New BMI Cut-Off Points for Obesity in Middle-Aged and Older Adults in Clinical Nutrition Settings in Italy: A Cross-Sectional Study

Obesity is a major health problem defined as an excess accumulation of body fat (BF). The World Health Organization (WHO) usually relies on a body mass index (BMI) ≥ 30 kg/m2 as an indicator of obesity. Due to changes in body composition that occur across the lifespan, with an increase in BF and a decrease in lean mass, we aimed to test the validity of this BMI cut-off point for adiposity in middle-aged and older adults. This cross-sectional study, composed of 4800 adults of mixed gender aged between 40 and 80 years, included (according to the WHO BMI classification) 1087 normal-weight, 1826 overweight, and 1887 obese individuals who were referred to the Department of Biomedicine and Prevention, University of Rome “Tor Vergata”, Italy. The sample was then categorized by adiposity status based on the total BF% as measured by dual-energy X-ray absorptiometry (DXA), and the best sensitivity and specificity were attained for predicting obesity according to the receiver operating characteristic curve (ROC) analysis. In a real-world clinical setting, a new BMI cut-off point (BMI = 27.27 kg/m2) has been identified for predicting obesity in middle-aged and older adults. Obesity guidelines in Italy therefore need to be revised accordingly.


Introduction
Ageing causes changes in body composition [1], such as a decline in lean mass (i.e., muscle), which decreases by approximately almost 5% each decade after the age of 30 [2], and this may lead to an overall reduction of the latter by nearly 20% between the ages of 30 and 70 years [3]. On the other hand, body fat (BF) increases across the lifespan [4], to reach its peak in middle age. The maximum amount of BF has been observed between 50 and 60 years, and that of BF% between 55 and 70 years [5]. These changes usually occur without significant alterations in body weight status or body mass index (BMI) [6]. In this context, obesity become a serious health problem with an increasing prevalence over the past three decades, currently affecting more than 650 million adults worldwide. Furthermore, since it is considered a significant risk factor for several medical [7,8] and psychosocial morbidities [9][10][11][12], as well as increased mortality rates [13], accurate screening for obesity in the early stages is vital [14].
Obesity is best defined as excessive fat deposition in the adipose tissue [15,16]; its identification based on BF quantification seems to be the most accurate method [17]. When classifying obesity in adults, the World Health Organization (WHO) mainly relies on a This single-center cross-sectional observational study was conducted in adherence to the strengthening the reporting of observational studies in epidemiology (STROBE) guidelines (STROBE checklist in Supplementary File) [27]. The participants were selected from a large pool of patients whom general practitioners referred and subsequently enrolled in the Division of Clinical Nutrition at the Department of Biomedicine and Prevention, University of Rome "Tor Vergata", Italy, during the period between June 2018 and May 2022. Patients were considered eligible to participate if they were adults and had completed a body composition measurement using dual-energy X-ray absorptiometry (DXA). A total of 8722 patients were considered eligible and were checked for inclusion and exclusion criteria. The inclusion criteria included being between 40 and 80 years old and therefore defined as middle aged or elderly, with a BMI ≥ 20 kg/m 2 . Patients were excluded if they were pregnant (among females), taking medication that affects body weight or composition, or presented with medical comorbidities associated with weight loss (i.e., cancers) or severe psychiatric disorders at the baseline assessment. A total of 4800 individuals of different genders and different body weight statuses according to the WHO BMI classificationnormal weight (n = 1087), overweight (n = 1826), or obesity (n = 1887)-were included. The research was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of the Calabria Region Center Area Section (Register Protocol No. 146 17/05/2018). All patients' personal data were treated according to European/Italian privacy laws, and informed written consent was obtained.

Body Weight and Height
Body weight and height were measured with the participants wearing light clothes and no shoes, using an electronic weighing scale (SECA 2730-ASTRA, Hamburg, Germany) and a stadiometer. BMI was then calculated according to the standard formula of body weight measured in kilograms divided by the square of the height in meters.

Body Composition
Body composition was determined using a DXA (Primus, X-ray densitometer; software version 1.2.2, Osteosys Co., Ltd., Guro-gu, Seoul, Republic of Korea) fan beam scanner. It assessed both whole and segmental composition regarding fat and lean mass. Patients were given complete and standardized instructions on the testing procedure, as described elsewhere [25]. The sample was categorized according to age-and gender-specific obesity cut-off points as follows [21]: • 40-59 years: BF% ≥ 40% for females and BF% ≥ 28% for males. • 60-79 years: BF% ≥ 42% for females and BF% ≥ 30% for males.

