitlements due to the service availability and competency issues around NCDs, as well as the inability of HEF to cover transport costs related to frequent outpatient visits. On the positive side, HEFs and SDGs also seem to steadily produce a virtuous circle on facility volumes and perceived quality. This virtuous circle also results in the amplification of the utilization of health facilities thereby adding to the impact of these health financing instruments. As such, HEF expenditure has remained below 5 percent of the total public health expenditure in the country, even after nation- wide rollout. This incremental performance-linked payment provides the much needed resources and incentives to create a significantly higher output from the underlying public health investments. • Quality of Care It is also evident that quality of care remained a large pending agenda at the time of the baseline survey. Early investments being made in improving facility infrastructure, amenities and supplies, using the fixed lump sum SDGs, are already beginning to reveal the enthusiasm and innovations that are being tried out by facilities. It was interesting to see examples where the small investments have created tools (such as newborn packs) that allow greater interaction opportunities with patients, which are being used for health promotion and counseling. Greater clarity on the exact modality and rules surrounding these funds, as sought and being progressively understood by the health facilities, would further help the health centers to use these grants to their advantage, for improved patient care. Quality is a key focus for the SDG system, and will remain so for this impact evaluation too. • Managerial Practices, Autonomy and Empowerment The baseline context shows that there was somewhat limited level of autonomy and satisfaction in the targeted areas for improvement at health center-level, such as the authority to manage resources. In what may well prove to be transformative and a game-changer in the Cambodian context, all 1200-odd health centers were opening their bank accounts to directly receive SDG funds electronically, around the time this baseline survey was being conducted - and so the impact of receiving this additional resource, in a timely manner, will be something to watch out for in future rounds of this study. The changing levels of empowerment and the changing revenue mix at health facilities and OD offices, most notably due to the implementation of performance-based SDG, HEF reforms, and reducing share of external funding, creates several favorable dynamics around equity and sustainability. Improving autonomy and motivation at health facilities is one of the expected impacts of the SDG intervention; refinement in these areas has already been captured via other means of early H-EQIP documentation. • Other Implications Cambodia’s health system will need to rapidly address noncommunicable diseases and the continued burden of chronic malnutrition, as it faces an epidemiological transition. HEFs and SDGs may well become the platforms to bring about systemic capabilities for such an evolution of the system, and will remain an area for the impact evaluation to explore and document. 12 With considerably reduced outreach in recent years, driven by the changed resource context, there is an increasing need for stronger community links and outreach, especially to meet the needs of very remotely located communities. This will require changes in the institutional arrangements, finances, health workforce and information system. Going forward, the impact evaluation would provide insights toward the changing situations and roles of health institutions vis-à-vis community-based entities. Ensuring equity is at the heart of the H-EQIP, transcending the highly intertwined and mutually reinforcing architecture of the SDG and HEF systems. In the impact evaluation, the household and patient exit surveys can be stratified by HEF status as well as other patient, individual and household characteristics, enabling comparisons of areas of impact through the equity lens. The OD and HF surveys capture essential aspects of the SDG and HEF interventions, which can again be stratified according to key characteristics to examine contextual factors that may hinder achieving the expected program outcomes. These will be invaluable pieces of information in guiding the current and future health policy in Cambodia, given the substantial efforts and resources that have been put into the evolution of these programs. 13 1. Introduction 1.1 Background and Genesis The objective of this report is two-fold. It presents an overview of health systems reform in Cambodia and an analysis of the findings emanating from an impact evaluation baseline study undertaken by the pooled fund partners of the Health Equity and Quality Improvement Project (H-EQIP). The primary goal of H-EQIP is to improve the quality of health care services and financial protection for vulnerable groups. Chapter 1 provides the background and country context of H-EQIP, details of the project as well as a synthesis of the literature on what is already known about the Health Equity Fund (HEF) system and Service Delivery Grants (SDGs), the two key interventions supported under H-EQIP. The impact evaluation measures outputs and short-term outcomes of the implementation of performance-based SDGs, which are a major component of H-EQIP. Performance-based SDGs under H- EQIP have been completely redesigned from the earlier version of the SDG program implemented under the Second Health Sector Support Project (HSSP2), the predecessor of H-EQIP. The conclusion of HSSP2 in June 2016, and the phased roll out of the redesigned SDG system from May 2017 onwards, have provided the opportunity to design a systematic impact evaluation. The baseline data collection was undertaken in the second half of 2016; this report tries to integrate the new knowledge from this baseline survey with the existing body of knowledge, to provide a deeper understanding of the context within which these health financing innovations have been rolled out in Cambodia. 1.2 Health Equity and Quality Improvement Project (H-EQIP) H-EQIP is the new flagship project of the Cambodian Ministry of Health (MoH, #50) with cofinancing from the Governments of Australia, Germany and Korea, and the World Bank. It builds upon the innovations and achievements of its predecessor projects, the Health Sector Support Project (HSSP) and the Second Health Sector Support Project (HSSP2); and in particular, consolidates and scales up proven, potentially transformative interventions such as the HEF system and SDGs. Key shifts in H-EQIP design from its immediate predecessor (HSSP2) include: (i) mainstreaming implementation of project activities through Royal Government of Cambodia (RGC) systems; (ii) increasing funding flows to the implementation level; (iii) building domestic capacity to take over project implementation support and monitoring; and introducing use of output-based payments through HEFs, performance-based financing through SDGs, and Disbursement Linked Indicators (DLIs). H-EQIP accelerates overall reforms in the health sector, improves financial protection for the poor and vulnerable groups and expands access and coverage of health services, while strengthening their quality and affordability, and creating sustainable government institutions for health care management. Component 1 of H-EQIP (US$74.2 million) focuses on Strengthening Health Service Delivery. This component invests in the redesigned SDGs, providing performance-based financing to different levels of the health system based on results achievement. A total of US$34.2 million is provided by the Royal Government of Cambodia as the fixed component of SDGs, and an additional US$40 million, is shared equally by the government and H-EQIP pooled fund partners (US$20 million each). It includes SDGs for Health Centers (HCs), Referral Hospitals as well as for the supervisory levels - the Provincial Health Departments (PHDs) and Operational Districts (ODs), measuring and incentivizing their performance (Figure 1.1). 14 Component 2 (US$70 million) focuses on Improving Financial Protection and Equity, and invests in further expanding the HEF system, increased domestic ownership and creating sustainable domestic institutional arrangements for managing HEFs. This component builds on the success of the home-grown HEF system, aiming to improve utilization by the poor and ensure sustainability by transferring implementation responsibilities to domestic institutions in a planned and progressive manner during the project implementation period. The newly created Payment Certification Agency (PCA) is expected to gradually enhance its capacity and skillsets to perform the HEF monitoring and payment verification roles, as well as manage the information system for HEFs. Component 3 (US$31 million) focuses on Ensuring Sustainable and Responsive Health Systems. It uses results-based instruments, known as Disbursement Linked Indicators (DLIs) to improve supply-side readiness and strengthen health sector institutions, enhance the quality of pre-service and in-service training programs for health workers, equip health facilities to meet minimum standards for providing obstetric and neonatal care, enhance health service quality monitoring, improve timeliness of SDG and HEF payments and establish sustainable health service purchasing arrangements. This component also finances civil works identified in MOH’s civil works plan for 2016-2020, prioritizing investments in remote areas, and addressing concerns of patient safety and improvement of maternal and neonatal survival. Support to project management activities is also included in this component, including day-to- day coordination, administration, procurement, financial management, environmental and social safeguards management and monitoring and evaluation. A grant of US$1 million from the World Bank- Government of Japan Policy and Human Resources Development trust funds provides complementary financing to this component focused on monitoring and evaluation activities (Figure 1.2). Figure 1.1 H-EQIP Cost by Component (in US$ million) H-EQIP was approved by the World Bank Board of Executive Directors on May 19, 2016 and became effective on November 9, 2016 with a five-year duration, through June 2021. 15 Figure 1.2 H-EQIP Cost by Financing Sources (in US$ million) Operationally, the framework for quality improvement initiatives under H-EQIP includes the following: - To provide additional, flexible financial resources to health facilities to improve the use and functionality of available infrastructure, and maintain availability of necessary supplies and consumables, through fixed lump sum SDGs for all levels of the health system. - To measure and reward improvements in performance of health facilities, health workers’ knowledge and clinical skills, hygiene and infection control, availability of medicines and consumables, through performance-based SDGs for health facilities. - To improve performance of health facilities through performance-based SDGs for supervisory levels, to improve monitoring and supervision, introduce standardized assessments of performance, and reward coaching and other measures taken by supervisors. - To improve the competencies and skills of health workers through pre-service as well as in- service training opportunities through disbursement linked indicators. - To promote access and availability of services, especially in remote areas through necessary augmentations to health infrastructure. - To improve health workers’ morale, motivation and remuneration in a manner closely linked to their performance on quality, improved productivity and higher patient satisfaction through an overall focus on results, performance and performance-linked payments. To understand the overall impact of H-EQIP in general, and of performance-based SDGs in particular, would need a combination of administrative data and survey data. Administrative data would include the periodical objective assessment and scoring of health facilities, deviations observed between original ex-ante assessment scores and cross-verification scores, and how these triangulate with datasets from HEF and the routine health information system data. The remaining information needs for the impact evaluation are being met using specially commissioned surveys at the baseline, midline and endline points, as detailed in Chapter 2. 