OUP user menu

Impact of Laboratory-Reported Urine Culture Colony Counts on the Diagnosis and Treatment of Urinary Tract Infection for Hospitalized Patients

Jennie H. Kwon DO, Maureen K. Fausone, Hongyan Du MB, MS, Ari Robicsek MD, Lance R. Peterson MD
DOI: http://dx.doi.org/10.1309/AJCP4KVGQZEG1YDM 778-784 First published online: 1 May 2012

Abstract

Reducing health care–associated urinary tract infection (UTI) is a National Patient Safety Goal. The purpose of this investigation was to establish a colony count threshold to predict clinically significant UTIs that develop in hospitalized patients. A total of 185 cases were reviewed sequentially by 2 physicians. The information extracted included subjective complaints, presence of an indwelling urinary catheter, clinical signs and symptoms, WBC count, urinalysis, and urine culture results. The first reviewer recorded whether the patient was diagnosed and treated for a UTI by the clinician. The second reviewer determined if the patient met National Healthcare Safety Network guidelines for nosocomial UTI. Compared with patients with colony counts less than 100,000 colony-forming units per milliliter (CFU/mL), patients with colony counts 100,000 CFU/mL or more were 73.86 times more likely to have a clinically significant UTI (odds ratio, 73.86; 95% confidence interval, 24.23 ∼ 225.15; P < .0001; c-statistic, 0.859). Reporting positive results only for patients with 100,000 CFU/mL or more would have reduced the number of positive cultures by 38%. These data suggest that reporting colony counts less than 100,000 CFU/mL encourages treatment of non–clinically significant UTIs in hospitalized patients, causing inappropriate antibiotic use.

Key Words:
  • Urinary tract infection
  • Urine culture colony count
  • Asymptomatic bacteriuria
  • Nosocomial infection

Urinary tract infections (UTIs) are among the most common form of health care–associated adverse events. They contribute more than 30% of health care–associated infections reported by acute care hospitals each year.1,2

In current practice, at virtually all US laboratories, culture colony counts of more than 1,000 or 10,000 colony-forming units (CFU)/mL are reported from the diagnostic culture of a urine specimen regardless of patient “type” or location. In a 2011 survey of the Clinical Microbiology Network, a list server of clinical microbiology laboratory directors, 100% of the 61 responding faculty indicated that their laboratory reported the results for nosocomial UTIs in the same manner as they do for all other types of UTI.3 Although colony counts less than 100,000 CFU/mL are infrequently associated with a clinically significant infection during hospitalization,4 reporting lower colony counts often results in the patient being treated with antibiotics. This may lead to unintended consequences such as unnecessary antibiotic exposure resulting in the development of multidrug-resistant organisms and potentially significant drug adverse events.5,6 Antibiotic use may also place a patient at risk for antibiotic-associated Clostridium difficile infection, causing preventable disease and expense to the patient and the hospital.58

The purpose of this quality improvement investigation was to document the amount of excess antibiotic prescribing and to establish a colony count threshold that can be used to predict clinically significant UTIs developing in hospitalized patients. Our hypothesis is that the laboratory cutoff for a clinically significant UTI affecting an inpatient that occurs after admission should be 100,000 CFU or more of bacteria per milliliter of urine.

Materials and Methods

Setting

The purpose of this quality improvement investigation was to determine the extent of excess antibiotic treatment that was being given to inpatients diagnosed with a health care–associated UTI and how laboratory reporting of urine culture results may be affecting that overprescribing. NorthShore University HealthSystem (NorthShore; Evanston, IL) is a 4-hospital organization and research institute with more than 900 inpatient beds, 60,000 annual admissions, 75 associated off-site offices, 450 staff physicians, and more than 1,000 independent physicians. It is affiliated with the University of Chicago’s Pritzker School of Medicine (Chicago, IL). North-Shore has a single comprehensive inpatient and outpatient electronic medical record (Epic, Verona, WI) and was the first multihospital health care organization to adopt such a system in the United States. The Epic medical record was the source of our medical record review.

Medical Record Review

The records of 185 sequential presumed UTI cases during January and February 2010 were reviewed by 2 physicians, the primary reviewer (J.H.K.) and secondary reviewer (L.R.P.). The goal of the primary reviewer was to determine which patients were diagnosed with a UTI by their principal care physicians, and the secondary reviewer determined if patients found to have UTI at primary review had the diagnosis based on standard criteria for nosocomial UTI.

