Article, Urology

A prospective study to assess the association of body mass index and contamination of urinalysis samples

a b s t r a c t

Background: There is a commonly held belief that overweight women are more likely to offer contaminated Urine samples (UAs) in the emergency department (ED) than women with normal body mass index (BMI). However, there is a paucity of research evaluating this potential concern.

Objective: We hypothesized that patients with higher BMI would be more likely to provide contaminated urine samples than women with low BMI. Methods: This was a prospective, observational, cohort study evaluating consenting, adult, women that provided a clean catch, mid-stream sample at an inner-city ED. UAs were ordered at the discretion of the caring physician, cultures based on standardized parameters. The primary outcome parameter was the presence of UA contamina- tion as defined by our microbiology lab. Demographic/historical data and BMI were recorded on a structured data sheet. Categorical data were analyzed by chi-square; continuous data by t-tests. Multivariable logistic regression was performed to control for confounding.

Results: There were 350 patients in the study group; 22% overweight, 35% obese, 17% morbidly obese, mean BMI

31. 5, and 60% provided contaminated specimens. The mean BMIs of the subjects with contaminated vs. uncon- taminated UAs were significantly different (32.7 +- 10.2 vs 29.7 +- 8.8, p b 0.01). Within our multiple variable lo- gistic regression model, obese and morbidly obese patients were more likely to provide contaminated UAs, while there were no significant associations for contamination with other variables except for hypertension (OR = 1.85, p = 0.02).

Conclusion: Obesity was significantly associated with contamination of clean catch mid-stream samples in our population.

(C) 2019

Introduction

Urinalysis (UA) is performed in the emergency department (ED) to identify a wide scope of potential pathologies. These pathologies include presence of calculi, malignancy, and the full range of urinary tract infec- tions (UTI) from urethritis to pyelonephritis [1]. Although physicians frequently order UA in the ED, thought must be given to whether or not the data obtained from the test are reliable for various subsets of pa- tients. Tintinalli’s Emergency Medicine notes that, “Catheterization is

? We acknowledge the Office of Research and Academics, CHRISTUS Health, for their administrative support of this project.

* Corresponding author at: Department of Emergency Medicine, Texas A&M Health Science Center/CHRISTUS Spohn, Corpus Christi, TX 78405, United States of America.

E-mail address: [email protected] (P. Richman).

indicated if the patient cannot void spontaneously, is too ill or immobilized, or is extremely obese” [2].

Obese women represent an important subset of emergency depart- ment patients. In 2010 N 35% of all men and women older than 20 were obese [3]. In addition, that number does not include those who were overweight. Obesity as a condition increases risk for several condi- tions including hypertension, type 2 diabetes mellitus, infections, and lower urinary tract symptoms including urinary incontinence [3-6].

To date, investigators have provided surprisingly limited evidence to support both seasoned practitioners’ gestalt and highly regarded text- books respectively that recommend that obese women be catheterized for sample collection [1-4,7-11]. Within our Medline search of the liter- ature, we identified only one prior retrospective study that investigated the relationship between BMI and UA contamination [7]. In view of the paucity of investigation for this concern, we conducted a prospective

https://doi.org/10.1016/j.ajem.2019.05.036 0735-6757/(C) 2019

M. Strawn et al. / American Journal of Emergency Medicine 38 (2020) 508511 509

study to examine to evaluate the association between BMI and UA contamination.

Methods

Study design

This was a prospective cohort study of patients who underwent uri- nalysis as a component of their ED work-up.

Setting

The study was conducted at XXXXXX. The facility is a teaching affil- iate of the XXXX Health Science Center, a level-two trauma center, and serves an inner-city population. The annual Emergency Department census is 45,000 patients. The study was approved by the XXXXX Insti- tutional Review Board prior to initiation of data collection.

Population

We enrolled a convenience sample of consenting, stable, oriented, female patients, age >=18 years who presented to the ED and had UA or- dered by their treating physician were eligible for inclusion. We ex- cluded pregnant and incarcerated patients.