1.3 Country and Sector Context Due to rapid and sustained growth, Cambodia has become one of the world’s leaders in economic growth, poverty reduction and shared prosperity. Cambodia had an average growth rate of 7.6 percent over the period 1994-2015, ranking sixth in the world. With strong economic growth, and a gross 16 national income (GNI) per capita that more than tripled from US$300 in 1994 to an estimated US$1,070 in 2015, Cambodia entered the ranks of a lower middle-income economy. In addition to strong economic growth, Cambodia has achieved dramatic poverty reduction. Poverty incidence under the national poverty line has decreased from 47.8 percent in 2007 to 13.5 percent in 2017. From 2004 to 2007, poverty reduction was driven by the movement of people out of agriculture and into the fast-growing garment and services sectors. Poverty reduction then became particularly dramatic during the 2007-2009 period, at the peak of the agriculture commodity boom, and in a context of expansion in cultivated area, when poverty declined by 25 percentage points and 3.3 million people escaped poverty. However, Cambodian households face a high degree of economic vulnerability. According to the 2017 Systematic Country Diagnostic (SCD), most Cambodians not in extreme poverty are either moderately poor or vulnerable according to international standards. Two-thirds of the population lives under US$5.50 per day PPP. There is high vulnerability to financial and weather shocks, with Cambodia ranking among the world’s top ten countries in terms of out-of-pocket (OOP) health expenditure. Meanwhile, social protection schemes are in nascent stages, with Cambodia spending less than 0.1 percent of GDP on social assistance compared to the world average of 1.6 percent. Cambodia remains a leading example of how a low-income country can quickly advance toward health goals; progress and innovations in health financing and in service delivery contributed to the achievement of all health-related Millennium Development Goals (MDG). The government is improving access to care through a larger health workforce, as well as by improving the infrastructure of health facilities. There are continuing efforts to improve the quality of health services, which encompass pre- service training, in-service training, an ambitious performance-based financing program, and efforts to strengthen regulation. The National Social Policy Protection Framework (NSPPF) 2016–2025, an inter-ministerial initiative, outlines the country’s direction toward Universal Health Coverage (UHC). Ongoing public financial management reforms include the implementation of a Financial Management Information System (FMIS), program-based budgeting in the health sector, and increasing levels of financial autonomy and greater funding for peripheral health facilities. Health coverage is expected to expand, with social health insurance for the private formal sector aiming to cover 1.2 million people through a mandatory contributory scheme. A new civil servants’ scheme (covering 235,000 beneficiaries) and a plan for covering informal workers have recently been launched (Table 1.1). So far, there has been notable improvement in the maternal and child health outcomes. The maternal mortality ratio decreased from 442 per 100,000 live births in 2005 to 170 per 100,000 live births in 2014, and under-5 mortality rate decreased from 83 per 1,000 live births in 2000 to 35 per 1,000 live births in 2014. Consequently, in 2010 the MDG Goals Progress Index was estimated that ranked Cambodia as the fifth best performer out of 76 countries. Strong political commitment - combined with a willingness to innovate - yielded significant improvements in service delivery including dramatic increase in facility- based deliveries (10 percent in 2000 to 83 percent in 2014), uptake of antenatal care, and coverage of other maternal and child health services. 17 Table 1.1 Human Development Indicators, Cambodia (2000-2014) Indicator 2000 2014 Total Population 12,152,354 15,270,790 Total Fertility Rate 3.8 2.6 Life expectancy 58 68 Infant mortality rate 80 29 Neonatal mortality rate 36 17 Maternal mortality rate 437 170 Source: WDI, 2017 Despite these improvements, there remains the need for stronger efforts to reduce inequities in care due to geography and income. For example, reduction in child mortality since 2005 was twice as high in urban areas compared to rural, and higher for the richest income quintiles compared to the poorest. Child mortality has remained unchanged at 3.3 times higher for the poorest quintile compared to the wealthiest quintile since 2005 and three times higher for rural children compared to urban children. Inequities in utilization of health services accounted for part of the inequities in health outcomes. Moreover, health spending is a major source of debt and impoverishment for the poor and near-poor, and the chronically ill. Despite an overall decline in health spending and catastrophic spending as a percentage of income in recent years due largely to rising incomes, an estimated 2 percent of Cambodians fell into poverty in 2011 due to health costs. Health spending remains a significant burden on the poor, with about 18 percent incurring debt because of health expenses. The three rounds of Cambodian Socio-Economic Surveys (CSES) undertaken between 2004 and 2014 provide a robust, survey-based evidence of how access to health care for the most vulnerable groups in 2014 was remarkably different from that in 2004. This is likely to have been caused by multiple initiatives that were ongoing in this period, but HEF and Special Operating Agencies or SOAs are likely to have been the most significant health financing interventions in that period, creating this observed impact. An analysis of the surveys reveals a two-fold increase in the proportion of care-seeking behavior at public health facilities during the ten-year period (Table 1.2). While this may be notable in itself, a similar trend was seen at private health facilities during the same period, indicating the overall increase in utilization of health care. Table 1. 2 Proportion of Households Seeking Medical Care (2004, 2009, 2014) Sought medical care 2004 2009 2014 At public health facility 10.7 18.4 23.4 At private provider 31.9 56.9 64.1 From 2004 to 2014, the equity gap associated with access to care was substantially reduced, as those residing in other urban and rural areas had more access when ill (Figure 1.3). There was approximately a two-fold increase in the proportion of those residing in rural areas seeking care and a 1.5-fold increase among those in other urban areas. By 2014, all populations had higher than 80 percent access rate to care regardless of the place of residence, indicating more equitable access to medical care across the country. 18 Figure 1.3 Access to Medical Care When Ill by Three Regional Groups (2004, 2009, 2014) Source: Antunes et al., 2018; GIZ, 2014. Similarly, there was a significant improvement in access for the poorest populations groups. Progressive surveys indicate a much higher access to care in 2014, across all consumption quintiles (Figure 1.4). In 2004, access to medical care was highly segregated according to economic status of households; only less than 45 percent of those at the lowest consumption quintile had access to medical care whereas more than 65 percent of those at the highest consumption quintile had access. By 2014, more than 80 percent of the lowest quintile group had access to care, and the difference between the proportion of ill population at the lowest quintile accessing care and that at the highest quintile had also narrowed down considerably, to approximately 10 percent in 2014 compared to 25 percent in 2004. Figure 1.4 Proportion of Access to Medical Care by Consumption Quintile (2004, 2009, 2014) Source: Antunes et al., 2018; GIZ, 2014. Incidence of impoverishment from health expenditure, which used to be at 3.5 percent in 2004 was reduced to 0.9 percent in 2014. Between 2004 and 2014, there was an overall two-fold increase in the average household capacity to pay (Figure 1.5). While it is plausible that the increase in both capacity to 19 pay and OOP expenditure was affected by the overall rise of national economic growth and prices, there was a decrease in OOP as a proportion of the capacity to pay and incidence of catastrophic expenditure, indicating that more households became protected against health-related expenditures. Overall, incidence of impoverishment from health expenditure, which used to be at 3.5 percent in 2004 was reduced to 0.9 percent in 2014, showing a 75 percent reduction over the 10-year period. Figure 1.5 Trends In Health-Related Household Expenditures (2004, 2009, 2014) Source: Antunes et al., 2018; GIZ, 2014. There is more that needs to be done in order to accelerate progress and close remaining gaps in basic outcomes (including nutrition, immunization, and neonatal mortality especially as seen with an equity lens to narrow health disparities) and lay the foundation to address emerging issues such as pandemics and noncommunicable diseases (NCDs). Strengthening the supply-side capacity to deliver these services, stronger links to community services and overall focus on capacity for health protection and promotion, including outreach services through health workers, community-based health workers and volunteers, are major areas for further development. Policy to reorient the health system toward strengthening preventive and primary care and addressing social determinants of health is progressing, but requires further investments. 1.4 Current Knowledge on Health Financing Innovations in Cambodia Two major health financing initiatives stand out in the Cambodian context, both of which have endeavored to enhance the performance of the public health system with a focus on the poor and aim to improve financial protection including addressing high OOP spending. They are the Health Equity Fund (HEF) system, which aims to improve access to health services for the poor, and Service Delivery Grant (SDG), a financing program that has rewarded health facilities with flexible resources to improve provision of quality health services at all levels of health facilities. This section aims to describe the history and impact of HEF and SDG based on the published literature and anecdotes. It is important to delineate that the SDGs described in currently available literature refer to the system operated through ‘Special Operating Agencies’ or SOAs, prior to the redesigned SDGs introduced in H-EQIP. 20 Health Equity Fund System The HEF system has grown from a series of small NGO-run pilots in the early 2000s to a government- owned nation-wide social health protection and health financing mechanism providing health insurance to about 3 million poor people in Cambodia. The HEF purchases services from public health facilities on a pay-for-performance (output) basis that has improved access to health services for the poor, leverages quality improvements, and provides a major source of flexible revenue within the health system. As of May 2015, the HEF system had reached nation-wide coverage covering over 1,200 health facilities, including all Health Centers, Former District Hospitals, Referral Hospitals and one of the eight National Hospitals. New institutional sustainability measures, such as the recently created autonomous payment certification agency for HEFs, will provide strong governance support for critical institutional functions. HEFs have been financed through a pooled fund managed by the World Bank, and with contributions from bilateral and multilateral donors. The share and ownership of the Royal Government of Cambodia (RGC) has steadily increased and now a minimum of 50 percent of HEF costs are provided by RGC, expected to rise to over 70 percent of costs by 2021. HEF coverage for the poor recognizes two different ways of beneficiary identification. The first is through the national poverty targeting process owned and managed by the Ministry of Planning (IDPoor) which pre-identifies households and provides them with an “Equity Card.” Commune and Village level local authorities conduct the pre-identification process every three years or so, nation-wide. All individuals identified as IDPoor qualify for HEF benefits for the full period of three years until the next round of pre-identification. The second mechanism is provided within the HEF system itself through a post-identification interview at an HEF-covered hospital that provides them with a “Priority Access Card.” Post-identification helps cover the patients who may become impoverished within two rounds of IDPoor, including sometimes by the health episode itself. An interview that follows the same criteria used by the pre-identification process is conducted on demand with patients who do not have IDPoor cards, but self-report as being poor and are unable to pay the user fees. HEF benefits are awarded after a successful post-ID interview and are subsequently also confirmed with community verification on a random sample basis. The Priority Access Card provides all members of the patient’s household with HEF eligibility which lasts until the next IDPoor exercise is undertaken in that region. Households that have been identified as poor through either of the two modalities, receive a pre- defined and standard set of benefits which have been standardized by an inter-ministerial Prakas in May 2018. These include payment of service fees for the beneficiaries at public health facilities, and additional HEF benefits such as transportation reimbursements, food allowances, and funeral support as detailed in Table 1.3. HEF benefits are fully portable: eligible poor (including the post-identified poor) can seek health care at any contracted public health facility nation-wide. HEF-supported patients in order to receive transportation reimbursements at Referral Hospitals, are required to have a referral letter from a Health Center or pre-arranged appointment in order to encourage patients to access the appropriate levels of care.2 Tertiary referrals to the national hospitals (earlier limited to the Khmer- Soviet Friendship Hospital and now expanded to all other national level hospitals in Phnom Penh) are required to be pre-arranged and have confirmation to ensure that the necessary services are available at the national hospital where the patient has been referred. 