WHAT IS ALREADY KNOWN ON THIS TOPIC

WHAT THIS STUDY ADDS

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

Patients were recruited at the University Hospitals of Leuven, Belgium, from November 2019 to October 2022. During this period, Belgium implemented a soft lockdown in response to the COVID-19 pandemic, which lasted from March 2020 to May 2020. Outdoor walking was never completely prohibited. Citizens were required to stay at home as much as possible, although outdoor PA was encouraged under strict conditions. Such activities were only permitted with members of one’s household or a single, consistent friend throughout the lockdown period. All patients provided written informed consent before study enrolment. The trial was approved by the local medical Ethics Committee (Ethics Committee Research UZ/KU Leuven (s62426)), registered at ClinicalTrials.gov and publicly available before the trial commenced (NCT04122768, registered: 09/10/2019) and is in compliance with the ISHLT Ethics statement.21 Patients at least 6 months and a maximum of 4 years after a first double LTX, aged ≥30 years and able to work with a smartphone, were eligible. Patients were excluded in case of problems prohibiting a normal gait pattern. Patients with multiorgan, second solid-organ or bone marrow transplantation, chronic rejection of the allograft before inclusion, medically not stable or a life expectancy of less than 1 year as judged by the treating physician were excluded. Patients with cystic fibrosis were excluded as they are younger and also have more pre-existing comorbidities that remain present after LTX.22 23

At baseline, patients were classified as physically active or inactive (daily step count ≥7500 or <7500, respectively) based on accelerometer data (figure 1).24 Inactive patients were randomised (1:1) into the intervention group (IG) or light intervention group (LIG), using computer-generated random blocks of 4 and 6 in numbered opaque sealed envelopes, prepared by an independent investigator. Stratification variables were baseline exercise tolerance (6-minute walk test (6MWT) distance < or ≥450 m) and time since transplantation (< or ≥9 months). After the baseline assessments, investigators were not blinded to group allocation. Patients were blinded by providing two active interventions. Follow-up visits took place after 3 months and 1 year. Adverse events were recorded during the study period.

All patients were informed about the importance of PA and the PA recommendations provided by the WHO for patients with chronic conditions in a face-to-face discussion (5–10 min). The usual medical care was continued and adjusted as needed by a chest physician blinded to the study group allocation.

Details of the (light) intervention can be found in online supplemental file 1. Patients in the IG received a 1 year semiautomated telecoaching intervention, based on the intervention of Demeyer et al.25 The intervention consisted of the following components: (A) A motivational face-to-face interview at baseline between patient and coach discussing PA. (B) A step counter (Fitbit Inspire, Fitbit, San Francisco, USA), connected to a patient-tailored smartphone application (mPAC). The step counter could be worn around the wrist or at the hip, as preferred. The application provided an individualised step goal that could be changed weekly, based on the steps registered. (C) Supportive coaching calls occurred every 2 weeks during the coaching phase (first 3 months) when insufficient PA progress was observed (ie, no increase of step goal or not reaching the step goal on most of the days, for two consecutive weeks), as supported by a prior study in our research group demonstrating positive effects in patients with chronic obstructive pulmonary disease (COPD).25 The next 9 months were considered a maintenance phase in which patients were not contacted to discuss PA progress. Contacts about technical problems or medication changes (to monitor adverse events) occurred throughout the year. Patients could contact the coach at any time.

Patients in the LIG received a light version of the intervention that consisted of: (A) A wrist or hip-worn step counter (Fitbit Inspire, Fitbit, San Francisco, USA), connected with the smartphone application, where patients could only see their daily step count and a general action plan. (B) Calls with the study team occurred in case of medication changes or technical problems. Patients could contact the study team at any time.

