Defining anthropometric thresholds (mid-arm circumference and calf circumference) in older adults residing in the community: a cross-sectional analysis using data from the population representative Longitudinal Aging Study in India (LASI DAD)


A comprehensive assessment of the nutritional status of older adults and effective intervention are essential to prevent adverse outcomes associated with malnutrition.1 As the Indian population experiences a rapid increase in life expectancy, it brings to the forefront a multitude of age-related challenges faced by older individuals. Assessing nutritional status through anthropometric measurements is an important way to ensure the well-being of individuals.2 Anthropometric measurements such as body mass index (BMI), mid-arm circumference (MAC) and calf circumference (CC) are simple, affordable and widely available tools for assessing nutritional status.3

Addressing the need for a comprehensive study on the ageing population in India, the Longitudinal Ageing Study in India (LASI) stands as a pioneering and nationally representative initiative. LASI focuses on various critical domains of ageing, including physical and cognitive health, economic well-being and social well-being.4 Within the LASI framework, the Harmonised Diagnostic Assessment of Dementia for LASI (LASI DAD) study specifically investigates late-life cognition and dementia. This study draws on a subsample of individuals aged 60 and above, meticulously selected from the LASI study. The LASI DAD study, along with the broader LASI study, not only offers valuable insights into the ageing population but also provides rich epidemiological data concerning various ageing issues. It encompasses a detailed geriatric assessment, enabling researchers to obtain comprehensive information necessary for a holistic understanding of older adults’ health and well-being.5 6

Malnutrition is often thought of as undernutrition, but it can also include overnutrition.7 Both these conditions can have serious health consequences. While undernutrition can be associated with increased risk of infections, slow wound healing, decreased muscle mass and strength, overnutrition can contribute to chronic health conditions such as hypertension, diabetes, cardiovascular disease, osteoarthritis, etc. Previous research conducted with older adults primarily aimed at identifying the presence of undernutrition and its associated risk factors,8 9 often neglecting the equally important issue of overnutrition.

Given the absence of extensive nationwide studies that use large-scale representative data in India, it becomes imperative to bridge this research gap. In this regard, we have undertaken an analysis of such data. Our study focuses on using the LASI DAD data to examine the relationship between various biosocial factors and the nutritional status of older adults. Specifically, our aim is to identify factors associated with malnutrition (undernutrition and overnutrition) and determine suitable cut-off values for anthropometric measurements such as MAC and CC among community-dwelling Indian older adults. Furthermore, our research findings have the potential to inform evidence-based recommendations for effective intervention strategies and policies aimed at improving the nutritional well-being of older adults in the country.



The LASI DAD study includes 4096 people aged 60 years and above. The study sample is a subsample of the LASI study. The samples were taken from 18 different states to capture the characteristics of the diverse population of India. The states included in the study were: Jammu and Kashmir, Punjab, Uttarakhand, Delhi, Haryana, Uttar Pradesh, Bihar, Madhya Pradesh, Rajasthan, West Bengal, Orissa, Assam, Karnataka, Tamil Nadu, Telangana, Maharashtra, Gujarat and Kerala. Methodological considerations of the LASI DAD study have been elucidated in a previous publication by Banerjee et al.6 For this analysis, we included 4092 community-dwelling older adults.

This study places emphasis on a range of sociodemographic variables, including age, gender, education, habitat and monthly per capita expenditure (MPCE) quintile. Additionally, comorbidities, geriatric syndromes, childhood financial and health status, and nutritional status assessed through anthropometric measures (BMI) are also examined.

Measurement of BMI, MAC and CC

Standard measuring instruments were employed to collect anthropometric measurements. Height and weight were measured using a stadiometer and weighing scale, respectively, enabling the calculation of BMI. BMI was further categorised using Asia-Pacific cut-off: underweight (<18.5 kg/m2), normal (18.5–22.9 kg/m2), overweight (23–24.9 kg/m2) and obesity (>25 kg/m2).10 For the purpose of addressing the research question, the groups were further simplified to normal (18.5–22.9 kg/m2), low BMI (<18.5 kg/m2) and high BMI (>23 kg/m2). Trained research nurses and field workers used a measuring tape to assess CC and MAC. Participants were instructed to stand upright for MAC measurement, allowing the arm to hang freely by the side. Biceps circumference was measured without applying pressure at the point of greatest bulge (middle of the arm). For measurement of CC, the participants were seated on the edge of the chair in a relaxed position with their feet flat on the floor, forming a right angle at the knee. They were instructed to pull up their trousers/bottom wear or saree. The widest part of the right calf was identified, and the circumference was measured using a seca measuring tape. CC was measured to the nearest 0.1 cm. LASI-DAD did not include bed-bound participants.


