Urine Trouble: Reducing Midstream Clean Catch Urine Contamination Among Children 3–17 Years Old

The University of Iowa Stead Family Children’s Hospital comprises a tertiary academic children’s hospital with multiple on- and off-site outpatient pediatrics clinics. The hospital consists of 190 pediatric inpatient beds. Three pediatric inpatient units serve patients admitted to multiple clinical services, including general pediatrics and pediatric and surgical subspecialties. The on-site pediatric subspecialty clinics serve both the local community and as a referral center, serving 77 000 patients annually from every county in Iowa, nearly every US state, and several other countries. The emergency department is a Level I Trauma Center that cares for nearly 10 000 pediatric patients annually. Two off-site University-affiliated clinics are located nearby and primarily serve the local community. The clinics serve >23 000 children with a variety of cultural and socioeconomic backgrounds annually.

In all locations, midstream clean catch urine culture collection was performed by children or their caregivers, with instruction and support from a nurse or nursing assistant. The person who collected the urine specimen was determined at the discretion of the nurse or nursing assistant involved (who might, for example, provide assistance to a preschool-aged child and that child’s caregiver but give a teenager privacy to collect the sample). This remained the same before and after our intervention.

All units, on- and off-site, used the same centralized medical microbiology unit to process urine culture samples. The microbiology unit is located in the University of Iowa General Hospital adjacent to the children’s hospital.

In October 2017, a quality improvement project was initiated by several pediatric residents who observed a high frequency of contaminated urine cultures collected by midstream clean catch technique. The residents assembled a multidisciplinary team consisting of physicians from general pediatrics, pediatric nephrology, pediatric infectious diseases, pediatric emergency medicine, and pathology, as well as staff nurses from several clinical practice settings and representatives of hospital and nursing quality improvement programs.

From October 2017 to January 2018, the team surveyed the included clinical units and formed collaborations with nursing staff and physicians from those units. Three inpatient units, 2 outpatient general pediatrics clinics, the outpatient pediatric subspecialties clinics, and the emergency department were included. The inpatient units are described by nurse staffing rather than physician staffing, and each included a mix of patients managed by the general pediatrics and pediatric and surgical subspecialty teams. On the basis of participant observations and local survey results, several potential opportunities for improved practices were identified and discussed with physician and nursing stakeholders. In Fig 1, we show the process chart developed by participants at these meetings with stars representing areas identified for intervention. Taking into account a review of the published literature and iterative feedback from 1:1:1 (“1 provider, 1 patient, 1 encounter”) tests of interventions, an implementation bundle was developed.

The bundle consisted of the following:

• provision of new cleansing towelettes (0.13% benzalkonium chloride) for urethral meatus cleansing. Previously, each unit had developed local practices of cleansing, and some did not routinely ensure meatal cleansing^5,6 ;

• requirement that nursing staff provide the child or caregiver with new written patient instructions and cleansing towelettes rather than the previous practice of leaving these materials in patient restrooms.⁷  Written instructions were approved by a hospital subcommittee tasked with ensuring the readability of patient materials;

• increased emphasis in nursing and patient instructions on midstream urine collection rather than first void⁶ ;

• recommendations to decrease contamination of the collected urine sample by avoiding inserting a dipstick into the original sample and ensuring sterile pipette technique;

• clarification that the sample should arrive at microbiology within 30 minutes or be refrigerated, after which a boric acid-containing tube should be used^6–9 ; and

• development of a real-time data resource made available to nursing leadership to track unit urine culture contamination data.

Written instructions and the nursing protocol developed from this work have been included as Supplemental Information.

Staff education materials, including a flow diagram with red stars to highlight modifiable sources of contamination (Fig 1), were posted in staff work areas, and presentations highlighting the initiative were given to all unit nurses and medical assistants. These presentations and the disbursement of bundle supplies (patient instructions and cleansing wipes) took place between January and March of 2018. Initial postintervention data collection to evaluate the success of the intervention began in April 2018.

