Relationship between body mass index and percentage of body fat, estimated by bio-electrical impedance among adult females in a rural community of North India: A cross-sectional study.

Introduction: Body adiposity measured by percentage of body fat (BF%) is found to be better predictor of cardiovascular morbidity and mortality than body mass index (BMI). Limited information exists showing relationship between BMI and BF% in North Indian population.
Objectives: To study the relationship between BMI and BF% among North Indian adult females across various age strata and level of BMI. Materials and
Methods: This was a cross-sectional study conducted at Ballabgarh Health and Demographic Surveillance Site (HDSS) among randomly selected females. BMI using standard techniques and BF% using bioelectrical impedance analysis was estimated. Linear regression was performed using general linear model with BF% as dependent variable and BMI as main independent variable.
Results: Mean (±2 SD) age of participants was 41.3 ± 15.7 years. Mean BMI (±SD) was 23.3 (±4.6) kg/m2, whereas mean fat mass (±2 SD) and BF% (±95% CI) was 19.2 (±7.9) kg and 33.6 (±6.9) %. BMI and BF% were highly correlated among obese (r = 0.77), whereas least correlated (r = 0.32) in underweight females. Across age strata, correlation between BMI and BF% was maximum in 18-35 years age group (r = 0.95), whereas least in females ≥56 years (r = 0.67). Age and BMI together predicted 73% of variability in BF% in hierarchical linear regression model. Conclusions: In this population, we have found strong correlation between BMI and BF% particularly at higher level of BMI and in younger females. There is need to conduct more robust prospective longitudinal studies to assess BF%, which is a better predictor of cardiovascular morbidity and mortality.

Introduction

Overweight and obesity defined as having body mass index (BMI) has significant impact on health and has reached epidemic proportion, globally. Global Burden of Disease (GBD) 2013 study estimated that number of overweight and obese individuals has increased from 921 million in 1980 to 2.1 billion in 2013.[1] Worldwide, overweight and obesity is attributed to 3.4 million deaths and 3.8% of Disability Adjusted Life Years.[2] High BMI, both directly and mediated through high blood pressure and cholesterol, is a major risk factor for cardiovascular deaths.[3]

BMI has been used traditionally as an anthropometric mean of measuring generalized obesity due to its ease to use, low cost, and convenience. However, BMI does not reflect the adiposity or BF%. Central adiposity as measured by waist height ratio, waist circumference, and BF% are known to be better predictor of diabetes and cardiovascular events than BMI.[456] Moreover, at same level of BMI, south Asians are believed to have high BF% (both central and generalized) and lesser lean, muscle and skeletal mass than Caucasians and thus being at a higher risk of cardiovascular events and deaths.[789]

Despite the utility of measuring body composition to predict metabolic syndrome and cardiovascular diseases, data showing relationship between BMI and BF% in India and other similar South Asian countries is scarce.[1011] In this context, it is important to understand the relationship between BMI and BF% in Indian population. This is of more importance among females as they have relatively higher BF% on a corresponding BMI as compared with males.[12] Since BF% is better predictor of metabolic disorder than BMI alone, women are likely to be at more risk of metabolic disorders than men having similar BMI. However, no study has been conducted in India to study this relationship specifically in women. Knowledge regarding the relationship between BMI and BF% can be used to understand the utility of BMI as a predictor of cardiovascular morbidity among Indian female population.

Various methods, such as dual energy x-ray absorptiometry (DEXA), air displacement plethysmography (ADP), and bio-electrical impedance analysis (BIA), have been used to estimate BF%. We have used BIA to estimate the percentage of body fat, because it was easy to use in rural setup in transporting machine, also it is less costly than others, portable, less time taking, and requires minimal training. BIA has been validated to measure BF% against other methods.[131415] BIA has been reported as a good predictor of DEXA-derived fat mass (r = 0.85–0.88).[16] In a study done in Chinese females, it was found that BIA provides a relatively accurate prediction of BF% as BIA underestimated BF% by only 5.1% (−2.4 to 12.7).[17] We studied female population residing in a rural community of North India to estimate the mean BF% and to identify the relationship of BMI with BF% (measured with noninvasive BIA machine) across various strata of age and BMI.

