Optimizing Provider Test Ordering and Patient Outcomes Through Best Practice Alerts and Doctorate in Clinical Laboratory Sciences (DCLS) Consultation for Urine Cultures

1. Introduction

Concerns regarding the emergence of antibiotic-resistant organisms remain at the forefront of the healthcare industry. The Centers for Disease Control and Prevention (CDC) have labeled it as a global threat stating that more than 2.8 million antibiotic-resistant infections occur in the US each year, and more than 35,000 people die as a result [1]. Hospitals have been encouraged to implement robust antimicrobial stewardship committees to help mitigate the overuse of antibiotics, thereby, reducing the emergence of multi-drug-resistant organisms. These factors not only lead to increased lengths of stay in the hospital, but also increase morbidity and mortality. When antibiotic use is inappropriate, the patient is exposed to unnecessary risk and the risk/benefit algorithm is unjustified.

For several years, programs such as the Choosing Wisely Campaign and guidelines from the American Society for Microbiology (ASM), Infectious Disease Society of American (IDSA) and the Society for Healthcare Epidemiology of America (SHEA) promoted the importance of antimicrobial stewardship for improved patient outcomes. These programs are designed to improve patient outcomes, reduce adverse events such as C. difficile infection (CDI), and reduce antimicrobial resistance specially to targeted antibiotics [2,3,4]. Antibiotic resistance is a serious and increasing worldwide threat to global public health and prolonged exposure to antibiotics decrease normal gut microbiota and increase susceptibility to gastrointestinal pathogens even though they play a critical role in fighting infections, long-term or overuse can cause patient harm [5,6].

Isolation of bacteria in culture is often used to diagnose infections and provide a basis for prescribing antimicrobial therapy. In clinical practice, however, it is common to isolate bacteria that may be colonizing or not causing an infection. This is one reason laboratories are encouraged to have well-defined specimen collection protocols in place. For example, blood culture collection recommendations have very strict cleansing techniques to remove the colonizing bacteria present on the skin prior to collection. For urine cultures, suprapubic aspiration, straight catheter, or mid-stream collections are used to help improve diagnostic accuracy [7]. Even when stringent collection techniques are employed, bacteria may still be isolated in culture which leads to diagnostic uncertainty. The definition of asymptomatic bacteriuria (ASB) has evolved over time and currently clinical practice guidelines from IDSA define asymptomatic bacteriuria as the quantitative isolation of ≥100,000 colony-forming units (CFU) per milliliter of bacteria in an appropriately collected urine specimen from a patient without urinary tract infection signs or symptoms [8]. Infectious Disease Society of America (IDSA), and the Society for Healthcare Epidemiology of American (SHEA) discourage treatment of asymptomatic bacteriuria (ASB) in certain patient populations stating that treatment has no clinical benefit, does not improve morbidity or mortality, and contributes to antibiotic overuse.

Additionally, culturing urine on asymptomatic patients may lead to inappropriate reporting of catheter-associated urinary tract infections (CAUTI) [9]. CAUTI is considered a preventable disease and hospital acquired condition and is a required reportable to Center for Medicare and Medicaid Services (CMS) which may ultimately be tied back to compensation for acute care hospitals [10].

Test selection and interpretation play a huge role in preventing diagnostic error. According to Laposata & Dighe, improved result interpretation leads to quicker and more accurate diagnosis [11]. When clinicians order laboratory tests such as urinalysis and urine cultures inappropriately, a pre-pre analytical error, the information provided by the results can lead to the inappropriate use of antibiotics and the emergence of resistant pathogens when ASB is present. Therefore, urine cultures and urinalyses that reflex a urine culture should not be ordered for certain patient populations when urinary symptoms, such as urgency, frequency, retention, dysuria, suprapubic pain, flank pain, pelvic discomfort, or acute hematuria, are absent. This identifies a gap in the ordering practices of providers which was addressed in this project using an LIS intervention at order entry and a follow-up DCLS consultation for over- and under-utilization.

