Association of nirmatrelvir for acute SARS‐CoV‐2 infection with subsequent Long COVID symptoms in an observational cohort study

Oral nirmatrelvir/ritonavir is approved as treatment for acute COVID‐19, but the effect of treatment during acute infection on risk of Long COVID is unknown. We hypothesized that nirmatrelvir treatment during acute SARS‐CoV‐2 infection reduces risk of developing Long COVID and rebound after treatment is associated with Long COVID. We conducted an observational cohort study within the Covid Citizen Science (CCS) study, an online cohort study with over 100 000 participants. We included vaccinated, nonhospitalized, nonpregnant individuals who reported their first SARS‐CoV‐2 positive test March–August 2022. Oral nirmatrelvir/ritonavir treatment was ascertained during acute SARS‐CoV‐2 infection. Patient‐reported Long COVID symptoms, symptom rebound and test‐positivity rebound were asked on subsequent surveys at least 3 months after SARS‐CoV‐2 infection. A total of 4684 individuals met the eligibility criteria, of whom 988 (21.1%) were treated and 3696 (78.9%) were untreated; 353/988 (35.7%) treated and 1258/3696 (34.0%) untreated responded to the Long COVID survey (n = 1611). Among 1611 participants, median age was 55 years and 66% were female. At 5.4 ± 1.3 months after infection, nirmatrelvir treatment was not associated with subsequent Long COVID symptoms (odds ratio [OR]: 1.15; 95% confidence interval [CI]: 0.80–1.64; p = 0.45). Among 666 treated who answered rebound questions, rebound symptoms or test positivity were not associated with Long COVID symptoms (OR: 1.34; 95% CI: 0.74–2.41; p = 0.33). Within this cohort of vaccinated, nonhospitalized individuals, oral nirmatrelvir treatment during acute SARS‐CoV‐2 infection and rebound after nirmatrelvir treatment were not associated with Long COVID symptoms more than 90 days after infection.


| INTRODUCTION
Symptoms after SARS-CoV-2 may persist as Long COVID, a type of postacute sequalae of SARS-CoV-2 infection (PASC) defined as unexplained symptoms attributed to COVID-19. 1 Vaccination reduces but does not eliminate risk of Long COVID. 2 Higher acute viral load or prolonged shedding may be associated with increased risk. 3,47][8][9][10] Among high-risk, unvaccinated, nonhospitalized individuals with symptomatic COVID-19, nirmatrelvir, a novel orally-administered SARS-CoV-2 main protease (M pro ) inhibitor in combination with ritonavir reduces viral load and progression to severe disease. 113][14] However, whether treatment with nirmatrelvir during acute infection reduces post-COVID conditions is uncertain, with two studies finding conflicting results in the same population. 15,16Use of EHR-based diagnosis of post-COVID conditions, as in those studies, relies on patient reporting and clinician documentation of medical diagnoses so may not capture the possible treatment effect on Long COVID symptoms.A third study, which surveyed individuals 4 months after SARS-CoV-2 infection, found no association between nirmatrelvir use and prevalent Long COVID symptoms, but reported a prevalence of symptoms of nearly 50% in both groups. 17erefore, the objectives were to test the hypothesis that nirmatrelvir treatment during acute SARS-CoV-2 infection is associated with a lower prevalence of patient-reported Long COVID symptoms >90 days after infection, and that rebound of acute symptoms or rebound test-positivity after treatment is associated with higher prevalence of Long COVID symptoms.

| Design, setting, and participants
The COVID-19 Citizen Science (CCS) Study is an online cohort study hosted on the Eureka Research Platform. 18Recruitment occurred through email invitations to existing Eureka participants, referrals from partner organizations, and word of mouth.To participate, individuals must register for an account, agree to participate in English and provide electronic consent.After consenting, participants complete baseline and follow-up surveys.We used data collected from March 26, 2020, to December 22, 2022.We included individuals who enrolled before first known SARS-CoV-2 infection first infected when nirmatrelvir was available in March 2022 through August 2022.Outcomes were ascertained only for those who responded to surveys about Long COVID symptoms in November and December 2022.The primary outcome was defined as at least one patient-reported Long COVID symptom >90 days after acute SARS-CoV-2 infection. 1ollowing the Emergency Use Authorization, we excluded hospitalized and pregnant individuals.Because >98% were vaccinated before infection we limited analysis to vaccinated individuals.We chose not to limit inclusion to "eligible" patients as in our clinical experience the "high risk" criteria are loosely interpreted but conducted sensitivity analyses limiting inclusion criteria to those defined as at high-risk by the Centers for Disease Control (CDC).

| Exposures
The primary exposure was self-reported nirmatrelvir treatment within 30 days of first reported positive SARS-CoV-2 antigen or PCR test on a weekly survey from March 2022 to August 2022.
Participants who reported taking nirmatrelvir were subsequently invited to respond to a survey in December 2022 to assess for rebound symptoms or test positivity.Rebound symptoms were defined as symptom worsening after initial improvement.Rebound test positivity was defined as testing negative and then positive on antigen test after completing treatment.

