Cambodia Essential Health Services Cost Study - Final Report
S. Fabricant WHO/USAID/POPTECH November 16, 2018
Cost Analysis of Essential Health Services in CambodiaMOH/WHO Health Sector Reform Phase III Project
Final Report of Data Analysis
Prepared by Steve Fabricant, MBA, Ph.D.
Executive Summary:
Cost and activity data from a sample of MOH hospitals and health centers were analyzed. Comparing only health centers without beds, contracting-in centers had only 628 total contacts per month versus 1313 contacts for contracting-out centers and 1159 contacts for non-contracting centers. 57% of the total number of contacts were curative visits in the health center, followed by 27% EPI outreach. Preventive contacts at the health center comprised the balance.
Health center utilization rates in catchment populations for the sample averaged 0.99 contacts per capita/year, ranging from 0.27 to 1.75. The annual per capita recurrent health center cost including depreciation averaged $1.09, with a range from $0.49 to $2.00. There is an obvious correlation between high costs and high utilization for health centers. Drug costs and staff costs are somewhat correlated, so that high total health center costs are a product of both factors rather than one or another. Good drug supplies and staff financial motivation may combine to increase quality, which could explain higher utilization rates.
HealthCenter costs including depreciation averaged $1,260 per month, with a range from $662 to $3,245, although the high cost of one HC was due to consumption of drugs which could not be verified. Excluding that HC, the average is $1,130, with the range of costs still quite wide, from $662 to $1,975. At an average total recurrent cost of $2,418 per month, bedded health centers (HC+) resemble health centers more than district hospitals even though they have a fairly high number of beds and inpatients. The total costs of contracting-out health centers was higher than those contracting-in or not contracting, but since they had a greater number of contacts, the costs per contact were only slightly higher.
Outputs varied widely among district hospitals, but contracting-out DHs had a mean shorter average length of stay for non-TB patients (4.1 days) than the contracting-in hospitals (5.1 days) or non-contracting hospitals (9.0 days). Provincial hospitals averaged 3,865 inpatient-days per month, compared to 1,746 at the average district hospital. TB/leprosy accounted for 41% of all inpatient-days at district hospitals, but only 23% at provincial hospitals. Surgical cases accounted for the most inpatient-days at provincial hospitals, while district hospitals were occupied mostly by medical or emergency cases.
The mean cost per total contact (in-HC plus outreach) for the 16 health centers was $1.19 excluding depreciation, or $1.33 including depreciation. The range was $0.46 (Sen Sok HC, Kralanh OD) to $2.48 per contact (Kampong Treas HC, Kampong Cham OD), or $0.56 to $2.75 including depreciation. The mean cost per contact for bedded Health Centers was $0.79 or $1.00 including depreciation. The cost per in-HC contact was estimated to be two to three times that of an outreach contact. For the sample of health centers, unit costs decrease with increasing utilization, reaching a limiting level of between $0.50 and $0.70 per contact. This level can be considered to be the marginal cost of a health center contact.
DistrictHospital total monthly costs including depreciation averaged $9,711, ranging from $4,367 to $15,789. Provincial hospitals averaged $30,763, with a range from $22,181 to $49,961. This factor of three times the cost corresponds well to the greater number of beds and staff in provincial hospitals. Depreciation costs comprise about 12% of total costs in provincial hospitals but 17% in district hospitals, suggesting the need for higher maintenance budgets.
Direct costs comprise a very high proportion of total costs, from over 90% for health centers and 76.9% in district hospitals to 72.3% in provincial hospitals, so the sampled facilities are operating rather efficiently in terms of budgets being used for direct patient care. Drugs and supplies comprise on average 50% (hospitals) to 64% (all HCs) of all direct costs, while staff costs range from 36% (HCs) to 42% (hospitals). Contracting-in and contracting-out hospitals and health centers were somewhat more efficient (85% direct costs versus 70%) than the non-contracting facilities. The average staff pay at contracting-in DHs is $48.00 compared to $125.20 at contracting-out DH’s. The contracting-out hospitals also use a greater amount of drugs and supplies.
