The Costing of the Proposed Chronic Disease List Benefits in South African Medical Schemes in 2001

Transcription

1 The Centre for Actuarial Research (CARE) A Research Unit of the University of Cape Town In collaboration with The Costing of the Proposed Chronic Disease List Benefits in South African Medical Schemes in 2001 By Professor Heather McLeod, Professor Alan Rothberg, Leslie Pels, Sean Eekhout, Deus Bazira Mubangizi and Dr Therese Fish December 2002 ISBN X A Research Report Prepared under Contract for the Council for Medical Schemes

2 Centre for Actuarial Research (CARE) A Research Unit of the University of Cape Town Centre for Actuarial Research University of Cape Town Private Bag Rondebosch 7701 SOUTH AFRICA Telephone: +27 (0) Fax: +27 (0) care@commerce.uct.ac.za for Professor McLeod: hmcleod@iafrica.com Medscheme Integrated Care Division Private Bag X124 Bryanston 2021 SOUTH AFRICA Telephone: +27 (0) Fax: +27 (0) for Professor Rothberg: alanr@medscheme.co.za CARE Costing of Chronic Disease List Page ii

3 Executive Summary The Council for Medical Schemes invited tenders for three projects in March This report is presented in fulfilment of the contract in respect of the second of the three projects and deals with the costing of the proposed Chronic Disease List (CDL) benefits in medical schemes. The report on the first of the three projects, The Costing of Existing Prescribed Minimum Benefits in South African Medical Schemes in 2001, developed a base price for the current Prescribed Minimum Benefits (PMBs) in medical schemes. This report on the CDL conditions considers the additional cost of those conditions being added to the package. Proposed amendments to the Regulations under the Medical Schemes Act No. 131 of 1998 made a significant addition to PMBs to include treatment and medication for 28 chronic conditions (the CDL conditions). The final Regulations in November 2002 provide for the diagnosis, medical management and medication in respect of 25 chronic conditions to become part of PMBs on 1 January This report is therefore intended to inform stakeholders and policymakers on the cost of the inclusion of the CDL conditions. Data Extraction, Methodology and Cluster Analysis The Consortium requested and received data from the Medscheme Data Warehouse in August and September 2002, covering the full calendar year The data for the CDL study covers 43 options in 26 schemes. The credible exposure is million beneficiary months, giving an average exposure of beneficiaries for the year. Medscheme have developed a proprietary method of analysis that involves using statistical techniques to cluster benefit options into distinct groups based on option and demographic characteristics, primarily age of members, family size and percentage of African-Black membership. Research shows that the different clusters experience different benefit utilisation, costs and disease profiles. Provider behaviour has been shown to differ by cluster, even within the same hospital facility. Four distinct clusters have been used in the PMB analysis: High contains options with older, 'whiter' members with high utilisation; Medium-older contains options with medium utilisation and older members; Medium-younger contains options with medium utilisation and younger members; Low contains options with younger, 'blacker' members with low utilisation. The CDL study also contains a fifth cluster that was not present in the PMB study. The age last birthday on 1 January 2001 was obtained for the beneficiaries in the study. This enabled analysis to be done based in five years bands. The age profile of the study has more children and fewer elderly people than the industry. The ethnicity profile shows that there are more African /Black lives in the study than the industry. It was on this basis that the Consortium recommends using a weighted total to reduce the dominance of low cluster (to 50%) in order to better represent the industry. CARE Costing of Chronic Disease List Page iii

4 Data from the Statutory Returns to the Registrar was received from the Council for Medical Schemes in respect of calendar year This was used to determine the relationship between these million beneficiaries and the rest of the industry in that year, in respect of demographics, contribution levels and benefits paid and also for validation purposes. Identification of Beneficiaries with CDL diseases and Extraction of Costs The MediServe database at Medscheme Integrated Care contains details of all beneficiaries registered for chronic medicine benefits. Information is captured at the time of registration for chronic medicine and includes clinical information provided by the doctor and details of the medicines authorised for payment. The first step in the data extraction was to interrogate the MediServe database using the ICD-10 diagnosis coding recorded at the time of registration. All beneficiaries with at least one of the CDL-conditions were selected. Those beneficiaries with multiple CDL diseases could also be identified. The National Pharmaceutical Product Interface (NAPPI) is a list of codes that medical schemes use to identify pharmaceutical products for claims. All NAPPIs used by beneficiaries with CDL conditions were isolated. These were classified into three groups : Primary : clinically appropriate for the CDL conditions, as determined by Medscheme Integrated Care. This amounts to some CDL-NAPPIs. Secondary : all NAPPIs ever claimed under chronic benefits by any of the people in the study. This amounts to some CDL-NAPPIs. Total Tariff Amount: all NAPPIs ever claimed, whether under chronic medicine benefits or acute medicine benefits. This covers all medicines claimed by the beneficiaries in the study. The three sets of CDL-NAPPI codes were used to extract items claimed in respect of the beneficiaries with CDL diseases. It is important to note that claims were extracted that were lodged under both chronic and acute benefit limits. By taking items claimed under both limits the problem of dealing with chronic limits has been substantially reduced. The items claimed were then linked to the full tariff amount for those items. Rather than work with benefit or paid amounts, which may be heavily influenced by limits and copayments, the Consortium worked with the tariff or Blue Book amount for the items claimed. Total claims extracted from the Medscheme database was records, representing beneficiaries with chronic conditions. The total number of beneficiaries who claimed for one or more CDL conditions was , which was 76.8% of the number of who claimed for chronic conditions. The Total Tariff Amount for all beneficiaries in the schemes was R The Total Tariff Amount in respect of those with one or more CDL diseases was R Only a portion of this expenditure relates directly to the CDL conditions as a portion is related to the Secondary CDL-NAPPIs and acute drugs. On the basis that medicine benefits are paid only if the beneficiary is active on the scheme at the time of the dispensing of the claim, an individual s total annual exposure was determined by using the specific number of months for which he/she was active during the study period. CARE Costing of Chronic Disease List Page iv

