Contents - Previous - Next

This is the old United Nations University website. Visit the new site at http://unu.edu

Nutrition

Child mortality in Bangladesh: Food versus health care

A. K. M. Alauddin Chowdhury

Mortality levels in less-developed countries have long been substantially higher than those in developed areas, principally due to differences in sanitation, public health measures, available health care, and probably life-style. In a variety of settings, such differences also have been attributed to inequitable distribution and use of overall health services facilities.

Children often are most at risk. It is widely believed that the level of maternal education greatly affects child mortality, as better-educated mothers are most likely to know about health-related matters and to have beliefs and practices, including use of health services, that lead to healthier children [1].

Bangladesh is one of the world's poorest countries, with an annual per capita income equivalent to US$120 [2]. More than one-eighth of the country's newborns die in the first year of life, and another one-eighth die in the next four years [3]. Moreover, a majority of its population lives below the poverty line, overall mortality has not been declining, and per capita food consumption has deteriorated during the last decade [4].

In rural Bangladesh, sanitation and water supply are primitive, public health facilities are non-existent, and health practices and beliefs may vary with socioeconomic status. Effective medical technologies such as oral rehydration therapy and antibiotics are being used in rural areas, however, and are expected to reduce deaths caused by diarrhoea, pneumonia, and so on [5, 6]. Evaluation of the use of these therapies, which have been available for a considerable length of time, may be a means by which socio-economic influences on child mortality can be assessed.

In addition to the lack of primary health care services, family poverty may lead to significant differences in child mortality. Poor families have a greater proportion of severely malnourished children than their wealthier counterparts. These severely malnourished children are most vulnerable to death from common infections as well as from continued food deprivation. Rich families, with proportionately fewer malnourished children, have the advantage of nutritional reserves.

Thus child mortality differentials in Bangladesh may result from vastly different levels of use of health care services, varying nutritional levels and nutritional backup, and other factors. This paper evaluates child mortality in terms of maternal literacy. Only children one to four years old are considered, because this group shows a maximum variation of mortality by socio-economic status in rural Bangladesh [7].

Causes of child mortality

The main health-related risks for Bangladeshi children one to four years old are infections, malnutrition, and accidents. Infections and accidents reflect the physical environment. Malnutrition is due to children's inadequate food consumption over periods of time, mostly because of poverty. Malnutrition and infections produce a vicious cycle that often results in death.

Some acute infections, if untreated, have high rates of fatality, but with adequate treatment, deaths can be reduced greatly. Given the current relatively simple therapy for diarrhoea and respiratory infections, and the fact that such problems can be managed effectively in home settings without a physician, it is possible to reduce their associated mortality dramatically. Thus fluid replacement during acute watery diarrhoea can reduce the fatality rate from 20% to 2%, and antibiotics given for respiratory infections can save significant numbers of children's lives [6]. Death from pneumonia in childhood is not particularly related to malnutrition [8].

Fever, which is one of the common symptoms of several types of infections, occurs in both well and malnourished children (although in the latter some infections produce hypothermia). Regardless of nutrition children with fever require system therapy. The risks caused by diarrhoeal disease (watery), and respiratory and other infections with fever can be reduced by the use of primary health care services, however, and are not particularly associated with malnutrition.

Nutrition-related deaths are mainly due to preexisting malnutrition or lack of nutritional reserves in chronic infections, although infection itself often plays an auxiliary role. A recent study in Bangladesh showed that severe malnutrition was a high risk factor for death of patients hospitalized with dysentery [9]. Another study in the same area, which also was based on lay reporting of causes of death, showed that severely malnourished children, when discharged from a hospital after treatment for diarrhoea, had high mortality in the subsequent year, with most of these deaths caused by dysentery [10].

Measles coupled with pre-existing malnutrition has been shown to be a high death-risk factor for children [11, 12]. If exposed to further food deprivation, children who are already malnourished may develop marasmus, with fatal outcome. This may also occur with some chronic infections. The underlying cause of death in children with oedema is, in most cases, prolonged food deprivation. Pre-existing malnutrition is a high risk factor for diseases such as dysentery, measles, and oedema syndrome. When such diseases occur, nutritional reserves, with or without health care, can reduce the number of fatalities.

Since malnutrition and lack of nutritional reserves during and after infection may play a significant role in child mortality in Bangladesh, and since both are direct consequences of socio-economic status, it would be helpful to study how a family's socioeconomic status affects child mortality - that is, whether the effect is based on nutritional aspects of the household or inadequate health care, or both.

