Introduction

With an estimated prevalence of 0.4–1.8% in the general pediatric population and 30% among those with a history of febrile urinary tract infection (UTI), vesicoureteral reflux (VUR) is a common urological diagnosis affecting children [1, 2]. VUR management is individualized based on a given patient’s likelihood of spontaneous resolution and UTI, underlying renal pathology and parental preferences. In many cases, VUR spontaneously improves or resolves over time; however, observation may not be an optimal option for those at risk for breakthrough febrile UTIs or renal scarring.

The initial management for the majority of children with VUR is nonsurgical with continuous antibiotic prophylaxis or observation until spontaneous resolution. The decision to undergo surgical intervention in a child with VUR is impacted by assessing multiple factors that influence spontaneous VUR resolution, including the initial grade of reflux, bladder volume at the onset of VUR, age, gender, anatomical abnormalities, and bladder and bowel dysfunction [3, 4]. Of the aforementioned variables, bladder volume at the time of VUR is one of the most important prognostic indicators of spontaneous resolution as VUR occurring at earlier filling volumes is associated with a lower likelihood of spontaneous resolution as well as an increased likelihood of breakthrough febrile UTI [5, 6].

Voiding cystourethrogram (VCUG) is the gold standard for diagnosing and evaluating the grade and severity of VUR. However, a recent survey revealed significant variability in VCUG technique and protocol among 65 children’s hospitals in the United States and Canada [7]. This inconsistency raised concerns about patient safety as VCUG is an invasive test and its interpretation influences the management of VUR. In 2016, the American Academy of Pediatrics (AAP) developed a recommended protocol for VCUG with the purpose of standardizing VCUG technique, including reporting the volume of contrast infused and the bladder volume at which the onset of reflux occurred [8]. Concurrently, the 2019 American College of Radiology (ACR) VCUG practice parameter states that bladder volume at the onset of reflux is only recommended, not required [9]. Despite the AAP recommended protocol, there is still variation in practice. If bladder volumes are not recorded during the study, the urologist often estimates VUR timing by reviewing the fluoroscopy images. Our study aimed to determine whether pediatric radiologists and pediatric urologists can accurately estimate the timing of reflux by examining VCUG images without prior knowledge of the volume of contrast instilled.

Materials and methods

Institutional review board approval was obtained for the study. Patients diagnosed with vesicoureteral reflux were retrospectively identified at an institution where the volume of contrast instilled is routinely recorded. Studies were selected from those performed between February 2006 and May 2013. Patients were included if they were younger than 18 years old at the time of the study, underwent a VCUG with recorded bladder volumes (both total and at VUR onset), and were diagnosed with primary VUR. Patients were excluded if they had a diagnosis of secondary VUR (neurogenic bladder, posterior or anterior urethral valves, ureterocele, urethral stricture, bladder bowel dysfunction). A total of 39 patients were selected to satisfy calculated power requirements.

Typical protocol for a VCUG involves a scout image, followed by images during estimated early, mid and late bladder filling. Once reflux is identified, the ureter and renal collecting system are imaged. A voiding image is captured, as well as post void image. Fluoroscopic screen capture is used for imaging. Bladder volumes were measured using the amount of contrast material instilled, as determined by volume markers on the contrast bottle label. Maximum bladder capacity was considered as the instilled contrast volume at which spontaneous voiding occurred. Cyclical VCUG studies were performed for all selected patients. Total bladder volume and volume at the time of reflux, age at VUR diagnosis, gender and VUR index were collected from the VCUG reports. The patients were sorted into three groups of volume-based VUR timing: early-/mid-filling reflux (VUR onset <75% bladder filling), late-filling reflux (VUR onset at 75–100% bladder filling), and voiding-only reflux (VUR onset only during micturition).

Three fellowship-trained pediatric urologists and two pediatric radiologists were shown all available fluoroscopic images from each of the 39 included patients. The pediatric urologist reviewers included a fellow (M.L.G.-R.) and 2 practicing pediatric urologists with 1 (A.J.K.) and 18 (M.A.L.) years of post-fellowship experience at the time of study. The pediatric radiologist reviewers had 17 (J.D.G.-S.) and 25 (J.P.W.) years of post-fellowship experience at the time of study. All available images from each study were compiled in slide show format and presented to the reviewers in sequential order. An example of VCUG images shown to each reviewer is provided in Fig. 1. Reviewers had not participated in the care of these patients and were blinded to the associated radiology report, recorded bladder volumes and each other’s answers. To account for bilateral VUR and cyclic studies, the onset of initial VUR in the image series presented was tested whether unilateral or bilateral, or on first or second cycle of VCUG. Reviewer answers were compared against the timing of onset of VUR as reported by the radiologist (% filling=volume at VUR/total bladder filling at VCUG).

