Effectiveness of Preanalytic Practices on Contamination and Diagnostic Accuracy of Urine Cultures: a Laboratory Medicine Best Practices Systematic Review and Meta-analysis

SUMMARY Background. Urinary tract infection (UTI) in the United States is the most common bacterial infection, and urine cultures often make up the largest portion of workload for a hospital-based microbiology laboratory. Appropriately managing the factors affecting the preanalytic phase of urine culture contributes significantly to the generation of meaningful culture results that ultimately affect patient diagnosis and management. Urine culture contamination can be reduced with proper techniques for urine collection, preservation, storage, and transport, the major factors affecting the preanalytic phase of urine culture. Objectives. The purposes of this review were to identify and evaluate preanalytic practices associated with urine specimens and to assess their impact on the accuracy of urine culture microbiology. Specific practices included collection methods for men, women, and children; preservation of urine samples in boric acid solutions; and the effect of refrigeration on stored urine. Practice efficacy and effectiveness were measured by two parameters: reduction of urine culture contamination and increased accuracy of patient diagnosis. The CDC Laboratory Medicine Best Practices (LMBP) initiative's systematic review method for assessment of quality improvement (QI) practices was employed. Results were then translated into evidence-based practice guidelines. Search strategy. A search of three electronic bibliographic databases (PubMed, SCOPUS, and CINAHL), as well as hand searching of bibliographies from relevant information sources, for English-language articles published between 1965 and 2014 was conducted. Selection criteria. The search contained the following medical subject headings and key text words: urinary tract infections, UTI, urine/analysis, urine/microbiology, urinalysis, specimen handling, preservation, biological, preservation, boric acid, boric acid/borate, refrigeration, storage, time factors, transportation, transport time, time delay, time factor, timing, urine specimen collection, catheters, indwelling, urinary reservoirs, continent, urinary catheterization, intermittent urethral catheterization, clean voided, midstream, Foley, suprapubic, bacteriological techniques, and microbiological techniques. Main results. Both boric acid and refrigeration adequately preserved urine specimens prior to their processing for up to 24 h. Urine held at room temperature for more than 4 h showed overgrowth of both clinically significant and contaminating microorganisms. The overall strength of this body of evidence, however, was rated as low. For urine specimens collected from women, there was no difference in rates of contamination for midstream urine specimens collected with or without cleansing. The overall strength of this evidence was rated as high. The levels of diagnostic accuracy of midstream urine collection with or without cleansing were similar, although the overall strength of this evidence was rated as low. For urine specimens collected from men, there was a reduction in contamination in favor of midstream clean-catch over first-void specimen collection. The strength of this evidence was rated as high. Only one study compared midstream collection with cleansing to midstream collection without cleansing. Results showed no difference in contamination between the two methods of collection. However, imprecision was due largely to the small event size. The diagnostic accuracy of midstream urine collection from men compared to straight catheterization or suprapubic aspiration was high. However, the overall strength of this body of evidence was rated as low. For urine specimens collected from children and infants, the evidence comparing contamination rates for midstream urine collection with cleansing, midstream collection without cleansing, sterile urine bag collection, and diaper collection pointed to larger reductions in the odds of contamination in favor of midstream collection with cleansing over the other methods of collection. This body of evidence was rated as high. The accuracy of diagnosis of urinary tract infection from midstream clean-catch urine specimens, sterile urine bag specimens, or diaper specimens compared to straight catheterization or suprapubic aspiration was varied. Authors' conclusions. No recommendation for or against is made for delayed processing of urine stored at room temperature, refrigerated, or preserved in boric acid. This does not preclude the use of refrigeration or chemical preservatives in clinical practice. It does indicate, however, that more systematic studies evaluating the utility of these measures are needed. If noninvasive collection is being considered for women, midstream collection with cleansing is recommended, but no recommendation for or against is made for midstream collection without cleansing. If noninvasive collection is being considered for men, midstream collection with cleansing is recommended and collection of first-void urine is not recommended. No recommendation for or against is made for collection of midstream urine without cleansing. If noninvasive collection is being considered for children, midstream collection with cleansing is recommended and collection in sterile urine bags, from diapers, or midstream without cleansing is not recommended. Whether midstream collection with cleansing can be routinely used in place of catheterization or suprapubic aspiration is unclear. The data suggest that midstream collection with cleansing is accurate for the diagnosis of urinary tract infections in infants and children and has higher average accuracy than sterile urine bag collection (data for diaper collection were lacking); however, the overall strength of evidence was low, as multivariate modeling could not be performed, and thus no recommendation for or against can be made.

Objectives. The purposes of this review were to identify and evaluate preanalytic practices associated with urine specimens and to assess their impact on the accuracy of urine culture microbiology. Specific practices included collection methods for men, women, and children; preservation of urine samples in boric acid solutions; and the effect of refrigeration on stored urine. Practice efficacy and effectiveness were measured by two parameters: reduction of urine culture contamination and increased accuracy of patient diagnosis. The CDC Laboratory Medicine Best Practices (LMBP) initiative's systematic review method for assessment of quality improvement (QI) practices was employed. Results were then translated into evidencebased practice guidelines.
Search strategy. A search of three electronic bibliographic databases (PubMed, SCOPUS, and CINAHL), as well as hand searching of bibliographies from relevant information sources, for English-language articles published between 1965 and 2014 was conducted.
Main results. Both boric acid and refrigeration adequately preserved urine specimens prior to their processing for up to 24 h. Urine held at room temperature for more than 4 h showed overgrowth of both clinically significant and contaminating microorganisms. The overall strength of this body of evidence, however, was rated as low. For urine specimens collected from women, there was no difference in rates of contamination for midstream urine specimens collected with or without cleansing. The overall strength of this evidence was rated as high. The levels of diagnostic accuracy of midstream urine collection with or without cleansing were similar, although the overall strength of this evidence was rated as low. For urine specimens collected from men, there was a reduction in contamination in favor of midstream clean-catch over first-void specimen collection. The strength of this evidence was rated as high. Only one study compared midstream collection with cleansing to midstream collection without cleansing. Results showed no difference in contamination between the two methods of collection. However, imprecision was due largely to the small event size. The diagnostic accuracy of midstream urine collection from men compared to straight catheterization or suprapubic aspiration was high. However, the overall strength of this body of evidence was rated as low. For urine specimens collected from children and infants, the evidence comparing contamination rates for midstream urine collection with cleansing, midstream collection without cleansing, sterile urine bag collection, and diaper collection pointed to larger reductions in the odds of contamination in favor of midstream collection with cleansing over the other methods of collection. This body of evidence was rated as high. The accuracy of diagnosis of urinary tract infection from midstream clean-catch urine spec-imens, sterile urine bag specimens, or diaper specimens compared to straight catheterization or suprapubic aspiration was varied.
Authors' conclusions. No recommendation for or against is made for delayed processing of urine stored at room temperature, refrigerated, or preserved in boric acid. This does not preclude the use of refrigeration or chemical preservatives in clinical practice. It does indicate, however, that more systematic studies evaluating the utility of these measures are needed. If noninvasive collection is being considered for women, midstream collection with cleansing is recommended, but no recommendation for or against is made for midstream collection without cleansing. If noninvasive collection is being considered for men, midstream collection with cleansing is recommended and collection of first-void urine is not recommended. No recommendation for or against is made for collection of midstream urine without cleansing. If noninvasive collection is being considered for children, midstream collection with cleansing is recommended and collection in sterile urine bags, from diapers, or midstream without cleansing is not recommended. Whether midstream collection with cleansing can be routinely used in place of catheterization or suprapubic aspiration is unclear. The data suggest that midstream collection with cleansing is accurate for the diagnosis of urinary tract infections in infants and children and has higher average accuracy than sterile urine bag collection (data for diaper collection were lacking); however, the overall strength of evidence was low, as multivariate modeling could not be performed, and thus no recommendation for or against can be made.

