Remdesivir and survival outcomes in critically ill patients with COVID-19: A multicentre observational cohort study

Dear editor

The role of remdesivir (RDV) in the treatment of critically ill COVID-19 patients remains ill-defined.¹ The impact of the systemic inflammation and other aspects adding to the potential severity of patients in the SARS-CoV-2 viral dynamics are not well elucidated, particularly in critically ill patients.2, 3, 4 Thus, we aimed to evaluate the effect of RDV on the outcomes of critically ill patients with severe COVID-19 and analyze such outcomes in certain subsets of patients to investigate whether RDV proved particularly beneficial in a particular group of critically ill patients.

Multicentre, observational cohort study including consecutive COVID-19 patients admitted to 55 Spanish ICUs between 5 February 2020 and 21 December 2021. We compared two main groups: patients receiving RDV versus those who did not receive RDV at any moment. The primary outcome was all-cause 90-day mortality. We examined outcomes according to the following categories: (1) overall population; (2) several patient subgroups based on baseline data at ICU admission: age, illness severity and organ damage (Acute Physiology and Chronic Health Evaluation [APACHE] -II and Sequential Organ Failure Assessment [SOFA] score, respectively), laboratory findings (lymphocyte count and C-reactive protein [C-RP]), respiratory support; (3) by corticosteroids therapy; (4) by SARS-CoV-2 viral load (viral RNAemia, viral antigenemia and viral RNA load [N1 region]); and (5) by RDV administration timing. To evaluate the effect of remdesivir on 90-day mortality, we used a Cox regression model stratified on the center variable, tested in univariable and propensity-adjusted multivariable analyses. A further analysis using the propensity score matching method was performed to confirm the results obtained with the propensity-adjusted multivariable model. On the other hand, effect modification by factors potentially associate with patient outcomes and remdesivir use were assessed by an interaction term. We also analysed the association between remdesivir therapy and the following variables: in-hospital and 30-day mortality (by means of a Fine-Gray competing risks model stratified on the center variable and a Cox regression model stratified on the center variable, respectively); and nosocomial bacterial pneumonia (by means of generalised estimating equations, considering a binomial distribution and accounting for the effect raised by the clustering of patients from the same center).

Among the 6225 COVID-19 patients that were admitted to 55 ICUs. We included 5004 patients in this analysis, of whom 4209 (84%) did not receive RDV and 795 (16%) received RDV during hospitalization (Fig. 1 ).

Fig. 1.

Flowchart of the study population.

Ninety-day mortality rate was lower in patients who received RDV than in those who did not receive RDV (34 % vs. 29%, p = 0.012). Also, in-hospital (31 % vs. 27%, p = 0.025) and 30-day mortality (25 % vs. 18%, p<0.001) were lower in the RDV group (Supplementary Table 1). In the propensity-adjusted multivariable analysis, RDV use was not significantly associated with 90-day mortality (HR 1.00, 95% CI 0.85 to 1.17; p = 0.970), nor with in-hospital (sHR 0.94, 95% CI 0.81 to 1.09, p = 0.41) and 30-day mortality (HR 0.86, 95% CI 0.71 to 1.05, p = 0.145) (Table 1 ). In the analysis using the propensity score matching (777 patients received RDV and 777 did not receive RDV), the Kaplan-Meier curves showed that there was no statistical difference between both groups for 90-day mortality (p = 0.280) (Supplementary Figure 1). Furthermore, Cox regression showed that RDV use was not associated with the risk of 90-day mortality (HR 0.95, 95% CI 0.79 to 1.16; p = 0.636), nor with in-hospital mortality (sHR 0.92, 95% CI 0.76 to 1.11; p = 0.36) and 30-day mortality (HR 0.81, 95% CI 0.65 to 1.02; p = 0.075).

Table 1.

Association of remdesivir therapy and 90-day mortality (Panel A) and early administration of remdesivir treatment from first symptoms and 90-day mortality (Panel B).

