|Year : 2022 | Volume
| Issue : 1 | Page : 11-21
Propensity score-matched case–control study of convalescent plasma in treatment of patients with moderate and severe COVID-19
Kislay Kishore1, Sandeep Rana1, Vasu Vardhan2, Nishant Raman3, Sandeep Thareja4, K.V. Padmaprakash5, J. Muthukrishnan3, K. S Rajmohan6, Monika Agarwal6, K.K. Ashta3, Anirudh Anilkumar7
1 Department of Respiratory Medicine, Base Hospital, Delhi Cantt, India
2 Department of Respiratory Medicine, Army College of Medical Sciences, Delhi Cantt, India
3 Department of Medicine, Base Hospital, Delhi Cantt, India
4 Command Hospital, Central Command, Lucknow, Uttar Pradesh, India
5 Department of Gastroenterology, Base Hospital, Delhi Cantt, India
6 Department of Pathology, Base Hospital, Delhi Cantt, India
7 House Surgeon, Sir Ganga Ram Hospital, New Delhi, India
|Date of Submission||24-Aug-2021|
|Date of Decision||21-Dec-2021|
|Date of Acceptance||06-Dec-2021|
|Date of Web Publication||19-Apr-2022|
Dr. Nishant Raman
Department of Medicine, Base Hospital, Delhi Cantt, Pin Code: 110010
Source of Support: None, Conflict of Interest: None
Background: Convalescent plasma (CP) in coronavirus disease 2019 (COVID-19) acts as a source of neutralizing antibodies that could provide passive immunity. There still exists uncertainty on the safety and efficacy of CP as a therapeutic option in COVID-19. This study reports on the effectiveness of CP therapy (CPT) in patients with moderate and severe COVID-19 in an Indian cohort. Methodology: This propensity score matched (PSM) case–control study included a total of 477 adult patients with COVID-19 (≥18 years of age) with moderate and severe disease out of which 181 patients received CP having neutralizing antibody titers ≥1:40 (PRNT/MNT). PSM with an optimal algorithm was performed, matching cases (CP recipients) on baseline patient characteristics to select (178) controls in a 1:1 ratio. Analysis included logistic regression for the whole-patient sample, conditional logistic regression for PSM sample, and competing risk approach for time-to-event analysis. Primary outcome was in-hospital all-cause mortality. Results: Greater odds of clinical improvement were observed among CP recipients on day 7, but no significant differences were observed in the clinical status on day 14. Mortality was lower in patients treated with CP; however, the difference was not statistically significant. Logistic-regression analysis for whole-patient sample and conditional logistic regression for PSM sample showed no significant mortality benefit of CP both in the unadjusted and covariate-adjusted models. No significant mortality benefit was observed among CP recipients in the survival analysis. Conclusion: CPT in COVID-19 appears to have some benefit as evinced by greater odds of clinical recovery on day 7, however, offers no survival benefit.
Keywords: Antibody, competing risk, convalescent plasma, COVID-19, propensity score matching, SARS-CoV-2
|How to cite this article:|
Kishore K, Rana S, Vardhan V, Raman N, Thareja S, Padmaprakash K, Muthukrishnan J, Rajmohan KS, Agarwal M, Ashta K, Anilkumar A. Propensity score-matched case–control study of convalescent plasma in treatment of patients with moderate and severe COVID-19. J Assoc Chest Physicians 2022;10:11-21
|How to cite this URL:|
Kishore K, Rana S, Vardhan V, Raman N, Thareja S, Padmaprakash K, Muthukrishnan J, Rajmohan KS, Agarwal M, Ashta K, Anilkumar A. Propensity score-matched case–control study of convalescent plasma in treatment of patients with moderate and severe COVID-19. J Assoc Chest Physicians [serial online] 2022 [cited 2022 Sep 25];10:11-21. Available from: https://www.jacpjournal.org/text.asp?2022/10/1/11/339690
| Introduction|| |
Coronavirus disease 2019 (COVID-19), a highly contagious and potentially fatal infection caused by coronavirus-2 (SARS-CoV-2), a positive-stranded RNA virus, presents with a spectrum of disease severity. With no proven interventions and limited treatment options available, the disease poses a unique challenge.
Convalescent plasma (CP) obtained from individuals who have recovered from infection has been used with varying degrees of success to treat Influenza, Ebola virus disease, and severe acute respiratory syndrome (SARS) coronavirus epidemics. CP has been proposed as a therapeutic intervention in COVID-19. It acts as a source of neutralizing antibodies that could provide passive immunity. In addition, immune mediators in plasma like anti-inflammatory cytokines, defensins, pentraxins along with other immune pathways, such as antibody-dependent cellular cytotoxicity, complement activation, and phagocytosis might contribute to alleviating the systemic hyperinflammatory response.,
There, however, still exists uncertainty on safety and efficacy of CP as a therapeutic option in COVID-19., This study reports on the effectiveness of CP therapy (CPT) in patients with moderate and severe COVID-19 in an Indian cohort.
| Methodology|| |
Type of study: Propensity score-matched (PSM) case–control study.
