Outpatient Urinary Tract Infections in an Era of Virtual Healthcare: Trends From 2008 to 2017

Abstract

Background

Urinary tract infections (UTIs) occur commonly, but recent data on UTI rates are scarce. It is unknown how the growth of virtual healthcare delivery affects outpatient UTI management and trends in the United States.

Methods

From 1 January 2008 to 31 December 2017, UTIs from outpatient settings (office, emergency, and virtual visits) were identified from electronic health records at Kaiser Permanente Southern California using multiple UTI definitions. Annual rates estimated by Poisson regression were stratified by sex, care setting, age, and race/ethnicity. Annual trends were estimated by linear or piecewise Poisson regression.

Results

UTIs occurred in 1 065 955 individuals. Rates per 1000 person-years were 53.7 (95% confidence interval [CI], 50.6–57.0) by diagnosis code with antibiotic and 25.8 (95% CI, 24.7–26.9) by positive culture. Compared to office and emergency visits, UTIs were increasingly diagnosed in virtual visits, where rates by diagnosis code with antibiotic increased annually by 21.2% (95% CI, 16.5%–26.2%) in females and 29.3% (95% CI, 23.7%–35.3%) in males. Only 32% of virtual care diagnoses had a culture order. Overall, UTI rates were highest and increased the most in older adults. Rates were also higher in Hispanic and white females and black and white males.

Conclusions

Outpatient UTI rates increased from 2008 to 2017, especially in virtual care and among older adults. Virtual care is important for expanding access to health services, but strategies are needed in all outpatient care settings to ensure accurate UTI diagnosis and reduce inappropriate antibiotic treatment.

Outpatient urinary tract infection (UTI) rates increased from 2008 to 2017, especially in virtual care settings, where most patients were empirically treated for UTI. Strategies in all outpatient care settings are needed to optimize UTI diagnosis and reduce inappropriate antibiotic treatment.

Urinary tract infections (UTIs) are among the most common bacterial infections [1, 2]. UTIs affect females and males of all ages, but incidence is highest in women. In the United States (US), UTIs result in >10 million outpatient visits per year [3] and are the third-leading cause for outpatient antibiotic prescription in adults and children [4, 5].

Despite the substantial burden of UTIs on the health system, there is a dearth of contemporary data on UTI rates in the US. Several studies in the past 2 decades have focused on UTIs in specific populations, such as children [6, 7], women [8–10], or elderly patients [11–13]. However, few of these studies have addressed outpatient UTI rates at the health system level or compared rates between patients with different demographic characteristics. Studies also use varying definitions of UTI, such as patient self report, diagnostic codes, or positive urine culture, complicating comparisons across study populations.

Furthermore, it is unknown how UTI rates have been affected by healthcare delivery transitions toward virtual care via phone, video, and the internet. In a study of insured patients concentrated in the US South, virtual visits for all conditions grew 50% annually from 2005 to 2017 [14]. For UTI, virtual visits (compared to office visits) may be more likely to result in diagnosis and antibiotic treatment without confirmatory urine testing [15]. Among low-risk women with symptoms of cystitis, a large subpopulation of patients with UTI, empirical treatment is routine practice across care settings [16, 17]. This is likely to contribute to overuse of antibiotics in those with UTI symptoms, despite growing efforts to strengthen antibiotic stewardship for treatment of UTI [18].

To assess trends in outpatient UTI rates including cystitis and pyelonephritis in the US over a period of increasing virtual healthcare delivery, we conducted a retrospective cohort study from 2008 to 2017, using multiple definitions of UTI to compare UTI rates by care setting, sex, age group, and race/ethnicity.

METHODS

The study was conducted among members of Kaiser Permanente Southern California (KPSC), an integrated healthcare organization that serves >4.5 million residents of Southern California. The diverse member population is largely representative of the underlying population [19]. Electronic health records (EHRs) store member health information such as sociodemographics, diagnoses, laboratory tests, procedures, and medications from all care settings (inpatient, primary care office visits, emergency department [ED], and virtual encounters. Data captured in the KPSC EHR are generally considered comprehensive, as the prepaid system motivates members to use services internally and claims for reimbursement of outside care must be submitted to the health plan with documentation. Ethical approval for the study was obtained from the KPSC Institutional Review Board.