Urine cultures were considered positive if they showed any growth of 1,000 CFU/mL or more. One patient with a yeast infection was not included in the review because there is poor correlation between CFU/mL and symptomatic UTI with yeasts.9,10 The information extracted included subjective complaints, presence of an indwelling urinary catheter, clinical signs and symptoms, WBC count, urinalysis, and urine cultures documented as CFU/mL. Subjective complaints included, but were not limited to, dysuria, urinary frequency, and abdominal pain. Clinical signs including a temperature less than 36°C or more than 38°C, a resting heart rate more than 90 beats per minute, a respiratory rate more than 20 breaths per minute, and a WBC count of less than 4,000/μL (4.0 × 109/L) or more than 12,000/μL (12.0 × 109/L) were considered abnormal. The urinalysis data collected specifically focused on presence or absence of leukocyte esterase, nitrite, squamous epithelial cells, presence of bacteria, and semiquantitative assessment of WBCs.

The primary reviewer recorded whether the patient’s treating clinician diagnosed a clinically significant UTI, which was the basis for this first reviewer deciding if the patient had a UTI, and also if the physician chose to treat with antimicrobial agents. The secondary reviewer applied the National Healthcare Safety Network (NHSN) criteria specified by the Centers for Disease Control and Prevention for nosocomial UTI to the patient data to determine if the diagnostic decision made by the treating physician followed published guidelines for diagnosis of a health care–associated UTI. The secondary reviewer evaluated all patient data collected for patients with colony counts less than 100,000 CFU/mL who were classified by the primary reviewer as having a UTI, and a random 10% of all patient records for patients with colony counts 100,000 CFU/mL or more. If the NHSN criteria were not met, the patient was not classified as having a UTI by the secondary reviewer.

Data Analysis and Statistical Methods

Patients were organized by CFU results into 2 groups: with urine colony counts 100,000 CFU/mL or more and with urine colony counts less than 100,000 CFU/mL. The patients with colony counts of less than 100,000 CFU/mL were further divided into subgroups of less than 10,000, 10,000 to less than 50,000, and 50,000 to less than 100,000 CFU/mL. Data are presented as frequencies and percentages. A χ2 test, a Fisher exact test, an exact Pearson χ2 test, a Mantel-Haenszel (MH) test, or an exact MH test was used to assess the association between covariates and reviewer’s classification of UTI. We further performed logistic regression analysis to quantify the predictive ability of a patient’s characteristics to the diagnosis of UTI. Construction of a multiple logistic regression model was conducted by including covariates with P values less than .25 in univariate analysis followed by backward removal at a .05 cutoff. Odds ratios (ORs) and 95% confidence intervals (CIs) were determined, and a c-statistic was reported to represent discrimination ability. All statistical analyses were carried out by using SAS 9.2 (SAS, Cary, NC), and statistical significance was reached when the P value was less than .05.

This investigation was approved by the NorthShore Institutional Review Board as a quality improvement project.

Results

There were 70 cases with urine samples that grew less than 100,000 CFU/mL and 115 cases with 100,000 CFU/mL or more. The secondary reviewer concurred with the primary reviewer on all UTI classifications for the random 10% of cases with colony counts 100,000 CFU/mL or more that both reviewed. The 10% was used as a surrogate for the 100,000 CFU/mL or more population, and from this sample it was concluded that the secondary reviewer would have agreed with the UTI classification recorded by the primary reviewer for the other patients in this group. The results for the comparison of the reviewers are in Table 1.

Table 2 and Table 3 demonstrate that colony counts are significantly related to UTI classification for patients in the hospital (both reviewers found similar patterns with P < .0001). The proportion of patients with colony counts less than 100,000 CFU/mL classified as having a UTI by the primary reviewer was 24.3% and by the secondary reviewer was 5.7%. Furthermore, the odds for patients with a colony count less than 100,000 CFU/mL to be classified as having a UTI by the primary reviewer was 92.3% lower than for patients with a colony count of 100,000 CFU/mL or more. In other words, patients with colony counts of 100,000 CFU/mL or more are 12.93 times more likely to be classified as having a UTI by the primary reviewer than patients with colony counts less than 100,000 CFU/mL (OR, 12.93; 95% CI, 6.33 ∼ 26.43; P < .0001; c-statistic, 0.776).

