American Journal of Epidemiology
© The Author 2013. Published by Oxford University Press on behalf of the Johns Hopkins Bloomberg School of
Public Health. All rights reserved. For permissions, please e-mail: [email protected].
Vol. 178, No. 9
DOI: 10.1093/aje/kwt142
Advance Access publication:
September 7, 2013
Original Contribution
Association of Antepartum and Postpartum Depression in Ghanaian and Ivorian
Women With Febrile Illness in Their Offspring: A Prospective Birth Cohort Study
Nan Guo, Carola Bindt, Marguerite Te Bonle, John Appiah-Poku, Rebecca Hinz, Dana Barthel,
Mathurin Koffi, Sarah Posdzich, Simon Deymann, Claus Barkmann, Lisa Schlüter, Anna Jaeger,
Samuel Blay Nguah, Kirsten A. Eberhardt, Eliezer N’Goran, Harry Tagbor, and Stephan Ehrhardt*;
for the International CDS Study Group
* Correspondence to Dr. Stephan Ehrhardt, Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, 615 North Wolfe
Street, Suite E6153, Baltimore, MD 21205 (e-mail: [email protected]).
Initially submitted January 31, 2013; accepted for publication June 7, 2013.
In low-income countries, perinatal depression is common, but longitudinal data on its influence on child health
are rare. We examined the association between maternal depression and febrile illness in children. There were
654 mother/child dyads in Ghana and Côte d’Ivoire that were enrolled in a prospective birth cohort in 2010–2011
and underwent 2-years of follow up. Mothers were examined for depression using the Patient Health Questionnaire depression module antepartum and 3 and 12 months postpartum. The hazard of febrile illness in children of
depressed and nondepressed mothers was estimated using a recurrent event Cox proportional hazards model.
The prevalences of antepartum depression in mothers from Côte d’Ivoire and Ghana were 28.3% and 26.3%,
respectively. The prevalences of depression at 3 and 12 months postpartum were 11.8% and 16.1% (Côte
d’Ivoire) and 8.9% and 7.2% (Ghana). The crude and adjusted (for country and socioeconomic status) hazard
ratios of febrile illness in children of depressed mothers compared with those in children of nondepressed mothers
were 1.57 (95% confidence interval: 1.20, 2.07) and 1.32 (95% confidence interval: 1.01, 1.74) respectively. Perinatal depression was frequent and associated with febrile illness in the offspring. Our results showed that a high
prevalence of depression in sub-Saharan Africa may pose a serious public health threat to women and their offspring.
Africa; children; depression; febrile illness; motherhood; pregnancy; puerperium; women
Abbreviations: CI, confidence interval; PHQ-9, Patient Health Questionnaire depression module; SES, socioeconomic status.
The global burden of mental disorders has been recognized as a top priority on the agenda of the World Health
Organization (1). Still, mental disorders are among the most
prevalent and yet most neglected diseases. Depression, a leading cause of disability worldwide, affects more than 350 million
people of all ages (2). More women than men are affected by
depression, and the puerperium is a time of particular vulnerability (2, 3). A systematic review showed that the prevalence
of depression in the first trimester is similar to that in agematched nonpregnant women (7%); however, it is substantially
elevated in the second and third trimester (3). In high-income
countries, the prevalences of ante- and postpartum depression are
approximately 10%–15% (4–6). Two recent systematic reviews
in low- and low-middle–income countries suggested that perinatal depression is even more prevalent (7, 8). We recently reported
prevalences of antepartum depression of 27% and 33% in samples from Ghana and Côte d’Ivoire, respectively (9).
Maternal depression has been associated with adverse birth
outcomes and reduced postpartum growth, poor cognitive outcome, and psychiatric morbidity in childhood and adolescence
(10–15). There are several plausible biological and psychosocial
mechanisms via which maternal depression may affect children’s
health, including through direct impact on fetal development
(16–18) and through maternal parenting behavior (19). Yet,
studies on the association between maternal depression and
childhood morbidity are extremely rare. We hypothesized that
1394
Am J Epidemiol. 2013;178(9):1394–1402
Maternal Depression and Febrile Illness in Children 1395
ante- and postpartum depression in women is associated with
an increased risk of infections in their offspring, particularly
in low-income countries in which the prevalence of depression is high and infectious diseases are the main drivers of
childhood morbidity (20). We examined longitudinal data of
mother/child dyads from the Child Development Study in
Ghana and Côte d’Ivoire (9).
MATERIALS AND METHODS
approved by the ethical committee of the Kwame Nkrumah
University of Science and Technology in Kumasi, Ghana, the
National Ethical Committee in Abidjan, Côte d’Ivoire, and
the ethical committee of the Chamber of Physicians in Hamburg, Germany. Generally, persons suffering from nonpsychotic depression were able to understand and consent to study
requirements.Allparticipatingmothersprovidedwritteninformed
consent.
Participants and procedures
Study design and setting
The Child Development Study is a birth cohort of mothers
and children in Ghana and Côte d’Ivoire. Details of the study
design have previously been described elsewhere (9). In brief,
women in their last trimester of pregnancy were consecutively
recruited in 2 large hospitals, the Komfo Anokye Teaching Hospital in Kumasi, Ghana, and the Abobo Community Hospital in Abidjan, Côte d’Ivoire, during antepartum care visits
between March 2010 and December 2011. Whereas the Komfo
Anokye Teaching Hospital serves a mixed population in Ghana’s
second largest city, the Abobo Community Hospital provides
healthcare to an underprivileged population affected by civil
war. Antepartum care coverage is approximately 87% in Ghana
and 60% in Côte d’Ivoire (21). Prevalences of low birth weight
and preterm birth are between 9% and 17% in the 2 countries
(22, 23). After birth, the children of included mothers were
enrolled into the prospective birth cohort for a 2-year followup period.
