Feasibility of Fitness Tracker Usage to Assess Activity Level and Toxicities in Patients With Colorectal Cancer

PURPOSE

Performance status (PS) is a subjective assessment of patients' overall health. Quantification of physical activity using a wearable tracker (Fitbit Charge [FC]) may provide an objective measure of patient's overall PS and treatment tolerance.

MATERIALS AND METHODS

Patients with colorectal cancer were prospectively enrolled into two cohorts (medical and surgical) and asked to wear FC for 4 days at baseline (start of new chemotherapy [± 4 weeks] or prior to curative resection) and follow-up (4 weeks [± 2 weeks] after initial assessment in medical and postoperative discharge in surgical cohort). Primary end point was feasibility, defined as 75% of patients wearing FC for at least 12 hours/d, 3 of 4 assigned days. Mean steps per day (SPD) were correlated with toxicities of interest (postoperative complication or ≥ grade 3 toxicity). A cutoff of 5,000 SPD was selected to compare outcomes.

RESULTS

Eighty patients were accrued over 3 years with 55% males and a median age of 59.5 years. Feasibility end point was met with 68 patients (85%) wearing FC more than predefined duration and majority (91%) finding its use acceptable. The mean SPD count for patients with PS 0 was 6,313, and for those with PS 1, it was 2,925 (122 and 54 active minutes, respectively) (P = .0003). Occurrence of toxicity of interest was lower among patients with SPD > 5,000 (7 of 33, 21%) compared with those with SPD < 5,000 (14 of 43, 32%), although not significant (P = .31).

CONCLUSION

Assessment of physical activity with FC is feasible in patients with colorectal cancer and well-accepted. SPD may serve as an adjunct to PS assessment and a possible tool to help predict toxicities, regardless of type of therapy. Future studies incorporating FC can standardize patient assessment and help identify vulnerable population.

INTRODUCTION

It is estimated that 147,950 people will be diagnosed with colorectal cancer (CRC) in the United States in 2020.¹ Treatment for CRC may include surgery, combined modality therapy, or chemotherapy alone.^2,3 Both CRC and its treatment are associated with deficits such as malnutrition, fatigue, deconditioning, neuropathy, and cachexia, which have been shown to increase the risk of short-term morbidity and may be associated with increased mortality.⁴ Routine assessment of a patient's physical activity (PA) may provide insight into the patient's cardiovascular fitness and may predict tolerance of treatment and identify where supportive services can be used.⁵ PA can improve physical functioning, reduce anxiety, address chronic fatigue during and after cancer therapy, and improve a patient's physical and psychological ability to undergo treatment, and it correlates with cancer outcomes.^4,6-9 Performance status (PS), an assessment of a patient's level of function and capacity for self-care, is routinely assessed and is an important prognostic factor for survival.^10-13 The ability of Eastern Cooperative Oncology Group (ECOG) PS scale to predict chemotherapy tolerance has also been studied and validated, although this measure is subjective and prone to bias.^14,15 PA level is very closely related to PS such that patients who achieve higher levels of activity have better physical function and PS.^16-19 There is a degree of subjectivity and considerable interobserver variability in assessment of PS.¹² A study evaluating interobserver reliability for PS measurement using different scales found low absolute interobserver agreement rates, with an agreement rate of 38%-50% for PS among trained medical oncologists.²⁰ Hurria et al¹⁵ demonstrated the inadequacy of Karnofsky performance status measures to predict chemotherapy toxicity in a prospective study of older patients with cancer. Existing evidence highlights the limitations of contemporary PS scales and provides rationale for exploring objective markers of PA as a surrogate of PS.^12,15,21

CONTEXT

• Key Objective

• Objective measures of physical activity via fitness tracker (Fitbit Charge [FC]) may be an accurate and reproducible assessment of a patient's overall performance status (PS). We studied the feasibility of using an FC for evaluating the level of physical activity in patients with colorectal cancer (CRC).

• Knowledge Generated

• We found that using FC as a diagnostic tool was feasible and acceptable by patients with CRC. The level of activity correlated well with PS, and the baseline activity (steps per day) level was associated with toxicities (absolute incidence), but findings were not significant as this study was not designed to study it.

• Relevance

• In patients with CRC, objective tracking of activity is feasible. It can serve as an adjunct to PS to identify patients at risk for adverse outcomes and those who may benefit from prehabilitation efforts.