Materials and Methods

Study design and population

This was a cross-sectional study done in June to July 2015 on randomly selected nonpregnant nonlactating women aged ≥18 years residing in the study area for at least last 12 months. The study area was spread over 28 villages in rural Ballabgarh Block of Faridabad district of Haryana, North India. These 28 villages were part of Health and Demographic Surveillance System (HDSS), Ballabgarh.[18] Population of this area was under continuous demographic and health surveillance and information obtained is retained in health management information system (HMIS), which is in use since 1991.[19]

Ethics

Ethical approval for the study was obtained from Institute Ethics Committee (IEC) of All India Institute of Medical Sciences, New Delhi, India.

Sample size

The required sample size was calculated assuming standard deviation (SD) of BF% among females as 8.25,[11] a precision of 0.9, alpha error of 5%, and nonresponse rate of 20%. The required sample size was calculated to be 420.

Study procedures

HMIS database has details of all the individuals residing in the 28 villages of HDSS Ballabgarh. List of all the nonpregnant, nonlactating females aged 18 years or more residing in the study area was used as the sampling frame. From this sampling frame, 420 females were selected for this study. Selected females were informed 1 day prior to fast in the morning before examination and to not involve in any physical activity overnight till the physical examination is over. The randomly selected individuals were then visited at their house next day and informed written consent was taken from all the participants. Eligibility was assessed at the household using following exclusion criteria; women having problem of weight loss, history of tuberculosis, history of drug intake such as diuretics, weight loss associated with cancer, any locomotor problem, which limited the use of BIA equipment, and nonavailability of females at their houses during two consecutive visits. Eligible participants who consented for the study, their sociodemographic information and morbidity profile were obtained using a pretested semi-structured interview schedule. They were then taken to a nearby primary health center, where anthropometric and body composition measurements were undertaken by the field staff. Weight, height, and BF% were measured using standardized equipment. Height was measured using a stadiometer (Seca) to the nearest 0.1 cm in standing position with no footwear. Weight was measured using an electronic scale (Seca 808 Germany). BMI was, thus, calculated using standard definition of weight (kg)/square of height (m). BF% was estimated using a portable, single frequency, eight-electrode bio-electrical impedance Analyzer (Tanita Corporation, Tokyo, Japan). The measurements were done as per manufacturer instructions and standard operating procedures. This was done during morning hours after ensuring that participants have not indulged in any vigorous physical activity.

Participants stood on the footplate and grasped the two handgrips with arms held straight forward at 90° to create a complete circuit. Quality control was done by measuring the interobserver and intrasubject reliability. Blood pressure was measured using digital sphygmomanometer (OMRON). All clinical measures were done by trained medical graduates.

Statistical analysis

Participants were grouped as young (18–35 years), middle age (36–55 years), and older adults (≥55 years). Participants were classified as underweight (<18.5 kg/m2), normal weight (18.5–22.9 kg/m2), overweight (23.0–24.9 kg/m2), and obese (≥25 kg/m2) based on consensus guidelines for diagnosis of obesity among Asian Indians.[20] Descriptive statistics was reported as mean ± SD. For comparison of mean across various age groups, one-way analysis of variance was performed. Correlation between BF% and BMI and total body fat and BMI were calculated using Pearson's correlation coefficient (r) across various BMI categories and in different age group category. Visual inspection of the relationship (BMI–BF%) was also made. Linear regression was performed using general linear model with BF% as dependent variable and BMI as main independent variable. We did hierarchical regression with introduction of age in Model 1 followed by BMI and age in Model 2. Along with age and BMI, lean mass, muscle mass, height, weight, and systolic and diastolic blood pressures were introduced in Model 3. Fat mass and fat-free mass were omitted from the hierarchical model as they were highly collinear to BF%. Statistical analysis of data was carried out using the Stata version 13.0 software for windows (StataCorp. 2013; Stata Statistical Software: Release 13. College Station, TX: StataCorp LP.).

Results

Descriptive characteristics

In total, 388 women participated in the study out of 420 sampled. Rest of the participants (7.6%) could not be contacted even after two visits. About 53.6% of women were illiterate and 67.3% were housewife. History of either hypertension or diabetes was present in 28.1%. There was significant difference in BMI and fat mass between housewife and working group women [Table 1].