Setting: Tertiary COVID-19 care center in northern India.
Between June 21, 2020 and October 19, 2020, a total of 3134 confirmed patients with COVID-19 were hospitalized. The patients included in the present study were (i) ≥18 years, (ii) diagnosed by positive reverse-transcriptase polymerase chain reaction (RT-PCR) for SARS-CoV-2 or a positive rapid antigen test, and (iii) having moderate or severe disease at admission or with radiologic evidence of pneumonia. Severe illness is defined as oxygen saturation by pulse oximetry (SpO2) <94% on room air, respiratory rate >30 breaths/minute, PaO2/FiO2 <300 mmHg, or lung infiltrates >50%. Moderate illness is defined as evidence of lower respiratory disease during clinical assessment or imaging, with SpO2 ≥94% on room air. The patients excluded from CPT and hence also excluded from the study included (i) pregnant or lactating women, (ii) patients with hypersensitivity to blood products, (iii) recipients of immunoglobulin during past 30 days, (iv) patients with positive Coombs test, (v) patients with conditions precluding infusion of blood products, and (vi) critically ill patients with PaO2/FiO2 <200 mmHg or shock (requiring vasopressors to maintain a mean arterial pressure ≥65 mmHg). A total of 477 patients were included in the analysis, out of which 181 patients were CP recipients.
Convalescent plasma transfusion
Eligible donors were (i) men or nulliparous women aged 18 to 60 years with body weight ≥50 kg and (ii) who were COVID-19 RT-PCR positive or rapid antigen test positive. Plasma donation was performed at minimum of 14 days after resolution of symptoms. Eligible donors were screened for ABO typing and blood-borne infections including human immunodeficiency virus (HIV), hepatitis B and C viruses, and venereal disease research laboratory. Antibody titers in eligible donors were estimated. The required concentration of immunoglobulin G (IgG) antibody against COVID-19 was neutralizing antibody titers ≥1:40 (PRNT/MNT). CP recipients were transfused with one or two units of ABO type compatible plasma. Out of the total of 181 CP recipients, 104 received two units of plasma and 77 received single dose of plasma. Patients who received two units were either from single or multiple donors. Each unit, measuring approximately 200 mL, was infused over 1 to 2 hours. CP recipients were monitored every 15 minutes for signs of transfusion-related reactions and then followed for outcomes after the transfusion.
PSM of controls to convalescent plasma recipients
A PSM analysis was performed with XLSTAT version 2021 for Windows (Addinsoft, New York, NY, USA) for data tabulated in Excel for Microsoft Office 365 (Microsoft Corporation, Redmond, WA, SA). Logistic-regression model was fit to predict the potential for CPT based on (i) baseline patient-characteristics inclusive of age and sex; (ii) pre-existing comorbidities including diabetes, hypertension, and chronic respiratory conditions; (iii) other comorbidities categorized system wise comprising of (a) cardiovascular conditions including coronary artery disease, congestive heart failure, heart blocks, and valvular heart diseases, (b) neurologic disorders including cerebrovascular diseases, seizure disorder, space-occupying lesions, central nervous system infections, and (c) renal diseases including acute kidney injury at presentation, chronic kidney disease, and renal transplant recipients; and (iv) vital parameters at admission including pulse rate and systolic blood pressure; and respiratory rate and oxygen saturation as a measure of COVID-19 severity assessment. Supplementary file 1 presents the results of the logistic model used to determine the propensity scores.
Cases are defined as eligible patients who received CP. Eligible patients who did not receive CP are termed controls. Whole-patient sample is defined as all cases and controls who were not PSM.
A 1:1 matching was performed without replacement using optimum algorithm, tolerance ≤0.005, and estimated with 95% confidence interval (CI) which yielded 181 cases (CP recipients) and 181 PSM controls. The C-statistic for constructed model differentiated well between the two groups [area under the curve (AUC) 0.710, 95% CI 0.657–0.763] [Figure 1]. The distribution of logit-propensity scores of cases and controls in the matched sample as against those in the whole-patient sample is represented by [Figure 2].
|Figure 1 Receiver operating characteristic (ROC) curve of propensity scores to determine power with which the model predicts treatment allocation. The C-statistic for constructed model differentiated well between the two groups (n = 181 cases, n = 181 controls) (AUC 0.710, 95% CI 0.657–0.763). AUC, area under the curve; CI, confidence interval.|
Click here to view
|Figure 2 Distribution of logit propensity scores. The logit propensity scores were more similarly distributed in the cases (n = 181) and their propensity score-matched controls (n = 181) than in the whole-patient sample (n = 477). V4, Treatment group (1=convalescent plasma recipient; 0=non recipient).|
Click here to view
After PSM, other data for matched cases and controls were retrospectively collected. PSM was blinded to outcomes. Matched case–control pairs (n = 3) in which event occurred before start of follow-up period (T0) were excluded from analysis.
