Reproduction, health and mortality in Gambian women
Rebecca Sear
London School of Economics
Ruth Mace, University College London
Relationships between reproduction, health and mortality
1. Reproduction requires minimum body condition – predicts body condition will be positively correlated with successful reproduction
2. But reproduction costly – predicts reproduction should cause reduction in body condition and ultimately increase mortality rates
– maternal depletion– costs of reproduction
Study site
The Gambia
Data
• Collected from 4 rural villages since 1950
• Demographic, anthropometric and genealogical data
• Until 1975 both fertility & mortality were high
– TFR ~ 7, 5q0 ~ 0.43
• After 1975 mortality dropped, but fertility remained high
Summary of anthropometry
Non-pregnant adult women (n~11,600 measurements):
Mean Min Max
Weight (kg) 51.48 24.09 99.34
Height (cm) 157.71 122.55 178.43
BMI 20.67 12.37 39.74
Hb (g/dl) 11.8 2.0 17.1
Age (yrs) 39.61 21 90
Evidence for costs of reproduction?
• Is reproductive effort correlated with mortality rate for women?
• Samples:
– reproductive-aged women, 15-49 years (~1000)
– post-reproductive women, 50+ years (~400)
Measures of reproductive effort
• Giving birth
• Parity
• Timing of reproduction – age at first birth – age at last birth
• Intensity of reproductive effort– pace of reproduction– twins– boys
Mortality: reproductive-aged women• Discrete-time event-history
analysis • Analysing effects of:
– giving birth– parity– timing: age at first birth– intensity:
• sex of most recent birth• twins
• Controlling for birth cohort
Results of mortality model: 15-49 yrs
Variable Model 1
OR p
Age 1.06 <0.01
Birth year 2.09 <0.01
Parity 0.83 <0.01
Age first birth (ref 18-19): <18 >19
0.731.37
NSNS
Mother of twins 0.66 NS
NS:
• intensity of reproductive effort: whether most recent birth was twin or male
Mortality: post-reproductive women
• Discrete-time EHA
• Analysing effects of:– parity– timing:
• age at first birth
• age at last birth
– intensity:• pace
• twins
• sons
Mortality and parity
Age (years)
1009080706050
Pro
port
ion li
ving
1.0
.8
.6
.4
.2
0.0
Births
8+
4-7
1-3
Median age at death (yrs)
80
76
69
Mortality and age at last birth
Age (years)
1009080706050
Pro
port
ion
surv
ivin
g1.0
.8
.6
.4
.2
0.0
40+
<40
Median age at death (yrs)
77
72
Results of mortality model: 50+ yrsVariable Model 1
OR pAge 1.12 <0.01Parity 0.97 NSNulliparous 1.46 NSAge first birth (ref 18-19): <18 >19
1.151.47
NS0.057
Age last birth (ref 30-39): <30 >39
1.120.42
NS<0.01
Mother of twins 3.55 <0.01
NS:
• sex composition
Conclusions
• Evidence for costs of reproduction inconclusive
• Mortality:
– Giving birth clearly a risk factor
– But other evidence suggests negative relationship between reproduction and mortality
– EXCEPT twin mothers have higher post-reproductive mortality
Selection effects?
• Potential explanation for negative relationship between reproduction and mortality:– only healthy women can reach high parities, start
reproducing early and stop reproducing late
– less healthy women start later, stop earlier and have fewer births
• Solutions?– control for health in mortality analysis
– use repeated measures anthropometric data
Controlling for health: 15-49 yrs
Variable Model 1 Model 2
OR p OR p
Age 1.06 <0.01 1.11 <0.01
Birth year 2.09 <0.01 2.67 <0.01
Parity 0.83 <0.01 0.74 <0.01
Age first birth (ref 18-19): <18 >19
0.731.37
NSNS
0.961.57
NSNS
Mother of twins
BMI before first birth
0.66 NS 0.98
0.90
NS
NS
Controlling for health: 50+ yrsVariable Model 1 Model 2
OR p OR pAge 1.12 <0.01 1.20 <0.01Parity 0.97 NS 0.94 NSNulliparous 1.46 NS 1.20 NSAge first birth (ref 18-19): <18 >19
1.151.47
NS0.057
1.231.46
NS0.070
Age last birth (ref 30-39): <30 >39
1.120.42
NS<0.01
1.120.46
NS<0.01
Mother of twins
Height
3.55 <0.01 2.99
0.99
<0.01
NS
Selection effects?
• Potential explanation for negative relationship between reproduction and mortality:– only healthy women can reach high parities, start
reproducing early and stop reproducing late
– less healthy women start later, stop earlier and have fewer births
• Solutions?– control for health in mortality analysis
– use repeated measures anthropometric data
Conclusions
• Controlling for health does not affect relationship between reproduction and mortality:
– inadequate measures of health?
– protective effects of reproduction?