Technical Report 44: Practical Guide to Using MCA NDEs in VA Research and Operations

Suggested Citation

Illarmo S, Su P, Gehlert E, Wagner TH. Practical Guide to Using MCA NDEs in VA Research and Operations. Technical Report 44. Health Economics Resource Center, VA Palo Alto Health Care System, U.S. Department of Veterans Affairs. September 2024. 

 

For a list of VA acronyms, please visit the VA acronym checker on the VA intranet at http://vaww.va.gov/Acronyms/fulllist.cfm.

1. Overview

The VA Managerial Cost Accounting (MCA) program provides detailed cost information for all VA provided services, including the cost of all VA inpatient and outpatient encounters. The Managerial Cost Accounting Office (MCAO) creates National Data Extracts (MCA NDEs) to facilitate access to workload and cost data.

MCA data quick facts

• MCA data are the national source of cost information for VA.
• MCA uses activity-based costing to determine costs, which means that costs are based on the volume of services provided.
• MCA NDEs report the total cost of each encounter as well as cost-subtotals, divided at the department and product level.
• Costs in the MCA NDEs are divided into three categories: fixed direct costs, variable direct costs, and indirect costs. The sum of these three cost components will equal the total costs.
• The MCA NDEs do not include the cost of care that is provided by community providers, often referred to as community care.

This document focuses on practical guidance for using the MCA NDEs in a VA research or operations project. Detailed background on the MCA NDEs is available in the Research Guide to the Managerial Cost Accounting National Cost Extracts. We provide guidance and sample code for several common tasks, such as merging MCA data with VA encounter data. Sample code is available to VA data users in VA's Enterprise GitHub: https://github.ec.va.gov/COIN-HSR/HERC/blob/main/mca-cost-cohort-example.sas (VA intranet only).

return to top

2. Which MCA NDEs to Use for Your Project

There are several NDE datasets that relate to outpatient care, inpatient care, and clinical departments; the most useful NDEs for a project depends on the research question. A full list of NDEs is available on the Managerial Cost Accounting (MCA) page and Managerial Cost Accounting (MCA) National Data Extracts (NDEs) Technical Guide, available on the MCAO SharePoint site (VA intranet only: https://dvagov.sharepoint.com/sites/vhamcao/SitePages/National-Data-Extracts.aspx).

These are the core datasets that researchers will find useful for most projects:

Inpatient Cost Data: The two main NDEs for inpatient data are Discharge (DISCH) and Treating Specialty (TRT). DISCH has one record for each hospital discharge in each fiscal year. DISCH includes the entire cost of these stays, even if they began before the beginning of the fiscal year. TRT separates the inpatient stay into segments based on treating specialty (the type of unit where care was provided). Within one stay, TRT includes a separate record for each treating specialty per month.

The TRT NDE is the best source for researchers interested in detailed cost information for specific specialties. For those only interested in the total inpatient costs for a stay without details on the specific types of care provided, DISCH is a good option.

The DISCH file is limited in that it does not include subtotals if patients are transferred across bedsections / treating specialties.  HERC creates another file, known as the HERC MCA DISCH, that uses the TRT NDE to create the DISCH with subtotals. This file creates a standardized set of subtotals. If more precision is wanted on the subtotals, researchers can use the TRT and create their own. See section 7 below.

Outpatient Cost Data: The Outpatient NDEs (OUT, OUT2) include one record per clinic stop per day. All records with a cost are included in OUT. All $0 records are found in OUT2. We suggest researchers include both OUT and OUT2 to ensure you don’t miss any encounters.

Researchers interested in the total costs per fiscal year (VA inpatient + VA outpatient care) can sum costs from TRT and OUT.

Pharmacy Data: The Pharmacy (PHA) NDE provides detailed prescription drug information. It contains a single record for each pharmacy item and includes inpatient and outpatient drugs. The cost data from the PHA NDE are included when creating the inpatient (DISCH) and outpatient (OUT) datasets.

return to top

3. Common Analytic Tasks: Merging and Summarizing Cost and Utilization Data for a Cohort

Many projects will need to summarize cost and utilization data for a cohort. Summarizing these data at the patient level allows researchers to identify the effects of an intervention on VA costs and utilization and identify trends over time. In sections 3.1-3.4 we provide guidance for conducting these tasks using MCA data.

The sample code we link to in sections 3.3 and 3.4 was created for cost analyses conducted alongside clinical trials. If data users do not have a predetermined list of patient identifiers (SCRSSNs), then they will first need to build their cohort file. The code, which is available in VA's Enterprise GitHub, is from the study Can Service Dogs Improve Activity and Quality of Life in Veterans With PTSD?. The full results of the cost analysis is available in chapter 1 of the monograph, “The Economic Impact and Cost Effectiveness of Service Dogs for Veterans with Post Traumatic Stress Disorder.” Summarized results are also available in the paper Therapeutic and Economic Benefits of Service Dogs Versus Emotional Support Dogs for Veterans With PTSD.

We provide this code as a launching point to help data users develop their analysis plans. Please check all code before beginning analyses.

return to top

3.1. Extracting MCA Data and VA Encounter Data for a Cohort

The MCA NDEs do not contain detailed clinical information such as procedures and diagnoses; therefore, researchers will need to extract the cost and utilization data separately and then merge the data files to crate their analytic dataset. Researchers should extract all encounters for their cohort using SCRSSN for the study time frame.

