i
The analysis of REF shadow returns 2010
September 2011
Page ii
Table of Contents
EXECUTIVE SUMMARY ................................................................................................................................................................ 1
1.1 THE REF SHADOW PERIOD .......................................................................................................................................................... 2 1.2 PURPOSE OF THE REPORT ........................................................................................................................................................... 2 2.1 CASE DEFINITIONS AND BENCHMARKS ............................................................................................................................................ 3 2.1.1 ENTRY AND VERIFICATION CRITERIA .......................................................................................................................................... 3 2.2 REF DATA SUBMITTED FOR ANALYSIS ............................................................................................................................................ 4 2.3 CATEGORISATION AND THE ASSESSMENT OF SUBMITTED DATA ............................................................................................................ 4 2.3.1 CATEGORISATION ................................................................................................................................................................. 4 2.3.2 EVALUATION OF CLINICAL CREDIBILITY OF SUBMISSIONS ................................................................................................................ 6 2.3.3 REF RISK FACTORS WITH DEVIATIONS WITH SIGNIFICANT FINANCIAL IMPACT ................................................................................... 10 2.3.3.1 BIPOLAR MOOD DISORDER ............................................................................................................................................... 12 2.3.3.2 CHRONIC OBSTRUCTIVE PULMONARY DISEASE ..................................................................................................................... 12 2.3.3.3 DIABETES MELLITUS 2 ..................................................................................................................................................... 12 2.3.3.4 HYPERLIPIDEAMIA .......................................................................................................................................................... 13 2.3.3.5 THREE SIMULTANEOUS CONDITIONS ................................................................................................................................... 13 2.3.4 EVALUATION OF REF SUBMISSIONS BY ADMINISTRATOR ............................................................................................................. 13 2.3.4.1 CATEGORISATION BY ADMINISTRATOR ................................................................................................................................ 13 2.3.5 REF PRICE BY AGE AND COMMUNITY RATE ANALYSES ................................................................................................................ 15
List of Tables Table 1: Percentage of beneficiaries included in 2009 REF returns ............................................................................ 4
Table 2: Categories and groups used in the analysis of REF returns ........................................................................... 4
Table 3: The 10 most frequently occurring chronic diseases (December 2010) .............................................................. 7
Table 4: Expected and actual estimated REF risk factor costs ................................................................................. 11
Table 5: Scheme categories by administrator (December 2010) .............................................................................. 14
Table 6: Risk rates by month .......................................................................................................................... 20
List of Figures
Figure 1: Data quality groups by month ................................................................................................................ 5
Figure 2: All Schemes: Total CDL count per 1 000 lives (2010) .............................................................................. 6
Figure 3: Distribution of chronic disease (December 2010) .................................................................................... 7
Figure 4: Total cost load by REF risk factor group (December 2010) ........................................................................ 8
Figure 5: Total cost load by REF risk factor group (December 2010) ........................................................................... 9
Figure 6: Age-specific REF risk factor cost pbpm (December 2010) ............................................................................ 9
Figure 7: Price by age: All administrators (2010) ............................................................................................... 15
Figure 8: Price by age: All administrators (2009) ................................................................................................... 16
Figure 9: Price by age: All administrators (2008) ................................................................................................... 16
Figure 10: Price by age: All administrators (2007) ............................................................................................... 17
Figure 11: Price by age: All administrators (2006) ............................................................................................... 17
Figure 12: Price by age: All administrators (2005) ............................................................................................... 18
Figure 13: Number of beneficiaries by payment band (December 2010): Alternative payment intervals ............................ 19
Page 1
Executive summary
The South African private health environment is characterised by the presence of many competing robust medical aid
schemes. Medical schemes are by law required to offer a minimum set of benefits called “Prescribed Minimum Benefits”
(PMB). The aim of PMBs is to reduce the incentive of medical schemes to select risk. Differentiation by age or medical
condition of the beneficiary is not allowed, and most medical schemes apply “open enrolment”, i.e. they must accept any
applicant at standard rates (community rating). Scheme-specific community rating implies that contributions are based
on risk exposure of each scheme, which is a function of, amongst other things, the age, gender, morbidity structure and
the size of membership. The unequal distribution of risk between medical schemes has remained despite the
introduction of open enrolment, community rating and PMBs. The evaluation of REF shadow data in the past five years
has shown that the distortion in risk exposure of medical schemes is too large to be considered random. The December
2010 calculations of the cost of community-rated PMBs based on age distribution, CDL, HIV and Maternity data show
that the cost of PMBs for a scheme with the most unfavourable age structure is about R232 above the industry average,
whereas the cost for a scheme with the most favourable age structure is about R148 below the industry average.
The major drive of the office in the past 5 years has been to ensure that the industry is able to produce high quality data.
Major improvements have been noted between 2005 and 2007. A big dip in the quality of submitted data was recorded
in 2009. No major improvements have been achieved in 2010. The same group of medical schemes representing nearly
30% of beneficiaries continue to submit sub-standard data.
