The Best of Both Worlds%3F Potential of Hybrid Prospective%2FConcurrent Risk Adjustment

8/4/2019 The Best of Both Worlds%3F Potential of Hybrid Prospective%2FConcurrent Risk Adjustment

1/14

The Best of Both Worlds?

Potential of Hybrid Prospective/Concurrent Risk Adjustment

R. ADAMS DUDLEY, MD,* CAROL A. MEDLIN, PHD, LISA B. HAMMANN, MIRIAM G. CISTERNAS,RICHARD BRAND, PHD, DEBORAH J. RENNIE, AND HAROLD S. LUFT, PHD

BACKGROUND. There remains considerableuncertainty about whether prospective or con-current risk adjustment (RA) is preferable.

Although concurrent models have better pre-dictive power than prospective models, thelarge payments associated with concurrent RAcreate incentives for fraudulent coding. A hy-brid strategyin which prospective paymentswere used for patients with low expected costsand concurrent payments were available uponthe diagnosis of a small number of common,expensive conditionsmight improve predic-tive performance while requiring less auditingthan fully concurrent RA. In addition, within-

condition RA (using clinical data) for the se-lected conditions could further improve pre-dictive power.

OBJECTIVES. To assess how such a hybridstrategy might perform, focusing on a smallnumber of chronic, expensive conditions thatare verifiable (hence auditable).

SUBJECTS AND MEASURES. All patients fromseven health plans who had two completeyears of utilization data were considered. RAmodels were estimated among patients

younger than 65 (n 319,209) using the Hier-

archical Coexisting Conditions (HCC) model

with and without stratification of the sample

based on the presence of one or more of 100

verifiable, expensive, predictive conditions(VEP100). R2 and predictive ratios were calcu-

lated for each model studied.

RESULTS. Patients with a VEP100 condition

(9.3% of the population) accounted for 84.3% of

the variation in cost. R2 was 0.08 using a

prospective HCC model on the entire popula-

tion, but increased to 0.26 for a hybrid using

prospective HCCs on the 90.7% of the sample

without a VEP100 condition and a simple con-

current model consisting of dummy variablesfor each of the VEP100 conditions.

CONCLUSION. Combined with targeted audit-

ing, a hybrid approach to RA could improve

our ability to match payments to costs. How-

ever, because this would require additional,

costly data collection, more research is needed

to determine whether this benefit justifies the

data collection and auditing burden.

Key words: Risk adjustment; risk assess-

ment; health care financing; chronic disease;

auditing. (Med Care 2003;41:5669)

The potential for differential risk selectionamong health plans is widely recognized, and

many payers have adopted or are considering riskadjustment.13 Nonetheless, controversy remains

From the *Department of Medicine, the Institute forHealth Policy Studies, the Institute for Global Health,and the Department of Epidemiology and Biostatistics,University of California, San Francisco (UCSF),

California.

From Genentech, South San Francisco, California.

From MCG Data Services, San Mateo, California.

Supported by the Robert Wood Johnson Foundation(Grant 29661).

Address correspondence and reprint requests to: R.Adams Dudley, MD, Department of Medicine and Insti-

tute for Health Policy Studies, University of California,San Francisco (UCSF), 3333 California Street, Suite 265,San Francisco, CA 94118. E-mail: [email protected]

Received October 11, 2001; initial review January,2002; accepted July 9, 2002.

MEDICAL CAREVolume 41, Number 1, pp 56692003 Lippincott Williams & Wilkins, Inc.

56


2/14

about how to do the risk assessment (RA) that isneeded to carry out risk adjustment.1,48 Compet-ing approaches vary in their choice of predictors ofcost (diagnoses,1,58 pharmacy data,9,10 health sta-

tus questionnaires11

) and whether they use hier-archical logic incorporating severity of illness.1,4,5

Also, RA can be done prospectively or concur-rently. In this paper we focus primarily on theissue of timing, and confine our analyses todiagnosis-based RA, because this is the approachfavored by policy makers.2,12

We argue that both prospective and concurrentRA have inherent problems that cannot be over-come by incremental changes to existing models. A hybrid approach, however, can enhance the

strengths and minimize the weaknesses of each.Moreover, if additional clinical detail is added tothe concurrent component of this hybrid, furtherimprovement in RA is possible, and the datacollected could also be used for quality measure-ment and improvement.

Prospective RA models have low predictivepower (usually measured by R2 , which rangesfrom 0.08 to 0.15). This creates incentives forhealth plans to encourage the disenrollment ofhigh risk patients within high risk categories,

because plans achieving such selection can garnersignificant profits.13,14 In addition, prospective RAcompensates plans poorly for some utilizationdriven by patients (eg, obstetrical utilization).

Changes in enrollment status are also difficultto address prospectively. None of the extantclaims-based RA models handle deaths well.15 Inbuilding their models, the developers eitherdropped patients who died from the popula-tion5,8which understates true costs or annual-ize, based on monthly expenditures up to the time

of death6,7

which overstates costs. This decisionis important because a high proportion of totalexpenditures occur at the end of life.16 Further-more, in some populations especially Medicaidand Medicarepatients frequentlychurnon andoff a health plans rolls, so plans may enrollpatients who are expensive, but have no prior datato submit for RA.

Concurrent RA handles enrollment changesbetter. Costs associated with patients who die canbe included in concurrent model building without

annualization, and for new enrollees there is noneed to have data from the previous insurer.Concurrent RA models also have higher predictivepower. This is achieved primarily by allowingpayments for the concurrent occurrence of a dis-

ease that are larger than prospective RA paymentsfor the same disease. For example, concurrentmodels give greater weight to you have heartfailure nowthan prospective models give to you

had heart failure last year.Critics of concurrent RA argue these large pay-

ments invite over utilization and upcoding (re-porting diagnoses more severe than the patientstrue condition). They argue that concurrent RA isessentially fee-for-service (FFS) with a time lag(especially concurrent models that include as pre-dictors the performance of medical or surgicalprocedures5). Although any RA creates incentivesto upcode, the larger payments associated withconcurrent RA create stronger incentives. Thus,

concurrent RA would require additional auditing(as Medicare currently audits physician bills inFFS). Even if it focused primarily on outlier plansreporting a high proportion of their population tohave expensive diseases, auditing all their encoun-ter data still would be costly.

