Özet:

Pediatric growth curves provide the determination of general health and nutritional status of children by following the change of theage-related anthropometric measurements.The LMS method is widelyused to construct these curves.The LMS method can model thelocation, scale and skewness measures of the measurements, but not the kurtosis measure.LMSP and LMST are morerecent method swhich can model bothfourparameters of measurement distributio. In this study, it is aimedtocompare the performances of LMS, LMSP and LMST methods in a realdataset. Methods: The data set contains 13 different anthropometric measurements of 2894 healthy 0-6 agedyearschildrenliving in Kayseri. All three methods were applied in the gamlss package of the R software, and the Akaikein formationcriterion (AIC), the generalized Akaikein formationcriterion (GAIC) and the Schwarz Bayes informationcriterion (SBC) were used for comparison. Results:The LMS method performedas the best model with 7 measures in boy sand 5measures in girls,based on AIC. This method performedas thebest model in 9 measures in boy sand 5 measures in girls, based on GAIC. Lastly, it seemed as the best performed method in 11 measures in boy sand 12 measures in girls, based on SBC.In other cases, the LMSP method was found to be the best model in general. Conclusion:The LMS method is a well-performedone. However, the other two methods also performed as best growth curve models for various measurements. While constructing growth curves, researchers should considerall three methods and should decide the best model by carrying out comparative analyses.

Anahtar Kelimeler: