Nirmatrelvir/Ritonavir Use and Hospitalizations or Death in a Previously Uninfected Nonhospitalized High-Risk Population With COVID-19: A Matched Cohort Study

Abstract

Objective

To determine the association of nirmatrelvir/ritonavir (NMV/r) with hospitalization or death within 30 days as compared with untreated controls previously uninfected and nonhospitalized.

Methods

We used a matched cohort design using inverse probability of treatment weight (IPTW). Individuals prescribed NMV/r within 3 days of COVID-19 diagnosis were compared with IPTW-based untreated controls. Variables for IPTW included age, race, sex, body mass index, geographic location, vaccination status, and multiple comorbidities. Additional analyses were conducted on NMV/r-treated and propensity score–matched untreated controls.

Results

Among 7615 individuals prescribed NMV/r and 62 077 controls identified between 1 January 2022 and 25 February 2023, the risk of hospitalization/death was lower among NMV/r-treated persons vs untreated controls (243 vs 3468 events; absolute risk difference [ARD], −2.36 [95% CI, −2.57 to −2.14]). The difference was significant for those >60 and ≤60 years old (ARD, −3.86 [95% CI, −4.19 to −3.54] vs −0.27 [95% CI, −0.51 to −0.03]) and for persons asymptomatic and symptomatic (ARD, −7.09 [95% CI, −7.62 to −6.55] vs −1.46 [95% CI, −1.66 to −1.25]). Significant benefit was observed among individuals unvaccinated and vaccinated, with or without a booster dose.

Conclusions

NMV/r is associated with a significant reduction in 30-day hospitalization or death among individuals previously uninfected and nonhospitalized.

In a large national cohort, the risk of hospitalization/death was lower among nirmatrelvir/ritonavir-treated persons vs untreated controls. The difference was significant for individuals younger and older, symptomatic and asymptomatic, and vaccinated and unvaccinated.

Coronavirus disease (COVID-19) remains a major global threat with considerable morbidity and mortality [1]. Highly effective vaccines are now available, which induce a strong immune response and are associated with high levels of protection against symptomatic infection, severe/critical disease, and death [2–9]. Vaccination is by far the most important strategy in reducing the burden of disease. However, immunity wanes over time [10], and breakthrough infections among fully vaccinated persons have been reported [11–15]. In December 2021, 2 novel oral antiviral agents—nirmatrelvir/ritonavir (NMV/r) and molnupiravir (MPV)—were granted emergency use authorization by the Food and Drug Administration for treatment of early symptomatic cases of mild to moderate COVID-19 at high risk of progression to severe disease [16–19].

In a phase 2/3 double-blind randomized trial in adults who were symptomatic, unvaccinated, and nonhospitalized and at high risk of disease progression (EPIC-HR), NMV/r treatment was initiated within 3 days of symptom onset and associated with an 89% relative risk reduction for hospitalization or death by day 28 as compared with the placebo group [20]. In an observational study, the use of NMV/r during the Omicron variant–predominant period was associated with a 73% and 79% reduction in hazards of hospitalization and death vs untreated persons, respectively [21]. The benefit observed in this study was limited to patients >65 years old. In this study, 90% of the treated persons and 78% of the untreated had previous immunity from natural infection, vaccination, or both. Other observational studies, often in select subpopulations at risk, have noted a benefit of NMV/r in reducing hospitalization or death [22–26]. We undertook this study to determine the effect of NMV/r treatment upon hospitalization and death in a previously uninfected, nonhospitalized, and high-risk population in the Veterans Affairs (VA) national COVID-19 Shared Data Resource database.

METHODS

Study Setting

The Veterans Health Administration created a national COVID-19 Shared Data Resource, which contains detailed demographic, clinical, laboratory, vital status, and episodes-of-care information on all veterans with a laboratory-confirmed diagnosis of SARS-CoV-2 infection and receipt of a SARS-CoV-2 vaccine within the VA. Veterans who are tested or vaccinated outside VA are captured by patient self-report (presentation of a vaccination card) or through insurance claims data. The VA COVID-19 Shared Data Resource is updated regularly with information derived from multiple validated sources [9, 12, 27–29].

