Inappropriate antibiotic use leads to increased bacterial resistance, increased risk of Clostridioides difficile infections, other antibiotic-related side effects, and increased costs. Antibiotics prescribed in ambulatory settings account for an estimated 80%–90% of human antibiotic use.Reference Duffy, Ritchie, Metcalfe, Van Bakel and Thomas1 Furthermore, up to 50% of ambulatory antibiotics may be unnecessary, and ambulatory antibiotics account for >60% of antibiotic costs.Reference Chua, Fischer and Linder2,Reference Suda, Hicks, Roberts, Hunkler and Danziger3
Ambulatory urgent-care clinics and their clinicians may face challenges in antibiotic prescribing. Urgent care practices may see patients of higher acuity than primary care clinics. They may lack an existing relationship with patients, may see patients who do not have regular access to primary care, and may not have a mechanism for effective patient follow-up.Reference Dosh, Hickner, Mainous and Ebell4 Urgent care practices have higher antibiotic prescribing rates than other ambulatory sites of care.Reference Palms, Hicks and Bartoces5–Reference Jones, Sauer and Jones8
Patient satisfaction is of particular concern at urgent care practices, which compete with each other and other sites of care based on convenience and patient experience.Reference Incze, Redberg and Katz9 Although studies of whether antibiotic prescribing is associated with higher patient satisfaction have been mixed,Reference Sharp, Shen, Kanter, Berman and Gould7,Reference Jerant, Fenton and Kravitz10–Reference Sanghavi, McWilliams, Schwartz and Zaslavsky16 patient satisfaction is a frequent reason clinicians cite for inappropriate antibiotic prescribing.Reference Patel, Pfoh and Misra Hebert17,Reference Kohut, Keller and Linder18
To measure, monitor, and improve ambulatory antibiotic prescribing, our health system formed an ambulatory antibiotic stewardship committee in 2018. Due to relatively high antibiotic prescribing and a perception of injudicious antibiotic prescribing in urgent care, we focused interventions on our urgent care practices. Interventions implemented in fiscal year 2019 (FY19) focused on improving antibiotic use for non–antibiotic-appropriate diagnoses, but we were also concerned about diagnosis shifting (ie, an increase in the number of related antibiotic-appropriate diagnoses), overall antibiotic prescribing, and maintaining patient satisfaction.Reference Linder19
Methods
Setting and organization
Northwestern Medicine is an integrated academic healthcare system in the Chicago area that includes 13 hospitals, 481 outpatient locations, and 23 urgent care clinics, also known as “Northwestern Medicine Immediate Care Centers.” Of the urgent care clinics, 17 were open and operational continually from FY19 to FY21 and are included in this analysis; 6 additional urgent care clinics opened between the beginning of FY19 and the end of FY21. The Northwestern Medicine fiscal year begins on September 1.
These urgent care clinics are overseen by 1 operational unit, Northwestern Medical Group (NMG), and they span 4 different regions in our area: central, north, west, and northwest. Each urgent care clinic is staffed with physicians and advanced practice professionals (APPs)—advanced practice nurses and physician assistants. The urgent care practices have their own urgent-care quality committees.
The Northwestern Medicine Ambulatory Antimicrobial Stewardship Committee (AASC) was established in FY18 comprising physician champions from primary care, infectious diseases, and urgent care, as well as the population health pharmacy director, practice managers, and quality improvement leaders.
Clinician survey
To help plan stewardship activities, we created and administered a survey to practicing clinicians between December 2019 and January 2020. We asked, “What do you believe to be the most common reasons why physicians/APPs prescribe antibiotics for viral respiratory diagnoses?” Respondents could choose 1 or more of the following responses: (1) fear of receiving a poor postvisit patient satisfaction score; (2) patient perception of the value of clinic visit; (3) to treat superimposed secondary bacterial infections; and (4) to instruct the patient to wait and fill if the symptoms worsen. Via an urgent-care listserv, we sent the survey to 93 physicians and APPs; responses were anonymous.