The project was determined to be exempt from human subjects research oversight as a quality improvement initiative.

The preintervention period consisted of April 2016 to September 2017. Development and piloting of the intervention took place from October 2017 to March 2018, during which there were multiple iterative tests of changes to midstream clean catch urine culture collection as well as education in each unit. The original goal of the resident-initiated quality improvement project was to reduce urine culture contamination during the subsequent 6 months, April 2018 to September 2018. To evaluate sustainability of the intervention, we also evaluated data from October 2018 to September 2020.

All urine cultures collected by midstream clean catch for patients ages 3 to 17 years in participating units were identified from the electronic medical record. Data extracted from the medical record included the following: date and time of collection, method of collection, and patient age (in years), sex (male, female), and unit (inpatient units 1, 2, and 3, outpatient general pediatrics clinics 1 and 2, outpatient pediatric subspecialties clinics, and emergency department). Culture contamination rates were reported monthly for the duration of the study.

By local policy, a contaminated urine culture was defined as one growing any nonpathogenic organism (reported by the laboratory as “skin flora” or “urogenital flora”) at any colony count or a mix of ≥3 pathogenic organisms with similar colony counts (reported as “multiple organisms present suggesting improperly collected specimen”). Urinary tract pathogens included Gram-negative bacteria (including Escherichia coli, Proteus, Klebsiella, Enterobacter, Citrobacter, Morganella, and Pseudomonas), Enterococcus, Staphylococcus aureus, β-hemolytic Streptococcus, and yeasts, and had growth of ≥10 000 colony-forming units per mL.

Repeat urine culture (within 5 days of the initial urine culture), positive urine culture results, and the organisms reported were collected as balancing measures. A positive culture result was defined as a urine culture that was not contaminated and that revealed a pathogenic organism by using the definition above. Although contamination of urine samples collected by urethral catheterization was not a focus of our study, rates of urethral catheter-obtained urine culture contamination for all children 3 to 17 years old were also collected during the study period as a negative control to assess whether there were any changes in contamination rate during the period of study not directly related to our intervention for midstream clean catch (eg, because of changes in laboratory culture-handling processes). Data for other forms of urine collection (eg, vesicostomy) were not included.

Monthly rates of midstream clean catch urine contamination before and after intervention were plotted on a p-chart with upper and lower control limits and evaluated for centerline shift. The mean contamination rate in the first 6 months postintervention was used as the centerline for computing 3-σ control limits to evaluate for sustainability during the subsequent 24 months. To indicate potential special cause variation, we considered 1 point outside of the control limits as significant. We considered 9 consecutive points on either side of the centerline to indicate a shift in the process mean.

To quantify the impact of the intervention bundle, we compared the postintervention midstream clean catch urine culture contamination rate with the preintervention rate using an unadjusted Pearson’s χ² test. Similar tests were also performed stratified by site, patient sex, and patient age (3–9, 10–15, and 16–17 years). Homogeneity in the effect of the intervention across subgroup categories was assessed by using the Breslow-Day statistic. To account for variations in the balance of covariates (site, patient age group, and patient sex) associated with urine culture contamination before and after the implementation of the intervention, adjusted analyses of the primary outcome were performed by using the Cochran-Mantel-Haenszel method.

Repeat urine cultures within 5 days and the rates of positive culture results were compared before and after the intervention as balancing measures. Rates of contaminated urine cultures obtained by urethral catheterization were also calculated for the study period and compared before and after the intervention.

All analyses were performed in SAS version 9.4 (SAS Institute, Inc, Cary, NC). P values of <.05 were considered statistically significant. Our results were reported according to the Standards for Quality Improvement Reporting Excellence 2.0 guidelines¹⁰  as well as recommendations by LaRocco et al⁶  for quality improvement studies reporting on urine culture contamination.