Some observers have suggested blending pro-spective RA with FFS.17 This has the advantage ofmore closely matching payments to costs, but doesnot fully fix incentives to over utilize and requiresthe collection of data for both RA and FFS pay-

ment. Furthermore, plans that believe they do agood job of keeping patients with chronic diseasesin optimal health do not support payment systemsthat are tightly linked to utilization.18

Another alternative is carve-outs of specifichigh cost conditions for either condition-specificcapitation or FFS.19 However, these approachesrequire some model to estimate the appropriatecapitation rate or FFS budget. Studies have shownthat a single rate for each condition usually createsthe opportunity for risk selection within a condi-

tion, and current RA software is insufficient toeliminate the incentive to avoid high riskpatients.14,19

We propose an alternative based on the obser- vation that a substantial fraction of health carecosts arise from the treatment of a relatively smallnumber of common, expensive chronic conditions,such as cardiovascular disease, cancer, and humanimmunodeficiency virus (HIV) infection. Develop-ing better predictions of cost for patients withthese conditions could greatly improve the predic-

tive power of RA models. Translating this obser- vation into a modeling strategy, we propose ahybrid system of RA that adjusts prospectively formost conditions, but provides additional concur-rent payments for patients with selected condi-

Vol. 41, No. 1 HYBRID PROSPECTIVE/CONCURRENT RISK ADJUSTMENT

57


3/14

tions. The concurrent payments would be fixedamounts based on the condition and its severity. Inthis sense, the concurrent payments are like Medi-cares DRG payments one knows prospectively

how much will be paid for a given diagnosis, butdoes not try to predict when a patient will developthe problem. However, they differ from DRGs inthat they are tied to a clinical event (the patientsdiagnosis), rather than a utilization decision (ahospital admission).

We would address upcoding through targetedauditing. To minimize both the data collectioncosts and the audit burden, we would offer con-current payments only for a small number ofcommon conditions that are clinically verifiable

(that is, for which a specific pathologic or labora-tory result or constellation of symptoms must bepresent before a clinician would accept the diag-nosis). Hence the data for verification are the sameas those already collected for clinical purposes.

This strategy may be able to better matchpayments to costs than prospective RA and, bylimiting the concurrent payments to a short list ofconditions, may allow clinicians and policy makersto focus attention, resources, and audits wherethey matter most. In addition, a new area for

improving RA is opened; one in which the goal isto perform adjustment among patients with aparticular condition. Rather than being limited todata relevant to a broad mix of patients, such asdiagnoses or drugs used, within-condition RAcould use any type of information eg, lab resultsmeasuring severityrelevant to patients with agiven diagnosis. This is an important differencebetween our proposal and that of Carter et al,4

who suggest blending prospective and concurrentRA using diagnosis codes only, because otherresearchers have shown that diagnosis-based RAis insufficient to capture the range of within-condition variation in concurrent costs.14,19

Prior research has shown that within-conditionmodels of utilization can be developed with ap-propriate clinical input,2022 and the ability tostratify patients at risk for high utilization within adisease category is a critical element of manydisease management programs.23 The current per-formance of RA models is limited primarily by the

failure to handle extremely expensive patientswell,5,24 which suggests that a focus on within-condition RA for the most expensive conditionscould have a large impact on both the predictivepower of models and market incentives.

In this paper, we investigate the potential of ahybrid RA approach. We created a list of condi-tions that were verifiable and chronic and ex-pected to contribute significantly to total popula-

tion costs. We then used health plan data toidentify 100 of these100 being simply a roundnumber of the right magnitudeto use as therelatively small (compared with the number ofdiagnoses in the International Classification ofDiseases-9, Clinical Modification [ICD-9-CM]system) number of verifiable, expensive, predictiveconditions (the VEP100). We then assessed thelikely impact on market incentives of the blend ofprospective RA plus concurrent payments for theVEP100.

Materials and Methods

Data and Patients

We used data collected by the Society of Actu-aries (SOA). This included enrollees diagnoses(ICD-9-CM codes) and costs during 1991 and1992 from seven insurers, with each contributingdata from one or more commercial (non-

Medicare) products. Costs, measured by chargesfrom submitted claims, were summarized for eachpatient annually.

From the initial sample of 3,103,253 patients,we set aside those who: were not enrolled for theentire 2 years study period (n 2,662,918 patientsor 85.81% of the original sample), and excludedthose who (1) had only invalid diagnosis codes oronly charges without diagnoses (n 45,867,1.48%), (2) had multiple, conflicting entries forannual cost (n 73,982, 2.38%), (3) were older

than 64 (n

1234, 0.04%), and (4) had missinggender information, or had inconsistent gender orage information between 1991 and 1992 (n 63,0.00%). Our sample for model testing included319,209 patients.

There were 1,937,808 patients who were en-rolled 1992 but not in 1991. These patients wereused to select candidate conditions for inclusion inthe VEP100.

Identifying the VEP100

Creating a list of conditions for which to pro-vide concurrent payments involved subjective andobjective evaluation of all ICD-9-CM codes. As a

DUDLEY ET AL MEDICAL CARE

58


4/14

first step, a physician (RAD) reviewed every codeand identified a list of candidate conditions basedon his perception of their verifiability, relativeexpense, and chronicity. Verifiability was judged by

whether the diagnosis is typically based on specificclinical criteria, eg, review of pathologic specimensfor cancer, absence of blood proteins for hemo-philia. Relative expense and chronicity were de-fined by likelihood of creating ongoing expendi-tures through the course of the current year andsubsequent periods. Expensive, acute conditionsthat do not require long-term expenditures (eg,cholecystitis) were not included in the VEP100.

Using the patients with clean data for 1992 only,concurrent mean charges for each of the candidate

conditions were based on mean total annualcharges for all patients with that VEP condition(patients with multiple VEP conditions were ex-cluded). The top 100 conditions in terms of meanconcurrent annual charges were included in theVEP100.