Ethics statement
The present study was conducted in accordance with the
ethical principles of the Declaration of Helsinki. It was
All women in their third trimester of pregnancy who
resided within 5 km or less from the hospitals were eligible
for participation. We restricted the sample to 1) mothers with
relatively low obstetric risk and 2) children without severe
birth complications so that we could address confounding by
birth-related factors. Exclusion criteria for mothers were an
age less than 18 years, pregnancy with multiples, and severe
pregnancy complications (e.g., hypertension, hemorrhage, preeclampsia, and diabetes). Children were excluded in cases of
severe malformation or life-threatening condition detected perinatally, an Apgar score less than 5 after 1 minute and less than 7
after 5 minutes, intrauterine death, early postpartum death, premature birth (<34 gestational weeks), a birth weight less than
2,500 g, or massive clinical jaundice. Because of political
instability and resumed fighting in Côte d’Ivoire during the
study period, 311 mothers did not give birth in the designated hospitals and could therefore not be followed-up
(Figure 1).
After the baseline visit during the third trimester, followup visits 3 and 12 months after birth were scheduled to
collect data on maternal depression, breastfeeding practices,
and children’s physical statuses. Mothers were encouraged
to visit the recruiting hospital when their child was sick, and
1,206 Pregnant women in their third
trimester of pregnancy were asked to
participate
55 Women refused to participate
121 Women fulfilled exclusion
criteria
1,030 Women participated in the
baseline examination
311 Women did not give birth in
the study hospitals
719 Mother/child dyads were screened
53 Children fulfilled exclusion
criteria
666 Mother/child dyads were included
12 Mothers and/or children were
lost to follow-up
654 Mother/child dyads were included in
the analysis
Figure 1. Recruitment flow chart of the Child Development Study, Ghana and Côte d’Ivoire, 2010–2011.
Am J Epidemiol. 2013;178(9):1394–1402
1396 Guo et al.
every effort was made to accommodate the needs of the study
participants to ensure compliance (e.g., very short waiting periods,
expert attending pediatricians, subsidized fees for service).
Exposure, outcome, and covariates
The exposure variable was maternal depression, and it was
assessed antepartum and 3 and 12 months postpartum to capture the time-varying nature of depression using the Patient
Health Questionnaire depression module (PHQ-9) (24). The
questionnaire was translated from English into the local language Twi in Ghana and back. In Côte d’Ivoire, the official
translations into French that were provided by the original authors
were used. Given that illiteracy is common in both settings,
questionnaires were adapted for interviewer administration.
The PHQ-9 refers to the past 2 weeks and assesses the presence and severity of the 9 depression criteria from the Diagnostic and Statistical Manual of Mental Disorders-IV, which
cover emotional, cognitive, and functional somatic symptoms.
Response options generate a continuous score that can range
from 0 (no symptoms) to 27 (all symptoms present nearly
every day); scores of 10–14 represent moderate depression
symptoms and scores of 15–27 represent moderately severe to
severe depression symptoms. The PHQ-9 has been validated
for use in the general population and in primary care, as well
as in obstetrics and gynecology samples in both high- and
low-income settings. In rural Ghanaian women postpartum,
the PHQ-9 proved superior to other common depression
screening measures when using a semi-structured clinical interview as reference standard (25). A threshold score of 10 or
higher had a sensitivity of 88% and a specificity of 88% for
major depression (24, 26) and was used for case classification.
The term “depression” thus refers to the result of a robust
screening procedure in an epidemiologic study and not to a
clinical diagnosis. Women who were screened positive were
referred to local mental health professionals. The reliability
of the PHQ-9 score was somewhat limited (for Ghanaian
women, Cronbach’s α = 0.69, and for Ivorian women, α =
0.64). This may indicate that interviewer influences, a different understanding of the items (and therefore the diagnostic
criteria), and/or the response categories might have been
additional sources of error. Yet, the results showed consistent
patterns. A nonoptimal reliability inflates standard errors and
may also have affected the validity of our results. However, if
misclassification occurred, it was likely nondifferential and
probably biased our measure of association towards the null.
The outcome was febrile illness (body temperature ≥37.5°C)
in children with an unambiguous medical diagnosis that was
severe enough to prompt a medical prescription by the attending expert pediatrician. This could have been, for example, an
antibiotic, antimalarial, or paracetamol to reduce fever. This
outcome was chosen because of its importance in low-income
countries in which infectious diseases are by far the most important drivers of morbidity and mortality in young children (20).
The “prescription criterion” captured disease severity in this
setting. We considered this conservative outcome classification
more reproducible than a classification as “severe” or “nonsevere” by the clinician. The pediatricians who diagnosed the
children were unaware of the mothers’ depression status. The
files of mothers and children were kept separated.