Study Population

We used a matched cohort design for the current study using 2 analytic approaches: inverse probability of treatment weight (IPTW) as the primary analysis and propensity score matching as the secondary analysis. Eligible individuals were those in the VA COVID-19 Shared Data Resource who had at least 2 episodes of care in the VA health care system within the last 2 years, a first confirmed SARS-CoV-2 infection between 1 January 2022 and 25 February 2023, at least 1 risk factor for progression to severe disease, and receipt of NMV/r within 30 days of their COVID-19 diagnosis. Those who were hospitalized or died before receiving NMV/r, those who received NMV/r and MPV, and those who received monoclonal antibody for COVID-19 or remdesivir were excluded, as were those who received treatment ≥3 days after the index diagnosis. We used an IPTW-based approach for our primary analysis, as detailed in our previous publications [30, 31]. Briefly, we fitted a logistic regression model for NMV/r prescription using age (5-year blocks), race, sex, body mass index, VA facility where diagnosis was made, and vaccination status, as well as the presence of diabetes, hypertension, cardiovascular disease, chronic kidney disease, chronic lung disease, and cancer diagnoses. The estimated probabilities from this model were used to compute IPTWs, which were used to weigh in subsequent analyses. To account for potential replications caused by IPTW, we used a robust (sandwich) variance estimator in a Cox regression model, which yielded conservative 95% CIs. Adequacy of weighting was tested by calculating the standardized mean difference for each variable after applying the weights. A value <0.2 indicates good matching for the variable tested.

Vaccination status was categorized by the status at the time of COVID-19 diagnosis into individuals who were unvaccinated or did not complete a primary series, who completed a primary series, and who completed a primary series and received at least 1 booster dose after that. Body mass index was calculated by the average of the 2 most recent height and weight values. Comorbidities were retrieved from the VA national COVID-19 database, where they are identified by ICD-10 codes.

Primary Outcome Measure

Our primary outcome measure was hospitalization or death within 30 days among those prescribed NMV/r vs those not prescribed NMV/r. Time at risk started from the date of index COVID-19 diagnosis.

Statistical Analysis

We calculated the absolute risk difference (ARD) and associated 95% CIs between the groups overall and for substrata of the population by age, sex, body mass index, presence of various comorbidities, vaccination status, and presence of symptoms. Kaplan-Meier curves were generated to demonstrate the difference in outcomes over time among those treated with NMV/r or untreated controls. A log-rank test was used to calculate P values between groups. P < .05 was considered statistically significant.

Additional Analyses

We conducted additional analyses to determine the validity of our primary results. Among the eligible population, we used propensity score matching to identify those prescribed NMV/r and 1:1 matched controls who were not prescribed NMV/r. Propensity score matching was done on age, race, sex, body mass index, multiple comorbidities, site of diagnosis, and vaccination status. We used matching without replacement using a caliper of 0.2 SD. We calculated the ARD and 95% CIs for hospitalization or death overall and for various subgroups. We also generated Kaplan-Meier curves for presentation of our results comparing NMV/r vs untreated controls for hospitalization or death.

RESULTS

Among 121,768 persons with a confirmed first SARS-CoV-2 positive test result between 1 January 2022 and 25 February 2023 in the VA national COVID-19 Shared Data Resource database, we identified 77 674 who met the eligibility criteria. Of those included in the IPTW analysis, 7615 were prescribed NMV/r and 62 077 were not prescribed NMV/r (Figure 1).

Figure 1.

Cohort construction. NMV/r, nirmatrelvir/ritonavir; RT-PCR, reverse transcriptase–polymerase chain reaction; VA, Veterans Affairs.

IPTW Analysis

Baseline characteristics of the cohort before and after matching are presented in Table 1, with the absolute standardized mean differences before and after application of IPTWs shown in Supplementary Figure 1. In the IPTW group, the median age was 63.2 years, 85% was male, and 24% was Black. The median body mass index was 30 kg/m², and the median Charlson Comorbidity Index was 1 (IQR, 0–3). A quarter (25%) was unvaccinated or had not completed a primary SARS-CoV-2 vaccination series, and almost half (49%) had completed a primary series and received a booster dose of the vaccine. The median number of days from diagnosis to first prescription of NMV/r was 0 (IQR, 0–1), and 42.4% of the treated group vs 25.9% of the untreated group was symptomatic at presentation.