Measures
Stewardship measures
Based on Healthcare Effectiveness Data and Information Set (HEDIS) measures, we developed our own non–antibiotic-appropriate prescribing measures.20 The committee created and validated the acute respiratory infection (ARI) stewardship measures in 2018. We developed 4 stewardship measures that we combined into “stewardship target visits” (Table 1). The 4 stewardship measures were antibiotic prescribing (1) for upper respiratory tract infections for children aged 3 months to 17 years, (2) for pharyngitis with negative Streptococcus (ie, strep) test for children aged 3–17 years, (3) for pharyngitis with negative strep test for adults aged ≥18 years, and (4) for URI, acute bronchitis, influenza for adults aged ≥18 years. For this article, we have combined the data for all 4 measures and labeled them “stewardship measures.”
Encounter diagnoses are chosen by the treating clinician based on International Classification of Diseases, Tenth Clinical Modification (Appendix 1.A online). Adult patients with multiple stewardship-qualifying diagnoses documented in the same encounter were be included in 2 different measures. For instance, adult patients with pharyngitis and negative strep test and either URI, acute bronchitis, or influenza were included in 2 measures. The stewardship measures only included in-person and synchronous video or telephone encounters with clinicians and diagnoses did not have to be the primary diagnosis.
Patients were excluded from measures if they had a competing infectious diagnosis (Appendix 2 online) and/or comorbid condition history (Appendix 3 online). We used multiple code sources, including HEDIS, to identify competing diagnoses and comorbid conditions. For all measures, we excluded patients with a competing, antibiotic-appropriate diagnosis based on the scheme from Chua et alReference Chua, Fischer and Linder2 documented 30 days prior through 7 days after the date of the encounter. For the adult acute bronchitis, URI, and influenza measure, we excluded patients with a comorbid condition such as chronic infections, chronic lung diseases, cancer, or immunosuppression documented 12 months prior through the date of encounter.
Diagnosis shifting measure
Because we were concerned that clinicians might try to evade the stewardship measures by shifting their diagnosis selection, adding comorbid diagnoses, or adding antibiotic-appropriate diagnoses, we developed a “diagnosis shifting measure” (Table 1 and Appendix 1.B online). The diagnosis shifting measure included visits with any of the following codes: stewardship measure codes omitting exclusions for competing diagnoses or comorbid conditions (to detect whether clinicians were evading the measure by adding competing or comorbid diagnoses), potentially antibiotic-appropriate codes (eg, sinusitis, otitis media), antibiotic-appropriate codes (eg, pneumonia), and nonspecific codes (eg, cough, fever, other respiratory disorders).
All-antibiotic measure
To observe the impact of stewardship interventions on overall antibiotic prescribing in urgent care clinics, we measured antibiotic prescribing regardless of diagnosis or visit type.Reference Linder19 Patients were not excluded from the all-antibiotic measure for competing diagnosis and/or comorbid condition history.
Antibiotics
We used a custom grouping within our electronic health record containing all antibiotics based on generic name. Antibiotics prescribed in the same encounter as the diagnoses were included in the numerator.
Patient satisfaction
We examined patient satisfaction with clinicians for visits that met the stewardship measure criteria (ie, had a diagnosis of URI, pharyngitis with a negative strep test, acute bronchitis, or influenza). We sent a patient satisfaction survey to all patients with available e-mail or mobile phone numbers after their urgent care visit. The survey asks questions about satisfaction with the location and with the clinician. We calculated the annual satisfaction rate for each clinician as the proportion of patients who ranked the clinician a 9 or 10 on a 1–10 scale divided by the number of satisfaction survey respondents. We also compared individual clinicians’ antibiotic prescribing rate with their satisfaction rate.
Urgent-care stewardship program implementation
In FY19, review of health system data indicated that antibiotic prescribing was generally the highest in the urgent care clinics. This resulted in an initiative presented to the NMG Urgent Care Quality Committee, which included regional medical directors. Following approval by this committee, next steps included multiple introductory meetings between the lead regional medical director and local urgent-care physician leaders. The meeting introduced a review of ARI measure definitions, inclusion and exclusion criteria, and baseline data. These meetings were effective in gaining understanding and support from these key leaders. All of the regional medical directors as well as these lead urgent-care physicians performed random chart audits, provided feedback, and verified the accuracy of the data being captured by the stewardship measures.