Study protocol

The primary investigator identified and enrolled eligible patients during a series of dedicated study hours that varied in day of the week

4) Both infection and contamination (growth of N100,000 cfu of typical pathogens but also with growth of mixed skin flora).

Statistical analysis

With respect to weight classifications, normal weight and under- weight were defined as BMI b25, overweight as >=25 to b30, obese as

>=30 to b40, 40+ as morbidly obese. Categorical data are presented as frequency of occurrence and were analyzed by chi-square. Continuous data are presented as means +- SD and were analyzed by t-tests. All tests were two-tailed with an alpha set at 0.05. Multivariable logistic re- gression was performed to control for confounding. The primary out- come parameter was to evaluate the association of obesity/weight on risk of UA contamination.

Results

The primary author identified 355 eligible patients and five refused to participate during the enrollment window. Descriptive statistics and summary of bivariate analyses appear in Table 1. The final study group was 350 patients had a mean age 53.7 +- 19.1 years and mean BMI 31.5 +- 9.8. Patients were predominantly Hispanic and from lower socioeconomic and education status. Overall, 60% of patients provided a specimen with evidence of contamination. The contamination rates

Table 1

Descriptive statistics and bivariate analyses (N = 350).

and time of day within a predetermined enrollment window of six- months from the study initiation till cessation of data collection (May- November 2017). To increase the likelihood of identifying consecutive, eligible subjects during study hours, he systematically utilized the ED

electronic Tracking system scrolling down repeatedly from Bed 1 nu- merically to repeatedly review all patients who had been placed in the

Contamination No

Yes

39.7

department for evaluation. UAs were ordered at the discretion of the

BMI

31.5 (9.8)

29.7 (8.8)

32.7 (10.2)

0.005

treating emergency physician based on clinical indication alone. As

Age

53.7 (19.1)

49.7 (18.1)

51.2 (16.6)

0.406

All sample (N = 350)

%/mean (std.)

No contamination (N = 139)

Contamination (N = 211)

p value

60.3

per ED routine, patients were instructed to properly provide the UA through two means. First, they received verbal instructions by a nurse.

weight status NL and underweight

26.0 31.6 22.3

All of our nurses undergo an in-service as part of their introduction to our department for proper patient counseling regarding Urine collection techniques. Second, patients also were directed as part of nursing rou- tine to read UA sample provision instructions posted in English and Spanish on the inside of the bathroom door for ambulatory patients and the walls of each room for those who provided a sample there. Once an eligible patient was identified, the author confirmed with ei- ther the patient or the patient’s nurse that the sample sent to the lab

was in fact a clean catch and not a catheterized sample.

Overweight 22.0 26.6 19.0

Obese 35.4 29.5 39.3

Morbidly obese 16.6 12.2 19.4

Education 0.809

>=High school 65.4 66.2 65.4

bHigh school 34.5 33.8 35.0

Income 0.782

>=$20K 46.6 47.5 46.0

b$20K 53.4 52.5 54.0

Race and ethnicity 0.798

Patients willing to provide written consent were enrolled after the

Non-Hispanic white

28.0 28.8 27.5

ordering of the UA in an effort to reduce influence of the study on phy-

sician ordering practices, nursing verbal instruction techniques, and pa- tient cleansing/collection efforts. They subsequently filled out a structured data form to provide basic demographic/historical informa- tion. The primary investigator also abstracted the height and weight of the patient from the medical record for the purpose of BMI calculation. By study protocol, urine cultures were performed on all patients unless all of the following negative conditions as defined by our laboratory had been met on microscopy/dipstick: negative Leukocyte esterase, negative nitrites, <=5 white blood cells , and <=5 Bacteria. In cases where all

