COUNTERPOINT
Whatever is only almost true is quite false and among the most dangerous of errors, because being so near truth, it is the more likely to lead astray.
—Henry Ward Beecher
The diagnosis of urinary tract infection (UTI) is not straightforward, as providers must be able to distinguish UTIs from other syndromes with a similar clinical presentation and to recognize UTIs that present with unusual manifestations. The practice of using UA results to reflex to culture continues to be debated. Supporters of reflex testing suggest that if a majority of urine cultures are negative, why not streamline the process by using UA to eliminate unnecessary cultures, avoid treatment in patients who have a negative UA result, and initiate early therapy in patients who have a positive UA result. This sounds like a clear-cut approach, no? On the contrary, UA testing is complex and institution specific, consisting of a variety of methods and criteria to define a positive UA (i.e., the presence of WBC, leukocyte esterase [LE], nitrite, and bacteria). Moreover, the absence of data to support the practice and the overall lack of consensus for reflex testing in laboratories indicates that the decision is multifaceted.
Patients asymptomatic for bacteriuria can have a positive UA result.
The issue is not whether we can conserve laboratory resources by enforcing a reflex testing algorithm. Rather, the issue is whether UA and/or urine culture orders are justified. Emphasis must be placed on optimizing providers' ordering practices through the adoption of guidelines to ensure that urine testing is ordered only when clearly indicated.
Asymptomatic bacteriuria (ASB) is common, particularly with advancing age or certain underlying conditions. Guidelines provided by the Infectious Diseases Society of America (IDSA) state that pyuria (i.e., the detection of LE or the presence of WBC) is not diagnostic of ASB, catheter-associated (CA) UTI, or non-CA UTI, as pyuria is prevalent in other infectious and noninfectious conditions. Similarly, the presence of bacteriuria (i.e., the detection of nitrites and the presence of bacteria) does not rule out contamination or ASB and is a poor predictor of UTI (1, 2).
Pyuria is present with ASB in 32% of young women, 90% of elderly patients in long-term care facilities, and 90% of hemodialysis patients (1), rendering UA useless in determining when a urine culture should be done. Thus, screening and antibiotic treatment is not indicated for patients with ASB, except pregnant women or individuals undergoing urologic procedures (1, 2). Despite these guidelines, inappropriate treatment of ASB is widespread, and only 53% of providers were reported to be practicing in accordance with these guidelines when diagnosing bacteriuria; the bulk of discordant signs/symptoms were associated with the presence of pyuria in urine (3).
How reliable is UA as a screening test for UTI?
For the majority of patients, a negative UA result is a good predictor of a negative urine culture. However, in a certain subset of patients, screening with UA is insufficient, and missing a UTI due to a negative UA result may be detrimental in cases of complicated UTIs and urosepsis. For example, a Canadian study demonstrated this in an elderly population, finding the sensitivity of a positive urine culture to be 73.7% in symptomatic patients. Five patients with negative UA (defined by the absence of LE or nitrite) had urosepsis proven by both positive urine culture and positive blood culture (4). In addition, it must be emphasized that a positive UA result is a poor predictor of a positive urine culture and UTI. Multiple studies have confirmed the inadequate performance of UA in predicting true-positive UTI cases, with positive predictive values (PPVs) ranging from 31 to 46% (5-7). These findings corroborate the finding that UA results are a poor predictor of the presence UTIs.
Contradictory impact on antimicrobial stewardship.
A major argument for urine reflex testing is that it will streamline testing and avoid the use of antibiotics in patients with negative UA results. However, as indicated by the high incidences of pyuria and bacteriuria in patients with ASB, screening by UA is not a sufficient gatekeeper to prevent unnecessary urine culture and inappropriate antibiotic administration. Rather, abnormal UA results may trigger initiation of antibiotics prior to the availability of urine culture results. This is particularly concerning, since the specificity of UA has been reported to be 54% of that of urine culture, meaning that more patients may potentially be prescribed antibiotics when positive UA results are available without culture results (6).
A retrospective study evaluating urine testing and antibiotic prescribing practice for 676 patients who were ≥12 years old and had a positive urine culture reported that 60% of urine tests were ordered without indication. One hundred eighty-four of 676 (27%) patients had ASB, and 37/184 (20%) were treated with antibiotics. Importantly, of the patients with ASB that were treated with antibiotics, 89% were given antibiotics based on positive UA results (8). The injudicious use of antibiotics in this setting was further evident in a prospective study of 343 adult women seen in the emergency department. That study reported overtreatment of 47% of patients when UA was positive for LE, nitrite, or trace blood but urine culture was negative. Using these UA criteria, 13% of patients with true signs and symptoms of UTIs (and positive urine culture) would not have been treated due to negative UA results (9). This study also demonstrated that UA performance characteristics are highly dependent on the cutoffs that are adopted; had a more stringent UA cutoff been used (e.g., an LE of >2 and positivity for nitrite), the overtreatment rate would have decreased to 13%, but the undertreatment rate would have escalated to 48%.
