Our study evaluated the long-term effects of a PR program in a community primary care setting with moderate COPD patients. All the groups that took part in PR showed a significant improvement in the emotional dimension after 3 months. This improvement was only maintained in the group that completed the one-year maintenance PR program but not in the 3-month intervention group. Benefits in the fatigue and mastery dimensions also observed after one year of program maintenance were not achieved in the other 2 comparison groups. Lack of improvement in exercise tolerance and overall QoL ratings between groups did not allow accurate determination of PR short- and long-term effectiveness.
Previous studies reported that emotional improvement after the intervention was not maintained after discontinuation or not even after complete PR maintenance programs [29, 30], being the first deteriorated QoL dimension. Our results are in line with those reported by Moullec and Chavannes where PR maintenance strategies carried out in the community were associated with long-term improvements in QoL [16, 31]. Emotional benefits were observed in both PR groups at 3 months and persisted in the group that continued pulmonary rehabilitation together with improvements in two domains that include socio-psychological components. One explanation could be the beneficial results of mixing with individuals with similar problems and sharing negative experiences with disease. The social support of other COPD patients and the long-term attention of a physiotherapist would promote a reduction in emotional reactions and development of adapted behaviors .
Otherwise, between-group findings are not in line with the results of Lacasse et al. meta-analysis , as we were not able to find a consistent improvement in any quality-of-life dimension. The patient’s QoL was generally good at baseline and therefore had little scope to show large improvements through low intensity PR activities. Only dyspnea improved significantly after 3 months of PR in the RHB group compared to controls. This results could appear paradoxical given the very low attendance rate (36%) in this group; much lower than the RHBM group (69%). Patients showed a low baseline level of dyspnea in the MRC scale and high exercise capacity in the 6MWT. In fact, 22 patients had dyspnea with intense exercise (MRC 0); of the 47 patients with higher scores, only 17% had a MRC scale score >1.
In common with most PR studies conducted on COPD patients, we found no improvements in pulmonary function parameters [10, 21, 32]. Our patients following the rehabilitation program did not show significant improvements in exercise capacity in contrast to those described in a previous meta-analysis , and in a more recent study  where PR, including a much more intensive exercise such as endurance training, was highly effective in improving the exercise capacity of patients with COPD. Our RHB program included low intensity peripheral muscle training but did not have any endurance training which could explain this lack of effect. As mentioned above, we included patients with a high exercise capacity with a mean walking distance between 436 and 466 meters while the Italian study baseline mean distance was around 300 meters .
By selecting patients with low basal symptoms, the effectiveness of the intervention can be limited as described by others . These findings suggest that the greatest room for improvement in primary care patients can be expected in those with tangible dyspnea and impaired health status (MRC score > 2 and/or CCQ score >1) across all GOLD stages [16, 34].
We found that the proportions of individuals with some exacerbation episodes were similar in the three groups, in agreement with previous studies [35, 36], while a recent PR primary care program found a reduction in exacerbation rate after a community PR program .
Most PR programs are based at hospitals or at home [10–12, 14, 33, 37]. The intervention we evaluated was developed in primary care centers, with the existing resources of those health centers. In so doing, we attempted to show that this intervention was feasible in a setting where PR is not a regular service but highly accessible for patients. Few PR integrated community programs have been developed with primary care resources and though they found benefits, the results are not conclusive [16, 17]. Low patient compliance with the intervention indicates the need for a more integrated approach in order to involve patients as active partners in their treatment process and achieve positive results in behavior modification [31, 38].
Our study showed persistent within-group improvements in QoL after only a 1-year maintenance program, with no similar effects in the 3-month program group. At this moment it is not known how long these programs should take and how they might influence behavior  or whether alternative management such as an action-plan or self-management strategies could result in important long-term benefits [32, 39]. It is generally believed that longer programs yield more durable training effects but the majority of existing studies looking for long-term results, such as ours, have shown small improvements which may not always be possible to replicate [11, 15–17, 19, 23, 32, 34].
One important limitation arises from the fact that we were not able to recruit the sample size required. Even with active review of included-centers’ registers and COPD patients records, fewer patients than expected were identified. This is consistent with the results of other studies showing COPD under-reporting  and the difficulties of recruiting patients with COPD to a PR trial. This affected the power of the study to identify significant changes in the PR groups and led to negative results.
The high rate of withdrawal was also an obstacle to obtaining fair results. Of the patients initially included, almost 50% withdrew after randomization and the remainder showed moderate adherence to PR sessions. These two limitations reduced the power of this study to identify PR benefits. The characteristics of patients lost after initial evaluation did not differ significantly among the three randomized groups, indicating that loss to follow-up did not alter our final results.
In analyzing the factors related to non-adherence, we found that one of the primary health care centers was located far from the city center thus limiting accessibility to patients living far away. Low adherence during the second year of PR in our first cohort study resulted in a deviation from protocol, which was designed to evaluate the long-term (24-month) effect of PR maintenance. Patient withdrawal due to morbidity and mortality also had a negative influence on our final sample. Most studies analyzing similar rehabilitation programs in COPD patients have also shown considerable loss to follow-up, even higher than ours [15, 41, 42]. In addition, the PR provided in groups of patients probably made it more difficult to adapt to the timetable than if it were performed individually and, as such, influenced adherence to PR sessions.