Identifying Claim Denials in Medicare Fee-for-Service Research Identifiable Files

Acknowledgement
The authors would like to thank Matt McFalls for their valuable contribution to this work.
Abstract

Medicare fee-for-service (FFS) research identifiable files (RIFs) include both approved and denied claims, as well as claims that are denied in part. Understanding how to identify claims that are denied in whole or in part is key to selecting claims for an analysis.

Background

Medicare fee-for-service (FFS) research identifiable files (RIFs) include both approved and denied claims as well as claims that are denied in part. Claim denials are common, surprisingly complicated, and important for researchers to consider. The choice to include or exclude them could impact study inference(1). Understanding how to identify claims that are denied in whole or in part is key to selecting claims for an analysis.

There is no single way to identify denials for all claim types, which is a byproduct of the complexity of medical billing. The methods sections of many scholarly articles say they studied or excluded denied claims but do not explain in detail how they were identified(2-4). Billing specifications for claim denials are scattered across documents like the Medicare Claims Processing Manual. The purpose of this article is to serve as a resource that captures the specifications for identifying whole and partial claim denials across FFS claim RIF types.

Here are common reasons entire claims are denied:

  • Exact duplicate claim
  • Benefit was exhausted
  • Claim is submitted for someone who is not Medicare enrolled
  • Claim is for someone enrolled in Medicare managed care
  • Claim is for someone who also has private comprehensive healthcare insurance as well as Medicare, but that insurer wasn’t billed first.
  • Another insurer such as a liability carrier (e.g., auto) insurer is responsible for the care.

Here are common reasons parts of claims are denied:

  • A particular service is not Medicare covered but other services also on the claim are Medicare covered. For example, a visit with an eye practitioner where a glaucoma test is covered but the eyeglass prescription is not,
  • The denied part is contained within a larger component that is already approved. This is sometimes called ‘bundling.’ For example, payments for most surgical procedures include reimbursement for all routine pre- and post-operative care provided by the same provider. If a claim separately bills for the surgery and pre-operative care by the same provider, the pre-operative component will be denied because it is included in the surgery payment.

It should be noted that RIFs contain final action claims only - they do not include adjustments or claim history. For example, a claim that was initially denied, resubmitted, then approved would only appear in the RIF as an approved claim. Also note that denied claims are distinct from rejected claims. A claim is rejected by the Medicare Administrative Contractor (MAC) when it is deemed unprocessable, for example when it contains invalid or missing codes. The RIFs do not contain rejected claims.

Claim denials are specific to FFS Medicare data. There are no denials per-se in the Part D files. The Part D data are in the form of events, which are not the same as claims but rather summary records that represent the final status of a drug fill after all claim adjustments, resubmissions, and deletions have been applied.

It is important to be aware of and know how to identify denied claims because they can impact inference. For example, when claims are denied as duplicates, including them would result in overcounts, which could bias estimates made from using those claims.

Methodology

How to identify claims that were totally denied

Denied/non-covered/non-paid in FFS claims data can be identified using claim-level codes:

FFS Institutional claims-- Inpatient, Outpatient, SNF, HHA, Hospice claim denials: Claim Medicare Nonpayment Reason Code (CLM_MDCR_NON_PMT_RSN_CD) will have a value reported. It will not be blank.

Table 1. Percent of no payment / non-covered claims by institutional file type (2019)

File

Percent (%)

Inpatient

3.10

Outpatient

1.23

Skilled Nursing Facility

12.78

Home Health

1.27

Hospice

0.57

For virtually all non-paid claims in Inpatient and SNF files, the Non-Covered Charge Amount (NCH_IP_NCVRD_CHRG_AMT) will equal the Claim Total Charge Amount (CLM_TOT_CHRG_AMT).

MedPAR files contain rolled up institutional claims that summarize inpatient or SNF care. The records are not individual claims like in the FFS Inpatient file and do not contain claim processing variables like those described above. Also, a large portion of the MedPAR consists of “information only” records for services covered by Managed Care Organizations for which information related to denials (e.g., allowed amount) is unknowable.

FFS non-institutional claims--the Carrier and DME files contain claims that were completely denied. These are identified by a Carrier Claim Payment Denial Code CARR_CLM_PMT_DNL_CD) of “0” or “D”.

Table 2. Percent of denied claims by non-institutional file type (2019)

File

Percent (%)

Carrier

5.28

DME

11.32

Determining reason for claim denial

Determining why an entire claim was denied is considerably more challenging than determining that it was denied. For example, while denied institutional claims include a non-payment reason code, the majority (over 80%) have a code value of N- All other reasons for non-payment.

In the institutional files, the Condition Code, Occurrence Code and Span Code files (linkable to the base file by BENE_ID and CLM_ID) sometimes contain additional information about claim denials:

  • The claim related condition code (CLM_RLT_COND_CD) located in the Condition Code file may indicate why a claim was submitted to Medicare even though denial was expected, or why it was not covered (e.g., codes 20, 21, 77, etc.). These may indicate something was medically unnecessary, but billing Medicare was required for secondary insurance purposes, for example.
  • The claim related occurrence code (CLM_RLT_OCRNC_CD) located in the Occurrence Code file may provide detail about why all or part of a claim was not covered by Medicare (e.g., codes 21, 22, 31, 32; there may be others). These often indicate the claim was for a non-covered service.
  • The claim occurrence span code (CLM_SPAN_CD) located in the Span Code file may provide detail on specific non-covered spans of care specified by the claim occurrence span from date (CLM_SPAN_FROM_DT) and claim occurrence span through date (CLM_SPAN_THRU_DT). Reasons services may be denied noted here include exhausted benefits and provider liability.

The MedPAR file doesn’t contain information about the reason for non-payment.

In the non-institutional files, Carrier Claim Payment Denial Codes (CARR_CLM_PMT_DNL_CD) E-Y and 00-43 (available as of 2011) indicate that Medicare does not have primary responsibility for the claim but may be the secondary payer (Medicare Secondary Payer (MSP) cost avoided claims). The situations surrounding MSP claims can be quite complex.

How to identify claims that were partially denied

As with fully denied claims, the method for determining partial denial varies between institutional and non-institutional files.

While technically there may be some partial denials for Inpatient claims, it is important to remember that inpatient stays are paid using a prospective payment system. Thus, individual revenue center codes will not be approved or denied; Revenue Center Non-Covered Charge Amount (REV_CNTR_NCVRD_CHRG_AMT) = Revenue Center Total Charge Amount (REV_CNTR_TOT_CHRG_AMT) in the inpatient revenue center file generally happens in the context of fully denied claims.

For other FFS institutional claims that are paid by the revenue center (outpatient, HHA, hospice), partial denials will be identified by Revenue Center Total Charge Amount > 0 and Revenue Center Non-Covered Charge Amount = Revenue Center Total Charge Amount in the revenue center file.

Some revenue centers have a Revenue Center Total Charge Amount = 0. For example, there are special revenue center codes used in association with HIPPS codes where the charge amount is zero. In these cases, the HIPPS code is used by the claims processing system to determine reimbursement. One revenue center code is set for each Medicare prospective payment system that uses HIPPS codes. For example, skilled nursing uses revenue center code ‘0022’, and home health uses code ‘0023’.

Table 3. Percent of denied revenue centers by institutional file type (2019)

Revenue Center File

% Denied

% No charge

Inpatient

2.37

0.24

Outpatient

4.45

0.03

Skilled Nursing Facility

3.55

17.38

Home Health

1.16

5.02

Hospice

0.57

0.00

FFS non-institutional files--the Carrier and DME: denied lines can be identified by Line Submitted Charge Amount (LINE_SBMTD_CHRG_AMT) > 0 and Line Processing Indicator Code (LINE_PRCSG_IND_CD) not = ‘A’ (allowed), ‘R (reprocessed)’, or ‘S’ (secondary) in the line file. Some lines have Line Submitted Charge Amount = 0. For example, CPT codes used for quality measures (e.g., G8427), patient history (e.g., 1036F) and assessments (e.g., G8730) are common Carrier zero charge lines.

All non-institutional denied lines have a Line Allowed Charge Amount (LINE_ALOWD_CHRG_AMT) = 0 and all lines that = 0 are denied.

Table 4. Percent of denied claim lines by non-institutional file type (2019)

Line File

% Denied

% No charge

Carrier

11.31

6.13

DME

12.73

0.00

Table 5. Most common Line Processing Indicator Codes indicating a denied/noncovered/no pay line (2019)

Line Processing Indicator Code

Carrier Lines %

DME Lines %

O - Other

54.08

35.57

C - Noncovered care

33.98

41.91

M - Multiple submittal—duplicate line item

4.47

2.12

N - Medically unnecessary

3.51

19.34

All others

3.95

1.06

Total

100.00

100.00

Resources

Medicare manuals and codebooks are excellent sources of detailed information. Because there is significant variation by provider type and setting about how denials are handled, the manuals can help researchers understand how denials/non-covered/non-paid claims occur within different files.  In general, variables and methods are largely consistent within Institutional claim types (Inpatient, Outpatient, Skilled Nursing Facility (SNF), Home Health and Hospice) and within Non-institutional claim types (Carrier and DME).

Medicare Claims Processing Manual
Medicare Fee-for-Service (FFS) Claims Codebook 
Medicare Secondary Payment (MSP) Manual 
CCW Technical Guidance: Getting Started with CMS Medicare Administrative Research Files

References
  1. Stuart EA, DuGoff E, Abrams M, Salkever D, Steinwachs D. Estimating causal effects in observational studies using electronic health data: challenges and (some) solutions. Egems. 2013 Dec 18;1(3):1038.
  2. Jarosek S, Elliott S, Virnig BA. Proton beam radiotherapy in the US Medicare population: Growth in use between 2006 and 2009. Data Points Publication Series [Internet]. 2012 May 7.
  3. Malaeb BS, Yu X, McBean AM, Elliott SP. National trends in surgical therapy for benign prostatic hyperplasia in the United States (2000-2008). Urology. 2012 May 1;79(5):1111-7.
  4. Kurth L, Casey M, Chin B, Mazurek JM, Schleiff P, Halldin C, Blackley DJ. Medical claims paid by workers' compensation insurance among US Medicare beneficiaries, 1999–2016. American journal of industrial medicine. 2021 Mar;64(3):185-91.