Overall, balance was well achieved between the two groups, that is, cases (n = 178) and controls (n = 178). The standardized difference for all matched and unmatched variables was approximately <10% except for endocrine disorders, malignancy, HIV, history of dyspnea, duration of symptoms before admission and T0, diastolic blood pressure, respiratory rate, low Glasgow coma scale at admission [Table 1]; and concomitant use of investigational antivirals (favipiravir or remdesivir), use of interleukin-6 (IL-6) receptor-antagonist (tocilizumab) and patients on invasive mechanical ventilation (IMV) [Table 4]. Standardized difference was <25% for all variables except investigational antivirals [Table 4].
|Table 1 Baseline patient characteristics (matched and unmatched covariates)|
Click here to view
Categorical including dichotomous variables were expressed as frequencies and percentages. Chi-squared test of significance was performed to compare cases and controls from the whole-patient sample. A two-sided P < 0.05 was considered statistically significant. Continuous variables were expressed as mean values and their standard deviations or medians and their interquartile ranges. Independent sample T test and Mann–Whitney U test were used for comparison of these variables among cases and controls in the whole-patient sample. All P-values are two-tailed, equal variances not assumed for all continuous variables. For PSM cases and controls, categorical and continuous variables were compared by their respective standardized differences. McNemar test was used for comparison. Asymptotic P < 0.05 was considered significant.
Analysis of outcomes
Primary outcome was in-hospital all-cause mortality. Analysis of outcome was performed separately for whole-patient sample (n = 477) and PSM sample (n = 178 cases, 178 controls). Time-independent analysis of outcome was performed by logistic regression for whole-patient sample and conditional logistic regression for PSM sample. Results of both unadjusted and covariate-adjusted models are presented as odds ratio (OR) and 95% CIs. A P-value of <0.05 was considered significant.
Competing-risk approach for survival analysis was preferred over standard survival analysis to evaluate the effect of CP on time from T0 to death or recovery. T0 was defined as the day of CPT for cases and the corresponding day of admission for controls. Recovery was considered as a competing risk for mortality.,,, In comparison, in a standard survival analysis, recovered patients are censored which causes bias by overestimating incidence of death. Cumulative incidence probabilities were estimated by fitting proportional subdistributional hazard model by Fine and Gray approach with Gray test being used to determine P-values., Clustered robust variance estimates were performed to account for variance within PSM case–control pairs. The subdistribution hazard ratios are reported with 95% CI (with 5% type 1 error). Proportional hazards assumption was tested by evaluation of the Schoenfeld residuals [Supplementary file 1]. Univariable models were constructed and covariates with significant associations in the univariable analysis along with covariates with high degrees of standardized differences between cases and controls of the PSM sample were included in the final multivariable model. Statistical analysis was performed using Stata Version 15.0 for Windows (Stata Statistical Software: Release 15; StataCorp LLC, College Station, TX, USA) and SPSS version 23 for Windows (IBM SPSS Statistics for Windows, Version 23.0;IBM Corp, Armonk, NY, USA). GraphPad Prisma version 9.1.1 for Windows (GraphPad Software, La Jolla, CA, USA) was used to plot graphs.
Clinical improvement was assessed by a composite endpoint comprising of discharge or improvement in the COVID-19 outcome severity scale (COSS) which consists of seven mutually exclusive categories: 1, death; 2, hospitalized, receiving extracorporeal membrane oxygenation, or IMV; 3, hospitalized, receiving noninvasive ventilation, or nasal high-flow oxygen therapy; 4, hospitalized, receiving supplemental oxygen without positive pressure, or high flow; 5, hospitalized, not receiving supplemental oxygen; 6, not hospitalized and unable to perform normal activities; and 7, not hospitalized and able to perform normal activities. Other outcomes included adverse transfusion reaction and cause of death.
| Results|| |
Primary outcome was in-hospital all-cause mortality. Overall, 100 (21%) of the 477 patients included in the study succumbed to illness. Mortality rate in the whole-patient sample among cases (34, 18.8%) and controls (66, 22.3%) was not statistically different (Pearson Chi-squared test, P = 0.417) [Table 2]. In the PSM cases and controls, fatal outcome was observed in 12 matched pairs, whereas 112 matched pairs survived. Mortality in matched cases and controls was not significant (McNemar test, P = 0.178) [Table 3].