The study population included individuals of all ages who sought care for UTI at KPSC from 1 January 2008 to 31 December 2017. Outpatient UTIs were identified from office visits, ED, and virtual care settings (care delivered through phone, video, and the internet) using different combinations of diagnosis codes for cystitis and pyelonephritis, antibiotic prescriptions, and/or urine culture results. From 1 January 2008 to 30 September 2015, UTI diagnoses were identified from the International Classification of Diseases, Ninth Revision codes 595.0, 595.9, 599.0, and 590.1x. From 1 October 2015 onward, International Classification of Diseases, Tenth Revision codes N30.0x, N30.9, N39.0, and N10 were used. These diagnosis codes include cystitis and pyelonephritis, as well as “UTI of unspecified cite.” Positive urine culture results were defined according to KPSC laboratory guidelines as isolation of ≥1000 colony-forming units (CFU)/mL for sterile samples and ≥10 000 CFU/mL for clean-catch samples. Laboratory results provided to KPSC clinicans are based on these thresholds; therefore, we used these thresholds in our definitions instead of higher thresholds (eg, ≥100 000 CFU/mL) to reflect real-world practice. Cultures that were identified by the laboratory as positive for contaminants were excluded. To minimize antibiotics prescribed for other conditions, antibiotic prescriptions for UTIs were identified from a predefined list of antibiotics (Supplementary Box 1) ordered on the same day as a UTI diagnosis code or urine culture order.

Six definitions of UTI were created: (1) UTI diagnosis code; (2) UTI diagnosis code with antibiotic; (3) positive culture; (4) positive culture with antibiotic; (5) positive culture with diagnosis code; and (6) positive culture with diagnosis code and antibiotic. Some UTIs fell under multiple definitions, for example, definition 1 (UTI diagnosis code) includes definition 2 (UTI diagnosis code with antibiotic). Definition 2 and definition 3 (positive culture) are most common in the literature [12, 20, 21] and were the main focus of analyses.

Occurrences of UTI for each of the 6 definitions were collapsed into UTI events, as multiple healthcare encounters may have occurred for the same UTI. Incident UTI events were defined as occurrences >30 days apart. Outpatient care setting (ie, office visit, ED, or virtual) was assigned based on the care setting of the first occurrence in the 30-day period. Recurrent UTIs were defined as ≥3 UTI events within 365 days [22].

UTI rates were calculated for each definition as the number of UTI events divided by the number of active KPSC members during the year, for each year from 2008 to 2017. Annual rates were calculated per 1000 people and stratified by sex, outpatient care setting, age, and race/ethnicity. Overall rates per person-year from 2008 to 2017 by definition, outpatient care setting, age, and race/ethnicity were estimated and compared using Poisson regression models with robust error variance. Trends in UTIs over the study period were assessed using average annual percentage change (AAPC), which is a weighted average of the annual percentage change. AAPC and 95% confidence interval (CI) for each trend line were estimated using Poisson regression with linear or piecewise linear time model. All analyses were conducted using SAS version 9.3 software (SAS Institute, Cary, North Carolina).

RESULTS

Characteristics of UTI Events

During the 10-year period, 1 065 955 individuals experienced at least 1 UTI defined by a diagnosis code with antibiotic (mean, 2.1 [standard deviation {SD}, 2.3] events per person), while 623 355 individuals experienced at least 1 UTI defined by a positive culture (mean, 1.7 [SD, 1.8] events per person). Recurrent UTI occurred in 9.2% of individuals with UTI by diagnosis code with antibiotic and 6.0% by positive culture.

