2023 i [ ii iii FOREWORD The Constitution of Bhutan mandates the State to provide access to free basic health services for its citizens. This constitutional mandate is further reinforced in the 2011 National Health Policy. However, in fulfilling this mandate, the Royal Government of Bhutan is confronted with the issue of financial sustainability, especially against the rising demand for quality and free basic health services. Bhutan has made significant improvements in health outcomes in recent decades. Sustained investment in the health sector has allowed for increased access and utilization of health services due to the expansion of health facilities and improved distribution of health workers. Overall, this has significantly improved life expectancy. However, the continued advancement of health-related technologies for the improved efficiency and delivery of quality health services poses significant financial barriers for the government. While health outcomes continue to become a priority for the Government, financial resources for health are receding as Bhutan phases out its reliance on development partners. Unlike other countries which have health sector funding alternatives such as health insurance and private providers, Bhutan is still largely dependent on the public revenue to finance the government’s free basic health services. A series of exercises on the National Health Accounts which were carried out since 2010 confirms that public financing remains a predominant source of financing Bhutan’s health care services. During the 2018-2019 financial year, public health expenditure was estimated at 81% as a proportion of total health expenditure, or about 4% of GDP. This report presents costing estimates using 2018-2019 data, revealing that on average, the unit cost for an outpatient department (OPD) visit is Nu.1011 for referral hospitals, Nu.590 for District hospitals, Nu.303 for 10-bedded hospitals and Nu.415 for primary health care centers. Similarly, the average unit cost of inpatient department (IPD) admission for referral hospital is estimated at Nu.33709, Nu.16331 for district hospitals and Nu.12934 for 10-bedded hospitals. Furthermore, the report highlights key cost estimates at the facility level. To effectively prioritize health spending and investment, policymakers should make informed decisions based on the availability and credibility of relevant health costing information. The costing exercise provided in this report provides just that-- key information necessary for health planners to efficiently mobilize financial resources to effectively improve health services across the country. In this vein, we encourage a thorough reading of this important and informative costing report to make informed financial policy decisions. Mr. Pemba Wangchuk Acting Secretary Ministry of Health iv ACKNOWLEDGEMENTS The Ministry of Health would like to acknowledge continued support of the World Bank in carrying out cost analysis of health care services in Bhutan in collaboration with the Ministry of Health, Bhutan. The World Bank team has rendered both cooperation and guidance through virtual platform services during the COVID-19 pandemic to ensure that the costing exercise efficiently makes use of time and resources. We are grateful to the Global Fund for funding to support this work. We would also like to express our gratitude and appreciation to Mr. Kim Gustavsen from Denmark who was recruited as an international consultant to carry out this costing exercise to build the capacity of the Ministry of Health. He was also engaged as a consultant for the previous costing exercise conducted in 2010. With the aforementioned support, the Ministry of Health is able to present a synthesized and comprehensive costing report. Most importantly, we would like to acknowledge the support of district administrators, and health facility staff who provided data collection and verification support. The Ministry of Health would also like to extend its appreciation to the Policy and Planning Division of the Ministry of Health for taking the lead in carrying out this costing exercise. Lastly, we would like to acknowledge the following team members for their leadership of the costing exercise: • Mr. Kim Gustavsen, Consultant, Denmark • Ms. Kathryn Andrews, Health Economist, World Bank • Ms. Mamata Ghimire, Health Economist, World Bank • Mr. Faraz Salahuddin, Consultant, World Bank • Mr. Ajay Tandon, Lead Economist, World Bank • Mr. Hideki Higashi, Senior Health Economist, World Bank • Mr. Ali Hamandi, Senior Health Economist, World Bank • Mr. Tashi Penjor, Chief Planning Officer, Policy & Planning Division, Ministry of Health, Bhutan. • Mr. Tshering Wangdi, Senior Planning Officer, Focal Point, Policy & Planning Division, Ministry of Health, Bhutan. • Mr. Karma Jurmin, Senior Program Officer, Health Financing Division, Ministry of Health, Bhutan. v TABLE OF CONTENTS FOREWORD ........................................................................................................................................................ iv ACKNOWLEDGEMENTS ...................................................................................................................................... v TABLE OF CONTENTS ........................................................................................................................................ vi LIST OF TABLES ................................................................................................................................................. vii LIST OF FIGURES .............................................................................................................................................. viii ABBREVIATIONS ................................................................................................................................................. ix 1. EXECUTIVE SUMMARY.................................................................................................................................... 1 2. INTRODUCTION .............................................................................................................................................. 4 3. THE BHUTANESE HEALTH SECTOR ......................................................................................................... 6 3.1. DELIVERY OF HEALTH CARE ....................................................................................................................... 7 3.2. FINANCING OF HEALTH CARE .................................................................................................................... 8 3.3. SELECTED HEALTH CARE INDICATORS ........................................................................................................ 8 4. METHODOLOGY ...........................................................................................................................................10 4.1. DEFINITION OF VARIOUS COSTS ..............................................................................................................11 4.2. SCOPE AND PERSPECTIVE OF COSTING ANALYSIS ......................................................................................12 4.3. DATA COLLECTION ................................................................................................................................14 4.4. STANDARD COSTING MODEL ..................................................................................................................15 4.5. CALCULATION OF DISEASE SPECIFIC COSTS ...............................................................................................17 5. RESULTS .........................................................................................................................................................19 5.1. NATIONAL AND REGIONAL REFERRAL HOSPITALS.....................................................................................20 5.1. DISTRICT HOSPITALS .............................................................................................................................26 5.2. 10-BEDDED HOSPITALS .........................................................................................................................34 5.3. PRIMARY HEALTH CENTERS ....................................................................................................................39 6. COST CHANGE ..............................................................................................................................................43 7. LIMITATIONS & ASSUMPTIONS ..................................................................................................................47 8. DISCUSSION & RESULTS ANALYSIS ............................................................................................................50 5.4. HOW CAN THE COSTING RESULTS BE USED? .............................................................................................52 9. CONCLUDING REMARKS ..........................................................................................................................55 ANNEX A: DETAILED COSTING METHODOLOGY .........................................................................................58 REFERENCES ......................................................................................................................................................76 vi LIST OF TABLES TABLE E1. AVERAGE UNIT COSTS AT DIFFERENT LEVELS, NU. TABLE 3.1. HEALTH FACILITIES AND STAFF, 2005-2019 TABLE 3.2. MAJOR HEALTH INDICATORS, 2000-2019 TABLE 4.1. INCLUDED HEALTH FACILITIES TABLE 4.2. SUMMARY OF HEALTH FACILITIES TABLE 4.3. COST CENTERS OF THE STANDARD COSTING MODEL TABLE 5.1. AVERAGE UNIT COSTS AT DIFFERENT LEVELS, NU. TABLE 5.2. SUMMARY OF REFERRAL HOSPITALS TABLE 5.3. TOTAL COSTS FOR REFERRAL HOSPITALS, NU. TABLE 5.4. TOTAL COSTS FOR FINAL COST CENTERS FOR REFERRAL HOSPITALS, NU. TABLE 5.5. UNIT COSTS (NU.) AND ACTIVITY FOR REFERRAL HOSPITALS TABLE 5.6. DISEASE SPECIFIC INPATIENT COST PER ADMISSION AND NUMBER OF ADMISSIONS FOR REFERRAL HOSPITALS TABLE 5.7. SUMMARY OF DISTRICT HOSPITALS TABLE 5.8. TOTAL COSTS FOR DISTRICT HOSPITALS, NU. TABLE 5.9. TOTAL COSTS FOR FINAL COST CENTERS FOR DISTRICT HOSPITALS, NU. TABLE 5.10. UNIT COSTS AND ACTIVITY FOR DISTRICT HOSPITALS TABLE 5.11. DISEASE SPECIFIC INPATIENT (UNIT) COST PER ADMISSION FOR DISTRICT HOSPITALS, NU. TABLE 5.12. DISEASE SPECIFIC NUMBER OF ADMISSIONS FOR DISTRICT HOSPITALS TABLE 5.13. SUMMARY OF 10-BEDDED HOSPITALS TABLE 5.14. TOTAL COSTS FOR 10-BEDDED HOSPITALS, NU. TABLE 5.15. TOTAL COSTS BY FINAL COST CENTER FOR 10-BEDDED HOSPITALS, NU. TABLE 5.16. UNIT COSTS AND ACTIVITY FOR 10-BEDDED HOSPITALS TABLE 5.17. DISEASE SPECIFIC NUMBER OF ADMISSIONS FOR 10 BEDDED HOSPITALS TABLE 5.18. SUMMARY OF PRIMARY HEALTH CENTERS TABLE 5.19. TOTAL COSTS (COST STRUCTURE) FOR PRIMARY HEALTH CENTERS, NU. TABLE 5.20. TOTAL COSTS FOR FINAL COST CENTERS FOR PRIMARY HEALTH CENTERS, NU. TABLE 5.21. UNIT COSTS FOR PRIMARY HEALTH CENTERS, NU. TABLE 6.1. TOTAL COST COMPARISON OF 2009/10 AND 2018/19 COSTING EXERCISE, NU. (FIXED 2018/19 PRICES) TABLE 6.2. UNIT COST COMPARISON OF 2009/10 AND 2018/19 COSTING EXERCISE, NU. (FIXED 2018/19 PRICES) vii LIST OF FIGURES FIGURE 5.1. FIXED VS. VARIABLE COSTS FOR REFERRAL HOSPITALS, NU. FIGURE 5.2. DIRECT VS. INDIRECT COSTS FOR REFERRAL HOSPITALS, NU. FIGURE 5.3. FIXED VS. VARIABLE COSTS FOR DISTRICT HOSPITALS, NU FIGURE 5.4. DIRECT VS. INDIRECT COSTS FOR DISTRICT HOSPITALS, NU. FIGURE 5.5. FIXED VS. VARIABLE COSTS FOR 10-BEDDED HOSPITALS, NU. FIGURE 5.6. DIRECT VS. INDIRECT COSTS FOR 10-BEDDED HOSPITALS, NU. FIGURE 5.7. FIXED VS. VARIABLE COSTS FOR PRIMARY HEALTH CENTERS, NU. FIGURE 5.8. DIRECT VS. INDIRECT COSTS FOR PRIMARY HEALTH CENTERS, NU. FIGURE 6.1. TOTAL COST COMPARISON OF 2009/10 AND 2018/19 COSTING EXERCISE, NU. (FIXED 2018/19 PRICES) FIGURE 6.2. UNIT COST COMPARISON OF 2009/10 AND 2018/19 COSTING EXERCISE, NU. (FIXED 2018/19 PRICES) viii ABBREVIATIONS RGOB The Royal Government of Bhutan OPD Outpatient Department IPD Inpatient Department ORC Outreach Clinic JDWNRH Jigme Dorji Wangchuck National Referral Hospital ERRH Eastern Regional Referral Hospital CRRH Central Regional Referral Hospital RRH Regional Referral Hospital BHMIS Bhutan Health Management Information System BHIS Bhutan Health Information System MCH Maternal and Child Health HTA Health Technology Assessment DRG Diagnosis Related Group ix 1. EXECUTIVE SUMMARY 1 Overview The Royal Government of Bhutan (RGoB) is examining alternative sustainable financing strategies in the hopes of improving the effectiveness and efficiency of its health spending, and to be able to continue providing free, quality health services to its population. This report is the outcome of a costing study undertaken to obtain information about the costs of delivering health services at different administrative health care levels in Bhutan, to: • Inform the Ministry of Health about the costs of delivering various types of services at different levels; • Increase cost-awareness and knowledge about healthcare delivery costs among policy researchers and the general public; • Support future evidence-based and informed policy decisions to increase efficiency of the Bhutanese health sector. The study reviewed healthcare costs and activities in 9 districts, covering a total of 12 facilities, including: the National Referral Hospital, the 2 Regional Referral Hospitals, 4 district hospitals, two 10- bedded hospitals and 3 primary health centers. The study made use of an easily replicable methodology to assess the cost of resources used to provide services covering the 2018/2019 financial year. The study also updates a similar study undertaken in 2011 for the 2009/2010 financial year. For the purposes of the study, a costing model was developed using internationally recognized methodologies that were adjusted to the Bhutanese health system. Main results The results provide detailed insight into the cost structures of the health facilities in Bhutan at different service delivery levels. In general, there are significant differences in cost-structures between facility types at different levels, and significant differences between the same level of facility. Furthermore, the findings indicate that the higher the total cost at the facility, the more patients are treated at the facility. Some of the key results of the exercise, in terms of unit costs at the various types of health facilities, are summarized in Table E1. TABLE E1. AVERAGE UNIT COSTS AT DIFFERENT LEVELS , N U. Referral District 10-Bedded PH Hospitals Hospitals Hospitals Centers Outpatient Department OPD-visit 1,011 590 303 415 Inpatient Department Admission 33,709 16,331 12,934 - Bed day 5,106 - - - Note: Unit Cost refers to the average cost of one outpatient visit, an admission or a bed day (cost per day admitted). Data on number of bed days was only available for referral hospitals. 2 The results of the costing exercise suggests that the higher the level of health facility, the higher the cost of providing care. The costing exercise also suggests that it is more cost effective to access health services from lower-level health facilities than it is to access the same services from a higher level facility, given that the quality of the services is equal. However, there is an observable trend that people are more likely to seek out health services at higher-level facilities, despite the same services being provided at lower-level facilities. This behavioral tendency is considered to cost-inefficient and can result in unnecessary stress on a healthcare system with limited resources if the same service is provided and available at a lower level. In the past 10 years, the total costs as well as the average unit costs have increased, even after accounting for price inflation. Among the main drivers for increasing costs are not only the rising cost of personnel, but also the cost of medicine and medical consumables and equipment. As such, Bhutan has experienced an increase in total costs – in fixed prices – of 62 percent for referral hospitals and 54 percent for district hospitals since 2009/10. Key takeaways from the study are: • Costs and cost-structures at health facilities. The study gives detailed insight into total costs and composition of costs of facilities at various levels. • Average Unit costs. The costs of OPD-visits, admissions in general, as well as disease-specific groupings of admissions have been calculated and generally shows that services are more costly at facilities at higher levels. For example, the average cost of an OPD-visit at a referral hospital is three times as high as the average cost at a 10-bedded hospital. Likewise, the average cost of an admission at a referral hospital is double the average cost of an admission at a district hospital. • Cost increases over time. The cost of health services increases over time. Compared to the costing exercise conducted in 2009/10, total costs have increased more than 50 percent for referral hospitals and district hospitals. There is also an increase, although with broader variations, in the average unit costs of OPD-visits and admissions. • Lowest effective level of service delivery is the most cost-effective. Providing services at the lowest effective level of service delivery (facilities) is the most cost-effective way of providing services in the long run. This means that if the service is available at the same level of quality at a 10-bedded hospital, it is more cost-effective to utilize the service there than seeking out the same service at a district hospital or a referral hospital. The results have the clear potential to be used to inform decisions makers and to be used for further analytical work to increase the efficiency and value-for-money utilization of resources in the health sector. The study points to a number of ways in which the results from the costing study can be used to improve efficiency in the health sector. Two areas with particularly large potential are: • Using the results to improve referral pathways and increase cost-efficiency of service delivery by delivering services at the lowest feasible point of service delivery. • More input for optimal planning and budgetary allocations. 3 2. INTRODUCTION 4 The RGoB is examining alternatives for a sustainable financing strategy in its commitment to improve the effectiveness and efficiency of its health spending. Rising public expectations have recently further pressured the RGoB to improve the efficacy of its public health services. The first comprehensive analysis of the cost of public health services (hereby referred to as a costing study) was undertaken in 2011. The study in this report aims to build upon the methodology established from the previous exercise and update it using new data. This report is the outcome of a study undertaken on behalf of the Ministry of Health and financed by The World Bank to further illuminate and present an updated analysis of the costs of delivering public health services in Bhutan. The results of this study aim to be used widely to assist relevant stakeholders in making informed policy decisions to: improve the delivery of health services in Bhutan, enhance the effectiveness and efficiency of health spending, and increase cost awareness for policy-makers, administrators and health care workers. The study reviewed the health activities and associated costs in 9 districts, covering a total of: twelve facilities (including the National Referral Hospital), Bhutan’s two Regional Referral Hospitals, four district hospitals, and five other health facilities. For each of these facilities, total costs are calculated in addition to the unit costs of outpatient visits, admissions, and the cost of a bed-day (cost per day of admission) in the major hospitals. The average admission cost by both major and selected disease groupings are also calculated. The study compares the results of the current costing exercise with the results of the exercise undertaken in 2011, analyzing changes over time. Finally, the report provides guidance on where and how the costing study can be used for informed policymaking to strengthen the effectiveness and efficiency of health spending in Bhutan. The report is structured in several chapters, outlining: the methodology used, the results of the study, and a discussion of the results and suggestions for how to operationalize the results. Chapters 1 and 2 presents the executive summary and introduction of the report. In Chapter 3 background information on the Bhutanese health sector is briefly presented. The methodology for the study is presented in Chapter 4. Chapter 5 presents the results of the study detailing the four different levels of service delivery studied. A comparison of the results of the costing exercise undertaken in 2011 and the development in these results are presented in Chapter 6. The key limitations and assumptions of the study are outlined in Chapter 7. A discussion of results takes place in Chapter 8, including a presentation of possible uses of the results to inform policy decisions. Finally, Chapter 9 summarizes the conclusions of the study. 