Health Economics Spotlight

Updates on VA data, health economics research, and analytic methods

September 2023

In This Issue

1. September seminar: Comparing Data from Multiple Health Systems to Estimate Disease Burden for VA-Medicaid Dual Enrollees
2. October seminar: Prediction of opioid-related overdose and suicide events using administrative healthcare data
3. Using Medicare and Medicaid Cost Data in VA Research
4. Discharge Dataset with Inpatient Categories of Care
5. Wage Index Updated for FY23
6. ICD-10 codes for specific conditions

Seminars

Comparing Data from Multiple Health Systems to Estimate Disease Burden for VA-Medicaid Dual Enrollees

HERC Health Economics Seminar

Wednesday, September 20 at 2pm ET

Register

Investigator, Center for Innovations in Quality, Effectiveness and Safety (IQuESt), Michael E. Debakey VA Medical Center, Houston, TX
Assistant Professor of Medicine, Baylor College of Medicine, Houston TX

Health Economist & Investigator, Center for Innovations in Quality, Effectiveness and Safety (IQuESt), Michael E. Debakey VA Medical Center, Houston, TX
Instructor, Baylor College of Medicine, Houston TX

For Medicaid-VA dual enrollees, we examine the differences in three commonly used risk-adjustment measures as calculated VA data, Medicaid data and both sources of data combined. Analysis shows little overlap between conditions recorded using VA-only and Medicaid-only data. VA data tends to be more closely correlated with combined scores than Medicaid-only data, but neither is a reliable substitute for dual enrollees.

Target audience: This presentation may be useful for researchers and clinicians who study and care for Veterans who are eligible to receive care from both the VA and non-VA programs like Medicare or Medicaid.

Prediction of opioid-related overdose and suicide events using administrative healthcare data

HERC Health Economics Seminar

Wednesday, October 18 at 2pm ET

Register

Research Scientist, Health Equity and Rural Outreach Center of Innovation, Ralph H Johnson VA Medical Healthcare System, Charleston, SC
Research Associate Professor, Department of Public Health Sciences, Medical University of South Carolina, Charleston SC

Veterans suffer disproportionate health impacts from the opioid epidemic, including overdose, suicide, and death. Prediction models based on electronic medical record data can be powerful tools for identifying patients at greatest risk of such outcomes. In this study we developed an improved prediction model that built on existing work by incorporated advanced statistical methods, additional data sources and new predictor variables in a longitudinal setting. Our proposed model achieved an area under the ROC curve (AUC) of 84% and sensitivity of 71%. The model performed particularly well in identifying patients at risk for suicide related events, where 72% of actual events were accurately predicted among patients with the 100,000 highest risk scores. Some predictors in the proposed model have markedly different associations with overdose and suicide risks, which will allow clinicians to better target interventions to the most relevant risks.

Target audience: Statisticians interested in EHR-based prediction model development and clinicians interested in suicide and overdose prevention.

Spotlight

Including Medicare and Medicaid Cost Data in VA Research

VA patients often obtain additional care outside the VA system through Medicare, Medicaid, or other forms of insurance. This “dual use” is particularly common among Veterans eligible for Medicare. Therefore, researchers may want to include Medicare or Medicaid data in their analyses to gain a more complete picture of their cohort’s health care use and costs.

VA data users interested in learning more about including Medicare or Medicaid cost data in VA research can find more information in HERC Technical Report 42: Including Medicare Cost Data in VA Research and on the webpage Medicare and Medicaid Cost Data.

Below are common reasons VA researchers may include Medicare or Medicaid data in their analyses.

How VA Researchers can use Medicare and Medicaid Data

1. A more comprehensive understanding of Veteran health care utilization and costs

A common reason for including Medicare or Medicaid data is to gain a more comprehensive understanding of Veteran health care utilization and costs. Because VA-paid community care is now a substantial portion of the VA budget, many researchers include both VA-provided and VA-paid community care as well as Medicare or Medicaid data.

2. Conducting economic evaluations and cost-effectiveness analyses

Health systems are often interested in understanding the value of the care they provide relative to the cost of providing that care. Such analyses are critical for informing policy decisions regarding which interventions a health system should implement, which treatment should be administered, and decisions related to whether a health system like VA should provide in-house care or reimburse Veterans for care outside VA.

A comprehensive analysis that includes costs of VA care, VA-paid care in the community, and care received from other insurers such as Medicaid/Medicare for a Veteran cohort can significantly improve the relevance and scope of these analyses.

