Relationship Between Clinical Factors and Duration of IV Antibiotic Treatment in Neonatal UTI

We performed a retrospective chart review at 2 academic pediatric hospitals located in the midwest and the southeast regions. This study was approved by the institutional review boards of both institutions. We included neonates aged ≤28 days who were hospitalized for a UTI from 2012 to 2018. These neonates had been previously discharged from a newborn nursery, NICU, or birthing center and were admitted from home. UTI was defined as presence of a discharge diagnosis of UTI by International Classification of Diseases, Ninth Revision and International Classification of Diseases, 10th Revision codes and a positive urine culture result. In collaboration with the Research Information Solutions and Innovation team at the primary site, we chose 12 International Classification of Diseases, Ninth Revision and International Classification of Diseases, 10th Revision codes (N39, 599, 771.82, P39.3, 041.49, B96.20, 041.9, A49.9, 590.80, N12, 595.0, N30.00). Urine culture results were considered positive if a catheterized urine specimen yielded a potential uropathogen in quantities of either (1) 10 000 to 49 000 colony-forming units per mL with pyuria or bacteriuria on urinalysis or (2) ≥50 000 colony-forming units per mL regardless of urinalysis results.^8,9  Neonates with a gestational age <37 weeks, preexisting urinary tract anomalies, bacteremia, meningitis, other serious bacterial infection, or PICU admission were excluded.

Clinical, laboratory, and imaging data were collected from manual chart review by trained research assistants. Clinical variables included neonates’ demographics, birth weight, gestational age, preexisting medical conditions, and presenting symptoms. Laboratory variables included complete blood count, blood glucose, metabolic profile, urinalysis, urine culture, blood culture, cerebrospinal fluid (CSF) studies, C-reactive protein, and procalcitonin. Imaging data included the results and timing (inpatient versus outpatient) of renal ultrasound and voiding cystourethrogram (VCUG). Other covariates analyzed included subspecialty consultation, time to uropathogen identification, and susceptibility results. Additional indications for hospitalization unrelated to UTI that could potentially extend the length of stay were tracked, such as respiratory viral illness requiring oxygen, hypoglycemia, or dehydration requiring IV fluids.

We evaluated antibiotic management and neonatal outcomes, including IV antibiotic therapy duration, total antibiotic therapy duration, and length of stay. IV antibiotic duration ≤48 hours was considered short duration, and >48 hours was considered long duration. Forty-eight hours was chosen because it closely matched the median duration of IV antibiotic therapy (49 hours) in our study group and allows for the identification and susceptibility testing of organisms isolated from urine culture.¹⁰  The duration was calculated from the start time of the initial antibiotic dose to the start time of the final antibiotic dose to account for the fact that some antibiotics require longer administration times. We tracked return visits to the emergency department and hospital readmissions within 30 days of discharge, and we determined if the returns and/or readmissions were related to UTI treatment failure. Treatment failure was defined as a repeat culture with growth of the same organism meeting the UTI criteria outlined above at the repeat visit.

Baseline characteristics of the neonates are presented as counts and percentages for categorical variables and as either means and SDs or medians and interquartile range (IQR) for continuous variables. Differences in median duration of IV antibiotic therapy between 2 sites were compared by using the Wilcoxon 2-sample test. Pearson’s correlation was used to measure correlation between length of stay and IV antibiotics duration. Logistic regression was used to identify associations between clinical and demographic variables and the outcomes of IV duration and treatment failure. The effect of each predictor was estimated in a model adjusted for site. A P value <.05 was considered statistically significant. All statistical analyses were performed by using SAS version 9.4 (SAS Institute, Inc, Cary, NC).

Of 310 neonates identified by using diagnostic codes and chart review, 112 met all criteria for inclusion. We excluded 86 neonates who did not meet our criteria for UTI, 81 neonates for prematurity or congenital diseases, 26 neonates for meningitis and/or bacteremia, and 5 additional neonates for preexisting genitourinary disease. The mean age at time of admission was 15.7 days (SD 6.6); 68 (61%) were boys, and 71 (63%) were white. The admitting service was largely determined by institution because neonates at one institution were primarily admitted to an inpatient infectious diseases service and the other to a hospitalist service. Most neonates were admitted for a diagnosis of fever (73%), and most neonates did not have concurrent medical complexity (82%; Table 1).

