Antibiotic Utilization and Symptom Improvement in a Retrospective Cohort of Women with Urinary Tract Infection Symptoms

Abstract

Introduction and Hypothesis:

Urinary tract infections (UTIs) are one of the most common bacterial infections in women. We hypothesized that over half of those treated empirically would receive inappropriate antibiotics, those treated expectantly would have lower symptom improvement without antibiotics, and that overall progression to sequelae would be uncommon.

Methods:

In this retrospective cohort study of women with UTI symptoms, we quantified the proportion who received inappropriate antibiotics in those treated empirically, defined as those with a negative urine culture or antibiotics that were changed according to culture sensitivities, and identified factors associated with symptom improvement during expectant management. Secondarily, we sought to determine the proportion of UTI sequelae in both groups. During the study time frame, a modified UTI Symptom Assessment (UTISA) questionnaire was administered at baseline and again, with a global rating for change instrument, when urine culture results were relayed.

Results:

Analyses included 152 women, mean age 66.5 (SD 15.0), 30 (20%) received empiric antibiotics, 122 (80%) expectant management. At baseline, the empiric group reported greater mean scores for dysuria (p<.01), urgency (p<.01), frequency (p<.01), and incomplete emptying (p<.01). Positive culture results were reported for 16 (53%) in the empiric group and 72 (59%) in the expectant group. Inappropriate antibiotics were prescribed to 18 (60%) of the empiric group. A negative urine culture was associated with improvement in symptoms in the expectant group. No subjects experienced UTI sequelae within 30-days of initial evaluation.

Conclusion:

In our cohort of older women with UTI symptoms, deferring antibiotics until urine culture results appeared to be safe and decreased the use of inappropriate antibiotics.

Brief Summary

In an older population of urogynecologic patients seeking treatment of UTI symptoms, deferring antibiotics until urine culture results appears to be a reasonable and safe approach.

Introduction

Urinary Tract Infections (UTIs) affect approximately 60% of women at least once in their lifetime and 20–40% will go on to have one or more recurrent episodes of UTIs.[1,2] UTIs are the second most common indication for antibiotic prescriptions in ambulatory care, accounting for 10–20% of all antibiotic prescriptions,[2–4] but more recent literature suggests that up to half of antibiotic prescriptions for suspected UTI may be unnecessary.[4] In women who suffer from chronic lower urinary tract symptoms, such as urgency and frequency, it is even more difficult to determine when UTI symptoms require antibiotic therapy.[5,6] Previous studies of younger women with UTI symptoms report only 64–77% of patients have a positive urine culture suggesting that a substantial number of patients should not receive antibiotic therapy for their symptoms.[2,7,8] It is currently unclear if there are clinical predictors to help determine who may need antibiotics to clear UTI symptoms. Furthermore, barriers to improved antimicrobial stewardship in the treatment of UTIs include patient and provider concern for increased mortality, despite evidence that progression of uncomplicated UTI to systemic infection is very low in older women.[2,4,9]

The majority of previous studies of initial conservative measures for treatment of UTIs have primarily enrolled younger women.[2,7,10–13] Therefore, we sought to evaluate older women with UTI symptoms who were managed both expectantly and empirically. Our goals were to determine the proportion of empirically managed patients that received inappropriate antibiotics, investigate baseline characteristics that were associated with symptom improvement in those managed without empiric antibiotics, and determine the proportion that progressed to UTI sequelae of pyelonephritis and/or sepsis. We hypothesized that over half of those treated empirically would receive inappropriate antibiotics, those with higher dysuria scores on baseline questionnaires would have a lower chance of symptom improvement without antibiotics, and that overall progression to sequelae would be very uncommon events.

