Effect of a Symptom Monitoring Intervention for Patients Hospitalized With Advanced Cancer: A Randomized Clinical Trial

Key Points

Question  What is the effect of an inpatient symptom monitoring intervention on symptom burden and health care use among hospitalized patients with advanced cancer?

Findings  In this randomized clinical trial of 321 hospitalized patients with advanced cancer, an inpatient symptom monitoring intervention did not have a significant effect on patients’ self-reported physical and psychological symptoms or their hospital length of stay and 30-day readmission rates.

Meaning  Among hospitalized patients with advanced cancer, this type of inpatient symptom monitoring intervention did not have a significant effect on symptom burden or health care use.

Importance  Symptom monitoring interventions are increasingly becoming the standard of care in oncology, but studies assessing these interventions in the hospital setting are lacking.

Objective  To evaluate the effect of a symptom monitoring intervention on symptom burden and health care use among hospitalized patients with advanced cancer.

Design, Setting, and Participants  This nonblinded randomized clinical trial conducted from February 12, 2018, to October 30, 2019, assessed 321 hospitalized adult patients with advanced cancer and admitted to the inpatient oncology services of an academic hospital. Data obtained through November 13, 2020, were included in analyses, and all analyses assessed the intent-to-treat population.

Interventions  Patients in both the intervention and usual care groups reported their symptoms using the Edmonton Symptom Assessment System (ESAS) and the 4-item Patient Health Questionnaire-4 (PHQ-4) daily via tablet computers. Patients assigned to the intervention had their symptom reports displayed during daily oncology rounds, with alerts for moderate, severe, or worsening symptoms. Patients assigned to usual care did not have their symptom reports displayed to their clinical teams.

Main Outcomes and Measures  The primary outcome was the proportion of days with improved symptoms, and the secondary outcomes were hospital length of stay and readmission rates. Linear regression was used to evaluate differences in hospital length of stay. Competing-risk regression (with death treated as a competing event) was used to compare differences in time to first unplanned readmission within 30 days.

Results  From February 12, 2018, to October 30, 2019, 390 patients (76.2% enrollment rate) were randomized. Study analyses to assess change in symptom burden included 321 of 390 patients (82.3%) who had 2 or more days of symptom reports completed (usual care, 161 of 193; intervention, 160 of 197). Participants had a mean (SD) age of 63.6 (12.8) years and were mostly male (180; 56.1%), self-reported as White (291; 90.7%), and married (230; 71.7%). The most common cancer type was gastrointestinal (118 patients; 36.8%), followed by lung (60 patients; 18.7%), genitourinary (39 patients; 12.1%), and breast (29 patients; 9.0%). No significant differences were detected between the intervention and usual care for the proportion of days with improved ESAS-physical (unstandardized coefficient [B] = −0.02; 95% CI, –0.10 to 0.05; P = .56), ESAS-total (B = −0.05; 95% CI, –0.12 to 0.02; P = .17), PHQ-4–depression (B = −0.02; 95% CI, –0.08 to 0.04; P = .55), and PHQ-4–anxiety (B = −0.04; 95% CI, –0.10 to 0.03; P = .29) symptoms. Intervention patients also did not differ significantly from patients receiving usual care for the secondary end points of hospital length of stay (7.59 vs 7.47 days; B = 0.13; 95% CI, –1.04 to 1.29; P = .83) and 30-day readmission rates (26.5% vs 33.8%; hazard ratio, 0.73; 95% CI, 0.48-1.09; P = .12).

Conclusions and Relevance  This randomized clinical trial found that for hospitalized patients with advanced cancer, the assessed symptom monitoring intervention did not have a significant effect on patients’ symptom burden or health care use. These findings do not support the routine integration of this type of symptom monitoring intervention for hospitalized patients with advanced cancer.

