Malnutrition among Women in Kerala : An Analysis of Trends, Differentials and Determinants

Malnutrition among Women in Kerala:
An Analysis of Trends, Differentials and Determinants

P. Ramesh

Research Fellow
Population Research Centre
Gokhale Institute of Politics and Economics
(Deemed University)
Pune – 411 004
Maharashtra, India

Abstract

While under-nutrition (underweight and stunting) is still prevalent in
most of the developing countries, the rates of overweight and obesity are
steadily
increasing,
especially
among
adults.
In
Kerala,
overweight/obese exceeded underweight. This paper attempts to
examine the trends in the shift from underweight to overweight and
identify the major determinants of the co-existence of ‘double burden’ of
malnutrition among women of reproductive age 15-49 years in Kerala
using the data from National Nutrition Monitoring Bureau (NNMB) and
the Second National Family Health Survey (NFHS-2, 1998-99). The
results of the multivariate logistic regression analyses show that
household standard of living, religion and age are significantly
associated with both underweight and overweight/obesity. On the other
hand, woman’s education, work status, residence and caste are not
significant on women’s nutritional status. In summary, both chronic
energy deficiency and overweight/obesity are widespread in Kerala and
there is a need for public health programs that are able to address both
simultaneously.

Introduction

Malnutrition refers to any disorder of nutrition — whether it is due to dietary
deficiency, called under-nutrition, or to excess diet, called over-nutrition (Britannica
Student Encyclopedia, 2005). Malnutrition results from imbalance between the needs of
the body's and the intake of nutrients. Malnutrition worldwide includes a spectrum of
nutrient–related disorders, deficiencies, and conditions such as intrauterine growth
retardation, protein-energy malnutrition, iodine deficiency disorders, vitamin A
deficiency, iron-deficiency anemia, and overweight/obesity and other diet-related non-
communicable diseases (Ratzan et al., 2000).

In earlier days, developing countries experienced high prevalence of under-nutrition,
but this era of transition has brought a double burden of under-nutrition and over-
nutrition. While under-nutrition (underweight and stunting) is still prevalent in most of
the developing countries, the rates of overweight and obesity are steadily increasing,
especially among adults. Hence, the countries in transition face today new public health
problems, while they are yet to eradicate completely the nutritional deficiencies. Once

1

---

considered a problem related to affluence, overweight and obesity is growing rapidly in
many developing countries. This is due in large part to increasing urbanization and
changes in diet and life style, in particular the “nutrition transition” away from fruit,
vegetables and greater consumption of more ‘energy-dense, nutrient–poor’ diets,
dependence on television for leisure along with reduced levels of physical activity
(World Health Organization 2000, 2003). Overweight and obesity is a risk factor for a
number of chronic non-communicable diseases, such as diabetes, hypertension, asthma,
cardiovascular disease, some cancers, gall bladder disease and osteoarthritis – all of
which are on the rise in developing countries, particularly among the middle-class,
urban populations (Gopalan, 1998; Popkin et al., 2001). On the other hand, the Chronic
Energy Deficiency (CED) is associated with impaired physical capacity (Durnin, 1994),
reduced economic productivity (Kennedy, 1994; Untoro, 1998), increased mortality
(National Institute of Nutrition, 1991) and poorer reproductive outcomes (World Health
Organization, 1995; Schieve et al., 2000). Some evidence in developing countries
indicates that malnourished individuals, that is, women with a Body Mass Index (BMI)
below 18.5 kg/m2, show a progressive increase in mortality rates as well as increased
risk of illness (Rotimi et al., 1999). The World Health Organization estimates that in
1995, about one million adult deaths resulted from health problems exacerbated by
over-nutrition, while half of it were associated with under-nutrition (WHO, 1998).