Logistic-regression analysis of the whole-patient sample showed no significant benefit of CP (OR 1.241, 95% CI 0.781–1.970, P = 0.361). In the covariate-adjusted model, no statistical significance was observed (adjusted OR 1.324, 95% CI 0.617–2.841, P = 0.471) on the survival benefit of CP [Supplementary file 1].
Conditional logistic-regression analysis of the PSM cases and controls also showed no significant benefit of CP on the primary outcome (Conditional Odds Ratio, COR 1.499, 95% CI 0.874–2.572, P = 0.141) [Supplementary file 1].
The use of other pharmaceutical interventions with potential benefit such as that of investigational antivirals (remdesivir and favipiravir) was more among CP recipients (174, 96.1%) compared to controls (38, 12.8%) in whole-patient sample (Pearson Chi-squared test, P < 0.001) as in the PSM sample (cases versus controls: 172, 96.6% versus 36, 20.2%; McNemar test, P < 0.001) [Table 4]. A total of 10 (2.1%) patients received IL-6 inhibitors (tocilizumab). Among cases (8, 4.4%) and controls (2, 0.7%) in the whole-patient sample, significantly greater proportion of CP recipients received tocilizumab (Pearson Chi-squared test, P = 0.008). Similarly, among cases (8, 4.5%) and controls (2, 1.1%) in the PSM sample, use of tocilizumab was more common in CP recipients, though the difference did not attain statistical significance (McNemar test, P = 0.109) [Table 4].
|Table 4 Pharmacologic interventions administered to cases and controls and supplemental oxygen requirement for both whole-patient sample and propensity score-matched sample|
Click here to view
Logistic-regression models were adjusted for the use of investigational antiviral drugs and tocilizumab. In the whole-patient sample, logistic-regression showed no significant benefit of CP when adjusted for concomitant use of antiviral drugs (OR 1.328, 95% CI 0.491–3.589, P = 0.576) and use of tocilizumab (OR 1.308, 95% CI 0.816–2.094, P = 0.305) [Supplementary file 1]. In the PSM sample, results of the conditional logistic regression showed no significant survival benefit of CP when adjusted for antiviral drug use (adjusted COR 0.554, 95% CI 0.143 2.141, P = 0.165) as well as upon adjusting for tocilizumab (adjusted COR 1.628, 95% CI 0.926–2.862, P = 0.090) [Supplementary file 1].
There was no difference between CP recipients (n = 178) and their PSM controls (n = 178) with respect to the primary outcome [hazard ratio (HR) 0.687, 95% CI 0.451–1.049, P = 0.082]. No significant benefit of CP when adjusted for concomitant use of antiviral drugs (adjusted HR 0.658, 95% CI 0.348–1.242, P = 0.197); however, when adjusted for concomitant use of IL-6 antagonist (tocilizumab), CPT was associated with a trend toward significant reduction of risk in mortality (adjusted HR 0.663, 95% CI 0.435–1.009, P = 0.056) [Figure 3].
|Figure 3 Competing-risk model (P-value calculated by Gray test). CI, confidence interval; SHR, subdistribution hazard ratio.|
Click here to view
In the PSM sample, results of the conditional logistic regression showed no significant survival benefit of CP when adjusted for antiviral drug use (adjusted COR 2.424 95% CI 0.694–8.457, P = 0.165). Analysis of the PSM cases and controls when adjusted for mechanical ventilation showed no significant benefit of CPT on the primary outcome (adjusted HR 0.782, 95% CI 0.514–1.189, P = 0.251) [Figure 3].
In the final multivariable analysis model, when adjusted covariates having significant association with mortality in respective univariable analyses [Supplementary file 1]. CPT showed no significance with respect to reduction in risk of mortality (adjusted HR 0.801, 95% CI 0.427–1.498, P = 0.487) [Figure 3].
Cumulative-incidence curves of in-hospital all-cause mortality cases (n = 178) and controls (n = 178) accounting for competing risk of death are presented in [Figure 4].
