Limitations and strengths
The study’s limitations should be considered. Women
whose male partners were not surveyed were excluded
from the analysis given that the outcome could only be
determined from the DHS men’s survey. The couples’
survey design was also restricted to married and cohabiting
women, and male prenatal involvement was not ascertained
among women who did not attend ANC. