Effectiveness of COVID-19 treatment with nirmatrelvir-ritonavir or molnupiravir among U.S. Veterans: target trial emulation studies with one-month and six-month outcomes

Background: Information about the effectiveness of oral antivirals in preventing short- and long-term COVID-19-related outcomes during the Omicron surge is limited. We sought to determine the effectiveness of nirmatrelvir-ritonavir and molnupiravir for the outpatient treatment of COVID-19. Methods: We conducted three retrospective target trial emulation studies comparing matched patient cohorts who received nirmatrelvir-ritonavir versus no treatment, molnupiravir versus no treatment, and nirmatrelvir-ritonavir versus molnupiravir in the Veterans Health Administration (VHA). Participants were Veterans in VHA care at risk for severe COVID-19 who tested positive for SARS-CoV-2 in the outpatient setting during January and February 2022. Primary outcomes included all-cause 30-day hospitalization or death and 31–180-day incidence of acute or long-term care admission, death, or post-COVID-19 conditions. For 30-day outcomes, we calculated unadjusted risk rates, risk differences, and risk ratios. For 31–180-day outcomes, we used unadjusted time-to-event analyses. Results: Participants were 90% male with median age 67 years and 26% unvaccinated. Compared to matched untreated controls, nirmatrelvir-ritonavir-treated participants (N=1,587) had a lower 30-day risk of hospitalization (27.10/1000 versus 41.06/1000, risk difference [RD] −13.97, 95% CI −23.85 to −4.09) and death (3.15/1000 versus 14.86/1000, RD −11.71, 95% CI 16.07 to −7.35). Among persons who were alive at day 31, further significant reductions in 31–180-day incidence of hospitalization (sub-hazard ratio 1.07, 95% CI 0.83 to 1.37) or death (hazard ratio 0.61, 95% CI 0.35 to 1.08) were not observed. Molnupiravir-treated participants aged ≥65 years (n=543) had a lower combined 30-day risk of hospitalization or death (55.25/1000 versus 82.35/1000, RD −27.10, 95% CI −50.63 to −3.58). A statistically significant difference in 30-day or 31–180-day risk of hospitalization or death was not observed between matched nirmatrelvir- or molnupiravir-treated participants. Incidence of most post-COVID conditions was similar across comparison groups. Conclusions: Nirmatrelvir-ritonavir was highly effective in preventing 30-day hospitalization and death. Short-term benefit from molnupiravir was observed in older groups. Significant reductions in adverse outcomes from 31–180 days were not observed with either antiviral.


Target Trials Specification Target Trials Emulation Eligibility Criteria for All Three Trials
Aged ≥18 years at the time of positive SARS-CoV-2 test performed January 1, 2022-February 28, 2022 Same VA enrollees (excludes VA employees who are not enrollees) Same Followed by a primary care provider in the VA healthcare system, defined as having a primary care outpatient encounter in the preceding 18 months Same Documented first positive laboratory-based SARS-CoV-2 NAAT or antigen test in a respiratory specimen performed between January 1-February 28, 2022. Patients with reinfection during this period (i.e., those who also have a documented positive SARS-CoV-2 NAAT or antigen test before January 1, 2022) were not included.

Same
We identified all VA enrollees who had a first positive laboratory-based SARS-CoV-2 NAAT or antigen test using the VA COVID-19 National Surveillance Tool (NST) as documented in CSDR. This includes all patients tested within the VA as well as patients with such tests performed outside the VA but documented in VA records. Tests performed outside the VA are identified by methods including NLP and confirmed by manual review of the EHR before being documented in the NST. Symptomatic infection (≥1 symptoms) at the time of diagnosis (1, 2) Not included.
CSDR captures 15 pre-specified COVID-19-related symptoms documented in the EHR in the 30 days prior to the first positive SARS-CoV-2 test. Because ascertainment of symptoms is incomplete and since CSDR does not distinguish whether symptoms were present at the time of testing positive, we did not require presence of ≥1 symptoms. Presence or absence of symptoms were included in propensity score matching instead.
Not hospitalized on or before the date of the positive SARS-CoV-2 test (test-positive date) and not hospitalized on or before the date of randomized assignment to nirmatrelvirritonavir, molnupiravir, or no treatment.