For social and biological reasons, women of the reproductive age are amongst the most
vulnerable to malnutrition (UNACC/SCN, 1992). Several reviews have also
emphasized the vulnerability of women throughout their life cycle (Tinker, 1995;
Merchant and Kurtz, 1993). Many factors have been associated with both forms of
malnutrition of women in the literature. These include the socioeconomic (e.g.,
occupation, educational background and the standard of living); cultural (e.g., religion
and caste); the demographic (e.g., age and marital status) and dietary characteristics (De
Vasconcellos, 1994; Shetty and James, 1994; Stunkard, 1996; Griffiths and Bentley,
2001; Monteiro et al., 2002, 2004b; Shukla et al., 2002; Shetty, 2002; Radhakrishna
and Ravi, 2004; Radhakrishna et al., 2004; Roy et al., 2004). Based on the analysis of
anhropometric measurements for women age 20-49 in 36 developing countries, Mentez
et al., (2005) observed that the proportion overweight exceeded the proportion
underweight in a majority of the counties in both urban and rural areas. These results
are contrary to the general belief that in developing counties overweight is less
prevalent than underweight and that it is primarily concentrated in urban, higher
socioeconomic status households. There are several studies on nutrition transition in
Asia and the Pacific, as well as the developing world, in general (Popkin 1994; 1998;
Popkin et al., 2001). In India, which is typically known for large incidence of under-
nutrition, significant proportions of overweight and obese now coexist with the
undernourished (IIPS and ORC Macro, 2000) and there is some evidence of even
emerging nutrition transition also (Shetty and James, 1994; Griffiths and Bentley, 2001;
Shetty, 2002; Shukla et al., 2002; Radhakrishna and Ravi, 2004; Radhakrishna et al.,
2004; Arnold et al., 2004).

Prevalence of Malnutrition among Women in India

Though, malnutrition among women has long been recognized as a serious problem in
India, but national-level data on levels and causes of malnutrition have been scarce.
However, with the availability of data from the Second National Family Health Survey
(NFHS–2) carried out in 1998-99 all over the country, we may be able to sort out such

2

---

limitations. The NFHS–2 collected anthropometric data from a nationally representative
sample of 92,486 households (IIPS & ORC Macro, 2000). This data provides an
opportunity for examining the prevalence of malnutrition among ever-married women
of reproductive age 15-49.

The Body Mass Index [BMI, weight (kg)/height (m)2] can be used to assess individual
and community nutritional status (Bailey and Ferro-Luzzi, 1995). As per the Second
National Family Health Survey, the mean body mass index for women of age 15-49 in
India is 20.3, varying within the narrow range of 19–23 among various states, from 19.2
in Orissa to 23.7 in Delhi (IIPS and ORC Macro 2000:244). Table A–1 shows the
percentage of women age 15–49 according to level of body mass index by state. More
than 60 per cent of women in Kerala and all the northeastern states (except Tripura) fall
in the healthy weight category (BMI, 18.50–24.99). Punjab, Orissa, West Bengal, Goa,
Gujarat, Maharashtra, Andhra Pradesh, Karnataka and metro cities of Kolkata and
Mumbai has a level lower than the national average of 53 per cent (Table A–1, Col. 5).
The BMI can also be applied to define chronic energy deficiency/ underweight (Shetty
and James, 1994; Ferro-Luzzi et al., 1992) and overweight/obesity (Garrow, 1988).
Underweight/CED is usually indicated by a BMI of less than 18.5 kg/m2 and
overweight and obese indicated by a BMI of more than 25.0 kg/m2. More than one-third
(36 per cent) of women in India have a CED (20 per cent have mild CED, 9 per cent
have moderate CED, and 7 per cent have severe CED). At state level, the disparities are
quite widespread and CED ranges from 11 to 48 per cent. Prevalence was ?20 per cent
in 19 (70 per cent) of the 27 states and ?30 per cent in 11 states (41 per cent) of India.
Among the states, the level of CED is the high in Orissa (48 per cent) followed by West
Bengal (44 per cent). Andhra Pradesh, Bihar, Gujarat, Karnataka, Madhya Pradesh and
Maharashtra have a CED in the range of 37-40 per cent. The lowest percentages (11–
12) of women with chronic energy deficiency/underweight (low BMI) are found in the
states of Arunachal Pradesh, Sikkim, and Delhi (Table A–1, Col. 4).

Although under-nutrition is still prevalent, there is an alarming prevalence of
overweight and obesity among Indian women of childbearing age. Eleven per cent of
ever-married women of age 15-49 in India are overweight or obese and it ranges from
4–34 per cent. Among the states, the level of overweight or obesity is much high in
Delhi (34 per cent) followed by Punjab (30 per cent) and least common (less than 10
per cent) in Rajasthan and all parts of central, eastern and northeastern (except Sikkim
and Manipur) states (Table A–1, Col. 8). In all the three metro cities and Delhi, Punjab,
Sikkim and Kerala states, the proportion of overweight or obese is exceeded the
proportion of underweight. In Kerala, about two-fifths (20 per cent) of underweight
women are coexisting with 21 per cent of overweight or obese women.

