Matern Child Health J DOI 10.1007/s10995-013-1294-x
Intimate Partner Violence and Breastfeeding in Africa Emily S. Misch • Kathryn M. Yount
Springer Science+Business Media New York 2013
Abstract We examined the associations of maternal intimate partner violence (IPV) victimization with early initiation and exclusive breastfeeding in eight African countries. For mothers 15–49 years with an infant aged less than 6 months from national Demographic and Health Surveys since 2007 for Ghana (n = 173), Kenya (n = 449), Liberia (n = 313), Malawi (n = 397), Nigeria (n = 2007), Tanzania (n = 549), Zambia (n = 454), and Zimbabwe (n = 480), logistic regression was used to estimate the unadjusted and adjusted associations of lifetime maternal emotional, physical, and sexual IPV victimization with early initiation (less than 1 hour of birth) and exclusive breastfeeding in the prior 24 hours. Maternal lifetime IPV victimization often was adversely associated with optimal breastfeeding practices. Physical IPV in Zimbabwe (aOR 0.40, p = 0.002), sexual IPV in Zambia (aOR 0.42, p = 0.017), and emotional IPV in Kenya (aOR 0.54, p = 0.050) and Tanzania (aOR 0.57, p = 0.088) were associated with lower adjusted odds of early initiation. Sexual IPV in Liberia (aOR 0.09, p = 0.026), Ghana (aOR 0.17, p = 0.033), and Kenya (aOR 0.34, p = 0.085) were associated with lower adjusted odds of exclusive breastfeeding. Atypically, physical IPV in Tanzania (aOR 2.11, p = 0.042) and sexual IPV in Zambia (aOR 2.49, p = 0.025) were
E. S. Misch Hubert Department of Global Health, Rollins School of Public Health, Emory University, 1518 Clifton Rd, NE, Atlanta, GA 30322, USA e-mail:
[email protected] K. M. Yount (&) Asa Griggs Candler Chair of Global Health, Hubert Department of Global Health, Rollins School of Public Health and Department of Sociology, Emory University, 1518 Clifton Rd, NE, Atlanta, GA 30322, USA e-mail:
[email protected]
associated with higher adjusted odds of early initiation and exclusive breastfeeding, respectively. Across several settings, maternal IPV victimization may adversely influence breastfeeding practices. Longitudinal research of these relationships is warranted. Screening for IPV victimization and breastfeeding counseling in prenatal and postpartum care may mitigate the potential intergenerational effects of IPV. Keywords Breastfeeding Demographic and Health Surveys Intimate partner violence Sub-Saharan Africa
Introduction Early initiation of breastfeeding (within 1 hour of birth) and exclusive breastfeeding until 6 months of age are recommended practices [1, 2]. Early initiation provides the infant with maternal colostrum, nutrient-dense breast milk produced in the first days after delivery that helps to develop the infant’s mucosal immune system [3] and may reduce infection-related neonatal mortality [4, 5]. Exclusive breastfeeding until 6 months of age also imparts health benefits, including reductions in gastrointestinal and lower respiratory tract infections [6, 7]. An estimated 1.4 million deaths to infants in lower-income countries could be averted by practicing optimal breastfeeding [7, 8]. Yet, levels of early initiation and exclusive breastfeeding remain low in many countries, including in Sub-Saharan Africa [9]. Many factors influence a woman’s decision to initiate and continue breastfeeding. One hypothesized factor is intimate partner violence (IPV) victimization [10]—physical, sexual, or psychological harm, or the threat of such harm by a current or former partner [11]. Two opposing hypotheses may explain the relationship between maternal IPV victimization and breastfeeding. The deficit hypothesis
123
Matern Child Health J
contends that maternal victims of IPV may have difficulties with breastfeeding [10]. The type of IPV that a mother experiences may influence why and how this difficulty manifests. Mothers who experience sexual IPV may attach negative sexual associations to the breasts and breastfeeding [12]; whereas, mothers who experience sexual, physical, or emotional IPV may engage in negative coping behaviors [13], or may be physically, psychologically, or cognitively impaired, weakening their ability to breastfeed [14]. The compensatory hypothesis posits that maternal victims of IPV may be more sensitive to the needs of their child and show positive parenting behaviors, including optimal breastfeeding [15]. Few published studies have investigated the association of maternal IPV victimization with breastfeeding practices. Three studies found adverse associations of exposure to IPV with breastfeeding initiation and duration [16–18] but one was qualitative [18]. Three other studies found no association with breastfeeding practices [19–21], two after adjustment for confounding [19, 20]. The scant available research has several limitations [14]. First, physical IPV is examined more often than sexual or emotional IPV, which may have differing effects on a woman’s breastfeeding practices. Second, sample sizes are small, and adjustment for confounding is inconsistent. Third, women in poor countries are underrepresented, with women in Sub-Saharan Africa unrepresented despite their high rates of lifetime IPV victimization [22] and suboptimal breastfeeding practices [9]. Finally, although most agree with the deficit hypothesis, the most common direction of the relationship between IPV and breastfeeding is unclear. To fill this gap, we examined the association of maternal emotional, physical, and/or sexual IPV victimization with early breastfeeding initiation and exclusive breastfeeding of infants less than 6 months old in eight African countries.
Methods Data Source Our data sources were national Demographic and Health Surveys (DHS) conducted in Sub-Saharan Africa. Funded mainly by the U.S. Agency for International Development, the DHS are conducted in diverse lower- and middleincome countries and collect comparable data on household demographics, maternal and child health and nutrition, and reproductive health. In 1990, the DHS started to include a module on IPV. Based on the Revised Conflict Tactics Scales (CTS2) [23], questions ask about ever and prior-year physical, sexual, and emotional IPV victimization, physical IPV perpetration, and IPV during any pregnancy (in some countries).