Total costs and cost per inpatient-day of contracting-out district hospitals were much more expensive than those either contracting-in or not contracting. The mean per capita cost for contracting-out district hospitals was $1.39 (including depreciation), compared to $0.36 for contracting-in hospitals and $0.88 for non-contracting hospitals.
Per capita annual costs of provincial hospitals ranged from $0.97 to $3.02, averaging $1.77 excluding depreciation, or $1.14 to $3.23 averaging $2.10 including depreciation. These are nearly three times more expensive per capita than district hospitals, which averaged $0.73 and $0.88 per capita respectively.
In terms of cost per inpatient-day, all provincial hospitals fall in a fairly narrow range with an average inpatient-day cost from $5.91 to $8.38, (mean of $7.97) including depreciation. TB/leprosy cases tend to have much lower daily costs and longer stays, so if these are excluded the mean inpatient-day cost rises to $8.42. These same costs vary widely among the district hospitals however, from $1.67 to $14.10 per average inpatient-day, with a mean inpatient-day cost of $6.57 ($8.67 excluding TB/leprosy). Inpatient costs at bedded health centers are much lower, at $2.96 per inpatient-day including depreciation.
Over the sample of district hospitals, unit costs tend to decline with higher outputs, allowing the marginal cost for a district hospital inpatient-day to be extrapolated to between $3.50 and $4.00. Marginal costs for the major inpatient services were also estimated. The marginal costs of services when provincial hospitals were included in the analysis were somewhat higher, in part due to the higher costs of TB/leprosy treatment at provincial hospitals.
Further studies are needed to determine the exact relationship of facility costs to other factors such as service delivery strategies, treatment guidelines, staff motivation, and local population factors.
I. Study Objectives and Methods Used
The main objective of the Cost Analysis of Essential Health Services in Cambodia study described here was to determine the costs of the Minimum Package of Activities (MPA) and the Complementary Package of Activities (CPA). The study was co-financed by The World Bank, DfID, WHO Cambodia, and USAID, and the field data collection was managed by the Ministry of Health Cambodia. This report presents the analysis of data collected for this study. Information on costs and outputs was collected from a sample of MOH facilities stratified by level of facility (provincial hospital, district/referral hospital, and health center), and by the type of relationship to external technical assistance through a contracting mechanism, and other support mechanisms. As a first approximation for budgeting, the total output of the health centers can be considered a proxy for the proposed MPA, while the output of district hospitals can be taken as a proxy for the CPA.
Field data was collected in late 2001 by MOH staff according to the design of Bitran and Associates. Data on staffing and costs were obtained from MOH salary data, NGO records of bonuses and incentive payments, and user fee reports. Staff worktime allocation was measured by direct observation and interviews. The cost of drugs and supplies was derived from monthly Central Medical Stores (CMS) report forms; other direct and indirect costs came from facility records. For depreciation, buildings were measured and their construction type noted in the field. Vehicle and equipment lists were compiled by observation. Outputs by type or service were derived from a Health Information System, using records for the months of January, April, and September 2001. (A full description of data sources is given in Annex Table 1.) The output data was verified through a separate validation study that relied on field audits of reported data from the MOH health information system. It should be noted that only costs of health facilities were collected; none of the other costs of running the MOH health system, such as provincial and operational-district (OD) administration, training, planning, etc., are included in this study. No expatriate salary costs from contracting districts were included in the analysis either.
Raw data on staffing, time allocation, salary, incentive payments, and other costs were entered on one Excel spreadsheet per health facility. One to three monthly HIS forms were collected in computer format from each sampled facility as well. Consumption of drugs and other items supplied to facilities from Central Medical Stores was recorded and summarized separately. Most of the facility data was complete, but where one or more months of the HIS output data was missing or suspect, only the remaining one or two months outputs were used. This occurred for about 10% of facilities.
The approach used to analyze this large volume of data was to focus on the major cost components, disaggregating them as far as possible, then compiling them to correspond to the major health facility outputs. The components of health center and hospital costs are medical and non-medical personnel costs, cost of drugs, medical supplies and other direct costs of patient care, plus indirect costs and depreciation costs. The outputs of health centers included general outpatient treatment and outreach, mainly immunizations and antenatal/postnatal care. The major hospital outputs are the inpatient days, which were condensed into surgical cases, medicine/emergency cases, maternity/ob-gyn cases, pediatrics, TB/leprosy inpatients, and ‘other’ inpatients. Some hospitals also provided significant amounts of outpatient treatment.
