Lead Investigator: Malathi Ram, Johns Hopkins University
Title of Proposal Research: ACEi/ARB medications for hospitalized patients with COVID-19 – an individual patient data (IPD) based pooled analysis
Vivli Data Request: 7520
Funding Source: Funding is via an administrative supplement mechanism of an existing NCATS award
Potential Conflicts of Interest: None
Summary of the Proposed Research:
ACEi/ARB medications might be beneficial in COVID-19 due to inhibition of viral entry into pneumocytes and alteration of local inflammatory and vasoconstriction/vasodilation balance. The current preclinical and clinical database, albeit limited, suggests equipoise or potential benefit for these medications in hospitalized patients with COVID-19.
This study is designed to confirm our hypothesis that treatment of hospitalized patients with SARS-CoV-2 infection with ACEi/ARB medications will improve clinical outcome in comparison with control treatments.
The necessity of this research is urgent given that over 650,000 Americans have died of COVID-19, daily deaths again surpassed 2,000 at the peak of the delta variant surge, and thus there is a continued need for effective therapeutics.
Currently, starting ACEi/ARB for COVID-19 de novo is not standard because supportive data is weak and evolving. If administration of simple and inexpensive ACEi/ARB medications in COVID-19 will decrease morbidity and mortality, the drugs can have an important effect on outcome of many thousands of hospitalized patients internationally.
This study will be the first IPD-meta analysis of ACEi/ARB medications in COVID-19 and only the third IPD-meta analysis in COVID-19 more broadly.
After down selection of targeted RCTs, finalization of data harmonization spreadsheets and deidentification/anonymization processes, completion of data sharing agreements, trialists will collate IPD data per supplied data harmonization spreadsheets, and upload the data to Vivil. Data will be transferred to JHU where analysis per our prespecified SAP will occur using Bayesian statistical methods. The study, including its SAP, were registered on PROSPERO (CRD42021254261).
Statistical Analysis Plan:
This study is an individual patient data (IPD) meta-analysis of pragmatic trials of ACEi/ARB medications for the treatment of hospitalized patients with COVID-19.
The primary objective of the study is to assess the overall efficacy of ACEi/ARB compared with control treatments as a treatment for COVID-19 in an inpatient hospital setting.
The primary endpoint is differences in the NIAID COVID-19 7-point ordinal outcome scale (NCOSS) at day 28-30 post-randomization.
Secondary objectives are to (1) assess how the effect of ACEi/ARB varies with key individual-level covariates and (2) assess between-study heterogeneity.
The study’s protocol, including this SAP, will be registered on the PROSPERO international prospective registry of systematic reviews and thus be published online before outcome data are received. Fulfilling the PRISMA 2020 checklist will be ensured.
2. Study Design, Data Sources, and Trial Eligibility
We identified trials of ACEi/ARB to treat COVID-19 via clinicaltrials.gov. We searched for all N. America-based trials in February, 2021 and again in May, 2021, and invited Principal Investigators of trials connected with Clinical and Translational Science Award (CTSA) grant institutions and others to contribute data. In many cases willingness to collaborate was related to not having achieved planned sample sizes; most of the trials were stopping or had stopped at the time they agreed to participate. We will include only inpatient trials, which recruited individuals with varying severities of illness.
Inclusion criteria for inclusion of trials were:
Patient population is hospitalized COVID-19 patients
Intervention is comparison of new prescription of ACEi/ARB vs. control (placebo, standard of care, other)
Outcome measurements match – primary or secondary outcome measurements are NCOSS at 28-30 days or such data is extractable from collected data
Informed consent form allows data sharing and/or trial institution IRB allows data sharing of requested deidentified/anonymized dataset
Trialist agrees to this collaboration as per its preplanned tentative data harmonization, data analysis, data extraction/upload, and data sharing plans
The following 7 trials met inclusion criteria 1 and 2 and thus were targeted for inclusion (requests for participation were sent to trialists):
Institution sponsoring trial – North America
1 University of Kansas Medical Center
2 University of British Columbia
3 University of Minnesota
4 Bassett Medical Center
5 Sharp HealthCare
6 Ottawa Heart Institute Research Corporation
7 University of Hawaii at Manoa John A Burns School of Medicine
The attached Excel spreadsheet entitled, ACE ARB Pooling Study Sites RCTs – Appendix A summarizes details of the trials identified for inclusion in this study.