Statistical Analysis
Descriptive statistics are presented as means and standard deviations for continuous variables and frequencies, with proportions for categorical variables. Pearson's correlation coefficient and scatter plots were used to assess the association between BF% and BMI. A cumulative sum (CUSUM) linearity test was conducted to confirm a linear association between BMI and BF% [28]. A p-value > 0.05 for the CUSUM test indicated a positive linear association. To evaluate the diagnostic performance of BMI in detecting obesity status defined by BF% for all subjects by age and gender, a classification analysis was performed by calculating sensitivity and specificity as well as the area under the curve (AUC) of the receiver operating characteristic curve (ROC). For this purpose, the definition of obesity based on age-and gender-specific BF% as indicated above was used as a gold standard [21]. The criterion value of BMI with maximum sensitivity and specificity was selected for the age-and gender-specific BMI cut-off points. An AUC > 0.8 indicates that the criterion value has an excellent discriminating ability [29]. All values were considered significant at p < 0.

Results
The study sample comprised 1850 (38.5%) males and 2950 (61.5%) females, with a mean age of 54.8 ± 9.7 years and 54.0 ± 9.1 years, respectively. The mean BMI was 29.3 ± 4.5 kg/m 2 in males and 28.8 ± 4.9 kg/m 2 in females, with almost 40% of both groups categorized as obese (38.4% vs. 40.6%) based on the WHO criteria. However, according to the classification based on BF%, almost two-thirds of the males (71.1%) and females (63.6%) were obese (Table 1). Table 1. Anthropometric and body composition characteristics of the study sample (n = 4800) Body composition was determined using a DXA (Primus, X-ray densitometer; software version 1.2.2, Osteosys Co., Ltd., Guro-gu, Seoul, Korea) fan beam scanner. It assessed both whole and segmental composition regarding fat and lean mass. Patients were given complete and standardized instructions on the testing procedure, as described elsewhere [25]. The sample was categorized according to age-and gender-specific obesity cut-off points as follows [21]: • 40-59 years: BF% ≥ 40% for females and BF% ≥ 28% for males. • 60-79 years: BF% ≥ 42% for females and BF% ≥ 30% for males.

Statistical Analysis
Descriptive statistics are presented as means and standard deviations for continuous variables and frequencies, with proportions for categorical variables. Pearson's correlation coefficient and scatter plots were used to assess the association between BF% and BMI. A cumulative sum (CUSUM) linearity test was conducted to confirm a linear association between BMI and BF% [28]. A p-value > 0.05 for the CUSUM test indicated a positive linear association. To evaluate the diagnostic performance of BMI in detecting obesity status defined by BF% for all subjects by age and gender, a classification analysis was performed by calculating sensitivity and specificity as well as the area under the curve (AUC) of the receiver operating characteristic curve (ROC). For this purpose, the definition of obesity based on age-and gender-specific BF% as indicated above was used as a gold standard [21]. The criterion value of BMI with maximum sensitivity and specificity was selected for the age-and gender-specific BMI cut-off points. An AUC > 0.8 indicates that the criterion value has an excellent discriminating ability [29]. All values were considered significant at p < 0.

Results
The study sample comprised 1850 (38.5%) males and 2950 (61.5%) females, with a mean age of 54.8 ± 9.7 years and 54.0 ± 9.1 years, respectively. The mean BMI was 29.3 ± 4.5 kg/m 2 in males and 28.8 ± 4.9 kg/m 2 in females, with almost 40% of both groups categorized as obese (38.4% vs. 40.6%) based on the WHO criteria. However, according to the classification based on BF%, almost two-thirds of the males (71.1%) and females (63.6%) were obese (Table 1). curve (ROC). For this purpose, the definific BF% as indicated above was used as a with maximum sensitivity and specificity MI cut-off points. An AUC > 0.8 indicates inating ability [29]. All values were conruncher Statistical Systems (NCSS) 12.0.2 was used for the statistical analysis. A as determined with Power Analysis and , Kaysville, UT, USA). For the sample of tegorized as not, with an alpha of 0.05 and ) males and 2950 (61.5%) females, with a rs, respectively. The mean BMI was 29.3 ± ales, with almost 40% of both groups catthe WHO criteria. However, according to thirds of the males (71.1%) and females racteristics of the study sample (n = 4800) Ŧ . Values are presented as means (SD) for continuous variables and n (%) for categorical variables. BMI-body mass index; BF-body fat. £ WHO classification for BMI. ¥ Age-and gender-specific obesity cut-off points according to BF%.