The 2010 REF Annual Report indicates that data was analysed for 99% of medical scheme beneficiaries. The
proportion of schemes submitting poor data ranged from 17% - 34%. Poor data quality was mainly because of major
differences between REF and Statutory Returns beneficiary numbers. The office was not able to confidently adjudicate
the accuracy of reported CDL data because of the outdated benchmark data. The approach here has mainly been to
look at trends over time, epidemiological and clinical soundness of the reported data. A new REF study is underway to
establish the correct CDL benchmarks.
Page 2
1 Introduction
The South African private health environment is characterised by the presence of many competing robust medical aid
schemes. Medical schemes are required by law to offer a minimum set of benefits called “Prescribed Minimum Benefits”
(PMB). The aim of PMBs is to reduce the incentive of medical schemes to select risk. Differentiation by age or medical
condition of the beneficiary is not permitted, and most medical schemes apply “open enrolment”, i.e. they must accept
any applicant at standard rates (community rating). Scheme-specific community rating implies that contributions are
based on risk exposure of each scheme, which is a function of, amongst other things, the age, gender, morbidity
structure and the size of membership. The unequal distribution of risk between medical schemes has remained despite
the introduction of open enrolment, community rating and PMBs. The evaluation of REF shadow data in the past five
years has shown that the distortion in risk exposure of medical schemes is too large to be considered random. The
December 2010 calculations of the cost of community-rated PMBs based on age distribution, CDL, HIV and Maternity
data show that the cost of PMBs for a scheme with the most unfavourable age structure is about R232 above the
industry average, whereas the cost for a scheme with the most favourable age structure is about R148 below the
industry average.
1.1 The REF shadow period
As part of the Risk Equalisation Fund (REF) shadow period, which started in January 2005, schemes submit
consolidated monthly REF returns to the Council for Medical Schemes (CMS) on a quarterly basis. The main purpose of
the shadow period is to give schemes and the CMS an opportunity to prepare for a system of risk equalisation and to
test the risk equalisation formula. This entails the development of specific skills and development of systems to
administrate the REF.
1.2 Purpose of the report
The purpose of this report is to assist individual schemes to interpret the scheme-specific results given on the statutory
returns portal on the CMS website1. Schemes should consider this report to assist in the adjustment of processes and
systems to meet the requirements of the REF before submitting future REF returns.
This report contains high-level information with more details provided in the various annexure.
1 The CMS statutory returns portal is available at:
https://www.medicalschemes.com/Returns/login.aspx Note that a username and password is required to access scheme-specific information
https://www.medicalschemes.com/Returns/login.aspx
Page 3
2 REF data and methods: 2009 REF submissions
2.1 Case definitions and benchmarks
2.1.1 Entry and verification criteria
Version 52 of the REF entry and verification criteria was used to identify qualifying beneficiaries for 2010. The following
changes were made to Version 5 since the publication of Version 4 guidelines published on 06 October 2008.
i. The determination of age bands in paragraph 3.4 has been improved by adding that the value will always be an
integer to assist in the calculation of age-bands.
ii. The discipline codes for Rheumatologist changed in Q3 2009 from discipline 31 to discipline 18 sub-disciplines
12. The affected algorithm, Systemic Lupus Erythematosus has been updated accordingly.
iii. The admission date as evidence of hospitalisation has been explicitly stated in the proof of treatment rules for
Multiple Sclerosis.
iv. The table layout for Diabetes mellitus, Hyperlipidaemia and Maternity has been simplified for ease of reference.
In addition, reference has been made to the version of the Framingham risk score to be used for authorisation
in the Hyperlipidaemia algorithm.
v. The Rheumatoid Arthritis table has been revised resulting in the removal of the disciplines where a diagnosis
needs to be confirmed for medicine management other than the Disease Modifying Agents (DMARDS).
vi. ICD-10 code M05.89: Other seropositive rheumatoid arthritis, unspecified has been inserted to complete the list
of ICD-10 codes for Rheumatoid Arthritis.
The 2010 REF weighting table3 is based on the 2005 REF study4. The method applied to adjust the table for inflation
has been described previously5.