5 Existing Coverage of CDL Conditions Ranchod & McLeod conducted a study of 169 open scheme options in the industry in They found that only eight options, which equates to 4.7%, were found to offer no chronic medicine benefits at all. The largest category of chronic medicine benefits was options that use monetary limits for chronic medicine, with 61% of options using this approach. A wide range of limits was found both in terms of level and structure. It was found that 36.1% of the options made use of a chronic ailment list in determining whether expenditure would be covered. The authors have no insight into the diseases on those chronic ailment lists, but it is anticipated that the major CDL conditions such as hypertension, asthma and diabetes mellitus would almost certainly be covered. In the CDL study, options in the Low cluster tend to make more use of benefit limits and co-payments in order to ration care. Options in the Middle and High clusters invariably use chronic disease management programmes to ration care. In the Medscheme environment, all options on disease management programmes provide cover for all the CDL diseases. Not all Low cluster options make use of chronic medicine management programmes. This does not affect the key parts of the CDL study as prevalence and cost have been determined directly from the NAPPI-coded medicine claims, rather than from registrations. Registrations for CDL Conditions In the paper review of the MediServe database, there were beneficiaries registered for chronic medicine in the schemes studied in The total number of beneficiaries with one or more CDL conditions was , which was 77.1% of the total chronic medicine registrations. The nine most common registrations accounted for 92.4% of all CDL registrations, with the other 18 conditions accounting for only 7.6%. Hypertension, asthma, diabetes mellitus and hyperlipidaemia were the most common conditions. The majority of registrations were for CDL conditions as a co-morbidity to another CDL condition. A single CDL condition is experienced by 62.5% of the registered beneficiaries. 25.0% have two conditions and the balance have multiple CDL conditions, with a maximum of eight separate conditions for a single beneficiary. There were combinations of the 27 CDL conditions. The existence of many multiple diseases makes the isolation of prevalence and costs for each CDL condition a complex task. The analytical work on the claims data was therefore performed separately for the single and multiple diseases Cost of Each CDL Condition The cost for each condition was determined based on the cost for providing medicine for each condition. A separate estimate for the cost of diagnosis and medical management is added later. For all Single CDL conditions, the average cost per case is R2 076 for the Primary Drugs. A further R1 645 was spent per case on Secondary drugs and R 450 on other acute drugs. CARE Costing of Chronic Disease List Page v

6 37.5% of registered beneficiaries had multiple CDL conditions. These are further separated into the Top 9 Multiple diseases and the Other Multiple diseases. The average cost per case of the Top 9 Multiple conditions is R4 361 for the primary drugs whereas that for the Other Multiple CDL conditions is R The average cost for all CDL conditions is R3 122 per case per annum. Using the beneficiaries claiming the Primary CDL-NAPPI set, adjusted for exposure of those beneficiaries during the year, gives a prevalence for all CDL diseases of per beneficiaries. Combining average cost and prevalence gives a raw price for the CDL package of R per beneficiary per annum. This needs to be adjusted for medical management, margins for uncertainty and non-healthcare costs. Cost of the CDL Package by Age The childhood and early adult years are characterised by single diseases and asthma dominates the prevalence and cost. Above age 30 the average cost of the CDL package increases with age due to the emergence of multiple diseases. From age 55 onwards, the proportion of people with multiple diseases exceeds 50% of in each age band. The prevalence of asthma, diabetes mellitus, hypertension and diabetes mellitus combined with hypertension is per beneficiaries. This is 66.6% of the prevalence for all single and the Top 9 multiple diseases combined. These conditions alone explain more than 80% of the age profile for all bands up to age 15, 60% up to age 45 and 50% up to age 55. Hypertension is also a factor in many of the multiple conditions that predominate from age 55 onwards. The average cost of the CDL package for all ages is R3 122 per case per annum. The average cost in the age band is 4.5 times that in the 0-1 age band. The average cost is higher than the all ages cost for all bands over age 50. Age impacts substantially on the price of the CDL package. The raw price for all ages is R pbpa with the highest price being R pbpa for the age band. The price of the CDL package for all those above age 40 exceeds the community-rated price. The raw price for a child is R pbpa while that for an adult is R pbpa. The steepness of the price curve, which exacerbates the steepness of the prevalence and average cost curves, shows the substantial incentive open schemes have to avoid members over the age of 40 with chronic conditions. CDL Package by Cluster The raw price of the CDL package varies by cluster, due only partly to differences in the age profile in each cluster. The raw price for the High cluster is 2.7 times the total study price and 4.0 times the Low cluster price. The raw price for the Low cluster is R pbpa and for the High cluster is R pbpa. The study isolated the Low cluster as a possible proxy for the extension of medical scheme benefits to lower income employees. The High cluster represents an upper limit of impact on medical schemes and is indicative of groups with high claiming behaviour. CARE Costing of Chronic Disease List Page vi

