- Split View
-
Views
-
Cite
Cite
Molly J Horstman, Andrew Spiegelman, Aanand D Naik, Barbara W Trautner, National Patterns of Urine Testing During Inpatient Admission, Clinical Infectious Diseases, Volume 65, Issue 7, 1 October 2017, Pages 1199–1205, https://doi.org/10.1093/cid/cix424
- Share Icon Share
Abstract
Overuse of urine testing is a driver of inappropriate antimicrobial use. Limiting wasteful testing is important for patient safety. We examined the national prevalence and patterns of urine testing during adult inpatient admission in the United States.
We performed a retrospective cohort study using a national dataset of inpatient admissions from 263 hospitals in the United States from 2009 to 2014. We included all adult inpatient admissions, excluding those related to pregnancy, urology procedures, and with lengths of stay >30 days. A facility-level fixed-effects quasi-Poisson regression model was used to examine the incidence of urinalysis and urine culture testing for select diagnoses and patient factors.
The cohort included 4473655 admissions. Charges for urinalysis were present for 2086697 (47%) admissions, with 584438 (13%) including >1 urinalysis. Charges for urine culture were present for 1197242 (27%) admissions, with 246211 (6%) having >1 culture. Urine culture testing varied by principal diagnosis. Heart failure and acute myocardial infarction had 29% and 35% fewer cultures sent on the first day of admission compared to all other admissions (P < .001). Female sex and receipt of antibiotics during the hospital admission consistently predicted increased culture testing, regardless of principal diagnosis or age.
Urine testing was common and frequently repeated during inpatient admission, suggesting large-scale overuse. The variation in testing by diagnosis suggests that clinical presentation modifies test use. The sex bias in urine testing is not clinically supported and must be addressed in interventions aimed at reducing excess urine testing.
Unnecessary testing is widespread in healthcare. In the United States, the Choosing Wisely Campaign was launched in 2012 to encourage providers to reduce commonly performed tests that are not supported by evidence and contribute to waste [1]. Reducing unnecessary screening for asymptomatic bacteriuria is endorsed by 6 professional societies in the United States and is recommended by many Choosing Wisely campaigns internationally [2, 3]. The presence of a positive urinalysis or urine culture, even in a patient without urinary symptoms, is a powerful stimulus for antibiotic treatment [4–8]. Unnecessary treatment exposes patients to significant morbidity, including adverse drug events, Clostridium difficile colitis, and infections with resistant organisms [9–13]. Judicious use of urine testing (urinalysis and urine culture) is important to reduce waste, protect patient safety, and conserve effective antibiotics [14].
Despite the interest in reducing waste associated with urine testing, information regarding the number of hospitalized patients that receive urinalyses and urine cultures is limited. The prevalence of urinalysis testing during adult inpatient admissions is largely unknown; a few studies from single institutions with academic affiliations have reported that 43%–62% of patients admitted through the emergency department receive a urinalysis [5, 15]. The validity of these results for a larger, national sample and their comparability to urine culture testing frequency are unknown. Although prior studies have identified patient-level factors that contribute to clinicians’ decisions to treat asymptomatic bacteriuria, it is unclear which factors are associated with the initial decision to order a urinalysis or urine culture [7, 8]. Based on the available literature, we hypothesized that urine testing would vary by patient age, sex, and presenting condition and that this variation might not be clinically appropriate [5, 7, 15].
The aims of this study were to describe the frequency and patterns of urine testing in the United States using a national dataset to describe the prevalence and timing of urinalysis and urine culture testing during adult hospital admissions in the United States. We hypothesized that urinalysis and urine culture testing would vary by admitting diagnosis, based on a model of clinical appropriateness of urine culture testing and need for antibiotics. Finally, we examined whether patient-level factors (age, sex, and antibiotic use) were associated with orders for urinalysis and urine culture on admission.
Methods
This study used an existing national dataset from The Advisory Board Company, a healthcare best practice, research, and technology firm. The database is based on hospital billing and administrative data across inpatient and outpatient settings in the United States, and contains detailed information around patient encounters, such as diagnoses, procedures, charges and costs, physician roles and participation, and administrative outcomes. Limited patient demographics and facility characteristics are collected. This dataset is deidentified in accordance with the Health Insurance Portability and Accountability Act Privacy Rule and may be used for research studies [16, 17]. Data include 22 million inpatient encounters and 52 million ambulatory encounters. The database offers a cross-payer perspective, including encounters covered by commercial insurance, government insurance, self-pay, and charity care. The dataset is generally representative of hospitals across the United States, other than a shift away from small (<50 bed) facilities and away from facilities in the western United States (differences are ≤5% from the national average from the American Hospital Association).
Study Sample
We identified all inpatient admissions for patients aged ≥18 years with a hospital length of stay of ≤30 days. The Infectious Diseases Society of America recommends withholding urinalysis and urine culture in patients who do not have symptoms specific to the urinary tract, unless the patient is a pregnant woman or undergoing urologic procedures where mucosal bleeding is anticipated [18]. Guidelines developed for specific groups, such as patients with indwelling urinary catheters, recommend urine culture testing only when signs or symptoms of a urinary tract infection are present [19]. To exclude admissions where tests may have been sent to screen for asymptomatic bacteriuria, we excluded all admissions for pregnant patients and hospital stays with a urological procedure code or a hospital diagnosis of hematuria using International Classification of Diseases, Ninth Revision (ICD-9) and Current Procedural Terminology codes (Supplementary Materials). Admissions where the admission type was unknown were excluded. We included all emergency, urgent, elective, and trauma center admissions that met our eligibility criteria [20].