There were several potential confounders. The first was socioeconomic status (SES). Maternal depression has been associated with low SES (27), and low SES is associated with poor
outcome in children (28, 29). A questionnaire comprising 15
items on different aspects of region-specific determinants of living conditions, such as educational level, occupation, religion,
access to cooking gas, access to a flush toilet, and ownership of
a number of goods, was completed by the pregnant women to
assess SES at baseline. Items with low discriminating power or
many missing values were disregarded. A principal component
analysis was conducted, and an SES index score was constructed based on 6 items: ownership of refrigerator, car, or bed
net, toilet type, maternal educational level, and spouse educational level (Cronbach’s α = 0.67). Tertiles were created (30).
The second potential confounder was anxiety. Comorbidity of
anxiety and depression in the perinatal period has been previously described (31, 32), and pre- and postpartum maternal
anxiety have been associated with unfavorable childhood outcomes (17, 33). Anxiety was assessed antepartum and 3 and
12 months postpartum using the Generalized Anxiety Disorder
7-item anxiety scale. The third potential confounder was country. The fourth was the season in which an infection occurred.
There is an increased risk of Plasmodium infections during the
rainy season in malaria-endemic areas. An event was allocated
to a particular season if it occurred at least 3 weeks after onset
of this season. The fifth was sex. In some cultures, male children may receive more care and attention than female children
(34). The sixth potential confounder was maternal age; a maternal age less than 20 years or greater than 30 years may be associated with an increased risk of illness in young children (35).
The seventh was maternal anemia. Maternal anemia may indicate conditions such as poor nutritional status, gynecological
problems, or chronic exposure to malaria parasites and may be
associated with altered maternal mood and mother-child interactions. The eighth potential confounder was the use of intermittent preventive malaria treatment in pregnancy, and the ninth
was whether the mother had been infected with the human
immunodeficiency virus.
Potential mediators of the association between maternal
depression and febrile illness in children were also examined.
These included children’s nutritional status, premature birth,
and breastfeeding practices.
Statistical analysis
The average rate of febrile illness in children was calculated as the total number of sick-child visits with fever, a medical diagnosis, and a pharmaceutical prescription divided by
the total person-years during which the children were under
observation. Time-to-event analysis was used, and followup was counted as the time from birth to the last visit. Participants were censored if they did not have any events until the
final (2 years after birth) visit or were lost to follow-up. A
recurrent-event Cox proportional hazards model (Prentice,
Williams, and Peterson model) (36) was used to evaluate the
overall hazard ratio of febrile illness in children of mothers
with and without depression. The recurrent-event Cox proportional hazards model was chosen because it allowed us to
account for the dependence of events within a person, as
well as the time-varying exposure. The proportional hazards
Am J Epidemiol. 2013;178(9):1394–1402
Maternal Depression and Febrile Illness in Children 1397
assumption was verified by plotting the estimated cumulative
hazards over time. The correlation of multiple events within the
same participant was accounted for by adjusting the estimates
of the standard errors using a robust variance estimator (37).
Maternal depression was dichotomized (PHQ-9 score ≥10
indicated depressionandPHQ-9 <10 indicated nondepression)
and the most recent antecedent exposure status was used for
each outcome episode. In the multivariate model, we adjusted
for potential confounders that were associated with both the
exposure and the outcome in the univariate model. To assess
whether breastfeeding, premature birth, or infant nutritional status mediated the association between maternal depression and
the outcome, the hazard ratio of maternal depression associated
with achild’s febrile illness was calculated with and withoutadjustmentforthesefactors.Toexplorethepossibleinteractionbetween
country and depression, we assessed the interaction term in the
multivariate model. In addition, we conducted analyses stratified by country. Sensitivity analyses examined simulated data
with even intervals between each exposure measurement, the
cumulative association between recurrent depression and the
outcome, and an analysis of 288 mother/child dyads who had
been followed up for 2 years at the time of analysis.
Multicollinearity was examined by calculating the variance
inflation factor. With no variance inflation factor greater than
10, we found no indication of multicollinearity. All tests were
performed with a 2-sided significance level of 0.05. Point
estimates were complemented by symmetric 95% confidence
intervals.
RESULTS
Few women (n = 55) refused to participate in the study
(Figure 1). Table 1 shows characteristics of mothers and children by maternal antepartum depression status. Prevalences of
maternal depression at baseline and at 3 and 12 months were
28.3%, 11.8%, 16.1%, respectively, in Côte d’Ivoire and 26.3%,
8.9%, and 7.2%, respectively, in Ghana.
Febrile illness in children
The mean follow up time was 1.3 years, and the total time
at risk was 878.5 years in 654 children. A total of 523 events
were recorded. The most frequent clinical diagnoses were
malaria, upper respiratory tract infection, and urinary tract infection (Table 2).
The rates of febrile illness were 0.74 per person-year (95%
confidence interval (CI): 0.60, 0.92) and 0.50 per personyear (95% CI: 0.46, 0.56), respectively, among children of
Table 1. Characteristics of Mothers and Children by Maternal Antepartum Depression Status in the Child Development Study, Ghana and Côte
d’Ivoire, 2010–2011
Nondepressiona (n = 475)
Depressiona (n = 179)
Total (n = 654)
Variable
Mean (SD)
No.
%
Mean (SD)
No.
%
Mean (SD)
No.