INTRODUCTION
T he most common infection occurring in the United States is urinary tract infection (UTI), accounting for nearly 7 million office visits, 1 million emergency room visits, and 100,000 hospitalizations per year (1,2). Significantly more women than men are likely to experience UTIs, with 1 in 3 women having at least 1 episode of UTI necessitating treatment with antibiotics by the age of 24 (3). Nearly half of all women will experience at least one UTI during their lifetime (3)(4)(5)(6). An increased risk of UTI occurs in certain population subgroups, including infants (7), pregnant women (8), the elderly (9), patients with spinal cord injuries and/or catheters (10), patients with diabetes (11) or multiple sclerosis (12), and patients with AIDS/human immunodeficiency virus (13,14). The most common nosocomial infection is catheter-associated UTI, with over a million cases in hospital and nursing home patients every year (15). Increasing duration of catheterization increases the risk of infection (16). Urinary tract infections are the second-most-common infection in noninstitutionalized elderly populations and account for nearly 25% of all infections (9). The financial impact of UTIs is significant, with costs of up to $2 billion per year (17).
While many uncomplicated UTIs in outpatients are diagnosed clinically, the diagnosis of recurrent or complicated UTI is commonly achieved by testing urine specimens for the presence of microorganisms. As a result, urine cultures often make up the largest portion of the workloads of clinical microbiology laboratories (18). The appropriate management of components of the preanalytic phase of urine culture, namely, collection, preservation, and storage of urine specimens, has an important influence on the generation of meaningful culture results, which ultimately affects patient diagnosis and management (19).

Quality Gap: Factors Associated with the Preanalytic Phase of Urine Culture
The major goal of proper specimen management is to ensure that specimen quality is maintained during collection and transport (20). Urine specimens can easily become contaminated with periurethral, epidermal, perianal, and vaginal flora. This contamination can be reduced with proper attention to techniques for urine collection, transport, preservation, and storage, the major components of the preanalytic phase of urine culture. A Q-Probe study conducted by the College of American Pathologists in 1998 (21) and again in 2008 (22) examined the frequency of urine culture contamination (defined as more than two isolates in quantities greater than 10,000 CFU/ml) and associated facility practices of urine collection and specimen management. Contamination rates of 41.7% (low-performance facilities), 15% (median performers), and 0.8% (high performers) correspond to the 10th, 50th, and 90th percentiles of facilities, respectively (22). Contamination rates had no correlation to collection site, use of collection kits, preservatives, or thermally insulated transport containers. However, contamination rates were substantially affected by postcollection processing, especially refrigeration of the specimen. Also, collection instructions given in the outpatient setting had a statistically significant impact on contamination rates in some cases. Based on the similarities of overall contamination rates between the two Q-Probe studies, the authors concluded that no significant progress in reducing urine culture contamination during the intervening years had been made (22). This may be a reflection of the inherent limitations of the Q-Probe methodology, which is based on one-time quality assessments dependent on the gathering of current data from large numbers of laboratories in order to establish provisional benchmarks for systematic quality improvement efforts. Many of these indicators are based primarily on selfreported surveys rather than on evidence-based scientific study designs and/or adequately specified, standardized, and consistently implemented data collection methods. Nonetheless, it is not cost-effective for laboratories to continue to waste valuable resources on the work-up of contaminated urine cultures (23). Furthermore, inappropriate reporting of contaminated urine cultures by the laboratory can result in patients receiving suboptimal or unnecessary therapy, producing poor patient outcomes and higher cost (18).
To address this important quality gap and its consequences, this research identified and evaluated practices associated with the collection, preservation, and storage of urine specimens for culture and their impact on the accuracy of urine culture microbiology. Rating criteria were used for evaluating these practices. Specific practices examined included collection methods for men, women, children, and infants; preservation of urine samples in boric acid solutions; and the effect of refrigeration on urine storage. The evidence supporting these practices for minimizing contaminated urine cultures and the impact on the accuracy of patient diagnosis were evaluated by applying the LMBP initiative's systematic review methods for quality improvement practices and by translating the results into evidence-based guidance (24). The methodology has recently been used to evaluate preanalytical practices for reducing blood culture contamination (25) and blood sample hemolysis (26).

A-6 CYCLE FOR SYSTEMATIC REVIEW
The CDC's LMBP "A-6 Cycle" systematic review methods for evaluating quality improvement practices was used for conducting this review. The methodology, reported in detail elsewhere (24), is derived from previously validated methods. It is designed to assess the results of studies of practice effectiveness that lead to best-practice recommendations that are evidence based. Using this method, a review coordinator (author Mark T. LaRocco) and individuals trained to apply the LMBP methods (authors Alice S. Weissfeld and Elizabeth K. Leibach) conducted the systematic review with guidance from an expert panel. The expert panelists (authors Nancy E. Cornish, Colleen S. Kraft, Vickie Baselski, Robert L. Sautter, Edward J. Peterson, and Debra Rodahl) were chosen based on their breath of experience and perspective in clinical microbiology and laboratory management. A description of their scientific credentials and professional affiliations can be found in the author biography section. Lastly, the team was supported by a statistician with expertise in evidence review methodologies and meta-analysis (author Jacob Franek). The expert panel reviewed the results of the evidence review and drafted the evidence-based best-practice recommendations. The recommendations were then approved by the LMBP Workgroup, consisting of 13 invited members with broad expertise in laboratory medicine, clinical practice, health services research, and health policy, as well as one ex officio representative from the Centers for Medicare and Medicaid Services. A list of the members of the LMBP Workgroup is provided in Appendix 1.