An educational initiative was developed and implemented at urgent care clinics across multiple regions. Data for the 4 measures were presented to all the clinicians by the lead medical director at multiple immediate-care regional quarterly meetings. Specialty and regional data were also presented at the quarterly Urgent Care Leadership meetings, Ambulatory Pharmacy and Therapeutics Committee meetings, and the monthly System Antimicrobial Stewardship Quality Committee meetings.
Intervention design
In designing our interventions, we considered the competing priorities of our integrated academic health system. The goal of these interventions was not only to improve clinician antibiotic use but also (1) to not interfere with clinician workflows at the point of care (eg, introducing a hard stop in the electronic health record if only a nonantibiotic appropriate diagnosis was documented during a patient visit), (2) to not cause undesirable behavior change like diagnosis shifting, and (3) to not decrease patient satisfaction. We catalogued stewardship-related activities from September 2019 through September 2021 (Appendix 1.C online). Different interventions were introduced, mainly starting in December 2019, among the regions and urgent care clinics based on operational staff availability as well as decision making by the AASC, quality directors, and medical directors. In this intervention design, we used comparative feedback, clinician and patient education, and commitment posters.
Interventions—Clinician comparative feedback and education
Comparative feedback
Regional-level comparative antibiotic prescription data were shared with clinicians and leadership through multiple regional and quality committee meetings. In addition, we developed a quality-measure summary report using the NMG Enterprise Data Warehouse (EDW), which provided individual clinician antibiotic prescribing rates and compared them with the overall regional rates (Appendix 1.D online). We leveraged clinicians’ competitive nature and sense of group pride by providing comparisons of health-system regions and specialties (ie, urgent care vs primary care).Reference Linder, Meeker and Fox21 This report included the 4 ARI stewardship measures, the denominator for the last 12 months, and individual and regional-level antibiotic prescribing rates. The report did not include the diagnosis shifting measure. Regional medical directors reviewed the report with individual physicians and/or APPs annually to discuss potential barriers and opportunities for improvement. Additional follow-up was conducted with individual clinicians who had greater opportunity for improvement in their antibiotic prescribing.
Video
All urgent-care clinicians were assigned and required to view an internally developed 3-minute video showcasing a physician–patient interaction.22 The video showed a physician explaining the diagnosis of a viral illness, why an antibiotic prescription was not necessary, the expected duration of illness, and symptom management options.
Presentation by subject-matter expert
A physician champion also conducted a series of educational sessions discussing appropriate management of viral illnesses for urgent care clinicians. This forum also provided the opportunity for dialogue, questions, and answers.
Interventions—Patient education
Commitment posters
The CDC “A Commitment to Our Patients About Antibiotics” poster was printed (18×24 inches), signed by clinicians working in each individual location, and posted in each exam room facing the examination table in each urgent care clinic.Reference Meeker, Knight and Friedberg23,Reference Sanchez, Fleming-Dutra, Roberts and Hicks24
Patient education pamphlets
A trifold patient education pamphlet was developed and made available to urgent care clinics explaining the potential harms of being prescribed and taking antibiotics for viral illnesses.
Impact of the COVID-19 pandemic
During the initial COVID-19 period, from March 2020 to July 2020, the visit volume decreased dramatically in all urgent care clinics. Many urgent care locations implemented drive-through testing sites for influenza, pharyngitis, and COVID-19 and eventually became the outpatient COVID-19 testing sites for the system. Urgent care clinics resumed in-clinic testing in June 2020 for influenza, pharyngitis, and COVID-19. Each clinic managed telemedicine visits from March 2020 onward.
Statistical analysis
We calculated and report means (with standard deviations) and percentages. We examined visits and antibiotic prescribing initially by year. For antibiotic prescribing, we calculated the difference in antibiotic prescribing rates from FY19 to FY21 and the 95% confidence interval (CI) of the differences. We examined trends in visit rates and antibiotic prescribing over time by month. To compare antibiotic prescribing and patient satisfaction within each study year, we calculated Pearson correlation coefficients and P values.
