Matern Child Health J

DOI 10.1007/s10995-015-1817-8

The Influence of Women’s Empowerment on Child Immunization

Coverage in Low, Lower-Middle, and Upper-Middle Income
Countries: A Systematic Review of the Literature
Sara Thorpe1 • Kristin VanderEnde1 • Courtney Peters1 • Lauren Bardin1 •

Kathryn M. Yount1

Ó Springer Science+Business Media New York 2015

Abstract geographic location, and most studies focused on women’s

Objectives An estimated 1.5 million children under five decision making rather than freedom of movement. No
die annually from vaccine preventable diseases, and 17 % included study came from Latin America or the Middle East.
of these deaths can be averted with vaccination. Predictors Conclusions Overall, women’s agency, typically mea-
of immunization coverage, such as maternal schooling, are sured by decision-making, was positively associated with
well documented; yet, preventable under-five mortality the odds of complete childhood immunizations. Yet, the
persists. To understand these patterns, researchers are concept of agency was inconsistently defined and opera-
exploring the mother–child relationship through an tionalized. Future research should address these inconsis-
empowerment framework. This systematic review assesses tencies and focus on under-represented geographic regions
evidence of the relationship between women’s agency as a including Latin America and the Middle East.
component of empowerment and vaccine completion
among children \5 years in lower-income countries. Keywords Child immunization
Vaccine coverage

Methods We searched in Socindex, Pubmed, Web of Vaccine preventable diseases
Women’s agency

Science and Women’s Studies International for peer-re- Women’s empowerment
viewed articles focused on two measures of women’s
agency—decision-making and freedom of movement—and
child vaccination. Our initial search identified 406 articles Significance
and abstracts for screening; 12 studies met the inclusion
and exclusion criteria. As researchers and practitioners explore the relationship
Results A majority (83 %) of studies revealed at least one between gender-based approaches to improving child
positive association of measures for women’s agency with health outcomes including, vaccine preventable disease,
immunization coverage. These relationships varied by systematic evidence is required to identify feasible path-
ways of enhancing health.
& Sara Thorpe
thorpesara@[Link]
Introduction
Kristin VanderEnde
[Link]@[Link]
Despite the known cost effectiveness and health benefits of
Courtney Peters
courtneympeters@[Link]
immunization for children under five, mortality and mor-
bidity due to vaccine preventable diseases (VPDs) remains
Lauren Bardin
a concern [1]. As governments aim to reach Millennium
[Link]@[Link]
Development Goal 4, to reduce the risk of child mortality
Kathryn M. Yount
by two-thirds, understanding why an estimated 19.3 million
[Link]@[Link]
children under five remain unvaccinated is crucial, espe-
1
Emory University, Atlanta, GA, USA cially because vaccines can avert 2.5 million deaths in this

123
 Matern Child Health J

age group [2–4]. Since the World Health Organization’s a potential pathway to enhance children’s vaccination
(WHO) Expanded Program on Immunization (EPI) began coverage [18].
in 1974 to recommend routine vaccinations for tuberculo-
sis, diphtheria, pertussis, tetanus, poliomyelitis, measles, Conceptual Framework
and polio, a growing share of children have gained access
to these life-saving vaccines [2–4]. The number of vaccine Globally, the term women’s empowerment has been used
preventable deaths among children under five declined interchangeably with terms like women’s status, gender
from approximately 9.6 million in 2000–7.6 million in equality, and women’s autonomy [17]. Following Kabeer
2010 [3]. Still, disparities in coverage, particularly in sub- [18], we define women’s empowerment as a ‘‘process by
Saharan Africa and South Asia, lead to persistent under five which those who have been denied the ability to make
mortality and morbidity due to VPDs [1, 3]. strategic life choices acquire such an ability’’ [17]. Our
National, household-level, and individual-level deter- adapted framework describes women’s empowerment as
minants may explain persistent gaps in vaccination cover- the process by which women acquire enabling resources,
age [3–5]. Lower-income countries are less likely than exercise agency, and attain life achievements (Fig. 1) [17].
higher-income countries to have the political stability, Agency, therefore, as a component of women’s empow-
infrastructure, and services required for effective vaccine erment, refers to a woman’s ability to state her goals and to
delivery [3]. The focus on geographic differences and act upon them with motivation and purpose [17]. Thus,
barriers is shifting. Rapid urbanization increases the num- because enabling resources like schooling are necessary
ber of individuals living in slum dwellings, which elevates but insufficient conditions for women to exercise agency,
a child’s risk of exposure to under-immunized populations, our literature review focuses on agency [17] to understand
resulting in higher rates of under-five mortality [5]. it as a potential pathway to improved children’s immu-
Parental biases in the decision to vaccinate sons and nization status.
daughters are reported in lower-income countries, includ-
ing in India, Bangladesh, Nigeria and Ethiopia [6–9].
Household characteristics, such as assets and expenditures, Methods
also are cited as predictors of child immunization, as are
maternal characteristics like age, use of antenatal services, Search Strategy and Selection of Studies
location of delivery, schooling, and marital status [5, 6,
10]. Mothers with more schooling, lower parity, and higher In 2014, we searched Socindex, Pubmed, Web of Science,
household wealth are more likely to have timely vaccinated and Women’s Studies International databases with defined
children [1, 11–13]. Yet, many other attributes of the search terms (Table 1) and retrieved the titles and abstracts
mother–child relationship remain unaddressed theoretically of peer-reviewed, quantitative studies published in English
and empirically in this literature. between January 1, 1970 and September 1, 2013. This
Increasingly, researchers are adopting gender-based search yielded 409 unique titles and/or abstracts that met
approaches to understand health inequities, such as the initial screening criteria (Fig. 2). One researcher (ST)
incomplete vaccination coverage in children [14, 15]. screened these titles/abstracts, identifying 89 articles for
These researchers have focused on explanatory variables full-text review and application of a priori inclusion and
related to the underlying enabling resources of mothers, exclusion criteria (Table 2). Two researchers (ST and LB)
such as their schooling, employment, and income [16, 17]. pilot tested the inclusion and exclusion criteria on a subset
Based on the findings, researchers have advised that of articles, independently read and reviewed the 89 articles,
empowering mothers is a means to increase vaccination and resolved by consensus any discrepancies in opinion
coverage in children; yet, the empowerment pathways of regarding the relevance of each article for review. Fourteen
the ‘‘maternal resource-child vaccination’’ relationship are articles met the inclusion and exclusion criteria. Before
under-studied, including pathways capturing a mother’s data extraction, one researcher (ST) conducted a key author
‘‘agency’’ or capacity to influence and enact decisions that and reference list search, yielding another three articles to
may enhance the vaccination coverage of children. To which the inclusion and exclusion criteria were applied and
address this gap, this systematic review examines (1) how of which one article was retained, yielding a total of 15
researchers have defined and measured women’s articles. An adapted Cochrane Review data collection form
‘‘agency,’’ (2) how such measures are used in studies of was applied to capture general information on the study
children’s vaccination coverage, and (3) empirical patterns and detailed information on outcome and exposure vari-
of association. This review informs researchers and prac- ables [17]. During the data extraction, three articles were
titioners about the state of the evidence and next steps for excluded based on the inclusion and exclusion criteria
research and practice regarding women’s empowerment as (Table 2), for a total of 12 studies included in the data

