Routine use of home-based symptom monitoring and management using electronic patient-reported outcomes (ePRO) to improve care delivery is on the horizon. Randomized clinical trials demonstrate that use of patient-reported symptoms can have marked impact on patient outcomes, including minimizing symptom burden, enhancing quality of life, reducing hospitalizations, increasing time receiving cancer treatments, and, in some studies, improving survival.1-4 As a result, these powerful tools are now recommended as part of value-based health care initiatives, including the proposed Oncology Care First Model by the Center for Medicare and Medicaid Innovation and the American Society of Clinical Oncology’s Oncology Medical Home Model.5 One might think that with this evidence, as well as the pressure from professional organizations and payers, that health care systems across the country would rapidly adopt this approach to patient care. However, this is not the case, and few health systems have successfully, fully integrated ePRO.6 While the lack of adoption of this practice is multifactorial, one key component of the implementation gap is the lack of knowledge about how the intervention itself should be delivered as part of routine care.
In the study by Daly and colleagues,7 the authors begin to tackle an important question of frequency of assessment administration in ePRO. This study used daily symptom assessment in contrast to the weekly schedule that has been used in many of the prior studies. With the daily assessments, patients completed a mean (SD) of 3.9 (2.5) assessments per week in the initial 6 months, but this tapered off over time, falling to 2.7 (2.1) assessments being completed per week after 1 year of enrollment. While Daly et al7 did not report a goal for completion, this falls well below the 80% to 85% completion rates observed in studies of weekly symptom monitoring and raises questions about the distribution of assessment responses.8 Additionally, frequent assessments have the potential to lead to survey fatigue. Further work is needed to understand optimal frequency over the course of illness. For example, the frequency of assessment may be higher initially and taper as patients transition to the survivorship phase. In a 2019 study by Denis et al2 of patients in surveillance for lung cancer, weekly symptoms were leveraged to identify new symptoms and recurrence, which was associated with a 19% survival difference at 2 years. While this may be ideal for a disease in which patients recur early after initial treatment, for diseases for which the course is more indolent with later recurrences (eg, metastatic hormone-sensitive breast cancer), a less frequent cadence may be more appropriate, given the length of time expected for completing assessments. Ultimately, robust studies are needed to understand both clinically meaningful timing and to garner patient perspectives on varying schedules.
Another key consideration is that patients who do not complete assessments may be systematically different than patients who do. Patients may not complete assessments owing to technical difficulties, such as login trouble, difficulties navigating the technology, or a phone being shut off. Importantly, logistical challenges are expected among more vulnerable populations, such as those with lower health literacy and lower computer literacy. It is important to note that in a 2016 study by Basch and colleagues,4 the greatest benefit of ePROs was observed in patients lacking computer experience. Thus, the lack of engagement of patients who do not complete assessments may not only limit the impact of the intervention, but also has the potential to widen existing disparities in care delivery. Alternatively, the lack of assessment completion may be secondary to patients being too sick to complete assessments, again, a scenario in which follow-up is warranted. This challenge might be addressed through protocols in which patients receive a call if they have not reported in a specific time period (eg, 1 week) and should be considered in future studies evaluating more frequent assessment administration.
Despite the challenges with assessment completion, the study by Daly et al7 provides important context into symptom trajectories. The presence of alerts was associated with unanticipated emergency department use or hospitalization; 9% of red (severe) alerts were followed by an acute event, with increased events occurring with a higher number of red alerts. Importantly, many of these events were not able to be prevented despite the clinical team knowing of the alert. This suggests that by the time a patient has a red alert, it may be too late to prevent an acute care episode. Although Daly et al7 note that 45% of red alerts did not have a yellow (moderate) alert in the preceding week, this suggests that 55% did have a prior yellow alert. This proportion would likely be higher if one were to look at the weeks leading up to the red alert, suggesting that greater focus on yellow alerts may have potential for early intervention to prevent progression to red alerts and ultimate hospitalization. The scoring of symptoms and alert thresholds remains another ripe area of research for ePRO implementation.
Daly et al7 note that patients experience symptoms that may vary through the week, thus daily alerts are necessary. However, daily monitoring generates an enormous number of alerts for clinical teams to respond to. In this study, 14 603 assessments occurred in a 16-month period, with a 50% alert rate for a total of approximately 7300 assessments with alerts. If a nurse were to spend 10 minutes per alert reviewing the health record and calling a patient, this would translate to 1216 hours or approximately 0.6 full-time equivalents of nursing time. Furthermore, weekend assessments still had more than a 50% response rate, warranting staff to be available for timely management in nonbusiness hours. While this might be feasible at well-resourced, large institutions, many institutions may not have the capacity to accommodate this additional workload. Thoughtful consideration of workforce costs is needed for implementation of practice transformation activities.9 Given that Daly et al7 observed a predominance of within-week shifts being between no alert and a yellow alert, a 1-week recall period may in fact be appropriate with these patients generating yellow alerts for the week. Ultimately, efforts will inevitably be needed to separate alerts needing further management from noise, to both minimize alert fatigue for nursing staff and maximize nursing efficiency, particularly when nursing staff is in short supply owing to the COVID-19 pandemic.10
The study by Daly et al7 scratches the surface of what will be needed scientifically for broad uptake of ePRO. This evidence-based intervention is uniquely poised for rigorous evaluation using the principles of implementation science, given the complexity and multilevel nature of the intervention. Hopefully, future analysis of the program assessed by Daly et al7 and others will allow the oncology community to collectively learn, avoiding duplication of unsuccessful implementation strategies, to ultimately unlock the potential of home-based symptom monitoring.
Published: March 4, 2022. doi:10.1001/jamanetworkopen.2022.1090
Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2022 Rocque GB. JAMA Network Open.
Corresponding Author: Gabrielle B. Rocque, MD, MSPH, University of Alabama at Birmingham, 1824 Sixth Ave S, Birmingham, AL 35924-3300 (firstname.lastname@example.org).
Conflict of Interest Disclosures: Dr Rocque reported receiving grants from Carevive, Genentech, and the National Institute of Nursing Research and grants and personal fees from Pfizer outside the submitted work.
Rocque GB. Learning From Real-world Implementation of Daily Home-Based Symptom Monitoring in Patients With Cancer. JAMA Netw Open. 2022;5(3):e221090. doi:10.1001/jamanetworkopen.2022.1090
Customize your JAMA Network experience by selecting one or more topics from the list below.