1000

Asia Pacific J Clin Nutr (1997) 6(3): 191-199

Weight, skinfolds and circumference characteristics of
poor elderly people in Mumbai, India

MC Manandhar¹, PS Anklesaria^2, and SJ Ismail¹

¹ Public Health Nutrition Unit, Department of Epidemiology and Population Health, London School of Hygiene and Tropical Medicine, London, UK

² Biomedical Gerontology Centre of HelpAge India, Department of Pharmacology, Seth GS Medical College and King Edward VII Memorial Hospital, Parel, Mumbai 400 012, India

---

This paper describes the anthropometric characteristics of 1,335 (males 545, females 790) people of low socio-economic classes aged 50-97 years (mean age 60 years) living in slums and tenement blocks around a major teaching hospital in central Mumbai (Bombay). Descriptive statistics for weight, mid-upper arm and calf circumferences, and biceps, triceps, subscapular and suprailiac skinfolds are presented. Subjects were much lighter, thinner and had smaller circumferences than their age- and sex- matched American counterparts but were similar to nationally representative Indian slum groups, as well as other Asian groups. Men were significantly heavier than women and had larger circumferences whereas women had significantly fatter skinfoldss. Age was significantly but non-linearly related to all variables in women but only to mid-upper arm and calf circumferences in men : there was a marked step effect with the age cut-off 70 years. Reliability for all measurements was high (R > 0.95), with technical errors of measurement highest for skinfolds, especially the suprailiac. The overall prevalence of oedema was 2.8%. In the whole sample, men were significantly older than women, probably because men are more likely to be out working than women, especially below 65 years old. A good participation rate was achieved (78%), with women more likely to participate than men. Almost half of the non-participancy can be attributed to work-related activity, particularly in men. Discussion focuses on practical issues of taking anthropometric measurements in elderly people living in the community, reliability, and non-participant bias.

Key words: Anthropometry, India, Mumbai, elderly, urban slums, poverty, weight, height, skinfolds, circumference, malnutrition, oedema

---

Introduction

Ageing and urbanisation phenomena in India are changing the course of national health and nutrition policies. Elderly people living in the cities must now rank among India’s priority concerns¹. As part of the process of developing new programmes to meet the needs of the urban poor, malnutrition and impaired functional ability amongst urban elderly people living in poor socio-economic conditions must receive attention and the gaps in 1000 knowledge filled without delay.

This paper presents the characteristics of weight, mid-upper arm and maximal calf circumferences, and four skinfold thicknesses, from a study of low-income elderly people living in urban India. Reflecting both lean and fat tissues, the measurement of weight is a crude measure of nutritional status but unusually low weight or rapid changes in weight are of interest, and weight is an essential component for the derivation of body mass index (BMI). A decrease in mid-upper arm circumference (MUAC) reflects recent weight loss of both adipose and lean tissue, and in combination with arm skinfolds, can be used to calculate other indices of muscle mass depletion which predict morbidity and mortality². Calf circumference is considered the most sensitive measurement of muscle mass in the elderly³. Skinfolds thicknesses, particularly triceps and subscapular, provide an estimate of total body fat in elderly people and are in common clinical use⁴.

This study forms part of a collaborative partnership between the Public Health Nutrition Unit, London School of Hygiene and Tropical Medicine (LSHTM) and HelpAge International, a non-governmental organisation committed to improving the lives of elderly people worldwide. Fieldwork took place in Mumbai between March and December 1993, and operated under the umbrella of an existing programme: the Biomedical Gerontology Centre of HelpAge India (BGCHI), based within the Department of Pharmacology of the Seth GS Medical College and King Edward VII Memorial (KEM) Hospital. The BGCHI was established in 1991 as a service-orientated research centre to assess the health status of poor elderly people living around the hospital, to diagnose, treat and monitor chronic and acute health problems. Ethics Committees of both the LSHTM and the KEM Hospital approved the nutrition research. The Central Drug Research Institute of India, a government body, from which the BGCHI had already received official sanction and under which its main scientific and managerial staff were employed, also gave its approval.

