VARIATION IN THE LEVEL OF DETAIL IN PEDIATRIC VOIDING CYSTOURETHROGRAM REPORTS

SUMMARY

INTRODUCTION:

Voiding cystourethrography (VCUG) provides a wealth of data on urinary tract function and anatomy, but few standards exist for reporting VCUG findings.

OBJECTIVE:

We aimed to assess variability in VCUG reports and to test our hypothesis that VCUG reports from pediatric facilities and pediatric radiologists are more complete than those performed at other facilities or by non-pediatric radiologists.

STUDY DESIGN:

We analyzed original VCUG reports from children enrolled in the Randomized Intervention for Children with Vesicoureteral Reflux (RIVUR) trial. A 23-item checklist was created and used to evaluate reporting of technical, anatomic, and functional information. Radiologists were classified as pediatric or non-pediatric radiologists. Facilities were categorized as free-standing pediatric hospitals (FSPH), pediatric “hospitals within a hospital” (PHWH), non-pediatric hospitals (NPH), or outpatient radiology facilities (ORF). Multivariate linear regression was used to analyze factors associated with the completeness of the VCUG reports (percent of items reported from the 23-item checklist).

RESULTS:

602 VCUGs were performed at 90 institutions. 76% were read by a pediatric radiologist, and 49% were performed at a FSPH. On average, less than half of the 23 items in our standardized assessment tool were included in VCUG reports (M=48%, SD=12). The completeness of reports varied by facility type: 51% complete at FSPH (SD=11), 50% at PHWH (SD=10), 36% at NPH (SD=11), and 43% at ORF (SD=8) (P< 0.0001). In multivariate analysis, VCUG reports generated at NPH or ORF had 8% fewer items included (95% CI [3.0, 12.8], P< 0.01), and those generated at PHWH did not differ from those generated at FSPH. Reports read by a non-pediatric radiologist had 6% fewer items included (95% CI [3, 9.7]; P<0.01) compared with those read by a pediatric radiologist.

DISCUSSION:

There is substantial underreporting of findings in VCUG reports when assessing a widely-represented sample of routine, community generated reports using an idealized standard. Though VUR was often reported, other crucial anatomic and functional findings of the VCUG were consistently underreported across all facility types.

One limitation of this study is the relatively subjective assessment tool. Also, the study focused on the content of the VCUG report itself but not the actual technique of the procedure.

CONCLUSION:

Although pediatric radiologist and pediatric hospitals generated more complete VCUG reports compared to those from non-pediatric origins, the differences are small when considering the substantial underreporting of VCUG findings in general. This underscores the opportunities for improvement in reporting of VCUG findings.

Study Summary

Characteristic
No. of VCUG reports analyzed	602
Gender of participants
Male	49 (8.1%)
Female	553 (91.9%)
Median age in months at VCUG [IQR]	11 [5,30]
VCUG read by1
Pediatric radiologist	459 (76.2%)
Non-pediatric radiologist	134 (22.3%)
VCUG performed at2,3
FSPH	297 (49.3%)
PHWH	184 (30.6%)
NPF	87 (14.5%)
ORF	33 (5.5%)
Overall frequency of item being reported (8 items selected from 23 total)
Indication from procedure	557 (93)
Presence of VUR	601 (99)
Grade of VUR	585 (97)
Cyclic study	95 (16)
Complete bladder emptying	429 (71)
Bladder appearance	517 (86)
Urethra appearance	447 (74)
Delayed images assessing upper tracts	151 (25)

¹Nine reports lacked information to identify the radiologist

²FSPH – Free-standing pediatric hospital; PHWH – pediatric ‘hospital within a hospital’; NPH – non-pediatric hospital; ORF – outpatient radiology facility

³One facility type was unknown

INTRODUCTION:

In the United States, the voiding cystourethrogram (VCUG) has been recommended for the workup of many children presenting with febrile urinary tract infection¹. In addition to diagnosing vesicoureteral reflux (VUR), this test also provides a wide range of information about the anatomy and function of the upper and lower urinary tracts. As with most imaging studies, clinicians rely on the radiologist’s written report to summarize the findings of the VCUG, particularly when the images are not available for direct review, or when the clinician is not familiar with radiological interpretation of the images themselves. Failure to include all of the relevant information in a VCUG report obliges the clinician to make management decisions with incomplete information.