7 The differences in cluster prices are due to a combination of factors. These include variation in prevalence rates of important conditions; differing presentation or manifestation of such conditions; provider choice (GP vs. specialist and the management or prescribing habits of each); and benefits available within the health care plan. More research is needed in this area. Full Price of the CDL Package The full price of the CDL package is derived from the raw price with adjustments for haemophilia and for diseases not included in the final Regulations. Adjustments are made for the uncertainty in the price when the package is made mandatory. These are in respect of compliance and the removal of limits and co-payments. The estimated cost of diagnosis and medical management of the CDL conditions is added, based on a protocol for a certain number of visits, tests and devices for each condition. Finally the costs of medicine management and administration are added (the non-healthcare costs). These prices should not be used for contracting or scheme pricing without the assistance of an actuary or skilled pricing professional to amend them appropriately for the situation. The full price of the CDL package using the weighted total is R pbpa or R pbpm. The total PMB package (including CDL), using the weighted total is R pbpa or R pbpm. This is made up of an Inpatient package of R pbpa, an Outpatient package of R pbpa and the CDL package of R pbpa in 2001 prices. The CDL package thus adds 45.8% to the price of the PMB package at it currently exists, before CDL. The CDL package represents 31.4% of the total PMB package (including CDL). Of note is the full price of the CDL package for the Low cluster, which is 33.1% of the price for the High cluster. The full price for the Low cluster is R pbpa while that of the High cluster is R pbpa. The full PMB package (including CDL) for the Low cluster is R pbpa or R pbpm delivered in the private sector. The private sector prices are for the delivery of the PMB package in a fee-for-service environment. To the extent that contracts with providers are entered into on a risk-sharing basis, the price is expected to be lower. The proportion of non-healthcare expenditure for the CDL package is 7.2%. When the CDL package is combined with the Inpatient and Outpatient PMB packages, the non-healthcare expenditure for the full PMB package is 5.3%. This is well below the 10% benchmark commonly used by the Registrar of Medical Schemes. CDL Package Delivered in the Public Sector The 1995 Report of the Committee of Inquiry into a National Health Insurance System considered the impact of a switch in private sector prescribing to the Essential Drugs List (EDL) medicines. They favoured a scenario in which prescription costs would be halved, rather than the best-case scenario in which an 80% reduction could be achieved if all drug prescriptions were switched to EDL medicines at State tender prices. CARE Costing of Chronic Disease List Page vii

8 The study by Rothberg & Walters compared private sector prescribing patterns to the EDL protocols. They showed that if the EDL medicines prescribed by GPs in the study were purchased through the State, an 18.3% saving would be achieved on the total medicines expenditure. An additional 15.7% would be saved if GPs switched other forms of EDL items to EDL items, and another 38.0% if supplementary formulary items and some out-of formulary items could be switched to EDL medicines. Switching of all items, other than an acceptable 10% of out-of-formulary items, would achieve a saving of some 72% on total GP medicine expenditure. The ability to achieve this saving depends on the ability and willingness of prescribers to switch patients from non-edl to EDL products, and the base from which the savings are calculated. The issue of private sector access to the Essential Drugs List medicines at State tender prices was again raised in the recent report of the Social Security Committee of Inquiry. Consideration needs to be given to making certain drugs available nationally, either free, or on a cost-recovery basis. This intervention can dramatically bring down the cost of treating certain conditions and will eliminate the incentive for medical schemes to discriminate against chronic sufferers. Such an intervention will be both market sensitive and have significant implications for the achievement of public health objectives. The Consortium recommends using an estimate of a 50% saving in the cost of the CDL package when delivered in the public sector. As policy unfolds in this area, so this estimate can be further refined. The full PMB package (including CDL) for the Low cluster is estimated to be R pbpa or R pbpm when delivered in the public sector. Effect of CDL Package on Contributions Analysis has been done to compare the final price of the PMB package with the data held by the Registrar on schemes in the study, as well as other open and restricted schemes not in the study. The unit of analysis is the amount per beneficiary per annum. The medicine component of the CDL package is about 60% of the medicine expenditure in the industry. The Registrar s report does not distinguish between acute and chronic medicine benefits. The final price of the PMB package excluding managed care and administration costs is compared to total benefits. The PMB package (including CDL) is found to be equivalent to 50% of total benefits in the other industry open schemes. The final price of the PMB package, including managed care and administration costs, is compared to contributions. The apparent affordability is reflected in that the PMB package (including CDL) is 42% of the contributions of other industry schemes. Preliminary conclusions are that the CDL package appears to be affordable compared to medicine benefits. The PMB package (including CDL) appears to be well covered when compared to the level of total benefits and contributions in the industry. However the analysis has used amounts on a per beneficiary per annum basis on average. There can be significant differences in affordability for particular schemes and particular options. The question of affordability for the Bargaining Council schemes also needs attention. CARE Costing of Chronic Disease List Page viii