Trial Registration  ClinicalTrials.gov Identifier: NCT03396510

Patients with advanced cancer experience a substantial symptom burden, which is often underrecognized by their clinicians.^1-5 Symptoms such as pain, fatigue, and nausea lead to poor quality of life and functional impairment, and patients with these symptoms report high rates of depression and anxiety.^6-9 In addition, patients’ symptoms contribute to their use of health care services, such as prolonged hospitalizations and readmissions.^10-18 Despite the high symptom burden of patients with advanced cancer, data suggest that clinicians frequently fail to accurately identify their patients’ symptoms, often underestimating the severity, and patients may underreport their symptoms to their oncology team.^4,19-25 Thus, inadequately treated symptoms represent a highly prevalent and problematic issue for patients with advanced cancer, underscoring the need for interventions to help monitor and address these patients’ symptom burden when seeking to improve care delivery and outcomes in oncology.

Symptom monitoring interventions using patient-reported outcomes have shown the ability to enhance the care of patients with cancer and have increasingly become the standard of care in oncology.^26-35 Specifically, data support the use of patient-reported symptom monitoring interventions for patients with cancer to improve symptom management, enhance quality of life, prevent hospitalizations, and potentially increase survival.^26-29,33-35 Based on the compelling results with symptom monitoring interventions in the outpatient setting to date, a growing interest exists to implement remote monitoring of patient-reported outcomes as part of routine oncology practice.^30-32 However, most of the existing data on patient-reported symptom monitoring interventions focus on patients in the outpatient setting, despite the higher symptom burden and worse clinical outcomes of hospitalized patients with cancer.^{10,12-15,35-37} We previously conducted a pilot randomized clinical trial of an electronic symptom monitoring intervention for hospitalized patients with advanced cancer, which we called Improving Management of Patient-Reported Outcomes Via Electronic Data (IMPROVED).³⁸ In that prior work, we showed the feasibility of delivering IMPROVED and highlighted the need for a larger randomized clinical trial to further investigate the efficacy of this intervention.

In the present study, we sought to assess the effects of IMPROVED on symptom burden and health care use among hospitalized patients with advanced cancer. We hypothesized that patients assigned to IMPROVED would report significantly improved symptom burden over the course of hospitalization (primary outcome), experience shorter hospital length of stay (LOS), and have lower readmission rates compared with those assigned to usual care. We chose these outcomes because they are meaningful to patients, clinicians, and health systems as key measures of patients’ care experience and outcomes and because they are frequent outcome measures for symptom monitoring interventions.^26-35 By showing the effects of a symptom monitoring intervention for hospitalized patients with advanced cancer, findings from this work have the potential to inform future efforts to enhance outcomes for this highly symptomatic population.

From February 12, 2018, to October 30, 2019, we enrolled patients at Massachusetts General Hospital in a nonblinded randomized clinical trial of an electronic symptom monitoring intervention (IMPROVED) vs usual care (Alliance for Clinical Trials in Oncology A20 Pilot7³⁹). Trained study staff identified and recruited patients with an unplanned hospital admission during the study period by screening the daily inpatient oncology service census. After providing informed consent, participants completed baseline study measures and were then randomly assigned in a 1:1 ratio to receive IMPROVED or usual care (randomization list generated by statistician and performed within randomly permuted blocks and stratified by cancer type). This study followed the Consolidated Standards of Reporting Trials (CONSORT) reporting guideline. The Dana-Farber/Harvard Cancer Center Institutional Review Board approved the study protocol (trial protocol in Supplement 1). Within 36 hours of hospital admission, eligible patients provided written informed consent. No one received compensation or was offered any incentive for participating in this study.

Eligible patients included those 18 years or older who were admitted to the Massachusetts General Hospital oncology service with a known diagnosis of advanced cancer (defined as receiving treatment with palliative intent as per chemotherapy order entry designation or oncology clinic notes, or not receiving chemotherapy but followed up for incurable disease as per oncology clinic notes). Study participants had to be able to read and respond to questions in English. We excluded patients with planned or elective hospitalizations, defined as hospital admissions for scheduled procedures or for chemotherapy administration or desensitization.

Patients assigned to receive the IMPROVED intervention reported their symptoms daily using tablet computers. Each day during morning inpatient rounds, trained study staff presented the symptom reports to the clinical team (nurses, advanced practice providers, and physicians) via both a printout version of the symptom reports and a computer-based projection screen as they were being discussed. Notably, morning rounds each day consisted of the clinical team meeting in a central room on the inpatient oncology ward to discuss each patient. The detailed symptom reports provided patients’ numeric symptom scores as well as alerts for any specific symptom worsening by 2 or more points from the previous assessment or for any symptom reaching an absolute score of 4 or higher. Those detailed symptom reports also contained graphs depicting patients’ symptom trajectory for the hospitalization. We offered no guidance about managing patients’ symptoms and left all symptom management decisions up to the treating oncology team per their clinical judgment.