| Outcomes
The primary outcome was ≥1 self-reported prevalent Long COVID symptom >90 days after first SARS-CoV-2 positive test on a crosssectional survey distributed in November and December 2022 that asked about presence, duration, and severity of Long COVID symptoms using a published nonvalidated instrument. 19We decided a priori to use the 90-day World Health Organization definition as a significant proportion experience symptom resolution in the 4-week to 90-day period. 19Symptoms queried included fatigue, shortness of breath, confusion, headache, altered taste and smell, joint pain, muscle aches, cough, chest pain, scratchy throat, nausea, vomiting, diarrhea, fever, chills, red or painful eyes, sore throat, and other.
Because the survey was developed initially based on symptoms during acute COVID-19 before thorough understanding of common Long COVID symptoms, the queried symptoms do not include every symptom reported in Long COVID (brain fog and postexertional malaise are excluded, for example), but the inclusion of an "other" symptom question helps overcome this limitation.Severity was assessed using a Likert-scale asked for each symptom (1-5, very mild to very severe).

| Covariates
Other variables were self-reported including demographics, medical history, vaccine history, and lifestyle factors. 19Only body mass index (BMI) had >1% of data missing (650 missing, 40%).For the propensity adjusted model, we imputed missing BMI from age, sex, race/ ethnicity, and past medical history.

| Statistical analysis
The analysis plan was developed before outcome data collection and was designed to emulate a randomized clinical trial using a "target trial" approach. 20

| Results reporting and informed consent
Results are reported in accordance with STROBE guidelines. 21All participants provided digitally-signed informed consent.The study was reviewed and approved by the UCSF Institutional Review Board (#17-21879).
Restricting to those at high risk yielded similar results for sensitive  now posted on clinicaltrials.govbut not published. 235][26][27][28][29] Higher viral loads and prolonged viral shedding have been associated with risk of Long COVID, 3,4 and nirmatrelvir results in faster viral clearance, 30,31 supporting the hypothesis that nirmatrelvir may prevent Long COVID.Second, in this cohort, the number of symptoms during acute infection is associated with Long COVID symptoms independent of vaccination and variant wave, 19 but In all models, treatment with oral nirmatrelvir was not associated with lower odds of patient reported Long COVID symptoms with confidence intervals that cross 1, which suggests that treatment is not beneficial in reducing the risk of Long COVID.Note: Among 634 who responded to questions about rebound and Long COVID, although 26% experienced rebound symptoms, rebound test positivity, or both, there was not a statistically significant difference in the presence of Long COVID among those who experienced rebound compared to those who did not (OR: 1.34; 95% CI: 0.74-2.41;p = 0.33).
| 7 of 10 whether reducing acute symptoms with antiviral therapy prevents Long COVID has not been demonstrated.
Three prior studies have considered the effect of nirmatrelvir during acute infection on EHR-ascertained post-COVID outcomes.
Two found contradictory results despite both including Veterans and examining EHR-diagnosed post-COVID conditions by ICD-10 diagnostic codes as the outcome of interest. 15,16Bajema et al. 16 found that nirmatrelvir improved 30-day outcomes but not EHR-diagnosed post-COVID conditions at 6 months, whereas Xie et al. 15  In contrast to EHR-based studies focused on ascertainment of post-COVID-conditions using ICD-10 codes, our study evaluated whether nirmatrelvir is associated with patient-reported symptoms.
Patient-reported symptoms is a closer approximation of the condition of interest (Long COVID, defined as unexplained symptoms persisting for >90 days following initial infection). 1 One other published study surveyed participants from a single medical center 4 months after SARS-CoV-2 infection and compared the prevalence of self-reported prevalent Long COVID symptoms among people who reported nirmatrelvir use to those who were not treated with nirmatrelvir. 17r findings are consistent with their finding of no association between nirmatrelvir use and subsequent Long COVID symptoms.

Our inclusion of participants already enrolled in the COVID Citizen
Science study without prior infection, use of a target-trial framework, and propensity-adjustment further strengthens the validity of our findings.In sum, our finding that nirmatrelvir treatment during acute infection is not associated with lower odds of Long COVID is consistent with the Bajema report of no difference in post-COVID conditions at 6 months 16 and with the Congdon survey study at 4 months after infection. 17 found a higher proportion with clinical rebound than previously reported, [33][34][35] but did not identify an effect of posttreatment rebound on Long COVID symptoms.Since antivirals entered widespread use, there has been intense interest in this phenomenon, and a case report suggested the development of Long COVID among those with posttreatment rebound. 12The observation that antiviral-associated rebound is not associated with Long COVID suggests that the potential for treatment-related rebound should not discourage antiviral use, and that individuals experiencing this phenomenon are not at elevated risk for Long COVID.Rebound test positivity was common in the early days of the pandemic 36 ; we were not able to determine whether rebound unrelated to treatment increases risk of Long COVID.
Our study does not consider whether nirmatrelvir may be effective at treating (rather than preventing) Long COVID, which is currently under investigation in one ongoing and two planned clinical trials (NCT05576662, NCT05595369, NCT05668091).

