Abstract
Background:
Coronavirus disease 2019 (COVID-19) pneumonia requires considerable healthcare resources.
Research Objective:
Examine if a single dose of infliximab or abatacept, in addition to remdesivir and steroids, decreased resource utilization among participants hospitalized with COVID-19 pneumonia.
Study Design and Methods:
Accelerating COVID-19 Therapeutic Interventions and Vaccines Immunomodulator (ACTIV-1 IM) master protocol was a randomized, placebo-controlled trial examining the potential benefit in time to recovery and mortality of immunomodulators infliximab, abatacept, and cenicriviroc. This observational study performs a secondary analysis of the infliximab, abatacept, and common placebo participants to examine resource utilization. Hospital days, intensive care unit days, days with supplemental oxygen, days with high flow nasal cannula or non-invasive ventilation, ventilator days, and days of extracorporeal membrane oxygenation were each examined. Proportional odds models were used to compare days alive and free of resource use over 28 days between infliximab and placebo groups and between abatacept and placebo groups.
Results:
Infliximab infusion, compared to placebo, was associated with greater odds of being alive and free of all interventions tested. Abatacept use was associated only with greater odds of days alive and free of hospitalization and supplemental oxygen.
Interpretation:
Infliximab and abatacept use were associated with decreased use of healthcare resources over 28 days compared to placebo, but the absolute differences were small.
Clinical trial registration:
Keywords: COVID-19, infliximab, abatacept, hospitalization, resource utilization, ventilator, supplemental oxygen
Introduction
Severe coronavirus disease 2019 (COVID-19) pneumonia leads to hospitalization and medical resource utilization (1). The Accelerating COVID-19 Therapeutic Interventions and Vaccines Immunomodulator (ACTIV-1 IM) master protocol was a randomized trial investigating immunomodulators added to a standard care treatment, including remdesivir and steroids, among hospitalized patients between October 2020 and December 2021. It is widely accepted that source control has decreased morbidity and mortality in sepsis, but up to a third of patients still succumb, and additive treatment beyond direct anti-infectives is needed (2). Both infliximab, a tumor necrosis factor (TNF)-alpha inhibitor, and abatacept, a T-cell costimulatory modulator, were included in the master protocol as agents with the potential to limit damage related to immune dysregulation caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) virus.
The ACTIV-1 IM trial did not show differences in the primary endpoint of time to recovery - defined as an improvement on an ordinal scale to a level where patients no longer need supplemental oxygen or ongoing in-hospital medical care - with a single dose of intravenous infliximab (5mg/kg) or intravenous abatacept (10mg/kg), each compared to a shared placebo arm (3). In secondary outcomes, both the infliximab and abatacept treated-as-assigned populations had a decrease in all-cause 28-day mortality. No significant differences were found in composite safety endpoints for either drug compared to placebo. The National Institutes of Health COVID-19 Treatment Guidelines include infliximab and abatacept as alternative second immunomodulator options for the management of hospitalized adults requiring conventional oxygen (4).
This study is a secondary data analysis of the differences in resource utilization between those who received infliximab, abatacept, and shared placebo. We sought to determine if days of hospitalization, intensive care unit (ICU) care, supplemental oxygen via nasal cannula (O2) use, non-invasive ventilation/high-flow oxygen nasal cannula (NIV/HFNC) use, invasive mechanical ventilation (MV) use, and extracorporeal membrane oxygenation (ECMO) use were lower in those who received treatment with immunomodulator therapy (infliximab or abatacept) as compared to those receiving placebo. Although findings from secondary data analyses on resource use may not directly impact clinical care, such research is vital to understand how disease and treatment choices affect resource allocation in resource limited settings.
Methods
Data Sources
Detailed methodology of the ACTIV-1 IM master protocol, including data collection and the primary outcome results, have been previously published (3). ACTIV-1 IM was a randomized, double-masked, placebo-controlled trial using a master protocol design to examine several different immunomodulators added to standard care for hospitalized patients with COVID-19 pneumonia. The trial enrolled from October 2020 to December 2021 at 85 sites (95 hospitals) in the United States of America (US) and Latin America, including Mexico, Peru, Argentina, and Brazil. Although participants were followed until day 60 for clinical status and safety endpoints, detailed data on resource utilization available for this secondary analysis was only collected through day 28. ACTIV-1 IM used an innovative design with a common, shared placebo arm that minimizes the number of participants receiving placebo and reduces the sample size needed to retain adequate power while still evaluating multiple agents (3, 5). ACTIV-1 IM was approved by the institutional review board at all sites, and informed consent was obtained for all participants.
In this manuscript, we conduct a secondary analysis of the primary data. A STROBE checklist is available (e1-Table 1).
Study Participants
Participants enrolled in ACTIV-1 IM were male and non-pregnant females ≥18 years with confirmed SARS-CoV-2 infection within the prior 14 days, evidence of pneumonia by radiography or oxygen saturation ≤94% on room air and the expectation to remain hospitalized for ≥72 hours (3). Potential candidates were excluded if they had elevated liver function tests (aspartate aminotransferase or alanine aminotransferase >10 times the upper limit of normal), chronic liver disease, acute kidney injury (estimated glomerular filtration rate <30 ml/min, stable chronic renal insufficiency was permitted), severe heart failure (new right heart failure from COVID-19 was allowed), severe neutropenia (<1.0 x 103/μl) or lymphopenia (<0.2 x 103/μl), or known untreated infection other than COVID-19. The study provided remdesivir (Gilead Sciences, Foster City, CA), and steroids were allowed per hospital standard care. Participants could not receive cytotoxic or biologic targeted immune-modulator treatments within 4 weeks or 5 half-lives prior to screening. SARS-CoV-2 vaccines were not yet available at the beginning of the trial, and their use was not consistently tracked once they became available.