Sample code for extracting MCA OUT/OUT2 data and MedSAS data for a cohort is available in VA's Enterprise Github: https://github.ec.va.gov/COIN-HSR/HERC/blob/main/mca-cost-cohort-example.sas (VA intranet only). 

return to top

3.2 Merging MCA Data with VA Encounter Data

As mentioned in section 3.1, researchers looking for cost and utilization data for a cohort will need to merge MCA data to VA encounter data. In VA's Enterprise Github, we provide sample code for merging MCA data to MedSAS data: https://github.ec.va.gov/COIN-HSR/HERC/blob/main/mca-cost-cohort-example.sas (VA intranet only). 

Removing Duplicates

Before merging, data users should remove any duplicate records. To remove any duplicates, you need to understand how the file is populated and whether duplicates are allowed. In our example using MCA and MedSAS, duplicate records are those within the same file that have the same person (SCRSSN), location (STA3N), and date (ADMITDAY/DISDAY, TXSPSDT/TXSPEDT/TXSP, or VIZDAY). For more detail see the HERC MCA guidebook Sections 7.1.3 (Deletion of duplicates) and 7.2.5 (Duplicated records).

Merging Files

The variables you use to merge MCA and MedSAS data may depend on your research needs. MCA and MedSAS should always be merged on scrambled social security number (SCRSSN). This may be sufficient if your goal is to identify costs and utilization over a given time period. If you want to match specific visit information (i.e., you want to understand costs and utilization for a visit on a given day), then MCA and MedSAS data can be merged on the following variables: (1) SCRSSN, (2) medical center identification number, and (3) visit date (outpatient) or admission date and discharge date (inpatient). Before merging, sort both files by SCRSSN. Otherwise, errors will occur.

Choosing a source of VA outpatient encounter data:

VA has several sources of outpatient encounter data: MedSAS, CDW, and CDW OMOP. While merging MedSAS and MCA is a relatively straightforward task, merging CDW to MCA is difficult due to different “unit” levels in the data. If researchers are using CDW outpatient data, we recommend using CDW OMOP, since the data has been transformed into a common model. Inpatient (TRT) and Outpatient (OUT) NDEs are available in OMOP. 

return to top

3.3 Summarizing Cost and Utilization Data into 30-day Periods

Most studies that follow individuals over time will want to create patient level summaries per period of time. For example, a clinical trial that follows people after randomization will need to track these people over time. The period of time should be standardized. We recommend using 30 days for the period of time. The main challenge is separating inpatient stays that span the time periods. These inpatient costs should be prorated based on length of stay in each time period. If data in the 30-day periods are noisy, the time periods can be easily summed together to get 90-day (quarterly) or 360-day (yearly) estimates. If you start with 360-day estimates, you cannot easily create 30-day periods.

In the Service Dogs for PTSD study, we included all VA cost and utilization data for the 540 days (i.e., 18 months) before and after participants were paired with a dog. We summarized the data into 30-day periods by creating a long dataset with one value per person per thirty-day period for each variable. The variables you include will depend on your study question. We included variables such as number of unique days with an outpatient visit, number of inpatient admissions, number of inpatient days, and total costs for inpatient, outpatient, and for each category of care. Sample code is available in VA's Enterprise GitHub https://github.ec.va.gov/COIN-HSR/HERC/blob/main/mca-cost-cohort-example.sas (VA intranet only). 

return to top

3.4 Summarizing Cost and Utilization Data into Categories of Care

Researchers may want to group encounters into clinical categories of care to compare cost and utilization by cohorts or treatment. In the Service Dogs for PTSD study, we grouped cost and utilization data into mutually exclusive categories of care using inpatient Treating Specialty or outpatient Clinic Stop codes available in the TRT and OUT/OUT2 MCA NDEs. We used the following categories: Inpatient Medical/Surgical, Inpatient Psychiatry/Mental Health, Inpatient Substance Use Treatment, Inpatient Other, Outpatient Medical/Surgical, Outpatient Pharmacy, Outpatient Psychiatry/Mental Health, Outpatient Substance Use Treatment, and Outpatient Other. A list of codes by category of care is available in Appendix B. Sample code is available in VA's Enterprise GitHub: https://github.ec.va.gov/COIN-HSR/HERC/blob/main/mca-cost-cohort-example.sas (VA intranet only). Additional notes to provide context for the sample code are described in Appendix A. 

return to top

Appendix A: Sample SAS code

Notes for the code for summarizing data into 30-day periods and categories of care:

• This code assumes that data users have already identified a list of SCRSSNs for their study cohort.
• Costs were adjusted to the 2018 general consumer price index from the Bureau of Labor Statistics.
• We included the following variables for each 30 day-period in our analytic dataset:
  • Outpatient: Number of unique days with an outpatient visit; Outpatient total costs; Outpatient variable total costs; Admissions, days and dollars for inpatient medical/surgical, psychiatry and substance use care; Number of visit days dollars for outpatient medical/surgical, psychiatry, substance use care; Outpatient pharmacy costs; All other outpatient costs.
  • Inpatient: Number of inpatient admissions; Number of inpatient days; Inpatient total costs; Inpatient variable total costs;  Admissions, days and dollars for medical surgical, psychiatry, substance use care, and all other.

return to top

Appendix B: Categories of Care

return to top

Acknowledgements

We gratefully acknowledge Juliette Hong and Sharon Dally for their contributions to and review of the sample SAS code. 

Last updated: September 13, 2024