Description

Kumaravel et al developed a model for identifying a patient with severe acute pancreatitis using body mass index and change in serum amylase. The authors are from the Cleveland Clinic and University of Pittsburgh.


 

Patient selection: acute pancreatitis

 

Parameters:

(1) body mass index in kg per square meter (BMI)

(2) serum amylase in U/L on day 2

(3) serum amylase in U/L day 1 (day of admission)

 

percent change in amylase =

= ((amylase day 2) - (amylase day 1)) / (amylase day 1) * 100

 

X =

= (0.14 * BMI) + (0.01 * (percent change in amylase day 2)) - 5.9

 

where:

• The 0.01 and 100 would cancel each other out.

 

probability of severe pancreatitis =

= 1 / (1 + EXP((-1) * X))

 

Interpretation:

• A probability > 9% was considered positive.

 

Performance:

• The area under the ROC curve was 0.73 to 0.79.

• Using a probability > 9% the sensitivity was 75% and specificity 65%. The negative predictive value was 95% vs positive predictive value of 22%.

 


To read more or access our algorithms and calculators, please log in or register.