Wearable fitness tracking devices are increasingly popular and are of interest in assessing activity level among patients.²² These devices provide data that are independent of the patient's cognition, literacy, language, and health status, as well as observer bias. They can be used remotely to assess the patients' true function at home and provide a longitudinal view of daily activity that surpasses what can be assessed in the context of a clinic visit. Finally, they are low-cost tools that do not require extensive technical skills.²³ Modern activity trackers can provide a detailed log of daily steps, heart rate data, and active minutes. As an alternative tool for assessing physical functioning and possibly PS for patients with cancer, these devices may also predict treatment tolerance.²⁴ In the general population, studies have additionally shown improvements in the level of PA after introduction of such activity trackers.^25,26 However, these trackers have been used sparsely for evaluation of the activity level of patients with cancer with limited clinical utility at the current time. Furthermore, limited data are available regarding the utility of this assessment, either in prediction of treatment-related toxicities or in informing healthcare providers of patients' true functional status.^27-29

This study was designed to test the feasibility of using a fitness tracker (Fitbit Charge [FC]) to obtain PA data in patients with CRC undergoing therapy, with further evaluation of potential correlation between PA, physician-assessed PS, and treatment toxicity using a cutoff of 5,000 steps per day (SPD). The cutoff value of 5,000 SPD was chosen as it is exactly half of the widely recommended daily goal of 10,000 SPD,^30-32 has previously been used as a cutoff in the defining of a sedentary lifestyle,^32,33 and falls between previously reported mean SPD values for oncologic patients with the ECOG PS of 0 (5,911 SPD) and 1 (1,890 SPD).³⁴

MATERIALS AND METHODS

With institutional review board approval, patients with CRC at a National Cancer Institute designated comprehensive cancer center were prospectively enrolled. Eligible patients were of age 18 years or older, had a histologically confirmed diagnosis of colon or rectal cancer (stages II-IV) with an ECOG PS of 0-2, and were undergoing CRC treatment at our institution (Fig 1). Patients were assigned to the medical arm if they were starting new systemic chemotherapy (± 4 weeks), or the surgical arm if they planned to undergo curative resection, even if metastatic. Demographic data collected for each patient included sex, age, marital status, race and ethnicity, diagnosis, stage, and provider-assessed ECOG PS.

FIG 1.

CONSORT diagram of enrolled patients. FC, Fitbit Charge.

At baseline evaluation, providers recorded ECOG PS and patients were provided an FC HR to wear daily for 4 days and were instructed to wear it for as many hours as was comfortable. All participants subsequently underwent a second FC assessment of 4 days. This was done at the time of postoperative discharge in the surgical cohort and after 4 weeks (± 2 weeks) of initial assessment in the medical cohort. Data collection was managed using Fitabase, an online research platform that aggregates and allows for easy export of FC data from multiple users without direct user participation. Patients were not given access to the online platform and could only see what data were available on the device itself, which was limited to SPD for any 1 given day. Baseline and follow-up evaluation was set to 4 days as the battery life of FC is approximately 5 days, allowing for continuous wear over that period and account for 1 day of calibration. Patients were also not allowed to sync the FC with any internet-enabled device. This ensured no reliance or burden on patients to charge the device or transfer any data. Similarly, this avoided protected health information (PHI) from being handled by a third-party vendor (study staff preregistered patients using a non-PHI identifier) and helped to reduce cognitive bias (patient could not access their metrics).

Patients' mean SPD was calculated for days, and FC was worn ≥ 12 hours. Any postoperative complication (surgery patients) or ≥ grade 3 toxicity (medical patients) was considered toxicity of interest (TOI); the risk of TOI was stratified by the cutoff of 5,000 SPD. Their incidence was compared between patients who achieved an average of more or less than 5,000 SPD.

Participants completed an Acceptability Questionnaire after their baseline FC assessment. Items assessed patients' ease and enjoyment of the FC using a five-point Likert scale. Providers of participating patients also completed questionnaires at baseline and follow-up assessing patient's ECOG PS, diagnosis and stage, and a TOI assessment.

The primary objective of this study was to assess the feasibility of using the FC to track patients' PA. This was defined by at least 75% of patients complying with their instructions to wear the device for at least 12 hours a day on at least 3 out of 4 assigned days. Time wearing the device was assessed by the number of hours per day with heartbeat data available. Since this was a feasibility study, most secondary end points were descriptive. Activity measures, including SPD and active minutes, were summarized by arm and time point, with differences assessed using Wilcoxon tests. We categorized baseline average SPD as above or below our prespecified threshold of 5,000. We assessed the relationship between baseline steps and PS (0 v ≥ 1) using Fisher's exact test. Stratified by treatment type, we tested the relationship between baseline SPD and toxicities (medical) or complications (surgical) using Fisher's exact tests. We also assessed relationships between TOIs and changes in SPD via Wilcoxon tests.