Table 1.

Baseline Characteristics of the NMV/r Analysis Cohort

	Before IPTW		IPTW
	NMV/r (n = 7615)	Control (n = 62 077)	NMV/r (n = 7615)	Control (n = 62 077)	SMD
Age, y	66.4 (56.3–74.5)	62.8 (50.7–73.1)	63.2 (52.2–73)	63.2 (51.2–73.3)	0.01
Male sex	85.2	84.5	84.9	84.58	0.01
Race					0.03
White	64.78	64.08	64.65	64.17
Black	25.9	23.79	24.23	24.02
Other/unknown	9.32	12.12	11.12	11.81
BMI, kg/m2	30.1 (26.5–34.3)	30 (26.5–34.2)	30.1 (26.5–34.3)	30 (26.5–34.2)	0
Charlson Comorbidity Index	2 (0–3)	1 (0–3)	1 (0–3)	1 (0–3)	−0.01
Comorbidities
Obesity (BMI >30 kg/m2)	50.86	49.62	50.98	49.67	0.03
Diabetes	39.41	34.23	35.27	34.8	0.01
Hypertension	70.14	64.13	65.58	64.8	0.02
Cardiovascular disease	41.92	40.34	41.12	40.52	0.01
Chronic kidney disease	13.17	15.7	15.58	15.42	0
Chronic lung disease	39.63	37.06	38.19	37.35	0.02
Cancer diagnosis	22.1	19.02	19.57	19.36	0.01
Vaccination status at baseline					0
Unvaccinated or primary series incomplete	15.27	26.36	25.04	25.15
Primary series complete	19.13	26.95	25.9	26.09
Primary series + booster	65.59	46.69	49.06	48.76
Geographic location					0.06
City/town	4.37	5.39	4.57	5.36
Small town/rural	3.13	3.72	3.14	3.71
Urban	74.52	70.98	74.47	71.01
Unknown	17.98	19.91	17.82	19.92
Days from
Symptoms to prescription	0 (0–1)	…	0 (0–1)	…
Diagnosis to prescription	0 (0–1)	…	0 (0–1)	…

Data are presented as median (IQR) or percentage.

Abbreviations: BMI, body mass index; IPTW, inverse probability of treatment weight; NMV/r, nirmatrelvir/ritonavir; SMD, standardized mean difference.

The incidence of hospitalization or death within 30 days was lower among those who were prescribed NMV/r overall (243 vs 3468 events; ARD, −2.36; 95% CI, −2.57 to −2.14). Benefit was observed among those aged >60 years (ARD, −3.86; 95% CI, −4.19 to −3.54) as well as those <50 years, though the magnitude of ARD was smaller (ARD, −0.27; 95% CI, −.51 to −.03). Benefit was observed among the unvaccinated (ARD, −2.5; 95% CI, −2.92 to −2.07), those who completed a primary series without a booster dose (ARD, −1.83; 95% CI, −2.25 to −1.42), and those who completed a primary series and received a booster dose (ARD, −2.56; 95% CI, −2.87 to −2.25). Individuals who were symptomatic experienced a larger benefit (ARD, −7.09; 95% CI, −7.62 to −6.55), though those who were asymptomatic also experienced a benefit (ARD, −1.46; 95% CI, −1.66 to −1.25; Figure 2A).

Figure 2.

Absolute risk difference in the incidence of hospitalization or death within 30 days among individuals who did and did not receive nirmatrelvir/ritonavir. A, Inverse probability of treatment weight analysis. B, Propensity score–matched analysis. BMI, body mass index; COPD, chronic obstructive pulmonary disease; NMV/r, nirmatrelvir/ritonavir.

Kaplan-Meier curves showed a significant reduction in the proportion of persons who were hospitalized or died among those treated with NMV/r as compared with untreated controls (log-rank P < .001; Figure 3A).

Figure 3.

