|
Indian Pediatr 2016;53: 867-869 |
 |
Moving Beyond a Maternal Perspective to Child
Survival
|
* S V Subramanian and
#Daniel J Corsi
From the *Department of Social and Behavioral
Sciences, Harvard TH Chan School of Public Health, United States; and
#Ottawa Hospital Research Institute, Canada.
Email:
[email protected]
|
N otwithstanding the significant improvements in
child survival in recent decades [1], India accounts for the largest
share of the global burden of under-five mortality with an estimated
1.2-5.9 million child deaths [2]. Consequently, scientific efforts
continue to identify factors and interventions that can help improve
child survival [3]. An overwhelming majority of studies are informed,
almost exclusively, by a ‘maternal perspective’, such that factors and
interventions have largely focused on mothers [3,4]. For instance,
increasing educational attainment among women has been identified and
targeted as a means to achieve rapid progress towards fourth millennium
development goal (MDG-4) [4]. Other maternal specific interventions that
have received considerable attention in the literature include family
planning and care targeted to mothers in the pre-conception period along
with micronutrient and folic acid supplementation and early initiation
of breastfeeding during pregnancy and in the postnatal period [5]. In
this issue, Sinha, et al. [6] consider maternal age at childbirth
(hereafter referred to as maternal age) as a potentially modifiable
social determinant of child survival within a large prospectively
followed cohort. They report that young motherhood is associated with an
increase in child mortality, leading to a conclusion that delaying age
at pregnancy would confer important survival benefits in this
population.
Maternal age at childbirth can influence offspring
outcomes, especially survival, in early days after birth through
biological and social mechanisms. Specifically, younger age at
childbirth could be a marker for biologic vulnerabilities, including
short stature, inadequate weight gain during pregnancy and potential
difficulties in delivery, leading to adverse outcomes [7]. In addition,
women in low- and middle-income countries who have children at younger
ages are also more likely to be poor and less educated, implying a
social disadvantage leading to adverse outcomes in their offspring [8].
The plausibility of biologic or/and social pathways linking maternal age
to child survival provides an important basis for interpreting such
epidemiologic associations. However, in the absence of rigorous
mediation studies substantiating the mechanisms linking the exposure
(maternal age) and child survival, caution and scepticism is warranted
before attributing causality to the observed associations. Sinha and
colleagues’ recommendation to prevent young motherhood as a method of
improving child survival implicitly assumes a causal relationship
between maternal age and offspring mortality, which given their
observational study design, is impossible to ascertain. At the same
time, conducting randomized controlled trials on social determinants
such as maternal age at child birth are not feasible.
The key question, therefore, is whether the observed
association between young motherhood (age <20 y) and an increased risk
of post-perinatal mortality is causal? Without the knowledge of whether
the observed association is causal, it remains unclear whether maternal
age is simply a marker of unmeasured aspects that matter for child
survival, or is maternal age truly an independent risk/exposure. In this
editorial, we highlight one approach to improving causal inference in
observational studies that consider exposures measured on mothers. Using
a publicly available national data, we then apply this approach to the
case of maternal age at childbirth in India. Based on other studies
conducted using this approach, and insights from our own analysis, we
conclude that an exclusive maternal focus for improving child survival
and development in India may be problematic and misleading.
Improving Causal Inference in Studies With Maternal
Exposures
One approach, which we refer to as ‘maternal-paternal
comparison’ can be considerably useful to improving causal inference in
observational epidemiologic studies with an interest in maternal
exposures [9,10]. The approach entails comparing the similarity in the
effect size of a defined maternal and paternal exposure on the outcome.
Under the framework where one anticipates a unique and substantial
maternal mechanism (e.g., intrauterine mechanisms or behavioral
mechanisms such as breastfeeding or providing care) – essentially
aspects that fathers do not experience or undertake – one should expect
that the size of effect for the maternal exposure should be
significantly greater than that of the paternal effect size on the same
exposure. If evidence supports a larger effect size for mothers as
compared to fathers, this substantially increases our belief to
attributing a causal interpretation to the maternal exposure. On the
other hand, if the effect size associated with maternal and paternal
exposures is highly similar, one ought to be sceptical of attributing
causality to maternal exposures. Similarity in effect sizes on the same
exposure for mothers and fathers are likely to imply residual
confounding at the household level, including assortative mating
(couples forming partnerships based on similar characteristics) [11].
While the similarity in effect size does not entirely rule out a
potential causal effect of the maternal exposure, causal interpretations
become highly implausible. For instance, if the effect of paternal and
maternal age at childbirth is similar, one would have to believe,
somewhat unrealistically, that whatever unique mechanisms that link
maternal age to offspring survival are exactly similar in magnitude as
the mechanisms that link paternal age to offspring survival.
