https://orcid.org/0000-0001-6644-717X
Abstract
Most guidelines recommend a midstream urine (MSU) or a midstream clean-catch (MSCC) sample for urinalysis. However, whether this sample is better than others is still controversial.
To assess the most adequate non-invasive method to collect a urine specimen for diagnosing urinary tract infections (UTI) in symptomatic non-pregnant women.
This review was conducted according to the Systematic Reviews of Diagnostic Test Accuracy guidelines (PROSPERO CRD42021241758). PubMed was searched paired sample studies and controlled trials. Studies comparing MSCC, MSU without cleaning, first-void urine, and random voiding samples were considered. Studies evaluating invasive methods were excluded. The main outcome was diagnostic accuracy of urine cultures. Contamination rates were evaluated. The risk of bias tool for systematic reviews on diagnostic accuracy (QUADAS-2) was assessed.
Six studies including 1,010 patients were evaluated. Only two studies used paired samples. No study was considered as having low risk of bias. There was no difference in contamination for MSU specimens collected with or without cleansing and between random void urine collection and MSCC. In one study comparing first-void urine with MSU samples, the contamination rate was lower in the latter, but the gold standard of urine culture was only used for one sampling collection.
To the best of our knowledge, this systematic review is the first to assess the evidence available from different exclusively non-invasive urine sampling. Despite being widely recommended, our review did not find consistent evidence that asking women to provide midstream samples with or without cleansing is better.
Lay Summary
Urine is one type of specimen that can be easily collected from a patient. Urinalysis testing can give the doctor valuable information about the presence of an infection in the urine and the type of microorganism causing this infection. The physician can also use the information from urine testing to diagnose and treat other diseases. The collection of the mid-stream of the urination has always been advocated. However, this recommendation has never been proven with good quality studies, and the results of the studies carried out so far have been controversial. In a systematic review, we recently determined that the use of any specimen during urination is as good as midstream collection when patients are requested to provide a urine sample and in terms of quality even specimens collected without proper cleansing are also comparable to mid-stream collection with cleansing. In the present systematic review, we evaluated the most adequate non-invasive method to collect a urine specimen for diagnosing urinary tract infections in symptomatic non-pregnant women. We identified only six studies comparing different urine sampling techniques and we did not observe any difference regarding the quality of the urine between them.
The midstream specimen is the recommended type for culture in current guidelines.
Despite being preferred, a midstream sample is still controversial in women with urinary symptoms.
There is a paucity of studies evaluating urine collection methods, with substantial heterogeneity.
Our review does not find consistent evidence about what urine sample is best.
New studies with rigorous methodologies, considering patient preferences, are needed.
Introduction
Acute urinary tract infections (UTIs) are very common among women in general practice. Urine culture is considered the gold standard method for the diagnosis of UTI, but improper sample collection can lead to contamination with normal urogenital commensals. Common uropathogens, such as Escherichia coli or enterococci, may act as contaminant flora as well. Contaminated samples may result in diagnostic misclassification, wrong therapy, overtreatment, unnecessary side events, and antibiotic resistance.
Urethral catheterization was a routine technique for urine specimens that was widely recommended in the past to decrease contamination from the skin, colon, or vaginal areas. This method along with suprapubic puncture was thought to possibly reduce contamination but these methods are associated with discomfort for the patient and risk of iatrogenic infection and other complications. In the late fifties, two papers carried out in the hospital setting reported that midstream clean-catch (MSCC) samples achieved the same results with regard to contamination rates, replacing the method to collect urine samples.1,2 The paradigm of the need to use a MSCC was later questioned, when some papers conducted mainly in primary care found no evidence that sampling technique affected the contamination or infection rate in urine specimens. A diagnostic accuracy review published in 2016 confirmed that other collection techniques, such as first-void urine (FVU), midstream urine without cleaning (MSU) or random samples delivered without instruction collected many times at home, which are much easier and more comfortable for patients, made no difference regarding the accuracy of the microbiological diagnosis in non-pregnant women with symptoms of suspected UTI.3 However, this study was exclusively carried out in the primary care setting and was limited by language restrictions. Another systematic review published in the same year showed similar results.4 Despite being recommended in men and children, MSCC samples do not yield better accuracy rates than other non-invasive collection methods in women with symptoms of suspected UTI. Some studies have demonstrated a lower contamination rate with MSCC compared to other non-invasive urine collections among asymptomatic women—for which urine cultures are not usually needed—but not in women with suspicion of having acute UTIs. A study by the College of American Pathologists showed that the collection site had no influence on the contamination rate, but post-collection processing, especially refrigeration of the specimen, had a substantial effect.5
Despite this, current guidelines continue recommending women with symptoms of UTI to provide a MSCC sample when collecting urine for culture.6, 7 However, these guidelines are based on studies including pregnant and asymptomatic patients and their conclusions do not necessarily apply to the average symptomatic women in general practice. Thus, we conducted the current systematic review aimed at evaluating the accuracy of urine culture from different non-invasive sampling techniques in symptomatic non-pregnant women.
