Adding a Physical Activity Program with Coaching to Usual Care for Patients with COPD

BACKGROUND: Observational studies consistently show that physical inactivity is associated with increased hospitalizations and mortality in chronic obstructive pulmonary disease (COPD). Despite the unequivocal evidence that pulmonary rehabilitation, a comprehensive guideline-recommended intervention, improves patient-centered outcomes, participation remains suboptimal, at approximately 5%. Health systems need alternative approaches. The “real-world” and long-term effectiveness of interventions to increase physical activity and positively impact health care use in a large representative sample of older adults at high risk for recurrent COPD exacerbations remains unknown. OBJECTIVES: To evaluate the effectiveness of a 12-month community-based physical activity coaching intervention (Walk On! [WO]) designed to be generalizable and scalable and to have sufficient flexibility to meet the needs of a large diverse sample of patients with a history of COPD exacerbations. The primary composite outcome was all-cause acute care use and mortality compared with standard care (SC). METHODS: We conducted a pragmatic trial with preconsent randomization to the 12-month WO intervention (n = 1358) or SC (n = 1349) at 6 sites in a large integrated health care system. Eligible patients with a COPD-related hospitalization, emergency department (ED) visit, or observational stay in the previous 12 months were automatically identified from the electronic medical record (EMR) and randomly assigned. Only patients assigned to WO were approached by respiratory therapists who served as physical activity coaches and obtained patient consent to participate in intervention activities. Enrollment occurred from July 2015 to July 2017 with follow-up ending in July 2018. The WO intervention included collaborative monitoring of physical activity/steps, semiautomated step goal recommendations, individualized reinforcement, and peer/family support. Any patient across the 2 groups could be referred to pulmonary rehabilitation. The primary outcome was a composite of all-cause hospitalizations, observational stays, ED visits, and mortality in the 12 months following randomization. Secondary outcomes included COPD-related acute care use, cardiometabolic markers, and self-reported physical activity. EMR data for these outcomes were available for all randomly assigned patients. Additional secondary outcomes included symptoms, health-related quality of life (HRQOL), and satisfaction for a subset of the sample; these data were obtained from mailed surveys for all WO participants and a randomly selected subset of SC participants at baseline, 6 months, and 12 months. The primary analyses followed the intent-to-treat (ITT) principle, where all randomly assigned patients, regardless of participation in WO, were included. The follow-up time was 12 months from the date of randomization. Secondary analyses included prespecified, as-treated analyses in which WO participants were compared with SC participants, and the follow-up time was 2 to 12 months from the date of randomization. Prespecified analyses used logistic regression as the primary approach to assess the intervention effect on the proportion of patients with occurrence of the primary composite outcome and secondary outcomes, and with survival analysis as a secondary approach to assess the intervention effect on the time to the first occurrence of each event. For the as-treated analysis, we used stabilized propensity score inverse probability of treatment weighting to balance baseline characteristics between patients who participated in the WO and SC groups. Linear regression was used to analyze the survey data and other continuous secondary outcomes. Missing data were not imputed, nor were P values adjusted for multiple comparisons with these exploratory analyses. RESULTS: Baseline patient characteristics were mean ± SD age, 72 ± 10 years; percentage forced expiratory volume in 1 second (FEV1%) predicted, 61% ± 23%; 54% female; and 74% White. A total of 321 patients randomly assigned to the WO group participated in any intervention activities (24% uptake; 14%-36% across 6 sites). Self-reported physical activity in the 12 months postrandomization was not significantly different between the WO and SC groups in the ITT analysis. However, in the as-treated analysis, a higher percentage of WO participants reported engaging in physical activity than did SC participants (47% vs 31%, respectively; P < .001). The ITT analysis showed no differences between the WO and SC groups on the primary all-cause composite outcome (odds ratio [OR], 1.01; 95% CI, 0.92-1.28) or in the individual primary outcomes over 12 months. Prespecified, as-treated analyses compared outcomes between all SC patients and 321 WO patients who participated in any intervention activities over 10 months of follow-up. There were non–statistically significant estimates in favor of WO participants compared with all SC participants on all-cause hospitalizations (OR, 0.84 [95% CI, 0.65-1.10]; P = .21) and deaths (OR, 0.62 [95% CI, 0.35-1.11]; P = .11) for the as-treated analyses. Adjusted time-to-event analyses showed similar trends with significant group differences in observational stays (hazard ratio, 0.72 [95% CI, 0.53-0.98]; P = .04). Survey data were missing for 29% to 42% of patients across groups at 6 and 12 months. Compared with SC patients, WO patients reported greater improvements in the Patient-Reported Outcomes Measurement Information System-10 (PROMIS-10)-physical HRQOL domain at 6 months (effect size, 0.25; P = .01) in the as-treated analyses. There were no group differences in other secondary outcomes. For the ITT analyses, there were no significant interactions between treatment groups and comorbidity (Charlson Comorbidity Index of <4 or ≥4, heart failure, diabetes, anxiety, or depression); social support (partnered or not partnered); White vs non-White race; sex; age <70 or ≥70 years; and internet access. The as-treated analyses similarly showed no significant interactions except for race (P < .001), where non-White WO participants appeared to have a lower risk of acute care use or death (47% vs 67%) from months 2 to 12 postrandomization than did non-White SC participants, but this was not the case for the White cohort. CONCLUSIONS: In this pragmatic trial of patients at high risk for recurrent COPD exacerbations, a potentially scalable and sustainable 12-month physical activity coaching intervention had poor uptake and did not have significant effects on the primary composite outcome of acute care use and mortality for both the ITT and as-treated analyses. The intervention was, however, associated with increased physical activity and improvements in the physical HRQOL domain in the as-treated analyses. The findings from this study call for more realistic expectations for frail, older patients with serious chronic conditions to engage in and meet physical activity recommendations and for increased physical activity to have an impact on distal outcomes. LIMITATIONS AND SUBPOPULATION CONSIDERATIONS: Given the need for automated identification and randomization of eligible patients using EMR data to ensure the broadest representation, we could not rely on spirometry measurements as inclusion criteria and thus may have included some patients without COPD. The high level of survey missingness (especially for patients who had higher comorbidities and lower baseline survey scores across both groups) and the multiple comparisons suggest caution in interpreting the significant findings. Despite efforts to adjust for selection bias in the as-treated analyses, residual confounding is an important limitation. Due to the low intervention uptake overall, there was insufficient power for the ITT subgroup analyses. However, the as-treated analyses suggest that non-White patients may benefit more from participating in WO than do White patients. This finding requires further exploration and confirmation.