Following a thorough review of the existing literature, the following variables were chosen from the available variables in LASI and LASI DAD: age (categorised as 60–74 and >75 years), sex (male and female), education (no formal education, less than secondary and secondary or higher), occupation skill classified as per the international standard classification of occupations (IV—legislator, senior officials, professionals; III—technicians and associate professionals; II—clerks, service workers and shop and market sales workers, skilled agricultural and fishery workers, craft and related trade workers and I—elementary occupations and others),11 habitat (rural and urban), MPCE (poorest, poorer, middle, richer and richest), alcohol consumption and ever smoked (yes and no), comorbidities (hypertension, diabetes, chronic lung disease, chronic heart disease, anaemia in the past 2 years and stroke), multimorbidity (presence of two or more comorbidities). Geriatric syndromes such as Basic12 and Instrumental activities of daily living,13 cognition (using Hindi-Mental Status Examination),14 depression (using the Centre for Epidemiology Studies Depression),15 anxiety (using Beck Anxiety Inventory).16 In addition to these, the financial status during childhood (poor, average, pretty well of financially and varied) was also included.

Patient and public involvement

Patients and the public are not involved in the design and recruitment of LASI-DAD.

Statistical analysis

All analyses were performed by using STAT V.14 (StataCorp. 2015. Stata Statistical Software: Release 14, StataCorp). Continuous variables were summarised using measures such as mean, SD, median and IQR. Categorical variables are representation as frequency and weighted percentage. In order to examine the association between BMI groups (normal, low and high) with various factors a χ2 test was performed. To further understand the direction of association, weighted multinomial regression analysis was performed with normal BMI as base outcome. Variables which showed significant associations in univariate analysis were included in multivariate analysis. The results are represented as relative risk ratio (RRR) and 95% confidence interval (CI). Scatterplots with fitted linear regression lines were calculated to evaluate the relationship between MAC, CC and BMI. Correlation analysis was performed using Pearson’s correlation.

Using receiver operating characteristic (ROC) curve analysis, the cut-off values of MAC and CC were determined for males and females, aligning with BMI cut points of 18.5 and 23 kg/m². The area under the receiver operating characteristic curves (AUROCC) with 95% CIs were generated. The optimal cut-off values were estimated from the ROC curves using the sensitivity-specificity equality method. The CI was set at 95% and the level of significance at 5%. All analyses used survey weights to account for the complex survey design.


Baseline characteristics

The total population had a median age of 67 years (IQR: 63–73 years). Among the population, 3190 participants were aged between 60 and 74 years. Females accounted for 51.07% of the participants. A significant majority of the study participants (3085) had not received any formal education. Hypertension was the most prevalent comorbidity, affecting 1605 participants (39.03%), followed by diabetes, which affected 717 participants (17.85%). A total of 945 participants (22.83%) had multimorbidity, meaning they had two or more comorbidities. Impairments in instrumental activities of daily living were observed in 2270 older adults (54.47%), while basic activities of daily living were impaired in 2167 participants (50.88%). Table 1 provides a detailed overview of the baseline characteristics of the study population.

Table 1

Association between BMI and other factors

Association with BMI

Of the 4092 community-dwelling older adults, 902 (20.55%) had low BMI, 1742 (44.25%) had high BMI. BMI showed significant associations with various sociodemographic factors (table 1). These factors included age category (p<0.001), sex (p<0.001), education status (p<0.001), habitat (p<0.001), MPCE quintile (p<0.001), alcohol consumption (p=0.001) and smoking habits (p<0.001). Additionally, several comorbidities were found to be associated with BMI. These included hypertension (p<0.001), diabetes (p<0.001), chronic lung disease (p=0.038), chronic heart disease (p<0.001), high cholesterol (p<0.001), presence of anaemia in the past 2 years (p=0.002) and multimorbidity (p<0.001). Geriatric syndromes such as cognitive status (p<0.001), depression (p<0.001) and anxiety (p<0.001) were also associated with BMI. Furthermore, childhood factors, specifically financial status during childhood, showed a statistically significant association with BMI.

To assess the strength of association with BMI, the variables that showed significance in the univariate analysis (refer to online supplemental table 1) were included in the multivariate analysis. The factors associated with an increased risk of low BMI included higher age (RRR 1.46, p=0.005), belonging to the poorest MPCE quintile (RRR 1.44, p=0.025) and having impaired cognition (RRR 1.49, p=0.002). On the other hand, residing in urban areas (RRR 0.64, p=0.001), having hypertension (RRR 0.69, p=0.018) and experiencing varied financial status during childhood (RRR 0.26, p=0.039) were significantly associated with lower risk of poor BMI (table 2).

