Identifying Emergency Department Visits in Medicare Claims Data

Introduction

Emergency department (ED) utilization serves as a foundational metric across health services research. Because ED visits are high-cost and high-intensity, the identification of these events is critical for measurement and evaluation of healthcare quality(1). Health researchers rely heavily on administrative datasets like Medicare claims to do this work. Because there are multiple conceptual frameworks for measuring utilization within administrative datasets, navigating these records requires careful methodological consideration.

This article explores the promises and pitfalls of various ED identification methods within Medicare administrative data with a focus on fee-for-service (FFS) Medicare. Specifically, I outline how federal policy, institutional Medicare billing requirements, and data file structure impact the calculation of facility-based ED metrics. I also provide guidance for ED identification in the Medicare Provider Analysis and Review (MEDPAR) file, I discuss considerations for including professional ED claims in measurements, and I talk about measuring ED in Medicare Advantage Encounter data.

Methods

Fee-For-Service Medicare Institutional Claims Files

The Centers for Medicare and Medicaid Services (CMS) offers Medicare claim files to researchers in the form of Limited Data Set (LDS) Standard Analytic Files (SAFs) and Research Identifiable files (RIFs)(2). Hospital claims containing institutional/facility ED charges are found in Inpatient (IP) and Outpatient (OP) Medicare SAFs and RIFs.

FFS Medicare ED billing guidelines for institutional/facility providers specify the following:

• OP services are treated as IP services when incurred up to three days before admission (3-day payment window policy) with some exceptions, such as Critical Access Hospitals (CAHs)(3,4).
• ED facility services are reported under a 045x revenue center code(5).
• Hospitals may bundle an ED professional fee component (revenue code 0981) into an institutional claim, for instance when the ED physician is a hospital employee or the hospital operates under an all-inclusive billing model(6,7).

Based on these CMS billing guidelines, the following “all inclusive” method was determined for capturing ED utilization in FFS Medicare claims:

Revenue Center Codes 0450-0459 (Emergency room) and 0981 (Professional Fees-Emergency room) identify ED claims in FFS Medicare Outpatient and Inpatient claim files.

Note that charges for one ED visit will be found in either Outpatient or Inpatient claims files - a visit should not generate facility ED charges on both an inpatient and outpatient claim(4,8,9).

When capturing hospital admissions preceded by an ED visit, it is important to understand that looking through IP claims alone is not sufficient - OP claims must also be searched. Examples of why this is the case include patients who were transferred to a different hospital for admission, and care that occurs at certain types of hospitals required to bill ED separately from IP (e.g., Critical Access Hospitals). To determine if an OP ED visit led to admission requires joining the OP ED record to the IP file using key variables.

Table 1 below provides a summary of IP claims preceded by an ED visit based on a search of both IP and OP claims. The first category - “ED” - counts IP claims that contain at least one ED revenue center code; “Overlaps with OP ED” measures IP claims that do not contain an ED revenue code but link to an OP claim that does; And “No ED” counts IP claims with no associated ED visit.

The method for determining “Overlaps with OP ED” was by identifying OP claims having at least one ED revenue code and joining them to the IP file. The join was based on matching BENE_ID and overlapping service dates. Specifically, when the OP “claim through date” fell within the time span of three days prior to the IP admit date (or “claim from date” when admit date was missing) and the IP discharge date (or “claim through date” when discharge date is missing), the IP and OP claim were categorized as being associated. The choice of three days was informed by the 3-day payment window policy. One limitation of this method is the assumption that any OP ED visit occurring within three days of the IP admit was medically related to the IP admit, which may lead to an overcount of admit-related ED visits.

Table 1. Proportion of IP Claims with an Associated ED Visit, 2023

_{*ED defined as a claim having at least one ED revenue center code 0450-0459 or 0981. 
**For IP claims that do not contain an ED revenue code.} 
Table 1 caption: Proportion of 2023 IP claims with an ED revenue code or overlaps with an OP claim with an ED revenue code.

Table 1 shows that most IP claims for stays originating in the ED contain an ED revenue center code (91.5%). This reflects CMS billing rules that require most hospitals to adhere to the 3-day payment window policy and bundle ED services with the IP claim.

There are exceptions. A notable one is Critical Access Hospitals (CAHs). The CMS Billing Manual says that CAHs must not bundle ED services on an IP claim(3). There are nuances to the rules dependent upon CAH ownership, meaning a subset of CAH ED claims will be subject to the payment window policy(3). However, as shown in Table 2, most CAHs bill ED services on an OP claim (92.8%).

