Abstract
The strongest risk factor for readmission to the hospital is impaired physical function. We sought to determine the usability and feasibility of a post-hospital behavioral intervention to improve mobility in older adults with significant morbidity and functional impairment. We conducted a two-arm pragmatic pilot randomized trial of a behavioral economics-informed intervention to increase daily steps in Veterans, age 60 or older, receiving home health (HH) services post-discharge. The intervention group received a multicomponent behavioral economics-informed intervention, including daily step count goals, performance feedback, social incentives, and a pedometer. The control group received usual care, which included the use of a pedometer without additional interventions. Both groups wore an ActivPAL device to continuously monitor activity. Outcomes were measured at 60 days post-discharge. The primary outcomes were feasibility (enrollment and completion rates) and usability (device compliance and participant satisfaction). Secondary outcomes included changes in mean daily steps and hospital utilization. Differences in daily steps between the intervention and control group were examined using a linear mixed effects model. Sixteen out of the 37 consented Veterans completed the study (9 intervention, 7 control). All Veterans rated as very satisfied with wearing the pedometer, while 3 did not like wearing the ActivPAL. The pedometer data was more complete, with a missing step data rate of 5%, compared to 29% for the ActivPAL. The median (interquartile range, IQR) baseline step counts were 926 steps (2744) in the control arm and 1131 steps (2952) in the intervention arm. Both groups increased steps during the intervention; however, there was no significant difference between groups (P = .18). Few older adults were able to complete the study, suggesting improvements to feasibility and acceptability are needed. Step counts were very low but did improve in both groups during the intervention. A pedometer was preferred by participants and provided more complete information than a research-grade device. Larger studies are needed to evaluate efficacy of such interventions.
Introduction
Functional impairment at hospital discharge is the strongest risk factor for readmission to the hospital and inability to return to the community.1,2 While in-hospital interventions to improve mobility have demonstrated preliminary effectiveness, few interventions are known to improve mobility in older adults after hospital discharge.3,4 Proven in-hospital interventions include structured mobility programs, early mobilization protocols, and regular physical therapy sessions, all administered within the hospital by medical professionals. However, these interventions are limited in their duration and often challenging to sustain after discharge due to shorter hospital stays and increasing medical acuity of inpatients.
Mobility interventions harnessing the principles of behavioral economics (e.g., social incentives, pre-commitment, feedback, and fresh start effect) have demonstrated effectiveness in increasing daily steps in a variety of clinical populations.5 -7 However, these trials have generally enrolled younger, less comorbid, and more mobile cohorts at baseline than what is typically observed in older adults receiving post-acute care.5,7
Given the promise of these behavioral interventions and the risk to older adults related to hospital deconditioning, we sought to develop a post-hospital intervention informed by behavioral economics to improve mobility. Our post-hospital intervention was built on the foundation of previous studies demonstrating the effectiveness of using behavioral economic principles to increase daily steps in clinical populations. This prior work provided the framework and evidence base for developing our current intervention. 7 As a first step, we sought to understand the usability and feasibility of such a behavioral intervention. Specifically, we sought to enroll older adults receiving home health (HH) post-hospitalization, using HH as a proxy for medical comorbidity and/or functional impairment. We also sought to establish baseline step counts and trajectories over time, and to provide baseline knowledge regarding recruitment, device monitoring, and step changes, to inform future large-scale trials.
Materials and Methods
Study Design
The study was a 2-arm, 60-day pilot pragmatic randomized trial of a behavioral intervention to increase daily steps in Veterans receiving home health (HH) services following discharge from an acute hospitalization. Data were collected between February 11, 2021, and April 20, 2022. The protocol was approved by the local Institutional Review Board (IRB). All participants provided written informed consent, our consent form was reviewed and approved by the local IRB/ethics committee.
EQUATOR Guidelines Compliance
This study adheres to the CONSORT (Consolidated Standards of Reporting Trials) guidelines for randomized controlled trials. We have completed the CONSORT checklist and included it as a Supplemental file.