Results
Clinician survey
To the survey, we received 38 responses (response rate, 41%) and 51 answers. Clinicians said the most common reasons for prescribing antibiotics for viral respiratory diagnoses were (1) fear of receiving a poor postvisit patient satisfaction score (42%), (2) to treat superimposed secondary bacterial infections (37%), (3) patient perception of the value of the clinic visit (32%), and (4) to instruct the patient wait and fill the prescription if the symptoms worsen (24%) (Appendix 1.E online).
Patient characteristics
From FY19 to FY21, 576,609 patients made 1,358,816 visits to the urgent care practices. Patients had a mean age of 42 years; 57% were women; 78% were white; and 9% were Latino (Table 2c and Table 3). Overall, patients had a median of 0 comorbidities, took a median of 3 prescription medications, and had a median of 1 physician visit during the study period. Also, 77% of patients had a primary care physician listed. Unique patients with a stewardship visit, compared to patients overall, were younger, were more likely to be women, to be single, and to have private insurance. They were less likely to make physician visits or emergency department visits or to be hospitalized (Table 2a and 2c). From FY19, the mean age of stewardship visit patients decreased from 35 to 30 years (Table 2a); the mean age of all patients increased from 40 to 44 years (Table 2c).
Note. FY, fiscal year and FY19 indicates fiscal year 2019.
a Percentages may not add up to 100% because of rounding.
b 1 patient was missing data for sex.
Note. FY, fiscal year and FY19 indicates fiscal year 2019.
a Percentages may not add up to 100% because of rounding.
b 4 patients were missing data for sex.
Note. FY, fiscal year and FY19 indicates fiscal year 2019.
a Percentages may not add up to 100% because of rounding.
b 42 patients were missing data for sex.
Note. Neg strep, negative for Streptococcus; FY, fiscal year and FY19 indicates fiscal year 2019.
Antibiotic prescribing
Examined by year, from FY19 to FY21, antibiotic prescribing for stewardship measures decreased from 34% to 12% (−22%; 95% CI, −23% to −22%) (Table 3). By absolute percentage, the largest decreases in antibiotic prescribing were in the adult acute bronchitis, URI, influenza measure, with a mean decrease of 28% (95% CI, −29% to −27%). The number of qualifying visits also notably declined from 21,531 in FY19 to 7,341 in FY21. For adult acute bronchitis, URI, and flu, the biggest decrease was for antibiotic prescribing for acute bronchitis (−25%; 95% CI, −28% to −23%).
Antibiotic prescribing for the diagnosis shifting measures decreased from 63% to 35% (−28%; 95% CI, −28% to −27%), and the number of visits that qualified for the diagnosis shifting measure also decreased. Antibiotic prescribing for any visit decreased from 30% to 10% (−20%; 95% CI, −20% to −20%) and the year-to-year total number of visits increased.
By month, superimposed on the overall downward trend in antibiotic prescribing for the stewardship measure, in the first 3 months of the COVID-19 pandemic, the number of visits that qualified for the stewardship measure and the diagnosis shifting measure notably decreased and the antibiotic prescribing rate increased (Fig. 1). Toward the end of FY21, the diagnosis shifting measure increased.
Patient satisfaction
The patient satisfaction scores for ARI visits in FY19, FY20, and FY21 were 83%, 89%, and 89%, respectively. Examined by year, there was no association between antibiotic prescribing rates of individual clinicians and patient satisfaction for ARI visits (Fig. 2).
Discussion
We conducted a multifaceted quality-improvement project to improve antibiotic prescribing in urgent care practices that was associated with achieving the stewardship target. In addition, we noted reductions in diagnosis shifting and overall antibiotic prescribing. Patient satisfaction at ARI visits increased over time, and patient satisfaction was not associated with clinicians’ antibiotic prescribing rate.