%
P Value
Age, years
28.8 (5.3)
29.7 (5.9)
29.1 (5.4)
0.06
Mothers’ hemoglobin
level, g/dL
11.1 (1.2)
11.0 (1.1)
11.1 (1.2)
0.28
3179.8 (422.4)
3219.6 (434.2)
3190.7 (425.7)
Children’s birth weight, g
0.29
Country
0.57
Ghana
216
45.5
77
43.0
293
44.8
Côte d’Ivoire
259
54.5
102
57.0
361
55.2
Top 20%
101
22.0
25
14.4
126
19.9
Middle 50%
233
50.7
90
51.7
323
51.0
Bottom 30%
126
27.4
59
33.9
185
29.2
Male
234
49.3
100
55.9
334
51.1
Female
241
50.7
79
44.1
320
48.9
Socioeconomic status
0.06
Child sex
0.13
Episodes of exclusive
breastfeeding
0.34
0
13
3.5
3
2.3
16
3.2
1
243
65.9
78
60.5
321
64.5
2
113
30.6
48
37.2
161
32.3
9
3.4
4
3.7
13
3.4
0.88
216
45.9
76
42.9
292
45.1
0.51
Maternal HIV infection
Maternal IPTp use
Abbreviations: HIV, human immunodeficiency virus; IPTp, intermittent preventive treatment in pregnancy; PHQ-9, Patient Health Questionnaire
depression module; SD, standard deviation.
a
Nondepression was defined as having a raw score less than 10 on the Patient Health Questionnaire depression module; depression was
defined as having a score of 10 or higher.
Am J Epidemiol. 2013;178(9):1394–1402
1398 Guo et al.
depressed and nondepressed mothers. Approximately half of
the children did not have any febrile illness during the followup period. More than 25% had 1 episode of febrile illness,
and approximately 10% had 2 episodes. Less than 10% of children had 3 or more episodes of febrile illness. Most (73%)
events happened during the first year of life. The rates of febrile
illness were 0.66 (95% CI: 0.58, 0.71) and 0.37 (95% CI: 0.31,
0.45) per person-year, respectively, in the first and second
year of life.
Table 2. Clinical Diagnoses in Febrile Children in the Child
Development Study, Ghana and Côte d’Ivoire, 2010–2011
Total No. of
Febrile Illness
No. of
Cases
Malaria
523
211
Upper respiratory tract
infection
523
176
Urinary tract infection
523
132
Diarrhea
523
68
Lower respiratory tract
infection
523
59
Skin infection
523
50
Influenza-like disease
523
39
Specific Diagnosisa
Maternal depression and febrile illness in children
Oral infection
523
51
Ear infection
523
17
Eye infection
523
13
Anemia
523
11
Nose infection
523
9
Other diagnoses
523
18
a
Multiple diagnoses can be made for 1 episode of febrile illness.
Table 3 presents the crude and adjusted hazard ratios of
febrile illness in children and maternal depression. In the univariate model, the hazard for Ghanaian children was one fifth that
for Ivorian children. The hazard decreased with increasing
SES in both countries. The hazard of febrile disease in children
of depressed mothers was 57% higher than the hazard in children
of nondepressed mothers.
Country and SES were identified as confounders. After
adjusting for both, the hazard of developing a febrile illness
was 32% higher in children of mothers with depression than
Table 3. Cox Proportional Hazards Regression Models of Febrile Illness in Children in the Child Development
Study, Ghana and Côte d’Ivoire, 2010–2011
Unadjusted
Variable
HR
95% CI
Depressiona
1.57
1.20, 2.07
Ghanab
0.20
Middle 50%
Top 20%
Adjusted
P Value
P Value
HR
95% CI
0.001
1.32
1.01, 1.74
0.04
0.16, 0.26
<0.001
0.24
0.18, 0.32
<0.001
0.65
0.52, 0.81
<0.001
0.97
0.78, 1.20
0.76
0.23
0.15, 0.34
<0.001
0.67
0.43, 1.05
0.08
Slight
1.08
0.86, 1.35
0.51
Moderate
1.59
1.05, 2.40
0.03
Severe
0.67
0.19, 2.45
0.55
Infection during the wet season
0.91
0.75, 1.10
0.33
Female child sexf
0.87
0.70, 1.08
0.20
Maternal age
1.00
0.98, 1.02
0.73
Maternal anemia
1.28
0.95, 1.74
0.11
Maternal IPTp use
0.91
0.72, 1.16
0.46
Maternal HIV infection
1.56
0.88, 2.75
0.13
Socioeconomic statusc
Maternal anxietyd
e
Abbreviations: CI, confidence interval; HIV, Human immunodeficiency virus; HR, hazard ratio; IPTp, intermittent
preventive treatment in pregnancy.
a
Depression was defined as having a raw score on the Patient Health Questionnaire depression module of 10 or
higher; nondepression was defined as having a raw score less than 10. Nondepressed mothers were the reference
group.
b
Subjects from Côte d’Ivoire were the reference group.
c
The bottom 30% were the reference group.
d
Anxiety was defined as follows: none, Generalized Anxiety Disorder 7-item anxiety scale raw score of 4 or less;
mild anxiety, a raw score of 5–9; moderate anxiety, a raw score of 10–14; and severe anxiety, a raw score of 15–21.
Subjects with no anxiety were the reference group.
e
The dry season was the reference group.
f
Male children were the reference group.
Am J Epidemiol. 2013;178(9):1394–1402
Maternal Depression and Febrile Illness in Children 1399
Table 4. Children’s Nutritional Status at Birth and 3 and 12 Month
After Birth in the Child Development Study, Ghana and Côte d’Ivoire,
2010–2011
Birth
(n = 654)
Nutritional Status
No.