123
Ten DHS conducted in Sub-Saharan Africa since 2007 had the requisite data for this analysis. To select the countries for inclusion, we grouped them by region [Eastern Africa (2), Western Africa (5), Southern Africa (3)], and we agreed to include all countries from each region with ‘‘adequate’’ power to detect a significant difference in the percentage of infants in sample dyads who were exclusively breastfed in the prior 24 hours by maternal lifetime IPV victimization status. The following estimates from each national DHS report were used to estimate power: prevalence of exclusive breastfeeding for infants less than 6 months old (value expected in unexposed mothers), prevalence of any lifetime IPV in all women 15–49 years (to estimate sample sizes of maternal IPV victims and nonvictims), an assumed 30 % difference in the prevalence of exclusive breastfeeding in maternal IPV victims and nonvictims, and a = 0.05 for a two-tailed test. Based on these estimates, power exceeded 65.0 % in eight countries, including two in Eastern Africa (Kenya 2008–2009, Tanzania 2010), three in Western Africa (Ghana 2008, Liberia 2007, Nigeria 2008), and three in Southern Africa (Malawi 2010, Zambia 2007, Zimbabwe 2010–2011). These countries are diverse economically and demographically and represent the range of breastfeeding practices observed in Sub-Saharan Africa since 2007 (Table 1). Sampling The DHS uses a stratified, multistage, cluster probability sample design. Details for the eight study countries are available elsewhere [24–31]. Each country’s population was stratified by urban or rural residence, and census enumeration areas were selected within strata. Households were selected randomly within enumeration areas, and all women 15–49 years in interviewed households were eligible to participate. A random subsample of interviewed households was selected for the IPV module (with proportions for the subsample varying across countries). Following international guidelines for research on IPV [32], only one randomly selected woman per household was administered the IPV module. Response rates for the household (96.0–98.8 %), woman (93.3–96.9 %), and IPV module (95.3—97.7 %) were high across all sample countries. Study Populations The study population for each country included women 15–49 years who were matched to an infant less than 6 months old, who had completed an IPV module with questions on psychological, physical, and sexual IPV, who also had answered questions about breastfeeding their youngest child, and who had complete data for all outcomes and exposure variables. Relatively few mother-infant dyads
Matern Child Health J Table 1 Selected characteristics of the study countries Country
GDP (PPP$ in billions)a
Human Development Index (out of 187)b
Total fertility rate (per woman 15–49 years)c
Infant deaths (per 1,000 live births)c
Ever use of contraception (%)c,d
Exclusive breast-feeding rank (of 23) and ratef
Ghana
74.8
135
4
50
50.4
4 (63.4 %)
Kenya
70.9
143
4.6
52
57.7
16 (31.9 %)
Liberia Malawi
1.8 14.6
182 171
5.2 5.7
71 66
35.3 65.2
18 (29.1 %) 2 (71.4 %)
Nigeria
414.5
156
5.7
75
28.6
23 (13.2 %)
Tanzania
73.5
152
5.4
51
28.8e
11 (49.9 %)
Zambia
21.9
164
6.2
70
63.1
6 (60.9 %)
5.9
173
4.1
57
58.5e
17 (31.8 %)
Zimbabwe
Ranks are 1 = highest to 23 = lowest for rates of exclusive breastfeeding of all infants less than 6 months old in the 23 Sub-Saharan African countries with a DHS completed since 2007. Rwanda ranks first, with a rate of 84.9 % a b c
GDP: Gross domestic product; PPP: Purchasing power parity; Source: CIA World Factbook (2011 estimates) Source: Human Development Reports, UNDP (2011) Source: Demographic and Health Surveys since 2007 [24–31]
d
Any use of modern or traditional forms of contraception
e
Only current contraceptive use is reported for Tanzania and Zimbabwe in the DHS online statistical database
f
Source: ICF International (2012). MEASURE DHS STATcompiler, http://www.statcompiler.com, February 6, 2013
were dropped because of missing data on outcomes or exposures (Ghana, n = 5; Kenya, n = 8; Liberia, n = 14; Malawi, n = 4; Nigeria, n = 164; Tanzania, n = 20; Zambia: n = 15). The final sample sizes were: Ghana, n = 173; Kenya, n = 449; Liberia, n = 313; Malawi, n = 397; Nigeria, n = 2007; Tanzania, n = 549; Zambia, n = 454; and Zimbabwe, n = 480. Measures Outcomes Two outcomes measured breastfeeding initiation soon after delivery and exclusive breastfeeding at the time of interview. To determine the timing of breastfeeding initiation, mothers were asked how long after the birth did they put their infant to the breast. A binary variable was coded 1 for initiation within 1 h and 0 for initiation after 1 h. To determine exclusive breastfeeding, mothers were asked about liquids or foods that their infant had consumed in the prior 24 h. A binary variable was coded 1 for exclusively breastfeeding (breastfed and no supplementary liquids or foods given) and 0 for not exclusively breastfeeding (any supplementary liquids or foods given) in this period. Exposures Three exposure variables measured whether (=1) or not (=0) the mother reported any lifetime emotional, physical, or sexual IPV victimization. The measure for emotional
IPV captured whether the mother’s (last) husband/partner had ever (a) humiliated her in front of others, (b) threatened to harm her or someone close to her, or (c) insulted her or made her feel bad about herself. The measure for any physical IPV captured whether the mother’s (last) husband/ partner had ever (a) pushed, shaken, or thrown something at her, (b) slapped her, (c) punched her with a fist or something that could hurt her, (d) kicked or dragged her, (e) tried to strangle or burn her on purpose, (f) threatened her with a knife, gun, or any other weapon, or (g) attacked her with a knife, gun, or any other weapon. The measure for sexual IPV captured whether the mother’s (last) husband/partner had ever (a) physically forced her to have unwanted sexual intercourse or (b) forced her to perform unwanted sexual acts. Covariates The multivariate analysis adjusted for ten child-, maternal-, and household-level covariates known to be associated with IPV [33–37] and breastfeeding practices [38–41] and so may confound the relationships of interest. Child attributes included his/her age in months. Maternal attributes (and household attributes, given one woman selected per household for the IPV module) included the mothers’ age at first union in years, age in years, total number of children ever born minus the index infant, lifetime perpetration of physical IPV, highest level of education attended (none [reference], primary, secondary and higher), partner’s highest level of education attended (none [reference],
123
Matern Child Health J
primary, secondary and higher), score for household wealth (derived from the first component of a principal components analysis of country-specific lists of household assets and amenities) [42], rural versus urban [reference] residence, and region of residence (Northern vs. Southern [reference]). Maps of DHS sampling districts were used to allocate districts evenly to Northern and Southern regions. Analysis The Stata/SE 12.0 software (StataCorp, College Park Station, TX, USA) was used for all analyses. Univariate analyses were conducted to explore the completeness and distributions of all variables. Logistic regression was used to estimate the relationships of maternal lifetime IPV victimization by type with breastfeeding outcomes, first unadjusted for other covariates then adjusted for other covariates. In all analyses, region-by-urban/rural residence population weights were used, and standard errors were adjusted for the stratified, multistage, cluster sample designs. Given the sample sizes (typically\500) and lower prevalences for some forms of IPV (\10 %), significant (p \ 0.05) and marginally significant (p \ 0.10) associations are discussed.
