Examining antibiotic prescribing and urine culture testing for urinary tract infections (UTIs) in a primary care spinal cord injury (SCI) cohort

A retrospective cohort study. To describe antibiotic prescribing and urine culture testing patterns for urinary tract infections (UTIs) in a primary care Spinal Cord Injury (SCI) cohort. A primary care electronic medical records (EMR) database in Ontario. Using linked EMR health administrative databases to identify urine culture and antibiotic prescriptions ordered in primary care for 432 individuals with SCI from January 1, 2013 to December 31, 2015. Descriptive statistics were conducted to describe the SCI cohort, and physicians. Regression analyses were conducted to determine patient and physician factors associated with conducting a urine culture and class of antibiotic prescription. The average annual number of antibiotic prescriptions for UTI for the SCI cohort during study period was 1.9. Urine cultures were conducted for 58.1% of antibiotic prescriptions. Fluroquinolones and nitrofurantoin were the most frequently prescribed antibiotics. Male physicians and international medical graduates were more likely to prescribe fluroquinolones than nitrofurantoin for UTIs. Early-career physicians were more likely to order a urine culture when prescribing an antibiotic. No patient characteristics were associated with obtaining a urine culture or antibiotic class prescription. Nearly 60% of antibiotic prescriptions for UTIs in the SCI population were associated with a urine culture. Only physician characteristics, not patient characteristics, were associated with whether or not a urine culture was conducted, and the class of antibiotic prescribed. Future research should aim to further understand physician factors with antibiotic prescribing and urine culture testing for UTIs in the SCI population.


INTRODUCTION
Urinary tract infections (UTIs) are the number one secondary complication and the second most common cause of mortality for individuals living with spinal cord injuries (SCI) [1]. It has been suggested the average rate of UTIs in an individual with SCI is 2.5 annually [2]. Individuals with SCI are at increased risk of experiencing frequent, recurrent or chronic UTIs due to reduced bladder sensation, incomplete bladder emptying, and catheterizations [1]. UTIs, if untreated or inappropriately treated, could lead to hospitalization, sepsis and in extreme cases mortality [1,3]. Untreated UTIs can also cause patient discomfort impacting their overall quality of life [3]. Hence it is important to appropriately diagnose and treat UTIs. It is also important not to misdiagnose UTI cases due to presence of asymptomatic bacteriuria [1]. Individuals with SCI exhibit higher incidences of asymptomatic bacteriuria than the average population, and treatment of asymptomatic bacteriuria should be avoided to decrease contributing to antimicrobial resistance(AMR) [4][5][6][7]. Although urine cultures can provide bacterial sensitivity information, they are insufficient to confirm UTIs, as positive results can also indicate asymptomatic bacteriuria [4][5][6][7]. Before prescribing antibiotics, it is important to detect signs and symptoms associated with UTIs. Typical signs and symptoms associated with UTIs may differ in individuals with SCI, making diagnoses more difficult [4][5][6][7].
Published practice guidelines may help providers diagnose and treat UTIs in individuals with SCI [4-6, 8, 9]. Although guidelines are not meant to replace clinical judgement, as they may not always be based on comprehensive evidence and capture real-life scenarios sufficiently, they can be a tool providers use to help guide clinical practice [10]. Guidelines from the American Spinal Injury Association (2020), Ghent University Hospital in Belgium (2009), and the SCI High Project working group (2019) are SCIspecific UTI guidelines [5,7,9]; whereas, guidelines from the European Association of Urology (2017), the Canadian Urological Association (2019), and the Infectious Diseases Society of America (2010) are more general with a wider focus either on the general population, neurogenic bladder, or catheter-associated UTIs [4,6,8]. These guidelines emphasize before antibiotics are prescribed to treat UTIs, urine cultures should be done to ensure the most appropriate antibiotic is prescribed for a given infection [4][5][6][7]. Ensuring the right antibiotic class is prescribed to treat UTIs in individuals with SCI is especially important as they may exhibit higher rates of AMR [1,11]. Recent guidelines recommend narrowspectrum antibiotics be prescribed as first-line therapy to treat UTIs to avoid contributing to AMR [4,6]. However, older guidelines developed before 2015 recommended broad-spectrum fluoroquinolone antibiotics as first-line therapy to treat complicated cases of UTI to ensure comprehensive therapy [8,9]. UTI occurrences in those who use catheters, are male, or have a urological abnormality/impairment would be classified as complicated [4], and many individuals with SCI would fall into these categories. Therefore, contrary to current guidelines which recommend narrow-spectrum antibiotics, older guidelines suggest when urine culture tests are unavailable, fluoroquinolones be prescribed as first-line antibiotic therapy to individuals with SCI [4,6,8,9].
Although guidelines suggest appropriate prescribing to treat UTIs in the SCI population, there is little published research that examines antibiotic prescribing or urine cultures for UTIs in outpatient settings, especially in primary care, for the SCI population. It is unclear if providers conduct urine cultures prior to antibiotic prescribing or if they are prescribing guideline suggested first-line antibiotics. As such, research examining urine culture testing and antibiotic prescribing patterns for UTIs in an outpatient SCI population is needed. Since healthcare providers are responsible for ordering urine cultures and prescribing antibiotics, it is important to investigate prescriber factors in addition to patient factors. There is a time-lag between antibiotic prescribing patterns and resistance patterns; as such to address current issues with AMR, research on previous antibiotic prescribing patterns to treat UTIs in the SCI population is needed [12,13]. Therefore, the primary objectives of this paper are: (1) To describe patterns of urine culture orders and antibiotic classes prescribed for UTIs; (2) Identify potential patient and physician factors associated with conducting a urine culture prior to antibiotic prescription; and (3) Identify potential patient and physician factors associated with class of antibiotics prescribed for UTIs.