Patients receiving usual care reported their symptoms each day, yet these participants’ clinical teams did not receive their symptom reports. Study staff instructed patients in both study groups to report their symptoms as they normally would to their clinical team.

Participants completed baseline study measures prior to randomization. To describe participant demographic characteristics, we asked patients to self-report their race, relationship status, employment, educational level, and annual income. We obtained information about participants’ age, sex, cancer, and comorbidities from the electronic health record.

We obtained patients’ self-reported symptom burden at baseline (at the time of study consent) and daily throughout their hospital admission, including weekends. We used the Edmonton Symptom Assessment System (ESAS), a validated tool for patients with cancer, to measure symptom burden.^40-43 The ESAS assesses pain, fatigue, tiredness, nausea, drowsiness, appetite, dyspnea, and well-being. We also included constipation and diarrhea because they are highly prevalent and modifiable symptoms among patients with cancer.^42,44-47 Patients rated their symptoms on a scale of 0 to 10 (0 reflecting absence of the symptom; 10, the worst possible severity). We categorized the severity of each symptom score as none (0), mild (1-3), moderate (4-6), and severe (7-10), consistent with prior work.¹⁶ In addition, we computed composite ESAS-physical and ESAS-total symptom scores, which included summated values of patients’ physical and total symptoms, as previously used in the oncology setting.^38,41,48 To evaluate patients’ self-reported psychological symptoms, we used the Patient Health Questionnaire-4 (PHQ-4), a 4-item tool with two 2-item subscales assessing depression and anxiety (scores ≥3 for each subscale denoting clinically significant depression or anxiety symptoms).^49,50 Both subscales (range, 0-6) and the composite PHQ-4 (range, 0-12) can be evaluated continuously, with higher scores indicating worse psychological distress.⁴⁹

We obtained information about hospital LOS and unplanned hospital readmissions from the electronic health record because most patients receiving their cancer care at Massachusetts General Hospital are admitted within this health system.³⁸ For hospital LOS, we calculated the number of days from admission to discharge. To determine the risk of hospital readmissions, we identified unplanned readmissions within 30 and 90 days of hospital discharge and calculated time to first unplanned readmission as the number of days from hospital discharge to the first unplanned admission within 30 and 90 days, consistent with prior work.^16,38,51

We included data obtained through November 13, 2020, in the analyses. The primary outcome was a comparison of change in ESAS-physical symptom burden, as measured by the proportion of days with improved symptoms for those who completed 2 or more days of symptom reports, between study groups.³⁸ To calculate the proportion of days that patients’ symptoms improved, we summed the total days that patients had an improvement in their symptom score and divided this value by the total number of days with symptom assessments completed after baseline. Based on data from the pilot study, we estimated that 320 patients would provide greater than 80% power to detect a mean (SD) difference of 0.09 (0.28) in the mean proportion of days with improved ESAS-physical symptoms between study groups using a t test with a 2-sided significance level of .05.³⁸ To determine differences in the mean proportion of days that patients’ symptoms improved, we used linear regression, adjusted for baseline symptom score. To further assess changes in patients’ symptom burden throughout their hospitalization, we used linear regression to evaluate the mean daily change in symptom scores during the hospitalization, adjusted for baseline symptom score. We also explored differences in the proportion of days in which patients reported any symptom worsening by 2 or more points from the previous assessment or any symptom scored 4 or higher because these are clinically important cutoffs to patients and would have resulted in an alert to the team for patients receiving the intervention.^43,48 To evaluate differences in hospital LOS, we used linear regression. To compare differences in time to first unplanned readmission within 30 and 90 days, we used competing-risk regression (with death treated as a competing event). All data analyses included the intent-to-treat population and were conducted using SAS, version 9.4 (SAS Institute Inc).

