Discussion

The observed decline in inappropriate ambulatory antibiotic prescribing from 2012 to 2017 is welcome, though further stewardship efforts are needed particularly in males and adults ≥40 years old. The most common diagnoses for both “not indicated” and “maybe indicated” antibiotic prescriptions were ARTI.

A 2015 meta-analysis showed that women in the 16–54 age group receive significantly higher number of prescriptions of cephalosporins and macrolides in primary care than men do, presumed by the authors to be related to the higher prescriptions in respiratory tract infections as the fluoroquinolones prescribed for primarily urinary tract infections remained balanced between the genders. ^{Reference Schröder, Sommer, Gladstone, Foschi, Hellman and Evengard5} Women have more acute visits for respiratory infection than men, despite men having more risk factors for chronic respiratory conditions related to social behaviors (e.g. smokers, illicit drug user, regular and heavy alcohol use) which put them at risk for acute infections. ^{Reference Barlam, Morgan, Wetzler, Christiansen and Drainoni6,Reference Pinkhasov, Wong, Kashanian, Lee, Samadi and Pinkhasov7} While women received more antibiotic prescriptions in our data (3:2 ratio), there was more appropriate antibiotic prescribing for diagnoses in males; this should be further researched with considerations for the differences in health-seeking behaviors and comorbid conditions.

In the 2010 US Census data, 33.7% of the population of South Carolina resided in a rural area. ⁸ Rural areas may be a particularly difficult area for antimicrobial stewardship. Rural visits, particularly in the southern United States, are a significant contributor to inappropriate antimicrobial prescriptions. ^{Reference Winders, Royer, Younas, Justo, Bookstaver and Weissman3} In a 2019 survey, 10% of providers in rural areas had not heard of antimicrobial stewardship but were generally (84.3%) interested in learning more and acknowledged that antibiotics are overprescribed and inappropriate use leads to resistance. ^{Reference Kufel, Mastro, Mogle, Williams, Jester and Snyder9} Providers who saw ≥50 patients a week felt more pressured to prescribe antibiotics, as well as if the visit were ≤20 minutes. Only 42.9% of prescribers selected that 90-98% rhinosinusitis are viral, with only 5.7% recommending supportive care without antibiotics; viral sinusitis was one of the most common “not indicated” diagnoses in this analysis. ^{Reference Kufel, Mastro, Mogle, Williams, Jester and Snyder9}

Behavioral and unmeasured factors, such as patient, prescriber, health care system, health care payer, and local/state policies, may impact prescribing rates. Additional factors such as the setting (e.g., urgent care or emergency department) or patient/provider relationship also likely play a significant role in prescribing patterns. ^{Reference Bizune, Tsay, Palms, King, Bartoces and Link-Gelles10} In an analysis of patient, physician, and practice characteristics, it was noted that family medicine providers, female gender, and self-report race/ethnicity of white or Hispanic were significantly associated with inappropriate antibiotic prescribing around ARTI. ^{Reference Barlam, Morgan, Wetzler, Christiansen and Drainoni6} They also noted that a group with low prescribers was likely to have overall low prescribers, whereas groups with high prescribers were 1.3 times more likely to have a second high prescriber. ^{Reference Barlam, Morgan, Wetzler, Christiansen and Drainoni6} In a study where patient characteristics were similar, it is the providers who had significant variability in prescribing. ^{Reference Shi, Barnett, Jena, Ray, Fox and Mehrotra11} Patients who saw high prescribers received 3.0 more fills per 100 people or 14.6% more fills over the subsequent year, compared with those seen by low prescribers. ^{Reference Shi, Barnett, Jena, Ray, Fox and Mehrotra11} The patients with prior fills were also more likely to have repeat upper respiratory tract visits in the subsequent year, with a relative increase in prescribing across quartiles at these subsequent visits at 2.8%. ^{Reference Shi, Barnett, Jena, Ray, Fox and Mehrotra11} Diagnoses were also different in high prescribers compared to low prescribers; the high prescribers tended to use diagnoses where antibiotics may be indicated compared to low prescribers using more viral conditions where antibiotics are not indicated. ^{Reference Shi, Barnett, Jena, Ray, Fox and Mehrotra11,Reference Martinez, Rood and Rothberg12}

This study has several limitations. Extrapolating from coding and claims data is limited in assessing accuracy of the data, and the provider may have coded the diagnoses inaccurately. ^{Reference Martinez, Rood and Rothberg12} Additionally, prescribers may have intentionally coded inaccurately to avoid scrutiny of inappropriate prescribing. There may be limited ability to generalize these data to other states or geographic regions. The ≥ 65-year-old cohort was excluded from analysis due to inability to capture them in this data set (only 1.0% initial data), which limits generalizability into that population. There was also ongoing national and international work to increase awareness of antimicrobial stewardship during the period of this study, which may confound results. Ecologic fallacy is also a concern in using administrative data such as this.

While the reduction in inappropriate antibiotic prescribing is encouraging, there is still significant work to achieve the goals of antimicrobial stewardship in the ambulatory setting.