Table 1

Distribution of sociodemographic characteristics with respect to BMI and fat mass of participants (n=388)

Variable	Category	Frequency (%)	BMI (mean±SD)	P	Fat mass (mean±SD)	P
Education	Illiterate	208 (53.6)	22.9±4.5	0.060	19.7±7.5	0.207
	Educateda	180 (46.4)	23.7±4.6		19.2±8.0
Occupation	Housewife	261 (67.3)	23.7±4.6	0.013	20.1±8.4	0.002
	Workingb	127 (32.7)	22.4±4.4		17.5±6.5
Type of family	Joint	263 (67.8)	23.1±4.3	0.412	19.2±7.8	0.993
	Nuclear	125 (32.2)	23.5±5.1		19.2±8.0
History of HTN/DMc	No	279 (71.9)	23.1±4.5	0.148	18.7±7.4	0.066
	Yes	109 (28.1)	23.8±4.8		20.4±9.0

BMI: Body mass index, SD: Standard deviation, HTN: Hypertension, DM: Diabetes mellitus. aPrimary up to graduation, bAgricultural laborer, manual laborer, and others, cHTN and/or DM

Mean age of participants was 41.3 ± 15.7 with >40% in young age group. Table 2 shows anthropometric and body composition of participants with respect to age group. Mean BMI was 23.3 (±4.6) kg, whereas mean fat mass and BF% were 19.2 (±7.9) kg and 33.6 (±6.9) %, respectively. No significant difference was observed in any of the anthropometric and body composition characteristics with respect to age [Table 2]. The population represented a wide range of BMI varying from 15.6to 38.1 kg/m2.

Table 2

Anthropometric and body composition with respect to age category (n=388)

Parameter	18-35 years	36-55 years	≥56 years	Total	Pa
Frequency (%)	176 (45.4)	131 (33.8)	81 (20.9)	388 (100)	-
Height (cm)	153.3 (6.2)	152.5 (5.7)	153.7 (5.9)	153.1 (6.0)	0.33
Weight (kg)	54.2 (11.6)	54.3 (12.0)	56.2 (12.4)	54.6 (11.9)	0.43
BMI (kg/m2)	23.0 (4.4)	23.5 (4.9)	23.6 (4.7)	23.3 (4.6)	0.49
Fat mass (kg)	18.5 (7.4)	19.2 (7.5)	20.6 (9.5)	19.2 (7.9)	0.14
Muscle mass (kg)	33.6 (4.4)	33.4 (4.5)	34.8 (5.9)	33.8 (4.8)	0.11
Bone mass (kg)	1.98 (0.46)	2.28 (3.29)	3.73 (11.2)	2.45 (5.5)	0.05
Fat-free mass (kg)	35.7 (4.7)	35.3 (4.9)	36.2 (6.4)	35.6 (5.2)	0.50
Fat (%)	33.0 (6.6)	33.7 (6.6)	34.6 (7.9)	33.6 (6.9)	0.19
TBW (kg)	46.7 (3.6)	46.3 (5.4)	45.9 (8.1)	46.4 (5.6)	0.33
BMR	4547.1 (660.1)	4507.1 (692.2)	4564.2 (756.2)	4537.7 (690.7)	0.81

a<0.001. Values are mean (±SD) unless otherwise indicated. P value for one-way test for analysis of variance. BMI: Body mass index, TBW: Total body water, BMR: Basal metabolic rate, SD: Standard deviation

Analysis of anthropometric and body composition characteristics with respect to BMI category revealed a highly significant (P-value < 0.001) increase in fat mass, muscle mass, and BF% with an increment in BMI category [Table 3].

Table 3

Anthropometric and body composition parameters with respect to body mass index category (n=388)

Variable	Underweight	Normal	Overweight	Obese	P
Frequency (%)	67 (17.3)	122 (31.4)	68 (17.5)	131 (33.7)	-
Weight (kg)	40.8 (4.5)	48.3 (5.1)	55.7 (4.7)	67.0 (9.3)	<0.001
Fat mass (kg)	9.8 (2.6)	15.6 (5.5)	19.7 (1.7)	27.1 (5.7)	<0.001
Fat (%)	23.8 (4.1)	31.0 (5.0)	35.1 (1.5)	40.1 (3.2)	<0.001
Fat free mass (kg)	30.9 (2.8)	33.6 (4.7)	36.4 (2.8)	39.7 (4.4)	<0.001
Muscle mass (kg)	29.4 (2.6)	31.8 (4.5)	34.3 (2.5)	37.6 (3.9)	<0.001

Values are mean (±SD) unless otherwise indicated. P value for one-way test for analysis of variance. SD: Standard deviation

Relationship between body mass index and body fat

Table 4 reports the Pearson's correlation coefficient for the relationship between BMI and BF% and BMI and fat mass. Redoing the analysis across various strata of BMI revealed significant correlation between BMI and BF% in all BMI categories except among underweight individuals, similar was true for correlation between BMI and fat mass. The correlation coefficient, in both the cases, increased with increase in BMI category. Similarly, analysis across age groups showed a highly significant correlation (P < 0.001) of BF% and fat mass with BMI in all the age groups with a decrease in correlation coefficient with an increase in age category. Visual inspection of scatter plot of relationship between BMI and BF% is shown in Figure 1. Figure 2 shows the scatter plot of different age groups. Figure 3 shows the receiver operating characteristics curve of BF%, assuming cut-off for BMI as 25.

