Discussion
To our knowledge, this is the first study focusing on the risk of hospitalisation due to pneumonia for patients with COPD related to ICS particle size. Among patients with COPD using ICS, we found a strong association between receiving ICS in an extrafine particle formulation and having a lower risk of pneumonia hospitalisation. The result was robust for the adjustments made in the main analysis. In addition, the subgroup analysis, excluding patients not receiving spray devices, showed an even lower HR of hospitalisation due to pneumonia for the patients treated with extrafine particle ICS. Furthermore, we did a propensity-matched cohort for sensitivity analysis with similar HRs to the main analysis.
The underlying mechanism behind the lower risk of pneumonia for patients with COPD treated with extrafine particles remains unclear. Patients receiving extrafine particle ICS were comparable to those receiving other formulations, but slightly younger and healthier, nevertheless the results were robust for the adjusted analysis. Patients receiving extrafine particle ICS did on average receive a higher median budesonide equivalent daily dose, potentially leading to a higher risk of pneumonia in contrast to our results. The risk of pneumonia may be lower due to smaller particle size leading to increased passive absorption or different liposolubility properties affecting elimination.
To date, no clinical trials have been conducted comparing extrafine particle ICS to standard particle size ICS for patients with COPD. Previous studies mainly focus on patients with asthma or obstructive pulmonary disease in general. Registry studies pooling obstructive lung diseases might confound the results, due to the different risks of pneumonia related to ICS treatment for patients with asthma and COPD17 and corresponding different prescription patterns for these two separate disease entities. A register study investigated the risk of pneumonia for patients with obstructive lung disease and showed a lower risk of pneumonia in patients stepping up to extrafine particle ICS compared with standard size particle ICS.11 However, this risk reduction was not significant in the adjusted analysis. Furthermore, this study was limited to a selected group of patients, excluding patients receiving budesonide. Another study examined the effectiveness of extrafine particles compared with standard size particles and did not show differences in the risk of pneumonia, however, they only found very few cases of pneumonia (four confirmed).18 Similarly, one-third study showed no difference in incidence of pneumonia for patients with COPD receiving triple therapy with or without extrafine particle ICS, although based on few outcomes.19
Our study has several strengths. We included more than 35 000 patients with COPD followed in outpatient clinics receiving ICS. We had approximately 1400 patients receiving extrafine particle ICS of which about 10% had a hospitalisation due to pneumonia. This dataset contains information on multiple confounders including age, BMI, smoking status and FEV1 enabling adjustments and matching. We have complete data for the primary outcome, redeemed prescriptions and death. The baseline characteristics of patients treated with extrafine particle ICS, compared with those receiving standard particle size ICS, were quite similar. Although there are no clear clinical guidelines for prescribing extrafine particle ICS, one might expect more ill patients to receive this treatment. However, the baseline data do not support this hypothesis. This similarity in baseline characteristics makes our study somewhat less reliant on adjustments. Data completeness was high for all the variables used for adjustment. Most patients with missing data had missing data on multiple parameters (FEV1, BMI, smoking status) at the same time, and these patients had a similar age to patients with full data completeness, this indicates most of the missing data is due to registration problems, indicating data are missing completely at random. The hypothesis was tested with different sensitivity analyses with similar robust results.
Although our study has several strengths, some limitations deserve consideration. The diagnosis of pneumonia was based on hospitalisation records and may be subject to misclassification by clinicians, leading to potential under-reporting or over-reporting of the outcome. However, it is unlikely that the ICS particle size would affect the accuracy of the pneumonia diagnosis, and as such, any bias attributed to this possible misclassification would most likely be non-differential. Furthermore, ICD-10 codes have previously been validated as a method for identifying hospitalisations due to pneumonia.20 The study used data on redeemed prescriptions rather than actual medication intake, which may probably result in an overestimation of ICS consumption. However, this error is likely to be small given that it is based on multiple collected prescriptions and is unlikely to significantly impact the primary outcome. A potential criticism of our study design is that the comparison between the two groups was not solely based on ICS particle size, but also on the type of ICS used (budesonide, fluticasone, etc). Previous studies have indicated a possible excess risk of pneumonia associated with fluticasone.21 However, when accounting for dose equivalency, such associations to fluticasone cannot be reproduced, and the biological correlate to such a notion seems less clear.4 While our study cohort consisted of patients with COPD followed in outpatient clinics by pulmonologists, treated with ICS, we believe that the influence of particle size on pneumonia risk is likely to hold relevance across broader COPD patient populations. However, caution should be exercised when generalising our findings to patients with COPD followed by general practitioners, as they may exhibit lower pneumonia hospitalisation rates. As with other prospective studies, it is not possible to rule out remaining residual confounding.
In conclusion, we found that extrafine particle ICS for patients with COPD is strongly associated with a lower risk of hospitalisation due to pneumonia compared with standard particle size ICS. The signal was robust through three different adjusted models. Until randomised controlled trials can inform us further, ICS formulations with extrafine particles, using the lowest possible dose, seem like a safe choice, in patients with COPD at high risk of this serious adverse effect to corticosteroids.