2 Except in cases of delivery, attempted delivery, post abortion care, permanent contraception or emergency. 21 Table 1. 3 Categories of HEF Benefits National Hospitals & Former District Hospital Health Center Item CPA 1-3 Referral Hospital IPD OPD IPD OPD OPD User Fees      Delivery, Delivery, Attempted Attempted Transport   Delivery, and Post No Delivery, and Reimbursement Abortion Care Post Abortion Only Care Only Delivery and Caretaker Food  No Attempted No No Support Delivery Only Funeral   No No No Support Public health facilities are paid by the HEF system on a case-based output payment system (Table 1.4) and are included in the latest inter-ministerial Prakas. Payments are made directly to facilities at the end of each month based on electronically documented utilization by HEF beneficiaries. Transportation reimbursements are calculated for HEF beneficiaries based on the actual traveling distance from their residence to the Referral Hospital. A patient allowance of US$1.25 per day is paid for HEF beneficiaries who are admitted for in-patient care and serves to cover incidental costs including their caretaker’s food. Table 1.4 HEF Schedule of Payments to Providers (May 2018) (in US$) Service Packages & Health CPA1 CPA2 CPA3 NHs Payment Rate Center OPD 1.00 2.50 4.00 8.00 10.00 Delivery/Attempted Delivery/PAC* 20.00 20.00 20.00 20.00 20.00 Long-Acting Reversible Contraception 5.00 5.00 5.00 5.00 5.00 (IUD/Implants)/ Family Planning IPD (CPA/NH: with minor surgery) 20.00 20.00 25.00 30.00 35.00 IPD with major surgery** N/A N/A 80.00 250.00 300.00 Minor Surgery N/A 40.00 50.00 100.00 100.00 Emergency 5.00 62.50 62.50 75.00 80.00 *Referral Hospital facilities at the CPA 1-3 levels will receive a case-based payment for attempted deliveries that are referred to a higher level of care if there is a documented referral that clearly states a valid reason for the referral. This payment is to ensure that there are no financial considerations influencing the decision to refer a delivery to higher care. **All surgical cases are defined by the use of general or epidural anaesthesia during the procedure. The Patient Management and Registration System (PMRS) is a customized and fully Cambodian web- based application designed to manage all HEF patient information nation-wide that was developed under the stewardship of the Department of Planning and Health Information (DPHI) of the MOH. Specific tools are built into PMRS for the management of the HEF system. Under USAID funding, University Research Co. (now called ‘URC’) has been engaged by MOH and the pooled fund partners cofinancing the Health Equity Fund system to function as the “HEF Implementer”. This includes 22 maintaining PMRS as well as managing a system of field-based monitors covering each of the country’s Operational Districts to conduct household and key informant interviews, bedside monitoring, and document reviews to verify HEF claims. In light of the overall increase in health services utilization, the recent body of evidence indicates a growth in public health service utilization among Health Equity Fund (HEF) beneficiaries compared to those without the entitlement (Bigdeli, 2016 #31;GIZ, 2014 #39). This held true not only between the rich and poor, but also when the non-beneficiaries had almost the same income status as HEF beneficiaries, or the so-called “near-poor”. For example, approximately 10 percent of the near-poor population chose to visit public health facilities compared to 25 percent of HEF beneficiaries in the year 2015, which was an increase from 20 percent of HEF beneficiaries who had access to free health care in the year 2011. (Figure 1.6) (Bank, 2016 #28). The growing utilization of public health services by HEF beneficiaries may have been due to the nation-wide coverage of HEF as well as the increased awareness of their entitlement to free health services, given that there was a 2.5-fold increase in utilization when the beneficiaries received instructions on how to use the entitlement card (Bank, 2016 #28). One analysis predicts that the access rate is expected to improve further with the increased use of the entitlement (Ensor, 2017 #30). The increase in utilization may be observed at different levels of public health services. Firstly, there is growing evidence highlighting the increasing role of Health Centers (HC) as a primary-level care and a referral system in Cambodia. Between 2006 and 2013, approximately 60 percent of the public health services utilization had been captured at HCs, while the remaining use was at the Referral Hospital (RH) inpatient departments (IPD) and outpatient departments (OPD), which was again higher for HEF beneficiaries than non-beneficiaries (Table 1.5) (Annear, 2016 #33). There was a higher utilization rate of IPD (40 percent) compared to OPD (10 percent) among HEF beneficiaries based on the World Bank 2013 rural health market study (Bank, 2016 #28). In addition, there was a higher increase in access rate to these services and new born services by HEF beneficiaries at district RHs than the provincial RHs (Annear, 2016 #33). These trends may indicate growing referrals made after an HC visit by HEF beneficiaries. Figure 1.6 Utilization by Service Type for the Near-Poor and HEF Beneficiaries, 2015 Source: World Bank, 2016. 23 Table 1.5 HEF Member Visits to Facilities by Frequency and Percent, 2004-2013 Department Frequency Percent (%) Health Center 1,651,627 62.7 Inpatient Department 485,472 18.4 Outpatient Department 498,938 18.9 Total 2,636,037 100 Source: Annear et al., 2016. There are concurrent schemes and programs that further augment the effect of HEF, especially by improving quality of care or further reducing access barriers. It is indeed difficult to attribute all changes to the HEF system alone, given that there are other coexisting schemes with similar aims of increasing the access and utilization of public health services among the poor, such as Integrated Social Health Protection Scheme (iSHPS), Community-Based Health Insurance (CBHI) and Vouchers. Many of these, though, work closely with HEF and leverage the impact of the program. In comparison to districts with only HEF scheme in place (13-40 percent), those that had implemented iSHPS showed higher proportion of HEF beneficiaries who sought care at a public facility (56 percent) and thus decreased direct costs (Table 1.6) (Jacobs, 2017 #29). This points towards the effectiveness of the supply-side quality improvement components embedded in the iSHPS planning, such as pay-for-performance and improved governance structure (Jacobs, 2017 #29). A significant impact on increased public health services utilization and decreased OOP spending was documented when HEF beneficiaries also had vouchers for certain services (Ensor, 2017 #30). The emerging evidence indicates that the concurrent impacts of different schemes have played a role in the overall improvement of health systems that may have not only affected public health facility visits among the poor, but also toward non-poor and private sectors to some extent (Ensor, 2017 #30). Table 1.6 Care Seeking at First Provider, 2014 HoHEF CHEF iSHPS Sought care at N (%) N (%) N(%) Health Center 34 (8.3) 130 (29.0) 559 (48.7) Public Hospital 21 (5.1) 47 (10.5) 80 (7.0) Private Facility 209 (50.8) 161 (36.0) 337 (29.3) Non-Medical* 147 (35.8) 110 (24.6) 172 (15.0) Total 55 (13.4) 177 (39.5) 639 (55.7) Source: Jacobs et al., 2018. HoHEF: HEF-covered hospitals only; CHEF: HEF-covered Health Centers and Hospitals (comprehensive HEF). *Non-qualified informal providers (drug shops, traditional healers, market vendors). Several challenges still exist, often arising from systematic factors that contribute to under-utilization among the poor. Firstly, several studies have identified distance as one of the fundamental barriers, especially for OPD visits (Figure 1.7 and 1.8) (Annear, 2016 #33;Bank, 2016 #28). Between July 2016 and March 2017, when the post-ID process was temporarily unavailable, as well as no transport allowances given, the utilization rate among HEF beneficiaries significantly decreased, even in reproductive, maternal and newborn health services that usually comprise the majority of the HEF patient visits (PSL, 2017 #52). Secondly, low awareness of the entitlement among HEF beneficiaries, or the weak post-ID process, has been identified as another problem preventing the poor’s access to public health facilities 24 (Ros, 2015 #27). Almost half of HEF beneficiaries indicated lack of knowledge of their benefits in a survey conducted in 2015, although HEF coverage had become nearly nation-wide (Bank, 2016 #28). Another study postulated that there may be a delay of 1-2 years for the beneficiaries to become familiarized with the entitlement (Annear, 2016 #33). Moreover, some populations with similar poverty level as HEF beneficiaries may have not been identified during the pre-ID process, thereby not being able to access public health facilities (Bank, 2016 #28). Figure 1.7 Distance Travelled to Facility by HEF Beneficiaries (% of visits) Source: Annear et al., 2016 Figure 1.8 HEF Beneficiary-Reported Reasons for Lack of Utilization of Public Health Facilities, 2015 There are notably several supply-side barriers identified through surveys and interviews with HEF beneficiaries and public facility staff that still need to be addressed. The perceived quality of care at public facilities continues to be lower than private facilities, as patients often rated lower staff skills and 25 attitudes for the former (Figure 1.9) (Bank, 2016 #28;Ros, 2015 #27). The issues around competency of providers and provision of medicines seem to be especially prominent for patients with noncommunicable diseases (NCD) (Bigdeli, 2016 #31;Jacobs, 2016 #37). Figure 1.9 HEF Beneficiaries' Recommendations for Public Providers, 2015 Source: World Bank, 2016. Several studies suggest that the HEF system has made access to health care more affordable. Interviews conducted in 2013 found that deliveries at public facilities reduced spending for HEF beneficiaries especially if they were more severe cases (Ros, 2015 #27). In 2013-2014, approximately 70 percent of HEF beneficiaries indicated on a survey that they had chosen to seek care at a private facility, incurring higher OOP expenditures than others who sought care at a public facility (Jacobs, 2017 #29). Reinforcing this finding, the HEF utilization survey 2015 revealed that HEF beneficiaries had a four-fold increase in spending when they did not use public health services and the near-poor had a six-fold higher spending when not identified as eligible for HEF benefits (Figure 1.10) (Bank, 2016 #28). Therefore, the findings consistently report a reduction in OOP spending among HEF beneficiaries who visit public health facilities. Figure 1.10 Medical Expenses and Monthly Income by HEF Beneficiaries and Near-Poor Source: World Bank, 2016. 