National survey study of nirmatrelvir-ritonavir prescribing processes and access barriers during the severe acute respiratory coronavirus virus 2 (SARS-CoV-2) omicron surge of the coronavirus disease 2019 (COVID-19) pandemic

Abstract

In this national survey, we found that individual patient assessments by pharmacists were more common at facilities using centralized prescribing for nirmetralvir-ritonavir (Paxlovid) than decentralized prescribing. Provider discomfort was initially less with centralized prescribing, but later, there was no difference in provider discomfort based on prescribing mechanism.

In December 2021, in the midst of a severe acute respiratory coronavirus virus 2 (SARS-CoV-2) omicron surge of the coronavirus disease 2019 (COVID-19) pandemic, nirmatrelvir-ritonavir (Paxlovid) received emergency use authorization (EUA) for treatment of mild-to-moderate COVID-19 based on data showing an 89.1% relative risk reduction in hospitalization or death in high-risk, unvaccinated patients.¹ Subsequent real-world studies have demonstrated its effectiveness in immunized populations in the setting of evolving variants.^2–5 Paxlovid’s oral formulation presents fewer logistical barriers to administration compared to intravenous COVID-19 treatments, for which delivery via outpatient infusion within a narrow eligibility timeframe was challenging.⁶ However, a number of unique challenges have been associated with Paxlovid prescribing, especially early after this drug became available in January 2022.

Paxlovid has numerous drug–drug interactions, predominantly through CYP3A inhibition, resulting in the potential need for medication adjustments to optimize safety.⁷ In addition, its EUA status led to medicolegal prescribing considerations unfamiliar to providers. Lastly, drug supply was initially limited, with variable federal allocations to each state at a time of high case volumes and significant staffing shortages. This situation presented concerns about access among individuals seeking treatment, prompting institutions to implement different strategies to address these challenges. In this study, we characterized the heterogeneity of Paxlovid prescribing processes utilized at multiple facilities across the United States, and we evaluated the association between prescribing processes and perceived barriers to access.

Methods

Study design and participants

This national cross-sectional survey study was conducted through the Society of Healthcare Epidemiology of America (SHEA) Research Network (SRN), a consortium of >100 healthcare facilities collaborating on multicenter research projects. A questionnaire was created to assess Paxlovid prescribing processes and access barriers (Supplementary Fig. 1 online). Responses were collected over 8 weeks starting in August 2022. This study was granted exempt-from-review status by the Dartmouth-Hitchcock Institutional Review Board.

Study exposure and covariates

The primary exposure was the mechanism of Paxlovid prescribing at the time of initial drug availability, which we classified as centralized or decentralized. Centralized prescribing was defined as utilizing a single mechanism, such as a telehealth clinic, to prescribe Paxlovid on behalf of a hospital or health system. Decentralized prescribing was defined as allowing providers to prescribe directly to pharmacies without oversight. Covariates included facility characteristics, pharmacist involvement in the prescribing process, and barriers to Paxlovid access at the times of initial drug availability and survey distribution. Barriers were ranked on a scale from 1 to 5 in descending order of significance and then dichotomized to “most significant” (if ranked 1 or 2) or “less significant” (if ranked 3, 4, or 5) for analysis. In addition, the SRN supplied demographic information about eligible and responding facilities.

Statistical analysis

For each covariate, we used univariate analysis to examine frequency for categorical variables and distribution for continuous variables. We used bivariate analysis to examine associations between the exposure and each of the covariates. We compared categorical variables using the Pearson χ² test or the Fisher exact test, and we compared continuous variables using the Wilcoxon rank-sum test. Changes in paired responses were compared using the Wilcoxon signed-rank test. All statistical analyses were conducted using SAS Studio version 3.8 software (SAS Institute, Cary, NC).

