The COVID‐19 ibuprofen controversy: A systematic review of NSAIDs in adult acute lower respiratory tract infections

In light of the recent safety concerns relating to NSAID use in COVID‐19, we sought to evaluate cardiovascular and respiratory complications in patients taking NSAIDs during acute lower respiratory tract infections.

almost 2.4 million deaths, making them the sixth commonest cause of mortality in all age groups. 1,2 The first line pharmacological treatment depends on the aetiology of the infection, i.e. bacterial, viral or fungal. However, non-steroidal anti-inflammatory drugs (NSAIDs) are commonly used to alleviate symptoms such as pain, fever and cough regardless of aetiology. All NSAIDs work by blocking inflammatory prostanoids produced by the enzyme cyclooxygenase-2 (COX-2). 3 COX-2 is induced at the site of inflammation 4 where it mediates swelling and pain and by both bacterial and viral mimetics 5 where it mediates fever. 5 A prospective study in France found that at least 50% of patients with pneumonia received an NSAID prescription. 6 Furthermore, this number is likely to be an underestimate as NSAIDs are readily available and can be purchased over the counter.
With the recent outbreak of COVID-19, NSAIDs have been in the spotlight as French health officials recommended avoiding NSAIDs such as ibuprofen over concerns of worsening the course and outcome of COVID-19 infection. This led to substantial controversy, with several experts suggesting that NSAIDs were best avoided in patients with COVID-19 given the drugs' safety profile. 7,8 However, a response by the European Medicines Agency (EMA) stated that there was no scientific evidence establishing a link between ibuprofen and worsening of COVID-19. 9 This is not the first time the safety of NSAIDs during pandemics has been bought into question-it has been hypothesised that the high fatality rate during the 1918 influenza pandemic, especially amongst young adults, might have been partly due to use of the NSAID aspirin. Early deaths often presented with "wet" and haemorrhagic lungs, which we now know is consistent with hyperventilation and pulmonary oedema secondary to salicylate toxicity due to the high doses used at that time (8-31.2 g per day). 10 While we do not yet know of the respiratory effects of therapeutic doses caused by NSAIDs in infections such as COVID-19, these drugs are associated with cardiovascular, 3 renal 11 and gastrointestinal 12 side effects and in people with a specific sensitivity, they can cause asthma. 13 Given that NSAIDs are a mainstay of symptomatic relief therapy in respiratory infections, it is now important to elucidate if short-term NSAID use is detrimental to not only those with COVID-19, but also for patients with lower respiratory tract infections in general. In light of the ibuprofen-COVID-19 debate and the rapidly unfolding situation of the current pandemic, the author group collaborated to systematically assess the evidence pertaining to the safety of short-term NSAID use in lower respiratory tract infections. Our objective was to evaluate cardiovascular and respiratory complications in patients taking NSAIDs during an acute lower respiratory tract infection.

| Selection of studies for inclusion
his systematic review and meta-analysis was performed according to the methods described in the Cochrane Handbook for Systematic Reviews of Interventions 14 and the PRISMA guidelines. 15 The full review protocol was registered on PROS-PERO (CRD42020176168) prior to data extraction. A systematic electronic search of Embase, Medline, Cochrane Controlled Register of Trials (CENTRAL) and Clinical Trials. Gov was performed to identify relevant publications from inception to 21 March 2020.
The full details of the search terms can be found in Appendix A. Citations were screened by title and abstract by two independent reviewers from a pool of six. Any disagreements were resolved by a third reviewer. Relevant full texts were then retrieved for screening by two independent reviewers from a pool of six. Any disagreements were resolved by a third reviewer and reasons for exclusion were documented. Any non-English articles were translated. References of included studies and review articles were manually screened for further eligible studies. Authors were contacted when necessary.

| Types of studies for inclusion
We included clinical randomised controlled trials (RCTs) assessing the safety profile of NSAID use during an acute lower respiratory tract infection. The search was extended to include the following quasi experimental and observational studies: controlled pre/post intervention, prospective and retrospective cohort, cross-sectional and casecontrol studies. We excluded case reports, pharmacodynamic or pharmacokinetic studies that did not report clinical outcomes and studies with less than five patients.

