Concordance between patient-reported (PR) and physician-documented (PD) comorbidities and symptoms among Stage 4 breast cancer patients (pts).

Abstract

e24089 Background: Comorbidities (Co) and symptoms (Sx) in metastatic breast cancer (MBC) pts impact treatment decisions, eligibility for clinical trials, and influence prognosis and quality of life. The aim of this study was to evaluate the concordance between PR and PD Co and Sx in an electronic medical record to document pts’ health, identify Co or Sx that may be more comprehensively reported by pts vs physicians (phy), and understand if PR, PD, or concordant Co or Sx were more predictive of pt survival. Methods: New pts at UCSF’s Breast Care Center (BCC) are administered an electronic intake survey assessing PR health history, Co and Sx. Chart reviews of the initial clinic visit were conducted for PD Co and Sx. Pt and phy concordance was summarized for 54 Co and 42 Sx. Agreement was quantified using Cohen’s kappa (κ) (poor (κ < 0.2), fair (0.2≤κ < 0.4), moderate (0.4≤κ < 0.6), substantial to high (κ ≥0.6)). Cox-proportional hazards models were used to determine hazard ratios (HR) for survival with PR, PD, and concordant Co and Sx, controlling for factors including age, sites of metastatic disease, tumor biology, etc. Results: Between Nov 2016 and Feb 2020, 168 pts with confirmed MBC seen at the BCC who consented to use of their clinical data for research were included in the analysis (median age, 56 years; median time from MBC diagnosis, 0.46 years). Highest PD Co were obesity, hypertension (HTN) and thyroid disease, while highest PR Co were HTN, depression and arthritis. 23 of 54 Co had a moderate to high level of agreement between physician and pt reports (κ≥0.40). Agreement was high for diabetes, HTN, and low for obesity, anxiety, and gastroesophageal reflux disease (GERD). After controlling for clinical covariates, of these Co, only PR GERD was significantly associated with survival (HR = 1.87, p < 0.05). Only 2 of 42 Sx (shortness of breath and cough) had a moderate to high agreement between PD and PR. PR Sx were the primary drivers for predicting survival (HR > 1, p < 0.05 for PR Sx including vomiting, fatigue, weight loss and others). PD and PR agreement for these sx was poor (κ < 0.2). Conclusions: In this review of data collected as part of routine care at an academic medical center, there was substantial variance in the concordance of Co and Sx reported by pt vs phy. Concordance was higher for Co, with phy documenting a higher number of Co that can be objectively measured by lab values (diabetes, HTN) while pts reported higher rates of Co that were more subjective (anxiety, GERD). Overall, pts reported more Sx than phy. PR Sx were also the highest predictors of survival. Intriguingly, Co such as diabetes and HTN did not predict survival in this metastatic population. This suggests that incorporating PR Sx, either secondary to their cancer or related to their Co, may provide a more informative estimate of a pt’s predicted survival, and assist phy in evaluating trial eligibility and reasonable treatment options.