Incidence and Management of Urinary Tract Infection in a National Medicaid Population, 2009-2019 Open Access

Urinary tract infection (UTI) is common in pediatric patients and is a leading cause of bacterial infections in infants.¹ A diagnosis of UTI drives numerous medical interventions beyond the initial antibiotic prescription, including radiographic imaging and, in some cases, prophylactic antibiotic use and a urology referral. The National Institute for Health and Care Excellence (NICE) in the United Kingdom released revised UTI guidelines in 2007 advocating for more stringent imaging, recommending ultrasonography and voiding cystourethrogram (VCUG) only for infants younger than 6 months with atypical or recurrent UTIs.² Additionally, the 2011 American Academy of Pediatrics (AAP) UTI guidelines for infants and young children aged 2–24 months (currently retired owing to concerns around the use of race-based recommendations) provided 2 noteworthy recommendations that were departures from the 1999 AAP UTI guidelines.³ One recommendation advocated for stricter diagnostic criteria for UTI, requiring a positive urinalysis result (ie, pyuria) in addition to a positive urine culture result. However, a recent meta-analysis has called for reconsideration of this requirement of pyuria for diagnosis of UTI, arguing that asymptomatic bacteriuria is a very rare cause of positive urine culture results (and therefore that urine cultures alone are an adequate gold standard).⁴ 

Additionally, although the 1999 guidelines recommended routine VCUG for all first-time UTIs (hereafter referred to as index UTIs [iUTIs]), the 2011 guidelines instead suggested obtaining a VCUG only if the renal and bladder ultrasonography (RBUS) result is abnormal.^3,5 However, this recommendation has been controversial. In response to the 2011 guidelines, the AAP Section on Urology stated that “the recommendation is based on a flawed interpretation of limited data and that this stands to potentially harm significant numbers of children because of delayed diagnosis of harmful urinary tract conditions.”⁶ This group expressed concern that missing vesicoureteral reflux (VUR) would lead to a reduction in the use of continuous antibiotic prophylaxis (CAP), one of the most common interventions that follows a VUR diagnosis. Although several important studies have demonstrated the effectiveness of CAP in preventing recurrent UTIs, the overall risk-to-benefit ratio remains uncertain as the effect size is small, the risk of subsequent resistant UTIs has been shown to increase with CAP use, and there has not been a documented benefit in the prevention of renal scarring.^7–9 

Given this ongoing controversy over the diagnosis and management of one of the most common bacterial infections in young children, a better understanding of the impact of these more restrictive recommendations on UTI incidence, management, and outcomes can help to inform current practice and future guidelines. The aims of this study were to assess recent trends in (1) UTI diagnosis in children younger than age 2 years with Medicaid; (2) downstream interventions following an iUTI, including RBUS and VCUG, CAP use, and subsequent urology referral; and (3) UTI recurrence within 24 months of an iUTI.

This study used the Medicaid MarketScan Database (Merative, Ann Arbor, MI), which contains de-identified data for all inpatient and outpatient managed care and fee-for-service health care encounters in 10–12 states (depending on the year).¹⁰ We reviewed inpatient and outpatient diagnosis codes as well as procedure and retail pharmaceutical claims. This study did not require institutional review board approval as we used pre-existing, de-identified data for analysis. We examined enrollees younger than age 2 years between January 1, 2009, and December 31, 2019. Age is only recorded in years (ie, “0 years” or “1 year”) in the Medicaid database.

We excluded patients with (1) complex chronic conditions with a corresponding International Classification of Diseases, Ninth or Tenth Revision (ICD-9/10) code at time of or before iUTI diagnosis,¹¹ or (2) with a genitourinary abnormality diagnosed either prenatally or before iUTI as defined by predetermined ICD-9/10 codes (Supplemental Table 1). We opted to follow patients for as long as they were continuously enrolled in the database, up to 24 months following an iUTI.

