Fuzzy risk factors in case–control studies; an example of relationship between obesity and hypertension based on National Surveillance of Risk Factors of Non-Communicable Diseases-2007 Data

Abstract

Background & Aim: There are some ambiguities in assessment of associations between continuous risk factors and different health outcomes usually from different cut points. Data loss, near to the cut point values different categorization, and no real definition of risk are important limitations for usual odds ratios (ORs). Fuzzy method considers a specific membership function for all numbers in range of the variable and can provide a similar OR. In this study, we used a large data set for these different measures calculation and making a comparison between them according to their privileges.
Methods & Materials: The study was conducted on noncommunicable diseases risk factors surveillance data set (National Surveillance of Risk Factors of Non-Communicable Diseases-2007) with regard to obesity and abdominal (central) obesity as risk factors for hypertension according to a “fuzzy risk factor” approach and usual approach based on regular cut points in different literature. OR of chances to have hypertension calculated by both methods and compared with each other.
Results: ORs with usual and fuzzy methods of calculations had similarities and some differences in amount, confidence interval and confidence length. With different cut points (for waist circumference), variation between different calculations was high. Fuzzy OR was more sensitive and resistant to minor change in individual data than the others.
Conclusion: OR{Fuzzy} measures the association of exposure to risk factors with different outcomes in a closer form of clinical reality with no dependency to any cut point selection, less variability and more resistance to data variation and can be suggested as a good estimator.