123
Matern Child Health J

Fig. 1 Conceptual framework

for women’s empowerment [18]

Table 1 Search terms for

Women’s empowermenta Search terms Child immunization
identifying the associations
between women’s Women’s agency OR Child vaccine preventable disease OR
empowerment and childhood
immunization Women’s mobility OR Child immunization coverage OR
Women’s empowerment OR Child passive vaccination OR
Women’s autonomy OR AND Child immunization OR
Women’s decision making OR Child routine vaccination OR
Women’s freedom of movement OR Child immunization uptake
Gender equality OR
Women’s status
NOTb
HPV OR
HIV OR
Cervical cancer
a
Following Kabeer’s framework [18], women’s empowerment includes resources, agency and achieve-
ments. Because much of the work on empowerment is conceptual, we included terms, such as autonomy,
gender equality, and women’s status that have been used to describe analogous constructs [15]
b
As our focus was not on HPV, HIV, or cervical cancer, we excluded studies focused on these outcomes

analyses. This systematic review follows the PRISMA country study was included [28]. India [15, 19, 21, 24, 25]
guidelines, including adherence to all items listed in the and Nigeria [9, 12, 22] were the most frequently repre-
PRISMA checklist. sented countries in the review. The sample sizes ranged
from around 1000 to around 15,000 respondents. The
majority of authors used multivariate (MV) logistic
regression in their analyses [19–21, 26], with the remainder
Results using multilevel (ML), stepwise (SW), or bivariate
regression analysis. Only two studies included a theoretical
Characteristics of Included Studies framework that addressed both empowerment and immu-
nization coverage. The most common age range for
Of the included studies, all were published on or after 2005 reporting immunization outcomes was 12–23 months [9,
and were based on data collected between 1991 and 2009 21, 23–25, 28].
(Table 3). Most studies were secondary analyses of
Demographic and Health Surveys (DHS) [9, 19–28] con- Measurement of Agency
ducted in urban and rural settings in lower-middle income
countries [9, 15, 19, 21, 23–25, 28] and representative of In general, the authors varied in their definition of agency.
two major geographic regions: Africa [9, 23, 24, 27] and The most common terms were ‘‘autonomy’’ [9, 20, 21, 23,
South Asia [15, 19–21, 25, 26, 28]. One African multi- 24, 26] and ‘‘empowerment’’ [25, 27] instead of ‘‘agency,’’

123
 Matern Child Health J

Fig. 2 Article identification

procedure

*3 articles excluded during data extraction.

as defined by Kabeer [18]. For clarity, we categorized the to see friends, family, or relatives (Table 3). The most
items used to measure agency into three domains: decision common DM items were DM4 regarding major/large
making (DM), freedom of movement (FM), and other (OT) household purchases/goods [9, 23, 27, 28], DM6 regarding
(Table 4), with the latter category including items that did visits to friends, family, or relatives [9, 20–22, 27, 28], and
not reflect only the DM or FM categories. Most often, DM3 regarding healthcare in general or for her [20, 27, 28],
agency items pertained to: health decisions of the woman; either alone [20, 22] or in combination with other items in
financial control; permission to go to/visit health centers, other domains [9, 23, 27, 28].
friends and family; as well as both large and daily house- The authors rarely measured agency using only freedom-
hold purchases. of-movement items (n = 2, Table 4) [15, 19]. The authors of
Most authors used one of 13 DM items to measure that the two studies that included freedom-of-movement items
domain of women’s agency (Table 3) [9, 22, 23, 25, 26, analyzed DHS data from India [15, 19]. As such, we cannot
28]. Two items (DM2 and DM8) captured a woman’s generalize findings for FM because too few studies system-
ability to decide to go to a health facility, and one item atically included FM studies. Otherwise, authors opera-
(DM1) asked if the mother was ‘‘the main decision-maker tionalized freedom of movement with items reflecting
when the child is ill’’ (Table 4). Otherwise, the remaining decisions about movement rather than the action itself [9, 22,
DM items pertained to more general decisions about 25, 28]. The remaining authors combined items across all
household purchases, health, finances, and ability to travel categories [20, 21] or used only OT items [24].