Patients randomized to receive infliximab, abatacept, or a placebo were included in this secondary analysis. ACTIV-1 IM included a cenicriviroc (CVC) substudy that was stopped for futility. Excluded participants were those that were randomized to the placebo arm of the CVC substudy or were not treated after randomization. As the CVC substudy was closed early due to a lack of benefit, we wanted to remove bias by also excluding the CVC arm placebo recipients that received placebo pills instead of a single infusion placebo, leading to a higher dropout rate in the study. Not-treated defines participants who were randomized but did not receive their assigned intervention.
Variables
Six healthcare resources for pneumonia care were examined: hospitalization, ICU care, O2 use, NIV/HFNC, MV, or ECMO. Resource use was captured daily for hospitalized participants. ACTIV-1 IM allowed select participants with baseline O2 use but measured only the need for any O2 use. O2 use after discharge was captured on days 8, 11, 15, 22, and 29 for the previous day. Last captured data was used for participants missing data O2 use (e1-Table 2).
Baseline covariates were selected a priori from among the variables collected in ACTIV-1 IM by the authors based on both literature review and clinical experience as factors that could be associated with increased resource use in severe COVID-19 infection. The selected variables were age, sex, geographic region, baseline clinical status (8-point ordinal scale), body mass index (BMI), diabetes mellitus, hypertension, coronary artery disease, chronic obstructive pulmonary disease, and asthma. Age was treated as an ordinal variable (<40, 40-64, ≥65 years) as the majority of participants were in the 40-64 year age range. The 8-point ordinal scale, a method to categorize COVID-19 clinical status, was described in the ACTIV-1 IM protocol (3). BMI was treated as an ordinal variable using standard BMI classifications for underweight (≤18.5 kg/m2), normal weight (18.50-24.99 kg/m2), overweight (25-29.99 kg/m2), and obese (≥30 kg/m2). Multiple imputation was performed for participants missing BMI in statistical models.
Outcomes
The primary outcomes of interest were: 1) days alive and free of each resource from day 1 to day 28. The primary exposure of interest was the receipt of infliximab or abatacept compared to placebo. The original study was not designed to compare infliximab and abatacept arms directly.
Statistical Analysis
Descriptive statistics for non-missing values are presented as counts (percentages) for categorical variables and as medians (25th,75th percentiles) or as means (standard deviations) for continuous variables.
To evaluate resource use outcomes between infliximab compared to placebo and abatacept compared to placebo, we examined resource-free days as an ordinal variable combining resource-free day outcomes with mortality and adjusted for covariates using a proportional odds (PO) model. This innovative modeling approach, previously used to study the outcome of oxygen-free days in COVID-19 (6), has two main advantages over other models used to examine resource utilization. First, combining resource-free days with mortality is particularly important since 12.2% of study participants died by day 29. Second, the PO model is appropriate for ordinal outcomes like resource-free days, where the levels are ordered, but the relative differences between levels are not defined. The PO model acknowledges that the days-free variable is not truly continuous: a patient who dies after using 20 days of resources is not the same as a patient who dies after using 5 days of resources (e2-Appendix 1).
First, we summarized the mean days alive and free of hospital resources among survivors. Days alive and free were analyzed as an ordinal outcome with 30 possible response levels (−1 to 28 days). Death was assigned the value of −1 to avoid attributing less resource use to those who died prior to day 28. A “first on, last off’ method was used to assign each participant their days-free outcome: all days between first initiation and last use of the resource were counted as “resource use days” (e2-Appendix 2). PO models were fitted to test the association between study groups and days alive and free outcomes. Unadjusted and adjusted odds ratios (ORs) were estimated from proportional odds models with a 95% confidence interval (CI). An OR of >1.0 indicates a greater number of resource-free days (benefit) in the intervention group than in the placebo group (e2-Appendix 3). The PO assumption was violated, but the ORs remained interpretable (7) and reflect a global assessment of treatment effectiveness (e2-Appendix 4).
Given that days of resource use, in and of itself, is an important clinical question for hospitals during a pandemic, we also summarized and compared mean differences in days that each resource was used without penalizing for death and adjusted for covariates using a generalized mixed linear model (e3-Appendix 1). With the generalized linear model, we were not assessing whether the intervention made the patient better or worse but whether the intervention changed the number of days of resources that the hospital had to provide for study patients over a 29 day period.
All data analyses were performed with SAS v9.4 (SAS Institute Inc., Cary, NC).
Results
A flow diagram of the study population is shown in Figure 1. All participants who were randomized to receive infliximab, abatacept, or a placebo (N=1570) were initially included in this secondary analysis. Participants who received the CVC arm placebo (n=124, 7.9%) or were not treated after randomization (n=38, 2.4%; abatacept arm 20, infliximab arm 18) were excluded from the final population (N=1446). Baseline characteristics comparing excluded patients and the final population are shown in e1-Table 3. Five hundred and eighteen participants received infliximab, 509 received abatacept, and 419 received a single saline infusion as the common placebo.
Figure 1. Study Population.
Flow diagram describing the study population.