Supplemental material

Table 2

Weighted complex multinomial regression analysis

Factors associated with an increased risk of high BMI included being female (RRR 1.83, p<0.001), having higher education (less than secondary (RRR 1.42, p=0.013), secondary education or higher (RRR 1.86, p=0.025)), residing in urban areas (RRR 1.83, p<0.001) and having comorbidities such as hypertension (RRR 1.46, p=0.002), diabetes (RRR: 1.51, p=0.019), chronic heart disease (RRR 1.58, p=0.047). Whereas, advancing age (RRR 0.66, p=0.002), belonging to the poorest MPCE quintile (RRR 0.64, p=0.006), having anaemia in the past 2 years (RRR 0.55, p=0.022) and cognitive impairment (RRR 0.66, p<0.001) were associated with lower risk of high BMI.

MAC and CC cut-off score determination

Figure 1 presents scatterplots showing the relationship between BMI and MAC (A) and CC (B) among older adults, segregated by sex, along with regression fits. The results revealed a significant positive correlation between BMI and MAC for both males (r=0.709, p<0.001) and females (r=0.715, p<0.001). Similarly, there was a significant positive correlation between BMI and CC for males (r=0.764, p<0.001) and females (r=0.761, p<0.001).

Figure 1
Figure 1

Scatter plot of BMI by mid-arm circumference (MAC) (A) and by calf circumference (CC) (B) of males and females. BMI, body mass index.

Table 3 describes the proposed cut-offs for males and females in the study population for MAC and CC. For males, the best statistically derived MAC cut-off corresponding to a BMI<18.5 was 23.9 cm (AUROCC: 0.86, sensitivity: 82.13%, specificity: 75.90%), while for females, it was 22.5 cm (AUROCC: 0.89, sensitivity: 83.88%, specificity: 75.98%) (table 3). Similarly, the derived MAC cut-off corresponding to a BMI<23 was 26.9 cm for males (AUROCC: 0.85, sensitivity: 70.31%, specificity: 84.71%) and 25 cm for females (AUROCC: 0.87, sensitivity: 80.83%, specificity: 80.58%).

Table 3

The cut-off scores of mid-arm circumference and calf circumference of community dwelling males and females

Regarding CC, the cut-off to detect low BMI (<18.5 kg/m2) was 28.1 cm for males (AUROCC: 0.86, sensitivity: 81.79%, specificity: 75.90%) and 26 cm for females (AUROCC: 0.88, sensitivity: 85.19%, specificity: 69.00%). The optimum cut-off corresponding to a BMI<23 kg/m2 was 31.5 cm for males (AUROCC: 0.85, sensitivity: 70.74%, specificity: 85.39%) and 29 cm for females (AUROCC: 0.87, sensitivity: 78.13%, specificity: 83.83%).

Figure 2 displays the ROC curves for MAC and CC separately for males and females.

Figure 2
Figure 2

Receiver operating characteristics (ROC) curve to determine optimum cut-offs for mid-arm circumference (MAC) of males (A, B) and females (E, F) and calf circumference (CC) of males (C, D) and females (G, H) to screen for undernutrition (BMI<18.5 kg/m2) and overnutrition (BMI>23 kg/m2). BMI, body mass index.


The study focused on assessing the nutritional status of community-dwelling older adults in India and provided valuable insights into the connections between BMI and various sociodemographic and health-related factors. The results emphasise the significance of addressing age-related concerns, such as functional dependence, disabilities, malnutrition, changes in psychosocial aspects and multiple health conditions, in order to establish an age-friendly society.

This study found that 44.25% (weighted percentage) of the population had a high BMI, while 20.55% (weighted percentage) had a low BMI. Undernutrition was associated with advancing age, being in the poorest wealth quintile and impaired cognition. Urban residency, hypertension and varied childhood financial status were linked to a lower risk of undernutrition. Overnutrition was associated with higher education, urban living and comorbidities such as hypertension, diabetes and chronic heart disease. It showed a negative correlation with increasing age, being in the poorest wealth quintile, experiencing anaemia in the past 2 years and impaired cognition. In addition to this, we also assessed the predictive ability of MAC and CC for underweight and overweight in older adults in India.

The prevalence of underweight in our study is comparatively lower than in previous studies. For instance, a study conducted in rural Uttarakhand reported a prevalence of 26.6%,17 while another study from rural area of Haryana found a prevalence of 33.6%.18 However, our study did reveal a higher prevalence of low BMI in rural areas (27.88%) compared with urban areas (12.46%). This difference was statistically significant in the regression analysis (RRR: 0.64 in urban, compared with rural areas). It is noteworthy that a prior analysis of cross-sectional data of 6372 older adults from WHO’s Study on global AGEing and adult health (SAGE), which included data from six states during 2007–2008 reported an overall prevalence of underweight of 38%.19 This comparison highlights a clear decline in the prevalence of underweight over the past decade. This decline in undernutrition rates could indicate improvements in various factors, such as access to adequate food resources, changes in dietary patterns, advancements in healthcare or interventions targeting malnutrition. The reduction in underweight prevalence signifies a positive trend towards better health outcomes.