RESULTS

Eighty patients were enrolled, 44 (55%) were male, with a median age of 59.5 years (range, 29-90 years), and 99% were ECOG PS 0 or 1. Most medical patients had metastatic disease (70%), whereas a majority (85%) of the surgical cohort had nonmetastatic disease. Few surgical patients with stage IV were undergoing resection with curative intent and hence were allowed to participate. Table 1 further highlights the baseline characteristics of participants.

TABLE 1.

Patient Characteristics (Unadjusted Values, n [%] Unless Otherwise Specified)

All patients were evaluated for the primary aim of feasibility. Overall, 68 (85%) patients met the feasibility criteria, with 35 (87.5%) medical and 33 (82.5%) surgical patients passing the threshold for feasibility (Table 2). This met our prespecified feasibility cutoff (defined by at least 75% of patients complying with their instructions to wear the device for at least 12 hours a day on at least 3 out of 4 assigned days). Three medical and one surgical patient never wore the device for > 12 hours in a day. Figure 2 shows the distribution of the length of time that participants wore the device, by study day. Many participants wore the device continuously (at or close to 24 hours), and the use of the device did not seem to decrease over time. Overall, FC testing was well received by patients. Of 85 patients approached to participate, only one declined participation in this study because of being overwhelmed with diagnosis and treatment. Almost all patients (91%) who participated in the study found the FC to be easy to wear and use. A majority (81%) reported that the device did not interfere with their daily activity. Few patients provided additional comments regarding barriers to use; however, three patients noted difficulty with the band length and comfort. For secondary analyses, we included 76 patients (37 medical and 39 surgical) who had available PA data for at least 1 day of 12 or more hours of FC usage. When analyzing these 76 patients by ECOG PS, those with a score of 0 had a mean SPD of 6,313 (122 active minutes) and those with a score of 1 had a mean SPD of 2,925 (54 active minutes) (P = .0003). The optimal SPD cutoff depicting ECOG PS 0 versus 1, using conditional inference trees, was calculated to be 4,236 SPD. This was very close to our prespecified cutoff of 5,000 SPD.

TABLE 2.

Feasibility and Activity Outcomes (Mean [SD] Unless Otherwise Specified)

FIG 2.

Distribution of the length of time participants wore Fitbit tracker, by study day.

We found that 12 medical patients experienced TOI compared with nine surgical patients. When evaluating SPD as a continuous variable, those who experienced a toxicity had a baseline average of 4,150 SPD compared with 5,660 SPD in those who did not. Stratifying by SPD, the absolute incidence of TOI was lower among the patients with SPD > 5,000 (7 of 33, 21%) compared with those with SPD < 5,000 (14 of 43, 32%), although this difference was not statistically significant (P = .31). Figure 3 contrasts the SPD-derived TOI distribution with ECOG PS–derived TOI distribution. We also analyzed change in SPD between the baseline and follow-up assessment in relation to TOI. In the surgical cohort, there was significant reduction in the mean SPD count between the pre- and postoperative assessment (−3,364 SPD, P < .001), as one would expect. This change was not observed in the medical cohort (− 473 SPD, P = .09) between the two time points. In both cohorts, the change in SPD was not associated with the rate of TOI (P > .05 for both).

FIG 3.

Incidence of TOIs by PS (A and B) and SPD (C and D) in medical and surgical cohorts (unadjusted values, n [%]). ECOG, Eastern Cooperative Oncology Group; PS, performance status; SPD, steps per day; TOI, toxicity of interest.

DISCUSSION

In patients with CRC, routine assessment of a patient's PA may provide useful information regarding the patient's fitness and capacity to tolerate therapy. We assessed feasibility of prospective activity tracking and ability to collect and analyze pooled data for objective assessment of PA. Our trial design required little to no patient interaction with the device once activated, and the results show that patients had minimal difficulty with its use. Baseline and follow-up evaluation set at 4 days allowed potential continuous use and data collection without dependence on patients to remove the device and charge it themselves during this period. We found our approach feasible in patients with CRC undergoing chemotherapy or cancer-related surgical interventions. Our results also illustrate the value of SPD as an adjunct to PS assessment and a possible tool to predict toxicities, regardless of type of therapy for CRC.