|Figure 4 Cumulative incidence curves of in-hospital all-cause mortality cases (n = 178) and controls (n = 178) accounting for competing-risk of death with P-value calculated by Gray test).|
Click here to view
The proportion of patients attaining the composite endpoint of clinical improvement (determined by discharge or improvement in COSS) on day 7 was observed to be higher among CP recipients (117, 65.7%) compared to controls (91, 51.1%), a difference which reached statistical significance (McNemar test, P = 0.008). By day 14, 144 (80.9%) cases attained clinical improvement compared to 125 (70.2%) controls (McNemar test, P = 0.023) [Supplementary file 1]. Similar results were observed for clinical improvement on day 14 (cases 144, 80.9%, controls 125, 70.2%) [Supplementary file 1]. Univariable conditional logistic regression showed significant increase in odds of clinical recovery at day 7 (COR 1.838, 95% CI 1.184 2.854, P = 0.007) and day 14 (COR 1.863, 95% CI 1.109–3.141, P = 0.019). Covariate adjusted conditional logistic-regression models (adjusted to covariates included in survival analysis) for clinical improvement on day 7 showed a marginal trend toward significance (adjusted COR 4.535, 95% CI 1.0002–20.569, P = 0.050). However, no statistical significance was observed in the multivariable conditional logistic regression for clinical improvement on day 14 (adjusted COR 9.989, 95% CI 0.491–204.564, P = 0.135) [Figure 5].
|Figure 5 Conditional logistic-regression model for clinical improvement on days 7 and 14. †Adjusted to antivirals, tocilizumab, invasive ventilation at T0 and covariates with significant association in univariate analysis (age, diabetes, hypertension, chronic respiratory diseases, cardiovascular diseases, malignancy, low Glasgow coma scale, respiratory rate and oxygen saturation). CI, confidence interval; COR, conditional odds ratio.|
Click here to view
Adverse reactions to transfusion
No adverse reactions were observed in any of the CP recipients.
Cause of death
Respiratory failure was the commonest cause of death followed by sepsis and multiorgan dysfunction.
| Discussion|| |
Ever since CPT received regulatory approval, it has extensively been used in treatment of COVID-19. Initial case reports, that observed potentially better clinical outcomes, were hindered by limited sample size and lack of controls.,,
Following the use of CP in COVID-19 becoming more rampant, larger observational studies subsequently reported potential benefit regarding the use of CP leading to reduced mortality, hospital stay, and improved viral clearance. In an Indian study, Budhiraja et al. observed reduced mortality in CP transfused patients with severe COVID-19, the effect being more profound in the elderly, females, those with comorbidities and those who required ventilation. Joyner et al., in an observational study involving 3082 hospitalized adults with COVID-19 reported that mortality within 30 days was lower in patients who received CP within 3 days of diagnosis, and no effect on risk of death was observed among patients on IMV.
Joyner et al. also provided robust demonstrated evidence that CP transfusion is safe in hospitalized patients with COVID-19 and purported that administration of plasma earlier in the disease course was more likely to reduce mortality.
The earliest randomized controlled trials (RCTs) were from China by Li et al. and The Netherlands by Gharbharan et al. Both were halted prematurely—the Chinese study due to inadequate patient recruitment and the Dutch study after interim findings raised concerns regarding transfusion. Neither study found a mortality benefit.
Other RCTs, peer reviewed or otherwise, provided mixed evidence on the efficacy of CP. At this point, observational studies demonstrate more positive trends than RCTs, suggesting modest clinical effects and improvement in measurable outcomes. However, it is essential to consider the potential biases and shortcomings inherent in observational studies.
Two large and noteworthy RCTs, whose results have recently been made available are the PlasmAr and the RECOVERY trial. PlasmAr, an RCT evaluating outcomes in CP treated versus placebo in 333 patients with severe COVID-19, showed no significant differences in either mortality or clinical status at day 30 after transfusion. RECOVERY, an RCT of CP versus usual care in 11,558 hospitalized UK COVID-19 patients of all ages, showed no significant differences in either mortality or clinical status at day 28 after transfusion.
In India, the PLACID trial, a multicentered RCT concluded that the use of CP was associated with resolution of dyspnea and fatigue in patients with moderate COVID-19 and led to higher negative seroconversion on day 7 post enrollment, but no reduction in 28-day mortality or progression to severe disease was observed.
The PSMs have been proposed as a potential solution to confounding of the treatment–outcome association and are widely used in observational studies of therapeutic interventions., PSM allows effects to be estimated with use of univariable models after covariates were adjusted for in the matching process. In addition, the use of PSM makes it possible to replicate the measures of effect that are commonly reported in RCTs with time-to-event outcomes.
In the present study, treatment with CP, though associated with a lower mortality rate (18.5% vs. 24.7%) did not attain statistical significance. In the PSM sample, conditional logistic regression showed no significant mortality benefit of CPT.
Liu et al. reported findings from a similar study involving 39 patients with severe COVID-19 transfused with CP having antibody titers >1:320 dilution. The outcomes measured were clinical condition and oxygen requirements at 14 days. There was no significant difference in clinical condition between CP recipients and controls. Similarly, in the present study, the odds of clinical improvement among cases and PSM controls on day 14 were not significantly different. However, greater odds of clinical improvement were observed among CP recipients on day 7 in the multivariable analysis, though the difference was only marginally significant.