Kaplan-Meier curves depicting the proportion of individuals without hospitalization or death among those treated with and without nirmatrelvir/ritonavir. A, Inverse probability of treatment–weighted analysis. B, Propensity score–matched analysis. NMV/r, nirmatrelvir/ritonavir.

Propensity Score–Matched Analysis

Baseline characteristics of the cohort after propensity score matching are presented in Supplementary Table 1, with the absolute standardized mean differences before and after application of IPTWs shown in Supplementary Figure 2. After matching, the median age was 66.4 years in the NMV/r group and 66.8 years in the control group (age was matched on 5-year blocks), 85% were male, and 26% were Black. The median Charlson Comorbidity Index score was 2 (IQR, 0–3); furthermore, 15% were unvaccinated or had not completed a primary SARS-CoV-2 vaccination series, whereas 66% had completed a primary series and received a booster dose of the vaccine. The median number of days from diagnosis to first prescription of NMV/r was 0 (IQR, 0–1), and 42.4% of the treated group vs 25.5% of the untreated group was symptomatic at presentation. The change in ARD for hospitalization or death within 30 days was similar to the IPTW analysis, except that no benefit was observed among those ≤60 years old (Figure 2B). Kaplan-Meier curves showed a significant similar reduction in the proportion of persons who were hospitalized or had died among those treated with NMV/r as compared with untreated controls (log-rank P < .001; Figure 3B).

Additional Analyses

We calculated the hazard ratios associated with the use of NMV/r (vs untreated individuals) in the IPTW and propensity score–matched analyses using Cox regression analysis. The results are shown in Supplementary Figure 3 for the IPTW groups and Supplementary Figure 4 for the propensity score–matched groups. These results confirm our primary analyses and provide additional assurance regarding the validity of our results.

DISCUSSION

Emerging data suggest the beneficial role of NMV/r in reducing hospitalization and death in early/mild symptomatic COVID-19 in persons at high risk of progression to more severe disease. Our data provide additional evidence of the benefit of NMV/r in a population previously uninfected, at high risk, and nonhospitalized and clarify its role in various demographic and clinical subgroups.

In our well-matched study population, a reduction in hospitalization or death within 30 days of COVID-19 diagnosis was observed in persons treated with NMV/r overall. In a previous study in the Omicron variant–predominant era, this benefit was limited to the older population, with no significant benefit observed among those who were <65 years old [21]. We observed a benefit among older and younger populations, though the magnitude of benefit was more pronounced in the older population. The reason for a lack of benefit or attenuated benefit in the younger population is not clear. It is plausible that the low rate of severe disease with the more recent SARS-CoV-2 variants [29, 32, 33] is even lower among the younger population due to better general health and a reduced burden of comorbidities. With such low event rates in the younger population, the studies may lack power to detect a significant difference between groups. Our study population was older and with a higher burden of comorbidities than the general US population. A benefit in the younger population, though smaller than what was observed in the older population, suggests reexamining the age-based criteria for the use of NMV/r in individuals with COVID-19.

Another important finding from our study is that the benefit of NMV/r was observed among the unvaccinated, as well as those who had completed a primary series with or without a booster dose. This has management and policy implications since a sizable proportion of the world's population remains unvaccinated. Our study demonstrates that this unvaccinated population can still benefit from NMV/r use in the appropriate setting, thus lowering the burden of hospitalization, mortality, and costs associated with more severe disease. This finding also underscores the need to identify infection at an early stage to avail the full benefit of NMV/r treatment. Whether prior natural immunity with or without additional vaccine-induced immunity has an effect on these outcomes is unknown.

Individuals both symptomatic and asymptomatic experienced a benefit from NMV/r in our study, though the magnitude of benefit was more pronounced among the symptomatic. It is possible that the asymptomatic persons had specific reasons for experiencing milder outcomes (demographic, clinical, or genetic), had a lower viral load, or had an enhanced immune response, which led to better outcomes independent of any benefit from NMV/r. It is also possible that individuals who were symptomatic were identified earlier in the course of the disease, at which time NMV/r may have been most beneficial. The benefit observed in the asymptomatic group in our study is in contrast to the current emergency use authorization for its use in persons with mild to moderate disease who are at risk of progression to more severe disease [34]. More studies are needed to confirm this finding and to determine if the current authorization criteria need to be modified.