Results Demographic Characteristics and Levels of IPV Perpetration in the Samples Mother-infant dyads across the study countries differed on all demographic variables (p \ 0.001) except mean age of infant (p = 0.114) and region of residence (p = 0.176) (Table 2, exact p values available upon request). The mean ages of infants were 2.5–2.8 months. On average, mothers were in their first customary or legal union at a young age (17.8–19.5 years), were young at the time of interview (26.1–28.8 years), and were multiparous (1.7–3.2 births, excluding the index infant). The percentage of mothers who were in a union at interview differed widely (54.4–97.3 %), as did their educational attainments, with 51.8 % of Liberian mothers having had no schooling but 64.9 % of Zimbabwean mothers having had some secondary or higher schooling. Similarly, more than one-third (34.5 %) of Nigerian partners had had no schooling, but more than three-fourths (82.4 %) of Zimbabwean partners had had some secondary or higher schooling. Mothers tended to be living in rural areas and in households that occupied the three lowest wealth quintiles, relative to other households in their respective countries. Relatively few mothers reported having ever perpetrated any physical IPV (1.6–6.7 %).
123
Breastfeeding Practices and Maternal Reported IPV Victimization The breastfeeding practices of mothers also differed across the study countries (Table 2, p \ 0.001, exact p values available upon request). Nigerian mothers least often practiced early initiation (37.1 %) and exclusive breastfeeding (14.4 %); whereas, mothers in Malawi most often engaged in both practices (95.6 and 73.4 %, respectively). Mothers in the remaining countries were intermediate but still variable in their levels of early initiation (45.1–64.6 %) and exclusive breastfeeding (29.1–61.7 %). Reported lifetime IPV victimization also differed across countries (Table 2, p \ 0.001, exact p values available upon request). Rates of reported lifetime emotional IPV victimization ranged from 19.2 % in Zambian mothers to 35.6 % of Liberian mothers. Levels of reported lifetime physical IPV victimization also varied, from 14.4 % in Ghana to 43.9 % in Zambia. Finally, reported lifetime sexual IPV victimization was similarly uncommon in the West African countries (3.9–4.1 %) and more common in the five remaining countries (12.0–15.5 %). Associations of Early Breastfeeding Initiation and Maternal IPV Victimization The unadjusted odds of early breastfeeding initiation by maternal lifetime IPV victimization varied in direction across countries, but the significant associations were uniformly negative (Table 3). Compared to their counterparts, Kenyan mothers reporting emotional IPV victimization (uOR 0.60, p = 0.077), Zimbabwean mothers reporting physical IPV victimization (uOR 0.47, p = 0.003), and Zambian mothers reporting sexual IPV victimization (uOR 0.44, p = 0.017) had 40–53 % lower odds of early initiation. These and more relationships were significant with adjustment for confounding (Table 3). Compared to their counterparts, Kenyan and Tanzanian mothers reporting emotional IPV victimization (aOR 0.54–0.57, p B 0.088), Zimbabwean mothers reporting physical IPV victimization (aOR 0.40, p = 0.002), and Zambian mothers reporting sexual IPV victimization (aOR 0.42, p = 0.017) had 43–60 % lower adjusted odds of early initiation. However, mothers from Tanzania and Malawi reporting physical IPV victimization had over 200 % higher adjusted odds of early initiation than did mothers not reporting such victimization (aOR 2.11–9.00, p \ 0.05). Associations of Exclusive Breastfeeding and Maternal IPV Victimization The unadjusted odds of exclusive breastfeeding by maternal exposure to IPV varied across countries, but again, only
Matern Child Health J Table 2 Sample characteristics, mother-infant dyads in eight Sub-Saharan African Countries Region
Eastern Africa
Western Africa
Southern Africa
Country (n)
Kenya (449)
Tanzania (549)
Ghana (173)
Liberia (313)
Nigeria (2007)
Malawi (397)
Zambia (454)
Zimbabwe (480)
Initiated \1 h of birth
57.5
45.1
54.4
64.5
37.1
95.6
57.1
64.6
Exclusive breastfeeding last 24 h Reported IPV victimization (%)
35.5
51.6
61.7
29.1
14.4
73.4
59.0
36.6
Ever emotional