Most of the research on determinants of malnutrition among women in India has
concentrated either on underweight (for example, Shetty and James, 1994; Singh, 1999)
or overweight (Gopalan, 1998; Dudeja et al., 2001). Very little has been done on the
emerging dual burden of both forms of malnutrition, particularly among women
(Griffiths and Bentley, 2001; Shukla, 2002; Roy et al., 2004). With this backdrop, this
paper sheds light on the emerging ‘double burden’ of malnutrition among women in
Kerala and its linkages with socioeconomic, demographic and other factors.
Specifically, this paper will attempt first, to explore the trends in the shift from
underweight to overweight among rural women during the last three decades and
finally, it analyses the differentials and identify the distinct factors that may influence

3

---

malnutrition among women of reproductive age 15-49. In this paper, malnutrition
covers problems of both under-nutrition and over-nutrition.

The Setting

Kerala is situated on the southwest coast of India with the Western Ghat Mountains
forming its eastern border. Various demographic and socioeconomic features of Kerala
and India are outlined in Table A–2. According to the 2001 Census, Kerala had a
population of 31.8 million, accounts for 3.1 per cent of India’s population and for 1.2
per cent of its land area. Kerala is predominantly an agricultural state with 73 per cent
of the population living in rural areas. Kerala has achieved remarkable progress in
human development, as reflected in the high levels of education and health of its
population. Kerala’s demographic experience has attracted wide attention both at the
national and the international levels (Bhat and Irudaya Rajan, 1990; Zachariah and
Irudaya Rajan, 1997; Irudaya Rajan and Aliyar, 2005). Crude death rate, infant
mortality rate, and life expectancy at birth are comparable even to those in the
developed countries. As of 2002, the birth rate in Kerala was estimated as 16.9 births
(per 1000 population), as against 25.0 for all-India. The crude death rate was 6.4 deaths
(per 1000 population), compared to the national average of 8.1 (India, Registrar
General, 2004). Thus, Kerala has apparently entered the third or final phase of the
demographic transition characterized by low death rate and declining birth rate leading
to a slow down in the growth rate of population. Kerala has made significant advances
not only in demographic transition but also in epidemiological, and health care
transitions (for more discussion on epidemiological and health care transitions in
Kerala, see Panikar and Soman, 1981; Panikar, 1999).

Trends of Malnutrition and Food Consumption in Rural Kerala

The National Nutrition Monitoring Bureau (NNMB) conducted diet and nutrition
surveys in three periods (1975-79, 1988-90 and 1996-97) and two separate rural diet
and nutrition surveys (1990-91 and 2000-2001) in a sub-sample of villages of NSSO,
adopting the sampling design of NSS consumer expenditure survey in rural areas of
nine states including Kerala. The availability of repeated, representative cross-sectional
surveys provides a basis for the careful understanding of secular trends in the nutritional
status of population in rural Kerala. From these NNMB surveys, among rural women, a
decreasing trend in the prevalence of underweight/CED and an increasing trend in the
proportions of ‘normals’ and overweight or obese were observed over the last two
decades. Between 1975-79 and 2000-2001, the extent of CED declined from 47 per cent
in 1975-79 to 25 per cent in 1996-97 (NNMB repeat surveys) and from 28 per cent in
1991-92 to 19 per cent in 2000-01 (NNMB–NSS sample design). On the other hand,
during the same period, a consistent increasing trend was observed in the proportions of
overweight or obese among rural women from 2 per cent to 12 per cent and then from
16 to 24 per cent, respectively (Table A–3).

Since the formation of the State in 1956 to 1970 (i.e., during the period 1950-70), the
nutrient intakes (both energy and proteins) had been consistently lower than the
recommended dietary allowances (Panikar and Soman, 1981: 24). The average intakes
of energy and protein observed during the three NNMB survey periods (during 1975-79
to 1996-97) reflect almost steady increase. Between 1975-79 and 1996-97, the average
intake of energy by rural households increased from 1,978 to 2,106 Kcal, but below the

4

---

recommended dietary allowances (RDA) of 2,425 Kcal. On the other hand, the average
intake of proteins in rural households increased from 46 g to 56 g, as against the RDA
of 60g. During the same periods (1975-79 and 1996-97), the proportion of households
with protein-adequacy and calorie-adequacy showed an increase of 28 per cent and 13
per cents, respectively. Thus, the proportion of households with adequate protein and
energy intake registered a moderate increase between the late 1970s and the 1990s
(Table A–4).

