Comparing the performance of body mass index, waist circumference and waist-to-height ratio in predicting Malaysians with excess adiposity

Abstract Background Body mass index (BMI) is a widely used surrogate tool to screen for obesity/adiposity, but it cannot differentiate between lean and fat mass. Thus, alternative tools to detect excess adiposity should be identified. Aim This study aimed to compare the performance of BMI, waist circumference (WC) and waist-to-height ratio (WtHR) in predicting Malaysians with excess body fat defined by dual-energy X-ray absorptiometry (DXA). Subjects and methods A total of 399 men and women aged ≥40 years were recruited from Klang Valley, Malaysia. The body composition of the subjects, including body fat percentage, was measured by DXA. The weight, height, WC and WHtR of the subjects were also determined. Results BMI [sensitivity = 55.7%, specificity = 86.1%, area under curve (AUC) = 0.709] and WC (sensitivity = 62.7%, specificity = 90.3%, AUC = 0.765) performed moderately in predicting excess adiposity. Their performance and sensitivity improved with lower cut-off values. The performance of WHtR (sensitivity = 96.6%, specificity = 36.1, AUC = 0.664) was optimal at the standard cut-off value and no modification was required. Conclusion The performance of WC in identifying excess adiposity was greater than BMI and WHtR based on AUC values. Modification of cut-off values for BMI and WC could improve their performance and should be considered by healthcare providers in screening individuals with excess adiposity.


Introduction
Obesity is defined as excessive fat accumulation that may impair health (World Health Organization 2021). The global prevalence of obesity rises continuously. The World Obesity Federation (2022) predicted that one billion people worldwide, or 1 in 5 women and 1 in 7 men, will be affected by obesity by 2030. The prevalence of obesity in Malaysia has been increasing from 15.1% in 2011 to 19.7% in 2019, with an estimated 3.3% annual increase in adult obesity from 2010 to 2030 (National Institutes of Health Malaysia 2019). Malaysia also has the highest prevalence of diabetes and obesity among the Southeast Asian countries and carries one of the highest body mass index (BMI)-related Disability-Adjusted Life Years and deaths per 100,000 population (2274) due to noncommunicable diseases in the Western Pacific region (Ministry of Health 2016). As a chronic, relapsing, multifactorial disease, obesity is linked to many noncommunicable diseases, such as diabetes, cardiovascular diseases (CVDs) and cancers (Poirier et al. 2006). BMI is the most widely used index to define obesity and predict CVD mortality due to its simplicity, affordability and convenience (Centers for Disease Control and Prevention 2022). Despite its clinical relevance, the reliability of BMI in measuring adipose tissue and its physical distribution remains debatable (Ode et al. 2007;Nuttall 2015;Ortega et al. 2016). Aside from its inability to distinguish between fat and lean tissues (Romero-Corral et al. 2006), BMI cannot discriminate CVD risk accurately in individuals with intermediate BMI values, particularly those with normal-weight obesity (Bosomworth 2019). BMI classifies fewer individuals as obese as compared to other obesity indices (Swainson et al. 2017). Numerous studies have analysed the performance of BMI in identifying adiposity with heterogeneous results. Some studies revealed a good diagnostic performance of BMI (Ranasinghe et al. 2013), while others reported insensitivity of BMI in identifying adiposity (Akindele et al. 2016). Thus, further studies are necessary to determine the performance of BMI in detecting excessive body adiposity to justify its use in clinical practice.
Although total body fat and lean soft tissue mass can be measured directly using dual-energy X-ray absorptiometry (DXA) (Shepherd et al. 2017), the machine is limited by its accessibility. Simple anthropometric measures like waist circumference (WC) have been proposed to examine body composition and disease risk given its sensitivity towards body size, body fat percentage and fat distribution (World Health Organization 2008). It is most useful in individuals with normal and overweight BMI. Waist-to-height ratio (WHtR) is another recommended simple abdominal obesity indicator (Ashwell and Gibson 2016). WHtR is defined as WC in centimetres divided by height in centimetres. BMI, WC, and WHtR have been suggested as tools to evaluate obese patients with excess fat mass percentage (FM%) defined by DXA (Moltrer et al. 2022).