24.3
35.3
28.9
35.6
23.7
20.3
19.2
24.5
Ever physical
35.4
35.6
14.4
31.3
16.1
19.5
43.9
25.0
Ever sexual
12.3
14.0
3.9
4.1
4.1
12.0
15.5
14.3
2.7 (0.1)
2.6 (0.1)
2.8 (0.1)
2.6 (0.1)
2.8 (0.0)
2.5 (0.1)
2.6 (0.1)
2.8 (0.1)
18.5 (0.2)
18.1 (0.2)
19.5 (0.3)
17.8 (0.3)
17.9 (0.2)
17.5 (0.2)
17.8 (0.2)
18.8 (0.2)
Mean age in years (SE)
26.2 (0.4)
27.4 (0.5)
28.8 (0.5)
28.3 (0.5)
27.7 (0.2)
26.5 (0.5)
27.8 (0.3)
26.1 (0.3)
Mean number non-index births (SE)
2.5 (0.1)
2.8 (0.1)
2.3 (0.2)
3.0 (0.2)
2.9 (0.1)
2.6 (0.2)
3.2 (0.1)
1.7 (0.1)
Breastfeeding practices (%)
Child characteristics Mean age of index infant in months (SE) Maternal characteristics Mean age at first union in years (SE)
Union status (%) Married (legal or customary)
85.9
88.2
80.7
54.4
97.3
84.2
93.1
90.0
Living together
5.8
4.1
15.7
39.7
2.1
13.8
1.1
5.1
Divorced/widowed
8.3
7.7
3.6
5.9
0.7
2.1
5.9
4.8
None
15.2
27.2
27.2
51.8
43.7
13.6
13.2
1.0
Primary
65.1
68.5
27.6
35.3
22.5
67.0
63.6
34.1
19.7 3.4
4.3 1.6
45.2 6.0
13.0 4.3
33.8 2.3
19.4 2.6
23.2 6.7
64.9 2.7
None
11.4
18.2
21.9
26.6
34.5
9.8
6.8
0.8
Primary
55.2
73.4
9.6
26.9
20.6
59.8
50.9
16.8
Secondary/higher
33.4
8.4
68.6
46.5
44.9
30.4
42.2
82.4
Poorest
29.6
18.6
22.5
24.6
23.7
15.9
24.3
23.5
Poorer
22.6
26.2
17.3
29.3
21.6
22.3
24.0
25.8
Middle
17.6
20.8
20.4
17.8
18.3
25.7
22.0
16.3
Richer
15.7
19.8
21.2
19.4
17.6
16.1
17.3
19.3
Richest
14.6
14.6
18.5
8.9
18.7
20.1
12.4
15.2
Rural residence (%)
84.5
77.4
55.8
74.8
68.4
85.9
70.5
72.3
Northern region (%)
56.3
58.3
42.1
42.9
60.2
56.4
61.6
55.2
Education attended (%)
Secondary/higher Ever physical IPV perpetration Household characteristics Partner education attended (%)
Household wealth quintile (%)
Observed differences in attributes across countries are significant at p \ 0.001 except for the age of index infants in months (p = 0.114) and region of residence (p = 0.176)
negative relationships reached significance (Table 4). Compared to their counterparts, Kenyan and Liberian mothers reporting emotional IPV victimization (uOR 0.46–0.55, p B 0.088) and Kenyan mothers reporting physical IPV victimization (uOR 0.63, p = 0.066) had 37–54 % lower odds of exclusive breastfeeding. In Ghana,
Kenya, and Liberia, mothers reporting sexual IPV victimization had 66–86 % lower odds of exclusive breastfeeding than their counterparts (uOR 0.14–0.34, p \ 0.05). The results with adjustment for confounding were broadly similar (Table 4). Compared to their counterparts, Kenyan and Liberian mothers reporting emotional IPV
123
123 0.17
0.68
Ever sexual
0.59
0.84 4.15
SE
0.660
0.804 0.198
p value
0.57
0.98 9.00 0.68
0.67 7.95
SE
0.640
0.980 0.013
p value
0.44
0.85 0.73
OR
0.15
0.25 0.15
SE
0.84
0.017
0.589 0.133
p value
0.616
0.85
0.53
0.22
SE
0.42
1.29 0.81
OR
0.15
0.45 0.20
SE
Adjusted
1.25
1.31
0.64
OR
Adjusted
0.857
0.526
0.315
p value
0.36
0.78
0.19
SE
0.017
0.457 0.382
p value
0.743
0.494
0.195
p value
0.89
2.11
0.57
OR
Adjusted
0.34
0.18
0.14
SE
0.470
0.162
0.763
p value
1.00
0.89 0.47
OR
0.32
0.22 0.12
SE
Unadjusted
0.998
0.656 0.003
p value
Zimbabwe, n = 480
1.22
1.23
1.04
OR
Unadjusted
Nigeria, n = 2007
0.763
0.042
0.088
p value
0.31
0.20
0.15
SE
1.28
1.23 0.40
OR
0.47
0.36 0.12
SE
Adjusted
1.03
1.16
0.95
OR
Adjusted
0.510
0.485 0.002
p value
0.915
0.384
0.758
p value
Italicized values: p \ 0.10, bold values: p \ 0.05, adjusted for complex sample designs for each country
Unadjusted and adjusted models control for each type of IPV victimization, and adjusted models also control for age of infant (months), age at first union (years), age (years), number of nonindex births, ever physical IPV perpetration, own education attended, partner education attended, household wealth score relative to other households in country, rural residence, and region
1.19 3.92
OR
OR
Ever emotional Ever physical
Unadjusted
1.36
0.633
0.265
Adjusted
0.588
0.28
0.22
Unadjusted
0.58
0.85
0.70
Zambia, n = 454
0.58
0.799
0.871
Malawi, n = 397
0.520
0.74
0.52
Southern Africa
0.50
1.18
0.91
0.57
0.978
0.650
Ever sexual
0.45
0.35
0.36
0.30
0.18
SE
p value
0.99
SE
0.83
OR
Ever physical
SE
Ever emotional
SE
p value
0.93
OR
0.280
1.18
0.80
OR
0.29
0.711
0.050
Unadjusted
0.58
0.36
0.17
Adjusted
0.486
1.13
0.54
OR
Unadjusted p value
0.077 0.778
p value
Liberia, n = 313