Methods
Search strategy
This systematic review was conducted in accordance with the Systematic Reviews of Diagnostic Test Accuracy guidelines and the protocol was registered in PROSPERO (CRD42021241758).8 We searched PubMed from the inception until 19 April 2022. Reference lists and citations of included studies were backward searched for additional studies. No restrictions were applied in ways of publication language. The search contained the following medical subject headings (MESH) and key text words: ((‘urinary tract infections’[MeSH Terms] OR (‘urinary’[All Fields] AND ‘tract’[All Fields] AND ‘infections’[All Fields]) OR ‘urinary tract infection*’[All Fields]) OR (‘cystitis’[MeSH Terms] OR ‘cystitis’[All Fields]) OR ‘UTI’[All Fields] OR (‘bacteriuria’[MeSH Terms] OR ‘bacteriuria’[All Fields]) OR (‘pyuria’[MeSH Terms] OR ‘pyuria’[All Fields])) AND ((‘Urine Specimen Collection’[MeSH Terms] OR ‘collect*’[All Fields]) OR ‘Specimen Handling’[MeSH Terms] OR ‘Urinalysis’[MeSH Terms] OR ‘urine/microbiology’[MeSH Terms] OR ‘midstream’[All Fields] OR ‘clean catch’[All Fields]) NOT ((‘Animals’[MeSH Terms] NOT ‘Humans’[MeSH Terms]) OR ‘Pregnancy’[MeSH Terms] OR ‘Child’[MeSH Terms] OR ‘Infant’[MeSH Terms] OR ‘Male’[MeSH Terms]) (Supplementary Appendix 1).
Review questions and outcomes of interest
We aimed to evaluate the accuracy of urine culture from different non-invasive sampling techniques in symptomatic non-pregnant women. The patient, index test, and target condition, known as PIT for a test accuracy question, were as follows9: P. the population was constituted by any woman aged 16 or older with symptoms of UTI with urine cultures collected; I. studies comparing exclusively non-invasive sample collections were considered, such as MSCC with soap or disinfectants, the MSCC technique with the use of only water, MSU samples, FVU samples, home-voided samples with instructions, and random voiding samples; and T. the target condition was definitive diagnosis of UTI. The main outcome was the diagnostic accuracy of urine culture, but we have also determined the contamination rates of the urine collection methods. We assumed an increasing contamination rate in the order of MSCC with water and soap, MSCC with only water, MSU, and random samples or home-voided urine samples. The least contaminated was used as the reference and the most contaminated as the index test. For example, if a study investigated both MSU and random urine sampling in a paired design, MSU was used as the reference standard and random samples as the index test.
Eligibility criteria
For inclusion, studies needed to use a paired design or a controlled trial to compare the result of urine culture obtained with two or more collection techniques in self-helped, non-pregnant adult women with symptoms of acute UTI in any healthcare setting. Studies were excluded if they compared invasive methods for obtaining a urine sample, such as the use of catheters, suprapubic aspirate, cystoscopy, ureteric, ileal conduit, urostomy, or nephrostomy urine. Studies investigating patients who were asymptomatic, pregnant, children, and/or men were excluded. In studies in which symptomatic and asymptomatic patients were included and the results were presented separately, only those with symptoms of UTI were included. Studies evaluating the use of external collection devices were also excluded from this systematic review.
Screening of title and abstracts, and full text was undertaken independently by CL and AM. Any disagreement between reviewers was resolved by discussion. If consensus could not be reached, review was undertaken by another reviewer (AGS.).
Data extraction and analysis
A standardized form was used to extract data from the studies included, including first author, year of publication, country, study design, inclusion and exclusion criteria for the study, population demographics, setting of the study, number of patients enrolled, and index (details of the non-invasive urine collection methods) and the reference test with assigned cut-off values for UTI vs. contamination. When two or more studies used the same data source, the study with the largest sample size was used. Two reviewers independently (CL and AM) extracted the required information from the primary studies. Disagreements were resolved by a third investigator by discussion (AGS).
A descriptive analysis was used in controlled trials and overall agreement rates were calculated for paired sample studies. We planned to combine the outcomes from the individual trials through meta-analysis whenever possible (comparability of intervention and outcomes between studies) using a random-effects model because a certain degree of heterogeneity was expected among individual studies. However, due to small number of studies included, the different types of urine collection methods used, and the heterogeneity of outcomes used in the different studies, a meta-analysis to explore the accuracy of the different sampling collection was not performed and a narrative synthesis was conducted.
Risk of bias assessment
The risk of bias of each study included was assessed by two independent authors (CL and AM) using the Quality Assessment of Diagnostic Accuracy Studies tool (QUADAS-2).10 Any disagreement was resolved by discussion. Four components were assessed: patient selection, index test, reference standard, flow, and timing. Each study was scored according to whether the assessment criteria are met or not, and then classified as being of “high risk,” “low risk,” or “unclear risk” of bias.
Data analysis
The specified dichotomized outcomes were used to calculate predictive values, sensitivity, and specificity in paired studies. The generated sensitivity and specificity values were used to create forest plots on the diagnostic accuracy.
Results
The MEDLINE search yielded 4,754 articles. After review of titles and abstracts, we included five full text articles presenting results of studies investigating non-invasive urine sampling collection techniques in non-pregnant women with symptoms of UTI. After screening the reference lists and citations of these five studies, one additional study was included for the final review out of five new studies, leading to a total of six studies included in the final synthesis.11–16 Twelve studies compared invasive methods and eleven included asymptomatic women. A flow diagram of the literature search and review of titles, abstracts, and articles is shown in Fig. 1.
Flowchart of study inclusion.