Same
Participants were not hospitalized on or within 7 days before test-positive date.
Participants receiving nirmatrelvir-ritonavir or molnupiravir were not hospitalized on or before the date of treatment initiation.
Matched untreated participants were not hospitalized on or before their assigned index date (which was the same number of days from the test positive date as the matched treated patient, see below) Did not initiate therapy before the test-positive date Same Persons treated with nirmatrelvir-ritonavir or molnupiravir did not receive more than one outpatient COVID-19 therapy within 7 days. Untreated persons did not receive any other outpatient COVID-19 therapy on or prior to the matched antiviral treatment date.

Same
Persons receiving nirmatrelvir-ritonavir or molnupiravir were not treated with other outpatient COVID-19 treatments (nirmatrelvir-ritonavir, molnupiravir, sotrovimab, remdesivir) on or prior to the antiviral treatment date. Untreated persons did not receive any other outpatient COVID-19 therapy on or prior to the matched antiviral treatment date. Having ≥1 risk factors for progression to severe COVID-19 by the FDA EUA/CDC criteria (Supplemental Table 3)

Additional Eligibility Criteria for Participants in Trials Including a Nirmatrelvir-ritonavir
Arm Persons with any of the following were excluded: • Moderate or severe liver disease • Advanced renal impairment (CKD IV or V, on dialysis, or eGFR<30) • Prescription contraindicated medications per FDA in the 90 days prior to test-positive date (Supplemental Table 4)

Additional Eligibility Criteria for Participants in Trials Including a Molnupiravir Arm Pregnant persons were excluded Same
Treatment Strategies Randomized to treatment with one of the following within 5 days of symptom onset: • Nirmatrelvir-ritonavir We determined the date of treatment with nirmatrelvir-ritonavir or molnupiravir from pharmacy fill date of the medication.
Since symptom onset date could not be ascertained in the EHR, we used treatment within 10 days of the test-positive date rather within 5 days of symptom onset. In clinical practice, adherence to EUA criteria (treatment within 5 days of symptom onset) is closely monitored and enforced by PBM.

Treatment Assignment
In each of the three RCTs eligible participants are randomly assigned to each of the two treatment options: • Propensity-score matching After exact-matching by these four factors, we performed an additional propensity score matching step ultimately aiming to identify the best-matching comparator. We used matching with replacement in a 1:k variable ratio, where k varied based on the number of propensity score ties. We included all ties to avoid imbalance due to random pruning. The characteristics included in the propensity score logistic regression model are shown in Supplemental • Cardiac, pulmonary, renal, GI, neurologic, cancer, mental health, musculoskeletal, endocrine • All-cause mortality

Follow-up
For each person, follow-up started on the day of randomization for each of the three target trials and continued until day 30 after treatment for short-term outcomes or from day 31-180 for post-COVID-19 condition outcomes.

Same
For comparisons with untreated patients, an index date was assigned to untreated patients which was the same interval from the test-positive date. Untreated persons had to fulfill enrollment criteria as of that index date (e.g., a person treated on day 3 after testing positive was matched to a person who was untreated, alive, and not hospitalized 3 days after testing positive). This study design ensured that dates of eligibility determination, treatment, and follow-up initiation were the same (Supplemental Figure 1).

Causal Contrasts
Intention-to-treat (ITT) effect Observational analogue of the per-protocol effect Per-protocol effect (i.e., the effect if all individuals had received their assigned treatment) Observational analogue of the per-protocol effect Statistical Analysis 30-day risk, risk difference and risk ratios for the primary outcome. 31-180-day incidence of any acute or longterm care admission or all-cause death 31-180-day incidence of post-COVID conditions Same * Includes treatments available in the VA during the study period (nirmatrelvir-ritonavir, molnupiravir, sotrovimab, and outpatient remdesivir).

Departures from the protocol for the comparison of nirmatrelvir-ritonavir versus molnupiravir
Due to insufficient matching for the comparison of nirmatrelvir-ritonavir with molnupiravir, we expanded the calendar time match window from +/-7 days to +/-10 days around the testpositive date of the matched comparator in the exact-matching step. For propensity-score matching, we used VA region rather than VISN(3) and did not include chronic kidney disease since differences in nirmatrelvir-ritonavir prescribing practices for persons with eGFR 30-59 ml/min (by either dose-adjusting nirmatrelvir-ritonavir or considering an alternative agent) were anticipated to result in difficulty matching on this variable.