26 Despite the overall reduction in OOP spending, concerns regarding unresolved effects of indirect costs still persist. Interviews in 2013 found that some HEF beneficiaries used loans or spent additional amounts to buy preferred food, medicine, and transportation, and those who could not address these additional indirect spending chose not to use the HEF entitlement altogether (Ros, 2015 #27). There are mixed responses towards informal payment for providers at a public facility, as some reported reduction while others reported in-cash or in-kind payments still being given to public facility providers (Ros, 2015 #27). In addition, having an NCD had a significant impact on the likelihood of catastrophic expenses, based on a multivariate logistic regression analysis (Table 1.7) (Jacobs, 2016 #36). Table 1.7 Odds Ratios of Associations between Household with a Chronic Patient and Catastrophic Health Expenditure Household member with a chronic disease 40% threshold 10% threshold None 1.00 1.00 At least one 5.04* 13.85* Source: Jacobs et al., 2016a. *Significant at 1 percent (p<0.01). As access and affordability improved, it was also evident that more poor populations, based on the IDPoor identification, had utilized public health facilities to an increasing extent over the years, thereby reducing OOP spending among the poorest Cambodians. Since the HEF scheme focuses on the poor, such effects correlate to the level of poverty of the household (Ensor, 2017 #30;Ros, 2015 #27). Within the HEF beneficiary populations, there exist factors that differentiate the level of care and OOP saving to which patients are entitled. So far, the analyses show that HEF has primarily been used by patients requiring maternal and child health services (Figure 1.11) (Annear, 2016 #33). While this itself is good progress and corroborates Cambodia’s trailblazing performance on MDG 4 and 5, the use of other services has been poor and addressing this may be necessary to address growing health needs of Cambodia, such as for NCDs. In fact, patients with NCD have not been able to benefit from their HEF entitlement, due to the inability of HEF to cover transportation for the frequent outpatient visits that are often required for these patients. The lack of training among providers to properly manage NCD treatment compounds this problem (Jacobs, 2016 #36). Moreover, there are no special services included in the HEF package, which target patients with NCD (Bigdeli, 2016 #31). This is worrisome, given the steadily high and increasing care seeking behavior among the older generation from 2004 to 2014 (87 percent to 97 percent), as well as the two-fold increase in their risk of indebtedness in rural areas compared to urban areas (Table 1.8) (Jacobs, 2016 #36). When these special needs and arrangements are not met under their HEF entitlement, patients reported the use of private facilities (Ros, 2015 #27). Therefore, further refinement of the HEF benefit package and eligibility criteria is worth considering, hand-in-hand with improved service readiness of public health facilities, amidst the rapidly changing burden of disease in Cambodia. 27 Figure 1.11 Distribution of HEF patients by Age at Admission, 2000-2012 Source: Annear et al., 2016. Table 1.8 Care Seeking Behavior by Age, and Indebtedness by Age and Region (Rural, Urban) (%), 2004-2014 2004 2009 2014 Care seeking (%) Individual ≥ 60 years 86.7 92.2 97.3 Individual < 60 years 90.8 91.3 98.2 Indebted for paying health care costs (%) Households without older people 5.0 4.0 2.5 Households with older people 4.8 3.3 2.2 Rural 5.3 3.6 2.5 Urban 2.0 2.0 1.0 Source: Jacobs et al., 2016a. Based on the above findings, there are several persisting challenges that have been identified by the existing body of evidence around the HEF system which suggest where future efforts at improving the program may be directed. • The competency and attitudes of providers at public facilities have not yet matched the expectation of HEF beneficiaries, which has led to underutilization of public health facilities. • Beneficiary identification process and beneficiaries’ awareness of their entitlement have a significant impact on access to public health services, and remain areas for further improvement. • Despite the overall reduction in OOP spending, some concerns were raised regarding the unresolved effects of indirect costs, which deterred HEF beneficiaries from seeking care at public health facilities. • Patients with NCD have higher risk of experiencing catastrophic expenses; however these patients have not been able to benefit from their HEF entitlement due to the inability of HEF to 28 cover indirect costs for their visits and the lack of competency of providers to properly manage NCD treatments. Service Delivery Grants In 2008, the Cambodian Ministry of Health (MoH, #50) established special operating agencies (SOAs), either based in a provincial referral hospital (PRH) or in an operational district (OD) with the aim to increase utilization and quality of care in identified locations. This was part of an initiative to pay performance-related grants, or Service Delivery Grants (SDG) under an internal contract between MOH/PHD and the SOA. As per the legal provisions in Cambodia, SOAs are designated organizational units of a Ministry with service delivery functions, which are granted some additional delegation of managerial authority and flexibility, jointly by the parent Ministry and MEF, under sub-decree and other policy issuances, in return for stronger accountability for performance. The arrangement required that SOA staff collectively and individually sign contracts, which set annual performance targets, and achievement of these targets triggered payments of SDGs. The SOA arrangement had several goals. Firstly, the main priority was to help facilities improve service quality, by means of providing additional funds to fill monetary gaps and issue bonus payments to individual staff when targets were achieved or surpassed (up to 20 percent of funds). This would increase output in public facilities, allow more staff to be employed as per need, and reduce stock-outs of important drugs and supplies. Secondly, it was anticipated that the subnational entities improve their management and accountability structure, by allowing more autonomy to the SOA designated organizations. Lastly, the monitoring procedure was intended to improve the responsiveness and performance at health facilities. Based upon the implementation experience until 2015, the SOA and SDG system underwent a major revision with the commencement of the Health Equity and Quality Improvement Project (H-EQIP) as described earlier in this chapter. H-EQIP aims to use performance-based payments under SDGs with much stronger links to performance and in particular focuses on quality of services. This is combined with fixed grants to health facilities, in addition to further streamlining the funds flow and reporting arrangements. The new fixed lump sum grants form part of the SDG system through a joint Prakas issued by the Ministry of Economy and Finance (MEF) and MOH, and are intended to complement the facilities’ operational budget. The fixed grants initially added up to an additional US$6.8 million for 2016, and then increased to US$9 million per year from 2017. In the redesigned system, the payment of SDGs to HCs and hospitals is more closely linked to performance in the delivery of basic and comprehensive packages of services, such as critical reproductive, maternal, neonatal, child, and adolescent health services. The MOH SDG manual outlines the intended use for the fixed lump sum grant and performance-based grants (MoH, 2016 #50): • Fixed Lump sum Grants are intended as a complement to the facility operational budget, to manage and implement direct spending for the purpose of promoting quality and equity in patient care. • Performance-based Grants are intended to reward health facilities for quality performance and to reward OD and PHD offices, and in particular certified assessors for conducting quality ex-ante assessment. Up to 80 percent of Performance-based Grants can be spent for staff incentives. At least 20 percent of Performance-based Grants are eligible for any other SDG-eligible expenditures. 29 Despite this progress in the SDG system, rigorous evaluation of SOA/SDG has been difficult. Notably, there was no baseline and no control or counterfactual that could be assessed in comparison to SOA performance. SOA ODs have a distinctive history of higher levels of funding on average and higher total staff remuneration on average, which makes it difficult to attribute changes in performance to the contracting design features. Moreover, there have been implementation problems with the performance component of the SDG and design problems with the grant allocation formula, leading to a series of revisions in design and implementation of design each year. Thus, it would be difficult to attribute evaluation findings to design features, as distinct from implementation issues. However, now that the new objective performance assessment system, National Quality Enhancement Monitoring Program (MQEMP) linked to the payment of performance-based SDGs has commenced in May 2017, new administrative data is expected to become available for future analytical work on this new modality. So far, considerable observation-based and anecdotal information has started becoming available on the use of these fixed lump sum grants, which have been disbursed since July 2016. To illustrate one such example, in March 2017, the H-EQIP cofinancing partners undertook the second Implementation Support Mission (ISM) in four provinces (Kampong Cham, Kampong Chhnang, Kampong Speu and Kampot), with the objective to review the overall progress and operational management of the project at the point of service delivery, including the implementation of the fixed lump sum SDGs (WorldBank, 2017 #49). It was observed that most health facilities had used the fixed lump sum grant according to the MOH guidelines and eligible expenditure list. In addition to those that were captured during the ISM visit, a few additional uses of fixed lump sum grant were noted at subsequent site visits, which included: hygiene and infection control supplies (for example, alcohol, hand-washing supplies, dustbins), medical equipment (for example, thermometers, fetal heart sound monitors or Dopplers), stethoscopes, tension meters, toilet bowls (in existing toilet stalls only), fans, and repair of incinerators. Some anecdotal experiences were shared by facility directors and patients at the health facilities, in relation to the impact of fixed lump sum grant on service quality improvement, and have been included as boxes in this section in view of their highly innovative and inspiring use of SDG funds (Box 1-1 and Box 1-2). 30 Box 1-1 Provision of Neonatal Kit at Tray Koh Health Center, Kampong Bay OD, Kampot Province The director and staff at the Tray Koh health center shared a story about the newborn kit that every new mother received upon her delivery in the health center, which included baby mittens, blankets, clothes, instructions on nutrition and breastfeeding. Following this initiative, the health center has seen increased utilization patterns for antenatal care and postnatal care by patients who were perceived to have appreciated the attention of health care staff. The practice of giving out and explaining the contents of the kits to the mother strengthened personal relationships between the midwives and patients as it allowed an additional counselling opportunity for midwives, which included instructions on breastfeeding and its implications on nutrition. Moreover, midwives had an opportunity to take ownership for this innovative idea that could be supported by their health center. Newborn kits for boys and girls – an innovation by Tray Koh HC (World Bank, 2017) Box 1-2 Use of Fetal Heart Sound Monitoring Device (Doppler) at Skun Health Center, Cheung Prey OD, Kampong Cham Province This health center in Skun (and many others like this one) has invested its fixed lump sum SDG resources in what the team has been wanting to do for a long time - to buy a fetal heart sound monitoring device, which costs US$150 to US$160. Pregnant mothers really like being able to hear the heart sounds of their unborn child, which comes out loud and clear using this device; this makes them look forward to their Antenatal Care (ANC) visits. The device has impacted positively on the interpersonal relationship between HC staff and patients, and also provided greater counselling opportunities. More importantly from a clinical perspective, the device allows the facility to monitor fetal well-being and detect any fetal distress when a woman is in labor at the health center. Timely detection of fetal distress can help save lives - both of the mother and of the child - through timely remedial action at the health center and also through timely referral. Fetal heart sound monitor and blood pressure monitor at the Skun HC (World Bank, 2017) 31 Findings from the field visits furthermore indicate that there is a need for more information on SDGs for the facility staff’s understanding of the use of funds in the early years of this modality , particularly related to the eligible