0.35
0.26
SE
Unadjusted
Tanzania, n = 549
Ghana, n = 173
0.71
Ever sexual
Western Africa
0.60
0.92
Ever emotional
Ever physical
OR
OR
p value
Unadjusted
Type of IPV victimization SE
Adjusted
Kenya, n = 449
Eastern Africa
Table 3 Unadjusted and adjusted odds of early breastfeeding initiation by maternal lifetime IPV victimization, mother-infant dyads in eight Sub-Saharan African Countries
Matern Child Health J
0.96
0.60
Ever sexual
0.25
0.35
0.23
SE
0.226
0.913
0.271
p value
0.43
1.19
0.91 0.26
0.52
0.36
SE
0.158
0.696
0.802
p value
1.51
0.92
0.79
OR
0.46
0.21
0.20
SE
0.13
0.176
0.721
0.356
p value
0.025
0.10
0.53
0.20
SE
2.49
0.69
0.74
OR
1.01
0.20
0.25
SE
Adjusted
0.09
1.34
0.53
OR
Adjusted
0.832
0.513
0.421
p value
0.44
0.80
0.25
SE
0.025
0.199
0.382
p value
0.026
0.458
0.099
p value
1.29
1.76
0.56
OR
Adjusted
0.32
0.22
0.15
SE
0.547
0.651
0.279
p value
0.62
0.93
0.78
OR
0.20
0.22
0.21
SE
Unadjusted
0.131
0.773
0.347
p value
Zimbabwe, n = 480
0.78
1.09
0.82
OR
Unadjusted
Nigeria, n = 2007
0.458
0.212
0.200
p value
0.43
0.24
0.23
SE
0.73
1.54
0.65
OR
0.26
0.47
0.21
SE
Adjusted
0.91
0.96
0.98
OR
Adjusted
0.372
0.159
0.178
p value
0.834
0.873
0.940
p value
Italicized values: p \ 0.10, bold values: p \ 0.05, adjusted for complex sample designs for each country
Unadjusted and adjusted models control for each type of IPV victimization, and adjusted models also control for age of infant (months), age at first union (years), age (years), number of nonindex births, ever physical IPV perpetration, own education attended, partner education attended, household wealth score relative to other households in country, rural residence, and region
0.69
Ever physical
OR
OR
Ever emotional
Unadjusted
0.15
0.526
0.088
Adjusted
0.033
0.30
0.19
Unadjusted
0.14
0.79
0.55
Zambia, n = 454
0.17
0.871
0.575
Malawi, n = 397
0.029
0.63
0.37
Southern Africa
0.12
1.10
0.76
0.14
0.816
0.569
Ever sexual
0.40
0.90
0.32
0.31
0.28
0.23
SE
p value
0.80
SE
Ever physical
OR
Ever emotional
SE
p value
0.93
OR
SE
p value
0.085
1.17
0.79
OR
0.21
0.712
0.071
Unadjusted
0.34
0.34
0.19
Adjusted
0.028
0.86
0.48
OR
Unadjusted
0.2
0.015 0.066
p value
Liberia, n = 313
Ever sexual
0.2 0.16
SE
Unadjusted
Tanzania, n = 549
Ghana, n = 173
0.34
Ever physical
Western Africa
0.46
0.63
Ever emotional
OR
OR
p value
Unadjusted
Type of IPV victimization SE
Adjusted
Kenya, n = 449
Eastern Africa
Table 4 Unadjusted and adjusted odds of exclusive breastfeeding by maternal lifetime IPV victimization, mother-infant dyads in eight Sub-Saharan African Countries
Matern Child Health J
123
Matern Child Health J
victimization had 47–52 % lower adjusted odds of exclusive breastfeeding than their counterparts (aOR 0.48–0.53, p B 0.099), and Ghanan, Kenyan, and Liberian mothers reporting sexual IPV victimization had 66–91 % lower adjusted odds of exclusive breastfeeding (aOR 0.09–0.34, p B 0.085). In Kenyan mothers, the significant, negative unadjusted association of physical IPV victimization with exclusive breastfeeding became insignificant in adjusted models. In Zambia, mothers reporting sexual IPV victimization had 2.49 times the adjusted odds of exclusive breastfeeding than their counterparts (p = 0.025). Associations of Covariates in Adjusted Models In adjusted models for early initiation, each covariate was significantly (p \ 0.05) associated with the outcome in at least one country (results available upon request). The number of non-index births was negatively associated with early initiation in Malawi and Liberia (aOR 0.30–0.74) and positively associated with early initiation Kenya (aOR = 1.39). All other covariates were significantly associated with early initiation in only one or two countries. In adjusted models for exclusive breastfeeding, the age of the index infant was negatively associated with this outcome in all countries (aOR 0.38–0.76, p B 0.001). Other covariates were significantly associated with exclusive breastfeeding in three or fewer countries (results available upon request). Maternal history of physical IPV perpetration was positively associated with breastfeeding initiation (aOR 8.27, p = 0.008) and exclusivity (aOR 4.12, p = 0.037) in Kenya only, but these findings warrant caution because reported perpetration is rare (Table 2).