Males
The correlation analysis revealed a significant positive correlation between BMI and BF% in the overall sample of males (ρ = 0.709, p-value < 0.0001) and females (ρ = 0.741, p-value < 0.0001). A significant positive linear association between BMI and BF% (Figure 1a  The results of the ROC analysis on the diagnostic performance of BMI by age group and gender are shown in Table 2  either overweight (34.7%) or obese (37.5%) based on the WHO criteria. Alternatively, almost two-thirds (64.0%) of females in this age group were categorized as obese according to BF% classification ( Table 1). The ROC analysis among females aged 40-59 showed that the most appropriate BMI cut-off point for identifying obesity based on BF% was 27.03 kg/m 2 . The BMI cut-off point for this group achieved high sensitivity (80.69%) and specificity (83.63%), indicating a low chance of false negatives and false positives ( Table 2). A comparison of the distribution of obesity determined by BF and the WHO criteria revealed a disagreement, with only 56.2% of those defined as obese based on BF% being correctly classified according to the WHO criteria, thus missing almost half of those who are obese (Table 3). After comparing the proportion correctly diagnosed by the WHO cut-off point (56.2%), the identified cut-off for females aged 40-59 years improved the detection of obesity by nearly 25% (24.5%), reaching 80.7% ( Table 3). The AUC (0.90) depicts the excellent discriminating ability of BMI, which has a 90% chance of detecting obesity (Table 2, Figure 2a).  The females in age group 2, ranging between 60 and 79 years, had a mean age of 66.6 ± 5.2 years and a mean BMI of 29.4 ± 4.9 kg/m 2 , with 40% being overweight (40.8%) or obese (41.0%) based on the WHO classification criteria. Alternatively, almost two-thirds (62.3%) of females in this age group were identified as obese based on BF% ( Table 1). The ROC analysis among females aged 60-79 demonstrated that the most appropriate BMI cut-off point for diagnosing obesity based on BF% was 27.11 kg/m 2 . The BMI cut-off point for this group achieved high sensitivity (85.06%) and relatively lower specificity (74.29%), indicating a low chance of false negatives and false positives ( Table 2). A comparison of the distribution of obesity determined by BF and the WHO criteria revealed a disagreement, with only 59.5% of those defined as obese based on BF% being correctly classified according to the WHO criteria, thus missing almost 40% of those who are obese (Table 3). After comparing the proportion correctly diagnosed by the WHO cut-off point (59.5%), the identified cut-off for females aged 60-79 improved the detection of obesity based on BF by 25% (25.3%), reaching 84.8% (Table 3). The AUC (0.87) indicates the excellent discriminating ability of BMI, which has an 87% chance of detecting obesity (Table 2, Figure 2b).
The age of the whole female sample ranged between 40 and 79 years, with a mean age of 54.0 ± 9.1 years and a mean BMI of 28.8 ± 4.9 kg/m 2 , with one-third either being overweight (36.3%) or obese (38.4%) according to the WHO classification criteria. Alternatively, almost two-thirds (63.6%) of females in this age group were identified as obese based on BF% ( Table 1). The ROC analysis among females aged 40-79 demonstrated that the most appropriate BMI cut-off point for diagnosing obesity according to BF% was 27.08 kg/m 2 . The BMI cut-off point for this group achieved high sensitivity (81.66%) and specificity (81.19%), indicating a low chance of false negatives and false positives ( Table 2). Comparing the distribution of obesity determined by BF and the WHO criteria revealed a disagreement, with only 57.0% of those defined as obese based on BF% being correctly classified according to the WHO criteria, thus missing almost 40% of those who are obese (Table 3). After comparing the proportion correctly diagnosed by the WHO cut-off point (57.0%), the identified cut-off for females aged 40-79 years improved the detection of obesity according to BF by 25% (24.7%), reaching 81.7% (Table 3). The AUC (0.89) illustrates the excellent discriminating ability of BMI, which has an 89% chance of detecting obesity (Table 2, Figure 2c).
