alculation of the diagnostic accuracy of the
scale: individual scores were added up and a
ROC curve was developed to determine the
best cut-off point and to calculate the area
under the curve. Stratification was chosen
using the items corresponding to values of 1-
sensitivity of the following ranges: < 0.5; 0.51-
0.75; 0.76-0.94 and ≥ 0.95.
f) Prospective validation of the scale: A
prospective cohort of newborn infants was
designed using data from January 2010 to
December 2012. Selection criteria were the
same as for the previous cohort. Follow up
started at the time newborn infants were
admitted at the NICU. Outcome measures
included in this scale were taken within the
first hour of admission Based on the first
phase of the study that included 424 patients
admitted in the NICU, with a 5% mortality
rate (a= 95% and d= 99%), the odds ratio for a
finite population was estimated resulting in a
required sample size of 90 newborn infants. A
10% was added for losses and 10 patients were
added per outcome measure (90), resulting in
a final size of 189 patients.