Integrated Veteran Care Consolidated Data Sets (IVC CDS) Cost Data
The VHA Office of Integrated Veteran Care Consolidated Data Sets (IVC CDS) is the primary source for VA community care data. IVC CDS contains referrals, claims, and eventually payment data from Community Care Reimbursement System (CCRS), Electronic Claims Adjudication Management System (eCAMS), Fee Basis Claims System (FBCS), Plexis Claims Manager (PCM), and VistA Fee Basis Package (Fee). Data are available for claims processed on or after October 1, 2018.
On this page, we provide an overview of cost data in CDS, and we present one method for identifying the cost of an inpatient hospital stay using CDS data. Investigations into outpatient cost data are underway.
Learn more:
- IVC CDS Data Review from VIReC (VA intranet only): https://vaww.virec.research.va.gov/Reports/DR/DR-IVC-CDS.pdf
- IVC CDS Data Dictionary from VIReC (Available on the VHA Data Portal; VA intranet only): https://vaww.vhadataportal.med.va.gov/Data-Sources/IVC-Consolidated-Data-Sets
- IVC Consolidated Data Sets on the VHA Data Portal (VA intranet only): https://vaww.vhadataportal.med.va.gov/Data-Sources/IVC-Consolidated-Data-Sets
- IVC CDS Data Review: Referrals from VIReC (VA intranet only): https://vaww.virec.research.va.gov/Reports/DR/DR-IVC-CDS-Referrals.pdf
- Finding Hospital Inpatient Stays in CDS from ACCENT
- Information on Specific Community Care Programs and Policies from ACCENT includes useful information on IVC CDS data including SQL scripts and cyberseminars.
- A Brief Introduction to a New Community Care Data Source: IVC_CDS (VA intranet only): https://vincicentral.vinci.med.va.gov/Shared%20Documents/Videos/A_Brief_Introduction_to_a_New%20Community_Care_Data_Source-IVC_CDS-092023.mp4
August 2024 update
Based on input from IVC, we have updated guidance with two changes: (1) We now recommend filtering IsCurrent=Y. IVC has informed us that claim total amounts are not cumulative for reprocessed claims, and the most current version of a claim should represent the full amount paid on that claim. (2) We removed the 3-day payment window lookback period to be consistent with other cost data and given the unknowns in the complicated billing process.
Cost Data Overview
Community care data include claims-level and other information about care submitted for reimbursement by a non-VA facility. All claims will have a header record and associated line detail records. The header record contains the total amount VA paid on the claim while the line records contain payments for specific services.
The granularity you need will depend on your research question, but for most research questions, we recommend using header level data. HERC researchers have found that payments at the header and line levels match in approximately 98% of claims, and the agreement between header and line level claim payments has increased over time. We’ve found that a large portion of the mismatched header and line claim payments are sourced from eCAMS and PCM and suggest that data users look out for claims from eCAMS with $0 payments at the line level, but payments greater than $0 at the header level.
Claims can be submitted using three different forms: institutional, professional, and dental. Institutional claims can be for inpatient or outpatient care and are typically submitted by hospitals and other healthcare organizations. Professional claims can also be for care delivered in an outpatient, inpatient, or other setting and are submitted by physicians and other healthcare providers. The information included in a claim varies depending on the type of claim form.
Many inpatient and outpatient services contain both an institutional component and a professional component, and researchers will need to sum these to calculate the total cost of the service. In addition, researchers interested in inpatient care specifically should be aware that multiple claims may need to be aggregated to capture the total cost of the inpatient stay.
Key Variables for Cost Analyses
Below are key variables for conducting cost analyses using CDS data. A full list of variables is available in the IVC CDS Data Dictionary, available on the VHA Data Portal (VA intranet only: https://vaww.vhadataportal.med.va.gov/Data-Sources/IVC-Consolidated-Data-Sets).
| Original_ClaimID | The original number assigned to the first submission of the claim. Original_ClaimID can be used to identify resubmitted claims within a source system. |
| Source_Key | The highest number assigned to the most recent claim within an Original_ClaimID. |
| Claim_Status_ID | Indicates whether a claim was paid. |
| IsCurrent | This flag indicates whether this record is the most current version of the claim. We recommend restricting analyses to IsCurrent=Y. We have identified select cases where ISCurrent=Y would not yield the most recent paid claims; however, this applies to a very small number of claims (less than 0.01%), mostly from eCAMS. Researchers needing to capture these claims may consider limiting records to IsCurrent not in (‘R’,’D’) (reversed or deleted) and taking additional steps to address duplicates. |
*Claims may be resubmitted (see Duplicate Claims). Therefore, for many inquiries, researchers will need to identify the most recent submission of a claim.
| Claim_Total_Amount | The total amount paid by VA. CDS contains two other cost variables: Amount_Allowed and Total_Charges. In many (but not all) cases, Amount_Allowed will be equal to Claim_Total_Amount. In cases where the values are not equal, HERC researchers have found that Amount_Allowed does not include adjustments to the final paid amount. We do not recommend using Total_Charges as these reflect charges not payments. |
| Service_Start_Date and Service_End_Date |
We recommend using Service_Start_Date and Service_End_Date from the claim header file to identify an inpatient stay’s window given that admission and discharge date variables are often missing. |
| Claim_Form_Type | This variable will allow you to distinguish between the institutional and professional components of an inpatient stay. HERC analysts have found some institutional outpatient community care claims from VistA (Fee) appear in CDS as claim_form_type=P; therefore, investigations into outpatient care should use Claim_Form_Type with caution. |
| Place_of_Service_ID | 2-digit code indicating where the service was performed for professional claims. |
| Bill_Type | Code that classifies the type of facility and type of care for institutional claims. Although the code is 4 digits, the first digit (0) is excluded, and we usually ignore the final digit for purposes of identifying type of care. |
| Billing_Provider_TaxID | This variable enables data users to identify the same billing provider across records. It can be used as a facility identifier in cost analyses. Billing_Provider_NPI and Service_Provider_NPI are other variables that can be used as facility identifiers; however, HERC analysts have found some missingness in both variables. Missingness varies by year, claim form type, and source system, with missingness by year related to source system. Billing_Provider_NPI is rarely missing in CCRS, while it is almost always missing in PCM and often missing in VistA (Fee). As the distribution of source system changes over time with the proportion of CCRS claims increasing, Billing_Provider_NPI missingness decreases, with missingness dropping to 0.2% by 2023. |
| Source system | Variable that indicates the claim’s source system (eCAMs, VistA, FBCS, etc). This variable can be used to de-duplicate claims. |
Duplicate Claims
Data users may find duplicate claims in the data due to resubmission of claims and claims from multiple source systems. Limiting to records for a unique patient admission where Claim_Status_ID in (‘71’) (Paid) and IsCurrent in (‘Y’) will eliminate most duplicate claims. However, you may still find duplicates after applying these filters. Therefore, in addition to the above filters, data users should look for and remove duplicate claims.
Within inpatient data, one method for removing duplicates within a source system is by using header level Patient_ICN + Original_ClaimID, and between source systems by using header level Patient_ICN + minimum of Service_Start_Date + maximum of Service_End_date and Bill_Type (institutional records) or Place_of_Service_ID (professional records), for claims with the same Claim_Total_Amount (see the section “Identifying the Cost of an Inpatient Hospital Stay“). The maximum and minimum service dates are aggregated across contiguous claims to form inpatient stays and serve as the admit and discharge dates. (We do not use the Admission_Date and Discharge_Date fields from CDS because of missingness.)
Users should always de-duplicate when the same claim appears in both FBCS and VistA (Fee) source systems; we recommend keeping claims from FBCS since they're generally more complete. Users may also encounter claims that appear to be duplicates originating from other systems or within the same system. We offer suggested solutions within our code to deal with these types of duplicates, but we leave the final decision to de-duplicate these claims to the individual researcher as we’re unable to definitively determine if duplicate sets of claims were actually paid (perhaps erroneously).
The number of duplicate claims are greatest in 2019 data and decrease over the years. While HERC analysts examining line level claims have identified approximately 30% of claims as potential duplicates in FY19, this drops to 5% in FY20, 0.07% in FY21, and <0.01% in FY22 and onwards. The large number of duplicates in FY19 is primarily due to duplication between FBCS and VistA (Fee).
Learn more about the reasons for duplicate claims in VIReC’s IVC CDS Data Review (VA intranet only: https://vaww.virec.research.va.gov/Reports/DR/DR-IVC-CDS.pdf).
Time to Claim Processing
Data users should allow enough time for all claims to be processed before pulling cost data. IVC recommends allowing at least a 4-month processing time from the last date of service before including claims in a cost analysis.
Identifying the Total Cost of an Inpatient Stay
In this section, we share an algorithm for identifying the cost of an acute inpatient hospital stay and sample SQL code. This is only one method for identifying inpatient costs and is meant to serve as a starting point for data users conducting their own evaluations.
