Pulmonary Rehabilitation After Exacerbation of Bronchiectasis

A Pilot Randomized Controlled Trial

James D. Chalmers; Megan L. Crichton; Gill Brady; Simon Finch; Mike Lonergan; Thomas C. Fardon


BMC Pulm Med. 2019;19(85) 

In This Article


Fifty one patients were screened and 48 patients were enrolled at baseline. 62.7% of patient were female and the median age was 68 years. Twenty seven patients attended with acute exacerbations and were randomized. Therefore 56% of patients enrolled could be randomized. Of those patients not attending within 12 months with exacerbations, 2 withdrew from the study prior to exacerbation while one patient was withdrawn by the investigators for failing to report exacerbations. The other patients did not have exacerbations during the observation period. The flow of patients through the study are shown in Figure 1.

Figure 1.

Flow chart of the study

The characteristics of patients enrolled and the patients randomized in the study are shown in Table 1.

6-min Walk Distance

The 6-min walk distance improved from post-exacerbation to 8 weeks in both groups, with no significant difference between pulmonary rehabilitation and standard care. 6 min walk distance improved by 26 m (− 2 to 54.01) in the pulmonary rehabilitation group and 15 m (− 16.6 to 46.7) in the standard care group. The mean difference of 11 m (− 34.3 to 56.3), p = 0.6 was not statistically significant (Figure 2).

Figure 2.

Improvement in 6-min walk distance from randomization at the end of antibiotic therapy to 8 weeks (primary outcome) and 12 weeks (secondary outcome). Data are presented as mean change from baseline with standard error

There was similarly no statistically significant difference in 6 min walk distance at 12 weeks and in fact the improvement in walking distance was numerically greater in the standard care group compared to pulmonary rehabilitation (mean difference 4.6 m 95% CI − 59.9 to 50.7, p = 0.9).

Secondary Endpoints

Time to the next exacerbation was not significantly different between the two groups. The median time to next exacerbation was 169 days in the standard care group and 190 days in the pulmonary rehabilitation group. The hazard ratio was 0.83 (95% CI 0.31–2.19, p = 0.7) Figure 3.

Figure 3.

Kaplain-Meier survival curve showing time to the next exacerbation after randomization

Cough symptoms improved by a mean of 0.83 points (+/− 1.24) in the pulmonary rehabilitation group compared to 0.91 points (+/− 0.61) in the standard care group at 8 weeks. The corresponding results at 12 weeks were 0.82 (+/− 1.1) and 0.51 (+/− 0.51), p = 0.8.

Symptoms measured using the CAT score improved by 2.1 points (+/− 1.6) in the pulmonary rehabilitation group and by 0.86 points (+/− 1.9), p = 0.6 at 8 weeks. Difference in symptoms was even greater at 12 weeks with an improvement of 2.89 points (+/− 1.9) in the pulmonary rehabilitation group and a worsening of symptoms by 0.6 points (+/− 2.5) in the standard care group. The mean difference of 3.5 points at this time point substantially exceeded the MCID for this scale but was not statistically significant (=0.3).

No changes in the St. Georges Respiratory questionnaire were statistically significant across the total score or the activity, impact or symptom domains (Figure 4).

Figure 4.

Changes from randomization in the St Georges Respiratory Questionnaire (SGRQ) across the 3 domains and the total score. No differences were statistically significant between the pulmonary rehabilitation and standard care groups at any time point. Data are presented as mean change with standard error

FEV1 improved by 100 ml (+/− 56 ml) in the pulmonary rehabilitation group compared to 21 ml (+/− 33 ml) in the standard care group at 8 weeks,p = 0.2 with no difference evidence at 12 weeks (p = 0.4).

Sputum microbiology showed no differences between groups.

The per-protocol analysis for the primary endpoint found a mean difference of 24 m (− 28.9 to 76.9 m), p = 0.4. No significant differences were observed in any other endpoints in the per protocol analysis. No safety issues were identified in the study.

Power Calculation

The data collected suggested that large differences between groups exceeding the reported MCIDs of 25 or 54 m in the 6- min walk test were unlikely. The probability of rehab leading to improved outcomes (based on the estimated mean difference and using a common standard deviation to calculate the standard error on the difference between the group means) is 0.77.

This translates into a 17% probability that rehabilitation improves mean 6 min walk test by more than 25 m, and a 0.2% chance that the difference is greater than 54 m.

A full assessment of the probability of the study producing a statistically significant result for a clinically meaningful effect size needs to allow for uncertainty in both the estimate of the difference between the mean outcomes for the two treatments and the estimated common standard deviation around those means. It also needs to respect the correlation between those estimates. Uncertainty in the proportion completing per protocol has a smaller effect. The figure below shows that this probability continues to climb as the sample size increases past 2000, but the only scenario where the probability reaches 80% occurs when only the per protocol data is used and any significant difference, no matter how small, is considered a success. Even under those conditions, more than 1000 participants would need to be recruited (Figure 5).

Figure 5.

Probability of a significant difference comparing Pulmonary rehabilitation with standard care. The Y-axis shows the probability of a positive result and x-axis shows the required sample size. The colour of lines are as follows BLACK: any difference between groups is considered meaningful RED = a difference of > 25 m is required, GREEN = a difference of > 54 m is required. The broken lines indicate per-protocol analysis with solid lines indicating the intention to treat population