expenditures (for example, renovations and civil works) and procurement procedures, including limits of spending using petty cash and other sources of funds. The mission also noted that while facilities had a good understanding of bookkeeping due to the nation-wide Finance Management (FM) training and effective coaching by PHDs and ODs, the understanding of accounting and financial management instructions given to the facilities varied. Providing such clarity will further build facilities’ confidence in planning and budgeting of these novel funds. As indicated earlier, the published literature primarily pertains to the previous configuration of the SOA and SDG system, prior to its revision in H-EQIP. It Indicates that the incentives targeting health workers (HW) have had positive impacts on recruiting or retaining more HWs in the public sector who may have otherwise gone to work for NGOs or the private sector (Witter, 2016 #38;Ensor, 2016 #45). Another set of interviews with SOA officials indicated that the amount of incentives given to primary care providers either diverted them to the private sector or kept them in public facilities, indicating whether the incentive offset the monetary benefit of working in the private sector was a determining factor for them to stay in public facilities (Khim, 2013 #47). Some expressed the benefits of the SOA managers’ monthly monitoring, as it affected staff incentives and worked to reduce misrepresentation of service outputs (Khim, 2013 #47). On the other hand, it was also repeatedly reported that the monitoring routines by the central and provincial teams were not kept regular, mainly due to low incentives (Vong, 2015 #51;Khim, 2013 #47). Moreover, the very nature of internal contracting in which the purchaser, provider, and monitoring roles are often unclear came with its own challenges in terms of accurate reporting of data (Khim, 2017 #35). Some indicated a problem in the centralized process of choosing and setting targets, which may not be realistic at the level of each facility (Khim, 2013 #47;Vong, 2015 #51). Moreover, inadequate competency, unequal distribution of HWs among rural and urban areas, insufficient remuneration have been documented (Khim, 2017 #35); (Witter, 2016 #38). Many of these emerging challenges were duly addressed in the revised SDG system implemented under H-EQIP. For example, the importance of performance linkages was fully restored, and supervision quality is also now undertaken in a structured manner as well as measured and incentivized in the revised system. SOAs in Cambodia stand out in their positive effects on local autonomy and ownership. The SOA arrangement in Cambodia was in a different league from most other contracting-in and contracting-out models operational in the previous decade that had engaged international organizations, which incurred higher costs and thereby were of questionable sustainability (Witter et al., 2016). In addition to the increased MOH discretion by means of internal contracting, the SOA implementation also required substantial engagement of the provincial and district entities in order to be selected, implemented, and monitored effectively (Vong, 2015 #51). In fact, a set of interviews with administrative officials at central, provincial and district level reveals that the SOA operations have given them an increased sense of ownership (Khim, 2013 #47). Another study expressed that the SOA arrangement was a key instrument in giving the MOH control to manage the Human Resources for Health (Witter, 2016 #38). Health managers also conveyed their increased ownership, allowing for creative measures taken at their discretion (Vong et al., 2015). The development of clearly defined performance indicators between different levels of government bodies created accountability structures that improved performance (Khim, 2017 #35;Vong, 2015 #51). 32 In one analysis, Kampong Cham was shown to have the highest productivity in service delivery given that approximately half of its ODs were early adopters of SOAs compared to other provinces included in the study that did not have SOAs (Ensor, 2016 #45). It was generally accepted across interviews with the SOA officials that the SOA arrangement facilitated better management at health facilities (Khim, 2013 #47). However, management and supervision capacity did need further improvements. The SOA system had been established with detailed manuals and institutional arrangements, and there was wide ownership and acceptance of the model. It translated into improved management, better provider behavior, and strengthened health service delivery. However, several challenges remained, particularly in relation to the SDG scheme’s complexity in monitoring and disbursement process. Despite the greater administrative autonomy and increased local capacity compared to previous contracting models, evidence elucidates weak management capacity in certain areas. Firstly, there was general confusion around clear roles of contracting bodies (Khim, 2017 #35). Secondly, although the SOAs received approximately 95 percent of their funding from the government budget and SDGs, some SOA directors raised concerns as to how to accurately allocate the SDG funding and expressed that they were given insufficient instructions to fulfill their duties (Khim, 2013 #47). Delays and inadequate amount associated with the SDG disbursement were commonly mentioned in interviews, which had a negative impact on executing their work plan (Khim, 2017 #35;Khim, 2013 #47). One source noted the complexity often required with performance-based financing mechanisms at the local level, indicating the need for continued support for capacity building during the HSSP2 phase (Witter, 2016 #38). The redesigned SDGs attempt to address the known concerns with the SDG system, and have improved linkages with performance, a robust assessment system, and independent cross-verification of assessed performance. The impact evaluation of these redesigned SDGs is expected to yield valuable information on the redesigned system, and will inform decisions on the continuation of these grants, as well as areas where the design and implementation of SDGs can be further improved. 33 References Annear, Peter Leslie, Keovathanak Khim, Por Ir, Ellen Moscoe, Tapley Jordanwood, and Thomas Bossert. 2016. National coverage and health service utilization by Health Equity Fund members, 2004- 2015. In ADRA Research Report. Antunes, Adélio Fernandes, Bart Jacobs, Richard de Groot, Kouland Thin, Piya Hanvoravongchai, and Steffen Flessa. 2018. "Equality in financial access to healthcare in Cambodia from 2004 to 2014." Health Policy and Planning:1-14. doi: https://doi.org/10.1093/heapol/czy073. Bigdeli, Maryam, Bart Jacobs, Chean Rithy Men, Kristine Nilsen, Wim Van Damme, and Bruno Dujardin. 2016. "Access to Treatment for Diabetes and Hypertension in Rural Cambodia: Performance of Existing Social Health Protection Schemes." PLoS One 11 (1):e0146147. doi: 10.1371/journal.pone.0146147. Ensor, Tim, Chhim Chhun, Ton Kimsun, Barbara McPake, and Ijeoma Edoka. 2017. "Impact of health financing policies in Cambodia: A 20 year experience." Social Science & Medicine 177:118-126. doi: 10.1016/j.socscimed.2017.01.034. Ensor, Tim, Sovannarith So, and Sophie Witte. 2016. "Exploring the influence of context and policy on health district productivity in Cambodia." Cost Effectiveness and Resource Allocation 14:1. doi: 10.1186/s12962-016-0051-6. GIZ. 2014. Out-of-Pocket and Catastrophic Expenditure on Health in Cambodia: Cambodian Socio- Economic Surveys 2004, 2007 & 2009 Analysis. Cambodia: Deutsche Gesellschaft für Internationale Zusammenarbeit. Jacobs, Bart, Ashish Bajracharya, Jyotirmoy Saha, Chhorvann Chhea, Ben Bellows, Steffen Flessa, and Adélio Fernandes Antunes. 2018. "Making free public healthcare attractive: optimizing health equity funds in Cambodia." International Journal for Equity in Health 17 (88). Jacobs, Bart, Richard de Groot, and Adélio Fernandes Antunes. 2016. "Financial access to health care for older people in Cambodia: 10-year trends (2004-14) and determinants of catastrophic health expenses." International Journal for Equity in Health 15:94. doi: 10.1186/s12939-016-0383-z. Jacobs, Bart, Peter Hill, Maryam Bigdeli, and Cheanrithy Men. 2016. "Managing non-communicable diseases at health district level in Cambodia: a systems analysis and suggestions for improvement." BMC Health Services Research 16:32. doi: 10.1186/s12913-016-1286-9. Khim, Keovathanak, and Peter Leslie Annear. 2013. "Strengthening district health service management and delivery through internal contracting: lessons from pilot projects in Cambodia." Social Science & Medicine 96:241-9. doi: 10.1016/j.socscimed.2013.02.029. Khim, Keovathanak, Por Ir, and Peter Leslie Annear. 2017. "Factors Driving Changes in the Design, Implementation, and Scaling-Up of the Contracting of Health Services in Rural Cambodia, 1997- 2015." Health Systems & Reform 3 (2):105-116. doi: 10.1080/23288604.2017.1291217. MoH. 2016. Service Delivery Grants Operational Manual. edited by Ministry of Health. Phnom Penh, Cambodia. PSL. 2017. Out of reach? The critical barrier of transporation to access reproductive, maternal and newborn health services for vulnerable women in northeast Cambodia (Draft Policy Brief). Cambodia. Ros, Bandeth, Suzanne Fustukian, and Barbara McPake. 2015. Health care seeking behaviour and impact of health financing policy on household financial protection in post-conflict Cambodia: A life history approach. In Working Paper. Liverpool, UK: ReBUILD RPC. Vong, Sreytouch, Joanna Raven, and David Newlands. 2015. Understanding contracting in Cambodia: Findings from interviews with key informants and health service managers and providers. In Research Report. Liverpool, UK: ReBUILD RPC. 34 Witter, Sophie, Maria Paola Bertone, Yotamu Chirwa, Justine Namakula, Sovannarith So, and Haja R. Wurie. 2016. "Evolution of policies on human resources for health: opportunities and constraints in four post-conflict and post-crisis settings." Conflict and Health 10:31. doi: 10.1186/s13031-016- 0099-0. World Bank. 2013. Cambodia's Rural Health Markets. Cambodia. World Bank. 2016. Utilization and Impact of Health Equity Funds: Improving Entitled Benefits Uptake by the Poor. Phnom Penh, Cambodia. World Bank. 2017. Health Equity and Quality Improvement Project: Second Implementation Support Mission (March 20-24, 2017) - Draft Aide Memoire. Cambodia. World Bank. 2017. "World Development Indicators." http://datatopics.worldbank.org/world- development-indicators/. 35 2. Impact Evaluation and Methodology 2.1 Intervention: Service Delivery Grants The purpose of the SDG Impact Evaluation is to measure the impact – outputs and short-term outcomes – of the implementation of performance-based SDGs. As outlined in Chapter 1, the SDG program is a component of H-EQIP, whose primary objective is to improve quality of health care services and financial protection for vulnerable groups. This section describes in more detail each component of performance-based SDGs that comprises the program’s overall objective. The performance-based SDGs are an additional layer of payments provided, based on assessed and verified performance, over and above the fixed lump sum SDGs. Quality health service delivery in the H-EQIP operational definition refers to infrastructure, managerial capacity, and clinical competency, achieved through the implementation of new performance-linked SDGs, which is in effect at three levels: PHD and OD Offices, Referral Hospitals and Health Centers (Figure 2.1). The performance of these entities is measured by National Quality Enhancement Monitoring Tools (NQEMT), which are applied quarterly by certified assessors from the OD and PHD offices. These payments supplement the revised fixed lump sum SDGs that complement health facilities’ existing operating budget. Figure 2.1. Components of SDG Program for Each of Three Types of Health Entities Directly Affected by the SDG Program PHD/OD Office Referral Hospital Health Center Do quarterly assessments & Balanced scorecard (CPA) Balaced scorecard (MPA) follow up targets Provide coaching & Indice tool Indice tool pre-/in-service trainings Maintain assessment Staff performance evaluation Staff performance evaluation equipment Monthly meetings Clinical vignettes Clinical vignettes Management & supply of Patient surveys Patient surveys vaccines/medicines The ex-post verification is initially carried out by an independent verification agency financed by KfW, and in early 2019, this function will be systematically handed over to the Payment Certification Agency (PCA). Once assessed and verified, the MOH releases the performance-based SDG payments directly to the health facilities, which provide a flexible source of funding that can be used as per the provisions of the SDG manual. Provincial Health Department and Operational District Level SDGs at this level aim to strengthen the management of Operational Districts (ODs) and Provincial Health Departments (PHDs). Performance of ODs and PHDs is measured every quarter against their self- reported activities on a scorecard measuring key supervisory processes and health system outputs. 