Discussion This study was the first to explore the association of maternal IPV victimization and breastfeeding practices in African women. Although the findings varied somewhat across the forms of IPV and eight African contexts, some general patterns are notable. First, maternal IPV victimization tended to be associated with suboptimal breastfeeding practices. In all countries except Malawi and Nigeria, at least one form of maternal IPV victimization was associated with suboptimal breastfeeding. Second, all forms of maternal IPV victimization were negatively associated with early or exclusive breastfeeding in at least one setting. Third, of all measured forms of IPV, sexual IPV victimization was most often negatively associated with suboptimal breastfeeding. Fourth, the finding that maternal lifetime IPV victimization is adversely associated with recent breastfeeding practices suggests that such violence may have long-term effects on mothers’ interactions with their children. Finally, the
123
analysis showed large, adjusted associations between IPV victimization and suboptimal breastfeeding. In Liberia, for example, maternal lifetime sexual IPV victimization was associated with 91 % lower odds of exclusive breastfeeding, and nearly all other associations reflected at least 50 % reductions in the odds of practicing early or exclusive breastfeeding. The magnitudes of these adjusted associations speak to their practical importance. In Sub-Saharan Africa, IPV victimization may substantially, adversely affect maternal breastfeeding practices. Surprisingly, no adjusted associations between maternal IPV victimization and breastfeeding were observed in Nigeria. Although these findings could be valid, they (and others in the analysis) could have resulted from some degree of differential misclassification either of maternal IPV victimization or breastfeeding practices. In the case of reporting IPV, the disclosure of victimization (and perpetration) may have been hampered by poor recall, misinterpretation of the question, or especially fear related to the consequences of disclosure [43]. Although multiple, behavioral questions on IPV victimization likely enhanced disclosure more than a single-question format [22, 44], IPV victimization still may be underreported. From this perspective, the estimated associations of maternal IPV victimization with breastfeeding practices may reflect lower bounds of the actual magnitude of effect. In the case of reported breastfeeding practices, self-reports of early initiation and exclusive breastfeeding are open to misclassification for reasons including poor recall and social desirability; yet, because misclassification can occur in both directions, its effects on associations with reported IPV victimization are uncertain. Also, ‘‘exclusive’’ breastfeeding was measured with respect to the prior 24 h, which may or may not reflect typical feeding patterns. The atypical findings for Malawi, Tanzania, and Zambia, that maternal physical or sexual IPV victimization were positively associated with early or exclusive breastfeeding, may be evidence of maternal compensatory responses in these contexts (see ‘‘Introduction’’ section). Qualitative research in victimized and non-victimized mothers in these countries would help to confirm this interpretation, and quantitative research across a wider range of settings would help to ascertain the frequency of compensatory breastfeeding in victimized mothers. Also, understanding the motivations of these mothers to breastfeed may be helpful in counseling all victimized mothers and facilitating a transition from deficit to compensatory parenting. Our analysis provides a foundation for further research. First, our study should be replicated with DHS from other regions to assess the generalizability of the findings across lower-income settings. Second, the DHS should include systematically questions about IPV victimization in specific (e.g., last) pregnancies to disentangle maternal lifetime
Matern Child Health J
victimization from joint maternal and fetal exposure to IPV. Third, the overall consistency of our cross-sectional associations across settings and forms of IPV victimization motivates the need for cross-cultural longitudinal research in cohorts of pregnant women to clarify the timing of IPV victimization around the index pregnancy, reciprocal influences [45–51], and important potential mediators, including the quantity and quality of prenatal care, maternal pre- and post-natal depression and substance abuse, birth outcomes, and newborn health [21]. Such longitudinal research may help to understand how the effects of IPV victimization may vary across contexts that themselves differ in the availability of high-quality prenatal care, popular beliefs about breastfeeding, and public health efforts to promote breastfeeding. These contextual differences may also help explain why our results varied across countries. The strengths of this analysis far outweigh its limitations. Unlike the few published studies discussed, ours included multiple national samples from poorer African countries, enabling us to identify cross-cultural patterns and countryspecific findings in probability samples of mothers with young infants in settings where early childhood morbidity and mortality remain high. Another strength was our ability to distinguish the associations of different types of IPV victimization with breastfeeding practices. In so doing, we discovered that the most common association across countries was that between sexual IPV victimization and suboptimal breastfeeding, especially less exclusive breastfeeding in Ghana, Kenya, and Liberia. As mentioned in the introduction, because breastfeeding involves the function of a sometimes sexualized organ, sexual IPV victimization may result in an acute aversion to early and exclusive breastfeeding [12]. Given that sexual IPV victimization was associated with suboptimal breastfeeding more often than the other forms of IPV victimization, experiencing sexual objectification and victimization may be a more powerful pathway of effect than those associated with emotional and physical IPV. In Africa and beyond, the cascading effects of IPV against mothers and associated suboptimal breastfeeding practices may be large [22]. In addition to their effects on infant mortality and morbidity [4–8], suboptimal breastfeeding [52] and sexual IPV against mothers [53] are associated with malnutrition in early childhood, suggesting that their negative effects may persist beyond infancy. Researchers should estimate the childhood morbidity and mortality attributable to suboptimal breastfeeding practices associated with IPV against mothers. Based on our findings, prenatal screening for IPV, with programs to address IPV and its after-effects in victimized women, are warranted to ensure women’s health and human rights and to mitigate cascading intergenerational effects.
Acknowledgments This paper is based on the thesis that EM completed under the direction of KY while EM was a Masters of Public Health student in the Hubert Department of Global Health, Rollins School of Public Health, Emory University. The authors gratefully acknowledge Ms. Trenise Stirrup for her assistance with the preparation of this paper and Dr. Sarah Zureick-Brown for her assistance to EM with the STATA programming. Any remaining errors are the responsibilities of the authors.