Abbreviations: ACTIV-1 IM = Accelerating COVID-19 Therapeutic Interventions and Vaccines Immunomodulator; IV = intravenous
Baseline characteristics of these 3 groups are shown in Table 1. Thirty-two participants (2.1%) were missing BMI values. Randomization worked well in the ACTIV-1 IM trial, and the 3 groups in our observational study remained similar with regard to demographic data, comorbidities, and clinical presentation. At the time of randomization, approximately 4% of participants were hospitalized without O2 (3.3% for placebo), ~52% required O2, ~33% required NIV/HFNC (34% for infliximab), and ~11% required MV/ECMO (9.2% for infliximab). More than 94% of participants received remdesivir, and 93% participants received either dexamethasone or prednisone (94.5% in the placebo group). A few participants received tocilizumab (32[2.2%]) or baricitinib (25[1.7%]) after randomization. At day 28, 176 (12.2%) participants had died, 157 (14.1%) were alive and remained hospitalized, and 1113 (77.0%) were alive and discharged. Days of resource use by day 28 mortality and discharge status is shown in e1-Table 4.
Table 1.
Participant Characteristics by Study Group
| Infliximab N=518 | Abatacept N=509 | Placebo N=419 | All Participants N=1446 | |
|---|---|---|---|---|
| Age, median years (q1, q3)a | 55.0 (44.0, 66.0) | 55.0 (44.0, 65.0) | 56.0 (45.0, 65.0) | 55.0 (44.0, 65.0) |
| Age (years) | ||||
| <40 | 94 (18.1%) | 87 (17.1%) | 68 (16.2%) | 249 (17.2%) |
| 40-64 | 281 (54.2%) | 289 (56.8%) | 235 (56.1%) | 805 (55.7%) |
| ≥65 | 143 (62.9%) | 133 (26.1%) | 116 (27.7%) | 392 (27.1%) |
| Sex | ||||
| Female | 192 (37.1%) | 188 (36.9%) | 175 (41.8%) | 555 (38.4%) |
| Male | 326 (62.9%) | 321 (63.1%) | 244 (58.2%) | 891(61.6%) |
| Race | ||||
| White | 320 (61.8%) | 318 (62.5%) | 277 (66.1%) | 915 (63.3%) |
| Black or African American | 77 (14.9%) | 70 (13.8%) | 47 (11.2%) | 194 (13.4%) |
| American Indian or Alaska Native | 5 (1.0%) | 5 (1.0%) | 5 (1.2%) | 15 (1.0%) |
| Asian | 11 (2.1%) | 16 (3.1%) | 12 (2.9%) | 39 (2.7%) |
| Native Hawaiian or Pacific Islander | 1 (0.2%) | 1 (0.2%) | 0 (0.0%) | 2 (0.1%) |
| Multiracial | 2 (0.4%) | 0 (0.0%) | 2 (0.5%) | 4 (0.3%) |
| Other | 76 (14.7%) | 71 (13.9%) | 59 (14.1%) | 206 (14.2%) |
| Unknown | 26 (5.0%) | 28 (5.5%) | 17 (4.1%) | 71 (4.9%) |
| Ethnicity | ||||
| Hispanic or Latino | 253 (48.8%) | 214 (42.0%) | 208 (49.6%) | 675 (46.7%) |
| Not Hispanic or Latino | 255 (49.2%) | 274 (53.8%) | 202 (48.2%) | 731 (50.6%) |
| Unknown | 10 (1.9%) | 21 (4.1%) | 9 (2.1%) | 40 (2.8%) |
| Geographic region | ||||
| Argentina | 56 (10.8%) | 52 (10.2%) | 49 (11.7%) | 157 (10.9%) |
| Brazil | 54 (10.4%) | 45 (8.8%) | 43 10.3%) | 142 (9.8%) |
| Mexico | 15 (2.9%) | 18 (3.5%) | 14 (3.3%) | 47 (3.3%) |
| Peru | 62 (12.0%) | 54 (10.6%) | 46 (11.0%) | 162 (11.2%) |
| United States of America | 331 (63.9%) | 340 (66.8%) | 267 (63.7%) | 938 (64.9%) |
| BMI (kg/m2)b | ||||
| <18.50 | 2 (0.4%) | 4 (0.8%) | 3 (0.7%) | 9 (0.6%) |
| 18.50-24.99 | 58 (11.4%) | 60 (11.9%) | 53 (13.1%) | 171 (12.1%) |
| 25.0-29.99 | 179 (35.2%) | 152 (30.2%) | 106 (26.1%) | 437 (30.8%) |
| ≥30.00 | 269 (53.0%) | 287 (57.1%) | 244 (60.1%) | 800 (56.5%) |
| Comorbidities | ||||
| Hypertension | 207 (40.0%) | 213 (41.8%) | 168 (40.1%) | 588 (40.7%) |
| Diabetes mellitus | 138 (26.6%) | 146 (28.7%) | 121 (28.9%) | 405 (28.0%) |
| Coronary artery disease | 38 (7.3%) | 33 (6.5%) | 18 (4.3%) | 89 (6.2%) |
| History of heart failure | 16 (3.1%) | 16 (3.1%) | 13 (3.1%) | 45 (3.1%) |
| History of cancer | 34 (6.6%) | 31 (6.1%) | 30 (7.2%) | 95 (6.6%) |
| Asthma | 35 (6.8%) | 36 (7.1%) | 47 (11.2%) | 118 (8.2%) |
| COPD | 24 (4.6%) | 19 (3.7%) | 20 (4.8%) | 63 (4.4%) |
| HIV infection | 2 (0.4%) | 3 (0.6%) | 3 (0.7%) | 8 (0.6%) |