There was a total of 912 midstream clean catch urine cultures obtained in the 18 months preintervention, of which 416 (45.6%) were contaminated (Fig 2). In the 30 months postintervention, a total of 1098 urine cultures were obtained (233 during April 2018 to September 2018, 480 during October 2018 to September 2019, and 385 during October 2019 to September 2020), of which 441 (72 [30.9%], 184 [38.3%], and 165 [42.9%], respectively) were contaminated. Both before and after the intervention, rates of contamination were higher among girls and older children (Table 1).

In the 6-month period postintervention (Table 2), there was a 14.7% (95% confidence interval [CI]: 8.0% to 21.5%) absolute decrease in the rate of midstream clean catch urine culture contamination compared with before the intervention. This corresponded to a 47% decrease in the odds of contamination (odds ratio [OR] = 0.53 [95% CI: 0.39 to 0.72]). There were no differences across subgroups in the effect of the intervention as measured by the Breslow-Day test (for clinical site [P = .57], patient age [P = .97], and patient sex [P = .71]). Because the ages and sexes of children differed somewhat between pre- and postintervention periods (Table 1), adjusted analyses to account for these potential differences were performed and did not affect the primary finding of a significant reduction in contamination rate in the initial 6-month postintervention period.

Assessment of sustainability for an additional 2 years revealed a decrease in the effect of the intervention over time (Fig 2). The overall rate of contamination for 30 months postintervention was 38.4%, a 7.3% (95% CI: 2.9 to 11.6) absolute decrease in contamination and 26% decrease in the odds of contamination (OR = 0.74 [95% CI: 0.62 to 0.89]) (Supplemental Table 4). However, there were important changes in rate during the postintervention period. In August 2019, 17 months after the intervention was implemented, the monthly contamination rate was outside of the control limits (Fig 2). In October 2019, 19 months after the intervention was implemented, the criteria for a shift in the process mean were met. From October 2019 to September 2020, the new baseline rate of contamination was 42.9%, compared with 45.6% preintervention.

The preintervention rate of repeat culture obtained within 5 days was 4.9% (5.0% [n = 28 of 560] for April 2016 to March 2017 and 4.8% (n = 17 of 352) for April 2017 to September 2017). The rate decreased to 0.9% (n = 2 of 233) in the 6 months after the intervention (P < .01). The rate increased during the 2-year evaluation for sustainability to 3.1% (n = 15 of 480) and 4.7% (n = 18/385) at 7 to 18 months and 19 to 30 months, respectively.

The rate of positive culture results did not change during the study, from 17.3% preintervention (18.2% [n = 102 of 560] for April 2016 to March 2017 and 15.9% [n = 56 of 352] for April 2017 to September 2017) compared with 14.6% (n = 34 of 233) during the first 6 months postintervention (P = .67). There were no additional changes during the 2-year evaluation for sustainability (16.2% [n = 78 of 480] at 7 to 18 months and 19.0% [n = 73 of 385] at 19–30 months). There were no changes in species identified in positive urine culture results during the postintervention period (Table 3).

Contamination rates of urine cultures collected by urethral catheterization for children 3 to 17 years at our institution, for which there were no changes in policy or procedures during the study period, did not significantly change during the study. The contamination rate was 13.2% (n = 9 of 68) and 11.7% (n = 7 of 60) in the preintervention time intervals, and 11.8% (n = 6 of 51) in the immediate 6-month postintervention period (P = .99). Similarly, there were no changes during the subsequent 2 years (8.6% [n = 10 of 117] at 7 to 18 months and 11.8% [n = 12 of 102] at 19–30 months).

The implementation of a low-cost quality improvement bundle by a resident-led team was effective in decreasing contamination of pediatric midstream clean catch urine samples for children 3 to 17 years old across a variety of clinical areas at our institution for 6 months after implementation. The effect was consistent across a range of practice environments and among both boys and girls of different ages. The intervention led to a clinically important decrease in repeat urine cultures and did not affect the rate of positive urine culture results.

