Hayriye Ertem Vehid1, Gökalp Eral2, Suphi Vehid3, Mustafa Şükrü Şenocak2

1İstanbul Bilim Üniversitesi Tıp Fakültesi Tıp Eğitimi ve Bilişimi Anabilim Dalı, İstanbul, Türkiye
2İstanbul Üniversitesi Cerrahpaşa Tıp Fakültesi Biyoistatistik Anabilim Dalı, İstanbul, Türkiye
3İstanbul Bilim Üniversitesi Tıp Fakültesi Halk Sağlığı Anabilim Dalı, İstanbul, Türkiye

Keywords: High-density lipoprotein cholesterol; low-density lipoprotein cholesterol; triglyceride; troponin


Objectives: This study aims to investigate the known methods showing whether there is any difference between low-density lipoprotein (LDL), high-density lipoprotein (HDL), and triglyceride values based on troponin values being under or above 0.014 ng/mL in cases suspected of a diagnosis of myocardial infarction (MI) and the effect size of such values.

Patients and methods: The study included 256 patients (135 males, 121 females; mean age 62.4±15.2 years; range, 46 to 78 years) who applied to the emergency department between 01 January 2013 and 31 December 2013 with suspected MI. Since the patients’ data were not applicable for normal distribution, Box-Cox transformation was performed on the LDL, HDL, and triglyceride data. Confidence intervals and effect sizes of the data were evaluated to show if there is any difference between the two data groups created according to a troponin value under or above 0.014 ng/mL.

Results: While comparison of the two Box-Cox transformed data groups did not demonstrate any difference in terms of LDL or triglyceride values, it revealed a difference in terms of HDL values. When the Box-Cox transformed data were transformed to their real values and reevaluated, the effect size was d=0.8356 and g=0.8331.

Conclusion: In medical studies, no statistically significant difference is observed between the obtained evaluation results and the transformed evaluation results when data are not applicable for normal distribution. However, optimization of data for normal distribution through transformation is essential since this will allow other evaluations such as confidence interval and effect size, in particular, that assist in a more definite interpretation of any difference between the data of two independent groups.