International Classification of Disease (ICD) Codes in Medicare Files


The purpose of this knowledge base article is to 1) present an overview of International Classification of Disease (ICD) codes and versions available, 2) describe where researchers can find these codes in the Medicare claims, and 3) understand how ICD codes appear in the data.
Current Version Date:
Overview of International Classification of Disease (ICD) Codes

The International Classification of Disease (ICD), developed and published by the World Health Organization, is the international standard for reporting diseases and health conditions. Medicare claims data include both ICD procedure and diagnosis codes across different files. ICD procedure codes are used for inpatient billing only, whereas ICD diagnosis codes are used to identify diseases and health conditions across all claims.

ICD Versions in the Medicare Data

Prior to October 2015, all ICD diagnosis and procedure codes on Medicare claims were reported using version ICD-9-CM. The Department of Health and Human Services mandated the use of ICD-10-CM beginning in October 2015. The Medicare claims include an indicator for each ICD code to identify if the reported procedure or diagnosis code uses ICD-9 or ICD-10.

Medicare File ICD Procedure Code Included in File Number of ICD Procedure Codes Reported* ICD Diagnosis Code Included in File Number of ICD Diagnosis Codes Reported*
Master Beneficiary Summary File No - No -
MedPAR Yes 25 Yes 26+
Inpatient Yes 25 Yes 26+
Outpatient Yes 25 Yes 26+
Carrier No - Yes 13^
Skilled Nursing Facility Yes 25 Yes 26+
Home Health Agency No - Yes 26+
Hospice No - Yes 26+
Durable Medical Equipment No - Yes 13^
Table 1 describes the availability and number of ICD codes reported across Medicare files.
Availability of ICD Codes in the Medicare Data

*Fewer total codes reported prior to 2010.
# While present, not the basis for payment.
+Principal diagnosis code plus 25 additional diagnoses. Diagnosis 1 typically populated with the Principal Diagnosis.
^Principal diagnosis code plus 12 additional diagnoses. Diagnosis 1 typically populated with the Principal Diagnosis.

Note: Beginning with Version “J” of the Medicare claims data, diagnosis E codes, which describe the mechanism of injury, are stored as separate variables in addition to the ICD diagnosis variables within the data.

How ICD Codes Appear in the Medicare Data

In the Medicare data, ICD diagnosis codes are included as character variables with the decimal points removed. In order to search Medicare data for records with a specific ICD diagnosis code, researchers would drop the decimal found in the ICD codebook. For example, for breast cancer, ICD-9-CM diagnosis codes range from 174.0 to 174.9. In the Medicare data, these codes appear as '1740' to '1749'.

One helpful approach ResDAC researchers use when trying to ensure that all necessary codes are included is to do a frequency of an ICD code field in the data to see the range of codes. Using this approach allows researchers to identify whether ‘subcodes’ (i.e., 174.XX) may have been added or deleted over the course of the study period.  For researchers who do not yet have access to the data, CMS maintains a list of ICD procedure and diagnosis codes by fiscal year for ICD-9-CM and ICD-10-CM. Finally, researchers should also be mindful of the transition from ICD-9 to ICD-10 in October 2015, as the codes to identify the same/similar diagnoses will have changed. Please see the attachment, "Managing the Transition from ICD-9 to ICD-10" for more information.