3 Month
(n = 595)
12 Month
(n = 422)
%
No.
%
No.
%
Wasting (weight-for-age
z score <−2)
1
0.2
25
4.2
25
5.9
Stunting (height-for-age
z score <−2)
45
6.9
38
6.4
26
6.2
Undernutrition (weight-forheight z score <−2)
65
10.3
41
6.9
19
4.5
it was in children of mothers without depression (Table 3).
We analyzed the interaction term between depression and country. The hazard ratio for interaction was 1.45 (95% CI: 0.72,
2.92; P = 0.29), meaning we did not find evidence of an interaction (Web Table 1, available at http://aje.oxfordjournals.
org/). The stratified hazard ratios in Ghana and Côte d’Ivoire
were 1.62 (95% CI: 0.86, 3.08; P = 0.14) and 1.21 (95% CI:
0.91, 1.60; P = 0.19), respectively (Web Table 2). Because
of the halved sample size, these stratified results lost significance. We concluded that we did not find evidence of an
interaction.
We did not detect a mediating effect of breastfeeding, late
prematurity, or children’s nutritional status. Table 4 shows the
frequencies of stunted, wasted, and undernourished children
at birth and 3 and 12 month after birth.
Cumulative maternal depression and febrile illness
in children
We constructed a cumulative depression exposure by categorizing mothers as never or once depressed and depressed
2 or 3 times. The crude and adjusted hazard ratios for children of recurrently depressed mothers compared with those
for children of mothers with fewer episodes were 2.20 (95%
CI: 1.51, 3.19) and 1.90 (95% CI: 1.32, 2.75), respectively.
Sensitivity analyses
Table 5 shows the results of the sensitivity analyses. The
association between maternal depression and febrile illness
in children was robust and the hazard ratios appeared to be
even larger in the subset of participants who have completed
the 2-year follow up.
DISCUSSION
To our knowledge, this is the first longitudinal study that has
explored 1) the prevalence of antepartum and postpartum
depressive episodes in African women, 2) their change over
time, and 3) the association with children’s health. The outcome was clinically confirmed, and the exposure was measured using an established questionnaire. Children of depressed
mothers had an increased risk of febrile illness. After adjustment
for confounders, the point estimate decreased only slightly. The
risk was even higher in children of recurrently depressed women.
Our findings support our hypothesis that maternal depression
affected not only the mother but also the health of the child in
our sample.
A few studies on maternal depression and illness in children
reported similar findings. In a cohort study from Pakistan,
Rahman et al. (38) found that among children born to depressed
mothers compared with children born to nondepressed mothers, the odds ratio of having 5 or more episodes of diarrhea
per year was 3.1 (95% CI: 1.8, 5.6) after adjustment for confounders. In a Nigerian study focused on infant growth during
the first 9 months of life, Adewuya et al. (10) reported a significant increase in the average number of episodes of diarrhea
and other infectious illnesses in infants of depressed mothers
(5.23 episodes; standard deviation, 2.37) compared with infants
of nondepressed mothers (3.70 episodes; standard deviation,
4.14). However, that study did not adjust for potential confounders. A study of a large population-based cohort in the
United Kingdom found that children of women with antepartum depression had a 40% increased risk (incidence rate
ratio = 1.40, 95% CI: 1.37, 1.42) of gastrointestinal infections
and a 30% increased risk (incidence rate ratio = 1.30, 95%
Table 5. Hazard Ratios From Sensitivity Analyses of the Association Between Maternal Depression and Febrile
Illness in children in the Child Development Study, Ghana and Côte d’Ivoire, 2010–2011
Unadjusted
Depression
Analysisa
Adjusted
HR
95% CI
P
HR
95% CI
Analysis Ab
1.35
1.05, 1.75
0.02
1.28
1.00, 1.65
0.05
Analysis Bc
2.39
1.56, 3.66
<0.001
1.83
1.21, 2.75
0.004
d
2.24
1.43, 3.53
<0.001
1.83
1.20, 2.83
0.005
Analysis C
P
Abbreviations: CI, confidence interval; HR, hazard ratio.
a
Depression was defined as having a raw score on the Patient Health Questionnaire depression module of 10 or
higher; nondepression was defined as having a raw score less than 10. Nondepressed mothers were the reference
group.
b
Analysis A featured a simulated dataset with even intervals between exposure assessment time points.
c
Analysis B contained a subset of participants who had been followed-up for 2 years (n = 288).
d
Analysis C featured a simulated dataset with even intervals between exposure assessment time points within
subset of participants who had been followed-up for 2 years (n = 288).
Am J Epidemiol. 2013;178(9):1394–1402
1400 Guo et al.
CI: 1.22, 1.32) of lower respiratory tract infection. That study
did not adjust for confounders (39). A cohort study in the United
States found that postpartum maternal depression was associated
with more hospital admissions and emergency department visits
in offspring for up to 30 months (odds ratio = 1.4, 95% CI: 1.2,
1.8).However,diagnosesofchildrenwereobtainedthrough medical records and from parent report only (40).
There are several possible mechanisms through which maternal depression may affect the health of offspring. Depressed
pregnant women more frequently exposed their fetuses to alcohol, nicotine, and drugs (41). Some studies suggested that the
high cortisol levels associated with antepartum depression
could have a direct influence on fetal development by affecting the hypothalamic–pituitary–adrenal axis, limbic system,
and the prefrontal cortex (16–18). Postpartum depression may
disturb the mother’s parenting behavior and mother-child interaction (19). Impeded maternal care poses a high risk in the first
year of life because infants require more care during this period
and are more susceptible to their mothers’ depressive moods
(42). Maternal depression may lead to disengagement or child
neglect, delayed healthcare-seeking (24), and decreased preventive behavior, such as hand-washing before feeding (38).
