We conducted a single-arm pilot study at the University of Rochester Medical Center (ClinicalTrials.gov identifier: NCT04035499) to evaluate the feasibility of a mobile health exercise intervention, Geriatric Oncology-Exercise for Cancer Patients (GO-EXCAP), for older adults with MNs. Patients were screened and recruited between February 2020 and July 2021. Eligible patients were approached after confirming with their treating oncology team. Eligibility criteria were (1) age of 60 years and older; (2) being English speakers; (3) diagnosis of an MN, such as AML, MDS, or MDS and myeloproliferative neoplasm overlap; (4) undergoing outpatient chemotherapy; (5) a physician-verified Eastern Cooperative Oncology Group (ECOG) performance status of 0 to 2; (6) ability to walk 4 m; and (7) capacity to provide informed consent. Individuals were excluded from the study if they (1) had medical contraindications to exercise as determined by their treating oncologists, (2) had a low platelet count on the most recent complete blood count (ie, ≤10,000 platelets per microliter), and (3) did not receive platelet transfusion (eg, due to platelet refractoriness). Findings from the primary study have been previously reported [11].

This study is a secondary analysis of data from the parent study [11]. We report this analysis following the TREND (Transparent Reporting of Evaluations With Nonrandomized Designs) guidelines for nonrandomized evaluations of behavioral and public health interventions (Table S1 in Multimedia Appendix 1).

GO-EXCAP is a mobile health exercise intervention developed to promote successful behavior change by using evidence-based approaches to improve exercise self-efficacy (ie, verbal persuasion, social modeling, imaginal experiences, and provision of regular monitoring and feedback on mastery as well as emotional and physiological states).

GO-EXCAP integrates the registered and copyrighted home-based exercise intervention Exercise for Cancer Patients (EXCAP) with the GO-EXCAP mobile app. The EXCAP intervention was designed and delivered by American College of Sports Medicine–certified clinical exercise physiologists and has been tested in multiple studies [12-14]. It consists of progressive aerobic walking and resistance band exercise prescribed by a clinical exercise physiologist during a one-on-one in-person teaching session. The exercise physiologist generated individually tailored exercise prescriptions for each participant, increasing their baseline daily steps weekly by 5% to 20%. Between 5% and 20% weekly increase in steps would allow individuals classified as sedentary (ie, <5000 steps) to progress toward or achieve an active classification (ie, ≥10,000 steps) over an 8- to 12-week intervention period. The 5% to 20% step progression rate was individualized based on three key criteria: (1) average daily step counts at baseline, (2) participants’ physical capability and perceived readiness, and (3) weekly performance during the intervention. The strategy used to determine the weekly progression rate was described in Multimedia Appendix 2. For resistance band exercise, participants also received individualized prescriptions based on their baseline fitness level. The protocol began with 1 set of 8 to 15 repetitions for 16 exercises, using color-coded resistance bands appropriate for individual strength. Under the instruction of exercise physiologists, participants progressed weekly by increasing sets and repetitions until reaching 4 sets of 15 repetitions. Upon reaching this milestone, participants transitioned to a heavier resistance band and repeated the progression. Throughout the intervention, participants were instructed to perform both aerobic walking and resistance band exercises at a low to moderate intensity, corresponding to a rating of perceived exertion (RPE) of 3 to 5 on a scale from 1 to 10 (1=“not tired at all”; 10=“extremely tired”) [15]. We provided participants with a GO-EXCAP kit containing a printed manual, an activity tracker to record daily steps (Garmin Forerunner 35), 3 therapeutic bands with resistance levels corresponding to 2 to 14 pounds, and a tablet preloaded with the GO-EXCAP app. The exercise physiologist monitored participant-entered exercise data on the app and adjusted prescriptions as needed.

The EXCAP intervention lasted 6 weeks in prior trials [12-14]. Because chemotherapy for MNs is typically 4 weeks per cycle, the intervention duration was extended to align with 2 chemotherapy cycles. Given that chemotherapy delays are common in clinical practice, the GO-EXCAP intervention lasted 8 to 12 weeks, or 2 chemotherapy cycles.

Participants with gait outcomes available at both baseline and the postintervention time point were included in the analyses. To evaluate potential selection bias, we compared baseline characteristics between participants included in this analysis and those who were not. For participants included in this analysis, we described their characteristics, exercise adherence, and gait outcomes. Depending on the data distribution, we used paired 2-tailed t tests or Wilcoxon signed-rank tests to evaluate within-subject changes in each gait outcome from baseline to the postintervention time point. We also calculated Cohen d effect sizes (ESs) to quantify changes from baseline to the postintervention time point (small effect: 0.20 to <0.50; moderate effect: ≥0.50 to <0.80; large effect: ≥0.80) [19]. Spearman correlation analyses were conducted to assess the correlations of changes in daily steps and the duration of resistance band exercise with changes in gait outcomes. Individual data were presented in scatterplots to detect potential influential observations. In sensitivity analysis, we repeated the Spearman correlation analyses after excluding one observation that was more than 1.5 IQRs below the first quartile and did not follow the general trend of the rest of the data.