The purpose of this study is to determine if an intervention at order entry coupled with a DCLS laboratory consultation for over- or under-utilization of urine cultures and urinalyses would reduce unnecessary laboratory testing. By determining over-utilization, the DCLS-led intervention could prevent the inappropriate use of antibiotics for ASB and, therefore, reduce the adverse effects of antibiotic therapy. Likewise, the DCLS-led intervention for under-utilization could alert clinicians of a potential need for urine cultures and antibiotic therapy when symptomatic cases are overlooked improving patient outcomes. Additionally, highlighting the inappropriate ordering practices could educate clinicians and modify his or her ordering behavior on future patients. The overall outcome of the project is to reduce inappropriate urine culture orders which could indirectly reduce the inappropriate use of antibiotics and improve patient outcomes.

2. Materials and Methods

The study is a quasi-experimental quantitative analysis that involved both retrospective and prospective chart reviews of patients with urinalysis and/or urine culture orders at a community hospital. The hospital, a 293-bed, level II trauma and primary stroke center in North Central Texas, serves a nine-county area. The study included non-pregnant adult patients (18 years or older) with urinalysis and/or urine culture orders from July 2021 to September 2022. Patients excluded from the study include pregnant females, patients undergoing urologic procedures, ICU admissions, dialysis patients, cancer patients, patients with foley catheters, and those with symptoms that could not be assessed at order entry. The timeline was July 2021 to September 2022, with samples from July 12, 2021, to September 30, 2021, as the retrospective group and samples from July 12, 2022, to September 30, 2022, as the prospective or post-intervention group.

The intervention in this study was a Best Practice Advisory (BPA) alert integrated into the EPIC electronic health record (EHR). This alert notified providers via a pop-up message of the possibility of an inappropriate order. An interprofessional team, including a DCLS laboratory professional, a pathologist, an infectious disease practitioner, a physician champion, an application services coordinator, and an application services analyst, collaboratively designed the BPA using the Choosing Wisely Recommendations from ASM and IDSA guidelines. This study was developed as a quality improvement request from the Antimicrobial Stewardship Team and received approval from the hospital’s Department of Compliance and the Project Governance Committee. It was also approved as an exempt study by the University of Texas Medical Branch Internal Review Board.

2.1. Data Collection

Retrospective data was manually collected and included patient demographics (age, race, sex, and ethnicity), provider types, patient location, antibiotic therapy at admission and discharge, test orders (urinalysis reflex to urine culture, urinalysis only & urine culture only), laboratory results for the urine culture when applicable, patient’s chief complaint and days since last admission (readmission is ≤30 days). For prospective review, data was extracted using a Business Intelligence SQL report available on the hospital server. This report included all variables from the retrospective review (patient demographics, provider types etc..) and added urinary symptoms, and BPA actions. Urinary symptoms were defined as the presence of one or more of the following: urgency, frequency, retention, dysuria, suprapubic pain, flank pain, pelvic discomfort and/or acute hematuria [12].

Providers for which the BPA triggered at least five times were reviewed. For those providers who bypassed the BPA and continued with the original order more than 80% of the time, direct communication was attempted using both the Epic chat feature and email. Based on the BPA report, the DCLS provided consultations to practitioners when appropriate.

2.2. Data Analysis

Pre- and post-intervention practices were evaluated across several variables (over-/under-utilization, appropriate), facility (provider type, patient location area, etc.), test order type (order, culture reflex) and symptom (symptomatic/asymptomatic). A chi-square statistical analysis was performed using SPSS software, with P values < 0.05 indicating statistical significance.

3. Results

In our study, we reviewed a total of 6,372 patients, 3,408 in the retrospective review group while 2,964 in the prospective review group. Many of the patients were female, White or Caucasian with a mean age of 50.6 years (range, 18-101 y) in the retrospective review and 53.5 years (range, 18-103 y) for the prospective review (Table 1).

Regarding urinalyses test orders, our study identified 60% (n=2053) of the test orders were inappropriate which after BPA intervention reduced to 20% (n=591). Chi-square test was statistically significant Χ² (1, N=6372) 1060.609, p <.001, Φ=.408, and indicated that the intervention had a moderate effect on urinalyses test orders. For inappropriate orders, overutilization accounted for 59% (n=2021) of orders in the retrospective review and only 12% (n=365) for the prospective review while underutilization percentages increased slightly from 1% (n=32) to 8% (n=226). Chi-square statistical analysis determined that a significant difference in the inappropriate test orders (overutilization, underutilization vs. appropriate) with a moderate to large effect between the two reviews, Χ² (2, N=6372) 1549.791, p <.001, Φ=.493. Lastly, analysis of data regarding types of test orders, from the 3,408 total test orders, 3,078 (90.3%) were urinalyses reflex to urine culture, 204 (6.0%) urinalyses only (no culture reflex), and 126 (3.7%) urine culture only for the retrospective review. For the prospective review, from the 2,964 total test orders, 2080 (70.2%) were urinalyses reflex to urine culture, 810 (27.3%) urinalyses only (no culture reflex), and 74 (2.5%) urine culture only.