The potential integration of wearable FC with oncologic treatment monitoring represents a growing area of academic interest. Gupta et al performed a pilot study in 32 patients with advanced cancer, which found minimum steps/day correlated strongly with quality-of-life measures, and patients reported a positive experience with the devices (75%).³⁴ Van Blarigan et al³⁵ conducted a smaller randomized controlled trial where FC was used over a 12-week period following the completion of curative-intent treatment among patients with nonmetastatic CRC. Similar to our findings, feasibility, defined by wear time adherence ≥ 70% and ≤ 20% attrition prior to completion of the 12-week regimen, was confirmed. As this study incorporated an exercise-based intervention goal, the correlation of baseline activity with patient outcomes was not done.

The results of our study highlight several important points. It is well-known that clinician- and patient-reported activity levels are inaccurate with significant interobserver variability and are prone to bias. In one study, the self-reported activity level was 366% higher than that measured by activity monitors in patients undergoing chemotherapy.³⁶ As mentioned earlier, interobserver reliability for PS measurement is fair at best, with an agreement rate of 38%-50% among trained medical oncologists.²⁰ Activity trackers may provide a simple, quick, and reproducible measure of activity level.³⁷ Existing data confirm that PA and exercise have beneficial effects on health-related quality-of-life domains in patients with cancer and cancer survivors.^6-8 Among patients with metastatic CRC, PA was associated with improved progression-free survival and lower adjusted risk for ≥ grade 3 toxicities in the CALGB 80405 trial.³⁸ Although prospective data examining outcomes among patients with CRC following in-treatment intervention are lacking, the results of our and other studies demonstrate that such an investigation could be performed using digital fitness tracking.

With regard to our secondary end points, SPD correlated well with physician-assessed PS and the finding persisted in both the medical and surgical cohorts. Correlation of SPD and toxicities was not the primary aim of this study, and the results were encouraging, although not significant. For example, patients who experienced toxicities have a numerically lower SPD compared with those who did not. Interestingly, medical patients who experience toxicities showed a paradoxical increase in SPD count at follow-up assessment, but this difference was not significant and carries low clinical relevance. It might have been possible that patients who experienced TOI and related hospitalizations were prescribed physical therapy, leading to improved mobility in the hospital or as an outpatient. These variables or interventions were not captured as part of the study.

An important area of consideration when studying digital health technologies pertains to issues surrounding patient privacy and data security. Social media platforms, wearable FC, and apps to manage treatment toxicities all collect health data that can possibly be misused by a third-party vendor for advertising and other purposes. A recent study was able to reidentify 95% of patients, from accelerometer-measured PA data, which had geographic and PHI removed using machine learning techniques.³⁹ As digital health technologies are further integrated into health care, developing regulations to restrict the sharing of such data by device manufacturers is required. Also aggregating data can help overcome some of these concerns. Finally, informed consent documents need to appropriately reflect these concerns to participating patients. Similar concerns also influenced our decision to restrict our activity tracking to 4 days, circumventing the need for patients to create accounts and sync their devices to online applications. Study staff was able to create generic accounts for patients and store all patient-specific data on internal database (Research Electronic Data Capture).

Our study has unique strengths, including prospective design that allowed for contemporaneous collection of objective PA data through reliable digital technology with simultaneous comparison of provider-assessed PS. The inclusion of medical and surgical patients enhanced the generalizability of our results. The willingness of patients to adopt digital health technology is clearly demonstrated by this study. Patients did not create their own personal account in association with the FC and therefore allowed for confidentiality within the online, third-party–managed database. In addition, patients were unable to access their data beyond daily SPD for a specific day, and therefore, our results are unlikely to be confounded by patients desire to outperform a previous day.

Our study does have several limitations, including our single institution study may not be representative of the national population. As primarily a feasibility study, our trial design prescribed a relatively modest sample size, limiting statistical power for the toxicity correlation secondary end point. Our inclusion of patients with both metastatic and nonmetastatic diseases across both cohorts reduced the specificity of our results but does show feasibility across a wide spectrum of patients. We relied on data obtained over a short duration of time to limit patient burden and capability of data synching, but this prevented us from studying how change in PA level acts as predictor of treatment tolerance. Also, we had a paucity of ECOG 2 patients on this study limiting the extent of interpretation. In addition, given the scope of this study, we were unable to determine if activity prevents toxicities or if decreased activity is a sign of an impending toxicity and any such correlations carry a risk of attribution bias. Finally, although modern trackers (like the one used in this study) have the ability to track swimming, bicycling, and other exercise as active minutes, the step count is not accurately calibrated to reflect such activities and this may affect the accuracy of the step counts to some degree. Nonetheless, it provides objective data that can help with reliable assessment of patient's activity level.