36 These include: (a) timely completion of quality checklists for health facilities in their jurisdiction, (b) contribution to capacity building activities for in-service and pre-service training, (c) drug stock-outs in health facilities, human resources availability, (d) submission of Health Management Information System (HMIS) reports, and (e) quarterly review meetings and system functionality. Funds received by these supervisory levels are predominantly intended to meet their supervisory travel costs (travel for assessment and coaching is also supported through this mechanism) as well as for performance-based incentives. Hospital Level: CPA-1, CPA-2 and CPA-3 This subcomponent aims to incentivize improvements in quality of care at the secondary level, improve performance in capacity building activities for in-service and pre-service candidates, and promote utilization of services by HEF beneficiaries. Quantitative tools are used to assess their performance on structure, process, and outcomes. Structural measures comprise the context in which care is delivered, including infrastructure, staff, financing and equipment. Process measures include the technical and interpersonal process and actions that make up health care as reflected in the transactions between patients and providers and staff throughout the delivery of health care. Outcomes refer to the effects of health care on the status of patients and populations and will be considered to be a result of inputs and processes of care. Up to 80 percent of the Performance-Based Service Delivery Grant (PBG) can be spent for staff incentives; at least 20 percent of the PBG is eligible for any other eligible expenditures from those laid out in the H-EQIP SDG manual, such as small civil works, equipment, drugs and operational costs. The NQEMT for the hospital level includes an assessment of: the hospital balanced scorecard (which comes in three different types, as per the type of hospital: CPA-1, CPA-2, CPA-3), indice tool, individual performance evaluation for hospital staff, selected and adjusted L2 clinical vignettes, neonatalie observational checklist, and content of care traces extracted from community client satisfaction surveys. Health Center Level: MPA SDGs to health centers help support the delivery of Minimum Package of Activities for health centers. The quality of service delivery is systematically measured quarterly, using standardized assessment tools. The NQEMT for health center level include: the health center balanced scorecard, indice tool, individual performance evaluation for health center staff, selected and adjusted L2 clinical vignettes, neonatalie observational checklist, and content of care traces extracted from community client satisfaction surveys. Eligible categories of expenditure for PBG at health center level are the same as hospital level and specified in the H-EQIP SDG manual. Since 2018, additional performance-based resources are available for the special needs of remote areas and for health centers with a higher proportion of indigenous population. 2.2 Theory of Change The quality of health service delivery is expected to be positively reinforced by the quarterly quality assessment using NQEMT, leading to the regular disbursement and use of performance-based SDG (Figure 2.2). The baseline survey was conducted at the level of OD office, health center and household to measure the existing quality of health service delivery prior to the implementation of the intervention. (In comparison to the level of intervention as outlined in the previous section, PHD office and Referral Hospitals were not chosen entities to measure impact.) The SDG results chain is outlined in Annex A. 37 Figure 2.2 Theory of Change and Short-Term Outcomes at the Level of OD Office, Health Center and Household Expected Outcomes Supply Demand I. Improved monitoring and evaluation D. Improved perceptions of quality of J. Improved managerial capacity and financial autonomy care by citizens K. Improved perceptions of quality of care by health worker E. Improved financial protection L. Increased motivation and morale of health worker F. Increased utilization of public health M. Improved health worker clinical competency facilities N. Increased availability of health worker at health facilities O. Improved use and functionality of available infrastructure P. Increased availability of necessary supplies and consumables OD office | Health Center | Health Worker | Exit Exit | Household SDG Impact Evaluation Survey Instruments Operational Districts The intervention will provide ODs with regular, verified and detailed data on the quality of care at health centers. This will raise awareness of performance issues, which interact with incentives for the OD to perform essential management and supervision tasks. This is expected to: • Increase OD awareness of performance issues and local constraints, including staffing and performance • Focus OD efforts on improving quality at lower-performing facilities • Lead ODs to increased, better informed and more regular/routine supervision of clinics The intervention directly encourages ODs to improve their management of clinics. It also creates indirect incentives, as health centers may more actively demand OD support so that they can improve their quality and hence obtain a larger bonus payment. As a result, the intervention may: • Improve and change OD management in line with the incentives under NQEMP • Improve targeting of supervision and resources to facilities that are most in need • Raise the overall performance of the OD, for example, with regard to training, stock-outs, HMIS reporting and indicators measured by the NQEM assessment tool 38 • Increase the motivation and satisfaction of OD and center staff Health Centers The intervention provides health centers with a range of incentives, as well as the requisite financing and autonomy that are expected to improve performance and motivation. For instance, the intervention is expected to: • Increase the productivity of center staff, as measured by improved performance on measured indicators, lower absenteeism and improved clinical quality • Improve the allocation of resources to deficient areas of care, and facility management, so as to improve quality scores • Increase health centers’ awareness of performance problems, because of increased supervision by the OD and regular data collection/reporting that is tied to bonus payments • Focus health centers’ attention on clients and improve attitudes toward clients. • Raise the motivation of staff, thus improving retention and lowering the rate of workers engaging in “dual practice”. Households Although the intervention does not directly target households, it is expected to indirectly impact households in the respective districts and catchment areas through the improved performance of the health facilities. Increases in quality and quantity of health care delivery are expected to increase the demand for such services. The objective of the study at the household level is therefore to measure health-seeking behavior at baseline, both amongst the general population, and in the HEF beneficiaries (as well as Community-Based Health Insurance (CBHI) and voucher beneficiaries). It is of particular interest to document changes in demand toward the public health sector, and especially, if there is a decrease in the utilization of informal health structures. In addition, self-reported perceptions of quality by clients are also being measured. Finally, the intervention at the OD and health center levels is expected to impact households in various ways, for example: • Increase the quality of care received, which in turn could affect perceptions of quality • Increase demand for health care services at the health center level, thus increasing care seeking and reducing the utilization rate for informal providers, private providers and higher-level public facilities. • Improve equity as HEF households may disproportionally benefit because of the added demand- side incentives to use health centers. Measurable Impact As summarized in this section, the following impact may be more likely to be detected in the impact evaluation period areas such as: • Management of ODs and health centers (for example, M&E, financial autonomy, managerial capacity) • Satisfaction of health worker and OD management staff • Short-term perception of quality of care by health workers and patients • Health worker clinical competency and availability • Infrastructure, supplies and consumables • Utilization (overall and public vs. private) • Out-of-pocket payments 39 However, the following areas of impact are less likely to be detected within the impact evaluation timeframe: • Health outcomes • Long-term demand and access at public health facilities • Sustainable equity gap reduction in financial protection and quality of care received • Nation-wide quality of care at public health facilities The SDG results chain is outlined in Annex A. 2.3 Impact Evaluation Design Research Questions Based on the theory of change, the overall goal of the evaluation is to assess the impact of PBGs through the implementation of the National Quality Enhancement Monitoring Program (NQEMP), in parallel with synergistic effects from fixed lump sum SDG (Box 2-1). The evaluation has three primary research questions as outlined below. For each question, other valid sources of information are listed that will complement the findings from the impact evaluation quantitative surveys to reveal more plausible explanations. 4. Does the National Quality Enhancement Monitoring Program (NQEMP) at the levels of the operational district and health centers have an average impact? The impact of the intervention is likely to evolve over time, as the procedures are firmly put in place and all stakeholders learn and understand how the intervention works. 5. Are the impacts of the intervention heterogeneous with regard to differences in the local context? This includes variations across ODs (for example, SOA vs non-SOA; ODs with low vs high-quality health centers; ODs with low or high-variability in center quality); health centers (low vs high quality); health care workers; and populations (for example, HEF vs non-HEF). 6. Does the intervention affect equity? In particular, does the intervention narrow the gap in the various outcomes across health centers (for example, by raising the performance of the initially lower-quality centers) and populations (especially HEF and non-HEF)? Measurement methods: Impact evaluation quantitative surveys and qualitative interviews, NQEMP quality scores, Implementation Support Mission aide-memoirs, SDG Process documentation, Cambodia Socio Economic Survey (CSES), Cambodia Demographic Health Surveys (CDHS), Cambodia Health Management Information System (HMIS), and Patient Management and Registration System (PMRS). 