How much do health services cost? Findings from three districts Annual Health Forum BMICH 9 th -10 th February 2007 Dr. Ravi P. Rannan-Eliya Institute for Health Policy
Outline Study TOR & Goals Approach & Scope Methods Problems encountered Results Implications Policy Future monitoring
Mandate and HPRA TOR Original TOR: To measure public and private sector unit costs in three districts Colombo, Badulla, Matale By levels of facilities Modified TOR: To measure unit costs in public sector by levels of institutions, and unit prices in combined private sector
General Approach & Scope Public sector Cost survey of government health facilities Three districts only Private sector National surveys of private sector hospitals, laboratories and doctors All districts
Public Facility Cost Survey Design of MoH-IHP Public Facility Cost Survey 2006 Based on previous Sri Lanka Public Facility Surveys (1992, 1997) to minimize development costs and maximize reliability by learning from previous experience General method Stratified sample survey in the three districts Field investigators used to collect data on activities and expenditures in each facility, supplemented by: Central MoH/Treasury data for MoH hospital expenditures Regional drug stores data for medical supplies Provincial/district office records for salaries/overtime Analysis of data using statistical software
MOH-IHP PFS Sampling Sample = 81 (Colombo - 30, Badulla - 28, Matale - 23) 69 hospitals/dispensaries, 12 MOOH units Response rates - 100% overall, >90% for most data
MOH-IHP PFS Data Collection Service activities Questionnaire used to collect data on service outputs in 2005: Inpatients, outpatients, operations, X-rays, lab tests, etc. Supplemented by IMMR returns where available Dependent on responses from key informants Time allocations of staff Questionnaire used to collect estimated time allocations of key staff groups Doctors, nurses, labourers, attendants, lab staff Expenditures Budgetary totals obtained from hospital directors Salary budgets re-estimated using staffing numbers where responses not reliable Overtime and other data collected from district and provincial offices Medical supplies based on MSD data, and sample survey of drugs dispensed in each facility Treasury data used for line ministry hospitals
MOH-IHP PFS Problems Non-line ministry facilities generally not responsible for budgets, so usually lack reliable information on actual budgetary expenditures Salary expenditures data not easily accessed at low level facilities Needed to supplement using other data sources Drug expenditures not responsibility of facilities Budget held by MoH MSD, estimates prepared by PDoHs MSD computerised inventory system only tracks supplies to regional drug stores. Further distribution to facilities not computerised, and no easily accessible data on actual drug expenditures by facility Reliability of IMMR returns in question in many facilities
Private sector price surveys Surveys conducted of private hospitals, private laboratories, private ambulance companies Survey of prices of private doctors found not to be feasible owing to reluctance of doctors to cooperate or provide accurate data Response rates for private hospitals and laboratories high, but not for other surveys Survey problems Private hospital respondents often did not understand or keep track of “average length of stay” or average bed occupancy Inconsistencies between revenue and activity data suggested reporting errors with many hospitals Non-responses and identified data errors handled using imputation techniques
Findings How do unit costs vary at different levels? How do unit costs vary by district? How do private sector prices compare?
Bed-day costs by facility type MOH-IHP Public Facility Survey 2006
Admission costs by facility type MOH-IHP Public Facility Survey 2006
Outpatient costs by facility type MOH-IHP Public Facility Survey 2006
Chest X-ray costs by facility type MOH-IHP Public Facility Survey 2006
Admission costs by district MOH-IHP Public Facility Survey 2006
Outpatient costs by district MOH-IHP Public Facility Survey 2006 *Private sector = Rs
Medical officer overtime costs by district (Rs per month) MOH-IHP Public Facility Survey 2006
Public-Private Comparison: Admission costs, small hospitals Bed size < 70 MOH-IHP Public Facility Survey 2006
Public-Private Comparison: Admission costs, large hospitals Bed size > 70 MOH-IHP Public Facility Survey 2006
Public-Private Comparison: WBC/DC Costs
Public-Private Comparison: Chest X-ray Costs
Key Findings (1) Variation in unit costs between districts is not great Significant variation in unit costs between individual facilities, but largest variation is between levels of facility Unit costs in public sector increase uniformly in all districts by level of hospital Costs increase by level with longer admissions, greater levels of service provision, more complex case loads Unit costs by themselves do not indicate inefficiencies. Must also look at case complexity, services provided, location and demand profile, etc.
Key Findings (2) Public sector costs generally the same or lower than in private sector Need to consider purpose of exercise: If concern is contracting-out, then overhead costs should be excluded Admission costs in private sector significantly higher than in public sector No compelling evidence that contracting routine clinical services out will produce significant cost savings - opposite might be true Actual overtime costs appear to be significantly less than implicit liabilities Variation in overtime costs may be due to many factors, including availability of overtime budget
Issues What is purpose of exercise? Need to clarify in order to interpret data Unit costs may be useful at facility level, not at district BUT … unit costs alone are not good measure for assessing facility efficiency or performance, see UK NHS experience Feasibility Measurement difficult owing to lack of routine financial data at level of institutions - need for surveys Short term priority should be improving information system Use of IHP-MOH PFS 2006 data Survey is potentially valuable data source for examining determinants of facility efficiency in combination with other information Further analysis should be done by IHP/MOH Results should be fed back to individual facilities