Materials and Methods

This is a retrospective analysis of women seeking treatment for urinary tract infection (UTI) symptoms at an academic urogynecology practice from November 1, 2020 to March 3, 2021. During this time, our division had undertaken a quality improvement project (QIP) that included administration of a modified Urinary Tract Infection Symptom Assessment (UTISA) questionnaire by phone or in person, to female patients with UTI symptoms prior to the collection of urine cultures. The UTISA is a screening tool that evaluates the severity of UTI symptoms and has been validated to assess uncomplicated UTI outcomes.[14] Our questionnaire specifically evaluated the severity of seven UTI symptoms: frequency, urgency, dysuria, incomplete bladder emptying, abdominal pain or pressure, low back pain, and hematuria as well as overall symptom severity. For each of the symptom domain questions, patients were asked to respond on a Likert-type scale, with options of: ‘did not have,’ ‘mild,’ ‘moderate,’ and ‘severe,’ which were scored 0–3, respectively. Higher numerical values signified increased perceived severity of that symptom.[14] Degree of bother items were excluded from the questionnaire. Of note, minimally important differences (MIDs) for symptom domains have been reported as follows: 1.75, 1.50, 1.25, and 0.50 for regularity of urination, urination problems, UTI-associated pain, and hematuria domains, respectively.[14]

The QIP sought to explore the safety of deferring antibiotics until culture results were available in women who were managed with and without empiric antibiotics. After urine cultures were resulted, but before they were shared with patients, the UTISA questionnaire was again administered. A global rating of change (GRC) question was also administered to those patients who were managed expectantly. The GRC, which has also been validated, evaluated change from baseline in UTI symptoms (same, better, or worse). At our institution, urinalysis is not ordered prior to culture in this setting. Patients are instructed to provide a midstream, clean-catch urine specimen, though in this study some patients provided a catheterized sample in the office. Additionally, urine cultures require 48–72 hours to result, after which patients received a phone call with the questionnaire and results.

For our retrospective analysis, women with UTI symptoms within the study time frame were included if they completed a baseline UTISA (either via telephone encounter or in the office) and submitted a urine specimen for culture. We excluded patients without a baseline UTISA or urine culture. A chart review was performed to obtain demographic, questionnaire responses, and urine culture results. Data on recent instrumentation, history of pyelonephritis or sepsis, and initial presentation of UTI symptoms to a primary care physician or outside office was not collected. Follow up data up to 30 days after initial presentation included office, emergency room or inpatient encounters to determine if adverse sequelae of pyelonephritis and/or sepsis were diagnosed. Demographic data included history of recurrent UTI (rUTI), defined as a previous experience of greater than 3 culture-proven UTIs in one year or at least 2 or more culture-proven UTIs in 6 months. Positive urine culture was defined as greater than or equal to 10³ colony-forming units from either a clean catch or catheterized specimen. Women were grouped by management strategy to either: 1) Empiric antibiotic group (Empiric) defined as those given an antibiotic prescription prior to urine culture results or 2) Expectant management group (Expectant) defined as those managed without antibiotics while awaiting urine culture results. Empiric antibiotics were prescribed at the discretion of providers in our practice, and not based on certain criteria nor the baseline UTISA. Women treated empirically were instructed to provide a urine specimen prior to starting the antibiotic course. The empiric group was further characterized as receiving appropriate therapy if the culture was positive and the antibiotics were sensitive to the resultant bacteria. Women were categorized in the inappropriate empiric antibiotics group if: 1) their urine culture was negative or 2) they required a change in antibiotics due to uropathogen resistance to the empiric antibiotic. “No growth” and “mixed flora” results were included in the “negative urine culture” group for analysis.

Our co-primary outcomes were to: 1) determine the proportion of inappropriate antibiotics prescribed to the empiric group and 2) identify factors associated with symptom improvement in the expectant group. We defined symptom improvement as responding ‘better’ on the GRC and those that were “not improved” as responding ‘about the same’ or ‘worse’ on the GRC. Our secondary aim was to assess the proportion of pyelonephritis and/or sepsis within 30-days of initial UTI evaluation in the empiric and expectant groups.