Although templates have been proposed for structuring the information in a VCUG report, previous work has shown great variability in the completeness of reports, which frequently omit many key elements². Pediatric-centered institutions were shown to produce more complete VCUG reports². Several reasons could explain improved reporting at pediatric facilities, such as greater familiarity with and higher volumes of the procedure, or greater pediatric experience and training among radiologists. However, the generalizability of these findings is unclear as the data were based on a regional sample.

The Randomized Intervention for Children with Vesicoureteral Reflux (RIVUR) trial is a randomized controlled trial of antimicrobial prophylaxis among children aged 2–71 months diagnosed with grade I-IV VUR following a first or second febrile or symptomatic urinary tract infection³. Prior to enrollment in the RIVUR trial, all children had a VCUG performed. As these VCUGs were obtained as part of standard of care rather than within the context of a research study, no standardization for reporting of these images was in place. We sought to use this widely-representative sample of procedures performed at 90 distinct institutions by over 200 radiologists to assess variability in VCUG reports and to test our hypothesis that VCUG reports from pediatric facilities and pediatric radiologists are more complete than those performed at other facilities.

MATERIALS AND METHODS:

We developed a 23-item assessment tool as an “inventory” of data-points obtained during a standard VCUG. As shown in Table 1, this instrument incorporated variables related to technical information (e.g., catheter size and residual volume), anatomical information (e.g., severity of VUR according to the International Reflux Study (IRS) grading system⁴) and functional information (e.g., emptying of the bladder). The RIVUR study was performed in North America and the assessment tool was used to address VCUGs obtained on these participants. Thus, certain items like route of bladder access (e.g. antegrade suprapubic puncture v. retrograde urethral catheterization) were not considered. All reports were compared using this 23-item assessment tool.

Table 1.

Frequency of reporting of elements of a voiding cystourethrogram

		Source of VCUG Report: 1,2
Item	All Facilities (N=602) No. (%)	FSPH (N=297) No. (%)	PHWH (N=184) No. (%)	NPH (N=87) No. (%)	ORF (N=33) No. (%)
Indication for procedure	557 (93)	281 (95)	179 (97)	79 (91)	18 (55)
Urethral catheter size	178 (30)	89 (30)	60 (33)	22 (25)	6 (18)
Residual urine (volume) at placement of catheter	63 (10)	24 (8)	38 (21)	1 (1)	0 (0)
Urine culture or urinalysis sent	61 (10)	46 (15)	9 (5)	2 (2)	4 (12)
Contrast type instilled	448 (74)	237 (80)	151 (82)	39 (45)	20 (61)
Scout film – bony structures	318 (53)	165 (56)	122 (66)	16 (18)	14 (42)
Scout film – abdominal structures	402 (67)	210 (71)	146 (79)	23 (26)	22 (67)
Cyclic study performed	95 (16)	49 (16)	39 (21)	5 (6)	2 (6)
Observed volume of bladder	503 (84)	262 (88)	163 (89)	50 (57)	27 (82)
Predicted capacity of bladder	47 (8)	35 (12)	10 (5)	2 (2)	0 (0)
Bladder appearance (e.g. smooth-walled, trabeculated, “Christmas tree”)	517 (86)	270 (91)	159 (86)	58 (67)	29 (88)
Presence of VUR	601 (99)	297 (100)	183 (99)	87 (100)	33 (100)
Grade of VUR	585 (97)	296 (100)	183 (99)	74 (85)	31 (94)
Bladder volume at onset of VUR	14 (2)	2 (1)	9 (5)	2 (2)	1 (3)
VUR onset during filling vs. voiding phase	344 (57)	174 (59)	96 (52)	52 (60)	22 (67)
Location of insertion of ureters reported	65 (11)	46 (15)	14 (8)	4 (5)	1 (3)
Duplication of collecting system reported	38 (6)	22 (7)	10 (5)	4 (5)	2 (6)
Voiding phase observed	571 (95)	286 (96)	176 (96)	76 (87)	32 (97)
Urethral appearance	447 (74)	246 (83)	136 (74)	41 (47)	24 (73)
Complete bladder emptying	429 (71)	208 (70)	151 (82)	47 (54)	22 (67)
Delayed images assessing drainage/obstruction of upper tracts	151 (25)	69 (23)	66 (36)	11 (13)	4 (12)
Fluoroscopy time	125 (21)	78 (26)	20 (11)	18 (21)	9 (27)
Radiation exposure	56 (9)	55 (19)	1 (1)	0 (0)	0 (0)