123
Matern Child Health J

Table 2 Inclusion and exclusion criteria

Criteria Included Excluded Rationale

Sampling Population-based Clinic-based, convenience- The study aims to understand population-level

method based etc. health outcomes
Analysis At least bivariate Anything less than bivariate Included as a minimum in order to capture the
various ways that empowerment and
vaccinations are measured and
operationalized
Date January 1, 1970–September 1, 2013 Anything below or above This date range covers the period during
range which WHO established the expanded
program on immunization (in 1974)
Geographic Low income; lower-middle income; upper- High-income Focus of this review is on geographic
(based on middle income locations that bear a the high burden of
World Bank vaccine preventable diseases (Black 2003)
definitions)
Outcome Complete vaccination or at least one of the Influenza and/or HPV This review followed the WHO guidelines for
variable recommended vaccines (BCG, Hepatitis recommended vaccines for children under
B, Polio, DTP, haemophilus influenza five
Type B, pneumococcal, rotavirus,
measles, and rubella)
Exposure Decision making and/or freedom of Items that fall within the The term empowerment serves as the
variable movement resource category according umbrella term under which agency is
to the conceptual framework conceptualized. We defined women’s
(Fig. 1) decision making and freedom of movement
as domains of agency as expressed in the
conceptual framework (Fig. 1)
Language English All other languages unless Majority of research in this area published or
translation was provided translated into English; linguistic limitations
of the authors
Peer reviewed Peer reviewed Non-peer reviewed This criterion is reflective of the focus on the
highest-quality research examining the
association between women’s agency and
child vaccination
Population of Women with children less than the age of Women with adolescents, This serves as the topic of interest for this
interest five teenagers review and captures the time period of
interest between the exposure and outcome

Measurement of Immunization Coverage 25, 26, 28]. The authors’ use of vaccine measures varied
within countries and across studies that used the same DHS
Definitions of immunization coverage varied, even though data sources [9, 21, 24, 26]. For example, among the five
83 % of the authors indicated that they followed WHO studies in India, all studies ostensibly used full immu-
guidelines at the time of the study for recommended vac- nization measures, but the specific immunizations included
cines for children under five (Table 3) [9, 15, 19–21, 23, varied (Table 5) [15, 20, 21, 25, 26]. The most commonly
24, 26, 27]. The WHO recommends vaccines for tubercu- measured vaccines in the VAC and VACI categories were
losis, diphtheria, pertussis, tetanus, poliomyelitis, measles, DPT [9, 15, 19, 20, 22, 23, 25–28] and measles [9, 15, 19,
and polio [3]. For clarity, we organized immunization 20, 23–28].
measures into two categories: full immunization (VAC) The measurement of ‘‘full’’ or ‘‘complete’’ immuniza-
and individual immunization (VACI). The authors often tion, specifically for VAC items 1–6, also varied in their
used the terms ‘‘full’’ and ‘‘complete’’ interchangeably to consideration of timing and spacing metrics, which are
describe immunization, as defined by the WHO VAC Items important factors for vaccine effectiveness [14]. A majority
1–6 (Table 5). One author used DPT3 as a proxy for full of the authors ostensibly met the WHO guidelines, but in
immunization coverage [23]. We assume that VAC 6 only two studies using data from Nigeria and India,
(complete vaccination) incorporates the aforementioned respectively [9, 15], were timing and spacing of the vac-
recommended vaccines. The most common measures for cines considered in their measurements, and only studies
full immunization included a range of immunizations that used VAC1 and 3–5 included age matrices for receipt
covering DPT, polio, BCG and measles [9, 15, 19, 20, 23, of immunizations, ranging from birth to 23 months

123
 Matern Child Health J

Table 3 Characteristics of
Characteristic %a Author reference number
included studies (N = 12)
Year published
1970–2000 0 N/A
2000–2005 8 [15]
2006–2010 25 [19, 22, 23]
2011–September 2013 67 [9, 20, 21, 23, 25–28]
World Bank classification
Low income 17 [19, 27]
Lower-middle income 83 [9, 15, 20–26, 28]
Upper-middle income 0 N/A
Data source
DHS 92 [9, 19–28]
Human Development Profile Index 8 [15]
Empowerment framework
Empowerment only 33 [19, 20, 23, 25]
Immunization coverage only 8 [21]
Empowerment and immunization coverage 17 [9, 27]
No framework 42 [15, 22, 24, 26, 28]
Type of analysis
Bivariate logistic regression 17 [21, 24]
ML logistic regression 25 [9, 22, 28]
MV logistic regression 42 [19, 20, 23, 26, 27]
SW logistic regression 8 [25]
HLM 8 [15]
Country
Nigeria 25 [9, 22, 23]
India 42 [15, 20, 21, 24, 25]
Bangladesh 8 [19]
Nepal 8 [27]
Kenya 8 [26]
Multiple countries 8 [28]
Temporality of reporting
12–23 months 50 [9, 21, 23, 24, 27, 28]
12–35 months 25 [15, 22, 26]
12–59 months 17 [19, 25]
Not reported 8 [20]
Study location
Urban 8 [24]
Rural 17 [15, 19]
Urban and rural 67 [9, 20, 21, 23, 25–28]
Not indicated in study 8 [22]
Empowerment measure
Decision-making (DM items used) 58 [9, 22, 23, 25–28]
Freedom of movement (FM items used) 17 [15, 19]
Other (OT items used) 8 [24]
Multiple domains (OT, MB and DM items used) 17 [20, 21]
Vaccine measure
Full immunization (VAC items used) 83 [9, 15, 19–21, 23–25, 27,
28]
Full immunization and individual immunizations (VAC and VACI 8 [27]
items used)

123
Matern Child Health J

Table 3 continued
Characteristic %a Author reference number

Individual immunization (VACI) 8 [22]