Evaluation typically included a complete blood count (n = 110; 98%) and a basic or complete metabolic panel (n = 89; 79%). Almost all neonates had a blood culture performed (n = 109; 97%) and lumbar puncture attempted (n = 109; 97%). Less than a quarter of neonates had inflammatory markers collected (procalcitonin n = 14 [13%] and C-reactive protein n = 21 [19%]), all from the same institution. Urine cultures grew Escherichia coli (n = 74; 66%), Enterococcus species (n = 11; 9%), Klebsiella pneumoniae (n = 7; 6%), other enteric Gram-negative bacilli (n = 6; 5%), group B Streptococcus (n = 4; 4%), Staphylococcus aureus (n = 4; 4%), coagulase-negative Staphylococcus (n = 3; 3%), and other organisms (n = 3; 3%). All isolates were susceptible to ≥1 available oral antibiotic.

All but 1 neonate had a renal ultrasound obtained (n = 111; 99%); however, there were institutional variations on whether this was done inpatient or after discharge. All renal ultrasounds were completed during hospitalization at 1 institution, compared to only 85% at the other institution. Approximately 53% (n = 59) of renal ultrasounds were abnormal. VCUGs were performed in 21 (19%) neonates during hospitalization and 44 (39%) neonates after discharge. We were unable to confirm for several neonates whether VCUG was obtained after discharge because they were not documented within the electronic health record.

The median duration of IV antibiotic therapy was 49 hours (IQR, 40–72 hours), and the median duration of total antibiotic therapy (IV and oral) was 10 days (IQR, 10–14 days). Fifty-seven (51%) neonates received IV antibiotics for >48 hours. Length of stay and duration of IV antibiotic therapy were found to be strongly correlated (r = 0.89; P < .001). The median IV antibiotic duration differed between the 2 hospitals (56.5 [IQR, 48–102] vs 46.5 [IQR, 37–60] hours; P < .05). Thus, all other analyses were adjusted for hospital site. After this adjustment, no demographic features, presenting laboratory values, or specific pathogenic organisms increased the odds of a long IV antibiotic duration (Table 2), apart from age >7 days (adjusted odds ratio [aOR] = 0.29; 95% confidence interval [CI], 0.09–0.97) as compared with neonates ≤7 days. However, the odds of a long IV duration increased if the neonate had additional indication for hospitalization (aOR = 3.2; 95% CI, 1.2–8.7), subspecialty consult (aOR = 4.79; 95% CI, 1.87–12.30), or an abnormal renal ultrasound (aOR = 2.26; 95% CI, 1.01–5.08) either during admission or after discharge. The most common subspecialty consultants included nephrology and urology.

In 16 (14%) neonates, clinicians transitioned from IV to oral antibiotics without knowing final antibiotic susceptibilities of the uropathogen. One neonate experienced treatment failure with UTI caused by K pneumoniae (Table 3). This neonate was treated with IV antibiotics for 48 hours and was discharged after 64 hours to complete a 14-day total course of antibiotics. The renal ultrasound was normal, and VCUG was not performed initially. There was no concomitant diagnosis or secondary illness noted during hospitalization. The neonate returned to the hospital 26 days after discharge with a fever, and the urine culture again yielded K pneumoniae. Both the initial and recurrent organism were susceptible to the discharge antibiotic. The neonate’s second hospital stay was unremarkable, and a VCUG performed after his second UTI was ultimately normal.

In this study, we found that, at 2 academic institutions, many neonates aged ≤28 days with a UTI were treated with ≤48 hours of IV antibiotics. Despite differences in diagnostic evaluation between institutions, few factors were associated with duration of IV antibiotic treatment. Even when controlling for site variation, providers who are choosing longer courses do not seem to be doing so on the basis of any meaningful clinical signifiers other than age <7 days, newly discovered genitourinary anomalies, consultations, and additional indications for hospitalization. Additionally, although we were underpowered to reliably detect a difference in treatment failures (given that this was an infrequent event), the incidence was similar to national norms.^3,5 

Although there were no statistically significant findings when comparing duration of longer IV therapy for neonates by individual weeks of life, when isolating neonates <7 days of age compared with their older cohorts 8 to 28 days of age, we found that this youngest group was more likely to receive longer IV therapy. Despite the variation in care of neonates presenting for febrile illness <90 days, there continues to be a trend toward more in depth and comprehensive evaluations and treatments in the youngest subset of this group.¹¹  It is possible that this group is considered more vulnerable because of their relative proximity to delivery and thereby higher susceptibility to perinatal infections. However, most authors examining febrile illness often group neonates by 0 to 30 or 0 to 28 days and do not examine age on such a granular level.^12,13  Further work remains to be done to determine if this trend was unique to our study population or may be more broadly applicable.