We defined iUTI using ICD-9/10 codes corresponding to UTI, cystitis, pyelonephritis, and renal/perirenal abscess (Supplemental Table 2) and evidence of a concomitant UTI-related antibiotic prescription (Supplemental Table 3) filled within 0–5 days of diagnosis.¹² We included both criteria to increase the likelihood of a true UTI as claims data do not provide urinalysis and urine culture results and ICD codes alone have suboptimal predictive value.¹³ For this study, we defined “recurrent” UTI as a second UTI (using same criteria as iUTI) between 30 days and 24 months of the iUTI diagnosis.⁷ 

VCUG completion after iUTI was defined as a VCUG obtained within 3 months of the iUTI. RBUS completion after iUTI was defined as ultrasonography within 60 days of the iUTI. Because of a lack of literature defining appropriate timeframes at which these studies should be obtained, these time periods were decided based on the clinical consensus of the authorship team. VCUGs and RBUSs were identified by reviewing claims for Current Procedural Terminology codes (Supplemental Table 4).¹⁴ VUR was identified using corresponding ICD-9/10 codes after UTI diagnosis (Supplemental Table 5). We identified CAP from pharmaceutical claims for antibiotics commonly used in UTI prophylaxis (Supplemental Table 3) that were prescribed for longer than 21 days within 12 months of an iUTI.^7,8,15 We elected to use a longer exposure window for CAP given that completion of all imaging and referral to urology can be a prolonged process, especially when care is fragmented. We chose to use a prescription duration of 21 days to differentiate patients who may have required prolonged antibiotic courses for persistent UTI from those who received antibiotics for UTI prophylaxis. Urology clinic visits were identified using the MarketScan Database dictionary to identify specific visit types.

We described UTI diagnoses as rates per 1000 patient months. We report subsequent use and diagnoses (imaging, VUR diagnosis, CAP, and urology visits) as percentages of patients with an iUTI. Changes in rates and proportions over time were assessed using Poisson and logistic regression models, respectively, using the year as a fixed effect. We reported outcomes as incidence rate ratios (IRRs) for rates and odds ratios (ORs) for binary outcomes. We opted not to conduct an interrupted time series analysis using the 2011 AAP guidelines as a focal point because of existing evidence that imaging patterns were changing even before the guideline release.^16,17 

From 2009 to 2019, a total of 4 296 966 children younger than age 2 years were enrolled in the MarketScan Database for some period of time, contributing 58 456 287 patient months of data. Nearly two-thirds (2 811 722; 65.4%) were enrolled for at least 24 months. Enrollment ranged from a low of 248 390 in 2011 to a high of 606 256 in 2009, with most children being younger than age 1 year (79%) and male (51.2%) (Table 1). Reported race and ethnicity data, determined by demographic information included in the claims database, showed the highest proportion of patients were non-Hispanic white (41.1%) (Table 1).

A total of 52 268 children were diagnosed with an iUTI during the study period, resulting in a rate of 0.89 UTIs per 1000 patient months (95% CI, 0.89–0.90). iUTI diagnoses decreased over time (IRR, 0.93; 95% CI, 0.92–0.93), most notably after 2013 (Figure 1). An RBUS was obtained in 19 499 (37.3%) children with an iUTI. RBUS imaging did not change significantly over time (OR, 1.0 per year; 95% CI, 0.99–1.01) (Figure 2). After an iUTI, 10 146 (19.4%) children received a VCUG. VCUG usage decreased over time from approximately 30% in 2009 to approximately 11% by 2019 (OR, 0.87 per year; 95% CI, 0.86–0.88) (Figure 2). A total of 4054 patients (7.8%) received a diagnosis of VUR, and this proportion decreased over time (OR, 0.92 per year; 95% CI, 0.91–0.93) (Figure 3). CAP was prescribed within 12 months of an iUTI in 10.1% of patients and decreased during the study period (OR, 0.95 per year; 95% CI, 0.94–0.96) (Figure 3). Urology clinic visits occurred in 3130 (6%) patients and increased on average over time (OR, 1.06; 95% CI, 1.05–1.08) (Figure 3). However, much of this increase was driven by a large spike in 2012. Urology visits steadily decreased after that.