2 Version 5: Guidelines for the Identification of Beneficiaries with REF Risk Factors.
http://www.medicalschemes.com/files/Risk%20Equalisation%20Fund/VersionFiveOfEnVGuidelines.pdf 3 9 March 2010, “Combined REF Grid Count in REF Contribution Table 2010”
http://www.medicalschemes.com/files/Risk%20Equalisation%20Fund/REFWTnCount2010.xls 4 03 May 2006, “Recommendations by the Risk Equalisation Technical Advisory Panel to the Council for Medical Schemes - Proposed
Methodology for the Risk Equalisation Fund Contribution Table 2007: RETAP Recommendations Report No. 8 (20 April 2006)”
http://www.medicalschemes.com/publications/ZipPublications/Risk%20Equalisation%20Fund/REFCT%202007%20Methodology%20March%
202006%20vFinal.pdf 5 09 March 2010, “Methodology to determine the REF weighting table and REF count table for 2010”
http://www.medicalschemes.com/files/Risk%20Equalisation%20Fund/MethodologyToDetermineREFWTFor2010.pdf
http://www.medicalschemes.com/files/Risk%20Equalisation%20Fund/VersionFiveOfEnVGuidelines.pdfhttp://www.medicalschemes.com/files/Risk%20Equalisation%20Fund/REFWTnCount2010.xlshttp://www.medicalschemes.com/publications/ZipPublications/Risk%20Equalisation%20Fund/REFCT%202007%20Methodology%20March%202006%20vFinal.pdfhttp://www.medicalschemes.com/publications/ZipPublications/Risk%20Equalisation%20Fund/REFCT%202007%20Methodology%20March%202006%20vFinal.pdfhttp://www.medicalschemes.com/files/Risk%20Equalisation%20Fund/MethodologyToDetermineREFWTFor2010.pdf
Page 4
2.2 REF data submitted for analysis
Table 1 indicates that by December 2010, the number of beneficiaries reported in the REF grids matched those reported
in the Statutory Returns for the same period. Big variations were observed in March and June 2010.
Table 1: Percentage of beneficiaries included in 2009 REF returns
Statutory returns submissions
REF submissions REF Beneficiaries as % SR
Beneficiaries
Mar 2010 7 807 264 8 051 355 103.13
Jun 2010 7 885 330 8 122 120 103.00
Sep 2010 8 102 360 8 115 155 100.16
Dec 2010 8 270 536 2 210 371 99.27
2.3 Categorisation and the assessment of submitted data
Similar to the previous analyses of REF returns, in assigning submissions to categories, the CMS considered the
deviation from expected count values, deviations from statutory returns, and the evaluation of clinical credibility. Each
submission was evaluated by at least two analysts. In instances where the analysts assigned discordant categories to a
scheme, the REF team evaluated the submission.
2.3.1 Categorisation
REF submissions were categorised by REF analysts in accordance with the categories listed in Table 2 below. The
table groups categories as representative of “fair data”, “CDL definitions applied poorly”, or “poor data”, in accordance
with the definitions in Annexure B.
Table 2: Categories and groups used in the analysis of REF returns
Category6 Short description Group
3 L Some concerns, CDLs are reported at very low levels
Fair data 3 Some concerns
3 H Some concerns, CDLs are reported at very high levels
4 Many more beneficiaries in REF returns than in statutory returns Poor data
5 No REF data or substantially less than in statutory returns
6 Much lower than expected CDLs CDL definitions applied poorly 7 Much higher than expected CDLs
8 Maternity data unlikely Poor data
9 Combinations of the above or other serious errors in submitted data
6 Note that categories 1 and 2, which were previously used to identify “good” datasets with minor and no concerns respectively, have been
discontinued.
Page 5
Figure 1 indicates that the proportion of schemes submitting fair data averaged 77% for January to December 2010.
The average in December 2009 was 73% and 71% in December 2008. The proportion of schemes reporting
problematic data or applied CDL definitions inadequately has remained the same for quite some time now.
Figure 1: Data quality groups by month
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Jan-10 Feb-10 Mar-10 Apr-10 May-10 Jun-10 Jul-10 Aug-10 Sep-10 Oct-10 Nov-10 Dec-10
Jan-10 Feb-10 Mar-10 Apr-10 May-10 Jun-10 Jul-10 Aug-10 Sep-10 Oct-10 Nov-10 Dec-10
Problematic 19% 20% 24% 16% 17% 25% 12% 10% 17% 18% 17% 26%
Definition 4% 6% 6% 6% 6% 6% 5% 5% 5% 2% 3% 3%
Fair 77% 74% 70% 78% 77% 69% 83% 85% 78% 80% 80% 71%
Page 6
2.3.2 Evaluation of clinical credibility of submissions
Figure 2 indicates that the actual rate of reported CDL counts was higher than expected levels throughout the year. The
upward trend seen in the 2010 submission has been observed in previous years. This trend may be due to the slow
uptake of CDL benefits by beneficiaries and new members joining the scheme at the beginning of the year. The age
profile submitted by medical schemes can also distort the expected rates, especially in cases where there are gross
errors or lack of consistency in the age profile.
Figure 2: All Schemes: Total CDL count per 1 000 lives (2010)
Total CDL Conditions 2010
0.0
20.0
40.0
60.0
80.0
100.0
120.0
140.0
Jan
-2010
Feb
-2010
Mar-
2010
Apr-
2010
Ma
y-2
01
0
Jun
-2010
Jul-
2010
Aug
-2010
Sep
-2010
Oct-
2010
Nov
-2010
Dec
-2010
rate
s pe
r 10
00 l
ives
Actual Expected
Page 7
Figure 3 shows the large burden of cardiac and associated conditions, highlighting that lifestyle diseases are prevalent.