(4) criteria were met, the urine was deemed neither infected nor con- taminated, and the laboratory did not perform urine culture. Otherwise,

the lab either performed a reflexive culture by routine or, by pre-study

CVA

No

88.0

89.2

87.2

arrangement, a culture was automatically performed for the purpose of

Yes

12.0

10.8

12.8

Hispanic 62.3 60.4 63.5

Others 9.7 10.8 9.00

0.014

Hypertension

No

57.4

65.5

52.1

Yes

Diabetes

42.6

34.5

47.9

No

66.6

66.9

66.4

Yes

33.4

33.1

33.6

CAD

No

92.9

94.2

91.9

Yes

Hyperlipidemia

7.1

5.8

8.1

No

67.7

69.1

66.8

Yes

32.3

30.9

33.2

0.914

0.413

0.661

0.572

the investigation. UA interpretation was defined by our laboratory as:

Cancer

0.915

No

83.7

83.4

83.9

Yes

16.3

16.6

16.1

1) Neither contamination nor infection (No Growth); 2) Contamination only (Mixed Skin Flora); 3) Infection only (N100,000 cfu/ml); and

510 M. Strawn et al. / American Journal of Emergency Medicine 38 (2020) 508511

were: 52% normal/underweight, 52% overweight, 67% obese, and 71% morbidly obese respectively. Patients with contaminated specimens had significantly higher mean BMI than patients with non- contaminated specimens (32.7 +- 10.2 vs. 29.7 +- 8.8, p b 0.01). Table 2 illustrates patient characteristics by weight subgroup.

Within our multivariate logistic regression model, with the excep- tion of hypertension (OR 1.85; 1.1-3.2, p = 0.02), all other patient de- mographic and historical features were not associated with UA contamination. We observed that obesity (OR = 1. 90;1.1-3.4, p = 0.03) and morbid obesity (OR = 2.42;1.2-5.1, p = 0.02) were signifi- cantly associated with UA contamination. However, there was no such association for those patients who were classified as overweight (OR

= 1. 05;0.56-2.0, p = 0.88).

Discussion

Although it is difficult to quantify the number of UAs physicians order in EDs on a national basis, local data provides some context of the potential scope of US-based ordering. At our county hospital ED, in 2014, there were 45,489 patient visits, and in 32% of those visits, a phy- sician ordered a UA. Our county hospital is part of a larger six-hospital system serving 12-counties, and our laboratory directors reported to us that there are UAs ordered on 34% of all patients who visit these EDs (N120,000 annual ED visits).

When one considers the frequency for which this test is ordered, we believe that it is of significant Clinical concern that our study question regarding the risk of UA contamination for overweight subjects has not been adequately addressed in the past. We observed that patients with contaminated specimens had higher mean BMI. Further, within our multivariate logistic regression model, both obesity and morbid obesity respectively were associated with contamination.

The overall 60% UA contamination rate in our study group is rela- tively high but within the range suggested by prior investigations [7,10,11]. Valenstein and Meier evaluated lab reports from 906 institu- tions each providing the results of 250 consecutive, outpatient urine culture specimens [10]. They found that 18.1% of UAs collected from outpatients were contaminated, and specimens from women were

Table 2

Characteristics by weight status b%/mean (std.)N.

over twice as likely as men to have contamination. Frazee et al. reported their findings from a crossover study of 40 female menstrual-age female ED staff without UTI symptoms who each provided specimens both by direct (non-clean) and midstream clean catch (ideal). They found that samples in these asymptomatic patients were frequently suggestive of infection and were contaminated in N60% of samples from both collec- tion techniques [11]. Thus, while our results suggest a need for physi- cians to focus on overweight patients, our overall findings are consistent with the general concern that UA specimens from women may not provide reliable diagnostic information.

