A global multicenter study on reference values: 2. Exploration of sources of variation across the countries

•Sources of variation (SVs) of reference values (RVs) was explored for 13 ethnic groups from 12 countries.•Specific sex and age profile of RVs were observed in majority of analytes regardless of countries.•BMI-related changes in RVs expressed as partial correlation coefficient (rp) were comparable across the counties.•However, the slope of regression line (x: BMI vs. y: RV) differed greatly across the ethnic groups.

ObjectivesThe intent of this study, based on a global multicenter study of reference values (RVs) for serum analytes was to explore biological sources of variation (SVs) of the RVs among 12 countries around the world.MethodsAs described in the first part of this paper, RVs of 50 major serum analytes from 13,396 healthy individuals living in 12 countries were obtained. Analyzed in this study were 23 clinical chemistry analytes and 8 analytes measured by immunoturbidimetry. Multiple regression analysis was performed for each gender, country by country, analyte by analyte, by setting four major SVs (age, BMI, and levels of drinking and smoking) as a fixed set of explanatory variables. For analytes with skewed distributions, log-transformation was applied. The association of each source of variation with RVs was expressed as the partial correlation coefficient (rp).ResultsObvious gender and age-related changes in the RVs were observed in many analytes, almost consistently between countries. Compilation of age-related variations of RVs after adjusting for between-country differences revealed peculiar patterns specific to each analyte. Judged from the rp, BMI related changes were observed for many nutritional and inflammatory markers in almost all countries. However, the slope of linear regression of BMI vs. RV differed greatly among countries for some analytes. Alcohol and smoking-related changes were observed less conspicuously in a limited number of analytes.ConclusionThe features of sex, age, alcohol, and smoking-related changes in RVs of the analytes were largely comparable worldwide. The finding of differences in BMI-related changes among countries in some analytes is quite relevant to understanding ethnic differences in susceptibility to nutritionally related diseases.