The 10 most common chronic conditions are listed in Table 3. This list has remained the same as in 2009 except for the
change in order for two of the conditions.
Figure 3: Distribution of chronic disease (December 2010)
Table 3: The 10 most frequently occurring chronic diseases (December 2010)
Top 10 CDL conditions 2010
Order in 2010
Order in 2009
CDL Number % of CDL % of population
1 1 Hypertension 326 755 30.6% 4.0%
2 3 Diabetes mellitus 2 154 494 14.5% 1.9%
3 2 Hyperlipidaemia 147 058 13.8% 1.8%
4 4 Asthma 101 038 9.5% 1.2%
5 5 HIV / AIDS 77 482 7.3% 0.9%
6 6 Ischaemic heart disease 49 694 4.7% 0.6%
7 8 Hypothyroidism 35 511 3.3% 0.4%
8 7 Cardiomyopathy 34 729 3.3% 0.4%
9 9 Epilepsy 28 968 2.7% 0.4%
10 10 Diabetes mellitus 1 20 127 1.9% 0.2%
Other - 91 236 8.5% 1.1%
Total 1 067 092 100.0% 0.0%
Dec 2010All Administrators Total CDL Conditions
0
20,000
40,000
60,000
80,000
100,000
120,000
140,000
Und
er 1
1-4
5-9
10-1
4
15-1
9
20-2
4
25-2
9
30-3
4
35-3
9
40-4
4
45-4
9
50-5
4
55-5
9
60-6
4
65-6
9
70-7
4
75-7
9
80-8
4
85+
Hypertension Hyperlipidaemia Coronary Artery Disease
12: DM1 & 13: DM2 Other Asthma
HIV/AIDS Total CDL Conditions
Page 8
Figure 5 shows that the bulk of REF risk factor costs7 are included in the “NON” column (49.4%), indicating the
importance of age as a risk factor in REF.
Figure 5 presents the total cost load by REF risk factor groups, indicating the importance of lifestyle diseases, maternity,
and multiple chronic diseases8.
Figure 5 demonstrates the risk factor group costs per beneficiary per month by age.
Figure 4: Total cost load by REF risk factor group (December 2010)
7 Note that REF risk factor costs are based on the weights published in the REF weighting tables, and that the weight of a specific risk factor (E.g. Hypertension),
includes the costs included in the “NON” column. The cost estimates published here are the numbers of actual cases reported in the industry in December 2008,
multiplied by the values in the REF weighting table. 8 For the purposes of the illustration, CDL are grouped together as follows: Lifestyle diseases HYP, IHD, HYL, DM2 Other cardiac CMY, CHF, DYS Multiple chronic diseases CC2, CC3, CC4 Psychiatric BMD, SCZ Respiratory AST, COP, BCE Endocrine DM1, TDH, ADS, DBI Neurologic EPL, MSS Autoimmune RHA, SLE, CSD, IBD Other HAE, PAR, GLC
NON, 49.4%
HYP, 7.3%
DM2, 5.1%
HYL, 3.9%
HIV/AIDS, 3.8%
IHD, 2.8%
CMY, 2.5%
Other, 27.6%
Page 9
Figure 5: Total cost load by REF risk factor group (December 2010)
Figure 6: Age-specific REF risk factor cost pbpm (December 2010)
All Administrators Jun-2010
R -
R 50,000,000.00
R 100,000,000.00
R 150,000,000.00
R 200,000,000.00
R 250,000,000.00 U
nd
er
1
1-4
5-9
10
-14
15
-19
20
-24
25
-29
30
-34
35
-39
40
-44
45
-49
50
-54
55
-59
60
-64
65
-69
70
-74
75
-79
80
-84
85
+
NON Lifestyle Diseases Respiratory Neurologic HIV MAT
Psychiatric Renal Autoimmune Other Cardiac Endocrine Other
All Administrators Dec-2010
-
200
400
600
800
1,000
1,200
Un
de
r 1
1-4
5-9
10
-14
15
-19
20
-24
25
-29
30
-34
35
-39
40
-44
45
-49
50
-54
55
-59
60
-64
65
-69
70
-74
75
-79
80
-84
85
+
NON Lifestyle Diseases Respiratory Neurologic HIV
MAT Psychiatric Renal Autoimmune Other Cardiac
Endocrine Other ICR
Page 10
2.3.3 REF risk factors with deviations with significant financial impact
Table 4 lists the expected as well as the estimated REF risk factor costs along with the actually reported count numbers.
The table shows the degree of deviation from the expected values. These are highlighted in red or blue in the A / E
column in Table 4. Risk factors that are highlighted in red are reported above expected levels and the trend increases
year on year. The numbers might be a true reflection of the industry‟s risk profile, a true epidemiologic shift, or PMB
„diagnosis creep‟ by providers. The reporting of two to multiple chronic conditions is consistently higher than at expected
levels year on year and not necessarily confined to schemes with aging populations.