Given the exploratory nature of this analysis and the small sample size, no imputation methods were applied. A complete-case approach was used for all analyses. We performed hypothesis testing at an α value of .10 (2-tailed). All analyses were conducted using SAS (version 9.4; SAS Institute).

The parent study was approved by the Research Subjects Review Board of the University of Rochester (study ID: STUDY00003945), and all participants provided informed consent. To ensure participant privacy and confidentiality, all data were deidentified and managed via REDCap (Research Electronic Data Capture; Vanderbilt University), a secure web application accessible only to authorized research personnel. Participants received US $50 in compensation after completing all baseline and postintervention activities. The original ethics approval covers secondary analysis, and thus, no additional consent was required for this study.

We included 13 participants in this analysis (Figure 2). Of the 25 participants enrolled, 3 (12%) had missing baseline gait data due to logistical issues (eg, Bluetooth connection error). Of the 22 participants with available baseline gait data, 13 (59.1%) completed postintervention gait measures. Reasons for missing postintervention gait data included missed in-person appointments with the exercise physiologist due to health decline (3/22, 13.6%), being unsafe to complete walking (1/22, 4.5%), withdrawal (1/22, 4.5%), death (1/22, 4.5%), G-Studio processing errors (2/22, 9.1%), and unknown reasons (1/22, 4.5%). Comparison of baseline characteristics between participants included in the analysis and those who were not did not reveal substantial differences (Table S1 in Multimedia Appendix 3).

Table 2 describes the characteristics and exercise adherence of the 13 participants with both baseline and postintervention gait data. These participants had a mean age of 71 (SD 4.8) years, and most were White (n=12, 92.3%) and non-Hispanic (n=13, 100%) individuals. More than half were male (n=8, 61.5%), married (n=9, 69.2%), and college graduates or of higher educational levels (n=7, 53.8%). Most participants were diagnosed with AML (n=7, 53.8%) or MDS (n=5, 38.5%). These 13 participants had a mean of 3084 (SD 1765.5) daily steps at baseline and 3757 (SD 2623.6) daily steps at the postintervention time point. On average, they performed resistance band exercises for 4 days per week, 25 minutes per day, with a mean RPE of 3/10 (SD 1.3). No adverse events were reported related to the intervention [11]. Overall, this intervention demonstrated excellent tolerability among older adults with MNs.

Table 3 describes gait outcomes at baseline and the postintervention time point and changes in these measures. At baseline, participants walked an average of 408.0 (SD 128.8) m in the 6-minute walk test, with a cadence of 110.9 (SD 12.4) steps per minute and a velocity of 82.0 (SD 21.8) meters per minute. The mean GSI was 95.1% (SD 3%). Stance duration accounted for an average of 59.9% (SD 1.9%) of the gait cycle, with a mean stance duration variability of 0.8% (SD 0.9%).

From baseline to the postintervention time point, we observed a numerical increase in average cadence (mean +4.6, SD 14.6 steps per minute; P=.24; ES=0.38) and velocity (mean +1.6, SD 8.2 meters per minute; P=.24; ES=0.07). We also observed a numerical increase in stance duration (mean +0.4%, SD 2.3%; P=.55; ES=0.20) and a numerical decrease in swing duration (mean −0.4%, SD 2.3%; P=.55; ES=0.20), which together brought stance and swing durations closer to their optimal ranges (Table 3). In terms of regularity outcomes, we observed a numerical decrease in variability for both normalized stride (mean −1.4%, SD 8.5%; P=.34; ES=−0.18) and swing duration (mean −0.1%, SD 1.1%; P=.54; ES=−0.15). None of these outcomes were statistically significant.

Table 4 presents the correlations between changes in daily steps and duration of resistance band exercise with changes in gait outcomes. Increased daily steps significantly correlated with decreased swing duration variability (r=−0.72; P=.01). Duration of resistance band exercise significantly correlated with increased distance covered in the 6-minute walk test (r=0.62; P=.03), cadence (r=0.54; P=.06), velocity (r=0.56; P=.05), and stance duration (r=0.58; P=.04). We did not observe significant correlations for other gait outcomes.

Scatterplots showing the relationship between changes in gait outcomes and changes in exercise are shown in Figures S1 and S2 in Multimedia Appendix 3. In sensitivity analysis excluding one potential influential observation (Table S2 in Multimedia Appendix 3), the correlations remained largely consistent. Increased daily steps continued to be correlated with reduced swing duration variability (r=−0.64; P=.02). Similarly, longer resistance band exercise duration was correlated with improved walking distance (r=0.59; P=.04) and velocity (r=0.54; P=.07). Correlations of resistance band exercise duration with cadence (r=0.45; P=.14) and stance duration (r=0.50; P=.10) remained in the same direction.