It is essential to approach obesity in a regional context as the current World Health Organisation criteria to classify obesity in European adults may not be accurate for Malaysians. As health risks associated with obesity occur at a lower BMI in Asian populations, some studies have recommended a lower BMI value for the Asian populations (Deurenberg-Yap Mabel et al., 1999;Hsieh et al. 2000;Lin et al. 2002). Ahmad et al. (2016) suggested WC as a better obesity indicator because the prevalence of abdominal obesity using WC was higher than using waist-hip ratio. A crosssectional study by Zaher et al. (2009) demonstrated a lower cut-off value of BMIs and WCs for defining overweight or obesity among Malaysian adults. The study also supported the use of WC in predicting obesity-related CVD risk factors compared to BMI. On the other hand, Jayvikramjit Singh et al. (2020) reported better performance of WHtR in screening health risks earlier compared to BMI and WC. Considering these heterogeneous results, this study aimed to compare the performance of the BMI, WC, and WHtR in identifying Malaysian adults with excess adiposity. The optimal cut-off levels of these three indices to define excess adiposity in Malaysian adults were also determined.

Subjects and methods
This cross-sectional study was conducted from April 2018 to January 2019. A quota sampling approach was used to recruit subjects at the screening centre at the Universiti Kebangsaan Malaysia Medical Centre and community centres in Kuala Lumpur, Malaysia. Participants were Malaysians aged !40 years residing in Kuala Lumpur and its environs. The inclusion and exclusion criteria have been summarised in previous publications (Chan, Subramaniam, Chin, Ima-Nirwana, Muhammad, Fairus, Mohd Rizal, et al. 2019;Chan, Subramaniam, Chin, Ima-Nirwana, Muhammad, Fairus, Ng, et al. 2019;Subramaniam et al. 2019;Chan et al. 2020). Briefly, the main study was a bone health study so subjects with strong risk factors of bone loss, recent previous fractures and a previous diagnosis of osteoporosis were excluded. However, subjects with metabolic disorders, such as obesity, hypertension, hyperglycaemia and hyperlipidaemia, were not excluded. The study protocol had been reviewed and approved by the Ethics Committee of Universiti Kebangsaan Malaysia Medical Centre (approval code: UKM PPI/111/8/JEP 2017-721). Subjects were briefed on the study and informed consent was obtained before their enrolment.
A standardised questionnaire was used to collect subjects' sociodemographic and lifestyle details. Subjects' sociodemographic details, including sex (male/female), date of birth and ethnicity (Malay/Chinese/Indian or others), were selfdeclared. For anthropometric measurements, the standing height of the subjects without shoes was measured using a stadiometer (Seca, Hamburg, Germany) and recorded to the nearest 1 cm. The body weight of the subjects with light clothing and without shoes was determined using a weighing scale (Tanita, Tokyo, Japan) and was recorded to the nearest 0.1 kg. The BMI of the subjects was calculated based on the formula: body weight in kg divided by the square of height in metres (Centers for Disease Control and Prevention 2022). Individuals with BMI <18.5 kg/m 2 were classified as underweight, 18.5-24.9 kg/m 2 as normal, 25.0-29.9 kg/m 2 as overweight and >30 kg/m 2 as obese. Obesity is further categorised into obese I (30.0-34.9 kg/m 2 ), obese II (35.0-39.9 kg/ m 2 ) and obese III (!40 kg/m 2 ) (Weir and Jan 2021). WC was measured between the lowest rib margin and the iliac crest using a soft measuring tape and recorded to the nearest 0.1 cm while subjects maintained a standing position. A WC !90 cm in men and !80 cm in women is associated with an increased risk of comorbidities (World Health Organization 2008). WHtR was calculated as WC in centimetres divided by height in centimetres (Moosaie et al. 2021). A WHtR ratio <0.5 indicates low risk, 0.5-0.6 indicates increased risk and >0.6 indicates high risk for CVD and diabetes (Ashwell and Gibson 2016). The body composition of the subjects in supine position was measured using a DXA machine (Discovery QDR Wi, Hologic, MA, USA) by a single trained technician during the study period. The machine was calibrated daily using a phantom following the manufacturer's instructions.