A triaxial accelerometer was worn before the baseline, 3-month and 12-month visits (DynaPort MoveMonitor, McRoberts BV, The Hague, the Netherlands). Patients were instructed to wear the accelerometer for 7 consecutive days during waking hours. At least 2 days with a minimum of 8 hours of wearing time were considered valid and included in the analysis.26 Change in average daily step count after 3 months was the primary endpoint. The secondary PA outcomes at 3 months and 1 year included total moving time (walking, taking stairs and cycling), sedentary time and movement intensity during walking. A PA responder was considered a patient with an increase of ≥1000 steps from baseline, based on the minimal important difference for patients with COPD, as specific LTX guidelines are missing.27 While the applicability to LTX recipients remains uncertain, 1000 steps corresponds to approximately 10 min of walking.28

Exercise tolerance was measured with the 6MWT following American Thoracic Society/European Respiratory Society (ATS/ERS) recommendations.29 The best distance of two tests was taken. Isometric quadriceps force was measured with a maximum voluntary isometric contraction of the quadriceps (Biodex System II, Biodex Corporation, New York, USA).30 The best of three attempts was selected. Exercise tolerance and isometric quadriceps force were expressed as percentages of normal reference values.31 Health-related quality of life, symptoms of anxiety, depression and fatigue were assessed using respectively the 36-item Short Form Survey Instrument (SF-36), the Hospital Anxiety and Depression Scale (HADS) and the Checklist Individual Strength (CIS).

Pulmonary function was performed according to the ATS/ERS guidelines.32–34 Results are shown in absolute values and percentages of reference values.35 All patients were asked about the importance, motivation and self-efficacy regarding improving PA before the results on the current PA level were known. These questions were scored on a 10-point Likert Scale ranging from not important (1) to very important (10). The experience of the patients with the delivered intervention was investigated using a study-tailored questionnaire (online supplemental file 1).

Sample size calculation and more details on the blinded sample-size re-estimation and other analyses are provided in the online supplemental file 1.36 The following analyses were performed:

(1) Between-group differences in primary and secondary outcomes were compared using mixed model analyses (unstructured covariance structure) to investigate the effectiveness of the telecoaching intervention compared with the light intervention programme. The outcome variables were used as dependent variables, the treatment group and visit as independent variables and the duration of daylight, as a proxy for seasonality, was tested as a possible confounder for PA outcomes. All models were adjusted for stratification variables (6MWT and time since transplantation). The interaction effect treatment group*visit was retrieved, with baseline as the reference. Within-group differences in primary outcomes were calculated using similar mixed model methods, using the outcome variable as dependent variable, the visit as independent variable and the duration of daylight as confounder. Assumptions for the mixed model analyses, including normality of residuals, homoscedasticity, linearity of relationships between predictors and the outcome and absence of influential outliers, were checked. (2) Outcomes after 1 year of active patients at baseline were compared with baseline with a paired t-test. (3) The OR to be a ‘responder’ was calculated between IG and LIG using χ². (4) The overall amount of daily step count over the year, measured with the step counter (Fitbit), was calculated for both groups using area under the curve. Based on the normal distribution, data were compared with an unpaired t-test. (5) To test potential effect modifiers, we explored the 12-month intervention effect in different subgroups, using the same mixed model analyses in each subgroup separately. In line with the main models, we retrieved the interaction effect treatment group*visit, in each subgroup. For this, the following subgroups were created: time since transplantation (<18 months or ≥18 months), 6MWT distance (<450 m or ≥450 m), symptoms of fatigue (≥27 points or <27 points) and age (≥62 years or <62 years) (online supplemental file 1). (6) Number of contacts with the coach, total contact time and experience with the intervention (median Q1–Q3) were compared between the IG and LIG using a Mann-Whitney U test. Adherence to the intervention was expressed as the percentage of weekly reports that were completed by the patient (IG only) and the percentage of days that the step counter was worn (IG and LIG) and was compared between groups using χ². (7) The per-protocol analysis included only patients that completed at least 50% of the weekly reports in the application. Finally, post hoc sensitivity analyses were performed to test the influence of the COVID-19 pandemic and baseline imbalances (online supplemental file 1).

Within and between-group changes are shown as marginal mean (CI). Statistical significance was set at p<0.05 for all analyses, which were performed using SAS V.9.4 (SAS Institute, Cary, North Carolina, USA).