| Participants/population
The search was limited to studies that reported the safety profile of any NSAID used during an acute lower respiratory tract infection in adults (age ≥18 years). The lower respiratory tract was defined as a distal airway infection including either the trachea, bronchi or lungs. Types of infections included viral, typical and atypical bacterial. We excluded chronic lung infections, aspiration pneumonia and fungal infections. Acute infections in patients with chronic lung diseases as specified by authors of identified studies, e.g. acute infections in patients with chronic obstructive pulmonary disease, were included. Patients with chronic NSAID use were excluded as we were primarily interested in short-term NSAID use.

| Interventions/comparator
The interventions included any non-steroidal anti-inflammatory drug including but not limited to: aspirin, ibuprofen, naproxen, diclofenac and celecoxib. The comparators included nonexposed group including placebo, alternative NSAID or any other drug.

| Outcome domains of interest
Our primary outcome was all cause mortality and longest available follow-up period. Our secondary outcomes included major adverse cardiac and cerebrovascular events (MACCE), renal complications and respiratory complications (including adult respiratory distress syndrome [ARDS], mechanical ventilation, tracheostomy, abscess/empyema). We did not include gastrointestinal complications.

| Data extraction
Data from included studies was extracted onto an online form using Covidence systematic review software (Veritas Health Innovation, Melbourne, Australia. Available at www.covidence.org). Data relating to summary estimates of baseline characteristics and the outcomes stated were extracted by two independent authors from a pool of six with any disagreements being resolved by consensus or consulting a third reviewer. The risk of bias for each included study was independently assessed by two review authors from a pool of six with any disagreements resolved by consensus or consulting a third reviewer. For RCTs, criteria from the Cochrane Handbook for Systematic Reviews of Interventions 14 was used and the ROBINS-I tool 16 for observational studies. The risk of bias from each study was taken into consideration when discussing any treatment effects.

| Subgroup analysis
We had planned to carry out the following subgroup analysis: • Types of NSAID used (e.g. Ibuprofen vs Aspirin) • Type of infection (e.g. viral vs bacterial) • Effect of any co-morbidities (e.g. smoking, hypertension) using meta-regression.
However, we did not perform any planned subgroup analysis as there was an insufficient number of studies identified for inclusion and we were unable to adjust for confounding using meta-regression.

| Data analysis
Due to significant heterogeneity and the poor quality of included studies, we were unable to pool data and conduct any meaningful metaanalysis and instead the data has been described narratively. Relative risks (RR) with 95% confidence intervals (CI) for the individual studies were presented and calculated using the modified Wald method.
Although analyses of the number needed to treat for an additional harm was planned, it was omitted in the final analysis as the authors felt this could be misleading for readers in view of all studies having a high risk of bias. Since less than ten studies were included, publication bias was not assessed.
Two independent review authors independently judged the overall quality of the evidence using the five GRADE criteria (study limitations, consistency of effect, imprecision, indirectness, and publication bias). 17 All P-values were two-sided, and P < 0.05 were considered statistically significant.

| Patient and public involvement
It was not possible to involve patients or the public in the design, conduct, reporting or dissemination plans of our research.

| Nomenclature of targets and ligands
Key protein targets and ligands in this article are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY, 18 and are permanently archived in the Concise Guide to PHARMACOLOGY 2017/18. 19 3 | RESULTS • The LIBERATE Trial (NCT04334629), which is an ongoing multicentre RCT comparing the use of lipid ibuprofen versus standard care for acute hypoxemic respiratory failure due to COVID-19; the trial has begun recruitment. Eight studies were included in the final analysis (Table 1).

| Risk of bias of the included studies
All six observational studies were deemed to have an overall high risk of bias, primarily driven by lack of adjustment for confounding and bias in selection of the reported result. The two RCTs were also deemed to have a high risk of bias. A summary for each domain is presented in Figure 2.

| Outcomes
A summary of the included studies and outcomes is presented in Table 1.