Liu et al. further reported a significant survival benefit of CP in the 1:4 matched data set (P = 0.048), with potential benefit in the 1:2 matched dataset (P = 0.14). In comparison, the results from the competing-risk survival analysis in the present study showed no significant benefit of CP in both unadjusted (P = 0.082) and covariate adjusted (P = 0.487) models.
A similar study by Salazar et al. included 136 cases matched to 251 controls and assessed the efficacy of CP transfusion versus standard of care as treatment for severe and/or critical COVID-19 reported a significant reduction (P = 0.047) in mortality within 28 days, specifically in patients transfused within 72 hours of admission with plasma having antispike protein receptor-binding domain (RBD) titer ≥1:1350. Relative to patients transfused within 72 hours of admission with anti-RBD IgG plasma with a titer ≥1:1350, patients transfused >72 hours after admission with anti-RBD IgG plasma with a titer <1:1350 had a significantly higher risk of mortality.
The study by Liu et al. and that by Salazar et al. both included patients with COVID-19 with severe and critical or life-threatening disease, whereas the present study included patients with moderate to severe disease at admission and excluded patients with immediately life-threatening disease. The study by Salazar et al. was prospective, whereas the present study and that by Liu et al. are retrospective in nature. The measures of outcome were similar across the three studies.
A substantial difference existed in the intervention. The CP recipients in the study by Salazar et al. received plasma with a higher neutralizing antibody titer >1:1350, compared to antibody titers >1:320 dilution in Liu et al.’s study. The cases in the present study were transfused with plasma having neutralizing antibody titers >1:40 (PRNT/MNT). The present study does not consider the effect of neutralizing antibody titer in CP or the timing of CP administration with respect to course of the disease. The differences in the results of the three studies, i.e, the one by Liu et al, the one by Salazar et al, and the present study, might possibly be accounted for by these factors.
The three studies, all made use of PSM to identify and compare controls. The study by Liu et al. employed generation of two sets of matched data based on 1:4 and 1:2 ratios for cases versus controls using the nearest neighbor-matching algorithm, with and without replacement, respectively. Salazar et al. performed a one to many nearest neighbor PSM without replacement using an initial case/control ratio of 1:3 and calipers of 1 between CP recipients (cases) versus controls. Further, a secondary matching was conducted based on the ventilation status at day 0 (ratio = 1:1, calipers <1).
In the present study, a 1:1 matching of cases and controls was carried out based on baseline patient parameters and clinical status at admission by using optimal algorithm. Other factors pertaining to pharmacologic interventions and oxygen requirement at T0 (day of CP transfusion for cases and the corresponding day of hospitalization for controls) were adjusted for in the multivariable analysis.
In the studies by Liu et al. and Salazar et al., day 0 was considered as the day of CP transfusion for cases and the corresponding day of admission for controls. In the present study as well, selecting T0 as the day of CP transfusion and the corresponding day of hospitalization was carried out for the purpose of time-to-event analysis.
Another difference was the use of various laboratory investigations as markers of disease severity. These included D-dimer and C-reactive protein at admission which were covariates used in propensity score generation in the study by Liu et al. These markers were not used for PSM in the present study, as most guidelines for use of CP in patients with COVID-19 do not consider these. The present study does not report the effect of CP on mortality in the presence of confounding effects of biomarkers of disease severity. This can be considered as a limitation of the present study.
The confounding effect of supplemental oxygen requirement on the outcomes associated with CPT was investigated. Supplemental oxygen requirement on T0 for cases and controls was compared. Among the mechanically ventilated patients, one patient in the CP recipient (case) group survived, whereas none survived in the control group. No survival benefit of CP was observed in the present study when adjusted for requirement of mechanical ventilation. The uneven representation of mechanical ventilation among case and control groups in the PSM sample precludes any further analysis for mechanically ventilated patients. Liu et al. reported in a subgroup analysis that CP recipients who were not mechanically ventilated at the time of transfusion were significantly less likely to die than their matched controls, whereas no significant association was observed in their study in patients who were mechanically ventilated at day 0. In comparison, Salazar et al. performed a secondary matching for ventilation status at day 0, balancing the significant difference between the case and control cohorts which existed with respect to ventilation after the first matching.
The time-independent logistic-regression analysis adjusted for concomitant use of investigational antivirals and that of tocilizumab with CP found no significant association of CPT with mortality. Similarly, the competing-risk survival analysis showed no significant benefit of CP when adjusted for administration of investigational antivirals and for tocilizumab. These findings are however confounded by a greater proportion of CP recipients having received investigational antivirals and a very low proportion of cases and controls having received tocilizumab. Further, the use of tocilizumab in the present study was under-represented and no generalizable conclusions can be drawn from it. These results corroborate with those of Liu et al.