Data and Methods

Data: The anthropometric data used for analysis in this paper were derived from the
Second National Family Health Survey, 1998-99 (NFHS–2) for the state of Kerala. The
objective of the survey were to provide state-level estimates on fertility, family planning
practices, infant and child mortality, reproductive and child health, quality of health and
family welfare services and nutrition of women and children. Data were collected
between March and July 1999 from 2,884 ever-married women of age group 15-49 in
2,834 households. The household response rate was 98 per cent and women’s response
rate was 93 per cent. In addition, the survey also collected measures of height and
weight from 2,770 ever-married women (96 per cent of those ever-married women
interviewed). The details of the study design as well as sampling frame and sample
implementation are provided in the national and state NFHS–2 reports (IIPS and ORC
Macro, 2000; 2001).

Measures: In NFHS–2, all ever-married women aged 15-49 years were weighed using a
solar-powered scale with an accuracy of ±100 g. Their height was measured using an
adjustable wooden measuring board, specifically designed to provide accurate
measurements (to the nearest 0.1 cm) in a developing country field situation. The
weight and height data were used to calculate the BMI. The BMI is defined as weight in
kilograms divided by the height in meters squared (kg/m2). According to the WHO
(1995) recommendations, the BMI variable was categorized into six groups for
preliminary analysis. The six groups identify women who are: severely thin (BMI,
<16.00 kg/m2), moderately thin (BMI, 16.00–16.99 kg/m2), mildly thin (BMI, 17.00–
18.49 kg/m2), normal weight (BMI, 18.5–24.9 kg/m2), overweight (BMI, 25.0–29.9
kg/m2) or obese, BMI?30 kg/m2. For the 15-17 years old females in the NFHS–2
sample, the cut-off points recommended by Cole et al., (2000) for adolescent obesity
and overweight were used because the BMI values change substantially with age in
adolescents. For obesity, these points are 29.29 for 15-years old, 29.56 for 16-years old
and 29.84 for 17-years old females. For overweight, the recommended cutoff points are
24.17 for 15 years old, 24.54 for 16-years old and 24.85 for 17-years old females.
Normal weight categories used for adolescent females were 18.5–24.17 for 15-years
old, 18.5–24.54 for 16-years old and 18.5–24.85 for 17-years old females. The same
definition of underweight/thinness was used for all women in the sample regardless of
age because no age-specific definitions of thinness for adolescents have been suggested
(Cole et al., 2000). The WHO (1995) overweight grade–III definition of BMI >40 was
not used because of only five women in the sample. Women who were pregnant at the
time of the survey or women who had given birth during the two months preceding the
survey were excluded from the analysis. This is to avoid the exaggerated BMI values
for the women due to their pregnancy status.



5

---

Table 1: Definition of Variables and Percentage Distribution of Ever-married Women
Age 15–49, NFHS–2, Kerala, 1998-99
Variable
Definition of the Variable
Per cent
Socio-cultural and Economic

Place of Residence


Rural
Women lives in Rural areas
76.9
Urban
Women lives in Urban areas
23.1
Religion


Hindu
Women lives in a household whose head is Hindu
51.3
Muslim
Women lives in a household whose head is Muslim
32.6
Others
Women lives in a household whose head is neither
16.1
Hindu nor Muslim
Caste


SC/ST
Woman lives in a household whose head belongs to
9.8
Scheduled Caste (SC) or Scheduled Tribe (ST)
OBC
Woman lives in a household whose head belongs to
43.1
Other Backward Class (OBC)
OC
Woman lives in a household whose head does not
47.0
belong to OBC or SC or ST
Education


Illiterate
Woman is illiterate
12.6
Literate, < middle
Woman literate with less than middle school education
30.2
Middle school
Woman is literate with middle school education
17.1
High school & above
Woman is literate with high school or above education
40.2
Work Status


Not working
Woman is currently not working besides from her own
76.9
household work
Working
Woman is currently working besides from her own
23.1
household work
Standard of Living


Low
Woman lives in a household with low standard of
15.5
living
Medium
Woman lives in a household with medium standard of
55.1
living
High
Woman lives in a household with high standard of
29.3
living of living
DEMOGRAPHIC


Current Age


15-24
Woman’s age is 15–24 years at the time of the survey
15.9
25-34
Woman’s age is 25–34 years at the time of the survey
37.4
35-49
Woman’s age is 35–49 years at the time of the survey
46.7
Marital Status


Currently married
Woman is currently in married status
92.8
Not currently married
Woman is widowed or divorced/separated or
7.2
abandoned
Life Style (Media & Food Habits)

Watches Television (TV)
Every week
62.4
Frequency of Eating Fruits
Daily
17.9
Weekly
38.6

Occasionally
41.8

Never
1.7
Frequency of Consuming
Daily
61.6
Meat/Chicken/Fish
Weekly
21.2
Occasionally
13.5

Never
3.6
N
Total Number of women
2884
Source: NFHS-2, 1998-99 Data files of Kerala.