Literate and illiterate households have similar patterns of usage of well water and fixed latrines. Given this uniformity, differences in child mortality due to conditions not strongly associated with malnutrition arise mostly from differences in health care services provided for those conditions. When there are no health care services (or they are evenly distributed), no differences in mortality are observed.

The differences in nutrition-related deaths may occur because of variations in pre-existing nutritional status of children and in nutritional reserves in case of repeated infections, in addition to differences in health care services.

Methods

This study used two types of data. The first were demographic surveillance data from the International Centre of Diarrhoeal Disease Research, Bangladesh, which has been operating a field hospital in Matlab, a rural area of the country. Using this hospital station as a base, the Centre has maintained a demographic surveillance system for the surrounding area since 1966. The system now covers 159 villages, which had a population of 173,000 in 1978. It consists of periodic censuses and bi-monthly registration of births, deaths, marriages, and migrations. This paper uses data for the three-year period 1978-1980.

The second data set is taken from a special study in the same area. In October 1975 nearly 2,000 children 13-23 months old from 86 villages were identified through the birth-registration system. Between November 1975 and January 1976, 1,966 children were examined for body weight and height and arm circumference. The cross-sectional nutritional assessment followed standard field procedures, and included body weight (on the Salter scale) measured to the nearest 50 g, recumbent body length (measured on a locally constructed two-track length board) to the nearest 0.1 cm, and left mid-arm circumference (using an oil-cloth tailor's tape) measured to the nearest 0.1 cm. The measurements were made by trained two-person teams of field-workers from the Centre. Scales were calibrated daily, and standardization was done on completion of the field-work [13].

As all births are registered in the demographic surveillance system, precise age data were available. Moreover, as the children's families were part of the study population, selected socio-economic information was available from the 1974 population census.

Deaths of the study children during the 24 calendar months following the anthropometry were registered from the surveillance system, which noted the date and age at death. Cause-specific mortality rates were calculated by socio-economic status as described in the Appendix, and an attempt was made to correlate these rates with child mortality.

Results

Table 1 shows the distribution of children during the mid-period of the study by mother's age and education. Of a total population of 175,887, 12.3% were one to four years old. It is estimated that 58.5% of these children had mothers under 25 years old, and that 76% of the mothers had no formal education. The numbers in this table were used as denominators when estimating cause-specific mortality rates.

TABLE 1. Distribution of children one to four years old by age and education of mothers, June 1979 (estimated on the basis of the Matlab Periodic Census, 1978)

Few data are available on specific medical causes of child mortality in Bangladesh, except for some hospital and death records. With experience and training, however, non-medical personnel can report causes of death with considerable precision. This is especially the case in Matlab for children one to four years old[14]. Moreover, it was found that, although registration data are based only on lay reporting, such causes as diarrhoea, dysentery (as distinguished from other diarrhoeas by the presence of mucus and/or blood in loose stools), measles, tetanus, and accidents are identified with few errors [9].

Oedema related to severe malnutrition also is reported as a cause of death for children one to four years old. When oedema is present in this age group, deaths are classified as having been related to malnutrition. When fever is also reported as a cause, the febrile illness might result from influenza or other infections.

Deaths due to diseases of the pulmonary and gastrointestinal tracts are not identified distinctly; however, such conditions may be important in classifying deaths from other causes [15]. The yearly pattern of deaths in these children, as reported by lay persons, has remained the same in recent years, thus reflecting the consistency of data and uniformity of the causes of death [16, 17].

The reported causes of deaths of children one to four years old from 1978 through 1980 are shown in table 2. The most common (25%) was dysentery. Other major causes, in descending prevalence, were measles, fever, drowning, nutritional oedema, respiratory diseases, and watery diarrhoea.

TABLE 2. Distribution of children one to four years old who died during 1978 and 1979, by cause of death (bimonthly death registration data, Matlab)

Tables 3 and 4 show the mortality rates and the odds ratios for the children of illiterate versus literate women in two age groups. Statistical tests (Z test) were done for all instances. An odds ratio significantly greater than 1 means that mortality is higher among the children of illiterate women.

TABLE 3. Three-year mortality rate per 1,000 children one to four years old and odds ratios, by causes of death related to malnutrition

1. p < .05.
2. p < .01.

TABLE 4. Three-year mortality rate per 1,000 children one to four years old and odds ratios, by causes of death in which fatality was related to health care

Table 3 presents the results for conditions related to malnutrition. For oedema in children of women in the young group, the odds ratio is 2.40, which is statistically significant. The odds ratio of 1.26 for children of women in the older age group is not statistically significant. Similarly, for measles the odds ratios reflect the fact that illiterate mothers younger than 25 years experienced significantly more child deaths than did their literate counterparts. For those 25 years old and over, only a slight difference in child mortality due to measles was noted. The figures for dysentery are the same, but the differences relate to the mothers' age.