Review Question, Analytical Framework, and Search Strategy
The review question addressed by this analytical review was as follows: "Are there preanalytic practices related to the collection, preservation, transport, and storage of urine for microbiological culture that improve the diagnosis and management of patients with urinary tract infection?" Components of the preanalytic phase of urine culture were studied in the context of an analytical framework for factors affecting specimen contamination and diagnostic accuracy, depicted in Fig. 1. The population, intervention, comparison, and outcome (PICO) elements are as follows.
• "Population" is any patients who have urine cultures collected. • "Intervention" is clinical practice.
• "Comparison" is made of X immediate versus delayed processing of urine held at room temperature, X immediate versus delayed processing of refrigerated urine or urine preserved in boric acid, X midstream clean-catch collection of urine without cleansing versus with cleansing (men and women), X midstream clean-catch collection of urine without cleansing versus with cleansing versus collection with a sterile urine bag versus diaper collection for infants and children. • "Outcomes" are the results of determining the contamination rate and the diagnostic accuracy of urine culture.
Specific practices involving the preanalytic phase of urine culture covered in this evidence-based review were addressed by asking the following eight clinical questions. 1. What is the difference in colony counts when comparing immediate versus delayed processing of fresh urine stored at room temperature after collection? 2. What is the difference in colony counts when comparing immediate versus delayed processing of urine kept refrigerated or preserved in boric acid? 3. What is the difference in contamination rates between midstream urine collected with cleansing versus without cleansing in women being tested for a UTI? 4. What is the diagnostic accuracy of midstream urine collected with or without cleansing compared to bladder catheterization for the diagnosis of UTI in women? 5. What is the difference in contamination rates between midstream urine collection, with or without cleansing, and first-void collection in men? 6. What is the diagnostic accuracy of midstream urine collected, with or without cleansing, compared to that of bladder catheterization or suprapubic aspiration for the diagnosis of UTI in men? 7. What are the differences in contamination rates between midstream collection with cleansing, midstream collection without cleansing, and sterile urine bag or diaper collection in children? 8. What is the diagnostic accuracy of midstream clean-catch, sterile urine bag, or diaper collection compared with that of suprapubic aspiration or catheterization for the diagnosis of UTI in children?
The search for studies of practice effectiveness was conducted to identify those with measurable outcomes collected to the rigor of review requirements. With input from the expert panel and assistance of a research librarian at the Jesse Jones Library at the Texas Medical Center in Houston, TX, a literature search strategy and set of terms were developed. A search of three electronic bibliographic databases (PubMed, SCOPUS, and CINAHL) for Englishlanguage articles published between 1965 and 2014 was conducted. In addition, hand searching of bibliographies from relevant information sources was performed. All search results were catalogued and maintained using a Web-based, commercial reference software package (RefWorks; ProQuest LLC, Ann Arbor, MI). Finally, solicitation of unpublished quality improvement studies was attempted by posting requests for data on both the Laboratory Medicine Best Practices website (https: //wwwn.cdc.gov/futurelabmedicine/) and two listservs supported by the American Society for Microbiology: clinmicronet (http://www.asm.org/index.php/online-community-groups/listservs) and DivCNet (http://www.asm.org/division/c/divcnet.htm).
Titles and abstracts were initially screened by the review coordinator, with assistance from the expert panel when necessary, to select studies for a full review. A study was included if it was considered likely to provide valid and useful information and met the PICO criteria previously discussed. Specifically, these inclusion criteria required that a study (i) address a defined population/ definable group of patients, (ii) evaluate a specific intervention/ practice included in this review, (iii) describe at least one finding for a relevant outcome measure (percent contamination, diagnostic accuracy) reproducible in comparable settings, and (iv) present results in a format which was useful for statistical analysis. Studies failing to meet the inclusion criteria (not considered to report a relevant practice, did not include a practice of interest, or did not present an outcome measure of interest) were excluded from further review.
Studies that cleared this initial screening were then abstracted and evaluated by the expert panel. For eligible studies, information on study characteristics, interventions, outcome measures, and findings of the study was extracted using a standardized form and assigned a quality rating derived from points awarded for meeting quality criteria. Individual quality ratings were based on four dimensions: study quality, practice effectiveness, defined outcome measure(s), and findings/results. The objective for rating individual study quality was to judge whether sufficient evidence of practice effectiveness was available to support inclusion in an overall body of evidence for evaluation of a best-practice recommendation (that is, a practice likely to be effective in improving one or more outcomes of interest in comparison to other commonly used practices).
The four study quality dimensions were rated separately, with a rating score assigned up to the maximum for a given dimension. The rating scores for all four dimensions were added to reach a single summary score reflecting overall study quality. A total of 10 points were available for each study. Reviewers assigned one of three quality ratings to each study: good (8 to 10 points), fair (5 to 7 points), or poor (4 points or less). Each study was reviewed and rated by two expert panel members to minimize subjectivity and bias. Any study ranked as poor by one reviewer but good by the second reviewer was assigned to a third expert panel member for resolution. More detail on the rating process of individual studies can be found elsewhere (24)(25)(26). Studies that did not meet a study quality rating of fair or good were excluded from further consideration. Data from published studies that passed a full review were transformed to a standardized, common metric according to LMBP methods (24). Summary data and quality scores for each publication included in this evidence-based review can be found in Appendix 3 below.
The study quality ratings and results from the individual studies for each clinical question were aggregated into bodies of evidence. The consistency of effects and patterns of effects across studies and the rating of overall strength of the body of evidence (high, moderate, low, suggestive, and insufficient) were based on both qualitative and quantitative analyses. Estimates of effect and the strength of the body of evidence were then used to translate results into one of three evidence-based recommendations (recommend, no recommendation for or against, recommend against). The ratings criteria are described in greater detail elsewhere (24).
While recommendations are based on the entire body of evidence, meta-analyses to generate summary estimates of effect were undertaken for outcomes that provided sufficient data for measurements of diagnostic accuracy and contamination, i.e., proportions of specimens containing periurethral, perianal, epidermal, or vaginal flora. For the outcome of contamination proportion, summary odds ratios were calculated using Mantel-Haenszel methods in a random-effects model performed using Review Manager (RevMan) software version 5.0 (2008; The Nordic Cochrane Centre, The Cochrane Collaboration, Copenhagen, DK). A contamination event was defined according to how individual studies defined contamination because definitions varied between studies. Wherever possible, contamination proportions were determined for the entire test population rather than a subset population (such as only among those individuals that tested negative for urinary tract infection). The I 2 statistic, which describes the percentage of variability in effects estimates due to statistical heterogeneity rather than sampling error, was used to assess between-study heterogeneity. For the outcomes of diagnostic accuracy, it was planned that point estimates of sensitivity and specificity would be summarized using the bivariate model when similar cutoff points were used; however, all models failed to converge due to a too-small number of study or sample sizes. Similarly, hierarchical summary receiver operator characteristic curves (HSROC) could not be generated because these models too failed to converge. Solutions for failure of convergence, including removing individual studies, were explored but did not improve convergence. Meta-analysis of diagnostic accuracy outcomes and curve fitting were not pursued further given the limitations of univariate methods. All work on summarizing diagnostic accuracy outcomes was performed using SAS software version 9.2 (2008; SAS Institute Inc., Cary, NC, USA) and the MetaDAS macro, version 1.3 (27). Significant growth (i.e., a positive sample) was defined according to how each individual study defined significant growth because cutoff points tended to vary among studies. All other growth, including contamination and no growth, were considered nonsignificant growth (i.e., a negative sample), as this most closely reflects actual clinical practice. Two-by-two tables were used to determine sensitivity and specificity, and exact 95% confidence intervals were calculated.