The males in age group 1, ranging between 40 and 59 years, had a mean age of 49.5 ± 5.7 years and a mean BMI of 29.3 ± 4.5 kg/m 2 , with 40% being overweight (40.5%) or obese (40.6%) according to the WHO criteria. Alternatively, over two-thirds (71.7%) of males in this age group were identified as being obese based on BF ( Table 1). The ROC analysis among males aged 40-59 demonstrated that the most appropriate BMI cut-off point for diagnosing obesity according to BF% was 27.36 kg/m 2 . The BMI cut-off point for this group achieved high sensitivity (80.0%) and good specificity (79.40%), indicating a low chance of false negatives and false positives ( Table 2). Comparing the distribution of obesity determined by BF and the WHO criteria revealed a disagreement. Only 54.4% of those defined as obese based on BF% were correctly classified according to the WHO criteria, thus missing almost half of those who are obese (Table 3). When comparing the proportion correctly diagnosed by the WHO cut-off point (54.4%), the identified cut-off for males aged 40-59 improved the detection of obesity based on BF by 25% (25.6%), reaching 80.0% (Table 3). The AUC (0.88) indicates the excellent discriminating ability of BMI, which has an 88% chance of detecting obesity (Table 2, Figure 2d).
The males in age group 2, ranging between 60 and 79 years, had a mean age of 66.8 ± 5.1 years and a mean BMI of 29.2 ± 4.2 kg/m 2 , with 40% being overweight (41.8%) or obese (41.1%) based on the WHO classification criteria. Alternatively, more than twothirds (70.3%) of males in this age group were categorized as obese according to BF% (Table 1). The ROC analysis among males aged 60-79 showed that the most appropriate BMI cut-off point for diagnosing obesity defined by BF% was 27.25 kg/m 2 . The BMI cut-off point for this group achieved high sensitivity (81.11%) and relatively low specificity (74.40%), indicating a low chance of false negatives and a relatively low chance of false positives ( Table 2). Comparing the distribution of obesity determined by BF% and the WHO criteria revealed a disagreement, with only 55.4% of those defined as obese based on BF% being correctly classified according to the WHO criteria, thus missing almost half of those who are obese (Table 3). After comparing the proportion correctly diagnosed by the WHO cut-off point (55.4%), the identified cut-off for males aged 60-79 improved the detection of obesity according to BF% by 25% (25.7%), reaching 81.1% (Table 3). The AUC (0.85) indicates the excellent discriminating ability of BMI, which has an 85% chance of detecting obesity (Table 2, Figure 2e).
The age of the whole male sample ranged between 40 and 79 years, with a mean age of 54.8 ± 9.7 years and a mean BMI of 29.3 ± 4.4 kg/m 2 , with 40% either being overweight (40.9%) or obese (40.8%) according to the WHO classification criteria. Alternatively, almost two-thirds (71.2%) of males in this age group were classified as obese based on BF% ( Table 1). The ROC analysis among males aged 40-79 showed that the most appropriate BMI cut-off point for diagnosing obesity based on BF% was 27.36 kg/m 2 . The BMI cut-off point for this group achieved high sensitivity (80.00%) and specificity (79.40%), indicating a low chance of false negatives and false positives ( Table 2). Comparing the distribution of obesity determined by BF% and the WHO criteria revealed a disagreement, with only 54.7% of those defined as obese according to BF% being correctly categorized according to the WHO criteria, thus missing almost half of those who are obese (Table 3). After comparing the proportion correctly diagnosed by the WHO cut-off point (54.7%), the identified cut-off for males aged 40-79 years improved the detection of obesity according to BF by almost 25% (24.4%), reaching 80.0% (Table 3). The AUC (0.88) indicates the excellent discriminating ability of BMI, which has an 88% chance of detecting obesity (Table 2, Figure 2f). Finally, considering the sensitivity and specificity of the WHO cut-off point for defining obesity (≥30 kg/m 2 ), overall, this cut-off point demonstrated poor sensitivity, reaching less than 60%, and high specificity, achieving more than 90% across all age and sex groups. Therefore, it has a higher chance of false positives and false negatives (Table 3).

