Arbel et al. (Sept. 1 issue)1 found that among adults 65 years of age or older, the rates of hospitalization and death due to Covid-19 were lower among those who were treated with nirmatrelvir than among those who were not. We are uncertain whether these results are generalizable to the untreated population for two reasons. First, the vast majority of patients (96%) were not treated, which points to the presence of marked selection bias. The cause of this observation may be difficult to determine; however, it must be elucidated. The considerably skewed distribution of known confounders impairs extrapolation of the results to the untreated group. The use of propensity-score–based methods would allow for the probability of treatment in both groups to be matched, thereby enabling a more balanced, and perhaps stratified, comparison of outcomes between treated and untreated patients.2 Second, despite the fact that the authors corrected for known confounders, there were probably unmeasured confounders (e.g., severity of symptoms) that influenced the treatment decision through presumed treatment benefit. Although the results of the study are encouraging, differences between the groups and the absence of data on adverse events impair astute translation of the findings of this observational study to clinical practice and policy making.

Maaike C. Swets, M.D.
Mark G.J. de Boer, M.D., Ph.D.
Geert H. Groeneveld, M.D., Ph.D.
Leiden University Medical Center, Leiden, the Netherlands
[email protected]

No potential conflict of interest relevant to this letter was reported.

1. Arbel R, Wolff Sagy Y, Hoshen M, et al. Nirmatrelvir use and severe Covid-19 outcomes during the omicron surge. N Engl J Med 2022;387:790-798.

2. Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res 2011;46:399-424.

Arbel et al. reports that among high-risk patients 65 years of age or older, the rates of hospitalization and death due to Covid-19 were lower among patients who received nirmatrelvir than among those who did not. Although a role for nirmatrelvir treatment in this high-risk older age group is possible, particularly in the now-dwindling older age group without previous immunity, further emphasis should be placed on the finding that treatment with nirmatrelvir did not appear to benefit high-risk patients 40 to 64 years of age and by extension, any high-risk patient younger than 65 years of age and any low-risk patient overall. In addition, on the basis of the data available in the article, the number of high-risk patients 65 years of age or older that would need to be treated to prevent one hospitalization is estimated to be 69 at a cost of $36,397 and, to prevent one death, 321 at a cost of $170,316. In a population of high-risk patients 65 years of age or older with previous immunity, the number needed to treat and the cost per hospitalization or death would increase substantially. At this stage in the global pandemic, it is time to contrast the benefits of these interventions with their cost effectiveness and with the number needed to treat.

Blake Nield, M.B., B.S.
Royal Prince Alfred Hospital, Sydney, Australia
[email protected]

No potential conflict of interest relevant to this letter was reported.

In a retrospective analysis, among high-risk patients 65 years of age or older who were infected with SARS-CoV-2 during the omicron surge, the rates of hospitalization and death were lower among those who received nirmatrelvir than among those who did not. However, the percentage of immunized patients was higher in the nirmatrelvir group than in the untreated group, given that infection- and vaccination-induced immunity confers protection against severe disease1 and given that nirmatrelvir therapy is reported to have no clinically significant beneficial effects in patients who had preexisting SARS-CoV-2 antibodies or in patients who have been vaccinated.2,3

The clinical spectrum of Covid-19 has changed with the emergence of the omicron variant, as evidenced by the dramatically reduced numbers of hospitalizations and deaths4 and by the fact that lung failure induced by Covid-19 has become a rarity. Nowadays, most patients are hospitalized with Covid-19 and another leading diagnosis and not because of Covid-19 alone. Therefore, information regarding the causes of hospitalization and death would result in a better estimate of the clinical risk–benefit profile of nirmatrelvir therapy. Such information would also provide details regarding the number of patients that would need to be treated to prevent one SARS-CoV-2–induced hospitalization or death and would take into account the many pharmacologic interactions of nirmatrelvir.