Developing and Applying Risk AssessmentModels

We used relative cost weights for 1992 (anindividuals actual charges divided by the popula-tion mean, without truncation or transformation)as the outcome variable for modeling. Althoughthere are several different RA packages available,we did not intend to compare them and used onlyDxCG software (release 4.2, DxCG, Boston, MA),which groups patients into Hierarchical CoexistingConditions (HCCs). We used linear regression topredict cost weights based on demographic vari-ables and binary variables representing the pres-

ence of one or more of a group of diagnosis codes. We estimated a demographic model and pro-spective and concurrent versions of HCCs. We alsoestimated a model in which patients were firststratified into those with and without a VEP100condition in 1992. A concurrent model was createdfor patients with a VEP100 condition in which thepredictor variables were age, gender, and a sepa-rate dummy variable for each VEP100 condition.The HCC prospective model was then recalibratedto patients without VEP100 conditions (the re-

mainder population, or REMAIN). Combining thepredictions generated from concurrent RA on pa-tients with a VEP100 condition in 1992 and pro-spective RA on REMAIN patients, we created ahybrid model. The sum of squared deviations

from population mean cost weight and R2 werecalculated for all models.

To calculate R2 for the hybrid models, the totalerror sum of squares for the combined populations

(VEP100 and REMAIN) was calculated as the errorsum of squares from the VEP100 regression plusthe error sum of squares from the REMAIN re-gression. The corrected total sum of squares wascalculated as the sum of squares adjusted for themean of the overall population. We then calculatedR2 as one minus the ratio of the error sum ofsquares to the corrected total sum of squares.

Results

Patients

Table 1 shows the demographic, clinical, andplan characteristics for our population. Only 9.3%had a VEP100 condition in 1992, and 13.8% had aVEP100 condition in 1991, 1992, or 1991 and 1992.

Distribution of Charges

Mean annual charges for all patients in 1992

was $1334, with a relative cost weight (by defini-tion) of 1.00. Mean relative cost weight was 5.23for patients with a VEP100 condition (Table 2) and0.57 for patients without a VEP100 condition. Among VEP100 patients, those with only oneVEP100 condition had a mean relative cost weightof 3.58, whereas those with more than one had amean of 12.33.

Almost all the variation in cost from the popu-lation average is attributable to VEP100 patients.This group accounts for nearly half of total expen-

ditures and 84.3% of the variation in cost (Table 2).The 1.7% of individuals with more than oneVEP100 condition account for 21.6% of expendi-tures and 62.5% of the variation.

Predictive Power of Models

Table 3 shows the R2 for several models. Thedemographic model has little power. Using HCCsimproves prospective predictive power (R2

0.08), but still leaves almost all variation in costunexplained. The concurrent HCC model dramat-ically improves R2 (to 0.37).

When the sample is stratified into those with a VEP100 condition in 1992 and those without


59


5/14

(REMAIN), the model using dummy variables foreach of the VEP100 conditions achieves an R2 of0.21 within the VEP100 population. Combining

these concurrent estimates for the VEP100 popu-lation with prospective HCC predictions for RE-MAIN yields an R2 of 0.26 (Model H1).

Impact of Stratification on HierarchicalCoexisting Conditions Models

Table 3 also shows that Model H2, a hybridmodel that applies prospective HCCs to REMAIN

and concurrent HCCs (rather than VEP100 dum-mies) to patients in the VEP100, generates an R2 of0.36. Thus, using concurrent HCCs on the 9.3% of

the population with a VEP100 diagnosis achievesalmost the same predictive power as applyingHCCs concurrently to the entire population (0.37).

Predictive Ratios

Table 4 shows the predictive ratios for the various models when calculated for clinical or

TABLE 1. Study Population Characteristics

Characteristic

Age on January 1, 1992 (mean SD) 30.4 16.7

Gender N (%)Male 160,895 (50.4)

Female 158,314 (49.6)

Plan Type N (%)

Indemnity 224,609 (70.4)

HMO 28,913 (9.1)

PPO 65,687 (20.6)

Patient total cost in 1991 (mean SD) $1,082 4,652

Patient total cost in 1992 (mean SD) $1,334 5,923

VEP100 Subgroup Status N (%)

Had a code for a VEP100 Diagnosis in 1991 26,050 (8.2)Did not have a code for a VEP100 Diagnosis in 1991 293,159 (91.8)

Had a code for a VEP100 Diagnosis in 1992 29,594 (9.3)

Did not have a code for a VEP100 Diagnosis in 1992 289,615 (90.7)

Had a code for a VEP100 Diagnosis in 1991 or 1992 44,138 (13.8)

Did not have a code for a VEP100 Diagnosis in 1991 or 1992 275,071 (86.2)

TABLE 2. Relative Cost Weights and Contribution to Variation in Cost, by VEP100 St aus in 1992

Number of VEP100Conditions

Mean AnnualRelative Cost

WeightSum Patients

N (%)

Sum Relative Cost Weight Sum(% of Total)

Contribution to Variationin Cost (Sum of Squared

Deviations from MeanPopulation Cost)

SS (%TSS)

0 0.57 289,615 (90.7) 164,434 (51.5) 987,174 (15.7)

1 or more 5.23 29,594 (9.3) 154,775 (48.5) 5,305,647 (84.3)

All enrollees 1.00 319,209 (100.0) 319,209 (100.0) 6,292,821 (100.0)

Among patients with a VEP100 condition:

1 condition 3.58 24,008 (7.5) 85,907 (26.9) 1,374,638 (21.8)

1 condition 12.33 5,586 (1.7) 68,868 (21.6) 3,931,009 (62.5)

Note: SS sum of squared deviations from the population mean for the subgroup; %TSS % of the total sumof squared deviations from the population mean attributable to the subgroup.


60


6/14

cost subgroups. Among the clinical subgroups,the VEP100 hybrid Model H1 has excellentpredictive ratios for those conditions that are inthe VEP100, including breast cancer, depressionthat is not in remission or mild, and cerebralpalsy. However, we have also shown for each ofthese conditions a related condition that is not

included in the VEP100 (melanoma, depressionin remission or mild, and mental retardation,

respectively), and the VEP100 model does notdo as well for these as concurrent models that use allICD-9-CM codes (Models A3 and H1). However,these conditions are either rare (melanoma, mentalretardation) or difficult to verify (depression in re-mission or mild). When predictive ratios are calcu-lated by quintile, the two hybrid models are similar,

whereas concurrent HCCs applied to all models hasbetter predictive ratios.