References 1. United Nations Children’s Fund/World Health Organization (2005). Baby friendly hospital initiative: Revised, updated, and expanded for integrated care. Geneva, Switzerland: United Nations Children’s Fund/World Health Organization. 2. United Nations Children’s Fund/World Health Organization (2003). Global strategy for infant and young child feeding. Geneva, Switzerland: United Nations Children’s Fund/World Health Organization. 3. Walker, A. (2010). Breast milk as the gold standard for protective nutrients. Journal of Pediatrics, 156(2 Suppl), S3–S7. 4. Edmond, K. M., Zandoh, C., Quigley, M. A., Amenga-Etego, S., Owusu-Agyei, S., & Kirkwood, B. R. (2006). Delayed breastfeeding initiation increases risk of neonatal mortality. Pediatrics, 117(3), e380–e386. 5. Edmond, K. M., Kirkwood, B. R., Amenga-Etego, S., OwusuAgyei, S., & Hurt, L. S. (2007). Effect of early infant feeding practices on infection-specific neonatal mortality: An investigation of the causal links with observational data from rural Ghana. American Journal of Clinical Nutrition, 86(4), 1126–1131. 6. Kramer, M. S., & Kakuma, R. (2004). The optimal duration of exclusive breastfeeding: A systematic review. Advances in Experimental Medicine and Biology, 554, 63–77. 7. Lauer, J. A., Betran, A. P., Barros, A. J. D., & de Onis, M. (2006). Deaths and years of life lost due to suboptimal breast-feeding among children in the developing world: A global ecological risk assessment. Public Health Nutrition, 9(6), 673–685. 8. Black, R. E., Allen, L. H., Bhutta, Z. A., et al. (2008). Maternal and child undernutrition: Global and regional exposures and health consequences. Lancet, 371(9608), 243–260. 9. Setty, V. (2006). Better breastfeeding, healthier lives. Population Reports, Series L, 1(14), 3–23. 10. Kendall-Tackett, K. A. (2007). Violence against women and the perinatal period: The impact of lifetime violence and abuse on pregnancy, postpartum, and breastfeeding. Trauma, Violence, & Abuse, 8(3), 344–353. 11. Saltzman, L. E., Fanslow, J. L., McMahon, P. M., & Shelley, G. A. (1999). Intimate partner violence surveillance uniform definitions and recommended data elements. Atlanta, GA: Centers for Disease Control and Prevention (CDC). 12. Klingelhafer, S. K. (2007). Sexual abuse and breastfeeding. Journal of Human Lactation, 23(2), 194–197. 13. Bailey, B. A., & Daugherty, R. A. (2007). Intimate partner violence during pregnancy: Incidence and associated health behaviors in a rural population. Maternal and Child Health Journal, 11(5), 495–503. 14. Yount, K. M., DiGirolamo, A., & Ramakrishnan, U. (2011). Impacts of domestic violence on child growth and nutrition: A conceptual review of the pathways of influence. Social Science and Medicine, 72(9), 1534–1554. 15. Levendosky, A. A., Huth-Bocks, A. C., Shapiro, D. L., & Semel, M. A. (2003). The impact of domestic violence on the maternalchild relationship and preschool-age children’s functioning. Journal of Family Psychology, 17(3), 275–287.
123
Matern Child Health J 16. Acheson, L. (1995). Family violence and breast-feeding. Archives of Family Medicine, 4(7), 650–652. 17. Lau, Y., & Chan, K. S. (2007). Influence of intimate partner violence during pregnancy and early postpartum depressive symptoms on breastfeeding among Chinese women in Hong Kong. Journal of Midwifery & Women’s Health, 52(2), e15–e20. 18. Lourenco, M. A., & Deslandes, S. F. (2008). Maternal care and breastfeeding experience of women suffering intimate partner violence. Revista de Saude Publica, 42(4), 615–621. 19. Silverman, J. G., Decker, M. R., Reed, E., & Raj, A. (2006). Intimate partner violence around the time of pregnancy: Association with breastfeeding behavior. Journal of Women’s Health, 15(8), 934–940. 20. Moraes, C. L., de Oliveira, A. S., Reichenheim, M. E., & Lobato, G. (2011). Severe physical violence between intimate partners during pregnancy: A risk factor for early cessation of exclusive breast-feeding. Public Health Nutrition, 14(12), 2148–2155. 21. Bullock, L. F. C., Libbus, M. K., & Sable, M. R. (2001). Battering and breastfeeding in a WIC population. The Canadian Journal of Nursing Research, 32(4), 43–56. 22. World Health Organization. (2005). WHO Multi-country study on Women’s Health and Domestic Violence against Women: Initial results on prevalence, health outcomes and women’s responses. Geneva, Switzerland: World Health Organization. 23. Strauss, M. A. (Ed.) (1979). Measuring intrafamily conflict and violence: The conflict tactics (CT) scales. Journal of Marriage and the Family, 41(1):75–88. 24. Ghana Statistical Service (GSS), Ghana Health Service (GHS), ICF Macro (2009). Ghana Demographic and Health Survey 2008. Accra, Ghana: GSS, GHS, ICF Macro. 25. Kenya National Bureau of Statistics (KNBS), ICF Macro (2010). Kenya Demographic and Health Survey 2008-09. Calverton, MD: KNBS, ICF Macro. 26. Liberia Institute of Statistics and Geo-Information Services (LISGIS) [Liberia], Ministry of Health and Social Welfare [Liberia], National AIDS Control Program [Liberia], Macro International, Inc. (2008). Liberia Demographic and Health Survey 2007. Monrovia: Liberia: LISGIS, Macro International, Inc. 27. National Statistical Office (NSO) and ICF Macro (2011). Malawi Demographic and Health Survey 2010. Zomba, Malawi, and Calverton, Maryland, USA: NSO and ICF Macro. 28. National Population Commission (NPC) [Nigeria], ICF Macro (2009). Nigeria Demographic and Health Survey 2008. Abuja, Nigeria: National Population Commission and ICF Macro. 29. National Bureau of Statistics (NBS) [Tanzania] and ICF Macro (2011). Tanzania Demographic and Health Survey 2010. Dar es Salaam, Tanzania: NBS and ICF Macro. 30. Central Statistical Office (CSO) [Zambia], Ministry of Health (MOH), Tropical Diseases Research Centre (TDRC), University of Zambia, Macro International Inc. (2009). Zambia Demographic and Health Survey 2007. Calverton, MD: CSO, Macro International Inc. 31. Zimbabwe National Statistics Agency (ZIMSTAT) and ICF International. (2012). Zimbabwe Demographic and Health Survey 2010–11. Calverton, Maryland: ZIMSTAT and ICF International Inc. 32. World Health Organization (WHO) Department of Gender, Women and Health (2001). Putting women first: Ethical and safety recommendations for research on domestic violence against women. Geneva, Switzerland: WHO Department of Gender, Women and Health. 33. Hindin, M. J., & Adair, L. S. (2002). Who’s at risk? Factors associated with intimate partner violence in the Philippines. Social Science and Medicine, 55(8), 1385–1399. 34. Vest, J. R., Catlin, T. K., Chen, J. J., & Brownson, R. C. (2002). Multistate analysis of factors associated with intimate partner
123
35.