SAGE data showed lower rates of overweight (10.7%) and obesity (2.7%) compared with our study’s higher prevalence of 44.25%. One possible explanation for this disparity is the utilisation of different criteria to define overweight. In our study, we employed the Asian-Pacific cut-off score, whereas the SAGE analysis employed the more lenient WHO cut-off score, which defines overweight as 25.0–29.9 kg/m2. Another study from Puducherry reported overweight prevalence of 41.4% and obesity prevalence of 4.5%.20 The growing prevalence of overweight among the elderly is a matter of great concern, as it is likely to result in a subsequent increase in the prevalence of multimorbidity within this vulnerable population.21 22

Our study not only strengthens the existing body of research but also expands on it by providing further evidence on the associations and interactions between BMI and various factors. Consistent with previous studies, we confirmed the association between low BMI and increasing age,23 poor cognition24 and wealth quintile.19 Additionally, our study demonstrated significant interactions between high BMI and factors such as female sex,19 higher education qualification,25 urban residency19 25 and comorbidities. These findings expand our understanding of the relationships between BMI and various factors, providing valuable insights for targeted interventions and public health policies.

Anthropometric measurement, such as MAC and CC can be used as practical indicators to screen for malnutrition in older adults. These measurements, with specific cut-off values differentiated by gender, enhance the accuracy and practicality of the findings. This is particularly valuable for early detection of malnutrition, especially in community settings where height and weight measurements may be difficult to obtain. Our study aimed to determine the optimal cut-off values for MAC and CC to screen for malnutrition, recognising the importance of early diagnosis and intervention. We used BMI as the reference standard for deriving these cut-off values. we determined that the lower limit of MAC for males was found to be 23.9 cm (AUROCC: 0.86). Similarly, for females, the lower limit of MAC was identified as 22.5 cm (AUROCC: 0.89). Additionally, we derived the MAC cut-off values corresponding to a BMI below 23, which were 26.9 cm for males (AUROCC: 0.85) and 25 cm for females (AUROCC: 0.87). Regarding CC, the identified cut-off value for detecting low BMI (<18.5 kg/m2) was 28.1 cm for males (AUROCC: 0.86) and 26 cm for females (AUROCC: 0.88). Furthermore, the optimal cut-off corresponding to a BMI below 23 kg/m2 was determined as 31.5 cm for males (AUROCC: 0.85) and 29 cm for females (AUROCC: 0.87). Multiple studies have emphasised the utility of these anthropometric measurements in evaluating the nutritional status of older adults across different setting, such as hospitals, outpatient departments and community-based settings.26 Moreover, MAC and CC have shown superior predictive accuracy in comparison to BMI when it comes to assessing nutritional status and overall health conditions.3

Our study highlights a noticeable difference in the cut-off values for MAC and CC when compared with cut-off identified for other populations.27–30 Additionally, among Asian populations and specifically the Indian population, the differences range from 0.5 cm to 3 cm as observed in various Asian studies.27 28 30 It is worth noting that only a limited number of studies have reported specific cut-off scores for MAC tailored to a particular population. Therefore, bridging this research gap and determining standardised cut-off values for MAC and CC specific to the Indian population hold great importance in addressing the nutritional assessment needs of this specific demographic.

Strengths and limitations

This study has several strengths that enhance the quality and generalisability of its findings. First, the utilisation of data from LASI-DAD, a nationally representative sample, ensures that the study’s results are applicable to the broader population. The consistency in study design and measurements across different sites minimises potential biases and allows for reliable comparisons between different groups and regions. Previous studies have focused on social and economic factors or comorbidities. This study takes a more comprehensive approach by considering multiple factors, including financial situation during childhood and geriatric syndromes. This allows for a better understanding of the complex interplay of factors that influence nutritional status in the older population. Additionally, our study addressed a gap in previous research by using the Asia-Pacific cut-off for BMI, which is more relevant to our population.31 32 The study’s findings provide sex-specific MAC and CC cut-off measurements that can be used in future research and clinical practice, improving the accuracy and relevance of nutritional assessments in Indian older adults.

While this study has several strengths, it is important to acknowledge its limitations. One significant limitation is its cross-sectional design, using data from only the first wave. This design restricts the study’s ability to establish causality or determine temporal relationships between variables. Another limitation stems from the reliance on previously diagnosed comorbidities, while several older adults live with undiagnosed hypertension, diabetes, etc. It is crucial to recognise that BMI does not provide an accurate measure of body composition. Additionally, it is important to highlight that LASI-DAD lacks data related to body composition. Finally, even though a wide range of socioeconomic factors, comorbidities and geriatric syndromes are included in this study, there may be other unknown factors that have the potential to affect the nutritional status.

This post was originally published on