The study was approved by the medical research ethics committee of the Amsterdam University Medical Centers, location Vrije Universiteit Medical Center (registration number 2018/112). All patients provided written informed consent.

A feasibility study with a mixed methods design was performed between April 2019 and April 2020 in patients with gastrointestinal or lung cancer scheduled for surgery. The formal sample size was not calculated. Instead, a convenience sample with a 1-year inclusion period was chosen. Self-monitoring of PA after hospital discharge was evaluated using a questionnaire and interviews conducted in April 2020, and the study procedures were evaluated using administrative data during the course of the study.

The inclusion criteria were adult patients with gastrointestinal or lung cancer who were invited for preoperative physiotherapy screening between April 2019 and April 2020 at the outpatient clinic of our tertiary teaching hospital (Amsterdam University Medical Center, location Vrije Universiteit Medical Center), which included patients scheduled for a lobectomy, esophageal resection, or hyperthermic intraperitoneal chemotherapy (HIPEC). Exclusion criteria were <7 days between inclusion and surgery, emergency procedures, patients who are nonambulatory, and no access to or not able to use a smartphone or tablet.

Potentially eligible patients were informed about the study by the treating physiotherapist during the preoperative consultation. Patients who were willing and eligible to participate received the Physical Activity Monitor (PAM) AM400 3-axis accelerometer (Pam BV Doorwerth) and were given access to the corresponding smartphone app called Atris (Peercode BV, Geldermalsen; Figures 1 and 2). The PAM was selected for this study as (1) the PAM AM400 was found to be a suitable movement sensor to validly measure activity minutes [19]; (2) the battery of the PAM lasts for approximately 1 year, eliminating the need for patients to recharge the device; (3) the data of the PAM can be synced directly to a web-based application, enabling remote monitoring by clinicians; and (4) the PAM can be worn around the ankle and is waterproof to allow 24/7 wearing. The PAM measures PA continuously and provides the total PA every 15 minutes. With the Atris app, patients were able to self-monitor their daily PA levels and received feedback on the number of active minutes per day. Patients were able to set personal activity goals in the app by themselves.

All patients received the usual pre- and postoperative physiotherapy care. During the standard preoperative consultation, potential risk factors (eg, smoking or sedentary lifestyle) for delayed postoperative recovery were identified, leading to personalized advice for improving preoperative physical fitness. In addition, the patients were given instructions about the PAM and Atris app. The physiotherapist informed the patient that the app provided insights into the recovery of PA and advised them to use the app to resume their daily activities after surgery. Patients were asked to start wearing the PAM 24 hours a day in a strap around the ankle for at least 7 days before surgery to familiarize themselves with its use.

During hospitalization, patients received standard physiotherapy consultations and were stimulated to mobilize according to a daily mobilization goal following the ERAS protocol [20]. In addition, the physiotherapist guided the patients in using the PAM and Atris app.

Personalized rehabilitation recommendations were provided at discharge. If indicated by the physiotherapist, the patients were advised to continue physiotherapy in primary care after discharge. For this study, the patients were asked to wear the PAM 24 hours a day for 6 weeks. A period of 6 weeks was chosen as it was expected that patients would be able to gain sufficient experience using the PAM and Atris app to assess feasibility. A hospital physiotherapist provided insights into the activity levels of the patient on the corresponding web application and monitored the activity data of the patients weekly. In case the activity levels decreased or no data were available, the physiotherapist contacted the patient. In the case of technical problems, the physiotherapist helped resolve them if possible. In cases where patients had problems resuming their PA level by themselves, the physiotherapist advised them to contact a physiotherapist in primary care. The patients also had the opportunity to contact the physiotherapist themselves.

To evaluate the feasibility of self-monitoring of PA using the PAM and Atris app, we set feasibility criteria a priori based on cutoff points described in previous studies (Textbox 1) [21-25].

To better understand the feasibility, additional qualitative data on acceptability, satisfaction, and experienced additional value were collected to explore user experiences.