Search Results
Search results produced 5,092 unique documents that were initially screened for eligibility to contribute to evidence of effectiveness for practices defined by the eight clinical questions posed (storage and preservation of urine, collection of urine from women, collection of urine from men, and collection of urine from infants and children). There was no response to requests for unpublished data. The reduction of studies through the screening process is detailed in Fig. 2. Initial screening for topic relevance eliminated 4,917 studies. From the remaining 171 studies, 124 were eliminated for not meeting the inclusion criteria (i.e., having elements potentially relevant to at least one topic area review question, reporting practices that are in use and available for adoption, reporting practices reproducible in other comparable settings, and addressing a defined population/definable group of patients). Forty-seven studies met the criteria for inclusion and were subjected to full abstraction and quality scoring. After an additional 12 studies were excluded because of insufficient quality scores, the remaining 35 were included in the statistical analysis: 10 studies on storage and preservation, 8 studies on collection from women, 3 studies on collection from men, and 14 studies on collection from infants and children.

STORAGE AND PRESERVATION OF URINE
Summary information on the 10 published studies comprising the body of evidence for the clinical questions on the storage and preservation of urine is presented in Tables 1 and 2. The publication dates for these studies range from 1969 (28) to 1999 (29). All studies were given a "fair" quality rating. Three studies examined the effect of prolonged storage of clean-catch    (29,31,32), glycerol-boric acid-sodium formate (32)(33)(34)(35)(36)(37), and sorbitol-boric acid-sodium formate (32,37). The length of delay of culture while samples were preserved in boric acid was assessed at various time points across studies, but 24 h was chosen for analysis as it was the most common endpoint. Three studies examined the effect of 24-h refrigeration of urine samples on changes in colony counts from those of immediate culture (33,35,37). The majority of studies used clean-catch midstream urine samples, although collection methods were undefined in five studies (30,32,34,35,37). Growth was defined as either "significant" or "nonsignificant." The definitions of significant growth varied among studies, but in general, a threshold of Ͼ10 5 CFU/ml of one or two species of bacteria was used.

Body-of-Evidence Qualitative Analysis
The difference in colony counts when immediate and delayed processing of urine specimens stored at room temperature were compared is shown in Table 3. Data from three observational studies (28,30,31) found a moderate increase (approximately 10%) in colony counts after 4 h of storage at room temperature and a large increase (Ͼ135%) in colony counts after storage for 24 h or more. The effect of delayed culture on urine specimens kept refrigerated or preserved in solutions of boric acid is shown in Tables 4, 5, and 6. Data from three observational studies (29,33,35) found 73 to 93% positive agreement (sensitivity) and 96 to 100% negative agreement (specificity) between the results of immediate culture and after a 24-h delay with specimens preserved in boric acid. Data from one study (35) found 93% positive agreement and 100% negative agreement between specimens cultured immediately upon receipt versus after a 24-h delay with refrigeration (Table 4). Colony counts in urine samples either refrigerated or chemically preserved showed similar results. Five studies (31,32,34,37,38) showed that urine samples preserved in boric acid solutions for 24 h (Table 5) or refrigerated for 24 h (Table 6) had only minor changes in the numbers of cultures with either significant or nonsignificant growth. These data suggest that both boric acid and refrigeration adequately preserve urine specimens prior to their processing for up to 24 h. Furthermore, the results suggest that urine held at room temperature for more than 4 h should not be processed due to overgrowth of both clinically significant and contaminating microorganisms. Based on statistical analysis of the data, however, the overall strength of this body of evidence was rated as low.

COLLECTION OF URINE FROM WOMEN
Summary information on the eight published studies comprising the body of evidence for the clinical questions on contamination rates and the diagnostic accuracy of midstream urine collection from adult females is presented in Tables 7 and 8. Three studies (39)(40)(41) were given a quality rating of "good," and five studies (38,(42)(43)(44)(45) were rated as "fair." Five studies (38-40, 42, 43) examined the impact of perineal cleansing on contamination and are summarized in Table 7. Patient settings included a clinic for adolescents (38), a general practice (38), an antenatal ambulatory-care clinic (39,43), and a health center for teenagers (40). Definitions of contamination varied among studies and included any growth of normal vaginal flora and/or small quantities (Ͻ2,000 CFU/ml) of pathogenic bacteria (38), the presence of epithelial cells (42), mixed growth in quantities of Ͼ10 5 CFU/ml (39) or at any quantity (43), and growth of any nonpathogen or pathogen in quantities of Ͻ10 4 CFU/ml (43) or Ͻ10 5 CFU/ml (40).
Three studies (41,44,45) examined the diagnostic accuracy of midstream urine collection with or without cleansing, with straight urinary catheterization as the reference standard (Table  8). Patient populations in these studies included women presenting to an emergency department (41) or ambulatory clinic (44) or admitted to a general medical ward (45). In two studies (43,44), each patient had urine collected by midstream collection with cleansing, followed by a second collection by urinary catheterization. In the third study (46), no cleansing was performed prior to midstream collection.

Body-of-Evidence Qualitative Analysis
The evidence examining the impact of perineal cleansing on contamination of midstream urine specimens collected from females is depicted in Fig. 3. Data from four observational studies (38,40,42,43) and one randomized control trial (39) found no difference  in the odds of contamination between midstream urine specimens collected with or without cleansing. The overall strength of this evidence was rated as high. The diagnostic accuracy of midstream urine collection with or without cleansing is shown in Table 9.
Using catheterization as the reference standard, midstream collection had a sensitivity of 98 to 100% and a specificity of 95 to 100%. However, the overall strength of this body of evidence was rated as low.

COLLECTION OF URINE FROM MEN
Summary information on the three published studies comprising the body of evidence for the clinical questions on contamination and diagnostic accuracy of midstream urine collection from adult males is presented in Tables 10 and 11. One study (46) was given a quality rating of "good," and two studies (47,48) were rated as "fair." Two studies (46,47) examined contamination in midstream clean-catch specimens compared to that in first-void collection specimens (Table 10). Patients in both studies were either ambulatory or hospitalized men with symptoms of urinary tract infection being seen at a VA Medical Center. In the first study (46), men had a first-void and/or midstream urine sample collected, but only half of the patients were asked to wash their glans penis prior to collection. In the second study (47), urine specimens from men were obtained by midstream clean-catch collection, firstvoid collection, straight catheterization, and suprapubic bladder aspiration, with 7 men being sampled more than once. Contamination was defined as either the growth of Ͼ10 3 CFU/ml of two or more colony types (46) or any growth of three or more microbial species (47). For the meta-analysis, only those samples obtained via midstream clean-catch collection and first-void collection without cleansing were compared.
Two studies (47,48) examined the diagnostic accuracy of midstream urine collection from men using either straight catheterization or suprapubic aspiration as the reference standard (Table  11). One study (47) compared midstream clean-catch specimens to those collected by suprapubic aspiration or straight catheterization in a group of hospitalized or ambulatory men, while the second study (48) compared midstream clean-catch specimens to specimens collected by suprapubic aspiration in a group of patients with spinal cord injury without indwelling catheters. Significant growth in one study (47) was defined as Ն10 4 CFU/ml of a single or predominant species for midstream clean-catch specimens or Ն10 3 CFU/ml for specimens collected by straight catheterization or suprapubic aspiration. Significant growth in the second study (48) was defined as any growth of Ն10 4 CFU/ml for either collection method.