40 Box 2-1 What is Impact Evaluation? Impact Evaluation is an analytical process that aims to identify the attributes and their degrees of effect for the changes in the intended outcomes due to the implementation of a program or policy. Simply put, it is an assessment of a causal relationship between the content, or inputs, of a program or policy and its outcomes of interest. The term outcomes is to be distinguished from the immediate outputs of a program or policy, which has been more commonly measured in the conventional assessments for a program or policy. Impact Evaluation is a complementary approach to Monitoring and Evaluation (M&E) whose primary objective is to monitor and measure target indicators over time, but not necessarily to reveal the reasons for success or failure of a program or policy – which is the central objective of Impact Evaluation. In order to identify the causes of certain outcomes, it is therefore essential to have an appropriate comparison group, or counterfactual, which did not receive the same implementation of a program or policy. There are various methods that could be employed in an impact evaluation, according to the operational characteristics of a program or policy – namely, the available resources, eligibility criteria for selecting beneficiaries, and timing for program implementation. A prospective impact evaluation could define the intended results through a theory of change as the program or policy is being designed, identify the comparison group prior to the implementation and collect baseline data. The lack or insufficiency of such information in a retrospective impact evaluation would often lead to weak or biased results and may fail to reveal the causal relationship between a program or policy and the outcomes of interest. Even with a well-designed impact evaluation, it is important to complement the findings with other sources of information around the same program or policy, in order to ensure technical quality and policy relevance (Gertler et al., 2016). The main purpose of Impact Evaluation is contingent on the importance of evidence-informed policy making. The demand for results is increasing to ensure the accountability and transparency of publicly- funded programs and policy. A carefully designed Impact Evaluation not only assesses the impact but also helps reaching the intended outcomes. Therefore, Impact Evaluation can be a useful tool in guiding the decision-making process for the program or policy in question. These can be categorized in two ways: The results can reveal the effectiveness of a given program or policy, and the causal relationship between specific inputs and outcomes. Having an appropriate counterfactual would ensure internal validity. The results can also reveal the effectiveness of a given program or policy in comparison to an alternative. This allows variations and innovations in a program or policy to determine the most effective or cost-effective version. In addition, the results from an impact evaluation can provide additional knowledge to a global platform if given sufficient external validity, in such a way that other countries could adopt a similar program or policy in their own context. Impact Evaluation, therefore, is becoming a global public good. A recent publication reveals the emerging trend that two-thirds of total development impact evaluations were published between 2010 and 2015. Health and nutrition, education, and social protection have been the most commonly evaluated sectors since 2000, although other sectors, such as energy, transportation and ICT, which were severely underrepresented prior to mid-2000 were extensively evaluated during the 2010-2015 timeframe (Sabet, 2018 #54). Bibliography DIME. 2015. What is IE. edited by WorldBank. Gertler, Paul J., Sebastian Martinez, Patrick Premand, Laura B. Rawlings, and Christel M. J. Vermeersch. 2016. Impact Evaluation in Practice. Second ed. Washington, DC: The World Bank. Sabet, Shayda Mae, and Annette N. Brown. 2018. "Is impact evaluation still on the rise? The new trends in 2010-2015." Journal of Development Effectiveness 10 (3):291-304. SIEF. 2017. Strategic Impact Evaluation Fund. edited by WorldBank. 41 Service Delivery Grant Treatment Assignment in Practice The treatment assignment for rolling out the NQEMP is at the OD level and the intervention is rolled out in three phases. In the first phase, the intervention was purposefully rolled out to SOA districts, as those districts were expected to be most “ready” to implement the intervention. This initial assignment was not randomized. The assignment of the remaining non-SOA districts was randomized. Half of these districts were assigned to the rollout in phase 2 (these are called “treatment” districts) while the other half w ere assigned to receive the intervention in phase 3 (these are called “control” districts) (Tables 2.1 and 2.2). This was done in several steps: 1. Within a province, four strata of ODs were created: • District average of the facility score in the Level 2 quality assessment (L2) is above or below the median of the district L2 in the province. • District standard deviation of the facility L2 is above or below the median of the district L2 in the province. 2. Within each province and strata, ODs were randomly assigned to treatment and control status. For example, if there were two ODs in a province that are “high mean and low SD”, one district was randomly assigned to treatment and the other to control. If there were three ODs in this stratum, the third was randomly assigned to either treatment or control. This could only be done where there are a lot of ODs, that is, in Phnom Penh and Kandal provinces. 3. For the remaining ODs, the randomization was done within strata but without accounting for province. Table 2.1 Number of Facilities by Phase, Stratum and Whether in Baseline Low SD High SD All Low mean High mean Low mean High mean In baseline Phase 1 (SOA) 8 12 4 18 42 Phase 2 (treatment) 12 12 18 12 54 Phase 3 (control) 10 10 16 8 44 Subtotal 30 34 38 38 140 Not in baseline Phase 1 (SOA) 54 72 32 108 266 Phase 2 (treatment) 31 45 64 58 198 Phase 3 (control) 69 43 84 48 244 Subtotal 154 160 180 214 708 42 Table 2.2 Number of Facilities in the Baseline Survey by Phase and Province Phase 2 Phase 3 Province Phase 1 (SOA) Total (treatment) (control) Banteay Meanchey 4 4 8 Battambang and Pailin 8 4 12 Kampong Cham 6 2 8 Kampong Chhnang 4 2 6 Kampong Speu 4 4 Kampong Thom 2 4 6 Kampot and Kep 2 4 6 Kandal 4 4 8 Kratie 4 4 Mondul Kiri 2 2 4 Oddar Meanchey 2 2 4 Phnom Penh 6 8 14 Prey Veng 2 6 4 12 Pursat 2 6 8 Ratanakiri 2 2 4 Siemreap 8 8 16 Sihanoukville 2 2 Stung Treng 2 2 4 Svay Rieng 6 2 8 Takeo 8 8 16 Tbong Khmum 4 4 2 10 Total 42 54 44 140 The impact evaluation uses a difference-in-difference design that compares the change between a baseline and follow-up time in districts that received the intervention versus districts that did not receive the intervention. The assumption to identify the intervention’s impact is that non-intervention districts would have experienced similar changes as the intervention districts, if the latter had not received the intervention. That is, the evaluation assumes that the non-intervention districts represent a good counterfactual for the intervention districts. As noted, the intervention was assigned to three groups of districts. The first phase was purposefully reserved for SOA districts, that are likely very different from other ODs. Therefore, the phase 2/3 districts by themselves are not ideal counterfactuals for the phase 1 ODs. However, the counterfactual for comparing phase 2 and phase 3 districts is strengthened by a random assignment of these districts (and hence clinics and households) to the intervention and non-intervention groups. The power calculation of samples is outlined in Annex B and Annex C. The list of provinces, ODs, and health centers is outlined in Annex D. Figure 2.3 shows how the difference-in-difference analysis could be done on the phase 2/3 districts, using baseline, midline and endline data. 43 Figure 2.3 Difference-in-Differences to Evaluate Impact of SDGs at Endline Sample Design The baseline sample was stratified by L2 score (OD and facility), distance from facility (villages) and HEF status (households). For more details on the stratification and randomization process and sampling representativeness, refer to Annex E. 2.4 Baseline Data Collection Sample The SDG baseline survey was conducted from June to July 2016 (supply side) and September to November 2016 (demand side). The total number of interviews completed was: • 140 health facilities with facility assessment and medical record review, 545 health worker interviews, and 1053 patient exit interviews in 140 villages with community interview • 70 OD Director interviews • 140 villages with community interviews • 2,506 households (HEF and non-HEF) with household survey and mother/child survey. Households were eligible for the survey if at least one household member was a woman 15-49 years old who has had at least one pregnancy event (live birth, stillbirth, miscarriage and/or abortion) in the 24 months directly preceding the survey. These numbers deviated from what had been the original plan, as shown in Table 2.3. 44 Table 2.3 Planned and Actual Data Collection Unit of observation Original plan Actual Completed Operational district offices (across 23 provinces) 70 70 69 Health facilities (across 23 provinces) 140 140 140 Health workers (4 per facility) 560 550 546 Patient exit interviews 1,120 1,061 1,053 (4 antenatal patients, 4 under-5-child patient caregivers) Medical record review (4 records per facility) 560 560 560 Households 2,800 2,506 2,474 (20 per facility catchment area: 10 HEF & 10 non-HEF) Community (2 per operational district) 140 140 140 The achieved sample sizes for health workers and patient exit interviews is low because of the low patient flow at some of the health facilities visited and also a few respondent refusals. As for the achieved sample size for households, a few of the villages (those below 200 households) in the sample did not have sufficient number of eligible households to fulfill all sample groups in addition to refusal and absence of a few respondents on the survey days. Survey Instruments The survey instruments are based on those proposed by the Health Results and Innovation Trust Fund (HRITF) Evaluation Toolkit, with the exception of the OD instrument, which was newly developed for this evaluation. The key sections for each instrument are listed in Annex E. 2.5 Ethics Ethical clearances for this evaluation study were duly obtained from the National Ethics Committee for Health Research (NECHR) and participation in the study was voluntary and with informed consent. 45 3. Findings This chapter discusses the analysis of survey responses from seven different data collection instruments, conducted as part of the baseline study, focused on supply as well as demand dimensions of the health system in Cambodia. On the supply side, instruments captured responses from health centers (usually involving the heads of health centers as the respondents, supported by their teams), from health workers, and from supervising agencies at the operational district level. The demand side surveys were implemented at the overall household level, including an additional instrument to capture the maternal and child health dimensions, at the community level, as well as exit surveys of patients using health facilities. Each of the first seven sections below summarizes the key findings and provides a description of the results from each of these seven tools. In order to complement the baseline findings and to provide a more comprehensive view of the health system prior to the H-EQIP, two additional analyses were incorporated using the L2 quality assessment scores and HEF utilization survey results, both of which pertain to a period just prior to the launch of H- EQIP. The L2 quality assessment scores were analyzed to measure any existing differences based on the SOA and HEF status, in the health center performance on clinical vignettes. In theory, being designated as an SOA does entail a different degree of investment and incentives, which would render these facilities different in terms of infrastructure, process and outcomes. Though the L2 dataset by itself is not part of the impact evaluation, it was analyzed in detail for a better understanding of the status quo of health facilities at the time of the baseline. The analysis uses different combinations of SOA and HEF status for additional insights in understanding the baseline variations, and possibly also for interpreting the midline and endline results. 