Persons with certain comorbidities are at higher risk of COVID-19 disease progression and complications: diabetes mellitus, cardiovascular disease, chronic lung and kidney disease, among others. Patients in our study with any of these comorbidities experienced a benefit from NMV/r treatment. While this suggests that some of those at a higher risk of disease progression based on the presence of these comorbidities are more likely to benefit from NMV/r given early in the course of the disease, this finding should be interpreted with caution. Diagnoses based on ICD-10 codes may not account for the severity or spectrum of these diseases, and the actual risk of COVID-19 disease progression and NMV/r benefit may be quite different according to the severity of these comorbidities.

Only 15% of our study population was female. While the absolute number of women in each group was substantial (1127 in the NMV/r group and 9621 in the control group in the IPTW analysis), caution should be observed in generalizing these results to the entire US population. When stratified by sex, men and women experienced benefit from NMV/r. In the IPTW analysis, men had a numerically superior benefit, while in the propensity score–matched analysis, the benefit was relatively similar. Whether this observation reflects differences in prescription patterns for men vs women requires further study.

While IPTW and propensity score matching mitigate some of the biases inherent in observational studies, they are not exact substitutions of randomized clinical trials. Residual confounding is always a limitation of observational studies. Other limitations of our study include a predominantly male population that was exclusively veterans, which may not be generalizable to the overall US population. We also did not evaluate the association of supplemental oxygen use at baseline and other treatment modalities for COVID-19 (eg, steroids, monoclonal antibodies, and remdesivir). Likewise, we did not examine the role of specific variants or sublineages. Our study was designed to determine the effect of NMV/r in persons previously uninfected. However, it is quite possible that a substantial proportion may have had undetected infection in the past. How this affects the effectiveness of oral antivirals such as NMV/r is unclear and requires further study. We did not have information on reasons to initiate treatment at an individual level. It is conceivable that such decisions were not uniform and may not have always conformed to the approved emergency use authorization guidelines. An IPTW approach was used to mitigate this bias. Another limitation is the use of ICD-10 codes for determining the presence of comorbidities, which may not account for the severity of the underlying comorbidities.

In conclusion, NMV/r use is associated with a reduction in hospitalization or death in a population previously uninfected and nonhospitalized with COVID-19 that is at a high risk of progression to severe disease. The benefit is evident in individuals older and younger, unvaccinated and vaccinated, and symptomatic and asymptomatic. These results suggest revisiting the treatment recommendations for the use of NMV/r in these populations.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Author contributions. Study concept and study design: A. A. B. Drafting of the manuscript: A. A. B. Data acquisition: P. Y. Data analysis: P. Y., A. A. B. Critical revision of the manuscript for important intellectual content: A. A. B., V. B. T., P. Y., O. S. S., S. B. O., F. B. M. Final approval of the article: A. A. B., V. B. T., P. Y., O. S. S., S. B. O., F. B. M.

Data access. A. A. B. and P. Y. had complete access to the data at all times and accept responsibility for the integrity of this article.

Data availability statement. This study used data created and maintained by the Veterans Health Administration, Department of Veterans Affairs. These data are freely available to approved individuals upon fulfilling the specified requirements.

Ethical review. The study was approved by the Institutional Review Board at the VA Pittsburgh Healthcare System. A waiver of informed consent was granted for the study.

Disclaimer. The content of this article is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The funding source had no role in the analysis, interpretation, or conclusions of this study or the decision to publish. The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of the Department of Veterans Affairs or the funding agencies.

Financial support. This work was supported by the National Institute of Allergy and Infectious Diseases at the National Institutes of Health (R21AI174041 to A. A. B. as principal investigator); by the National Institutes of Health (K23GM132688 to F. B. M.); by data created in the VA COVID-19 Shared Data Resource and with resources and facilities of the Department of Veterans Affairs Informatics and Computing Infrastructure (VA HSR RES 13-457); by resources from and facilities at the VA Pittsburgh Healthcare System and Veterans Health Foundation of Pittsburgh; and by the central data repositories maintained by the VA Information Resource Center, including the Corporate Data Warehouse.