Secondary Medicare Diagnosis Coding Expansion, Electronic Health Records Associated with Increase in Illness Severity

Is the expansion of secondary diagnosis codes in January 2011 and incentive payments for health information technology associated with changes in measured severity of illness?

Bottom line: Expansion of secondary diagnosis coding positions is associated with a statistically significant increase in measured severity of illness among hospitalizations for all diagnoses, diagnoses commonly targeted by incentive programs, and untargeted diagnoses.

Comment: This is another interesting paper. There have been similar publications in the past few years regarding how you cannot use electronic medical record (EMR) data to assess severity of comorbidities because it’s becoming a coding game involving trying to make patients look as sick as possible. You could also argue that we have been chronically undercoding and that recent changes are more reflective. Regardless of why we may hypothesize these changes occurred, this has ramifications for hospital ranking, predicted/actual mortality, and (I believe) billing/revenue. It is also reflected in compliance seminars where lectures are given on the importance of entering a diagnosis code for every possible thing the patient has going on (again, this may or may not be appropriate). It also has an impact on those of us who do outcomes research, especially if we need to retrospectively include and analyze data from hospital charts. You could argue that cohorts from pre and post 2011 are not comparable (at least if calculating different comorbidities indices from listed comorbid conditions) due to significant coding variability. —Jennifer A. Villwock, MD

BACKGROUND: In January 2011, the Centers for Medicare and Medicaid Services (CMS) expanded the number of secondary diagnosis coding slots, capturing up to an additional 15 diagnoses in inpatient claims. Similarly, the expansion of electronic health record (EHR) capabilities may have allowed hospitals to capture more detail about patient risk.

STUDY DESIGN: Cohort study of 47,951,443 Medicare fee-for-service beneficiary discharges between Jan. 1, 2008, and Aug. 31, 2015, at 2,850 hospitals.

SETTING: University of Michigan Medical School, Ann Arbor, Mich., USA.

SYNOPSIS: Approximately 14% of all discharges were for targeted diagnoses. Between 2008 and 2015, the mean number of condition categories increased from 1.70 to 2.67, the Medicare severity diagnosis related group (MS-DRG) weight increased from 1.50 to 1.64, and the Hierarchical Condition Category (HCC) score increased from 1.23 to 1.69. Over the same period, the mean number of condition categories increased from 2.24 to 3.44 for targeted diagnoses and from 1.61 to 2.54 for untargeted diagnoses. Larger hospitals experienced a greater absolute and relative change in the number of condition categories. Incentives for meeting health information technology criteria were associated with a modest change in the number of condition categories for all diagnoses. A sensitivity analysis found that, as the degree of EHR use increased, the differential change in measured severity of illness increased. The differential change in the C statistic from adding condition category indicators before and after the expansion of secondary diagnoses was 0.90% among all diagnoses, 0.95% among targeted diagnoses, and 0.81% among untargeted diagnoses. Limitations included a deviation from the traditional condition category derivation and lack of measured participation in the Medicaid meaningful use program.

CITATION: Sukul D, Hoffman GJ, Nuliyalu U, et al. Association between Medicare policy reforms and changes in hospitalized Medicare beneficiaries’ severity of illness. JAMA Netw Open. 2019;2:e193290.