Characteristics of the studies included
The studies included are shown in Table 1. The quality of the studies included is summarized in Table 2. Generally, the studies were judged to be of moderate or high risk of bias as no study was considered as having low risk of bias. The full quality assessment is described in Supplementary Appendix 2. The oldest report was from 1979, whereas the latest report was from 2018. Four of the studies were carried out in general practice, one in university clinics and another study was performed in emergency departments, including a total of 1,010 women. Only two of the studies considered random samples.
Characteristics of the studies included.
| Author | Setting, country | Population | Comparison | Diagnostic accuracy results | Contamination results |
|---|---|---|---|---|---|
| Morris,11 1979 | General practice, UK | n = 180. Age not mentioned | Unclear how patients were allocated. Two groups (unclear if there were 40 paired samples): MSU collected in surgeries (n = 140) vs. MSCC after cleansing with sterile water at home and supervised by nurses (n = 40) | Definitive infection (≥105 CFU/ml of a single microorganism) in 91% in the MSU group without cleansing and 92% in the MSCC group after cleansing | Contamination was defined as mixed growths of ≥ 105 CFU/ml or 104–105. Contaminated samples in 9% in the MSU group and 8% in the MSCC group |
| Bradbury,12 1988 | General practice, UK | n = 158 aged 16-75 yr. | RCT. MSU without cleansing (n = 65) vs. MSCC after cleansing with water and soup (n = 93) | Definitive infection (≥105 CFU/ml) in 16 (24.6%) cases in MSU samples and 23 (24.7%) specimens with MSCC | Contamination was defined by the presence of epithelial cells. There were 6 contaminated samples (9.2%) in the MSU group vs. 8 (8.6%) in the MSCC group |
| Bærheim,13 1990 | General practice, Norway | n = 73 aged 18-60 yr. | Paired samples. Home voided urine samples vs. MSCC after cleansing with water (n = 73) | Bacteriuria, considered with a cut-off point of ≥ 104 CFU/ml, was observed in 52 (71.2%) specimens in home voided samples and 54 (73.9%) in MSCC samples. The overall agreement rates were Κ=0.70 with a cut-off point of 104 CFU/ml and Κ=0.74 with 105 CFU/ml | Contaminated samples (coagulase-negative staphylococci and other gram-positives) in 3 (4.1%) and 7 (9.6%) samples, respectively |
| Lifshitz,14 2000 | University clinic, US | n = 242 aged 17–50 yr | RCT. Three groups: 1. Urine into a clean container (no cleansing, no midstream) (n = 77); 2. MSCC after cleansing with water and bactericidal wipe (n = 84); 3. MSCC after cleansing with bactericidal wipe and insertion of a vaginal tampon prior to urine collection (n = 81) | Definitive infection defined as the growth of a germ ≥ 104 CFU/ml in 44 specimens of the group without cleansing (57.1%), 42 in the MSCC group without tampon (50%) and 46 in the MSCC group with tampon (56.8%) | Contamination defined as mixed growth and low levels (<104 CFU/ml) of organisms commonly found on the skin and external and internal genitalia. Contaminated samples in 22 samples of the group without cleansing (28.6), 27 in the MSCC group without tampon (32.1%) and 25 patients with MSCC + tampon insertion (30.9%) |
| Eley,152016 | Emergency department, Australia | n = 240a, aged 18 or over | Pseudo-RCT. The two groups provided MSCC samples after cleansing with water and a towelette. Only verbal instructions (n = 120) vs. illustrated instructions (n = 120) | Definitive infections confirmed in 11 cases in the group assigned to verbal instructions (9.2%) and in 15 cases among those with illustrated instructions (12.5%) | Contaminated samples were defined as the presence of 10 or more epithelial cells per high power field. Differences in contamination. Contamination rates in 47 (39.2%) and 30 (25%) cases, respectively |
| Hølmkjær,16 2018 | General practice, Denmark | n = 117 aged 18 or older | Paired samples. FVU vs MSU (n = 117) | Definitive infection was considered with a cut-off point of ≥ 103 CFU/ml analysed immediately after collection. Overall agreement of the FVU and MSU collection was observed in 90 cases (76.9%) and 98 cases (83.8%), respectivelyb | No data |
| Author | Setting, country | Population | Comparison | Diagnostic accuracy results | Contamination results |
|---|---|---|---|---|---|
| Morris,11 1979 | General practice, UK | n = 180. Age not mentioned | Unclear how patients were allocated. Two groups (unclear if there were 40 paired samples): MSU collected in surgeries (n = 140) vs. MSCC after cleansing with sterile water at home and supervised by nurses (n = 40) | Definitive infection (≥105 CFU/ml of a single microorganism) in 91% in the MSU group without cleansing and 92% in the MSCC group after cleansing | Contamination was defined as mixed growths of ≥ 105 CFU/ml or 104–105. Contaminated samples in 9% in the MSU group and 8% in the MSCC group |
| Bradbury,12 1988 | General practice, UK | n = 158 aged 16-75 yr. | RCT. MSU without cleansing (n = 65) vs. MSCC after cleansing with water and soup (n = 93) | Definitive infection (≥105 CFU/ml) in 16 (24.6%) cases in MSU samples and 23 (24.7%) specimens with MSCC | Contamination was defined by the presence of epithelial cells. There were 6 contaminated samples (9.2%) in the MSU group vs. 8 (8.6%) in the MSCC group |