Some studies reported that maternal depression was associated with reduced frequency or duration of breastfeeding
(6). Our data did not confirm this finding. We assessed breastfeeding behavior at 2 time points postpartum and may not
have fully captured this practice. Breastfeeding behavior is
complex and differs across cultures and with social norms
(43). Previous studies have suggested that maternal depression predicts poor infant nutritional status (10, 38, 44), which
may mediate increased risk of illness. Our analysis did not confirm this pattern. We restricted our sample to babies without a
very low birth weight and therefore may have missed children
that could be prone to nutritional deficiencies due to unknown
causes. Generally, children in our study were rarely malnourished during the follow-up period.
Ivorian and Ghanaian women showed changes in depression
over time, suggesting that depression is a relatively dynamic process in our sample and subject to change in the first year postpartum. This dynamic pattern has also been found in the United
States and China (45, 46). Some studies have suggested that antepartum depression or a prior episode of postpartum depression
were predictive of recurrent postpartum depression, and approximately 50%–70% of women depressed in the puerperium were
at risk of chronic depression (6, 38). We observed that around
one-fourth of women in our study who were depressed antepartum had 1 or 2 episodes of postpartum depression.
The present study has limitations. First, although we
restricted our sample to pregnant women who were physically healthy at recruitment and children without serious birthrelated complications to avoid confounding by pre-existing
maternal physical illnesses and very poor birth outcome, residual confounding is possible. We used multivariate regression analysis to control for potential confounders, but there
may be unknown confounders that we did not measure. Second, selection bias is possible because 323 mothers either did
not give birth in the study hospitals or were lost to follow up
from the beginning of the study. They were almost exclusively from Côte d’Ivoire and lost to observation because of
the confusion during and after the armed conflict in that country that resumed 6 months after recruitment started. The missing mothers were, on average, more depressed, younger, and
of lower SES. We may therefore have underestimated the prevalence of maternal depression. Third, information bias may
occur when using imperfect screening instruments like the
PHQ-9. Forth, the findings are likely not generalizable to all
African women because our sample was restricted to urban,
relatively healthy pregnant women who had access to healthcare.
This large, prospective cohort established the correct temporal sequence, that is, the exposure had to occur before the
outcome. Yet, it is theoretically possible that febrile childhood illness from month 3 on influenced depression status in
the mother. We believe that our directly observed outcome
is particularly meaningful in low-income settings in which
febrile diseases strongly drive infant morbidity during the
first 2 years of life (20). We used a recurrent-event Cox proportional hazards model to account for the dependence of
events within a person and the time-varying exposure. We
modeled our exposure variable quite dynamically, that is,
each episode of maternal depression was assumed to have an
immediate impact on children’s health. To measure the cumulative influence of repeated maternal depression, we conducted
additional analyses with repeated maternal depression as the
exposure and found that the association was stronger, which
suggests a possible dose–response relationship. Our results were
robust to several sensitivity analyses.
Our findings underline and expand the knowledge on depression in women and illness in their children in sub-Saharan
Africa. This association between 2 major public health problems (mental health and infectious diseases) in this region of
the world emphasizes the need for a multidisciplinary approach
to assessing the relationship between the health of mothers and
their children and to integrating diagnosis and treatment of
depression into primary healthcare. Should further studies confirm the high prevalences of ante- and postpartum depression in
sub-Saharan African women, the development and rigorous testing of effective, locally acceptable, feasible, and safe interventional strategies is recommended.
ACKNOWLEDGMENTS
Author affiliations: Department of Epidemiology, Bloomberg School of Public Health, Johns Hopkins University,
Baltimore, Maryland (Nan Guo, Stephan Ehrhardt); Department of Child and Adolescent Psychiatry, University Medical
Center Hamburg-Eppendorf, Hamburg, Germany (Carola
Bindt, Claus Barkmann); Centre de Guidance Infantile, Institut National de Santé Publique, Abidjan, Côte d’Ivoire (Marguerite Te Bonle); Department of Behavioural Sciences, School
of Medical Sciences, Kwame Nkrumah University of Science
and Technology, Kumasi, Ghana (John Appiah-Poku); Clinical
Research Unit, Bernhard Nocht Institute for Tropical Medicine,
Hamburg, Germany (Rebecca Hinz, Dana Barthel, Sarah
Posdzich, Simon Deymann, Lisa Schlüter, Kirsten A. Eberhardt,
Am J Epidemiol. 2013;178(9):1394–1402
Maternal Depression and Febrile Illness in Children 1401
Stephan Ehrhardt); Jean Lorougnon Guede University, Daloa,
Côte d’Ivoire (Mathurin Koffi); Infectious Disease Epidemiology, Bernhard Nocht Institute for Tropical Medicine, Hamburg,
Germany (Anna Jaeger); Department of Child Health, Komfo
Anokye Teaching Hospital, Kumasi, Ghana (Samuel Blay
Nguah); Research Unit of Parasitology and Parasite Ecology
at Unité de Formation et de Recherche en Biosciences, Université de Cocody, Abidjan, Côte d’Ivoire (Eliezer N’Goran);
and Department Community Health, School of Medical Sciences, Kwame Nkrumah University of Science and Technology, Kumasi, Ghana (Harry Tagbor).