2.1 Risk of bias assessments for each trial
Two investigators will independently assess risk of bias of the primary outcome using the Risk of Bias 2 (RoB 2) tool, with disagreements resolved through discussion. We will assess the effect of assignment to intervention (the “intention-to-treat” effect). We will consider the following domains of bias, using the trial protocols and other information provided by the trial investigators:
bias arising from the randomization process (methods used to generate and conceal the allocation sequence);
bias due to deviations from intended interventions (whether participants and health professionals were masked to assigned intervention and methods used to ensure that participants received their allocated intervention);
bias due to missing primary outcome data;
bias in measurement of the primary outcome.
We will follow the recommended algorithms to reach an overall ‘risk of bias’ assessment for each trial. The overall risk of bias will have three categories: ‘low risk of bias,’ ‘high risk of bias,’ or ‘some concerns’.
See the attached Excel spreadsheet, entitled ACE ARB Pooling Study Data Harmonization Variable List v1 – Appendix B, for a full list of the variables requested of each participating trial.
3.1 Primary outcome
The primary outcome is the National Institute of Allergy and Infectious Diseases (NIAID) COVID-19 7-point ordinal outcome scale (NCOSS) measured anytime between day 28 and day 30 post-randomization. (For individuals with multiple measurements, the outcome will be taken as the earliest.)
Levels of the scale are 1) death; 2) hospitalized, on invasive mechanical ventilation or extracorporeal membrane oxygenation (ECMO); 3) hospitalized, on non-invasive ventilation or high flow oxygen devices; 4) hospitalized, requiring supplemental oxygen; 5) hospitalized, not requiring supplemental oxygen; 6) not hospitalized, limitation on activities; 7) not hospitalized, no limitations on activities.
3.2 Secondary outcomes
Our secondary outcomes are hospitalization length of stay and duration of mechanical ventilation. Analyses of these outcomes will be interpreted in light of any effect of treatment on mortality. To the extent possible, we will also examine the lowest NCOSS score achieved prior to day 28-30.
3.3 Safety outcomes
Number of serious adverse events (SAEs) (between enrollment and day 28/30).
All-cause mortality (between enrollment and day 28/30).
Acute kidney injury (AKI) SAEs/AEs
With the exception of the all-cause mortality outcome, analyses of these outcomes will be interpreted in light of any effect of treatment on mortality.
3.4 Merging trial arms
Trial arms will be coded as “treatment” or “control” as indicated in the attached Excel spreadsheet entitled, ACE ARB Pooling Study Sites RCTs – Appendix A.
4. Risk factors and covariates
From each trial, we will request treatment assignment, whether treatment arm included corticosteroids, and baseline covariates including age (binned in 5-year intervals, and truncated at age 90), sex, race and ethnicity, BMI, number of days before enrollment that symptoms began, mechanical ventilation status at time of enrollment, and baseline NCOSS status.
We will also request patients’ status for the following Charlson comorbidity variables (all coded as Yes/No): AIDS, cerebrovascular disease, prior myocardial infarction, congestive heart failure, dementia, chronic obstructive pulmonary disease (COPD), asthma, history of hypertension, HIV status (without AIDS), solid tumor, liver disease, diabetes mellitus, cigarette or tobacco smoking, and vaping.
See the attached Excel spreadsheet entitled, ACE ARB Pooling Study Data Harmonization Variable List v1 – Appendix B, for a complete list.
5. Populations: eligibility and subgroups
All participating studies recruited patients in an inpatient hospital setting. The studies included patients with varying degrees of illness. We will impose no additional exclusion criteria beyond those imposed by the participating trials; see Supplementary Table 1C for details of these.
6. Statistical analyses
6.1 Design-stage checks
Before the outcome data are shared with the analysis team, we will examine the distributions of baseline covariates within and between trials. If there are individuals with extreme values of baseline covariates, we will check their data with trial investigators, and potentially exclude them from the outcome analysis.
If there are substantial covariate imbalances (with respect to covariates not included in our model) between treated and control groups, either within a larger trial or overall, we will consider including that covariate in our outcome model.
