WHR had the strongest association with DMT2, followed by WHtR, WC and BMI. The predictable values differed between men and women, showing lower cut-offs for women for WHR and WC. The potential confounders adjusted for in the regression models (sex, age and smoking) did not to affect the OR statistically except in a few cases. Notably, OR in the unadjusted model was highest for WHtR, but in the adjusted model higher for WHR, which in total placed WHR as the strongest predictor for DMT2.
While the cross-sectional design of this study does not make it possible to draw any conclusions about causality, it still provides valuable insight in the association between anthropometric indices and DMT2. The result suggests that we need to be aware of the different predictive potentials of anthropometric indices for screening prevalent DMT2.
This result is in line with what a previous meta-analysis has concluded [20]. The prevalence of DMT2 in the sample (3.9%) is slightly lower than figures mentioned in the current literature [19]. Selection bias could however explain this difference. A high proportion of individuals were categorized either overweight or obese in the sample (67.3%) with and an average BMI of 27.2.
The AUC suggests that the predictive potential of WHR is highest in this sample, followed by WHtR, WC and BMI. The order is the same for both men and women. This result is in line with data from systematic reviews, even though AUC values were higher in this study than previously reported in other studies [11, 20]. All the tested anthropometric indices worked statistically significantly better for women compared to men. Whether this could be explained by the difference in anthropometry between men and women or by a potential bias is not clear.
The optimal cut-off values presented in this study are similar to what previous studies have reported for men in a Caucasian-white population and the proposed universal recommendations suggested by WHO [13, 17]. However, the present result indicates that for Norwegian women, the cut-offs for WHR and BMI should be higher than what other studies have suggested for an European populations[13]. This could possibly be explained by the fact that Norwegian women are in average taller than central and south Europeans [21]. The present result indicates that the difference between men and women could be small for BMI and WHtR, but that men should have a higher cut-off point than women for WC and WHR. This is in line with the current litterateur even though the difference in WHR seem to be smaller for the Norwegian population than previously suggested for a general Caucasian-white population [22].
The suggested optimal cut-off estimates found based on this large population based sample, should be taken into considiration when revising national guidelines for Norway respectivly, and caution should be exercised when applying universal cut-offs for WC and BMI since these differ from the findings in this country specific sample.
To our knowledge, there are no studies that have investigated, the predictive potential of anthropometric indices for DMT2 on a large population-based sample in Norway, and there is currently incoherent evidence regarding the association of anthropometric indices and prevalent DMT2 presented in the literature. One strength of this study is the large population-based sample with standardized measurements that increase the generalizability of the results.
In the regression model, potential confounders from the available variables were applied (sex, age, and smoking). The regression analysis showed that the covariates did not have a statistical impact on the association in for the majority of cases, therefore they were not included in the ROC curve analysis and were only stratified for sex. This study did not include confounders like physical activity and diet due to the reliability of this data that often have substantial limitation when self-reported.
4.1 Conclusions
Our study confirms that WHtR, WHR, WC and BMI perform differently as predictive tools for prevalent DMT2 in the Norwegian population. The result contributes to clarify which overweight index are strongest associated with prevalent DMT2 and suggest that WHR may provide the best index to detect DMT2 in Norway. This study also indicates that anthropometric indices detect DMT2 better in Norwegian women than men. The gender difference should be considered when screening for DMT2 in a Norwegian population. We have concluded that the currently mostly used anthropometric index, BMI has not the strongest association with DMT2 compared to WHR, WC and WHtR which is important to consider when designing screening programs for DMT2 in a Norwegian population.