Contraceptive failure: levels, trends and determinants in Matlab, Bangladesh.

This study investigated the levels, trends and determinants of contraceptive use-failure in Matlab, Bangladesh, using a set of prospective data on 25,960 women of reproductive age. The data were extracted from the Record Keeping System (RKS) of Matlab for the period 1978-94. If there was any live birth during the use or within 7 months after the discontinuation of use, it was considered as a failure. The life table technique and hazard model were used as analytical tools. The results suggest that use-failure for pills, IUDs (TCu 200) and injectables and other temporary methods increased from 1978 to 1988, but began to decline after 1988. The cumulative probability of first-method failure within 1 year of method acceptance of the cohort of 1990-94 acceptors was 12.9% for pills, 2.0% for IUDs, 0.5% for injectables, 22.0% for condoms and 13.4% for 'other' methods (sampoon, foam, jelly and traditional methods). For pills, condoms and 'other' methods, the likelihood of failure declined with the duration of use; by contrast, the probability of an IUD failure increased over time, peaking at 3 years of use. The injectables maintained a low likelihood of failure regardless of the duration of use. The quality of Community Health Workers' (CHWs) performance was associated with the risk of failure of all temporary methods except condoms; women's background characteristics associated with failure varied by method. The effect of the quality of the CHWs' performance and the background variables on failure did not change much over time. It is felt that contraceptive failure deserves the serious attention of programme managers and policy makers to make the Bangladesh national family planning programme more successful.

Improvement in female survival: a quiet revolution in Bangladesh.

PIP: This article examined the time trends of mortality and made a relative comparison of the mortality changes between males and females in different age groups in a rural area in Bangladesh between 1970 and 1995. Data from the Matlab Demographic Surveillance System were used for the study. Two areas were compared: 1) the maternal and child health/family planning (MCH/FP) area and 2) the other area known as the comparison area. Data showed that the mortality situation improved in both areas in every age group within the study period. However, the improvement was much greater for females in infancy and at ages 1-4 and 5-14. While improvement was seen in both areas, greater improvement was observed in the MCH/FP area, indicating a gain in life expectancy that is greater in females than in males. These improvements, however, were still below expectation for infants and children in the 1-4 age group. Thus, more efforts need to be exerted to maintain the improvement in female survival and eliminate excess female child mortality in Bangladesh.^ieng

Inconsistencies in the findings of child nutrition surveys in Bangladesh.

BACKGROUND: The Bangladesh Bureau of Statistics (BBS) has been conducting nutrition surveys of young children at 2-3-y intervals since 1985. However, data on malnutrition determined on the basis of weight-for-age and height-for-age in different surveys are inconsistent. OBJECTIVE: The aim of this study was to investigate the reasons for inconsistencies in the findings of these surveys. DESIGN: Although the regular personnel of the BBS collected data on anthropometry and age in the previous surveys, 1995-1996 data were collected by locally recruited workers. Data on 100 children included in the 1995-1996 survey were also independently collected by BBS personnel. A comparison of the data from these 2 sources and the inconsistencies therein form the basis of this study. RESULTS: The average difference between the 2 sources (bias) was found to be approximately 1.8 mo for age and 0.13 cm for midupper arm circumference. Differences in weight and height were negligible; however, random error was found to be substantial in data on height. CONCLUSIONS: Bias in age was thought to be mainly responsible for the inconsistencies in the measures of malnutrition in the different child nutrition surveys in Bangladesh. A method for accurately collecting age data needs to be developed on the basis of a study in a demographic surveillance area where accurate age data are available. Until such a method is fully developed, midupper arm circumference should be routinely collected in all nutrition surveys. The importance of accuracy in weight and height data is emphasized.

Fertility and its proximate determinants in Bangladesh: evidence from the 1993/94 Demographic and Health Survey.