Measures and Variables
To measure the number of urinalyses and urine cultures sent during inpatient admissions, we identified charges submitted for urine testing. We recorded the day the charge was submitted (hospital day 1, day 2, etc). Codes for urinalysis and urine culture charges included in the analysis were identified by manually reviewing an aggregate list of charge names for urine tests. Urinalysis data were aggregated to determine the prevalence of urinalysis ordering, the total number of urinalyses ordered during each admission, and the distribution of urinalyses ordered by the day of admission. Urine culture data were aggregated similarly.
Quasi-Poisson Regression Model for Urinalyses and Urine Cultures
A facility-level fixed-effects quasi-Poisson regression model was used to examine the impact of covariates on urine testing during admission. The facility-level fixed-effects quasi-Poisson model was chosen to control for the facility effects on the outcome and to account for any overdispersion in the data [21]. A multilevel Poisson model with random intercepts and a negative binomial regression were performed; results were substantively similar to the fixed-effects quasi-Poisson model. Unique patients could have >1 admission in the analysis. Separate models were developed for the following outcomes: the number of urinalyses sent on the first day of admission, the number of urinalyses sent during the entire hospital stay, the number of urine cultures sent on the first day of admission, and the number of urine cultures sent during the entire hospital stay. Each quasi-Poisson model was developed using a random sample of approximately 30000 hospitalizations from the cohort of hospital stays meeting our inclusion criteria. Any predictor that failed to reduce the model deviance by >1 in a given 30000 sample was deemed inconsequential and excluded from the model. This was done repeatedly until ten 30000 samples in a row produced the same set of coefficients. Dispersion was close to 1, ranging from 0.8 to 0.9 for all models. Incidence rate ratios were calculated from the exponentiation of the coefficients from the quasi-Poisson model. Models were developed using R statistical software (version 3.2.2).
Analyses by Principal Diagnoses
To examine the differences in testing by principal diagnosis, we first classified diagnoses by the clinical likelihood that a urine test would be part of a routine diagnostic evaluation and antibiotics part of a routine treatment plan. The diagnoses were discussed and agreed upon by the clinician investigators (a hospitalist, infectious disease specialist, and geriatrician who provide inpatient care) prior to starting the data analysis. Three broad categories were identified: (1) routine culture testing not indicated and antibiotics not indicated for the principal diagnosis (acute myocardial infarction, congestive heart failure); (2) routine culture testing not indicated, but antibiotics indicated for the principal diagnosis (cellulitis, pneumonia); and (3) culture testing likely indicated for evaluation of the principal diagnosis (urinary tract infection, diabetic ketoacidosis). The specified principal diagnoses were identified by ICD-9 codes from definitions used by the Centers for Medicare and Medicaid Services or published literature (Supplementary Materials) and included as covariates in the quasi-Poisson regression model.
Analyses by Patient-Level Characteristics
Patient-level characteristics included as covariates were patient age, sex, ethnicity, hospital length of stay, number of discharge diagnoses from admission, Elixhauser comorbidity index, All Patient Refined Diagnosis Related Groupings (APR-DRG) severity level, APR-DRG mortality level, presence of a hospital readmission within 30 days, and receipt of antibiotics during the hospital stay [22, 23]. To explore the differences in urine culture testing for different combinations of patient-level factors, we used the regression model to simulate the predicted number of urine cultures per 100 patients by age for each of the prespecified principal diagnoses. We simulated the model under 4 different conditions based on sex and antibiotic use: women with antibiotic use, women without antibiotic use, men with antibiotic use, and men without antibiotic use. For each of the principal diagnoses, we first modeled the number of urine cultures per 100 patients using the estimates from the quasi-Poisson regression to create a best-fit line for each of the 4 conditions. We then simulated each of the 4 conditions 100 times using almost-best-fit estimates from the normal distribution for each variable. This simulation method provides a visual display of uncertainty around the best-fit line. All other variables in the model were held constant.
Ethics Statement
This study was reviewed by the Institutional Review Board at Baylor College of Medicine and the Research and Development committee at the Michael E. DeBakey Veterans Affairs Medical Center, which determined that this study did not constitute human subjects research.
RESULTS
Between 1 January 2009 and 31 December 2014, the database contained 4473655 adult admissions from 263 hospitals that met our inclusion criteria (Supplementary Materials). Patients’ mean age was 62 (±18) years and 2055407 (46%) were men. Median length of stay was 3 days (interquartile range, 2–6 days). The majority of admissions included in the cohort were at hospitals with a teaching affiliation (n = 2662426 [60%]), and half of the admissions were at hospitals with ≥400 beds (n = 2286796 [51%]) (Table 1). The majority of admissions were emergency or urgent (n = 3515394 [79%]).
