The use of statins was associated with reduced COVID-19 mortality: a systematic review and meta-analysis

Abstract Background Statins are widely used to treat people with metabolic and cardiovascular disorders. The effect of statins on coronavirus disease 2019 (COVID-19) is unclear. To investigate the association between statins and COVID-19 outcomes and, if possible, identify the subgroup population that benefits most from statin use. Materials and methods A systematic review and meta-analysis of published studies that included statin users and described COVID-19 outcomes through 10 November 2020. This study used the generic inverse variance method to perform meta-analyses with random-effects modelling. The main outcomes were evaluation of the need for invasive mechanical ventilator (IMV) support, the need for intensive care unit (ICU) care and death. All outcomes were measured as dichotomous variables. Results A total of 28 observational studies, covering data from 63,537 individuals with COVID-19, were included. The use of statins was significantly associated with decreased mortality (odds ratio [OR] = 0.71, 95% confidence interval [CI]: 0.55–0.92, I2=72%) and the need for IMV (OR = 0.81, 95% CI: 0.69–0.95, I2=0%) but was not linked to the need for ICU care (OR = 0.91, 95% CI: 0.55–1.51, I2=66%). Subgroup analysis further identified five types of studies in which statin users had even lower odds of death. Conclusions The use of statins was significantly associated with a reduced need for IMV and decreased mortality among individuals with COVID-19. Statins may not need to be discontinued because of concern for COVID-19 on admission. Further randomized controlled trial (RCTs) are needed to clarify the causal effect between statin use and severe COVID-19 outcomes. Key messages Participants in five types of studies were shown to have even lower odds of death when taking statins. The use of statins was significantly associated with a reduced need for invasive mechanical ventilation and decreased all-cause mortality among individuals with COVID-19. However, statin use did not prevent participants from needing care in the intensive care unit. The results justify performing randomized controlled trials (RCTs) to validate the benefits of statins on COVID-19 outcomes.


Introduction
Coronavirus disease 2019 (COVID-19) has been a tremendous threat to human life since the pandemic began. People at high risk of unfavourable outcomes caused by COVID-19 include elderly persons and individuals who have chronic metabolic and cardiovascular comorbidities [1,2]. Statins are a class of medications that are widely used to manage metabolic syndrome and cardiovascular diseases. A nationwide report showed that up to 30% of the population aged !45 years took statins [3]. Whether the use of statins will affect the clinical course and prognosis of COVID-19 is gaining increasing attention [4,5]. Statins are primarily used to lower serum lipids by inhibiting HMG-CoA reductase. In addition, statins are also known to have pleiotropic effects, including (1) anti-inflammation, (2) immunomodulation, (3) upregulation of angiotensin-converting enzyme 2 (ACE2) receptor expression, (4) antithrombosis and (5) antioxidation [4,[6][7][8]. Some of the effects may have complicated interactions with COVID-19. For example, ACE2 is the main entry receptor of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Chronic use of statins may theoretically increase susceptibility to COVID-19. On the other hand, ACE2 mediated conversion of angiotensin II to angiotensin (1-7) peptide, which can protect the lung from acute injury [6,9,10]. The overall influence of statin use on COVID-19 remains unclear.
An increasing number of researchers are focussing on the relationship between statins and COVID-19 [11][12][13][14]. Currently, these studies are all observational, and there are no published randomized controlled trials (RCTs). To date, most studies agree that statins do not have to be discontinued after the onset of COVID-19 [11,13,15]. However, there is still no concrete conclusion about the impact of statins on the prognosis of COVID- 19. This study aims to explore the relationship between statins and COVID-19. Through a comprehensive systematic review and meta-analysis, we expect to answer two clinical questions: (1) will the use of statins lead to a favourable/poor outcome of COVID- 19, and (2) can statins decrease the mortality of COVID-19 in a subgroup of statin users? The study results will provide valuable information to health care workers and guide the choice of potential participants in future RCTs.

Search strategy
This study was a systematic review and meta-analysis conducted following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) recommendations [16] (Supplementary Table S3). We searched the PubMed, Cochrane Library, Embase and MEDLINE Online databases for relevant articles published through 10 November 2020, using the keywords and related MeSH terms as follows: "COVID-19", "statin' and "HMG-CoA reductase". There was no restriction on language or publication status. The reference lists of eligible studies were also manually searched to identify any additional relevant studies. Detailed search strategies are described in Supplementary  Table S1.