Table 4

Relationship between BMI and BF% and BMI and fat mass; overall and across different BMI and age category

Variable	BMI and BF%		BMI and fat mass
	r*	P	r*	P
Overall	0.85	<0.001	0.88	<0.001
Underweight	0.32	0.17	0.21	0.07
Normal	0.42	<0.001	0.42	<0.001
Overweight	0.60	<0.001	0.57	<0.001
Obese	0.77	<0.001	0.79	<0.001
18-35	0.95	<0.001	0.97	<0.001
36-55	0.87	<0.001	0.91	<0.001
≥56	0.67	<0.001	0.73	<0.001

BF%: Percentage of body fat, BMI: Body mass index, *r: Pearson’s correlation coefficient

Figure 1

Scatter plot showing distribution of BF% with respect to body mass index (overall)

Figure 2

Scatter plot showing distribution of BF% with respect to body mass index (age group category wise): (a) age 18–35 years, (b) age 36– 55 years, and (c) age ≥56 years

Figure 3

Receiver operating characteristic curve of BF%: Body fat percentage (area under the curve: 68.8%, P = 0.07)

Table 5 shows the results of linear regression using general linear mode with hierarchical introduction of variables. Linear regression showed predictability of Model 1 to be very less (R2 = 0.002, P = 0.14) with maximum increment in Model 2 (R2 = 0.73, P < 0.001) followed by marginal increment in Model 3 (R2 = 0.80, P < 0.001). This showed that age had very less effect on BF% and BMI had the maximum effect. But, coefficient of BMI that was 1.27 (1.20–1.35) in Model 2 reduced to 0.39 (0.17–0.60) after introduction of other variables in Model 3.

Table 5

Linear regression showing effect of BMI, age, and other variables on BF%

Variable	Coefficient	SE	95% CI	β
Model 1 (R2=0.002, P=0.14)
Age	0.03	0.02	−0.01 to 0.07	0.07
Constant	32.2	0.98	30.2-34.2
Model 2 (R2=0.73, P<0.001)
Age	0.02	0.01	−0.002 to 0.04	0.04
BMI	1.27	0.03	1.20 to 1.35	0.85
Constant	2.97	1.04	0.93 to 5.02
Model 3 (R2=0.80, P<0.001)
Age	0.01	0.01	−0.001 to 0.03	0.02
BMI	0.39	0.11	0.17 to 0.60	0.25
Weight	0.26	0.05	0.15 to 0.37	0.46
Height	−0.35	0.05	−0.04 to−0.25	−0.25
Muscle mass	0.40	0.11	0.17 to 0.61	0.28
Bone mass	0.15	0.04	0.07 to 0.22	0.11
SBP	0.02	0.01	−0.004 to 0.04	0.05
DBP	0.009	0.02	−0.034 to 0.053	0.014
Constant	2.97	1.04	0.93 to 5.02

SBP: Systolic blood pressure, DBP: Diastolic blood pressure, BMI: Body mass index, SE: Standard error, CI: Confidence interval. Model 1: age, Model 2: age and BMI, Model 3: age, BMI, weight, height, muscle mass, bone mass, SBP, and DBP

Discussion

Our study mainly intended to understand the relationship between BMI and BF% (measured by bio-electrical impedance analysis [BIA]) among females of rural north India and effect of age group and BMI category on the quantum of correlation between BMI and BF%. The relationship between BMI and BF% has been studied across various ethnic groups; particularly in Western countries.[1421222324252627] Body composition of South Asians are different from other ethnic groups, such as Caucasians, Blacks, and even Mongolians.[28] Thus, it is needed to understand the relationship between BMI and BF% in native Indian population as it has been reported that for the same degree of obesity as measured by BMI, the BF% among South Asians may be much more than population of other ethnicities. To the best of our knowledge, only few studies have been published from India and other Asian nations with an intent to answer this research question.[42930]