PIP: This study examined determinants of fertility in Bangladesh. Data were obtained from the 1993-94 Bangladesh Demographic and Health Survey among a nationally representative 2-stage sample of 9640 ever-married females aged 10-49 years. Findings indicate that the age-specific marital fertility rate was highest among women aged 15-19 years. The total fertility rate was 3.44 births/woman in 1993-94 and 5.12 births/woman in 1989. The interval between marriage and first birth declined more for younger cohorts. The proportion of women who had a child within 5 years increased. Over 60% were married under the age of 14 years. The proportion currently married has remained stable since 1981. The number of those never married has increased, especially among women aged 15-19 years. 44.6% of currently married women used family planning; 36.2% used modern methods and 8.4% used traditional ones. Prevalence was highest for the pill, followed by female sterilization. 48% of infants were breast-fed on the first day. Breast-feeding duration averaged 30 months. Duration of postpartum amenorrhea averaged 12 months. 0.5% reported induced abortion. Analysis of proximate determinants indicates that contraception accounted for 39.0% of fertility decline; lactational infecundability accounted for 34.7%. Marriage patterns accounted for 23.9%. The fertility inhibition of contraception varied by religion. Contraception had the highest impact among higher educated, upper class, urban, and non-Muslim women. Lactational infecundability had the highest impact among poor, nonworking, illiterate, and non-Muslim women.^ieng

Estimating childhood mortality trends from routine data: a simulation using the preceding birth technique in Bangladesh.

The Preceding Birth Technique (PBT) has been proposed as a method suitable for ascertaining the prevailing level of under-2 mortality in countries without full vital registration. It is a monitoring tool rather than a method that will replace other established approaches to measuring childhood mortality levels and differentials that other demographers have developed over the last 30 years. The principle obstacle to the wider adoption of the PBT is the low proportion of women who give birth in maternity clinics and hospitals. A larger proportion of mothers, however, visit clinics and hospitals for antenatal care and to vaccinate their newborn. We used data from the Matlab surveillance system to test the accuracy of mortality estimates derived using the PBT with data obtained from mothers at antenatal visits and at the vaccination of their youngest children. The study shows that the PBT estimates under-3 rather than under-2 mortality in Bangladesh due to the long birth intervals. The data when used to stimulate the collection of the information at antenatal or postnatal visits, nonetheless provide an accurate description of under-3 mortality trends and differences for the two periods examined--before 1984 and before 1989.

An evaluation of the 1993-94 Bangladesh Demographic and Health Survey within the Matlab area.

PIP: The 1993-94 Bangladesh Demographic and Health Survey (DHS) reported substantial declines in vital rates, especially the fertility rate, which needed confirmation. The demographic database of the International Center for Diarrheal Disease Research, Bangladesh (ICDDR,B) contains the birth and death records for 200,000 people whose households have been visited every 2 weeks since 1966. In addition, the system kept records on the pregnancy and contraceptive use status of women of reproductive age since 1977. A validation study was conducted, which entailed the comparison of fertility and infant mortality rates from a special DHS survey conducted in the Matlab treatment area in 1994, with rates obtained by the Demographic Surveillance System (DSS) over the 5 years prior to the survey and also the comparison of the current contraceptive use rate. The records of 2628 women were examined. The Matlab DHS was found to be accurate in estimating fertility both in the treatment and comparison areas. The Matlab DHS infant mortality rates for the 5 years prior to the survey were also consistent with the estimates derived from the DSS. However, the Matlab DHS seemed to have underestimated contraceptive prevalence, which underestimate was substantial for modern temporary methods, especially pills and injectables. Since contraceptive prevalence may also be higher at the national level as a result of this, the total fertility rate for Bangladesh of 3.4 children/woman may be plausible. Although the Matlab DHS figures on vital rates seem to be reliable, the national level DHS estimates may not be as reliable, because women elsewhere in the country may not have reported their children's births and deaths as accurately as did women in the Matlab area.^ieng

Socioeconomic and anthropometric status, and mortality of young children in rural Bangladesh.

