The Impact of the Verification Criteria on the REF Grid Count April 2007

REF Study 2005 Tables derived from data in REF Study 2005: Four administrators: Discovery Health, Medscheme, MHG and Old Mutual Healthcare. 63.4% of beneficiaries in industry. Data on prevalence and PMB expenditure for calendar 2005. CASES and TREATED data using REF Entry and Verification Criteria v2, in force from 1 January 2007. Graphs show final tables published with REFCT2007: Prevalence: all beneficiaries with condition Revised Prevalence: after application of multiple disease rules Count: as in REF Grids, with allocation to highest cost disease Final REFCT2007 used for order of diseases. The tables are effectively for calendar 2007. HIV has been adjusted to the expected level of the epidemic in 2007. Comparison: REF Study 2002 Count, adj. to 2006. Source: REF Study 2005

Definition of TREATED and CASES Two sets of data were extracted: The first used the full Entry and Verification definitions and was called the “Treated Patient Data set” or “TREATED” The second set was extracted without the test for “treated patient” and was called the “Total Cases Data set” or “CASES”. While this meant a doubling of the extractions, it provided a powerful tool to investigate potential prevalence and cost if compliance improves and to be able to determine the impact if more people in future fall within the definition of “treated patient”. Most important comparison for REF financial sensitivity is CASES Count vs. TREATED Count. Difference represents “bubbling under” for each disease. Source: REF Study 2005

Ranking of Diseases in Multiple Disease Rules Effectively uses an approach similar to hierarchical co-exiting conditions methodology. Order of diseases from REFCT2007 using gender as a risk factor. Only one disease in the following groups may be selected. Highest cost disease in bold: respiratory: COP+AST+BCE cardiac: CMY+CHF+IHD+DYS+HYP renal: CRF+HYP gastro: CSD+IBD diabetes: DM1+DM2 (always default to DM2) mental: BMD+SCZ neuro: MSS+BMD+EPL skeletal: SLE+RHA (other way around in REF Study 2002) Source: REF Study 2005

Amounts above NON for Diseases Source: REF Study 2005

Explanation of graphs using AST All with AST diagnosis After respiratory multiple rule Allocation to highest cost disease. Potential AST count if compliance improves. REF Grid Count: “treated patient” respiratory: COP+AST+BCE COP>BCE>AST Source: REF Study 2005

Comparison to REFCT2006Ver REF2006Ver has disease patterns largely taken from the REF Study 2002. HAE updated with clinical information from the Haemophilia Society. HIV updated to expected level for 2006 with advice from Leigh Johnson at the Centre for Actuarial Research and using the ASSA AIDS models. MAT updated to levels seen in industry in Q3 2005. In retrospect, the 2002 MAT levels were probably too low. Comparison to this table effectively gives us a comparison to the previous REF Study shape. Source: REF Study 2005

Respiratory Disease

AST At older ages AST reduces as COP > AST. respiratory: COP+AST+BCE Source: REF Study 2005

BCE At older ages BCE reduces as COP > BCE. respiratory: COP+AST+BCE Source: REF Study 2005

COP COP Verified is same as COP prevalence because COP is most expensive respiratory condition. respiratory: COP+AST+BCE Source: REF Study 2005

Diabetes

DBI TREATED very low compared to CASES Source: REF Study 2005

Already applied to data during cleaning before this analysis. DM1 with DM2 Rule Version 2 of REF Entry and Verification Criteria: 3.9.1.8 Note that, in accordance with the Diabetes Mellitus table in section 6, Diabetes Mellitus Type 1 and Type 2 cannot co-occur. 5.7.2 Categorise Diabetes Mellitus cases as either Type 1 or Type 2 diabetes. Table 11: For REF purposes, Type 1 and Type 2 diabetes cannot occur concurrently. Evidence of use of oral hypoglycaemic or euglycaemic medicines automatically leads to the classification of a diabetic case as Type 2. NOT just a count issue as with other verification rules. If DM1 and DM2 found together, must default to DM2. Already applied to data during cleaning before this analysis. Source: REF Study 2005

DM1 Treated about half of previous levels due to application of Verification criteria and rule. Not due to DM1 and DM2 overlap: very little in 2002. Source: REF Study 2005

DM1 and DM2 in 2002 DM1 and DM2 would be defaulted to DM2 now. Source: REF Study 2005

DM2 DM2 Count dramatically higher than before. Result of change in ranking of common multiple diseases, as prevalence is reasonable. Source: REF Study 2005

DM2 Prevalence DM2 levels after application of criteria seem reasonable compared to 2002 Study prevalence. Source: REF Study 2005

Change in Order of Diseases DM2 now more expensive than RHA, AST, HYL, HYP hence many more swinging to DM2 in REF Grid Count Source: REF Study 2005

Cardiac and Renal Disease

CMY CHF no longer exists – combined with CMY. New combined disease exceeds CHF+CMY in 2002. cardiac: CMY+CHF+IHD+DYS+HYP Source: REF Study 2005

DYS DYS substantially reduced by Verification criteria and cardiac rule. cardiac: CMY+CHF+IHD+DYS+HYP Source: REF Study 2005

HYL Why was not included in cardiac rule ? [cardiac: CMY+CHF+IHD+DYS+HYP] Source: REF Study 2005

IHD Impact of cardiac rule at older ages. cardiac: CMY+CHF+IHD+DYS+HYP Source: REF Study 2005

HYP Impact of renal and cardiac rules at older ages. cardiac: CMY+CHF+IHD+DYS+HYP renal: CRF+HYP Source: REF Study 2005

CRF Not verified due to “treated patient” definition Very large numbers not verified. CRF essentially requires dialysis for “treated patient”. Unlikely to be significant movement to becoming “treated”. renal: CRF+HYP Source: REF Study 2005

Gastro-intestinal Disease

CSD No impact from gastro rule as CSD > IBD gastro: CSD+IBD Source: REF Study 2005

IBD Only very small impact from gastro rule. gastro: CSD+IBD Source: REF Study 2005

Skeletal Disease

RHA SLE now greater than RHA but very little impact from rule. skeletal: RHA+SLE Source: REF Study 2005

SLE SLE now greater than RHA skeletal: RHA+SLE Source: REF Study 2005

Neurological Disease

BMD Neuro rule has almost no impact neuro: MSS+BMD+EPL Source: REF Study 2005

EPL Neuro rule has little impact neuro: MSS+BMD+EPL Source: REF Study 2005

MSS neuro: MSS+BMD+EPL Source: REF Study 2005

SCZ Increase in Prevalence and Count at oldest age is odd. Very little data. Source: REF Study 2005

PAR Verification criteria produce slightly lower result. Source: REF Study 2005

Other Disease

ADS Large divergence at older ages from Prevalence. Source: REF Study 2005

HAE HAE Count expected in 2006 was from REF Grids. Source: REF Study 2005

GLC Reasonable shape. Source: REF Study 2005

TDH Massive difference between Prevalence and Count: TDH is very low cost disease. Source: REF Study 2005

HIV Expected Count in 2007 slightly lower than Expected in 2006. Note REF Grids have been about half of Expected to date. Source: REF Study 2005

Multiple Disease

CC2 CC2 CASES Count is higher than Prevalence – impact of multiple disease rules pushing more CC3 and CC4 down to CC2. As predicted. Source: REF Study 2005

CC3 As predicted, reduction in CC3 from application of multiple rules. Large “bubbling under”. Source: REF Study 2005

CC4+ As predicted, massive reduction in CC4 from application of multiple rules. Large “bubbling under”. Source: REF Study 2005

NON TREATED NON higher than CASES NON. Difference is “chronic not verified”. Source: REF Study 2005

Summary In some disease, very large differences between CASES Count and TREATED Count. This represents a financial risk to schemes in REF: if more beneficiaries are compliant and hence meet the “treated patient” criteria, then the numbers with chronic disease go up substantially. Affects some very high cost diseases like CRF (although not many likely to become “treated”) and MSS. Affects numbers with multiple disease. If REF had an annual Industry Community Rate, REF Fund would take the risk during the year. If REF has a monthly Industry Community Rate, then schemes immediately impacted. Source: REF Study 2005