Table 4 shows the same calculations for conditions not considered to be strongly related to malnutrition, that is, respiratory diseases, fever, diarrhoeas, and drowning, for which fatalities are assumed to be related more to child health care. In no case was the odds ratio significantly greater than 1. This appears to indicate that, for these four causes of death, children's mortality risk is not affected by mothers' literacy. Perhaps this reflects the minor differences between availability and non-availability of effective child health care for both illiterate and literate mothers.

In summary, table 3 suggests that malnutrition severe enough to result in children's death is significantly influenced by mothers' education. Supporting evidence is available from data on 1,966 children who were analysed by nutritional status, maternal education, and subsequent mortality. In general, children of illiterate mothers had a significantly higher death rate than did those of literate mothers (p < .02) (table 5). This was true for both young and older mothers, although the difference by age group was not significant.

TABLE 5. Two-year mortality rate for children 12-23 months old, by education and age of mother

Figures in parentheses are number of children.

TABLE 6. Mortality rate over the subsequent two years for one-year-old children, by the child's nutritional status and the mother's age and education

Figures in parentheses are number of children.
a. Percentage of Harvard median.
b. Fisher's exact test.

Table 6 presents childhood mortality based on mothers' education, controlling for children's nutritional status. Nutritional status was based on deviation from the median of standard weight for age. Nutritional status below 60% of the Harvard standard was considered severely malnourished. The table clearly shows that malnutrition was the predictor of subsequent risk of mortality. Its effect was essentially equal within each of the four age/education categories.

It is noteworthy that the proportion of severely malnourished children was considerably higher for the illiterate than for the literate group: 19% versus 8% respectively for young mothers, and 26% versus 19% for older mothers. This confirms the fact that child mortality differences observed in this study depend on nutritional status. The mortality rate of severely malnourished children in this population was three times higher than that of their better-fed counterparts. Hence the reason table 5 appears to show that children of illiterate mothers experienced more deaths is that this group had proportionately more malnourished children than did the literate mothers.

Discussion

Two points in the study merit discussion here. First, in classifying the causes of deaths into two broad groups, there are probably some misclassifications. However, since misclassifications operate to reduce any differences observed, the general conclusions are considered to be valid.

Second, in the group for which deaths were related to malnutrition, the difference in mortality based on the mothers' education was not as striking among the children of older mothers as among those of younger mothers (table 3). This phenomenon may be explained by the differences in the proportion of malnourished children by maternal education: it was less for older than for younger mothers (table 6). This may result in a smaller difference in disease-specific child mortality by education among the older mothers.

Conclusion

It is widely believed that, within a given society, educated mothers acquire health attitudes and practices that lead to better health care for their children, including use of health facilities, and that the result is lower child mortality. Our study shows little evidence of this being the case in Bangladesh. The reason may be that health services available in a rural area are negligible and that whatever is available may be used equally by women irrespective of education. Moreover, use of health care services also depends on many factors, such as cost, distance, operational problems, social status, and the like, and the effect on mortality of a single factor such as education may not be apparent.

When food is unequally available, lower socioeconomic groups can be expected to have proportionately more malnourished children. One study has shown that maternal education was one of the important determinants of child malnutrition in rural Bangladesh [18]. As malnourished children have a higher risk from such common diseases as measles, dysentery, or oedema, this itself may result in different mortality rates by socio-economic status. This study suggests such a hypothesis.

Another study of a different area of Bangladesh evaluated the effect on child mortality of gross food shortage in a subsistence-level agrarian population. The results showed that differences based on landholding had a similar pattern. Families with no arable land had four times higher child mortality than families having three acres or more during famine; in the non-famine periods the ratio was only 1:1.5 [8].

Thus this study confirms that in Bangladesh, nutritional status is the main reason for differences in child mortality in different social strata. When studies control for children's nutritional state, either by anthropometric measures or by nutritionally related causes of death, differences in risk between educational groups disappear. The fact that malnutrition is more common among the less educated is one of the determining factors.

A food programme assisting malnourished children of Bangladesh, as well as nutritional supplements for those who suffer from repeated infections, should be a part of a primary health care programme. This will not only reduce the number of deaths but will decrease the differences in child mortality. Selective primary health care programmes without a nutritional component probably will have limited impact on the latter.