METHODS
This was a retrospective cohort study to evaluate antibiotic prescribing patterns in relation to urine culture testing patterns for UTIs in individuals with SCI conducted using linked health administrative and EMR databases at ICES, in Ontario, Canada. ICES is an independent, non-profit research institute funded by an annual grant from the Ontario Ministry of Health and the Ministry of Long-Term Care. As a prescribed entity under Ontario's privacy legislation, ICES is authorized to collect and use healthcare data for the purposes of health system analysis, evaluation, and decision support. Secure access to these data is governed by policies and procedures approved by the Information and Privacy Commissioner of Ontario. The use of the data in this project is authorized under section 45 of Ontario's Personal Health Information Protection Act (PHIPA) and does not require further review by a Research Ethics Board. The use of the data in this project is authorized under section 45 and approved by ICES' Privacy and Legal Office.

Cohort development
This study used a cohort of 432 individuals with SCI previously identified in the Electronic Medical Records Primary Care (EMRPC) (formally EMRald) database at ICES for another study investigating antibiotic prescribing in primary care [14]. A tested algorithm (sensitivity = 88.1, PPV = 52.6, Fscore = 65.9), developed based on keyword searches, was used, followed by chart review to verify SCI cases [15,16]. All individuals with a confirmed SCI were included in this study based on a clear diagnoses and associated impairments [17]. Clinical and injury characteristics for this SCI cohort were abstracted manually during the chart review, which included injury duration, type of SCI injury (traumatic SCI (TSCI) vs. non-traumatic SCI (NTSCI)), catheter use, UTI-preventative antibiotic prophylaxis use, ASIA classification, and type of impairments (sensorimotor, sensory only, bladder and bowel). In the EMRPC database terms used to identify sensorimotor impairments included loss of fine motor movement, weakness, use of mobility aids, functional loss of arms or legs, paralysis, spasticity, paraplegia, tetraplegia, and quadriplegia [17]. Terms used to identify sensory only impairments included pain, burning sensations, tingling, loss of sensation, and numbness [17]. Bladder impairments included incontinence, urinary retention, bladder urgency, neurogenic bladder, and spastic bladder [17]. Finally, bowel impairments terms included need for a bowel routine, bowel incontinence, loss of sensation in bowel area, and neurogenic bowel [17].
The Registered Persons Database (RPDB), which contains demographic information about anyone who has ever received an Ontario Health Insurance Plan (OHIP) card, provided information on sex, date of birth, and postal code for the SCI cohort [18]. Rurality was classified based on the Rurality Index for Ontario (RIO) which is based on an individual's postal code with an RIO score less than 10 for urban, 10-39 for suburban, and 40 and above for rural [19]. Please see previous study for further information on cohort development [14].