When analyzing urinalyses reflex to urine culture for test order appropriateness, there was a significant difference between the two reviews with a moderate effect size, Χ² (1, N=5158) 1019.458, p <.001, Φ=.445. Neither group had test orders which demonstrated underutilization, however, overutilization accounted for 62% (n=1913) of the retrospective review and after intervention dropped to 17% (n=357) in the prospective review. Regarding urinalysis only test order, there was a significant difference between the two reviews with a small to medium effect size, Χ² (2, N=1014) 61.744, p <.001, Φ=.247. However, post hoc tests demonstrated there was only a small to no effect when comparing appropriate and underutilization which accounted for most of the data in this test group, Χ² (1, N=1001) 9.843, p <.005, Φ=.099. The statistical differences occurred when including overutilization (assessment variable) which was only present in the retrospective data (n=13). Of note, there were four times as many test orders for urinalysis only in the prospective data when compared to the retrospective data. For urine cultures ordered alone, there was a significant difference in the appropriate test orders between the two reviews demonstrating a large effect size, Χ² (1, N=200) 74.686, p <.001, Φ=.611. After intervention, appropriate ordering practices increased from 25% (n=31) to 88% (n=65) (Table 2).

Furthermore, data regarding appropriateness of test orders in asymptomatic patients was assessed based on presenting clinical symptoms. When comparing appropriate test orders (overutilization vs appropriate; no underutilization for this test group) for asymptomatic patients, a significant difference was present with a large effect size, Χ² (1, N=3129) 1079.144, p <.001, Φ=.588. Our study identified inappropriate ordering practices which were all categorized as overutilization accounted for 93% (n=2021) of the retrospective data which dropped to 38% (n=366) after the intervention. For urinalyses that reflex to a urine culture by symptoms, there was significant difference between the two reviews which demonstrated a moderate effect, Χ² (1, N=1964) 318.554, p <.001, Φ=.403. In asymptomatic patients, urinalyses that reflex to urine culture decreased from 51% (n=501) in the retrospective review (n=990) to 13% (n=127) after intervention.

Finally, for asymptomatic patients, the presence of antibiotic therapy at discharge dropped from 54% (n=509) to 25% (n=219) after intervention. Statistical analysis of the two review groups showed a significant difference in antibiotic therapy for asymptomatic patients at discharge with a moderate effect size, Χ² (1, N=1841) 171.377, p <.001, Φ=.305 (Table 3).

Next, the data was assessed based on the provider type. Physicians included both MDs and DOs while advanced practice providers included physician assistants and nurse practitioners. Differences in pre- and post-intervention ordering practices were statistically significant for both provider types. For physicians, a medium to large effect size was calculated, Χ² (2, N=5310) 1130.699, p <.001, Φ=.461; appropriate orders accounted for 42% of the retrospective data and 80.5% of the prospective data (Table 4). The intervention demonstrated a large effect for advanced practice providers, Χ² (2, N=1062) 1079.144, p <.001, Φ=.646, who used appropriate orders only 27.6% of the time on the retrospective review, but 78.0% on the prospective review. Overutilization decreased for both provider types: physicians went from 57% overutilization down to 13% while advanced practitioners went from 72% down to 9%. Unexpectedly, underutilization increased from 1% to 7% for physicians and 0 to 13% for advanced practitioners.

4. Discussion

The objective of this quality improvement project focused on the question: Will an intervention at order entry coupled with a DCLS consultation for inappropriate orders of urinalyses and urine cultures reduce unnecessary laboratory testing of asymptomatic patients? Our study established the presence of over and under-utilization of urinalyses and urine cultures and after the intervention (BPA alerts and DCLS consults), we saw a reduction in inappropriate test orders from 60% (n=2053) to 20% (n=592). Similar results were seen in studies where they recommend interventions at order entry to prevent overutilization of cultures and require symptom documentation including inappropriate order alerts to providers to be effective in achieving improved provider practices and laboratory test utilization [9,13].