This article investigates the interrelationship of socioeconemic status, anthropometric status and mortality of young children in rural Bangladesh. Data for this study come from Matlab, the vital registration area of the International Centre for Diarrhoeal Disease Research, Bangladesh, where anthropometric and socioeconomic data concerning 1976 children aged 12-23 months were collected in November-December 1975. From these data, anthropometric indices are created, and the relationships of socioeconomic status measured by dwelling space and anthropometric indices with mortality during the 2 years following measurement of these children are investigated. It is found that both socioeconomic status and anthropometric indices are related to mortality. However, the relationship of anthropometric indices is much stronger. Among the anthropometric indices considered, weight-for-age, height-for-age and arm circumference reflect socioeconomic status better than weight-for-height does; and, the first three indices are equally good, and individually better than weight-for-height, as predictors of mortality. The degree of the effect of socioeconomic status (dwelling space) on mortality explained by the best performing anthropometric index, weight-for-age, was not more than 25%. It is concluded that an anthropometric index that can classify socioeconomic status more efficiently is a better predictor of 2-year mortality than any other anthropometric index.

Effects of family sex composition on fertility preference and behaviour in rural Bangladesh.

The effects of family sex composition on fertility preferences and behaviour during the period 1977-88 are examined using longitudinal data from Matlab, Bangladesh. The sex composition of living children was found to be systematically related to fertility preferences and behaviour, with a higher number of sons at each family size associated with a higher percentage of women wanting no more children, a higher percentage currently using contraception, and lower subsequent fertility. However, the highest percentage wanting no more children, using contraception and having lowest subsequent fertility was found among women who already had one daughter as well as at least one son. The effects of sex preference on fertility preferences and behaviour were measured using an index developed by Arnold (1985). The results suggest that while sex preference remained largely unchanged during the study period, its effect on contraceptive use declined and its impact on actual fertility remained modest and fairly stable.

PIP: The hypothesis of this study is that the effect of sex preference on fertility is dependent on the level of contraceptive use, and the relative importance given to sex composition and family preference in determining contraceptive use. The effect of sex composition under different contraceptive prevalence conditions is examined by comparing in the Matlab intervention area 2 different periods: 1977-82 and low contraceptive use, and 1984-88 and higher use. The study area comprised 149 villages and a total population of 187,523. This research used data on 2111 women in the 1977 survey and 3140 women in the 1984 survey. Survey data were linked with the Matlab demographic surveillance system. The results showed that between 1977 and 1984 there was an increase in the percentage of women desiring no more children, a decline in family size preference, and an increase in contraceptive use. Contraceptive use was 20% for modern methods in 1977 and 38% in 1984. The general finding was that the higher the number of sons at any family size level, the higher the percentage of women desiring no more children. The effect of sex preference on the percentage of women desiring no more children did not change between 1977 and 1984 (9.5% during 1978-82 and 19.5% during 1984-88); what did change was the percentage of women desiring no more children, from 39.0% to 54.7%. Change would have occurred is there had been no sex preference; expected changes would have decreased the desire for no more children from 14.6% in 1977 to 16.4% in 1984. Without sex preference, contraceptive use would also have declined from an increased prevalence in 1977 of 20.4% and 9.9% in 1984. the sex preference effect measure (SPEM) of Arnold (1985) was used and modified to quantify the effect of sex preference on fertility preferences and behavior. Fertility was consistently and negatively related to the number of living children. Without sex preference, fertility would have decreased 9.5% during 1978-82 and 10.5% during 1984-88. Family size preference had increased in its importance in determining contraceptive use. The effect of sex composition on subsequent fertility remained constant between the 2 periods.

Mortality-discriminating power of some nutritional, sociodemographic, and diarrheal disease indices.

This study compared the predictive power of selected nutritional (anthropometric), socioeconomic, and diarrheal disease morbidity variables for subsequent childhood mortality over a 1-year period. The data consisted of observations of approximately 1,900 children aged 6-36 months obtained from a longitudinal demographic surveillance system located in a rural area of Bangladesh in 1988-1990. The results suggested that weight-for-age (%) was the best predictor of subsequent mortality over a 1-year period, followed by weight velocity (monthly weight gain or loss in grams). Standardization of weight velocity by the US National Center for Health Statistics standard did not improve the mortality-discriminating power of this variable. Reported diarrheal morbidity was also a useful criterion for predicting mortality. Neither maternal education nor sex of the child had significant mortality-discriminating power.