| Severe liver disease | 3 (0.6%) | 1 (0.2%) | 1 (0.2%) | 5 (0.3%) |
| Severe kidney disease | 3 (0.6%) | 4 (0.8%) | 6 (1.4%) | 13 (0.9%) |
| Time prior to study treatment | ||||
| Days of symptoms; median (q1, q3) | 10.0 (7.0, 13.0) | 9.0 (7.0, 12.0) | 9.0 (7.0, 12.0) | 9.0 (7.0, 12.0) |
| Days from diagnosis; median (q1, q3)c | 3.0 (1.0, 7.0) | 3.0 (1.0, 6.0 | 3.0 (1.0, 7.0) | 3.0 (1.0, 7.0) |
| Day from hospitalization; median (q1, q3) | 2.0 (1.0, 3.0) | 2.0 (1.0, 3.0) | 2.0 (1.0, 3.0) | 2.0 (1.0, 3.0) |
| Clinical status at randomization (8-point ordinal score) | ||||
| 1: Dead | 0 (0%) | 0 (0%) | 0 (0%) | 0 (0%) |
| 2: Hospitalized, on ECMO or MV | 58 (11.2%) | 47 (9.2%) | 46 (11.0%) | 151 (10.4%) |
| 3: Hospitalized, on NIV or HFNC | 171 (33.0%) | 173 (34.0%) | 139 (33.2%) | 483 (33.4%) |
| 4: Hospitalized, on O2 | 268 (51.7%) | 266 (52.3%) | 220 (52.5%) | 754 (52.1%) |
| 5: Hospitalized, not requiring O2 | 21 (4.1%) | 23 (4.5%) | 14 (3.3%) | 58 (4.0%) |
| 6: Hospitalized, not requiring O2 and no longer requiring hospital level care | 0 (0%) | 0 (0%) | 0 (0%) | 0 (0%) |
| 7: Not hospitalized, with limitations on activities and/or requiring home O2 | 0 (0%) | 0 (0%) | 0 (0%) | 0 (0%) |
| 8: Not hospitalized, no limitations on activities | 0 (0%) | 0 (0%) | 0 (0%) | 0 (0%) |
| COVID-19-directed therapy on study | ||||
| Remdesivir (at entry) | 488 (94.2%) | 485 (95.3%) | 397 (94.7%) | 1370 (94.7%) |
| Dexamethasone or prednisone (at entry) | 481 (92.9%) | 473 (92.9%) | 396 (94.5%) | 1350 (93.4%) |
| Tocilizumab (post-randomization) | 7 (1.4%) | 14 (2.8%) | 11 (2.6%) | 32 (2.2%) |
| Baricitinib (post-randomization) | 6 (1.2%) | 6 (1.2%) | 13 (3.1%) | 25 (1.7%) |
Abbreviations: BMI = body mass index; COPD = chronic obstructive pulmonary disease; COVID-19 = coronavirus disease 2019; ECMO = extracorporeal membrane oxygenation; HFNC = high-flow nasal canula; HIV = human immunodeficiency virus; MV = invasive mechanical ventilation; N = number; NIV, non-invasive ventilation; O2 = supplemental oxygen by nasal canula; q1 = quarter 1; q3 = quarter 3
(q1, q3) = (25th, 75th percentiles)
BMI with 2.1% missing data: infliximab (N=508), abatacept (N=503), placebo (N=406), all participants (N=1417).
Diagnosis with documented positive SARS-CoV-2 test
Table 2 shows days alive and free of the following resources among survivors: 1) hospitalization; 2) ICU care; 3) ECMO/MV use; 4) ECMO/MV/NIV/HFNC use; and 5) O2 use (in hospital or at home). With descriptive statistics, few differences were observed between study groups for days alive and free of resource use over 28 days of observation. Overall, 1270 survivors had a mean of 16.9 days alive and free of hospitalization, 24.2 days alive and free of ICU care, 25.4 days alive and free of MV/ECMO, 22.4 days alive and free of ECMO/MV/NIV/HFNC, and 1.0 day free of O2. In the PO model comparing infliximab and placebo use, statistically significant differences in resource utilization were noted in several unadjusted models and all adjusted models (Table 3, e2-Table 1). Specifically, after adjusting for covariates, those receiving infliximab compared to those receiving placebo had 48% greater odds (OR 1.48; 95% CI 1.18–1.86) of having more hospital-free days, 50% greater odds (OR 1.50; 95% CI 1.17–1.96) of having more days alive and free of ECMO/MV/NIV/HFNC, and 42% greater odds (OR 1.42; 95% CI 1.11–1.84) of having more alive and O2-free days. Less advantage was noted with abatacept receipt compared to placebo except for alive and hospital-free days (Table 4, e2-Table 2). After adjustment, those receiving abatacept had 40% greater odds (OR 1.40; 95% CI 1.11–1.84) of days alive hospital-free compared to those receiving placebo.
Table 2.