TABLE 3. R2 for Various Risk Assessment Models

Model Predictors Clinical PopulationPercent of

Patients Timing R 2

A1 age and gender only All patients 100% Prospective 0.02 A2 HCCs All patients 100% Prospective 0.08

A3 HCCs All patients 100% Concurrent 0.37

V1 VEP100 dummies VEP100 patients 9.3% Concurrent 0.21

V2 HCCs VEP100 patients 9.3% Concurrent 0.33

R1 HCCs REMAIN patients 90.7% Prospective 0.07

H1* HCCs VEP100(mixed)

All patients 90.7% 9.3% Hybrid 0.26

H2 HCCs HCCs All patients 90.7% 9.3% Hybrid 0.36

*Component models: R1 V1.

Component models: R1

V2

TABLE 4. Predictive Ratios for Alternative Risk Adjustment Models among Clinical or Cost Subgroups

Age-sexProspective(Model A1)

ProspectiveHCC

(Model A2)

ConcurrentHCCs

(Model A3)

HCC/VEPHybrid

(Model H1)

HCCHybrid

(Model H2)

By Clinical Condition (Those that are among theVEP100 are in plain text, similar conditions

that are not in the VEP100 are in italics):

Breast Cancer 0.21 0.43 0.91 1.00 0.94

Melanoma 0.26 0.49 1.18 0.76 0.80

Cerebral Palsy 0.07 0.61 1.23 1.00 1.27

Mental Retardation 0.08 0.39 1.00 0.28 0.93

Depression (not in remission or mild) 0.22 0.44 0.88 1.00 0.94

Depression (in remission or mild) 0.38 0.68 1.03 0.71 0.81

By Quintile:

Lowest quintile 1992 Expenditures* N/A N/A N/A N/A N/A

Second quintile 1992 Expenditures 47.46 40.51 2.50 27.51 27.03

Third quintile 1992 Expenditures 6.61 5.89 2.94 4.64 4.21Fourth quintile 1992 Expenditures 2.36 2.54 2.43 2.31 2.00

Highest quintile 1992 Expenditures 0.29 0.42 0.79 0.61 0.63

*The sum actual costs for this quintile is zero, so it is not possible to calculate this ratio.


61


7/14

Discussion

RA is essential to establishing a health caremarket that functions well. Because the point of

RA is to explain variation in cost and becauseclinicians often know which patients costs couldbe high, one might expect RA to be easy. Unfor-tunately, it is not. Although clinicians (and healthplans) may know before expenditures are madewhich patients will be expensive, they usually donot know so far ahead of time that they cananswer this question with much accuracy duringthe preceding annual enrollment period. Predict-ing costs a year in advance is difficult. In theabsence of adequate risk adjustment, plans arelikely to prefer to avoid cases that may beexpensive.

Choosing the Timing of Risk Assessment

A key question, then, is: Is it necessary topredict cost at annual enrollment (prospective RA)or can RA wait until a diagnosis is made (concur-rent RA)? From the standpoint of preventingupcoding and over utilization, we should predict

costs prospectively. If thousands of diagnoses canbe associated with large concurrent payments, theauditing effort required to prevent gaming wouldbe substantial. However, if one could limit audit-ing to a small number of important conditions thatare clearly defined and can be verified relativelyeasily, then substantial payment for selected con-ditions is more plausible. Auditing costs could alsobe reduced by focusing on situations in whichreported diagnoses are inconsistent with normalepidemiologic patterns. That is, a plan reporting a

prevalence of VEP100 conditions that is consistent with its demographic mix could be audited lessfrequently, although a plan claiming a large pro-portion of such patients might be audited moreoften.

Having concurrent payments for only a fewconditions also will give health plans more budgetstability than concurrent RA for all conditionswould, as the prevalence of VEP100 conditions isless likely to vary annually than acute conditionssuch as trauma. The hybrid approach we propose

would provide plans payments based on prospec-tive HCC predictions for all patients at the begin-ning of the contract period, and then provideadditional concurrent payments as patients withVEP100 conditions are reported by the plan. Thus,

the plan can estimate its expected revenues withreasonable accuracy by determining how manypatients it expects to have with VEP100 condi-tions, but does not have to tie its revenue stream to

rare or unpredictable events.Finally, there are advantages to not providing

large concurrent payments for events that could becomplications. Many infections such as pneumoniaare complications of chronic diseases. By using pro-spective payments for chronic conditions and notincluding acute infections among the VEP100, ourapproach rewards plans that avoid complications.

The VEP100 we describe is an initial effort to listconditions for this hybrid approach to RA. The listappears to have performed well in identifying

patients who contribute to cost variations. How-ever, it is not clear that these are the best 100conditions or that 100 conditions is the rightnumber. Further research will be needed to deter-mine what the selection criteria should be forreceipt of concurrent payments. These criteriashould include the expected costs of auditing andthe expected benefits in terms of reducing riskselection.

Improving the Predictive Power of RiskAdjustment

The R2 values and predictive ratios reported inTables 3 and 4 prompt the question: If concurrentHCCs do better than a hybrid model using pro-spective HCCs and concurrent VEP100 dummyvariables, why bother developing the VEP100 con-cept? The answer lies in part in reducing auditingcosts and focusing on priority conditions, but alsoin the potential of within-condition RA to improve

predictive power.Prior research has shown that within-conditionRA is possible. For example, Farley et al25 havegenerated models predicting costs among patientswith end stage renal disease. Within-condition RAamong VEP100 conditions would reduce predic-tive errors in the patient group that explains 84.3%of the total variation in cost in the population. Thismight significantly improve the ability of RA tomatch payments to actual costs.

Within-condition RA seems feasible for specific

VEP100 conditions. Many have numerical indicesof severity. For example, the clinical status ofpatients with HIV is known to correlate with twoquantitative markers: the CD4 count (more CD4cells implies a more intact immune system) and


62


8/14

viral load. These markers also correlate with cost,because the prescription of anti-HIV drugs isbased on changes in CD4 count and viral load andcomplications of HIV become more frequent as

immune status worsens. Similar continuous mea-sures of severity exist for heart, lung, and liverdisease.

Other conditions such as sickle cell disease andcancer have categorical/ordinal severity scales thatmay be amenable to within-condition RA. Forinstance, every cancer has a staging system, andtreatment and cost often vary by stage. Thus,focusing RA on a short list of conditions like the VEP100 and pursuing within-condition RA maybe a fruitful approach to developing the next

generation of RA tools.

