Regional Variation in healthcare costs in South Africa. Linda Kemp Shirley Collie

Regional Variation in healthcare costs in South Africa Linda Kemp Shirley Collie

Agenda Private healthcare insurance in South Africa The argument for analysing healthcare consumption regionally Methodology applied to obtain South African healthcare drainage districts Methodology to calculate disease burden index Are South African regional healthcare consumption patterns explained by the underlying burden of disease and access to benefits? Cost of death in the last six months by region Cost efficiency by region Supply of beds per region Are the regional supply of beds commensurate with the underlying demand Is there a relationship between competition and the variation in supply? Concluding remarks

Percentage of People 1-4 5-9 U 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85+ 1-4 5-9 U 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85+ Private healthcare insurance in South Africa Public healthcare available to all with cost in line with ability to pay Can opt for private cover through medical aid Legislative framework for medical aids: Open enrolment, community rating No risk equalisation or mandatory enrolment Schemes must deal with selective joining and withdrawals Different risk profiles for different schemes and benefit options 6.5% 6.0% 5.5% 5.0% 4.5% 4.0% 3.5% 3.0% 2.5% 2.0% 1.5% 1.0% 0.5% 0.0% Coverage: Insurable Families Total Population Current Voluntary Medical Schemes Mandatory from Tax Threshold Mandatory Formal Wage Earners Female Gender and Age Bands Reimbursed on a fee for service basis Private healthcare expenditure per insured life has increased 3-4% above inflation for several years There are long terms concerns regarding the affordability and sustainability of private healthcare given the regulatory environment Male

Private healthcare insurance in South Africa South Africa Medical schemes are not-for-profit funders of private healthcare services 8.7 million lives were covered by medical schemes at end of 2012 Discovery Health Medical Scheme Roughly 2.5 million lives under administration Fastest growing open medical scheme (average growth of 5.5% p.a. since 2005) More than half the lives have been on the scheme for five years or longer Claims data provides opportunity for deep analysis

The argument for regional healthcare analysis Patients access local healthcare for the majority of their needs Secondary and tertiary services may be further away Patterns of how general practitioners choose to refer to specialists and hospitals allows for consideration of a region as a healthcare system Dartmouth Atlas Project considers variations in how medical resources are distributed and used in the US based on Medicare data Improve their understanding of the efficiency and effectiveness of health care systems Regional variation in cost of providing healthcare can exist due to disease burden, access issues, technology etc. Where variation is not due to disease burden: Dartmouth atlas promotes learning from regions that have attained sustainable growth rates and consumption levels

Obtaining South African Drainage Districts Patients allocated to a district based on where they access the majority of their primary care Hospital referral regions defined as where patients receive the majority of major cardiovascular and neurosurgery care Hospital service areas are defined as areas where at least 60% of policyholders receive cardiovascular and neurosurgery care within the region Adjacent magisterial districts are collapsed into the hospital service areas where the majority of patients receive their care

Obtaining South African Drainage Districts

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 56 58 60 62 64 66 68 70 72 74 76 78 80 82 84 Axis Title Development of disease burden index 6.00 Indexed costs by age and gender 5.00 4.00 3.00 2.00 1.00 F M 0.00 3.50 Indexed costs by plan type 3.00 2.50 2.00 1.50 1.00 0.50 Costs Costs adjusted for demographic differences 0.00 Plan 1 Plan 2 Plan 3 Plan 4 Plan 5 Plan 6 Plan 7 Plan 8 Plan 9

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 56 58 60 62 64 66 68 70 72 74 76 78 80 82 84 Development of disease burden index 8.00 7.00 6.00 5.00 4.00 3.00 2.00 1.00 0.00 Females on plan 5 by chronic registration status Not Registered for a chronic condition Registered for a chronic condition 3.00 Indexed costs adjusted for age, gender and plan by registered chronic status 2.50 2.00 1.50 1.00 0.50 0.00 Plan 1 Plan 2 Plan 3 Plan 4 Plan 5 Plan 6 Plan 7 Plan 8 Plan 9 Not registered for chronic conditions Registered for chronic conditions

Age 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69 71 73 75 77 79 81 83 90 80 70 60 50 40 30 20 10 0 Development of disease burden index Indexed costs for females registered for a chronic condition on plan 5 by RUB 1 2 3 4 5 20 18 16 14 12 10 8 6 4 2 0 Indexed costs adjusted for age, gender, chronic and plan by RUB Plan 1 Plan 2 Plan 3 Plan 4 Plan 5 Plan 6 Plan 7 Plan 8 Plan 9 0 1 2 3 4 5

100 300 500 700 900 1100 1300 1500 1711 1721 1731 1741 1751 1761 1771 1800 2000 2200 2400 2600 2800 3000 3200 3400 3600 3800 4000 4210 4310 4330 4420 4510 4610 4710 4730 4820 4910 4930 5010 5030 5050 5070 5200 5311 5321 5331 5341 Disease burden index results 40.00 35.00 30.00 25.00 20.00 15.00 10.00 5.00 0.00 Case weights for claimed ACG in 2010 Plan Group 2 Plan Group 3 Plan Group 4 Plan Group 5 3.00 Disease burden index in 2010 by plan group 2.50 2.00 1.50 1.00 0.50 0.00 Plan group 1 Plan group 2 Plan group 3 Plan group 4 Plan group 5 Plan group 6