The maternal-paternal comparison approach has been
validated using the example of parental smoking in the Avon Longitudinal
Study of Parents and their Children (ALSPAC) [10]. In this study,
maternal smoking during pregnancy was inversely and strongly associated,
as should be the case given the strong mechanistic linkages, with
offspring birthweight while paternal smoking was not, suggesting a clear
maternal specific pathway for the effect of smoking on offspring
birthweight. Using this approach, studies using data on Indian
populations have investigated whether maternal exposures such as Body
Mass Index (BMI) [12,13], height [14], education [15], commonly
considered to be causally associated with child mortality and
undernutrition, are robust to this test of sensitivity. In each of the
instances, the maternal effect sizes were found to be no different from
the paternal effect size for the same exposure.
It is not clear to us if Sinha, et al. [6] can
examine this in their data; however, if they can, a supplementary
follow-up analysis would be valuable to strengthen the interpretation of
their study. Meanwhile, using the publicly available 3 rd
Indian National Family Health Survey (NFHS) [16], we assessed the
similarity in the effect size related to age at birth for both mothers
and fathers on child mortality as well as child undernutrition.
Maternal and Paternal Age at Childbirth and Child
Mortality/Undernutrition
In the NFHS, 50,248 births were available over a
10-year period covering 1995/6 to 2005/6 where maternal and paternal age
and other covariates as used by Sinha, et al. were available. We
used the same age categorizations as used by them for maternal and
paternal age at childbirth (<20 y, 20-24 y, 25-29 y, 30-34 y, and
³35 y), and examined
their association with perinatal mortality (<7 d), post-perinatal
mortality (7 d-59 mo), child mortality (birth-59 mo), and an additional
category for neonatal mortality (<1 mo). We present models with maternal
age and paternal age that are further adjusted for child age and sex,
household wealth index, maternal and paternal education, and place of
delivery, which is a covariate set close to Model 3 specified by Sinha,
et al. (Web Fig. 1a).
For perinatal and neonatal mortality, the maternal
and paternal effects on offspring mortality were similar. For example,
maternal age < 20 y was associated with an odds ratio of 1.31 (95% CI
0.99,1.75) for offspring mortality while paternal age <20 y at child
birth had an odds ratio of 1.19 (95% CI 0.77, 1.85), with the test of
difference being not statistically significant (P=0.74).
Interestingly, the effect for paternal age <20 y was stronger compared
to maternal age <20 y for post-perinatal mortality (P=0.006) and
child mortality, although the effect for child mortality was not
statistically significantly stronger compared to mothers (P=0.08).
Younger mothers and fathers at childbirth seemed to show the largest
effects on child mortality, with the effects of older ages being less
consistent and failed to reach statistical significance.
We further examined associations between maternal and
paternal age at childbirth and anthropometric failure, defined according
to the 2006 WHO growth standards as stunting, wasting, and underweight (Web
Fig. 1b). First, the magnitude of maternal and
paternal effects was very weak, even for the youngest age group,
compared to mortality. Second, a comparison of the maternal and paternal
effects indicated that the associations were largely similar, with some
evidence of maternal effects being stronger than paternal at ages >25 y.
Conclusion
In summary, given the similarity of effects of
maternal and paternal age at childbirth, it seems that maternal age at
childbirth may be more likely a marker than a causal exposure in its own
right. Consequently, caution must be exercised while attributing
causality to the maternal age at childbirth. As expected, there exists a
strong correlation between maternal and paternal age at birth in India
(r=0.72, P<0.0001) suggesting substantial residual confounding at
the family/household level including support for a strong presence of
assortative mating [11]. While it has been hypothesized that there may
be mechanisms where paternal age could affect birth outcomes [17] (e.g.,
quality of sperm, epigenetic or DNA changes), it would mainly be
observed among older males (>50 years) [18]. In the absence of any
biologic mechanisms for younger fathers, we should expect the maternal
effects to be much larger at younger ages.
While improving the conditions of the mother has
intrinsic significance (including perhaps delaying age of marriage or
childbirth), their instrumental role in improving child outcomes may be
exaggerated. Scrutiny and scepticism is warranted on existing
observational studies with an interest in maternal exposures. Perhaps
most critically, there is an urgent need to move away from an exclusive
maternal lens to a more household perspective to addressing child
survival and development [19-21], since more often than not, in
countries such as India, the vulnerabilities at household level often
are considerably greater in magnitude than vulnerabilities between
individuals within a household.
Funding: None; Competing interest: None
stated.
References
1. Corsi DJ, Subramanian SV. Revisiting the discourse
on accomplishing MDG-4. Int J Epidemiol. 2013;42:648-53.