Günter Weiss, M.D.
Medical University of Innsbruck, Innsbruck, Austria
[email protected]

Dr. Weiss reports receiving lecture fees from Astro Pharma, Lilly, Menarini, Shionogi, Takeda, and Vifor Pharma and advisory board fees from AstraZeneca, Insmed, MSD, Pfizer, and Shionogi. No other potential conflict of interest relevant to this letter was reported.

1. Nordström P, Ballin M, Nordström A. Risk of SARS-CoV-2 reinfection and COVID-19 hospitalisation in individuals with natural and hybrid immunity: a retrospective, total population cohort study in Sweden. Lancet Infect Dis 2022;22:781-790.

2. Hammond J, Leister-Tebbe H, Gardner A, et al. Oral nirmatrelvir for high-risk, nonhospitalized adults with Covid-19. N Engl J Med 2022;386:1397-1408.

3. Pfizer. Pfizer reports additional data on PAXLOVID supporting upcoming new drug application submission to U.S. FDA. June 14, 2022 (https://www.pfizer.com/news/press-release/press-release-detail/pfizer-reports-additional-data-paxlovidtm-supporting).

4. Lewnard JA, Hong VX, Patel MM, Kahn R, Lipsitch M, Tartof SY. Clinical outcomes associated with SARS-CoV-2 omicron (B.1.1.529) variant and BA.1/BA.1.1 or BA.2 subvariant infection in Southern California. Nat Med 2022;28:1933-1943.

The authors reply: In response to Swets and colleagues: The applied method aimed to provide results that are generalizable to the treatment-eligible population (approximately 10% of SARS-CoV-2–infected patients). Eligible patients were further prioritized for therapy on the basis of their evaluated risk, which resulted in differences between the characteristics of the treated and untreated patients. We accounted for these differences by using multivariate Cox regression models. This approach was chosen to maximize the statistical power of the study, given the relatively small treatment group and low event rate. Moreover, we applied matching as an alternative method, and this method produced point estimates that were similar to those of the published analysis but with lower statistical power that did not allow for further stratification according to immunity status.

A primary limitation of the study is the lack of data regarding the specific Covid-19 symptoms and their severity, which may act as a confounder. A higher proportion of patients with more severe symptoms receiving treatment would increase the risk of severe Covid-19 outcomes among treated patients. Hence, our results may underestimate the effectiveness of the treatment. The alternative direction of this bias (i.e., if patients who were unwell enough to receive treatment at home were hospitalized shortly after the infection) was addressed by an additional analysis described in our article. This sensitivity analysis excluded hospitalizations that occurred during the first 2 days of follow-up. The fact that the effect size beginning on day 3 of follow-up was similar to that of the main analysis suggests that this bias does not explain a substantial proportion of the effectiveness.

In response to Nield and Weiss: We used the adjusted hazard ratios and corresponding confidence intervals to calculate the number needed to treat, as recommended in studies with varying follow-up times,1 and the associated costs. Among patients 65 years of age or older, the cost associated with the number of patients that would need to be treated with nirmatrelvir to prevent one Covid-19–related death is estimated to be $171,276; in contrast, the cost associated with the number of patients that would need to be vaccinated with a first booster of the BNT162b2 mRNA vaccine (Pfizer–BioNTech) to prevent one such death is estimated to be $22,744.2 Indeed, our study did not show benefit in younger adults at the population level. However, the decision regarding the use of therapy in an individual patient should take into account the risk of severe Covid-19 outcomes, as well as the patient’s coexisting conditions and concomitant medications.

In response to Weiss: The Israeli Ministry of Health has developed a national reporting system according to National Institutes of Health guidelines.3 The system includes reporting for hospitalization and death due to Covid-19. A systematic international analysis of Covid-19–related deaths showed that the Covid-19–related mortality rate reported in Israel was highly correlated with the observed excess mortality rate in our study.4

Ronen Arbel, Ph.D.
Yael Wolff Sagy, Ph.D.
Ariel Hammerman, Ph.D.
Clalit Health Services, Tel Aviv, Israel
[email protected]

Since publication of their article, the authors report no further potential conflict of interest.