Selection criteria
The inclusion criteria for this study are described as follows in the order of population, intervention, comparator, outcome, study design (PICOS) formulation. The population were individuals with positive/confirmed COVID-19; the intervention was the use of statin before or during the hospital course of COVID-19; the comparators were those who did not take statin before or during COVID-19. This study set up three outcomes to evaluate the relationship between statin use and COVID-19, including intensive care unit (ICU) care, use of an invasive mechanical ventilator (IMV) and death. Any study that described at least one of the outcomes was included for assessment and analysis. Eligible study designs included RCT, cohort, clinical trial, case-cohort and cross-over design. The following types of articles were excluded, including review articles, hypothesis, case reports, articles in a non-English language, articles focussing on paediatric populations (17 years of age or younger), articles with a Newcastle-Ottawa scale (NOS) score of 6 or below, articles providing inadequate information and articles not relevant to the study goal. Two investigators (TCP and PLT) independently performed a systematic review using the same criteria and included studies on the basis of agreement. Upon disagreement, a third investigator (PCL) joined and helped make the final decision.

Data extraction
Eligible studies were first deduplicated by EndNote X9. Then, two investigators (KSW and PLT) independently extracted data from the included studies using an established data collection form. Collected variables included the first author's name, publication year, study country, study design, sample size, study period, demographics of participants, follow-up duration, diagnostic method of COVID-19, study quality and outcomes. We also contacted corresponding authors to gather missing data when needed. Both raw data and results presented as relative risk/odds ratio (OR)/hazard ratio were included. If a study provided unadjusted and adjusted results, we extracted the adjusted results. The case fatality rate (CFR) of each study was calculated using the number of deceased individuals divided by the number of all individuals [17]. Any discrepancies in data extraction or quality assessment were discussed with a third investigator (TCP) to reach a final agreement.

Quality and risk of bias assessments
This study used the Cochrane risk of bias tool [18] and NoS [19] to appraise the quality of RCTs and observational studies. The NoS consists of eight items covering three broad perspectives: The selection of the study groups, the comparability of the groups and the ascertainment of either the exposure or outcome of interest. The total maximum score of these three perspectives is nine. A study that scores equal to or higher than seven is considered high-quality research. Besides, a funnel plot was used to evaluate the existence of publication bias.

Data synthesis and statistical analysis
This study used the generic inverse variance method to perform meta-analysis since some of the enrolled studies reported results with an inverse design. We conservatively chose random effect modelling for analysis since differences in the study population and study design between studies were expected. If an article presented both unadjusted and adjusted results, we retrieved the adjusted results for meta-analysis. This study used the odds ratio to present the overall effect estimates in forest plots and subgroup analysis. The results of the meta-analysis are shown as forest plots with heterogeneity tested by I 2 analysis. An I 2 > 50% indicated the existence of substantial heterogeneity, and subgroup analysis was further performed. All meta-analyses were conducted by Review Manager version 5.3 software (Cochrane Collaboration, Oxford, UK) provided by Cochrane. In addition, a funnel plot was generated with Stata Statistical Software version 15.0 (StataCorp LLC, College Station, TX) to evaluate the possibility of publication bias.

Statins and severe outcomes of COVID-19
Four articles, two presenting unadjusted outcomes [21,32] and another two presenting adjusted outcomes [23,39], reported the association between statins and ICU care. The meta-analysis showed that the use of statins was not significantly associated with the need for ICU care (OR ¼ 0.91, 95% confidence interval [CI]: 0.55-1.51, p¼.73, I 2 ¼ 66%) (Figure 1). Four studies were included to evaluate how statin use affects the need for IMV in COVID-19-confirmed individuals. Three of these four studies presented adjusted outcomes regarding IMV use [12,14,39]. The use of statins was significantly associated with a decreased risk of the need for IMV support (OR ¼ 0.81, 95% CI: 0.69-0.95, p¼.010, I 2 ¼0%) ( Figure 2).

Subgroup analysis of statins and COVID-19 mortality
Since substantial heterogeneity existed in the metaanalysis of statins and COVID-19 mortality (I 2 ¼72%),   we performed subgroup analysis to explore the source of heterogeneity. We identified five subgroups of studies in which statin users had even lower odds of death, including studies that (1)  The heterogeneity in these five subgroups of studies was lower than that before the subgroup analyses. On the other hand, one single study that included only individuals with DM revealed that statin use was associated with increased COVID-19 mortality (OR ¼ 1.55, 95% CI: 1.07-2.25, p¼.02). Table 2 describes the detailed results of the subgroup analyses.