2. You D, Hug L, Ejdemyr S, Idele P, Hogan D, Mathers
C,
et al. Global, regional, and national levels and trends in under-5
mortality between 1990 and 2015, with scenario-based projections to
2030: a systematic analysis by the UN Inter-agency Group for Child
Mortality Estimation. Lancet. 2015;386:2275-86.
3. Bhutta ZA, Ahmed T, Black RE, Cousens S, Dewey K,
Giugliani E, et al. What works? Interventions for maternal and
child undernutrition and survival. Lancet. 2008;371:417-40.
4. Gakidou E, Cowling K, Lozano R, Murray CJ.
Increased educational attainment and its effect on child mortality in
175 countries between 1970 and 2009: a systematic analysis. Lancet.
2010;376:959-74.
5. Bhutta ZA, Das JK, Rizvi A, Gaffey MF, Walker N,
Horton S, et al. Evidence-based interventions for improvement of
maternal and child nutrition: what can be done and at what cost? Lancet.
2013;382:452-77.
6. Sinha S, Aggarwal AR, Osmond C, Fall CHD, Bhargava
SK, Sachdev HS. Association between maternal age at childbirth and
perinatal and under-five mortality in a prospective birth cohort from
Delhi. Indian Pediatr. 2016;53:871-7.
7. Alam N. Teenage motherhood and infant mortality in
Bangladesh: maternal age-dependent effect of parity one. J Biosoc Sci.
2000;32:229-36.
8. Finlay JE, Ozaltin E, Canning D. The association
of maternal age with infant mortality, child anthropometric failure,
diarrhoea and anaemia for first births: evidence from 55 low-and
middle-income countries. BMJ Open. 2011;1:e000226.
9. Richmond RC, Al-Amin A, Davey Smith G, Relton CL.
Approaches for drawing causal inferences from epidemiological birth
cohorts: a review. Early Human Dev. 2014;90:769-80.
10. Davey Smith G. Assessing intrauterine influences
on offspring health outcomes: can epidemiological findings yield robust
results? Basic Clin Pharmacol Toxicol. 2008; 102:245-56.
11. Davey Smith G. Epidemiology, epigenetics and the
‘Gloomy Prospect’: embracing randomness in population health research
and practice. Int J Epidemiol. 2011;40:537-62.
12. Subramanian SV, Ackerson LK, Smith GD. Parental
BMI and childhood undernutrition in India: an assessment of intrauterine
influence. Pediatrics. 2010;126: e663-71.
13. Corsi DJ, Subramanian SV, Ackerson LK, Davey
Smith G. Is there a greater maternal than paternal influence on
offspring adiposity in India? Arch Dis Child. 2015;100:973-9.
14. Subramanian SV, Ackerson LK, Davey Smith G, John
NA. Association of maternal height with child mortality, anthropometric
failure, and anemia in India. JAMA. 2009;301:1691-701.
15. Vollmer S, Bommer C, Krishna A, Harttgen K,
Subramanian SV. The association of parental education with childhood
undernutrition in low- and middle-income countries: comparing the role
of paternal and maternal education. Int J Epidemiol. 2016 Aug 8. pii:
dyw 133.[Epub ahead of print]
16. International Institute for Population Sciences
(IIPS) and Macro International. National Family Health Survey (NFHS-3),
2005–06: India: Volume I. Mumbai: IIPS; 2007.
17. Shah PS, Knowledge Synthesis Group on
determinants of preterm/low birthweight b. Paternal factors and low
birthweight, preterm, and small for gestational age births: a systematic
review. Am J Obstet Gynecol. 2010;202: 103-23.
18. Sharma R, Agarwal A, Rohra VK, Assidi M, Abu-Elmagd
M, Turki RF. Effects of increased paternal age on sperm quality,
reproductive outcome and associated epigenetic risks to offspring.
Reprod Biol Endocrinol. 2015;13:35.
19. Subramanian SV, Mejia-Guevara I, Krishna A.
Rethinking policy perspectives on childhood stunting: time to formulate
a structural and multifactorial strategy. Mat Child Nutr. 2016;12(Suppl
1):219-36.
20. Corsi DJ, Mejia-Guevara I, Subramanian SV.
Improving household-level nutrition-specific and nutrition-sensitive
conditions key to reducing child undernutrition in India. Soc Sci Med.
2016;157:189-92.
21. Corsi DJ, Mejia-Guevara I, Subramanian SV. Risk
factors for chronic undernutrition among children in India: Estimating
relative importance, population attributable risk and fractions. Soc Sci
Med. 2016;157:165-85.
|
|
 |
|