Regarding inappropriate urinalysis test orders, the most common overutilized test order was a urinalysis with culture reflex which supports the findings in similar study and the need for diagnostic stewardship [9]. Overutilization of this specific test accounted for 62% (n=1913) within the retrospective review and after intervention reduced to 17% (n=357). One reason this test was ordered most often is because it automatically reflexes to urine culture when indicated and the indications are based on urinalysis results rather than the presence of urinary symptoms. Adding the BPA intervention highlighted the need to consider symptoms along with the urinalysis order thereby reducing unnecessary urine culture [9,13].

For urinalysis only, overutilization of urinalysis only tests went from 6% (n=13) to 0% after intervention while underutilization increased from 16% (n=32) to 28% (n=225). Our study showed that the urinalysis only test was least affected by the BPA. Appropriate orders in the urinalysis only were at 78% (n=159) and decreased to 72% (n=585) after the intervention, which was unexpected, but supported by Sullivan, et al. [9]. This is likely because urinalyses that automatically reflex to urine culture were the predominant order used for physician order sets. When only urinalysis was ordered, the physicians were aware of the exact urinalysis ordered and likely did not want urine culture results. This would have been a conscious additional step and likely not part of the routine order set workflow for the retrospective study. For the prospective review, there was a four-fold increase in the urinalysis only orders indicating awareness of the need to reduce urinalysis with urine culture reflex orders on the asymptomatic population. For urine culture only, after intervention, overutilization of this test type dropped to 12% (n=9) from 75% (n=95) and saw an increase in appropriate test orders from 25% (n=31) to 88% (n=65) highlighting the effectiveness of BPA and DCLS consult intervention. The reduction in inappropriate urine culture orders between the two review periods showed a large effect highlighting ASB awareness.

Next, we will discuss the effect of the BPA on the treatment of ASB since recognizing test selection errors (pre-pre-and post-post-) and the need for test interpretations play a huge role in prevention of diagnostic errors [11]. A pre-preanalytical error occur when clinicians order inappropriate tests or fail to order appropriate tests and, when applied to ordering practices of urine cultures and urinalyses in the hospital setting, this would include the ordering of urine cultures on non-pregnant adult patients who are asymptomatic. In our study, for asymptomatic population, inappropriate test orders were found in 93% of the patients in the retrospective review and, due to the intervention, decreased to 38% demonstrating a large effect and reduction in diagnostic errors.

As discussed previously, results from inappropriate urine cultures led to treatment of ASB and improper reporting of CAUTIs. To demonstrate the significant reduction, we looked at urine culture reflexes compared to symptoms present which confirmed that less culture results were reported for the asymptomatic population in the post-intervention data, dropping from 51% to only 13%. Furthermore, we compared initiation of antibiotic therapy at discharge in the presence of symptoms and a significant difference with a moderate effect size was identified. This validates the reduction in treatment of ASB in the post-intervention population – reducing ASB treatment by more than 55%. Overall, this intervention led to a significant reduction in the treatment of ASB which has been linked to the overuse of antibiotics leading to the development of antimicrobial resistance.

Additional studies are needed to assess whether the reduction in inappropriate test orders leads to the reduction of adverse events such a C. difficile infection and readmissions. Our study documented the days since the patient’s last admission but did not go forward to determine if the patient had another admission in the next 30 days. This information would be useful to determine if readmission rates were affected with the reduction of inappropriate orders. Our study’s findings highlight the importance of using technological developments to reduce errors and provide easy and rapid access to test interpretation [14]. The study further details that improvements in test request appropriateness and results interpretation should occur due to implemented initiatives that seek to improve knowledge about laboratory tests and the correct interpretation of test results. Therefore, reducing pre-preanalytical errors by improving knowledge at the point of care using technology presents a distinct and readily available opportunity for process improvement.

Lastly, all provider types showed significant improvement in appropriate orders. For both provider types, overutilization of lab tests dropped from 59.3% (n=2021) to 12.3% (n=366) while appropriate test orders improved from 39.8% (n=1355) to 80.1% (n=2373). One unexpected significant finding, although there was no effect size, was the increase in underutilization. A reason for this could be the lack of physician understanding of the BPA as mentioned above. Specifically, one provider did not realize he or she should have been modifying the order rather than “accepting” or proceeding with the original order. Post-study follow-up of the specific provider’s BPA information shows a decrease in BPA bypasses from the 100% BPA bypass rate during the last month of the study.