Body-of-Evidence Qualitative Analysis
The evidence comparing levels of contamination after midstream urine collection and uncleansed first-void collection is shown in Fig. 4A. Summary data from both studies (46,47) found a large (77%) reduction in the odds of contamination in favor of midstream clean-catch over first-void specimens. The strength of this evidence was rated as high. Only one study (46) compared midstream collection with cleansing to midstream collection without cleansing (Fig. 4B). Results showed no difference in contamination between the two methods of collection. However, imprecision was largely due to the small event size. The diagnostic accuracy of midstream urine collection from men, with straight catheterization or suprapubic aspiration used as the reference standard, is shown in Table 12. Data for both studies found high diagnostic sensitivity (82 to 100%)   and specificity (92 to 100%) for midstream clean-catch collection. However, the overall strength of this body of evidence was rated as low.

COLLECTION OF URINE FROM CHILDREN
Summary information on the 14 published studies comprising the body of evidence for the clinical questions on contamination rates and the diagnostic accuracy of midstream urine collection from children is presented in Tables 13 and 14. Four studies (49)(50)(51)(52) were given a quality rating of "good," and 10 studies (53-62) were rated as "fair." Six studies (49,50,(53)(54)(55)(56) compared differences in contamination rates in urine collected by midstream collection (with or without cleansing), collected with a sterile urine bag, or collected from diapers (Table 13). Patients studied ranged in age from 1 month to 18 years. Definitions of contamination varied among studies and included mixed growth in any concentration (54), mixed growth in any concentration or any growth of Ͻ10 5 CFU/ml (49)(50)(51), and mixed growth at a concentration of Ͼ10 5 CFU/ml (56); any specimen interpreted as contaminated by the clinical microbiology laboratory was also included (53). Eight studies (51,52,(57)(58)(59)(60)(61)(62) examined the accuracy of midstream clean-catch, sterile urine bag, or diaper collection, with suprapubic aspiration or straight catheterization used as the reference standard for diagnosing urinary tract infections in children (Table 14). Patient age ranged from 0 to 10 years. Definitions of significant growth varied across studies, particularly for the reference standards. All studies except one defined significant growth for midstream clean-catch, sterile urine bag, or diaper collection as Ն10 5 CFU/ml. The remaining study (52) defined significant growth by sterile urine bag collection as "pure growth." Significant growth for suprapubic aspiration or straight catheterization  was defined as any growth in one study (51), growth of Ն10 2 CFU in one study (59), growth of Ն10 3 CFU/ml in one study (62), growth of Ն10 5 CFU/ml in three studies (57,60,61), and "pure growth" in one study (52). In one study (58), the definition of significant growth was unclear.

Body-of-Evidence Qualitative Analysis
The evidence comparing contamination rates for midstream urine collection with cleansing, midstream collection without cleansing, sterile urine bag collection, and diaper collection is shown in Fig. 5. Data obtained from five observational studies (49,(53)(54)(55)(56) and one cluster-randomized controlled trial (50) found larger reductions (68 to 73%) in the odds of contamination for specimens obtained by midstream collection with cleansing than for specimens obtained by the other methods of collection. Data from three observational studies (49,54,55) found no significant differences in the odds of contamination between specimens collected with sterile urine bags and specimens taken from diapers. This body of evidence was rated as high.
The accuracy of results for midstream clean-catch urine specimens, sterile urine bag specimens, or diaper specimens, with straight catheterization or suprapubic aspiration used as the reference standard for the diagnosis of urinary tract infection in children, is shown in Fig. 6. Data from eight observational studies showed varied results. The inability to meta-analyze the point estimates of sensitivity and specificity due to small sample and study sizes, together with heterogeneity in positivity thresholds, made interpretation difficult. Similarly, HSROC curves could not be generated, and thus it is unclear which method of noninvasive urine collection is most accurate for the diagnosis of urinary tract infection in children.

ADDITIONAL CONSIDERATIONS
This section addresses additional considerations for evaluating preanalytical practices associated with urine cultures and the impact of these practices on contamination and diagnostic accuracy.

Clinical Applicability
The studies included in this review reported collection, storage, and preservation of urine samples through commonly used methods for both children and adults in both inpatient and outpatient settings; results are therefore likely to apply to other health care environments. Many of the methods for collection, storage, and preservation are widely recommended (18,63) and are typically used in most hospitals, outpatient clinics, and clinical microbiology laboratories today (21,22). The focus of this review, however, is largely on clean-catch midstream urine collection because this method remains the most commonly used in most patient populations and settings (18). This is primarily due to its noninvasiveness; i.e., it has no risk of producing iatrogenic infection, despite the paucity of data supporting its use as a standard (63).
Controversy remains among clinical microbiologists and infectious disease physicians about the most accurate means for diag- Not given Contamination was defined as specimens containing more than 2 microbial species. None (0%) of the 75 specimens in the MSCC group were contaminated. 5 (6.9%) specimens of the 72 in the UFV group were contaminated. a MS, unclean midstream urine collection; MSCC, midstream clean-catch collection; UFV, first-void urine collection without cleansing; CFV, first-void collection with cleansing; SPA, suprapubic aspiration; CATH, urethral catheterization. The setting for these studies was the VA Medical Center, Seattle, WA. nosing urinary tract infections, including the best methods of specimen collection for women, men, children, and infants (19,63). A recent collaboration between the American Society for Microbiology (ASM) and the Infectious Disease Societies of America (IDSA), designed to assist physicians in the appropriate use of laboratory tests for infectious diseases, addressed methods of specimen collection, as well as guidelines for testing patients for urinary tract infections (64). A recommendation was made for collection of urine in a manner to minimize contamination and included midstream collection with cleansing and immediate refrigeration of samples upon collection, although the lack of supporting data was cited (64).
In applying the findings of this review, a strength assessment of the overall body of evidence should be weighted by the quality of findings from individual reports most closely resembling populations and settings of particular interest. For instance, an overall body-of-evidence quality rating may decrease because of the aggregate number of included studies omitting study parameters of little applicability to a particular clinical setting. Researchers may take guidance, with a higher degree of confidence than the overall quality rating might indicate, from individual included studies of high or moderate strength which address specific clinical populations or settings directly comparable to their research interests. The conduct of evidencebased practice would guide clinicians to assess both the quality and the "goodness of fit" of studies relevant to their own particular questions before applying findings in support of their decisions.
This review has directed attention to the need for reexamination of preanalytic factors affecting urine culture. A great number  of the studies covered in the review predate the regionalization and other significant restructuring of the delivery of microbiological services in the United States, which portend increased variation in collection, storage, and preservation methods. More studies are needed to support recommendations for specific pop-ulations, e.g., nursing facility residents needing skilled care. Important also is the growing need for documentation of health outcomes and cost-effectiveness of current practices through the implementation of well-designed, system-wide quality improvement studies. Of equal importance is the need to expand (and   communicate) the literature on diagnostic testing algorithms to include nonanalytic variables, such as those measured in the included studies reported here. This systematic review provides a current and substantial literature base from which to begin investigations not only to address these gaps in current knowledge related to the effects of preanalytic factors on urine culture but also to validate these best-practice recommendations in additional settings and populations.

