CONSTRUCTION AND VALIDATION OF AN EARLY-P REGNANCY RISK PREDICTOR INSTRUMENT

An early pregnancy risk scale, with scores ranging from 11 to 66 point s from lower to higher risk, was constructed using variables associate d with teenager's pregnancy. This scale was applied to 3000 female tee nagers, coming from Metropolitan Santiago public schools. The sample w as divided in three risk groups: group A (high risk) with scores equal or over 35 points, group B (low risk) with scores equal or below 20 p oints and group B (intermediate risk) with scores between 20.1 and 34. 9 points. These girls were followed during 2 years. During this period , 84 girls became pregnant, 24 of 184 (13%) in group A, 60 of 2332 (2. 6%) in group C and none of 307 in group B. There were 104 school deser tions in group A and 37 in group B. To study associations and analyze risk, the sample was divided in two risk groups: high, with scores ove r 27 and low, with scores below 27. There was a high association betwe en pregnancy risk score and the occurrence of pregnancy (RR 5.25 p < 0 .0001) and school desertion (RR 3.32 p < 0.0001). Pregnancy was predic ted with a 78% sensitivity and 55.6% specificity. School desertion was predicted with a 74% sensitivity and 56% specificity. The importance variable weighing using multiple regression models, to improve the pre dictor's sensitivity and specificity, is discussed.