The biggest potential impact on appropriate antimicrobial usage is to adopt the mantra that “less is more” to mitigate unnecessary urine testing and/or reporting (10). This was demonstrated in a recent prospective pre- and postintervention comparison study conducted on adult patients with urinary catheters who were admitted to acute-medicine and long-term care wards (11). At the intervention site, extensive educational programs were introduced and included case-based audits, direct feedback, and interactive presentations to train clinicians to use a streamlined diagnostic algorithm based on IDSA guidelines for CA UTI and ASB. In contrast, the comparison site received traditional education methods and an email with the full text of the IDSA guidelines. The intervention site demonstrated a significant reduction in urine culture orders from 41.2 to 23.3 per 1,000 bed days. In addition, the rate of ASB overtreatment decreased by 1.6 per 1,000 bed days to 0.6 per 1,000 bed days, and the overall likelihood of ASB overtreatment decreased by 50%. In contrast, urine culture order rates and treatment of ASB were comparable to baselines at the comparison site. A proof-of-concept study conducted at an acute-care hospital in Canada demonstrated that intervention at the time of laboratory reporting can significantly decrease unnecessary antimicrobial usage. During the intervention period, rather than routinely reporting positive urine cultures from noncatheterized inpatients, technicians appended a comment to electronic medical records to request clinicians to call the laboratory for culture results if UTI is strongly suspected. Of the 37 modified reports, results were requested for only 5 (14%) patients, decreasing the rate of antibiotic therapy from 48% to 12% in noncatheterized patients with ASB (12).
Reflex urine culture and laboratory resources.
There are limited data on the impact of reflex culture on laboratory utilization. Depending on the institution, UA may be rather complex, consisting of one or more of the following: dipstick, urine microscopy, and urine Gram staining. A study found that performing dipstick analysis and culture on samples from all pediatric patients was the most cost-effective compared to performing microscopy UA and urine Gram staining on all urine specimens with reflex to culture ($3.70 versus $6.66/patient). Performing only microscopic UA with reflex to culture was found to reduce costs to $3.48/patient, but 18% of patients with UTIs would be missed due to the absence of culture (13). The cost reduction stated in this study is rather insignificant when you take into account the number of samples where testing was likely not warranted. Hence, the biggest impact on laboratory resources would be to restrict urine studies to only those from patients with clinical indications for testing.
Reflex to culture based on symptoms rather than urinalysis.
As discussed, it is not uncommon for providers to test and treat patients based on positive UA results, even in the absence of signs and symptoms of UTI. The Canadian study described above reported treatment of 71.4% of culture-positive elderly patients, despite the absence of symptoms compatible with UTI (4). A contributor to excessive testing and treating is the incorrect association of certain clinical features, namely, pyuria and bacteriuria, with UTI, resulting in poor diagnostic accuracy and reliability. Providers must identify correct signs and symptoms of UTIs prior to ordering UA or urine culture and be conscious of the high prevalence of ASB when interpreting a positive result.
How can these evidence-discordant norms of practice and biases be shifted to allow for evidence-based decision-making practices that use signs and symptoms associated with UTI to guide testing? One approach is asymmetric paternalism, defined as assisting individuals who are more inclined to make irrational decisions without actually limiting their freedom of choice (14). This has been applied in recent studies to assist providers in achieving their goals (i.e., correctly diagnosing and treating UTIs) in the presence of biases (discordant signs and symptoms of UTI). A common intervention strategy used by advocates of asymmetric paternalism is to exploit the same biases that would normally result in harmful behavior to instead promote healthy behavior (14). Trautner et al. (3) conducted a two-part study that first confirmed the inaccuracies of the “providers' mental models” in diagnosing CA UTIs. This was followed by the redirection of the providers' mental model through creation of a valid diagnostic “kicking CA UTI” algorithm that began with ordering only urine culture for patients who presented with symptoms of CA UTIs. The use of the diagnostic algorithm enhanced the reliability of differentiating between CA UTI and CA ASB diagnosis (3), and its success was demonstrated in the study described above (11). A large U.S. study of 1,469 females of <2 years of age presenting to the ED applied a similar approach. A clinical-prediction model was derived to identify children at high risk for UTIs, using clinical factors highly associated with positive urine cultures. Clinical factors that were associated with UTIs in females of <2 years of age included <12 months of age, white race, urinary symptoms, and the absence of gastrointestinal symptoms. Of note, a history of UTIs was not included in the prediction model, as clinicians may be influenced to obtain urine culture for these patients, even in the absence of other indications. Using this model, 95% of the patients with UTIs were identified and 30% of unnecessary urine cultures would have been eliminated (15).
If the correct diagnostic paradigm is utilized, then reflex culture may be limited to certain patient populations. In fact, there is consensus among multiple practice guidelines that culture should be limited to infection of the upper urinary tract or complicated UTIs, that culture is unnecessary in the vast majority of uncomplicated cystitis cases, and that treatment may be initiated based on symptoms alone. This typically includes premenopausal, nonpregnant women in the absence of urological abnormalities or comorbidities (16, 17). In contrast, culture may be warranted for other patient groups, including elderly patients in long-term care facilities and patients with hematologic malignancies or urological abnormalities.
Conclusions.
Reflexing urine culture from UA results does not directly improve antimicrobial stewardship or conserve laboratory resources. Rather, reeducating providers on the appropriate diagnostic algorithm associated with potential UTI signs and symptoms and adopting utilization restrictions are imperative. This will in turn reduce UA and culture orders, thereby decreasing the utilization of laboratory resources, reducing unnecessary antimicrobial therapy, and improving overall health care costs.
Jennifer Dien Bard