9 Estimated Impact with no Mandatory Treatment Policy A margin in the CDL package has been included for that portion of price for which uncertainty exists until the package is fully defined and for the impact of the package being mandatory. Effectively there is a 30% margin on the medicine component of the CDL package. While this may seem to be a large amount, the change in member and provider behaviour from the existing environment to one with a mandatory CDL package is difficult to forecast. The Consortium is of the opinion that the collective margin of 30% on the medicine component is sufficiently conservative to cover this uncertainty in the pricing. A further aspect of the issue of a mandatory package is the potential in the industry for riskselection if the package is not mandatory. The analysis of price by age shows the substantial incentive open schemes have to avoid members over the age of 40 with chronic conditions. There is a real danger in the current medical scheme environment that open schemes will pursue more aggressive self-seeking behaviour and limit the amount of chronic medicine benefits in order to discourage older members and thus improve their community rate relative to their competitors. The substantial broker activity and churning of members that occurs each year at present, worsens this incentive. An environment where membership of medical schemes is compulsory, rather than voluntary, for medium to higher income groups is one essential component of stabilising the system. A mandatory minimum package of chronic medicine and management benefits, as costed in the report, is essential for reducing opportunistic behaviour by some medical schemes. A risk equalisation system between medical schemes, based on the Prescribed Minimum Benefit package (including CDL), will reduce the opportunistic profiting from risk selection still further. The Composition of the CDL List and the Package of Benefits The Consortium recommends that there should be a comprehensive process for chronic disease prioritisation in medical schemes in order to inform the rationing process in future. This would assist in deciding which conditions are included on the CDL list. The Consortium, with the assistance of Medscheme Integrated Care, has included a document entitled Draft Treatment Guidelines for Chronic Disease List Conditions in order to facilitate the process of developing the therapeutic algorithms required by Regulation. It is suggested that a task team be appointed to be responsible for documenting and maintaining the treatment algorithms for the CDL conditions. Ideally this team should work closely with the Standard Treatment Guidelines and the EDL committee. The task team should have representation from private funders, relevant clinical disciplines and the public sector. Actuarial and pricing expertise should be sought once initial clarity has been achieved in order to estimate the price of the algorithms. This may result in an iterative process of refining the algorithms and sufficient time needs to be allocated to this possibility. A project manager should be identified to ensure that the process moves forward in time for schemes to incorporate the benefits in their pricing in August 2003, in order to ensure a smooth introduction of benefits from 1 January CARE Costing of Chronic Disease List Page ix

10 The Danger of Restricting Consumer Choice in Healthcare Despite our inclusion of recommendations and information for the process of developing therapeutic algorithms, members of the Consortium and clinical advisors have expressed serious concerns about the implications of legislating the algorithms for CDL conditions. Effectively this will mean that there is only one approach to treatment that will receive funding from medical schemes. On the one hand there is no doubt that funders, providers and members need clarity on what will be covered as a Prescribed Minimum Benefit. To this end, we support the process of the creation of the therapeutic algorithms. However the process will almost certainly result in the entrenchment of an allopathic approach to treatment, largely based on drug interventions. The movement to complementary medicine is being lead worldwide, not by healthcare professionals or the funders of healthcare, but by consumers. The conclusions from a review of the practice of complementary medicine and African traditional medicine in South Africa are presented. The results reflect the deep-seated needs that patients are expressing; needs that include access to a much broader range of healthcare interventions. The inclusion of other treatment modalities in the therapeutic algorithms is unlikely to be simple and the debate is expected to be vigorous. It is important that hard won legal freedoms to operate alongside allopathic medicine are not negated by effectively preventing the funding of complementary medicine and African traditional medicine as treatment for the CDL conditions. Other Issues A further aspect of the question of the effect of the CDL package on contributions is the effectiveness of chronic medicine management programmes in managing costs. Medicines management is essentially divided into rules-based formulary management and clinicallybased member management. Experience with a large membership base shows that there is no question that a combination of the two yields the best results. This needs to be taken into account in the development of the therapeutic algorithms. The data from this study and the PMB project will be used to answer further issues arising from the studies. A number of projects have been initiated at the Centre for Actuarial Research to continue to explore the richness of this data in CARE Costing of Chronic Disease List Page x

