The impact of patient-reported outcomes in cancer patients in Croatia.

Abstract

e18659 Background: Perception of symptoms is essential for optimal cancer care and improving the quality of life but an inappropriate interpretation may lead to treatment outcomes failure or may leave the patients undertreated. This variation in the perception of adverse events (AEs) can lead to wrong assumptions and potentially unnecessary hospitalizations (H) or emergency room visits (ERV). Methods: Patients considered for inclusion were 18 years of age or older receiving systemic therapy and were able to read and understand questionnaire via tablet computers. All patients had a verified diagnosis of malignant disease of breast (31%), lung (17%), colorectal (16%), hematological (8%) or other site cancers (28%). Patients were asked on the same day as physician assessment, to complete 10 PRO-CTCAE symptoms (including fatigue, pain, nausea, vomiting, diarrhea, dysgeusia, appetite, sleep disturbance, fever, and alopecia). Participants remained on study until discontinuation of cancer treatment, withdrawal, or death. Informed consent was obtained from all patients and the study was approved by the institutional review board of the General Hospital Pula, Croatia. Concordance of symptoms reporting between patients and health care providers at the individual level was compared, and intraclass correlation coefficients were computed by a two way mixed effect model and an absolute agreement definition. Univariate and multivariate COX proportional hazard models adjusted for covariates including age, gender, cancer type and performance status were constructed for unplanned H and ERV. Results: The median age of 308 included patients in this trial was 62 years. Cohen's coefficient of agreement that corrects for chance for AEs gr. 1-5 and AEs gr. 3-4 was 0.115 (95%, CI, 0.15 to 0.02), indicating only slight agreement and 0.320 (95%, CI, 0.428 to 0.448) indicating fair agreement, respectively. Physicians underestimated symptoms at the group level (p < 0.013) for all symptoms. The highest level of agreement was found for pain but fatigue and alopecia were most underestimated. The median times of follow-up were 13 months, and 8,1 % of patients had H or ERV. Female patients had worse PRO scores, with higher levels of pain, fatigue, and alopecia across disease groups. In multivariate analysis, patient symptoms was not predictive, but we noticed trend for higher patient reports of pain (p = 0.05) for unnecessary H or ERV. Conclusions: The failure of the physician to identify a substantial number and severity of symptoms identified by the PRO instrument is not surprising considering the published experience in clinical trials. To our knowledge, this is the first such analysis aimed specifically at examining the reliability of physician identification of AEs in a real world practice in Croatia. The incorporation of PRO measures into routine practice might help to avoid unnecessary hospitalizations and emergency room visits. Clinical trial information: 2019-000855-15.