PIP: The predictive power of selected nutritional (anthropometric), socioeconomic, and diarrheal disease morbidity variables for subsequent childhood mortality were compared over a 1-year period. The data were obtained from the Sample Registration System of the International Centre for Diarrhoeal Disease Research, Bangladesh, located in a rural area of Bangladesh in 1988-90. In 1988 the weights of approximately 1900 children aged 6-36 months were measured to the nearest 100 gm on a bar scale to assess the impact of the flood on the nutritional status of children in the area. The weights of the children were measured again after 3 months for examination of any changes in nutritional status. Information on diarrheal disease during the previous 24-hour period and also during the previous week, including the day of the interview/weight measurement, was also collected. Measurements for 1874 children were available from the 1st weight measurement, and measurements for 1900 children were available from the 2nd weight measurement. The 1664 children for whom measurements were available from both time periods were included in this study; 27 of these died during the 12-month period following the last weight measurement. Below a threshold point of 70% mortality increased very sharply for weight-for-age (percent) and around 70 gm month for the weight velocity index. The sensitivity and specificity curves weight-for-age (percent) appeared to be superior to the weight velocity index as a predictor of mortality. Weight velocity was a weaker predictor than was weight-for-age (percent), and the power of the weight velocity index was lower than that of weight velocity. Neither maternal education nor sex of the child had significant mortality-discriminating power. The discriminating powers of the variables of diarrhea on the day of assessment and diarrhea during the previous week were significant (p 0.05).

The influence of nutritional status on age misstatement for young children in rural Bangladesh.

PIP: The influence of nutritional status on age misstatement in 1981 among 679 children aged 22-59 months in Companyganj, Bangladesh, is examined. The age limits were established to represent the most vulnerable age group; age information 22 months was unavailable. Calibrated scales were used for measurement. Weight was measured within 100 g, and height and arm circumference within .1 cm. Age error was calculated as actual age minus reported age. Weight for age (WA) and height for age (HA) were also calculated for actual and reported age. The Polish standard was used to calculate arm circumference for age (ACA). This rural area revealed findings different from those previously reported for the Matlab area. Systematic and random error was higher in Companyganj. Interviewer bias needs to be investigated as 1 possible explanation for the differences. There was a difference of 4.1 months between the best and worst interviewer. Although child's sex and mother's age were associated with age error in the Matlab study, child's sex was found to be insignificant. The Matlab study included children 0-14 years, which may explain the difference. The nutritional status of children influenced different interviewers, and literate and illiterate mothers report ages differently. The reasons given are the varying expectations of both the interviewers and the mothers of normal growth patterns by age. A malnourished child might be underestimated and a well-nourished child overestimated. The illiterate mother may be influenced by the opinions of the interviewer or not know the child's actual birth date, and then understate the age. The results are that there was age overreporting for each age group. The standard deviation of the age error reveals random error. The standard error increases with age. The systematic error and random error are significantly higher than in comparable Matlab data. In the analysis of nutritional status, mother's education, and interviewer, nutritional status was associated with a linear and significant (p.05) relationship to age error, as was mother's education. The mean difference between literate and illiterate age error was .55 months. The analysis of variance substantiated that the 3 independent variables had a significant interaction with age error.^ieng

Does malnutrition predispose to diarrhoea during childhood? Evidence from a longitudinal study in Matlab, Bangladesh.

