ANALISIS RECEIVER OPERATING CHARACTERISTIC (ROC) DALAM PENILAIAN STATUS GIZI (Aplikasi pada data JPS Sidang Kesehatan Provinsi Jawa Timur)

ABSTRACT Receiver Operating Characteristic (ROC) curve analysis is popular tool in medical and imaging research. It conveniently displays diagnostics accuracy expressed in term of sensitivity (or true positive rate) against (1- specificity) (or false positive rate) at all possible threshold values. An ROC plot is obtained by graphing sensitivity values against their equivalent 1- specificity values for observed threshold. The resulting curve ilustrated the trade-off between sensitivity and specificity. An important problem in ROC analysis concerns the comparison of two (or more) diagnostic tests. In diagnostic accuracy study, if such a test is repeated on different occasions, or if different test are administrated to the same set of subjects, then the test results are typically correlated. The purpose of the study was to evaluate the performance og diagnostic tests of nutritional anthropometrics by using reference standard recommended by WHO-NCHS in three forms of presentation ; percent median, percentile, z-score (i.e. index BB/U, PB/U, BB/PB) using analysis ROC under in varying conditions simulation technique. Result indicated that percent median had the largest area under charcateristic (0.992), and had the smallest probability value (asymptotic Sig' = 0.000) a < 0.05, then with sensitivity had the highest and positive predictive value and likelihood test ratio, too. This value was followed by percentile with specificity also had the highest and negative predictive value. The conclusion of this study was that to take the percent median for detection malnutrition and then the percentile for detection nutrition well.