We will also consider either simplifying or expanding our outcome data, depending on the number of patients in the study population.
We will also examine baseline covariates for missing data (See section 8).
6.2 Primary outcome – model
We will fit a Bayesian proportional odds ordinal regression model for NCOSS measured at day 28-30. For individuals with multiple measurements in this time interval, the outcome will be taken as the earliest measurement. In addition to treatment, whether the treatment included corticosteroids, and study, the model will include the following individual-level covariates: sex, age, number of baseline comorbidities, BMI, and baseline COVID outcome scale.
The coding and reference levels for these variables are as follows:
Covariate Coding Reference level Binned version
sex female = +1/2, male = -1/2 0 (midpoint) n/a
age (age in years – 60)/10 60 years old <18, 18-29, 30-49, 50-69, 70-79, 80+
Corticosteroids 1 = corticosteroids, 0 = no corticosteroids 0 = no corticosteroids n/a
Count of baseline comorbidities Raw count of baseline comorbidities 0 0, 1, 2, 3, ≥4
BMI (BMI – 25)/5 BMI of 25 ≤20, 20-25, 25-30, 30-35, ≥35
baseline COVID ordinal scale (NCOSS) indicators for levels 2-5 of the NCOSS (see Table 2) plus (5 – the numeric NCOSS score) 5 = hospitalized, not requiring supplemental oxygen (the highest possible value for inpatients) n/a
Model description. Let individual patients be indexed by i. Each patient has a vector of baseline covariates ci and is assigned to treatment ti, either ACEi/ARB (ti = 1) or control/standard of care/placebo (ti = 0). Let the primary outcome for individual i be denoted yi with levels indexed by l. Finally, let yi(0) and yi(1) be the potential outcomes for individual i under control t = 0 and treatment t = 1 respectively, regardless of whether in fact ti = 0 or ti = 1.
For each individual i, the model fits a vector of predicted probabilities that individual i has outcome yi, based on their covariates ci and treatment status ti. Changing the value of an individual’s ti in the model generates a prediction for the counterfactual outcome for a similar individual. In a cumulative ordinal model, the probabilities are linked to a linear predictor ηi via
P(y_i≤l”given” c_i,t_i)=〖”logit” 〗^(-1) (θ_l-η_i)
for l = 1, …, 6. The θl are cut points (also called intercepts) that are common to all individuals.
The linear predictor takes the form
■(η_i&=&x_i^T β+α_”study” +δ_”baseline NCOSS” @&+&t_i×[τ+x_i^T β^”treat” +α_”study” ^”treat” +δ_”baseline NCOSS” ^”treat” ])
Where xi (a function of ci) contains the predictors sex; natural cubic splines of age, BMI, and number of baseline comorbidities (with 3 degrees of freedom, generated with the R function ns()); 5 minus the numeric baseline NCOSS status; and an indicator for corticosteroids use;
the coefficients β and βtreat are given uniform priors;
the coefficients αstudy, , δbaseline NCOSS, and the corresponding treatment terms, are all modelled as independent mean-zero Normal random effects, each with their own standard deviation parameter distributed as a half Student-t distribution with 3 degrees of freedom and scale parameter 10.
The model will be fit using R, and the library ‘brms’.
To assess the sensitivity of our conclusions to modeling choices, we will repeat the core analysis with weakly informative N(0, 52) priors on the fixed effect coefficients, and more conservative half Student-t (df = 3, scale = 5) priors on the group-level standard deviations, repeat the core analysis with a stopping ratio (i.e., hazard) model allowing for hazard-specific effects, repeat the core analysis with study-level covariates, for example mean age and/or number of baseline comorbidities, explore the impact of adding time between symptom onset and enrollment to the model.
6.2 Primary outcome – effects of interest
For all of our effects of interest—the overall effect of treatment; the between-study heterogeneity; the individual-level treatment-covariate interactions—we will produce two kinds of effect estimates:
Conditional effect estimates. These correspond to coefficients and fitted curves from the regression model. They admit conditional interpretations, with the other included covariates set to their reference levels (see above). When reporting conditional estimates, we will average over the super-population distribution of between-study effects.