Nan Guo and Carola Bindt contributed equally to this work.
This work was supported by an academic grant to Stephan
Ehrhardt from the German Research Council (Deutsche
Forschungsgemeinschaft; grant EH 384/1-1e).
The International Child Development Study Study Group
comprises (in addition to the authors and in alphabetical
order): Jana Baum (Hamburg, Germany), Gerd D. Burchard
(Hamburg, Germany), Lisa Claussen (Hamburg, Germany),
Heike Ewert (Hamburg, Germany), Torsten Feldt (Hamburg,
Germany), Daniel Fordjour (Kumasi, Ghana), Andreas Hahn
(Hamburg, Germany), Jean E. Koffi (Abidjan, Côte d’Ivoire),
Esther Kra (Abidjan, Côte d’Ivoire), Wibke Loag (Hamburg, Germany), Jürgen May (Hamburg, Germany), Yasmin
Mohammed (Kumasi, Ghana), Yaw Osei (Kumasi, Ghana),
Birgit Reime (Hamburg, Germany), Stefanie Schoppen (Hamburg, Germany), and Egbert Tannich (Hamburg, Germany).
The results of this study were in part presented at the 2013
Meeting of the Society for Epidemiologic Research, June
18–21, Boston, Massachusetts, and published abstract form
(Am J Epidemiol. 2013;177(11 suppl):S178).
This work forms part of the MD, PhD, or master theses of
D.B., J.B., L.C., L.S., S.D., and S.P.
The funders had no role in study design, data collection and
analysis, decision to publish, or preparation of the manuscript.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
Conflict of interest: none declared.
18.
REFERENCES
1. Prince M, Patel V, Saxena S, et al. No health without mental
health. Lancet. 2007;370(9590):859–877.
2. World Health Organization. Depression Fact Sheet. Geneva,
Switzerland: World Health Organization; 2012. (http://www.
who.int/mediacentre/factsheets/fs369/en/index.html).
(Accessed November 21, 2012).
3. Bennett HA, Einarson A, Taddio A, et al. Prevalence of
depression during pregnancy: systematic review. Obstet
Gynecol. 2004;103(4):698–709.
4. Beck CT. Predictors of postpartum depression: an update.
Nurs Res. 2001;50(5):275–285.
5. Gavin NI, Gaynes BN, Lohr KN, et al. Perinatal depression: a
systematic review of prevalence and incidence. Obstet
Gynecol. 2005;106(5):1071–1083.
6. Marcus SM. Depression during pregnancy: rates, risks and
consequences—Motherisk Update 2008. Can J Clin
Pharmacol. 2009;16(1):e15–e22.
7. Fisher J, de Mello MC, Patel V, et al. Prevalence and
determinants of common perinatal mental disorders in
Am J Epidemiol. 2013;178(9):1394–1402
19.
20.
21.
22.
23.
24.
women in low- and lower-middle-income countries: a
systematic review. Bull World Health Organ. 2012;90(2):
139–149H.
Sawyer A, Ayers S, Smith H. Pre- and postnatal psychological
wellbeing in Africa: a systematic review. J Affect Disord.
2010;123(1-3):17–29.
Bindt C, Appiah-Poku J, Te Bonle M, et al. Antepartum
depression and anxiety associated with disability in African
women: cross-sectional results from the CDS study in Ghana
and Côte d’Ivoire. PLoS One. 2012;7(10):e48396.
Adewuya AO, Ola BO, Aloba OO, et al. Impact of postnatal
depression on infants’ growth in Nigeria. J Affect Disord.
2008;108(1-2):191–193.
Patel V, DeSouza N, Rodrigues M. Postnatal depression and
infant growth and development in low income countries: a
cohort study from Goa, India. Arch Dis Child. 2003;88(1):
34–37.
Grote NK, Bridge JA, Gavin AR, et al. A meta-analysis of
depression during pregnancy and the risk of preterm birth, low
birth weight, and intrauterine growth restriction. Arch Gen
Psychiatry. 2010;67(10):1012–1024.
Traviss GD, West RM, House AO. Maternal mental health and
its association with infant growth at 6 months in ethnic groups:
results from the Born-in-Bradford birth cohort study. PLoS
One. 2012;7(2):e30707.
Avan B, Richter LM, Ramchandani PG, et al. Maternal
postnatal depression and children’s growth and behaviour during
the early years of life: exploring the interaction between physical
and mental health. Arch Dis Child. 2010;95(9):690–695.
Murray L, Arteche A, Fearon P, et al. Maternal postnatal
depression and the development of depression in offspring up
to 16 years of age. J Am Acad Child Adolesc Psychiatry.
2011;50(5):460–470.
Field T, Diego M, Hernandez-Reif M. Prenatal depression
effects and interventions: a review. Infant Behav Dev. 2010;
33(4):409–418.
Van den Bergh BR, Mulder EJ, Mennes M, et al. Antenatal
maternal anxiety and stress and the neurobehavioural
development of the fetus and child: links and possible
mechanisms. A review. Neurosci Biobehav Rev. 2005;29(2):
237–258.
Weinstock M. The potential influence of maternal stress
hormones on development and mental health of the offspring.
Brain Behav Immun. 2005;19(4):296–308.