11 Table of Contents Executive Summary... iii Table of Contents... xi 1. Introduction and Background Purpose and Objectives Revision to Legislation on Prescribed Minimum Benefits Acknowledgements Data and Methodology Data Extraction Schemes, Options and Beneficiaries Identification of Beneficiaries with CDL diseases CDL-NAPPI Categorisation Beneficiary Counts and Tariff Amounts Methodology and Adjustment for Exposure Analysis by Cluster Applicability of the Study to the Industry Conclusions on Applicability of the PMB Study Beneficiaries by Cluster in the CDL Study Age and Ethnicity Profiles of the Additional Cluster Conclusions on Applicability of the CDL Study Existing Coverage of the CDL Conditions Chronic Medicine Benefit Design in the Industry Proportion of Industry Options Covering the CDL Proportion of Study Options Covering the CDL Proportion of Study Beneficiaries Potentially Covered for CDL Prevalence of CDL Condition Registrations The Cost for Each CDL Condition Cost for Single CDL Conditions Cost for Multiple CDL Conditions Prevalence and Raw Price of CDL Conditions The CDL Package by Age Age Distribution of Beneficiaries Age Pattern for Specific CDL Conditions Prevalence of CDL Conditions by Age Average Cost of CDL Package by Age Raw Price of CDL Package by Age Raw Price of CDL Package by Wider Age Banding The CDL Package by Cluster Age Distribution of Beneficiaries by Cluster Prevalence of CDL Conditions by Cluster Average Cost of CDL Package by Cluster Raw Price of CDL Package by Cluster Price Differential between High and Low Clusters...45 CARE Costing of Chronic Disease List Page xi

12 9. Items Affecting the Full Price of the CDL Package Cost of Treating Haemophilia Removal of Diseases from CDL List in Final Regulations Anti-coagulating Therapy Cushing s Disease Osteoarthritis Estimate of Cost of Diagnosis and Medical Management Adjustment for Compliance Adjustments for Co-payments and Limits Medicine Management and Administration Costs Delivery of the CDL Package in the Public Sector Comparison of EDL to Private Sector Prescribing Private Sector Access to the Essential Drugs List Estimate for CDL Package Delivered in the Public Sector Full Price of the CDL Package and the Effect on Contributions Full Price of CDL Package by Cluster Non-Healthcare Expenditure in the CDL Package Proportion of Expenditure on CDL and PMB Packages Preliminary Conclusions on Impact on Contributions Estimated Impact with no Mandatory Treatment Policy Issues About the CDL Package Raised During the Project The Composition of the CDL List The Definition of the Package of Benefits The Danger of Restricting Consumer Choice in Healthcare The Effectiveness of Chronic Medicine Management Programmes Further Analysis Using the Data Extracted for the Project Bibliography...78 Appendix A : Proposed Methodology for the Chronic Disease Project...80 Appendix B : Revised 2002 Regulations on Prescribed Minimum Benefits...82 Appendix C : ICD-10 Codes for Chronic Disease List...85 Appendix D : Primary 2000 CDL NAPPI Set...88 Appendix E : Registered Beneficiaries with Multiple CDL Diseases...88 Appendix F : Beneficiaries Registered for Chronic Medicine Benefits by Cluster...89 Appendix G : Registrations for Multiple Conditions with Anti-coagulating Therapy 91 Appendix H: Registrations for Multiple Conditions with Osteoarthritis...91 Appendix I: Data Extract for Medical Management of Selected Diseases...92 Appendix J : Spreadsheet : Cost Estimate for CDL Medical Management Protocols.95 Appendix K: Draft Treatment Guidelines for CDL Conditions...95 CARE Costing of Chronic Disease List Page xii

13 1. Introduction and Background 1.1 Purpose and Objectives The Council for Medical Schemes (the Council) invited tenders for three projects in notices in national newspapers on 15 March The Council set out the scope of the projects and the expected outcomes. A consortium of researchers, led by the Centre for Actuarial Research at the University of Cape Town (CARE PMB Consortium), was awarded the tender for these projects in a letter dated 20 May The Consortium consists of: Professor Heather McLeod of the Centre for Actuarial Research at UCT Professor Alan Rothberg of the Faculty of Medicine of the University of the Witwatersrand and Executive Director of Medscheme Dr Therese Fish of the University of Stellenbosch Graduate School of Business Deus Bazira Mubangizi of the School of Public Health and Primary Health Care at UCT. Medscheme Integrated Care Division, which is responsible for managed care initiatives for Medscheme clients, supplied data for these projects. One of the objectives of the Centre for Actuarial Research is to build capacity in actuarial research through the contracting and mentoring of young researchers. To this end, a number of UCT Honours and Masters students assisted with aspects of this project. Their contributions are acknowledged in Section 1.3. The Consortium proposed a number of changes to the methodology of the projects and the revised expected outcomes formed part of the agreed contract signed in June The proposed methodology for the second project, the Chronic Disease Project, is set out in Appendix A. This report is presented to the Council for Medical Schemes in fulfilment of the contract in respect of the second of the three projects. This report deals with the costing of the proposed Chronic Disease List (CDL) conditions as part of the Prescribed Minimum Benefit (PMB) package, as described in Section 1.2 and Appendix B. The report on the first of the three projects, The Costing of Existing Prescribed Minimum Benefits in South African Medical Schemes in 2001, by Fish, McLeod, Rothberg et al, developed a base price for the current Prescribed Minimum Benefits (PMBs) in medical schemes. This report on the CDL conditions considers the additional cost of those conditions being added to the PMB package. Issues on the impact on affordability for medical schemes are dealt with in a third report, The Impact of Prescribed Minimum Benefits on the Affordability of Contributions, which draws together the findings of the two studies in the context of low-cost option design. CARE Costing of Chronic Disease List Page 1

