Understanding Missing Patterns in Standardized Patient Functional Data at Inpatient Rehabilitation Facilities

Abstract

<h3>Research Objectives</h3> The Improving Medicare Post-Acute Care Transformation Act of 2014 adopts standardized functional assessment, known as Section GG, across post-acute settings. Section GG has multiple options for missing and much higher percentages of missing values compared to Functional independent Measure ® (R) (FIM) at the inpatient rehabilitation facilities (IRFs). This study examines missing patterns of Section GG in stroke to enhance our understanding of how to interpret and use Section GG scores. <h3>Design</h3> We analyzed 2018 Medicare data. Each individual was assessed by both Section GG and FIM at admission and discharge. We categorized patients with any Section GG missing into one of five missing patterns: refuse, no attempt, not applicable, no value and tie (patients had the same highest frequency for more than two types of missing). We compared FIM scores across five missing patterns. Multiple imputation was applied to calculate missing GG scores. We compared correlations and agreements between Section GG and FIM before and after multiple imputation. <h3>Setting</h3> IRFs. <h3>Participants</h3> Patients with stroke older than 66 years (N=42,789). <h3>Interventions</h3> None. <h3>Main Outcome Measures</h3> Section GG and FIM. <h3>Results</h3> The mean age of the patients was 78.7 (7.44), with the majority as male (52.3%), non-Hispanic White (80.3%), overweight (34.7%) and married (49.4%). Patients in the missing categories of ‘refused' and ‘not applicable' had higher FIM scores compared to other missing types (e.g. FIM self-care scores at discharge: refuse [25.4], not applicable [23.7], no value [20.6], not attempt [18.8] and tie [18.4]). Correlations and agreements between Section GG and FIM were improved after multiple imputation (correlations at discharge: 084 vs. 0.90, before and after imputation). <h3>Conclusions</h3> Patients had different functional levels when missing Section GG items, indicating different levels of assistance, resources and treatment approaches may be needed. Multiple imputation may be a potential solution to calculate missing Section GG values. Future study needs to examine how to adequately report variations of missing Section GG items, to ensure using standardized functional assessment can provide accurate patient assessment outcomes. <h3>Author(s) Disclosures</h3> No conflicts or interest declared by the authors.