Antibiotic prescribing information
Antibiotic information was identified and collected in EMRPC database using a previously developed method of keyword searches in the medication field of patient charts, followed by subsequent chart review [14,20]. For this study only antibiotics that were prescribed for a written diagnoses associated with a UTI were included. Antibiotic names were sorted into the following antibiotic classes: (1) fluoroquinolones; (2) nitrofurantoin; (3) sulfonamides, trimethoprim & combinations; (4) penicillins; and (5) other (cephalosporins, metronidazole & tetracyclines). Duration and dose for antibiotic prescriptions were abstracted manually by chart reviews. During chart reviews, UTI indicated antibiotics were determined based on the written indications in the patients' EMRPC chart [14]. For this study, antibiotics categorized as prophylaxis or were prescribed for 90 days or more were excluded (n = 93, 10.2%), as the purpose of this study is to understand patterns of antibiotic treatment in relationship to urine culture testing, whereas prophylaxis prescriptions are intended to prevent UTI occurrences.

Urine culture information
Urine culture information was abstracted by manual chart review in EMRPC. Urine culture information was abstracted for seven days before and after an antibiotic was prescribed for a UTI indication. If the antibiotic was prescribed before the urine culture results were available, the antibiotic was labelled as "pre-culture prescription". If the antibiotic was prescribed after the urine culture results, the antibiotic was labelled as "postresults prescription". If an antibiotic was originally prescribed before the urine culture results, and a revised new antibiotic was prescribed after the test results, the antibiotic was labelled as "revised-culture prescription". If the antibiotic did not have a urine culture associated with it, then the antibiotic was labelled as "no-culture prescription".

Rostered primary care physician description
The cohort of 432 individuals with SCI were rostered to 211 unique primary care physicians (PCPs) [14]. The general demographic information about these rostered PCPs was obtained from the ICES Physician Database (IPDB). The IPDB has information about physician's date of birth, sex, rurality of practice, years of practice and Canadian Medical School Graduate (CMG) or International Medical Graduate (IMG) [18]. For this study, years of practice was categorized as early-career (less than 11 years), mid-career (11-25 years), and late-career (over 25 years) [21].

Data analysis
Means, standard deviations and proportions were used to describe the demographic, injury and clinical characteristics for the SCI cohort, and their rostered PCP. Differences between individuals who received at least one A. Senthinathan et al.
UTI antibiotic and individuals with no UTI antibiotic prescribed were calculated using chi-square, t-test, and the Mann-Whitney U test. Twotailed t-test was conducted to calculate differences between ages and twotailed Mann-Whitey U test for injury duration.
A multinomial logistic regression was conducted to determine if patient, or rostered PCP factors were associated with the class of antibiotic prescribed for UTI indications, with clustering within each individual to account for multiple antibiotics prescribed to the same individual. The outcome categories (independent variable) for these logistic regressions were organized into the top three antibiotics prescribed, and all other antibiotics were grouped into a category labelled "other". A binary clustered logistic regression was conducted to determine if patient or rostered PCPs' factors were associated with urine culture or no urine culture for UTI indicated antibiotics, with the independent variables being (1) urine culture conducted or (2) urine culture not conducted. An assessment of multicollinearity was conducted and factors that exhibited multicollinearity were not modeled together in the regression analyses. Dependent variables used in iterations of the regression analyses included patient (age, sex, duration of injury, rurality, bladder impairment, bowel impairment, sensorimotor impairment, NTSCI vs. TSCI, prophylaxis use, and catheter use) and PCP (sex, age, years of practice, and training location) factors. Multiple iterations of the regression analyses were conducted to identify key factors to include in the final analyses, with only significant factors included in the final model. Analysis was conducted using SAS 9.4 (SAS Institute). Regression analyses used "proc genmod/gee" commands to account for clustering, as such outputs were reported as RR values.