Applications and Limitations

The failure of RA models to predict well within-condition has negative effects on several financialaspects of health care, such as carve-out ratesetting14 and the ability to do specialty providerprofiling, in addition to health plan premiumadjustment. Our approach could help in each of

these cases, because it would allow for morevariation in carve-out rates (by using clinical detailto estimate within-condition severity) and wouldpermit profiling of specialists utilization patternsthat addresses case mix variation within diagnosticcategories (as would be necessary, eg, in evaluat-ing oncologists efficiency in managing lungcancer).

There are limitations to our analysis. We re-quired that patients have a full 2 years of eligibility,excluding all who died in the second year. (Con-

current models could include these people, butfully prospective ones cannot without annualiza-tion.) This could have created bias tending tooverestimate (if patients who disenrolled had highutilization but did not have VEP100 conditions) orunderestimate (especially if patients who died hadhigh costs attributable to VEP100 conditions) thepredictive potential of the VEP100 approach. Anyattempt to implement the VEP100 approachshould use models developed on data appropriatefor the population to which they will be applied,

using data that includes partial-year enrollees andis more recent than ours.Nonetheless, the limitations of our data are

unlikely to have had much effect on our mainpoint: that it may be possible to reap most of the

benefits of concurrent RA while generating fewerof the incentives to game by limiting concurrentpayments to a small number of patients withpredictably expensive, clinically defined, and veri-

fiable conditions. We may not have chosen thebest conditionsthere may be better choicesamong the high cost categories in HCCs or othermodels but this is a matter for further research.Our strategy of focusing on a small number ofchronic conditions is consistent with the Instituteof Medicines recent call for the identification ofpriority conditionsfor quality improvement, pay-ment reform, and policy attention,26 and with therecent announcement by CMS that it is consider-ing focusing RA in Medicare on a list of 25 to 100

conditions.12

The key issue, then, is: How much is it worth toimprove predictive power? Our approach wouldadd data collection and auditing costs. Nonethe-less, it offers an intermediate solution betweenprospective and concurrent RA and may be supe-rior to either, because it has greater predictivepower than prospective and lower cost but similarpredictive power with greater potential than con-current RA using claims data. The final advantageof our approach is that the clinical data we collect

could be used for other purposes, so that the costsincurred should be weighed against more benefitsgained than just improved RA.

Conclusions

The VEP100 may not be exactly the right set ofconditions, but there can certainly be a short list ofconditions that are verifiable, predictably expen-sive, and account for a high percentage of total

cost variation. The data collection and auditingburdens associated with concurrent RA for se-lected conditions is much less than it would be forconcurrent RA for all conditions, so the predictivepower is gained at lower cost. In addition, identi-fying these conditions suggests a direction forfurther RA research to generate additional predic-tive power and further reduce incentives to selectrisk.

We cannot claim these benefits justify the ad-ditional costs. Further empirical efforts to develop

within-condition models and determine data col-lection costs are needed, as is debate about thecost-benefit tradeoffs. There may be, however, anRA strategy that may be superior to those previ-ously considered.


63


9/14

References

1. Kronick R, Gilmer T, Dreyfus T, et al. Improv-ing health-based payment for Medicaid beneficiaries:CDPS. Health Care Financ Rev 2000;21:2964.

2. Hash M. From the Health Care Financing Ad-ministration: Medicare Choice Plans. JAMA2000;284:2988.

3. Knutson D. Case study: the Minneapolis BuyersHealth Care Action Group. Inquiry 1998;35:171177.

4. Carter GM, Bell RM, Dubois RW, et al. Aclinically detailed risk information system for cost.Health Care Financ Rev 2000;21:6591.

5. Ellis RP, Pope GC, Iezzoni L, et al. Diagnosis-based risk adjustment for Medicare capitation payments.Health Care Financ Rev 1996;17:101128.

6. Kronick R, Dreyfus T, Lee L, et al. Diagnosticrisk adjustment for Medicaid: The disability paymentsystem. Health Care Financ Rev 1996;17:733.

7. Meenan RT, OKeeffe-Rosetti C, Hornbrook MC,et al. The sensitivity and specificity of forecasting high-costusers of medical care. Med Care 1999;37:815823.

8. Weiner JP, Dobson A, Maxwell SL, et al.Risk-adjusted Medicare capitation rates using ambula-tory and inpatient diagnoses. Health Care Financ Rev1996;17:7799.

9. Roblin DW. Physician profiling using outpatient

pharmacy data as a source for case mix measurement andrisk adjustment. J Ambul Care Manage 1998;21:6884.

10. Gilmer T, Kronick R, Fishman P, et al. TheMedicaid Rx model: Pharmacy-based risk adjustment forpublic programs. Med Care 2001;39:11881202.

11. Parkerson GR, Jr., Harrell FE, Hammond WE,et al. Characteristics of adult primary care patients aspredictors of future health services charges. Med Care2001;39:1170 1181.

12. Centers for Medicare and Medicaid Services.Medicare Choice Risk Adjustment Public Meeting -

Meeting Materials. Available at: http://www.hcfa.gov/medicare/rskadjmtrls.htm. March 15, 2002.

13. Newhouse JP, Manning WG, Keeler EB, et al. Adjusting capitation rates using objective health mea-sures and prior utilization. Health Care Financ Rev1989;10:4154.

14. Ettner SL, Frank RG, McGuire TG, et al.Risk adjustment alternatives in paying for behavioralhealth care under Medicaid. Health Serv Res2001;36:793 811.

15. van Vliet RC, Lamers LM. The high costs ofdeath: should health plans get higher payments whenmembers die? Med Care 1998;36:14511460.

16. Gaumer GL, Stavins J. Medicare use in the lastninety days of life. Health Serv Res 1992;26:725742.

17. Newhouse JP. Patients at risk: Health reformand risk adjustment. Health Aff 1994;13:132146.

18. Porell FW, Gruenberg L. Discretionary hospitaluse and diagnostic risk adjustment of Medicare HMOcapitation rates. Inquiry 2000;37:162172

19. Conviser R, Murray M, Lau D. Medicaid man-

aged care reimbursement for HIV and its implications foraccess to care. Am J Manag Care 2000;6:990999.

20. Ashton CM, Petersen NJ, Souchek J, et al.Geographic variations in utilization rates in VeteransAffairs hospitals and clinics. N Engl J Med 1999;340:3239.

21. Berlowitz DR, Ash AS, Hickey EC, et al.Profiling outcomes of ambulatory care: Casemix affectsperceived performance. Med Care 1998;36:928 933.

