Association between ibuprofen exposure and severe COVID-19 infection: a nationwide register-based study

Background: Initially, the World Health Organization (WHO) and national health boards issued a warning against NSAID use in corona virus disease 2019 (COVID-19) patients and recommended that paracetamol or acetaminophen instead should be administered. However, later and current WHO and European Medicine Agency recommendations do not call for avoidance of ibuprofen in COVID-19 patients. Given the current uncertainty of drug safety for NSAID use in patients with COVID-19, we performed a nationwide register-based study on the association of recent ibuprofen exposure and COVID-19 severity. Methods: Using national administrative databases, we identied COVID-19 patients in Denmark between end of February 2020 and April 2, 2020. Patients <30 years of age and heart failure, for whom ibuprofen is not recommended, were excluded. Ibuprofen exposure was dened using the Danish National Prescription Registry, where we had information until January 31, 2020. Endpoint was a composite of severe COVID-19 diagnosis with acute respiratory syndrome, intensive care unit admission or death. Results: Among 1,872 patients, 46 (2.5%) were exposed to ibuprofen prior to COVID-19 infection. Patients with recent ibuprofen exposure tended to be older and more likely to have hypertension, diabetes, myocardial infarction, chronic obstructive pulmonary disease, and cancer, though all insignicant (P>0.05). When adjusting for these covariates, odds ratio was 1.57 [95% CI 0.72-3.38], with 12 ibuprofen-exposed patients meeting the endpoint (26.1% [95% CI 13.4-38.8%]) versus 272 unexposed patients (14.9% [95% CI 13.4-16.4%]), P=0.15. Conclusion: The association between ibuprofen and severe COVID-19 was insignicant, although with a trend towards increased disease severity risk.


Background
The corona virus disease 2019 (COVID-19) pandemic caused by severe acute respiratory syndrome corona virus 2 (SARS-CoV-2) has led to substantial changes in healthcare utilization, morbidity, intensive care resources, and mortality (1)(2)(3)(4)(5)(6)(7). The entry point for this novel coronavirus for its further pathogenesis in humans is thought to be through an angiotensin converting enzyme 2 (ACE2) found in cell surfaces in the lungs, arteries, heart, kidney and intestines (8). It has been hypothesized that a number of drugs including non-steroidal anti-in ammatory drugs (NSAIDs) might facilitate and aggravate infection with COVID-19 (9-10). The mechanisms through which ibuprofen and other NSAIDs may accelerate  infection and lead to further deterioration include upregulation of the ACE2 enzyme expression as well as these drugs may delay diagnostics by masking in ammation and fever (9)(10). At rst, the World Health Organization (WHO) and national health boards issued a warning against the use of NSAIDs in COVID-19 patients and that paracetamol or acetaminophen instead should be administered. However, later and current WHO and European Medicine Agency recommendations do not call for avoidance of ibuprofen to treat COVID-19 symptoms, based on the lack of evidence to support the initially recommended warning (10). Therefore, given the current uncertainty of drug safety for NSAID use in patients with COVID-19, we performed a nationwide register-based study on the association of recent ibuprofen exposure in COVID-19 positive patients and severe infection de ned as a diagnosis of COVID-19 infection with severe acute respiratory syndrome and/or intensive care unit utilization and/or fatal COVID-19 infection. Since very few patients were prescribed other NSAIDs, we only studied the association between ibuprofen exposure and outcome.

Study setting and population
In Denmark, all citizens hold a unique civil registration number used in all healthcare and social contacts and recorded in national administrative databases. From the Danish National Patient Registry, we identi ed COVID-19 positive patients between end of February 2020 and April 2, 2020, through use of ICD-10 diagnosis codes DB342, DB342A and DB972 for unspeci c COVID-19 infection and DB972A for COVID-19 infection with severe acute respiratory syndrome. Patients below 30 years of age as well as patients with heart failure, for whom ibuprofen is not recommended, were excluded.

Exposure
From the Danish National Prescription Registry, data on prescription medication was available until February 1, 2020; thus, we assessed ibuprofen exposure in a one-month span from January 1 to January 31, 2020.

Covariates
We included information on patient age and sex from the Danish Civil Registration System. We assessed the following comorbidities: Hypertension, diabetes, prior myocardial infarction, chronic obstructive pulmonary disease, hypertension, any malignancy, and rheumatic disease. Hypertension was de ned as redemption of at least two antihypertensive drugs in two consecutive 100-day periods prior to the COVID-19 diagnosis, as done previously (11). All other comorbid conditions were identi ed from the Danish National Patient Registry up to ve years prior to COVID-19 diagnosis.

Endpoint
The study endpoint was a composite of severe COVID-19 diagnosis (ICD-10: DB972A), admission to an intensive care unit (ICU), or death. The Danish National Patient Registry was used to identify the rst two components of the composite endpoint, whereas vital status was identi ed from the Danish Civil Registration System.

Statistics
Continuous variables are reported using median and 25-75 percentiles. Categorical variables are reported using counts and percentages. Propensity for ibuprofen treatment was evaluated using logistic regression. Furthermore, we used unadjusted, age-and sex-adjusted, and fully adjusted logistic regression to associate the risk of severe COVID-19 infection, death or intensive care unit admission with ibuprofen exposure. Fully adjusted models included the following covariates: age, sex, diabetes, prior myocardial infarction, chronic obstructive pulmonary disease, hypertension, any malignancy, and rheumatic disease. Age was divided into the following categories: 30-50, >50-60, >60-70, and >70 years. Reported are odds ratios with 95% con dence intervals (CI). Data management and analyses were performed using R, version 3.6.1 (12).