Previous studies have shown a positive correlation between BMI and BF% in various population and our study has confirmed the direction of correlation among females of a north Indian population.[142021222324] In a study done among young adults of Sri Lanka, a strong and significant positive correlation was observed between BMI and BF%; overall (r = 0.82, P < 0.01 among females) and also by age with correlation in younger age group (r = 0.84, P < 0.01) more than the elderly females (r = 0.75, P < 0.01).[31] We have also got similar results with r = 0.95 and 0.67 (both significant at P < 0.001) among young females and elderly females, respectively. Similarly, the Pakistan study showed a positive correlation between BMI and BF% with r = 0.74, P < 0.001.[27] They did not report the correlation with respect to age. In both the studies, the relation was linear at lower level of BMI and became quadratic at higher BMI as observed in our study by the visual inspection of the scatter plot. Significantly positive relationship was also seen in a multicentric study done in various ethnic groups, such as Europeans, natives of Maori and Pacific Island, and Asian Indians.[28]

Other studies done among Caucasians and Blacks have also confirmed similar relationship. In our study, the relationship was significant across all the BMI categories except among individuals whose weight were below normal. Also, the strength of correlation increased with the increase in BMI category with a nonsignificant r of 0.32 in underweight category to r of 0.60 and 0.77 in overweight and obese category, respectively. Similar picture was also observed among UK adults, where relationship was not significant particularly for BMI <25 kg/m2.[14] The pattern of correlation between BMI and BF% with respect to level of BMI reported in this study and confirmed with minor differences in other previous studies from south Asian as well as western population has biological plausibility and has been discussed extensively in previous studies. We also reported relationship of BMI with fat mass and found that this relationship was almost similar to the relation obtained between BMI and BF%. This is contrary to an earlier observation which noted that relation between BMI and BF% was not as stronger as the relation between BMI and fat mass. Another study among children and adolescents also reports that BMI is not as strongly correlated with BF% measured by DEXA than with total body fat. The hypothesis stated that BMI is indicative of body fat mass rather than the relative measure of proportion of body fat, and thus, increment in BMI may not have equivalent increment in BF%. However, both these studies were done in western population. It is known that BF% in western population for a same level of obesity is less than the BF% in the Asian population. Thus, it is likely that higher proportion of body fat among Asian population may have greater influence on BMI than their western counterparts.

We also tried to understand the relationship between BMI and BF% across various age groups. Though the relationship was significant across all the age groups, the correlation coefficient showed a declining trend while moving from younger age group to elderly. Linear regression showed age to be nonsignificant predictor variable whereas BMI, muscle mass, and bone mass were significant predictors with 80% predictability after including all the variables in the model. However, we have studied only few variables in the regression model and there may be many more predictor variables which if included in the model can change model equation and predictability of the model. Also, among younger females, BF% showed larger increase with per-unit increase in BMI in comparison to the elderly females. This observation was contrary to the results available from Sri Lanka in a similar study. In the Sri Lankan study, the elderly females had reduced BMI and increased BF% than their younger counterparts. The contrary results might be due to different body composition characteristics in two populations and need to be interpreted with caution.[4] In our study, there was no significant change in any of the body composition characteristics across the age group, whereas in the Sri Lankan study, there was significant increase in BF% between younger and elderly while same was not true for BMI. Various theories, such as sarcopenia, resulting in loss of muscle mass and subsequent increase in fat mass, in elderly population has been attributed to this phenomenon. Our study population did not show this kind of phenomena and the muscle and fat mass was almost constant even in the elderly age group. This could be one of the reasons for the contrary results. Another Indian study done in urban areas of Lucknow showed similar results.[32] Although there was increase in abdominal fat in the elderly group, there was no significant change in BF% with the age in this study. Similar pattern of increase in abdominal fat while no change in BF% was observed in the Asian Indian females recruited in the multicountry study.[25] Since we have not measured abdominal fat percent or waist circumference as an index of central adiposity, we could not comment on its change with advancing age. However, it could be interpreted with little doubt that Indian population do not exhibit similar change in body composition characteristic with advancing age as seen in western population. Ethnic differences in body composition characteristics, such as muscle mass, degree of adiposity, skeletal mass and leg length, need to be taken in consideration while drawing any correlation between body fatness and BMI. Since south Asian population is different from their western counterparts in these parameters for a same level of BMI, the BF% and BMI correlation with respect to age category may not stand similar to the western population. Lack of any significant difference in BMI across the age group could also be due to high mortality rates with age in obese individuals. This result is important in Indian context, where age of development of cardiovascular diseases and diabetes is skewing toward left. With a steeper increase in BF% at a particular increase in BMI among younger age group, they would be more prone for metabolic derangement and insulin resistance even in earlier period of life leading to increase in incidence of these diseases and the resulting complications.

