Constipation on abdominal radiograph as potential risk factor for recurrent urinary tract infection development

Abstract

Background

To compare clinical history and measurements of fecal load on abdominal radiography (AR) in the prediction of urinary tract infection (UTI) recurrence in children.

Methods

We combined data from two multicenter longitudinal studies in which children less than 6 years of age with a first or second UTI were followed for recurrence of UTI. Two radiologists reviewed the scout abdominal radiographs of initial voiding cystourethrograms obtained at enrollment from children at two participating sites and measured stool visible in various parts of the colon. We examined how well clinical variables (e.g., voiding and bowel history, use of laxatives at enrollment) and measurements of fecal load predicted recurrence of UTI within 12 months of enrollment.

Results

One hundred and ninety-two children were included. On univariate analyses, age, vesicoureteral reflux (VUR), cecal diameter, rectal diameter, and total stool length on AR were associated with recurrence of UTI. After controlling for age, the odds of recurrent UTI in children with VUR at baseline was 3.85 (95% CI: 1.62, 9.14) higher than in children without VUR. Recurrent UTI was 2.57 (95% CI: 1.01, 6.55) times more likely in children with cecal diameter > 3.10 cm than children with lower cecal diameters; time to first recurrent UTI was shorter in children with elevated cecal diameters (p = 0.0023).

Conclusions

Cecal diameter on abdominal radiographs predicts UTI recurrence in children with a previous UTI. However, its accuracy is suboptimal to serve as a screening test. Accordingly, its routine use for this indication is not supported. If cecal diameter on an AR ordered for another indication is > 3.10 cm, then management of constipation could be considered.

Introduction

Constipation and UTIs have been linked; several studies have reported an association between constipation and the recurrence of UTI in children; some studies have also reported that treatment of constipation appears to reduce recurrences of UTI (Table 1).

Table 1.

Characteristics and results of previous studies reporting outcomes of children with constipation presented in reverse chronologic order

Author, year	N	Age (years)	Constipation prevalence	Constipation definition	Study design	More recurrent UTIs in...
Mingin 2004 [1]	850	0–14	--	Clinical	Retrospective chart review of children with febrile UTI	Females > 5 years old with dysfunctional elimination
Koff 1998 [2]	143	--	43%	Clinical	Prospective observational study of children with VUR	Children with dysfunctional elimination
Loening-Baucke 1997 [3]	234	5–18	100%	Clinical	Prospective interventional (without a control group) of children with constipation	Those whose constipation resolved
Blethyn 1995 [4]	61	2–14	38–50%	Radiographic	Retrospective chart review of children with UTI	Children with higher fecal loading on abdominal X-ray

VUR vesicoureteral reflux, UTI urinary tract infection

– Data not reported in manuscript

What is less clear is how to best assess constipation in young children. In addition to clinical history, some clinicians continue to obtain abdominal radiographs to assess constipation and its severity. On radiographs, severity can be assessed by measuring distention of key regions of the colon [5].

The objective of this study was to determine whether constipation by parental history or measurement of specific radiographic parameters in different areas of the colon best predicts development of subsequent UTI.

Methods

For this observational study, we used data and images from the previously completed Randomized Intervention for Children with Vesicoureteral Reflux (RIVUR) and Careful Urinary Tract Infection Evaluation study (CUTIE) studies. Both were prospective longitudinal studies, and the study forms and follow-up were largely identical; UTI recurrence was one of the main outcomes in both studies. The RIVUR [6] trial was a 2-year multisite, randomized, placebo control trial that evaluated the efficacy of antibiotic prophylaxis in children diagnosed with VUR and UTI; participants were randomized to antimicrobial prophylaxis vs. placebo group. The CUTIE [7] study was a parallel observational clinical study in children 2 months to 6 years old, aimed at understanding which children are at the greatest risk of developing kidney scarring following a UTI without history of VUR.

We included children enrolled in the above-mentioned studies, from the Pittsburgh, PA, and Chicago, IL sites, who had a scout abdominal X-ray as part of their voiding cystourethrogram (VCUG). Study approval by the University of Pittsburgh and the University of Chicago Institutional Review Boards were obtained. Children at the remaining 17 study sites were excluded due to lack of access to the radiographic images. We also excluded children in the RIVUR study who were randomized to antimicrobial prophylaxis because relatively few of these children had a recurrent UTI during the first year of the study. To avoid possible selection bias, the included children matched demographic characteristics when compared to the two source studies.