An interesting finding from the medical staff meeting that supported findings from other studies was that physicians were already aware that they should not treat ASB, and agreed they needed something to reinforce what they already knew [15,16]. The best practice alert provided this reassurance and corrected many of the inappropriate ordering practices. Treatment of ASB, therefore, was not due to lack of education but due to providers having access to unnecessary culture information. When a provider was given unnecessary culture information, he or she tended to err on the side of caution and treat it [4]. By reducing overutilization of the urine culture, practitioners withheld antibiotic treatment as evidenced by the reduction of new antibiotic therapy at discharge in the asymptomatic population. Antimicrobial stewardship continues to play a huge role in the prevention of the development of multi-drug-resistant organisms and the reduction of other adverse events. Ongoing studies are necessary to learn the best way to reduce antibiotic usage while also reducing patient harm and adverse events. Education alone is not always the appropriate tool. Practitioners must learn to rely on the evidence-based practice and quit erring on the side of caution, especially when that caution leads to patient harm.

Further studies are needed to assess the differences in ordering practices of physicians receiving a consultation compared to those that did not. Finally, although extremely important to process improvement projects, practitioner feedback is severely lacking. Future studies are needed to determine the cause of this disconnect and to determine the best way to increase practitioner feedback.

Grol & Grimshaw acknowledge the existence of a gap between evidence and practice stating that education materials, such as continuing medical education sessions, alone may not be effective in implementing practice changes [15]. The article highlights common barriers and strategies, one specific barrier noted was that the clinical environment must be conducive to change. They concluded that change requires heavy preparation which includes involvement of appropriate people, evidence-based best practices, concrete descriptions of desired performance, defining appropriate indicators of success and thorough monitoring.

Clinical decision support (CDS) is a term used to describe a wide-array of tools available in the electronic health record (EHR) to support clinicians by providing information such as evidence-based clinical guidelines at the point of care [17]. Largely due to the Affordable Care Act mandate, this technology has been incorporated and readily available in most EHRs. One CDS tool available is the BPA which is a method of communication built within the EHR that notifies the provider of specified information through messages, storyboard alerts, and/or interruptive or passive alerts when defined criteria are met [18]. While it has been shown that CDS tools like the BPA have a positive impact on patient care, many barriers exist across healthcare facilities preventing the implementation. For CDS to be successful, one must get the right information to the right person in the right format through the right channel at the right point in the workflow naming these the “5 rights”. Along with the 5 rights BPA design, one must also have support from administration, clinicians, and information services and use reporting tools to enhance provider communication, evaluation, and follow up [16]. Another study supported the need for CDS; however, it highlighted the existence of alert fatigue as a barrier noting that alerts regarding inappropriate culture orders may be deemed as a nuisance by some nurses [19]. Technology alone is not enough, and monitoring quality indicators does not necessarily achieve improvement [20].

There were several limitations to our study. Antimicrobial stewardship practices and ongoing provider education may have made physicians more aware of appropriate ordering practices and proper indications for antibiotic prescribing and could be responsible for skewing the data. Also, antibiotics may be initiated for non-urinary indications which is not accounted for in this study. There was an instrumentation effect in this study. Determination of the presence of urinary symptoms on the prospective review was based on the provider answered order entry question while the retrospective review was based on the chief complaint and chart review. Additionally, this analysis assumes the practitioner selected the appropriate answer to the order entry question and does not account for inappropriate or wrong answers. Both of which could lead to inconsistency in symptom determination affecting the assessment of order appropriateness. Finally, our data was a convenience sampling collected at one specific hospital and may not be reflective of the entire patient population.

5. Conclusions

Our study reinforces the value of direct interventions in guiding appropriate ordering behaviors in community hospital settings. It highlights the importance of providing direct feedback to end users to ensure compliance and understanding of the intervention. Educating providers about appropriate orders at the point of order entry proved effective, highlighting the need for diagnostic stewardship. Post-intervention ordering practices improved significantly, reducing the number of inappropriate orders across all patient and provider types. Overall, this initiative led to significant reduction in the treatment of asymptomatic bacteriuria which has been linked to the overuse of antibiotic therapy.