22. Thomas JW, Bates EW, Hofer T, et al. Inter-preting risk-adjusted length of stay patterns for VA

hospitals. Med Care 1998;36:16601675.

23. Rossiter LF, Whitehurst-Cook MY, Small RE,et al. The impact of disease management on outcomesand cost of care: a study of low-income asthma patients.Inquiry 2000;37:188202.

24. Fowles JB, Weiner JP, Knutson D, et al.Taking health status into account when setting capitationrates: A comparison of risk-adjustment methods. JAMA1996;276:1316 1321.

25. Farley DO, Carter GM, Kallich JD, et al.Modified capitation and treatment incentives for endstage renal disease. Health Care Financ Rev1996;17:129 142.

26. Institute of Medicine. Crossing the qualitychasm: A new health system for the 21st century.Washington, DC: National Academy Press; 2001:335.


64


10/14

APPENDIX A. TABLE 1. VEP100 Conditions and Associated ICD-9-CM Codes

VEP # Short Description ICD-9-CM Codes

1 Anoxic brain damage 348.1

2 Stomach cancer 230.2, 151, 151.0, 151.1, 151.2, 151.3, 151.4, 151.5, 151.6, 151.7,151.8, 151.9

3 AMI 410.xx, 411.0

4 Leukemias 204.xx, 205.xx, 206.xx, 207.xx, 208.xx

5 Peritonitis 567, 567.0, 567.1, 567.2, 567.3, 567.4, 567.5, 567.6, 567.7, 567.8,567.9

6 Chronic benign panc. dz 577.1, 577.2, 577.3, 577.4, 577.5, 577.6, 577.7, 577.8, 577.9,251.4, 251.5, 251.6, 251.7, 251.8, 251.9

7 Sickle cell anemia 282.5, 282.6

8 Lung/pleural cancer 195.1, 231.1, 231.2, 235.7, 235.8, 162, 162.0, 162.1, 162.2, 162.3,162.4, 162.5, 162.6, 162.7, 162.8, 162.9, 163, 163.0, 163.1,

163.2, 163.3, 163.4, 163.5, 163.6, 163.7, 163.8, 163.99 CNS trauma 806.xx, 851.xx, 852.xx, 853.xx

10 Neonatal/chronic resp dstrss 769, 770.7

11 Premature/low bwt babies (764.xx, 765.xx) and age (0, 1)

12 Alzheimers/cer. deg/dementia 331.xx, 290.xx, 294, 294.1, 294.2, 294.3, 294.4, 294.5, 294.6,294.7, 294.8, 294.9

13 Chronic benign bil trct dz 576, 576.0, 576.1, 576.2, 576.3, 576.4, 576.5, 576.6, 576.7, 576.8,576.9

14 Osteomyelitis 730.xx

15 Cancer of nervous system 237.0, 237.1, 237.3, 237.5, 237.6, 237.7, 237.8, 237.9, 191, 191.0,191.1, 191.2, 191.3, 191.4, 191.5, 191.6, 191.7, 191.8, 191.9,192, 192.0, 192.1, 192.2, 192.3, 192.4, 192.5, 192.6, 192.7,192.8, 192.9

16 Chronic renal failure 585, 586, 588, 588.0, 588.1, 588.2, 588.3, 588.4, 588.5, 588.6,588.7, 588.8, 588.9

17 Mediastinal cancer 164, 164.0, 164.1, 164.2, 164.3, 164.4, 164.5, 164.6, 164.7, 164.8,164.9

18 Cancer of ovaries/adj strctrs 236.2, 183, 183.0, 183.1, 183.2, 183.3, 183.4, 183.5, 183.6, 183.7,183.8, 183.9

19 Aseptic necrosis of bone 733.4x

20 Pancreatic cancer 157, 157.0, 157.1, 157.2, 157.3, 157.4, 157.5, 157.6, 157.7, 157.8,157.9

21 Prostate cancer 233.4, 236.5, 185

22 Cancer: sm/lg intestine/anus 159.0, 159.8, 159.9, 195.2, 230.3, 230.4, 230.5, 230.6, 230.7,230.9, 235.2, 235.5, 239.0, 152, 152.0, 152.1, 152.2, 152.3,152.4, 152.5, 152.6, 152.7, 152.8, 152.9, 153, 153.0, 153.1,153.2, 153.3, 153.4, 153.5, 153.6, 153.7, 153.8, 153.9, 154,154.0, 154.1, 154.2, 154.3, 154.4, 154.5, 154.6, 154.7, 154.8,154.9

23 A1 Antitrypsin df/hered angio 277.6

24 Esophageal cancer 230.1, 150, 150.0, 150.1, 150.2, 150.3, 150.4, 150.5, 150.6, 150.7,150.8, 150.9

(Continues)


65


11/14


25 Cerebrovasc dz and comps 431, 438, 432, 432.0, 432.1, 432.2, 432.3, 432.4, 432.5, 432.6,432.7, 432.8, 432.89, 433, 433.0, 433.1, 433.2, 433.3, 433.4,433.5, 433.6, 433.7, 433.8, 433.9, 434, 434.0, 434.1, 434.2,434.3, 434.4, 434.5, 434.6, 434.7, 434.8, 434.9, 435, 435.0,435.1, 435.2, 435.3, 435.4, 435.5, 435.6, 435.7, 435.8, 435.9,436.xx, 437.xx

26 HIV 042.xx, 043, 043.0, 043.1, 043.2, 043.3, 043.4, 043.5, 043.6, 043.7,043.8, 043.9, 440, 044.1, 044.2, 044.3, 044.4, 044.5, 044.6,044.7, 044.8, 044.9, 795.8, 079.53, V08

27 Testicular cancer 236.4, 186, 186.0, 186.1, 186.2, 186.3, 186.4, 186.5, 186.6, 186.7,186.8, 186.9

28 Sepsis, prob. nosocomial pneumonia 038.xx, 482.8x, 482.0, 482.1, 790.7

29 Female genital fistula 619, 619.0, 619.1, 619.2, 619.3, 619.4, 619.5, 619.6, 619.7, 619.8,619.9

30 Endometriosis 617, 617.0, 617.1, 617.2, 617.3, 617.4, 617.5, 617.6, 617.7, 617.8,617.9

31 Hepatobiliary cancer 230.8, 235.3, 155, 155.0, 155.1, 155.2, 155.3, 155.4, 155.5, 155.6,155.7, 155.8, 155.9, 156.0, 156.1, 156.2, 156.3, 156.4, 156.5,156.6, 156.7, 156.8, 156.9