We asked two radiologists who were blinded to outcomes, treatment group, and site, to assess the abdominal radiographs of the initial VCUG obtained at baseline. First, each radiologist independently assessed 25 scans. Because the lower limit of the confidence interval for Cronbach’s intraclass correlation coefficient (ICC) was above 0.75 m, the remaining 167 radiographs were randomly split between the two [8, 9]. Using radiographic parameters previously described by Marks et al. [5], the radiologists measured (a) cecal diameter, (b) length of stool in rectum, and (c) total length of stool throughout the colon using previously described methods.

At the time of enrollment in either study, possible constipation was clinically assessed by asking parents four questions: (1) average number of bowel movements per week during the last 2 months, (2) history of ever being diagnosed with constipation before enrollment, (3) history of ever being treated for constipation in the past 2 months, and (4) use of laxatives (prescription or over the counter) at the time of enrollment. These questions were not validated, and each question was considered separately.

We used logistic regression models to examine the association between predictors (demographics, four questions on clinical constipation, and three measurements on AR) and outcome (UTI recurrence during the first year of the study). Prediction was limited to 1 year after enrollment because most recurrences occurred during this period. For each predictor, we constructed receiver operation characteristic (ROC) curves and calculated the area under the curve (AUC). The cutoff that maximized the Youden index (sensitivity + specificity − 1) was calculated. To construct the multivariate logistic model, we selected variables that had a p value < 0.2 from the univariate logistic regression model (Table 3). A stepwise model selection method was applied. The criteria for inclusion of potential independent variables was p < 0.05. We also performed the Hosmer and Lemeshow test for goodness-of-fit. To examine whether constipation, as determined by cecal diameter, was associated with first recurrent UTI, we used Kaplan-Meier (KM) estimates and used the log-rank test to assess for differences. A significance level of p < 0.05 (two-sided) was considered statistically significant. All statistical analyses were performed by SAS version 9.4 and R version 3.6.1.

Table 3.

Odds of recurrent urinary tract infection in first year of follow-up from univariate analysis (N = 192)

	Odds ratio	95% CI	p
Age, months	1.03	1.01, 1.05	0.0014
Sexa	----		0.2203d
Racea			0.2794d
Average number bowel movements per week	0.96	0.90, 1.02	0.2174
History of constipation	2.06	0.94, 4.52	0.0716
Ever been treated for constipation	1.65	0.69, 3.92	0.2580
Laxative use at the time of enrollment	0.79	0.17, 3.67	0.7584
Vesicoureteral reflux at baseline (VUR)
None (reference)	---	---	---
Any grade	2.89	1.32, 6.53	0.0086
Cecal diameter at baseline
≤ 3.10 cm (reference)	---	---	---
> 3.10 cm	3.44	1.49, 7.95	0.0039
Rectal stool diameter at baseline, cm
≤ 4.00 cm (reference)	---	---	---
> 4.00 cm	2.57	1.17, 5.65	0.0185
Total length of stool in colon at baseline, cm
≤ 27.41 cm (reference)	---	---	---
> 27.41 cm	2.74	1.24, 6.04	0.0125
Constipation assessment on abdominal radiographb	1.91	0.87, 4.19	0.1072

^aNo males had a recurrent UTI. No black child had a recurrent UTI

^bModerate to severe (vs. mild)

Results

A total of 192 patients (Pittsburgh n = 188; Chicago n = 4) was included. Their median age was 13.0 months, 94% were female, and 80% were white. VUR was identified in 40% of subjects, and 31% had a clinical history of constipation. On AR, the median cecal diameter was 2.5 cm, and 48% of children were rated as having moderate or severe constipation by the radiologists (Table 2). We found no significant differences in age, sex, or race between the 192 children included and the 610 excluded children (Table in supplement).

Table 2.

Baseline demographic, clinical, and radiographic characteristics of 192 children included

	N = 192
Demographic characteristics
Age, months, median [IQR]	13.0 [6.0, 42.0]
Female sex	94% (180)
Race
White	80% (153)
Black	8% (15)
Other	12% (23)
Clinical characteristics
Average number bowel movement per week, median [IQR]a	7.0 [6.8, 10.2]
History of constipationb	31% (60)
Ever been treated for constipationc	21% (41)
Laxative use at the time of enrollmentd	8% (15)
Baseline bladder and bowel dysfunction
No BBD	17% (32)
BBD	15% (28)
Not toilet trained	68% (129)
Vesicoureteral reflux at baseline (VUR)
None	60% (116)
Grade I	6% (11)
Grade II	18% (35)
Grade III	11% (22)
Grade IV	4% (8)
Radiographic assessment of stool burden on baseline abdominal radiograph obtained during VCUGe
Cecal diameter at baseline, cm, median [IQR]	2.5 [1.8, 3.0]
Rectal stool diameter at baseline, cm, median [IQR]	2.8 [1.2, 4.4]
Total stool length in colon at baseline, cm, median [IQR]	24.0 [17.1, 32.2]
Overall radiographic assessment of constipationf
Mild	52% (100)
Moderate/severe	46% (89)/2% (3)

^aExact wording of question posed to parent: “During the last 2 months, how many bowel movements did your child have per week on average?”

