Nirmatrelvir Plus Ritonavir for Early COVID-19 in a Large U.S. Health System: A Population-Based Cohort StudyFREE

We used data from Mass General Brigham, a large, nonprofit, integrated health care system (1.5 million patients annually) that includes 2 large academic hospitals, 7 community hospitals, and a network of ambulatory clinics and community health centers throughout Massachusetts and southern New Hampshire. During the study period, Omicron lineages BA.1.1, BA.2, BA.2.12.1, and BA.5 were predominant in the region (7). Nirmatrelvir plus ritonavir became available for prescription by Mass General Brigham providers for the highest-risk persons on 21 January 2022 and for all patients who were eligible according to the emergency use authorization on 22 February 2022. The study was approved by the Mass General Brigham Human Research Committee institutional review board, and requirement for informed consent was waived.

Mass General Brigham hospitals and clinics use a shared electronic health record (EHR) (Epic Systems). Dates of positive SARS-CoV-2 test results (including both polymerase chain reaction tests and home antigen tests that were documented in the EHR), inpatient admissions at any of the 9 Mass General Brigham hospitals, patient demographic characteristics, comorbidities, concomitant home medications, COVID-19 treatments, COVID-19 vaccination status, and deaths were obtained from an integrated COVID-19 data repository. Data were validated by manual review of medical records by 2 physicians. Similar to all other prescriptions, outpatient orders for nirmatrelvir plus ritonavir were required to be electronically transmitted to pharmacies per regulations in Massachusetts and New Hampshire (verbal and paper prescriptions were permitted only during system failures). Patients who were electronically prescribed courses of nirmatrelvir plus ritonavir were categorized as being in the treatment group.

Persons aged 50 years or older who resided in Massachusetts or New Hampshire and had a new diagnosis of COVID-19 (no positive molecular test result in the preceding 90 days) between 1 January and 17 July 2022 were included. Patient medical conditions, vaccination history, medications used at the time of diagnosis, height and weight, self-reported race and ethnicity, and home address were obtained from the EHR. Recorded medical conditions and age were used to calculate the Monoclonal Antibody Screening Score, a comorbidity index that predicts risk for COVID-19 hospitalization (8, 9), for each patient. Patients were considered immunocompromised if they were receiving immunosuppressive medications (such as prednisone, tumor necrosis factor inhibitors, calcineurin inhibitors, mechanistic target of rapamycin inhibitors, or anti-CD20 antibodies); had active cancer; had received a stem cell or solid organ transplant; or had HIV infection, regardless of CD4 cell count. Neighborhood disadvantage was used as a marker for socioeconomic status for each patient and was determined via geolocation of the home address to the census block group (2010 U.S. Census definitions) and Area Deprivation Index (2020 version) (10, 11). Patients with an estimated glomerular filtration rate less than 30 mL/min and those taking common medications for which coadministration with nirmatrelvir plus ritonavir is not advised (cyclosporine, tacrolimus, everolimus, sirolimus, clopidogrel, rivaroxaban, amiodarone, carbamazepine, phenytoin, and ranolazine) were excluded. Patients with no weight measurement recorded in the 2 years before their COVID-19 diagnosis were also excluded.

The primary objective was to assess the effectiveness of nirmatrelvir plus ritonavir, using a composite end point of reduction in risk for hospitalization within 14 days or death within 28 days after an outpatient COVID-19 diagnosis among persons aged 50 years or older. Secondary objectives were to assess effectiveness at preventing hospitalization within 14 days and, separately, death within 28 days. Although nirmatrelvir plus ritonavir became the preferred therapy at Mass General Brigham in January 2022, limitations in supply, pharmacy locations, patient acceptance, provider awareness, and clinical capacity to prescribe resulted in staged patient access. We compared patients who were prescribed nirmatrelvir plus ritonavir versus those who were not to estimate the effectiveness of the treatment. The study included recorded COVID-19 cases occurring between 1 January and 17 July 2022 and any subsequent hospitalizations (through 31 July 2022) or deaths (through 14 August 2022).

We sought to emulate a hypothetical randomized trial similar to EPIC-HR (Evaluation of Protease Inhibition for Covid-19 in High-Risk Patients) (1), using observational data where nirmatrelvir plus ritonavir was randomly assigned among outpatients with early COVID-19 in the context of high prevalence of prior immunity and the currently circulating variants. To mitigate immortal person-time bias and align with the design of EPIC-HR, only patients who were alive, had not received other authorized treatments for early COVID-19 (molnupiravir, remdesivir, or the anti–SARS-CoV-2 monoclonal antibodies sotrovimab and bebtelovimab), and were not hospitalized through 2 calendar days (day of diagnosis and subsequent day) were included. Eligibility was assessed and study group was assigned at the end of the second calendar day, and hospitalizations and deaths after that time were considered end points. Patients prescribed nirmatrelvir plus ritonavir or another authorized treatment after 2 calendar days from diagnosis were retained in their originally assigned study group. The trial emulation is described further in the Supplement Table.