The increased odds of longer IV antibiotic duration associated with obtaining a renal ultrasound before hospital discharge is also both statistically and clinically relevant. The current American Academy of Pediatrics guidelines suggest that a renal ultrasound should be obtained after the first febrile UTI, noting that ultrasounds can be misleading early in the acute phase of infection.^14,15  However, the guidelines do not address this specific population, which may explain why for some neonates there was a variation in the timing of ultrasounds in our study group. An abnormal renal ultrasound also seemed to influence clinicians’ decision to treat with longer courses of IV antibiotics. The longer duration of treatment may be related to additional workup that ensues including consulting subspecialists, scheduling a VCUG, and potentially obtaining the VCUG while hospitalized. Providers may also be concerned about associations between recurrent UTI and renal scarring and therefore may be apt to keep neonates on IV treatment longer.¹⁶  Additionally, those with abnormal ultrasounds may also be at risk for increased illness severity, which may lead to a longer length of stay.

The presence of an additional indication for hospitalization may increase the length of stay and therefore possibly the duration of IV antibiotics. Many of these neonates required additional evaluation and supplemental care outside of UTI management, including oxygen and IV fluids. Clinicians may be more apt to continue IV antibiotics while the neonate is hospitalized, even if unnecessary, given the ease of administration. The use of IV antibiotics may also be for practical purposes if a neonate has concomitant dehydration, risk for aspiration, or respiratory distress that precludes oral intake. Other reasons for prolonged hospitalization were often also infectious in nature, such as suspicion for meningitis, given either a lack of CSF or initial growth on a culture that was ultimately determined to be a contaminant. Providers may have been more likely to continue courses of IV antimicrobial agents until this uncertainty was resolved. This population of neonates with potential meningitis or bacteremia should be analyzed separately in subsequent studies in which the association with longer IV antibiotic therapy could be explored further.

Finally, we noted that the presence of subspecialty consultation increased the odds of IV antibiotic duration. However, this association is potentially confounded by the fact that consultants are typically involved with the most complex patients who may already have an increased length of stay. Additionally, neonates who are persistently febrile and have ongoing emesis and other complications of UTI may be more likely to trigger a consultation but also may be more likely to remain on IV antibiotics for longer given an inability to tolerate oral therapy. More work remains to be done to determine if these practice variations are changing with recent research on neonatal UTI or whether they are unique to geographic locations or single institutions.

Our study has several limitations. First, analysis of neonates at only 2 hospitals may lack generalizability to other hospitals. In fact, the practice pattern differences between the 2 hospitals suggests that UTI management varies among institutions, a finding that is consistent with previous studies and reveals a need for better national consensus on the management and treatment of neonatal UTIs. At some institutions, these neonates may be primarily cared for in NICUs, a population that was not included in our study because of practice patterns. Future work should be dedicated toward reviews of larger populations in geographically distinct areas because there may be additional factors that influence treatment duration at other institutions.

Second, although we collected data on a variety of demographic, clinical, laboratory, and imaging variables, there may be others that were not included that could impact IV treatment duration, including response to treatment, resolution of fever, vomiting, or other clinical symptoms. Third, there continues to be controversy surrounding the diagnostic criteria for UTIs, particularly in this age group, and therefore some cases included in our study may have represented contamination or asymptomatic bacteriuria. Finally, our study was not designed to detect a difference in failure rates between short and long IV antibiotic durations. Although clinical failure was rare in our study regardless of IV treatment duration, we cannot draw definitive conclusions about the safety and efficacy of shorter IV antibiotic courses for neonatal UTI, and larger studies are needed.

Despite institutional differences in the evaluation and management of UTI, physicians are commonly prescribing 48-hour courses of IV antibiotic therapy with minimal initial clinical factors driving the decision for extended courses of treatment. Shorter courses of IV antibiotics not only reduce the risk of potential iatrogenic complications but also reduce the tremendous cost to the health care system associated with inpatient treatment¹⁷  and allow for the return of the intact mother-infant dyad.¹⁸  Future work, therefore, should be focused on the development of larger studies that evaluate rates of treatment failure in this population to provide evidence of the safety and efficacy of earlier transitions to oral therapy.

The relationship between longer antibiotic duration and imaging results, concomitant illness, and consultation of a subspecialist should be explored further because these factors were significantly associated with longer antibiotic duration. The percentage of abnormal renal ultrasound and VCUG results does seem to suggest that there was a significant proportion of preexisting renal anomalies not captured on prenatal ultrasound because we excluded those with known abnormalities. Children with discovered genitourinary anomalies may warrant special consideration and longer hospitalization or expedited outpatient evaluation. Future work should seek to determine if these factors merit longer duration of IV therapy, in addition to other potentially important clinical factors, including PICU admissions, prematurity, and concomitant bacteremia.

We thank The Ohio State University Center for Clinical and Translational Science for their collaboration.