The data allowed cost, output and efficiency comparisons among the individual hospitals on the basis of cost components and total and per capita costs, and costs per inpatient-day for the main services. The cost per in-HC and outreach contact was calculated for health centers. Comparisons are also possible on the basis of the contracting arrangement in use by the health facilities. A caveat must be made at this stage of analysis, that additional information on the specific situation of each health facility and its catchment population should be incorporated in order to fully understand the cost and efficiency results presented here. Little information of this nature was collected because of limitations in financial resources and capacity of enumerators. This study was empirical and done without full information on service delivery strategies and treatment guidelines, and therefore it is not always valid to compare the unit cost results in such an information vacuum. These specific details and those concerning the contracting arrangements, population density, accessibility, and the socio-economic environment of the population, are among variables of possible interest which could be the basis of subsequent research.
Data analysis methods and estimations:
The cost component for which the data was the most complete is staffing costs, which because of the “step-down” study design could be allocated directly to the various outputs. Other direct and indirect costs, and depreciation usually had to be allocated among outputs on the basis of assumptions and weightings that were based on logic and the experience of the author. Costs of CMS drugs and medical supplies, which comprised about half of all direct costs, were the largest component that had to be allocated so approximately. A brief study at a few hospitals could confirm the assumptions used, or else would call for a re-analysis. The assumptions and allocation formulas are discussed in connection with the results derived.
Depreciation is the amount by which the value of a physical asset decreases continuously due to its productive use. Since it is almost impossible to know or forecast with certainty the useful life of, say, a building or a car, standard lifespans are usually used to calculate depreciation costs of many kinds of physical assets. For the purpose of budget allocation, the annual depreciation of an asset is a good approximation of either the amount that should be set aside for the eventual replacement of the asset, or the amount that should be budgeted for maintenance so the asset can be kept in full productive condition. While the sizes and construction materials used in buildings was carefully observed and noted, and some health facility records indicated the cost of equipment, very little other information was available. A standard construction cost of $180/square meter and a standard life of 30 years was used for concrete buildings; $100/square meter and 20 years life for wood buildings. Vehicle replacement costs were estimated and a lifetime of 5 years used to calculate depreciation regardless of the age of an existing vehicle. Equipment costs were estimated, and a life of 10 years applied for medical equipment and 15 years for non-medical equipment such as furniture.
The average costs of outputs were obtained simply by dividing total costs by total outputs. The calculation of marginal costs from the available data, i.e., the cost of an additional unit of output, was derived by graphing the dataset of unit costs and utilization for the sampled health centers and hospitals. Short-run marginal costs assume that services can be expanded to a certain level without increasing fixed costs. The short-run marginal cost then equals the lowest possible unit variable cost of providing that service, e.g., staff costs plus drugs, supplies, and other direct costs, at a high utilization rate which reduces the fixed cost per unit to nearly zero .
In the case of expanding health services, however, a realistic approach to marginal costs would need to examine the costs of increasing the demand, utilization, and cost-efficiency of health services. As with the production of other goods, marginal costs of health services decrease as production expands due to economies of scale, but eventually they will start to rise. Attracting many additional patients may require the population as a whole to have a perception of improved quality of health services, which may be costly to achieve. Outreach to the least accessible parts of the catchment areas will be more expensive and find fewer and fewer people. After a certain point the remaining potential patients will be the more difficult chronic ambulatory cases (although it may be as likely that the last few patients to use the health facilities are those with minor problems that are inexpensive to treat.) The overcrowding of health facilities when utilization increases in the absence of planned expansion can also result in inefficiency. The data supports some interpolations of how expanding utilization affects total and marginal costs.
Sampled Health Facilities
The sampled health facilities are shown in Table 1, together with information about catchment areas, staffing, contracting, and external support.