¹FSPH = Free-standing pediatric hospital; PHWH – pediatric ‘hospital within a hospital’; NPH – non-pediatric hospital; ORF – outpatient radiology facility

²One facility was unknown

Each RIVUR participant had an index VCUG performed prior to enrollment. After de-identification, we reviewed these VCUG reports using the standardized assessment tool, and noted the institution where each report was generated and the reporting radiologist. The Children’s Hospital Association (CHA) website was used to categorize the facility where each VCUG was generated as a free-standing pediatric hospital or pediatric specialty hospital (FSPH), or a pediatric hospital within a larger general hospital (PHWH)⁵. If the institution was not on the CHA website but was clearly identified as a hospital, it was categorized as a non-pediatric hospital (NPH). If the institution did not match any of the above criteria, it was categorized as an outpatient radiology facility (ORF). We used the Check Board Certification tool within the American Board of Radiology (ABR) website to identify whether the attending radiologist had pediatric subspecialty certification; those with certification were classified as pediatric radiologists⁶. For radiologists not identified by the ABR tool, the website of the radiologist’s employer was searched to identify whether he/she had completed a pediatric radiology fellowship or his/her research and clinical focus was dedicated solely to the practice of pediatric radiology. This ensured that radiologists exempted from pediatric subspecialty certification were classified appropriately.

Using the assessment tool, we assigned one point for each item on the tool, with each item given equal weight. Items were given equal weight because we felt that all components (not just grade of VUR) were critical to extracting the maximal information from the VCUG. A score was generated by dividing the number of points achieved for each report by 23, the maximum number of points, and multiplying the result by 100. This score represented the percentage of elements assessed in each VCUG report and represented the primary outcome for this study.

We used descriptive statistics to display relationships between participants, hospital and provider characteristics, and the percentage of elements assessed. Non-pediatric hospitals and outpatient radiology facilities were grouped into the same category for statistical analysis. Unadjusted and adjusted models were considered to study the association between outcome and type of institution where the report was generated, and whether the report was read by a pediatric radiologist. Age at VCUG, participant’s gender, and VUR grade affect the VCUG study and its reporting, and accordingly, were included as covariates. Generalized estimating equations were used to account for correlations among reports generated by the same radiologist and within the same institution. All statistical tests were two-sided, and P<0.05 was considered significant. SAS software v. 9.3 (SAS Institute Inc., Cary, NC) was used for data management and analysis.

RESULTS:

Of 607 RIVUR participants, 602 had VCUG reports available for review. The sample included 49 males and 553 females; median age at the time of VCUG was 11 months (Table 2). These reports were generated at 90 distinct institutions (12 FSPH, 19 PHWH, 52 NPH, 7 ORF) by 221 different radiologists, 127 (57%) of whom were classified as pediatric radiologists. The institution’s name and type was not available for one report, and the radiologist’s name and type were not known for nine reports. One report was generated from a NPH in Canada.

Table 2.

Baseline characteristics of subjects and VCUG reports

Characteristic	N=602 No. (%) or Median [IQR]
Gender
Male	49 (8.1%)
Female	553 (91.9%)
Age at VCUG (months)	11 [5, 30]
Reference Radiologist VUR Grade 1
0	319 (26.5%)
I	104 (8.6%)
II	400 (33.2%)
III	321 (26.7%)
IV	54 (4.5%)
V	1 (0.1%)
VCUG read by a pediatric radiologist 2
Yes	459 (76.2%)
No	134 (22.3%)
VCUG performed at 3,4
FSPH	297 (49.3%)
PHWH	184 (30.6%)
NPF	87 (14.5%)
ORF	33 (5.5%)

¹Reference grade not available for 5 ureters

²Nine reports lacked information to identify the radiologist

³FSPH – Free-standing pediatric hospital; PHWH – pediatric ‘hospital within a hospital’; NPH – non-pediatric hospital; ORF – outpatient radiology facility

⁴One facility type was unknown

Amongst the whole cohort, VCUG reports included an average of 47.9 ± 11.8% of the items assessed. Elements that were reported almost uniformly included VUR status (99% reported), grade of VUR using the IRS grading system (97%), whether the voiding phase was observed (95%), and a statement about the indication for the procedure (93%) (Table 1). Examples of items infrequently reported include the residual urine volume at placement of the urethral catheter (10% reported), radiation exposure and fluoroscopic time (9% and 21%, respectively), location of the insertion of refluxing ureters into the bladder (11%), whether delayed images assessing drainage of the upper tracts were performed (25%), duplication of collecting systems (6%), and bladder volume at onset of VUR (2%). Even for VUR grade, not all reports were complete: 15% and 6% of reports from NPH and ORF, respectively, did not grade VUR using the IRS grading system. Cyclic studies are important for maximizing the sensitivity of VUR detection in infants. Among the 306 participants aged 2–12 months for whom knowing whether a cyclic study was performed is critical, only 66 reports (22%) included this information.