5 3. THE BHUTANESE HEALTH SECTOR 6 The predominantly public financed and managed health system in Bhutan has evolved and grown remarkably in fifty years. Health services are available through a three-tier structure, i.e., primary, secondary and tertiary levels. Traditional and allopathic medicine services are integrated and delivered under one roof. Health services are free as enshrined in the Constitution of Bhutan. Despite the difficult geographical terrain and dispersed population settlements, access to health services has improved remarkably. Bhutan is now among the top global performers in life expectancy gains within the past 40 years and has reached set targets for reducing child mortality and improving maternal health. Additionally, immunization levels are today very high, especially in comparison to quite low levels of immunization coverage in 1995. 1 During the COVID-19 pandemic Bhutan managed to inoculate more than 95 percent of its eligible population during two rounds of nationwide COVID-19 vaccination campaigns. 3.1. Delivery of health care Health care in Bhutan is delivered through several types of health facilities, ranging from outreach clinics (ORCs) and sub-posts (the smallest health facilities) to regional referral hospitals (and one national referral hospital in the nation’s capital). Service delivery is to a high degree decentralized with districts overseeing planning, including: staffing, training, supplies and construction. However, human resource planning and employment, the purchase of drugs and equipment, and the construction of larger hospitals are still handled centrally. Vertical health and disease control programs, including those for tuberculosis, leprosy, malaria, immunizations, and maternal and child health is also being managed and monitored centrally. The table below outlines the major developments in health facilities and staff in Bhutan in the past 15 years. TABLE 3.1. H EALTH FACILITIES AND STAFF , 2005-2019 2005 2010 2019 Facilities Hospitals 29 31 49 Primary Health Centers 176 181 186 Thromde Health Centers N/A N/A 3 Sub-posts N/A N/A 53 Outreach Clinics 485 518 542 Doctors 145 187 376 1 Source: WHO and UNICEF (who.int/immunization/monitoring_surveillance/en). 7 Staff Doctors per 10,000 2.3 2.7 4.9 population Health Assistants 171 366 620 Nurses 538 556 1,364 Source: Health Bulletins, various years. Between 2005 and 2019, Bhutan has seen an increase in its number of health care facilities, especially hospitals (29 to 49). An even greater increase is observed in staffing (doctors, health assistant, and nurses) that has doubled or tripled within 15 years. However, despite this growth, it is still considered low compared to recommendations set by the World Health Organization (WHO). Specifically, in 2019 Bhutan had almost 5 doctors per 10,000 people, while the recommendation by WHO is double that amount at 10 per 10,000 people. 3.2. Financing of health care Health care in Bhutan is predominantly financed by the government, aligned with the constitutional mandate of providing free access to basic public health services. The total health expenditure as a share of GDP has remained around 4 to 4.5 percent in recent years, reaching 4.5 percent in 2019/20, amounting to Nu. 8.7 billion. The National Health Accounts for the 2019/20 fiscal year show that government funding is the main source of health financing in Bhutan at 73.4 percent, with households providing 15.4 percent, donors 5.1 percent and other sources 6 percent (NGOs, insurance companies, private corporations and non- profit institutions serving households). 3.3. Selected health care indicators Select major health indicators between 2000 and 2019 are presented in the table below. There were positive gains in all indicators during the past twenty years. The positive trends can be attributed to the expansion of health services in the period providing better access for the entire population as well as improvements in quality of services. 8 TABLE 3.2. MAJOR HEALTH INDICATORS , 2000-2019 Indicator 2000 2005 2010 2015 2019 Infant Mortality Rate (per 1,000 live births) 60.5 40.1 47.0 30.0 30.0 Under 5 Mortality Rate (per 1,000 live births) 84.0 61.5 69.0 37.3 34.0 Maternal Mortality Ratio (per 100,000 live births) 255 - - 86 89 Births attended by trained health staff (percent) 23.7 49.1 64.5 74.6 97.0 Source: Annual Health Bulletin, various years. Improvements in water and sanitation, expansion of primary education and general socio-economic development has likewise been major factors that have contributed to improvements in health. However, it should be noted that maternal mortality rates are to be interpreted with caution. Due to the small size of the population, a single maternal death can drastically change the indicator’s value. In the past 20 years, a decline in infectious diseases (diarrhea, dysentery, etc.) has been observed, as well as an increase in non-communicable diseases (especially diabetes and hypertension). Likewise, the areas of alcohol-induced liver diseases and traffic accidents have contributed to a substantial portion of the disease burden in Bhutan. 9 4. METHODOLOGY 10 This chapter presents a brief discussion of the methodology used for the costing analysis. A full and more detailed version is available as an annex to the report. 4.1. Definition of various costs Costing theory distinguishes between different types of costs. In the short term, a distinction is made between fixed and variable costs, which together constitute the total cost. Fixed costs cover resources such as staff and buildings, and are independent of and do not vary with specific activities and care at the health facility (i.e. number of patient visits and admissions). The variable costs, on the other hand, are costs that vary with various activities and care, such as the cost of medicine and food for admitted patients. In this regard, ‘activities’ refers broadly to both in-patient and out-patient treatments. Average costs are the total fixed and variable costs calculated per unit. Marginal costs constitute the cost increase for a rise in the number of activities and care (such as patients treated) and thus do not affect fixed costs. The definition of the various terms relevant for this analysis are presented in Box 4.1 below. BOX 4.1. KEY COSTING TERMS • Total Costs = Fixed costs + Variable Costs • Fixed Costs = independent of production/services (buildings, fixed staff) • Variable Costs = increases with production/services (medicine, utensils, care) • Total Average Costs = Total Costs / Number of produced units • Marginal Costs = Cost increase of a small production increase • Direct Costs = Costs that can be attributed directly to the production of services / the production site (medicine, equipment, staff at service delivering departments, etc.) • Indirect Costs = Costs that cannot be attributed directly to the production of services / the production site – overhead costs and costs of cross-cutting service centers (administration, security, x-ray, laboratories, etc.) Note: The concept of Direct and Indirect Costs are used differently in the literature depending on whether it is "cost accounting" or "health economic evaluation". For this analysis the definition for Cost Accounting was used. 11 4.2. Scope and perspective of costing analysis The costing analysis was carried out using a top-down approach and designed to calculate unit costs by relating activities carried out in a number of select health facilities to the costs of undertaking those activities. It is important to note that the study was not designed to assess the quality of care or the outcome of care provided, and hence no judgment can be made on the relative cost-benefit of care provided by different facilities. The unit cost represents the average cost for OPD-visits and admissions. In this regard an OPD-visit is a visit to the outpatient department that does not include an admission or overnight stay. An admission is when a patient is admitted at a ward for treatment that requires one or more overnight stays in the facility. 4.2.1. Sample Size A total of 12 facilities and associated out-reach clinics2 have been selected, as shown in the table below. TABLE 4.1. I NCLUDED H EALTH FACILITIES Dzongkhag Facility Name Facility Type /District Jigme Dorji Wangchuck National Referral National Referral Thimphu Hospital Hospital (JDWNRH) Eastern Regional Regional Referral Referral Hospital Mongar Hospital (ERRH) Regional Referral Central Region (CRRH) Sarpang Hospital Paro District Hospital Paro Wangdicholing District Hospital Wangdicholing Damphu District Hospital Tsirang Punakha District Hospital Punakha Gyelposhing 10-bedded Hospital Mongar Kanglung 10-bedded Hospital Trashigang Genekha Primary Health Center Thimphu Mendelgang Primary Health Center Tsirang Thinleygang Primary Health Center Punakha 2 Many facilities have associated outreach clinics in order to provide services in remote and underserved areas. The outreach clinics are only manned at specific times during the month and services provided are often related to health promotion activities. The cost of sheds and staff time spent doing outreach services was assessed and included in this study. However, these activities are not always recorded as OPD-visits, potentially resulting in additional ORC costs that are attributed to maternal and child health services, despite not directly affecting the unit costs of OPD-visits. 12 The main criterion for facility selection in the study was whether or not the facility was included in the 2009/10 analysis. A preference was set for facilities that were included in the past study, with the aim of being able to analyze and compare changes over time whenever possible. One facility from the 2009/10 analysis however has been excluded as it was upgraded to a higher-level facility, and one other facility has been excluded as it has been repurposed into a facility that provides traditional medicine only. To make up for the loss of these two 10-bedded hospitals in the sample, Kanglung hospital has been included as a new facility. Although the study only includes 12 facilities (including their associated out-reach clinics) from a total of 235 facilities in 2020, this still covers a large share of the total OPD-visits and admissions in Bhutan, as this sample size includes all major hospitals. For example, out of 1.4 million total OPD-visits and 73,000 total admissions in the country in 2019, the study manages to cover 1 million OPD-visits, with 44,678 total admissions of the same year. In this regard, the study effectively covers 60 to 70 percent of all cases in Bhutan. Table 4.2 below summarizes characteristics of the 12 facilities in terms of staffing, type and number of services delivered. TABLE 4.2. SUMMARY OF HEALTH FACILITIES Number OPD-visits Admissions per Facility Name OPD-visits Admissions of staff per staff staff JDWNRH 1,100 514,502 13,480 468 12 ERRH 380 53,650 5,168 141 14 CRRH 460 128,867 4,206 280 9 Paro 140 167,940 15,411 1,200 110 Wangdicholing 70 48,763 738 697 11 Damphu 130 30,330 3,028 233 23 Punakha 100 54,970 1,791 550 18 Gyelposhing 20 15,082 474 754 24 Kanglung 35 15,628 382 447 11 Genekha 2 3,000 NA 1,500 NA Mendelgang 5 5,954 NA 1,191 NA Thinleygang 4 6,170 NA 1,543 NA Total 2,446 1,044,856 44,678 427 18 Source: Bhutan HMIS and data collected through costing tool. Note: The staff number used is the total staff at the facility, including non-health personnel, and is presented to illustrate huge variations across facilities. 13 4.2.2. Type of costs Both capital and recurrent expenditure were considered in the study to derive the total costs. A unit cost analysis that ignores capital costs assumes that physical assets will be available forever. The daily use of these assets (e.g., buildings, equipment, etc.) and the depreciation of capital items is considered an expense – even though it is not an expenditure (e.g., salaries and drugs are an expenditure). The cost of capital items is represented as an annualized depreciation cost. This represents the yearly cost of the given item. 4.2.3. Perspective The perspective adopted for the study is from the viewpoint and interest of the facility’s administration – and not from the perspective of the patient or larger public interest. The cost analysis is thus limited to the costs related to providing services at a given facility. This ensures comparability and uniformity of the results, as well as the ability to provide institutional-level information to be used as input for policymaking. This means that costs incurred by patients outside of facility care (e.g., travel, cost of an escort, etc.) have not been considered. Likewise, the supervisory costs of the central MOH and other various vertical programs have not been estimated nor used as overheads for the individual health facilities. Finally, costs of out-of-country referrals are not covered by the study as the study scope is limited to the costs of activities and care in health facilities in Bhutan. 4.3. Data collection The study required the collection of a considerable amount of data from all participating health facilities as well as central agencies. This included interviews and guidance on using the data collection templates with staff at all sampled facilities and from all participating central agencies. The data section in the methodology annex describes this process and the data sources in further depth. To support data collection, a data collection template was developed in excel. This template was used to estimate the staffing costs at the facilities, in addition to where in the facility resources were deployed to establish weights used for the final cost calculations. The final cost calculations took place in a model also designed in excel that allowed for automatic and efficient calculations and predictions. The improve the tool’s predictive accuracy, multiple rounds of data validation were conducted, with assistance from the MoH. The MOH also facilitated the use of the data collection template in all the facilities covered by the sample size. In general, data required for the study were available and of good quality. Public facilities are fully financed by the government, with the Ministry of Health supplying medicine, medical supplies, consumables, and equipment. Detailed records of these transactions were in principle available. The Bhutanese Health Management Information System (BHMIS) is generally well developed, and routinely collects data from all facilities across Bhutan. Reporting templates and forms are standardized, and data are collected at the district level, summarized, and then put into the system. The referral hospitals also keep their own records at a more detailed level than what is required to be reported in the HMIS. Records from JDWNRH were also made available and used for the study. Finally, 14 yearly financial statements at the facility level were available and used as the main source for calculating recurrent costs at the facility level. However, some data were not able to be collected and instead were estimated. This includes: • Cost of medicine and consumables. Data on the actual use of drugs and consumables within facilities were not available for the full observation period of the study, nor were they typically recorded into a system. In the 2009/10 costing exercise, it was assumed that what was distributed during the observation period was fully consumed. With this assumption, facility distribution lists were used for item costing. For this current costing exercise, however, it was not possible to get a full facility distribution list of all the different medicines and medical consumables with their associated usable unit prices. However, the total cost of drugs and consumables for 2018/19 for all facilities was available and used in the absence of detailed individual facility level data. It was then assumed that the cost for the facilities in the sample could be estimated using a share of the total cost of all 12 facilities in the sample, divided by each facility’ share of total OPD-visits from the 2009/10 costing exercise3. • Cost of medical equipment. Detailed lists of thousands of different medical equipment at the various facilities in the sample were available. Though ideally each medical equipment should have been costed with an estimated annual depreciation value, not all unit prices were readily available. Instead, the annual expenses on equipment were used. By using the same approach as with medicine and consumables, the total cost for all sample facilities were derived and then divided according to their shares of the 2009/10 exercise to determine costs at the facility level3. This meant that no annual depreciation cost was utilized in the calculation, but as the annual cost was used, the total assumes a constant rate of investment in equipment. • Cost of buildings and vehicles. The cost of buildings and vehicles were estimated in the 2009/10 exercise and was used in the current exercise, expressed now in 2018/19 prices. 4.4. Standard costing model A standard costing model was developed and used uniformly across all 12 facilities. The model is described in full detail in Annex A. The model used in the current exercise is the same model used in the 2009/10 costing exercise, which has allowed for direct comparisons between the two time periods based on a standard top-down costing methodology. A top-down costing approach involves allocating an aggregated total cost to a pre-defined number of cost centers in the facility, according to a specified allocation criteria. A cost center represents a function at the facility. The model has cost centers identified at three levels: overhead, intermediate, and final cost centers. Overhead cost centers refer to the cost for administration, security, etc.; intermediate cost centers cover various supportive functions like diagnostic services, operating theatres and kitchens. Final cost centers include areas of final service provision such as outpatient and inpatient departments. 3 A new facility was introduced in the study – Kanglung – that was not a part of the 2009/10 exercise. For this facility the share of a facility of similar type was used. 15 Inputs to each cost center were identified, quantified and given a numerical, financial value in Nu. A step-wise approach was used. First, all overhead costs were allocated to intermediate cost centers. Second, all intermediate costs were then allocated to final cost centers. Then, the unit costs for final cost centers were calculated. All data presented in the model is based on collected data from health facilities, district administrative offices, and Ministry of Health. In general, the model utilized the collected data directly. However, for some inputs, as outlined in the section above, it was necessary to utilize estimations, especially where data was not available. 