Overall, there were 2 227 576 UTI events by diagnosis code with antibiotic and 1 068 390 UTI events by positive culture, with 540 162 UTIs included in both definitions (ie, definition 6, positive culture with diagnosis code and antibiotic). Approximately 90% of UTIs occurred in females (Table 1). Most of the UTI burden in females by both definitions was spread across women aged 25–84 years. In males, more than half of UTI events occurred in men aged ≥65 years. The majority of UTI events were seen at office visits for both females and males, and the proportion seen at the ED was higher in males than females. Virtual care accounted for >1 in 4 UTIs by diagnosis code with antibiotic in females (27.6% vs 18.2% in males). Because urine cultures can be ordered as part of virtual care, 17.6% of positive culture events in females and 14.7% in males occurred in virtual care (Table 1).

Table 1.

Characteristics of Outpatient Urinary Tract Infection Events at Kaiser Permanente Southern California, 2008–2017

	Events by Diagnosis Code With Antibiotic Definition		Events by Positive Culture Definition
	Females	Males	Females	Males
Characteristic	(n = 2 007 573)	(n = 220 003)	(n = 919 923)	(n = 148 467)
Outpatient care setting
Office visit	1 276 995 (63.6)	141 805 (64.5)	679 900 (73.9)	103 809 (69.9)
Emergency department	177 369 (8.8)	38 165 (17.4)	77 734 (8.5)	22 908 (15.4)
Virtual	553 209 (27.6)	40 033 (18.2)	162 289 (17.6)	21 750 (14.7)
Age, y
0–14	90 853 (4.5)	8542 (3.9)	66 225 (7.2)	9531 (6.4)
15–24	249 359 (12.4)	8081 (3.7)	102 943 (11.2)	3071 (2.1)
25–44	606 735 (30.2)	28 406 (12.9)	235 251 (25.6)	12 399 (8.4)
45–64	591 651 (29.5)	64 264 (29.2)	236 346 (25.7)	40 312 (27.2)
65–84	392 088 (19.5)	89 073 (40.5)	229 488 (25)	66 915 (45.1)
≥85	76 887 (3.8)	21 637 (9.8)	49 670 (5.4)	16 239 (10.9)
Race/ethnicity
Asian or Pacific Islander	153 133 (7.6)	13 233 (6.0)	70 490 (7.7)	9138 (6.2)
Black	166 433 (8.3)	24 177 (11)	68 375 (7.4)	15 959 (10.8)
Hispanic	815 934 (40.6)	74 187 (33.7)	366 762 (39.9)	48 019 (32.3)
Othera	73 785 (3.7)	6285 (2.9)	30 559 (3.3)	3495 (2.4)
White	798 288 (39.8)	102 121 (46.4)	383 737 (41.7)	71 856 (48.4)

^aIncludes other, missing or unknown, or multiple races/ethnicities.

More than 70% of uropathogens isolated in positive cultures were Escherichia coli. Of UTI by diagnosis code with antibiotic in all care settings, 36.7% had a urine culture order, and 24.5% were positive (63.7% of those with a urine culture order). However, in virtual care settings, 31.6% had a urine culture order, and only 8.9% were positive (28.9% of those with a urine culture order).

UTI Rates and Trends

During the 10-year period, UTI rates varied considerably by definition but increased from 2008 to 2017 by all definitions (Figure 1 and Supplementary Table 1). For example, rates per 1000 person-years were higher by diagnosis code with antibiotic (53.7 [95% CI, 50.6–57.0]) than by positive culture (25.8 [95% CI, 24.7–26.9]). Rates increased 3.1% per year (95% CI, 2.9%–3.3%) by diagnosis code with antibiotic and 2.2% per year (95% CI, 1.4%–3.0%) by positive culture.

Figure 1.

Rates of outpatient urinary tract infection by different definitions at Kaiser Permanente Southern California, 2008–2017. Abbreviation: UTI, urinary tract infection.

Virtual care settings experienced large increases in UTI rates compared to office visits and ED settings, although rates remained highest for office visits (Figure 2 and Supplementary Table 2). By diagnosis code with antibiotic, rates for virtual care increased in females by 21.2% per year (95% CI, 16.5%–26.2%) and 29.3% per year (95% CI, 23.7%–35.3%) in males. Rates for office visits declined 2.8% per year in females and 1.1% per year in males. By positive culture, moderate increases were observed in virtual care (AAPC, 6.2% [95% CI, 5.5%–6.9%] in females and 5.6% [95% CI, 4.7%–6.6%] in males). Lesser increases (AAPC range, 2.8%–4.4%) were observed for males in office visits and for both females and males in ED settings.

Figure 2.

Rates of urinary tract infection by sex and outpatient care setting at Kaiser Permanente Southern California, 2008–2017. Abbreviation: UTI, urinary tract infection.

UTI rates also varied by age group in both females and males (Figure 3 and Supplementary Table 3). Rates were highest among those aged ≥85 years, even though this age group represented a relatively small proportion of total UTI events (Table 1). By diagnosis code with antibiotic, rates increased by 1.1% to 6.4% per year for females and males of all age groups, except males aged 25–44 years for whom rates declined by 1.2% per year. By positive culture, rates increased only in females and males ≥45 years (AAPC range, 1.9%–4.4%). For both definitions, the largest increases in UTI rates occurred in the oldest individuals. The mean age in years of the underlying KPSC population increased from 37 (SD, 22) in 2008 to 39 (SD, 22) in 2017.

Figure 3.

Rates of urinary tract infection by sex and age group at Kaiser Permanente Southern California, 2008–2017. Abbreviation: UTI, urinary tract infection.

There were also differences in UTI incidence by race/ethnicity (Figure 4 and Supplementary Table 4). By diagnosis code with antibiotic and by positive culture, rates were highest in white and Hispanic females and white and black males. By diagnosis code with antibiotic, rates increased in all race/ethnicity categories (AAPC range, 1.7%–3.3% per year for females and 0.6%–4.7% per year for males). By positive culture, increases were also observed in all race/ethnicity categories (AAPC range, 0.6%–2.5% per year in females and 1.8%–4.1% in males).

Figure 4.

Rates of urinary tract infection by sex and race/ethnicity at Kaiser Permanente Southern California, 2008–2017. Abbreviation: UTI, urinary tract infection.

DISCUSSION

In a large integrated healthcare organization in Southern California, we found increases in UTI rates from 2008 to 2017 in outpatient care settings by multiple definitions of UTI. In recent years, virtual care visits have been rapidly increasing across the region. In this setting, where culture orders are uncommon, we observed large increases in UTI rates, especially by the diagnosis code with antibiotic definition. This study adds to limited data on UTI rates by age group and race/ethnicity, reporting increases in nearly all age and race/ethnicity categories.

There are few previous recent studies on outpatient UTI rates. A study by Foxman and colleagues is frequently cited for UTI incidence; this random digital dialing survey of 2000 women in 1995 to 1996 found that 1 in 3 women reported a UTI in their lifetime and 10%–12% reported a UTI in the prior year [9, 23]. In a 2014 survey of >2400 randomly sampled women, Butler and colleagues found a similar result, that 11% of women reported a UTI in the prior year, most of whom had sought outpatient care for their symptoms [8]. Our study using EHRs found comparable rates by UTI diagnosis code with antibiotic among females of all ages (114.0 per 1000 females, or 11.4%), although rates among older females in our study were significantly higher (up to 23%).

Several other studies have examined trends in UTI rates, with differing results. One study in the United Kingdom used an electronic database of routinely collected primary care data among adults ≥65 years of age and found that UTI diagnoses increased from 2004 to 2014 by approximately 1% to 9% per year [12]. However, another United Kingdom study using a different database of primary care diagnosis codes observed a decrease in UTIs from 2008 to 2015 but noted that hospitalizations for UTIs were steadily increasing in England [24]. In the US, increases in hospitalization for UTIs have also been observed [25], but no recent data on trends in outpatient UTIs exist.