In conclusion, objective PA monitoring using a wearable FC is feasible in patients with CRC undergoing both medical and surgical therapies, and SPD correlate with ECOG PS. A high rate of compliance can be expected, as most patients find these devices nonintrusive and easy to manage. Our results also illustrate the value of SPD as an adjunct to PS assessment as a possible tool to predict toxicities, regardless of type of therapy for CRC. Although validation in larger cohorts is needed, these findings provide rationale to study SPD in conjunction with PS for risk stratification and can possibly identify at-risk individuals who benefit from supportive services and prehabilitation efforts.

AUTHOR CONTRIBUTIONS

Conception and design: William H. Ward, Crystal S. Denlinger, Sanjay S. Reddy, Elin R. Sigurdson, Efrat Dotan, Matthew Zibelman, Joshua E. Meyer, Jeffrey M. Farma, Namrata Vijayvergia

Administrative support: Caitlin R. Meeker

Provision of study materials or patients: R. Katherine Alpaugh, Igor Astsaturov, Crystal S. Denlinger, Michael J. Hall, Elin R. Sigurdson, Namrata Vijayvergia

Collection and assembly of data: William H. Ward, Caitlin R. Meeker, Margret Einarson, R. Katherine Alpaugh, Igor Astsaturov, Crystal S. Denlinger, Michael J. Hall, Sanjay S. Reddy, Efrat Dotan, Jeffrey M. Farma, Namrata Vijayvergia

Data analysis and interpretation: William H. Ward, Caitlin R. Meeker, Elizabeth Handorf, Maureen V. Hill, Thomas L. Holden, Crystal S. Denlinger, Michael J. Hall, Sanjay S. Reddy, Efrat Dotan, Matthew Zibelman, Joshua E. Meyer

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/cci/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).

Elizabeth Handorf

Honoraria: NCCN/Pfizer

Research Funding: Pfizer, NCCN/Lilly

R. Katherine Alpaugh

Research Funding: Merck

Patents, Royalties, Other Intellectual Property: An inflammatory breast cancer cell line was established and is being provided by Fox Chase Cancer Center for research purposes. Royalties are received yearly dependent on the number of requests.

Igor Astsaturov

Honoraria: Caris Life Sciences

Consulting or Advisory Role: Caris Life Sciences, Guardant Health

Speakers' Bureau: Caris Life Sciences

Research Funding: Lantern Pharma

Crystal S. Denlinger

Consulting or Advisory Role: Merck, Bayer, Astellas Pharma, Exelixis, Lilly, Taiho Pharmaceutical, Bristol-Myers Squibb

Research Funding: Lilly, MedImmune, Bristol-Myers Squibb, AstraZeneca, Macrogenics, Agios, Array BioPharma, BeiGene, Amgen, Sanofi, Exelixis, Zymeworks, Genmab

Michael J. Hall

Research Funding: Invitae, Foundation Medicine, Caris Life Sciences, Myriad Genetics, AstraZeneca, Ambry Genetics

Patents, Royalties, Other Intellectual Property: I share a patent with several Fox Chase investigators for a novel method to investigate hereditary CRC genes

Travel, Accommodations, Expenses: Foundation Medicine, Caris Life Sciences, Myriad Genetics, AstraZeneca

Other Relationship: Myriad Genetics, Foundation Medicine, Invitae, Caris Life Sciences

Efrat Dotan

Honoraria: Pfizer, Boston Biomedical

Consulting or Advisory Role: ARMO BioSciences

Research Funding: Pfizer, Bayer, Incyte, Boston Biomedical, Merck, MedImmune, GlaxoSmithKline, Lilly, AstraZeneca

Matthew Zibelman

Honoraria: Pfizer

Consulting or Advisory Role: EMD Serono, Horizon Pharma, Pfizer, Janssen

Research Funding: Horizon Pharma, Bristol-Myers Squibb, Pfizer, Exelixis

Other Relationship: Association of Community Cancer Centers (ACCC)

Joshua E. Meyer

Honoraria: Varian Medical Systems, Sirtex Medical

Consulting or Advisory Role: Quantigic Genomics

Research Funding: Varian Medical Systems

Patents, Royalties, Other Intellectual Property: Patent PCT/US18/18110

Travel, Accommodations, Expenses: Varian Medical Systems, Sirtex Medical

Jeffrey M. Farma

Consulting or Advisory Role: Novartis, Castle Biosciences

Other Relationship: Delcath Systems

Namrata Vijayvergia

Consulting or Advisory Role: Lexicon, Sun Pharma HalioDx

Research Funding: Merck, Bayer

No other potential conflicts of interest were reported.