Associated Harms
Methods of collecting, storing, and preserving urine specimens for the diagnosis of urinary tract infections have a critical influence on culture results. Poorly collected or preserved specimens can become easily contaminated with perineal, vaginal, and periurethral flora, which can inhibit or obscure the presence of true urinary tract pathogens. Conversely, the use of high concentrations of boric acid as a preservative has been known to inhibit urinary pathogens such as Escherichia coli and Klebsiella pneumoniae (65).
Midstream urine collection may be the preferred choice for collection for most patients; however, there are patient populations and clinical scenarios where a more invasive method of collection is preferred (63). All of these issues can produce incorrect culture results, misdiagnosis, especially in asymptomatic patients, poor patient management, including the use of inappropriate or ineffective antibiotics, and potentially more complicated urinary tract infection in the long term (2, 3).

Additional Benefits
Urine specimens that are appropriately collected, transported, stored, and preserved benefit patients by producing more-accurate culture results. In addition, such practices can provide benefit to the laboratory by allowing technologists to focus on the work-up of clinically significant pathogens rather than the growth of contaminants. Urine cultures are often a major component of the typical clinical microbiology workload (18); therefore, minimizing the processing of poor-quality urine specimens can allow the laboratory to focus its resources in a more cost-effective manner (22).

Economic Evaluation
Proper attention to the preanalytic phase of urine cultures should decrease the number of contaminated urine specimens processed by the laboratory. It may also decrease the time it takes for microorganism identification and susceptibility testing of pathogens in infected patients by reducing the number of recollected specimens. Both of these scenarios would likely reduce health care costs for both patients and institutions by reducing the time to appropriate targeted therapy and by making more-effective use of laboratory and hospital resources. However, no economic evaluation analyses were found for the studies covered in this review.

Feasibility of Implementation
The methods of specimen collection and handling covered in this review are feasible in all settings and patient populations and are, in fact, commonly used in most medical environments today. There are data showing the benefit of either refrigerating or chemically preserving urine samples that are not immediately processed (28)(29)(30)(31)(32)(33)(34)(35)(36)(37). Furthermore, midstream urine collection, with or without cleansing, is common practice for most clinical settings and patient populations . For facilities that have historically paid little attention to the preanalytic aspects of urine culture, there may be some resistance on the part of patients and staff that is typically associated with quality improvement initiatives. Appropriate education regarding the proper collection of urine specimens may be needed for both patients and health care workers. The additional costs associated with chemical preservatives, such as boric acid, would also need to be budgeted and justified.

Future Research Needs
The findings of this systematic review highlight the lack of recent high-quality studies that evaluate components of the preanalytical phase of urine culture. For example, the relative paucity of rigorous studies evaluating methods of storage and chemical preservation of urine specimens is troublesome considering the widespread use of these practices in many laboratories and a general consensus among microbiologists as to their benefit. A large number of the studies suffered from small sample sizes, limiting the precision of the results and reducing the likelihood that findings are applicable across a larger population. Studies also used various or unclear definitions of contamination or positivity thresholds, making meta-analysis or qualitative summary analysis problematic. Studies further suffered from missing data. For example, most studies were cross-sectional or otherwise observational (without randomization) in design, but many, particularly those retrospective in nature, did not obtain or report the results of samples from all patients obtained by all collection methods under study. These inconsistencies lead to significantly uneven comparison groups in some cases. Future studies should strive for statistically sufficient sample sizes, use common and clearly defined definitions of contamination and thresholds for positivity, and report accuracy results across several common positivity thresholds to aid subsequent meta-analysis. An example is the number of positive/negative samples calculated if reviewers use a threshold of Ͼ10 4 versus Ͼ10 5 CFU per ml of urine. Studies should also be more rigorous in design, include more randomized controlled trials, and ensure paired sampling when possible in prospective or cross-sectional studies. Moreover, for all methods under evaluation, patients should have urine collected within a reasonable time frame, and the time delay between collection and culture should be clearly reported.
Finally, future studies should strive to obtain data on down- stream patient-centered outcomes as influenced by different methods of collection, preservation, or storage of urine that are under evaluation. This broader measurement pool includes system-oriented outcomes, such as time to targeted therapy, cost of antibiotic use, number of UTI discharge diagnoses, or number of Clostridium difficile cases avoided, such that the direct or indirect impact of implementing a particular preanalytic practice can be measured at the patient and organizational level. Information provided in Appendix 2 can be used as a guide to organize and plan studies as well as collect data for any quality improvement project that examines preanalytical practices associated with urine cultures.

Limitations
The LMBP systematic review method is compatible with other standards of practice for systematic reviews (24) but includes some unique elements, such as the rating of study quality. Rating study quality is based on attributes such as facility description, study setting and design, practice description, outcome measures, and results. How studies are ultimately considered for inclusion in the review depends on consensus assessments that may be influenced by such things as a rater's professional background and experience. Indeed, several on-topic studies were excluded because of limitations identified during quality evaluation, mostly related to poor reporting of important study, practice, or outcome details. This is likely somewhat explained by the publication dates of many of the studies, with several of both the excluded and included studies having been published in the late 1960s. As is the case with most systematic reviews, attempts were made to limit publication bias by soliciting unpublished data; however, no unpublished data were submitted. Moreover, restricting the review to English-language studies may also introduce bias.
Outside the limitations of the review process, there were a number of limitations in this review that affected the ability to draw firm conclusions and make recommendations. Most of these limitations were addressed above in the context of future research, but additional limitations will be discussed here. The study settings varied across included studies. Both inpatient and outpatient settings were included, and the specific setting examined in each study-emergency department, adolescent clinic, obstetric clinic, etc.-may not be generalizable to other settings. Some settings may be better equipped to perform certain collection methods or to educate patients or parents on how to perform certain collection methods. Similarly, the patient populations under study varied. Some studies included healthy asymptomatic patients, while others included patients with more-severe conditions, such as spinal cord injury patients. This too might affect the generalizability of results. Within the body of evidence for children, studies often included patients ranging in age from 0 to 16 years. Unfortunately, there were not enough data available to properly stratify children, such as infants, into smaller age groups, and because of this, results may not be generalizable to patients of a specific age. Finally, as discussed above, an important limitation was the variability in positivity thresholds and definitions of contamination used across studies. Although several guidelines have been developed to address definitions of significant bacteriuria for culture (18,63,(66)(67)(68), these guidelines are not always consistent, and this lack of consistency is reflected in the studies and results reported in this review.