We found that postintervention results were not sustained over the subsequent 2 years. In a simple pre- and postintervention comparison, the overall rate of urine culture contamination for 30 months postintervention was 7.3% (95% CI: 2.9 to 11.6) lower than the 18 months before. However, use of a p-chart revealed a shifting process mean during the 30 months postintervention.

The initiative described here, initiated and led by pediatrics residents as part of a quality improvement curriculum through their residency training program, highlights the importance of evaluating the long-term sustainability of resident-led quality improvement. Although the Accreditation Council for Graduate Medical Education requires opportunities for experiential learning in quality improvement for medical residents,¹¹  follow-up for such programs beyond the duration of residency training may be limited. In one report describing published resident-led quality improvement projects, researchers noted that only 2 of 18 projects included follow-up beyond 1 year.¹²  In the case of this project, the successful results of the intervention, based on 6 months of data after implementation, were presented at a local quality improvement conference in November 2018, and the intervention was incorporated into nursing policies and procedures in the various units. However, the gains realized by the project diminished over the subsequent 2 years. Upon reviewing the results of the sustainability analyses, nursing and other leadership involved in the project noted that, despite updated standards of practice and an electronic dashboard for displaying urine culture contamination rates, after the residents who directed this project graduated, urine culture contamination data were no longer routinely discussed in the units and attention was shifted to other priorities. Our findings support the recommendations of others that further consideration by educators, clinicians, and administrators be given to the unique challenges of sustainability in resident-led quality improvement.¹² 

Strengths of our study include its large size, spanning 4.5 years, allowing for long-term assessment of sustainability as well as the consideration of subgroup-specific effects based on patient characteristics. Our findings of high rates of contamination among girls and teenagers are consistent with the published literature and suggest the need for future efforts to focus specifically on these populations.^5,13  Our findings are consistent with published interventions to reduce midstream clean catch urine culture contamination. A randomized clinical trial among children revealed a reduction in midstream clean catch urine culture contamination with the use of meatal cleansing compared with no meatal cleansing.⁵  At least 1 study in adults has revealed independent effects of written and verbal instructions on reducing contamination rates.¹⁴  Decreased contamination from the use of boric acid and refrigeration of urine samples has also been reported.^6–8 

Limitations of our study include that it was restricted to a single center. However, this was a large medical center with several practice environments, so our findings may be relevant to other institutions. Importantly, we did not measure how the reduction in urine culture contamination affected urinalysis, antibiotic use, or patient outcomes. Additionally, given the private and personal nature of aspects of the intervention (eg, cleaning the urethra, urinating into a cup), the number of units participating, and the lack of existing documentation surrounding many aspects of the intervention (the addition of which would have been a substantial burden to providers), data on process measures were not systematically collected. Our results may have varied somewhat if we had used a different definition of urine culture contamination; we used a pragmatic definition used in our clinical practice. Our intervention was a bundle that included several interventions implemented simultaneously. We were unable to discern which aspects of the bundle were most effective. Additionally, our study was an observational study and so subject to confounding. However, we found homogeneity in the effect of the intervention across subgroups, a consistent effect of the intervention over time after accounting for potential confounding variables, and no change in the rate of contamination in urine cultures obtained by urethral catheterization. These additional observations strengthen our finding that decreased urine culture contamination was associated with our intervention.

Our study demonstrates that implementation of a simple bundled intervention was associated with a meaningful reduction in contamination of urine cultures collected by midstream clean catch. Effects of the resident-led intervention decreased over time. Our experience highlights the need for further attention to sustainability in resident-led quality improvement. The standardized bundle materials described here may be transferrable to other institutions.

We thank Hermann Hounkponou, Benjamin Doyle, and Jeffrey Kritzman, who assisted with the aggregation of urine culture data for this initiative, and the numerous physicians, nurses, medical assistants, and others who participated in the process of improving local practices. We also thank Michael Edmond and Marin Schweizer for their critical reviews of this manuscript.

CI

confidence interval

OR

odds ratio