Sample
Veterans were screened at daily interdisciplinary rounds at the Corporal Michael J. Crescenz VA Medical Center, in Philadelphia, PA, USA, to identify those who might be eligible for enrollment prior to discharge. Our study exclusively focused on a veteran population due to the distinct healthcare needs and challenges faced by this group, which differ significantly from those of the general population. We included Veterans, 60 years of age or older, discharged home with HH services after an acute hospitalization, and had access to a phone with texting capabilities. HH services are often prescribed to individuals with significant medical comorbidities or functional impairments, helping us identify a population at higher risk of reduced mobility and hospital readmission. Veterans were excluded if they had significant cognitive impairment that would prevent them from being able to consent, were non-ambulatory prior to the hospitalization, had medical conditions that would prevent participation in rehabilitation activities in the opinion of the treating clinician, or were positive for COVID-19 acute infection. We identified the age, race and ethnicity, gender, Care Assessment Needs (CAN) score, and inpatient unit (surgical or medical) of the Veteran using the electronic health record. The CAN score is a risk score calculated for all enrolled Veterans in the VA that predicts 1-year hospitalization or death on a percentile scale of 0 to 99, with 99 representing a risk that is at the 99th (highest) percentile of all VA enrollees.
Randomization and Study Procedures
Veterans who were identified as eligible for the study were approached prior to hospital discharge by a member of the research team to discuss the study. Interested Veterans then provided written informed consent and were issued 2 wearable devices.
Both groups received an activPAL4TM monitor (PAL Technologies Ltd., Glasgow, UK), a small device attached to the thigh using a waterproof medical bandage. The ActivPAL is a research-grade accelerometer and gyroscope that accurately measures lying/sitting time, standing time, and steps 24 h/day.8 -10 In addition, both groups received a pedometer (Accusplit AX2710) to be worn on the hip. We elected to use both devices because they provide a trade-off in terms of potential accuracy and burden. The ActivPAL is a research-grade device but requires attaching to the body and removing to be recharged, while the pedometer is easy to wear on the hip, but its accuracy compared to the ActivPAL was unclear. 8 We elected not to use FitBit or other automatically-syncing devices because wrist-worn devices may substantially underestimate steps in older adults who use an assistive device that limits arm swing while walking. 8
After the ActivPAL devices were returned, data was downloaded for processing with ActivPAL software (PALconnnect, PALanalysis, PALbatch). Participants were instructed on how to use and report data from both the ActivPAL and Accusplit pedometer. Because the Accusplit pedometer did not have Bluetooth or texting capabilities, the VA Annie platform was used to obtain daily step counts from all participants. VA Annie is a text messaging service that allows users to receive broadcast messages from their local VA medical center. 11 Participants were sent a text message every morning that asked them to send a return text with the prior day’s step count and to reset their pedometer. Passive step monitoring was the only interaction with the control group.
Intervention
After discharge, participants were randomly assigned to the intervention or control group through a random number generator. Step data from the initial 7-day post-hospital period were used to calculate a baseline average for all participants. On day 7, participants in the intervention arm were contacted via telephone to establish a daily step goal for the 60-day intervention period. Participants selected a goal that was either 25%, 33%, or 50% step increase above their baseline for the end of the 60-day period. In prior work, participants have chosen ambitious goals (more than half selected a 50% step increase) 12 Participants in the control arm were not contacted to establish a step goal.
Participants who were randomized to the intervention group received a multicomponent intervention that leveraged insights from behavioral economics to increase daily steps. First, participants provided verbal agreement that they would try their best to achieve a daily step goal during the study, which leverages the behavioral economic principle of pre-commitment. 13 Participants were then entered into an intervention using performance feedback, social incentives, and the fresh start effect (as needed) to increase daily steps.
Intervention participants received performance feedback weekly, by telephone, about their progress toward their step goal. The weekly telephone call served as a social incentive in that participants were aware that the study team could see their step goal achievement and would contact them to discuss this throughout the study. Social incentives can increase motivation to achieve goals because individuals care about others’ perceptions of their behavior, which can increase motivation. 14 We averaged each participants’ step count over 7 days to compare to their goal. If intervention participants were successful in meeting their step goal for that week, they continued on their trajectory of gradual step increases to meet their ultimate 60-day goal. If intervention participants did not reach their goal every week, they were allowed to reset their 60-day goal to make this more achievable. This leveraged the “fresh start effect” – the behavioral economic principle demonstrating that individuals are more motivated to change their behavior around temporal landmarks. 15 Both arms received $50 in compensation, $25 after consenting and $25 after the completion of the study, and all participants kept the pedometer.