CONCLUSIONS AND RECOMMENDATIONS
A summary of the findings of this evidence-based review of urine culture preanalytics can be found in Table 15. Conclusions are categorized as "recommended," "not recommended," or "no recommendation for or against" and refer to studies of urine collected by noninvasive methods: 1. No recommendation for or against is made for delayed processing of urine that is stored at room temperature, refrig- erated, or preserved in boric acid due to insufficient evidence. Data from nine studies receiving a "fair" quality rating suggest that both refrigeration and boric acid adequately preserve urine specimens for up to 24 h prior to their being processed. Furthermore, data from three studies receiving a "fair" quality rating suggest that urine held at room temperature for more than 4 h should not be processed due to overgrowth of both clinically significant and contaminating flora. However, because the overall strength of the body of evidence was rated as low, no recommendation for or against can be made due to insufficient evidence. This does not preclude the use of refrigeration or chemical preservatives in clinical practice. It does indicate, however, that more systematic studies evaluating the utility of these measures are needed. 2. If noninvasive collection is being considered for women, midstream collection with cleansing is recommended, but no recommendation for or against is made for midstream collection without cleansing due to insufficient evidence. Data from two studies, including one randomized controlled trial receiving a "good" quality rating and three studies receiving a "fair" quality rating, show that contamination rates are similar between specimens obtained by midstream collection with and without cleansing. The overall strength of this body of evidence was rated as high. However, whether midstream collection can be routinely used in place of straight catheterization is unclear. Data from three studies, two with a quality rating of "fair" and one with a rating of "good," suggest that clean-catch midstream urine collection is highly accurate for diagnosing urinary tract infections in women; however, because the overall strength of this body of evidence was rated as low, no recommendation for or against can be made. 3. If noninvasive collection is being considered for men, midstream collection with cleansing is recommended and collection of first-void urine is not recommended. No recommendation for or against is made for collection of midstream urine without cleansing due to insufficient evidence. Data from two studies, one with a quality rating of "good" and one with a rating of "fair," found a large reduction in the level of contamination in specimens obtained by midstream collection with cleansing compared to the level of contamination after collection of first-void urine. This body of evidence was rated as high. Although data from one study rated as "good" quality found no difference in contamination between midstream urine collected with and that collected without cleansing, imprecision was large due to the small event size, and no recommendation can be made as to which method is superior. Whether midstream collection can be used routinely in place of straight catheterization or suprapubic aspiration is unclear. Data from two studies receiving a "fair" quality rating suggest that midstream collection with cleansing is highly accurate for the diagnosis of urinary tract infections in men; however, because the overall strength of the body of evidence was rated as low, no recommendation for or against can be made. 4. If noninvasive collection is being considered for children, midstream collection with cleansing is recommended and collection with sterile urine bags, from diapers, or midstream without cleansing is not recommended. Data from six studies, two with a quality rating of "good" and four rated as "fair," found large reductions in contamination in midstream clean-catch urine specimens compared to contamination after other noninvasive methods of collection. This body of evidence was rated as high. Whether midstream collection with cleansing can be routinely used in place of catheterization or suprapubic aspiration is unclear. Data from eight studies, two with a quality rating of "good" and six rated as "fair," suggest that midstream collection with cleansing is accurate for the diagnosis of urinary tract infections in infants and children and that midstream collection with cleansing has higher average accuracy than sterile urine bag collection (data for diaper collection was lacking). However, the overall strength of evidence was low, as multivariate modeling could not be performed; thus, no recommendation for or against can be made due to insufficient evidence.

APPENDIX 2 LMBP Evaluation of Preanalytic Practices for the Contamination and Diagnostic Accuracy of Urine Cultures
Suggested guidance for future studies. This review identified and rated practices associated with the collection, preservation, and storage of urine specimens for culture and assessed the impact of these preanalytic practices on the diagnostic accuracy of urine culture microbiology. In theory, the design, description, methods, data collection, and analysis for any study should be written and documented so that other investigators can reproduce exactly the same study in their laboratory, with their results validating or verifying those of the original study.
The following organizational plan with instructions can be used as a guide for quality improvement project design, implementation, and evaluation of preanalytic practices associated with urine cultures. Figure A1 shows a form for use in collecting data for any QI project that examines preanalytical practices associated with urine cultures.

Problem/quality issue description.
A. Practices and equipment. Describe what preanalytic practices associated with the collection and preservation of urine for culture were studied and exactly how specimens were collected. Include examples of educational material handed to patients or displayed on walls in areas where patients were seen. i.
Collection. Indicate whether specimens were obtained by midstream clean-catch collection versus midstream collection without cleansing, midstream clean-catch collection versus straight catheter collection and/or suprapubic aspiration, midstream collection without cleansing versus straight catheter collection and/or suprapubic aspiration, midstream collection with or without cleansing versus first-void urine collection, midstream collection with or without cleansing versus diaper collection and/or sterile urine bag collection, or other means. ii. Preservation/storage/transport. Include the time from collection of the specimen to the addition of preservative and how long it took the specimen to reach the lab after collection, as well as how long it took from receipt in the lab to setup of culture. Indicate whether boric acid, glycerol boric acid, or sorbitol boric acid was used as a preservative, whether the specimen was stored in a refrigerator or at room temperature, and any other relevant preservation or storage information. B. Population under study and age ranges. Include physical differences which may affect the collection of the specimen, such as physical disability, the presence of a foreskin in males, the presence of diapers, etc. With infants and neonates, consider tighter age ranges, such as 0 to 2, 2 to 4, etc. C. Collection personnel. Indicate whether the specimen was collected by a nurse, nurse's aide, physician, specialized urine collection team, lab technologist or technician, or other staff member. 2. Submitter(s) and organization affiliation. For additional questions concerning the quality improvement (QI) study, contact information is required. 3. Funding source(s). Refer to the chart in Fig. A1. Check all boxes that apply.

