This large-sized Real-World Data (RWD) study suggests that outpatient glucocorticoid treatment is consistently associated with a higher risk of severe COVID-19 outcomes, and that this effect is seemingly independent of the active ingredient. Furthermore, our results suggest that, as use approaches the index date and as the dose increases, the risk of susceptibility, progression, hospitalisation, and mortality due to COVID-19 rises. Specifically, risk of death is almost fourfold higher among subjects who use glucocorticoids in the previous month.

The increased risk found in our study for different outcomes is in line with the findings of some studies conducted on patients with autoimmune or respiratory disease, among whom ambulatory glucocorticoid use has been associated with an increased risk of susceptibility (Singh et al. 2021), hospitalisation (Brodin et al. 2022), disease progression (Ungaro et al. 2021) and mortality (Calderón-Parra et al. 2022; Malekpour et al. 2023). In addition, our results suggest that this effect is independent of the approved active ingredient, and is observed for both high and low doses (≤ 0.34 DDDs/day), unlike other studies on patients with autoimmune disease, in which the greatest risk of susceptibility and severe COVID-19 outcomes was exclusively reported for users of high doses of glucocorticoids (Gianfrancesco et al. 2020; Shin et al. 2021).

The increased susceptibility observed in our study (aOR 1.29; 95% CI 1.19–1.41, p < 0.001) is consistent with the increased risk of infections found by Dixon et al. in their meta-analysis (Dixon et al. 2011). This could be accounted for by the effect of these medications on the suppression of cellular immunity and alterations in phagocyte function (Cutolo et al. 2008). Insofar as disease progression is concerned, other studies have reported that SARS-CoV-2 infection could present with more severe symptoms in patients with autoimmune (Khan et al. 2021) or respiratory disease (Adir et al. 2021) treated with glucocorticoids, due in part to the reduction in SARS-CoV-2 RNA clearance (FakhriRavari et al. 2021). Our results would support this hypothesis, since risk of disease progression (aOR 1.69; 95% CI 1.43–2.00, p < 0.001) is greater among patients with ambulatory glucocorticoid use than among non-users.

The 1.8-fold higher risk of hospitalisation (aOR 1.79; 95% CI 1.56–2.05, p < 0.001) observed in our study could be due to suppression of immune T-cell response, which would hinder initial diagnosis of COVID-19 (Migita et al. 2013), along with the harmful effect of treatment with glucocorticoids in the initial stages (Wagner et al. 2022). In this respect, this increased risk of hospitalisation shows a high degree of internal consistency in our study, since we also found an increased risk of susceptibility (aOR 1.29; 95% CI 1.19–1.41, p < 0.001) and disease progression (aOR 1.69 95% CI 1.43–2.00, p < 0.001).

Our results show that the risk of mortality due to COVID-19 among hospitalised patients with ambulatory glucocorticoid treatment in the preceding month is fourfold higher than that of non-users (aOR 3.85; 95% CI 2.63–5.62, p < 0.001). This higher risk of mortality is in line with what has been suggested in the previous studies but, unlike ours, this evidence is based on the studies conducted on very defined populations, such as subjects with autoimmune (Ward et al. 2022) or respiratory disease (Adir et al. 2021), users of high doses (Brodin et al. 2022), hospitalised patients (Suárez-García et al. 2021), and subjects with a PCR + test (Ku et al. 2023). This important risk of mortality could be accounted for: (i) By the increased risk of developing serious (George et al. 2020) or invasive fungal infections (Lionakis and Kontoyiannis 2003); (ii) By the harmful effect of glucocorticoid treatment in the initial stages of the disease (Li et al. 2020); and (iii) By the lack of effectiveness of glucocorticoid treatment in hospitalised patients, due to possible resistance to glucocorticoids (Bruscoli et al. 2022), as a possible consequence of prolonged exposure to inflammatory cytokines (Quax et al. 2013).