14 1.2 Revision to Legislation on Prescribed Minimum Benefits The Medical Schemes Act No. 131 of 1998 introduced prescribed minimum benefits as a policy instrument for defining minimum allowable levels of medical scheme cover. Regulations made in terms of the Act were promulgated on 20 October 1999 and came into force on 1 January Annexure A to the Regulations defines the Prescribed Minimum Benefits in terms of a positive list of 270 diagnosis and treatment pairs that must be provided by each scheme, without financial limits and in at least one provider setting. Proposed amendments to the Regulations were published on 30 April Changes included clarity on the provision of Prescribed Minimum Benefits, the inclusion of emergency medical conditions, the concept of a designated service provider and changes to the wording of 13 of the diagnosis-treatment pairs. A significant addition to PMBs was the proposed inclusion of treatment and medication for 28 chronic conditions (the CDL conditions). The Department of Health is required to monitor the impact, effectiveness and appropriateness of the PMB provisions. A review is to be conducted at least every two years by the Department. These reviews are to provide recommendations for the revision PMBs on the basis of: (i) inconsistencies or flaws in the current regulations; (ii) the cost-effectiveness of health technologies or interventions; (iii) consistency with developments in health policy; and (iv) the impact on medical scheme viability and its affordability to members. In the publication of the proposed amendments, no further information was given to the industry on the methodology for choosing the 28 CDL conditions or on the costeffectiveness of treating these conditions. The determination of the impact on medical scheme viability and affordability is the subject of the three reports commissioned by the Council for Medical Schemes. Comment on the Draft Regulations was received by 31 July 2002 and the Minister of Health promulgated revised Regulations in the Government Gazette of 4 November The wording of the revised Regulations is given in Appendix B. In the revised Regulations, three of the proposed CDL conditions were removed from the list, namely: Anti-coagulating Therapy Cushing s Disease Osteoarthritis. On a point of clarity, the revised Regulations contain the condition Dysrhythmias twice. The Medical Advisor to the Council for Medical Schemes, Professor Jan van der Merwe, advised that Ulcerative Colitis should remain in the revised Regulations and that this had been inadvertently removed by the repetition of Dysrhythmias (Personal communication, 25 November 2002). This report will consider the 28 conditions in the draft Regulations (the CDL conditions), will comment on the exclusion of the three conditions above in the revised Regulations and will prepare a final price for CDL using the final 25 conditions. CARE Costing of Chronic Disease List Page 2

15 The inclusion of the CDL list in PMBs is to come into operation on 1 January The prescribed treatment for the CDL conditions is described as diagnosis, medical management and medication, to the extent that this is provided for by way of a therapeutic algorithm for the specified condition, published by the Minister by notice in the Gazette. The development of treatment protocols for the 25 CDL conditions is still to be discussed with industry stakeholders. A starting point for those protocols and an initial costing of the medical management component is provided as part of this study. 1.3 Acknowledgements Specific mention is made of Shivani Ranchod, who in her last year at UCT in 2001 acted as a research assistant to Professor McLeod. Her work on simplifying the comparability of medical scheme benefits, on low-cost option design and chronic medicine benefit design has formed a solid and essential starting point for this project. Other students whose projects provided valuable material used in the development of the report were Rikki Barnes, Samora Adams and Shamim Aghdasi. Lance Osburn did pioneering work on correcting the age profile of the medical scheme industry. Vimbayi Mutyambizi assisted in graphics work on the cluster analysis and chronic registrations. Bettina Taylor of Medscheme Integrated Care provided substantial input on the EDL protocols as a starting point for protocols for the CDL conditions, as well as on a possible methodology for rationing a CDL package. We sincerely thank Dr Laubi Walters for providing and facilitating unfettered access to the people, knowledge and resources in Medscheme Integrated Care. The authors wish to express their gratitude to Kirsty Davies and Adrienne Gurnell for their role in the smooth running of our respective offices and in helping us to meet very tight deadlines. CARE Costing of Chronic Disease List Page 3

16 2. Data and Methodology 2.1 Data Extraction Schemes, Options and Beneficiaries The Consortium requested and received data from the Medscheme Data Warehouse in August and September 2002, covering the full calendar year This has the advantage of being the most recent data in terms of medical costs and member and provider behaviour. It is also reflective of the effect of recent benefit designs under the Medical Schemes Act of It was expected that the run-off period to allow the majority of claims for 2001 to have been recorded would be three months into the new calendar year. By drawing the data in August, there should be complete run-off of the 2001 claims. Data in respect of schemes administered previously by Sanlam Health and Sovereign, and schemes in Namibia and other countries was excluded, as the nature of the information was not identical. The data for the PMB study covered 91 options in 29 schemes, however, whereas Medscheme had immediate hospital and related claims data for some schemes that came under administration in 2001, they did not have immediate registration and chronic medicines management details for those schemes. Consequently, for the CDL study, four schemes were excluded. Furthermore, one large scheme was added where chronic medicine management was provided as a stand-alone service but not hospital benefit management. The newly administered schemes had unusually large numbers of options, unlike the typical Medscheme benefit design, so that the CDL study covers 43 options in the 26 schemes. The numbers of beneficiaries in the schemes each month was made available and an adjustment was made for beneficiaries who were not present for a full year (see Section 2.2). The credible data for the PMB study covered million beneficiary months, giving an average exposure of beneficiaries for the year. For the CDL study, the credible exposure is million beneficiary months, giving an average exposure of beneficiaries for the year. Demographic information was received for all options and thereby for the clusters (see Section 2.3). The definition of age used was age last birthday as at 1 January Beneficiaries were then grouped into the following age bands: Under 1 year of age 1 to 4 years 5 year age bands from 5 years of age up to 84 years of age 85 years and older. Medscheme also provided contribution and benefit tables for most of the schemes. Data confidentiality has been maintained. At no time were the Consortium members outside of Medscheme able to identify specific patients. Data from the Statutory Returns to the Registrar was received in September 2002 from the Council for Medical Schemes in respect of calendar year This was used to determine the relationship between these million beneficiaries and the rest of the industry in that year, in respect of demographics, contribution levels and benefits paid. CARE Costing of Chronic Disease List Page 4

