Risk adjustment is essential for understanding resource utilization, provider profiling, resource allocation, and quality. Performance monitoring efforts in the VA have been hindered by the lack of an adequate risk adjustment method for patients with mental health and/or substance abuse (MH/SA) disorders. The number of veterans receiving VA mental health services has increased 36 percent in the last decade, totaling over 650,000 veterans in FY99 at a cost of almost $2 billion. Budget restrictions and staffing limitations have heightened the need for accurate methods for profiling services, assessing quality, and allocating resources, but these efforts have been limited by lack of an adequate MH/SA case mix measure. Whereas most prior psychiatric case mix research has focused on outcomes in inpatient settings or variations of particular MH/SA disorders, VA mental health services have undergone a major shift from inpatient to outpatient care.
Our principal objectives are to: 1) derive and validate a risk-adjustment methodology for veterans with MH/SA disorders using psychiatric diagnosis-based case-mix system (PDCM) piloted by one of our team members; 2) to compare the predictive power of this method and existing models to explain concurrent (FY 99) and prospective (FY 99 to FY 00) MH/SA and total utilization and costs among patients with psychiatric diagnoses; 3) to profile variation in MH/SA utilization and expenditures across VA facilities nationally; and 4) merge VA data with Medicare data to determine the extent to which these case-mix measures that we develop may be affected by access barriers or use of Medicare services.The PDCM will be added to each of the three leading case-mix measures Adjusted Clinical Group (ACGs), Diagnostic Cost Group (DCGs), and Chronic Illness and Disability Payment System (CDPS) to compare the ability of these newly combined case-mix measures in improving concurrent and prospective prediction of resource consumption.
We identified all veterans who received health care services in FY’99 and had a MH/SA diagnosis (ICD-9-CM codes 290.00-312.99, 316.00-316.99). The two main sources of data for this project are VA and Medicare files.
At the VA, we accessed four databases via the Austin Automation Center: the Outpatient Clinic File, the Patient Treatment File, the Extended Care File, and the Beneficiary Identification and Record Locator Subsystem File. We also obtained utilization cost estimates from the VA Health Economics Resource Center, and total pharmacy costs from DSS. From Medicare, we obtained denominator, inpatient, outpatient, and provider files from the same time period, which will be merged with VA files using patients’ social security numbers. We developed concurrent risk-adjustment models on a 60% random sample of MH/SA patients and validated them on 40%. We choose three outcomes variable (total cost, MH/SA cost, and MH/SA outpatient cost) for analyses.
We found small differences in models’ performance in predicting both total and MH/SA costs in our preliminary analyses of the three leading case-mix systems; mean absolute prediction errors and predictive ratios did not demonstrate any clear ranking of model performance. The PDCM performed reasonably better than other case-mix systems for both MH/SA and MH/SA outpatient cost concurrently and prospectively. We selected combined model (ACG+PDCM) and two single models (DCG and CDPS) for concurrent and prospective total cost, respectively.
This project is of great relevance to the VA. We have derived a diagnosis-based, psychiatric case mix measure, specific to the VA, that can be used to profile and predict concurrent and future resource consumption among patients with mental health and substance abuse disorders across VA facilities.
- Carey K, Montez-Rath ME, Rosen AK, Christiansen CL, Loveland S, Ettner SL. Use of VA and Medicare services by dually eligible veterans with psychiatric problems. Health services research. 2008 Aug 1; 43(4):1164-83.
- Montez-Rath M, Christiansen CL, Ettner SL, Loveland S, Rosen AK. Performance of statistical models to predict mental health and substance abuse cost. BMC medical research methodology. 2006 Oct 26; 6:53.
- Sloan KL, Montez-Rath ME, Spiro A, Christiansen CL, Loveland S, Shokeen P, Herz L, Eisen S, Breckenridge JN, Rosen AK. Development and validation of a psychiatric case-mix system. Medical care. 2006 Jun 1; 44(6):568-80.
- Rosen AK, Christiansen CL, Montez ME, Loveland S, Shokeen P, Sloan KL, Ettner SL. Evaluating risk-adjustment methodologies for patients with mental health and substance abuse disorders in the Veterans Health Administration. International Journal of Healthcare Technology. 2006 Jan 2; 7(1/2):43-81.
- Dossa A, Berlowitz DR, Loveland S, Hoenig H. Impact of Psychiatric Comorbidity on Rehospitalization, Mortality, and Functional Outcomes in Stroke Patients Following Inpatient Rehabilitation. Poster session presented at: VA HSR&D National Meeting; 2009 Feb 13; Baltimore, MD.
- Chatterjee, Rosen AK, Seal, Glickman, Spiro. Predicting Mental Health Outcomes in the VA: What is the Best Model for Risk. Poster session presented at: VA HSR&D National Meeting; 2008 Feb 13; Baltimore, MD.
- Loveland SA, Christiansen CL, Zhao S, Rivard PE, Rosen AK. Risk Adjustment in VA populations - A Method for Partial Recalibration. Presented at: VA HSR&D National Meeting; 2007 Feb 21; Arlington, VA.
- Montez-Rath ME, Christiansen CL, Ettner SL, Loveland SA, Rosen AK. Performance of Statistical Models to Predict Mental Health and Substance Abuse Cost. Paper presented at: Joint Statistical Annual Meeting; 2006 Aug 1; Seattle, WA.
- Montez M, Christiansen C, Loveland S, Ettner S, Shokeen P, Rosen A. Statistical Model Comparisons for Prediction of Mental Health and Substance Abuse Cost in the Veterans Health Administration. Paper presented at: American Statistical Association Joint Statistical Annual Meeting; 2004 Aug 1; Toronto, Canada.
- Rosen A. How Well Does the CDPS Predict Utilization: A Comparison to Other Case-mix Measures. Paper presented at: VA HSR&D National Meeting; 2002 Feb 14; Washington, DC.
Sensory Loss, Mental, Cognitive and Behavioral Disorders, Substance Abuse and Addiction, Health Systems
Epidemiology, Technology Development and Assessment
Dual diagnosis – substance abuse and mental health, Research measure, Risk adjustment