Prevalence of Pain as a Function of Self-Reported vs. Observer-Rated Pain Assessments Among Long-Stay US Nursing Home Residents

Abstract

To examine changes in prevalence of pain among long-stay US Nursing Home (NH) residents as a function of being assessed via self-reported vs. observer-rated pain assessments. We conducted a retrospective study of long-stay residents age 65 and over in Medicare / Medicaid certified NHs in 2014-2015, using the Minimum Dataset 3.0. The MDS allows for residents to self-report pain or be assessed by a clinician if unable to self-report. We examined changes in the prevalence of pain between two consecutive quarterly assessments 90 and 180 days after admission (Q1, Q2). We estimated a Poisson regression with robust covariance to examine the prevalence of pain as a function of having an observed assessment at Q2 for those self-reporting at Q1, and of having a self-reported assessment at Q2 for those assessed by staff at Q1. Models were stratified by cognitive impairment, controlling for pain presence/treatment at baseline, age, sex, and clustering at the facility level. Four response patterns were found: self-reporting both assessments (SR2, n=156,749), switching from self-reported to observed (SR-OB, n=9,111), observer-rated at both assessments (OB2, n=15,841), and switching from observed to self-reported (OB-SR, n=7,580). SR2 residents were less cognitively impaired than those with other response patterns. The adjusted Poisson regression model revealed that, among residents without cognitive impairment, the prevalence ratio of pain as a factor of switching from SR to OB was 0.88, as opposed to 1.75 for the same switch among those with cognitive impairment. Residents without cognitive impairment switching from OB to SR had a prevalence ratio of pain of 1.45, and those with cognitive impairment of 0.76. Changes in the prevalence of pain as a function of assessment mode indicates that these approaches do not yield similar results. Including a single constant observer-rated measure of pain among NH residents with cognitive impairment would address this measurement issue.