Discussion

Our study is the first to evaluate the accuracy of a required indication field when prescribing antibiotics in the outpatient setting. Overall, 92.5% of selected indications were accurate and matched the infectious syndrome on chart documentation. Accuracy was similarly high among the ED and other outpatient settings, and these indication data can serve as a foundation for benchmarking and systematic stewardship interventions.

Through the course of this review, we identified ways to optimize the indication field listing. Half of the indication inaccuracies were due to prescribers selecting “genital tract infection” instead of “urinary tract infection” for patients with UTIs. Therefore, we have eliminated “genital tract infection” and replaced it with “sexually transmitted infection” to further direct prescribers to the appropriate choice. To provide additional actionable data, we also changed “urinary tract infection” to “UTI, pyelonephritis” and “UTI, cystitis,” and we split “otitis/sinusitis” into 2 unique headings. Furthermore, because the “surgical site infection” indication was selected for only 1 patient and because these infections can be captured by the “skin and soft-tissue infection” indication, we eliminated the former indication.

Prior studies have evaluated the accuracy of the indication field for inpatient antibiotic orders. One of the earliest was conducted at a large academic medical center and evaluated indications entered via CPOE. ^{Reference Patel, Esterly and Scheetz6} These researchers reported that 43 (86%) of the 50 randomized CPOE indications were accurate, with most inaccurate designations being “other.” With the addition of an indication to account for the selection of “other,” it was estimated that accuracy would have improved to 92%. This validation study allowed for additional benchmarking and assessment institutional guideline adherence.

Heil et al ^{Reference Heil, Pineles and Mathur7} assessed the accuracy of prescriber-selected indications in a pediatric ICU and adult medicine step-down unit. This analysis evaluated the frequency at which antibiotic indications chosen by the prescriber matched the true indications 48–72 hours after the initial order. The study reported a 90% concordance between the provider selected indication and independent review by infectious disease physicians. Similar to our study, one of the highest rates of discordance was seen with genitourinary and urinary tract infections, but this finding was not further elaborated.

In another inpatient medical center, Timmons et al ^{Reference Timmons, Townsend and McKenzie8} reported a method for monitoring prescribing patterns by generating indication lists for selected broad-spectrum antibiotics. Indication matching was determined by ensuring the chart-documented diagnosis mirrored the indication in the order. The matching rate was 74%–80%. Notably, prescribers chose “other” for 41% of orders, but in most cases the true indication was not available on the prepopulated indication list. Timmons et al attempted to characterize clinical appropriateness of prescribing as well, and they reported that 13.6% of antibiotic orders were inappropriate. This study sheds light on the scope of potential opportunities to assess clinical appropriateness with validated indication fields.

A recent study by van den Broek et al ^{Reference van den Broek, Beishuizen and Haak9} focused on respiratory tract infection and UTIs in 3 hospitals in the Netherlands. Similar to our study, prescription indications and chart notes were manually screened, and accuracy ranged from 78.2% to 96.7%.

Through our EHR, we can query specific indication fields used and key elements of the prescription including dose, duration, prescriber, clinic, and other information. These data, when combined with our now-validated indication fields can facilitate benchmarking of outpatient antibiotic use and can serve as a foundation from which to investigate opportunities to improve care. Stratifying these data by prescriber and/or department can shed light on prescribing practices amenable to systematic intervention or provider comparisons. Some system-oriented solutions may include generating indication-based order sets, review and modification of specialty-specific preference lists, modification of order defaults, and implementation of clinical decision support. More targeted antibiotic stewardship interventions may include educational efforts to address inappropriate prescribing patterns for specific prescribers or indications.

Generalizability across institutions is a limitation of this single-center study; additional data are needed from other outpatient centers. Additionally, appropriate use of the indication fields may change over time, which may warrant routine monitoring of this tool. Notably, not all outpatient antibiotic orders have an associated note. Although we did not encounter this within our study, scenarios in which providers enter orders without EHR documentation may occur.

In this retrospective quality assurance project, outpatient prescribers appropriately utilized the mandatory indication function in the CPOE. This project, along with the aforementioned studies, support the importance of a required indication field because it can inform quality care. The indication field offers a method to standardize and support reliable and efficient assessments of antibiotic use. This method provides institutions the opportunity to identify inappropriate antibiotic use and to prompt changes in prescribing practice. Other outpatient centers are encouraged to implement mandatory indication fields for antibiotic orders.