Characteristic . | No. (%) . |
---|---|
Patient characteristics | |
Patient sex | |
Female | 2418258 (54) |
Male | 2055407 (46) |
APR-DRG severity level | |
1—Minor | 1066407 (24) |
2—Moderate | 1807531 (40) |
3—Major | 1264553 (28) |
4—Extreme | 293330 (7) |
Not defined | 41844 (1) |
APR-DRG mortality level | |
1—Minor | 1946977 (44) |
2—Moderate | 1342139 (30) |
3—Major | 895428 (20) |
4—Extreme | 247492 (6) |
Not defined | 41629 (1) |
Claim inpatient admission type | |
Emergency | 2970017 (66) |
Elective | 928887 (21) |
Urgent | 545377 (12) |
Trauma center | 29379 (1) |
Hospital characteristics | |
AHA teaching affiliationa | |
Major | 1112633 (25) |
Minor | 1549793 (35) |
Non-teaching | 1372266 (31) |
AHA hospital locationa | |
Urban | 3770187 (84) |
Rural | 264505 (6) |
AHA hospital sizea | |
≥500 beds | 1654161 (37) |
400–499 beds | 632635 (14) |
300–399 beds | 522860 (12) |
200–299 beds | 484716 (11) |
100–199 beds | 576842 (13) |
<100 beds | 163478 (4) |
Characteristic . | No. (%) . |
---|---|
Patient characteristics | |
Patient sex | |
Female | 2418258 (54) |
Male | 2055407 (46) |
APR-DRG severity level | |
1—Minor | 1066407 (24) |
2—Moderate | 1807531 (40) |
3—Major | 1264553 (28) |
4—Extreme | 293330 (7) |
Not defined | 41844 (1) |
APR-DRG mortality level | |
1—Minor | 1946977 (44) |
2—Moderate | 1342139 (30) |
3—Major | 895428 (20) |
4—Extreme | 247492 (6) |
Not defined | 41629 (1) |
Claim inpatient admission type | |
Emergency | 2970017 (66) |
Elective | 928887 (21) |
Urgent | 545377 (12) |
Trauma center | 29379 (1) |
Hospital characteristics | |
AHA teaching affiliationa | |
Major | 1112633 (25) |
Minor | 1549793 (35) |
Non-teaching | 1372266 (31) |
AHA hospital locationa | |
Urban | 3770187 (84) |
Rural | 264505 (6) |
AHA hospital sizea | |
≥500 beds | 1654161 (37) |
400–499 beds | 632635 (14) |
300–399 beds | 522860 (12) |
200–299 beds | 484716 (11) |
100–199 beds | 576842 (13) |
<100 beds | 163478 (4) |
Abbreviations: AHA, American Hospital Association; APR-DRG, All Patient Refined Diagnosis-Related Group.
aAHA hospital characteristics were not available for 438973 hospital admissions (10% of the cohort).
Characteristic . | No. (%) . |
---|---|
Patient characteristics | |
Patient sex | |
Female | 2418258 (54) |
Male | 2055407 (46) |
APR-DRG severity level | |
1—Minor | 1066407 (24) |
2—Moderate | 1807531 (40) |
3—Major | 1264553 (28) |
4—Extreme | 293330 (7) |
Not defined | 41844 (1) |
APR-DRG mortality level | |
1—Minor | 1946977 (44) |
2—Moderate | 1342139 (30) |
3—Major | 895428 (20) |
4—Extreme | 247492 (6) |
Not defined | 41629 (1) |
Claim inpatient admission type | |
Emergency | 2970017 (66) |
Elective | 928887 (21) |
Urgent | 545377 (12) |
Trauma center | 29379 (1) |
Hospital characteristics | |
AHA teaching affiliationa | |
Major | 1112633 (25) |
Minor | 1549793 (35) |
Non-teaching | 1372266 (31) |
AHA hospital locationa | |
Urban | 3770187 (84) |
Rural | 264505 (6) |
AHA hospital sizea | |
≥500 beds | 1654161 (37) |
400–499 beds | 632635 (14) |
300–399 beds | 522860 (12) |
200–299 beds | 484716 (11) |
100–199 beds | 576842 (13) |
<100 beds | 163478 (4) |
Characteristic . | No. (%) . |
---|---|
Patient characteristics | |
Patient sex | |
Female | 2418258 (54) |
Male | 2055407 (46) |
APR-DRG severity level | |
1—Minor | 1066407 (24) |
2—Moderate | 1807531 (40) |
3—Major | 1264553 (28) |
4—Extreme | 293330 (7) |
Not defined | 41844 (1) |
APR-DRG mortality level | |
1—Minor | 1946977 (44) |
2—Moderate | 1342139 (30) |
3—Major | 895428 (20) |
4—Extreme | 247492 (6) |
Not defined | 41629 (1) |
Claim inpatient admission type | |
Emergency | 2970017 (66) |
Elective | 928887 (21) |
Urgent | 545377 (12) |
Trauma center | 29379 (1) |
Hospital characteristics | |
AHA teaching affiliationa | |
Major | 1112633 (25) |
Minor | 1549793 (35) |
Non-teaching | 1372266 (31) |
AHA hospital locationa | |
Urban | 3770187 (84) |
Rural | 264505 (6) |
AHA hospital sizea | |
≥500 beds | 1654161 (37) |
400–499 beds | 632635 (14) |
300–399 beds | 522860 (12) |
200–299 beds | 484716 (11) |
100–199 beds | 576842 (13) |
<100 beds | 163478 (4) |
Abbreviations: AHA, American Hospital Association; APR-DRG, All Patient Refined Diagnosis-Related Group.
aAHA hospital characteristics were not available for 438973 hospital admissions (10% of the cohort).