36.
37.
38.
39.
40.
41.
42. 43.
44. 45. 46.
47.
48.
49. 50.
51.
52.
53.
violence. American Journal of Preventive Medicine, 22(3), 156–164. Ntaganira, J., Muula, A. S., Siziya, S., Stoskopf, C., & Rudatsikira, E. (2009). Factors associated with intimate partner violence among pregnant rural women in Rwanda. Rural and Remote Health, 9(3), 1153. Makayoto, L. A., Omolo, J., Kamweya, A. M., Harder, V. S., & Mutai, J. (2012). Prevalence and associated factors of intimate partner violence among pregnant women attending Kisumu District Hospital, Kenya. Maternal and Child Health Journal. Published Online First: 09 May 2012. doi:10.1007/s10995-0121015-x. Harrykissoon, S. D., Rickert, V. I., & Wiemann, C. M. (2002). Prevalence and patterns of intimate partner violence among adolescent mothers during the postpartum period. Archives of Pediatrics and Adolescent Medicine, 156(4), 325–330. Baxter, J., Cooklin, A. R., & Smith, J. (2009). Which mothers wean their babies prematurely from full breastfeeding? An Australian cohort study. Acta Paediatrica, 98(8), 1274–1277. Kehler, H. L., Chaput, K. H., & Tough, S. C. (2009). Risk factors for cessation of breastfeeding prior to six months postpartum among a community sample of women in Calgary, Alberta. Canadian Journal of Public Health, 100(5), 376–380. Petraro, P., Duggan, C., Msamanga, G., Peterson, K. E., Spiegelman, D., & Fawzi, W. (2011). Predictors of breastfeeding cessation among HIV-infected women in Dar es Salaam, Tanzania. Maternal & Child Nutrition, 7(3), 273–283. Demetrio, F., Pinto, E. D. J., & Assis, A. M. (2012). Factors associated with early breastfeeding cessation: A birth cohort study in two municipalities in the Reconcavo region, Bahio State, Brazil. Cadernos de Saude Publica, 28(4), 641–650. Rutstein, S. O., & Johnson, K. (2004). The DHS Wealth Index. Calverton, MD: ORC Macro. Ellsberg, M., Jansen, H. A., Heise, L., Watts, C. H., & GarciaMoreno, C. (2008). Intimate partner violence and women’s physical and mental health in the WHO multi-country study on women’s health and domestic violence: An observational study. Lancet, 371, 1165–1172. Kishor, S., & Johnson, K. (2004). Profiling domestic violence—a multi-country study. Calverton, MD: ORC Macro. Jasinski, J. L. (2004). Pregnancy and domestic violence: A review of the literature. Trauma, Violence, & Abuse, 5(1), 47–64. Taylor, R., & Nabors, E. L. (2009). Pink or blue…black and blue? Examining pregnancy as a predictor of intimate partner violence and femicide. Violence Against Women, 15(11), 1273–1293. Decker, M. R., Martin, S. L., & Moracco, K. E. (2004). Homicide risk factors among pregnant women abused by their partners. Violence Against Women, 10(5), 498–513. Jasinski, J. L. (2001). Pregnancy and violence against women: An analysis of longitudinal data. Journal of Interpersonal Violence, 16(7), 713–734. Burch, R. L., & Gallop, G. G., Jr. (2004). Pregnancy as a stimulus for domestic violence. Journal of Family Violence, 19, 243–247. Martin, S. L., Harris-Britt, A., Li, Y., Moracco, K. E., Kupper, L. L., & Campbell, J. C. (2004). Changes in intimate partner violence during pregnancy. Journal of Family Violence, 19, 201–210. Richardson, J., Coid, J., Petruckevitch, A., Chung, W. S., Moorey, S., & Feder, G. (2002). Identifying domestic violence: Cross sectional study in primary care. BMJ, 324(7332), 274. Kumar, D., Goel, N. K., Mittal, P. C., & Misra, P. (2006). Influence of infant-feeding practices on nutritional status of under-five children. Indian Journal of Pediatrics, 73(5), 417–421. Sobkoviak, R. M., Yount, K. M., & Halim, N. (2012). Domestic violence and child nutrition in Liberia. Social Science and Medicine, 74(2), 103–111.