Data analysis was performed with SPSS 28 software. We compared variables between groups utilizing Chi-squared test, Fisher’s exact test, Mann-Whitney U and Kruskal Wallis Test. A logistic regression was performed to determine factors associated with symptom improvement in the expectant group. This research was approved under the University of Pittsburgh Institutional Review Board (STUDY21030043) as well as the University of Pittsburgh Quality Improvement Review Board (2963).

Results

A total of 152 women with UTI symptoms were included in analyses; 122 (80%) were managed expectantly and 30 (20%) received empiric antibiotics. Their mean age was 66.5 years (SD 15.0), mean BMI 28.2 (SD 6.2), and 79 (52%) reported a history of recurrent UTI. Groups did not differ significantly on demographic variables, however at baseline, the empiric group reported greater mean scores on the UTISA questionnaire for dysuria (p<.01), urgency (p<.01), frequency (p<.01), and incomplete emptying (p<.01). Positive urine cultures were reported for 16 women (53%) in the empiric group and 72 (59%) in the expectant group. (Table 1)

Table 1:

Characteristics and Baseline UTI Symptoms of Women Managed Expectantly and with Empiric Antibiotics

Variable	Overall (n=152)	Expectant (n=122)	Empiric (n=30)	p-value
Demographic Variables
Age	66.2 (15.2)	69 [58–76]	71.5 [61.8–79]	.25
BMI	28.3 (6.3)	26.9 [23.7–32.1]	26.2 [23.2–28.9]	.39
History of diabetes	30 (20%)	27 (22%)	2 (7%)	.05
Caucasian	145 (97%)	117 (96%)	28 (93%)	.63
Current antibiotic suppression	18 (11%)	17 (14%)	2 (7%)	.37
Current vaginal estrogen	92 (61%)	67 (55%)	21 (70%)	.13
Current Smoker	13 (9%)	9 (7%)	4 (13%)	.47
History of recurrent UTI	79 (52%)	61 (50%)	18 (60%)	.33
Positive urine culture	88 (58%)	72 (59%)	16 (53%)	.57
Baseline UTISA Responses
Frequency	1.8 (1.0)	1.7 (0.8)	2.5 (0.6)	<.01
Urgency	1.8 (0.9)	1.7 (0.9)	2.3 (0.9)	<.01
Dysuria	1.4 (1.1)	1.2 (1.1)	1.9 (1.2)	<.01
Incomplete bladder emptying	0.8 (1.1)	0.6 (0.9)	1.5 (1.3)	<.01
Pain/abdominal pressure	1.1 (1.0)	1.0 (1.0)	1.5 (1.0)	.06
Low back pain	0.7 (0.9)	0.6 (0.9)	0.9 (1.1)	.18
Hematuria	0.1 (0.5)	0.1 (0.4)	0.2 (0.6)	.49
Overall symptom rating	2.1 (0.7)	2.0 (0.7)	2.5 (0.7)	<.01

Data is presented as n (%), mean (SD), or median [IQR] unless otherwise specified

Abbreviations: UTI, Urinary Tract Infection, UTISA, Urinary Tract Infection Symptom Assessment; BMI, Body Mass Index

In the empiric group, nitrofurantoin was the most commonly prescribed oral antibiotic (18, 60%) followed by cephalexin, (5, 17%) and cefuroxime (3, 10%). For 18 of 30 women (60.0%), these were inappropriate choices. Four women required a change in antibiotics due to resistance of uropathogen to originally prescribed antibiotic and 14 women had negative urine cultures. Compared to those receiving appropriate empiric antibiotics, the inappropriate empiric group had higher BMI (27.3 vs 24.7, p=.04) and higher mean urgency and frequency scores (2.5 vs. 1.9, p=.03 and 2.7 vs 2.2, p=.02, respectively). The inappropriate empiric group had lower dysuria scores, though this did not meet statistical significance (1.7 vs. 2.3, p=.68). (Table 2) All 72 women with a positive culture in the expectant group were prescribed antibiotics based upon sensitivity test results.