Statistical analysis
Data were analysed using Statistical Package for the Social Sciences version 27 (SPSS Inc, Chicago). Descriptive statistics were used to summarise the characteristics of the study population. Means and standard deviations (SD) were used to express continuous variables or frequencies and percentages to express categorical variables. The normality of the data was determined using the Kolmogorov-Smirnov test. The agreement between each adiposity index and body fat % was analysed using Cohen's kappa. The kappa value is interpreted as poor (<0 points), slight (0.01-0.20), fair (0.21-0.40), moderate (0.41-0.60), substantial (0.61-0.80), or almost perfect (0.81-1) agreement. Receiver operating characteristic (ROC) curves were plotted and the area under the curve (AUC) was conducted with a 95% confidence interval (95% CI) to assess the performance of anthropometric indicators (BMI, WC, and WHtR) to predict adiposity as defined by body fat percentage >25% in men and 35% in women. AUC was compared between men and women for each index. In general, an AUC of 0.5 suggests no discrimination, 0.7-0.8 is considered acceptable, 0.8-0.9 is considered excellent, and more than 0.9 is considered outstanding. The Youden index (sensitivity þ specificity À 1) was calculated to determine the ideal cut-off point for each anthropometric index in predicting adiposity.
Using the standard BMI cut-off (>25 kg/m 2 ) to detect excess adiposity (body fat percentage >25% for men; >35% for women), the ROC analysis rendered a sensitivity value of 55.7%, specificity value of 86.1% and area under curve (AUC) of 0.709 (95% CI: 0.649 À 0.769). Based on the Youden Index, optimal cut-off values of BMI in detecting excess adiposity were !24.2 kg/m 2 in men and !22.2 kg/m 2 in women. The new BMI cut-off values improved AUC to 0.792 (95% CI: 0.734 À 0.850), sensitivity to 76.5%, but reduced the specificity to 81.9%. The agreement between the new BMI cut-off values and body fat percentage based on DXA in defining adiposity also improved from 0.241 to 0.434 (p < 0.001; Table 2).
Using the standard WC cut-off values (men !90 cm and women !80 cm) to detect excess adiposity, the ROC analysis revealed a sensitivity of 62.7%, specificity of 90.3% and AUC of 0.765 (95% CI: 0.711 À 0.819). Based on the Youden Index, the optimal WC cut-off values to identify excess adiposity were ! 85.0 cm in men and !76.0 cm in women, with a sensitivity of 80.7%, specificity of 73.6% and AUC of 0.772 (0.708 À 0.836; Table 2).
Using the standard WHtR cut-off values (!0.5) to detect excess adiposity, the ROC analysis revealed a sensitivity of 96.6%, specificity of 36.1% and AUC values of 0.664 (95% CI: 0.585 À 0.743). Based on the Youden Index, the optimal WHtR cut-off was 0.50, consistent with the conventional cutoff values (Table 2).

Discussion
The current study revealed that lower BMI (!24.2 kg/m 2 in men and !22.2 kg/m 2 in women) and WC (!85.0 cm in men and !76.0 cm in women) cut-off values would increase the performance and sensitivity of these tools in detecting subjects with excess adiposity. However, the standard WHtR cut-off value (!0.5) is optimal in detecting subjects with excess adiposity.