4.4.1. Defining cost centers The different cost centers are depicted in the table below. TABLE 4.3. C OST CENTERS OF THE STANDARD COSTING MODEL Overhead Intermediate Final Administration Imaging Outpatient department Transport Kitchen Inpatient department Staff quarters Laboratory • Medical • Surgical and Security & Maintenance Pharmacy/Dispensary medical Operating Theatre Maternal & Child Health Traditional Medicine Unit For the purposes of the model, the inpatient department was divided into ‘Medical’ and ‘Surgical and Medical’ wards. Based on disease groupings data, patients could be classified under either one of these two wards. Since it is not possible with certainty to know whether surgical or medical (or both) procedures were used in the treatment by disease grouping, the ‘surgical and medical’ subgroup covers diseases where both types of procedures can be used, while the ‘medical’ subgroup only covers diseases where no surgical procedures were conducted and where treatment was conducted only through health care services and the provision of medicine. However, not all facilities have all cost centers identified in Table 4.3, while more advanced facilities that do could further benefit from a more detailed division of the currently identified cost centers. Despite that, a uniform model was used to ensure comparability of results across all the different types of facilities. A tailor-made model for each facility was outside the scope of the study. Using one uniform model made it possible to assess the 12 facilities in the sample uniformly. 16 The final cost centers constitute service-delivery departments like OPD and inpatient wards. It also includes Maternal & Child health, where a number of promotive and preventive health activities are conducted. The results of the overhead and intermediate cost centers were utilized to determine the final – service delivery – cost centers, using a step-wise allocation method. The various mappings used for this allocation varies and a detailed description can be found in Annex A. After final costs centers were determined, unit costs were calculated by dividing the total cost of the cost centers with the number of units produced in those cost centers. For example, the total cost of the outpatient department (once all costs were allocated in a step-wise manner) was divided by the number of outpatient visits to derive the average unit cost of an OPD-visit at the given facility. 4.4.2. Period covered The period chosen for the study was the 2018/19 financial year. A period of one year was chosen in order to equalize for seasonal variations, as well as accommodate for some key data sources that are only available on an annual basis, such as the facility’s annual financial statements. 4.4.3. Annual depreciation of capital items The annual depreciation cost of capital items is difficult to measure if certain information is not available, such as the purchase price and life expectancy of items. The resulting cost estimate may therefore be very sensitive based on the assumptions made. Ideally, all capital items should be given a financial value and a life expectancy. The yearly depreciation cost of capital items can then be calculated using a depreciation rate of six percent, which is based on the Bhutanese consumer price- index. Details of these calculations are presented in Annex A. As outlined in the data collection section above, for a number of the capital items it was necessary to develop estimates based on total expenditure data for the 2018/19 financial year, and to use distributions from the 2009/10 study. Annex A includes further details on how annual depreciation of capital items were calculated. 4.5. Calculation of disease specific costs In addition to providing overall cost calculations of OPD and IPD activity, disease specific calculations were made according to the major diagnostic classifications used in the Bhutan Health Information System (BHIS). The data only supported calculating the average cost by diseases for inpatient care. In order to perform this calculation, disease aggregated data on inpatient bed-days was used to calculate the average bed-days by disease at the three referral hospitals included in the study. This resulted in calculations showing the aggregate total of bed-days at the three referral hospitals. Based on the unit cost of bed-day at each facility, an average admission cost for the various disease categories could then be calculated across hospitals. 17 In the current study, the detailed bed-day data of the hospitals was not available. By using the weights of the three referral hospitals for bed-days from the 2009/10 costing study, the bed-days was able to be calculated for the current study. With these bed-day estimations, the disease specific costs of admissions for the three referral hospitals could also be calculated. For the district hospitals and 10-bedded hospitals, the combined weight of the three referral hospitals was used to calculate bed-days and subsequently the disease specific costs of admissions. This is the most precise method to assess disease specific costs given the available data and utilizing a top-down costing exercise methodology. However, since cost data of expensive medicine, equipment and diagnostic services for different disease groupings were not recorded or available, these costs were split broadly across all disease groups, resulting in the length of stay at the hospital becoming the defining factor in determining the average cost of disease-specific admissions. 18 5. RESULTS 19 In this chapter the main results of the study across all levels of service delivery are summarized. The key results of unit costs at the four different levels of service delivery are presented in Table 5.1 below. In the following sub-sections, each level of service delivery is described in more detail. From Table 5.1, a general trend is observed that the higher the level of facility, the higher the unit cost, with the exception that unit costs for OPD-visits at primary health care centers are higher – at Nu. 415 – than at 10-bedded hospitals where it is Nu. 303. The average cost of an OPD-visit at referral hospitals are Nu. 1,011 compared to roughly half the cost at Nu. 590 at district hospitals, and down to Nu. 303 at 10-bedded hospitals. The higher unit cost of OPD-visits at primary health care centers is driven by relatively lower levels of activity in terms of patient visits, in addition to the higher costs of drugs and medical supplies at primary health care centers compared to 10-bedded hospitals. TABLE 5.1: AVERAGE UNIT COSTS AT DIFFERENT LEVELS , N U. Referral District 10-Bedded PH Hospitals Hospitals Hospitals Centers Outpatient Department OPD-visit 1,011 590 303 415 Inpatient Department Admission 33,709 16,331 12,934 - Bed day 5,106 - - - Source: Data summary from costing model. Note: Unit Cost refers to the average cost of one outpatient visit, an admission or a bed day (cost per day admitted). Data on number of bed days was only available for referral hospitals. This summary excludes results from Paro Hospital, as the data differed too much from the other district hospitals and was considered an outlier. The same pattern is seen with regards to inpatient care. The higher the facility level, the higher the cost per admission. The average cost of an admission is Nu. 12,934 at 10-bedded hospitals, Nu. 16,331 at district hospitals, and Nu. 33,709 at referral hospitals. The average cost of a bed day at referral hospitals were calculated to be Nu. 5,106. The following sections will cover in more detail the four different levels of service provision in terms of cost composition, unit costs and disease specific unit costs. 5.1. National and Regional Referral Hospitals Some of the key characteristics of the three referral hospitals are summarized below. Not surprisingly, as the largest referral hospital, the Jigme Dorji Wangchuck National Referral Hospital (JDWNRH) has the highest number of admissions as well as OPD-visits. Inpatient care capacity at the two regional referral hospitals (Eastern and Central RRH) is more or less equal. However, when looking at number of OPD-visits, Eastern RRH has less than half of Central RRH’s OPD-visits, at around 53,000 for the 20 period covered. The number of OPD-visits for Eastern RRH is curiously more on par with district hospitals (excluding Paro) which varies between 30,000 to 55,000. TABLE 5.2. SUMMARY OF REFERRAL H OSPITALS Number Number OPD- OPD-visits Admissions Facility Name Admissions of beds of staff visits per staff per staff JDW NRH 350 1,100 514,502 13,480 468 12.3 Eastern RRH 150 380 53,650 5,168 141 13.6 Central RRH 150 460 128,867 4,206 280 9.1 Total 650 1,940 697,019 22,854 359 11.8 Source: HMIS and health facilities. 5.1.1. Total Costs The total costs of referral hospitals are depicted below in Table 5.3. The capital costs are annual depreciation values. In Table 5.3, the total costs according to the costing model’s final cost centers are presented. The total costs at JDWNRH are about five to six times as high as that of the other two referral hospitals. This is expected from the size and level of activity at JDWNRH compared to the other two hospitals. The share of recurrent costs varies from approximately about 72 percent to 85 percent. The cost of drugs and medical supplies at JDWNRH is almost 33 percent of the total costs, whereas it is only 7 and 14 percent at CRRH and ERRH, respectively. JDWNRH is the most sophisticated hospital offering treatments for a number of complex diseases that are not offered at the other hospitals. Thus, a number of advanced and expensive medical therapies and supplies are used at JDWNRH for services not offered at the other facilities. The proportion of capital costs ranges between 15 percent and 28 percent. The highest share of capital cost is due to equipment costs in JDWNRH, whereas in Central RRH and Eastern RRH, building costs constitute the larger share of capital costs— at 22 percent and 18 percent, respectively. 21 TABLE 5.3. TOTAL COSTS FOR REFERRAL HOSPITALS , N U. -- JDWNRH -- -- Central RRH -- -- Eastern RRH -- Nu. % Nu. % Nu. % Recurrent cost 1,029,542,023 84.8 136,534,012 71.8 170,091,248 73.3 • Staff 457,405,000 37.7 108,794,000 57.2 104,048,000 44.9 • Drugs & medical 397,233,023 32.7 12,312,012 6.5 31,903,248 13.8 supplies • Other 174,904,000 14.4 15,428,000 8.1 34,140,000 14.7 Capital cost 184,900,032 15.2 53,528,666 28.2 61,809,208 26.7 • Buildings 66,761,802 5.5 41,547,775 21.9 40,589,472 17.5 • Equipment 111,208,977 9.2 7,361,388 3.9 19,833,886 8.6 • Vehicles 6,929,254 0.6 4,619,502 2.4 1,385,851 0.6 TOTAL COST 1,214,442,056 100.0 190,062,678 100.0 231,900,456 100.0 Source: Data summary from costing model. In Table 5.4 below, the costs are distributed according to the final cost centers identified in the costing model. The cost share of outpatient departments varies from 27 percent at Eastern RRH to almost 48 percent at JDWNRH, with Central RRH slightly higher than Eastern RRH at 34 percent. Taking into consideration the large number of outpatient visits for JDWNRH, that is not surprising. The cost share of inpatient departments varies from 43 percent at JDWNRH to 61 percent at Eastern RRH, with Central RRH at 58 percent. The Maternal & Child Health share of total costs are around 10 percent at both JDWNRH and Eastern RRH, and 6 percent at Central RRH. TABLE 5.4: TOTAL COSTS FOR F INAL COST CENTERS FOR REFERRAL HOSPITALS , N U. -- JDWNRH -- -- Central RRH -- -- Eastern RRH -- Nu. % Nu. % Nu. % Outpatient department 577,759,190 47.6 65,284,136 34.3 61,695,098 26.6 Inpatient department 519,250,645 42.8 109,409,988 57.6 141,733,786 61.1 • All Medical 124,405,392 10.2 34,812,689 18.3 40,911,996 17.6 • Surgical & Medical 394,845,253 32.5 74,597,300 39.2 100,821,791 43.5 Maternal & Child Health (MCH) 117,432,221 9.7 11,057,109 5.8 22,795,681 9.8 Indigenous Unit - - 4,311,444 2.3 5,675,891 2.4 Total 1,214,442,056 100.0 190,062,678 100.0 231,900,456 100.0 Source: Data summary from costing model. 22 5.1.1. Fixed and Variable Costs Figure 5.1 presents fixed and variable costs. The share of fixed costs varies from about 65 percent to 90 percent and the variable costs vary from 10 percent to 36 percent. Not surprisingly, the share of fixed costs is relatively high. Since only variable cost differs with activities, the marginal cost of treating one more patient is relatively low, but up to a point where it would be necessary to recruit additional staff and purchase additional capital items as well. Hence, it can be very cost-effective to fully utilize fixed assets, and in the longer run adapt it to the expected demand for services. FIGURE 5.1. FIXED VS. VARIABLE COSTS FOR REFERRAL HOSPITALS , N U. 1,400 1,200 431.976 1,000 (35.5%) NU, MILLIONS 800 600 783.475 400 46.576 (64.5%) 18.821 (20.1%) (9.9%) 200 171.241 185.324 (90.1%) (79.9%) 0 JDWNRH CENTRAL RRH EASTERN RRH Fixed costs Variable costs Source: Data summary from costing model. 5.1.1. Direct and Indirect Costs As reflected by Figure 5.2, the share of direct costs varies from 33 percent at JDWNRH to 53 percent at the two regional referral hospitals. The relative share of direct vs. indirect costs can give an indication of costing estimate’s precision. As a rule of thumb, the higher the share of direct costs, the higher the precision in the costing estimate, since there would be fewer assumptions on how to allocate costs. Indirect cost calculations are determined from the overhead and intermediate cost centers, and the direct cost calculation is determined from final cost centers. The direct costs of JDWNRH are relatively low at only 30 percent. This is mainly due to the share of costs for drugs and medical supplies (one-third of total costs) that cannot be linked directly to the final cost centers. 23 FIGURE 5.2. DIRECT VS. I NDIRECT COSTS FOR REFERRAL HOSPITALS , N U. 1,400 1,200 1,000 NU, MILLIONS 821.499 800 (67.6%) 600 400 108.814 88.827 (46.9%) 392.949 (46.7%) 200 (32.4%) 101.234 108.814 (53.3%) (46.90%) 0 JDWNRH CENTRAL RRH EASTERN RRH Direct costs Indirect costs Source: Data summary from costing model. 5.1.1. Unit Costs and Disease Specific Costs Unit costs for OPD-visits, admissions, and bed days are presented in Table 5.5 below. The average cost of an OPD-visit is around Nu. 1,100 at JDWNRH and Eastern RRH with the cost of Central RRH being around half of that at Nu. 507. It is worth noting that the total costs at the two regional referral hospitals are relatively close with Eastern RRH at Nu. 232 million and Central RRH at Nu. 190 million. The lower unit cost for OPD-visits at Central RRH can be attributed to the fact that Central RRH has more than double Eastern RRH’s outpatient visits. TABLE 5.5: UNIT COSTS (N U .) AND ACTIVITY FOR REFERRAL HOSPITALS Central Central Eastern JDWNRH Eastern RRH JDWNRH RRH RRH RRH Unit Unit Cost Unit Cost Cost Number Number Number (Nu) (Nu) (Nu) OUTPATIENT DEPARTMENT • OPD-visit 1,123 507 1,150 514,502 128,867 53,650 INPATIENT DEPARTMENT • Admissions 38,520 26,013 27,425 13,480 4,206 5,168 • Bed days 5,508 4,626 4,299 94,277 23,650 32,967 Source: Data summary from costing model. The average cost of an inpatient admission is similar at the two regional referral hospitals being at Nu. 26,013 and 27,425 for Central RRH and Eastern RRH, respectively. The average cost at JDWNRH for an 24 admission is somewhat higher at Nu. 38,520. In Table 5.6 below the number of admissions and cost of each admission are grouped into the major diagnostic groups. TABLE 5.6: DISEASE SPECIFIC I NPATIENT COST PER ADMISSION AND NUMBER OF ADMISSIONS FOR REFERRAL HOSPITALS Unit Cost Admissions Disease Grouping JDWNRH C.RRH E.RRH JDWNRH C.RRH E.RRH INFECTIONS - - - - - - • Diarrhea 17,660 15,620 14,583 145 22 72 • Tuberculosis 99,549 88,046 82,203 210 62 56 • Other infections 21,102 18,663 17,425 108 105 143 VIRAL, PROTOZOAL & HELM DIS. 25,168 22,260 20,782 163 86 136 NEOPLASM 47,240 39,413 - 740 40 42 BLOOD DISEASE 35,347 31,263 29,188 313 198 129 ENDOCRINE, METABOLIC & NUTR. - - - - - - • Diabetes 53,380 47,212 44,079 164 68 119 • Other endocrine etc. 52,708 46,618 43,524 122 27 16 MENTAL DISORDERS 49,718 43,973 41,054 660 84 77 DISEASE OF NERVOUS SYSTEM 44,813 39,635 37,004 281 50 117 EYE & EAR DISEASES - - - - - - • Cataract 52,919 44,152 40,752 19 5 66 • Other Eye & Ear 41,504 34,628 31,961 181 22 93 DISEASE OF CIRCULATORY SYSTEM - - - - - - • Hypertension 29,809 26,365 24,615 242 146 260 • Other circulatory etc. 44,400 37,044 34,191 614 138 241 RESPIRATORY DISEASE - - - - - - • Common Cold 17,733 - 14,643 40 - 13 • Pneumonia 30,905 27,334 25,520 304 72 79 • Other respiratory 35,496 29,615 27,334 761 275 278 DISEASES OF THE DIGESTIVE SYSTEM - - - - - - • Peptic Ulcer Syndrome 27,476 24,302 22,689 66 37 16 • Alcohol Liver Diseases 36,269 32,078 29,949 104 65 126 • Other digestive 28,653 23,906 22,065 1,528 206 313 SKIN DISEASES 43,302 36,128 33,345 195 94 263 DISEASES OF MUSC-SKEL. ETC. 60,460 50,444 46,559 453 39 265 GENITO-URINARY DISEASES 33,988 28,357 26,173 1,085 289 401 PREGNANCY, CHILDBIRTH ETC. - - - - - - • Abortions 16,184 13,502 12,463 212 102 90 • Other pregnancy etc. 19,421 16,203 14,956 1,241 1,102 1,069 PERINATAL CONDITIONS 36,773 30,680 28,318 2,063 360 457 MALFORMATIONS 55,807 46,561 42,976 420 19 26 INJURIES AND TRAUMA 51,408 42,891 39,588 1,046 213 205 OLD CASES 280 ALL 38,520 26,013 27,425 13,480 4,206 5,168 Source: Data summary from costing model. 25 The calculations are based on the unique treatment structure of each hospital with variations in patient case-mix and length of stay. In general, the most expensive admission to treat is TB, ranging from Nu. 82,203 to almost Nu. 100,000 at JDWNRH. These results should be interpreted with caution, especially due to the low number of cases these calculations are based on, in addition to the potential difference in the case-mix of patients. It is a fact that the most complicated cases will be found at JDWNRH, and hence also be more expensive to treat. Finally, the cost of medicine, utilities, expensive diagnostic equipment, etc. have been allocated broadly to disease groupings, not taking the particular disease into account. Hence the main cost driver for an admission ends up being the length of stay for the type of admission. 5.1. District Hospitals The analysis included four district hospitals of various sizes. Paro District Hospital is the largest district hospital in the sample and in the country, where the number of inpatients is higher than each of the three referral hospitals and the number of outpatient visits are likewise higher than the two regional referral hospitals. A summary of the four district hospitals is presented below. TABLE 5.7. SUMMARY OF D ISTRICT H OSPITALS Number Number OPD-visits Admissions Facility Name OPD-visits Admissions of beds of staff per staff per staff Paro 40 140 167,940 15,411 1,200 110 Wangdicholing 20 70 48,763 738 697 11 Damphu 40 130 30,330 3,028 233 23 Punakha 40 100 54,970 1,791 550 18 Total 140 440 302,003 20,968 686 48 Source: HMIS and facilities. The activity data from Paro shows a very high number of OPD-visits and admissions within the time period observed. Considering Paro Hospital is a 40-bed hospital, it is assumed to have a maximum capacity of 14,600 yearly admissions (40 beds times 365 days), given that all beds are fully utilized, and each patient only stayed one day. The data, however, indicate that the number of admissions is 15,411, which is not deemed possible for a 40-bed hospital. The data were double-checked and is consistent with data from the Bhutan Health Information System. This suggests that there may be procedural issues with regards to registering patients at Paro hospital. Since the activity numbers from Paro Hospital is a significant outlier compared to the other three hospitals, and may of be dubious quality, the results from Paro were excluded when determining total average calculations at the district hospital level for calculating unit costs. The specific results from Paro Hospital however are still included in the sections below where the hospitals are more thoroughly described, but it should 26 be noted that due to the potentially inaccurate and very high number of patients at Paro Hospital, the average unit costs are relatively low and most likely incorrect. 