Table 2. Proportion of Critical Access Hospital (CAH) IP claims with an associated ED visit, 2023

_{*ED defined as having at least one revenue center code 0450-0459 or 0981. 
**For IP claims that do not contain an ED revenue code.} 
Table 2 caption: Proportion of 2023 Critical Access Hospitals IP claims that have an ED code or that overlap with an OP ED claim.

Given the complexity of IP billing rules, it is advisable to check both IP and OP claims for evidence of ED utilization related to an IP stay. This exercise will also help the researchers understand ED billing patterns for their cohort.

A note on OP dates of service: the date of service on the ED claim line is the date the patient entered the ED even if the visit spanned multiple days. For all other services related to the ED visit (e.g., lab, radiology) the line-item date of service is the date the service was actually rendered (5).

MedPAR File

The Medicare Provider Analysis and Review (MedPAR) dataset is an administrative database maintained by CMS. It consolidates individual inpatient claims and “informational records” into summarized stay-level records from Medicare covered hospital admissions. Records with the same beneficiary id, admit date and provider number are consolidated into stays(10).

The MedPAR data dictionary describes the Emergency Room Charge Amount field as, “The charge amount (rounded to whole dollars) for emergency room services provided during the beneficiary's stay.” Based on this description, one can infer that ED-associated stays in the MedPAR file are identified when the Emergency Room Charge Amount field is > $0 (8).

Table 3 shows that the total number of MedPAR hospital stays (STAY_INDICATOR = ‘S’ or ‘L’) associated with an Emergency Room Charge in 2023 is 72.6%.

Table 3. Total Number of MedPAR Hospital Stays with ED Charge Amounts >0, 2023

Table 3 caption: Total number and percentage of MedPAR hospital stays with ED charge amounts >0, 2023.

There are known differences between the data and methods that may explain the difference in percentage of ED in Tables 1 and 3:

1. The MedPAR includes both FFS Medicare and Medicare Advantage stays.
2. The data in the MedPAR are based only on IP claims/records – OP is not included.
3. Revenue code 0981 is not included in the MedPAR Emergency Room Charge Amount algorithm (a very small number of FFS IP claims contain this code (<0.02%)).
4. The MedPAR consolidates claims into stays (multiple IP claims were submitted for approximately 0.03% of FFS IP stays in 2023 based on matching BENE_ID CLM_ADMSN_DT and PRVDR_NUM).

What About Professional Claims?

and professional claims. One example is the Healthcare Data Effectiveness Data Set (HEDIS)(11) used to measure Medicare Advantage health plan performance. HEDIS includes an emergency department utilization (EDU) measure whose methodology identifies ED on a professional claim using HCPCS codes and/or place of service (POS) codes. HEDIS counts ED claims incurred on the same day as a single visit.

Based on a “back of the envelope” analysis of 2019 FFS Medicare ED visits using the 2015 HEDIS definition, 4.6% of professional ED claims had no associated ED facility claim when joined by beneficiary ID and exact service date. Researchers can consider including professional ED claims that lack a date-matched facility claim in their study; however, it is difficult to understand why a facility would fail to submit a claim for an ED visit. Furthermore, some claim elements like place of service code that do not impact payment tend to be less reliable than those that do. Researchers may do well to dig into stand-alone professional claims to evaluate their characteristics or conduct sensitivity analyses with and without them.

What About Medicare Advantage Encounter Data?

CMS refers to Medicare Advantage utilization records as encounters. CMS disseminates encounter data separately from FFS in the form of Encounter RIFs (they are not available in LDS SAF format). When measuring ED utilization in Encounter data, applying FFS methods is a good place to start; however, Medicare Advantage Organizations (MAOs) are not beholden to Medicare billing rules, therefore there is no guarantee that all MAOs' will follow those rules. One can evaluate Encounter results (e.g., ED visits/1,000) generated using FFS-based methods and check for outliers by contract and/or plan (contract ID+ plan ID). Plans with zero or low counts may not be following FFS billing rules, or the data they sent to CMS may be incomplete.

Conclusion

Accurate measurement of ED utilization requires a comprehensive approach that checks for ED revenue center codes in both inpatient and outpatient files to capture complex institutional billing nuances (such as CAH exemptions). Validating standalone professional ED claims, and Medicare Advantage encounter records by contract and plan when measuring ED utilization are recommended.