We anticipated bias in the prescription of nirmatrelvir plus ritonavir related to vaccination, medical history, health care access, and other patient factors. To better achieve exchangeability between patients prescribed and not prescribed nirmatrelvir plus ritonavir, we used an inverse probability–weighted design. Treatment weights were generated using a logistic model that included a priori–determined factors of age (50 to 64, 65 to 79, or ≥80 years), comorbidity score (Monoclonal Antibody Screening Score ≤3, 4 or 5, or ≥6), vaccination status (unvaccinated, partially vaccinated, vaccinated, or vaccinated and ≥1 booster dose), recency of last vaccine dose (<20 weeks or >20 weeks, based on observed decreased protection against severe disease [12]), self-reported race and ethnicity (Hispanic or Latinx, White non-Hispanic and non-Latinx, Black non-Hispanic and non-Latinx, Asian non-Hispanic and non-Latinx, or other or unavailable), study period (January to April or May to July), and neighborhood disadvantage (<75th or >75th percentile of Area Deprivation Index [10, 11] of participant cohort census block groups). The treatment weights were used to generate a pseudo-population (13) to estimate the effect of nirmatrelvir plus ritonavir if prescribing was not biased on the basis of the included factors. Modified Poisson models using robust error variance (14) and accounting for the weighted design (15, 16) were used to estimate relative risk reduction with nirmatrelvir plus ritonavir compared with no treatment in the pseudo-population. Adjusted cumulative incidence curves were generated using inverse probability weights (17, 18).

The funding source had no role in the design or conduct of the study; collection, management, analysis, or interpretation of the data; preparation, review, or approval of the manuscript; or the decision to submit the manuscript for publication.

Between 1 January and 17 July 2022, a total of 117 385 Mass General Brigham patients were diagnosed with COVID-19, of whom 51 842 were aged 50 years or older, resided in the region, and had a weight measurement in the preceding 2 years. Among these patients, 7291 had contraindications for nirmatrelvir plus ritonavir or were first diagnosed or treated with COVID-19 therapy within 1 calendar day of hospitalization or death. The remaining 44 551 patients were diagnosed as outpatients and were eligible for nirmatrelvir plus ritonavir (Figure 1). Cases, prescriptions of nirmatrelvir plus ritonavir, and hospitalizations and deaths throughout the study period are shown in Figure 2.

The baseline characteristics of the analysis population are shown in the Table. Nirmatrelvir plus ritonavir was prescribed to 12 541 patients and was more likely to be prescribed to patients with older age, higher comorbidity scores, full vaccination, and recent receipt of a vaccine dose. Findings from the multivariable logistic model used to calculate treatment weights showed that outpatients reporting Hispanic or Latinx ethnicity (adjusted odds ratio, 0.77 [95% CI, 0.68 to 0.86]) or Black race (adjusted odds ratio, 0.53 [CI, 0.46 to 0.62]) were less likely than patients reporting White race to be prescribed nirmatrelvir plus ritonavir. Residents of neighborhoods in the highest quartile of disadvantage were prescribed nirmatrelvir plus ritonavir at rates similar to those for other patients (adjusted odds ratio, 0.99 [CI, 0.95 to 1.04]).

The primary composite end point of hospitalization within 14 days or death within 28 days of incident SARS-CoV-2 infection occurred in 69 (0.55%) patients who were prescribed nirmatrelvir plus ritonavir and 310 (0.97%) who were not (Figure 3). None of the hospitalizations among recipients of nirmatrelvir plus ritonavir were attributable to the previously described rebound syndrome (19). The adjusted risk ratio for nirmatrelvir plus ritonavir was 0.56 (CI, 0.42 to 0.75). In a sensitivity analysis that excluded 1592 patients who received 1 or more outpatient COVID-19 therapies (1038 received nirmatrelvir plus ritonavir, 447 received monoclonal antibodies, 77 received remdesivir, and 37 received molnupiravir) after study group assignment, the estimated effect of nirmatrelvir plus ritonavir on risk for hospitalization or death was unchanged (adjusted risk ratio, 0.56).

The observed reduction in risk was similar across age groups, comorbidity scores, and body mass index groups. However, nirmatrelvir plus ritonavir was associated with increased protective activity among incompletely vaccinated persons and patients who had received their most recent vaccine dose more than 20 weeks prior (Figure 4). An 81% risk reduction was seen in the subgroup analysis of unvaccinated persons.

In secondary analyses assessing effectiveness at preventing hospitalization and, separately, death, prescription of nirmatrelvir plus ritonavir was associated with reduced risk for both end points. Fewer hospitalizations (adjusted risk ratio, 0.60 [CI, 0.44 to 0.81]) and deaths (adjusted risk ratio, 0.29 [CI, 0.12 to 0.71]) occurred among recipients of nirmatrelvir plus ritonavir.