| Bærheim,13 1990 | General practice, Norway | n = 73 aged 18-60 yr. | Paired samples. Home voided urine samples vs. MSCC after cleansing with water (n = 73) | Bacteriuria, considered with a cut-off point of ≥ 104 CFU/ml, was observed in 52 (71.2%) specimens in home voided samples and 54 (73.9%) in MSCC samples. The overall agreement rates were Κ=0.70 with a cut-off point of 104 CFU/ml and Κ=0.74 with 105 CFU/ml | Contaminated samples (coagulase-negative staphylococci and other gram-positives) in 3 (4.1%) and 7 (9.6%) samples, respectively |
| Lifshitz,14 2000 | University clinic, US | n = 242 aged 17–50 yr | RCT. Three groups: 1. Urine into a clean container (no cleansing, no midstream) (n = 77); 2. MSCC after cleansing with water and bactericidal wipe (n = 84); 3. MSCC after cleansing with bactericidal wipe and insertion of a vaginal tampon prior to urine collection (n = 81) | Definitive infection defined as the growth of a germ ≥ 104 CFU/ml in 44 specimens of the group without cleansing (57.1%), 42 in the MSCC group without tampon (50%) and 46 in the MSCC group with tampon (56.8%) | Contamination defined as mixed growth and low levels (<104 CFU/ml) of organisms commonly found on the skin and external and internal genitalia. Contaminated samples in 22 samples of the group without cleansing (28.6), 27 in the MSCC group without tampon (32.1%) and 25 patients with MSCC + tampon insertion (30.9%) |
| Eley,152016 | Emergency department, Australia | n = 240a, aged 18 or over | Pseudo-RCT. The two groups provided MSCC samples after cleansing with water and a towelette. Only verbal instructions (n = 120) vs. illustrated instructions (n = 120) | Definitive infections confirmed in 11 cases in the group assigned to verbal instructions (9.2%) and in 15 cases among those with illustrated instructions (12.5%) | Contaminated samples were defined as the presence of 10 or more epithelial cells per high power field. Differences in contamination. Contamination rates in 47 (39.2%) and 30 (25%) cases, respectively |
| Hølmkjær,16 2018 | General practice, Denmark | n = 117 aged 18 or older | Paired samples. FVU vs MSU (n = 117) | Definitive infection was considered with a cut-off point of ≥ 103 CFU/ml analysed immediately after collection. Overall agreement of the FVU and MSU collection was observed in 90 cases (76.9%) and 98 cases (83.8%), respectivelyb | No data |
CFU, colony-forming unit; FVU, first voided urine; MSCC, midstream clean-catch; MSU, mid-stream urine; RCT, randomized clinical trial.
Women presenting symptoms of infection (e.g., flank pain, painful urination, and fever) for which urinalysis was clinically required.
Only the MSU samples were sent to the microbiology lab for urine culture.
Characteristics of the studies included.
| Author | Setting, country | Population | Comparison | Diagnostic accuracy results | Contamination results |
|---|---|---|---|---|---|
| Morris,11 1979 | General practice, UK | n = 180. Age not mentioned | Unclear how patients were allocated. Two groups (unclear if there were 40 paired samples): MSU collected in surgeries (n = 140) vs. MSCC after cleansing with sterile water at home and supervised by nurses (n = 40) | Definitive infection (≥105 CFU/ml of a single microorganism) in 91% in the MSU group without cleansing and 92% in the MSCC group after cleansing | Contamination was defined as mixed growths of ≥ 105 CFU/ml or 104–105. Contaminated samples in 9% in the MSU group and 8% in the MSCC group |
| Bradbury,12 1988 | General practice, UK | n = 158 aged 16-75 yr. | RCT. MSU without cleansing (n = 65) vs. MSCC after cleansing with water and soup (n = 93) | Definitive infection (≥105 CFU/ml) in 16 (24.6%) cases in MSU samples and 23 (24.7%) specimens with MSCC | Contamination was defined by the presence of epithelial cells. There were 6 contaminated samples (9.2%) in the MSU group vs. 8 (8.6%) in the MSCC group |
| Bærheim,13 1990 | General practice, Norway | n = 73 aged 18-60 yr. | Paired samples. Home voided urine samples vs. MSCC after cleansing with water (n = 73) | Bacteriuria, considered with a cut-off point of ≥ 104 CFU/ml, was observed in 52 (71.2%) specimens in home voided samples and 54 (73.9%) in MSCC samples. The overall agreement rates were Κ=0.70 with a cut-off point of 104 CFU/ml and Κ=0.74 with 105 CFU/ml | Contaminated samples (coagulase-negative staphylococci and other gram-positives) in 3 (4.1%) and 7 (9.6%) samples, respectively |
| Lifshitz,14 2000 | University clinic, US | n = 242 aged 17–50 yr | RCT. Three groups: 1. Urine into a clean container (no cleansing, no midstream) (n = 77); 2. MSCC after cleansing with water and bactericidal wipe (n = 84); 3. MSCC after cleansing with bactericidal wipe and insertion of a vaginal tampon prior to urine collection (n = 81) | Definitive infection defined as the growth of a germ ≥ 104 CFU/ml in 44 specimens of the group without cleansing (57.1%), 42 in the MSCC group without tampon (50%) and 46 in the MSCC group with tampon (56.8%) | Contamination defined as mixed growth and low levels (<104 CFU/ml) of organisms commonly found on the skin and external and internal genitalia. Contaminated samples in 22 samples of the group without cleansing (28.6), 27 in the MSCC group without tampon (32.1%) and 25 patients with MSCC + tampon insertion (30.9%) |