To the best of our knowledge, there presently exists no large-scale observational or randomized controlled study comparing the concomitant use of investigational antivirals or that of tocilizumab along with CP on COVID-19 mortality or disease progression. The RCT by O’Donnell did match for use of remdesivir, but no conclusions were reported.
The results of the present study suggest a limited effectiveness of CP in the management of moderate and severe COVID-19. CP was associated with an earlier improvement in clinical status among recipients; however, this did not translate into improved survival. The study, however, is subject to certain limitations including low power, insufficient recruitment of patients on invasive mechanical ventilation, or those on other pharmaceutical interventions such as tocilizumab and investigational antivirals. These limitations need to be considered while interpreting results of the present study.
| Conclusion|| |
CPT in COVID-19 appears to have some benefits as evinced by greater odds of clinical recovery on day 7, however, offers no survival benefit.
Dr K.K. Astha, Dr J. Muthukrishnan, Dr K.S. Rajmohan, and Dr Monika Agarwal were involved with concept and design of study or acquisition of data and compiling of data.
Dr Kislay Kishore, Dr Sandeep Rana, Dr K.V. Padmaprakash, Dr Nishant Raman, and Dr Anirudh Anilkumar were involved with analysis and interpretation of data and draft of manuscript.
Dr Vasu Vardhan and Dr Sandeep Thareja were involved in manuscript drafting, editing, and preparation of final approval of the version to be published.
Each author has contributed to the conduct of study or preparation of manuscript. Due to the nature of the study and the duration it was conducted in, the compilation of data and its analysis required a multidisciplinary effort from various departments including internal medicine, respiratory medicine, pathology, blood transfusion, and biostatistics. The combined efforts from authors of these various departments led to the preparation of this manuscript. Furthermore, in the current scenario of COVID-19 where authors of this study were deployed at various COVID-19 hospitals throughout the country, efforts by different teams of researchers (who now constitute authors of the present study) was required at different periods during the conduct of this study.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| References|| |
Cascella M, Rajnik M, Cuomo A, Dulebohn SC, Di Napoli R. Features, Evaluation and Treatment Coronavirus (COVID-19). StatPearls Publishing; 2020.
Agarwal A, Mukherjee A, Kumar G et al.
Convalescent plasma in the management of moderate covid-19 in adults in India: open label phase II multicentre randomised controlled trial (PLACID Trial). BMJ 2020;371:m3939.
Marano G, Vaglio S, Pupella S et al.
Convalescent plasma: new evidence for an old therapeutic tool? Blood Transfusion 2016;14:152-7.
Rojas M, Rodríguez Y, Monsalve DM et al.
Convalescent plasma in Covid-19: possible mechanisms of action. Autoimmun Rev 2020;19:102554.
Katz LM. (A Little) Clarity on convalescent plasma for Covid-19. N Engl J Med 2021;384:666-8.
Piechotta V, Chai KL, Valk SJ et al.
Convalescent plasma or hyperimmune immunoglobulin for people with COVID-19: a living systematic review. Cochrane Database Syst Rev 2020 10:CD013600.
National Institutes of Health. COVID-19 Treatment Guidelines Clinical Spectrum of SARS-CoV-2 Infection. Coronavirus Disease 2019 (COVID-19) Treatment Guidelines. 2021;1-5.
Mandrekar JN. Receiver operating characteristic curve in diagnostic test assessment. J Thoracic Oncol 2010;5:1315-6.
Austin PC, Lee DS, Fine JP. Introduction to the analysis of survival data in the presence of competing risks. Circulation 2016;133:601-9.
Noordzij M, Leffondré K, van Stralen KJ, Zoccali C, Dekker FW, Jager KJ. When do we need competing risks methods for survival analysis in nephrology? Nephrol Dial Transplant 2013;28:2670-7. doi: 10.1093/ndt/gft355. Epub 2013 Aug 24. PMID: 23975843.
Beyersmann J, Friede T, Schmoor C. Design aspects of COVID-19 treatment trials: Improving probability and time of favourable events. November 27, 2020. Available at http://arxiv.org/abs/2012.02103
. Accessed June 16, 2021.
Zuccaro V, Celsa C, Sambo M et al.
Competing-risk analysis of coronavirus disease2019 in-hospital mortality in a Northern Italian centre from SMAtteo COvid19 REgistry (SMACORE). Scientific Reports [Internet]. December 1, 2021;11:1137. Available at https://doi.org/10.1038/s41598-020-80679-2
. Accessed June 16, 2021.