Investigators have provided evidence of anatomic influences on con- tamination that suggest possible solutions to the contamination prob- lem. The reports of Baerheim et al. and Shalom et al. respectively found that extra tissues blocking flow of urine from the urethra when attempting a sample collection increased contamination rates [13,14]. Researchers in the former study evaluated 864 urine samples collected from 110 women and found that simply spreading the labia before col- lection led to a 50% reduction in contamination rate compared to no in- tervention [13]. Shalom et al. investigated the contamination rates for urine samples provided by women with pelvic organ prolapse. They found that samples obtained by catheterization had lower contamina- tion rates than those that were obtained by the midstream clean catch method [14]. Pelvic organ prolapse, like obesity, provides extra tissue in close anatomic proximity to the urethra. Given the similarly altered anatomy, it is sensible to believe that catheterization of obese women may improve the accuracy of urine specimens.

Our study question has been investigated in a retrospective fashion previously. Zwank and Bourdon analyzed a convenience sample of 7134 clean-catch urine cultures collected over a 7-month period for fe- male, ED patients that were provided a clinical diagnosis of UTI. Similar to our results, they found that over 50% of the specimens were contam- inated and that contamination rates were significantly higher in obese women. Further, they observed that typical markers for UTI on the UA dipstick/microscopy were higher for women with BMI N35. Compara- tively, we note that our study reduces bias inherent to their retrospec- tive method for contamination rate analysis as, prospectively, we included all UAs ordered for ED patients by emergency physicians, not simply those obtained from patients with an ED-based, clinical diagno- sis of UTI. Both reports highlight the significant limitations of urine sam- ples obtained in the ED to accurately diagnose UTI for obese women,

Normal & underweight

Overweight Obese Morbidly

obese

p value

and we suggest that future studies should evaluate potential methods to lower the rate for these patients.

Age 53.7 (19.1) 52.0 (18.0) 50.1

(16.24)

Education

0.414

>=High school

70.3

66.2

65.3

56.9

bHigh school

29.7

33.8

34.7

43.1

Income

0.437

>=$20K

42.9

50.6

50

39.7

b$20K

Race and ethnicity

57.1

49.4

50

60.3

0.220

Non-Hispanic

36.3

31.2

25

17.2

white

Hispanic

52.7

61.0

65.3

72.4

Others

11.0

7.8

9.7

10.3

Hypertension

0.028

No

68.1

62.3

50

50

Yes

31.9

37.7

50

50

Diabetes

b0.001

No

85.7

67.5

61.3

46.6

Yes

14.3

32.5

38.7

53.4

CAD

0.963

No

93.4

93.5

92.7

91.4

Yes

6.6

6.5

7.3

8.6

Hyperlipidemia

0.819

No

87.9

89.6

88.7

84.5

Yes

12.1

10.4

11.3

15.5

CVA

0.298

No

83.5

79.2

83.1

91.4

Yes

16.5

20.8

16.9

8.6

45.3

(13.8)

0.045

Limitations

We note that our study has several limitations that warrant discus- sion. First of all, due to the processes of our hospital laboratory (e.g. cul- tures not obtained for “clean” UAs by the criteria described in our methods), we may have underestimated the overall contamination rate of the study group. However, we are unclear as to how this limita- tion might have influenced differences observed for lower and higher weight patient subgroups.

Additionally, though we believe our laboratory characterization of contamination and infection are representative of typical definitions, clinicians utilizing alternative definitions at other centers might observe somewhat different outcomes. In contrast to the Zwank and Bourdon analysis, we were unable to provide an assessment of the predictive value for contamination and/or infection of any single UA component value as our laboratory did not perform microscopy and a culture on every patient. For example, the utility of squamous epithelial cells as a predictor of contamination has been a source of conflicting reports in the literature, but our data does not provide an opportunity to add to this debate [11,13-24].

Finally, we note the additional concern for external validity in view of the somewhat unique characteristics of our patient population. The majority of patients in our study group were Hispanic and from lower

M. Strawn et al. / American Journal of Emergency Medicine 38 (2020) 508511 511

socioeconomic and educational levels, and there is an unusually high prevalence of diabetes in our region. We did attempt to isolate the rela- tionship of these variables to the risk of contamination within our logis- tic regression model.