View this table:
Table 1

When classified by the secondary reviewer, patients with less than 100,000 CFU/mL have a 98.6% reduced risk of being classified as having a UTI compared with patients with 100,000 CFU/mL or more. Thus, patients with colony counts 100,000 CFU/mL or more are 73.86 times more likely to be accurately classified as having a UTI according to the secondary reviewer than patients with colony counts less than 100,000 CFU/mL (OR, 73.86; 95% CI, 24.23 ∼ 225.15; P < .0001; c-statistic, 0.859). A c-statistic approaching 0.8 or more indicates very good to excellent discrimination of colony counts less than 100,000 compared with 100,000 CFU/mL or more to differentiate UTI by either reviewer.11

Multiple logistic regressions were attempted, but no significant multivariate models were found. Once adjusted for colony count, there were no longer any significant predictors of UTI other than colony counts for the primary or secondary reviewer. This indicates there were no reliable markers for clinically significant UTI other than colony counts from urine culture in the population studied. Figure 1 depicts the breakdown of appropriately diagnosed UTI based on the CFU/mL thresholds from the urine culture results.

Only 1 patient with a colony count less than 10,000 CFU/mL was classified as having a UTI by the primary reviewer. The secondary reviewer did not classify the patient as having a UTI, and while the patient was treated for 3 days, the treating physician documented doubts of a clinically significant UTI, but prescribed antibiotics nevertheless.

View this table:
Table 2

Five patients with colony counts ranging from 10,000 to less than 50,000 CFU/mL were classified as having a UTI by the primary reviewer. One of these patients was also classified as having a UTI by the secondary reviewer but can be excluded from any inpatient reporting rule because the patient was actually an outpatient upon further investigation. Of the 4 patients in this group for whom the secondary reviewer disagreed with the diagnosis, 2 were asymptomatic but still treated for a UTI, 1 was receiving combination antimicrobial therapy for cellulitis with bacteremia at the time of the culture, and 1 was receiving antibiotics for treatment of aspiration pneumonia at the time of the urine culture. Treatment with antibiotics at the time of the urine culture may have changed the colony count but would not affect the therapy for any infection for either of the last 2 patients.

Eleven patients with colony counts in the range of 50,000 to less than 100,000 CFU/mL were recorded as having a UTI by the primary reviewer. For 3 of these patients, the secondary reviewer agreed with the diagnosis and treatment of a UTI. All 3 of these patients had indwelling urinary catheters; 2 of the 3 were receiving antibiotics at the time of the urine culture, predisposing them to lower colony counts, and the third had been receiving antibiotics for another condition by the time the culture results were reported. Of the 8 patients for which the secondary reviewer did not agree with the diagnosis and treatment of a UTI, 5 never had documented symptoms, 1 was given antibiotics for pneumonia at the time of the culture, the culture for 1 patient showed mixed flora, and 1 patient likely had a contaminant in the culture because the patient was receiving ciprofloxacin for another infection at the time the specimen was obtained and the culture grew enterococci that were susceptible to ciprofloxacin. Thus, no adverse consequence would have occurred for any patient if the laboratory reported language to the effect “No evidence for health care–associated UTI” as the result for all specimens with a CFU/mL of less than 100,000.

Discussion

The results of our quality improvement study show bacterial colony counts 100,000 CFU/mL or more are more likely associated with clinically significant UTIs that should result in treatment than do colony counts of less than 100,000 CFU/mL for hospitalized patients. Using the NHSN criteria, patients in this study with colony counts 100,000 CFU/mL or more were 73.86 times more likely to be appropriately diagnosed as having a clinically significant UTI than patients with colony counts less than 100,000 CFU/mL. The data strongly suggest that reporting all colony counts from clinical microbiology laboratory testing encourages treating patients with low-colony-count urine specimens, providing exposure to unnecessary antibiotics that leads to the development of multidrug-resistant bacteria and antibiotic-related side effects.5,6 Reporting colony counts less than 100,000 CFU/mL also artificially raises the rate of health care–associated (eg, nosocomial) UTI because physicians frequently treat positive test results that are not clinically significant. Had we only reported positive results for the patients with 100,000 CFU/mL or more, we would have reported about 38% fewer positive cultures in the study population, and by review, none of these were clinically significant. Thus, overreporting of likely nosocomial UTIs leads to incorrect physician diagnosis, and nontargeted laboratory reporting criteria can have an adverse impact on patients plus a negative financial impact on hospitals.1,7,12,13

View this table:
Table 3
Figure 1

The percentages of patients with and without a urinary tract infection (UTI) associated with varying thresholds (in colony-forming units per milliliter) for reporting urine culture results.