17 2.1.2 Identification of Beneficiaries with CDL diseases The MediServe database at Medscheme Integrated Care contains details of all beneficiaries registered for chronic medicine benefits as part of the Pharmacy Benefit Management programme. Information is captured at the time of registration for chronic medicine and includes clinical information provided by the doctor and details of the medicines authorised for payment. The first step in the data extraction was to interrogate the MediServe database using the ICD-10 diagnosis coding recorded at the time of registration on the chronic medicine programme. All beneficiaries with at least one of the CDL-conditions were selected. Those beneficiaries with multiple CDL diseases could also be identified. Beneficiaries with haemophilia are not usually registered on the chronic medicine programme unless they have a co-existing chronic illness. This is because the blood products or factors necessary to treat haemophilia are obtained from a benefit category other than medicines. In the data extraction and analysis of the cost of medicine, haemophilia is thus not included with the other CDL conditions. The estimate for haemophilia is dealt with separately in Section 9.1. It was also not possible to separate those beneficiaries with diabetes mellitus into Type 1 and Type 2. Thus information was extracted for 27 separate CDL conditions. In the paper review of the MediServe database, there were beneficiaries registered for chronic medicine in the schemes studied in The total number of beneficiaries with one or more CDL conditions was , which was 77.1% of the total chronic medicine registrations. Initially, the count of single ICD-10 codes (i.e. single CDL diseases) was 45.0% of all beneficiaries with CDL diseases. The maximum number of CDL-ICD-10 codes for a single beneficiary was 17 (i.e. the person had been registered for medicine benefits for 17 separate diseases from the list of 28 CDL conditions). However the ICD-10 coding system uses several related codes to describe what is expected to be included in the definition of a CDL condition. For example, asthma may appear as asthma or wheezing within different alpha-numeric categories, and these two descriptions of the same diagnosis were therefore aggregated. Professor Rothberg manually collated the ICD-10 codes into the 27 CDL-diseases (see Appendix C). This categorisation increased the number with single diseases to or 62.5% of CDL conditions. The maximum number of CDL diseases for a single beneficiary reduced to eight separate diseases. There is a notable difference between medicines authorised for payment and medicines actually claimed from the scheme. Of the beneficiaries registered for any CDL condition, only made any claim for acute or chronic medicine. In other words 77.2% claimed some form of medicine. Some of these prescriptions may also not be in respect of the CDL condition. This apparent anomaly was the result of the lag between issue of new or changed membership numbers and initiation of claims, terminated members still on the database, or member numbers not linked backwards over member or plan number changes. CARE Costing of Chronic Disease List Page 5

18 In order to assess the impact of the CDL list on schemes, the full tariff amount for medicines used for those conditions needs to be isolated, as discussed below. The issue of compliance, as it impacts on pricing, is explored further in Section 9.4. A further issue in the data extraction is that a few options do not make use of the chronic medicine management programme but still provide benefits for chronic medicine. In those instances, the prevalence and cost of the CDL conditions can be ascertained by identifying beneficiaries claiming medicines that are prescribed for those conditions CDL-NAPPI Categorisation The National Pharmaceutical Product Interface (NAPPI) is a list of codes that medical schemes use to identify pharmaceutical products for claims. A NAPPI code consists of a six-digit number, followed by a three-digit suffix. The suffix differs when a product has more than one pack size. A product that occurs in various forms, e.g. tablets and syrup, would have separate NAPPI codes. In this study, three sets of data were extracted, based on the NAPPI codes: Primary: using only selected NAPPIs that are clinically appropriate for the CDL conditions, as determined by Medscheme Integrated Care. This amounts to some CDL-NAPPIs. Secondary: using all NAPPIs ever claimed under chronic benefits by any of the people in the study. This amounts to some CDL-NAPPIs. Total Tariff Amount: using all NAPPIs ever claimed, whether under chronic medicine benefits or acute medicine benefits. This covers all medicines claimed by the beneficiaries in the study. Drugs that fall into the Secondary category are considered to be supportive, rather than therapeutic in treating the CDL condition. Thus in order to isolate the costs of the medication used for treating the CDL conditions, the analysis in this report will predominantly make use of the Primary data set. The CDL-NAPPIs developed and extracted for this study are given in Appendix D on CD- ROM. The file also includes Anatomic Therapeutic Classification (ATC) codes that divide the products into chapters Beneficiary Counts and Tariff Amounts The three sets of CDL-NAPPI codes were used to extract items claimed in respect of the beneficiaries with CDL diseases. It is important to note that claims were extracted that were lodged under both chronic and acute benefit limits. Thus if a person was unable to get all their medication under the chronic limit, they could have utilised their acute limit. By taking items claimed under both limits the problem of dealing with chronic limits has been substantially reduced. The items claimed were then linked to the full tariff amount for those items. Rather than work with benefit or paid amounts, which may be heavily influenced by limits and copayments, the Consortium worked with the tariff or Blue Book amount for the items claimed. The impact on pricing is discussed in Section 9.5. CARE Costing of Chronic Disease List Page 6

