Urine cultures are commonly ordered among hospitalized patients, and they are often ordered repeatedly, leading to laboratory overutilization and increased cost. Reference Trautner1,Reference Foong, Munigala and Jackups2 Positive urine cultures are a major driver of unnecessary antibiotic use. Reference Hartley, Kuhn and Valley3–Reference Trautner, Bhimani and Amspoker5 Previous studies have used multiple methods to prevent unnecessary urine testing including education, reflex urine culture cancellation and 2-step urine-culture ordering, Reference Trautner, Grigoryan and Petersen6,Reference Gupta, Hooton and Naber7 Prior interventions at our institution to prevent unnecessary urine cultures, including modification of “reflex” urine-culture criteria for specimens submitted for urinalysis and modifications of emergency department and inpatient electronic order sets, decreased urine cultures performed by 46.6% and 45%, respectively. Reference Munigala, Jackups and Poirier8,Reference Munigala, Rojek, Wood, Yarbrough, Jackups, Burnham and Warren9 However, data on the effect of changes in electronic order sets and its role on repeated urine testing practices are limited.
We evaluated the impact of duplicate order alerts for urine culture orders in a computer physician order entry (CPOE) system on urine culturing practices in a large urban, academic medical center (Supplementary Fig. 1 online).
Methods
We included patients admitted to Barnes-Jewish Hospital (BJH) between June 2, 2018, and December 31, 2021, who had ≥1 urine culture ordered during their hospitalization. Patients admitted during the study period but who did not have a urine culture ordered during their stay were excluded.
Intervention
To prevent providers from ordering duplicate urine cultures, a notice that a laboratory-generated urine culture was in process, based upon urinalysis results (ie, as part of a urinalysis reflexed to urine culture order), began to appear in the urinalysis results section of the electronic medical record (EMR) starting on April 25, 2019. A second intervention was initiated on October 28, 2020. If a provider attempted to order a urine culture and a culture had been collected and processed within the past 72 hours, a duplicate order alert appeared that displayed the status of the in-lab culture. Providers had the option of overriding the alert (Supplementary Fig. 2 online). E-mail notification was sent to providers prior to intervention implementation. We compared the daily number of inpatient urine cultures performed, the daily number of positive urine cultures, and daily isolated urine culture count before and after the changes: period 1 (June 2, 2018, to April 24, 2019) versus period 2 (April 25, 2019, to October 28, 2020) versus period 3 (October 29, 2020, to December 31, 2021).
Patient and laboratory data were abstracted from the hospital informatics database. Data included patient demographics, urine culture results including specimen type and order type (reflex, based upon urinalysis results, vs standalone or isolated). For patients with multiple urine cultures during an admission, each sample was treated as an independent observation. This study was approved by the Washington University Human Research Protection Office.
Results
During the 42-month period, 104,965 urine cultures were performed at BJH, of which 31,134 (29.7%) were from inpatient locations (mean, 24 cultures per day). Of the inpatient urine cultures, 7,661 (24.6%) were abnormal: 7,049 (22.6%) were reported as positive and 865 (2.8%) were reported as contaminated. Most of the cultures (62.1%, n = 19,325) were ordered and performed as part of a reflex algorithm, based upon abnormal urinalysis results, and 33.0% (n = 10,259)were isolated or were standalone cultures. The most common indication selected by providers for isolated urine cultures was a recent positive urinalysis (43.3%). Significant reductions in the mean daily urine cultures per 100 inpatients (3.34 vs 2.34; P < .001; 3.34 vs 2.22; P < .001), mean daily positive urine cultures per 100 inpatients (0.80 vs 0.49; P < .001; 0.80 vs .52; P < .001), and mean daily isolated urine cultures per 100 inpatients (1.05 vs 0.86; P < .001; 1.05 vs 0.66; P < .001) were noted after EMR notification of laboratory-generated urine-culture orders (period 2) on April 25, 2019, and after repeated urine culture alerts (period 3) went live on October 28, 2020 (reference period, June 2, 2018, to April 24, 2019). The median duration of time to repeated urine culture significantly increased during the intervention periods: 24 hours versus 96 hours (P < .001) and 24 hours versus 120 hours (P < .001). No significant differences in mean daily, isolated, and positive urine cultures were observed when comparing period 2 with period 3. However, median time to repeated urine culture was significantly increased for period 3: 96 hours versus 120 hours (P < .001) (Table 1). Our intervention resulted in $83,194 savings in laboratory costs for urine cultures avoided, based on the Medicare Clinical Laboratory Fee Schedule using a national median Medicare payment rate of $16.16 per urine culture. 10
Table 1. Urine Culture Testing Before and After Changes in the Hospital CPOE System
Note. Average daily urine cultures for period 1, period 2 and period 3 were 29.2, 21.8 and 22.2 cultures, respectively. Average daily positive urine cultures for Period 1, Period 2 and Period 3 were 7.0, 4.5 and 5.2 cultures, respectively. Laboratory-generated urine culture orders to appear on the hospital computer physician order entry (CPOE) system went live on April 25, 2019, and repeated urine culture alert went live on October 28, 2020. The hospital used EPIC (Epic Systems, Verona, WI) as its CPOE system.
a P-value comparison for period 1 vs period 2.
b P-value comparison for period 2 vs period 3.
c Tested by autoregressive integrated moving average (ARIMA) analysis.
d Logistic regression analysis with June 2, 2018, to April 24, 2019, as a reference period.
e Includes specimen collection by suprapubic aspiration, kidney aspiration, percutaneous nephrostomy tube placement, ileal loop, and cystoscopy.
Discussion
We observed a significant reduction in the mean daily urine cultures and positive urine cultures per 100 inpatients and a significant increase in the median duration of time to repeated urine culture in the postintervention periods versus the preintervention period. During period 2, the urine cultures per 100 inpatient days and the proportion of positive cultures decreased and the proportion of isolated cultures increased. This finding suggests that urine cultures were being ordered on patients with a lower pretest probability for a positive culture during this period.
This study had several limitations. The study had a retrospective design. These results may not be generalizable to facilities with different CPOE systems. The metric of daily cultures per 100 inpatients was chosen to adjust for wide swings in hospital census seen because of the COVID-19 pandemic; however, we cannot rule out unrelated temporal changes, which might have influenced our results.
In conclusion, relatively minor changes in the order notification system could have significant impact on provider ordering practices. The CPOE system plays an important role in diagnostic stewardship, and continued optimization, along with antimicrobial stewardship efforts, can reduce the incidence of unnecessary urine cultures as well as lower healthcare costs and improve the use of diagnostic tests. Ongoing monitoring and surveillance is required to identify areas of improvement and to prevent unintended consequences.
Supplementary material
For supplementary material accompanying this paper visit https://doi.org/10.1017/ice.2022.283
