Results
Fifty-four children with CSLD or bronchiectasis, and isolation of Pa in at least one clinical sample, were identified through the clinical records. The patient’s clinical characteristics are shown in table 1. There were only 16 patients who had valid spirometries both at the time of Pa isolation and at the end of the follow-up period with no difference between the ppFEV1 values between the two time points (93.2±1.3, and 94.0±1.5, respectively, p= 0.43). The same comparison was also performed in the subgroup of 8 patients who attained remission and no difference was found (93.7±2.0, and 94.4±2.2, respectively, p= 0.68).
The Cox proportional hazards model showed that the use of nebulised colistin and the presence of bronchiectasis at HRCT scan were statistically significant predictors of remission (HR:3.99; 95%CI:1.12-14.14; p= 0.032, and HR:0.24; 95%CI:0.08-0.71; p= 0.010). The estimated HRs indicated that patients on colistin were, on average, approximately four times more likely to achieve remission compared to patients not treated with this drug whereas patients with radiologically confirmed bronchiectasis were, on average, approximately four times less likely to achieve remission compared with patients without bronchiectasis. Age, use of azithromycin, and hypertonic saline inhalation were also included as covariates in the Cox model but no significant correlation between them and remission was established (HR:1.13; 95%CI:0.99-1.28; p=0.06, HR:1.34; 95%CI:0.46-3.93; p= 0.58 , HR:0.80; 95%CI:0.22-2.87; p= 0.74, respectively). The differences in time to remission for the two significant predictors are shown in Figure 1.
The Schoenfeld tests showed that there was no violation of the proportional assumption in our model (all p >0.05).