Study site and population

In 1991 the number of people over 60 years old in India stood at about 55 million, or about 6.5% of the total population of 844 million⁵. This is forecast to rise to 150 million (12%) by 2025. In 1991, the population of Mumbai, India’s foremost industrial and commercial centre, was approximately ten million⁶. The 1981 Census of India reported the proportion of over 50 year olds in Greater Mumbai as approximately 9% of the total population, large numbers of whom are living in slums and government-owned tenement blocks where most of the projected rise in the city’s population is likely to occur.

BGCHI study site was a defined administrative area (F/South Ward of the Municipality of Greater Mumbai) covering an area of 8.8 square kilometres located around the KEM Hospital. The study population consisted of people over 50 years old living in the largest slum and government-built tenement block (chawl) settlements within the Ward boundaries. Based on available Municipal data, BGCHI estimated that there were approximately 37,000 people over the age of 50 years living in the Ward, but as much of the area is occupied by industrial mills and factories and large medical and teaching institutions, and because pavement dwellers and smaller slum pockets were not included in the study, BGCHI calculated their target population to be about 3,100 people over 50 years of age. Because of the timing of the nutrition study, we aimed to cover just below half of that number.

Visiting door-to-door, BGCHI social workers conducted a household census of the largest slum and tenement blocks in F/South Ward from 1000 which a list of all those over 50 years old was extracted. There may have been some underestimation of the true population of the over 50’s as some were missed during the survey and there may also have been some falsification due to hostility or fear. However, we believe that the subjects seen in this study are generally representative of low-income elderly people living in the large slums and government-built tenement blocks in F/South Ward of Greater Mumbai.

Our socio-economic survey revealed that well over half (58%) of elderly people in the study area had no education at all, the rest having only basic schooling and low levels of literacy, especially amongst women. Only a third of the subjects were self-supporting, the rest being dependent on others. The overwhelming majority (98%) lived with other family members or friends, with only 2% living alone. Most elderly people (95%) lived in a single room, with five other people on average. Hinduism was the dominant religion (90%). Nearly three quarters of subjects had lived in Mumbai for more than 20 years.

Both slum and tenement blocks areas are overcrowded, and suffer from poor environmental and sanitation conditions, irregular water and electricity, and inadequate access to health care and social services. Preliminary analysis of the prevalence of health problems in the sample revealed the following : bilateral cataracts (54%), musculo-skeletal problems (40%), low haemoglobin (39%), hypertension (17%), skin infections (11%), chronic obstructive pulmonary disease and gastro-intestinal problems (each 6%), cardiovascular problems (5%), tuberculosis (4%), diabetes, neurological and upper respiratory tract infections (each 3%), and urogenitary problems (1%).

Methodology

Study implementation

Fieldwork was coordinated by one of the authors (MM) who recruited and trained a local team (doctor, social scientist supervisor, six field workers, two social workers) most of whom had previous experience of social work or primary health care in slums. The field team set up a temporary clinic in each community by turn in F/South Ward. About 30 elderly people were contacted daily from the census list and invited to the community clinic. Arrangements were made for home visits to elderly people who were too sick or frail to attend the clinic. Each subject signed (or made a thumb impression) an informed consent form after both social workers and medical officers in the field had fully explained the BGCHI objectives, facilities, examinations and confidentiality on previous occasions in home visits. An awareness programme using local organisations and word of mouth was used to publicise the clinics.

Anthropometric data collection

A range of 11 anthropometric measurements (weight, standing height, mid-upper arm and maximal calf circumferences, biceps, triceps, subscapular and suprailiac skin-folds, and armspan, demispan and knee height as approximates of stature) were conducted according to standard methodologies. Only weight, circumferences and skinfolds will be covered in this paper. Arm skinfolds were taken in triplicate but all other measurements were repeated twice only. Subjects moved between replicate measurements.

Weight was measured on digital weighing scales (Soehnle model S Sport no. 770102), calibrated in 100g units and recorded to the nearest 0.1kg. The scales were placed on level ground adjacent to a wall and footprints drawn on the standing platform to indicate the correct standing position. Subjects removed any heavy items in their pockets or sari waist pouches such as keys and coins. No correction 1000 was made for voiding prior to the measurement or the weight of clothing (very light cotton, weighing less than 0.5kg).