There are several possible explanations for increasing rates of outpatient UTI. First, increases may be due to changing patterns of healthcare. Virtual care, which is increasingly used in many health systems across the US, is cost-saving and convenient, and may be of great benefit to persons unable to travel to a medical office or ED. We observed large increases in UTI rates in the virtual care setting (20% or more per year by diagnosis code with antibiotic), which was not completely offset by a decline in UTI rates in office visits. The large increase in UTI rates in virtual care may be driven by greater healthcare access provided by virtual care options. Increases may also be due to changes in clinically associated bacterial strains, such as the emergence of multidrug-resistant E. coli strains [26, 27].

Our data suggest that empiric antibiotic treatment, which is recommended for low-risk women with suspected cystitis [16, 17], may contribute to overtreatment. Among UTIs identified by diagnosis code with antibiotic across care settings, approximately 1 in 4 had a positive culture (<2 in 3 of those with a urine culture order). Some studies have shown that antibiotics are more likely to be inappropriately prescribed at virtual visits [15, 28, 29], as virtual care providers may order antibiotics as a “conservative” approach. We found that in virtual care settings, <1 in 10 UTIs by diagnosis code with antibiotic were confirmed with a positive culture (<1 in 3 of those with a urine culture order). Although empiric treatment in a virtual setting may be appropriate for some low-risk patients, the consequent low culture rate may be of concern, leading to overdiagnosis of UTI due to presumed diagnoses. Moreover, virtual visits, if not done appropriately, could potentially create a care gap if alternative or codiagnoses, such as benign prostatic hyperplasia, prostatitis, or interstitial cystitis were not pursued. Future work should explore strategies for optimizing UTI diagnosis and treatment in different outpatient care settings, including consideration of urine culture indications and costs/benefits [30]. Prediction models for identifying UTI [31], clinical pathways to standardize management of UTI [32, 33], and follow-up of patients with negative cultures to promote antibiotic discontinuation are important next steps [34].

There were several potential limitations to this study. The KPSC laboratory thresholds used for the positive culture definition are lower than the common diagnostic threshold of ≥10⁵ CFU/mL; these lower thresholds are more sensitive for acute cystitis [35–37], but they may detect subclinical bacteriuria. Because the definitions did not include symptoms, which are poorly captured in structured data, definitions including positive culture may have also included asymptomatic bacteriuria [38]. This could have overestimated UTI rates for individuals with a urine culture. Furthermore, we were unable to ascertain if the increase in positive cultures was due to an increase in symptomatic UTI or to variations in urine culture practices. The definitions also did not include dipstick or microscopic urinalysis due to inconsistent practice by providers and capture in the EHR. Instead, we created parsimonious definitions using routinely collected EHR data that would be feasible for use across health systems. For definitions including antibiotics, we limited antibiotic orders to the same day as a UTI diagnosis code or culture order to avoid capturing antibiotics prescribed for other conditions in the surrounding days, an approach that could miss a small proportion of UTIs. Finally, some members with UTI may not have sought care for less severe symptoms, leading to underestimated rates using any of our definitions. However, UTI symptoms usually require care, and KPSC patients who seek care are motivated to stay within the health system.

In conclusion, this study found that UTI rates in outpatient care settings increased from 2008 to 2017, particularly in virtual care and among older individuals. Virtual healthcare is important for expanding access to UTI care, but strategies are needed in all outpatient care settings to improve UTI diagnosis and reduce overtreatment.

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

Acknowledgments. The authors acknowledge the contributions of Harpreet Takhar, Peggy Hung, Meghan Davis, Chris Heaney, Cindy Liu, Keeve Nachman, and Lance Price.

Financial support. This study was supported by Kaiser Permanente Southern California internal research funds and by the National Institute of Allergy and Infectious Diseases, National Institutes of Health (grant number 1R01AI130066-01A1). A. P. has received grants from the National Center for Complementary and Integrative Health and the National Institute of Diabetes and Digestive and Kidney Diseases.

Potential conflicts of interest. The authors report no potential 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.