Resource-free Days by Intervention among Survivors Over 28 Days
| Outcome | Infliximab N=466 | Abatacept N=453 | Placebo N=351 | All Participants N=1270 |
|---|---|---|---|---|
| Alive and free of hospitalization (days) | ||||
| Mean (SD) | 17.1 (7.95) | 16.9 (8.22) | 16.5 (8.15) | 16.9 (8.10) |
| Median (q1, q3)a | 20.0 (14.0, 23.0) | 20.0 (13.0, 23.0) | 19.0 (13.0, 23.0) | 20.0 (13.0, 23.0) |
| Alive and out of ICU (days) | ||||
| Mean (SD) | 24.2 (7.65) | 24.3 (7.36) | 24.0 (8.22) | 24.2 (7.71) |
| Median (q1, q3) | 28.0 (24.0, 28.0) | 28.0 (24.0, 28.0) | 28.0 (25.0, 28.0) | 28.0 (24.0, 28.0) |
| Alive and free ECMO/MV (days) | ||||
| Mean (SD) | 25.2 (6.97) | 25.6 (6.45) | 25.3 (7.30) | 25.4 (6.88) |
| Median (q1, q3) | 28.0 (28.0, 28.0) | 28.0 (28.0, 28.0) | 28.0 (28.0, 28.0) | 28.0 (28.0, 28.0) |
| Alive and free of ECMO/MV/NIV/HFNC (days) | ||||
| Mean (SD) | 22.6 (8.47) | 22.3 (8.31) | 22.2 (8.94) | 22.4 (8.54) |
| Median (q1, q3) | 27.0 (21.0, 28.0) | 27.0 (20.0, 28.0) | 27.0 (20.0, 28.0) | 27.0 (21.0, 28.0) |
| Alive and free of O2b (days) | ||||
| Mean (SD) | 0.9 (2.50) | 1.2 (3.65) | 0.8 (1.71) | 1.0 (2.81) |
| Median (q1, q3) | 1.0 (0.0, 1.0) | 1.0 (0.0, 1.0) | 1.0 (0.0, 1.0) | 1.0 (0.0, 1.0) |
Abbreviations: ECMO = extracorporeal membrane oxygenation; HFNC = high-flow nasal canula; ICU = intensive care unit; MV = invasive mechanical ventilation; N = number; NIV = non-invasive ventilation; O2 = supplemental oxygen by nasal canula; q1 = quarter 1; q3 = quarter 3; SD = standard deviation
(q1, q3) = (25th, 75th percentiles)
Includes in hospital or at home O2. Last captured data was used for participants missing data on supplemental O2 use at home.
Table 3.
Days Alive and Free of Resource Use for Infliximab Compared to Placebo Over 28 Days (N=937)a
| Outcome | Infliximab N=518 Mean (SD) | Placebo N=419 Mean (SD) | Unadjusted OR (95% CI) | p-value | Adjusted ORb (95% CI) | p-value |
|---|---|---|---|---|---|---|
| Days alive and free of hospitalization | 17.1 (7.95) | 16.5 (8.15) | 1.36 (1.08-1.70) | 0.01 | 1.48 (1.18-1.86) | 0.00 |
| Days alive and free of ICU care | 24.2 (7.65) | 24.0 (8.22) | 1.22 (0.95-1.57) | 0.12 | 1.32 (1.00-1.75) | 0.05 |
| Days alive and free of ECMO/MV | 25.2 (6.97) | 25.3 (7.30) | 1.28 (0.97-1.68) | 0.08 | 1.41 (1.03-1.94) | 0.03 |
| Days alive and free of ECMO/MV/NIV/HFNC | 22.6 (8.47) | 22.2 (8.94) | 1.29 (1.02-1.62) | 0.03 | 1.50 (1.17-1.96) | 0.00 |
| Days alive and free of O2c | 0.9 (2.50) | 0.8 (1.71) | 1.37 (1.08-1.75) | 0.01 | 1.42 (1.11-1.84) | 0.01 |
Abbreviations: CI = confidence interval; ECMO = extracorporeal membrane oxygenation; HFNC = high-flow nasal canula; ICU = intensive care unit; MV = invasive mechanical ventilation; N = number; NIV = non-invasive ventilation; O2 = supplemental oxygen by nasal canula; OR = odds ratio; SD = standard deviation
Proportional odds model (reference is placebo). Death is penalized as −1.
Covariates: age, sex, geographic region, baseline clinical status (8-point ordinal scale), body mass index, diabetes mellitus, hypertension, coronary artery disease, chronic obstructive pulmonary disease, and asthma. Body mass index was imputed for 32 patients with missing data.
In-hospital and at-home O2 use. Last captured data was used for participants missing data on supplemental O2 use at home.
Table 4.
Days Alive and Free of Resource Use for Abatacept Compared to Placebo Over 28 Days (N=928)a
| Outcome | Abatacept N=509 Mean (SD) | Placebo N=419 Mean (SD) | Unadjusted OR (95% CI) | p-value | Adjusted ORb (95% CI) | p-value |
|---|---|---|---|---|---|---|
| Days alive and free of hospitalization | 16.9 (8.22) | 16.5 (8.15) | 1.30 (1.04-1.62) | 0.02 | 1.40 (1.11-1.76) | 0.05 |
| Days alive and free of ICU care | 24.3 (7.36) | 24.0 (8.22) | 1.17 (0.91-1.50) | 0.22 | 1.18 (0.89-1.56) | 0.24 |
| Days alive and free of ECMO/MV | 25.6 (6.45) | 25.3 (7.30) | 1.30 (0.98-1.72) | 0.06 | 1.36 (0.98-1.87) | 0.06 |
| Days alive and free of ECMO/MV/NIV/HFNC | 22.3 (8.31) | 22.2 (8.94) | 1.15 (0.91-1.45) | 0.25 | 1.19 (0.93-1.54) | 0.16 |
| Days alive and free of O2c | 1.2 (3.65) | 0.8 (1.71) | 1.31 (1.03-1.66) | 0.03 | 1.34 (1.04-1.72) | 0.02 |
Abbreviations: CI = confidence interval; ECMO = extracorporeal membrane oxygenation; HFNC = high-flow nasal canula; ICU = intensive care unit; MV = invasive mechanical ventilation; N = number; NIV = non-invasive ventilation; O2 = supplemental oxygen by nasal canula; OR = odds ratio; SD = standard deviation
Proportional odds model (reference is placebo). Death is penalized as −1.
Covariates: age, sex, geographic region, baseline clinical status (8-point ordinal scale), body mass index, diabetes mellitus, hypertension, coronary artery disease, chronic obstructive pulmonary disease, and asthma. Body mass index was imputed for 32 patients with missing data.