| Eley,152016 | Emergency department, Australia | n = 240a, aged 18 or over | Pseudo-RCT. The two groups provided MSCC samples after cleansing with water and a towelette. Only verbal instructions (n = 120) vs. illustrated instructions (n = 120) | Definitive infections confirmed in 11 cases in the group assigned to verbal instructions (9.2%) and in 15 cases among those with illustrated instructions (12.5%) | Contaminated samples were defined as the presence of 10 or more epithelial cells per high power field. Differences in contamination. Contamination rates in 47 (39.2%) and 30 (25%) cases, respectively |
| Hølmkjær,16 2018 | General practice, Denmark | n = 117 aged 18 or older | Paired samples. FVU vs MSU (n = 117) | Definitive infection was considered with a cut-off point of ≥ 103 CFU/ml analysed immediately after collection. Overall agreement of the FVU and MSU collection was observed in 90 cases (76.9%) and 98 cases (83.8%), respectivelyb | No data |
| Author | Setting, country | Population | Comparison | Diagnostic accuracy results | Contamination results |
|---|---|---|---|---|---|
| Morris,11 1979 | General practice, UK | n = 180. Age not mentioned | Unclear how patients were allocated. Two groups (unclear if there were 40 paired samples): MSU collected in surgeries (n = 140) vs. MSCC after cleansing with sterile water at home and supervised by nurses (n = 40) | Definitive infection (≥105 CFU/ml of a single microorganism) in 91% in the MSU group without cleansing and 92% in the MSCC group after cleansing | Contamination was defined as mixed growths of ≥ 105 CFU/ml or 104–105. Contaminated samples in 9% in the MSU group and 8% in the MSCC group |
| Bradbury,12 1988 | General practice, UK | n = 158 aged 16-75 yr. | RCT. MSU without cleansing (n = 65) vs. MSCC after cleansing with water and soup (n = 93) | Definitive infection (≥105 CFU/ml) in 16 (24.6%) cases in MSU samples and 23 (24.7%) specimens with MSCC | Contamination was defined by the presence of epithelial cells. There were 6 contaminated samples (9.2%) in the MSU group vs. 8 (8.6%) in the MSCC group |
| Bærheim,13 1990 | General practice, Norway | n = 73 aged 18-60 yr. | Paired samples. Home voided urine samples vs. MSCC after cleansing with water (n = 73) | Bacteriuria, considered with a cut-off point of ≥ 104 CFU/ml, was observed in 52 (71.2%) specimens in home voided samples and 54 (73.9%) in MSCC samples. The overall agreement rates were Κ=0.70 with a cut-off point of 104 CFU/ml and Κ=0.74 with 105 CFU/ml | Contaminated samples (coagulase-negative staphylococci and other gram-positives) in 3 (4.1%) and 7 (9.6%) samples, respectively |
| Lifshitz,14 2000 | University clinic, US | n = 242 aged 17–50 yr | RCT. Three groups: 1. Urine into a clean container (no cleansing, no midstream) (n = 77); 2. MSCC after cleansing with water and bactericidal wipe (n = 84); 3. MSCC after cleansing with bactericidal wipe and insertion of a vaginal tampon prior to urine collection (n = 81) | Definitive infection defined as the growth of a germ ≥ 104 CFU/ml in 44 specimens of the group without cleansing (57.1%), 42 in the MSCC group without tampon (50%) and 46 in the MSCC group with tampon (56.8%) | Contamination defined as mixed growth and low levels (<104 CFU/ml) of organisms commonly found on the skin and external and internal genitalia. Contaminated samples in 22 samples of the group without cleansing (28.6), 27 in the MSCC group without tampon (32.1%) and 25 patients with MSCC + tampon insertion (30.9%) |
| Eley,152016 | Emergency department, Australia | n = 240a, aged 18 or over | Pseudo-RCT. The two groups provided MSCC samples after cleansing with water and a towelette. Only verbal instructions (n = 120) vs. illustrated instructions (n = 120) | Definitive infections confirmed in 11 cases in the group assigned to verbal instructions (9.2%) and in 15 cases among those with illustrated instructions (12.5%) | Contaminated samples were defined as the presence of 10 or more epithelial cells per high power field. Differences in contamination. Contamination rates in 47 (39.2%) and 30 (25%) cases, respectively |
| Hølmkjær,16 2018 | General practice, Denmark | n = 117 aged 18 or older | Paired samples. FVU vs MSU (n = 117) | Definitive infection was considered with a cut-off point of ≥ 103 CFU/ml analysed immediately after collection. Overall agreement of the FVU and MSU collection was observed in 90 cases (76.9%) and 98 cases (83.8%), respectivelyb | No data |
CFU, colony-forming unit; FVU, first voided urine; MSCC, midstream clean-catch; MSU, mid-stream urine; RCT, randomized clinical trial.
Women presenting symptoms of infection (e.g., flank pain, painful urination, and fever) for which urinalysis was clinically required.
Only the MSU samples were sent to the microbiology lab for urine culture.