Mid-upper arm, and maximal calf circumferences were measured on the left side of the body whenever possible using a flexible steel tape (3 m Stanley tape model 32-031). The arm midpoint was first marked using the distance between the tip of the shoulder (acromial process) and the tip of the elbow (olecranon process), and the mid-arm circumference was then measured with the arm hanging loosely at the side of the body. Calf circumference was measured in the recumbent position with the maximal point determined visually. Both circumference measurements were taken to the nearest 0.1 cm. Care was taken to ensure that the tape fitted snugly around the skin but not so tight that it compressed the tissue.

Triceps, biceps, subscapular and suprailiac skinfold measurements were taken in rapid succession using Holtain Tanner/ Whitehouse calipers. The skinfold was released between each replicate measurement and fingers remained holding the skinfold whilst the calipers were applied. Measurements were recorded to the nearest 0.2mm after at least 3 seconds to allow for the increased compressibility of skin in older subjects⁷.

As fluid retention and oedema are known to affect the accurate measurement of weight, circumferences and skin-fold measurements, depending on the severity and location in the body, any visible oedema (on feet, ankles, legs, arms or face) was noted by the medical officer and field observer.

Quality control and reliability

Each weighing scale was calibrated regularly against local cast-iron weights (total 40kg), and skinfold calipers were checked against wood of known and constant thickness. Recording error was minimised by daily checking of recording forms for obvious mistakes, with any suspect measurement repeated later. As a safeguard against imprecision, we used pre-set limits for repeated measurements^4,7. These were 0.2kg for weight; 0.5cm for arm and calf circumferences; 1mm for arm skinfolds; and 4mm for trunk skinfolds. If a repeat set of measurements did not fall within its specified pre-set limit the measurement set was repeated.

Intra- and inter- observer errors were calculated for all measurements at the end of the initial training period and mid-way through the fieldwork on 20 elderly volunteers from a day-care centre near the study site. Measurement sessions were conducted blind so that no reference could be made to previous results, and anatomical sites were not marked. The two error estimates recommended to determine reliability, the technical error of measurement (TEM), and the coefficient of reliability (R), were calculated using the equations for more than two observers⁸.

Definition of "elderly" and age determination

There is no consistent definition of "elderly". Many international organisations such as the UN refer to the elderly as those over 60 years of age^9,10, often based on the upper quintile of the population. However, the equivalent upper quintile constituting the "elderly" population in some developing country may mean including people as young as 45 or 50¹¹. Moreover, given long-term malnutrition, disease exposure, physical work patterns and generally harsh life conditions common in many developing countries, the process of biological ageing occurs earlier and proceeds faster than in developed populations so that an individual may be biologically "old" at a chronological age lower than 60 years¹². Taking these factors into 1000 account, we took the cut-off 50 years to define our "elderly" population.

There are problems in determining age with accuracy amongst elderly people with low levels of education^13,14. Some older people, particularly women, are unused to, or suspicious of, completing forms or answering questions about their lives, and many do not have any formal registration of birth. Assuming that self-reported age alone would be inaccurate, we asked a series of questions related to well-known historical events (e.g. the "Quit India" movement, dock explosion in 1944, religious riots) and a secondary series about age at birth of first surviving child, and that child’s age. A "best guesstimate" was the mean of all the responses. Out of 1,398 subjects measured, 63 cases were judged to be below 50 years old and have been excluded from the analysis.

Data analysis

All statistical analyses were performed with the Statistical Package for Social Sciences, SPSS/PC version 4.0¹⁵. Histograms and box plot procedures were used to determine whether or not variables were approximately normally distributed. Pearson’s Chi Squared (c ²) was used to determine whether the prevalence of oedema was related to sex and age, and Yates corrected c ² to explore demographic differences. T-tests were used to determine whether sex and community type differences in measurements were significant. The relationships between anthropometric variables and age were first explored using linear regression and plotting residuals, and subsequently in the case of non-linear relationships with t-tests. Pearson product moment correlation coefficients (r) were obtained for the examination of the relationships between independent anthropometric variables by gender.