| Limitations
The primary limitations arise from its observational nature which is at for example.Our study differs from EHR-based reports in that we assessed Long COVID symptoms rather than ICD-10 codes.We therefore could not determine whether treatment had an impact on post-acute diagnoses.We also did not include objectively measured post-COVID outcomes (e.g., exercise capacity, 37 neurocognitive performance, 38 or other measurable physiologic perturbations).We used propensity scores and inverse probability of treatment weighting to adjust for baseline differences between propensity of treatment between the treated and untreated, but residual confounding may still bias the results.Rebound testing was not performed systematically, and results were based on participant test self-report.

| CONCLUSIONS
Among vaccinated, nonhospitalized adults in the COVID Citizen Science online cohort, nirmatrelvir treatment during acute SARS-CoV-2 infection was not associated with Long COVID symptoms >90 days after infection.Among those treated, rebound was not associated with Long COVID symptoms.
We included CCS participants enrolled March 2020-August 2022 (median March 2021, interquartile range [IQR] August 2020-September 2021) with first-reported positive SARS-CoV-2 antigen or PCR test starting in March 2022.We excluded those treated with molnupiravir, remdesivir, and monoclonal antibodies.
First, we compared treated and untreated individuals using Chi-squared tests for categorical variables and t-tests for continuous variables.We modeled nirmatrelvir use propensity with logistic regression models including all individuals who met the inclusion/exclusion criteria including age, sex, race, ethnicity, socioeconomic status, education, employment, past medical history, substance use, number of vaccines.After checking the Hosmer-Lemshow goodness of fit and balance of key covariates by propensity score quintile, we used logistic regression models to model the association between nirmatrelvir and Long COVID among survey respondents adjusted for age, sex, time since SARS-CoV-2 test positivity, and the cubic spline of the propensity score.As a sensitivity analysis, we repeated the analysis using inverse probability of treatment weighting, truncating extreme weights >95th percentile.As a post hoc sensitivity analysis we restricted the inclusion criteria to United States residents at risk for severe COVID-19 defined according to the CDC using two definitions (age ≥50, one or more comorbid condition, or BMI ≥ 25 kg/m 2 ; age ≥65, one or more comorbid condition, or BMI ≥ 30 kg/m 2 ).The next post hoc sensitivity analyses used propensity matching to estimate the average treatment effect using 1:1 nearest neighbor matching by propensity score or by Mahalanobis distance (inverse sample covariate covariance).For analyses of rebound, we used Fisher's exact test and logistic regression for unadjusted analyses given the lack of evident confounders.Analyses were conducted in SAS version 9.4 and STATA 17.0.
T A B L E 1 Baseline characteristics among Long COVID survey respondents.

4 |
DISCUSSION Within an online observational cohort, treatment with nirmatrelvir among vaccinated, nonhospitalized individuals during first known SARS-CoV-2 infection was not associated with a lower prevalence of patient-reported Long COVID symptoms >90 days after infection.Treatment was not associated with fewer Long COVID symptoms or severe symptoms, although these endpoints were limited by rarity of these outcomes.Rebound symptoms or test positivity after nirmatrelvir treatment were not associated with Long COVID symptoms.Prior studies of nirmatrelvir, including two randomized clinical trials, EPIC-HR and EPIC-SR, focused on acute outcomes of SARS-CoV-2 infection among unvaccinated individuals.EPIC-HR demonstrated a reduction in hospitalization and mortality by day 28 among those at high risk of disease progression treated with nirmatrelvir compared to placebo, 11 but the EPIC-SR study of standard risk individuals was stopped early for lack of benefit with negative results Model 3, the prespecified primary result, includes age, sex, and time since SARS-CoV-2 infection, and the restricted cubic spline of the propensity score.Model 1 includes age, sex, and time since SARS-CoV-2 infection.Model 2 includes age, sex, time since SARS-CoV-2 infection, race/ethnicity, past medical history and substance use.Model 4 is the propensity-adjusted version of Model 2. Finally, Model 5 incorporates inverse probability of treatment weighting and age, sex, and time since SARS-CoV-2 infection.Results were similar in additional sensitivity analyses.CI, confidence interval.T A B L E 3 Number and percentage among rebound and no rebound.
risk for selection bias and confounding.The cohort was relatively homogeneous; most individuals identified as White with advanced education, in large part because we limited this study to those privileged enough to avoid infection until March 2022.We relied on self-report of treatment and Long COVID symptoms.Baseline data was collected before infection, and SARS-CoV-2 testing and nirmatrelvir use were collected at the time of testing and treatment, respectively, limiting the impact of recall bias.Nonetheless differential reporting of test positivity or survey response (which was nondifferential by treatment) may induce selection bias.The survey did not include all Long COVID symptoms, such as postexertional malaise, insomnia, decreased exercise tolerance, and menstrual cycle changes, For race and ethnicity, participants could select all that apply so these are not mutually exclusive categories.Number and percentage among the treated and untreated.