A cluster analysis of clinician distress trajectories when caring for seriously ill hospitalized patients.

Abstract

9012 Background: Millions of Americans are hospitalized every year; many of whom are seriously ill with one or more co-morbidities. Clinicians, including physicians and advanced practice providers (APPs) care for these complex patients while also juggling competing clinical demands from fielding multiple specialty recommendations to navigating interprofessional relationship. But clinicians are distressed, which has the potential to impact the quality of healthcare delivery at the moment of care and in the future. To date there is limited empirical inquiry examining the longitudinal trajectory of clinician distress and its potential impact on healthcare quality. Study objective was to describe unique clinician distress trajectories in general medicine hospital clinicians caring for seriously ill patients based on their level of distress over time through mobile ecological momentary assessments (mEMAs). Methods: Latent class cluster analysis of prospective serial mEMAs. Exploratory analysis of patient and clinician variables was then performed using generalized estimating equations univariate ordinal logistic regression. Total participants consisted of 184 hospital encounters for hospital clinicians (n=68) caring for seriously ill patients (n=151). Results: The main outcome of clinician distress typology was identified by latent class cluster analysis. Distress was measured by serial mEMA distress thermometer levels over two days. The sample included more physicians (60.3%) than APPs (39.7%) and clinicians had an average of 8.4 years’ experience (range 0-31 years). Patients average age was 65.4 years, majority were male (53.6%) and White (61.6%) The majority of patients had a primary serious illness of a solid tumor malignancy (50%), followed by hematologic malignancy (27.9%) then non-cancer chronic illness (22.1%). Clinicians fell into four typology clusters: low distress (23.2%), moderate distress (33.1%), variable distress (19.7%) or high distress (23.9%). Credentials (APP vs. physician; x 2 =9.11, p=0.0025) and clinician emotional experience (x 2 =11.29, p=0.0008) were significantly associated with clustering by typology. Compared to physicians, APPs were six times more likely to be in a higher distress typology (OR=6.16, p=0.003). Clinicians who had reported more emotions were more likely to be in a higher distress typology (OR=1.90, p=0.001). Mid-career clinicians were more likely to be distressed than either early or late career clinicians (OR=1.80, p 0.370). Patient and clinician demographics were not otherwise significantly related to clusters. Conclusions: Clinicians experience distress throughout their workday. This study identifies unique distress trajectories measured in real-time and specific characteristics of those trajectories that can be leveraged by healthcare systems when designing interventions and support resources for hospital clinicians.