We enrolled 390 patients (76.2% of patients approached; Figure). Our study analyses included 321 of 390 patients (82.3%) who had 2 or more days of symptom reports completed (usual care, 161 of 193; IMPROVED, 160 of 197), as we sought to assess change in symptom burden. We found no significant differences in baseline characteristics between patients who did or who did not complete 2 or more days of symptom reports. Patients completed a total of 1270 daily symptom reports during a total of 1723 hospitalized days in which symptom collection was possible. The average rate of daily symptom report completion per patient was 81.0%, with a mean (SD) of 3.74 (2.20) symptom reports completed per patient. Participants had a mean (SD) age of 63.6 (12.8) years and were mostly male (180; 56.1%), self-reported as White (291; 90.7%), and married (230; 71.7%) (Table 1). The most common cancer type was gastrointestinal (118 patients; 36.8%), followed by lung (60 patients; 18.7%), genitourinary (39 patients; 12.1%), and breast (29 patients; 9.0%). At baseline, patients reported experiencing a mean (SD) of 3.31 (1.79) moderate or severe symptoms, with only 17 patients (5.3%) reporting no moderate or severe symptoms. More than one-fifth of patients had clinically significant symptoms of depression (71 patients; 22.2%) and anxiety (82 patients; 25.6%) based on their PHQ-4 scores. Few patients died during hospitalization (12 patients; 3.7%), with the remaining being discharged to home (253 patients; 78.8%), hospice (29 patients; 9.0%), or another facility (27 patients; 8.4%).

We found no significant differences between the intervention and usual care groups for the primary outcome of the proportion of days with improved symptoms (Table 2). We found no significant intervention effect on the proportion of days with improved ESAS-physical (unstandardized coefficient [B] = –0.02; 95% CI, –0.10 to 0.05; P = .56), ESAS-total (B = –0.05; 95% CI, –0.12 to 0.02; P = .17), PHQ-4–depression (B = –0.02; 95% CI, –0.08 to 0.04; P = .55), PHQ-4–anxiety (B = –0.04; 95% CI, –0.10 to 0.03; P = .29), and PHQ-4–total (B = –0.06; 95% CI, –0.13 to 0.01; P = .11) symptoms. The mean proportion of days with improved symptoms for each of the individual ESAS symptoms by study group is shown in eFigure 1 in Supplement 2.

In addition, we found no significant differences in the mean day-to-day change in symptom scores during patients’ hospitalization (Table 3). We found no significant intervention effect on the mean daily change in ESAS-physical (B = 0.57; 95% CI, –1.10 to 2.24; P = .50), ESAS-total (B = 1.24; 95% CI, –0.69 to 3.16; P = .21), PHQ-4–depression (B = 0.03; 95% CI, –0.15 to 0.20; P = .76), PHQ-4–anxiety (B = 0.11; 95% CI, –0.09 to 0.31; P = .27), and PHQ-4–total (B = 0.13; 95% CI, –0.17 to 0.42; P = .41) symptoms. Furthermore, we did not find significant differences in the proportion of days in which patients reported any symptom worsening by 2 or more points from the previous assessment or any symptom scored 4 or higher (usual care, 95.4% vs IMPROVED, 92.9%; P = .19).

We did not find significant intervention effects on health care use. We found no significant differences in hospital LOS (usual care, 7.47 days vs IMPROVED, 7.59 days; B = 0.13; 95% CI, –1.04 to 1.29; P = .83) or risk of unplanned hospital readmissions (competing-risk regression) in 30 days (usual care, 33.8% vs IMPROVED, 26.5%; hazard ratio, 0.73; 95% CI, 0.48-1.09; P = .12) and 90 days (usual care, 45.5% vs IMPROVED, 38.7%; hazard ratio, 0.78; 95% CI, 0.55-1.10; P = .15) (eFigure 2 in Supplement 2).