In-hospital and at-home O2 use. Last captured data was used for participants missing data on supplemental O2 use at home.
The results and tables from the second analysis summarizing and comparing mean differences in days of resource use from day 1 to day 28 without penalizing for death and adjusting for covariates using a generalized linear model are in e3-Appendix 2 and e3-Tables 1–5.
Discussion
The Randomized Evaluation of COVID-19 Therapy (RECOVERY) trial found that among patients hospitalized with COVID-19 infection requiring supplemental oxygen, dexamethasone 6mg daily for 10 days lowered 28-day mortality (5). National Institutes of Health COVID-19 Treatment Guidelines suggest adding a second immunomodulator to dexamethasone, such as baricitinib or tocilizumab, which may provide a survival benefit to patients who require conventional oxygen, especially those with rapidly increasing oxygen requirements and systemic inflammation (4).
Infliximab, a TNF-alpha inhibitor, and abatacept, a T-cell costimulatory modulator, were also explored as COVID-19 therapeutics with the potential to neutralize destructive cytokine signaling and restore immunoregulatory cascades (9). Infliximab was selected as TNF-alpha is a pro-inflammatory cytokine involved in many acute inflammatory reactions and previously associated with more severe COVID-19-related disease (10). Abatacept, a fusion protein of the extracellular portion of the T-lymphocyte CTLA-4 domain and a fragment of the Fc portion of the human immunoglobin, modulates T-cell costimulatory signaling to restore balance to the immune system (11).
In the ACTIV-1 IM trial, neither infliximab nor abatacept arms met the primary endpoint of time to recovery with an estimated rate-to-recovery ratio. However, patients randomized to infliximab and abatacept had improvement in secondary outcomes, including mortality at day 28 and day 60 and clinical status improvement at day 14 and day 28 (3). Infliximab and abatacept are now included as alternative second immunomodulator options, in addition to remdesivir and dexamethasone, for the management of hospitalized adults requiring conventional oxygen (4).
Therefore, we postulated that receipt of infliximab and/or of abatacept compared to receipt of placebo may decrease resource utilization as measured by hospital days, ICU days, MV or ECMO use, NIV/HFNC use, and O2 use over 28 days of observation. Although days used of each resource are the outcomes most easily understood by clinicians, we elected to use PO modeling with penalization for death as our primary statistical model to avoid attributing less resource use to those who died. Findings with an OR >1.0 indicate a greater number of resource-free days in the infliximab and/or abatacept group compared to the placebo group. After adjusting for covariates, those receiving infliximab compared to placebo and those receiving abatacept compared to placebo had greater odds of having more days alive and hospital-free. After adjustment, those receiving infliximab also had greater odds of having more days alive and free of any form of O2 support and of having more days alive and O2-free compared to placebo.
Why the benefit was more pronounced with infliximab is unknown. However, these findings align well with the primary study report (3), which noted slightly greater survival and clinical improvement with infliximab compared to placebo than abatacept did compared to placebo, although differences were not statistically significant. Infliximab remains a compelling therapeutic target since TNF activates a wide variety of immune cells, and aberrant TNF signaling is a central feature of cytokine release syndrome, which is a component of advanced COVID-19 infection (12).
Compared to earlier in the pandemic, hospital resource utilization for COVID-19 has decreased. The percentage of US patients requiring ICU admission and mean total length of hospital stay decreased from 25.1% and 9.3 days, respectively, in April-June 2020 to 20.6% and 8.4 days, respectively, in December 2020-February 2021 (13). More recent data from December 2021- April 2022 found that 18.9% of US patients required ICU care and that the mean length of hospital stay was 4.9 days (14); however, each hospital day is a notable use of resources.
To date, the studies reporting the effects of immunomodulators on healthcare resources have been mixed in their results. COV-BARRIER trial’s key secondary outcomes revealed that baricitinib did not significantly reduce the number of ventilator-free days, nor did it reduce the duration of hospitalization (15). However, the RECOVERY trial demonstrated that patients who received dexamethasone had a shorter duration of hospitalization (8). Similarly, those allocated to tocilizumab had a shorter median time to discharge (16). Our work suggests that additive treatment with infliximab does make a difference in resource utilization. Continued study of resource allocation, both as it pertains to patient outcomes with regard to treatment and as a limited or finite entity that must be shared, can aid us in preparing for the next epidemic.
This study is limited in its conclusions as we performed a secondary analysis of a clinical trial designed to study time to clinical recovery rather than to directly examine resource utilization. Trial participants are often different from the general population, although the demographics and comorbidities of the ACTIV-1 IM population were reasonably similar to those of hospitalized patients with COVID-19 in observational studies (13, 17). Moreover, 35.1% of ACTIV-1 trial participants were enrolled from international settings where culture for admission and discharge may differ from the practice in the US.
The improvement in survival noted in the ACTIV-1 trial (3) complicates this analysis since improved survival may come at the cost of increased resource utilization. Therefore, we examined the days alive and free of resource use with a PO model penalizing for death. While we reported O2 use, this outcome had limitations in this analysis. Following hospital discharge, data on O2 use was collected at intervals rather than daily, and last data captured was used for those who missed post-discharge follow-up visits.
Finally, ACTIV-1 IM enrolled from October 2020 until December 2021, when the beta and delta SARS-CoV-2 variants were circulating, and vaccination was not entirely routine. Vaccination began in December 2020 in the US and Argentina and slightly later in Brazil (January 2021), Peru (February 2021), and Mexico (April 2021). By mid-December 2021, near the conclusion of enrollment, many citizens throughout all countries had received at least one vaccine dose (US 71.7%, Argentina 84.2%, Brazil 76.8%, Peru 71.1%, Mexico 64.1%) (18). Therefore, our findings may not be reproducible in the current climate of different SARS-CoV-2 variants and widespread immunity. However, patients requiring hospitalization are more likely to be unvaccinated (19), and one large US study in 2021 found that unvaccinated patients accounted for 84.2% of COVID-19 hospitalizations (20).