RESULTS
In the SCI cohort of 432 individuals, 158 (36.5%) received 812 antibiotics to treat a UTI between Jan 1, 2013, and Dec 31, 2015. See Table 1 for cohort description. Less than 15% of individuals had an ASIA classification recorded in their chart, hence this information was not reported in Table 1. Those who received at least one antibiotic to treat a UTI were significantly more likely to have a longer duration of injury (U = 39,130, p < 0.0001), have a TSCI (X 2 (df (1), N = 432) = 10.5, p = 0.001), use catheters (X 2 (df (1), N = 432) = 80.7, p < 0.0001), on prophylaxis (X 2 (df (1), N = 432) = 19.1, p < 0.0001), have a bladder impairment (X 2 (df (1), N = 432) = 68.4, p < 0.0001), and bowel impairment (X 2 (df (1), N = 432) = 14.5, p = 0.0001). For the 211 rostered-PCPs included in this study their average age was 44.0 (SD: 0.9) years old, average years of practice was 21.8 (SD: 11.6) years, 50.2% were female, and 10.4% were IMGs. *Indicates significant (p < 0.05) difference between individuals with zero and at least one UTI-indicated antibiotic prescription. **To avoid small cells (less than six) from being re-calculated, missing data for rurality was listed as urban, and missing data for duration of injury as over 20 years. Unclassifiable cases of SCI were listed under the TSCI category.
The average number of UTI indicated antibiotic prescriptions for the SCI cohort of 432 individuals over the three-year study period (Jan 1, 2013-Dec 31, 2015) was 1.9 (SD:4.9, Q1:0, Median:0, Q3:1) ( Table 2). Seventy-six percent of antibiotics were confirmed to be prescribed by a patient's rostered PCP. If not prescribed by the rostered-PCP, they were either prescribed by another practitioner at the same clinic, or this information was missing from the database. Only 25 (5.8%) individuals with SCI had 3 or more annual antibiotic prescriptions for UTIs.
The 158 (36.6%) individuals with SCI who received antibiotics for UTIs were rostered to 108 unique PCPs. The average age of the PCPs was 44.9 (SD: 11.4) years old, and the average years of practice was 21.9 (SD:12.4) years. Forty-five (41.7%) were female and 97 (89.8%) were Canadian medical school graduates. The average rostered-PCP who prescribed antibiotics for UTIs had 1.4 (SD:0.6) patients with a SCI with a range of 1-3 patients with a SCI.
Of those 812 antibiotic prescriptions indicated for UTI, 472 (58.1%) were linked to an antibiotic urine culture. For the antibiotic prescriptions linked to a urine culture, 257 (54.4%) were prescribed before the test results, 151 (31.9%) were prescribed after waiting for the test results, 54 (11.4%) were revised after the test results, and 10 (2.1%) had unknown timing.
Overall, fluroquinolone and nitrofurantoin antibiotics were the most frequently prescribed at 271 (33.4%) prescriptions each. Table 3 outlines the antibiotic classes prescribed based on timing of antibiotic prescription. Fluoroquinolones were the most frequently prescribed antibiotics if no urine culture was done, if the prescription was given before the urine culture results were available, and if the antibiotic prescription was revised post-urine culture results. Nitrofurantoin was the most frequently prescribed antibiotic post-urine culture test results only.
The clustered logistic regression to determine factors associated with class of antibiotic prescribed to treat UTI indications found that male physicians (RR = 2.1, 95% CI: 1.3-3.3, p = 0.003), and physicians who were IMGs (RR = 2.2, 95% CI: 1.2-4.0, p = 0.01) were more likely to prescribe fluoroquinolones, compared to nitrofurantoin (Fig. 1). The clustered logistic regression to determine factors associated with urine culture found PCPs' years of practice (RR = 2.3 95% CI: 1.2-4.2, p = 0.01) was a significant factor with early-career physicians, compared to late-career physicians, more likely to order a urine culture when prescribing a UTI indicated antibiotic (Fig. 1).
Due to multicollinearity between physician age and years of practice, only years of practice was included in the model. As well, multicollinearity was found between patient sex, duration of injury, bladder impairment, bowel impairment, NTSCI vs. TSCI, and catheter use; hence only one of these factors were included in different iterations of the regression model. Patient factors of rurality and sensorimotor impairment had no issues of multicollinearity with other variables and were also included in iterations of regression model. However, no patient related factors were found to be significant in both logistic regression analyses for any of the iterations.