BF% is a better predictor of diabetes and cardiovascular events than BMI.[456] In our study, it was found that BMI was associated with BF% and fat mass across all age groups. Hence, BMI can be used as a predictor of diabetes and cardiovascular events across all age groups of Indian female population. However, the association of BMI with BF% and fat mass was not uniform across all levels of BMI. In this study, BMI was not found to be associated with BF% and fat mass among underweight females. Magnitude of correlation increased with the increase in BMI. Hence, BMI as a risk factor of diabetes and cardiovascular events among underweight females should be interpreted with caution in population-level screening in Indian population. However, BMI can be used as a predictor of cardiovascular events in normal and overweight Indian females. Findings of this study can be used for predicting cardiovascular risk among Indian females in future. Large-scale longitudinal studies should be conducted to further explore this relationship.

Though our results are in alignment with the already available evidence, it needs to be interpreted cautiously in following context: method of assessment of percentage of body fat and body composition characteristics of the population. We used commercially available portable Bio-electric Impedance Analyzer for measurement of BF%. The pre-requisite for body composition measurements such as 3 hours of fasting and 12 hours of nonindulgence in any physical activity was self-reported and was not confirmed by our field staff. Though, BIA has been compared with various reference body composition measurement techniques, such as DEXA and air displacement plethysmography, and is found to be valid and reliable but is not a gold standard.[33] Moreover, BIA has disadvantage in assessing the location of fat deposition when compared to DEXA technique. BIA also has a limited utility at extreme levels of BMI as its algorithm has not been validated in clinical use under such circumstances.[2122] Since, BMI in our population ranged from as low as 15 Kg/m2 to 38 Kg/m2, this factor needs to be taken into account while interpreting the result, particularly at extremes of BMI. Also, our assessment of abnormal hydration was totally based on use of diuretics and their exclusion from the study. It is worth mentioning that we did not perform any hydration study. Since, we used single frequency equipment rather than multiple frequency, abnormal hydration may limit the validity of our results. In this study, we did not ask about the menstrual history or menopause. Menopause is known to affect body fat and increases the chances of developing obesity.[34] However, considering all the limitations, in population based epidemiological studies portability of the equipment is an important concern and ease of using portable BIA machine in community settings makes it more suitable than any of the other methods. Regarding body composition characteristics, as already detailed, the South Asian population is different from their western counterparts in terms of BF% and BMI distribution. However, the body composition of our population was similar to the one recruited from Sri Lanka and Pakistan in terms of BMI and BF% (mean BMI: this study – 23.3 (4.6), Sri Lanka – 23.8 (4.2), Pakistan – 24.1 (6.3); mean BF%: this study – 33.6 (6.9), Sri Lanka – 31.7 (3.8), Pakistan – 34.9 (8.0)).[427] It could be inferred without much doubt that our population was fairly representative of the south Asian population recruited in similar studies and the relationship estimated in our study could be extrapolated for the south Asian population.

We could recognize several strengths in our study. Our study was among the first studies to demonstrate the relationship between BMI and BF% among north Indian rural females, the population which is relatively unexplored in this context. We recruited a fairly representative sample that even matched with the body composition of participants of similar study of other south Asian nations, such as Pakistan and Sri Lanka. Also, our sample had a wide spectrum of age and BMI enabling us to infer the relationship across various age and BMI categories. We measured the BF% using BIA method that may not be as valid and reliable as other sophisticated methods, such as DEXA or ADP. However, BIA is the best bet for community-based rural settings taking convenience of its use in account.

Conclusions

Our results demonstrate a significant positive correlation between BMI and BF% using BIA among north Indian rural females. We could also establish the effect of age and level of BMI on the magnitude of correlation with maximum positive correlation of BMI and BF% in the youngest (18–35 years) age group and in individuals with high BMI, that is, obese (≥25 kg/m2). Our findings provide initial information on relationship of BMI and BF% in north Indian population, which may be different from individuals of other ethnic groups and geographical regions. However, it is important to see whether assessment of BF% will be a feasible screening tool than BMI as a predictor of metabolic disorder. This study provides a platform for further research to provide more understanding in this context through prospectively planned longitudinal studies.

Financial support and sponsorship

Nil.

Conflicts of Interest

There are no conflicts of interest.