Health care providers underestimate symptom intensities of cancer patients: A multicenter European study

Abstract

Background: Many patients with advanced cancer depend upon health care providers for symptom assessment. The extent of agreement between patient and provider symptom assessments and the association of agreement with demographic- and disease-related factors was examined. Methods: This cross-sectional study included 1933 patient-health care provider dyads, from 11 European countries. Patients reported symptoms by using the four-point scales of the European Organization of Research and Treatment of Cancer Core Quality of Life Questionnaire (EORTC QLQ-C30) version 3, and providers used corresponding four-point categorical scales. Level of agreement was addressed at the group level (Wilcoxon Signed-Rank test), by difference scores (provider score minus patient score), at the individual level (Intraclass Correlation Coefficients, ICCs) and visually by Bland-Altman plots. Absolute numbers and chi-square tests were used to investigate the relationship between agreement and demographic-, as well as disease-related factors. Results: The prevalence of symptoms assessed as moderate or severe by patients and providers, respectively, were for pain (67 vs.47%), fatigue (71 vs. 54%), generalized weakness (65 vs. 47%), anorexia (47 vs. 25%), depression (31 vs. 17%), constipation (45 vs. 30%), poor sleep (32 vs. 21%), dyspnea (30 vs. 16%), nausea (27 vs. 14%), vomiting (14 vs. 6%) and diarrhea (14 vs. 6%). Symptom scores were identical or differed by only one response category in the majority of patient-provider assessment pairs (79-93%). Providers underestimated the symptom in approximately one of ten patients and overestimated in 1% of patients. Agreement at the individual level was moderate (ICC 0.38 to 0.59). Patients with low Karnofsky Performance Status, high Mini Mental State-score, hospitalized, recently diagnosed or undergoing opioid titration were at increased risk of symptom underestimation by providers (all p < 0.001). Also, the agreement was significantly associated with drug abuse (p = 0.024), provider profession (p < 0.001), cancer diagnosis (p < 0.001) and country (p < 0.001). Conclusions: Considerable numbers of health care providers underestimated symptom intensities. Clinicians in cancer care should be aware of the factors characterizing patients at risk of symptom underestimation.