Table 2:

Demographics and Baseline Variables Associated with Appropriate Empiric Therapy for UTI symptoms in Women

Variable	Appropriate Antibiotics in Empiric group (n = 12)	Inappropriate Therapy in Empiric Group (n=18)	P-value
Demographic Variables
Age (median, IQR)	75 [65–87.5]	69.5 [60.8–75]	.15
BMI (median, IQR)	24.7 [23.1–27.1]	27.3 [24.8–32.9]	.04
Diabetes	0 (0%)	2 (11%)	.50
Caucasian	12 (100%)	16 (89%)	.50
Current Antibiotic Suppression	0 (0%)	2 (11%)	.50
Current Vaginal Estrogen	9 (75%)	12 (67%)	.70
Current Smoker	0 (0%)	4 (22%)	.13
History of recurrent UTI	9 (75%)	9 (50%)	.26
Baseline UTISA Responses
Frequency	2.2 (0.6)	2.7 (0.5)	.02
Urgency	1.9 (0.9)	2.5 (0.8)	.03
Dysuria	2.3 (0.6)	1.7 (1.4)	.68
Incomplete bladder emptying	1.6 (1.1)	1.4 (1.4)	.71
Pain/abdominal pressure	1.5 (1.0)	1.6 (1.0)	.59
Low back pain	1.0 (1.3)	0.9 (0.9)	.90
Hematuria	0 (0)	0.3 (0.8)	.14
Overall symptom rating	2.3 (0.5)	2.6 (0.8)	.02

Data is presented as n (%), mean (SD), or median [IQR] unless otherwise specified

Abbreviations: UTI, Urinary Tract Infection, UCx, Urine Culture, UTISA, Urinary Tract Infection Symptom Assessment; BMI, Body Mass Index

For those managed expectantly, 78/122 (64%) completed a follow-up UTISA questionnaire after their urine culture was resulted. (Table 3) There were no significant differences in demographic and baseline variables between those who experienced symptom improvement and those who did not. The improved group reported significantly greater reduction in severity of UTISA symptoms in the areas of frequency (−0.7 vs. 0.2, p=.01), urgency (−0.8 vs. 0.1, p=.03), dysuria (−1.0 vs. −0.2, p=.01), and overall symptoms (−0.5 vs. 0.1, p=.01), however both groups did not meet the prespecified MIDs per domain. Positive cultures were present in the vast majority, 54 of 58 (93%), of those who did not report improved symptoms, compared to 11 of 20 (55%) patients reporting symptom improvement. Although 46% of the total study cohort did not have available UTISA follow-up, there were no differences in baseline variables between those with and without follow-up data, so sensitivity analysis was not performed.

Table 3:

