Urinalysis (UA) commonly consists of physical, chemical, and microscopic evalsuations with increasing complexity, cost, and turnaround time. UA is also multidisciplinary in that it may be performed in the core lab or by point-of-care and may be a manual or automated assay. Bacteria causing urinary tract infections (UTIs) may further be identified and characterized by urine cultures in the microbiology lab.
To boost efficiency, many laboratories have implemented reflex testing approaches. If paired with carefully designed electronic order options and clinical decision-support systems, this strategy has the potential to optimize test utilization, improve result turnaround times, and reduce laboratory costs for reagents and labor.
Reflex testing generally takes two forms. In reflex-to-microscopic approaches, the laboratory first performs a chemical UA to detect abnormalities such as blood, protein, glucose, and indirect indicators of bacterial infection (e.g., leukocyte esterase and nitrite). Abnormal chemical UA results then trigger subsequent microscopic UA to look for cells, bacteria, yeast, casts, and crystals. Some laboratories also use a reflex-to-culture approach and only perform urine cultures if infection-relevant chemical or microscopic UA findings, or both, are detected.
Our hospital system’s leadership was interested in implementing a reflex-to-culture approach to reduce the reporting of clinically insignificant catheter-associated urinary tract infections (CAUTI), alleviate misinterpretation or over-interpretation of clinically insignificant positive culture results, and support antibiotic stewardship. However, there are currently no evidence-based guidelines on how to define UA reflex criteria or implement reflex approaches.
Although UA reflex approaches are now widely used and discussed in the literature, data involving modern automated UA methods is limited. While most studies that investigated reflex-to-culture criteria compared manual microscopy results to urine culture results, we were interested in implementing a compounded reflex approach—chemical UA with reflex-to-microscopic UA for general UA orders, and chemical UA with reflex-to-microscopic UA followed by reflex-to-culture for reflex culture UA orders. We also wanted to use an automated UA system as the primary UA method.
We performed our own retrospective study to determine how well chemical UA and microscopic UA results compared with each other and with urine culture results. Our goal was to estimate the significance of the diagnoses that would be missed and the numbers of microscopic UA and cultures that could be avoided by our proposed reflex UA approaches.
Using our proposed reflex-to-microscopic reflex criteria (hazy or cloudy appearance, positive hemoglobin, glucose ≥1,000 mg/dL, positive protein, positive leukocyte esterase, or positive nitrate), the chemical UA showed a sensitivity of 93%, specificity of 57%, and negative predictive value of 91% for positive microscopic UA results, defined as red blood cell or white blood cell (WBC) count ≥ 4/hpf or any detectable presence of bacteria (J Appl Lab Med 2020; doi:10.1093/jalm/jfaa011). A reflex-to-microscopic approach would have led to a 34% reduction in the number of microscopic UA performed, and the frequency of missed positive microscopic UA was 3%. Of the samples with urine culture results available (n=3,127), 6% were negative for all chemical UA criteria but had clinically significant positive urine cultures, indicating fairly good performance of negative chemical UA alone in ruling out culture-positive UTIs. Based on our data, along with support from the literature, we ultimately decided to implement a standardized reflex-to-microscopic approach using the aforementioned criteria and to use WBC count ≥ 10/hpf for implementing a new reflex-to-culture approach.
Standardizing the urine reflex testing workflow across our health system yielded many lively discussions about sample collection and transportation between laboratory areas. One major issue we had to tackle first was the urine specimen container. Some of our hospitals already used a urine collection kit with a boric acid preservative tube for culture, while others used standard urine collection cups. Because the UA would have to be performed and resulted prior to determining whether the urine culture should be started, all of our laboratories had to switch to the standard collection kit.
We also debated whether the preemptive urine culture tube should be held in the core lab or the microbiology lab, and how to alert the microbiology lab of the need to perform the culture. We decided that having a new specimen label for urine culture automatically printed in the microbiology lab upon a positive UA result would be a good trigger.
Laboratories setting up their own urine reflex workflows should keep their specific patient populations in mind. Importantly, complete UA and urine culture should typically be concurrently tested for pregnant women, neonates, and immunocompromised individuals regardless of chemical or microscopic UA results. Our laboratories designed a clinical decision support message (pop-up alert) on the reflex-to-culture order in the electronic order system to inform clinicians of which patient populations should have urine culture ordered directly instead. Similarly, the standalone urine culture order, if visible to providers, may include an alert indicating its appropriate criteria. Other laboratories have described having multiple order options for different patient populations, such as microscopic UA and urine culture reflexed simultaneously for neutropenic patients.
Overall, it is important to keep key clinical stakeholders involved: Our planning involved frequent communication with leadership in infectious diseases, infection prevention, hospital quality, and patient safety. Successful urine reflex testing workflows require a balance of stewardship enforcement with room for clinical judgement. Finally, although our labs can appreciate the reduction in labor, improvements in patient outcomes or CAUTI rates remain to be determined.