For the conditional effect estimates, our scale will be the relative risk of “mechanical ventilation or death” (levels 1 and 2 of the NCOSS outcome scale).
Standardized (post-stratified) effect estimates. These represent the effect of the treatment in a given population, averaged over that population’s distribution of individual-level covariates.
For examining treatment heterogeneity within categorical covariates and between studies, we will standardize to the empirical covariate distribution. For treatment interactions with continuous covariates, we will divide the study population into covariate-based bins (see table above) and standardize to the empirical covariate distribution within each bin.
We will produce standardized estimates by drawing from the posterior predictive distribution of unobserved, counterfactual outcomes, combining these with the (fixed) observed outcomes, and regressing these potential outcomes against the corresponding treatment indicator.
The advantage of standardized over conditional effect estimates is that they are directly comparable to unadjusted “plug-in” maximum likelihood effect estimates. We will present both of these in our results, with the maximum likelihood confidence intervals uncorrected for multiple testing.
For the standardized effect estimates, our effect measure will be the “proportional odds ratio” corresponding to the estimate from a proportional odds model fit to the data by maximum likelihood. We will present this estimate as our primary assessment of the effect of treatment.
We will also consider presenting selected absolute effects: probabilities of the different outcome categories under the treatment and control conditions.
6.2 Primary outcome – communicating results
We will present our estimates of standardized (post-stratified) effects with forest plots that contrast model-adjusted and plug-in estimates. These will be presented along with the associated tables.
We will present conditional effect estimates with figures and tables appropriate to the nature of the covariate (e.g., line graphs for continuous covariates; “point-range” plots for discrete covariates).
6.3 Secondary outcomes and safety outcomes (excluding death and lowest NCOSS score)
Because both secondary outcomes are potentially compromised by truncation-by-death, we will not pre-register a model-based analysis; our analysis of these outcomes will be primarily exploratory.
6.4 Model checking and robustness
For the primary outcome model, we will conduct and report posterior predictive checks and other model checking diagnostics, and comment on the extent to which they cast doubt on the main analysis. We may conduct follow-up analyses if model fit is poor.
For each treatment-covariate interaction in the model, we will examine, using exploratory plots, the association between the estimated study-specific treatment effect and the study-specific mean or median level of the covariate.
We may also examine in an exploratory manner the effect of treatment duration and dose.
7. Missing data
We will assess the extent of missing baseline covariate data at the design stage. If the missingness is judged to be minor and sporadic, the missing values will be filled-in using multiple imputation based on a selection of other baseline covariates, but excluding treatment assignment and outcome. If the missingness is judged to be severe, the corresponding baseline variables will be excluded from the analysis.
Any missing outcome data will be modeled under the assumption that it is missing at random conditional on the covariates included in our regression model.
8. Quality Assurance of Statistical Programming
A second statistician will verify the code and re-run all analyses used in fitting the model and generating the subsequent tables and figures.
All code will be made publicly available.
An Open Label Phase 1 Trial of Losartan for Worsening Respiratory Illness in COVID-19
Data Contributor: University of Kansas Medical Center
Study ID: NCT04335123
Sponsor ID: STUDY00145514
Randomized Open Label Study of Standard of Care Plus an Angiotensin II Receptor Blocker Compared to Standard of Care Alone to Minimize the Progression to Respiratory Failure in SARS-CoV-2 Infection
Data Contributor: Sharp Healthcare
Study ID: NCT04340557
Sponsor ID: COVID-ARB
Randomized Controlled Trial of Losartan for Patients With COVID-19 Requiring Hospitalization
Data Contributor: University of Minnesota
Study ID: NCT04312009
Sponsor ID: SURG-2020-28675
COVIDMED-Comparison Of Therapeutics for Hospitalized Patients Infected With SARS-CoV
Data Contributor: Bassett Medical Center
Study ID: NCT04328012
Sponsor ID: NCT04328012
Di Stefano, L., Ram, M., Scharfstein, D. O., Li, T., Khanal, P., Baksh, S. N., … & Freilich, D. A. (2023). Losartan in hospitalized patients with COVID-19 in North America: An individual participant data meta-analysis. Medicine, 102(23). Doi: 10.1097/MD.0000000000033904