In this randomized clinical trial of hospitalized patients with advanced cancer, IMPROVED did not have a significant effect on patients’ symptoms or health care use. Despite a well-designed and rigorously conducted study, this trial did not meet the primary end point of showing improvements in patients’ symptom burden, as measured by the proportion of days with improved symptoms, with an inpatient symptom monitoring intervention. In addition, IMPROVED did not significantly reduce patients’ hospital LOS or risk of unplanned readmission. Collectively, findings from this work do not support the routine integration of this type of symptom monitoring intervention for hospitalized patients with advanced cancer. However, as the largest randomized clinical trial to date investigating a global symptom monitoring intervention in the inpatient oncology setting, this study provides important insights to inform future efforts to integrate symptom monitoring interventions using patient-reported outcomes into the care of patients with cancer.

Although the field of oncology has begun adopting efforts to integrate patient-reported outcomes into routine practice, the present study shows a lack of benefit for a symptom monitoring intervention in the inpatient setting. In our prior pilot study of IMPROVED, we found encouraging preliminary efficacy for this intervention, despite similar enrollment rates in the same population; thus, our present findings illustrate how pilot studies may overestimate benefits.⁵² Moreover, findings from the present study highlight the need for population-specific interventions because different patient populations have unique care needs meriting interventions designed to meet these needs.^53-55 Hospitalized patients with advanced cancer clearly represent a highly symptomatic population, as evidenced by the high symptom burden among the patients in the present study.^10-18,36 However, hospitalized patients uniformly receive intensive symptom management in the inpatient setting, which may negate the effect of systematic symptom monitoring.¹⁸ In addition, supportive care interventions may not have the same utility in the hospital setting as that of the outpatient setting, considering that the goals and care needs of inpatients and outpatients often differ.^56-58 Thus, our findings in the present study underscore the importance of developing and testing population-specific supportive care interventions, particularly in the field of oncology, as patients’ care needs often differ by cancer type and stage as well as across care settings.

In addition, when designing patient-centered, population-specific interventions, researchers must carefully consider the details of the intervention and usual care provided. In the present randomized clinical trial, patients in both study groups reported their symptoms daily throughout their hospitalization using validated measures, and only patients assigned to the IMPROVED intervention had their symptoms presented to their clinical team. We did not provide guidance or feedback about managing patients’ symptoms or ask clinicians to document clinical actions in response to the symptom reports, which other studies have done,^33,35,59 thus possibly limiting the potency of IMPROVED. In addition, we reported intervention patients’ symptoms to all members of their inpatient care team each day, including nurses, advanced practice providers, and physicians. Although this ensured that all members of the care team remained updated about patients’ symptoms each day, we never assigned a specific clinician to take responsibility for addressing the symptoms, and this could result in a diffusion of responsibility.⁶⁰ For this study, we asked patients in the usual care group to also report their symptoms each day, which could have encouraged these patients to more proactively report their symptoms to their clinical teams than they otherwise would have done in the absence of the study. Thus, our findings with the IMPROVED symptom monitoring intervention should be interpreted within the context of the details of the care provided in both the intervention and usual care groups.

The selection of study outcomes may play a major role in determining the utility and meaningfulness of interventions involving patient-reported outcomes. We used the ESAS tool to monitor patients’ symptoms, and we selected the ESAS-physical symptom score as our primary outcome, which may have influenced our results, but other studies have successfully used this strategy.^34,38 Prior studies of symptom monitoring interventions have used other symptom assessment tools, as no uniform tool exists for this work.^26-29,59 Notably, symptom burden may not represent a readily modifiable study outcome in the inpatient oncology population, and IMPROVED may have resulted in benefits to other unmeasured outcomes, such as patient activation or satisfaction with care, even despite the lack of an effect of IMPROVED on symptom burden.^10-18,35,36 Moreover, we did not power the current study to find significant effects on health care use, yet the differences in the rates of hospital readmissions appeared to favor the intervention, whereas hospital LOS was longer for patients assigned to IMPROVED, consistent with our prior work.³⁸ Ultimately, IMPROVED did not result in detectable benefits in the present study, despite our selection of clinically important outcomes, which further supports the lack of a meaningful impact from this type of symptom monitoring intervention in the inpatient oncology setting.

