Impact of assessment frequency of patient-reported outcomes: an observational study using an eHealth platform in cancer patients

The evaluation of patient-reported outcomes (PRO) in cancer has proven relevant positive clinical impact on patients' communication with healthcare professionals, decision-making for management, well-being, and overall survival. However, the optimal frequency of PRO assessment has yet to be defined. Based on the assumption that more frequent sampling would enhance accuracy, we aimed at identifying the optimal sampling frequency that does not miss clinically relevant insight. We used pilot data from 31 advanced cancer patients who completed once daily the 19-item MD Anderson Symptom Inventory at home. The resulting dataset allowed us to compare different PRO assessment frequencies to daily sampling, i.e., alternate days (q2d), every third day (q3d), or once a week (q1w). We evaluated the sampling frequencies for two main outcomes: average symptom intensity and identification of severe symptoms. The majority of the differences between corresponding averages of daily data and those for q2d, q3d, and q1w datasets were close to 0, yet the extremes exceeded 5. Clinically meaningful differences, i.e., > 1, were observed in 0.76% of patient items for q2d, in 2.72% for q3d, and in 11.93% for q1w. Moreover, median values of missed instances of a severe symptom (i.e., > 6) were 14.6% for q2d, 27.8% for q3d, and 55.6% for q1w. Our analysis suggests that in patients receiving chemotherapy for advanced cancer, increasing the density of PRO collection enhances the accuracy of PRO assessment to a clinically meaningful extent. This is valid for both computations of averages symptom burden and for the recognition of episodes of severe symptom intensity.