DISCUSSION
For the SCI cohort, we found individuals received less than 2 UTIindicated antibiotic prescription over the three-year study period. However, it was also found that less than 60% of antibiotics prescribed for UTIs in the SCI cohort were associated with a urine culture. Fluoroquinolones were the most prescribed antibiotic to treat UTIs for the cohort if no urine culture was conducted or prior to urine culture results. Only when an antibiotic was prescribed after urine culture tests become available was nitrofurantoin the most frequently prescribed antibiotic. Physician characteristics, rather than patient characteristics, were found to determine if a urine culture was conducted and the class of antibiotic prescribed.
A frequently cited review suggested the average number of UTIs per year for individuals with a SCI is 2.5 [2]. It should be noted that this statistic seems to have stemmed from a single study conducted prospectively in an inpatient hospital in Spain from 1993-1996 in a cohort of 136 patients with acute SCI, with over 90% having a TSCI [22]. Our study found the majority of individuals with SCI did not receive a UTI-indicated antibiotic prescriptions over 3 years. Our study was conducted in an outpatient setting and had more individuals with NTSCI. Additionally, our study did not evaluate the average number of UTIs, but the average number of antibiotics prescribed to treat  UTIs. Our study only investigated primary care outpatient antibiotics, and it is possible individuals received antibiotics for UTIs elsewhere in the community. Furthermore, there is also a 20year difference between the study time periods. These variations between our study and this previous study make the findings difficult to compare. Although guidelines for SCI suggest that a urine culture should be conducted before an antibiotic is prescribed for UTIs, our study found over 40% of antibiotics were prescribed without a urine culture. Our study also found early-career physicians, compared to late-career physicians, were significantly more likely to order a urine culture when prescribing an antibiotic to treat UTI for individuals with a SCI. Similarly, a study conducted in Denmark from 2004-2013 in general practitioners found point-of-care testing for a general population prior to antibiotic prescribing decreased with physician's age [23]. This is in line with our finding that late-career physicians are less likely to conduct a urine culture prior to prescription, as age usually positively correlates to years of practice, as was the case for our study.
The SCI population frequently exhibits high rates of antimicrobial resistance; hence, conducting a urine culture, as suggested by multiple guidelines, before antibiotic prescriptions may be important [1,[5][6][7][8]11]. Our study found less than 60% of antibiotic prescriptions for UTIs adhered to this recommendation, identifying it as a potential area for improvement. However, with the SCI population, there are most likely many circumstantial factors not investigated in this study when it comes to conducting urine culture testing. Individuals with SCI frequently experience mobility issues, and may not have appropriate transportation or face physical barriers when accessing their PCP in a timely manner [24,25]. It has been suggested that individuals with SCI prefer the use of phone or virtual consultations when accessing their PCP [25,26]. Hence, it is possible many of these UTI-indicated antibiotic prescriptions may not happen in-person, but over the phone or virtually, which would make obtaining a urine sample to conduct a urine culture or dip not possible. However, even if prescriptions are given virtually and/or there are physical barriers to obtaining a sample, urine cultures should still be prescribed as they are important to ensuring appropriate antibiotic prescriptions and there are alternatives methods to obtain a sample if needed. There are most likely other factors associated with the physicians' decision to order a urine culture or not before prescribing UTIindicated antibiotics that require further investigation, potentially through qualitative methods, before any interventions can be created.