Outcomes and Follow-Up
The primary outcomes were feasibility of the intervention and usability of the devices. We measured feasibility as the number of Veterans identified as eligible during the interdisciplinary rounds, the proportion of those who consented to participate out of those screened, and the proportion who completed the study out of those who consented. The completion rate is defined as the proportion of participants who completed the study out of those who consented however but the enrollment and completion rate reflect the primary outcome, feasibility. We measured usability through how often Veterans reported their steps and wore the devices, how complete the step data was, as well as Veteran survey responses at the end of the study. The survey inquired about satisfaction with the pedometer and ActivPAL devices, interest in continuing to use the devices, if the devices and intervention increased motivation to improve daily steps, and the frequency of wearing devices throughout the study. Participants in both groups were texted daily to input their steps using the VA’s Annie platform. They were required to report the previous day’s step counts from the pedometer. The intervention group additionally received the ActivPAL device, which recorded steps continuously and did not require daily texting. Instead, this device needed to be sent back at regular intervals for data extraction. Providing this detailed reporting structure ensures timely and accurate data collection from both devices.
Exploratory secondary outcomes included the correlation between the ActivPAL and pedometer daily step count measurements, and an effectiveness measure: the change in median daily steps across weeks of the intervention between the intervention and control groups. Participants were withdrawn from the study if they had 3 consecutive weeks without any data reported and were unreachable. All study procedures were approved by the Corporal Michael J. Crescenz VA Medical Center IRB.
Survey Instruments and Validation
The survey for our study was developed based on a prior published survey which assessed Veterans’ perceptions and experiences with wearable devices. 16 The survey includes sections to assess the acceptability and usability of the ACCUSPLIT and ActivPAL devices, participants’ interest in continuing to use the devices, and the role of the intervention in motivating mobility. This survey included questions about the most and least effective components of the intervention and participants attitudes to using wearable devices, with responses measured on a Likert scale.
Analysis
We calculated feasibility and usability as simple proportions using enrollment and survey data, respectively. To understand the relationship between the pedometer-reported step counts and ActivPAL-reported steps, we conducted a Pearson correlation test.
Statistical Analysis
To measure our exploratory effectiveness outcome, we used a linear mixed effects regression using intervention group and baseline steps as our fixed effects and a participant random effect. All daily step counts > 1 step were included in the analysis. To manage missing data, we included all step counts greater than 1 step in the primary analysis. Days with step counts less than 100 were excluded in a sensitivity analysis to understand the potential impact of our step count in the exploratory outcomes. We conducted a sensitivity analysis that excluded days with less than 100 reported steps (6 patient days were excluded) to examine whether excluding these low-activity days would impact our results.
Results
Thirty-seven Veterans consented of 104 screened (36% enrollment rate). Of the 37 who provided informed consent, 16 completed the study (9 interventions, 7 control; 43% completion rate). Nine participants self-revoked from the study (4 control, 5 intervention) after hospital discharge. In addition, 12 Veterans were lost to follow-up for the following reasons: 3 declined home health services, 4 were hospitalized before the 60-day period ended, and 5 did not report data and were unable to be reached. Most patients were 70 to 74 years old, half were Black, and 94% were male (Table 1). The intervention group had a lower CAN score on average (Control = 0.52; Intervention = 0.18). Two serious but unrelated adverse events were reported during the study: 1 Veteran experienced chest pain and another experienced a fall, but neither were related to the intervention as adjudicated by study team and the IRB.
Study Characteristics of Participants by Study Group.
Other races are not included as only black and white were reported in the sample.
Care Assessment Need (CAN) Score reflects estimated probability of admission or death within a specified time.
Usability and Feasibility
The missing data rate for the ActivPAL was 29%, in contrast, there was only 4% missing data using the Accusplit pedometer. Every participant reported a preference for the Accusplit pedometer over the ActivPAL. Nearly all (15 of 16) participants would recommend a pedometer or similar device to track steps to other Veterans. The correlation between daily steps recorded with the Activpal and the Accusplit was moderate, but significant for the overall sample (Pearson’s Rho = .47, P < .01). Since we had more missing data with the ActivPAL device, we elected to evaluate our effectiveness results based on the Accusplit pedometer readings and did not impute for the 4% of data that was missing. Instead, our analysis was based on the available complete data sets. This decision was made to maintain the integrity of the collected data and avoid introducing potential biases through imputation.
In unadjusted results, the control group achieved more steps during the intervention period (+3221 steps in control vs. +959 in intervention). However, results from the regression model when adjusted for baseline steps and CAN score demonstrated no significant difference between groups in the change in steps achieved over the study (P = .095, Figure 1). The point estimate refers to the estimated difference in the mean change in daily steps between the intervention and control groups over the study period. In our analysis, the point estimate was 882 steps. This means that, on average, the intervention group showed an increase of 882 more steps per day compared to the control group, although this difference was not statistically significant (P = .19). This point estimate provides an indication of the potential effectiveness of the intervention. A sensitivity analysis removing days with step counts less than 100 did not meaningfully change results.