Several limitations of this study merit discussion. First, we conducted the trial at an academic center with limited sociodemographic diversity, thereby limiting the generalizability of our results. Patients with other racial and ethnic and socioeconomic characteristics as well as patients in other clinical settings may have experienced differential effects from this type of intervention. Second, we lack data regarding some factors that could influence the effect of IMPROVED, such as patients’ health literacy, prognostic understanding, physical and cognitive function, and level of caregiver involvement.^61-63 Future studies should investigate whether those, and other important factors, such as cancer type and use of additional support services (eg, nutrition, palliative care, and physical or occupational therapy) influence the effect of IMPROVED. We chose symptom burden and health care use as important outcomes in the current study, but additional outcomes could be considered in future work, such as patient-clinician communication and coordination of care. In addition, we lacked information about how often clinicians discussed the symptom reports or developed a plan to address patients’ symptoms. Third, patients in our study had various cancer types with varying times since diagnosis with advanced disease, thereby limiting our ability to determine how best to personalize IMPROVED according to patients’ distinct care needs. Fourth, we did not collect information about postdischarge patient-reported outcomes or patients’, caregivers’, and clinicians’ perceptions of IMPROVED. Future efforts to integrate symptom monitoring with patient-reported outcomes into the care of patients with cancer should consider these key outcomes.

In this randomized clinical trial, we sought to determine the effects of IMPROVED on symptom burden and health care use among hospitalized patients with cancer. Few studies of inpatient symptom monitoring interventions exist, and the present study addressed this gap by investigating daily symptom monitoring in the inpatient oncology setting. The IMPROVED intervention did not show improvements in patients’ symptom burden, hospital LOS, or unplanned readmission. Notably, our data highlight the critical importance of efforts to address the symptomatic needs of hospitalized patients with cancer, as evidenced by these patients’ high symptom burden and willingness to participate in a symptom monitoring intervention, based on the high rates of study enrollment and completion of daily symptom assessments. Collectively, findings from this trial provide key insights to help guide future work investigating symptom monitoring interventions using patient-reported outcomes in oncology and demonstrate the lack of benefit for this type of symptom monitoring intervention for hospitalized patients with advanced cancer.

Accepted for Publication: November 22, 2021.

Published Online: February 10, 2022. doi:10.1001/jamaoncol.2021.7643

Corresponding Author: Ryan D. Nipp, MD, MPH, Division of Hematology & Oncology, Department of Medicine, Massachusetts General Hospital, 55 Fruit St, Yawkey, Boston, MA 02114 (rnipp@mgh.harvard.edu).

Author Contributions: Dr Nipp had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Nipp, Kay, Ryan, Jackson, Greer, El-Jawahri, Temel.

Acquisition, analysis, or interpretation of data: Nipp, Horick, Qian, Knight, Kaslow-Zieve, Azoba, Elyze, Landay, Kay, Ryan, Jackson, El-Jawahri, Temel.

Drafting of the manuscript: Nipp, Temel.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Horick, Qian, El-Jawahri.

Obtained funding: Nipp.

Administrative, technical, or material support: Nipp, Qian, Azoba, Landay, Kay, Ryan, Jackson, Temel.

Supervision: Nipp, Ryan, Jackson, Greer, El-Jawahri, Temel.

Conflict of Interest Disclosures: Dr Knight reported stock ownership in Agilent Technologies, Assembly Biosciences, Illumina, IQVIA, and Thermo Fisher Scientific outside the submitted work. Dr Ryan reported receiving grants from Stand Up To Cancer; equity from MPM Capital and Exact Sciences Equity; and personal fees from Boehringer Ingelheim, UpToDate, and McGraw Hill outside the submitted work. Dr Greer reported receiving grants and personal fees from Blue Note Therapeutics and royalties from Springer (Humana Press) outside the submitted work. Dr El-Jawahri reported consulting for Novartis, GlaxoSmithKline, and Aim Pathway outside the submitted work. Dr Temel reported receiving grants from AstraZeneca through funding from the National Comprehensive Cancer Network outside the submitted work. No other disclosures were reported.

Funding/Support: This study was supported by Alliance NCORP grant UG1CA189823 and the Alliance NCORP Research Base.

Role of the Funder/Sponsor: The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Meeting Presentation: This paper was presented at the 2020 ASCO Virtual Scientific Program; May 29, 2020.

Data Sharing Statement: See Supplement 3.