In addition to low rates of inclusion for many items generally, univariate analyses showed substantial variability in the completeness of VCUG reports among institutions and radiologist types (Table 3). Specifically, 50.7 ± 11.4% of elements were reported in FSPH, compared to 50.3 ± 9.5% in PHWH (FSPH vs. PHWH: P = 0.86), and 37.8 ± 10.4% in NPH and or ORF (FSPH vs. NPH/ORF: P< 0.01). Pediatric radiologists reported 50.6 ± 10.7% of items versus 39.2 ± 10.7% of items reported by non-pediatric radiologists (P< 0.01). After controlling for participants’ gender, age at the time of VCUG, and grade of reflux, reports generated at non-pediatric hospitals or outpatient radiology facilities included 8% (95% CI [−3.0, −12.8], P< 0.01) fewer items compared with those generated at FSPH (Table 3). There was no difference in the completeness of reports between PHWH and FSPH. Multivariate analysis also showed that VCUGs reported by pediatric radiologists had, on average, 6.3% more items (95% CI [2.9, 9.7], P< 0.01), compared with those reported by non-pediatric radiologists.

Table 3:

Unadjusted and adjusted associations between percentage of items assessed in VCUG reports and participant and report characteristics and hospital type

Characteristic	N	Mean percentage of variables assessed (SD)	Unadjusted Difference in percentage of variables assessed in VCUG report (95% CI)	P-value	Adjusted Difference in percentage of variables assessed in VCUG report (95% CI)	P-value
Overall Study Sample	602	47.9 (11.8)	-	-	-	-
Patient age at time of VCUG
≥12 months	277	47.8 (11.5)	−0.5 (−2.1,1.2)	0.57	0 (−1.3, 1.4)	0.95
<12 months	314	48.2 (12.0)	Referent		Referent	-
Gender
Male	47	49.5 (11.9)	1.7 (−2.2, 5.5)	0.40	2.0 (−1.5, 5.5)	0.26
Female	546	47.9 (11.7)	Referent		Referent	-
Worst Adjudicated VUR grade
High grade (Grades IV, V)	48	48.5 (13.3)	0.4 (−3.0, 3.8)	0.82	0.3 (−2.6, 3.1)	0.85
Moderate grade (Grades III)	229	47.8 (11.8)	−0.3 (−2.2, 1.6)	0.74	0.1 (−1.4, 1.7)	0.87
Low grade (Grades 0, I, II)	315	48.1 (11.5)	Referent		Referent	-
VCUG read by a pediatric radiologist
Yes	459	50.6 (10.7)	11.4 (7.9, 14.9)	<0.01	6.3 (2.9, 9.7)	<0.01
No	134	39.2 (10.7)	Referent		Referent	-
Hospital Type
Non-pediatric facility or imaging facility	118	37.8 (10.4)	−12.8 (−17.2, −8.5)	<0.01	−7.9 (−12.8, −3.0)	<0.01
Pediatric HWH	183	50.3 (9.5)	−0.4 (−5.0, 4.1)	0.86	0.7 (−4.0,5.5)	0.76
Free-standing Pediatric Hospital	291	50.7 (11.4)	Referent	-	Referent	-

DISCUSSION:

This study shows that substantial underreporting of findings occurs in VCUG reports when comparing this sample of routine, community generated reports to a list of 23 technical, functional, and anatomic findings that are important in VCUG reporting. When analyzing the reports based on the type of facility and radiologists, these data suggest that reports generated at pediatric hospitals by pediatric radiologists are 6 and 8% more complete, respectively. Yet these differences are small when considering that about 50% of reports excluded key items irrespective of where and by whom the VCUG report was generated. Importantly, nearly all VCUG reports commented on the presence of VUR, although as many as 15% and 6% of reports generated at non-pediatric hospitals or outpatient radiology facilities, respectively, did not grade reflux using the IRS grading system. Even if one views the presence and grade of VUR as the only important finding on a VCUG, these results are noteworthy.