Financially relevant conditions are bipolar mood disorder, chronic obstructive pulmonary disease, chronic renal failure,
diabetes mellitus, hyperlipidaemia, hypertension, HIV/AIDS and three simultaneous conditions. These are defined by
deviations larger than 0.5% of the total expected cost of the respective risk factors.
Page 11
Table 4: Expected and actual estimated REF risk factor costs
* “Diff (A-E)” means the difference between actual and reported values while “A / E” means actual divided by expected
Dec-2010
Diff (A-E) Expected Actual A/E
No CDL disease -65,680,191 1,534,139,716 1,468,459,525 96%
Addison's Disease -33,714 139,074 105,360 76%
Asthma -8,470,146 73,900,285 65,430,139 89%
Bronchiectasis -55,154 384,635 329,481 86%
Bipolar Mood Disorder 21,926,630 8,108,560 30,035,190 370%
Cardiac failure 118,287 - 118,287 0%
Cardiomyopathy -10,024,075 83,929,154 73,905,079 88%
CHF&CMY -9,905,788 83,929,154 74,023,365 88%
Chronic Obs. Pulmonary Disease -24,759,744 50,659,380 25,899,636 51%
Chronic Renal Disease 14,447,071 44,000,586 58,447,657 133%
Crohn's Disease -83,715 2,679,978 2,596,263 97%
Diabetes Insipidus 73,759 121,554 195,313 161%
Diabetes Mellitus 1 -643,407 42,939,769 42,296,362 99%
Diabetes Mellitus 2 68,585,492 83,191,836 151,777,329 182%
Dysrhythmias 8,683,759 14,286,955 22,970,714 161%
Epilepsy 3,302,132 31,315,315 34,617,448 111%
Glaucoma 1,596,977 8,598,148 10,195,125 119%
Haemophilia 333,713 1,457,930 1,791,644 123%
Hyperlipidaemia 20,598,306 96,141,727 116,740,034 121%
Hypertension 18,921,170 199,565,401 218,486,571 109%
Ulcerative Colitis 149,332 1,823,821 1,973,153 108%
Coronary Artery Disease 1,323,389 81,619,464 82,942,853 102%
Multiple Sclerosis -888,483 11,729,430 10,840,947 92%
Parkinson's Disease 2,093,606 7,922,299 10,015,906 126%
Rheumatoid Arthritis 701,919 14,774,083 15,476,003 105%
Schizophrenia 970,358 2,237,915 3,208,273 143%
Systemic LE 780,718 2,770,820 3,551,538 128%
Hypothyroidism 533,211 17,293,593 17,826,803 103%
HIV/AIDS 32,737,646 81,069,904 113,807,550 140%
- -
Two simultaneous conditions 16,128,673 52,427,887 68,556,560 131%
Three simultaneous conditions 30,360,877 32,604,320 62,965,197 193%
Four or more simultaneous conditions 14,125,109 8,815,903 22,941,011 260%
Maternity Events 12,276,439 223,352,091 235,628,530 105%
Total CDL Conditions 120,181,391 881,591,715 1,001,773,106 114%
Multiple CDL Conditions 60,614,659 93,848,110 154,462,769 165%
Total 160,129,944 2,814,001,536 2,974,131,479 106%
Amount from REF by Condition
Page 12
2.3.3.1 Bipolar mood disorder
Levels of BMD are reported at rates two to four times higher than expected across most schemes. This trend has been
steadily increasing year-on-year. The therapeutic algorithm which guides the treatment of this illness was published late
in 2009 by the department of health.
WHO ranks depression, alcohol related disorders, schizophrenia and bipolar mood disorder to be the top ranking
disease. It is projected that unipolar depression will be the highest t ranking mental illness in 2030 (Mathers et al, 2006).
In low and middle income countries HIV is linked to increasing burden of depression. In South Africa the most common
mental health conditions are anxiety disorder, alcohol disorders followed by major depression (William et al, 2008). The
most prevalent mental health conditions are either not included in PMB conditions or the benefits are not adequate. The
increased variance between expected and actual costs in BMD can be attributable to misclassification of conditions i.e.
patients with unipolar depression and other mental health illnesses such as substance misuse, anxiety disorders etc
may be reported as bipolar is a PMB, actual increase in burden of BMD (meaning that the expected rates under-
estimates the true rates of BMD).
Bipolar mood disorder peaks between 35 and 44 years. This is similar to the mental health survey in South Africa that
reported a peak at this age. Although for most medical schemes age peak is at 35-44, the BMD rates in Metropolitan
Health group peak in the 20‟s. Both Discovery Health (PTY) LTD and Metropolitan health group have high rates of
Bipolar Mood disorder and Medscheme the lowest rates. Total cost of Bipolar Mood disorder for the whole industry is
R30 Million and cost per beneficiary was around R1 700.
2.3.3.2 Chronic obstructive pulmonary disease
The reported respiratory conditions, notably asthma and chronic obstructive pulmonary disease, have persistently lower
count rates than expected across submissions throughout the REF shadow period.
