orange publication summary banner small

Document Type
Working Papers
Publication Topic(s)
Child Health, Infant and Child Mortality, Nutrition
Shea O. Rutstein and Macro International Inc. Calverton, Maryland, USA
Publication Date
September 2008
Publication ID

Order a Hard Copy: Please use electronic copies of DHS publications whenever possible. Hardcopies of publications are intended primarily for those in developing countries where internet connections are limited or unavailable.


In response to my 2005 article on the effect of birth spacing on infant and child mortality, a number of suggestions were made to improve the research. This paper, which is based on a much larger, more recent set of Demographic and Health Surveys (DHS) revises that research and incorporates the suggestions for analytical improvement. A handful of new studies have come out since the 2005 article, all of which indicate that too short intervals (either birth or pregnancy) are associated with adverse pregnancy outcomes, morbidity in pregnancy, and increased infant and child mortality. Long birth intervals were also seen as contributing to adverse pregnancy outcomes. This study pools birth history data from all 52 DHS surveys conducted from 2000 through 2005. Utilizing life tables and Cox hazard multivariate regression, the effects of the birth-to-pregnancy interval are studied for infant and child mortality, broken down into several periods – early neonatal, neonatal, post-neonatal, infant, child (one to four years), and under-age five years. The birth-to-pregnancy interval is classified into groups that will be harmonized with those of forthcoming studies. In the analyses, intervals based on imputed dates of birth are excluded. The resulting data set includes 1,123,454 births. Confounding relationships are statistically removed by including control variables for birth-specific characteristics (duration of the pregnancy, maternal age at birth, birth order, sex, survival of preceding child, sex of preceding child, multiplicity of the birth, use of health care service and wantedness of the birth), mother-specific characteristics (education, area of residence), and household characteristics (an index of wealth, source of drinking water, type of toilet facility and possession of a refrigerator). Birth weight, size at birth, and nutritional status at the time of the survey (stunting, wasting and underweight) are studied using means and logistic regressions. Controls used in addition to the above were breastfeeding status and types of other child feeding. Bivariate results indicate that 36 to 47 months between a birth and the next conception is the interval with the lowest risk of neonatal, infant, and under-five mortality. There are weak relationships of the birth-to-pregnancy interval with small size at birth and low birth weight, with children conceived in the six months following a birth most likely to be small and of low birth weight. Bivariate results show substantial declines in stunting as the birth-to-pregnancy interval increases among children conceived more than 18 months after the birth of a preceding child. The multivariate results confirm and clarify the bivariate findings. The relationship between the preceding birth-to-pregnancy interval and under-five mortality is highly significant (p<0.001) and each interval group under 36 months is significantly different from the reference group of 36 to 47 months as well as from the group 60 to 96 months. For intervals of less than 36 months, the adjusted risk ratios are always substantially higher than those of the reference group. The risk of mortality trends downwards with increasing birth interval, rapidly until 24 to 29 months and then more slowly, with longer intervals and with a final upturn for intervals of 96 or more months duration. For child mortality (one to four years), the effect of the duration of the preceding birth to pregnancy interval is almost a constant decline in mortality with an increase in interval length. The risk ratios for all interval groups are significantly different from the reference group. The risk of child death falls from 2.2 times that of the 36 to 47 month group for conceptions within six months of the preceding birth to 0.7 for conceptions at 96 months or more. There is no upturn in child mortality for the open-ended long interval as is noticed for under-five mortality. The analysis of infant mortality is based on births that occurred 0 to 59 months prior to the survey to ensure full use of available data and control variables. Compared with a preceding birth-to-pregnancy interval of 36 to 47 months, children conceived with interval durations less than 24 months have significantly higher risks of mortality. There is little difference in mortality risk for the duration groups between 24 and 59 months. However, the mortality risk increases for births with durations of 60 months or more. The shorter the duration of the interval for intervals less than 24 months, the higher is the risk of dying during infancy. The risk of neonatal mortality by preceding birth-to-pregnancy interval is U-shaped with the lowest point at the reference group (36-47 months). All interval groups outside the period 24 to 47 months have adjusted relative risks that are significantly higher than that of the reference group. Intervals shorter than 24 months have adjusted relative risks that are from 19 percent to 146 percent higher than the risk of the reference group’s mortality. Intervals longer than the reference group have risks that are from 20 percent to 79 percent higher. For children conceived less than six months after a prior birth, the odds of low birth weight are 42 percent greater than that of the reference group (ci=1.29-1.56). For children conceived with an interval of 6 to 11 months, the odds are 16 percent higher (ci=1.07-1.26). The relationship for small or very small size at birth is somewhat similar to that with low birth weight. One notable difference is the larger relative odds for children conceived within six months of a prior birth. The increase in the odds of being small or very small at birth for interval group 12 to 17 months is also statistically significant (rr =1.04, ci=1.01-1.07). From the multivariate analyses, it is clear that chronic and overall undernutrition decline substantially with the longer the interval. Children conceived after an interval of 12 to 17 months are 23 percent more likely to be stunted (ci=1.19-1.26) and 19 percent more likely to be underweight (ci=1.15-1.24) than children conceived after an interval of 36 to 47 months. Children conceived after even longer durations are less likely to be stunted (up to 18 percent less, ci=0.77-0.87) and underweight (up to 18 percent less, ci=0.75-0.88) than children conceived during the reference period. In general, the findings of this study confirm those of the author’s preceding study on 17 DHS surveys. While the excess risk of mortality is highest for very short intervals (less than 12 months birth to pregnancy), there are relatively few children conceived at such intervals (14 percent). Combining both the increased risk of death for children conceived between 12 and 35 months with the great number of children with such intervals (42 percent) results in substantial declines in mortality by avoiding these intervals. The population attributable risk (PAR) for under-five mortality for avoiding conceptions at less than 24 months after a birth is 0.134. In other words, if all women would wait at least 24 months to conceive again, under-five deaths would fall by 13 percent. The effect of waiting 36 months to conceive again would avoid 25 percent of under-five deaths. The impact of avoiding these high risk intervals (less than 36 months) would be a total of 1,836,000 deaths avoided annually in less developed countries, excluding China (where there is a one child policy). Thus, parents who want their children to survive and thrive would do well to wait at least 30 months after a birth to conceive another child.