Prevalence and Frequency of Urinalyses and Urine Cultures
At least 1 urinalysis was sent in 2086697 (47%) hospital stays, with 584438 (13%) including >1 urinalysis (range, 2–42; total number of urinalyses, 2872019). Urine cultures were sent in 1197242 (27%) admissions, with 246211 (6%) including >1 urine culture (range, 2–19; total number of urine cultures in sample, 1513785). Among admissions that included a urine culture, 12.3% had >1 urine culture sent every 2 days on average. The majority of urine tests were sent within the first 2 days of the hospital stay (n = 2004578 urinalyses; n = 1010939 urine cultures) (Supplementary Materials). Among admissions at teaching hospitals, 45.6% had at least 1 urinalysis (n = 1215102) and 25.5% had at least 1 urine culture (n = 678612). Among admissions at nonteaching hospitals, 49.0% had at least 1 urinalysis (n = 672446) and 29.2% had at least 1 urine culture (n = 401245). For admissions at hospitals with ≥400 beds, 46.9% had at least 1 urinalysis (n = 1073041) and 25.3% had at least 1 urine culture (n = 579972).
Urinalyses and Urine Cultures by Principal Diagnosis
Table 2 provides the number of admissions with a urinalysis or urine culture for each of the 6 prespecified principal diagnosis groups. The incidence rate ratio for urinalyses and urine cultures sent on the first day of the admission varied by principal diagnosis (Table 3 and Supplementary Materials). Diagnoses where urine culture and antibiotic use would not be routinely indicated were associated with fewer urine cultures sent on the first day of admission. For acute myocardial infarction and heart failure admissions, there were 35% and 29% fewer urine cultures sent on the first day of the admission compared to all other admissions in the sample (P < .001 for both). For diagnoses where urine culture could be indicated as part of an inpatient workup (diabetic ketoacidosis and urinary tract infection), there were significantly more urine cultures sent on the first day of admission (P < .001 for both). Results were mixed for our 2 representative diagnoses where antibiotics would be indicated, but urine culture would not be routinely needed. For cellulitis admissions, fewer urine cultures were sent on the first day of admission, while admissions for pneumonia had no statistically significant difference in urine culture testing compared to all other admissions in the sample. Similar findings were seen in the association between principal diagnosis and the total number of urine cultures sent during the entire hospital stay, with the exception of acute myocardial infarction, in which there was no longer a statistically significant decrease in the number of urine cultures sent (Table 4).
Principal Diagnosis . | All Patients . | Heart Failure . | Acute Myocardial Infarction . | Cellulitis . | Pneumonia . | Diabetic Ketoacidosis . | Urinary Tract Infection . |
---|---|---|---|---|---|---|---|
(N = 4473655) . | (n = 190383) . | (n = 107874) . | (n = 98694) . | (n = 137263) . | (n = 27463) . | (n = 71692) . | |
Any urinalysis | 2086697 (46.6) | 88379 (46.4) | 34871 (32.3) | 32101 (32.5) | 72641 (52.9) | 22383 (81.5) | 65032 (90.7) |
Any urine culture | 1197242 (26.8) | 45099 (23.7) | 19681 (18.2) | 18329 (18.6) | 45782 (33.4) | 10493 (38.2) | 62953 (87.8) |
Urinalysis and urine culture | 1089968 (24.4) | 41563 (21.8) | 17884 (16.6) | 16299 (16.5) | 41402 (30.2) | 9735 (35.4) | 60590 (84.5) |
Women | (n = 2418258) | (n = 94429) | (n = 41248) | (n = 47765) | (n = 73275) | (n = 13592) | (n = 52731) |
Women with urinalysis | 1200072 (26.8) | 47439 (50.2) | 15959 (38.7) | 17431 (36.5) | 40082 (54.7) | 11369 (83.6) | 47890 (90.8) |
Women with urine culture | 738590 (16.5) | 27489 (29.1) | 10105 (24.5) | 10707 (22.4) | 26335 (35.9) | 6166 (45.4) | 46168 (87.6) |
Men | (n = 2055407) | (n = 95954) | (n = 66626) | (n = 50929) | (n = 63988) | (n = 13871) | (n = 18961) |
Men with urinalysis | 886625 (19.8) | 40940 (42.7) | 18912 (28.4) | 14670 (28.8) | 32559 (40.9) | 11014 (79.4) | 17142 (90.4) |
Men with urine culture | 458652 (10.3) | 17610 (18.4) | 9576 (14.4) | 7622 (15.0) | 19447 (30.4) | 4327 (31.2) | 16785 (88.5) |
Principal Diagnosis . | All Patients . | Heart Failure . | Acute Myocardial Infarction . | Cellulitis . | Pneumonia . | Diabetic Ketoacidosis . | Urinary Tract Infection . |
---|---|---|---|---|---|---|---|
(N = 4473655) . | (n = 190383) . | (n = 107874) . | (n = 98694) . | (n = 137263) . | (n = 27463) . | (n = 71692) . | |
Any urinalysis | 2086697 (46.6) | 88379 (46.4) | 34871 (32.3) | 32101 (32.5) | 72641 (52.9) | 22383 (81.5) | 65032 (90.7) |
Any urine culture | 1197242 (26.8) | 45099 (23.7) | 19681 (18.2) | 18329 (18.6) | 45782 (33.4) | 10493 (38.2) | 62953 (87.8) |
Urinalysis and urine culture | 1089968 (24.4) | 41563 (21.8) | 17884 (16.6) | 16299 (16.5) | 41402 (30.2) | 9735 (35.4) | 60590 (84.5) |
Women | (n = 2418258) | (n = 94429) | (n = 41248) | (n = 47765) | (n = 73275) | (n = 13592) | (n = 52731) |
Women with urinalysis | 1200072 (26.8) | 47439 (50.2) | 15959 (38.7) | 17431 (36.5) | 40082 (54.7) | 11369 (83.6) | 47890 (90.8) |
Women with urine culture | 738590 (16.5) | 27489 (29.1) | 10105 (24.5) | 10707 (22.4) | 26335 (35.9) | 6166 (45.4) | 46168 (87.6) |
Men | (n = 2055407) | (n = 95954) | (n = 66626) | (n = 50929) | (n = 63988) | (n = 13871) | (n = 18961) |
Men with urinalysis | 886625 (19.8) | 40940 (42.7) | 18912 (28.4) | 14670 (28.8) | 32559 (40.9) | 11014 (79.4) | 17142 (90.4) |
Men with urine culture | 458652 (10.3) | 17610 (18.4) | 9576 (14.4) | 7622 (15.0) | 19447 (30.4) | 4327 (31.2) | 16785 (88.5) |
Data are presented as No. (%).