46 3.1 Health Facility Key Findings • On average, about 21 percent of the health facility’s total income per mo nth came from HEF reimbursement; 23 percent of health facilities with HEF income revealed that HEFs were usually or always late. • The average amount of funding received from user fees, HEF, and MOH per month was approximately 11 million Riels, 6 million Riels, and 18 million Riels, respectively (1 million Riel= approximately US$250). • Health centers with lower HEF volume are located closer to higher-level facilities. Eighty percent of referral hospitals are located beyond the 5km radius from the health center. • 99 percent of high-HEF volume facilities reported that they offer delivery services in general, compared to 77 percent of low-HEF volume facilities. • About 58 percent of all facilities reported having had electric power outages in the last seven days. The average duration of absence of electric power in the last seven days was 5.8 hours. • Almost one-fourth of facilities had rainwater (12 percent) and unprotected wells (10 percent) as primary water sources. • Fifty percent of facilities owned a functioning computer, of which 44 percent had access to internet connection. • Overall, 86 percent of health centers had access to working vehicles; 23 percent had an ambulance owned by the facility, while 81 percent had an ambulance owned by OD. The likelihood of having an ambulance owned by the facility was higher at high-HEF volume facilities (41 percent) than at low-HEF volume facilities (7 percent). • 28 percent of respondents chose health facility head or health facility staff as entities with the authority to procure drugs and equipment for the facility. The most selected entities were NGO staff (62 percent) and local government (43 percent). • 33 percent of respondents disagreed or strongly disagreed on having choice over types of services provided at the facility. Twenty-seven percent disagreed or strongly disagreed on having enough authority to obtain the resources for the facility, namely drugs, supplies or funding. There were 140 facilities surveyed in the target areas from 23 provinces and 70 operational districts. The assessment was carried out over a period of 24 days from June 22 to July 15, 2016. These facilities were classified by the actual HEF reimbursement for May 2016, based on the Cambodia HMIS into two equal groups of 70 health facilities with low and high HEF volumes (lower- and upper-half of the median of HEF reimbursement, US$111.30). They were also classified by start of the HEF program into two facility groups as before or since 2015, so correspondingly had two groups of 65 and 75 health facilities. The sample weights of the health facility, which were stratified by average and variability of L2 scores of each OD, are applied in all calculations. According to the HMIS, the average total patients, pregnant women, and under-5 patients were respectively 523, 100, and 140 in the last completed calendar month (or around 18 total patients, four pregnant women, and five under-5 patients per day). About 38 percent and 9 percent of all facilities had more than 500 and 1,000 total patients respectively in the last month. Specifically, 51 percent of facilities had 101-500 total patients, 29 percent had 501-1000 total patients, and 9 percent had more than 1,000 total patients. In less than 20 percent cases, the referral facilities3 were located within a 5km radius from the health center, another 20 percent within 6-10 km, and about 60 percent beyond 11km (Figure 3.1). It is interesting to note that more than 70 percent of health centers with higher HEF claims (greater than or 3 District/provincial referral hospitals or national hospitals 47 equal to median) (henceforth “high-HEF”) are located further than 11km from higher-level facilities, whereas this was 56 percent in the case of those with lower HEF claims than median (henceforth “low - HEF”) (Figure 3.2). Therefore, health centers with lower HEF volume seem to be located relatively closer to higher-level facilities, with 13 percent of them within a radius of less than 1km , compared to 4 percent of high-HEF facilities under 1km. The relatively shorter distance between facilities may act as an enabler for patients to easily access secondary or tertiary health services. Figure 3.1 Distance to the Nearest Higher-Level Health Facility from Health Centers (in percent) Figure 3.2 Distance to the Nearest Higher-Level Facility from High-HEF Volume and Low-HEF Volume Facilities (in percent) There are degrees of financial dependence on HEF, but a number of facilities experienced delays in receiving HEFs. On average, about 21 percent of a health facility’s total income in the last month came 48 from HEF. Furthermore, 11 percent of all health facilities mentioned that their HEF income was more than half of total income. Unfortunately, as of mid-2016, 23 percent of health facilities with HEF income revealed that HEFs were usually or always late. About 34 percent of health facilities in the upper half of the median of HEF volumes and about 40 percent of health facilities commissioned before 2015 had experienced delays in receiving HEFs. Almost all the facilities with HEF income (98.9 percent) received HEF through a bank account; so in principle HEF should reach the facility’s account on time, unless there are issues in upstream funding (and this was indeed the case in the past, where different phases of the project cofinancing had temporary delays in availability of funds at the national level). Almost all HCs identified user fees (96 percent) and HEF (92 percent) as their main source of income (multiple responses were permissible for this question), while approximately half of respondents identified MOH (47 percent). The average amount of funding received from user fees, HEF and MOH, in the month prior to the survey was approximately 11 million Riels, 6 million Riels, and 18 million Riels, respectively.4 According to the volume of HEF and duration of HEF operations, however, facilities reported having received different amounts of user fees in the month prior to survey. For example, as expected, high- HEF facilities reported higher HEF income (10.6 million Riels) and user fees (13.5 million Riels) than low- HEF facilities’ HEF income (1.7 million Riels) and user fees (8.6 million Riels). A similar trend was seen between pre-/post-2015 HEF adopters: pre-2015 HEF adopters reported higher HEF income (9.7 million Riels) and user fees (12.3 million Riels) than post-2015 HEF adopters’ HEF income (3.5 million Riels) and user fees (10 million Riels) respectively (Figure 3.3). Figure 3.3 Amount of User Fees and HEF In a Month Prior to Survey (In thousand Riels) Overall, 86 percent of health centers had access to working vehicles . Among these facilities, 23 percent had an ambulance owned by the facility, while 81 percent had an ambulance owned by OD. The likelihood of having an ambulance owned by the facility was noticeably different between high-HEF facilities (41 percent) and low-HEF facilities (7 percent) (Figure 3.4). There was also a difference between 4 1 million Riel= approximately US$250 49 pre-2015 HEF adopters (30 percent) and post-2015 adopters (17 percent), although to a lesser degree (Figure 3.5). Figure 3.4 Proportions of High-HEF Volume and Low-HEF Figure 3.5 Proportions of Pre-2015 HEF and Post-2015 HEF Volume Facilities with Ambulance Adopters with Ambulance In addition, the average number of ambulances among these facilities also differed according to HEF volume and HEF adoption year. On average, high-HEF facilities had 1.5 ambulances, compared to 1 ambulance for low-HEF facilities; pre-2015 HEF adopters had 1.6 ambulances, compared to 1.1 ambulances for post-2015 HEF adopters. These facilities with more experience with HEF patients could demonstrate a positive impact and prompt procurement of vehicles to meet specific services in a timely manner. Even for very common services such as immunization and maternity delivery, high-HEF volume facilities and early adopters of HEF were better placed. All health centers with high-HEF volume and those that started HEF operation prior to 2015 provided immunization services. However, just about 90 percent of facilities that started HEF operation after 2015 and had lower volume of HEF provided immunization services for the community. Similarly, there were a higher proportion of facilities providing delivery services at high-HEF facilities and pre-2015 HEF adopters, although the differences were larger in both cases. Ninety-nine percent of high-HEF facilities were providing delivery services, compared to 77 percent of low-HEF facilities (Figure 3.6), and 95 percent of pre-2015 HEF adopters were providing delivery services, compared to 81 percent of post-2015 HEF adopters (Figure 3.7). Figure 3.6 Proportions of High-HEF and Low-HEF Volume Figure 3.7 Proportions of Pre-2015 HEF and Post- Facilities Providing Delivery Services 2015 HEF Adopters Providing Delivery Services 50 Sixty-two percent of respondents indicated nongovernmental organization staff as an entity with the authority to procure drugs and equipment for the facility; 43 percent indicated local government (Figure 3.8). Health facility manager and health facility staff were chosen by 20 percent and 8 percent respectively of the respondents, showing that about one-third of the respondents, who are health facility head or staff themselves, expressed their ability to procure drugs and equipment at the time of the baseline. Figure 3.8 Average Independent Proportion of Entities with Authority to Procure Drugs and Equipment (in percent) Infrastructure A high number of health facilities had challenges with available infrastructure, especially those that were built a while ago. The last major investment in infrastructure in 16 percent of the health facilities was more than 10 years ago. For health facilities commissioned more than 10 years ago, almost one-fourth had the last major investment in the infrastructure at least 11 years ago. This suggests that several older facilities (which tend to serve more patients than the newer facilities) need to find resources to implement major improvements in infrastructure. It is critical to fix problems of electricity outage and water scarcity. About 58 percent of all facilities reported having electric power outages in the last seven days. The average duration of absence of electric power in the last seven days was 5.8 hours. About 8 percent of health facilities had no water available in the last seven days. The average duration of non-availability of water in the last seven days was 43 hours. Almost one-fourth of facilities had rainwater as a primary water source (12 percent), unprotected well (10 percent), and surface water such as lake, river, or stream (2 percent). Transportation services for patients were limited. Twelve percent of large volume facilities (defined by their size being larger than the median of total patients in the last month) faced transportation difficulties in serving patients. Six percent of facilities reported having no transportation for patients. The average unavailability of transportation for health care services was about a day for all facilities with transportation shortage. Access to computers and internet connectivity would be very useful and efficient for the health facilities, even though they are not yet expected to use Patient Management and Registration System (PMRS) at 51 the health center level. Fifty percent of facilities owned a functioning computer. However, connectivity remained erratic. Only 44 percent of those that reported owning a computer had access to internet connection. For facilities with internet connection, about 27 percent reported having no internet connection in the last seven days. Almost all facilities had a reception room and a waiting area (98 percent and 99 percent, respectively). However, only 74 percent of all facilities have a room with auditory and visual privacy for patient consultations. Only half of the facilities (49 percent) had a minor surgery theater. Furthermore, 16 percent of facilities had no electric fan or air-conditioner, 8 percent had no observation bed, and 18 percent had no separate wards for ANC and deliveries. The average number of observation beds, beds for general medicine, and beds for ANC/delivery were 3.7, 2.2, and 2.8, respectively. All facilities had a functional toilet facility available for patients, but almost one-fifth (19 percent) of the toilets were rated somewhat unclean or very unclean. Only 17 percent of all facilities provide separate toilet facilities for female and male patients. Although accommodation had been provided for health workers required to be on call in most HCs, 12 percent of facilities reported having no accommodation for health workers on call during non-routine hours, for example, during night shift. This was the case despite the fact that all facilities are expected to provide round-the-clock services all the time or sometimes (93 percent and 7 percent, respectively). Most of the national protocols are available in about 80 percent of HCs. Some national protocols are, however, less available, and available in less than 50 percent of facilities. These include the protocol for malaria diagnosis and treatment, protocol for reducing unsafe abortion morbidity/mortality, prevention of mother to child transmission of HIV (PMTCT) guidelines, HIV treatment (antiretroviral therapy, ART) guidelines, HIV treatment (antiretroviral therapy, ART) for children/infants guidelines, national list for essential drugs, protocol for drug procurement, detecting and reporting adverse drug or vaccine reaction, and national health strategy. Autonomy No respondents disagreed or strongly disagreed on having autonomy on assigning tasks to staff as needed or on OD’s support to the facility to make decisions (Table 3.1). On the other hand, 33 percent of respondents disagreed or strongly disagreed on having choice over types of services to be provided at the facility. Similarly, 27 percent disagreed or strongly disagreed on having enough authority to obtain the resources for the facility, namely drugs, supplies or funding. 52 Table 3.1 Autonomy on Decision-Making and Authority In the Facility (in percent) HEF claim HEF started Size (total patients) Autonomy All HFs >=median =median