Eligibility criteria and study population
Participants in all three emulated trials were limited to test-positive VA enrollees. We excluded persons who received nirmatrelvir-ritonavir or molnupiravir before or after 10 days following the test-positive date, determined to be a reasonable proxy treatment window relative to the 5-day window from symptom onset per FDA EUA. While it is difficult to accurately ascertain symptom onset from the EHR, VA Pharmacy Benefits Management Services (PBM) monitors and enforces adherence to EUA. We also limited eligibility to Veterans with at least one risk factor for progression to severe COVID-19, including hospitalization or death, according to the Centers for Disease Control and Prevention (CDC)(4) as shown in Supplemental Table 2 below.

Exclusions for nirmatrelvir-ritonavir comparisons
For comparisons involving nirmatrelvir-ritonavir, we excluded persons with advanced renal impairment, defined as any of following: • Estimated glomerular filtration rate (eGFR) less than 30 milliliters per minute based on labs obtained in the six months prior to positive SARS-

Cohort matching
We used two matching steps, exact and propensity score matching to achieve balance of covariates between comparator groups and reduce confounding. NIH tiers of prioritization used for exact-matching are shown in the

COVID-19 Vaccination Status
We aggregated all administered vaccine doses documented in VA-CDW, CMS-Medicare and VA Community Care data. Vaccine records with service dates prior to December 11, 2020, the earliest date of EUA for COVID-19 vaccination in the United States, were excluded. To ensure that vaccine doses documented in more than one source were not counted more than once, after combining records from all sources, we treated 2 vaccine doses as duplicates if they were documented within 7 days of each other. We included Moderna, Pfizer-BioNTech, and Janssen vaccine types, which were approved in the United States during the period of study and comprised most of all vaccine types. To allow for complete categorization of vaccination, we also included Novavax (authorized after the end of this study period). Vaccine doses of unknown or other type (e.g., Oxford-AstraZeneca) were categorized as other.
Non-immunocompromised (6) 1. Veterans were considered unvaccinated if they did not receive any COVID-19 vaccine or received a vaccine dose other than Janssen less than 14 days prior to the first positive SARS-CoV-2 test (test date). 2. Partial vaccination was indicated by receipt of a single mRNA dose (Pfizer-BioNTech or Moderna) or a single Novavax dose alone or in combination with another vaccine other than Janssen <14 days prior to the index date or a Janssen (Johnson & Johnson) dose <14 days before the test date. 3. Primary vaccination was indicated by receipt of 2 doses of any mRNA or Novavax vaccine or a single dose of Janssen ≥14 days before the test date. 4. Booster vaccination was indicated by any primary regimen above, followed by an additional dose(s) of mRNA, Janssen, or Novavax vaccine ≥7 days before the test date. 5. Other was indicated by any vaccination not captured above.
Immunocompromised (7) Veterans were considered immunocompromised if they had recently received immunosuppressive or cancer medications as described in Supplementary Table 3).
1. Veterans were considered unvaccinated if they did not receive any COVID-19 vaccine or received a vaccine dose other than Janssen less than 14 days prior to the test date. 2. Partial vaccination was indicated by receipt of 2 doses of an mRNA vaccine, a single dose of Janssen, or a single dose of Novavax. It was also indicated by receipt of 3 doses of an mRNA vaccine, a single dose of Janssen followed by a single dose of an mRNA vaccine, or 2 doses of Novavax <7 days before the test date. 3. Primary vaccination was indicated by receipt of 3 doses of an mRNA vaccine, a single dose of Janssen followed by a single dose of an mRNA vaccine, or 2 doses of Novavax ≥7 days before the test date. 4. Booster vaccination was indicated by any primary regimen above, followed by an additional dose(s) of mRNA, Janssen, or Novavax vaccine ≥7 days before the test date. 5. Other was indicated by any vaccination not captured above.
The  Table 7) Chronic kidney disease ‡ Diabetes Immunosuppressive and cancer medications (Supplemental Table 3) Cardiovascular disease Chronic lung disease Dementia Care Assessment Needs (CAN) score Healthcare utilization: number of clinical encounters * Unless otherwise specified, factors were ascertained as of the index date. Urban/rural residence and ADI were determined using ZIP codes for Veteran's most recent place of residence in the one year prior to index date. Substance use and underlying conditions were ascertained based on documentation in the two years prior to index date. Number of outpatient visits of any type were counted in the one year prior to index date. † Any of 15 pre-specified COVID-19-related symptoms present on the day of positive SARS-CoV-2 test or within the preceding 30 days. ‡ Not included in molnupiravir vs nirmaltrevir trial. Table 7. List of underlying conditions totaled for inclusion in propensity score matching models for COVID-19 treatment * Any immunosuppressive or cancer medications (Supplemental Table 3 Table 2 are organized into more discrete groupings. Age, tobacco, and substance use are included separately in propensity score models.

