Overall, our analyses found that short preceding birth intervals (<18 months, <24 months) have higher odds of neonatal and under-five mortality, a finding that is consistent with most previous findings.[2, 13] However, in contrast to previous findings, our analyses found that long preceding birth intervals (≥60 months, ≥72 months) have lower odds of mortality. By making mortality comparisons among births of the same mother, we believe we have controlled for most socioeconomic and other background factors that could have confounded previous cross-sectional analyses on the impact of birth intervals on child mortality. We do not expect large or systematic changes in the mother’s socioeconomic or other background characteristics throughout her reproductive life to affect our findings, and we believe that any maternal age confounding is sufficiently limited by examining only births that occurred when the mothers were age 18-<35.
While the overall effects of short intervals on neonatal and under-five mortality were consistent with findings from previous studies, our analyses also found major differences in odds ratios related to fertility history of mothers. When the associations were stratified by maternal completed fertility, the association between short birth intervals and child mortality largely disappeared for mothers with low completed fertility (had four or fewer live births at the end of their reproductive period). Furthermore, odds ratios tend to report larger magnitude associations than relative risks, a more easily interpretable measure of association. While we do not expect the effect sizes to differ widely because of the rarity of the mortality outcomes, even the statistically significant associations that remained may be minimal or non-existent if we had examined relative risks instead.
In contrast, children of mothers with high completed fertility (had five or more live births by the end of their reproductive period) had strong associations between short intervals and mortality. However, when we limited the births included in the analysis for high fertility mothers to birth order 2-4, the magnitude of the associations were attenuated while remaining statistically significant. This suggests that there may be complex interactions between short birth intervals, differences across low- and high-fertility women, and parity.
These findings seem to be consistent with the theory that maternal depletion is responsible for the higher mortality risks associated with short birth spacing.[6, 7] Low fertility mothers may have better nutritional status and access to care, hence short birth intervals may not deplete the mother’s nutritional resources to a level that results in increased risk of mortality for the child. For high fertility mothers, the women may be starting out with worse nutritional status and access to care. A short interval birth can deplete the mother’s resources to a level that results in increased risk of mortality for the child, but this effect may be even greater for later births; later births may be more vulnerable to short birth intervals as a result of mothers’ repeated exposure to nutritional depletion.
The maternal depletion explanation is also consistent with our findings for longer birth intervals. The protective effect against mortality of long intervals was primarily seen among the high fertility women. However, among these women, the protective effective was similar comparing all births to just their earlier births; considering our previous theory regarding the cumulative adverse effect of short intervals, we expected the protective effect to also be smaller among the earlier births and greater for later births.
Winkvist et al.  classified women into five different patterns of energy balance during a reproductive cycle in order to better define maternal depletion syndrome. Those who maintain equilibrium or have increased energy during pregnancy make up the first two categories, and a group who has decreased energy but has a long enough potential repletion phase (PRP) to regain equilibrium belongs in a third category. The fourth category of women has both negative energy intake and a short PRP, and Winkvist et al. defines this particular pattern where the short PRP is the driver of depletion as “maternal depletion syndrome.” Finally the fifth category comprises of women who are so undernourished that regardless of how long the repletion intervals are, they remain negative in energy balance. From the associations we saw in our analyses, we suspect that most of the high fertility women included in our study belong to the third and fourth categories. However, the much higher fertility women may have characteristics that place them in the non-repleteable fifth category.
In contrast to our findings, previous studies have reported increased risk of neonatal and infant mortality associated with long birth intervals, even using cut-offs higher than the ones we used here. However, the previously reported adverse impact of long birth intervals may be largely affected by residual confounding. For instance, Rutstein  uses a cut-off of 96 months for birth-to-conception interval, or approximately 105 months birth-to-birth interval, a 9-year gap between births. A large proportion of mothers experiencing such long intervals may be those who have experienced reproductive difficulties rather than those who consciously used family planning to delay the birth. Women who belong to the former category may be malnourished, be experiencing infections, or have other negative characteristics, and it could be those characteristics, rather than the long birth intervals, that drive their children’s heightened mortality risk.
Our findings imply that lengthening birth intervals may only be physiologically beneficial to higher birth orders. Reproductive health programs may benefit more from targeting interventions to women with high parity. In contexts like India where there is a high prevalence of sterilization after a woman achieves her ideal of two or three children , lengthening birth intervals may not be a high priority. However, Winkvist’s interpretation of the maternal depletion syndrome also implies that for maximum impact, reproductive health and nutritional interventions would need to be tailored to the distinct types of mothers to improve their birth outcomes, as lengthening birth intervals may not be enough.
Our study presents the impact of birth intervals on child mortality, using an innovative method of controlling for residual confounding by comparing short and long interval births to regular interval births of the same mother. Furthermore, other literature has not addressed or identified how maternal background characteristics modify the association between birth interval and mortality. This leads us to reevaluate previous findings on this topic, and reexamine the actual impact of short or long birth intervals on a child’s mortality risk.
For LiST, these analyses suggest that there is a causal linkage between short birth spacing and mortality in children. Other papers in this supplement will try to further identify this mechanism and help quantify the effects to be included in the model.