Crosswalk Algorithms for Cognitive and Functional Outcomes Among 2013–2018 Medicare Beneficiaries With Dementia

Before 2019, the Minimum Data Set (MDS) and Outcome and Assessment Information Set (OASIS) had incongruent response categories for rating cognitive impairment and activities of daily living (ADLs), hindering direct comparisons between nursing facilities and home health. We devised rule-based algorithms to compare cognitive impairment and ADL limitations between these 2 care settings among people with Alzheimer's disease and Alzheimer's disease–related dementias (ADRD). A retrospective cohort study. Included fee-for-service Medicare beneficiaries (2013–2018) transitioning from nursing facilities to home health, with 1-year of continuous enrollment, aged ≥65 years, diagnosed ADRD, and with complete MDS discharge and OASIS admission assessments (N = 398,496). We identified target phenotypes using the Cognitive Function Scale (CFS) and ADL items from the MDS discharge assessment as reference standards. We compared 6 OASIS-based algorithms for cognitive impairment and 1 for each ADL limitation by estimating sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). The average age was 83.5 (SD = 7.5) years and 82.3% transitioned from nursing to home health within 3 days. In the MDS discharge assessment, 42.2% had moderate-to-severe cognitive impairment. ADL limitations ranged from 71.4% for feeding to 97.8% for bathing. Compared with the moderate-to-severe cognitive impairment (CFS ≥3) on the MDS, the OASIS cognitive assessment indicating “considerable assistance to total dependence in routine situations” had 24% sensitivity, 94% specificity, 75% PPV, and 63% NPV. The ADL limitation algorithms exhibited high sensitivities (>96%) and PPVs (>94%) except for feeding (Sensitivity: 82%; PPV: 74%). Despite the short time frame between the 2 assessments, the OASIS admission assessment showed a higher prevalence of ADL limitations than the MDS discharge assessment. We highlighted differences in patient function between post-acute care settings. Our algorithms can help researchers, clinicians, and policymakers standardize patient-centered outcomes for comparative effectiveness research or quality initiatives.