Results

Coverage

Table 1 shows the demographic characteristics of the elderly subjects covered by age group, sex and by type of settlement, presented in five year age groupings. Out of the total study sample of 1,335, the overall sex ratio was 41:59 in favour of females (545 males; 790 females). More than half (53%, n=719) are under 60 years old and 47% (n=616) over 60. Only 158 (12%) are over 70 years old. Males in the whole sample are significantly older than women (c ²=31.8, df=4, p<.0001). Out of the total sample, 58% (n=777) were from the slums and 42% (n=558) from the tenement blocks. The male : female ratios for slum and tenement block dwellers were not significantly different from the sex ratio for the entire sample. However, slum dwellers were significantly younger than the tenement block dwellers (c ²=17.1, df=5, p<.001), with 35% of the entire sample made up of slum dwellers under 60 years. The over 60 year-olds accounted for the same proportion in both types of communities (23%). The smallest group was tenement block dwellers under 60 years old (19% of the total sample). The number of housebound elderly subjects seen was 32 (2.4%). There was no association with sex but, as expected, housebound subjects were significantly older than non-housebound subjects, (c ²=73.4, df=5, p<.0001) with more than half (56%) over 70 years old.

Table 1. Demographic characteristics.

		Totals		M	F	M : F
		n	%	n	N	ratio
Totals	All	1335	100	545	790	41 : 59
	Slum	777	58	307	470	40 : 60
	Chawl	558	42	238	320	43 : 57
Age
50 - 54	All	390	29	121	269	31 : 69
	Slum	254	19	72	182	28 : 72
	Chawl	136	10	49	87	36 : 64
55 - 59	All	329	25	125	204	38 : 62
	Slum	211	16	83	128	39 : 61
	Chawl	118	9	42	76	36 : 64
1000 60 - 64	All	298	22	142	156	48 : 52
	Slum	165	12	78	87	47 : 53
	Chawl	133	10	64	69	48 : 52
65 - 69	All	160	12	85	75	53 : 47
	Slum	84	6	47	37	56 : 44
	Chawl	76	6	38	38	50 : 50
> 70	All	158	12	72	86	45 : 55
	Slum	63	5	27	36	43 : 57
	Chawl	95	7	45	50	47 : 53

A total of 297 eligible subjects did not participate, giving a participation success rate of 78%. We found a significant sex difference with women more likely to be participants than men (c ²=22.8, df=2, p<.0001), and the overall male : female ratio amongst the non-participants was 56:44, a reversal of that in the participant sample. For both sexes, non-participation was significantly higher in the younger old (below 60 years) than in the older old (over 60 years) (c ²=8.1, df=2, p<.001 in males: c ²=21.5, df=2, p<.0001 in females). From socio-economic information recorded in the initial survey, we determined that approximately 40% of all 297 non-participants were working, of which 82% were men. Non-participant subjects of both sexes under 60 years were far more likely to be working than those over 60 years (c ²=27.1, df=2, p<.0001 for men; c ²=7.1, df=2, p <.001 for women). We followed up 101 non-participants over 60 years of age (34% of all non-participants) and found that indeed work, or the search for work, were the main reasons given for non-participation in the study.

Anthropometric descriptives

There was noticeable oedema in 38 subjects (2.8%) which was more common in the feet and ankles (74%) than in the upper body (26%). No associations were found between the presence of oedema and age or sex. Oedemat 1000 ous subjects were significantly heavier than non-oedematous subjects (t=-4.9, df=1326, p<.0001), and had significantly greater skinfolds and circumference (p<0.0001). Thus we excluded cases with oedema from the analysis, leaving a total sample size of 1,297 (531 males and 766 females). Excluding the cases with oedema eliminated much of the skew and kurtosis in the distributions of most variables, which then approximated normality. However, some degree of positive skew still persisted in the distributions of trunk skinfolds in women.

Descriptive statistics (mean, standard deviation, and range) for all variables by age group and gender are presented in Table 2. Overall percentiles are presented by gender in Table 3.

Table 2. Descriptive statistics of anthropometric variables by age group and sex.