At least 1 recurrent UTI occurred in 6783 children, resulting in a rate of 8.24 recurrent UTIs per 1000 patient months. Recurrent UTI increased slightly without statistical significance (IRR, 1.01; 95% CI, 1.00–1.02) in the population of patients with an iUTI. To account for the decrease observed in iUTIs during the study period, we also analyzed recurrent UTIs as a function of the entire population. The overall population rate of a recurrent UTI decreased (IRR, 0.95; 95% CI, 0.94–0.95) (Figure 4).

In this retrospective study of more than 4 million children younger than age 2 years enrolled in a national Medicaid claims database, we report notable decreases in UTI diagnoses from 1.25 per 1000 patient months in 2009 to 0.58 per 1000 patient months in 2019. Additionally, we observed decreases in related downstream interventions including VCUG imaging and prophylactic antibiotics. These decreases did not appear to come at the expense of an overall increase in rates of recurrent UTI. The exact mechanisms behind these trends are not clear. The decrease in incident UTIs may have been driven by prior investigations and related recommendations for stricter definitions for UTI diagnosis, whereas recommendations for a more selective approach to imaging may have contributed to the decrease in VCUGs.³ If indeed the use of costly and potentially harmful tests and medications has decreased without negatively impacting outcomes, then these trends are suggestive of substantial improvements in health care value around the diagnosis and management of UTI.

Updated population estimates on the diagnosis and treatment of common diseases such as UTI are useful to help target and prioritize educational, research, and quality improvement initiatives. Although prior literature has provided insight into the prevalence of UTIs in specific pediatric groups (eg, febrile infants and uncircumcised infants), there are minimal data on the overall population incidence of UTIs.¹ The incidence reported from our investigation (0.89 iUTIs per 1000 patient months) is similar to the incidence described by Liang et al in commercially insured patients younger than age 1 year (0.86 per 100 patient years or 0.72 per 1000 patient months).¹² Although Liang et al did not analyze data before 2016, the authors did note a gradual decrease in UTI incidence from 2016 onwards, similar to the findings from our study.¹² These estimates on UTI incidence might reasonably be questioned given the reliance on claims data rather than microbiologic data. However, they are likely to reflect actual initial treatment patterns. Although some patients who are initially diagnosed and treated as having a UTI may wind up having negative or inconclusive urine culture results, there is considerable evidence that many such patients do not have their antibiotic courses truncated.^18–21 Nonetheless, this bias suggests that the UTI rates calculated from claims data likely exceed the true incidence of UTI.

One potentially concerning finding from our investigation was the low proportion (35.3%-39.7%) of patients who underwent an RBUS within 2 months of iUTI. In both the 1999 and 2011 AAP UTI guidelines, RBUS is recommended for all children aged 2 months to 2 years with a first febrile UTI.^3,5 Although many evaluations for UTI in children are triggered by fever, we are unable to discern whether patients in our sample were febrile, and some UTIs may have been diagnosed in children with symptoms aside from fever. Additionally, as stated above, some children with suspected initial UTI may wind up having negative urine culture results and as such would not warrant an RBUS. Our findings around lower than expected RBUS obtainment warrant more evaluation, especially given the potential vulnerability and lack of a medical home for many children with public insurance.²² 

We observed a decrease in VCUG rates from 2009 to 2019, which is consistent with prior studies and is hypothesized to be occurring for a variety of reasons.^16,23–28 Although the 2011 AAP UTI guidelines recommendation for more selective VCUGs was likely an important factor, the trend appears to have preceded these recommendations. A study by Courchia et al showed a trend toward decreased UTI hospitalizations, especially after 2009, along with decreased hospital length of stay for UTI admissions, providing less opportunity to obtain VCUGs during hospitalization.²⁸ McDonald et al reported that scheduling VCUGs more than 1 week after UTI diagnosis was more likely to result in a failure to perform the procedure.²⁹ Additionally, NICE released revised UTI guidelines in 2007 recommending imaging with ultrasonography and VCUG only for infants younger than age 6 months with either recurrent or atypical UTIs,² an approach that was adopted and studied in at least 1 US center before the 2011 AAP guidelines.¹⁷ 

