Abstract
The “paradoxical” effect of inhaled corticosteroids increasing pneumonia incidence in COPD but reducing post-pneumonia and all-cause mortality is no paradox but due to selection and immortal time biases in the observational studies of mortality http://bit.ly/2Yd3hDh
From the authors:
I thank P. Almagro and co-workers for their letter, which provides updated data regarding the association between inhaled corticosteroids (ICS) and pneumonia mortality in chronic obstructive pulmonary disease (COPD). It confirms the shortage of, and need for, rigorous studies on this question. My article, on the other hand, was focusing on a methodological, causal inference explanation for the paradoxical phenomenon of observational studies reporting that ICS use increases the incidence of pneumonia in contrast with other observational studies reporting that ICS use prior to a pneumonia event decreases mortality [1].
Nevertheless, their conclusion that in using ICS in COPD “clinicians must balance the pneumonia risk with the reduction in both severe exacerbations and the mortality” is not founded on the evidence available to date. This conclusion implies that while ICS increase the risk of pneumonia, they are also effective at reducing the incidence of both severe exacerbations and mortality. The latter is not supported by the totality of evidence. Indeed, a network meta-analysis of 99 randomised controlled trials shows that ICS do not reduce the incidence of severe exacerbations in COPD, on the contrary they appear to increase it, when compared with mono or dual bronchodilators [2]. This meta-analysis also corroborates the increased risk of pneumonia with ICS [2]. With respect to mortality, only observational studies reported a reduction in all-cause death with ICS [3, 4]. However, these studies were shown to be seriously affected by immortal time bias, which tends to exaggerate a drug's beneficial effect [5, 6]. Observational studies that were designed to avoid this bias showed no reduction in mortality [7]. To date, no randomised trial has shown that ICS are associated with a reduction in mortality [2, 8].
Therefore, the evidence-based balance that clinicians should be concerned with when prescribing ICS versus long-acting bronchodilators in patients with COPD is between the risk of pneumonia and perhaps a reduction of mild and moderate exacerbations, but certainly not the severe ones or mortality, as suggested by P. Almagro and co-workers.
The more recent observational studies conducted in real-world clinical practice settings suggest that ICS may be effective at reducing moderate exacerbations only in the subset of patients with significant eosinophilia or frequent exacerbators [9, 10]. Consequently, with the established risk of pneumonia with ICS, this therapy should only be used sparingly in the subset of COPD patients for whom the benefit may outweigh the risk, such as patients with asthma, frequent exacerbators and those with significant eosinophilia [11, 12]. More research is needed to better identify such markers for a more accurate precision medicine approach to the treatment of COPD patients [13].
Shareable PDF
Supplementary Material
This one-page PDF can be shared freely online.
Shareable PDF ERJ-01276-2019.Shareable
Footnotes
Conflict of interest: S. Suissa reports personal fees for board membership from Novartis and Boehringer Ingelheim, grants from Boehringer Ingelheim/Pfizer, personal fees for lectures from AstraZeneca, outside the submitted work.
- Received June 26, 2019.
- Accepted June 27, 2019.
- Copyright ©ERS 2019