Conclusion

Within our prospective cohort of ED patients, obesity was signifi- cantly associated with contamination of clean catch mid-stream sam- ples. Investigators should evaluate methods to specifically reduce contamination risk in this patient subgroup in future studies.

Financial disclosures

None.

References

  1. Simerville Jeff A, Maxted William C, Pahira John J. Urinalysis: a comprehensive re- view. Am Fam Physician 2005;71.6:1153-62 [Web].
  2. Stapczynski JStephan, Tintinalli Judith E. Tintinallis emergency medicine: a compre- hensive study guide. . 8th ed.McGraw-Hill Education; 2016.
  3. Ogden Cynthia L, et al. Prevalence of obesity in the United States, 2009-2010; 2012.
  4. Falagas Matthew E, Kompoti Maria. Obesity and infection. Lancet Infect Dis 2006;6 (7):438-46.
  5. Moller Lars Alling, Lose Gunnar, Jorgensen Torben. Risk factors for lower urinary tract symptoms in women 40 to 60 years of age. Obstet Gynecol 2000;96(3): 446-51.
  6. Mommsen S, Foldspang A. Body mass index and adult female urinary incontinence. World J Urol 1994;12(6):319-22.
  7. Zwank MD, Bourdon R. Increased body mass index among women correlates with increased rates of urine contamination. Ann Emerg Med 2013;62(4). https://doi. org/10.1016/j.annemergmed.2013.07.057.
  8. Vessey MP, et al. Urinary tract infection in relation to diaphragm use and obesity. Int

    J Epidemiol 1987;16(3):441-4.

    Semins Michelle J, et al. The impact of obesity on urinary tract Infection risk. Urology

    2012;79(2):266-9.

    Valenstein Paul, Meier Frederick. Urine culture contamination. Arch Pathol Lab Med 1998;122:123-9.

  9. Frazee Bradley W, et al. Abnormal urinalysis results are common, regardless of spec- imen collection technique, in women without urinary tract infections. J Emerg Med June 2015;48(6):706-71.
  10. Baerheim A, Digranes A, Hunskaar S. Evaluation of urine sampling technique: bacte- rial contamination of samples from women students. Br J Gen Pract 1992;42(359): 241-3.
  11. Shalom Dara F, et al. The effect of collection method on urinalysis results in women with pelvic organ prolapse. Urol Nurs 2014;34(3):128.
  12. Lifshitz Edward, Kramer Liane. Outpatient urine culture: does collection technique matter? Arch Intern Med 2000;160(16):2537-40.
  13. Walter Frank G, Knopp Robert K. Urine sampling in ambulatory women: midstream clean-catch versus catheterization. Ann Emerg Med 1989;18(2):166-72.
  14. Prandoni Dana, et al. Assessment of urine collection technique for microbial culture. Am J Infect Control 1996;24(3):219-21.
  15. Morris RW, Watts MR, Reeves DS. Perineal cleansing before midstream urine, a nec- essary ritual? Lancet 1979;314(8134):158-9.
  16. Leisure Marketa Kyncl, Dudley Sharon M, Donowitz Leigh G. Does a clean-catch urine sample reduce bacterial contamination? N Engl J Med 1993;328(4):289-90.
  17. Immergut Mark A, et al. The myth of the clean catch urine specimen. Urology 1981; 17(4):339-40.
  18. Fabio Manoni, et al. Mid-stream vs. first-voided urine collection by using automated analyzers for particle examination in healthy subjects: an Italian multicenter study. Clin Chem Lab Med 2012;50(4):679-84.
  19. Rashmi Shrestha, et al. Effect of urogenital cleaning with paper soap on bacterial contamination rate while collecting midstream urine specimens. J Lab Physicians 2013;5(1):17.
  20. Mie Morimoto, et al. Effects of midstream collection and the menstrual cycle on urine particles and dipstick urinalysis among healthy females. Clin Chem 2003;49 (1):188-90.
  21. Pfau A. Bacteriuria–sampling methods and significance. Infection 1994;22:S42-3.