2.3.3.3 Diabetes mellitus 2
Schemes consistently reported DM2 above the expected levels. In the 2000 projection of DM2, WHO estimated that the
prevalence of DM2 would almost double in 2030, increasing prevalence of obesity may actually increase the prevalence
even more. SA has reported high rates of obesity (SADHS, 2003) and the difference between the actual and expected
rates can be attributable to the increasing burden of obesity and diabetes mellitus. In a burden of disease study by
Bradshaw et al, Chronic disease of lifestyle where rated second to infectious diseases.
The Metropolitan Health Corporate (Pty) Ltd administered schemes have reported the highest rates of DM2, at
approximately two times the expected levels. Discovery (Pty) Ltd reported the lowest rates of DM2.
Page 13
2.3.3.4 Hyperlipideamia
In the age groups younger than 50 years, expected and actual number of cases is similar. In the beneficiary over the
age of 50 years, expected rates are lower than the actual rates. The deviance in the older age group can be attributes to
the under-estimation of expected values. Momentum medical scheme administrators and Medscheme Holdings are
reporting higher rates of hyperlipideamia and discovery the lowest. The cost hyperlipideamia is 116 Million. Schemes
paid 20 Million more than expected to manage this condition.
2.3.3.5 Three simultaneous conditions
Three simultaneous conditions are reported at 193% of expected levels. Metropolitan Health Corporate (Pty) Ltd
administered schemes have reported higher rates when compared with other administrator groups. Schemes
administered by Discovery Health (Pty) Ltd reported low rates of multiple conditions. It should be noted that chronic
diseases of lifestyle tends to co-occur. With both increasing burden of chronic disease of lifestyle and HIV/AIDS rates of
co-occurrence of these conditions are increasing.
2.3.4 Evaluation of REF submissions by administrator
2.3.4.1 Categorisation by administrator
Table 5 shows the number of schemes by administrator and category in December 2010. Twenty schemes (21%) were
classified as category 9 schemes. Most of category 9 schemes were self-administered (35%). Allcare Administrators
(Pty) Ltd and Status Medical Aid Administrators (Pty) Ltd had 3 scheme each classified as category 9, making up the
second largest group of administrators with medical schemes that submitted poor data (30%).
Page 14
Table 5: Scheme categories by administrator (December 2010)
Administrator Category
Frequency Row Pct 3 3H 4 5 6 7 9 Total
AGILITY GLOBAL HEALTH SOLUTIONS AFRICA 1 100.00
0 0.00
0 0.00
0 0.00
0 0.00
0 0.00
0 0.00
1
ALLCARE ADMINISTRATORS PTY LTD 1 25.00
0 0.00
0 0.00
0 0.00
0 0.00
0 0.00
3 75.00
4
DISCOVERY HEALTH PTY LTD 10 83.33
0 0.00
0 0.00
0 0.00
1 8.33
0 0.00
1 8.33
12
ETERNITY PRIVATE HEALTH FUND ADMINISTRATORS PTY LTD
1 100.00
0 0.00
0 0.00
0 0.00
0 0.00
0 0.00
0 0.00
1
MEDSCHEME HOLDINGS PTY LTD 15 71.43
5 23.81
0 0.00
0 0.00
0 0.00
0 0.00
1 4.76
21
METROPOLITAN HEALTH CORPORATE PTY LTD 4 44.44
3 33.33
0 0.00
1 11.11
0 0.00
0 0.00
1 11.11
9
METROPOLITAN HEALTH PTY LTD 3 75.00
0 0.00
0 0.00
1 25.00
0 0.00
0 0.00
0 0.00
4
MOMENTUM MEDICAL SCHEME ADMINISTRATORS (PTY) LTD
8 72.73
1 9.09
0 0.00
1 9.09
0 0.00
0 0.00
1 9.09
11
PRIVATE HEALTH ADMINISTRATORS (A DIVISION OF SWEIDAN TRUST (PTY) LTD)
0 0.00
0 0.00
0 0.00
0 0.00
0 0.00
0 0.00
1 100.00
1
PROFESSIONAL MEDICAL SCHEME ADMINISTRATORS (PTY) LTD
1 100.00
0 0.00
0 0.00
0 0.00
0 0.00
0 0.00
0 0.00
1
PROVIDENCE HEALTHCARE RISK MANAGERS PTY LTD
3 60.00
2 40.00
0 0.00
0 0.00
0 0.00
0 0.00
0 0.00
5
SANLAM HEALTHCARE MANAGEMENT (PTY) LTD 0 0.00
0 0.00
0 0.00
0 0.00
0 0.00
0 0.00
1 100.00
1
SECHABA MEDICAL SOLUTIONS (PTY) LTD 1 100.00
0 0.00
0 0.00
0 0.00
0 0.00
0 0.00
0 0.00
1
SELF-ADMINISTERED 5 38.46
0 0.00
0 0.00
0 0.00
1 7.69
0 0.00
7 53.85
13
SIGMA HEALTH FUND MANAGERS (PTY) LTD 0 0.00
0 0.00
0 0.00
0 0.00
0 0.00
1 100.00
0 0.00
1
STATUS MEDICAL AID ADMINISTRATORS PTY LTD 2 40.00
0 0.00
0 0.00
0 0.00
0 0.00
0 0.00
3 60.00
5
THEBE YA BOPHELO HEALTHCARE ADMINISTRATORS PTY LTD
0 0.00
0 0.00
0 0.00
0 0.00
0 0.00
0 0.00
1 100.00
1
V MED ADMINISTRATORS (PTY) LTD 2 50.00
0 0.00
1 25.00
1 25.00
0 0.00
0 0.00
0 0.00
4
Total 57 11 1 4 2 1 20 96
Page 15
2.3.5 REF price by age and community rate analyses
The REF price by age curve demonstrates the combined risk of each of the reported REF risk factors on schemes in
comparison to the expected risk attributable to the REF risk factors. It should be noted that the expected REF risk factor
rates applied in this section are the 2005 REF study rates. These curves, therefore express the deviations from the
expected study rates. A small administrator that administrates a single scheme with a very low (or high) risk might have
very low (or high) price by age curves that could, in fact, be a true reflection of the particular scheme‟s true risk. Large
fluctuations and trends should however, not be influenced by this single standard benchmark for REF risk factors.