Urgent care clinics may have particularly high antibiotic-prescribing rates because of case mix (ie, urgent care practices see a higher proportion of patients with respiratory complaints) as well as high antibiotic prescribing for non–antibiotic-appropriate diagnoses. For example, Palms et alReference Palms, Hicks and Bartoces5 reported that urgent care centers have particularly high antibiotic prescribing rates compared to medical offices, emergency departments, and retail clinics.Reference Palms, Hicks and Bartoces5 In addition, the focus on convenience and patient satisfaction may encourage inappropriate antibiotic prescribing.Reference Incze, Redberg and Katz9
Other antibiotic stewardship efforts in urgent care clinics have been successful,Reference Link, Townsend, Leung, Kommu, Vega and Hendrix25–Reference Ha, Ong’uti and Chang30 but to our knowledge, none have simultaneously measured stewardship, diagnosis shifting, and overall antibiotic prescribing while striving to maintain patient satisfaction.
Our survey confirmed prior work; clinicians’ main concern in prescribing antibiotics for viral respiratory diagnoses was maintaining patient satisfaction and meeting patients’ expectations.Reference Patel, Pfoh and Misra Hebert17,Reference Kohut, Keller and Linder18 Clinicians also cited use of “delayed” prescriptions despite many reasons why delayed antibiotics are inadvisable.Reference Rowe and Linder31,Reference Rowe and Linder32 With a focus on clinician–patient communication, individual antibiotic prescribing rates for ARIs were not associated with decreased patient satisfaction.
Our intervention and measurement study had several limitations. First, this was not a randomized trial, we did not use a comparison or control group, and the changes we noted happened during the COVID-19 pandemic. At the beginning of the COVID-19 pandemic, we observed an increase in the antibiotic prescribing rate. We attributed this trend to the inability of clinicians to perform a physical examination due to the introduction of telemedicine visits and empirical treatment. Because the number of visits during this period was lower, the overall impact on the cumulative annual prescription rate was minimal. Excluding rapid shifts in visits and antibiotic prescribing during the initial phase of the COVID-19 pandemic, there appeared to be a reasonably consistent downward trend in antibiotic prescribing over the study period.
Second, the intervention was not implemented in a uniform way among the urgent care practices, making it impossible to attribute changes to any one activity or component of the intervention. Third, diagnoses depended on clinician coding. By examining a diagnosis shifting measure, we tried to ensure that improvements in the stewardship measure were not a result of “gaming.” This intervention improved diagnosis shifting as well as overall antibiotic prescribing. Fourth, the COVID-19 pandemic and the staggered implementation among practices precluded our use of a more robust primary analysis, like an interrupted time series.Reference Schweizer, Braun and Milstone33 However, a post-hoc interrupted time series demonstrated a significant step decrease in antibiotic prescribing (Appendix 1.F online).
Despite the apparent success of these efforts, the antibiotic prescribing rate for non–antibiotic-appropriate diagnoses remained ∼10% by the end of the measurement period. In addition, in the last year of the intervention, there appeared to be a dissociation between the stewardship and the diagnosis shifting measures that should be monitored.
In conclusion, it is possible to decrease antibiotic prescribing for presumed viral illnesses in the urgent-care setting and maintain patient satisfaction. Key elements of success may have included an ambulatory antimicrobial stewardship committee, physician champions, having data for analysis and reporting, and educational resources for patients and prescribers. Decreasing antibiotic prescribing for non–antibiotic-appropriate diagnoses has the potential to start a virtuous cycle for patients because nonreceipt of antibiotics may be associated with less antibiotic seeking in the future.Reference Shi, Barnett, Jena, Ray, Fox and Mehrotra34
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/ice.2022.164
Acknowledgments
We thank the members of the Ambulatory Antimicrobial Stewardship Subcommittee, System Antimicrobial Stewardship Quality Committee, and Immediate Care Center Quality Committee for their contributions. We are grateful to Immediate Care Center clinicians who participated in the survey, performed chart audits, and shared their insights.
Financial support
No funding specifically supported this work. Dr. Linder is supported by a contract from the Agency for Healthcare Research and Quality (grant no. HHSP233201500020I) and grants from the National Institute on Aging (grant nos. R33AG057383, R33AG057395, P30AG059988, and R01AG069762), the Agency for Healthcare Research and Quality (grant nos. R01HS026506 and R01HS028127), and the Peterson Center on Healthcare.
Conflicts of interest
No author reports potential conflicts of interest related to this article.