Principal Diagnosis . | All Patients . | Heart Failure . | Acute Myocardial Infarction . | Cellulitis . | Pneumonia . | Diabetic Ketoacidosis . | Urinary Tract Infection . |
---|---|---|---|---|---|---|---|
(N = 4473655) . | (n = 190383) . | (n = 107874) . | (n = 98694) . | (n = 137263) . | (n = 27463) . | (n = 71692) . | |
Any urinalysis | 2086697 (46.6) | 88379 (46.4) | 34871 (32.3) | 32101 (32.5) | 72641 (52.9) | 22383 (81.5) | 65032 (90.7) |
Any urine culture | 1197242 (26.8) | 45099 (23.7) | 19681 (18.2) | 18329 (18.6) | 45782 (33.4) | 10493 (38.2) | 62953 (87.8) |
Urinalysis and urine culture | 1089968 (24.4) | 41563 (21.8) | 17884 (16.6) | 16299 (16.5) | 41402 (30.2) | 9735 (35.4) | 60590 (84.5) |
Women | (n = 2418258) | (n = 94429) | (n = 41248) | (n = 47765) | (n = 73275) | (n = 13592) | (n = 52731) |
Women with urinalysis | 1200072 (26.8) | 47439 (50.2) | 15959 (38.7) | 17431 (36.5) | 40082 (54.7) | 11369 (83.6) | 47890 (90.8) |
Women with urine culture | 738590 (16.5) | 27489 (29.1) | 10105 (24.5) | 10707 (22.4) | 26335 (35.9) | 6166 (45.4) | 46168 (87.6) |
Men | (n = 2055407) | (n = 95954) | (n = 66626) | (n = 50929) | (n = 63988) | (n = 13871) | (n = 18961) |
Men with urinalysis | 886625 (19.8) | 40940 (42.7) | 18912 (28.4) | 14670 (28.8) | 32559 (40.9) | 11014 (79.4) | 17142 (90.4) |
Men with urine culture | 458652 (10.3) | 17610 (18.4) | 9576 (14.4) | 7622 (15.0) | 19447 (30.4) | 4327 (31.2) | 16785 (88.5) |
Principal Diagnosis . | All Patients . | Heart Failure . | Acute Myocardial Infarction . | Cellulitis . | Pneumonia . | Diabetic Ketoacidosis . | Urinary Tract Infection . |
---|---|---|---|---|---|---|---|
(N = 4473655) . | (n = 190383) . | (n = 107874) . | (n = 98694) . | (n = 137263) . | (n = 27463) . | (n = 71692) . | |
Any urinalysis | 2086697 (46.6) | 88379 (46.4) | 34871 (32.3) | 32101 (32.5) | 72641 (52.9) | 22383 (81.5) | 65032 (90.7) |
Any urine culture | 1197242 (26.8) | 45099 (23.7) | 19681 (18.2) | 18329 (18.6) | 45782 (33.4) | 10493 (38.2) | 62953 (87.8) |
Urinalysis and urine culture | 1089968 (24.4) | 41563 (21.8) | 17884 (16.6) | 16299 (16.5) | 41402 (30.2) | 9735 (35.4) | 60590 (84.5) |
Women | (n = 2418258) | (n = 94429) | (n = 41248) | (n = 47765) | (n = 73275) | (n = 13592) | (n = 52731) |
Women with urinalysis | 1200072 (26.8) | 47439 (50.2) | 15959 (38.7) | 17431 (36.5) | 40082 (54.7) | 11369 (83.6) | 47890 (90.8) |
Women with urine culture | 738590 (16.5) | 27489 (29.1) | 10105 (24.5) | 10707 (22.4) | 26335 (35.9) | 6166 (45.4) | 46168 (87.6) |
Men | (n = 2055407) | (n = 95954) | (n = 66626) | (n = 50929) | (n = 63988) | (n = 13871) | (n = 18961) |
Men with urinalysis | 886625 (19.8) | 40940 (42.7) | 18912 (28.4) | 14670 (28.8) | 32559 (40.9) | 11014 (79.4) | 17142 (90.4) |
Men with urine culture | 458652 (10.3) | 17610 (18.4) | 9576 (14.4) | 7622 (15.0) | 19447 (30.4) | 4327 (31.2) | 16785 (88.5) |
Data are presented as No. (%).