3. Understanding where Veterans choose to receive care for different services

VA researchers can use Medicare and Medicaid data to determine where dually eligible Veterans choose to receive care: either at commercial hospitals (paid for by Medicare or Medicaid) or VA. Understanding how this choice differs by type of service and factors influencing this decision (e.g., distance or wait time) can inform quality improvement efforts. It also has implications for allocating VA resources and VA’s decision about whether to continue providing a certain service at VA hospitals or to purchase care from outside VA (i.e., make vs buy decisions).

4. Understanding the impact of payer on quality of care, outcomes, and costs

Comparing quality, outcomes, and costs between systems can provide insights for VA quality improvement efforts and evidence to guide VA Community Care policies.

However, cost data in VA and Medicare aren’t directly comparable due to differences in how costs are defined, and the elements included in the calculations of costs. Before comparing costs between VA and Medicare data, researchers should take steps to make the data more comparable and understand that even after doing so, there will be limitations with these comparisons. More information is available on the pages Medicare and Medicaid Cost Data and Comparing VA vs. Non-VA Costs.

Because providers in traditional Medicare are compensated on a fee-for-service basis, while providers in VA are salaried, researchers can also use VA and Medicare data to examine how different pay structures may impact patient care.

5. Understanding prescription drug use across systems

Medicare-eligible Veterans often receive prescription drugs from both VA and Medicare. Therefore, combining these data can provide a more complete picture of prescription drug use, increase awareness of overlaps in prescribing, and highlight any prescription safety risks.¹

Access to Medicare and Medicaid Cost Data

Medicare and Medicaid data are available to VA data users through VIReC for research projects or the Medicare and Medicaid Analysis Center (MAC) for operations projects.

Information about requesting these data is available on the VA intranet (vaww).

• Operations data users: visit the MAC Data Request Process page (https://vaww.va.gov/MEDICAREANALYSIS/mac_data_requesting_process.asp)
• Research data users: Visit the VIReC VA/CMS Data Request Overview page (https://vaww.virec.research.va.gov/VACMS/Requests/Overview.htm)

¹Thorpe JM, Thorpe CT, Schleiden L, et al. Association Between Dual Use of Department of Veterans Affairs and Medicare Part D Drug Benefits and Potentially Unsafe Prescribing. JAMA Intern Med. 2019 Nov 1;179(11):1584-1586.

Data

Discharge Dataset with Inpatient Categories of Care

HERC creates a dataset identical to the MCA discharge NDE with additional information on cost and length of stay (LOS) subtotals for inpatient categories of care.

The MCA discharge NDE includes information on the entire span of an inpatient hospitalization. While it includes the discharge bed section, it doesn’t have details on other treating specialties used during an inpatient stay. The HERC Discharge Dataset includes additional fields containing cost and length of stay subtotals for 11 categories of care.

The HERC Discharge Dataset is useful to researchers interested in specific treating specialty segments of an inpatient stay (e.g., surgery or ICU). The dataset, available for VA research and operations access, has been updated for FY22. More information is available in the guidebook HERC's MCA Discharge Dataset with Subtotals for Inpatient Categories of Care.

Resources

Wage Index Updated for FY23

Health care costs are more expensive in geographic areas that have higher wages (e.g., Boston or San Francisco). Therefore, researchers may need to adjust their cost analyses for these wage differences. The best known method involves using the Medicare wage index. To assist VA researchers, HERC creates a wage index specific to VA facilities. The wage index, updated for FY2023 is saved as an excel file in the guidebook Medicare Wage Index for VA Facilities.

Data Q&A

Q: Where can I find a comprehensive set of ICD-10 codes for a specific condition such as heart failure? 

A: The Clinical Classification Software Refined (CCSR) for ICD-10-CM Diagnoses aggregates 1CD-10 codes into clinically meaningful categories. It is developed as part of the Healthcare Cost and Utilization Project (HCUP) sponsored by the Agency for Healthcare Research and Quality (AHRQ).

The CCSR includes a downloadable excel workbook with broad categories by body system, and within each of those broad categories, there are sub-categories, known as the CCSR categories, with greater granularities. These CCRS categories are mapped to all ICD-10-CM codes.

For example, the CCSR category for heart failure is CIR019, and this maps to 35 ICD-10-CM codes.

The Excel file with CCSR category to ICD_10-CM mapping is available for download on the HCUP website. The downloadable zip folder also includes a user guide and SAS program.