18 (17%) of the 108 included LTX recipients were active. 90 inactive patients were randomised into the IG (n=48) or LIG (n=42) (figure 2). Baseline characteristics are presented in table 1 and online supplemental table S1. Exercise tolerance and quadriceps force were below predicted normal values. After transplantation and before inclusion, 98% of the inactive patients had followed an exercise programme under the supervision of a physiotherapist or a pulmonary rehabilitation programme. In the IG, sedentary time was higher (p=0.03) and total moving was lower (p=0.009) compared with the LIG. No other significant differences in baseline characteristics were observed. In the IG and LIG, 77% and 67% of patients chose to wear the step counter at the wrist (p=0.27). Baseline characteristics and changes in outcomes of the active patients are provided in online supplemental table S2.

The participants wore the accelerometer 6.6±0.9 days, which included both week and weekend days. Daylight duration was 750±177 min and the wearing time was 826±106 min. These parameters were comparable between groups at baseline and were not different at the follow-up visits. No PA measurement had to be excluded due to invalid wearing time.

During the first 3 months, patients in the IG tended to increase their PA by 750 (−96 to 1596) steps per day (p=0.08) more as compared with patients in the LIG, measured with the activity monitor (DynaPort) (table 2, figure 3). The IG showed a statistically significant increase in daily steps and moving time compared with baseline, whereas the LIG had small, non-significant, within-group increases. After 1 year, the intervention effects on daily step count and moving time were maintained. Movement intensity did not change in both groups. A significant reduction in sedentary time, expressed as a percentage of wearing time, was observed in favour of the IG. After 3 months, 43% (n=19) of the patients in the IG had responded to the intervention, compared with 19% (n=7) in the LIG (p=0.02), OR 3.3 (1.2–9.0). After 1 year, this proportion was 37% (n=16) and 22% (n=8) for the IG and LIG respectively (p=0.13), OR 2.1 (0.8–5.8).

The overall amount of mean daily step count per week, measured with the step counter (Fitbit), improved in the IG, whereas the LIG remained stable (figure 4). The overall amount of activity was not statistically significantly different between the groups (p=0.15).

Time since transplantation, exercise tolerance, symptoms of fatigue and age at baseline were no significant effect modifiers (figure 5). Changes within IG and LIG in the specific subgroups are shown in online supplemental figure S1.

The 6MWT and quadriceps force improved similarly in both groups over 1 year, with no between-group differences (table 3). No changes in SF-36, CIS or HADS were observed between groups after 3 months and 1 year.

The IG had more contacts with the coach in the coaching phase (median (Q1; Q3): 7 (4; 9)) compared with the LIG (3 (2; 5), respectively), p<0.05. These contacts were mostly triggered by difficulties with PA improvement (online supplemental table S3). During the maintenance phase in the IG and in the LIG during the complete trial, health-related contacts were most prevalent, followed by contacts trying to enhance adherence. Calls related to PA in the LIG were limited (5% of all calls) and initiated by the patient. Adherence to using the step counter was excellent in both groups (97% IG, 95% LIG, p<0.05). On average, participants in the IG answered 71% of the weekly feedback reports (online supplemental table S3).

Both interventions were reported as motivational to increase PA by patients and patients were willing to continue using the step counter. More than 80% of both groups reported subjective PA improvement. The supportive coaching calls in the IG were well-received, with a median (Q1; Q3) score of 9 (8; 10) out of 10 and 55% of patients would like to continue receiving calls in combination with the step counter.

The IG reported 2.5±2 mild/moderate/severe events per patient, compared with 2.2±2 in the LIG (p=0.48). Of all 212 adverse events, 12 were possibly related to the intervention (6 IG and 6 LIG). Most common were overuse injuries and falls during PA, resulting in a sprained ankle or fracture (online supplemental table S4).