5.1.1. Total Costs In Tables 5.8 and 5.9 below, the total costs at district hospitals are presented according to recurrent and capital costs in Table 5.8 and according to the final cost centers of the costing model in Table 5.9. The total cost of the hospitals varies from about Nu. 55 million to Nu. 108 million, with Paro having the highest total cost. It is interesting to note that the cost structures are relatively similar across facilities, with recurrent costs constituting 53 percent to 62 percent and capital costs constituting 38 percent to 47 percent of total costs. In Punakha, of a total recurrent cost of 53 percent, only 7 percent is attributed drugs and supplies, the lowest cost share for this category among all district hospitals in this study. Meanwhile, buildings accounted for over 75 percent of total capital costs, the highest share cost in this category out of all the district hospitals. As indicated in Table 5.9, the cost share of outpatient departments varies from about 32 percent at Paro Hospital to 43 percent at Wangdicholing Hospital. The cost share of inpatient departments varies from about 35 percent at Wangdicholing to almost 54 percent at Paro Hospital. They are also the two hospitals with the highest and lowest number of inpatients. The cost share for Maternal & Child Health are almost the same across facilities, varying from about 10 percent to 14 percent with Paro being lowest and Wangdicholing the highest. TABLE 5.8: TOTAL COSTS FOR DISTRICT H OSPITALS , N U. PARO WANGDICHOLING DAMPHU PUNAKHA Nu. % Nu. % Nu. % Nu. % Recurrent cost 66,581,787 61.7 32,843,098 59.3 58,395,962 58.7 36,903,479 53.4 • Staff 41,804,000 38.8 20,995,000 37.9 25,650,000 25.8 27,514,137 39.8 • Drugs & suppl. 19,493,787 18.1 7,001,098 12.6 26,175,962 26.3 4,884,342 7.1 • Other 5,284,000 4.9 4,847,000 8.8 6,570,000 6.6 4,505,000 6.5 Capital cost 41,271,442 38.3 22,520,504 40.7 41,110,958 41.3 32,221,799 46.6 • Buildings 37,618,204 34.9 19,718,150 35.6 37,142,222 37.3 25,379,068 36.7 • Equipment 1,805,436 1.7 1,416,503 2.6 2,120,935 2.1 4,071,030 5.9 • Vehicles 1,847,801 1.7 1,385,851 2.5 1,847,801 1.9 2,771,701 4.0 TOTAL COST 107,853,229 100.0 55,363,602 100.0 99,506,920 100.0 69,125,278 100.0 Source: Data summary from costing model. 27 TABLE 5.9: TOTAL COSTS FOR F INAL COST CENTERS FOR DISTRICT H OSPITALS , N U. PARO WANGDICHOLING DAMPHU PUNAKHA Nu. % Nu. % Nu. % Nu. % Outpatient 30,849,110 32.2 21,281,249 43.3 32,907,727 37.5 24,920,492 36.1 department Inpatient 51,266,976 53.5 16,997,908 34.6 41,235,685 47.0 32,517,409 47.0 department • All Medical 12,971,204 13.5 5,487,789 11.2 14,619,756 16.7 10,767,166 15.6 • Surgical & 38,295,772 39.9 11,510,119 23.4 26,615,928 30.3 21,750,243 31.5 Medical MCH 9,885,277 10.3 6,961,263 14.2 10,809,747 12.3 8,080,588 11.7 Indigenous Unit 3,886,951 4.1 3,858,547 7.9 2,758,334 3.1 3,606,790 5.2 TOTAL COST 95,888,314 100.0 49,098,967 100.0 87,711,493 100.0 69,125,278 100.0 Source: Data summary from costing model. 5.1.1. Fixed and Variable Costs Figure 5.3 presents fixed vs. variable costs. The share of fixed costs varies from about 70 to 90 percent and the variable costs vary from 10 to 30 percent. The relatively high share of variable costs at Damphu hospital can be attributed to a relatively high cost for medicine and medical supplies. Since data were not available at facility level for the current exercise, cost of medicine at the facility level was determined by dividing the total cost for medicine across all facilities in accordance with facilities’ percentage shares from the 2009/10 exercise. In 2009/10, Damphu had a high total cost for medicine due to a spike in malaria medication, which may not be reflective of its spending on medicine today. Hence a level of caution should be used when interpretating the numbers. 28 FIGURE 5.3. FIXED VS. VARIABLE COSTS FOR DISTRICT H OSPITALS , N U 120 100 21.883 (29.3%) 29.045 6.857 80 29.2%) (9.9%) 9.551 NU, MILLIONS 60 (17.3%) 5.969 40 (79.7%) 70.460 62.627 (70.8%) 45.812 (90.1%) 20 (82.7%) 0 PARO WANGDICHOLING DAMPHU PUNAKHA Fixed costs Variable costs Source: Data summary from costing model. Generally, high-levels of fixed costs will – to a certain point – mean low marginal costs. Thus, the cost of treating one extra patient is much lower than the calculated average cost of treating one patient. The difficulty lies in adapting and implementing an optimal level of activity to appropriately meet the demand of services to a point where fixed assets are fully utilized. 5.1.2. Direct and Indirect Costs In Figure 5.4 below direct vs. indirect costs of the four district hospitals are presented. The share of direct costs varies from about 40 percent at Damphu Hospital to more than 56 percent at Punakha Hospital. The direct costs of Damphu Hospital are the lowest at about 40 percent. This is a result of the relatively high-cost share of drugs that cannot be linked directly to the final cost centers. 29 FIGURE 5.4. DIRECT VS. I NDIRECT COSTS FOR DISTRICT H OSPITALS , N U. 120 100 80 51.936 NU, MILLIONS (54.2%) 52.949 60 (60.4%) 30.257 (43.8%) 40 25.596 (52.1%) 43.951 38.867 20 34.761 (45.8%) 23.501 (56.2%) (60.4%) (47.9%) 0 PARO WANGDICHOLING DAMPHU PUNAKHA Direct costs Indirect costs Source: Data summary from costing model. 5.1.3. Unit Costs and Disease Specific Costs Unit costs are presented in Table 5.10 below. As stated previously, data from Paro Hospital has been excluded from the unit cost calculations for district hospitals due to its questionable data quality (it is unusually high, and its patient numbers do not correspond to its capacity). Looking at the other three district hospitals, it is noted the average cost of an OPD-visit is between Nu. 436 and Nu. 1,085. The highest average unit cost is at Damphu, which also has the lowest number of OPD-visits, undoubtedly driving up the average unit cost. The average cost of an inpatient admission also varies in the three district hospitals (excluding Paro) ranging from Nu. 13,618 to Nu. 23,032. The highest inpatient admission cost is at Wangdicholing, which when compared to the other two hospitals have a relatively low number of admissions. As with the case at Damphu, the lower number of admissions at Wangdicholing drives up the average cost of an admission. No data on the number of bed-days was available for the hospitals and hence no bed-day cost has been calculated. 30 TABLE 5.10: UNIT COSTS AND ACTIVITY FOR DISTRICT HOSPITALS PARO WANGDICHOLING DAMPHU PUNAKHA Unit Unit Unit Unit Activity Activity Activity Activity Cost Cost Cost Cost OUTPATIENT DEPARTMENT OPD-visit 184 167,940 436 48,763 1,085 30,330 453 54970 INPATIENT DEPARTMENT Admissions 3,327 15,411 23,032 738 13,618 3,028 18,156 1,791 Source: Data summary from costing model. In Tables 5.11 and 5.12 below, the number of admissions and associated costs are categorized by diseases, grouped into the major diagnostic labels used in HMIS reports. The very high number of old cases at Damphu hospital without any diagnostics attached (1,471 admissions out of a total of 3,028) meant that it was not feasible to calculate disease specific cost per admission for that hospital. Additionally, as mentioned above, the very low cost per admission for Paro Hospital compared to the other hospitals is presumably due to the very high number of patients at Paro Hospital. In general, as we saw with the referral hospitals, the most expensive disease to treat is TB, with costs ranging from Nu. 50,071 to Nu. 71,189. These results should be interpreted with caution, especially for Table 5.14, noting how many cases the calculations are based on. Additionally, the potential case-mix differences of patients should be taken into consideration. 31 TABLE 5.11: DISEASE SPECIFIC I NPATIENT (UNIT) COST PER ADMISSION FOR DISTRICT HOSPITALS , NU. DISEASE GROUPINGS PARO WANGDICHOLING DAMPHU PUNAKHA INFECTIONS - - - - • Diarrhea 1,905 12,629 - 8,883 • Tuberculosis 10,740 71,189 - 50,071 • Other infections 2,277 15,090 - 10,614 VIRAL, PROTOZOAL & HELM. DIS. 2,715 17,998 - 12,659 NEOPLASM - - - - BLOOD DISEASE 3,813 25,277 - 17,779 ENDOCRINE, METABOLIC & NUTR. - - - - • Diabetes 5,759 38,173 - 26,849 • Other endocrine etc. 5,686 37,693 - 26,511 MENTAL DISORDERS 5,364 35,554 - 25,007 DISEASE OF NERVOUS SYSTEM 4,835 32,046 - 22,540 EYE & EAR DISEASES - - - - • Cataract - 32,250 - - • Other Eye & Ear 3,924 25,294 - 21,472 DISEASE OF CIRCULATORY SYSTEM - - - - • Hypertension 3,216 21,317 - 14,993 • Other circulatory etc. 4,198 27,059 - 22,970 RESPIRATORY DISEASE - - - - • Common Cold 1,913 12,682 - 8,920 • Pneumonia 3,334 22,101 - 15,545 • Other respiratory 3,356 21,632 - 18,363 DISEASES OF THE DIGESTIVE SYSTEM - - - - • Peptic Ulcer Syndrome 2,964 19,649 - 13,820 • Alcohol Liver Diseases 3,913 25,936 - 18,242 • Other digestive 2,709 17,462 - 14,823 SKIN DISEASES 4,094 26,389 - 22,402 DISEASES OF MUSC-SKEL. ETC. 5,717 36,846 - 31,279 GENITO-URINARY DISEASES 3,214 20,713 - 17,583 PREGNANCY, CHILDBIRTH AND - - - - PUERP. • Abortions 1,530 9,863 - 8,373 • Other pregnancy etc. 1,836 11,836 - 10,047 PERINATAL CONDITIONS 3,477 22,410 - 19,024 MALFORMATIONS 5,277 - - - INJURIES AND TRAUMA 4,861 31,330 - 26,596 ALL 3,327 23,032 13,618 18,156 Source: Data summary from costing model. Note: It was not possible to calculate disease specific admission cost for Damphu Hospital due to a high number of old cases with no diagnostic information. 32 TABLE 5.12: DISEASE SPECIFIC NUMBER OF ADMISSIONS FOR DISTRICT HOSPITALS DISEASE GROUPINGS PARO WANGDICHOLING DAMPHU PUNAKHA INFECTIONS 722 5 2 104 • Diarrhea 37 5 12 13 • Tuberculosis 116 5 172 56 • Other infections 77 11 81 10 VIRAL, PROTOZOAL & HELM. DISEASES 13 6 10 17 NEOPLASM 83 16 8 20 BLOOD DISEASE - - - - ENDOCRINE, METABOLIC & NUTR. 92 9 16 14 • Diabetes 6 5 4 5 • Other endocrine etc. 226 16 54 59 MENTAL DISORDERS 318 16 26 39 DISEASE OF NERVOUS SYSTEM - - - - EYE & EAR DISEASES 2 15 9 • Cataract 308 5 4 13 • Other Eye & Ear - - - - DISEASE OF CIRCULATORY SYSTEM 326 25 30 34 • Hypertension 191 40 66 89 • Other circulatory etc. - - - - RESPIRATORY DISEASE 1,790 2 11 42 • Common Cold 71 27 138 189 • Pneumonia 1,371 143 231 177 • Other respiratory - - - - DISEASES OF THE DIGESTIVE SYSTEM 637 6 36 56 • Peptic Ulcer Syndrome 127 58 16 36 • Alcohol Liver Diseases 1,601 65 82 115 • Other digestive 999 16 63 102 SKIN DISEASES 698 16 35 88 DISEASES OF MUSC-SKEL. ETC. 648 62 127 135 GENITO-URINARY DISEASES - - - - PREGNANCY, CHILDBIRTH AND PUERP. 75 9 25 43 • Abortions 606 73 207 127 • Other pregnancy etc. 126 45 4 70 PERINATAL CONDITIONS 2 - - - MALFORMATIONS 3,176 37 97 129 INJURIES AND TRAUMA 967 - 1,471 OLD CASES 15,411 738 3,028 1,791 ALL 722 5 2 104 Source: Data summary from costing model. 33 5.2. 10-bedded Hospitals The analysis of 10-bedded hospitals comprises of two facilities. Compared to the 2009/10 costing exercise, there has been a change in the facilities included. In the 2009/10 exercise, Bajo and Bali hospitals were included. Since then, Bajo has been upgraded to a district hospital and Bali repurposed to a traditional medicine facility. For these reasons they have been excluded from the current study and Kanglung hospital has been added instead. TABLE 5.13 SUMMARY OF 10-BEDDED H OSPITALS OPD- Number Number OPD- visits Admissions Facility Name Admissions of beds of staff visits per per staff staff Gyelposhing 10 20 15,082 474 754 24 Kanglung 10 35 15,628 382 447 11 Total 20 55 30,710 856 558 16 Source: HMIS and facilities. The two facilities have similar patient loads, with Gyelposhing hospital having only a slightly higher number of inpatients compared to Kanglung hospital (15,628 vs. 15,082). However, there are more significant differences in staffing quantity at the two facilities, with approximately 20 staff members at Gyelposhing compared to 35 at Kanglung. 5.2.1. Total Costs The total costs of the two facilities are presented in Table 5.14 below. As explained in the methodology section, the cost of drugs, and medical supplies and equipment, were derived by allocating a share of the facility’s financial year costs, based on their weights from the 2009/10 costing exercise. Since Kanglung is a new facility, the weights from Gyelposhing were used, resulting in identical costs for these items. As outlined above, Kanglung is a larger facility with more staff members. This is also evident from Table 5.14 below, which indicates that the total cost for Kanglung is about Nu. 15 million, compared to about Nu. 10 million for Gyelposhing. Recurrent costs range from 82 percent at Gyelposhing and 84 percent at Kanglung. Capital costs range between 16 percent at Kanglung and 18 percent at Gyelposhing, which is on average lower than the share observed in referral and district hospitals. 34 TABLE 5.14: TOTAL COSTS FOR 10-BEDDED H OSPITALS , N U. GYELPOSHING KANGLUNG Nu. % Nu. % Recurrent cost 8,601,562 82.4 12,853,562 84.2 • Staff 6,278,000 60.2 9,800,000 64.2 • Drugs and 731,562 7.0 731,562 4.8 medical supplies • Other 1,592,000 15.3 2,322,000 15.2 Capital cost 1,831,000 17.6 2,411,552 15.8 • Buildings 1,112,697 10.7 1,693,248 11.1 • Equipment 256,353 2.5 256,353 1.7 • Vehicles 461,950 4.4 461,950 3.0 TOTAL COST 10,432,562 100.0 15,265,114 100.0 Source: Data summary from costing model. Looking at table 5.15 below the total costs are divided into final cost centers of the costing model. For the two facilities the total costs of the outpatient department make up for 35 percent and 39 percent respectively. For inpatients the shares are 46 percent and 43 percent respectively with Gyelposhing having the higher share which matches that Gyelposhing also had more admissions than Kanglung in the covered period. TABLE 5.15: TOTAL COSTS BY F INAL COST CENTER FOR 10-BEDDED H OSPITALS , N U. GYELPOSHING KANGLUNG Nu. % Nu. % Outpatient department 3,604,569 34.6 5,713,697 38.8 Inpatient department 4,760,044 45.6 6,311,440 42.8 • All Medical 2,775,048 26.6 2,006,028 13.6 • Surgical & Medical 1,984,995 19.0 4,305,412 29.2 Maternal & Child Health (MCH) 1,510,432 14.5 1,816,305 12.3 Indigenous Unit 556,808 5.3 897,388 6.1 Total 10,431,853 100.0 14,738,830 100.0 Source: Data summary from costing model. 35 5.2.2. Fixed and Variable Costs Figure 5.5 presents fixed vs. variable costs of the two facilities. The share of fixed costs is 83 percent and 89 percent for Gyelposhing and Kanglung, respectively. As also noted in previous sections of the report, the high levels of fixed costs will, to a certain point, result in a relatively lower marginal cost. FIGURE 5.5. FIXED VS. VARIABLE COSTS FOR 10-BEDDED H OSPITALS , N U. 18 16 1.643 14 (10.8%) NU, MILLIONS 12 1.751 10 (16.8%) 8 13.621 6 (89.2%) 8.681 4 (83.2%) 2 0 GYELPOSHING KANGLUNG Fixed costs Variable costs Source: Data summary from costing model. 5.2.3. Direct and Indirect Costs In Figure 5.6 below, direct vs. indirect costs of the two facilities are presented. The share of direct costs is similar at 55.1 and 56.5 percent for Gyelposhing and Kanglung, respectively. FIGURE 5.6. DIRECT VS. I NDIRECT COSTS FOR 10-BEDDED H OSPITALS , N U. 16 14 12 6.407 NU, MILLIONS 10 (43.5%) 4.688 8 (44.9%) 6 8.331 4 5.743 (56.5%) 2 (55.1%) 0 GYELPOSHING KANGLUNG Direct costs Indirect costs Source: Data summary from costing model. 36 5.2.4. Unit Costs and Disease Specific Costs Unit costs are presented in Table 5.16 below. The average cost of an OPD-visit is Nu. 239 at Gyelposhing and Nu. 366 at Kanglung. For inpatients, the average cost of an admission is Nu. 10,042 and 16,522 for Gyelposhing and Kanglung,respectively. As also previously mentioned, it is observed that the hospital with the lowest staffing (and total costs) is also the hospital with the highest number of admissions, despite the number of outpatients being almost identical. TABLE 5.16: UNIT COSTS AND ACTIVITY FOR 10-BEDDED H OSPITALS GYELPOSHING KANGLUNG Unit Unit Activity Activity Cost Cost OUTPATIENT DEPARTMENT OPD-visit 239 15,082 366 15,628 INPATIENT DEPARTMENT Admissions 10,042 474 16,522 382 Source: Data summary from costing model. In Table 5.17 below the number of admissions and cost of these are grouped into the major diagnostic groups following the labels and format used in HMIS reporting. As mentioned in the above sections on both referral and district hospitals, the results on disease specific admissions costs should be interpreted with caution, especially for 10-bedded hospitals as many of the calculations are based on a small number of admissions. 37 TABLE 5.17: DISEASE SPECIFIC NUMBER OF ADMISSIONS FOR 10 BEDDED HOSPITALS Costs Admissions DISEASE GROUPING GYELPOSHING KANGLUNG GYELPOSHING KANGLUNG INFECTIONS - - - - • Diarrhea 7,069 10,568 31 20 • Tuberculosis - - - - • Other infections 8,447 12,628 1 40 VIRAL, PROTOZOAL & HELMINTHIC DIS. 10,074 15,061 1 2 NEOPLASM - - - - BLOOD DISEASE 14,149 21,153 1 10 ENDOCRINE, METABOLIC & NUTR. - - - - • Diabetes - 31,945 - 6 • Other endocrine etc. 21,099 31,542 6 3 MENTAL DISORDERS 19,902 29,753 17 7 DISEASE OF NERVOUS SYSTEM 17,938 26,817 27 4 EYE & EAR DISEASES - - - - • Cataract - - - 1 • Other Eye & Ear 14,336 21,145 1 1 DISEASE OF CIRCULATORY SYSTEM - - - - • Hypertension 11,932 17,839 5 10 • Other circulatory etc. 15,336 22,620 3 14 RESPIRATORY DISEASE - - - - • Common Cold 7,099 - 97 - • Pneumonia 12,371 18,495 41 8 • Other respiratory 12,260 18,084 16 43 DISEASES OF THE DIGESTIVE SYSTEM - - - - • Peptic Ulcer Syndrome 10,999 16,443 25 2 • Alcohol Liver Diseases 14,518 21,704 3 4 • Other digestive 9,897 14,598 52 12 SKIN DISEASES 14,956 22,061 6 21 DISEASES OF MUSC-SKEL. ETC. 20,883 30,802 9 4 GENITO-URINARY DISEASES 11,739 17,316 30 29 PREGNANCY, CHILDBIRTH AND PUERP. - - - - • Abortions - 8,245 - 7 • Other pregnancy etc. 6,708 9,894 5 67 PERINATAL CONDITIONS - - - - MALFORMATIONS - - - - INJURIES AND TRAUMA 17,756 26,191 31 46 OLD CASES 66 21 ALL 10,042 16,522 474 382 Source: Data summary from costing model. 38 5.3. Primary Health Centers The analysis of Primary Health Centers included three facilities that were all utilized in the 2009/10 exercise as well. A summary of the facilities is presented below. TABLE 5.18. SUMMARY OF P RIMARY H EALTH CENTERS Numbe Numbe OPD- OPD- Admission Facility Name r of r of Admissions visits per visits s per staff beds staff staff Genekha 0 2 3,000 0 1,500 NA Mendelgang 0 5 5,954 0 1,191 NA Thinleygang 0 4 6,170 0 1,543 NA Total 0 11 15,124 0 1,375 NA Source: HMIS and facilities. 5.3.1. Total Costs In Table 5.19 below, total costs of the three facilities are presented. The total costs vary from Nu. 1.8 million to Nu. 3.9 million, with Genekha having the lowest cost and unsurprisingly the lowest number of patients seeking services. TABLE 5.19: TOTAL COSTS ( COST STRUCTURE ) FOR PRIMARY H EALTH CENTERS, N U. GENEKHA MENDELGANG THINLEYGANG Nu. % Nu. % Nu. % Recurrent cost 1,227,146 66.6 3,273,710 84.2 2,841,153 81.8 • Staff 700,000 38.0 1,825,000 46.9 1,068,660 30.8 • Drugs and 309,146 16.8 1,087,710 28.0 830,182 23.9 medical supplies • Other 218,000 11.8 361,000 9.3 942,311 27.1 Capital cost 615,069 33.4 615,069 15.8 631,041 18.2 • Buildings 468,504 25.4 468,504 12.0 484,476 14.0 • Equipment 146,565 8.0 146,565 3.8 146,565 4.2 • Vehicles - - - - - - TOTAL COST 1,842,215 100.0 3,888,779 100.0 3,472,194 100.0 Source: Data summary from costing model. 