Association of Changes in Seasonal Respiratory Virus Activity and Ambulatory Antibiotic Prescriptions With the COVID-19 Pandemic

Abstract

This pre-post study examines the association between respiratory virus incidence and ambulatory antibiotic orders and the COVID-19 pandemic.

The COVID-19 pandemic led to numerous measures to mitigate the spread of SARS-CoV-2, including cancellations of gatherings, closure of businesses and schools, social distancing, wearing face masks, and other hygiene measures.¹ These may have unintended positive associations with reducing other respiratory infections. As antibiotics are frequently inappropriately prescribed for viral respiratory diseases,² we hypothesized that a decreased respiratory virus incidence would be associated with reduced ambulatory antibiotic orders.

Methods

We conducted a pre-post study consisting of a pre–COVID-19 pandemic period (July 2018 to February 2020), a 1-month run-in period (March 2020), and a COVID-19 pandemic period (April 2020 to February 2021). This study received an institutional review board exemption from the University of Wisconsin because it used aggregate prescribing data only, without any patient or clinician identifiers. The Wisconsin State Laboratory of Hygiene provided weekly statewide surveillance polymerase chain reaction data for the following respiratory viruses: influenza, respiratory syncytial virus, human parainfluenza virus, human metapneumovirus, seasonal coronavirus, adenovirus, and enterovirus/rhinovirus. University of Wisconsin Health is a large, academic health system that serves southern Wisconsin, with more than 80 ambulatory sites and 7 000 000 annual ambulatory encounters. Ambulatory antibiotic prescribing data were collected for all University of Wisconsin Health ambulatory clinics, visit types (eg, in person, telemedicine, and telephone), and ages. We calculated the number of monthly antibiotic prescriptions per 1000 patient encounters to account for patient volume variability. Indication is required for all ambulatory antimicrobial prescriptions using an order tool within the electronic health record. We examined changes in (1) respiratory virus detections, (2) antibiotic prescriptions per 1000 patient encounters, and (3) antibiotic prescriptions for respiratory tract infections (RTIs) per 1000 patient encounters. Pre-post comparisons were performed using the t test and Mann-Whitney U test; correlations between prescription rates and virus detection was performed by the Pearson correlation coefficient (Sigma Plot, Systat Software, Inc). Further, we constructed a generalized additive model to compare antibiotic prescribing patterns before and during the COVID-19 pandemic with a spline to account for seasonality (R, version 1.3; R Foundation for Statistical Computing). Statistical significance was set at P < .05.

Results

Respiratory virus detections demonstrated seasonal variation during the prepandemic, but not pandemic, period (Figure 1). Winter seasonal viruses (influenza, respiratory syncytial virus, and seasonal coronavirus) currently average 12 per month compared with 4800 per month in previous seasons (P < .001). Other respiratory virus detections also decreased from 560 per month prepandemic to 228 per month during the pandemic (P < .001).

Figure 1. Respiratory Virus Testing Results From July 2018 Through February 2021.

The total number of polymerase chain reaction–positive respiratory virus test results by week are shown by stacked boxes. Data are from the State of Wisconsin Respiratory Virus Surveillance Program. The blue line represents the total testing volume on a weekly basis. The gray background indicates the COVID-19 pandemic period. hMPV indicates human metapneumovirus; PIV, human parainfluenza virus; RSV, respiratory syncytial virus; sCoV, seasonal coronavirus.

Ambulatory encounters were similar during both periods (637 000 vs 661 000 per month; P = .24). During the prepandemic period, antibiotic prescribing rates increased during winter respiratory viral seasons. In contrast, antibiotic prescribing rates decreased in the short term and remained low throughout the pandemic period (Figure 2A). Adjusting for seasonality, monthly antibiotic prescriptions for RTI fell 79% from 10.5 to 2.2 prescriptions per 1000 patient encounters (P < .001) (Figure 2, B and C). Noninfluenza virus detections demonstrated the strongest correlation with antibiotic prescribing for RTI (r = 0.82; P < .001).

Figure 2. Ambulatory Antibiotic Prescribing Rates July 2018 Through February 2021.

Antibiotic prescriptions were normalized by the number of patient encounters. Indication was captured using a required order form element within the electronic health record. A, Ambulatory antibiotic prescribing rates for all indications (orange), non–respiratory tract infection (RTI) indications (eg, urinary tract infection; black), and RTI indications (dark gray). Circles indicate the prepandemic period, and the triangles and blue background indicate the COVID-19 pandemic period. The dotted line denotes the onset of the COVID-19 pandemic. B, Ambulatory antibiotic prescription rates (95% CI) for RTI during the pre-COVID-19 and COVID-19 pandemic periods, adjusted for seasonality. C, Ambulatory antibiotic prescription rates for RTI by month stratified by prepandemic period (circles) and pandemic period (triangles). Individual points denote observed prescription rates. The black line represents the generalized additive model prediction (shaded area, 95% CI) for antibiotic prescriptions for RTI adjusted for seasonality.

Discussion

Consistent with our hypothesis, this study demonstrated a marked decrease in respiratory virus detections, with a concomitant 79% decline in ambulatory antibiotic prescribing rates for RTI, during the COVID-19 pandemic. Prior studies conducted soon after the onset of the pandemic and over shorter durations showed relatively quick declines in rates of viral infections or antimicrobial prescribing.^3,4,5,6 However, to our knowledge, this study is the first to combine respiratory virus activity and antibiotic prescriptions that were indexed to the number of encounters over a prolonged period that includes the typical respiratory virus season. The data suggest that COVID-19 transmission mitigation strategies may help curb respiratory viral diseases beyond SARS-CoV-2 and, indirectly, decrease antibiotic prescribing. Notably, during COVID-19 surges in Wisconsin (October to December 2020), we did not observe increased ambulatory antibiotic prescriptions. Additionally, noninfluenza virus activity demonstrated a stronger correlation with antibiotic prescription rates. We hypothesize that this may be because of highly accessible, rapid diagnostics for SARS-CoV-2 and influenza viruses to avert unnecessary antibiotic prescriptions. These findings may have important implications for future stewardship and public health strategies.