It is posited that diarrhoeal illness during one period has influence on diarrhoeal illness in a subsequent period. This relationship may potentially mask the association between malnutrition and subsequent diarrhoea. To test this, we analysed data on cross-sectional anthropometry in combination with data on diarrhoeal morbidity collected longitudinally in a community-based study of 1262 children (aged 6-60 months) during March-December, 1976, in Matlab, Bangladesh. The results confirmed the posited relationship between diarrhoeal morbidities in two consecutive periods and showed that the risks of diarrhoeal attack and longer diarrhoeal illness increased more than threefold during the 2 months following diarrhoeal illness during the preceding 2 months (previous diarrhoea). Children with no previous diarrhoea indicated a positive association between malnutrition and subsequent diarrhoea, but the pattern found among children with previous diarrhoea was not understandable. Logistic regression analyses performed separately for younger and older children showed that controlling for effects of previous diarrhoea, maternal illiteracy and household poverty, severe malnutrition as assessed by weight-for-age was found to be strongly associated with the risk of longer diarrhoeal illness in a 2-month interval in the age group 24-60 months; in the same age group the association with the risk of diarrhoeal attack was significant at the 10 per cent level. No such association for malnutrition, however, was found in the age group 6-23 months.

Sex differences and sustained excess in mortality among discordant twins in Matlab, Bangladesh: 1977-1985.

This paper reports findings from a study that compared survival of 204 pairs of liveborn twins discordant for sex and a random sample of 2371 singleton live births from the same population in Matlab, Bangladesh during the period 1977-1985. The discordant twins showed no evidence of sex differential in neonatal survival. They had neonatal mortality of 287 and infant mortality of 468, both per 1000 live births. These rates were respectively five and four times the rate for singletons in the study. Survival from 12 months through age nine years was similar overall for both discordant twins and singletons for each sex. However, higher female than male mortality was present among discordant twins after the first few months of life. This differential was most prominent during the second year of life and appeared wider than that observed for the singletons.

PIP: This paper reports findings from a study that compared survival of 204 pairs of liveborn twins of different sex and random sample of 2371 singleton livebirths from the same population in Matlab, Bangladesh during 1977-85. The different-sex twins showed no evidence of sex differential in neonatal survival. They had a neonatal mortality of 287 and an infant mortality of 468, both/1000 livebirths. These rates were respectively 5 and 4 times the rate for singletons in this study. Survival from 12 months through 9 years was similar overall for both different-sex twins and singletons of each sex. However, higher female mortality was seen among different-sex twins after the 1st months of life. This differential was most prominent during the 2nd year of life and appeared wider than that observed for the singletons.

Parental sex preference and its effects on fertility intention and contraceptive use in Calcutta.

This study investigated parental sex preference and its effects on fertility intention and contraceptive use in Calcutta. It also investigated the socioeconomic differentials of these effects. Data on 4536 married women of childbearing age came from a cross-sectional survey of slum and non-slum areas conducted in 1970. Desired numbers of sons and daughters in the slum area were 1.89 and 1.30, respectively, and in the non-slum area, 1.63 and 1.17, respectively. 35% of slum women and 30% of non-slum women wanted to have more children, and 22% of slum women and 45% of non-slum women were reported to use contraceptives. The authors estimate that in the absence of parental sex preference, 11% fewer slum women and 12% fewer non-slum women would desire more children. On the other hand, 20% more slum women and 8% more non-slum women would use contraceptives in absence of parental sex preference.

The association between malnutrition and diarrhoea in rural Bangladesh.

The interaction between diarrhoeal disease and nutritional status measured by anthropometry was investigated in approximately 1000 children aged 1 to 4 years during April-December 1976 in a rural area of Bangladesh. Data on diarrhoeal disease were provided by the mothers interviewed at seven-day intervals. Weight and height data were collected bimonthly. Children classified using anthropometric criteria--weight-for-age, height-for-age, and weight-for-height--were prospectively evaluated for incidence and duration of diarrhoea during a short (two-month) period and a long (eight-month) period. Incidence of diarrhoea was not found to be related to nutritional status measured by any of the anthropometric criteria for any of the periods. But duration of diarrhoea was found to be related consistently to nutritional status measured by weight-for-age and weight-for-height. Diarrhoea in the short term affected weight increment in the short term, but not in the long term and did not affect height increment for any of these periods. Diarrhoea in the long term affected both weight increment and height increment in the long term.