6

---

Dependent variables: Based on the WHO cutoffs, a three-category variable of
nutritional status of women was created, indicating underweight (BMI <18.50 kg/m2),
normal weight (BMI 18.5–24.9 kg/m2) and overweight or obese (BMI ?25.00 kg/m2).
Among 2770 ever-married women from whom anthropermetric data were collected,
measurements of height and weight for six women were incomplete or affected by
measurement errors. Women who were pregnant at the time of the survey are 145 and
31 women gave birth during the two months preceding the survey. These 181 women
(6+145+31) were excluded from the analysis. Thus, 2589 ever-married women were
included in the analysis, of whom, 489 were underweight, 1560 were normal weight
and 540 were overweight or obese.

Explanatory variables: The effect of one variable on the prevalence of malnutrition is
likely to be confounded with the effects of other variables. Therefore, socioeconomic
demographic and life style characteristics were controlled statistically. The variables
included as controls are: residence, religion, caste/tribe, education, work status,
standard of living (measured by an index based on the housing conditions and
ownership of consumer durables, which is used as a proxy for economic status),
current age, marital status, media and food habits. For definition and categories of
these variables, see Table 1.

Analysis: In analyzing the data, both bivariate and multivariate analyses were carried
out. In the bi-variate analysis, the chi-square test is employed to see the association
between each of the independent variables and the nutritional status of women. The chi-
square test in the bi-variate analysis does not consider confounding effects. Therefore,
multivariate analysis is used to examine the net effect of each independent variable on
malnutrition of women, while controlling for the other independent variables. Two
separate logistic (logit) regression models were used to identify and to compare the
factors associated with women being underweight (BMI <18.50 kg/m2) in the first
model (n=2,039) and women being overweight or obese (BMI ?25.00 kg/m2) in the
second model (n=2,112). In both of the models, underweight and overweight women
were compared with normal weight (BMI 18.5–24.9 kg/m2) women. Details of the
multivariate statistical technique used in the analysis are provided in the Appendix–2. In
the survey, certain geographical regions of Kerala and certain categories of households
were over-sampled. Therefore, survey–specific sample weights are used to restore the
representativeness of the sample. Further, sample sizes varied between analyses as a
result of missing data for certain variables, and sample sizes were lower in the
multivariate analyses than in the bi-variate analyses.

Results

Characteristics of women

The percentage distribution of ever-married women of age 15-49 according to selected
socioeconomic, demographic characteristics (i.e., residence, religion, caste/tribe,
education, work status, standard of living, age, and marital status) and life style
indicators such as media and food habits are presented in Table 2. More than three-
fourths (77 per cent) of respondents were from rural areas. Fifty-one per cent of the
respondents were Hindus and 33 per cent were Muslims. Forty-three per cent of
respondents belong to the other backward classes, ten per cent belong to the scheduled
castes/tribes and nearly half (47 per cent) of women did not belong to any of these caste

7

---

or class groups. Thirteen per cent of women were illiterates and two-fifths (40 per cent)
of women completed high school and above level of education. More than three-
quarters of women (77 per cent) did not participate in work other than their regular
housework during the 12 months preceding the survey. Twenty-nine per cent of women
hailed from high household standard of living group and 16 per cent from low standard
of living group. Nearly half of respondents (47 per cent) were in the age group 35-49
and more than one-third (37 per cent) of respondents were in the age group of 25-34.
The very small share (16 per cent) of the age group 15-24 was due to the relatively high
age at marriage in Kerala. Ninety-three per cent of respondents were currently married
and seven per cent were not currently married i.e., widowed, divorced, separated, or
abandoned women. With regard to lifestyle indicators, such as media and food habits,
62 per cent watched television at least once a week. More than three-fifths (62 per cent)
of women consumed chicken or meat or fish daily and 83 per cent consumed these
items at least once a week. In addition, 18 per cent of women consumed fruits daily and
57 per cent consumed them at least once a week.