Predictors . | Regression Estimate (Standard Error) . | Incidence Rate Ratio . | P Value . |
---|---|---|---|
Principal diagnosis of acute myocardial infarction | –0.44 (0.12) | 0.65 | <.001 |
Principal diagnosis of heart failure | –0.34 (0.07) | 0.71 | <.001 |
Principal diagnosis of cellulitis | –0.46 (0.11) | 0.63 | <.001 |
Principal diagnosis of pneumonia | 0.37 (0.21) | 1.45 | .08 |
Principal diagnosis of diabetic ketoacidosis | 0.81 (0.13) | 2.26 | <.001 |
Principal diagnosis of urinary tract infection | 2.10 (0.17) | 8.13 | <.001 |
Age | –0.04 (0.004) | 0.97 | <.001 |
Female sex | 0.24 (0.03) | 1.27 | <.001 |
Antibiotics during hospital admission | 0.47 (0.09) | 1.60 | <.001 |
Length of staya | 0.32 (0.05) | 1.38 | <.001 |
Elixhauser comorbidity indexb | 0.15 (0.04) | 1.16 | <.001 |
Predictors . | Regression Estimate (Standard Error) . | Incidence Rate Ratio . | P Value . |
---|---|---|---|
Principal diagnosis of acute myocardial infarction | –0.44 (0.12) | 0.65 | <.001 |
Principal diagnosis of heart failure | –0.34 (0.07) | 0.71 | <.001 |
Principal diagnosis of cellulitis | –0.46 (0.11) | 0.63 | <.001 |
Principal diagnosis of pneumonia | 0.37 (0.21) | 1.45 | .08 |
Principal diagnosis of diabetic ketoacidosis | 0.81 (0.13) | 2.26 | <.001 |
Principal diagnosis of urinary tract infection | 2.10 (0.17) | 8.13 | <.001 |
Age | –0.04 (0.004) | 0.97 | <.001 |
Female sex | 0.24 (0.03) | 1.27 | <.001 |
Antibiotics during hospital admission | 0.47 (0.09) | 1.60 | <.001 |
Length of staya | 0.32 (0.05) | 1.38 | <.001 |
Elixhauser comorbidity indexb | 0.15 (0.04) | 1.16 | <.001 |
Results repeated with 10 different random samples of 30000 hospital admissions from the original dataset of admissions. Results shown were robust in repeated samples. An incidence rate ratio of 0.65 indicates that a patient with a principal diagnosis of acute myocardial infection was 35% less likely than the entire sample of patients to receive a urine culture on admission.
aLog transformation.
bSquare root transformation.
Predictors . | Regression Estimate (Standard Error) . | Incidence Rate Ratio . | P Value . |
---|---|---|---|
Principal diagnosis of acute myocardial infarction | –0.44 (0.12) | 0.65 | <.001 |
Principal diagnosis of heart failure | –0.34 (0.07) | 0.71 | <.001 |
Principal diagnosis of cellulitis | –0.46 (0.11) | 0.63 | <.001 |
Principal diagnosis of pneumonia | 0.37 (0.21) | 1.45 | .08 |
Principal diagnosis of diabetic ketoacidosis | 0.81 (0.13) | 2.26 | <.001 |
Principal diagnosis of urinary tract infection | 2.10 (0.17) | 8.13 | <.001 |
Age | –0.04 (0.004) | 0.97 | <.001 |
Female sex | 0.24 (0.03) | 1.27 | <.001 |
Antibiotics during hospital admission | 0.47 (0.09) | 1.60 | <.001 |
Length of staya | 0.32 (0.05) | 1.38 | <.001 |
Elixhauser comorbidity indexb | 0.15 (0.04) | 1.16 | <.001 |
Predictors . | Regression Estimate (Standard Error) . | Incidence Rate Ratio . | P Value . |
---|---|---|---|
Principal diagnosis of acute myocardial infarction | –0.44 (0.12) | 0.65 | <.001 |
Principal diagnosis of heart failure | –0.34 (0.07) | 0.71 | <.001 |
Principal diagnosis of cellulitis | –0.46 (0.11) | 0.63 | <.001 |
Principal diagnosis of pneumonia | 0.37 (0.21) | 1.45 | .08 |
Principal diagnosis of diabetic ketoacidosis | 0.81 (0.13) | 2.26 | <.001 |
Principal diagnosis of urinary tract infection | 2.10 (0.17) | 8.13 | <.001 |
Age | –0.04 (0.004) | 0.97 | <.001 |
Female sex | 0.24 (0.03) | 1.27 | <.001 |
Antibiotics during hospital admission | 0.47 (0.09) | 1.60 | <.001 |
Length of staya | 0.32 (0.05) | 1.38 | <.001 |
Elixhauser comorbidity indexb | 0.15 (0.04) | 1.16 | <.001 |
Results repeated with 10 different random samples of 30000 hospital admissions from the original dataset of admissions. Results shown were robust in repeated samples. An incidence rate ratio of 0.65 indicates that a patient with a principal diagnosis of acute myocardial infection was 35% less likely than the entire sample of patients to receive a urine culture on admission.
aLog transformation.
bSquare root transformation.