Discussion
The current study aimed to provide benchmark data regarding the validity of the WHO BMI cut-off point of 30 kg/m 2 for obesity in middle-aged and older patients of both genders and, in the case of non-validity, to determine a new BMI cut-off point for more accurate screening for obesity in a clinical nutrition setting in Italy.

Findings and Concordance with Previous Studies
Our main finding was the identification of a new BMI cut-off point for obesity in a mainly clinical sample of participants enrolled in a nutritional management setting. This new cut-off point (i.e., BMI = 27.27 kg/m 2 ) in our population was found to be significantly lower than the WHO BMI cut-off point (i.e., BMI = 30 kg/m 2 ) [16]. Even though we are not in a position to determine the exact reason behind this discrepancy between the BMI cut-off for obesity in our population and that suggested by the WHO, we can, however, speculate and suggest that the changes in body composition across one's lifespan (i.e., the increased BF and reduction in lean mass), which seem to occur without a relevant variation in body weight, lead to higher adiposity at a lower BMI. Our investigation is not the first to suggest lower cut-off points for identifying obesity in middle-aged and older populations [30,31]. Other studies have suggested a BMI of approximately 27 kg/m 2 as a cut-off point for obesity in Western people [32,33]. For instance, Evans et al. identified a BMI > 27.0 kg/m 2 as defining obesity in white, middle-aged women in the United States [32]. Fernández-Real et al. recognized a BMI > 27.5 kg/m 2 (more specifically, 27.5 kg/m 2 in males and 27.4 kg/m 2 in females) as being indicative of obesity in the Spanish population [33]. The populations studied in the aforementioned studies are similar to a certain extent to our population in terms of several aspects (i.e., age, ethnicity, etc.) [32,33]. Outside of Western populations, Jahanlou and Kouzekanani detected obesity among Iranian adults and found that the appropriate cut-off points for BMI to define obesity were 27.70 kg/m 2 for females and 27.30 kg/m 2 for males [22], which overlap with our cut-off points.

Potential Clinical Implications and New Directions
Our findings have several clinical implications. Firstly, policy-makers in Italy are invited to take our results as at least preliminary evidence for considering this cut-off point as a new one for identifying obesity in clinical settings. Second, awareness should be raised among all healthcare professionals dealing with obesity with regard to recognizing this new cut-off point when screening for obesity and sharing or discussing this new information with their patients.
In addition, some new directions for future research are needed. Firstly, other investigations should replicate our findings to confirm this cut-off point (i.e., 27.27 kg/m 2 ) in middle-aged and older Italians. Moreover, future investigations are also needed to determine the cut-off point for leanness (i.e., underweight), especially in older adults, since previous research has shown that adults over 60-65 years who are underweight experience more health issues [34,35] and shorter life expectancy [35,36] than those with higher BMI (i.e., <27 kg/m 2 ) [37]. Secondly, on a national level, large sample studies must also assess the validity of normal (≥18.5 kg/m 2 ) and overweight (≥25 kg/m 2 ) cut-off points and, if necessary, determine new ones in both clinical and general populations. Finally, other works should extend the aim of our analysis to other European countries (i.e., Central, Eastern, Northern, Southern, and Western Europe).

Study Strengths and Limitations
Our study has several strengths. To the best of our knowledge, it is one of the few evaluations, if not the first, to test the validity of the WHO BMI cut-off point (i.e., 30 kg/m 2 ) and determine a new one for obesity in a large group of patients composed of middle-aged and older adults in a "real-world" clinical setting in Italy. Second, body composition was measured using DXA, which is known to exhibit a high level of precision [38]. However, our investigation also had some limitations. Most importantly, the data were obtained from a single unit, so external validation in other populations is necessary [39]. Finally, our study was of cross-sectional design; therefore, it was unable to detect BMI trends or changes (i.e., during COVID pandemic), which usually requires longitudinal assessment [40].

Conclusions
Obesity is a chronic disease associated with other severe comorbidities over time. Early identification is, therefore, crucial for managing the progression of the latter [41]. In our study, we provide evidence that the optimal BMI cut-off point (i.e., ≥27 kg/m 2 ) corresponding to obesity in a middle-aged and older mixed-gender clinical population in Italy varies from the widely used one (i.e., ≥30 kg/m 2 ). Therefore, we recommend that this new cut-off point be applied in clinical settings when screening individuals for obesity in Italy.

Supplementary Materials:
The following supporting information can be downloaded at https: //www.mdpi.com/article/10.3390/nu14224848/s1. File S1: STROBE Statement-Checklist of items that should be included in reports of cross-sectional studies  Informed Consent Statement: All patients' personal data were treated according to European and Italian privacy laws, and informed written consent was obtained.