A comparison of five anthropometric indices for identifying factors of malnutrition.

Five anthropometric indices--weight-for-age, height-for-age, weight-for-height, weight velocity, and height velocity--are compared for identifying factors of nutritional status of children. Data come from Matlab, the field station of the International Centre for Diarrhoeal Disease Research, Bangladesh. Weights and heights on approximately 1,400 children aged 12-60 months were taken 11 times at two-month intervals. Dwelling floor space, a proxy for socioeconomic status, is considered a long-term factor; the 1974-1975 Bangladesh famine, a medium-term factor; and season, a short-term factor of nutritional status. The figures of the indices by socioeconomic status at different times show their power to identify the factors visually; the regression analyses test statistical significance of the factors; and the standardized regression coefficients provide relative power of the indices to identify the factors. Weight-for-age and height-for-age appear best for identifying long- and medium-term factors; weight velocity, for identifying short-term factors. Weight-for-height is third in identifying long- and medium-term factors, second for identifying short-term factors. Height velocity identifies short-term factors only. The effect of short-term factors on height velocity appears approximately four months later than it appears on weight velocity. This study clarifies why certain indices do not predict mortality in this study population and others.

Food crises and female children in rural Bangladesh.

PIP: Although almost all nations show lower female than male mortality, Bangladesh and certain other developing countries show higher female mortality rates. Among children aged 1 to 4 in Bangladesh, female mortality rates are 45% higher for girls than for boys. This paper examines whether 1) sex biased attitudes toward nutrition (as expressed in terms of food intake) are more marked during food crises, and 2) these biases are related to the socioeconomic status of the family. The study measured weight and height of approximately 1400 children aged 1 to 4 in Bangladesh from April 1975 (10 months after the famine began) through December 1976 (14 months after the famine ended). The findings clearly indicate that sex and social status are strong correlates of nutritional status. Children of higher status families with larger homes fared better throughout the time period. Within each status category, boys fared better than girls. While poor families were harder hit by famine than wealthier ones, male-female nutritional discrimination was stronger among the higher classes. These differences were accentuated during the famine period. Policy makers and planners in Bangladesh must be made aware that such sex biases exist and that these patterns are exacerbated during food shortages.^ieng

Effects of age misstatement on the utility of age-dependent anthropometric indicators of nutritional status in rural Bangladesh.

We report the effects of age error on use of weight-for-age and height-for-age for assessing and screening malnutrition, and for identifying factors of malnutrition in 679 children aged 22-59 months in Companiganj, in rural Bangladesh. Overreporting and random error in age and correlation of age error with each of the anthropometric indices are observed. As a result, the proportion of children having less than or equal to 60 per cent of median weight-for-age is overestimated by five percentage points and the proportion having less than 80 per cent of median height-for-age is overestimated by six points. Loss in sensitivity (proportion of malnourished children correctly identified) for the above cutoff points is 20 per cent for weight-for-age and 28 per cent for height-for-age, compared to the situation in the absence of age error. Although mother's education is not a significant determinant of weight-for-age and height-for-age, age error makes mother's education appear artifactually significant in the analysis of variance.

Sex preference for children and its implications for fertility in rural Bangladesh.

In this study, the level and pattern of sex preference for children and its effects on fertility intention, fertility-regulating behavior, and fertility implications for women are investigated. Data are from a 1976 cross-sectional KAP (knowledge, attitude, and practice of family planning) survey and a three-year longitudinal study, conducted between December 1976 and 31 December 1979, of vital events for 860 married women of childbearing age from Companiganj, Bangladesh. The results show that although son preference is very strong in this area, more than 98 percent of women desire at least one daughter. However, women with a higher proportion of sons are less likely to want more children and are more likely to practice contraception and to be sterilized. Although women with a higher proportion of sons have somewhat lower fertility in the prospective follow-up period, the net effect of son preference on fertility is not significant, possibly because of the low level of contraceptive use in the population.