Table 1: Health Facilities Sampled in Costing Study
Name of Facility / Province / Operational District / Population served by facility / Contract in/out / Contractor / NGO support / Total beds including TB wards / Total full-time equivalent staffProvincial Hospitals
Kampong Cham PH / Kampong Cham / Kampong Cham / 265,500 / - / - / Cuba / 260 / 192Pursat PH / Pursat / Sompoey Meas / 237,131 / - / - / - / 202 / 129
Takeo PH / Takeo / Don Keo / 185,495 / - / - / MSF / 176 / 165
Siem Reap PH / Siem Reap / Siem Reap / 111,226 / - / - / - / 340 / 180
District/Referral Hospitals
Choueng Prey DH / Kampong Cham / Choueng Prey / 164,733 / in / SCFA* / SCF / 100 / 38Kroch Chhmar DH / Kampong Cham / Kroch Chhmar / 111,750 / control / - / ADB**** / 80 / 26
Memut RH / Kampong Cham / Memut / 105,708 / out / SCFA / - / 50 / 55
Kralanh DH / Siem Reap / Kralanh / 98,819 / none / - / - / 79 / 27
Sotr Nikum RH / Siem Reap / Sotr Nikum / 227,696 / none / - / MSF***** / 98 / 66
Kirivong RH / Takeo / Kirivong / 228,231 / in / EED** / - / 59 / 16
Ang Roka DH / Takeo / Ang Roka / 116,295 / out / AMDA*** / - / 39 / 24
Bakan DH / Pursat / Bakan / 120,764 / none / - / RACHA / 64 / 41
HealthCenters with beds
Batheay HC / Kampong Cham / Choueng Prey / 11,201 / in / SCFA / SCF / 20 / 10Rominh HC / Takeo / Kirivong / 17,524 / In / EED / - / 44 / 23
Health Centers
Ang Tasom / Takeo / Ang Roka / 13,713 / Out / AMDA / - / 0 / 6Trapeing Andeuk / Takeo / Ang Roka / 15,167 / Out / AMDA / - / 0 / 8
Kampong Treas / Kampong Cham / Kroch Chhmar / 13,158 / Control / - / ADB / 0 / 7
Rokar Khnol / Kampong Cham / Kroch Chhmar / 11,547 / Control / - / - / 0 / 6
Sen Sok / Siem Reap / Kralanh / 14,881 / None / - / - / 0 / 7
Kampong Thkov / Siem Reap / Kralanh / 21,140 / None / - / - / 0 / 6
Som Roung / Siem Reap / Sotr Nikum / 12,778 / None / - / MSF / 0 / 6
Sang Veuil / Siem Reap / Sotr Nikum / 21,078 / None / - / MSF / 0 / 5
Daun Dom / Kampong Cham / Choueng Prey / 9,954 / In / SCFA / - / 0 / 7
Tum Nup / Kampong Cham / Choueng Prey / 13,622 / In / SCFA / - / 0 / 10
Chan Moul / Kampong Cham / Memut / 9,996 / Out / SCFA / - / 0 / 9
Choam Treak / Kampong Cham / Memut / 15,583 / Out / SCFA / - / 0 / 9
Kampong Krosong / Takeo / Kirivong / 3,987 / In / EED (SCF) / - / 0 / 3
Prambeymon / Takeo / Kirivong / 9,250 / In / EED (SCF) / - / 0 / 5
Trapeing Chong / Pursat / Bakan / 18,954 / None / - / - / 0 / 9
Metouk / Pursat / Bakan / 12,384 / None / - / - / 0 / 7
*Save The Children Australia **Enfant et Developpement ***Asian Medical Doctors Ass’n. ****Asian Development Bank *****Medicins sans Frontieres
II. Health Facility Outputs:
a. Health Centers
For health centers the major outputs were curative visits and outreach contacts (most of which were immunizations, antenatal/at-risk, and birth-spacing). The outputs of district and provincial hospitals are inpatient-days of patient care, plus outpatient visits for some hospitals.
The health center outputs reported were visits in the health center and outreach contacts. These are shown in Table 2a below. Bedded health centers (HC+) were somewhat more active than other health centers and also have more staff. Comparing only health centers without beds, contracting-in HCs had only 628 total contacts per month, versus 1313 contacts for contracting-out HCs and 1159 contacts for non-contracting HCs. Table 2b shows specific outputs for one month. 57% of the total number of cases were curative visits in the health center, followed by 27% EPI outreach. Preventive contacts at the health center comprised the balance.