Objective: To determine if acid base status predicts which vomiting patient
s have pyloric stenosis
Design: Retrospective chart review.
Setting: Tertiary paediatric hospital.
Methodology: We compared the clinical and biochemical parameters of 100 pat
ients with a discharge diagnosis of pyloric stenosis and 84 patients of a s
imilar age who presented to the emergency department with vomiting and who
had an acid base determination. Patients were included from January 1995 to
January 1997. Clinical correlates consisted of age, duration of vomiting,
weight loss, gestation, and family history of pyloric stenosis. Biochemical
correlates were pH, bicarbonate, base excess (BE), chloride, potassium, an
d sodium.
Results: Independent variables of significance were pH, BE, chloride, bicar
bonate, potassium, weight loss (all of which had a P value <0.0001), and se
x (P = 0.006). Each variable was placed in a logistic regression equation w
ith pyloric stenosis being the dominant variable. Variables of significance
were pH (P = 0.0001), BE (P = 0.0001), and chloride (P = 0.009). A model f
or predicting pyloric stenosis using these variables was then created with
pH > 7.45, chloride < 98, and BE > +3, with a positive predictive value of
88%.
Conclusion: Acid base determination is a useful screening tool when conside
ring pyloric stenosis. This model now needs to be validated on a prospectiv
e series of patients with vomiting.