32 Lymph/sarcomas, otr lymph tmrs 200.xx, 201.xx, 202.xx, 238.4, 238.5, 238.6, 238.7

33 Coagulation defects 286, 286.0, 286.1, 286.2, 286.3, 286.4, 286.5, 286.6, 286.7, 286.8,286.9

34 Pulmonary fibrosis 508.1, 515, 516.3

35 Breast cancer 233.0, 238.3, 239.3, 174, 174.0, 174.1, 174.2, 174.3, 174.4, 174.5,174.6, 174.7, 174.8, 174.9, 175, 175.0, 175.1, 175.2, 175.3,

175.4, 175.5, 175.6, 175.7, 175.8, 175.9

36 Coronary artery dz (not AMI) (411, 411.1, 411.8, 412, 429.2, 413, 413.0, 413.1, 413.2, 413.3,413.4, 413.5, 413.6, 413.7, 413.8, 413.9, 414, 414.0, 414.1,414.2, 414.3, 414.4, 414.5, 414.6, 414.7, 414.8, 414.9) andnot (410.x or 411.0)

37 Primary thrombocytopenia 287.3

38 Head and neck cancer 140, 140.0, 140.1, 140.2, 140.3, 140.4, 140.5, 140.6, 140.7, 140.8,140.9, 141, 141.0, 141.1, 141.2, 141.3, 141.4, 141.5, 141.6,141.7, 141.8, 141.9, 142, 142.0, 142.1, 142.2, 142.3, 142.4,142.5, 142.6, 142.7, 142.8, 142.9, 143, 143.0, 143.1, 143.2,143.3, 143.4, 143.5, 143.6, 143.7, 143.8, 143.9, 144, 144.0,

144.1, 144.2, 144.3, 144.4, 144.5, 144.6, 144.7, 144.8, 144.9,145, 145.0, 145.1, 145.2, 145.3, 145.4, 145.5, 145.6, 145.7,145.8, 145.9, 146, 146.0, 146.1, 146.2, 146.3, 146.4, 146.5,146.6, 146.7, 146.8, 146.9, 147, 147.0, 147.1, 147.2, 147.3,147.4, 147.5, 147.6, 147.7, 147.8, 147.9, 148, 148.0, 148.1,148.2, 148.3, 148.4, 148.5, 148.6, 148.7, 148.8, 148.9, 149,149.0, 149.1, 149.2, 149.3, 149.4, 149.5, 149.6, 149.7, 149.8,149.9, 160, 160.0, 160.1, 160.2, 160.3, 160.4, 160.5, 160.6,160.7, 160.8, 160.9, 161, 161.0, 161.1, 161.2, 161.3, 161.4,161.5, 161.6, 161.7, 161.8, 161.9, 165, 165.0, 165.1, 165.2,165.3, 165.4, 165.5, 165.6, 165.7, 165.8, 165.9, 195.0, 230.0,231.0, 231.8, 231.9, 235.0, 235.1, 235.6, 235.9, 239.1

39 Paralysis 344.xx, 342, 342.0, 342.1, 342.2, 342.3, 342.4, 342.5, 342.6, 342.7,342.8, 342.9

(Continues)


66

APPENDIX A. TABLE 1 (Continued)


12/14


40 Cerebral degen of childhood 330, 330.0, 330.1, 330.2, 330.3, 330.4, 330.5, 330.6, 330.7, 330.8,330.9

41 Bladder cancer 233.7, 236.7, 239.4, 188, 188.0, 188.1, 188.2, 188.3, 188.4, 188.5,188.6, 188.7, 188.8, 188.9

42 Multiple myeloma 203.0x

43 Depression: not remission or mild 298.0, 301.12, 296.0, 296.1, 296.2, 296.3, 296.4, 296.5, 296.6,296.00, 296.02, 296.03, 296.04, 296.10, 296.12, 296.13,296.14, 296.20, 296.22, 296.23, 296.24, 296.30, 296.32,296.33, 296.34, 296.40, 296.42, 296.43, 296.44, 296.50,296.52, 296.53, 296.54, 296.60, 296.62, 296.63, 296.64, 296.7,296.8, 296.9

44 Chronic athero periph vasc dz 443.8x, 441.2, 441.4, 441.7, 441.9, 443.9, 440, 440.0, 440.1, 440.2,440.3, 440.4, 440.5, 440.6, 440.7, 440.8, 440.9, 557, 557.0,557.1, 557.2, 557.3, 557.4, 557.5, 557.6, 557.7, 557.8, 557.9

45 Disorders of porphyrin metab 277.1

46 Cholesteatoma 385.3x

47 Chronic pelvic inflam. dz 614.1, 614.2, 614.4, 614.6, 614.7, 614.8, 614.9, 615.1

48 Drug abuse 292.xx, 304.xx, 305.2x, 305.3x, 305.4x, 305.5x, 305.6x, 305.7x,305.8x, 305.9x

49 Organ transplantation 996.8x, V42, V42.0, V42.1, V42.2, V42.3, V42.4, V42.5, V42.6,V42.7, V42.8, V42.9

50 Other chronic GU disease 596.5x, 788.2x, 596.4, 591, 593, 593.xx

51 Spinocerebellar dz 334, 334.0, 334.1, 334.2, 334.3, 334.4, 334.5, 334.6, 334.7, 334.8,334.9

52 Pituitary disorders 253, 253.0, 253.1, 253.2, 253.3, 253.4, 253.5, 253.6, 253.7, 253.8,253.9

53 Aplastic anemias 284, 284.0, 284.1, 284.2, 284.3, 284.4, 284.5, 284.6, 284.7, 284.8,284.9

54 Acquired hemolytic anemias 283, 283.0, 283.1x, 283.2, 283.3, 283.4, 283.5, 283.6, 283.7, 283.8,283.9

55 Inflam bowel dz 556, 555, 555.0, 555.1, 555.2, 555.3, 555.4, 555.5, 555.6, 555.7,555.8, 555.9

56 Ant. horn cell dz/ALS 335.xx

57 Chronic liver dz 571.xx, 456.0, 456.1, 456.2, 572.2, 572.3, 572.4, 572.5, 572.6,572.7, 572.8, 789.5, 452, 453.0

58 Cerebral palsy 343, 343.0, 343.1, 343.2, 343.3, 343.4, 343.5, 343.6, 343.7, 343.8,343.9

59 Cervical cancer 233.1, 180, 180.0, 180.1, 180.2, 180.3, 180.4, 180.5, 180.6, 180.7,180.8, 180.9

60 Diab Mel w/chronic compl if 250.4x or 250.5x or 250.6x or 250.7x or 250.8x or 362.0x

61 Schizophrenia 295.xx

62 Spondylopath/spondylos/disc 720.xx, 721.xx, 722.xx

63 CHF, cardiomyopathies 429.3, 425, 425.0, 425.1, 425.2, 425.3, 425.4, 425.5, 425.6, 425.7,425.8, 425.9, 428, 428.0, 428.1, 428.2, 428.3, 428.4, 428.5,428.6, 428.7, 428.8, 428.9