A recent supplement published by the Infectious Diseases Society of America found overtreatment of asymptomatic UTI in hospitalized patients as a major target for reducing antibiotic use in any antimicrobial agent stewardship campaign.14 The authors cited numerous randomized trials not demonstrating benefit of treating most asymptomatic bacteriuria in hospitalized patients, and several reports found the practice leads to the development of bacterial resistance.1520 Similar to our data, these studies found that inappropriate treatment of asymptomatic bacteriuria occurs in one third to one half of patients, regardless of a patient’s catheter status; cases are likely treated because clinicians have difficulty ignoring a positive urine culture report.2123

For many of the asymptomatic patients who were treated for a UTI, a routine urine culture was ordered as part of a standard postsurgical or rehabilitation evaluation, and the patients were apparently treated when their health care provider saw the results were positive. It is interesting that the differing results for univariate logistic regression indicate that treating physicians may not routinely use the criteria specified on the NHSN guidelines, or a similar standardized document, to diagnose a UTI and may instead use nonstandard criteria that are likely not optimal for disease diagnosis. It is important to remember that fewer than 5% of patients with UTIs develop bacteremia; thus treating non–clinically significant UTIs is not an effective prevention for bacteremia,2 and overtreatment of patients with low colony counts of bacteria in their urine has not been shown to improve patient care.

For our patients, 4 (2.2%) were bacteremic at the time their urine sample was sent for culture. One was being treated for Listeria monocytogenes bacteremia at the time a urine specimen grew more than 50,000 to less than 100,000 CFU/mL Escherichia coli, another had Pseudomonas aeruginosa in the bloodstream and was undergoing therapy when the urine sample grew 50,000 to less than 100,000 CFU/mL of the same organism, the third had cellulitis and bacteremia (noted earlier in the “Results” section), and the fourth patient had an E coli bloodstream infection associated with a UTI having 100,000 CFU/mL or more in the urine culture. Our results for bacteremia are consistent with prior reports and reinforce the low risk of bacteremia from these potential UTI diagnoses in hospitalized patients, which demonstrates that a positive report with a low-colony-count urine culture does not mitigate treatment for other causes.

When diagnosing any disease, it is important to use recognized standards for disease definition. The NHSN is a surveillance system of the Centers for Disease Control and Prevention that provides data for tracking, prevention, and mandatory public reporting of health care–associated infections.24 The NHSN guidelines for the diagnosis of symptomatic UTIs are the accepted standard for diagnosing and reporting nosocomial UTI prevalence in health care organizations. Nosocomial UTIs must develop more than 48 hours after a patient is admitted to a hospital or less than 30 days after the end of a hospital stay.25 The diagnosis is based on findings such as the presence of an indwelling urinary catheter, signs and symptoms of infection, urinalysis, and culture evidence of microbial growth.1,25 We considered UTIs classified by the secondary reviewer to be the final discriminator where nosocomial UTIs are defined by an accepted standard such as the NHSN guideline.25

The significance of colony count varies depending on the patient population.26 Young, sexually active women often experience symptoms of UTI with colony counts less than 100,000 CFU/mL. In this setting, the presence of pyuria and a colony count less than 100,000 CFU/mL are sufficient to diagnose a clinically significant UTI. Pyuria is an important discriminatory feature of infection in this population as approximately 90% of the women experience pyuria with infection.26 Only the urine of patients with symptoms of a UTI and very selected patients expected to have asymptomatic bacteriuria (eg, patients in the first trimester of pregnancy) should be cultured in anticipation of therapy.5,26,27 Asymptomatic bacteriuria is primarily a concern in pregnant women because the infection has the potential to harm the fetus.5,28

Ii is important to note that asymptomatic bacteriuria should not be treated in elderly people because treatment does not change morbidity or mortality, does not alter the course of asymptomatic bacteriuria, and there is a high recurrence rate of bacteriuria after antimicrobial treatment.5,27,2931 Most cases of symptomatic UTIs are not preceded by bacteriuria for more than 1 day, and monitoring asymptomatic bacteriuria has not been shown to be an effective prevention method for symptomatic UTIs.2 Furthermore, 75% to 90% of patients with asymptomatic bacteriuria do not develop a systematic inflammatory response or any other signs to suggest infection.2 However, treatment of asymptomatic bacteriuria has been associated with the development of drug-resistant organisms,2,27 and decreasing opportunities for the development of drug-resistant bacteria protects other patients in the hospital from exposure to resistant bacteria that can cause infection. For this reason, it is important to treat only patients with a significant colony count and symptoms.1,5 Catheterized patients sometimes have low levels of bacteriuria and are typically asymptomatic.4 In this setting, levels of bacteriuria less than 100,000 CFU/mL often (96% of the time) can progress to levels 100,000 CFU/mL or more within 3 days of the initial culture if significant infection occurs. This implies that while low levels of bacteriuria in catheterized patients may signify the beginning of an infection, such levels of reporting have low specificity and that a diagnosis of a UTI will not be missed because low-level bacteriuria quickly progresses to levels 100,000 CFU/mL or more in these cases, leading to symptoms of infection, at which time the urine can be cultured for UTI diagnosis.