Our study also found fluoroquinolones accounted for nearly 1/3 of all antibiotics prescribed, which is in line with suggestions from guidelines developed before 2015 that fluoroquinolones be used as first-line therapy for cases of complicated UTIs [8,9]. Most of these fluoroquinolone antibiotics were prescribed without a urine culture or before the availability of urine culture results; hence, these prescriptions were not informed by microbiology susceptibility test results. This study found only physician characteristics determined which class of antibiotic was prescribed, with male physicians and IMGs more likely to prescribe fluoroquinolone antibiotics, rather than nitrofurantoin. This was in line with another study conducted in long-term care facilities in Ontario, Canada, which found female physicians were less likely to prescribe fluoroquinolone antibiotics [27]. A study conducted in Iran also found a significant difference in types of antibiotics classes prescribed between male and female general physicians for a general population [28]. A potential explanation for differences in prescribing patterns between male, and female physicians is female physicians may adopt newer guidelines more quickly or attend continuing medical education, as found in a 2001 United states study [29]. As such female physicians may be early adopters of new prescribing guidelines which endorse the use of narrow-spectrum nitrofurantoin antibiotics to treat UTIs for individuals with SCI [4,6,29]. Another study conducted on the beliefs and attitudes of SCI healthcare providers found when deciding the type of antibiotic class to prescribe, less than 61% of providers considered the risk of contributing to antimicrobial resistance or if the patient already had a resistant organism [30]. Instead the survey found providers chose an antibiotic class based on its ease of use and the efficacy, as well as their previous experience, and knowledge regarding the antibiotic [30]. As such, previous training and exposure to certain antibiotics may be the reason IMGs were more likely to prescribe fluoroquinolones to treat UTIs. It should also be noted different regional guidelines may outline the prescription of nitrofurantoin to treat UTIs may not be appropriate in all circumstances, including UTIs in males, as well as recommend shorter durations of nitrofurantoin due to potential side-effects [4,6,8,9].
There are limitations to this study, which include analysis of data retrospectively from an EMR database, which may have missing data. Additionally, there may be antibiotics prescribed to treat UTIs outside of primary care, which was not captured in this study. Since the work is conducted retrospectively, the study cannot make comments about appropriateness of UTI diagnoses, as information about the steps taken by physicians to make the Fig. 1 Characteristics associated with prescribing antibiotic class and conducting urine culture. Relative risk for significant characteristics associated with (1) prescribing fluoroquinolone, compared to nitrofurantoin, antibiotics for UTIs, and (2) conducting a urine culture when prescribing an antibiotic for UTIs. Error bars outline 95% CI.
diagnoses is unavailable. Additionally, the study attributed antibiotic prescriptions in EMRPC to the rostering PCP. Although it was found at least 76% of prescriptions for the SCI cohort were by their rostering PCP, it is possible that misattribution may have impacted our findings on physician characteristics.
This study found that-despite the availability of published guidelines to help providers appropriately diagnose and treat UTIs in individuals with SCI-guidelines are not being followed in many cases. Further research is needed to understand why providers do not always conduct urine cultures before prescribing antibiotics to treat UTIs for the SCI population, and if first-line antibiotic therapy reflects updated guidelines. Overall, this study has outlined some key areas for improvement and further investigation in primary care when it comes to prescribing and testing patterns for UTIs in individuals with SCI.

DATA AVAILABILITY
The dataset from this study is held securely in coded form at ICES. While legal data sharing agreements between ICES and data providers prohibit ICES from making the dataset publicly available, access may be granted to those who meet pre-specified criteria for confidential access, available at www.ices.on.ca/DAS (email: das@ices.on.ca). The full dataset creation plan and underlying analytic code are available from the authors upon request, understanding that the computer programs may rely upon coding templates or macros that are unique to ICES and are therefore either inaccessible or may require modification.