Characteristics and Urinary Symptoms in Expectant Group by Improvement Status^*

Variable	Not improved (n=58)	Improved (n=20)	p-value
Demographic Variables
Age (mean, SD)	68 [59–76]	72 [59–79]	.29
BMI (mean, SD)	27.3 [24.1–30.7]	27.6 [23.7–32.1]	.57
History of diabetes	13 (22%)	6 (30%)	.55
Caucasian	56 (97%)	20 (100%)	.61
Current antibiotic suppression	10 (17%)	1 (5%)	.27
Current vaginal estrogen	34 (59%)	13 (65%)	.62
Current Smoker	4 (7%)	1 (5%)	1.0
History of recurrent UTI	35 (60%)	11 (55%)	.68
Positive urine culture	54 (93%)	11 (55%)	<.01
Baseline UTISA Responses
Frequency	1.6 (0.9)	1.7 (0.9)	.76
Urgency	1.7 (0.9)	1.9 (1.0)	.45
Dysuria	1.4 (1.0)	1.5 (1.1)	.54
Incomplete bladder emptying	0.6 (0.9)	0.5 (1.1)	.36
Pain/abdominal pressure	1.2 (1.1)	1.1 (0.9)	.74
Low back pain	0.8 (1.0)	0.6 (0.9)	.25
Hematuria	0.1 (0.4)	0.1 (0.2)	.57
Overall symptom rating	2.0 (0.7)	1.9 (0.6)	.69
Change in UTISA Responses
Frequency	0.2 (1.1)	−0.7 (0.9)	.01
Urgency	0.1 (1.1)	−0.8 (1.3)	.03
Dysuria	−0.2 (1.0)	−1.0 (1.0)	.01
Incomplete bladder emptying	0.3 (1.3)	0.0 (0.8)	.54
Pain/abdominal pressure	−0.2 (1.0)	−0.4 (0.9)	.34
Low back pain	0.0 (0.8)	−0.1 (0.7)	.51
Hematuria	0.0 (0.4)	−0.1 (0.3)	.44
Overall symptom rating	0.1 (0.8)	−0.5 (1.0)	.01

^*We defined symptom improvement as responding ‘better’ on the Global Rating for Change (GRC) and those who were “not improved” as responding ‘about the same’ or ‘worse’ on the GRC.

Data is presented as n (%), mean (SD), or median [IQR] unless otherwise specified

Abbreviations: UTI, Urinary Tract Infection, UTISA, Urinary Tract Infection Symptom Assessment; BMI, Body Mass Index

A logistic regression analysis was carried out to investigate factors associated with improvement in symptoms. Only negative urine culture was found to be significantly related to likelihood of improvement (aOR = 9.40, 95% CI (1.77–50.06), p <.01) when controlling for age (aOR 1.03, 95% CI (0.98–1.08), p=.19), baseline UTISA domain dysuria (aOR 1.07, 95% CI (0.63–1.82), p=.80), and history of recurrent UTI (aOR 0.79, 95% CI (0.25–2.48), p=.68).

Finally, there were 0 (0%) women in either group who experienced pyelonephritis and/or sepsis within 30 days of initial presentation.

Discussion

In our cohort of women with UTI symptoms, none developed serious sequelae including pyelonephritis or sepsis despite the majority waiting for urine culture results prior to receiving antibiotics. Importantly, of women treated with empiric antibiotics, over half were prescribed inappropriate antibiotics due to a negative culture or resistant uropathogens. Considering the lack of UTI sequelae, our study suggests that deferring antibiotics until urine culture results may be a reasonable and safe approach especially in older populations or in those with a higher chance for resistant uropathogens. Finally, our data suggests that in an older woman with predominant urgency and/or frequency symptoms, empiric antibiotics may not be necessary.

Our finding that over 50% of women treated empirically had negative urine cultures is consistent with the literature.[2,4] Additionally, over 25% of those who completed follow-up questionnaires were improved by the time their urine cultures resulted. Other studies in the literature report that this number may be even higher with 40–60% of women reporting improvement with conservative management and no antibiotics.[2,7,10–13]

Compared to those managed expectantly, women receiving empiric antibiotics had significantly higher overall baseline UTISA symptom scores. Those who received inappropriate antibiotics scored the highest on frequency and urgency on the baseline UTISA compared to the expectantly managed patients, which is surprising given that 14/18 (78%) should not have received antibiotics at all due to a negative culture. Although both groups did not meet the MID of 1.50 for problems with urination (dysuria) or 1.75 for urination regularity (urgency and frequency) based on the follow up UTISA, higher baseline frequency and urgency scores are still clinically valuable. While these symptoms have been associated with UTI and warrant an evaluation, dysuria is a better clinical indicator of UTI because of the significant symptom overlap between UTI and overactive bladder symptoms in this population.[5,6,15] However, even when dysuria was evaluated as a factor associated with clinical improvement, along with age and history of rUTI, there was no association with improvement except for a negative urine culture, which reinforces the importance of waiting for culture results in symptomatic women. The poor predictive value of irritative voiding symptoms supports the importance of patient education, reassurance, fear mitigation and encouragement to utilize symptom mitigation interventions such as phenazopyridine, non-steroidal anti-inflammatory drugs or acetaminophen while awaiting culture results.[16,17]