39 Recurrent costs make up 66.6 to 84.2 percent of the total costs. There is no data available on equipment costs for primary health care centers. This was also the case in the 2009/10 exercise, where it was assumed that the annual depreciation cost of equipment was Nu. 100,000. The same number has been used and regulated to 2018/19 prices. Table 5.20 below shows the total costs for final cost centers. A major part of the work taking place at primary health centers is related to health promotion and prevention activities, reflected by relatively high shares of the MCH cost centers, which vary from 24 to 38 percent. This is considerably higher than any of the higher tier facilities. The cost share of outpatient departments varies from about 62 percent at Thinleygang to 76 percent at Genekha. TABLE 5.20: TOTAL COSTS FOR FINAL COST CENTERS FOR PRIMARY H EALTH CENTERS, N U. GENEKHA MENDELGANG THINLEYGANG Nu. % Nu. % Nu. % Outpatient department 1,400,238 76.0 2,734,015 70.3 2,139,720 61.6 Inpatient department - - 117,696 3.0 - - • All Medical - - 117,696 3.0 - - • Surgical & Medical - - - - - - Maternal & Child Health (MCH) 441,977 24.0 1,036,391 26.7 1,332,004 38.4 Indigenous Unit - - - - - - Total 1,842,215 100.0 3,888,103 100.0 3,471,724 100.0 Source: Data summary from costing model. Note: Although PHC do not have inpatients many have a few beds available. In the case of Mendelgang the facility reported they spend a fraction of their time in their inpatient department and hence a small cost has been attributed to it. 40 5.3.2. Fixed and Variable Costs From Figure 5.7 below, it can be seen that fixed costs vary from 52.4 percent to 75.1 percent. The relatively low share of fixed costs at Thinleygang at 52.4 percent, can mainly be attributed to a high transport cost that is considered a variable cost. FIGURE 5.7. FIXED VS. VARIABLE COSTS FOR PRIMARY H EALTH CENTERS, N U. 5 4 NU, MILLIONS 1.197 3 (30.8%) 1.652 0.456 (47.6%) 2 (24.9%) 2.691 1 (69.2%) 1.819 1.383 (52.4%) (75.1%) 0 GENEKHA MENDELGANG THINLEYGANG Fixed costs Variable costs Source: Data summary from costing model. 5.3.3. Direct and Indirect Costs The composition direct and indirect costs are presented in Figure 5.8 below. The share of direct costs ranges between 36.7 percent at Mendelgang to 52.1 percent at Genekha. FIGURE 5.8. DIRECT VS. I NDIRECT COSTS FOR PRIMARY H EALTH CENTERS, N U. 5 4 3 NU, MILLIONS 2.460 2.019 (63.3%) (58.2%) 2 0.882 1 (47.9%) 1.428 1.452 0.959 (36.7%) (41.8%) (52.1%) 0 GENEKHA MENDELGANG THINLEYGANG Direct costs Indirect costs Source: Data summary from costing model. 41 5.3.4. Unit Costs Unit costs for an outpatient visit for the three facilities are presented in Table 5.21 below. The unit cost varies from Nu. 347 at Thinleygang to Nu. 467 at Genekha. TABLE 5.21: UNIT COSTS FOR PRIMARY H EALTH CENTERS, N U. Unit Cost OPD-visits GENEKHA MENDELG. THINLEYG. GENEKHA MENDELG. THINLEYG. OUTPATIENT DEPARTMENT OPD-visit 467 459 347 3,000 5,954 6,170 Source: Data summary from costing model. Genekha has the highest average cost of an outpatient visit, but also only about half the patient load compared to the other two facilities. A general observation from the study is that the unit costs are sensitive to the flow rate of patients at the facilities. A sample of more than one facility per category is used to mitigate this potential sensitivity. 42 6. COST CHANGE 43 This costing exercise builds on the same costing model as the exercise that was undertaken for 2009/10. It is thus possible to compare the results directly and see the change in costs across the 9- year time period. It is well-known that the cost of healthcare increases over time, and also when observing fixed prices. Healthcare quality and costs develop with more advanced treatments and therapies, as well as diagnostic capabilities. When compared to years ago, more efficient and higher quality equipment, medicine and other therapeutic instruments now exist. This increase in quality naturally comes with an increase in price and Bhutan is no exception, as the data below will reveal. The change in costs is presented in Table 6.1 and Figure 6.1 below. The change data presented focuses on total costs and on unit-costs within the observation periods. The 2009/10 numbers have been price-regulated to 2018/19 prices using the Bhutan non-food consumer price index for the period in question. The changes seen in the tables below are thus real changes. For the change in total costs, we only look at referral and district hospitals since the composition of facilities has only changed slightly for 10-bedded hospitals. Here we see a fixed price increase in total costs of 62 percent for referral hospitals and 54 percent for district hospitals. The increases are especially prominent in staff costs that have increased from 87 to 93 percent within the period. TABLE 6.1: TOTAL COST COMPARISON OF 2009/10 AND 2018/19 COSTING EXERCISE , N U. (FIXED 2018/19 PRICES ) 2009/10 2018/19 Referral District District Referral Hs Hs Hs Hs Recurrent Cost 723,748,957 123,607,212 1,336,167,283 194,724,327 • Staff 359,378,860 60,192,056 670,247,000 115,963,137 • Drugs and medical supplies 249,489,493 41,682,024 441,448,283 57,555,190 • Other 114,880,604 21,733,131 224,472,000 21,206,000 Capital Cost 285,899,565 92,424,135 300,237,906 137,124,703 • Buildings 155,271,766 76,334,188 148,899,049 119,857,644 • Equipment 113,024,301 9,996,429 138,404,250 9,413,905 • Vehicles 17,603,498 6,093,518 12,934,607 7,853,154 TOTAL COST 1,009,648,522 216,031,347 1,636,405,189 331,849,030 Source: Data from 2009/10 costing model and 2018/19 costing model 44 FIGURE 6.1. COMPARISON OF TOTAL COSTS FROM 2009/10 TO 2018/19, N U. (FIXED 2018/19 PRICES ) 95% 93% 85% 87% 77% 58% 57% 48% 38% 29% 22% 5% -4% -2% -6% -27% REFERRAL HOSPITALS DISTRICT HOSPITALS Recurrent cost-total Staff (Recurrent) Drugs and medical supplies (Recurrent) Other (Recurrent) Capital cost-total Buildings (Capital) Equipment (Capital) Vehicles (Capital) Source: Data from 2009/10 costing model and 2018/19 costing model The total cost increase within the period of observation is very high, and is primarily driven by increased salary costs. Additionally, there has also been a considerable increase in activity at the facilities in terms of both out-patients as well as admissions. This is reflected in an increase, albeit lower, in the unit costs, as illustrated in Table 6.2 below. The unit-prices at all levels, and for all services have increased, in fixed prices. There is a significant pattern in which the highest cost increases are found at lower-level facilities. This is highlighted by the fact that an average OPD-visit at a primary health care center increased by 75 percent, compared to only 9 percent at a referral hospital. TABLE 6.2: UNIT COST COMPARISON OF 2009/10 AND 2018/19 COSTING EXERCISE , N U. (FIXED 2018/19 PRICES ) OPD-visit Admission Bed day Referral Hospitals 2009/10 930 25,435 4,097 2018/19 1,011 33,709 5,106 District Hospitals 2009/10 450 14,827 2018/19 590 16,331 PH Center 2009/10 236 2018/19 415 Source: Data from 2009/10 costing model and 2018/19 costing model Note: 10-bedded hospital not included due to different sample of facilities. 45 FIGURE 6.2.: COMPARISON OF UNIT COSTS FROM 2009/10 TO 2018/19, N U. (FIXED 2018/19 PRICES ) 75% 33% 31% 25% 9% 10% 0 REFERRAL HOSPITALS DISTRICT HOSPITALS PRIMARY HEALTH CENTER OPD-visit Admission Bedday Source: Data from 2009/10 costing model and 2018/19 costing model The average cost of an admission has increased by 33 percent at referral hospitals, but only by 10 percent at district hospitals. In the 2009/10 exercise, the lowest average unit cost for an OPD-visit was found at the lowest level of facility – the primary health care center. This is no longer the case with an increase in unit price at 75 percent. A reasonable explanation for this is related to patient volume. Whereas other health facility levels have experienced an increase in patient volume, OPD visits at primary health care centers have actually fallen in the period, which has then driven up the average unit cost. 46 7. LIMITATIONS & ASSUMPTIONS 47 When conducting a costing study such as this there will naturally be limitations to data and assumptions and estimates will have to be made. Key limitations and assumptions are listed in this chapter and are further elaborated in Annex A. The design of a costing study based on a uniform model has its own natural limitations and utilizes several assumptions. Though it also has many strengths in its approach, it is important to understand and consider its limitations when analyzing the results of such a study. One of the main strengths of this study design is that by basing the costing on a uniform model, comparisons between facilities at both the same and different levels can be made. Additionally, they are all subject to the same assumptions and limitations. This uniformity allows a seamless way to update the analyses with newer data, and to include other facilities within the country. It also allows for an analysis of changes over time, such as is the case of this exercise which compared results with those from the previous 2009/10 costing exercise. The costing study is to a high degree based on data that is regularly registered and maintained at health facilities and at the central Ministry of Health. This availability and uniformity of data is an obvious strength of the study, but on the other hand, this data dependency also makes the accuracy of results very sensitive to data quality issues, which is further discussed below. The main assumptions and limitations of the study are: • Quality of services has not been assessed. The focus of the study has been on quantification in terms of cost of health services at various levels of service delivery. The quality of the health care services, or the outcome of care has not been assessed. • Model design treats health facilities as uniform. The uniformity of the model design treats the facilities as uniform entities producing the same output. Even though there is a high degree of uniformity of health facilities in Bhutan, there are still differences among facilities. These issues need to be taken into consideration when interpreting the results. • Top-down model design. The top-down design of the model allows for a macro-perspective of cost structures and the unit costs of various facilities. This includes an overview of cost from total costs down to individual service delivery departments, according to identified categorizations. However, this broad and sweeping approach may lose out on smaller details. • Unit Cost sensitivity. As evident in the results analysis, the calculated unit costs are very sensitive to the activity level and number of patients at a facility. This is evident from the very high number of patients at Paro Hospital and its associated low unit costs. This relationship is also the reason why the unit cost of OPD-visits at primary health care centers have increased by 75 percent in real prices compared to the 2009/10 exercise. As the number of patients decreased, the unit cost increased. • Limited validation of results. The exercise was initiated just prior to the COVID-19 pandemic, which delayed the finalization of the exercise. The travel restrictions imposed by the pandemic 48 limited the possibility to do on-site follow-up visits to facilities to further validate the data collected, as well as the subsequent results to qualify the costing tool. • Data issues. The costing study is very data intensive. All results are based on data from health facilities, MOH and district authorities. This reliance on data also means it is sensitive to the quality of data. While collecting, processing and analyzing the data, various issues became apparent. These are summarized below. o Data availability. Not all the data needed were available, and estimations had to be made to fill in gaps. This adds a level of uncertainty to the results. o Output data quality (data on patient visits and admissions). The main output data source is the Bhutan Health Management Information System (BHMIS). In addition to this, detailed patient records from JDWNRH hospital were available and utilized. As outlined previously, the unit cost results are very sensitive to the number of patients at the facilities. Although the BHMIS provides detailed data on activities conducted at the facilities, varying quality and consistency of staff’s registration practices within facilities can lead to issues with data quality and validity. 49 8. DISCUSSION & RESULTS ANALYSIS 50 The results provide a detailed insight into the cost structures (unit cost and total cost) of health services delivered: (i) within the same facility, (ii) among facilities within the same levels, and (iii) among facilities at different levels in Bhutan. In general, there are significant differences in cost-structures between facility types at different levels, including the same level of facilities. As a rule of thumb, the higher the total cost at the facility, the higher the amount of activity in terms of the treatment of patients. Generally, the recurrent costs are considered a more significant cost driver compared to capital costs. Additionally, a high degree of the costs at the facilities are fixed costs, meaning that these costs do not in the short- or medium term vary with the activity types and levels at the facility. The costing exercise results suggest that a higher level of health facility is associated with higher costs in providing health services (when compared to other facilities at lower levels). The OPD unit costs of services at referral hospital are significantly more costly than facilities at lower levels. Even in terms of scale of services, the higher facilities also provide more comprehensive health services, and are equipped with advanced medical equipment and skilled health workforce as per the National Human Resource and Service standards. Facilities at the lowest level, however, may not always have the lowest unit cost. Specifically, the OPD unit cost of primary health centers—the lowest level among the facilities in the study —is not necessarily the lowest in terms of cost. Services at primary health centers are more expensive than at 10-bedded hospitals. This may be attributed to factors related to economies of scale and low utilization volume at some sites. For example, Geneka primary health center has the highest OPD unit cost while having the lowest number of patients visiting after Mendelgang primary health center. There is also a difference in the composition of cases treated at the facilities, which is more prominent between facilities of different levels, with more complicated cases and treatments being carried out at higher level facilities. Despite this, the costing exercise suggest that it is more cost effective to access health services from lower-level health facilities than to access the same services from higher-level facilities, specifically in the cases where the required services are available at lower-level facilities. However, studies have shown that patients prefer to receive care at higher-level facilities with the expectation of receiving higher quality services, despite lower-level facilities potentially providing the same level of care (Wangmo et al. 2018; Yangchen et al. 2016). In the case that lower-level facilities do provide the same level of care, public advocacy and awareness programs should be conducted to promote its utilization to avoid over capacity of higher-level health facilities. This would also lead to a more cost-effective use of resources, enabling the treatment of more people. Furthermore, services provided at higher-level facilities which are also provided and within the capacity of lower-level facilities should be shifted to lower-level facilities in a progressive manner for cost savings and efficient service delivery. It is well-known that the cost of healthcare services increases over time, and this is consistent when looking at fixed prices. Healthcare facilities quickly adapt and adopt more advanced treatments, therapies, and diagnostic capabilities as they become mainstreamed. Today we have more efficient 51 and higher quality equipment, medicine and other therapeutic instruments than we did 10 or more years ago. This has also been the case in Bhutan. Since the last costing exercise almost 10 years ago, the total costs as well as the average unit costs have increased, even when taking into consideration differences in price levels and inflation. One of the main drivers for the increase in cost is not only the cost of personnel, but also the cost of medicine and medical consumables and equipment. As such, Bhutan has experienced an increase in total costs since 2009/10 of about 62 percent for referral hospitals and 54 percent for district hospitals. These increases are mainly due to rising staff costs which has increased up to 93 percent in terms of fixed prices since 2009/10. Although the total cost increase in the observation period is primarily driven by increased salary costs, there has also been a considerable increase in activity levels at the facilities in terms of both outpatients and admissions. This is reflected in an increase, albeit lower, in unit costs. These findings and results have clear potential to inform decisions makers and be used in further analytical work to increase the health sector’s efficiency and value-for-money. 8.1. How can the costing results be used? Knowing the cost and cost structures of healthcare activities is crucial to make many informed policy decisions. In this section, a number of key potential uses of the results are presented. However, it is beyond the scope of this report to delve into each one. 8.1.1. Planning purposes and budget allocations Having knowledge of the cost structures of health facilities is invaluable for planning and developing budget allocations. Even without looking at the unit costs, the total costs of various facilities, especially when compared with patient load at these facilities can be used to optimize and increase both the effectiveness and efficiency of future budget allocations. The results of the exercise have shown many examples in which a mismatch between the total costs of the facility and the number of patients seen at the facility exists. There are examples of facilities with lower total costs but that see a higher flow of patients than other similar facilities with higher costs. One valid reason for such a case could be that facilities in more remote areas simply have smaller catchment populations. With the costing results, additional analysis in these areas can be undertaken to make more informed planning and budget allocation decisions. 8.1.2. Disease burden analysis Knowing the cost of services for various disease groupings allows for further analytical work into more specific burden of disease studies. Having this knowledge can help, from a cost perspective, to prioritize health resources to fight selected and priority diseases. It can also inform cost studies of health promotion and prevention initiatives targeting, for example, certain lifestyle related diseases or specific infectious diseases. 52 8.1.3. Introduction of payment for services In case Bhutan wants to introduce individual payment for healthcare services, the costing exercise can also help to inform price-setting procedures for different services. However, Bhutan has a history of providing health services for free, and this is unlikely to change under its decree to secure basic health care access to all peoples. However, there are some elements of a payment for services scheme that could be considered or is already in effect. For example, the off-hours clinic at JDWNRH allows patients to receive prioritized care or skip the wait queue by paying a fee. The costing exercise could inform these initiatives and help set a price that reflects the actual cost of associated services. In many countries with free healthcare, foreigners are charged for health services. In the case that Bhutan wishes to conduct similar measures, the results from the costing exercise can help set prices for these services or aid in designing private insurance schemes for visitors. 