Quality of studies assessed using QUADAS 2.
| Study, year | Risk of bias | Applicability concerns | |||||
|---|---|---|---|---|---|---|---|
| Patient selection | Index test | Reference standard | Flow and timing | Patient selection | Index test | Reference standard | |
| Morris, 1979 | |||||||
| Bradbury, 1988 | |||||||
| Bærheim, 1990 | |||||||
| Lifshitz, 2000 | |||||||
| Eley, 2016 | |||||||
| Hølmkjær, 2018 | |||||||
| Study, year | Risk of bias | Applicability concerns | |||||
|---|---|---|---|---|---|---|---|
| Patient selection | Index test | Reference standard | Flow and timing | Patient selection | Index test | Reference standard | |
| Morris, 1979 | |||||||
| Bradbury, 1988 | |||||||
| Bærheim, 1990 | |||||||
| Lifshitz, 2000 | |||||||
| Eley, 2016 | |||||||
| Hølmkjær, 2018 | |||||||
Quality of studies assessed using QUADAS 2.
| Study, year | Risk of bias | Applicability concerns | |||||
|---|---|---|---|---|---|---|---|
| Patient selection | Index test | Reference standard | Flow and timing | Patient selection | Index test | Reference standard | |
| Morris, 1979 | |||||||
| Bradbury, 1988 | |||||||
| Bærheim, 1990 | |||||||
| Lifshitz, 2000 | |||||||
| Eley, 2016 | |||||||
| Hølmkjær, 2018 | |||||||
| Study, year | Risk of bias | Applicability concerns | |||||
|---|---|---|---|---|---|---|---|
| Patient selection | Index test | Reference standard | Flow and timing | Patient selection | Index test | Reference standard | |
| Morris, 1979 | |||||||
| Bradbury, 1988 | |||||||
| Bærheim, 1990 | |||||||
| Lifshitz, 2000 | |||||||
| Eley, 2016 | |||||||
| Hølmkjær, 2018 | |||||||
Main results
Two studies examined the impact of perineal cleansing on contamination, comparing MSCC and MSU samples.11,12 A meta-analysis of the two studies could not be performed as the definitions of contamination varied between the two studies and included the presence of mixed growth in quantities of >105 colony-forming units (CFU)/ml or 104–105 CFU/ml,11 and the presence of epithelial cells.12 No differences in contaminated samples were observed between the two collection methods. Bærheim et al.13 examined the diagnostic accuracy of home voiding specimens with MSCC among women with suspected UTIs, with similar results between the two groups. Lifshitz et al.14 conducted a randomized clinical trial allocating adult women with suspected UTIs to any of these three arms: a first group with random voiding samples and two groups of women assigned to MSCC with two levels of intervention: the first group carried out the normal procedure of cleansing the perineal area with the use of a bactericidal wipe and the second group was requested to do the same but also inserting a vaginal tampon prior to the urine sample collection. The number of contaminated samples was slightly lower when random voiding specimens were collected, albeit without statistical differences.
Eley et al.15 compared two groups of women in a pseudo-randomized clinical trial. The first group of women was asked to provide MSCC samples after receiving verbal instructions and the second group was given illustrated instructions. The number of contaminated samples, defined as the presence of 10 or more epithelial cells per high power field, was significantly higher in the first group (47, 39.2% vs. 30, 25%). Patient populations in this study included women aged 18 or older presenting to an emergency department with symptoms of infection—flank pain, painful urination, or fever—for which a urinalysis was clinically required. However, a clear-cut UTI was observed in 26 women, accounting for 10.8% of all the patients included.
A recent Danish paper provided new evidence about the accuracy of MSU samples over FVU samples in a paired sample study performed in general practice.16 This study examined the overall agreement of urine collection at different times with the current endorsed gold standard of considering UTI above the cut-off point of 1,000 CFU/ml. When the analysis was performed immediately after urine collection the overall agreement was much higher when a MSU sample was considered (83.8% vs. 76.9% observed in FVU specimens). The reason for the lower accuracy with FVU was mainly due to the presence of Enterococcus spp. in the sample. However, the gold standard of urine culture was only considered for MSU samples and not for those specimens which were collected with FVU.
Table 3 shows the accuracy parameters of the paired sample studies. However, this information could not be obtained from one of the studies. In addition, because of the randomized design, accuracy could not be calculated from the other two studies.
Diagnostic accuracy of the different studies (95% confidence intervals in brackets).