Unadjusted mean step count over 8-week study period. Line graph displaying changes in weekly mean step count for control and intervention participants. There was no significant difference in step count change over 60 days of participation in the study.
Discussion
This pilot trial demonstrated both challenges and opportunities for future interventions to improve mobility among older adults after hospital discharge. First, we found that – as expected – recruiting and retaining study participants who were receiving home health services was difficult. Only 43% of Veterans who consented to participate completed the study. However, data capture using the VA’s Annie automated text-messaging platform and pedometers was surprisingly complete (96%), positively received by participants, and more complete than use of a research-grade device. There was moderate to strong correlation between a research-grade device (ActivPAL) and waist-worn pedometer on daily steps, suggesting use of a pedometer alone in the future may be more cost-effective and lead to more complete data capture. We were not powered to demonstrate a difference in steps and had large baseline differences in the characteristics of our intervention and control groups, suggesting larger samples – potentially in more homogenous populations - will be required to determine intervention efficacy.
However, we faced significant challenges with retention. Two other interventions to improve mobility after hospital discharge have experienced similar difficulty identifying suitable patients for enrollment, but have had less loss to follow-up. In 1 case, the population was younger and much healthier than our sample which may have played a role. 5 In the other, the intervention was delivered by the home health agency directly to patients referred to that specific agency. 17 Future interventions in a similar population to ours may need to partner directly with a home health agency and/or support partners of the patients to retain participants. Support partners can assist with the setup and maintenance, provide motivation and reminders to the participants to engage in regular physical activity, and troubleshoot any technical issues with the devices. Thus, they may contribute to more reliable and comprehensive step count data.
Strengths of the study include robust assessment of steps over a longer study period than previous studies and enrollment of a more functionally and medical complex population than prior trials.3,4 Limitations of the study include a small sample from a single site and a low retention rate. In addition, our sample was predominantly male which could limit the generalizability of the study. Although participants were randomized to control and intervention groups, participants were not blinded which could have resulted in performance bias. Our survey designed to assess device acceptability and usability was tested prior but not formally validated.
This study, taken in context with other recent developments in the field, suggest there is much to learn about how to effectively improve daily steps after hospital discharge. More trials are needed to determine the optimal design, timing, and motivating factors for older adults to improve mobility after hospital discharge. Using the results of our trial, we estimate it would require 650 total patients (325 in each group) to detect a change of 500 steps with alpha of 0.05 and power of 0.80; 500 steps are what the literature suggests is a minimally clinically relevant change.
Conclusion
Our study revealed that while step counts improved in both groups, feasibility and acceptability require enhancement. Participants preferred pedometers and provided more complete step data than research-grade devices. Addressing identified limitations in future research will enable more robust findings. Larger studies are necessary to evaluate the intervention’s efficacy and to refine methodological approaches for effective mobility improvement post-hospital discharge.
Supplemental Material
sj-docx-1-inq-10.1177_00469580251384787 – Supplemental material for Feasibility and Usability of a Post-Hospital Behavioral Intervention to Improve Mobility in Veterans
Supplemental material, sj-docx-1-inq-10.1177_00469580251384787 for Feasibility and Usability of a Post-Hospital Behavioral Intervention to Improve Mobility in Veterans by Jacqueline A. Benson, Matthew Wilson, Aidan J. Flynn, Julie Stutzbach, Kimberly J. Waddell and Robert E. Burke in INQUIRY: The Journal of Health Care Organization, Provision, and Financing
Footnotes
Ethical Considerations
All study procedures were approved by the Corporal Micheal J. Crescenz VA Medical Center IRB # 01829 on December 3, 2019.
Consent to Participate
Interested Veterans provided written informed consent.
Author Contributions
Study concept and design (R.B., J.B) study recruitment (J.B., A.F.), analysis and interpretation of data (R.B, J.B., M.W., K.W., J.S.), preparation and critical revision of manuscripts all authors.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Internal funding from VA Veteran Integrated Service Network 4 Competitive Career Development Fund. The funder had no role in the design, analysis, or interpretation of the results.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Data Availability Statement
The authors confirm that the data supporting the findings of this study are available within the article and its supplementary materials.
Supplemental Material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
