Search results

A total of 18,846 citations were identified and screened, 217 were retrieved for full consideration, and 14 studies included - 12 RCTs [10,11,12,13,14,15,16,17,18,19,20,21] and 2 comparative observational [22, 23] studies (from 18 papers). The PRISMA diagram for the review is shown in Fig. 1 (additional information in online supplementary appendix 2).

Fig. 1
figure 1

Preferred Reporting Items for Systematic Review and Meta-analysis (PRISMA) flow diagram for the systematic review and meta-analysis

Characteristics of studies

Nine studies compared HBPR to ‘usual care’ [10, 11, 14,15,16,17,18, 20, 22], four to OPR [12, 13, 19, 21, 23], and one [19] compared HBPR to OPR or ‘usual care’. They were conducted between 2007 and 2017 in Australia (2 studies) [10, 12], Brazil (2) [16, 19], China (2) [14, 18], Iran (1) [15],Spain (2) [17, 22], the United States (1) [11], and the United Kingdom (4) [13, 20, 21, 23] (Table 2). Collectively, they included 2293 patients and all but four [10, 12, 13, 22] were conducted in single centers.

Table 2 Characteristics of included studies included in the systematic review

HBPR program characteristics

All HBPR programs lasted at least 8 weeks. Except in three [11, 12, 18, 24,25,26,27] studies, programs began with one or more in-person training/education session(s) at an outpatient clinic or an initial in-home visit. The three [11, 12, 18, 24,25,26,27] exceptions delivered introductory sessions online or by video. Where described, sessions and in-home visits were conducted by nurses or physiotherapists. Additionally, all programs provided information booklets, manuals or workbooks and exercise prescription information, about which reported information varied. There were differences in specific components of and schedule for weekly in-home exercise, but most programs incorporated both strength and endurance training into activities to be performed unsupervised three times per week. However, four [12, 17, 22, 23] of the programs included supervised exercise. In the single [18] online HBPR, participation was monitored through the system itself, which collected log-on and log-off information. Patients logging on regularly were flagged and contacted by a nurse. Most programs asked patients to keep track of daily activities and symptoms in a diary or workbook. In one [15] case, patients used checklists. Programs employing special equipment or devices supplied them (hand and ankle weights [19], heart rate monitors [19], and pedometers [10, 12],). Weekly follow-up telephone calls by a nurse or physiotherapist to encourage or motivate patients and monitor progress were a part of almost all programs. Comparator interventions in studies were ‘usual care’ or OPR. ‘Usual care’, when described, varied significantly across studies, but typically comprised at least some form of in-person self-management/clinical needs assessment and advice on staying active and taking medications as prescribed, delivered through in-person education sessions and/or information booklets. OPR involved community-based supervised group sessions held two to three times weekly for seven to 12 weeks, and included exercise and education (online supplementary appendix 3, 4 and 5).

Risk of bias

Results of risk of bias assessment

The majority of RCTs provided a clear description of the objectives, interventions, outcomes and findings (online supplementary appendix 6). However, in two [15, 22] trials, ‘usual care’ was not defined. Four [10,11,12,13] of the RCTs had published or registered protocols pre-specifying outcomes, and in all five, such outcomes were the same as those reported in the trial results. In the remaining RCTs, it was not possible to determine if the results reflected all outcomes measured (Fig. 2).

Fig. 2
figure 2

Cochrane risk of bias summary for included RCTs. (1) random sequence generation (selection bias); (2) allocation concealment (selection bias); (3) blinding of participants and personnel (performance bias); (4) blinding of outcome assessment (detection bias) (patient reported outcomes); (5) blinding of outcome assessment (detection bias) (other outcomes); (6) incomplete outcome data (attrition bias); (7) selective reporting (reporting bias); (8) other bias

All non-randomized studies [22, 23] provided a clear description of the objectives, interventions, differences in patient characteristics and potential confounding variables between the HBPR and comparator groups, outcomes and findings (online supplementary appendix 7) (Fig. 3).