Additional exclusions applied during matching
We applied additional exclusions during matching. Due to lack of overlap, persons aged >115 years or who were categorized in the other vaccination category were excluded. Untreated persons who died or were hospitalized on or before their assigned index date were also excluded as were untreated persons who received other outpatient COVID-19 treatments (nirmatrelvir-ritonavir, molnupiravir, sotrovimab, remdesivir) on or prior to the antiviral treatment date.

Sample size and power calculations
For comparisons of nirmatrelvir-ritonavir or molnupiravir versus no treatment (trials 1 and 2), given a 1:4 match and assuming a 3% incidence of 30-day hospitalization or death among untreated groups, a sample size of 2,650 persons (530 in treatment, 2,120 in no treatment group) was determined to have 80% power to detect a 2% difference in 30-day hospitalization or death (3% untreated, 1% treated).
For comparisons of nirmatrelvir-ritonavir versus molnupiravir (trial 3), given a 1:1 match and assuming a 3% incidence of 30-day hospitalization or death, a sample size of 826 persons in each group was determined to have 80% power to detect a 2% difference in 30-day hospitalization or death (3% molnupiravir, 1% nirmatrelvir-ritonavir). Calculations are based on a two-way Fisher's exact test for equality of two independent proportions with a significance level of 0.05.        A. Absolute standardized mean differences and variance ratios of baseline characteristics between nirmatrelvir-ritonavir treatment versus no treatment shown for the raw and matched data.

Supplemental
Prior to matching, the absolute standardized difference in baseline characteristics between the nirmatrelvir-ritonavir treatment versus no treatment groups ranged from 0.003-0.435 with a median of 0.078 (IQR: 0.031-0.147). After matching, the absolute standardized differences ranged from 0.001-0.062 with a median of 0.012 (IQR: 0.005-0.025).
B. Cumulative distribution of propensity score between nirmatrelvir-ritonavir treatment versus no treatment shown for the raw and matched data demonstrate balance after matching. Figure 2. Distribution of baseline characteristics of persons who were treated with molnupiravir and their comparators with no COVID-19 antiviral or monoclonal antibody treatment, before and after matching.

Supplemental
A. Absolute standardized mean differences and variance ratios of baseline characteristics between molnupiravir treatment versus no treatment shown for the raw and matched data.

Supplemental Figure 3. Distribution of baseline characteristics of persons who were treated with molnupiravir and their comparators treated with nirmatrelvir-ritonavir, before and after matching.
A. Absolute standardized mean differences and variance ratios of baseline characteristics between molnupiravir treatment versus nirmatrelvir-ritonavir treatment shown for the raw and matched data.
Prior to matching, the absolute standardized difference in baseline characteristics between the molnupiravir treatment versus nirmatrelvir-ritonavir treatment groups ranged from 0.001-0.145 with a median of 0.033 (IQR: 0.023-0.079). After matching, the absolute standardized differences ranged from 0.002-0.105 with a median of 0.023 (IQR: 0.011-0.045).
b. Cumulative distribution of propensity score between molnupiravir treatment versus nirmatrelvirritonavir treatment shown for the raw and matched data demonstrate balance after matching. * Sub-hazard ratios (SHR), derived from proportional hazards regression accounting for the competing risk of death, are presented for acute or long-term care admissions and post-COVID outcomes. Hazard ratios are presented for death and acute or long-term care admission or death.