Predictors . | Regression Estimate (Standard Error) . | Incidence Rate Ratio . | P Value . |
---|---|---|---|
Principal diagnosis of acute myocardial infarction | –0.10 (0.08) | 0.91 | .21 |
Principal diagnosis of heart failure | –0.16 (0.05) | 0.86 | .001 |
Principal diagnosis of cellulitis | –0.55 (0.08) | 0.58 | <.001 |
Principal diagnosis of pneumonia | –0.01 (0.05) | 0.99 | .85 |
Principal diagnosis of diabetic ketoacidosis | 0.51 (0.12) | 1.66 | <.001 |
Principal diagnosis of urinary tract infection | 2.28 (0.20) | 9.74 | <.001 |
Age | –0.02 (0.003) | 0.98 | <.001 |
Female sex | 0.24 (0.02) | 1.27 | <.001 |
Antibiotics during hospital admission | 0.71 (0.03) | 2.02 | <.001 |
Length of staya | 0.36 (0.03) | 1.44 | <.001 |
Elixhauser comorbidity index | 0.02 (0.001) | 1.02 | <.001 |
Predictors . | Regression Estimate (Standard Error) . | Incidence Rate Ratio . | P Value . |
---|---|---|---|
Principal diagnosis of acute myocardial infarction | –0.10 (0.08) | 0.91 | .21 |
Principal diagnosis of heart failure | –0.16 (0.05) | 0.86 | .001 |
Principal diagnosis of cellulitis | –0.55 (0.08) | 0.58 | <.001 |
Principal diagnosis of pneumonia | –0.01 (0.05) | 0.99 | .85 |
Principal diagnosis of diabetic ketoacidosis | 0.51 (0.12) | 1.66 | <.001 |
Principal diagnosis of urinary tract infection | 2.28 (0.20) | 9.74 | <.001 |
Age | –0.02 (0.003) | 0.98 | <.001 |
Female sex | 0.24 (0.02) | 1.27 | <.001 |
Antibiotics during hospital admission | 0.71 (0.03) | 2.02 | <.001 |
Length of staya | 0.36 (0.03) | 1.44 | <.001 |
Elixhauser comorbidity index | 0.02 (0.001) | 1.02 | <.001 |
Results repeated with 10 different random samples of 29855 hospital admissions from the original dataset of admissions. Results shown were robust in repeated samples. An incidence rate ratio of 0.86 indicates that a patient with a principal diagnosis of heart failure was 14% less likely than the entire sample of patients to receive a urine culture at any time during the hospital admission.
aLog transformation.
Predictors . | Regression Estimate (Standard Error) . | Incidence Rate Ratio . | P Value . |
---|---|---|---|
Principal diagnosis of acute myocardial infarction | –0.10 (0.08) | 0.91 | .21 |
Principal diagnosis of heart failure | –0.16 (0.05) | 0.86 | .001 |
Principal diagnosis of cellulitis | –0.55 (0.08) | 0.58 | <.001 |
Principal diagnosis of pneumonia | –0.01 (0.05) | 0.99 | .85 |
Principal diagnosis of diabetic ketoacidosis | 0.51 (0.12) | 1.66 | <.001 |
Principal diagnosis of urinary tract infection | 2.28 (0.20) | 9.74 | <.001 |
Age | –0.02 (0.003) | 0.98 | <.001 |
Female sex | 0.24 (0.02) | 1.27 | <.001 |
Antibiotics during hospital admission | 0.71 (0.03) | 2.02 | <.001 |
Length of staya | 0.36 (0.03) | 1.44 | <.001 |
Elixhauser comorbidity index | 0.02 (0.001) | 1.02 | <.001 |
Predictors . | Regression Estimate (Standard Error) . | Incidence Rate Ratio . | P Value . |
---|---|---|---|
Principal diagnosis of acute myocardial infarction | –0.10 (0.08) | 0.91 | .21 |
Principal diagnosis of heart failure | –0.16 (0.05) | 0.86 | .001 |
Principal diagnosis of cellulitis | –0.55 (0.08) | 0.58 | <.001 |
Principal diagnosis of pneumonia | –0.01 (0.05) | 0.99 | .85 |
Principal diagnosis of diabetic ketoacidosis | 0.51 (0.12) | 1.66 | <.001 |
Principal diagnosis of urinary tract infection | 2.28 (0.20) | 9.74 | <.001 |
Age | –0.02 (0.003) | 0.98 | <.001 |
Female sex | 0.24 (0.02) | 1.27 | <.001 |
Antibiotics during hospital admission | 0.71 (0.03) | 2.02 | <.001 |
Length of staya | 0.36 (0.03) | 1.44 | <.001 |
Elixhauser comorbidity index | 0.02 (0.001) | 1.02 | <.001 |
Results repeated with 10 different random samples of 29855 hospital admissions from the original dataset of admissions. Results shown were robust in repeated samples. An incidence rate ratio of 0.86 indicates that a patient with a principal diagnosis of heart failure was 14% less likely than the entire sample of patients to receive a urine culture at any time during the hospital admission.
aLog transformation.
Impact of Patient-Level Factors on Urine Culture Testing
Women were more likely to have a urinalysis or a urine culture sent on the first day of admission and during the entire admission compared to men (Tables 3 and 4, Supplementary Materials). Increased length of stay, higher Elixhauser comorbidity index, and receipt of antibiotics during the admission were also positively associated with urine culture testing. For each of prespecified principal diagnoses, women were more likely to have urine cultures on the first day of admission than men, and patients who received antibiotics during the hospital stay were more likely to have a urine culture sent on the first day of admission than patients who did not receive antibiotics (Figure 1). The association between age and urine testing demonstrated higher rates of urine testing on the first day of admission for younger (<30 years) and older patients (>80 years). Trends were similar for urine cultures sent during the entire admission (Supplementary Materials).