Revisions to Definitions Important: Revisions to Definitions

Date of Treatment not Payment Original definition using “payment date” was contrary to general pricing practice and also the way PMBs and REF are priced. Original definition open to being gamed. Possible to manipulate payment dates for a multi-line scrip across two months. Thus more beneficiaries qualifying for REF shadow payments than envisaged. “Treatment date” can not be manipulated in this way. Guiding principals for REF risk factors: "Invulnerability to Manipulation: The risk factors should not be subject to manipulation by medical schemes, managed care organisations, administrators, providers, intermediaries or the beneficiaries.". "evidence is required that a patient has received the specified treatment during at least two preceding treatment months in the three treatment months preceding the current treatment month” “Proof of payment is still required to be able to count a particular treatment date. Payment towards a specific treatment must have been made from the risk benefits in order to qualify as a treatment.” Source: REF Study 2005

Components of BCE One Under 1 case at OMHC: DTP_Hospital was R873,000 over four months. Baby died. Although met BCE criteria, diagnosis was uncertain. After investigation, recommend treat as NON. Source: REF Study 2005

Components of CC3 Diagnosis of multiple chronic for Under 1 ? Two cases found. MS: EPL+CMY+HYP. Meds 4 months, no DTP. MHG: AST+CMY+HYP. 7 months, meds 1 month, R423,400 DTP Source: REF Study 2005

Prof Alan Rothberg on Neo-nates "As for causes of neonatal costs, some general principles are that neonatal problems mostly fall into one of four categories viz. birth asphyxia, infections, congenital abnormalities and prematurity. In the funded environment asphyxia is uncommon (better prenatal care), and infections occur but are not really drivers, so your high cost cases really fall into the other two categories. Congenital abnormalities may be expensive especially if cardiac because of the corrective and often complex surgery involved and the need for pre- and post-op ICU, but the highest costs are related to prematurity, mainly in infants below 1500g and within that group particularly the ones weighing less than 1000g." Note that congenital abnormalities are a PMB but would fall into the NON column. The problem arises when we try to use adult criteria for treatment of CMY and CRF on these cases. Source: REF Study 2005

Counts for Under 1s “… the algorithms and protocols are adult-based and one cannot simply squeeze kids in, especially those below 1 year of age.  In terms of the commonest conditions …, clearly schemes are including wheezing as asthma and seizures as epilepsy, and while some wheezers may go on to be asthmatic and the occasional fitter will become an epileptic, the majority will not.”  Small children get repeated viral infections and because their airways are narrow they wheeze easily, and as for seizures, many kids have a fit when they have an infection and associated high temperature.  Bronchiectasis is also very unlikely below one year of age, as is what we usually understand by the term chronic obstructive pulmonary disease.   Inflammatory bowel disease is probably also over-represented, and is probably picking up chronic diarrhoeas.  Cardiac failure, renal failure, diabetes are present in low numbers and may be realistically represented.” Source: REF Study 2005

Counts for Under 1s cont. “Getting to multiple diseases, I'd guess that the most common scenario relates to ex-prem babies (mostly below 1500gm birthweight) who were in ICU on ventilators, developed so-called chronic lung disease (or bronchopulmonary dysplasia) and ended up with low thresholds for lung infection, wheezing, pneumonia, and possibly associated heart failure.   So bottom line is that I think you're being hoodwinked and will need additional evidence before you pay out for these alleged disease burdens (e.g. history of very low birthweight, prolonged ventilation/ICU stay, frequent admissions, nature of medications etc.).” Prof Alan Rothberg, previously Professor and Head of Dept of Paediatrics and Child Health, University of the Witwatersrand Source: REF Study 2005

Important Decision on Under 1s for REFCT2007 Considered whether to allow: (A) the current chronic conditions for Under 1s (B) to allow only a restricted set of conditions (from his comments these would be CMY, CRF and DM1)  (C) to allocate all the cost to the Under 1 age band. Pricing team, with clinical assistance, have chosen (C). Confirmed by RETAP and recommended to Council. Pricing for REFCT2007 on the basis that there is no allocation of Under 1s to any chronic disease or multiple chronic disease. All Under 1s to be defaulted to NON for payment from 2007. Need to communicate in next REF Entry and Verification Criteria and for 2007 data. Source: REF Study 2005