^bExact wording of question posed to parent: “Does your child have a history of constipation?”

^cExact wording of question posed to parent: “Has your child ever been treated for constipation?”

^dLaxatives included: Benefiber/Docusate/Golytely/lactulose/lactulose syrup/mineral oil enema/Senna/Senna syrup/fleet laxative/fleets enema/magnesium citrate/methyl cellulose powder/mineral oil/MiraLAX/Miralx/Stool softener

^eVCUG voiding cystourethrogram

^fRadiologist assigned overall impression

A total of 31 children had recurrent UTIs; variables identified as having a significant association with UTI recurrence included older age, VUR, higher cecal diameter, higher rectal diameter, and higher total length of stool in colon (Table 3). No males had a recurrent UTI. No black child had a recurrent UTI. For measurements on AR, the cutoff was done by selecting the value that maximized sensitivity and specificity (cecal diameter = 3.10 cm, rectal diameter = 4.00 cm, total length of stool in colon = 27.41 cm). Children with cecal diameter > 3.10 cm had a higher risk to develop a recurrent UTI than those who had cecal diameter ≤ 3.10 cm (OR = 3.44, CI: 1.49, 7.95). Children with rectal diameter > 4.00 cm had a higher risk to develop a recurrent UTI than those who had rectal diameter ≤ 4.00 cm (OR = 2.57, CI: 1.17, 5.65). Children with total length of stool in the colon of > 27.41 cm had a higher risk to develop a recurrent UTI than those who had total length of stool in the colon ≤ 27.41 cm (OR = 2.74, CI: 1.24, 6.04). Clinical history of constipation at the time of enrollment (i.e., average number of bowel movements per week, history of constipation, having ever been treated for constipation, and laxative use) was not significantly associated with risk of recurrence of UTI. Recurrent UTI was developed in 32% (n = 12/37) of children who had cecal diameter > 3.10 cm. For children who had cecal diameter ≤ 3.10 cm, 12% (n = 19/155) developed a recurrent UTI. Of the 31 children who had a UTI recurrence, 12 (39%) had a cecal diameter > 3.10 cm, 14 (45%) had a history of constipation, nine (29%) had ever been treated for constipation, and two (6%) reported taking laxative at the time of enrollment; thus, a clinical history of constipation had the highest sensitivity (i.e., of 45%). Of the 161 children who did not have recurrence, 25 (16%) had a cecal diameter > 3.10 cm, 46 (29%) had history of constipation, 32 (20%) were ever treated for constipation, and 13 (8%) were taking laxative at time of enrollment (Table 3); thus, cecal diameter of < 3.10 cm had the highest specificity (of 84%).

A multivariate model that included age, VUR, and cecal diameter had a predictive power (as measured by area under the ROC curve) of 0.77 (Table 4). After controlling for age, both VUR and cecal diameter at baseline were associated with recurrent UTI. No other significant associations were found. The odds of recurrent UTI in children with VUR at baseline was 3.85 (95% CI: 1.62, 9.14) higher than in children without VUR. Recurrent UTI was 2.57 (CI: 1.01, 6.55) times more likely in children with cecal diameter > 3.10 cm than children with lower cecal diameters (Table 4).

Table 4.

Multivariable model for predicting odds of recurrent urinary tract infection in first year of follow-up (n = 192)

	Odds ratios	95% CI	p
Age, months	1.03	1.01, 1.05	0.0054
Vesicoureteral reflux at baseline (VUR)
None (reference)	---	---	---
Any grade	3.85	1.62, 9.14	0.0022
Cecal diameter at baseline
≤ 3.10 cm (reference)	---	---	---
> 3.10 cm	2.57	1.01, 6.55	0.0489

AUC = 0.77, p value from Hosmer and Lemeshow Goodness-of-Fit test = 0.4609

On time to event analysis, we found that recurrent UTI occurred earlier (p = 0.002) in children with cecal diameter > 3.10 cm compared to children with lower cecal diameter (Fig. 1).

Fig. 1.