The price by age curve for all administrators put together is gradually moving away from the expected price by age per
beneficiary per month (see Figure 7 - Figure 12). This is mainly due to the use of an expected count rate benchmark
that is likely to be outdated.
Figure 7: Price by age: All administrators (2010)
Quarter EndREF CurvesAll Administrators
0
100
200
300
400
500
600
700
800
900
1,000
1,100
1,200
1,300
1,400
Un
de
r 1 1-4
5-9
10
-14
15
-19
20
-24
25
-29
30
-34
35
-39
40
-44
45
-49
50
-54
55
-59
60
-64
65
-69
70
-74
75
-79
80
-84
85
+
Pri
ce
pe
r b
en
efi
cia
ry p
er
mo
nth
Expected pbpm
Mar 10
Jun 10
Sep 10
Dec 10
Page 16
Figure 8: Price by age: All administrators (2009)
Figure 9: Price by age: All administrators (2008)
Quarter EndREF CurvesAllSchemesAdmin: All Administrators
0
100
200
300
400
500
600
700
800
900
1,000
1,100
1,200 U
nd
er 1
1-4
5-9
10
-14
15
-19
20
-24
25
-29
30
-34
35
-39
40
-44
45
-49
50
-54
55
-59
60
-64
65
-69
70
-74
75
-79
80
-84
85
+
Pri
ce
pe
r b
en
efi
cia
ry p
er
mo
nth
Expected pbpm
Mar 09
Jun 09
Sep 09
Dec 09
0
100
200
300
400
500
600
700
800
900
1,000
1,100
1,200
Un
de
r 1
1-4
5-9
10
-14
15
-19
20
-24
25
-29
30
-34
35
-39
40
-44
45
-49
50
-54
55
-59
60
-64
65
-69
70
-74
75
-79
80
-84
85
+
Pri
ce
pe
r b
en
efi
cia
ry p
er
mo
nth
Expected pbpm
Mar 08
Jun 08
Sep 08
Dec 08
Quarter EndREF CurvesAllSchemesAdmin: All Administrators
Page 17
Figure 10: Price by age: All administrators (2007)
Figure 11: Price by age: All administrators (2006)
Quarter EndREF CurvesAllSchemesAdmin: All Administrators
0
100
200
300
400
500
600
700
800
900
1,000
1,100U
nd
er
1
1-4
5-9
10
-14
15
-19
20
-24
25
-29
30
-34
35
-39
40
-44
45
-49
50
-54
55
-59
60
-64
65
-69
70
-74
75
-79
80
-84
85
+
Pri
ce
pe
r b
en
efi
cia
ry p
er
mo
nth
Expected pbpm
Mar 07
Jun 07
Sep 07
Dec 07
0
100
200
300
400
500
600
700
800
900
1,000
1,100
1,200U
nd
er
1
1-4
5-9
10
-14
15
-19
20
-24
25
-29
30
-34
35
-39
40
-44
45
-49
50
-54
55
-59
60
-64
65
-69
70
-74
75
-79
80
-84
85
+
Pri
ce
pe
r b
en
efi
cia
ry p
er
mo
nth
Expected pbpm
Sep 07
Dec 07
Jun 07
Mar 07
Quarter EndREF CurvesAllSchemesAdmin: All Administrators
0
100
200
300
400
500
600
700
800
900
1,000
1,100
Un
de
r 1
1-4
5-9
10
-14
15
-19
20
-24
25
-29
30
-34
35
-39
40
-44
45
-49
50
-54
55
-59
60
-64
65
-69
70
-74
75
-79
80
-84
85
+
Pri
ce
pe
r b
en
efi
cia
ry p
er
mo
nth
Expected pbpm
Mar 06
Jun 06
Sep 06
Dec 06
0
100
200
300
400
500
600
700
800
900
1,000
1,100
1,200
Un
de
r 1
1-4
5-9
10
-14
15
-19
20
-24
25
-29
30
-34
35
-39
40
-44
45
-49
50
-54
55
-59
60
-64
65
-69
70
-74
75
-79
80
-84
85
+
Pri
ce
pe
r b
en
efi
cia
ry p
er
mo
nth
Expected pbpm
Sep 06
Dec 06
Jun 06
Mar 06
Page 18
Figure 12: Price by age: All administrators (2005)
Price by age graphs and community rate analyses for the major administrators appear Annexure E9 of this report.