This method uses the CDS_Claim_Header file. We use Billing_Provider_TaxID as the facility identifier; therefore, without de-duplicating, this method may lead to overcounting of stays. Data users should de-duplicate to address potential overcounting.
Disclaimer 1: This method applies to inpatient cost data only as analytic decisions made for inpatient care may not be applicable to outpatient care. For example, some VistA (Fee) outpatient institutional claims have been incorrectly categorized as professional claims in CDS. As of this writing, we recommend de-duplicating outpatient CDS records at the line level.
Disclaimer 2: These algorithms were developed using a cohort that was not randomly selected. Validation of these algorithms using a larger cohort is underway. We encourage those using this algorithm to check their data after each step for any erroneous or unexpected values.
Disclaimer 3: In the sample SQL code, we include billing_provider_npi and service_provider_npi as additional variables to help distinguish facilities. However, as described in the section, Key Variables for Cost Analyses, there is missingness that varies by year, claim form type, and source system. The importance of this missingness depends on the research question and cohort.
Steps for Identifying the Total Cost of an Inpatient Stay
- Allow for sufficient claim processing lead time, e.g., 4 months.
1. Identify the Institutional Component of the Stay
- Filter header records by Claim_Form_Type = I (institutional), IsCurrent = Y, Bill_Type 11x, 12x, 41x, 42x, 44x*, and Claim_Status_ID = 71 (PAID).
- Create a new variable: AdmitID. Define AdmitID as unique admission: Patient_ICN + Billing_Provider_TaxID + [Contiguous Service Dates].
- With unique AdmitID, calculate Service_Start_Date_Min and Service_End_Date_Max (Minimum of Service_Start_Date and Maximum of Service_End_Date).
- De-duplicate claims with the same Claim_Total_Amount and within the same source system using Patient_ICN + Original_ClaimID.
- De-duplicate FBCS and VISTA claims with the same Claim_Total_Amount using Patient_ICN + Service_Start_Date_Min + Service_End_Date_Max + Bill_Type
- With AdmitID, roll-up Claim_Total_Amount by Bill_Type x Source_System.
- For each AdmitID, sum Bill_Type x Source_System … this is the Institutional Component.
2. Identify the Professional Component of the Stay
- Within AdmitID, roll-up Claim_Total_Amount by Place_of_Service_ID x Source_System x Service_Start_Date with Claim_Status_ID = 71 (PAID) and IsCurrent = Y and Claim_Form_Type = P (Professional) and Place_of_Service_ID = 21**
- Link Patient_ICN + Service_Start_Date to the get the AdmitID (SQL join with Patient_ICN + Service_Start_Date_Min + Service_End_Date_Max)
- De-duplicate claims with the same Claim_Total_Amount and within the same source system using Patient_ICN + Original_ClaimID.
- De-duplicate FBCS and VISTA claims with the same Claim_Total_Amount using Patient_ICN + Service_Start_Date_Min + Service_End_Date_Max + Place_of_Service_ID
- Collapse Claim_Total_Amount by AdmitID x Source_System … select the sum as the Professional Component.
3. Calculate Total Costs
- Match the inpatient component to the professional component using Service_Start_Date_Min + Service_End_Date_Max
- Sum Claim_Total_Amount for the Institutional Component + Professional Component to calculate Hospital Stay Total Cost.
*The bill types researchers include may vary depending on their research question. A full list of bill type codes is available from ResDAC.
**Researchers may choose to add additional place of service codes such as ambulance or psychiatric residential treatment center depending on their study goals. A full list of place of service codes is available on the CMS website.
***This algorithm does not take any additional steps to account for Medicare’s 3-day payment window. Researchers interested in addressing this window may need to take additional steps when calculating the professional component of a stay.
SQL Code for Identifying the Total Costs of an Inpatient Stay
Sample SQL code for identifying the total costs of an acute inpatient stay is available within VA's Enterprise GitHub: github.ec.va.gov/COIN-HSR/HERC/blob/main/HERC-sop-cost-inpatient-hospital-stay.sql (VA intranet only).
The code is based on the steps outlined above. This is only one method and is meant to serve as a starting point for data users conducting their own evaluations.
Acknowledgements
We gratefully acknowledge Erin Beilstein-wedel from Access and Community Care Engagement Network Team (ACCENT) for her contributions and review of the contents of this page. We also thank the VHA Office of Integrated Veteran Care for their review.
Last updated: August 29, 2024