Conclusions
In this secondary data analysis of the ACTIV-1 IM trial, a single infliximab infusion compared to a placebo was associated with greater odds of being alive and free of all the interventions tested when used in addition to remdesivir and steroids over 28 days of observation. Abatacept use was associated only with greater odds of days alive and free of hospitalization.
Supplementary Material
Take Home Points.
Study question:
Does a single dose of intravenous infliximab or a single dose of intravenous abatacept, when added to the standard care treatment of hospitalized patients with COVID-19 pneumonia, change resource utilization compared to a placebo infusion?
Results:
In comparison to a placebo, the use of infliximab infusion was linked to higher odds of being alive and free of all the interventions tested (hospitalization, ICU care, supplemental oxygen, non-invasive and invasive ventilation, and extracorporeal membrane oxygenation). On the other hand, the use of abatacept was associated only with higher odds of days alive and free of hospitalization and supplemental oxygen.
Interpretation:
In a secondary data analysis of the ACTIV-1 immunomodulator trial, the receipt of infliximab or abatacept, in addition to the standard care treatment for COVID-19 pneumonia, was associated with changes in resource utilization, after taking into account changes in survival.
Impact of Research:
Infliximab and abatacept are alternative second immunomodulator options for the management of hospitalized adults in the NIH COVID-19 Treatment Guidelines. This secondary data analysis of the ACTIV-1 Immunomodulator trial examines whether these agents were associated with changes in resource utilization, including hospitalization, ICU care, supplemental oxygen, non-invasive and invasive ventilation, and extracorporeal membrane oxygenation. Compared to placebo, infliximab infusion was associated with greater odds of being alive and free of all the interventions tested, while abatacept use was associated only with greater odds of days alive and free of hospitalization and supplemental oxygen.
Acknowledgements
All authors contributed to the conception and design of the analysis. Miloni Shah and Derek Cyr performed the statistical analysis. Anne Lachiewicz prepared an early draft of the manuscript. All authors revised the manuscript for important intellectual content and approved the final version for submission.
We thank the members of the ACTIV-1 Study Team (e1-Appendix 1) for their many contributions in conducting the trial, and the participants and their families for their altruism in participating in this trial. We would also like to thank Cynthia Gonzalez, Jessica Springer, and Soju Chang from the National Center for Advancing Translational Sciences, Bethesda, MD; and Emily Honeycutt from University of North Carolina and Megan Roebuck, Judi Willhide, and Erin Campbell from Duke Clinical Research Institute, Durham, NC for their support in completing the manuscript.
Sources of support:
This project has been funded with federal funds from the U.S. Department of Health and Human Services, Office of the Assistant Secretary for Preparedness and Response, Biomedical Advanced Research and Development Authority, under contract number HHSO100201400002I/75A50120F33002. The findings and conclusions in this publication are those of the authors and do not necessarily represent the views of the Department of Health and Human Services or its components. Janssen provided infliximab, Bristol Myers Squibb provided abatacept and AbbVie provided cenicriviroc for use in this trial but did not provide any financial support. Gilead Sciences provided remdesivir for use in this trial but did not provide any financial support. Employees of AbbVie, Bristol Myers Squibb, Janssen and Gilead Sciences participated in discussions about protocol development and in weekly protocol team calls. The final trial protocol was developed by the protocol chair, Dr. William Powderly, the IND sponsor Dr. Daniel K. Benjamin, Jr, and a protocol development committee including representatives from the National Center for Advancing Translational Sciences (NCATS). NCATS had a collaborative role in the trial design, management, and interpretation of the data along with the protocol chair, the study statisticians and the study writing group.
Infrastructure support and resources for research reported in this publication were provided in part by NCATS of the National Institutes of Health under award number(s): UL1TR002345 to Washington University in St. Louis; Duke University – Vanderbilt University Medical Center Trial Innovation Center (U24TR001608) (NCATS, Trial Innovation Network) and ULTR003017- Rutgers University. Dr. Lachiewicz received support from Doris Duke Charitable Foundation grant #2020143.
Author disclosures:
Miloni Shah, Tatyana Der, Derek Cyr, Hussein R. Al-Khalidi, Vivek Iyer, Reynold Panettieri, Adriana M. Rauseo, Martin Maillo, Andreas Schmid, Sugeet Jagpal, and Samuel A. Bozzette have no conflicts of interest to report. Anne Lachiewicz receives research support from Shionogi, Contrafect, Novartis, Veru, Cidara, and Crestone. Christopher Lindsell receives grants and contracts to institution for research from NIH, Novartis, Cytokinetics, and Biomeme. He also sits on DSMBs unrelated to the current work, holds stock options in Bioscape Digital unrelated to the current work, and is incoming Editor-in-Chief of the Journal of Clinical and Translational Science. Akram Khan has received grant funding from Dompe Pharmaceuticals, Roche, Lilly, BOA Pharmaceuticals, and Direct Biologic. William Powderly has been a member of Merck Labs HIV advisory Board unrelated to the current work.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Data sharing statement:
De-identified participant data from the overall ACTIV-1 trial are available from the NIAID Clinical Trials Data Repository.
References
- 1.Centers for Disease Control and Prevention. COVID data tracker: COVID-19 update for the United States. Available at https://covid.cdc.gov/covid-data-tracker/#datatracker-home. Accessed December 4, 2023.