Disease burden - conclusions Disease burden is a function of: Age Gender Chronic conditions Other clinical interactions Access to benefits (including data considerations) Adjusting for the calculated disease burden allows all of these factors to be taken into account

Are South African regional healthcare consumption patterns explained by the underlying burden of disease and access to benefits? Applications

Axis Title Healthcare costs in the last six months of life 1.30 1.20 1.10 1.00 0.90 0.80 0.70 0.60 0.50 0.40 East Rand Johann esburg North & Surroun ds Johann esburg Potchef Pretoria stroom Last 6 months cost index by Tertiary Referral Region Bloemfo ntein Durban Rustenb urg West Coast & Karoo Cape Nelsprui Peninsu t la Vaal Triangle Maritzb urg Port Polokwa Overber Elizabet ne g h PLPM 24,508 24,486 23,971 23,415 23,194 19,287 19,186 19,000 18,133 17,210 16,586 16,572 16,233 16,108 14,756 14,072 13,739 12,077 Cost_index 1.19 1.19 1.17 1.14 1.13 0.94 0.93 0.92 0.88 0.84 0.81 0.81 0.79 0.78 0.72 0.68 0.67 0.59 Garden Route East London 30,000 25,000 20,000 15,000 10,000 5,000-70% Proportion of deaths in hospital 65% 60% 55% 50% 45% 40%

Regional variation Proportion of DHMS lives Paid PLPM Disease Burden Index adjusting for access to benefits Paid PLPM Disease Burden Adjusted

Supply of hospital beds Estimate the demand for hospital beds in South Africa Find areas with oversupply Estimate impact Intervene

Actual beds per referral region 6,000 5,000 4,000 3,000 2,000 1,000 - Variability in supply of hospital beds

Actual beds per 1,000 lives per region 6.00 Actual beds per 1,000 lives in 2012 5.00 4.00 3.00 2.00 1.00 - Variability in supply of hospital beds per 1,000 lives

Expected (required) beds methodology Use ACGs as risk adjustment tool Based on 2008 bed days per 1,000 lives per ACG Calculate required overall bed days in each region for 2012 Assumptions: ACG (disease burden) distribution of lives for DHMS is representative of population 2008 provides a good benchmark for hospitalisation need of members by ACG Compare actual bed days per region in 2012 to required Back test for earlier years

ACG model results (70% occupancy) A/E 2.50 2.00 1.50 1.00 0.50 - (0.50) (1.00) (1.50) (2.00) 2012 Over/under supply of beds per 1,000 lives per region assuming 70% occupancy

Over/under supply of beds at different occupancy rates Over/under supply of beds per 1,000 lives per region assuming different levels of occupancy

Understanding the impact of competition Other Herfindahl concentration index Measure of competition among hospital networks in regions Concentration index measures representation of network by number of beds relative to industry Example Representation Concentration Index Indication JHB & Surrounds Combination 0.28 Moderate concentration East London 100% Life Healthcare 1 High concentration

Herfindahl concentration index results Drainage region Concentration index 2012 Major network in region Oversupply of beds in 2012 East London 1.00 Life Healthcare -0.61 Polokwane 0.86 Mediclinic -1.47 Overberg 0.80 Mediclinic -0.06 Potchefstroom 0.55 NHN 1.27 Garden Route 0.54 Mediclinic 0.24 Nelspruit 0.51 Mediclinic -0.77 Port Elizabeth 0.50 Netcare -1.09 West Coast & Karoo 0.49 Mediclinic 0.27 Durban 0.44 Life Healthcare -0.02 Maritzburg 0.42 NHN 2.07 East Rand 0.38 Netcare 0.62 Johannesburg 0.36 Netcare 0.38 Bloemfontein 0.32 NHN 1.67 Vaal Triangle 0.32 Mediclinic 1.79 Rustenburg 0.32 Life Healthcare and Netcare -0.49 Johannesburg North & Surrounds 0.28 Netcare -0.37 Pretoria 0.26 Even split of networks 0.53 Cape Peninsula 0.25 Even split of networks 0.51 Total 0.25 Even split of networks 0.18

Admission rate Does competition impact the admission rate? 35% 30% 25% 20% 15% 10% 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 Competitive Admission rate Linear (Admission rate) Concentration index High concentration Admission rate adjusted for disease burden Linear (Admission rate adjusted for disease burden) After adjusting for disease burden, the admission rate is higher in areas with high competition (low concentration)

Beds/1,000 lives Does competition impact the supply of beds? 7.00 6.00 5.00 4.00 Correlation: -60% 3.00 2.00 1.00-0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 Competitive Concentration index High concentration More beds in highly competitive areas Is this required based on disease burden?

Beds/1,000 lives Does competition impact the supply of beds? 7.00 6.00 5.00 4.00 3.00 2.00 1.00-0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 Competitive Concentration index High concentration Actual beds per 1,000 lives Linear (Actual beds per 1,000 lives) Required beds per 1,000 lives Linear (Required beds per 1,000 lives) Disease burden does not explain the difference in number of beds between competitive and concentrated areas

Concluding remarks Understanding referral regions gives insight into healthcare costs and throughput Costs are variable across referral regions Variation in costs may be driven by various underlying factors such as disease burden, supply of beds, concentration and competition mix