Date of Award

January 2012

Document Type


Degree Name

Medical Doctor (MD)



First Advisor

Christopher Moore

Subject Area(s)

Medicine, Medical imaging and radiology


Background: The incidence of nephrolithiasis has increased across all age groups, with CT now considered a first-line test for adults with suspected kidney stones. CT imaging for suspected stones in pediatrics has also increased, despite evidence that ionizing radiation may be more damaging in younger patients.

Objectives: To identify clinical characteristics in pediatric patients that can aid in the diagnosis of kidney stones without reliance on CT imaging.

Methods: This retrospective study included all patients under the age of 21 who received a CT flank pain protocol (FPP) in one of three ED settings at our institution (YNNH Adult and Pediatric EDs and Shoreline ED) between 04/01/2005-02/07/2009. One hundred and one variables were identified a priori and extracted, using an explicit approach, from a scanned, templated handwritten ED record by a single reviewer blinded to CT results. Dictated CT findings were extracted (blinded to clinical variables) for presence of symptomatic ureteral stone or other findings requiring intervention. Presence of hydronephrosis (hydro) was extracted from the CT report and used as a predictor variable. Recursive partitioning using Classification and Regression Tree (CART) analysis with 10-fold cross-validation was used to determine optimal splits and derive decision trees.

Results: Of 2973 CT FPPs in the ED setting over approximately four years, 165 (5.5%) were done in patients under age 21. Of these 165 scans, there were 157 eligible scans in 152 individuals. The median age was 18 (range 6-20; interquartile range 16-20) and 66% were female (n=101). Of the CT scans, 55/157 (35%) identified a symptomatic kidney stone, all 5mm or less. Males were more likely than females to have symptomatic stones (Odds Ratio: 2.3, 95% CI: 1.2-4.2). Two patients were diagnosed with conditions requiring emergent intervention (ovarian torsion and appendicitis). Hydro (on CT dictation) was overwhelmingly predictive of kidney stone, and CART analysis of possible predictor variables yielded a decision tree that identifies patients with hydro mild or greater as likely to have symptomatic kidney stones with a specificity of 0.96 (95% CI: 0.90-0.99) and sensitivity of 0.78 (95% CI: 0.65-0.88). Excluding hydro, CART analysis yielded a decision tree that identified patients with pain duration less than 24 hours and with hematuria as likely to have symptomatic kidney stones with a specificity of 0.81 (95% CI: 0.72-0.88) and sensitivity of 0.73 (95% CI: 0.59-0.88).

Conclusion: Hydronephrosis was overwhelmingly predictive of kidney stones in our pediatric cohort, and even excluding hydronephrosis, flank pain for less than a day and hematuria were predictive of kidney stones and discriminated from other significant findings in this predominantly adolescent population.


This thesis is restricted to Yale network users only. This thesis is permanently embargoed from public release.