| Home-voided samples vs. midstream-clean-catch samples | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Study | Cut-off point | n | TP | TN | FP | FN | Sens. | Spec. | PPV | NPV |
| Bærheim,13 1990 | ≥104 CFU/ml | 73 | 48 | 15 | 6 | 4 | 0.92 (0.81–0.98) | 0.71 (0.48–0.89) | 0.89 (0.80–0.94) | 0.79 (0.58–0.91) |
| Home-voided samples vs. midstream-clean-catch samples | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Study | Cut-off point | n | TP | TN | FP | FN | Sens. | Spec. | PPV | NPV |
| Bærheim,13 1990 | ≥104 CFU/ml | 73 | 48 | 15 | 6 | 4 | 0.92 (0.81–0.98) | 0.71 (0.48–0.89) | 0.89 (0.80–0.94) | 0.79 (0.58–0.91) |
| Midstream urine without cleansing vs. midstream-clean-catch samples | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Study | Cut-off point | n | TP | TN | FP | FN | Sens. | Spec. | PPV | NPV |
| Bradbury,12 1988 | ≥105 CFU/ml | 158 | 12 | 131 | 11 | 4 | 0.75 (0.48–0.93) | 0.92 (0.87–0.96) | 0.52 (0.37–0.67) | 0.97 (0.37–0.67) |
| Midstream urine without cleansing vs. midstream-clean-catch samples | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Study | Cut-off point | n | TP | TN | FP | FN | Sens. | Spec. | PPV | NPV |
| Bradbury,12 1988 | ≥105 CFU/ml | 158 | 12 | 131 | 11 | 4 | 0.75 (0.48–0.93) | 0.92 (0.87–0.96) | 0.52 (0.37–0.67) | 0.97 (0.37–0.67) |
| First-void urine vs. midstream urine samples | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Study | Cut-off point | n | TP | TN | FP | FN | Sens. | Spec. | PPV | NPV |
| Hølmkjær,16 2018a | ≥103 CFU/ml | 117 | 89 | 18 | 9 | 1 | 0.99 (0.94–1.00) | 0.67 (0.46–0.83) | 0.91 (0.85–0.94) | 0.95 (0.72–0.99) |
| First-void urine vs. midstream urine samples | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Study | Cut-off point | n | TP | TN | FP | FN | Sens. | Spec. | PPV | NPV |
| Hølmkjær,16 2018a | ≥103 CFU/ml | 117 | 89 | 18 | 9 | 1 | 0.99 (0.94–1.00) | 0.67 (0.46–0.83) | 0.91 (0.85–0.94) | 0.95 (0.72–0.99) |
CFU, colony-forming units; FN, false negatives; FP, false positives; NPV, negative predictive value; PPV, positive predictive value; Sens., sensitivity; Spec., specificity; TN, true negatives; TP, true positives.
Only the MSU samples were sent to the microbiology lab for urine culture.
Diagnostic accuracy of the different studies (95% confidence intervals in brackets).
| Home-voided samples vs. midstream-clean-catch samples | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Study | Cut-off point | n | TP | TN | FP | FN | Sens. | Spec. | PPV | NPV |
| Bærheim,13 1990 | ≥104 CFU/ml | 73 | 48 | 15 | 6 | 4 | 0.92 (0.81–0.98) | 0.71 (0.48–0.89) | 0.89 (0.80–0.94) | 0.79 (0.58–0.91) |
| Home-voided samples vs. midstream-clean-catch samples | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Study | Cut-off point | n | TP | TN | FP | FN | Sens. | Spec. | PPV | NPV |
| Bærheim,13 1990 | ≥104 CFU/ml | 73 | 48 | 15 | 6 | 4 | 0.92 (0.81–0.98) | 0.71 (0.48–0.89) | 0.89 (0.80–0.94) | 0.79 (0.58–0.91) |
| Midstream urine without cleansing vs. midstream-clean-catch samples | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Study | Cut-off point | n | TP | TN | FP | FN | Sens. | Spec. | PPV | NPV |
| Bradbury,12 1988 | ≥105 CFU/ml | 158 | 12 | 131 | 11 | 4 | 0.75 (0.48–0.93) | 0.92 (0.87–0.96) | 0.52 (0.37–0.67) | 0.97 (0.37–0.67) |
| Midstream urine without cleansing vs. midstream-clean-catch samples | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Study | Cut-off point | n | TP | TN | FP | FN | Sens. | Spec. | PPV | NPV |
| Bradbury,12 1988 | ≥105 CFU/ml | 158 | 12 | 131 | 11 | 4 | 0.75 (0.48–0.93) | 0.92 (0.87–0.96) | 0.52 (0.37–0.67) | 0.97 (0.37–0.67) |
| First-void urine vs. midstream urine samples | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Study | Cut-off point | n | TP | TN | FP | FN | Sens. | Spec. | PPV | NPV |
| Hølmkjær,16 2018a | ≥103 CFU/ml | 117 | 89 | 18 | 9 | 1 | 0.99 (0.94–1.00) | 0.67 (0.46–0.83) | 0.91 (0.85–0.94) | 0.95 (0.72–0.99) |
| First-void urine vs. midstream urine samples | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Study | Cut-off point | n | TP | TN | FP | FN | Sens. | Spec. | PPV | NPV |
| Hølmkjær,16 2018a | ≥103 CFU/ml | 117 | 89 | 18 | 9 | 1 | 0.99 (0.94–1.00) | 0.67 (0.46–0.83) | 0.91 (0.85–0.94) | 0.95 (0.72–0.99) |
CFU, colony-forming units; FN, false negatives; FP, false positives; NPV, negative predictive value; PPV, positive predictive value; Sens., sensitivity; Spec., specificity; TN, true negatives; TP, true positives.
Only the MSU samples were sent to the microbiology lab for urine culture.
Discussion
Summary
To the best of our knowledge, this systematic review is the first to assess the evidence available from different exclusively non-invasive urine sampling techniques in symptomatic adult women with suspected UTI. Overall, we did not find consistent evidence to suggest important differences in diagnostic accuracy or the percentage of contaminated samples among the different sampling collection techniques in the studies included. The overall strength of evidence was low, as multivariate modelling could not be performed, and thus, no recommendation for or against can yet be made.
Limitations and strengths
Only six studies evaluated the diagnostic accuracy of different sampling collection techniques among women with symptoms of UTI, which is the population with the greatest potential for clear benefit from the collection of urine specimens. It is disappointing that so many of the studies originally identified had to be excluded, but there were clear reasons for the exclusion. The majority of studies excluded used invasive collection methods or included patents without symptoms suggestive of having an UTI. The inclusion of only these six studies, three of which published in the last century, constitutes the major limitation of this systematic review as evidence is insufficient for drawing reliable conclusions about which urine collection specimen best correlates with the microbiological diagnosis of UTI and is associated with the lowest contamination rate.