DISCUSSION
In our exploration of the national prevalence and patterns of urine testing in the United States, we demonstrated that almost half of the admissions in our sample included a urinalysis, and nearly one-third included 1 urine culture. More than 12% of admissions with a urine culture charge had >1 urine culture sent every 2 days on average, a testing frequency that exceeds the bounds of time (typically 48 hours) required to determine the results of a urine culture. Testing frequency was lower for diagnoses without an indication for routine testing. The impact of female sex and extremes of age on increased urine culture testing was consistent across principal diagnoses and may reflect clinician-level factors that bias toward suspecting urinary tract infections in these specific patient populations.
Our findings confirm a high prevalence of urine testing across a widely representative sample of United States hospitals, with repeated and frequent testing during many admissions. Among admissions with a charge for urine culture, 1 in every 5 hospital stays included charges for >1 urine culture. Prior studies have demonstrated that 58%–68% of urine cultures sent in the hospital are not clinically indicated based on published guidelines [4, 24]. Assuming that the initial and repeat testing within our cohort include a similar degree of unnecessary urine culture testing, adherence to guideline recommendations would have resulted in up to 1000000 fewer urine cultures being sent.
Our findings demonstrate evidence for clinical discrimination in deciding which patients require a urinalysis and urine culture [4, 5, 24]. Patients with principal diagnoses that did not have an indication for routine urine culture testing or antibiotic use were less likely to receive a urinalysis or urine culture test on the first day of admission compared to the rest of the sample. For diagnoses that required antibiotics but did not have an indication for routine testing (cellulitis and pneumonia), the results were mixed. This may be explained by the fact that cellulitis is often diagnosed by clinical examination alone, while pneumonia admissions may undergo a systemic infectious workup that includes urine testing. The discrimination found in testing by diagnosis is important as it demonstrates the role of presenting condition in the mental models clinicians use to determine who requires testing [8, 25]. Future interventions to reduce urine testing may be able to build upon these mental models that recognize that a patient with heart failure is less likely to need a urine culture than one with pneumonia.
Our findings demonstrate that patients who receive antibiotics during their hospital admission are more likely to have also received a urine culture on the first day of admission. Recent literature suggests that inappropriate, often reflexive, urine testing performed in the emergency room contributes to excess antibiotic use during admission [5, 15]. A substantial number of urinalysis and urine cultures in our national sample were ordered on the second day of the admission or later. This evidence suggests that testing stewardship interventions will have to focus on both emergency room providers and inpatient providers.
In our analysis, female sex was associated with a higher likelihood of having a urine culture sent on the first day of the admission regardless of use of antibiotics during the hospital stay, diagnosis, or age. Although women are more likely than men to be hospitalized for urinary tract infection, there is no clinical reason to expect that among patients admitted for congestive heart failure or cellulitis, women would require urine testing in excess of men [26]. While urinary tract infections and hospital admissions for urinary tract infections are common in elderly patients, sex differences in urinary tract infection rates narrow as patients age [27, 28]. The persistence in the sex difference in urine culture testing across ages demonstrates the potential role of cognitive biases as a source for the inconsistencies that may exist between current practice patterns and clinical need [8, 29].
Our study has several limitations. Urine tests were measured by charges for a test, not by tests with an actual result. Only charges that were clearly a urinalysis or urine culture were included in the analyses, which may have resulted in the exclusion of some urine tests with ambiguous charge names and an underestimation of the prevalence of testing. We do not know the indication for testing, how many of the tests were positive, and whether the antibiotic use we observed was in response to a urine test. Prior studies have demonstrated that urine testing in the hospital often does not meet clinical guidelines and that patients are often treated in the hospital inappropriately for asymptomatic bacteriuria [4, 5, 7, 15, 24]. We also do not know whether the patients who were tested had an indwelling urinary catheter; however, catheters are used in a minority of patients nationally [30]. Strengths of our study include the broadly representative sample, both of hospitals and of patients.
This study provides the first national estimate of the frequency and patterns of urinalysis and urine culture testing at the time of and during inpatient admissions in the United States. The harms of unnecessary medical tests have become increasingly clear. Studies repeatedly have shown that inappropriate urine testing leads to inappropriate antibiotic use. Despite this knowledge, urine testing continues to be routine for hospitalized patients nationally. Interventions to decrease unnecessary urine testing are needed, and the predilection for testing that we observed based on patient sex, age, and diagnosis can help guide the design of these interventions [31]. Our work suggests that successful antibiotic stewardship will need to consider aspects of “testing stewardship” as part of a comprehensive program focused on urinary tract infections and asymptomatic bacteriuria.
Supplementary Data
Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.
Notes
Author contributions. All authors contributed equally to the study design and data interpretation. A. S. had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the analysis. M. J. H. was responsible for drafting the manuscript, and all authors contributed equally to critically revising the manuscript.
Disclaimer. The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of the Department of Veterans Affairs, the United States government, or Baylor College of Medicine.
Financial support. This material is based upon work supported in part by the Department of Veterans Affairs, Veterans Health Administration, Office of Research and Development, and the Center for Innovations in Quality, Effectiveness and Safety (CIN 13-413) at the Michael E. DeBakey Veterans Affairs Medical Center, Houston, Texas. M. J. H. was supported by the Veterans Affairs Office of Academic Affiliations Advanced Fellowship in Health Services Research at the Center for Innovations in Quality, Effectiveness, and Safety. A. S. is supported by The Advisory Board Company.
Potential conflicts of interest. All authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.
References
Comments