Vaccine measures follow WHO guidelines
Yes 83 [9, 15, 19–21, 23, 24, 26–
28]
No 17 [22, 25]
Vaccine measure accounts for timing and spacing of vaccine in definition
Yes 17 [9, 15]
No 83 [19–27, 29]
Vaccine measurement tool
Household survey ONLY 8 [20]
Immunization card or mother’s recall 50 [9, 15, 22, 25–27]
Does not report method 42 [19, 21, 23, 24, 28]
(Adjusted) association between measure of agency and all immunization measures (N = 10)
Consistently positive relationship(s) 50 [9, 15, 19, 22, 25, 28]
Inconsistent relationship (s) 17 [20, 23]
No association 17 [26, 27]
(Unadjusted) association between measure of agency and all immunization measures (N = 2)
Consistently positive relationship(s) 50 [24]b
Inconsistent relationship(s) 0 N/A
No association 50 [21]
a
Due to rounding, percentages may not sum to 100 %
b
Due to referencing of the exposure and/or outcome these are positive relationship(s)

(Table 3). The sources of data on immunization varied, Associations with Decision-Making
with 50 % of authors reporting to use vaccines cards, and
in their absence, maternal recall (Table 3). The most commonly analyzed item was DM6 (n = 3,
decisions on visiting friends, family or relatives) [9, 22,
Associations between Measures of Women’s Agency 28]. Of the seven [10, 16, 20, 23, 25, 26, 29] studies that
and Childhood Immunization reported consistent relationships between agency and
immunization alone and not in combination with other
Overall Patterns exposure measures of women’s agency, most analyzed
items related to DM (n = 4) [10, 23, 26, 29]. Of studies
Among included studies, a majority (58 %) documented showing positive relationship(s) for at least one measure of
consistent relationships [9, 15, 19, 23, 25–27] and 42 % agency and immunization, four were based in South Asia
documented inconsistent or no relationship [20, 21, 23, 26, [15, 19, 24, 25] and three in Africa [9, 22, 28]. Four authors
27] between measures of agency and immunization cov- exclusively found a positive relationship between DM and
erage. However, among the studies with inconsistent immunization alone and not in combination with other
findings, the authors typically found at least one positive measures of agency [9, 22, 25, 28]. In sum, the results
relationship between a measure for maternal agency and suggest that low immunization often is associated with a
immunization. So, overall, the majority of authors showed lack of decision-making agency among mothers, mainly in
at least one positive relationship between a measure for India and Nigeria.
women’s agency and child immunization (n = 10). Two of Two studies included reports on vaccine completion
these authors found inconsistent relationships, one of before age 9 months and found significant positive rela-
which was not significant [20, 23]. Here, we define sig- tionships with women’s agency [9, 15]. Five authors
nificance as P B .05 and marginal significance as P B .10. reported vaccination on any child born in the past 5 years
In general, the larger the sample (n [ 3000), the more and reported significant positive relationships between at
often an association was seen between at least one measure least one measure of agency and immunization [9, 22–25,
of agency, particularly decision making, and child immu- 28]. Three of these studies were based in Nigeria [9, 22,
nization (n = 5) (Table 6) [15, 20, 23, 24, 26]. 23].

123
 Matern Child Health J

Table 4 Agency measures and response categories

Agency measurement items Response itemsa

Decision making (DM)

DM 1. The main decision maker when the child is ill RC 1. Dichotomous (Yes/No)
DM 2. Can you go to the local health center without seeking RC 2. Index for contribution to household decision making (0–12)
permission
DM 3. Who usually makes decisions about health care for you/ RC 3. Decisions alone or jointly = high No participation = low
decisions regarding healthcare
DM 4. Decisions on major/large household purchases/goods RC 4. Who makes the decision, Index ranging from (0–2)
DM 5. Decisions on daily household needs RC 5. Whether mother has money for own use (reference: has money for own
use)
DM 6. Decisions on visiting friends, family or relatives RC 6. Index categorized into low and high autonomy
DM 7. 12 Indicator index of contribution to decision-making RC 7. If woman made decisions or her opinion was included in decision in any
one item = 1 otherwise coded as 0
DM 8. Decisions to go to a health facility RC 8. Women that made all decisions either alone or jointly = high; not
involved in all four items = low
DM 9. Involvement in/decision on daily and major household RC 9. Self; jointly; others
decisions
DM 10. Decisions on how to spend husbands money/what to do RC 10. Unrestricted or restricted
with husbands money
DM 11. Purchasing of daily household goods
DM 12. How to spend money in the household
DM 13. Decisions on contraception
Freedom of movement
FM 1. Any HH female mobility
FM 2. Allowed to go market, health facilities and outside the
home/village/community
FM 3. Permission to go to the health center alone
Other
OT 1. Whether mother has money for own use or not
OT 2. Control over spending
a
Response categories reflect author ST’s categorization of the responses

The authors of the remaining studies, who found either varied by the venue to which a woman traveled and the
inconsistent relationships or no association between DM measurement scale for freedom of movement. Across the
and immunization outcomes, used different DM items than two studies whose authors analyzed freedom of movement
those reported above [20, 21, 23, 24, 27]. These authors alone, one found a significant relationship between ‘‘any
included items pertaining to the woman’s role in household HH female mobility’’ (FM1, Table 4) with immunization,
finances, which led to inconsistent findings or no associa- and the other found a significant relationship with ‘‘per-
tion [20, 23, 27]. One Nepali study combined DM items mission to go to the health center alone’’ (FM3, Table 4)
related to finances, health, and freedom of movement and with immunization [15, 21]. One of the studies in India did
reported a negative relationship between agency and all 8 not show a bivariate (unadjusted) association, so the
vaccines [27]; whereas, one Nigerian study found signifi- authors did not apply a regression analysis [21]. This study
cant associations of items related to household DM (DM characterized autonomy through freedom of movement,
3,4,11, Table 4) and vaccination but not financial DM (DM being ‘‘allowed to go to market, health facilities and out-
10, decisions on how to spend husband’s money/what to do side the home/village/community’’ (FM 2, Table 4) and
with husband’s money) and vaccination [23]. immunization, along with DM 6, 8 and 9 [21]. Two authors
operationalized freedom of movement with items that other
Associations with Freedom of Movement authors used to measure decision making and found a
significant positive relationship between FM and immu-
The authors of all studies measuring FM, whether alone or nization [20, 25]. In our analysis, these studies were
with other agency exposures, did not find consistent rela- grouped with those related to decision making. In general,
tionships between FM and immunization. The relationships the studies showing a positive relationship between

