Methods

Intervention

The CDS tool was integrated into the CPOE workflow for inpatient urinalysis alone (UA), urinalysis-with-reflex-to-culture (UARC), and urine culture alone (UCx), requiring providers to select a clinical indication prior to order completion. The CDS presented educational messaging on inappropriate indications for urine culturing but did not trigger indication-specific advice or warnings. For orders not meeting criteria for UARC or UCx, UA was available as an alternative. Indications were derived from institutional guidance aligned with national recommendations (Figure 1). Prior to implementation, clinicians were notified through a systemwide email that included screenshots of the CDS orders and ordering guidance, with targeted emails to department chairs to facilitate distribution.

Figure 1.

Clinical decision support tool interface and indication selection for inpatient urine testing.

Figure 1. Long description

A table detailing urine testing guidelines and criteria for hospitalized patients. The table has multiple sections including Information, Urethral Catheter Status, Frequency, Indicate reason for Urine Testing, Only order a urine specimen to evaluate for infection when the patient meets one of the following indications, Testing to be performed, Comments, Specimen Type, Specimen Source, and Add-on. The Information section warns about the detection of non-harmful bacteria and the risks of unnecessary antibiotic use. It lists circumstances where urine testing is not warranted. The Urethral Catheter Status indicates no active urethral catheter. The Frequency section allows selection of testing frequency and date. The Indicate reason for Urine Testing section lists various reasons such as Urine testing in a patient with possible UTI, Urinalysis only, Asymptomatic bacteriuria screening, and others. The Only order a urine specimen section lists indications for urine testing including urinary signs or symptoms, septic shock, spinal cord injury, inability to verbalize symptoms, and suspected UTI with neutropenia. The Testing to be performed section specifies Urinalysis Reflex Urine Culture. The Comments section allows adding comments. The Specimen Type and Source sections specify the type and source of the urine specimen. The Add-on section indicates no add-on specimen found.

Results

During the study period, 124,762 urine tests were ordered across 225,614 patient-days. There were no significant preintervention trends for UA, UARC, or UCx.

ITS analysis demonstrated an immediate increase in UA utilization after CDS implementation (41.4; 95% CI, 20.1 to 62.8), with no significant postintervention slope change (−2.8; 95% CI, −5.9 to 0.3). In contrast, UARC utilization showed an immediate decrease (−62.7; 95% CI, -−85.4 to −39.9), followed by a modest increase in postintervention slope (4.1; 95% CI, 0.8 to 7.4). UCx utilization also demonstrated an immediate decrease (−1.8; 95% CI, −2.9 to −0.7), with no significant postintervention slope change(−0.03; 95% CI, −0.2 to 0.1) (Figure 2; Supplemental Table 1).

Figure 2.

Monthly urine testing rates before and after implementation of clinical decision support tool at Tufts Medicine, August 2023–July 2025.

Of the 153 reviewed UARC orders, 46.4% originated in the emergency department (ED), 41.2% in intensive care units, and 12.4% on inpatient floor units. Documentation supported appropriate testing in 9.9% of ED encounters, 19.2% of inpatient floor encounters, and 27.8% of intensive care unit encounters. Antibiotics for presumed UTI were initiated in 29 patients (19%), whereas only 9 (5.9%) had documentation consistent with symptomatic infection.

Discussion

Implementation of an indication-based CDS tool for inpatient urine testing was associated with immediate, intended reductions in UARC and UCx ordering. However, these effects were incomplete and not fully sustained. UARC utilization partially rebounded, and UA ordering increased immediately after implementation, suggesting substitution toward less restricted testing rather than a true reduction in low-value testing. Indication-based CDS alone was therefore insufficient to produce sustained, guideline-concordant practices in our setting.

Several factors may explain these results. First, our intervention functioned as a required prompt without a hard stop, and softer interventions are more vulnerable to bypass or workaround behavior, ^{Reference Powers, Shiffman, Melnick, Hickner and Sharifi4,Reference Chaparro, Beus and Dziorny5} potentially explaining the immediate reduction in culture-based testing without consistent improvement in appropriateness. Second, the rise in UA ordering suggests substitution to a less restricted test. For clinicians remain concerned about missing infection, indication-based prompts may shift ordering without changing the underlying diagnostic threshold, consistent with qualitative work identifying diagnostic uncertainty, fear of undertreatment, and ingrained practice norms as barriers to stewardship. ^{Reference Durkin, Schmitz, Hsueh, Ney and Politi6} Third, required CDS fields may also fail to capture clinical reasoning when the tool is perceived as administrative rather than a meaningful decision aid, as alert fatigue promotes routinized selection. ^{Reference Hussain, Reynolds and Zheng7,Reference Olakotan and Yusof8} In this context, requiring indication selection does not ensure accurate indication selection.

Our audit findings reinforce this. Documentation-supported appropriateness remained low across inpatient settings, particularly in the ED, where prior work has identified ED ordering as a strong predictor of inappropriate urine culturing. ^{Reference Foong, Munigala, Jackups, Yarbrough, Burnham and Warren9} Antibiotics for presumed UTI were often initiated without documented symptoms. These findings suggest that interventions targeting test ordering alone may have limited downstream effect unless paired with complementary strategies aimed at interpretation and prescribing, such as prospective audit and feedback. ^{Reference Morgan, Malani and Diekema2,Reference Barlam, Cosgrove and Abbo10}

This study has several limitations. It was a single healthcare system study, and chart review was limited to a sample of 153 postimplementation UARC orders at Tufts Medical Center, which may limit generalizability. Although our ITS design uses each period as its own control, accounting for stable or gradually changing population characteristics, we did not tabulate baseline characteristics across periods, and interrater agreement for the chart review was not formally assessed. Appropriateness assessments were based on chart documentation and may have unrecorded clinical reasoning. Lastly, because review was performed only after implementation, we could not directly compare pre and postintervention appropriateness.

In conclusion, implementation of an indication-based CDS tool produced immediate reductions in culture-based urine testing, offset by test substitution, limited sustainability, low documentation-supported appropriateness, and antibiotic treatment without documented symptoms. These findings suggest that indication-based CDS, particularly without a hard stop or complementary stewardship measures, may be insufficient to achieve sustained improvement in urine testing and downstream prescribing. Multifaceted approaches combining diagnostic stewardship with behavioral, workflow, and prescribing-focused interventions are likely needed.