Results

In total, 70 SRN facilities in the United States were eligible to participate in the survey, of which 31 (44%) responded. Of these, 2 facilities were excluded because they answered an insufficient number of questions pertaining to the study covariates. Thus, 29 facilities were included in the primary analysis (Table 1). The median bed size was 500 and the median number of antimicrobial stewardship staff was 1.25 full-time equivalents. Nine facilities (31%) were academic medical centers. Facility characteristics did not differ between survey respondents and nonrespondents (Supplementary Table 1 online). Twelve facilities (41%) initially utilized a centralized Paxlovid prescribing process, with no differences in prescribing mechanism based on facility characteristics.

Table 1. Survey Responses Comparing Centralized and Decentralized Prescribing Mechanisms

Note. N/A, not available.

^a The Pearson χ² test or Fischer exact test was used for categorical variables. The Wilcoxon rank-sum test was used for continuous variables.

^b Median (interquartile range) reported for continuous variables.

^c Barriers were ranked on a scale from 1 to 5, in descending order of importance. Responses were dichotomized to those of 1 or 2 (ie, “most significant”) versus 3, 4, or 5 (ie, “less significant”).

Pharmacists were involved in several global and individual facets of the Paxlovid prescribing process, including guideline development (97%), direct teaching to providers (52%), and individual patient assessment for drug eligibility (48%). On bivariate analysis, pharmacists assessed individual patients more commonly at facilities with centralized compared to decentralized prescribing (75% vs 29%; P =.02).

The most common barriers at the time of initial drug availability were window of time for authorized use (55%), provider discomfort (55%), and drug supply (52%). By the time of survey distribution, window of time (76%) and provider discomfort (52%) remained significant barriers. Although supply became less of a barrier over this period (52% to 0%; P < .001), lack of patient interest emerged as a new barrier (10% to 62%; P < .001). On bivariate analysis, at the time of initial drug availability, provider discomfort was less of a barrier at facilities with centralized prescribing (33% vs 71%; P = .05). By the time of survey distribution, there were no differences in provider discomfort based on prescribing mechanism (50% vs 53%; P = .88).

Discussion

To our knowledge, this national survey is the first published study describing variations in Paxlovid prescribing at multiple SRN facilities across the United States, while correlating these differences with perceived access barriers. Our survey identified provider discomfort with Paxlovid prescribing as a significant barrier, possibly owing to its numerous drug–drug interactions, initial limited supply, and uncertainty with navigating its EUA status. To circumvent these issues, some facilities utilized centralized prescribing mechanisms for Paxlovid in the first weeks after drug availability, shifting away the burden of familiarity from individual providers in the short term. Our study demonstrated that early on, prescriber discomfort was significantly less of a barrier at facilities with centralized prescribing than those with decentralized prescribing. At the time of survey distribution, however, there was no difference in provider comfort based on initial prescribing strategy, indicating that a centralized mechanism did not hinder the ability of providers to gain expertise with this drug over time. Although supply became less of a barrier over time, patient interest also decreased significantly, which could be explained by the progressively decreasing severity of SARS-CoV-2 variants. These barriers did not differ by prescribing mechanism at any point in time.

At almost all facilities, pharmacists were integral to Paxlovid prescribing. These findings are consistent with previously published studies describing the role of pharmacists in the COVID-19 pandemic response.^8,9 In addition, our survey revealed that pharmacists were more likely be involved with individual assessments at facilities that adopted centralized prescribing strategies, highlighting their important responsibility within multidisciplinary clinics for tasks such as medication reconciliation and assessment of drug–drug interactions. The unique role of pharmacists in handling Paxlovid has since been recognized through expansion of its EUA to allow direct prescribing by pharmacists.¹⁰

Our study had several limitations. This study did not consist of a simple random sample and was not powered to assess for confounders. Our response rate was relatively low and could have resulted in selection bias, though facility characteristics were similar between respondents and nonrespondents. Responses were subject to recall bias, and we assumed that the responses of survey participants accurately reflected the perceptions of their institution.

In conclusion, through this national survey study, we characterized facility-level variations in Paxlovid prescribing processes and correlated these to the evolution of access barriers over time. These findings could inform future large-scale distribution for novel therapeutics with complex prescribing considerations for COVID-19 or other pandemic diseases.