123
Matern Child Health J

Table 5 Immunization measures and response categories

Measurement item Response item

Full immunization measures

VAC 1. 3 doses of diphtheria, pertussis and tetanus (DPT), 1 dose each RC 1. Dichotomous (Yes/No)
of BCG and measles vaccine before 12 months of age
VAC 2. 1 dose of BCG, 3 doses of DPT,3 doses of Polio vaccine, and 1 RC 2. Dichotomous (Full = 1/otherwise = 0)
dose of measles vaccine
VAC 3. Children aged 12–23 months who received 1 dose each of RC 3. Dichotomous (1 dose = 1/otherwise = 0)
BCG and measles, and 3 doses each of DPT and polio vaccine
VAC 4. 3 doses of oral polio vaccine, 3 doses of diphtheria, pertussis, RC 4. Dichotomous (3 doses = 1/otherwise = 0)
and tetanus, one dose each of BCG and measles vaccine before
12 months of age
VAC 5. 1 BCG vaccine at birth, 3 doses each of DPT and oral polio at RC 5. Dichotomous (1 = received DPT3/otherwise = 0)
6, 10, and 14 weeks of age, and finally, 1 measles vaccine at
9 months or soon thereafter
VAC 6. Complete Vaccination RC 6. None = if the child did not receive any immunizations at the
time of the survey; Some = if the child received at least one but not
all eight immunizations; All = if the child received all eight
immunizations -1 BCG, 3 DPT, 3 polio, 1 measles
Individual immunization
VACI 7. Measles immunization
VACI 8. Full series DPT3
VACI 9. DPT immunization
VACI 10. Polio immunization

freedom of movement and immunization were based in Data Quality

India [15, 20, 25]. However, only two of these studies
actually found a positive relationship between DM and In general, the authors of included articles adequately and
immunization due to the authors’ definition of freedom of appropriately reported their results. Often, the authors did
movement. The inconsistent findings outside of India not fully explain the study design and methods. Moreover,
suggest a need for more research on the relevance of the authors often were vague and brief in their descriptions of
measures for maternal freedom-of-movement and child the measures for women’s agency and the sources of data on
vaccination. immunization, which may affect the estimated associations
between these measures. As a methodological strength, most
Findings by Measure of Immunization included studies (n = 11) captured the temporality of the
mother’s reporting on the immunization of her child, which
Among the authors that reported consistently positive helps to explain recall bias [9, 15, 19, 21–28].
relationships between agency and immunization, five used
full immunization items, VAC1 [24], VAC4 [27] VAC6
[24], and VAC5 [9, 15] and one used the full series of Discussion and Recommendations
DPT3 VAC8 as a proxy for full immunization [22]. The
only study that included multiple vaccine measurements This systematic review is the first to investigate the rela-
found no association between agency and any immuniza- tionship between women’s agency and childhood immu-
tion measure [28]. The authors that analyzed vaccine nization in lower-income settings. Based on the included
measures incorporating timing and spacing of immuniza- studies, which represented countries in South Asia and
tions found a positive relationship with agency in terms of Africa, we observed the general pattern that higher agency
either DM or FM [9, 15]. Thus, despite some inconsis- among mothers was associated with higher odds of child-
tencies within and across countries in definitions of, mea- hood immunizations. This pattern of association was most
surement of, and relationship with women’s agency, we apparent when women’s agency was measured by items
generally found that women’s agency was positively capturing their ability to make decisions and immunization
associated with complete immunization of children in the was measured by items reflecting complete immunization.
countries represented. This pattern corroborates one from eight African countries

123
 Table 6 The associations between women’s agency and child immunization (n = 12)
Article Data source Sample Agency Agency Agency measurement Vaccine Temporality Type of Outcomes Summary
size definition measure instrument measure and of vaccine analysis of

123
and response reporting relationship
response items
items

Decision making (DM 1–13)

22 Nigeria DHS (2003) 1472 Mother DM 6 Summative continuous VACI 8 RC 5 Any child MV logistic Conjugal DM Positive
conjugal RC2 index (0–12, vague) born in regression power range aOR
power past 1.22*
5 years
20 Nigeria DHS (2008) 3725 Decision- DM 10, Dichotomous: If VAC 5 RC 1 All children ML logistic Square score for Positive
making 11, 6, responded alone or born since regression conjugal DM
autonomy 4; RC1 with husband to one 2003 power range aOR
or several = yes if (5 years) .97*
responded with other
person to all = no
26 Kenya DHS (2003) 2169 Autonomy DM 12 Summative continuous VAC 2 RC 2 Women who MV logistic DM Autonomy Consistent
and 13 index 0 = no gave birth regression aOR.76 SE (.11)
RC 4 autonomy 2 = high in past
autonomy 35 months
23 Nigeria DHS (2008) 3250 Autonomy DM 3, Categorical autonomy VAC 1 RC 1 Any child MV logistic Household DM Inconsistent
10, 11, index born in regression (ref: low) aOR
4 RC 3 past 1.64 CI
5 years (1.25–2.14)*
Financial DM (ref:
low) aOR .98 CI
(.76–1.27)
25 India Human Development 5287 Empowerment DM 1 Dichotomous VAC 1 RC 1 Most recent SW Logistic Visit Health Center Positive
Survey (2004–2005) and 2 birth Regression aOR 1.25 SE
RC 1 (.103)**d
Decision making
aOR 1.21 SE
(.090)f**
27 Nepal DHS (2006) 1056 Empowerment DM 3, 4, Unidimensional VAC 2 RC 2 First child MV Logistic Financial, Health, Inconsistent
5, 6 RC dichotomous VACI7 RC3 born Regression and Mobility
7 Decision making
VACI 9 RC 4 All 8 Vaccines
VAC I10 RC aOR .78
4
3 Doses DPT aOR
.76
3 Doses Polio aOR
.76
Matern Child Health J
 Table 6 continued
Article Data source Sample Agency Agency Agency measurement Vaccine Temporality Type of Outcomes Summary
size definition measure instrument measure and of vaccine analysis of
and response reporting relationship
response items
items
Matern Child Health J