8.1.4. Results-based financing and increasing efficiency The costing results can be used in designing result-based financing, performance-based financing, or activity-based financing schemes for public health facilities. A common practice for these types of schemes is that financial incentives are given to facilities that deliver specific services at a desired level of quantity or quality. These incentives can have many forms. One is by financing the facility according to a fixed cost for every patient treated or service provided. This incentivizes greater cost-efficiency and lower service delivery costs. Another form could be to reward quality services with financial rewards. The possibilities are abundant and are all aimed at increasing the efficiency of service delivery. The results have a clear potential to be used to inform decision makers, and to be used in further analytical work to increase the efficiency and value-for-money of the health sector in Bhutan. The costing study can also be used as a tool to measure the efficiency of health facilities. This could be done by using the costing results to set an expected unit cost for various services, which can then be compared against the observed costs at the facility level. If the observed facility-level cost is lower than the country average, then the facility may be considered more efficient and if it is higher, then the reverse is true. It should be noted that ‘efficiency’ in this case is defined as strictly output in relation to input and does not capture the quality component of service delivery. 8.1.5. Improve Bhutanese referral system The patient flow has increased in Bhutan in the past ten years beyond what can be attributed to the country’s changing demographics. In the 2009/10 exercise there was a demonstrated tendency that people sought out services at higher-level facilities. The current exercise supports this trend. There is additional evidence that the level of activities at primary health care centers are lower in the current exercise than in the 2009/10 exercise, despite a total increase in the numbers of patients. 53 The results generally show that the higher the level of facility, the higher the cost of services. From a strictly financial point of view, the most effective point to receive service delivery is where the cost is lowest. Hence, if a patient can receive the same services at a 10-bedded hospital, it would not be efficient or desirable to receive the same service at a more expensive referral or district hospital. The knowledge of the various costs across the different types of facilities can help inform the design of effective referral systems, with the right incentives and/or procedures to guide patients to the most appropriate facility for their needs. Doing so would not only be prudent from a cost perspective, but would also help to minimize bottlenecks in the system and shorten queues for patients seeking out services. 8.1.6. Input to Health Technology Assessments (HTAs) of new treatments and therapies The utilization of Health Technology Assessments (HTAs) is a commonly recognized method to evaluate new treatments, therapies, and changes in clinical practices. Besides the clinical considerations, HTAs also include economic evaluations of the intervention being studied. The costing exercise can assist in qualifying these economic evaluations, for HTAs undertaken in Bhutan. These costing studies are already being utilized for this purpose, and new and updated cost figures can qualify these evaluations even further. 54 9. CONCLUDING REMARKS 55 The 2018/19 costing study served a number of objectives. This includes informing the Ministry of Health about the cost of delivering various types of services at different levels, increasing general cost- awareness and knowledge, and supporting future evidence-based and informed policy decisions to increase the efficiency of the Bhutanese health sector. The report’s results cover the above objectives, and present the following summarized main observations: • Costs and cost-structures at health facilities. The study gives a detailed insight into total costs, cost-structures and composition of costs of facilities at various levels. • Average Unit costs. The costs of OPD-visits, admissions in general, as well as disease-specific groupings of admissions have been calculated, showing that services are generally more costly at higher facility levels. For example, the average cost of an OPD-visit at a referral hospital is three times higher than at a 10-bedded hospital. Likewise, the average cost of an admission is twice as high at a referral hospital than at a district hospital. • Cost increases over time. The cost of health services increases over time. Compared to the costing exercise in 2009/10, the total costs (measured in the same price levels) have increased by more than 50 percent for referral hospitals and district hospitals. An increase with broader variations is also observed in the average unit costs of OPD-visits and admissions. • The lowest level of service delivery is the most cost-effective. Providing services at the lowest effective level of service delivery (facilities) is the most cost-effective way of providing services in the long run. This means that if the service is available at a 10-bedded hospital, it is more cost-effective to utilize the service there than seeking out the same service at a district hospital or a referral hospital. • Results should be interpreted with caution. When interpreting the results, several factors should be taken into consideration. The top-down costing model has some weaknesses and the quality and precision of some of the data used for the exercise calls for a certain degree of caution when interpreting the results. Some of the key issues are data quality and the model’s high sensitivity to the level of activity at the facility when calculating the average unit costs. This study is the second comprehensive costing analysis in Bhutan and follows the same model as the exercise undertaken almost 10 years ago. The methodology has been described in detail in the report and annex and the standard costing model can be used to further include other facilities, and to update costing data in order to analyze changes in the cost of services. There is room to improve the model, and especially room to improve the quality of data and data sources used. The study points towards a number of ways in which the results from the costing study can be used in Bhutan to improve efficiency in the health sector. The two areas assessed as having the highest potential are: to use the results to improve referral pathways and as input and learnings to further optimize planning and budgetary allocations. 56 57 ANNEX A: DETAILED COSTING METHODOLOGY COSTING THEORY AND METHODOLOGY This annex presents a number of basic concepts of costing theory in relation to healthcare, which forms the foundation of the costing model developed and used for the costing analysis. In this context, it is essential to distinguish between different types of methods for conducting health economic analyses. One method is health economic evaluation, which analyzes the costs and benefits of various interventions. In this method, various health economic evaluation frameworks are utilized to measure costs, with the goal of answering the question "what does it cost?". This cost is then weighed against an effect that can be measured, which can be done in different ways depending on the type of analysis conducted (Alban et al. 1999). Another method of health economic analysis is cost analysis – or commonly referred to in the international literature as "cost-accounting (Flessa 2009; Mogyorosy and Smith 2005; Drummond et al. 2005; Tan et al. 2011). This method seeks to identify, quantify and evaluate resources used to provide different types of healthcare services. The purpose of this costing study is not to compare costs against their effects, as done in health economic evaluation. The study focuses only on considerations of cost, excluding impact, with the goal to compare costs across different types of organizational units (i.e., health centers and hospitals). This costing study thus utilizes the particular branch of health economic analysis called "cost- accounting” as its theoretical foundation for analysis. Costs Costing theory distinguishes between different types of costs. The ones relevant for this analysis are presented in the box below, which also provides examples of the individual types of costs. In the short term, a distinction is made between fixed and variable costs, which together constitute the total cost. The fixed costs (i.e. staff and buildings) are independent of the production/activity at the facility, which can, for example, be measured as the number of admissions. On the other hand, the variable costs (i.e., medicines) are costs that vary with the volume of production. 58 Average costs are calculated as the total (per unit) fixed and variable costs. Marginal costs only constitute the cost increase of an increase in production (i.e., patients treated) and thus do not affect fixed costs. Finally, the discipline of cost-accounting makes a distinction between direct and indirect costs. Direct costs are costs that contribute directly to the production of services, while indirect costs are not directly associated with production but rather associated with various types of overhead and administrative costs (Drummond et al. 2005). It is very important to note here that direct and indirect costs have different meanings in cost- accounting and health economic evaluation. In the field of health economic evaluation, indirect costs instead refer to an expression of loss of production or loss of the value added to society (Drummond et al. 2005). KEY COSTING TERMS • Total Costs = Fixed costs + Variable Costs • Fixed Costs = independent of production/services (buildings, fixed staff) • Variable Costs = increases with production/services (medicine, utensils, care) • Total Average Costs = Total Costs / Number of produced units • Marginal Costs = Cost increase of a small production increase • Direct Costs = Costs that can be attributed directly to the production of services / the production site (medicine, equipment, staff at service delivering departments, etc.) • Indirect Costs = Costs that cannot be attributed directly to the production of services / the production site – overhead costs and costs of cross-cutting service centers (administration, security, x-ray, laboratories, etc.) Note: The concept of Direct and Indirect Costs are used differently in the literature depending on whether it is "cost accounting" or "health economic evaluation". For this analysis the definition for Cost Accounting was used. Methods for Cost Analysis Mogyorosy and Smith (2005) have conducted a major literature analysis of various methods of cost analyses in health care. The study specifically analyzes different methods to estimate the costs associated with providing different services for both inpatients and outpatients at health facilities. The analysis drew upon published scientific literature in the field and summarized common features and differences of the methods used. Mogyorosy and Smith concluded that a general consensus exists amidst the studies, regarding the basic principles and elements of a cost analysis. A cost analysis exercise is said to consist of the following elements: 59 Analysis frame 1) A well-defined decision-making problem that includes the purpose of the cost analysis, perspective of the analysis and the time horizon 2) Description of the services provided that are the object of the analysis (cost object) Costing method 1) Identification of resources used to provide services 2) Quantifying resource consumption in natural units 3) Coupling of value to resource consumption There are several different basic cost analysis methods that can be used. There are top-down approaches that measure gross costs, and through a series of steps, allocate these to different types of productions (Mogyorosy and Peter 2005). Another method is the bottom-up method, or micro- costing method, which selects representative examples of specific types of services and measures resource consumption for them at the level at which the services are provided. This is also often referred to as Activity Based Costing (Mogyorosy and Smith 2005). Drummond et. al. (2005) establishes these two cost analysis approaches in a continuum of precision, with the bottom-up method being the most precise but also the most demanding and resource- intensive method to use (Drummond et al. 2005). Flessa (2009) points out that in developing countries it is generally very difficult to use the bottom-up approach simply because the available data often do not have the quality necessary to make such an analysis. Additionally, it would take considerable resources to do this at all selected facilities. Given these considerations, it is deemed most appropriate for this analysis to adopt a top-down approach. Furthermore, this was also the approach used for the 2009/10 analysis, which allows comparisons across time possible. Methodology and Data Collection In the above section, the theoretical starting point for the cost analysis was presented together with the standard usual steps of such an analysis. This section then describes the methodology used based on the steps presented by Mogyorosy and Smith (2005) and (Christiansen 2007). The five steps are presented in the sections below. They are: 1. A well-defined decision-making problem that includes the purpose of the cost analysis, perspective of the analysis and the time horizon 2. Description of the given services that are the object of the analysis (cost object) 3. Identification of resources used to provide the services provided 4. Quantifying resource consumption in natural units 60 5. Coupling of value to resource consumption Decision-making problem, perspective, and time horizon The decision-making problem should be identified thoroughly. This is presented in the introduction to this report. A cost analysis can be done from several perspectives. It can be from a socio-economic perspective, patient perspective, facility perspective, or other perspectives The perspective chosen determines the various cost indicators that would be included in the analysis (Mogyorosy and Smith 2005; Drummond et al. 2005; Christiansen 2007). For example, transport costs to/from the health facility are relevant from a patient perspective, while irrelevant from a facility perspective, unless transport takes place in a facility-owned ambulance. The purpose of the study should be the determining factor in defining the perspective to be taken for the analysis. Since the purpose of this study is to compare differences in costs of services across different types of health facilities, the most relevant perspective to be taken would be the facility perspective. Specifically, we are interested in the costs associated with providing services for each health facility. The adoption of these perspectives naturally excludes a number of other cost indicators. These include but are not limited to transport costs for patients, costs associated with informal care, costs of a number of vertical government disease programs run directly by the Ministry of Health, and the cost of treatments abroad. Likewise, the costs of the central monitoring and supervision that Bhutan's Ministry of Health conducts are also not included. The general recommendation on the time horizon for a cost study is one year (Mogyorosy and Smith 2005; Jegers et al. 2002). Some data to be used in the analysis are also only available on an annual basis. Extending the study to several years is described in the literature as problematic (Mogyorosy and Smith 2005), as it introduces additional methodological dilemmas in terms of measuring the cost of present value. Thus, the time horizon chosen for this costing study is one year, the same time period of the previous 2009/10 costing analysis. This costing analysis looks at the 2018/19 Financial Year. Cost object The cost analysis shall relate activities undertaken at health facilities, in the form of different treatments, to the cost of those activities. As the purpose of the analysis is to benchmark and compare different types of health facilities, the chosen cost objects are: • Unit cost per outpatient visit • Unit cost per inpatient admission • Unit cost per bed-day • Disease specific costs of admissions according to a categorization of diseases These cost objects to a certain degree offers a simplistic representation of the activities that actually take place at the health facilities. Other significant services provided at health facilities relate to health promotion and prevention (including outreach, immunizations, etc.) and maternal and child health (counselling for pregnant women, ongoing health visits related to pregnancy, etc.). 61 This costing study calculates the unit costs above for a sample of Bhutan's health facilities. A total of 12 facilities and their associated out-reach clinics were selected, and is shown in Table A1 below. TABLE A1. H EALTH FACILITIES INCLUDED Facility Name Facility Type Dzongkhag/District JDWNRH National Referral Hospital Thimphu Eastern RRH (ERRH) Regional Referral Hospital Mongar Central RRH (CRRH) Regional Referral Hospital Sarpang Paro District Hospital Paro Wangdicholing District Hospital Wangdicholing Damphu District Hospital Tsirang Punakha District Hospital Punakha Gyelposhing 10-bedded Hospital Mongar Kanglung 10-bedded Hospital Trashigang Genekha Primary Health Center Thimphu Mendelgang Primary Health Center Tsirang Thinleygang Primary Health Center Punakha The main criterion for the choice of facilities, is whether or not the facilities were also used in the previous 2009/10 costing analysis. This is done so that comparisons and analysis of cost changes over time are possible. However, two facilities from the 2009/10 analysis have been excluded, as one had been upgraded to a higher-level facility, and the other transformed to a traditional medicine facility. To make up for the loss of those two 10-bedded hospitals in the sample, Kanglung hospital has been added to the sample list. Although only 12 facilities (and their attached out-reach clinics) out of a total of 235 facilities are covered by the study, the sample still covers a large share of total OPD-visits and admissions in Bhutan, as all major hospitals are included in the sample. While the total number of OPD-visits in 2019 was about 1.4 million, the study with its sample facilities covers 1 million visits. Additionally, while the total number of hospital admissions in 2019 was about 73,000, the study manages to cover 44,678. Thus, between 60 to 70 percent of all cases in Bhutan is covered by the sample. To collect information on admissions, visits and bed days, two different data sources were used: Bhutan's national health information system and electronic patient records from JDWNRH. Bhutan's national health information system contains information on admissions and visits on a monthly basis. 62 A sample of around 80 different diagnostic codes based on ICD10 4 is also used at national level to classify individual hospitalizations and outpatient visits. Each health facility reports monthly data to the central level on a wide range of forms. In addition, some of the bigger hospitals maintain their own electronic patient records, which are more detailed and include bed days for admissions, as well as other information related to each individual patient contact. Among the smaller hospitals, no estimates for the number of bed days were available. Moreover, no diagnostic information was recorded or used at the national referral hospital in the registration of outpatient visits, which may seem odd given that the larger hospitals tend to record greater levels of detail for inpatient admissions. The other hospitals and health centers do not use electronic records of patients, but record patient information into a number of manual records or are typed into a web-based version of the health information management system. These two sources (the Bhutan health information system and electronic patient records) constitute the data sources used for measuring the previously defined cost objects for the of 2018/19 Financial Year time period. The next step is to identify and quantify the resources used to provide these services. Resources used to deliver services The above sections defined what is the object of the analysis (cost object) as well as the sampled facilities for the analysis. This section defines the resources used to provide the services at the facilities. The literature identifies a wide range of resources used to provide services at health facilities. However, this can also vary depending on the perspective of the analysis that is adopted (Alban et al. 1999; Mogyorosy and Smith 2005; Drummond et al. 2005). Mogyorosy and Smith (2005) summarize some of the main resources identified across the literature, and those which align to a facility perspective, which is used for this analysis, is presented below: • Labor (staff) • Other goods and services (Medical devices, medicines, consumable equipment, diagnostic equipment) • Furniture • Buildings and land • Staff quarters • Administration /management (Office hold, security, cleaning, office supplies, IT, repairs and maintenance) Based on the above list, the various resources could then be further identified and quantified. 