64 Chronic humoral deficiencies 279.0x

65 Alcohol abuse 303.xx, 305.0x, 291, 291.0, 291.1, 291.2, 291.3, 291.4, 291.5,291.6, 291.7, 291.8, 291.9

66 Lipidoses 272.7

(Continues)


67



13/14


67 Cancer: kidney/upper urin trct 236.9x, 239.5, 189, 189.0, 189.1, 189.2, 189.3, 189.4, 189.5, 189.6,189.7, 189.8, 189.9

68 Other female genital cancers 233.3, 236.3, 184, 184.0, 184.1, 184.2, 184.3, 184.4, 184.5, 184.6,184.7, 184.8, 184.9

69 Other endocrine malignancies 234.8, 237.2, 237.4, 239.7, 194, 194.0, 194.1, 194.2, 194.3, 194.4,194.5, 194.6, 194.7, 194.8, 194.9

70 Cataract 366.xx

71 Chronic benign esoph dz 530.0, 530.3, 530.5, 530.6

72 Cystic fibrosis 277.0x

73 Multiple sclerosis, etc. 340, 341, 341.0, 341.1, 341.2, 341.3, 341.4, 341.5, 341.6, 341.7,341.8, 341.9

74 Spinal cord injury 952.xx

75 Polio 045.xx, 138

76 Diab mel w/acute compl if (250.1x or 250.2x or 250.3x or 250.9x) and 250.4x and NOT(250.5x and 250.6x, 250.7x and 250.8x)

77 Uterine cancer 179, 181, 233.2, 236.0, 236.1, 182, 182.0, 182.1, 182.2, 182.3,182.4, 182.5, 182.6, 182.7, 182.8, 182.9

78 Seizure disorders 345.xx

79 Disorders of plsma prot & min metab 273, 273.0, 273.1, 273.2, 273.3, 273.4, 273.5, 273.6, 273.7, 273.8,273.9, 275.xx

80 Chronic renal dz w/o chrn ren failure (581.xx, 582.xx, 583.xx, 590.0x, 587) and not (585, 586, 588,588.0, 588.1, 588.2, 588.3, 588.4, 588.5, 588.6, 588.7, 588.8,588.9)

81 Malabsorption syndrome 579, 579.0, 579.1, 579.2, 579.3, 579.4, 579.5, 579.6, 579.7, 579.8,

579.982 Congenital anomalies 740.xx, 741.xx, 742.xx, 743.xx, 744.xx, 745.xx, 746.xx, 747.xx,

748.xx, 749.xx, 750.xx, 751.xx, 752.xx, 753.xx, 754.xx, 755.xx,756.xx, 757.xx, 758.xx, 759.xx

83 Chronic cellular deficiencies 279.1x

84 Myelopathies 336, 336.0, 336.1, 336.2, 336.3, 336.4, 336.5, 336.6, 336.7, 336.8,336.9

85 Diffuse conn tissue dz 710, 710.0, 710.1, 710.2, 710.3, 710.4, 710.5, 710.6, 710.7, 710.8,710.9

86 Cancer of the eye 234.0, 190, 190.0, 190.1, 190.2, 190.3, 190.4, 190.5, 190.6, 190.7,190.8, 190.9

87 Cardiac rhythm disturbances 426.xx, 427.0x, 427.1x, 427.2x, 427.3x, 427.4x, 427.5x, 427x,427.8x, 427.9x

88 Non-psych org mental disorder 310, 310.0, 310.1, 310.2, 310.3, 310.4, 310.5, 310.6, 310.7, 310.8,310.9

89 Advanced syphilis 090.4x, 093.xx, 094.xx, 095.xx, 090.5, 090.6, 090.7, 090.8, 090.9

90 Chronic skin ulcer 707, 707.0, 707.1, 707.2, 707.3, 707.4, 707.5, 707.6, 707.7, 707.8,707.9

91 Chronic arteritis/related conds 447.6, 446.xx

92 Parkinsons/degen of bsl gag/etc 333.9x, 333.0, 332, 332.0, 332.1, 332.2, 332.3, 332.4, 332.5, 332.6,332.7, 332.8, 332.9

93 Inflam polyarthropathies 714.xx

94 Cancer: unspec/uncertain site 234.9, 199, 199.0, 199.1, 199.2, 199.3, 199.4, 199.5, 199.6, 199.7,199.8, 199.9

95 Anorexia/eating disorders 307.1x, 307.5x

(Continues)


68



14/14


96 Sarcoidosis 135

97 Chronic peptic ulcer dz 530.1, 530.2, 531.4x, 531.5x, 531.6x, 531.7x, 532.4x, 532.5x,

532.6x, 532.7x, 533.4x, 533.5x, 533.6x, 533.7x98 Adrenal disorders 255, 255.0, 255.1, 255.2, 255.3, 255.4, 255.5, 255.6, 255.7, 255.8,

255.9

99 Cancer of bone or soft tissue 195.3, 195.4, 195.5, 195.6, 195.7, 195.8, 235.4, 238.0, 238.1,239.2, 170, 170.0, 170.1, 170.2, 170.3, 170.4, 170.5, 170.6,170.7, 170.8, 170.9, 171, 171.0, 171.1, 171.2, 171.3, 171.4,171.5, 171.6, 171.7, 171.8, 171.9

100 Bronchiectasis 494

x any value from 0 to 9.


69


The Best of Both Worlds%3F Potential of Hybrid Prospective%2FConcurrent Risk Adjustment

Documents