	Univariable analysis		Adjusted analysisa
	HR (95% CI)	P-value	Adjusted HR (95% CI)	P-value
Panel A - Remdesivir therapy
All patients (N = 5004)	0.78 (0.67 to 0.91)	0.001	1.00 (0.85 to 1.17)	0.970
Subgroup analysesb
Age group				0.284c
Age <40 years (n = 281)	0.13 (0.02 to 1.05)	0.055	-	-d
Age ≥40 - <65 years (n = 2419)	0.75 (0.58 to 0.98)	0.033	1.02 (0.77 to 1.36)	0.885
Age ≥65 years (n = 2304)	0.89 (0.74 to 1.08)	0.238	0.96 (0.78 to 1.17)	0.681
Severity of illness at ICU admission group				0.368c
APACHE-II score <12 (n = 1411)	0.57 (0.37 to 0.89)	0.012	0.97 (0.60 to 1.57)	0.893
APACHE-II score ≥12 (n = 1476)	0.86 (0.65 to 1.12)	0.260	0.96 (0.72 to 1.29)	0.793
Organ dysfunction and failure at ICU admission group				0.465c
SOFA score <5 (n = 1667)	0.76 (0.55 to 1.05)	0.095	0.88 (0.62 to 1.26)	0.494
SOFA score ≥5 (n = 1803)	0.89 (0.70 to 1.13)	0.333	1.01 (0.79 to 1.31)	0.922
Laboratory findings at ICU admission
Lymphocyte count group				0.694c
Lymphocyte count <0.724 × 109/L (n = 2613)	0.84 (0.69 to 1.03)	0.087	0.97 (0.78 to 1.20)	0.768
Lymphocyte count ≥0.724 × 109/L (n = 2221)	0.73 (0.57 to 0.93)	0.010	1.04 (0.80 to 1.36)	0.753
C-reactive protein group				0.333c
C-reactive protein <150 mg/L (n = 2627)	0.84 (0.68 to 1.03)	0.099	1.00 (0.80 to 1.26)	0.974
C-reactive protein ≥150 mg/L (n = 2048)	0.71 (0.56 to 0.91)	0.006	0.93 (0.72 to 1.21)	0.614
Respiratory support at ICU admission group				0.133c
Conventional oxygen therapy (n = 371)	0.85 (0.45 to 1.63)	0.633	0.69 (0.28 to 1.74)	0.436
High-flow nasal cannula / Non-invasive mechanical ventilation (n = 2046)	0.99 (0.78 to 1.24)	0.902	1.25 (0.97 to 1.62)	0.088
Invasive mechanical ventilation (n = 2571)	0.77 (0.62 to 0.96)	0.020	0.90 (0.72 to 1.14)	0.397
Corticosteroid therapy during ICU admission group				0.225c
No (n = 684)	0.71 (0.38 to 1.33)	0.285	0.73 (0.36 to 1.46)	0.370
Yes (n = 4271)	0.81 (0.69 to 0.95)	0.008	1.05 (0.89 to 1.24)	0.561
SARS-CoV-2 viral load
Viral RNAemia in plasma group				0.367c
No (n = 167)	0.70 (0.20 to 2.51)	0.588	3.32 (0.14 to 80.96)	0.462
Yes (n = 584)	0.88 (0.60 to 1.29)	0.507	0.83 (0.55 to 1.26)	0.384
Viral antigenemia in plasma group				0.238c
No (n = 400)	1.08 (0.62 to 1.87)	0.791	1.51 (0.81 to 2.84)	0.195
Yes (n = 335)	0.77 (0.47 to 1.27)	0.310	0.74 (0.41 to 1.34)	0.320
Viral RNA load in plasma (N1 region) groupe				0.753c
0 copies/mL (n = 130)	1.33 (0.27 to 6.53)	0.723	0.97 (0.04 to 26.90)	0.987
>0 - <3255 copies/mL (n = 376)	0.89 (0.50 to 1.59)	0.692	1.08 (0.55 to 2.13)	0.824
≥3255 copies/mL (n = 245)	0.90 (0.53 to 1.54)	0.702	1.25 (0.68 to 2.28)	0.469
Panel B - Early administration of remdesivir treatment from first symptoms
All patients receiving remdesivir (N = 733)f
<5 days	1.65 (1.13 to 2.40)	0.009	1.53 (1.02 to 2.31)	0.042
<7 days	1.19 (0.88 to 1.60)	0.253	1.21 (0.85 to 1.72)	0.285

Abbreviations: HR indicates hazard ratio; CI, confidence interval; APACHE, acute physiology and chronic health evaluation; SOFA, sequential organ failure assessment; RNA, ribonucleic acid.

^aAdjusted for variables (age, sex, body mass index, diabetes mellitus, chronic liver disease, chronic heart disease, chronic lung disease, chronic renal failure, immunosuppression, APACHE-II score at ICU admission, lymphocyte count at ICU admission, C-reactive protein at ICU admission, respiratory support at ICU admission, septic shock at ICU admission, lopinavir/ritonavir administration, tocilizumab administration, other antiviral administration, corticosteroids, COVID-19 wave and the propensity score).

^bAPACHE-II score was assessed in 2887 patients; SOFA score in 3470 patients; lymphocyte count in 4834 patients; C-reactive protein in 4675 patients; respiratory support in 4988 patients; corticosteroids in 4956 patients; viral RNAemia in plasma in 751 patients; viral antigenemia in plasma in 735 patients; and viral RNA load in plasma (N1 region) in 751 patients.

^cInteraction effect for the subgroup and treatment group.

^dEstimation failed due to numerical problem. Because the coefficients did not converge, no further models were fitted.

^eCut-off value obtained from ROC curve for 90-day mortality.

^fThe time of administration of remdesivir treatment from first symptoms was not available for 62 patients.

To examine mortality risk for particular types of patients, we explored modification effects by age, APACHE-II score, SOFA score, lymphocyte count, C-RP, respiratory support, corticosteroids, viral RNAemia, viral antigenemia and viral RNA load in plasma. No significant effect modification was found after adjustment for covariates using propensity score (Table 1).