Cummings MJ, O’Donnell MR. Study of critically ill patients with COVID-19 in New York City − Authors’ reply [Internet]. Vol. 396, The Lancet. Lancet Publishing Group; 2020. p. 1064. Available at http://ees.elsevier.com/
. Accessed June 16, 2021.
Austin PC, Cafri G. Variance estimation when using propensity-score matching with replacement with survival or time-to-event outcomes. Statistics in Medicine [Internet]. 2020 May 20;39(11):1623-40. Available at https://onlinelibrary.wiley.com/doi/full/10.1002/sim.8502
. Accessed June 16, 2021.
Self WH, Semler MW, Leither LM et al.
Effect of hydroxychloroquine on clinical status at 14 days in hospitalized patients with COVID-19: a randomized clinical trial. JAMA − Journal of the American Medical Association [Internet]. 2020 December 1;324(21):2165-76. Available at https://jamanetwork.com/
. Accessed April 2, 2021.
Shen C, Wang Z, Zhao F et al.
Treatment of 5 critically ill patients with COVID-19 with convalescent plasma. JAMA 2020;323:1582-9.
Duan K, Liu B, Li C et al.
Effectiveness of convalescent plasma therapy in severe COVID-19 patients. Proceedings of the National Academy of Sciences of the United States of America [Internet]. April 28, 2020;117:9490-6. Available from: https://osf.io/gahz5.
Accessed June 1, 2021.
Salazar E, Perez KK, Ashraf M et al.
Treatment of coronavirus disease2019 (COVID-19) patients with convalescent plasma. Am J Pathol [Internet]. August 1, 2020;190:1680-90. Available at https://pubmed.ncbi.nlm.nih.gov/32473109/
. Accessed June 1, 2021.
Abolghasemi H, Eshghi P, Cheraghali AM et al.
Clinical efficacy of convalescent plasma for treatment of COVID-19 infections: Results of a multicenter clinical study. Transfus Apher Sci 2020;59:102875.
Budhiraja S, Dewan A, Aggarwal R et al.
Effectiveness of convalescent plasma in Indian patients with COVID-19. Blood Cells Mol Dis 2021;88:102548.
Joyner MJ, Carter RE, Senefeld JW et al.
Convalescent plasma antibody levels and the risk of death from Covid-19. N Engl J Med 2021;384:1015-27.
Joyner MJ, Bruno KA, Klassen SA et al.
Safety update: COVID-19 convalescent plasma in 20,000 hospitalized patients. Mayo Clin Proc 2020;95:1888-97.
Li L, Zhang W, Hu Y et al.
Effect of convalescent plasma therapy on time to clinical improvement in patients with severe and life-threatening COVID-19: a randomized clinical trial. JAMA 2020;324:460-70.
Gharbharan A, Jordans C, Geurtsvankessel C et al.
Convalescent plasma for COVID-19. A randomized clinical trial. medRxiv. 2020; 2020.08.26. 20182444.
O’Donnell MR, Grinsztejn B, Cummings MJ et al.
A randomized double-blind controlled trial of convalescent plasma in adults with severe COVID-19. J Clin Investig 2021; 2021.03.12. 21253373.
Katz LM. (A Little) Clarity on convalescent plasma for Covid-19. N Engl J Med 2021;384:666-8.
Simonovich VA, Burgos Pratx LD, Scibona P et al.
A randomized trial of convalescent plasma in Covid-19 severe pneumonia. N Engl J Med 2021;384:619-29.
The RECOVERY Collaborative Group. Convalescent plasma in patients admitted to hospital 2 with COVID-19 (RECOVERY): a randomised, The RECOVERY Collaborative Group*. medRxiv 2021; 2021.03.09. 21252736.
Dahabreh IJ, Sheldrick RC, Paulus JK et al.
Do observational studies using propensity score methods agree with randomized trials? A systematic comparison of studies on acute coronary syndromes. Eur Heart J 2012;33:1893-901.
Collins GS, Le Manach Y. Comparing treatment effects between propensity scores and randomized controlled trials: improving conduct and reporting. Eur Heart J 2012;33:1867-9.
Austin PC. The use of propensity score methods with survival or time-to-event outcomes: Reporting measures of effect similar to those used in randomized experiments. Stat Med 2014;33:1242-58.
Liu STH, Lin HM, Baine I et al.
Convalescent plasma treatment of severe COVID-19: a propensity score-matched control study. Nat Med 2020;26:1708-13.
Salazar E, Christensen PA, Graviss EA et al.
Treatment of coronavirus disease2019 patients with convalescent plasma reveals a signal of significantly decreased mortality. Am J Pathol 2020;190:2290-303.
[Figure 1], [Figure 2], [Figure 3], [Figure 4], [Figure 5]
[Table 1], [Table 2], [Table 3], [Table 4]