| Mortality
Five studies reported mortality of which one was an RCT. The mortality data in Basille et al. 27 included data for chronic NSAID use and was therefore excluded from the analysis. Hung et al. 20 reported a reduction in 30 day mortality; RR 0.11 (95% CI 0.01-0.89). It is important to note that the study investigated the effects of both clarithromycin and naproxen in combination with oseltamivir and therefore any benefits seen may be due to the use of concomitant antibiotics and not NSAID use. Basille et al. 26 and Voiriot et al. 28 reported an uncertain effect on 30 day mortality, Messika et al. 22 reported an uncertain effect on in-hospital mortality, and Epperly et al. 24 reported an uncertain effect on 90 day mortality. Although the mortality data could not be pooled due to heterogeneity, the overall direction of effect suggests there may be a trend towards a reduction in mortality.

| Cardiovascular complications
Cardiovascular complications were not reported in any of the included studies. Whilst screening references of included studies, two case-crossover design studies were identified and reported myocardial infarction 29 and stroke. 30 The authors made a decision not to include these as they used unconventional methodology in the study design.    NSAID, non-steroidal anti-inflammatory drug.

| Dialysis
Two small observational studies in patients with pneumonia reported the need for dialysis. Both Messika et al. 22 ; RR 1.1(95% CI 0.1-9.1) and Voiriot et al. 28 ; RR 0.8 (95% CI 0.2-2.8) reported an uncertain effect on the need for dialysis. Such small sample sizes and the extremely wide 95% CI barred any clinically meaningful conclusions to be drawn.

| Other adverse events (excluding gastrointestinal complications)
Two observational studies in pneumonia patients reported uncertain effects on rates of organ failure. Llor et al. 22 reported uncertain effects on rates of adverse events; RR 1.8 (95% CI 0.55-6.15), although the majority of these were reported as mild.

| Summary
An overall summary of the treatment effects for each outcome measure with GRADE quality of evidence is presented in Table 2.

| DISCUSSION
In this systematic review of NSAID use during acute lower respiratory tract infections in adults, we found that the existing evidence for mortality, pleuro-pulmonary complications, rates of mechanical ventilation and need for dialysis or organ failure is of extremely poor quality and very low certainty, with most reports being observational/registry data with high risks of bias due to lack of adjustment for confounding variables. Moreover, we found no studies, subject to our selection criteria, where cardiovascular outcomes were documented. The apparent lack of quality clinical studies addressing benefits and risk of these drugs in acute lower respiratory tract infections was unexpected considering the frequency of NSAID use and the known cardio-renal and gastrointestinal side effect of NSAIDs.
The included studies demonstrated a trend towards higher rates of pleuro-pulmonary complications following short-term NSAID use.
Although this review focused on adults, these findings have been echoed in paediatric studies. [31][32][33][34][35][36] There are essentially two hypotheses that may explain these findings, both of which warrant further investigation: 1. Indirect: NSAIDs are associated with renal dysfunction but in our analysis we found insufficient evidence to draw any meaningful conclusions..
Data relating to only 196 patients with 15 events in total across two studies were included, so this should also be interpreted cautiously, especially since the nephrotoxicity of NSAIDs is well known. 11 Importantly, none of the studies identified in this review reported cardiovascular events. This is unexpected since, with the exception of aspirin, all NSAIDs including older (traditional) drugs such as ibuprofen and newer COX-2 inhibitors, such as celecoxib, are known to increase cardiovascular risks 40 and that the increased risk of myocardial infarction is seen as early as after 1-2 weeks of use. 41 It is therefore crucial that further studies consider cardiovascular and renal side effects. heterogeneity, we were unable to pool the data and conduct any meaningful meta-analysis. Despite these, until more data become available including from the ongoing ENACOVID trial (NCT04325633) and The LIBERATE Trial (NCT04334629), this report represents a summary of the best available evidence that is clinically relevant to frontline staff.

| CONCLUSION
In this systematic review of NSAID use during acute lower respiratory tract infections in adults, we found that the existing evidence for mortality, pleuro-pulmonary complications and rates of mechanical ventilation or organ failure is of extremely poor quality and very low certainty. Cardiovascular or renal complications could not be adequately addressed. There may be an overall trend towards an increase in pleuro-pulmonary complications but a reduction in mortality. However, these results must be interpreted with extreme caution given the high risk of bias of the included studies, the very low quality of evidence and the results' generalisability to COVID-19 is unclear.
Despite this, the summary represents the best available evidence for NSAID use in adult patients with acute lower respiratory tract infections. Mechanistic studies and clinical studies, especially pertinent to COVID-19, are urgently required.