Among the overall population receiving RDV, there was significant association observed between early administration (<5 days since symptoms’ onset) and the propensity-adjusted risk of 90-day mortality (HR 1.53, 95% CI 1.02 to 2.31, p = 0.042) (Table 1). In contrast, there were no significant association between <7 days since symptoms’ onset and the risk of 90-day mortality (HR 1.21, 95% CI 0.85 to 1.72, p = 0.285).

When compared to the non-RDV group, patients in the RDV group less frequently presented myocarditis, cardiac ischemia, delirium, coagulation disorder, anemia, acute renal failure and liver dysfunction (Supplementary Table 2). The propensity-adjusted analysis showed no association between RDV treatment and nosocomial bacterial pneumonia (OR 1.06, 95 CI 0.84 to 1.33, p = 0.640). In the analysis using the propensity score matching, RDV use was not associated with nosocomial bacterial pneumonia (OR 1.22, 95 CI 0.98 to 1.52, p = 0.068) (Supplementary Table 3).

In this large, multicentre study involving over five thousand critically ill patients with COVID-19 admitted to 55 Spanish ICUs, we observed that the use of RDV was not associated with an overall reduced risk of 90-day mortality, nor when analyzing by subgroup populations; patients treated with RDV overall presented longer lengths of ICU stay, which seems to be due to longer length of mechanical ventilation; early administration of RDV from symptoms’ onset was associated with a higher risk of 90-day mortality in the overall population; and patients treated with RDV showed lower significant rates of organic damage associated with severe COVID-19 such as cardiac, neurological, coagulation, renal and liver complications.

The underlying pathophysiological mechanisms associated with severe COVID-19 are not fully elucidated,^{5 , 6} nor are therefore the tools to identify early phenotypes with high risk of developing more severe cases of COVID-19 that might potentially benefit more of early and intensive antiviral treatment.⁷ This has become a major challenge for the scientific community, as patients developing severe COVID-19, and in particular those requiring ICU admission have poor prognosis.⁸ Our hypothesis that certain subsets of patients with either pro-inflammatory phenotypes or presenting with high viral loads⁹ might particularly benefit from RDV treatment seems to be refuted by our findings. The lack of influence of RDV on mortality independently of the viral load in plasma and the inflammatory status could reflect an inability of RDV to mediate a significant inhibitory activity of viral replication and/or clinical benefit in patients already exhibiting a large burden viral replication (both groups, RDV treated and non-treated presented to the ICU with high median levels of viral load in plasma), which in turn is associated to a strong stimulation of the innate immunity leading to exacerbated inflammation. Whether new, more specific/potent antivirals could mediate a beneficial effect in this context remains to be elucidated. Our results are in accordance of the recent published update of a living review about remdesivir in adults hospitalized with COVID-19 that confirm that remdesivir probably results in little to no difference in mortality.¹⁰

A possible limitation of the propensity score methods is their inability to control for unmeasured confounding. Another limitation is the different waves of the pandemic, which could have influenced our results. We have however adjusted all of our analyses for this confounder.

In summary, treatment with RDV was not associated with improved outcomes in critically ill patients with severe COVID-19, neither overall nor when stratifying by clinically relevant variables such as age, illness severity, organ damage, laboratory findings, respiratory support or SARS-CoV-2 viral load in plasma. Moreover, RDV treatment was associated with longer lengths of ICU admission. Early administration since symptoms onset may prove harmful. Our study adds to the mounting evidence suggesting that RDV is not efficacious in treating severe COVID-19, although further studies are warranted to elucidate whether certain subsets of patients might benefit from it.

Authors' contributions

Conceptualization: CC, AM, AT; Data Curation: CC, AM, TC, Formal analysis: AG; Methodology: all authors; Investigation: CC, AM, TC, FB, AT; Project administration: CC, AM, AT; critical revision of the manuscript for important intellectual content: CC, AM, FB, and AT; and Funding acquisition: AT; Resources: CC, AM, AT; Sofware: AG; Supervision: AT; Validation:all authors; Visualization: all the authors, Writing-original draft: all authors; Writing-review and editing: CC, AM, FB, TC, AT. AThad full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. All authors read and approved the final manuscript. CiberesUCICOVID consortium participated in data collection.

Funding

This study was supported by the Instituto de Salud Carlos III de Madrid (COV20/00110, ISCIII); Fondo Europeo de Desarrollo Regional (FEDER); "Una manera de hacer Europa"; and Centro de Investigación Biomédica En Red – Enfermedades Respiratorias (CIBERES). DdGC has received financial support from the Instituto de Salud Carlos III (Miguel Servet 2020: CP20/00041), co-funded by European Social Fund (ESF)/ “Investing in your future”. CC received a grant from the Fondo de Investigación Sanitaria (PI19/00207), Instituto de Salud Carlos III, co-funded by the European Union.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.