9 Annexure available on http://www.medicalschemes.com/Publications.aspx
REF Curves Quarter EndAllSchemesAdmin: All Administrators
0
100
200
300
400
500
600
700
800
900
1,000
1,100
1,200
Un
de
r 1
1-4
5-9
10
-14
15
-19
20
-24
25
-29
30
-34
35
-39
40
-44
45
-49
50
-54
55
-59
60
-64
65
-69
70
-74
75
-79
80
-84
85
+
Pri
ce
pe
r b
en
efi
cia
ry p
er
mo
nth
Expected pbpm
Mar-2005
Jun-2005
Sep-2005
Dec-2005
http://www.medicalschemes.com/Publications.aspx
Page 19
3 The potential financial impact on schemes
The financial impact of REF on a particular scheme is dependent on the difference between the scheme‟s community
rate and the industry community rate. This implies that even if a scheme did submit good data, but the rest of the
industry submitted poor data, the scheme risk estimate will be incorrect.
The financial impact by payment band on the beneficiaries is illustrated in Figure 13. For December 2010, 275 950
(3.36%) beneficiaries will receive R150.00 or more from REF and 38 570 (0.47%) will have to pay in between R125.01
and R150.00. (Theoretically, 70.44% beneficiaries will be net payers into REF.)
Figure 13: Number of beneficiaries by payment band (December 2010): Alternative payment intervals
0 38
57
0
25
59
6
77
65
0 79
44
57 16
74
56
0
31
72
28
9
42
03
14
78
77
29
23
66
62
27
17
47
16
04
63
27
43
84
27
59
50
0
500 000
1 000 000
1 500 000
2 000 000
2 500 000
3 000 000
3 500 000
Pay: M
ore
than
R150,0
0
PB
PM
Pay: R
125,0
1 t
o R
150,0
0
PB
PM
Pay: R
100,0
1 t
o R
125,0
0
PB
PM
Pay: R
75,0
1 t
o R
100,0
0
PB
PM
Pay: R
50,0
1 t
o R
75,0
0 P
BP
M
Pay: R
25,0
1 t
o R
50,0
0 P
BP
M
Pay: R
0 to
R25,0
0 P
BP
M
Receiv
e: R
0,0
1 to
R25,0
0
PB
PM
Receiv
e: R
25,0
1 to
R50,0
0
PB
PM
Receiv
e: R
50,0
1 to
R75,0
0
PB
PM
Receiv
e: R
75,0
1 to
R100,0
0
PB
PM
Receiv
e: R
100,0
1 to
R125,0
0
PB
PM
Receiv
e: R
125,0
1 to
R150,0
0
PB
PM
Receiv
e: M
ore
than
R150,0
0
PB
PM
Nu
mb
er
of b
en
efi
cia
rie
s
Full table
Page 20
For March the scheme risk varies from –R574.50 to R148.09. This means that the highest risk scheme will receive
R574.50 per beneficiary from REF and the lowest risk scheme has to pay R148.09 per beneficiary to REF. For
December the scheme risk varies from –R232.58 to R148.79. The standard deviation of the scheme risk for Q3 and Q4
is also smaller compared to Q1 and Q2, as shown in Table 6.
Table 6: Risk rates by month
Statistic Full Contribution Table
(Amount in rand)
March 2010 June 2010 September 2010 December 2010
Industry community rate 359.43 361.90 362.92 361.21
Minimum risk rate -574.50 -639.00 -238.72 -232.60
Maximum risk rate 148.09 161.45 174.72 148.79
Standard deviation 108.53 107.77 89.03 85.91
4 Conclusions
About twenty percent of schemes continue to submit data that is clearly of poor quality. The data quality issues mainly
involved scheme demographics. The age profile reported for Statutory Returns and REF differed significantly for nearly
21% of medical schemes in December 2010. The measurement of scheme risk is highly dependent on good quality
data.
4.1 Clinical credibility of submissions
As reported in previous reports, the number of schemes reporting poor quality clinical data has reduced dramatically.
Only three medical schemes were classified as either grossly under-reporting or over-reporting their CDL data. The
challenge the office faced in this analysis was the absence of a reliable chronic disease benchmark. Work is underway
in the office to update both the count and weighting tables.
000 - END - 000