In our study, the risk of mortality among users in the previous 3 months (aOR 2.30; 95% CI 1.68–3.15, p < 0.001) would appear to be higher than the risk of hospitalisation (aOR 1.79; 95% CI 1.56–2.05, p < 0.001). This could be accounted for by possible resistance to glucocorticoid treatment (Barnes and Adcock 2009) among inpatients or the appearance of side-effects associated with prolonged exposures to or high doses of glucocorticoids (Rhen and Cidlowski 2005). This lack of a glucocorticoid-induced effect on hospitalised patients taking glucocorticoids could give rise to a higher risk of mortality. It has been suggested that prolonged exposures to inflammatory cytokines, such as that which occurs in cases of severe COVID-19, could develop resistance to glucocorticoids (Quax et al. 2013).

The effects observed on all the outcomes analysed in our study are greater when they are associated with higher doses and use closer to the time of infection. This finding lends great internal consistency to our results, and suggests that the effect of glucocorticoids is maintained throughout the clinical disease course and favours progression to severe clinical profiles. It has been suggested that this could be due to the fact that use of glucocorticoids might delay the manifestation of the initial symptoms of COVID-19, (Calderón-Parra et al. 2022) or alternatively, to the harmful effect that these drugs have on the initial stages of the disease (Li et al. 2020). Moreover, as our results show, the closer glucocorticoid use is to the index date, and hence to infection, the higher the risk for all outcomes. This harmful effect in the initial disease stages, has also been suggested in other viral infections, such as that caused by SARS-CoV (Lee et al. 2004) or influenza-virus associated pneumonia (Ni et al. 2019).

Our study’s main strength lies in the internal consistency of our data for all outcomes studied. Another aspect of note is its large sample size, which made it possible to carry out separate analyses of the effect of dose, window of exposure and different glucocorticoids, and thereby rule out differences between them for the first time. Similarly, the fact that this is an RWD study and includes all the diagnosed cases of COVID-19 means that there is no selection bias, and that we were able to adjust the outcomes for socio-demographic variables, comorbidities, and previous treatments with other drugs, which might have otherwise confounded the effect observed. Lastly, the variables of exposure were obtained from dispensing rather than from prescription data, which reduces the risk of misclassification.

Our study has a series of limitations which should be considered when it comes to interpreting the results. Firstly, the fact that it is an observational study with secondary databases means that one cannot rule out the existence of confounding factors not included in our analysis or the possibility of misclassification in some variables due to prescriptions issued by private healthcare sources. Secondly, it could be thought that not having information about some indications for glucocorticoids could lead to confounding by indication bias, since glucocorticoids are used as treatment for chronic inflammatory and immune diseases (Barnes 2011), such as COPD, associated with a higher risk of severe outcomes (Christenson et al. 2022). We do not consider this to be a limitation, however, since an important part of indications for glucocorticoids are either not related to an increased risk (Choi and Seeger 2005), are not associated with a specific disease (Fardet et al. 2011), or are included in the variables of adjustment of our results (Waljee et al. 2017). Thirdly, the lack of matching in the susceptibility and disease progression substudies might be a limitation. That said, this lack of matching in the case − control substudies implies a decrease solely in efficacy, and not in the overall validity of the results (Rothman et al. 2008; Rose and van der Laan 2009). Lastly, it is likely that in the group of subjects without a PCR + test, asymptomatic subjects or subjects with mild symptomatology may have been included due to the difficulty of access to COVID-19 diagnostic tests in Spain, especially at the outbreak of the pandemic.

The increased risk of severe COVID-19 outcomes should be taken into account by clinicians when it comes to prescribing these medications, especially at high doses. Due to the many side-effects associated with glucocorticoids (Rhen and Cidlowski 2005; Moghadam-Kia and Werth 2010; Waljee et al. 2017), their use should be optimised by employing the smallest possible doses. Our results will add evidence to this recommendation, since the risk of severe outcomes and susceptibility increases among users of high doses.

Our results suggest that ambulatory use of systemic glucocorticoids is associated with an increased risk of severe COVID-19 outcomes, regardless of the active ingredient consumed, even at low doses. The effect of ambulatory glucocorticoid use increases as the clinical course of the disease advances. Furthermore, all risks become consistently higher as drug use approaches closer to the index date. Given the large effect magnitude found in our study and the fact that glucocorticoids are one of the medications used in intrahospital treatment of COVID-19, our results suggest that it would be of interest to conduct a specific intrahospital follow-up of patients who have consumed glucocorticoids on an ambulatory basis.