Our study has limitations. Because the secondary reviewer examined a random sample (10%) of the records for patients with 100,000 CFU/mL or more, it is possible that some of the patients classified as having a UTI by the primary reviewer in this group would not have met the published guidelines for diagnosis and would not have been classified as having a UTI. However, the principal focus of this investigation was to demonstrate that reporting low-colony-count urine cultures is not necessary for inpatients tested more than 2 days after their admission, and this conclusion would not be impacted by a small number of discrepancies in the group of patients with colony counts of 100,000 CFU/mL or more.

Another limitation is that the secondary reviewer did not review charts that the primary reviewer classified as not having a UTI. If a patient with a colony count less than 100,000 CFU/mL had a clinically significant UTI according to the guidelines used by the secondary reviewer but was not diagnosed or treated by the physician, the discrepancy would not have been noted. Mindful of that, the purpose of the investigation was to assess the overtreatment of UTI in patients with low-colony-count specimens so that once the primary care physician determined a UTI was not present, this patient was not at risk for receiving unnecessary antibiotics owing to a mistaken diagnosis of UTI. Thus, none of the patients who were not diagnosed as having a UTI by their primary care physician received antimicrobial agent therapy for a UTI, and there was no adverse consequence of this during the hospitalization evaluated by the primary reviewer.

Finally, the data in this study were obtained retrospectively from patient medical records. It is possible that not all clinical signs and patient complaints for UTI were recorded in all records by the treating physicians. If signs of clinical significance were underreported, clinically important UTIs would also be underreported by the secondary reviewer in this study. However, even following the decision of the primary care physician (ie, conclusion from the primary reviewer’s data analysis), our investigation indicates the potential to greatly enhance performance by using 100,000 CFU/mL or more as the optimal threshold when diagnosing UTI for inpatient health care–associated UTI.

Urine culture colony counts of 100,000 CFU/mL or more are highly associated with clinically significant UTIs in the inpatient setting. Colony counts of less than 100,000 CFU/mL are infrequently associated with clinically significant disease (P < .0001), and reporting these results as positive likely does more harm than good. The risk of reporting colony counts less than 100,000 CFU/mL is that physicians will typically treat patients who do not have a clinically significant UTI because they are considered to have a positive urine culture. In our investigation, we found no evidence that any patient with a urine culture growing less than 100,000 CFU/mL would have been adversely impacted by not receiving antibiotic therapy for a nosocomial UTI and, indeed, the vast majority had no signs or symptoms of any UTI. The much lower risk of UTI for patients with colony counts less than 100,000 CFU/mL provides evidence that it is not necessary or useful for the microbiology laboratory to report colony counts less than 100,000 CFU/mL and that reporting only colony counts of 100,000 CFU/mL or more as positive urine cultures for hospitalized patients can encourage more accurate diagnoses and restrict treatment to clinically significant UTIs.

CME/SAM

Upon completion of this activity you will be able to:

  • describe at least 3 unintended negative consequences of unnecessary antibiotic use.

  • define the bacterial colony count best correlated with a clinically significant urinary tract infection (UTI).

  • list the clinical signs and symptoms appropriate to guide testing for the diagnosis of UTI.

  • cite the risk of bacteremia associated with UTI.

The ASCP is accredited by the Accreditation Council for Continuing Medical Education to provide continuing medical education for physicians. The ASCP designates this journal-based CME activity for a maximum of 1 AMA PRA Category 1 Credit ™ per article. Physicians should claim only the credit commensurate with the extent of their participation in the activity. This activity qualifies as an American Board of Pathology Maintenance of Certification Part II Self-Assessment Module.

The authors of this article and the planning committee members and staff have no relevant financial relationships with commercial interests to disclose.

Questions appear on p 837. Exam is located at www.ascp.org/ajcpcme.

Footnotes

  • Disclosure: The authors are solely responsible for this report. During the past 5 years, Dr Peterson has received research grants from BD Diagnostics, Cepheid, MicroPhage, Nanogen, Nanosphere, NIAID, Roche, Synetzza, 3M, AHRQ, Wyeth (Pfizer), and the Washington Square Health Foundation for work in molecular diagnostics and has consulted for BD Diagnostics, Cepheid, Nanosphere, Wyeth (Pfizer), and Roche. Dr Robicsek has received research grants from BD Diagnostics and consulted for BD Diagnostics and Roche.

References

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
View Abstract