SPSS (version 26; IBM Corp) was used for quantitative data analysis. Study population characteristics were presented descriptively as mean (SD), median (IQR), and percentage. Quantitative data were analyzed descriptively. The study procedures were presented as percentages. PAM data were considered missing if they were not available for ≥3 days in a given week. The available PAM data are presented as the median (IQR) of the total active minutes per day of each week and as a percentage of the preoperative PA baseline level. The mean (SD) SUS was calculated using the method described in the study by Brooke [21]. Additional questions about user experience are presented as percentages.

The research software ATLAS.ti (version 8) was used for qualitative analysis. Qualitative data analysis was performed following the steps of thematic analysis by 2 researchers (MvdL and MB) [26]. The interviews were read several times to familiarize with the data. Data were open coded line by line to segment them into the initial codes. Axial coding was used to define the definitive codes. Definitive codes were classified and described under different themes [26].

A total of 90 potentially eligible patients were invited to participate between April 2019 and April 2020, of whom 68 (76%) were willing to participate. After the final eligibility check, of the 90 patients, 57 (63%) were included in the study, resulting in a participation rate of 63%. The retention rate was 72% (41/57); 28% (16/57) of patients dropped out during the study, of whom 44% (7/16) were related to the intervention. The reasons for nonparticipation, exclusion, and dropout are presented in the flowchart (Figure 3). Ultimately, 41 patients were included in the analysis. The median age of the patients in this study was 68 (IQR 60-73) years, and 58% (25/41) were male. The most common types of surgery were lobectomy (23/41, 56%) and HIPEC (10/41, 24%). The median length of hospital stay was 7 (IQR 6-11) days. Other relevant demographic and clinical data are presented in Table 1.

Approximately, 31% of the PAM data were missing. Missing value analysis suggested that the data were missing at random, as missing data increased during the time blocks of the postoperative phase. The amount of missing data increased during the 6 postoperative weeks: 27% of the PAM data were missing in the first postoperative week and 44% in the sixth postoperative week. Figure 4 shows an overview of the reasons for the missing data. The most common reasons for missing data were technical problems or withdrawal of wearing the PAM. Table 2 shows the median preoperative and postoperative PA levels in minutes per day and the median percentage of recovery in PA compared with the preoperative levels.

Self-monitoring of PA after discharge appears to be feasible based on good system usability and predominantly positive user experiences in patients with cancer after lobectomy, esophageal resection, or HIPEC. These findings are consistent with those of other studies [15-17,22]. Wu et al [15] found good feasibility of self-monitoring using a wrist-worn accelerometer and an app in patients after gastric cancer surgery. Low et al [17] reported good usability of a real-time mobile technology–based sedentary behavior intervention for patients with abdominal cancer in the perioperative period using a smartwatch. However, feasibility was considered moderate in that study as adherence to wearing the smartwatch decreased significantly from before to after the surgery. In our study, adherence also seemed to decrease based on an increase in missing data during the intervention period. Solving technical problems and improving the comfort of the ankle strap may reduce the number of dropouts and missing data in clinical use and follow-up studies. In addition, improving self-efficacy and self-motivation and engaging in more social support could enhance user adherence, as suggested in a systematic review of predictors of adherence in home-based physical rehabilitation [27].

The system usability of the PAM and Atris app is similar to that of devices used in other studies [17,25]. Jonker et al [25] reported good system usability (mean SUS 73.1) for a wrist-worn activity tracker and mobile app in older adult patients after oncological surgery. In a study by Low et al [17], the system usability of a Fitbit smartwatch with an accompanying smartphone app during the perioperative period in patients scheduled for abdominal cancer surgery was also found to be good (mean SUS 83.8). The qualitative data, in addition to the quantitative data, provided insights into the facilitators of and barriers to the use of the Atris app and PAM. The user experiences were largely positive. The interviewed patients mentioned that the PAM and Atris app were easy to use, motivated them to be more physically active, and provided support after discharge. However, most of the patients recommended the design of a more comfortable ankle strap, and some were annoyed about technical problems. Only a few patients did not experience the added value of the PAM and Atris app as, in their opinion, they were already sufficiently active and, therefore, did not feel the need for additional support. In contrast, some other patients indicated the need for more support, such as through occasional telephone contact with a physiotherapist or motivational messages. Therefore, the tailoring of interventions to individual needs and preferences should be considered.