1000 1000

5.0

		Age group in years
		50 - 54		55 - 59		60 - 64		65 - 69		> 70
		M	F	M	F	M	F	M	F	M	F
Weight (kg)	n	119	262	121	194	137	149	83	72	68	85
	mean	53.6	46.5	52.8	46.1	53.8	45.3	53.0	45.6	49.7	39.4
	SD	13.2	10.1	11.6	9.9	10.5	11.7	10.8	9.7	9.5	9.9
	minimum	26.8	28.5	34.8	25.4	30.2	23.9	30.5	26.3	29.9	23.4
	maximum	117.9	73.2	101.8	75.8	86.5	93.0	85.4	70.3	71.0	77.3
MUAC (cm)	n	120	262	121	195	138	151	83	72	69	86
	mean	24.3	23.8	24.3	23.2	24.3	22.8	23.8	22.6	22.6	20.6
	SD	< 1000 font size="2">4.0	3.6	3.3	3.7	3.2	4.2	3.3	3.4	2.7	3.8
	minimum	15.1	15.5	16.0	15.3	18.0	13.5	16.5	15.8	16.0	14.8
	maximum	37.4	34.3	34.5	34.0	31.3	34.0	31.5	31.3	27.5	30.3
Calf circum (cm)	n	120	262	121	194	138	151	83	72	69	84
	mean	29.6	27.7	29.4	27.3	29.3	26.8	29.1	26.8	28.0	24.5
	SD	3.7	3.4	3.4	3.3	3.4	3.8	3.3	3.3	3.0	3.1
	minimum	20.0	21.3	23.0	20.3	21.0	17.8	21.3	20.3	21.0	17.8
	maximum	39.4	37.3	38.4	36.8	38.0	37.3	36.5	36.3	36.0	31.8
Triceps (mm)	n	120	262	121	195	138	151	83	72	69	86
	mean	10.0	15.3	9.7	14.5	10.2	14.1	10.8	13.3	9.6	10.8
	SD	5.2	6.6	4.3	6.5	4.1	6.5	5.0	5.3	4.2	5.7
	minimum	3.0	3.9	3.0	2.0	3.1	2.0	2.9	4.3	3.0	2.5
	maximum	37.4	38.7	27.9	33.5	23.4	35.9	26.2	30.0	21.0	27.5
Biceps (mm)	n	120	262	121	195	138	151	83	72	69	86
	mean	4.6	5.8	4.4	5.8	4.6	5.6	4.7	5.3	4.3	4.5
	SD	2.6	3.1	2.0	3.1	2.1	3.0	2.1	2.3	1.8	2.4
	minimum	1.9	2.0	1.9	1.5	1.7	1.7	2.1	2.1	2.1	1.8
	maximum	20.1	23.7	11.9	22.2	13.2	19.1	13.9	14.0	10.9	14.5
Subscap (mm)	n	120	262	121	195	138	151	83	72	69	86
	mean	17.3	23.5	16.0	21.7	16.0	20.9	16.0	21.0	13.9	15.5
	SD	8.6	9.8	8.1	10.7	7.5	10.6	7.3	9.9	6.3	9.6
	minimum	4.5	3.9	5.1	3.2	4.8	2.8	5.3	5.7	4.9	4.0
	maximum	40.0	40.0	39.4	40.0	38.6	40.0	25.1	39.6	36.3	40.0
Suprailiac (mm)	n	120	261	121	193	138	151	83	72	69	83
	mean	15.1	21.3	14.4	20.8	14.3	19.4	13.6	19.9	12.3	15.3
	SD	8.0	9.4	7.4	9.7	6.6	9.9	5.8	8.4	5.9	8.7
	minimum	3.0	4.5	3.2	2.3	3.5	3.1	6.8	3.6	3.9
	maximum	39.0	40.0	38.8	39.6	33.1	39.4	28.1	38.4	30.7	40.0

All cases with oedema are excluded

Table 3. Percentiles for anthropometric variables by sex.

1000

	Weight (kg)		Muac (cm)		Calfcir (cm)		Triceps (mm)		Biceps (mm)		Subscap (mm)		Suprail (mm)
%ile	M	F	M	F	M	F	M	F	M	F	M	F	M	F
5	36.7	30.0	18.2	16.9	23.8	21.8	4.1	5.2	2.1	2.4	6.1	6.2	5.0	6.5
10	39.6	32.7	19.5	18.2	25.0	22.8	4.9	6.2	2.4	2.7	7.5	8.0	5.9	7.8
20	42.6	35.9	20.9	19.5	26.2	24.0	6.0	8.0	2.9	3.3	8.9	10.5	7.7	10.6
30	46.0	38.8	22.0