Fig. 3
figure 3

Of the 15 RCTs, eight [10,11,12,13, 17, 19,20,21] provided a clear description of the random sequence generation. In five [14,15,16,17, 19] RCTs, no details around allocation concealment were included. In the seven [10,11,12,13, 18, 20, 21] RCTs describing methods for ensuring blinding of patients and investigators, four [12, 18, 20, 21] used opaque sealed envelopes, two [11, 20] assigned the task to a researcher uninvolved in the trial, and two [10, 13] relied on web-based/online programs. In the two [10, 23] non-randomized studies, the risk of bias due to confounding was moderate, although known confounders were measured and controlled for through statistical analyses.

Six [10,11,12,13, 18, 21] RCTs reported that outcome assessors were blinded to the type of intervention. No information was provided in all but one [19] of the remaining trials. One [20] RCT explicitly stated that assessors were not blinded, rendering it at high risk for observer bias. None of the [10, 23] non-randomized studies mentioned blinding of assessors.

In five [10, 11, 13, 14, 19] RCTs, the risk of attrition bias was high but was low in six [12, 15,16,17,18, 21] RCTs, where missing data and reasons for withdrawals were similar between groups.

Results from GRADE assessment

GRADE assessment was conducted on selected outcomes (Tables 3 and 4). The GRADE level was ‘very low’ or ‘low’ for all outcomes in HBPR vs. ‘usual care’ studies and 2 of 3 outcomes in HBPR vs OPR studies (Tables 3 and 4).

Table 3 Studies comparing HBPR to ‘usual care’
Table 4 Studies comparing HBPR to OPR

Summary results of effectiveness


Three [11, 12, 21] studies reported adverse event rates, but none specified the type of adverse event. One [11] study comparing HBPR to usual care, and two [12, 21] studies comparing HBPR to OPR showed no statistically significant differences between groups (online supplementary appendix 8).

Health care resource utilization (hospital admissions, ER visits, and physician visits)

HBPR compared to ‘usual care’

In one of two [11, 22, 24, 25] studies comparing the percentage of patients with one or more COPD-related hospital admissions over a 12 to 18 month period, there were no statistically significant differences between groups. In the second [11, 24, 25] study, the percentage of patients with admission in the HBPR group was almost half that of the ‘usual care’ group (19% versus 30%; statistical significance of the difference not reported).

HBPR compared to OPR

One [12] study concluded that the impact of HBPR and OPR on health services utilization was similar.

Health-related quality of life (HRQoL)

Different disease-specific instruments were used, including the COPD assessment test (CAT), chronic respiratory disease questionnaire (CRQ), and the St. George’s respiratory questionnaire (SGRQ). Measurements were also at different points in time as the duration of rehabilitation programs varied among studies (online supplementary appendix 9, 10 and 11).

HBPR compared to usual care

In the study [14] using the CAT, short-term improvements in scores were statistically significantly greater in the HBPR group, but both groups experienced clinically meaningful improvements (a change in scores of at least 2 points [28]). However, 6 months after completion of HBPR, improvements from baseline were similar between groups. Two studies [10, 11, 24, 25] using CRQ found no statistically significant differences in changes between treatment groups on any of the 4 CRQ domains from baseline to several months after HBPR or ‘usual care’. However, in one [10], both groups reported clinically meaningful improvements in the dyspnea, fatigue and mastery domains at 2 and 6 months of follow-up (a change in score of at least 0.5 points). In the other [11, 24, 25], neither group experienced clinically meaningful improvements in the dyspnea domains. In contrast, the single study [20] that measured HRQoL directly following completion of HBPR reported statistically significant improvements in dyspnea, emotional function and fatigue among patients who had HBPR, but not among those who received ‘usual care’. In two [16, 17] of the three [16,17,18] studies using the SGRQ there were statistically significantly greater improvements in total scores (from baseline to end of treatment) with HBPR. However, for individual domains, scores varied between studies; in one study, changes in all domain scores were similar between groups and in the other, those for the ‘activity’ and ‘impact’ domains were greater among HBPR patients. The third [18] study assessed HRQoL 10 months post HBPR or ‘usual care’. There were no statistically significant differences between groups, except for social functioning and psychological disturbances resulting from COPD (i.e., ‘impact’ domain), which improved with HBPR and worsened with ‘usual care’.