In this study, we explored the course of recovery in PA in the first 6 weeks after surgery. These results may be supportive in clinical practice to gain more objective insights into patient recovery and identify which patients may need more support in improving their PA levels. We found that most (19/25, 76%) of our study population did not return to baseline PA levels 6 weeks after surgery, although these results should be interpreted with caution, given the relatively large amount of missing PA data. In previous observational pilot studies using objective PA data after (cancer) surgery, most patients did not reach preoperative PA levels even at 3 months after surgery [28,29]. Similarly, a study using questionnaires to investigate the course of recovery in physical functioning 6 and 12 weeks after lung cancer surgery showed that patients were still recovering between 6 and 12 weeks after surgery [4]. The patients in our study underwent major surgeries, including HIPEC and major lung resections. These procedures are both associated with prolonged functional recovery compared with less invasive procedures such as minimally invasive segmental colectomies or video-assisted small lung resections. Therefore, it is suggested that for most of these patients, the period of supportive care to improve PA should be >6 weeks after surgery. However, to increase user adherence for longer-term use, the previously mentioned improvements to the ankle bracelet and resolution of technical problems are necessary. In addition, to better understand all dimensions of user adherence, an in-depth analysis of adherence to ambulant monitoring in this patient population should be performed, taking into account the 5 dimensions of adherence as described by the World Health Organization [30].

By conducting this feasibility study, barriers and enablers were identified for the use of the PAM and Atris app after hospital discharge in patients after cancer surgery. However, proper technical functioning and comfort in wearing are important prerequisites for all activity trackers. Moreover, the enablers found in our study, for example, that it motivates patients to be more physically active and that it provides more insight into PA recovery, are also generalizable to other activity trackers. The cost-effectiveness and effectiveness of interventions using PA self-monitoring during cancer treatment is largely unknown, and the conduct of randomized clinical trials is warranted [31]. In addition, not all patients seem to require the same amount of postoperative support. Further research should take into account the risk of functional decline after surgery, as well as the needs and preferences of individual patients.

This study had some limitations. First, this feasibility study was conducted in a single hospital setting. In addition, not all diagnoses within gastrointestinal and lung cancer surgery were represented in our study population as preoperative physiotherapy screening was not part of the care pathway for all patients in our hospital. Therefore, the results of this study cannot be generalized to other diagnoses. Second, all patients were contacted to ask whether they were willing to participate in an interview after PA data collection had already ended. However, some patients did not respond or were unwilling to participate in the interviews. This could have caused a selection bias. To reduce selection bias among the patients interviewed, they were selected based on whether they perceived the PAM and Atris app as adding value. However, the selected interviewees had a somewhat higher mean SUS score than that of the entire study population (83.8 vs 77.3). Thus, this approach probably did not sufficiently eliminate the selection bias. To gain a full understanding of feasibility, future studies should also interview nonparticipants. In addition, data saturation may not have been achieved as the interviews were conducted in a small and partly selective sample. Third, for some patients, there was a period of several months between the end of the self-monitoring period and the completion of the questionnaires (and interviews), which may have led to recall bias. Finally, as the study was conducted in the usual care setting, the perioperative instructions from the physiotherapist were not strictly followed as per protocol, which hindered reproducibility. Our research group is currently working on a protocol to guide physiotherapists in using this intervention.

The results of our study showed good system usability and predominantly positive user experiences in patients with cancer after lobectomy, esophageal resection, or HIPEC. Most patients mentioned that the PAM and Atris app motivated them to be more physically active after discharge. The retention rate and amount of missing data need to be improved in follow-up studies. Solving technical problems and improving the comfort of ankle straps may enhance user adherence, thereby reducing the number of dropouts and missing data. Randomized clinical trials should be conducted to investigate whether interventions using accelerometers indeed improve the recovery of PA and physical functioning after surgery in this population.