In addition, the two papers that have been published more recently had some methodological issues that must be considered. Eley et al.15 did not compare collection samplings as they only compared verbal instructions with illustrative instructions in MSCC samplings and they managed to include only clear UTIs in a small percentage of cases. We cannot assume that the results obtained would have been the same if the authors had only included women with suspected UTI. In the recent study by Hølmkjær et al.,16 MSU collection was significantly more accurate than FVU sampling immediately after voiding when the results were compared with point of care tests as gold standard, such as the use of FlexiCult or the phase-contrast microscopy reading. However, the correlation between these rapid tests and urine culture is suboptimal.17,18 In addition, in this study, the diagnostic accuracy was only analysed in only one of the sampling collection methods. Due to omissions of the latter two studies and the different definitions of definitive infection used across the different studies as well as the different contamination definitions, we were unable to carry out a meta-analysis of the two outcomes. Notwithstanding, our systematic review is comprehensive, with the screening of nearly 5,000 references and actively reviewed all the references of the most important studies, in the search for other studies that might not have been identified in the initial search.
Comparison with the existing literature
In general, practice simplicity should be adopted to avoid overcomplicating patient management.19 Obviously, a non-invasive collection method should be recommended, but not all these collection methods are simple to perform. The different techniques differ extensively in preparation time and discomfort. We have always been told to recommend the use of midstream samples when collecting a specimen of urine for culture, with or without previous cleansing and with or without soap or disinfectants. When collecting a MSCC sample, women are required to wash their hands first and then use cottonwool swabs provided to cleanse the genital area with soap and water. They are told to spread the labia and wipe from front to back several times using a clean swab each time. Keeping the labia parted they pass some urine into the toilet discarding the first sample, then squeeze off and stop urinating, and then catch the next urine sample—the midstream sample—in a sterile container. Therefore, a midstream urine sample is not always easy to collect, mainly among elderly patients, who might experience unpleasant and uncomfortable cleansing procedures and collection postures. For instance, Bærheim et al.20 observed that only 10% of patients hold the labia apart during voiding. The usage of these instructions is therefore variable, and we certainly do not know how women collect the urine samples despite being instructed to collect a midstream urine sample.
There is no doubt that the percentage of contaminated urine samples should be reduced as this hampers proper diagnosis and subsequent treatment of a possible UTI. When comparing collection with non-invasive procedures with invasive sampling techniques, such as suprapubic puncture and catheter, Holm et al.3 observed that 5%–10% of healthy patients were over-diagnosed. This small percentage between non-invasive and invasive methods does not offset the degree of discomfort and risk of complications associated with these allegedly sterile techniques from a patient-centred perspective. As one is reliant on patient cooperation to produce samples, the easier and more acceptable the method the better.
Urine specimens can easily become contaminated with periurethral, epidermal, perianal, and vaginal flora. However, this systematic review fails to answer why there are so many differences regarding the contamination rates across microbiology departments. A study conducted by the College of American Pathologists in 2008 examined the frequency of urine culture contamination, ranging from 42% in the low-performance facilities to 1% in the highest performers.5 The authors concluded that contamination rates were mainly affected by post-collection processing, especially refrigeration of the specimen, and not by the way these urine samples were collected. Current recommendations about the time a urine sample can sit at room temperature suggest that samples should not be of more than half an hour.7
Implications of this study
The relative paucity of rigorous studies evaluating methods of urine collection is troublesome considering the widespread use of urine collection for dipstick analysis and urine cultures. Few studies have been performed in patients with symptoms with suspected UTIs, and most of the studies presented substantial heterogeneity regarding study designs, outcomes, and the diagnostic cut-offs applied. These differences suggest that the overall results regarding their diagnostic accuracy should be considered with caution. We do not know if the results obtained with the studies performed in the twentieth century would still apply today with the current microbiological procedures.
New studies are needed and they should strive for statistically sufficient sample sizes, use common and clear definitions of contamination and updated thresholds for positivity. Studies should be more rigorous in design, by using paired samples to better evaluate the comparison of the different non-invasive collection methods in the same patients, with the use of the same gold standard for all the specimens collected within a reasonable time frame. Very importantly, patient preferences should be taken into account. Until new studies shed new evidence, each healthcare professional should freely choose the most appropriate sampling technique, depending on the ease and convenience for patients and practices.
Authors’ contributions
CL and MA-S designed the review. RMor coordinated the review. AM, AGS, and RMon designed the search strategy. AM undertook the searches, and CL and AM screened the search results against eligibility criteria. AGS resolved conflicts. CL and AM appraised quality. CL and AM extracted the data, and CL analysed the data. CL, AM, MA-S, and RMon interpreted the data. RMor provided general advice on the review. RMor is the guarantor of this manuscript. All authors reviewed the final manuscript. CL attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted.
Funding
This study did not receive any specific funding.
Ethical approval
Not needed. This systematic review was conducted in accordance with the Systematic Reviews of Diagnostic Test Accuracy guidelines and the protocol was registered in PROSPERO (CRD42021241758).
Conflict of interest
CL declares having reported funds for research from Abbott Diagnostics. The other authors declared no conflicts of interest.
Data availability statement
The data underlying this article will be shared on reasonable request to the corresponding author.