Appendix Calculation of mortality rates

1. The mortality rates for the children one to four years old covered by the first set of data for the three-year study period were calculated as follows:

P_jkl = n_jkl/d_jk

where:

n_jkl = the number of deaths of children with mother's age j, mother's education k, and cause of death 1;
d_jk = the number of children at mid-year of the study with mother's age j and mother's education k, as estimated by the above procedures;
jk^djk = 22,050

The relative-odds ratios were calculated by

s = P_I(1 - P_E)/P_E(1 -P_I)

where:

P_I = the mortality rate from a specific cause for the children of illiterate mothers;
P_E = the mortality rate from the same cause for the children of literate mothers.

2. The mortality rates for the one-year-old children covered by the second set of data over the following 24 months of their lives were calculated on a cohort basis:

P_jkl = n_jkl/d_jkl

where:

n_jkl = the number of children who died during the 24 months with mother's age j, mother's education k, and anthropometry status l;
d_jkl = the number of children at the beginning of the study with mother's age j, mother's education k, and anthropometry status l;
jkl^djkl = 1,996

References

1. Caldwell JC. Education as a factor in mortality decline: an examination of Nigerian data. Paper presented at the conference on socio-economic determinants and consequences of mortality, Mexico City, 19-25 July 1979.
2. Regional performance figures. In: Far Eastern Review yearbook, 1982. Hong Kong: Far Eastern Review, Ltd., 1982.
3. Economic and Social Commission for Asia and the Pacific. Population of Bangladesh. Country monograph series no. 8. Bangkok, Thailand: ESCAP, 1981.
4. Institute of Nutrition and Food Science. Nutrition survey of rural Bangladesh, 1980-81. Mimeograph, Dhaka: University of Dhaka, 1984.
5. Hossain MM, Glass RI, Khan MR. Antibiotic use in a rural community in Bangladesh. Int J Epidemiol 1982;11(4):402-405.
6. Mosley WH. Social determinants of infant and child mortality: some considerations for an analytical framework. Discussion paper for a conference on health and mortality in infancy and early childhood, Cairo, Egypt, 18-20 May 1980.
7. D'Souza S. Abbas B. Socio-economic mortality differentials in a rural area of Bangladesh. Pop Devel Rev 1982; 8(4):753-769.
8. McCord CW, Chowdhury SA, Khan HA, Ashraf. Death rate, land and the price of rice 1975-78. Companigonj Health Project. Evaluation unit report no. 4. Christian Commission of Development in Bangladesh, 1980.
9. Engleberg CN, Mitra AK, Haque E. Chowdhury MK, Glass Rl. Dysentery-related deaths in rural Bangladesh. Atlanta, Ga, USA: Centers for Disease Control, 1983.
10. Roy SK, Chowdhury AKMA, Rahaman MM. Excess mortality among children discharged from hospital after treatment for diarrhoea in rural Bangladesh. Br Med J 1983;281(6399):1097-1099.
11. Ogbeide Ml. Measles in Nigerian children. Trop Pediatr 1967:71(5):737-741.
12. Koster FT, Curlin GT, Aziz KMA, Hoque A. Synergistic impact of measles and diarrhea on nutrition and mortality in Bangladesh. Bull WHO 1981;59(6):901-908.
13. Huffman SL, Chowdhury AKMA, Chakraborty J. Mosley WH. Nutrition and post-partum amenorrhea in rural Bangladesh. Pop Studies 1978;32(2):251-260.
14. D'Souza S. A population laboratory for studying disease processes and mortality: the demographic surveillance system, Matlab, Comilla, Bangladesh. Dhaka: International Centre for Diarrhoeal Disease Research, 1981.
15. Zimicki S. Nahar L, Sarder AM, D'Souza S. Cause of death reporting in Matlab source book of cause-specific mortality rates 1975-1981. In: Demographic surveillance system, vol. 13. Scientific report series no. 63. Bangladesh: International Centre for Diarrhoeal Disease Research, 1983.
16. Chowdhury MK, Razzaque A, Becker S. et al. Demographic surveillance system, vols. 9-11. Bangladesh: International Centre for Diarrhoeal Disease Research, 1979, 1980, 1981.
17. Shaikh AK, Mostafa G. Sarder MA, Wojtynia B. Demographic surveillance system, vols. 12, 13. Bangladesh: International Centre for Diarrhoeal Disease, 1982, 1983.
18. Bairagi R. Is income the only constraint on child nutrition in rural Bangladesh? Bull WHO 1980;58(5):767772.

Contents - Previous - Next