Should Fitness Trackers Be Used to Assess Performance Status in Patients With Cancer?

Gillian Gresham, PhD

Gillian Gresham, PhD

Using wearable activity monitors may eventually supplement standard assessments of performance status (PS) and functionality that could inform clinicians, especially because objective evaluation of PS is difficult to determine. Patients spend most of their time outside of the clinic, self-report to providers, and undergo changes throughout treatment. Findings from a recent study at Cedars-Sinai Medical Center (CSMC) in Los Angeles, California, demonstrated the feasibility of using these wrist-worn devices to correlate with Eastern Cooperative Oncology Group (ECOG) PS and Karnofsky PS (KPS) scales.1 PS is affected by bias, resulting in patient activity being over- or underreported, which can affect short- and long-term treatment plans and clinical trial eligibility.

“Before we can ask if these devices can change practice or replace these standard assessments, we need to determine if it’s feasible for the patient to use them, [and] to wear them, and if there are any challenges with their use,” said Gillian Gresham, PhD, lead author and postdoctoral student at CSMC. “After establishing short-term feasibility, we can begin to explore whether these measurements correlate with standard assessments, patient-reported outcomes, and predict clinical outcomes,” she added.

Additional objectives of the study included measuring patient reported outcomes (PROs). “We were able to correlate PROs such as improved physical functioning, better pain tolerance, improved sleep, and lower levels of depression with increased activity, as measured using the devices,” Gresham said.

Thirty-seven patients (20 men) with stage IV or unresectable advanced stage III cancer agreed to participate in the single-center, single-cohort study that evaluated the Fitbit Charge HR device to measure daily activity. Patients of varying ECOG PS and KPS ratings participated, with participants agreeing to wear the Fitbit for 3 consecutive clinic visits over 2 weeks, in which ECOG PS and KPS were assessed. Associations between metrics (steps, distance, and stair climbing) and PS, clinical outcomes (adverse events [AEs], hospitalizations, and survival), and PROs were determined.

Heat Map of Average Activity Intensity for Each Patient Over a 24-hour Period, as Measured by a Wearable Activity Monitor and Sorted by ECOG PS Categories

Median age was reported as 62 years (range, 34-81). At baseline, patients’ ECOG-PS scores were 0 (24%), 1 (35%), 2 (24%), or 3 (16%). The majority of patients were diagnosed with gastrointestinal cancer (n = 27) and had stage IV disease (n = 34). There were 2 patients with locally advanced stage IV pancreatic disease and 1 patient with stage IIIB endocervical serous carcinoma.

The highest correlations were observed between average daily steps and both PS scores. Each 1000 steps/day increase was associated with reduced odds for AEs (odds ratio [OR], 0.34; 95% CI, 0.13-0.94), hospitalizations (OR, 0.21; 95% CI, 0.56-0.79), and hazard for death (hazard ratio [HR], 0.48; 95% CI, 0.28-0.83). On average, patients walked about 3700 steps, or 1.7 miles, a day, climbed 3 flights of stairs daily, and slept 8 hours/night as measured by their wearable device (Figure).

Although the device doesn’t measure heart rate the same way electrocardiograms do, it estimates heart rate by measuring blood flow through pulse readings. “We lined up measurements of the heart rate at a specific time in the clinic and matched it during the time that the patients were wearing the Fitbit and looked at the comparison of those 2 findings for each clinic visit,” said Gresham. “They were very closely related. That was encouraging but warrants further investigation.”

The investigators were, however, intrigued by the poor correlation between sleep quality and duration (as measured with the device), said Gresham. “I think we need more granular-level data with regards to sleep,” Gresham said. “[Although] Fitbit has improved its sleep recording quality since we completed the study, it opens up a new and interesting facet of research,” she said. “It also highlights the importance of the patient’s voice in their management, and is an example of how a patient’s perception of their sleep quality may not always match what is measured. Perhaps in this case, quality does not mean quantity.”

Using the device did allow patients to engage more with their physicians, according to the authors. Patients could discuss symptoms, provide information about their level of activity, and ask questions that providers don’t usually get a chance to talk about with the patient. “It’s a way to engage both with the patient and providers—something that I think is highly valuable in patient care,” added Gresham.

The device data also revealed interesting details about the patients’ daily lives that may not always be apparent during clinic visits, according to Gresham. For example, an older patient appeared frail during clinic visits, and scored a lower PS rating. The investigators found that after a week of wearing the Fitbit, the patient averaged almost 15,000 steps a day. “It turns out, this particular patient was probably one of the most active patients we followed…and they didn’t just start walking more steps because of the study…they said it was part of their daily routine. These observations suggest that the use of devices can help inform and even change the course of treatment decisions and importantly, help inform treatment decisions in the future. It also reminds us that assessments should not be made based solely on age or physical appearance, but on a patient’s overall activity levels and functionality.”

The data collected by the devices can be used to fill in the gaps between clinic visits, said Gresham. “Throughout their care, we’re seeing that performance status assessments tend to be rather static and are often missing from the patient medical chart. We can now detect changes in real time by anticipating potential [adverse] effects or try to intervene a little bit sooner if complications or adverse events occur. It also allows for the patient to self-monitor themselves and notify their care team if there are changes to their usual activity levels.”

Data were collected through the device’s online dashboard, and each patient was provided an anonymous, de-identified email account that was used to register the Fitbit, access their activity data, and receive email summaries of weekly activity levels. Investigators observed a range of attitudes towards the devices. Some patients “explored the technological abilities of the device, and others just wore it like a watch,” Gresham said.

The data collected by devices like these could give oncology fellows access to information that they otherwise would not have had before, said Gresham. “Currently, along with the enterprise information services department here at CSMC, we’re working to integrate the wearable data into the patient’s medical chart,” she said. “The more knowledge that can be gathered about the patient beyond the clinic assessment could give the fellow an idea about the patient’s daily routine, levels of activity, and the patient’s standard for normal.”

In the long term, the data can be tracked from initial diagnosis, treatment, potential cure, and recurrence. The fellow has a way to monitor the patient, especially for patients who might have difficulty going to the clinic for various reasons such as socioeconomic or reliable transportation, she added.

The findings provide new information regarding the use of an emerging technology in cancer clinical settings. The incorporation of wearable activity monitors to correlate with functional and clinical outcomes are only beginning to be reported, according to the authors.

It is not clear whether activity monitor data can replace ECOG-PS or KPS assessments, but they can provide supplemental data for current functionality tools. Not only were step counts and other activity metrics correlated with PS, but they also provided a more detailed and continuous account of patients’ activity levels with the added benefit of being recorded in the patients’ free-living environments.

In the future, Gresham said the investigators would like to see if the data could be used as stand-alone functional outcomes to evaluate the impact of a particular intervention or treatment on the patient’s daily activity. “Like patient reported physical function and activities of daily living, frailty status, or provider assessed performance status, we would like to see ‘activity,’ as something clinicians can use to assess a patient’s functionality. Maybe some day in the future, we can incorporate physical activity recommendations or ‘activity prescriptions’ into the routine care of patients with the goal of improving health outcomes,” she said.

Gillian Gresham completed this work under the supervision of Steven Piantadosi, MD,PhD, Curtis Meinert, PhD, and Arvind Shinde, MD, MBA, MPH (senior author of the paper).

Gresham G, Hendifar AE, Spiegel B, et al. Wearable activity monitors to assess performance status and predict clinical outcomes in advanced cancer patients. Digital Medicine. 2018;1(27):1-8. doi:10.1038/s41746-018-0032-6.