4 ICD10 is WHOs 10th version of International Classification of Diseases 63 Quantification of resources The fourth step in the cost analysis is quantification of resource consumption. In this context, it is important to measure (quantify) all identified resources (Mogyorosy and Smith 2005; Lee et al. 2003). To quantify resource consumption, a major undertaking for data collection was carried out. The data collection procedure consisted of the amalgamation and processing of data from a variety of sources, including use of a data collection template as well as subsequent interviews to qualify and validate the collected data. This was done with the assistance and facilitation of staff from the Ministry of Health. The main purpose of the data collection template was to collect data on staff at the individual facilities and their designation and location at the health facility. Unlike other nations of similar levels of development, Bhutan has a significant amount of health care data available. The following section describes the six resource areas identified previously and the data sources that are associated with them. Labor Staff at the health institutions is considered an essential resource. There was adequate information on staff for the period analyzed, including information on labor costs, both at individual institutions and in the dzongkhags (districts). Other goods and services As all medical equipment and medicines are supplied centrally to individual institutions, the Ministry of Health could in theory provide detailed information on what has been delivered, when it has been delivered, and where it has been delivered. Each institution regularly obtains the necessary equipment and medicines from a central unit in the Ministry of Health, except for JDWNRH that manages this procurement themselves. However, there are no recorded inventory consumption data at the individual institutions, and thus the consumption of goods and services are estimated based on deliveries conducted or received for a given period. For the type of equipment that is not immediately consumed (medical equipment, diagnostic equipment and various appliances, etc.), records and reporting are made on an institution-level basis. These records are also available at central level and should ideally be used for the quantification of these resources but were not used in this current costing analysis due to data unavailability. Buildings and land The department of medical supplies and health infrastructure of the Ministry of Health is responsible for building new health facilities. However, since Bhutan’s decentralization reform in 2008, it has been the responsibility of the dzongkhags to build the lower-level facilities while hospitals remain a national responsibility. However, in practice, the decision is taken at the district level, but it is the central level that is responsible for building them. 64 Thus, at central level, there is detailed information on all of the country’s health infrastructure buildings. Likewise, many of the facilities have been built according to identical blue-prints and are thus structurally very similar. Staff Quarters In addition to the building itself, the majority of the country's health facilities also provide living accommodations for their staff. Health facility staff are regularly moved around the country, and thus official residence made available for staff transferred to new duty stations is essential. Information on these accommodations were collected through the costing template questionnaire. Administration / management The current administrative and management expenses such as for cleaning, security, office supplies, IT, repairs and maintenance, etc. are provided in the annual financial accounts of the facilities. Connecting value to use of resources After identifying and quantifying resources, a financial value must be determined. In practice, the valuation is usually based on the market value of a used resource (i.e., the wage costs of employees and the purchase price of medicines). The literature describes how markets are not always (rarely) perfect and therefore prices will often deviate from the price that would be in an ideal efficient market with ‘perfect’ competition (Drummond et al. 2005; Christiansen 2007; Pedersen 2013). This is especially the case for the purchase price of medicines, in addition to other areas where there is a monopoly. Drummond et.al. (2005) points out that this problem may create some bias if the evaluation is not adjusted to accommodate, but also that costing studies that do use actual market prices and adjusts for existing market prices is in practice difficult to do and can often create new problems (Drummond et al. 2005). This costing analysis thus applies the actual market prices to individual resources and goods. These prices are to some degree available for the different resources identified. Another important part of the valuation is that costs can fall at different times. This applies, for example, to different medical devices, buildings, furniture, etc. which have a period of use that goes beyond the period of the analysis (one year). For these capital goods, a calculation of the return and depreciation of the acquisition cost is made, which can then be distributed over the period in which the given resource is used. By doing this, an annual cost can be calculated. The annual depreciation cost of capital goods can be difficult to measure unless information such as purchase price (market price) and lifetime is available. Bhutan keeps records of purchase prices at the central level and these have been used for various equipment. However, for buildings, a market price similar to what it would cost to build them today has been used. The building price used the same cost as the 2009/10 study adjusted to the 2018/19 price level. Thus, ideally, all capital goods have been given a financial value equal to the recorded price information on the goods, or a market price equal to a present value of the resource. Likewise, all resources will be given a lifetime. The annual depreciation cost is thus calculated using an interest rate of the average development of Bhutan's consumer price index over the past five years. In the case of this study, a value of 6 percent was used. 65 Drummond et. al. (2005) indicates a formula for calculating this, where the annual cost E over a period of n year (lifetime), at an interest rate r, corresponds to the market value K for the resource (Drummond et al. 2005). It is expressed by the following formula: 1 − (1 + )− K = The level of detail behind these calculations for the analysis is in theory very high. The records for each facility include thousands of different types of equipment. However, the unit prices for those equipment were not readily available, and attaching them to all the individual items was a task that was assessed to be beyond the scope of this study. Had the information been readily available, the unit prices could be categorized into groups with an assumed life span of 1, 5, 10 and 15 years. The equipment with a life span of 1 year would be considered fully consumed since the time period for the cost analysis is 1 year. It is generally assumed in the literature that clinical equipment has a lifespan of 5 years (Drummond et al. 2015). Since it was not possible for the exercise to meaningfully attach unit prices to all the equipment, the annual expenses on equipment were used instead. The total cost of equipment for the financial year however was available, and by using the sample facilities’ share of total outpatient visits as a distributive weight, the total cost for the sample facilities was able to be derived. These were then allocated to the facilities by using the shares of equipment from the 2009/10 costing exercise as an allocation key. This meant that no annual depreciation cost was calculated, but the cost of the financial year was used assuming a constant level of investment in equipment. Vehicles were given a life span of 10 years while the life span of buildings was set at 30 years. Again, the literature recommends the life span of buildings as being 20 years (Drummond et al. 2015), which given the Bhutanese context was judged to be too low and thus set to 30 years. This adjustment was also made in the 2009/10 costing exercise. 2. DESIGNING THE STANDARD COSTING MODEL To calculate the desired unit costs, a uniform costing model that can be applied to all 12 facilities was developed based on the model from the 2009/10 exercise. As the same model was used, the same calculation methods were applied to all sample facilities, and thus results are able to be directly compared between facilities as well as over time against the 2009/10 costing analysis results. The starting point for the model is a top-down approach as described above. It is also based on the type of model developed at Yale University in the late 1970s and early 1980s, led by the economist Fetter (Wiley 2011; Thompson et al. 1978). This model is often referred to as the Yale cost model (or Yale Costing Methodology) and involves allocating an aggregated total cost to a pre-defined number of cost centers in the facility according to specified allocation criteria. A cost center in this context represents a function at the facility. Thereafter, based on the costs and knowledge (specifically quantity) of production, unit costs can be calculated for different types of services. This type of model is one of the most widely used in "cost- 66 accounting" and has also been used in a number of similar studies and analyses in developing countries, including but not limited to: Malawi, Tanzania, Uganda, the Philippines and Egypt (Mills et al. 1993; Tsolmongerel 2009; HERA 1999; Abt. Associates 2010, Maurizio et al. 2003). This basic model is also used in different forms to calculate costs in the diagnosis-related group (DRG)-systems in many countries where individual hospitals' costs are allocated through various mechanisms to different cost centers [15]. In other words, the chosen method for establishing a standard cost model is a common and well- tested method for this type of analysis. The model consists of a number of cost centers which are grouped into three categories or levels: the overhead level, intermediate level and the final level. The top two levels (overhead and intermediate) are not ‘producing’ cost centers (i.e., they do not directly provide services to patients) for the cost object (unit cost for outpatient visits and admissions). Whereas final cost centers are considered service- producing entities, meaning they are cost center that directly provide services to patients. Inputs to each cost center were then identified, quantified and given a financial value. This is described above and is also further illustrated below. Costs are then allocated from the overhead and intermediate level according to specified allocation rules to the final level, which includes service producing departments such as outpatient and inpatient departments. Finally, the unit costs of the final cost centers can be calculated. This process is described in more detail in the following sections. Cost centers The cost centers were defined by the functional structures of health facilities in Bhutan. As mentioned above, the model for all 12 facilities consists of a number of cost centers in a three-tiered structure: • Overhead cost centers. This level includes administration, accounting, security, maintenance, reception, etc. • Intermediate cost centers. This level are various support functions and transverse clinical service departments like diagnostics (X-ray and ultrasound), laboratory, etc. • Final cost centers. Final cost centers are the actual service-providing departments such as various outpatient and inpatient departments, department for child and maternal health and department for traditional (indigenous) medicine The cost centers for the three levels are depicted below. 67 TABLE A2. COST CENTERS OF THE STANDARD COSTING MODEL Overhead Intermediate Final Administration Imaging Outpatient department Transport Kitchen Inpatient department Staff quarters Laboratory • Medical Security & Maintenance Pharmacy/Dispensary • Surgical & medical Operating Theatre Maternal & Child Health Traditional Medicine Some of the defined cost centers will only be found, and thus relevant, at the hospital level. Similarly, the model, particularly for the larger hospitals, could benefit from further detail. However, it has been essential for the comparability of the results across different facility types to use a uniform standard model and methodology for calculating the desired unit costs. The fact that the model for primary health care centers contain a number of non-relevant cost centers (typically no operating theatre and diagnostics, etc. at minor facilities) simply means that no costs are allocated to these cost centers. Four final cost centers have been defined. It reflects the different types of services provided by the health system in Bhutan and the way in which it is organized. There are the two traditional outpatient and inpatient units. Additionally, there is a maternal and child health cost center. This is particularly found at the larger facilities which have these services organized into separate departments. Other tasks taking place here are immunizations and various screening programs. This is also where preventive and health promotion efforts are rooted. Ultimately, although the total cost of these will be calculated as they constitute essential services, they are not subject to further analysis. In Bhutan, traditional medicine plays an important role in healthcare and, unlike many other developing countries, is an integral part of public health services. Thus, in most hospitals there is a special department for traditional medicine, where services consist primarily of different kinds of natural medicines and herbal methods for treatment of various conditions. As with maternal and child health, traditional medicine is not the subject of further analysis here, but the total cost of these entities is calculated. The core of the analysis is focused on outpatient visits and inpatient admissions, and their associated unit costs. However, it is necessary to include the other entities in the model to ensure that only relevant costs are looked at for the specific services analyzed. First allocation of costs to cost centers In setting up and defining the different cost centers, the relevant identified and quantified costs shall be allocated to the different centers. Their identification and quantification were described in the previous chapter. 68 The costs allocated to overhead and intermediate cost centers are, as defined in the previous chapter, indirect costs. The costs allocated to the final cost centers are direct costs as these are directly related to the production of the services covered by the analysis (outpatient visits and inpatient admissions). In connection with the data collection from the participating facilities, interviews were conducted with selected senior staff at the facilities. These were conducted to identify whether there were any particular circumstances or events at their particular facility during the period analyzed, which were significantly different from other periods, and thus could distort the results. It was likewise used to validate the collected data. Similarly, each facility was asked to fill in data for the developed costing tool, indicating where each employee spent his or her working time. For a number of employees (i.e. pharmacists, laboratory technicians and administrative staff) this was unique to the individual. However, some staff groups were linked to different locations, and thus the cost of these were distributed across different cost centers. This could be staff (i.e., doctors, nurses and healthcare assistants) who were often distributed across outpatient and inpatient departments, or doctors spending time doing administration, ward duty and time in the operating theater. In the costing template that was filled in by each facility, each individual staff was able to indicate how much time they spent in the various departments. For example, a staff member could be listed as working in the outpatient department 40 percent of the time and inpatient department 60 percent of the time. The wage expense would then be distributed based on this percentage to their respective departments. Based on the costing tool of all sampled facilities, the distribution of staff was analyzed and the wage costs for each facility could then be allocated to the relevant cost centers. As also mentioned in the previous chapter, there were no data recorded for the consumption of medicines. Inventory of medicine deliveries were instead used as an estimate for consumption. All costs of medicine were allocated to the pharmacy cost center. Consumables (i.e., syringes, wire, gauze, etc.) were likewise allocated to cost centers following the above method. The depreciation costs of the buildings were also allocated to the individual cost centers. This was done using the proportion of staff costs as an allocation key. The literature recommends the use of square meters as a key (Flessa 2009). There were available floorplans of health centers, but not of hospitals. It was therefore not feasible to use this as a key. The same allocation key was used uniformly for all facilities, namely share of labor costs. Other costs from the accounts such as transport, office utilities, etc. were allocated to the respective relevant cost centers. The next step is to allocate the indirect costs (from overhead and intermediate cost centers) to the final cost centers. This is described in the section below. 