Grace SL, Evindar A, Stewart DE. The effect of postpartum
depression on child cognitive development and behavior: a
review and critical analysis of the literature. Arch Womens
Ment Health. 2003;6(4):263–274.
World Health Organization. Global Burden of Disease: 2004
Update. Geneva, Switzerland: World Health Organization; 2008.
(http://www.who.int/healthinfo/global_burden_disease/2004_
report_update/en/index.html). (Accessed November 21, 2012).
Arthur E. Wealth and antenatal care use: implications for
maternal health care utilisation in Ghana. Health Econ Rev.
2012;2(1):14.
Msaki BP, Mshana SE, Hokororo A, et al. Prevalence and
predictors of urinary tract infection and severe malaria among
febrile children attending Makongoro health centre in Mwanza
city, North-Western Tanzania. Arch Public Health. 2012;
70(1):4.
Nkyekyer K, Enweronu-Laryea C, Boafor T. Singleton
preterm births in korle bu teaching hospital, accra, ghana—
origins and outcomes. Ghana Med J. 2006;40(3):93–98.
Kroenke K, Spitzer RL, Williams JBW. The PHQ-9. J Gen
Intern Med. 2001;16(9):606–613.
1402 Guo et al.
25. Weobong B, Akpalu B, Doku V, et al. The comparative
validity of screening scales for postnatal common mental
disorder in Kintampo, Ghana. J Affect Disord. 2009;
113(1-2):109–117.
26. Kroenke K, Spitzer RL, Williams JBW, et al. The patient
health questionnaire somatic, anxiety, and depressive symptom
scales: a systematic review. Gen Hosp Psychiatry. 2010;
32(4):345–359.
27. Miech RA, Caspi A, Moffitt TE, et al. Low socioeconomic
status and mental disorders: a longitudinal study of selection
and causation during young adulthood. Am J Sociol. 1999;
104(4):1096–1131.
28. Dutton DB. Socioeconomic status and children’s health. Med
Care. 1985;23(2):142–156.
29. Currie J, Stabile M. Socioeconomic status and child health:
why is the relationship stronger for older children? Am Econ
Rev. 2003;93(5):1813–1823.
30. Fotso J-C, Kuate-Defo B. Measuring socioeconomic status in
health research in developing countries: should we be focusing
on households, communities or both? Soc Indic Res. 2005;
72(2):189–237.
31. Austin MP, Tully L, Parker G. Examining the relationship
between antenatal anxiety and postnatal depression. J Affect
Disord. 2007;101(1-3):169–174.
32. Ross LE, Gilbert Evans SE, Sellers EM, et al. Measurement
issues in postpartum depression part 1: anxiety as a feature of
postpartum depression. Arch Womens Ment Health. 2003;
6(1):51–57.
33. Schreier A, Wittchen HU, Hofler M, et al. Anxiety disorders in
mothers and their children: prospective longitudinal
community study. Br J Psychiatry. 2008;192(4):308–309.
34. Lovel H, Sabir N, Cleland J. Why are toddler girls at risk of
death and undernutrition in a slum area of Pakistan? [letter].
Lancet. 1984;323(8380):797.
35. Rutstein SO. Factors associated with trends in infant and child
mortality in developing countries during the 1990s. Bull World
Health Organ. 2000;78(10):1256–1270.
36. Prentice RL, Williams BJ, Peterson AV. On the regression
analysis of multivariate failure time data. Biometrika. 1981;
68(2):373–379.
37. Lin DY, Wei LJ. The robust inference for the Cox
proportional hazards model. J Am Stat Assoc. 1989;84(408):
1074–1078.
38. Rahman A, Bunn J, Lovel H, et al. Maternal depression
increases infant risk of diarrhoeal illness: a cohort study. Arch
Dis Child. 2007;92(1):24–28.
39. Ban L, Gibson JE, West J, et al. Association between perinatal
depression in mothers and the risk of childhood infections in
offspring: a population-based cohort study. BMC Public
Health. 2010;10:799.
40. Minkovitz CS, Strobino D, Scharfstein D, et al. Maternal
depressive symptoms and children’s receipt of health care
in the first 3 years of life. Pediatrics. 2005;115(2):306–314.
41. Hay DF, Pawlby S, Waters CS, et al. Antepartum and
postpartum exposure to maternal depression: different effects
on different adolescent outcomes. J Child Psychol Psychiatry.
2008;49(10):1079–1088.
42. Rahman A, Harrington R, Bunn J. Can maternal depression
increase infant risk of illness and growth impairment in
developing countries? Child Care Health Dev. 2002;
28(1):51–56.
43. Walker AR, Adam FI. Breast-feeding in sub-Saharan Africa:
outlook for 2000. Public Health Nutr. 2000;3(3):285–292.
44. Rahman A, Bunn J, Lovel H, et al. Maternal depression
increases infant risk of diarrhoeal illness: a cohort study. Arch
Dis Child. 2007;92(1):24–28.
45. Gjerdingen D, Crow S, McGovern P, et al. Changes in
depressive symptoms over 0–9 months postpartum. J Womens
Health. 2011;20(3):381–386.
46. Xie YC, Yuan HW, Zhuang RJ, et al. Study on the status
and related socio-psychological factors of maternal
depression among high-risk pregnancy women in Harbin city
(Chinese). Zhonghua Yu Fang Yi Xue Za Zhi. 2012;46(6):
543–546.
Am J Epidemiol. 2013;178(9):1394–1402