QI Project/Study
4. QI project study design/type. With similar patient populations, describe the methods/approaches used for your project with regard to age, sex, ethnicity, and/or diagnosis to limit bias. A. Pre-and postimplementation. Observations are made before and after the implementation of an intervention. B. Split implementation design. Indicate whether multiple sites were used to conduct the QI study. C. Case-control study. Indicate whether the study compared subjects with a specific outcome of interest (cases) with subjects from the same source population but without that outcome (controls) to examine the association between the outcome and prior exposure (for which there was an intervention). D. Cross-sectional associations. Collect information on interventions (past or present) and current health outcomes, i.e., those that are restricted to health states, for a group of people at a particular point in time, to examine associations between the outcomes and exposure to interventions. E. Cohort. A defined group of people (the cohort) is followed over time to examine associations between different interventions received and subsequent outcomes. F. Randomized assignment. Patients are randomly selected to receive the intervention practice or the comparator practice. G. Other study design used in this QI project. Describe the study design selected. 5. Facility description. Provide a complete description of the facility type and the number of beds (or patients if the facility is an outpatient facility). 6. Study setting. Select the unit(s) within the facility where the practice was implemented, e.g., inpatient, outpatient, emergency department, pediatric unit, neonatal intensivecare unit, or other. 7. Overall project/study time frame. Record the start and end dates for the new and usual practices; if pilot testing was conducted, include start/end dates for pilot testing of the new practice. Note that this is not the same as the QI study period but rather the dates during which these practices were being used in the unit(s) in which the study was done. Optimally, a power analysis should be performed prior to confirmation of sample size. Statistical power is the probability of concluding that there is a difference when there is, in fact, a difference between your standard method and your new method (i.e., the probability that your study will detect a difference, given that one truly exists). An example of a nomogram for sample size calculation can be found in reference 69.
Statistical power is the probability of concluding that there is a true and significant difference between your comparator and intervention, thus minimizing type I and type II errors (sensitivity and specificity).
For sample description, refer to the following list.
A. Random sampling. Subjects (patients) are selected for study inclusion using a formal random selection process applied to the census. B. Convenience sampling. Some subset of the census is selected since it is easy to access. For example, using only data from records of patients whom you can easily reach would be a convenience sample. C. Census sampling. All participants within a specified time period or location are used in census sampling. D. Other. Describe whether you are using a different sampling method. If you are using anything other than random sampling, convenience sampling, or census sampling, you need a statistician to identify sampling strategy.

QI Practice
9. Describe the original (usual) practice. Describe the original (usual) practice(s) that will be compared to the new practice/policy/technology implemented. 10. Describe the alternate/intervention (new) practice. Describe the new practice/policy/technology, including the characteristics and components for ongoing day-to-day operations. 11. Intervention duration dates (pilot, pre/postintervention, etc.). Record the start and end dates for the pilot testing, usual practice, and new practice. This is the date on which the particular QI project was implemented and the date on which it ended. Note that a pilot test may not have been used in this study. There should be no gaps in the QI project data collection once it begins. 12. Resource requirements/costs. Describe the requirements and costs for starting and sustaining the new practice during the study. If this information is not available, list "not known." Do not list the cost of the practice that is currently being used to do patient testing.

Outcome Measures
13. Outcome measure(s). Describe how the impact of the practice was measured. Provide the specific outcome(s) and corresponding specifications/definitions used to assess or track the impact of the practices implemented. An example is a description of how urine culture contamination rates were affected or how they had an impact on the diagnostic accuracy of urine culture.
14. Recording method. Describe each method used to collect data and to which practice (usual or new) it refers. 15. Potential sources of bias. Bias is the tendency to produce results that depart systematically from the "true" results. Bias is any nonrandom factor in the conduct of a study that can influence the results of a study. A. Selection bias. Selection bias occurs when studies are conditioned on (that is, they differentially select for) common effects of the exposure and the outcome. Selection bias occurs after exposure and arises when the associations between exposure and outcome are different for those who participate and those who do not participate in a study (i.e., all those who are theoretically eligible). This bias includes inappropriate selection of controls in a case-control study, different losses to follow-up for groups being compared (attrition bias), incidence/prevalence bias, nonresponse bias, and inclusion or exclusion of specific groups for study. B. Performance bias. Performance bias includes systematic differences in the types of care provided to participants and protocol deviations. Examples include contamination of the control group with the exposure or intervention, unbalanced provision of additional interventions or cointerventions, a difference in cointerventions, and providers and participants not being adequately blind to the study results. C. Detection bias. Detection bias includes systematic differences in outcome assessments among groups being compared. Reasons for this bias include misclassification of the exposure, intervention, covariates, or outcomes because of varying definitions, timings, diagnostic thresholds, and memories of an event; assessors not being adequately blind to the study results; and faulty measurement techniques. Erroneous statistical analysis might also affect the validity of effect estimates. D. Confounding bias. Confounding bias is the presence of systematic differences between baseline characteristics of the groups that arise when patient prognostic characteristics, such as disease severity or comorbidity, influence both treatment source and outcomes. Confounders are the common cause for intervention and exposure; they occur before exposure. Confounding by indication can occur from self-selection of treatments or physician-directed selection of treatments. E. Reporting bias. Reporting bias is the presence of systematic differences between reported and unreported findings (e.g., differential reporting of out-comes or harms, incomplete reporting of study findings, and potential for bias in reporting through source of funding).

APPENDIX 3
Refer to Tables A1 to A4 for evidence summaries of results for storage (refrigeration versus room temperature) and boric acid preservation of urine, contamination and diagnostic accuracy of urine collected from women, contamination and diagnostic accuracy of urine collected from men, and contamination and diagnostic accuracy of urine collected from children.       Description: All specimens were collected in the same room by the same 2 nurses. Both the nurses and patients wore sterile rubber gloves during the entire procedure of urine collection. The perineum was carefully scrubbed with green soap for 2 to 3 min. The labia majora were then separated, and the vulva was washed with green soap, with a fresh cotton swab used after each downward stroke. The same maneuver was repeated with an aqueous solution of Zephiran (benzalkonium; 1:1,000). The subjects were then instructed to void after separating the labia majora. After the stream was well started, a sterile screw-cap jar was placed into the path of the stream and a small sample of urine collected. Subjects from whom paired specimens were obtained were told to stop voiding as soon as the cleanly voided specimen was obtained. A sterile rubber French no. 8 catheter lubricated with a small amt of sterile lubricant was then inserted into the urethra, and a final specimen was collected in another screw-cap jar. The specimens were taken immediately to the laboratory, where they were processed for bacterial counts, routine bacterial identification, and Gram staining.
Description: Specimens were processed in the laboratory for quantitative culture with identification of uropathogens.   Description: Specimens were collected in the following order: suprapubic aspiration (10 ml), uncleansed first void (the first 10 ml was voided without prior cleansing of the urethral meatus), clean-catch midstream void (10 ml was voided after cleansing the urethral meatus with povidoneiodine and voiding ϳ100 ml), and urethral catheterization (10 ml       The urine was gently aspirated in a 5or 10-ml sterile syringe and the needle withdrawn. Cleanly voided urine from infants was collected in a sterile polyethylene urine bag after previous proper cleansing of the vulva, preputial folds, and perineum. Irrigation of the vulva and prepuce was performed twice with 5 to 10 ml tepid physiologic saline. A cleanly voided midstream specimen was obtained from children after a thorough cleansing as described above.   Description: all infants had urine collected either by CATH or SPA and by extraction from a disposable diaper. The urine was extracted from the diapers by removing the lining layer of the diaper under aseptic conditions using sterile tweezers, and then pushing the damp fibers into the barrel of a standard 20-ml disposable syringe from which the plunger had been removed. By replacing the plunger and compressing the fibers, urine was easily obtained from the diapers. Ultraabsorbent diapers that contain a gellike material were excluded from the study because extracting urine from them is difficult and time-consuming. In addition, diapers contaminated with feces or those that had been on the infant for longer than 3 h were excluded. The urine samples were sent to the laboratory and were cultured using standard bacteriologic techniques.