Fig. 1
figure 1

Example of a voiding cystourethrogram (VCUG) report shown to reviewers. Reviewers were asked to estimate volume at the onset of vesicoureteral reflux (VUR) and sort patients into three groups: early-/mid-filling reflux (VUR onset <75% bladder filling), late filling (75–100%) and voiding only (during micturition). Sections circled in red represent VUR during VCUG. Red circles indicate onset of VUR during the study

Full size image

A weighted kappa statistic with associated 95% confidence interval (CI) was calculated and used to assess rater agreement with the gold standard volume-based interpretation of VUR timing among the three pediatric urologists and two pediatric radiologists. The following ranges proposed by Landis and Koch [10] were used to interpret the degree of agreement between raters based on the value of the weighted kappa: (<0, poor), (0–<0.2, slight), (0.2–<0.4, fair), (0.4–<0.6, moderate), (0.6–<0.8, substantial), (0.8–1.0, almost perfect). Exact agreement was also tabulated. Analysis was conducted using SAS v. 9.4 (SAS Institute Inc., Cary, NC), and statistical significance was assessed at the 0.05 level.

Results

Patient demographics

A total of 39 patients were randomly selected from a historical cohort of children with VUR, with 13 patients in each VUR timing category (early-/mid-filling reflux, late-filling reflux, and voiding-only reflux) meeting inclusion criteria. Detailed demographics for the patients meeting inclusion criteria are outlined in Table 1.

Table 1 Demographic information for patients whose VCUG images were included in the survey (A), and number of studies included with early/mid, late or voiding VUR timing as defined by bladder volume during the study (B)
Full size table

Rater agreement

Answers provided by the three pediatric urologists and two radiologists were compared to the recorded volume-based onset of VUR. Details of rater agreement among pediatric urologists and pediatric radiologists are outlined in Table 2. Overall agreement among all five raters was moderate (k=0.43, 95% CI 0.36–0.50). Individual agreement between rater and the volume-based result was slight to moderate with kappa values ranging from 0.13 to 0.43. Interpretation among pediatric urologists (kappa scores ranging from 0.13 to 0.36) was less accurate than interpretation among pediatric radiologists (kappa scores ranging from 0.32 to 0.43) compared to the gold standard. However, both were fair at best. Interobserver variability was similar among pediatric radiologists and urologists (k=0.53 vs. 0.50). Pediatric radiologists and urologists did not consistently identify any of the VUR timing groups. The percentage of exact agreement among pediatric urologists ranged from 52.6% to 72.5% while the percentage of exact agreement among pediatric radiologists was 70%; the percentage of exact agreement with pediatric urologists and volume-based VUR timing ranged from 37.5% to 55% while the percentage of exact agreement with pediatric radiologists and volume-based VUR timing ranged from 47.5% to 57.5% (Table 2). As demonstrated in Table 3, the pediatric urologists’ accuracy, represented as percent agreement, was generally consistent across the volume-based VUR timing groups (between 41% and 53% agreement on average). However, the pediatric radiologists were more accurate when interpreting studies for the late-filling and voiding-only reflux groups compared to the early-/mid-filling reflux groups (56% and 69% agreement vs. 31%, respectively).

Table 2 Agreement among pediatric urologists and pediatric radiologists
Full size table
Table 3 Percent agreement by volume-based VUR timing groups
Full size table

Discussion

Our study reveals that pediatric radiologists and urologists are unable to retrospectively estimate VUR timing solely based on observing static fluoroscopic VCUG images. Accurate interpretation of VUR timing requires recording the volume of contrast instilled at the time of reflux and maximum bladder capacity by the radiologist or radiology technician during VCUG. Our findings support the recently published American Academy of Pediatrics (AAP) protocol recommending the routine recording of bladder volume at the onset of VUR as a standard component of all VCUGs to more accurately assess the likelihood of resolution and risk of recurrent UTI [8].

When discussing treatment options for VUR with families, it is important to have an accurate assessment of the patient’s condition as options range from continuous antibiotic prophylaxis to surgical intervention depending on the outcome and interpretation of the VCUG study. VCUGs are used to guide management and treatment of an individual child with VUR; therefore, our study supports the recommendation that the broader community of pediatric health care providers adopt into practice the current VCUG protocol established by the AAP. Standardizing an evidence-based protocol in medicine provides a key strategy for improving health care by reducing the variance in practice, minimizing patient risk, improving validity of the imaging results, and allowing outcomes to be accurately compared between individuals and institutions [11, 12].

This study is not without limitations. Our retrospective study of children with VUR identified by querying our electronic medical records carries inherent limitations in the ability to accurately identify all patients meeting the inclusion criteria. Additionally, our study is limited by a sample size of 39 patients. In addition, the number of raters was small (three urologists and two radiologists), which may not be an accurate representation of the variability among all subspecialists. However, the current study design is common when assessing the accuracy of a diagnostic test [13,14,15]. An additional limitation is that all raters were from the same institution. Further, it is unclear how these results apply to contrast-enhanced voiding urosonography as VUR timing was defined with standard fluoroscopic VCUG. This is a consideration for future study.

Conclusion

Our study supports the 2016 VCUG protocol established by the AAP. Pediatric radiologists and technicians should routinely record bladder volume at the onset of VUR and at maximum bladder capacity as a standard component of performing VCUG in order to provide a more accurate interpretation of the test outcome. Implementing the AAP’s standard protocol will improve patient care by assisting clinicians, patients, and families in making accurate and informed decisions regarding VUR management.