We also noted a decrease in VUR diagnoses, a finding that was likely driven simply by the observed reduction in VCUGs. This finding was also reported by the US center that adapted a more selective approach based on the NICE UTI guidelines,¹⁷ where the reduction occurred exclusively in low-grade VUR. However, 1 single-center study noted a higher rate of renal scarring in 2015 vs 2005 in children referred for a VCUG, which the authors suggest may be a result of delayed diagnosis of VUR and increased risk of recurrent UTI. Although these findings were limited by the fact that dimercaptosuccinic acid scans (to detect scarring) were only performed in 6%-7% of this population and that the overall number of UTIs in the referral catchment area was unknown, the study highlights the importance of continuing to track renal outcomes at the population level.³⁰ 

The decrease in VUR diagnoses mirrored the observed decrease in the provision of CAP, which is not surprising given that VUR is one of the most common drivers of CAP. CAP use for patients with VUR remains a widely debated topic. The American Urological Association recommends CAP in children younger than age 1 year who have febrile UTI and VUR but recommends a more selective approach in older children.³¹ The European Association of Urology and multiple other European groups advise a selective approach based on age, severity of VUR, and renal scarring.^2,32 Both the RIVUR and PRIVENT trials demonstrated modest decreases in recurrent UTI with CAP but no impact on renal scarring.^7,8 Other studies raised a concern that CAP in children with VUR increased the amount of non–Escherichiacoli isolates and multidrug-resistant organisms.^15,33 In our cohort, VUR diagnoses and CAP prescriptions decreased without a significant increase in recurrent UTI rates in patients with an iUTI but a reduction in overall recurrent UTIs at the population level. These findings provide some reassurance that decreased use of CAP may not be associated with increased harm.

We noted an increase in urology referral rates from 2011 to 2012, with a steady decline thereafter but with recent rates still well above pre-2011 rates. Although the exact cause of the 2012 spike is unclear, 1 potential explanation could be that the 2011 AAP guidelines led some front-line providers to refer to urology after an abnormal RBUS result rather than ordering a VCUG themselves, or possibly even referring to urology routinely after a first febrile UTI, a practice that was described to occur by 28% of providers in a survey study published in 2023.³⁴ 

This study has several limitations. Most importantly, we relied on insurance claims for our definition of UTI and for the other outcome measures. This approach could lead to an overestimate or underestimate of true UTI incidence as errors in coding and lack of microbiologic data could skew the results. Year-to-year variation in statewide participation in the database may have biased results if there is geographic variation in UTI diagnosis and management. We were also limited in identifying CAP based on prescription duration, which may have yielded antibiotic prescriptions used for other conditions. Additionally, findings from a publicly insured population may not be generalizable to patients with other forms of insurance. There is also the potential for bias given that the Affordable Care Act was released in 2010, which resulted in increased public coverage for Hispanic and Black patients compared with white patients. This shift may have influenced the demographics of our study population, which in turn could have influenced health care access patterns and thus impacted some of our observed trends.³⁵ Finally, we are unable to definitively assess the impact of these trends on renal scarring and kidney function. Future investigations analyzing overall trends in kidney dysfunction in children and young adults with a history of UTI are greatly needed to ensure that current practices are, in fact, safe.

In this retrospective study of children enrolled in a Medicaid claims database from 2009 to 2019, we observed decreasing trends in UTI diagnoses and downstream interventions in children younger than age 2 years without any apparent negative impact on recurrent UTIs. These findings reflect improving health care value for children with febrile UTI. However, further research is needed to confirm our findings, better understand the observed low rates of RBUS, and assess renal outcomes.

AAP

American Academy of Pediatrics

CAP

continuous antibiotic prophylaxis

ICD-9/10

International Classification of Diseases, Ninthor Tenth Revision

IRR

incidence rate ratio

iUTI

index urinary tract infection

NICE

National Institute for Health and Care Excellence

OR

odds ratio

RBUS

renal and bladder ultrasonography

UTI

urinary tract infection

VCUG

voiding cystourethrogram

VUR

vesicoureteral reflux