The development of correction factors for self-reported body mass index values in adults

INTRODUCTION AND AIM: Reputable national or international institutions and establishments like World Health Organization (WHO) utilize two main methods to determine Body Mass Index (BMI) levels and overweight and obesity prevalence in societies. The first method, the Direct Measurement Method (DMM) is the measuring of individuals' height and weight, which requires investment in time, man power, logistical support and economy. The second one is the Self-Report Method (SRM), which entails using individuals' own reports of their height and weight, which they report through surveys, phone or e-mail. The probable errors (biases) in BMI data collected through SRM, can be corrected by using Simple Correction Equation Method (SCEM), which requires utilizing correction factors that have been calculated in earlier studies. The aim of this study is to do a preliminary study on developing Simple Correction Equations (SCE) that can be used to correct the errors in BMI data that rely on SRM throughout Turkey. The study can guide future work with the aim of developing a SCEM which is unique to Turkey and can be used as an alternative to DMM. MATERIALS AND METHODS: The methodological study was conducted between September 2017 and November 2018 in Sakarya with the approval of the ethics committee. A 29-question survey was conducted face to face with individuals between the ages of 18 and 74 by physicians that participated in the study. The survey was composed of questions about certain socio-demographic features and the individuals' attitudes and behavior on measuring their weight and height. The first questions in the survey were the individuals' own weight and height as they know them. Upon the completion of the survey, their height and weight were measured. Ordinary Least Squares Regression (OLS) was used in order to be able to obtain two separate Simple Correction Equations (SCE) for women and men. Statistical package programs, SPSS v20.0 (IBM SPSS Statistics for Windows, Version 20.0; Armonk, NY, USA) and MedCalc v18.11 (MedCalc Software; Ostend, Belgium) were used for analyzing the data. RESULTS: The 1696 individuals that participated in the study composed of 923 women (%54,4) and 773 men (%45,6), with the mean age and standard deviation 40,6 ± 15,3 SD. The mean difference in "cm" and the standard deviation between women's own reports of their height and the calculated height was found to be 1,72 ± 3,12 SD. The mean difference in "kg" was identified as -1,31 ± 2,91 SD and the mean BMI difference in "kg/m2", -1,11 ± 1,67 SD. On the other hand, the mean difference in cm and the standard deviation between men's own reports of their height and the calculated height was found to be 1,20 ± 2,54 SD. Weight difference in "kg" was identified as -0,86 ± 2,80 SD and the BMI difference in "kg/m2", -0,66 ± 1,22 SD. These differences were obtained by subtracting the measured values from the reported values. As a result of the OLS Regression, SCE for women was determined as [Corrected BMI(Female) = (1,043 X Self-Reported BMI(Female)) – 0,040], and [Corrected BMI(Male) = (1,014 X Self-Reported BMI(Male)) + 0,285] for men. According to the gold standard method DMM, the sensitivity of SCEM in obese women was 88.03% and the specificity was 94.62%. Sensitivity was 90.32% and specificity was 97.04% in obese men. Significant increase in sensitivity and specificity were detected according to SRM levels especially in obese individuals. CONCLUSION: It can be observed that the SCEM obtained from this study is successful in correcting probable errors that stem from SRM. The sensitivity and specificity analyzes of the SCEM were performed to verify its accuracy and the SCEM was found to be successful. It is considered that the SCEM will be beneficial in a populationsense rather than in an individual sense. In order to develop a SCEM that can be generalizable to Turkish population, future studies that use probability sampling methods throughout Turkey with wider samples are needed. Keywords: Bias, Body Mass Index, Obesity, Self Report, Sensitivity and Specificity