1 Dose Measles
aOR .91
28 DHS; Democratic Republic 14,150 Gender equality DM 3, Decision making VAC 4 RC 1 Any child ML Logistic Household DM Consistent
of Congo (2008); Ghana (Pooled 11, 6, 4 summative born in Regression (ref: low) aOR
(2008); Liberia (2007); Data) RC 8 categorical household past 1.31 CI (.92,
Mali (2006); Nigeria decision making 5 years 1.87)
(2009);Uganda (2006); (High/Low)
Zambia (2007)
Freedom of Movement (FM 1–3)
19 Bangladesh DHS (2004) 3530 Mobility FM 3 RC Unidimensional VAC 2 RC 1 Last child MV Logistic Permission to go to Positive
characteristics 10 Dichotomous born Regression Health Center
Alone (ref:
unrestricted)
aOR .921 CI
(.736–1.51)
15 India Human Development 5623 Mobility of FM 1 RC Dichotomous freedom VAC 5 RC 6 Most recent HLM Any female HH Positive
Profile Index (1994) and household 1 to move outside the child who mobility aOR .17
Indian Census (1991) women home (no permission is alive SE (.06)* e
needed = 1; not
allowed to go
outside = 0)
Other (OT 1–2)
24 India DHS-NFHS-3 1527 Mother’s OT 1 RC Dichotomous: mother’s VAC 6 RC 1 Women who Bivariate Mother’s Positive
(2005–2006) autonomy 1 lack of autonomy gave birth logistic autonomy (ref:
reverse coding in the past regression has money for
5 years own use) OR
.6267 (.4084–
.9615) (ref: ever
vaccinated)c
Combination DM, FM, and OT items
21 India DHS-NFHS-3 Survey 1607 Autonomy DM6,8,9,10; FM2 RC 6 Categorical VAC 3 RC 1 Bivariate Autonomy: Chi- Inconsistent
(2005–2006) autonomy Analysis square (1.65)
index Most recent Low: 40.56
(high/low) birth of High: 43.81 b
women who
had teen
pregnancies

123
 Table 6 continued
Article Data source Sample Agency Agency Agency measurement Vaccine Temporality Type of Outcomes Summary
size definition measure instrument measure and of vaccine analysis of

123
and response reporting relationship
response items
items

20 India DHS NFHS—3f (not Only Autonomy OT 2 RC 9; DM 3 RC 9; DM 13 Principal VAC 2 RC 2 MV logistic OT control over Inconsistent
reported) entire RC 9; DM 6 RC 9; DM 9 RC 9 component regression Spending (ref:
NFHS-3 analysis Does not Self) aOR Jointly
sample report .974, aOR.764
reported Others
DM on own
healthcare (ref:
Self) aOR .795
Jointly, aOR
1.013 Others;
DM on large
household
purchases (ref:
Self) aOR .887
Jointly, .754
Others
DM on daily
household
purchases (ref:
Self) aOR .682
Jointly, aOR.972
Other
Mobility DM (ref:
self) aOR1.695*
Jointly, aOR
1.261 Others
NS not significant
a
Same data source
b
All the variables identified as significant in the bivariate analyses using the Chi-square test were included in the binary logistic regression model
c
Positive association due to reference coding even though alpha not reported
d
Transformed beta coefficients provided by author
e
Non-transformed
f
Date not reported in study, secondary source indicates data was collected from 2005 to 2006
* P \ .01
** P \ .0324
Matern Child Health J
Matern Child Health J

showing a significant, positive relationship between A third limitation, which highlights a key area for
women’s decision-making agency and the adjusted odds research, involves the content areas of DM and FM items.
that a child under two with acute respiratory infection visits In theory [17], the concept of women’s agency involves
a health facility (aOR 1.31 CI 1.12, 1.54) [29]. In general, decisions in an array of family domains, including those
fewer studies included measures for women’s freedom of historically reserved for men, and also involves the free-
movement, and when such measures were included, their dom to visit or travel to a range of public venues. Yet, in
associations with child immunization were inconsistent. the studies included in our review, only one DM item
Thus, our systematic review suggests, for lower-income captured decisions specifically related to children’s health,
settings, that specific dimensions of women’s agency may which may be more strongly associated with the outcome
enhance vaccination coverage for children, and that of interest. Indeed, the authors of one study in India
empowering women in such settings shows promise as a included ‘‘The main decision maker when the child is ill’’
means to improve child health. in there measure of DM1 and found a significantly positive
Our review also highlights important limitations in the relationship between decision making and immunization
literature reviewed that inform key recommendations for (Table 4) [25]. Moreover, the DM items used to measure
future research. First, a majority of included studies analyzed women’s agency in this review captured decisions that
DHS data, revealing a heavy reliance on this source for historically have been relegated to women, such as cooking
information on women’s empowerment. The DHS is consid- and small household purchases. Such items may not reflect
ered a gold standard for cross-national comparison, yet mea- agency, as Kabeer has defined, in that they do not capture
sures of agency still differ in the DHS, both across countries decisions historically reserved for men. Thus, measures of
and within countries over time [29]. Context-specific mea- women’s agency that may be more relevant for children’s
sures of agency have the advantage of capturing locally rel- health outcomes may include women’s capacity to make
evant domains and manifestations of agency [30], but they health-related and major financial decisions and to travel
complicate our ability to generalize more broadly about the unaccompanied to associated venues, such as clinics, hos-
relationship of women’s agency to child immunizations, and pitals, pharmacies, and local markets for food.
ultimately child health. Efforts to develop scales for women’s The measurement of vaccination coverage is a fourth
agency that include comparable and context-specific items in limitation. In half of the studies, immunizations were mea-
multiple domains would advance both comparative and con- sured first by vaccine cards, and in the absence of a complete
text-specific studies of these relationships. vaccine card, mother’s recall, which sometimes spanned
Second, the limited use of measures to capture women’s 5 years. A heavy reliance on maternal reports of child vac-
freedom of movement as a domain of their agency also cinations over several years may lead to systematic error in
informs our recommendations. The authors who measured the reported occurrence and timing of vaccinations. The two
FM alone showed a positive relationship [15, 20], and studies showing statistically significant relationships
those who measured women’s freedom of movement as a between included measures of agency and immunization
component of decision making found positive relationships before the age of 9-month could reveal the utility of shorter
[20, 25]. While both sets of findings are suggestive, the recall windows in limiting recall bias [9, 15]. Only two
studies had limited geographic scope, cautioning against authors measured the timing and frequency of vaccinations
broader generalizations at this time. Contemporary [9, 15], which is important because as evidence suggests,
research on women’s agency suggests that women’s free- repeated vaccinations decrease the effectiveness of vaccines
dom of movement remains a salient dimension of agency in [9, 15, 28]. Improvements in such data may better inform
particular contexts, such as the Middle East [30] and South global strategies and efforts that aim to address gaps in cov-
Asia [31]. In such settings, researchers should systemati- erage such as the Global Vaccine Action Plan and EPI [2–4].
cally measure and include women’s freedom of movement A fifth limitation concerns the cross-sectional design of
as one of multiple domains of agency in studies of child most of the included studies. In such studies, agency is
health. Such efforts may mitigate potential bias in the measured with respect to the date of interview; whereas,
estimated associations of women’s decision-making with information on vaccinations refers to events that occurred
child immunization. In addition, these efforts would some months or years before the interview. Thus, the
improve an understanding of the pathways by which estimates from such studies are purely correlational, and
women’s enabling resources may enhance child health appropriate temporal ordering is needed to establish the
through the multiple domains of women’s agency. Suc- causal direction from women’s agency to child immu-
cessful efforts like those outlined above would require nization. Thus, longitudinal studies—ideally embedded in
clarity regarding the definition of freedom of movement, its randomized women’s empowerment interventions—are
comparable and context-specific component items, and needed to test the ‘‘impact’’ of increases in women’s
scales derived from these items. agency on child vaccination coverage.

123
 Matern Child Health J

Finally, because all included studies in this review came 6. Mutua, M. K., Kimani-Murage, E., & Ettarh, R. R. (2011).
from South Asian and African countries, future research Childhood vaccination in informal urban settlements in Nairobi,
Kenya: Who gets vaccinated? BMC Public Health, 11, 6.
should include under-researched settings where vaccination 7. Corsi, D. J., Bassani, D. G., Kumar, R., Awasthi, S., et al. (2009).
in childhood remains problematically low. This recom- Gender inequity and age-appropriate immunization coverage in
mendation is especially important, as 70 % of the 22.3 mil- India from 1992 to 2006. BMC International Health and Human
lion children who did not receive DTP3 in 2011 lived in the Rights, 9(1), S3.
8. Chowdhury, A. M., Bhuiya, A., Mahmud, S., Abdus Salam, A.
Democratic Republic of Congo, Ethiopia, India, Indonesia, K., et al. (2003). Immunization divide: Who do get vaccinated in
Iraq, Nigeria, Pakistan, Philippines, Uganda, and South Bangladesh? Journal of Health, Population and Nutrition, 21(3),
Africa at the time of survey, and only half (n = 6) of these 193–204.
countries were represented in this review [1]. 9. Antai, D. (2012). Gender inequities, relationship power, and
childhood immunization uptake in Nigeria: A population-based
In sum, this review assessed the strength of existing cross-sectional study. International Journal of Infectious Dis-
evidence on the association of two dimensions of women’s eases, 16(2), e136–e145.
agency—freedom of movement and decision making— 10. Fatiregun, A. A., & Okoro, A. O. (2012). Maternal determinants
with child immunization. With growing interest among of complete child immunization among children aged
12–23 months in a southern district of Nigeria. Vaccine, 30(4),
funders—such as the Gates Foundation and UK Depart- 730–736.
ment for International Development—on the impact of 11. Babirye, J. N., Rutebemberwa, E., Kiguli, J., Wamani, H., et al.
women’s empowerment on maternal and child health, this (2011). More support for mothers: A qualitative study on factors
review is timely and offers a baseline of evidence on which affecting immunisation behaviour in Kampala, Uganda. BMC
Public Health, 11, 723.
researchers may build to improve an understanding of these 12. Fadnes, L. T., Nankabirwa, V., Sommerfelt, H., Tylleskär, T.,
relationships. Our findings suggest that women’s capacity et al. (2011). Is vaccination coverage a good indicator of age-
to make family decisions is positively related to child appropriate vaccination? A prospective study from Uganda.
immunization in Nigeria and India, the most populous Vaccine, 29(19), 3564–3570.
13. Pande, R. P., & Yazbeck, A. S. (2003). What’s in a country
countries in Africa and South Asia, respectively, and where average? Wealth, gender, and regional inequalities in immu-
child immunization remains low. Longitudinal research nization in India. Social Science and Medicine, 57(11),
across diverse contexts would confirm the external validity 2075–2088.
of our findings. As countries strive to reduce under-five 14. Jamil, K., Buhiya, A., Streatfield, K., & Chakrabarty, N. (2005). The
immunization program in Bangladesh: Impressive gains in coverage,
mortality and morbidity and to meet MDG 4, our findings but gaps remain. Health Policy and Planning, 1(14), 49–58.
suggest that programs to empower women should not be 15. Parashar, S. (2005). Moving beyond the mother–child dyad:
ruled out as a strategy to improve child health. Women’s education, child immunization, and the importance of
context in rural India. Social Science and Medicine, 61(5),
Acknowledgments This paper was developed in collaboration with 989–1000.
a group of mentors and colleagues who the principal researcher would 16. Topuzoğlu, A., Ay, P., Hidiroglu, S., & Gurbuz, Y. (2007). The
like to acknowledge and thank for their contributions. The authors barriers against childhood immunizations: A qualitative research
acknowledge Emory University, Rollins School of Public Health, and among socio-economically disadvantaged mothers. European
the Hubert Department of Global Health for its continuous support in Journal of Public Health, 17(4), 348–352.
the advancement of global health research and public health profes- 17. Kabeer, N. (1999). Resources, agency, achievements: Reflections
sionals. The views expressed in this article do not necessarily reflect on the measurement of women’s empowerment. Development
those of Emory University Rollins School of Public Health. and Change, 20(3), 435–464.
18. Moher D, Liberati A, Tetzlaff J, Altman DG, The PRISMA
Group. (2009). Preferred reporting items for systematic reviews
and meta-analyses: The PRISMA statement. PLoS Medicine,
References 6(6), e1000097. doi:10.1371/journal.pmed1000097.
19. Rahman, M., & Obaida-Nasrin, S. (2010). Factors affecting
1. Harris, J., Gacic-Dobo, M., Eggers, R., Brown, D., & Sodha, S. acceptance of complete immunization coverage of children under
(2014). Global immunization data (pp. 1–4). Geneva: WHO. five years in rural Bangladesh. Salud Publica de Mexico, 52(2),
2. Harahap, J. (2000). Factors affecting childhood immunization in 134–140.
North Sumatra Province, Indonesia (pp. 1–63). Master of Arts 20. Mahapatro, S. R. (2012). Utilization of Maternal and child health
Thesis, Mahidol University. care services in India: Does women’s autonomy matter? The
3. World Health Organization (WHO). (2011). Global Vaccine Journal of Family Welfare, 58(1), 22–33.
Action Plan 2011–2020 (NLM classification: WA 115). Geneva: 21. Singh, L., Rai, R. K., & Singh, P. K. (2011). Assessing the uti-
WHO Press. [Link] lization of maternal and child health care among married ado-
cine_action_plan/GVAP_doc_2011_2020/en/. lescent women: Evidence from India. Journal of Biosocial
4. Burton, A., Monasch, R., Lautenbach, B., Gacic-Dobo, M., et al. Science, 44(1), 1–26.
(2009). WHO and UNICEF estimates of national infant immu- 22. Babalola, S. (2009). Determinants of the uptake of the full dose of
nization coverage: Methods and processes. Bulletin of the World diphtheria–pertussis–tetanus vaccines (DPT3) in Northern Nige-
Health Organization, 87, 535–541. ria: A multilevel analysis. Maternal and Child Health Journal,
5. Kamau, N., & Esamai, F. O. (2001). Determinants of immu- 13(4), 550–558.
nization coverage among children in Mathare Valley, Nairobi. 23. Singh, K., Haney, E., & Olorunsaiye, C. (2012). Maternal
East African Medical Journal, 78(11), 590–594. autonomy and attitudes towards gender norms: Associations with

123
Matern Child Health J

Childhood immunization in Nigeria. Maternal and Child Health 28. Singh, K., Bloom, S., & Brodish, P. (2013). Gender equality as a
Journal, 17(5), 837–841. means to improve maternal and child health in Africa. Health
24. Agarwal, S., & Srivastava, A. (2009). Social determinants of Care for Women International, 36(1), 57–69.
children’s health in urban areas in India. Journal of Health Care 29. Upadhyay, U. D. & Karasek, D. (2010). Women‘s empowerment
for the Poor and Underserved, 20(4 Suppl), 68–89. and achievement of desired fertility in Sub-Saharan Africa (DHS
25. Vikram, K., Vanneman, R., & Desai, S. (2012). Linkages Working Papers No. 80). Calverton, Maryland, USA: ICF Macro.
between maternal education and childhood immunization in 30. Yount, K. M., VanderEnde, K., Dodell, S., & Cheong, Y. F.
India. Social Science and Medicine, 75(2), 331–339. (2015). Measurement of women’s agency in Egypt: A National
26. Abuya, B. A., Onsomu, E. O., Kimani, J. K., & Moore, D. (2011). Validation Study. Social Indicators Research. doi:10.1007/
Influence of maternal education on child immunization and s11205-015-1074-7.
stunting in Kenya. Maternal and Child Health Journal, 15(8), 31. Agarwala, Rina, & Lynch, Scott M. (2006). Refining the mea-
1389–1399. surement of women’s autonomy: An international application of a
27. Pandey, S., & Lee, H. N. (2012). Determinants of child immu- multi-dimensional construct. Social Forces, 84(4), 2077–2098.
nization in Nepal: The role of women’s empowerment. Health
Education Journal, 71(6), 642–653.

123