Time to first recurrent UTI during the first year of follow-up (n = 192, p = 0.0023)

Discussion

To our knowledge, this is the first prospective study to compare clinical and radiographic parameters in predicting the recurrence of UTI. Most studies to date (Table 1) have been retrospective and have included children older than 2 years of age. Only a single case-control study to date has assessed the association between radiographic criteria of constipation and recurrent UTIs [4]. In that study, ARs of 61 children with culture-proven UTI were compared to ARs of 30 controls without UTI (who had an AR for another unrelated diagnosis). A significant association between the degree of fecal loading (scores 0–3) and the number of previous UTIs (p < 0.05) was noted. Our study differs in that it was longitudinal (and thus avoids the limitations of retrospective case-control design) but similarly concluded that recurrent UTI occurred earlier (p = 0.002) in children with a higher degree of fecal loading. Two of the available studies provide data for children with dysfunctional elimination, many of whom were constipated, but stratified data on the rate of recurrences in children with constipation was not provided [1, 2].

We did not explore bowel bladder dysfunction (BBD) or alternative definitions of constipation (e.g., the Paris Consensus on Childhood Constipation Terminology [10]) as predictors or confounders because these were only assessed in a small subset of toilet-trained children. Most subjects in our cohort were not toilet trained. In our study population, older age, presence of VUR, and cecal diameter > 3.10 cm were better predictors of UTI recurrences than clinical history of constipation (Table 3). In our study, all UTI recurrences were in girls, probably because the study included predominantly females.

Although the radiographic measurements used in this study were reported as having excellent accuracy in predicting clinical constipation in a previous study [5] unlike rigorous diagnostic accuracy studies in which a cross-sectional design is used, these estimates were obtained from a case-control study comparing children with likely severe constipation presenting to a gastroenterology clinical at a tertiary center and controls with successfully treated constipation. We found that in children less than 6 years of age, cecal diameter was a weak predictor of UTI recurrence. Accordingly, this does not mean that an AR should be routinely ordered for the purpose of predicting future UTIs. In our cohort, 68% children with cecal diameter measurements > 3.10 cm did not have UTI recurrences, whereas 12% with diameters ≤ 3.10 cm did. Even though the likelihood of a recurrent UTI was higher in children with a high cecal diameter, the accuracy was not high, and the confidence interval around this estimate was wide, indicating the uncertainty around the estimate. In addition to its insufficient accuracy to serve as a diagnostic test, exposure to unnecessary radiation is another issue to consider [3, 11–13]. Ultrasound examinations of rectal diameters have also been proposed as alternatives [14]. However, the sensitivity of this technique has not been high enough to promote widespread use.

This study has several limitations. Only a subset of patients from the source studies were included; however, there were no significant differences in age, race, or sex between included and excluded subjects. In the original two studies, 91% were females in comparison to 94% in our sample. The clinical assessment of constipation was assessed through four non-validated questions; thus, lack of association between the answers to these questions and recurrence of UTI may be generalizable. Because of this limitation, the predictive value of clinical assessments deserves further study. Another limitation is the lack of detailed data regarding treatment of constipation during the study period; treatment of constipation could have affected the number of recurrent UTIs. Finally, most children were not old enough to self-report their symptoms (median age of included children was 13 months); the association between the examined predictors (demographic, clinical, and X-ray findings) and recurrent UTIs may be different in older children.

In conclusion, data from this study does not seem compelling enough to warrant routine use of ARs in young children with UTI. However, if young children in whom an abdominal radiograph is already available for other reasons (e.g., children who already had a VCUG), this study supports the need to consider proactive management of possible constipation. Further studies are needed to validate our findings.

Supplementary Material

Supplementary Information The online version contains supplementary material available at https://doi.org/10.1007/s00467-021-04973-5.

What’s known on this subject

Many articles have described a potential association between constipation and development of UTIs. We examined the presence and severity of fecal loading on abdominal radiographs of children diagnosed with UTIs as a potential predictor for recurrent infections.

What this study adds

We found that an increased cecal diameter on abdominal radiography was associated with UTI recurrence. These findings have important implications for the screening and treatment of constipation.

Abbreviations

UTI

Urinary tract infection

VUR

Vesicoureteral reflux

RIVUR

Randomized Intervention for Children with Vesicoureteral Reflux

CUTIE

Careful Urinary Tract Infection Evaluation study

BBD

Bladder and bowel dysfunction

AR

Abdominal radiograph

VCUG

Voiding cystourethrogram

DMSA

Dimercaptosuccinic acid renal scan

PACCT

Paris Consensus on Childhood Constipation Terminology

OR

Odds ratio

ROC

Receiver operation characteristic curve

AUC

Area under the ROC curve

KM

Kaplan-Meier curve