HBPR compared to OPR

In one [13] study using CAT before and after completion of the programs, actual scores were not presented, but the reported p-value was not statistically significant (Fig. 4). Three [12, 21, 23] studies used CRQ and measured HRQoL at program completion, and meta-analyses were possible in 4 domains of the instrument: dyspnea, emotional, fatigue and mastery. Two [15, 26] found no statistically significant differences between groups in these domains after 2 months of pulmonary rehabilitation. By contrast, in the third [21] study, scores were statistically significantly lower in the emotional, fatigue and mastery domains among HBPR patients. However, 4 months after the end of the program, no statistically significant differences were reported.

Fig. 4
figure 4

Mean differences in health-related quality of life after completion of 2-month active pulmonary rehabilitation phase in studies comparing HBPR with OPR

Adherence to/compliance with treatment

HBPR compared to usual care

With one exception [17], at least 79% of patients completed HBPR [10, 11, 15, 16, 18,19,20, 22, 24, 25] Of the three [17, 19, 20] studies reporting lower rates of adherence with HBPR, ‘usual care’ did not include an exercise component or diary for patients to log symptoms and/or activities on a daily basis.

HBPR compared to OPR

The percentage of patients completing HBPR was higher in three studies [12, 19, 21] but lower in two [13, 23], but the statistical significance was not reported. In one study [13], with lower adherence to HBPR, patients withdrew because it was too time-consuming or preferred a classroom setting or had computer problems affecting access (online supplementary appendix 12).

Frequency of exacerbations

HBPR compared ‘usual care’

Two studies [14, 22] showed no statistically significant differences in exacerbation rates during the rehabilitation phase between groups (online supplementary appendix 13).

Functional exercise capacity and activity levels

Different assessment tools were used to measure functional exercise capacity; the most commonly applied measure, the six-minute walk test. Others were the incremental shuttle walk test (ISWT) and the endurance shuttle walk test (ESWT).

HBPR compared to usual care

Across studies [11, 14, 18, 19, 22, 24, 25], the distance walked in 6 min statistically significantly increased among patients who received HBPR, indicating an improvement in exercise capacity. In some studies, it also increased among patients who received ‘usual care’, but the amount of increase was statistically significantly less than that reported for HBPR.

Five [10, 11, 14, 17, 19, 24, 25] out of six studies demonstrated that patients receiving HBPR had clinically meaningful improvements by the end of follow-up (change in distance of at least 30 m). However, one study [11, 24, 25] reported that at 18 months, only patients with serious COPD showed meaningful improvements in this test.

HBPR compared to OPR

In all four studies [13, 19, 21, 23], the distance walked in 6 min statistically significantly increased after HBPR and OPR, and the gains were similar between programs. However, one study [21] that measured ISWT reported no clinically meaningful improvements (change in distance of at least 48 m [29]) after HBPR or OPR.

Mental health

Studies used the Beck depression inventory, and the Hospital Anxiety and Depression scale (HADS) to assess changes in depression and anxiety (online supplementary appendix 14).

HBPR compared to ‘usual care’

In one [14] study, statistically significant improvements in scores on the Beck Depression Inventory were observed in both groups, but no statistical estimates of between-group measures were provided.

HBPR compared to OPR

Two [13, 21] of the five studies used the HADS. Within groups, post-program scores were similar to baseline, indicating that neither form of pulmonary rehabilitation reduced anxiety and depression among patients.


HBPR compared to ‘usual care’

One [15] study assessed self-efficacy changes between baseline and 2 months (the length of the active rehabilitation). HBPR patients had statistically significantly greater confidence in their ability to manage or avoid dyspnea during events across all five domains of the questionnaire compared to ‘usual care’.

HBPR compared to OPR

Three [12, 13, 21] studies used the PRAISE (Pulmonary Rehabilitation Adapted Index of Self-efficacy) tool [30]. However, only two [13, 21] reported statistical estimates of data variability between groups when administered to patients before and after the active rehabilitation phase (3 months in one study and 2 months in the other). Within and between groups, pre and post-treatment scores were similar, indicating that the confidence level among patients to self-manage their disease did not increase with HBPR or OPR (online supplementary appendix 15).

Source link