However, the VCUG provides a wealth of additional information about the urinary tract that may be of interest to both pediatricians and urologists. This includes information about the efficiency of bladder emptying, radiological signs like a “Christmas tree” shaped bladder suggesting neurogenic bladder, and the presence of ureteral duplications, among others. Thus, if the decision has been made to subject a child to a somewhat invasive procedure (and its associated ionizing radiation), it is incumbent upon clinicians and radiologists to ensure that the maximum information possible is derived from the study, and made available to guide management decisions. Our results underscore the opportunities for improvement in the reporting of VCUG findings.

What findings should be reported on a VCUG, and why are they important? Numerous findings may be of interest. Residual volume at placement of the catheter (particularly in a toilet-trained child) can be an indication of incomplete emptying due to bladder outlet obstruction, bowel and bladder dysfunction, or other causes. Delayed drainage films can show retention of contrast in upper tracts indicative of ureteropelvic junction or ureterovesical junction obstruction. Duplex refluxing systems are less likely to resolve spontaneously⁷. Performance of a cyclic study increases the sensitivity of a VCUG for VUR, particularly in infants⁸. Noting whether VUR occurred during filling or voiding phase, and the bladder volume at which VUR was first noted, have been shown to be predictive of resolution^{9, 10}. Inclusion of such parameters in the VCUG report does not just represent an academic exercise; these data points have real influence on decision-making for families and clinicians. Even when such parameters are assessed as negative by the radiologist, omitting them from the report leaves an open question in the mind of the reader: did the radiologist make no mention of the duplication status in the renal moiety with grade III VUR because it was a single system, or because the duplication status was not assessed at all? In such cases, omission cannot be distinguished from failure to assess.

The findings of this study largely confirm a previous report using regional data that showed that VCUG reports usually contain information on VUR status and grade, but often omit many other elements². The prior study also showed that reports generated at non-pediatric facilities and PHWH had on average 17 and 9% fewer items, respectively, compared with reports generated at FSPH. The present, nationally representative study supports the hypothesis that pediatric hospitals and pediatric radiologists have more complete reporting of key elements, irrespective of geographic location.

One might anticipate that a standardized VCUG reporting template, defined and agreed upon by key stakeholders, could improve the quality of VCUG reports. Marcovici and Taylor compared chest radiograph reports using structured reporting templates to unstructured stream-of-consciousness dictations¹¹, and found that structured reports were more complete. One outcome of the Radiological Society of North America’s (RSNA) initiative to improve radiology reporting practices was a VCUG template¹². This template includes some elements of our report assessment tool, but could be improved with the inclusion of the IRS VUR grade, whether upper tract drainage occurred, and location of ureteral insertion. It is possible that a comprehensive structured template, if widely disseminated and implemented, would dramatically improve the performance of VCUG tests and documentation of its findings.

The results of this study should also be viewed in light of its limitations. Foremost among these is the relatively subjective 23-item report quality assessment tool. This tool was generated by two senior authors (one pediatric urologist and one pediatric radiologist), and although the items included are generally evidence-based (or at least common sense-based), it represents the clinical biases and preferences of the study authors. However, the assessment tool used in the current study contained all items on the RSNA template, plus additional, clinically-significant items, thus making our tool relatively more stringent than the RSNA template. There may be other findings (e.g. presence of intrarenal reflux) which are important to other clinicians that were not included in our assessment tool. All of these differences in assessment tools would likely serve to magnify the findings of our study, if they change them at all. A final limitation is that our study focused on the content of the VCUG report itself; we did not attempt to assess the actual technique of the test, or the accuracy or reliability of the interpretation of the anatomic findings.

CONCLUSION:

Although pediatric radiologist and pediatric hospitals generated more complete VCUG reports compared to those from non-pediatric origins, the differences are small when considering the substantial underreporting of VCUG findings in general. This underscores the opportunities for improvement in reporting of VCUG findings.

Abbreviations:

RIVUR

Randomized intervention for children with vesicoureteral reflux

UTI

Urinary tract infection

VCUG

Voiding cystourethrogram

VUR

Vesicoureteral reflux

IRS

International Reflux Study

FSPH

Free-standing pediatric hospital

PHWH

Pediatric hospital within a larger general hospital

NPH

Non-pediatric hospital

ORF

Outpatient radiology facility