While BMI remains the most used tool to measure adiposity, the current findings demonstrated that the standard BMI cutoffs for overweight/obese (>25 kg/m 2 ) might underestimate the condition. A previous study in Singapore with population demography similar to Malaysia reported that at a given BMI, the Asian population had a higher amount of body fat compared to the Caucasians (Deurenberg-Yap et al. 2000). Another study by Aizuddin indicated that a lower BMI cut-off (>24.8 kg/m 2 ) was better in detecting Malaysian subjects with increased adiposity, but the body fat in the subjects of their study was measured using the bioelectric impedance method (Aizuddin et al. 2021). The current findings echoed the recommendation of an earlier meta-analysis that ethnic-specific BMI cut-offs are needed to define obesity due to the difference in body fat percentage (Deurenberg et al. 1998).
The current study reported that the performance of WC at standard cut-off values was better than BMI and WHtR in predicting excess adiposity in Malaysians based on AUC values. This observation concurred with the findings of Zaher et al. (2009), wherein WC was better than BMI in predicting obesity-related CVD risk factors among Malaysians. However, another study showed that WC and BMI showed similar performance in predicting metabolic syndrome among Malaysians (Cheong et al. 2015). The current study also revealed that lowering the WC cut-offs to !85.0 cm in men and !76.0 cm in women improved its performance in predicting subjects with excess adiposity. Using data from the National Health Morbidity Survey in 2006, Kee et al. (2011) showed that the optimal WC to detect overweight subjects (BMI !25 kg/m 2 ) was 86.0 cm for men, and 79.1 cm for women in Malaysia. These values are close to values reported in the current study, despite the benchmark they used being defined by BMI. In a similar study, Ahmad et al. (2016) found the optimal WC cut-offs to be 92.5 cm in men and 85.5 cm in women for abdominal obesity. WC is a better predictor of excess adiposity because it can reflect central obesity, unlike BMI which does not differentiate body fat distribution.
A systematic review supported WHtR as a predictor of diabetes, CVD and related risk factors, and the use of cut-off value !0.5 across different populations (Browning et al. 2010). The findings of this study are congruent with the Malaysian Ministry of Health's recommendation to use WHtR at 0.5 to identify Malaysians at increased metabolic health risk regardless of race, gender and age (Ministry of Health 2016). Earlier studies in Malaysia reported that WHtR demonstrated excellent performance in detecting hypertension among adolescents (Tee et al. 2020) and metabolic syndrome among vegetarians (Ching et al. 2020). However, the performance of WHtR was lower than BMI and WC in predicting excess adiposity among Malaysians.
Body fat accumulation is associated with numerous health consequences. However, tools for direct measurement of body fat, such as DXA and computed tomography, are not easily accessible (Duren et al. 2008), thus necessitating surrogate tools to identify patients at risk for excess adiposity. These tools could aid in the timely management of patients with a high cardiometabolic risk profile. However, inaccurate cut-off values of a screening tool would lead to misdiagnosis of excess adiposity, subsequently delaying further risk assessment and management. The current study is important because it showed that at the current cut-offs, WC performed better than BMI and WHtR in detecting excess adiposity. Besides, a lower BMI and WC cut-offs are better in identifying Malaysians at risk of excess adiposity. This study has some limitations. This is a cross-sectional study and the findings should be justified in a well-designed cohort study. The participants were Malaysian adults and thus the optimal cut-off values may not be applicable in other countries in view of the sociodemographic difference. The subjects recruited were aged 40 years and above, thus the findings might not be applicable to younger Malaysians. Sub-analyses based on age group and ethnicity were performed but the number of subjects was very limited to derive a valid result (Supplementary Table S1). Since the study is limited to BMI, WC, and WHtR only, it is suggested that more extensive research is conducted on evaluating different algorithms to identify better indices to detect excess adiposity. Nevertheless, the current findings are important because they suggest lowering the cut-offs of BMI and WC to optimise their performance in detecting adiposity, which is a major cause of mortality and morbidity associated with CVDs.

Conclusions
At the standard cut-off values, WC performed better than BMI and WHtR in identifying Malaysians with excess adiposity. The performance of BMI and WC in adiposity detection could be improved by lowering the standard cut-off values, while WHtR performed optimally at the current cut-off values. Healthcare providers who wish to use these indices in their practices should consider modifying the cut-offs for BMI and WC in detecting individuals with excess adiposity.