69 Stepwise allocation of indirect costs There are several methods for allocating the indirect costs to the final cost centers. Flessa (2009), Drummond et. al. (2005) and Tan et. al. (2011) describe different methods. The most commonly used method and the method recommended for an analysis of this type is the stepwise allocation method (Flessa 2009). The method follows a sequential step-by-step procedure. The first step is to allocate the top-level cost centers to all levels below it. Specifically, this means allocating overhead cost centers to intermediate cost centers and then to final cost centers. The second step is to allocate from the intermediate cost centers, which now consist of the original costs as well as what has been allocated from the overhead cost centers, to the final cost centers. The table below illustrates the allocation mechanism/key used for the different cost centers. TABLE A3. ALLOCATION RULES OF COST CENTERS Cost Center Allocation rule Overhead Allocated to intermediate and final cost centers based on Administration share of direct cost Allocated to intermediate and final cost centers based on Transport share of direct cost Allocated to intermediate and final cost centers based on Staff quarters share of direct cost Allocated to intermediate and final cost centers based on Security & Maintenance share of direct cost Intermediate Allocated to final cost centers: OPD, IPD and MCH based on Imaging share of direct cost Allocated fully to IPD and IPD medical and surgical weighted Kitchen based on number of admissions of each type Allocated to final cost centers: OPD, IPD and MCH based on Laboratory share of direct cost 10 percent allocated to MCH, 90 percent to IPD and OPD based Pharmacy/Dispensary on activity giving 4 OPD-visit the same weight as 1 IPD- admission. Operating Theatre Fully allocated to surgical IPD 70 The literature points to different ways of defining the allocation keys. Some of these were in practice not feasible due to a lack of data. Hence the share of direct costs is a widely used allocation mechanism as described by Flessa (2009). It is important to note that the above overview of the various allocation mechanisms provides a general guideline, but in practice has some caveats. For example, the allocation of medicines etc. from the pharmacy cost center caused several difficulties when attempting to perform allocations in the absence of records of consumption patterns for the three relevant cost centers. In the absence of this information, the literature recommends that activity/output can be used as an allocation key (Pedersen 2013). However, only activity data from outpatient and inpatient departments are available. There are no suitable activity data for this purpose in maternal and child health departments (MCH). Based on interviews with staff at JDWNRH and in the Ministry of Health during the 2009/10 analysis, it was decided to use 10 percent of pharmacy costs for MCH with the remaining 90 percent distributed according to activity levels, to IPD and OPD where four OPD visits equal 1 inpatient admission. For comparability, the same allocation key was used in the current analysis. Finally, it should be noted that all costs from the operating theater have been allocated to the IPD department cost center. Interviews revealed that operating theatres are only used for admitted patients, while less complicated interventions, etc. for outpatients are carried out directly in dedicated treatment rooms in the outpatient clinics. After all costs are allocated to the final cost centers then unit costs can be calculated. Calculation of unit costs The calculation of unit costs is relatively simple since unit costs correspond to the total average cost (see previous chapter for definition) of the cost object of the given cost center (outpatient visits or inpatient admissions). Where there is information on the number of bed days, the bed-day cost can be calculated similarly. This gives the following formulas: • Unit cost of an outpatient visit to a given facility = final cost outpatient cost center /total number of outpatient visits. • Unit cost of a IPD admission at a given facility = final cost IPD department cost center /total number of admissions. • Unit cost for a bed day at a given facility = final cost IPD department cost center /total number of bed days. Calculation of disease specific costs In addition to providing overall cost calculations of OPD and IPD activities, disease specific calculations were made according to the major diagnostic classifications used in the Bhutan Health Information System. 71 The data only supported calculating the average cost by diseases for inpatients. In order to do this calculation, disease aggregated data on inpatient bed-days was used to calculate the average bed- days by disease at the three referral hospitals. Based on the unit cost of a bed-day at the facility, an average admission cost for the various disease categories could be calculated across hospitals. In the current study, the detailed bed-day data of the hospitals was not available. By using the weights of the three referral hospitals for bed-days from the 2009/10 costing study, the bed-days was calculated for the current study. With these bed-days estimation, the disease specific costs of admissions for the three referral hospitals could be calculated. For the district hospitals and 10-bedded hospitals, the combined weight of the three referral hospitals were used to calculate bed-days and thus the disease specific costs of admissions. This is the most precise method to assess disease specific costs with the available data of a top-down costing exercise. However, since the information of costs of expensive medicines, equipment’s and diagnostic services for different disease groupings are not recorded or available, these are spread broadly across all disease groups, making the length of stay at the hospital the defining factor in the average cost of the disease specific admissions. 3. Limitations and assumptions When conducting a study like this there will naturally be some limitation and assumptions. The key ones are listed in this section and should be considered when using the results for policy analysis and decision-making. The design of a costing study based on a uniform model is based on a number of assumptions and naturally has its limitations. It is important to understand and consider these when analyzing the results. It also has certain strengths. One of the main strengths of this study design is that by basing the costing on a uniform model it allows for comparisons between facilities at same as well as different levels as all are subject to the same assumptions and limitations. Likewise, it can be expanded to include other facilities of the country as well as being updated with newer data in order to analyze developments over time is the case of this exercise comparing the results with the results of the 2009/10 costing exercise. The study is to a high degree based on data regularly registered and maintained at facilities and at the Ministry of Health. This is an obvious strength of the study, but on the other hand, this data dependency also makes the results very sensitive to issues with data quality as will be further discussed below. The main assumptions and limitations of the study are: • Quality of services has not been assessed. The focus of the study has been on quantification in terms of cost of health services at various levels of service delivery. The quality of the health care services, or the outcome of care has not been assessed. This implies when making comparisons there is an underlying assumption, ceteris paribus, that the results or quality of 72 the services we compare are the same. At the more specialized and higher-level facilities there are more clinical expertise present, and this could also mean a higher quality of services. On the other hand, the services offered at the lower levels reflect the expertise available at the facilities. When looking at patient-flows there is a tendency for people seeking out services at higher level facilities which can lead one to believe the general observation at the population is that quality is higher there. This tendency seems to have increased since the 2009/10 costing exercise since the patients visiting primary health care centers is lower now than in 2009/10, even when a general increase in services delivered has increased. • Model design uniforms health facilities. The uniformity of the model design treats the facilities as uniform entities producing the same output. Even though there is a high degree of uniformity of health facilities in Bhutan, there are still differences among facilities. These issues need to be taken into consideration when interpreting the results. For example, we know that various specialized diagnostic services are available only in some facilities as well as case-mix of patients vary within the same diagnostic groups. Here it is likely to assume that the more severe cases within the same diagnostic groups are referred to more specialized treatment at the higher-level facilities. • Top-down model design. The top-down design of the model allows for giving a macro-picture of cost structures and unit costs of various facilities. This spreads some of the cost from the top down to service delivery departments according to defined keys. Doing it this way however causes some level of detail to disappear. For example, the cost of medicine and medical equipment will not be spread out according to the diagnostic groups they are used for. To retain that level of detail would ultimately require more detailed data. A more detailed costing of specific services (micro-costing) would also require a bottom-up approach like Activity- Based-Costing. There was an interest in doing this in a few selected areas such as for the costing of specific laboratory and diagnostic services, and could be done using an ABC-analysis approach. However, this kind of analysis would require another comprehensive study in itself, and was not fit for the purpose of the current exercise. Additionally, it would only be a fragment of the various diagnostic services that are reported in the health information system, requiring a detailed output analysis at selected facilities in this area as well. However, since the costing model have defined cost centers for supportive services like laboratory and imaging, the facility-wise total cost of these areas is available and can be found in the more detailed data annex for all facilities. • Unit Cost sensitivity. As mentioned in the results chapter, the unit costs are very sensitive to the activity level / number of patients at a facility. This becomes evident from the very high number of patients at Paro Hospital and the resulting very low unit costs. This is also the reason why the unit cost of OPD-visits at primary health care centers have increased by 75 percent in real prices compared to the 2009/10 exercise, since the unit cost has been driven 73 up by fewer patients. Therefore, when analyzing the results, it is important to look at the total costs as well as the unit costs. • Average vs. Marginal Costing. The study analyzes average cost of services. Cost structures of fixed and variable costs imply that the marginal cost of treating more patients, to a certain point, is lower than the average cost – and often much lower when looking at the relatively high share of fixed costs at the facilities. When interpreting results, it is therefore important not to conclude that you could save, an amount corresponding to differences in the average unit costs, by changing patient flows from more expensive facilities to less expensive facilities. While this could be the case in the long-run where all costs are variable, in the short run the changes and possible savings would only be in the differences in the marginal cost, which is complicated to calculate. • Limited validation of results. The exercise was initiated just prior to the COVID-19 pandemic. That amongst other reasons delayed the finalization of the exercise. The travel restrictions imposed by the pandemic limited the possibility of consultants to do on-site follow-up visits to facilities to further validate data and results, and to qualify the costing tool. This was instead done remotely with tremendous efforts by the Ministry of Health. This was a circumstance that could not be avoided, but when considering and using the results of the exercise, it should be kept in mind that validation of the results was not optimally executed, which could add extra uncertainty to the results in addition to what has already been mentioned. • Data issues. The costing study is very data intensive. All results are based on data from facilities, MOH and district authorities and thus sensitive to the quality of data. During the course of collecting, processing and analyzing the data, various issues became apparent. These are summarized below. o Data availability. In theory, all the needed data for the study was available from various institutions and facilities. This is especially unique for a country at the stage of development that Bhutan is in. This is possibly due to Bhutan’s almost purely public health system as well as the centralization of key functions like health infrastructure management, procurement and supply of drugs and medical equipment and a national health information system. These data were collected and assessed for usability in the study. Unfortunately, some of the data was not possible to use although detailed datasets were available. o Medicine and consumables; medical equipment; and buildings and vehicles . Due to limitations in the available data these were, to a large degree, necessary to estimate as outlined in the data collection section above. 74 o Output data (data on patient visits and admissions) The main output data source is the Bhutan Health Management Information System (BHMIS). In addition to this, detailed patient records from JDWNRH hospital were available. As outlined previously, the unit cost results are very sensitive to the number of patients at the facilities. The BHMIS provides detailed data on the activity at the facilities, but when delving into the results, some quality issues are noted. A general irregularity and difference in reporting was observed. How reporting of activities was registered at the facilities, to some degree, was not executed uniformly. For example, when looking at old cases for both inpatients and outpatients, there seem to be huge variations in the share of old cases compared to the total between facilities. This implies different practices in registering information. For example, old cases in inpatients refer to readmissions of patients. About half of the facilities however report no old cases while the other half report old cases. In one hospital the old cases amount for half of the inpatient, which is very unlikely to be possible. Additionally, the old cases have no disease coding attached to them in the system, which would be required to analyze activity based on contacts with the health facility. When this is not done, tracking actual morbidity burden is difficult. Improving health information systems is a big exercise that reaches far beyond adapting it to the needs of doing costing. A holistic approach is recommended considering all possible uses of these important data to both inform policy makers but also to improve clinical practice and results. The study noted these irregularities in the system, but the analysis has used the data from the system that in some areas can have led to diverted results. This is for example the case with the data from Paro Hospital that has seen a huge increase in patient load since the previous exercise and having more inpatients that JDWNRH. JDWNRH is a 350-bed hospital compared to Paro Hospitals that is a 40-bed hospital. The data were verified to be correct from a system point of view, but the registration practice at the hospital could have resulted in an overreporting of cases. 75 REFERENCES 1. Abt. Associates. 2010. “Cost and Cost Analysis Report for Seven Hospitals in Egypt”. United States Agency for International Development (USAID). 2. Alban, Anita. 1999. ”Sundhedsøkonomi – principper og perspektiver”. DSI - Institut for Sundhedsvæsen. 3. Christiansen, Terkel. 2007. “Sundhedsøkonomisk evaluering” in Vallgårda S and Krasnik A. ”Sundhedsvæsen og sundhedspolitik”. Copenhagen: Munksgaard Denmark. 4. Drummond, Michael F., Mark J. Sculpher, Karl Claxton, Greg L. Stoddart, and George W. Torrance. 2005. “Methods for the Economic Evaluation of Health Care Programmes”. Oxford: Oxford University Press. 5. Fetter, Robert B., John D. Thompson, and Ronald E. Mills. 1976. “A system for Cost and Reimbursement Control in Hospitals”. Yale University, New Haven, Ohio, USA. Academic Press, Inc. 6. Flessa, Steffen. 2009. “Costing of Health Care Services in Developing Countries: a prerequisite for affordability, sustainability and efficiency. Challenges in public health 1863-768X. https://catalogue.nla.gov.au/Record/4587667. 7. Health Research for Action (HERA). 1999. ”Health Care Financing in Tanzania – Costing Study of Health Services”. Belgium. 8. Jegers, M, D L Edbrooke , C L Hibbert, D B Chalfin, and H Burchardi. 2002. “Definitions and methods of cost assessment: An intensivist’s guide”. Intensive Care Med. 28(6): 680-685. https://pubmed.ncbi.nlm.nih.gov/12107670/. 9. Jigme Dorji Wangchuck National Referral Hospital (JDWNRH) of the Kingdom of Bhutan. Patient records for 2018/19. 10. Kingdom of Bhutan. Financial records from select health facilities. 11. Lee, Robert H., Marjorie J Bott, Sarah Forbes, Linda Redford, Daniel L Swagerty, Roma Lee Taunton. 2003. “Process-based costing”. Journal of Nursing Care Quality 2003 Oct-Dec 18(4):259-66. https://doi.org/10.1097/00001786-200310000-00003. 12. Mills, A J, J Kapalamula, and S Chisimbi. 1993. “The cost of the district hospital: a case study in Malawi”. Bulletin of the World Health Organization, 71 (3/4): 329-339. https://pubmed.ncbi.nlm.nih.gov/8324852/. 13. Ministry of Health of the Kingdom of Bhutan. “Annual Health Bulletin 2010”. 14. Ministry of Health of the Kingdom of Bhutan. “Annual Health Bulletin 2015”. 15. Ministry of Health of the Kingdom of Bhutan. “Annual Health Bulletin 2019”. 16. Ministry of Health of the Kingdom of Bhutan. “Annual Health Bulletin 2020”. 17. Ministry of Health of the Kingdom of Bhutan. Various data from the Health Management Information System (HMIS), and medicine and equipment inventories. 76 18. Mogyorosy, Zsolt and Peter Smith. 2005. “The main methodological issues in costing health care services: a literature review”. CHE Research Paper 7. Centre for Health Economics. https://pure.york.ac.uk/portal/en/publications/the-main-methodological-issues-in-costing- health-care-services-a--2. 19. Murru, Maurizio, Corrado Bruno, John Odaga, Ahairwe Denis, Akulu Ernesta, Bavcar Alessandro, Bonane Emmanuel, Kirunda David, Mwesezi Henry, Nagujja Angela, and Ndindayino Kalire. 2003. “Costing of health services in Lacor Hospital, Uganda”. Department of Health Sciences, Uganda Martyrs University, Uganda. http://www.bioline.org.br/pdf?hp03012. 20. National Statistics Bureau of the Kingdom of Bhutan. https://www.nsb.gov.bt. 21. Pedersen, Kjeld M. 2013. “Sundhedsøkonomi”. Copenhagen: Munksgaard Denmark. 22. Rice, Dorothy P. 1967. "Estimating the cost of illness". American Journal of Public Health Nations Health 57(3): p. 424-40. https://ajph.aphapublications.org/doi/abs/10.2105/AJPH.57.3.424. 23. Tan, Siok Swan, Lisbeth Serdén, Alexander Geissler, Martin van Ineveld, Ken Redekop, Mona Heurgren, and Leona Hakkaart-van Roijen. 2011. ”DRGs and cost accounting: Which is driving which?” in Busse R, Geissler A, Quentin W. 2011. “Diagnosis-Related Groups in Europe”. World Health Organization, European Observatory on Health Systems and Policies. McGraw-Hill. 24. Thompson, J D, R F Averill, and R B Fetter. 1978. “Planning, Budgeting, and Controlling – One Look at the Future: Case-Mix Cost Accounting”. AUPHA Faculty Institute on Financial Management, Cincinnati, Ohio, USA. 25. Tsolmongerel, Tsilaajav. 2009. “Costing study for selected hospitals in the Philippines”. Technical Assistance to the Health Sector Policy Support Programme in the Philippines (EC-TA HSPSP). Philippines. https://doh.gov.ph/node/2829. 26. Wangmo Sangay, Walaiporn Patcharanarumol, Tandin Dorji, Kinzang Wangmo, and Viroj Tangcharoensathien. 2018. “. “Bypassing Primary Health Care in Bhutan: Complex Interplays between Demand and Supply-side Influences”. Quality in Primary Care 26(5): 117-124. https://www.primescholars.com/articles/bypassing-primary-health-care-in-bhutan-complex- interplays-between-demand-and-supplyside-influences-100474.html. 27. Wiley, Miriam. 2011. ”From the origins of DRGs to their implementation in Europe” in Busse R, Geissler A, Quentin W. 2011. “Diagnosis-Related Groups in Europe”. World Health Organization, European Observatory on Health Systems and Policies. McGraw-Hill. 28. World Health Organization (WHO), Kingdom of Bhutan. 2021. “National Health Accounts 2018/19 & 2019/20”. 29. World Health Organization (WHO), Regional Office for South-East Asia. 2017. “The Kingdom of Bhutan Health System Review”. Health Systems in Transition 7(2). https://apps.who.int/iris/handle/10665/255701. 30. World Health Organization (WHO). 2009. ”Global Health Risks – Mortality and burden of disease attributable to selected major risks”. World Health Organization. Geneva. WHO Press. 31. Yangchen, Sonam, Tashi Tobgay, Bjørn Melgaard. 2016. “Bhutanese Health and the Health Care System: Past, Present and Future”. The Druk Journal. http://drukjournal.bt/bhutanese-health-and- the-health-care-system-past-present-and-future/. 77