Discussion
Our study demonstrated that the use of statins was associated with a reduced mortality risk of COVID-19. When excluding one study enrolling only individuals with DM [14], the association between statin use and COVID-19 mortality was significant. In addition, the subgroup analysis showed that statin use prevented more deaths in studies with a CFR below 20% than in those with a CFR above 20%. This result implies that the benefit of statins seems to be greater when the COVID-19 mortality rate in the community is reduced.
Our meta-analyses also revealed that statin use was linked to a reduced need for IMV but could not prevent individuals from needing ICU admission. Our finding that statins were associated with reduced COVID-19 mortality confirmed the results from previous reports [15,45,46]. Compared with nonstatin users, statin users are usually older and have more comorbidities, which both lead to an increased risk of COVID-19 mortality [11,12]. Thus, when the meta-analysis pooled the adjusted and unadjusted results, the association between statin use and mortality might not be obvious because the protective effect of statins might be countered by increased risks from host factors. In our study, statins showed benefits even when the meta-analysis pooled both the adjusted and unadjusted results, making the results more robust and convincing.
One large meta-analysis performed a comprehensive systematic review and concluded statins were not associated with reduced mortality of COVID-19. However, the I 2 of the forest plot demonstrating the association of statin with mortality was as high as 90%, indicating the substantial heterogeneity across the included articles [47]. Therefore, one essential goal of our study was to find out the subgroups that had lower heterogeneity. After the subgroup analysis, this study identified five subgroups in which statins were associated with an even lower COVID-19 mortality. The five subgroups also had a lower heterogeneity than overall participants. Our results may be helpful for participant selection in future RCTs.  [11,12,20,23,29,33,34,36,38,39,41,43,44]  The measure of mortality outcome is provided for each subgroup, and the interaction p value and I 2 are provided for subgroup differences. All analyses were conducted using inverse variance and a random-effects model.
One of our subgroup analyses showed that statins had survival benefits in studies with an overall CFR of less than 20% but not in those with a CFR above 20%. There are three possible explanations. First, the beneficial effect of statins was mild. When an individual bears a high risk of COVID-19 mortality, the statins' impact might be overwhelmed by the host factor and turn out to be statistically nonsignificant. Second, the effect of statins was influenced by competing medical issues. A high overall mortality rate often reflects a more severe epidemic in the community because only individuals with serious conditions can be admitted for treatment. Under such circumstances, medical resources may be insufficient. The health care system may sort medical help to those who had a better prognosis than to those who had the most severe illness. Statin users, often with more comorbidities, may have poorer outcomes than those with sufficient medical resources. Third, all confirmed COVID-19 individuals' average mortality rate in the USA was 1.82% by 6 March 2021 [48]. The higher the overall mortality rate is in a study, the lower the generalizability of the result is. As more people acquire immunity against SARS-CoV-2, either by vaccination or infection, and the health care system gains more experience in treating COVID-19, we can expect a lower overall mortality rate in the future. We believe that the analysis results focussing on low CFR studies may better apply to future conditions. According to the study results, statins were associated with a reduced need for mechanical ventilators but did not influence the possibility of ICU admission. Both of the above results had low heterogeneity. Two possible reasons may explain that. First, the criteria for the use of mechanical ventilators or ICU care might vary among different areas, hospitals and even physicians. Insufficient medical resources could also play a role in some cohorts. Second, statins provide lung protection and thus decrease the need for IMV. However, statins could not reduce other nonpulmonary events, such as cardiovascular complications or sepsis. The mechanism of lung protection is most likely via upregulation of ACE2 expression, leading to increased angiotensin (1-7) production.
Although one of our subgroup analyses showed that in-hospital use of statins had a greater benefit, we cannot interpret that to mean that starting statins on admission to treat COVID-19 was helpful. According to clinical experiences, most in-hospital statin users were already taking statins before contracting COVID-19. Besides, the majority of articles included in our study emphasized the use of statin was prior to COVID-19 infection, and the results showed statin had a trend of protective effect from getting COVID-19. Retrospective observational studies had an intrinsic limitation in recording clear-cut timing of statin use among a large group of participants. However, our results at least confirmed that statin users can continue taking the medication during COVID-19 infection.
There are several limitations to the study. First, all articles included in our meta-analyses were observational study designs. The baseline demographics of statin users vs. nonusers exhibited substantial differences. Furthermore, some studies reported only crude odds ratios, which might lead to an even more significant bias. Second, the type, dose and duration of statins were by no means identical across studies. The majority of the studies did not detail the prescriptions of statins. Thus, even though the current evidence showed that statins might be beneficial for patients with COVID-19, it was almost impossible to provide a clear suggestion about the drug name, dose and duration. Third, a substantial portion of participants in this study had comorbidities of diabetes or cardiovascular diseases, and might have taken other concomitant medication, such as metformin, ACE inhibitors and angiotensin II receptor blockers (ARB). Metformin has shown to be associated with reduced COVID-19 mortality [49], and ACEI/ARB also had close relationship with COVID-19 [50]. The influence of interaction between statins and co-medications on COVID-19 remains unclear and might impact the estimate of statins' effect. Future RCTs are needed to solve the limitations mentioned above.

Conclusions
Our results demonstrated that the use of statins was associated with a reduced need for IMV and mortality rate among individuals with COVID-19. Subgroup analyses further identified individuals enrolled in studies with a lower mortality rate who benefitted more from statins. This study also confirmed that statins could be safely used during COVID-19 hospitalization.