Results for the exploratory per-protocol analysis are shown in online supplemental table S5. 12 patients of the IG (27%) were excluded. A between-group difference of 1154 (221 to 2087) (p=0.013) and 872 (−130 to 1875) (p=0.076) steps per day was observed in favour of the IG, after 3 months and 1 year, respectively. Significant differences were observed for total moving time and sedentary time, both in favour of the IG. Movement intensity remained unchanged between groups.

The results for the post hoc sensitivity analysis to adjust baseline PA imbalances between IG and LIG are shown in online supplemental table S6. No significant between-group differences in PA outcomes were observed after 3 months and 1 year.

Patients who started the programme before the first COVID-19 lockdown period (March 2020) reported more limitations to come outside (n=15, 6 out of 10) than those who started later (n=75, 4 out of 10) (p=0.01). It did not change the intervention effect (results of exploratory subanalysis not shown).

To the best of our knowledge, this is the first powered randomised controlled trial that investigated the effectiveness of a PA behavioural programme in LTX recipients. Second, a triaxial accelerometer was used to examine the intervention’s effectiveness and a consumer-based wearable served as a PA coaching tool, as recommended.38 Third, the intensive coaching contained frequently used techniques in effective behaviour change programmes.10 39 Next, by including two active treatment interventions, we were able to blind the patients and we could investigate the added value of intensive coaching on top of the use of an activity tracker. This also resulted in a lower dropout rate (12%) than initially anticipated (20%), but precluded comparing the full intervention to usual care. Finally, data of clinical parameters are widely distributed, providing generalisability to the general LTX population, if they meet eligibility criteria.

Some limitations should be considered. First, the COVID-19 pandemic slowed down the inclusion rate and reduced the number of LTX procedures, thereby reducing the number of eligible patients. Still, 50% of all eligible patients followed up in our centre were recruited into the trial. Different pandemic-related restrictions occurred during the intervention, which may have affected the PA behaviour, as seen in other populations.40 Second, although this study was a priori powered, the effect size of 1400 steps per day was not achieved. This can be attributed to the absence of a true control group, making it difficult to account for the natural evolution of PA. Besides, expecting an increase of 1400 daily steps may have been too optimistic, as a meaningful increase in PA for respiratory patients has been identified as 1000 steps per day.27 The expected effect size of 1400 daily steps was based on a trial in which LTX recipients in an early phase postdischarge increased their activity by pulmonary rehabilitation with 1400 steps. In contrast, our cohort had a median time of 13 months since LTX at the start of our study. Third, the lack of information on adverse events prior to inclusion, and the absence of data on participation in pulmonary rehabilitation during the intervention period, prevented us from including these as possible indications for success or not. Next, despite proper randomisation and stratification, baseline data are not well balanced for our primary outcome. However, our post hoc sensitivity analysis showed that this imbalance at baseline did not affect the intervention effect (online supplemental file 1). Lastly, these results cannot be generalised to patients with cystic fibrosis undergoing LTX. This exclusion was a priori defined since these patients are younger and face disease-specific problems, potentially influencing their physical function and PA. The intervention may not have been optimally tailored to meet their needs.

In the last years, step counters became easily available and measure several outcomes.41 It has been shown that using step counter interventions in daily life is an effective PA intervention and is easy to implement.20 Interestingly, 32% of the patients who were randomised had a step counter before the start of the study. In our study, one out of five participants in the LIG showed an increase of at least 1000 steps per day after 3 months, compared with baseline. However, following the intensive coaching showed an added value compared with light coaching, as patients in the IG were three times more likely to substantially increase their PA in the short term and two times more likely to show such improvements in the long term. Given the relatively low total contact time with the coach, this intervention seems feasible to implement in an expert transplant centre and could be proposed to patients who have already followed rehabilitation but experience difficulties in enhancing PA. Yet, spontaneous recovery cannot be ruled out. Unfortunately, there were no improvements in our secondary outcomes, which is in line with the lack of hard evidence in the literature for health benefits of light PA.

Not applicable.

This study involves human participants and was approved by Ethics Committee Research UZ/KU Leuven, reference number s62426. Participants gave informed consent to participate in the study before taking part.