Ethics
Register-based studies do not require ethical committee approval or patient consent according to Danish legislation. In accordance with the General Data Protection Regulation (GDPR), the data responsible institute in the Capital Region of Denmark has approved the use of the data sources for research purposes (approval number P-2019-191). Data are accessed on secure servers under Statistics Denmark and cannot be shared according to Danish legislation.

Patients and characteristics
Of a total of 2,220 patients with COVID-19 infection between end of February and April 2, 2020, 1,872 patients without heart failure and age ≥30 years formed the study population ( Figure 1). A total of 46 patients (2.5%) were exposed to ibuprofen prior to COVID-19 infection. Patients with recent ibuprofen exposure tended to be older and more likely to have hypertension, though insigni cant, as well as the prevalence of diabetes, myocardial infarction, chronic obstructive pulmonary disease, and cancer were numerically higher for patients with ibuprofen exposure ( Table 1).

Propensity of ibuprofen treatment
Propensity of ibuprofen treatment for variables listed in Table 1 are shown in Table 2. None of these factors were signi cantly associated with a propensity for ibuprofen use, although there was a trend towards an increased use for hypertension patients (  Table  3.

Discussion
In this nationwide study on the association between ibuprofen exposure and severe COVID-19 infection, de ned as a COVID-19 diagnosis with severe acute respiratory syndrome, intensive care unit admission, or fatal trajectory, we found no signi cant association between recent ibuprofen drug exposure and the composite endpoint.
Respiratory, septic and cardiovascular complications and prolonged recovery have been linked to NSAID use in COVID-19 infectious disease (9,13). Initially, the WHO and several national health boards and experts advocated against the use of NSAID in COVID-19 infected patients and recommended that paracetamol or acetaminophen instead should be administered (13). Later, the WHO changed track and now currently do not call for avoidance of ibuprofen to treat COVID-19 symptoms. The American Food and Drug Administration (FDA) has stated their awareness of reports of potential worsening of COVID-19 infection on the basis of NSAID use, but the agency also stresses that there is currently insu cient evidence to support that NSAID should worsen the clinical course of COVID-19 infection. Our results do not provide su cient evidence to change the current recommendations. However, although the relation between recent ibuprofen exposure and severe COVID-19 infection was insigni cant in our study, the odds ratio was 1.57 with a 95% con dence interval between 0.72-3.38. As such, we cannot rule out an insigni cant result due to a type II error (insu cient power).
On the other hand, a signi cant relation between ibuprofen and severe COVID-19 infection could be a result of confounding by indication, given the observational nature of our study. In our study, ibuprofen users were de ned on the basis of prescription lls and since ibuprofen also can be purchased as an over-the-counter (OTC) drug in Denmark, there are number of limitations related to ibuprofen exposure in our study. Apart from potential misclassi cation of ibuprofen exposure as unexposed, when in fact patients were taking OTC ibuprofen, patients on prescription ibuprofen are likely to be different than patients on OTC ibuprofen or patients not treated with ibuprofen with regard to underlying conditions both related and unrelated to the drug indication for ibuprofen. Therefore, although insigni cant, the trend we see towards an increased risk of severe COVID-19 infection in our data may be related to conditions associated with ibuprofen use. However, in our analysis of propensity of ibuprofen treatment, no factors shown in Table 1 were signi cantly related to ibuprofen use, although a trend towards an increased use in patients with hypertension was seen.
Another potential source of misclassi cation of the ibuprofen exposure in our study is related to the time period during which we were able to assign patients as exposed versus unexposed to ibuprofen. We were only able to assess NSAID drug redemptions until the end of January, 2020, and the rst positive COVID-19 case in Denmark was in the end of February, 2020. Although we can argue that ibuprofen prescription was not made on the basis to treat COVID-19 infection, we cannot exclude that there were patients exposed to ibuprofen in February and onwards, but not in January. Since those patients then would be incorrectly registered as unexposed, results would be directed towards no association between ibuprofen and COVID-19 severity, and we cannot rule out that this misclassi cation has in uenced the insigni cant results seen in our study.
In addition to the limitations mentioned above, follow-up data on outcome was limited to April 2, 2020, which may limit the registration of severe COVID-19 infection for patients diagnosed with COVID-19 in the end of the study period. Given the observational design, results are associations, and no conclusions with regard to causality can be made. Although we adjusted for several confounding factors, we cannot rule out residual or unmeasured confounding. Despite the lack of laboratory data to con rm that each case had a positive swab test, coding of both tested persons with tentative diagnosis codes and eventually those positive swabs with de nite diagnosis codes have been and are systematically performed in Denmark. Although most COVID-19 infections in Denmark are located to the capital and more densely populated areas, the use of nationwide data with consecutively included patients with positive COVID-19 infections minimizes selection bias.

Conclusions
In conclusion, we did not nd ibuprofen drug exposure to be associated with severe and fatal trajectory in COVID-19 infection, but further investigations are needed to con rm our preliminary results.

Con icts of interest
None of the authors have any con icts of interest to report.    Figure 1