Clinicians can be reassured that delaying antibiotic treatment until confirmatory culture results are available appears to be safe in a population like ours (ambulatory women with average age in their 60s, many with a history of rUTI). This data is encouraging for providers who may be skeptical of waiting for culture results, given concerns for increased symptom burden and progression of disease.[2,7,10–13] Recent data on progression of symptoms after evaluation for UTI are mixed. In a large claims-based data analysis Gharbi et al. demonstrated that in a population of women with mean age 76.7 years, deferred or no antibiotics was associated with a 7.12 fold higher adjusted odds of bloodstream infection and a 8.08 fold higher adjusted odds of all-cause mortality compared to empiric antibiotics.[4] On the other hand, Gulliford et al. reported overall low probability of sepsis after evaluation for UTI and a number needed to treat of 282 to prevent one episode of sepsis when considering women who are evaluated for UTI.[9] These claims-based studies may have limited ability to discern if the subject actually had sepsis and also what their initial symptomatology was during their UTI evaluation. Our study demonstrates that expectant management appears to be a safe option in a well described population, but larger prospective studies that can capture these rare outcomes are necessary.

Strengths of this project include the cohort of older women in a urogynecologic practice. Many women seek urogynecologic care for urge urinary incontinence, overactive bladder, rUTI, and pelvic organ prolapse among other diagnoses, of which symptoms may overlap with UTI symptoms.[5,6,18] An additional strength of our study was the utilization of the UTISA and GRC questionnaires, which have previously been validated. This questionnaire enables investigators to characterize populations by their subjective UTI symptoms as well as enabling us to utilize the minimally important differences for the various UTISA domains. Future studies can consider the validation of our modified UTISA, which was less cumbersome for the office staff to use as compared to the full version.

Limitations of our study include the retrospective nature, whereas there is always a chance of error in coding and data extraction. Moreover, follow-up data may have been absent from our electronic medical record if patients presented to an outside hospital or system. Another limitation is the small, convenience sample restricted to women from a urogynecology practice with access to urine cultures and prompt treatment based on culture results. As a small sample it is underpowered to evaluate our secondary objective of pyelonephritis and/or sepsis, though these outcomes are so uncommon very large studies are needed. Given that the mean age of our patients was 66.5 (SD 15.2) and the vast majority of patients were Caucasian (97%), these results may not be generalizable to younger or markedly older populations of women seeking care for UTI symptoms given the homogeneity of the population or women with less access to evaluation with urine cultures. While the QIP was conducted with a validated instrument, it has not been validated over the phone, which can be a focus of future studies. Women were not randomized into empiric or expectant management groups which may have introduced treatment bias. Additionally, we did not collect information on the use of pain medications such as acetaminophen or phenazopyridine which may have altered symptoms. However, the results reflect typical practice patterns. Finally, many women did not complete the follow-up questionnaire, often due to high volume of clinical calls for office staff, which is representative of a busy tertiary care clinic.

In conclusion, in an older population of urogynecologic patients seeking treatment of UTI symptoms, deferring antibiotics until urine culture results appears to be a reasonable and safe approach. Of those prescribed empiric antibiotics, 60% were inappropriately prescribed due to a negative culture or resistant uropathogens. Importantly, our data suggests that in an older woman with predominant urgency and/or frequency symptoms, empiric antibiotics may not be necessary, and symptoms may self-resolve. Future studies should evaluate clinical predictors to help determine who may need antibiotics to clear UTI symptoms.