Risk adjustment using automated ambulatory pharmacy data: the RxRisk model.

Center for Health Studies, Group Health Cooperative, Seattle, Washington 98101, USA.
Medical Care (Impact Factor: 2.94). 02/2003; 41(1):84-99. DOI: 10.1097/01.MLR.0000039830.19812.29
Source: PubMed

ABSTRACT Develop and estimate the RxRisk model, a risk assessment instrument that uses automated ambulatory pharmacy data to identify chronic conditions and predict future health care cost. The RxRisk model's performance in predicting cost is compared with a demographic-only model, the Ambulatory Clinical Groups (ACG), and Hierarchical Coexisting Conditions (HCC) ICD-9-CM diagnosis-based risk assessment instruments. Each model's power to forecast health care resource use is assessed.
Health services utilization and cost data for approximately 1.5 million individuals enrolled in five mixed-model Health Maintenance Organizations (HMOs) from different regions in the United States.
Retrospective cohort study using automated managed care data. SUBJECTS All persons enrolled during 1995 and 1996 in Group Health Cooperative of Puget Sound, HealthPartners of Minnesota and the Colorado, Ohio and Northeast Regions of Kaiser-Permanente. MEASURES RxRisk, an algorithm that classifies prescription drug fills into chronic disease classes for adults and children.
HCCs produce the most accurate forecasts of total costs than either RxRisk or ACGs but RxRisk performs similarly to ACGs. Using the R(2) criteria HCCs explain 15.4% of the prospective variance in cost, whereas RxRisk explains 8.7% and ACGs explain 10.2%. However, for key segments of the cost distribution the differences in forecasting power among HCCs, RxRisk, and ACGs are less obvious, with all three models generating similar predictions for the middle 60% of the cost distribution.
HCCs produce more accurate forecasts of total cost, but the pharmacy-based RxRisk is an alternative risk assessment instrument to several diagnostic based models and depending on the nature of the application may be a more appropriate option for medical risk analysis.

1 Bookmark
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Mortality prediction models generally require clinical data or are derived from information coded at discharge, limiting adjustment for presenting severity of illness in observational studies using administrative data. To develop and validate a mortality prediction model using administrative data available in the first 2 hospital days. After dividing the dataset into derivation and validation sets, we created a hierarchical generalized linear mortality model that included patient demographics, comorbidities, medications, therapies, and diagnostic tests administered in the first 2 hospital days. We then applied the model to the validation set. Patients aged ≥18 years admitted with pneumonia between July 2007 and June 2010 to 347 hospitals in Premier, Inc.'s Perspective database. In hospital mortality. The derivation cohort included 200,870 patients and the validation cohort had 50,037. Mortality was 7.2%. In the multivariable model, 3 demographic factors, 25 comorbidities, 41 medications, 7 diagnostic tests, and 9 treatments were associated with mortality. Factors that were most strongly associated with mortality included receipt of vasopressors, non-invasive ventilation, and bicarbonate. The model had a c-statistic of 0.85 in both cohorts. In the validation cohort, deciles of predicted risk ranged from 0.3% to 34.3% with observed risk over the same deciles from 0.1% to 33.7%. A mortality model based on detailed administrative data available in the first 2 hospital days had good discrimination and calibration. The model compares favorably to clinically based prediction models and may be useful in observational studies when clinical data are not available.
    PLoS ONE 01/2014; 9(1):e87382. · 3.53 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Background Risk adjustment instruments applied to existing electronic health records and administrative datasets may contribute to monitoring the correct prescribing of medicines. Objective We aim to test the suitability of the model based on the CRG system and obtain specific adjusted weights for determined health states through a predictive model of pharmaceutical expenditure in primary health care. Methods A database of 261,054 population in one health district of an Eastern region of Spain was used. The predictive power of two models was compared. The first model (ATC-model) used nine dummy variables: sex and 8 groups from 1 to 8 or more chronic conditions while in the second model (CRG-model) we include sex and 8 dummy variables for health core statuses 2 to 9. Results The two models achieved similar levels of explanation. However, the CRG system offers higher clinical significance and higher operational utility in a real context, as it offers richer and more updated information on patients. Conclusions The potential of the CRG model developed compared to ATC codes lies in its capacity to stratify the population according to specific chronic conditions of the patients, allowing us to know the degree of severity of a patient or group of patients, predict their pharmaceutical cost and establish specific programmes for their treatment.
    Health Policy 06/2014; · 1.73 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Currently-used risk-equalization models do not adequately compensate insurers for predictable differences in individuals' health care expenses. Consequently, insurers face incentives for risk rating and risk selection, both of which jeopardize affordability of coverage, accessibility to health care, and quality of care. This study explores to what extent the predictive performance of the prediction model used in risk equalization can be improved by using additional administrative information on costs and diagnoses from three prior years. We analyze data from 13.8 million individuals in the Netherlands in the period 2006-2009. First, we show that there is potential for improving models' predictive performance at both the population and subgroup level by extending them with risk adjusters based on cost and/or diagnostic information from multiple prior years. Second, we show that even these extended models do not adequately compensate insurers. By using these extended models incentives for risk rating and risk selection can be reduced substantially but not removed completely. The extent to which risk-equalization models can be improved in practice may differ across countries, depending on the availability of data, the method chosen to calculate risk-adjusted payments, the value judgment by the regulator about risk factors for which the model should and should not compensate insurers, and the trade-off between risk selection and efficiency.
    The European Journal of Health Economics 02/2014; · 2.10 Impact Factor

Full-text (2 Sources)

Available from
May 27, 2014