- 2.Dyck B, Unterberg M, Adamzik M, Koos B. The Impact of Pathogens on Sepsis Prevalence and Outcome. Pathogens. 2024;13(1):89. doi: 10.3390/pathogensl3010089 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.O’Halloran JA, Ko ER, Anstrom KJ, Kedar E, McCarthy MW, Panettieri RA Jr., et al. Abatacept, cenicriviroc, or infliximab for treatment of adults hospitalized with COVID-19 pneumonia: A randomized clinical trial. JAMA. 2023;330(4):328–339. doi: 10.1001/jama.2023.11043 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.National Institutes of Health COVID-19 Treatment Guidelines. Therapeutic management of hospitalized patients with COVID-19 pneumonia. Available at https://www.covid19treatmentguidelines.nih.gov/management/clinical-management-of-adults/hospitalized-adults--therapeutic-management/. Updated October 10, 2023. Accessed December 4, 2023.
- 5.LaVange L, Adam SJ, Currier JS, Higgs ES, Reineck LA, Hughes EA, et al. Accelerating COVID-19 Therapeutic Interventions and Vaccines (ACTIV): Designing master protocols for evaluation of candidate COVID-19 therapeutics. Ann Intern Med. 2021;174(9):1293–1300. doi: 10.7326/M21-1269 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Moskowitz A, Shotwell MS, Gibbs KW, et al. Oxygen-Free Days as an Outcome Measure in Clinical Trials of Therapies for COVID-19 and Other Causes of New-Onset Hypoxemia. Chest. 2022;162(4):804–814. doi: 10.1016/j.chest.2022.04.145 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Harrell Frank. Statistical Thinking: Violation of porportional odds is not fatal. Available at https://www.fharrell.com/post/po/. Published September 20, 2020. Accessed April 16, 2024.
- 8.RECOVERY Collaborative Group; Horby P, Lim WS, Emberson JR, Mafham M, Bell JL, et al. Dexamethasone in hospitalized patients with Covid-19. N Engl J Med. 2021;384(8):693–704. doi: 10.1056/NEJMoa2021436 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Feldmann M, Maini RN, Woody JN, Holgate ST, Winter G, Rowland M, et al. Trials of anti-tumour necrosis factor therapy for COVID-19 are urgently needed. Lancet. 2020;395(10234):1407–1409. doi: 10.1016/S0140-6736(20)30858-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Del Valle DM, Kim-Schulze S, Huang HH, Beckmann ND, Nirenberg S, Wang B, et al. An inflammatory cytokine signature predicts COVID-19 severity and survival. Nat Med. 2020;26(10):1636–1643. doi: 10.1038/s41591-020-1051-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Vincenti F, Luggen M. T cell costimulation: a rational target in the therapeutic armamentarium for autoimmune diseases and transplantation. Annu Rev Med. 2007;58:347–358. doi: 10.1146/annurev.med.58.080205.154004 [DOI] [PubMed] [Google Scholar]
- 12.Guo Y, Hu K, Li Y, et al. Targeting TNF-α for COVID-19: Recent advanced and controversies. Front Public Health. 2022;10:833967. doi: 10.3389/fpubh.2022.833967 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Moon RC, Brown H, Rosenthal N. Healthcare resource utilization of patients with COVID-19 visiting US hospitals. Value Health. 2022;25(5):751–760. doi: 10.1016/j.jval.2021.12.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Liu G, Done N, Song Y, et al. 534. Healthcare Resource Utilization and Costs among Patients with COVID-19 in the United States. Open Forum Infect Dis. 2023;10(Suppl 2):ofad500.603. Published 2023 Nov 27. doi: 10.1093/ofid/ofad500.603 [DOI] [Google Scholar]
- 15.Marconi VC, Ramanan AV, de Bono S, et al. Efficacy and safety of baricitinib for the treatment of hospitalised adults with COVID-19 (COV-BARRIER): a randomised, double-blind, parallel-group, placebo-controlled phase 3 trial [published correction appears in Lancet Respir Med. 2021 Oct;9(10):e102], Lancet Respir Med. 2021;9(12):1407–1418. doi: 10.1016/S2213-2600(21)00331-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.RECOVERY Collaborative Group. Tocilizumab in patients admitted to hospital with COVID-19 (RECOVERY): a randomised, controlled, open-label, platform trial. Lancet. 2021;397(10285):1637–1645. doi: 10.1016/S0140-6736(21)00676-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Ko JY, Danielson ML, Town M, et al. Risk Factors for Coronavirus Disease 2019 (COVID-19)-Associated Hospitalization: COVID-19-Associated Hospitalization Surveillance Network and Behavioral Risk Factor Surveillance System. Clin Infect Dis. 2021;72(11):e695–e703. doi: 10.1093/cid/ciaa1419 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Centers for Disease Control and Prevention. COVID data tracker: COVID-19 update for the United States. Available at https://covid.cdc.gov/covid-data-tracker/#global-vaccinations. Accessed March 21, 2024.
- 19.Centers for Disease Control and Prevention. Hospitalizations by vaccination status report. Available at https://www.cdc.gov/coronavirus/2019-ncov/covid-data/covid-net/hospitalizations-by-vaccination-status-report.pdf. Accessed March 21, 2024.
- 20.Tenforde MW, Self WH, Adams K, et al. Association Between mRNA Vaccination and COVID-19 Hospitalization and Disease Severity. JAMA. 2021;326(20):2043–2054. doi: 10.1001/jama.2021.19499 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
De-identified participant data from the overall ACTIV-1 trial are available from the NIAID Clinical Trials Data Repository.