19 Total claims extracted from the Medscheme database was records, representing beneficiaries with chronic conditions. The total number of beneficiaries who claimed for one or more CDL conditions was , which was 76.8% of the number of who claimed for chronic conditions. The Total Tariff Amount for all beneficiaries in the schemes was R The Total Tariff Amount in respect of those with one or more CDL diseases was R Only a portion of this expenditure relates directly to the CDL conditions, as discussed in Section 6. An investigation into beneficiaries registered for single CDL conditions showed that there were no medicine claims for 21.0% of them in This issue of compliance and of registrations relative to usage is reviewed in Section 9.4 in determining the impact on price. 2.2 Methodology and Adjustment for Exposure The costs were extracted separately for each of the clusters as described in Section 2.3. Total tariff amounts and the count of beneficiaries associated with those claims were isolated in each case. In the case of hospital claims, average cost per case is simply the total cost divided by the number of admissions. In the case of chronic medicine claims, the expectation is that once a person is on chronic medication, they would take it regularly throughout the year. This can be accommodated by inflating the tariff amount or reducing the number of beneficiaries in the average cost calculations. The Consortium preferred to make an adjustment to beneficiary numbers based on the exposure of each beneficiary to each option in the year. On the basis that medicine benefits are paid only if the beneficiary is active on the scheme at the time of the dispensing of the claim, an individual s total annual exposure was determined by using the specific number of months for which he/she was active during the study period. Thus the number of beneficiaries claiming any medicine was reduced from a count of to an exposure of The original count of beneficiaries in the study was reduced after the adjustment to beneficiaries exposed for the period. The raw price of the CDL package was estimated using the Primary CDL-NAPPI set and the total exposure of all beneficiaries for each month of the calendar year. 2.3 Analysis by Cluster The concept of analysis by cluster was introduced in the report on the PMB study. The material is repeated here for the sake of clarity for those reading only this CDL report. Medscheme have developed a proprietary method of analysis of the data, the results of which have been shared with the Consortium. The approach involves using statistical techniques to cluster the benefit options into distinct groups based on option characteristics and demographic characteristics, primarily age of members, family size and percentage African-Black membership. In the absence of details on socio-economic status, the reality is that ethnicity still plays a part in differentiating claiming behaviour. CARE Costing of Chronic Disease List Page 7

20 Research by Leslie Pels and Professor Alan Rothberg shows that the different clusters experience different benefit utilisation, costs and disease profiles. Provider behaviour has been shown to differ by cluster, even within the same hospital facility. The most recent analysis by Medscheme suggests that the data is best understood in four distinct clusters: High contains options with older, 'whiter' members with high utilisation; Medium-older contains options with medium utilisation and older members; Medium-younger contains options with medium utilisation and younger members; Low contains options with younger, 'blacker' members with low utilisation. The CDL study also contains the Additional cluster that was not present in the PMB study. The Additional cluster is from a single large scheme where the ethnicity is predominantly so-called Coloured. As demographic details were held separately, it was not possible to incorporate this in the tables and graphs in this section. The impact on the applicability of the study is dealt with in Section 3. The table below shows a number of key differences between the clusters used in the PMB study. In some instances, the comparative data is from the first quarter of Table 1: Key Cluster Differentiating Characteristics and Utilisation High Mediumolder Mediumyounger Low All Clusters Cluster data from 2001 PMB study Average age of beneficiaries (in years) Proportion of beneficiaries over the age of 55 years Proportion of African/Black beneficiaries Illustrative cluster data for Q % 18.4% 9.7% 5.0% 8.5% 15.8% 16.3% 22.2% 77.1% 61.0% Family size Proportion of beneficiaries registered for chronic medication Hospital admission rate per 1000 lives 29.7% 16.9% 9.3% 11.4% 13.2% Contributions pmpm R2 010 R1 593 R1 188 R1 074 Benefits pmpm R1 787 R1 121 R 902 R 732 Note that benefits paid out from capitation arrangements are 'hidden' from the analysis above, which only captures claims from providers. Prime Cure and MediCross arrangements, which are more common in lower clusters, add roughly a further R 250 pmpm to the effective benefits. The same issue affects the graph on contributions and benefits below. CARE Costing of Chronic Disease List Page 8