Obesity Among Postmenopausal Women

What Is the Best Anthropometric Index to Assess Adiposity and Success of Weight-loss Intervention?

Ahmed Ghachem, PhD; Alexis Marcotte-Chénard, MSc; Dominic Tremblay, MSc; Denis Prud'homme, MD, MSc; Rémi Rabasa-Lhoret, MD, PhD; Eléonor Riesco, PhD; Martin Brochu, PhD; Isabelle J. Dionne, PhD

Disclosures

Menopause. 2021;28(6):678-685. 

In This Article

Discussion

The main objective of the present study was to perform external validation of RFM, as a new proxy of BF% measured by DXA, which is widely used in clinical research. Specifically, we aimed to compare three anthropometric indexes (RFM vs BMI and WC), for their ability to: 1) correctly identify postmenopausal women living with obesity; and 2) predict BF% to assess success of weight-loss interventions in a subpopulation generally affected by obesity and cardiometabolic issues, for whom a lifestyle change intervention can be recommended.

Our first hypothesis was that RFM would be more accurate to correctly identify obese and nonobese postmenopausal women compared with the BMI and WC. Overall, our results showed that while the ability of RFM to correctly identify women living with obesity was similar to WC, BMI showed the best performance (area under curve [AUC]; sensitivity + specificity), with the lowest misclassification error (6%) compared with RFM (40%) and WC (27%) (data not shown). These results are not in agreement with those of Woolcott and Bergman in 2018 showing that for all age, sex, and ethnicity subgroups, obesity total misclassification was lower with RFM than with BMI. Some methodological key points may partly explain the discordance between both studies. For example, in the study of Woolcott and Bergman in 2018, women living with obesity (cut-off: BF% ≥ 33.9) were identified based on arbitrary threshold corresponding to the cutoff between the first and the second quintiles of BF%. Given that there is no consensus on the diagnosis of obesity based on BF%, in our study, we decided to use the WHO (1995) cutoffs (BF% > 35) to identify women living with obesity, which makes our results more comparable with previous and future studies. Subanalyses showed however that such a difference in obesity classification cutoffs could change the ROC analysis results such as AUC, sensitivity, and specificity as well as the percentage of obesity misclassification (data not shown). Nevertheless, even if we used the same methodology as Woolcott and Bergman in 2018 to identify women living with obesity, BMI still had higher sensitivity compared with RFM and WC (94.5% vs 53.6% vs 76.5%, respectively) (data not shown).

Our second hypothesis was that RFM would be more accurate to predict BF% measured by DXA before and after a weight-loss intervention than BMI and WC. Results of the present study refute this hypothesis. Indeed, RFM consistently showed inferior performance to predict BF% measured by DXA before and after a weight-loss intervention in postmenopausal women compared with BMI and WC. RFM consistently underestimated BF% in this subpopulation (based on bias from Bland and Altman plots). Actually, all performance indicators suggest that BMI is the best predictor of BF% measured by DXA before and after a weight-loss intervention. Our results are not in line with those from Woolcott and Bergman in 2018 showing that RFM was more accurate than BMI to predict BF% in different sex, age, and ethnicity subgroups.

While it is difficult to explain these discrepancies, some studies suggest that characteristics of the population (sex, age, and level of fat mass) could impact (underestimate or overestimate) BF% estimated by DXA which could consequently influence the performance of anthropometric indexes to predict BF%.[30] Compared with the population studied by Woolcott and Bergman in 2018, women from our study had higher (35.85 ± 9.94 vs 30.8 ± 0.3 kg) absolute whole-body fat. Our results are however in agreement with findings from other studies aimed to validate RFM in other subpopulations.[17–19] Results from these studies showed that BMI is more accurate to predict BF% (higher r2) and correctly identify women living with obesity (higher AUC, sensitivity, and specificity) than RFM. These results could be explained by several factors. First, it has been shown that the association is stronger between BF% and BMI in women, and between BF% and WC in men.[31] Second, the sex differences in body composition could partly explain these discrepancies.[32] Indeed, it has been shown that men store fat mass mainly at the abdominal level, which is well represented by WC measurement, while in women, during and after menopause, decrease in estrogen leads to an overall increase in total and central body fat, which is better represented by BMI.[3,32] Furthermore, it has been shown that body weight is more correlated with fat mass in women compared to men while this relationship was modulated by muscle mass among men.[33,34] Ultimately, although our results seem to disagree with those of Woolcott and Bergman in 2018, in reality this is not totally true. In fact, when closely looking at results from the study of Woolcott and Bergman in 2018, we notice that the difference in performance between RFM and BMI is subtle and seems to be nonsignificant in women, particularly among those aged 60 years and older. Thus, the conclusion of Woolcott and Bergman in 2018 should be interpreted with caution, at least for older women.

Taken together, results from the present study suggest that, despite the promising findings from Woolcott and Bergman in 2018 showing a high performance of RFM to predict whole BF% and reduce total obesity misclassification compared to BMI, its performance in postmenopausal women seems to be inferior compared to BMI and WC, both before and after a weight-loss intervention. Finally, findings from this study have substantial implications for clinical practice. Although WC might be more appealing in the clinical setting as it requires only a single measurement, the use of BMI with an optimal threshold is better to assess adiposity. For example, our results suggest that a BMI threshold of ~26 kg/m2 showed a better diagnostic accuracy to correctly identify postmenopausal women living with obesity, which is lower than the currently used ≥ 30 kg/m2 (Figure 6). These results support findings from previous studies showing that a BMI ≥ 25 kg/m2 displayed a moderately high sensitivity and specificity to correctly identify older adults living with obesity compared to a BMI ≥ 30 kg/m2.[10,35] Thus, for routine clinical assessment, the use of a BMI threshold ≥ 25 kg/m2 to assess obesity among postmenopausal women may be more judicious. However, although it is well established that overweight and obese individuals (BMI ≥ 25 kg/m2) have high risk of mortality compared to those with healthy weight (BMI between 18.5 and 24.9 kg/m2),[36–40] this relationship remains unclear for older adults. In this sense, some studies have shown that being overweight (BMI between 25 and 29.9 kg/m2) was not found to be associated with an increased risk of mortality in men and women aged 65 years and older, and mortality risk began to increase for BMI > 33 kg/m2.[41,42] Taken together, these suggest that a BMI threshold of 25 kg/m2 may be optimal to assess adiposity in postmenopausal women, but it could appear overly restrictive to assess obesity associated risk of death. Given that this hypothesis is not part of the objective of the present study, further work is needed to respond this question. Hence, despite criticism relative to the use of BMI and WC, such anthropometric indexes remain simple, inexpensive, and noninvasive screening tools with good accuracy to predict BF%. Furthermore, despite their often-reported lack of diagnostic accuracy to correctly identify middle-aged and older men and women living with obesity, we believe that such measures should still be used in clinical settings in the absence of more objective tools such as DXA, CT-scan, or magnetic resonance imaging, which are costly for routine assessment.

Some limitations of this study should be noted. All women were White; thus, our findings cannot be generalized to other ethnic groups and need to be replicated in large multiethnic populations. Also, our results concern postmenopausal women living with obesity, thus further studies are needed to confirm these observations among overweight and postmenopausal women. Despite these limitations, the present study is strengthened by the use of DXA as a reference method to assess adiposity and the use of robust statistical performance indicators to compare between models. To the best of our knowledge, the present study is the first to investigate the validity and the usefulness of the new proxy of BF%, RFM index, first, to estimate BF% to assess success of weight-loss intervention, and second, to correctly identify postmenopausal women living with obesity, compared to the most widely used indexes, BMI and WC.

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....