Adjusting a Patient-Reported Scale for Overlapping Symptoms of Co-Occurring Conditions

• Activity Number: 443 - Latent Variables, Causal Inference, Machine Learning and Other Topics in Mental Health Statistics
• Type: Contributed
• Date/Time: Thursday, August 12, 2021 : 4:00 PM to 5:50 PM
• Sponsor: Mental Health Statistics Section
• Author(s): Douglas Gunzler*
• Companies: Case Western Reserve University
• Keywords: structural equation modeling; patient-reported outcomes; factor analysis; multiple indicator multiple cause modeling; depression; hemodialysis

Abstract

A typical objective in scoring a multi-item patient reported scale is to assess the level of an underlying trait. For example, the patient health questionnaire (PHQ)-9 is used as a screening tool for depression. However, a problem in the use of patient-reported scales is the inflation or deflation of scoring due to the presence or absence of overlapping symptoms when individuals suffer from co-occurring conditions. In this talk, structural equation modeling (SEM) approaches will be discussed for analysis of overlapping symptoms in co-occurring conditions using a patient reported scale. Factor scores extracted from the model can be transformed using a probability integral transformation back into the metric of the original scale. These transformed scores are adjusted scale scores that account for overlapping symptoms of the co-occurring conditions. We also discuss approaches for the validation of this adjusted scale. We illustrate these approaches for a hemodialysis-specific PHQ-9 in a sample of 924 hemodialysis subjects, many of which have co-occurring depressive symptoms.