Bi-variate Analysis: Differentials of Malnutrition

The percentage of ever-married women of age 15-49 in Kerala with chronic energy
deficiency/underweight and the percentage with overweight or obese by selected
background characteristics are presented in Table 2. The results of the bi-variate
analysis using chi-square test showed a significant association between the BMI groups
and each of the explanatory variables. More than one-fifths (21 per cent) of women
were overweight or obese, but this percentage is considerably higher for some groups of
women (30 per cent of women who consume fruits daily, 26 per cent of women of age
35-49, 25 per cent of urban women, 23 per cent of women who consume chicken/meat/
fish daily, or who have completed at least high school education). Moreover, it has
been observed that the prevalence of overweight or obese varies significantly
according to the household’s living standard. Among the higher economic group, the
prevalence of overweight or obese was 33 per cent and it declines to 9 per cent for the
lower economic group.

The total prevalence of underweight/CED among the Kerala women was 19 per cent.
CED was particularly serious for women who belong to households with a low
standard of living i.e., women from households with a low standard of living were
more than thrice as likely to have a low BMI as women from households with a high
standard of living. In addition, underweight or suffering from CED was particularly
pronounced among women of scheduled caste (28 per cent), illiterate women (26 per
cent), younger women (25 per cent) Hindu or workingwomen (22 per cent), and rural
women (20 per cent). Women who eat fruits or chicken, meat, or fish on a daily basis
were less likely than the other women to have CED. Moreover, it has been observed
that the prevalence of overweight or obese exceeded underweight among women from
urban areas, Muslim, Christian and ‘other’ caste (OC) women, well educated, non-
working, women from higher standard of living group, older women, who watches
television at least once in a week and who consumes fruits daily (Table 2).






8

---

Table 2: Percentage Distribution of Non-pregnant Women Aged 15-49 Years, Classified as
Underweight (BMI, <18.50 kg/m2), Normal Weight (18.50–24.99 kg/m2) and Overweight or
Obese (BMI, ?25.00 kg/m2), according to Selected Background Characteristics, Kerala,
NFHS-2, 1998-99

Under
Normal
Over
Number of
2
? value &
Characteristic
weight
weight
weight
Women2
level of
(<18.50)
(18.5-24.99)
(?25.00)
(N)
Significance
Level of BMI1
Place of Residence




16.35***
Rural
20.2
60.2
19.5
1987

Urban
14.3
60.2
25.5
601

Religion




36.32***
Hindu
21.8
61.1
17.1
1351

Muslim
17.4
58.0
24.6
814

Christian
12.5
61.8
25.7
424

Caste





Scheduled caste/tribe
28.2
59.9
11.9
252

Other backward class
19.9
59.6
20.5
1125

Other caste
16.0
60.9
23.1
1212

Education




26.31***
Illiterate
26.5
59.9
13.6
339

Literate, < middle
19.8
59.4
20.9
805

Middle school
17.3
60.9
21.8
445

High school and above
16.2
60.9
22.9
999

Work Status




12.22**
Not working
17.8
60.0
22.2
1971

Working
22.2
61.2
16.6
616

Household Standard of Living



165.66***
Low
33.7
57.1
9.1
406

Medium
19.3
62.9
17.9
1422

High
10.3
57.0
32.8
760

Current Age




76.09***
15-24
25.4
66.6
8.0
350

25-34
22.4
59.0
18.7
953

35-49
14.5
59.5
26.0
1285

Marital Status




10.287**
Currently married
18.2
60.5
21.3
2397

Not currently married
26.9
57.5
15.5
193

Watches TV Every Week




58.22***
No
25.5
59.2
15.3
961

Yes
15.0
60.8
24.2
1627

Fruits Daily




31.78***
No
19.8
61.4
18.9
2141

Yes
14.7
54.8
30.5
449

Chicken/Meat or Fish Daily



9.66**
No
20.2
62.0
17.8
989

Yes
18.1
59.2
22.8
1601

Total3
18.9
60.3
20.9
2589

Note: 1. The body mass index (BMI) is the ratio of the weight in kilograms to the square of the height in metres
(kg/m2). For the 15-17-year-old adolescent females in the sample, the cut-off points recommended by Cole
et al. (2000) for adolescent obesity, over-weight and normal weight was used (For details, See Data &
Methods).
2. Table excludes women who are pregnant and women with a birth in the preceding two months.
3. Total includes a small number of women with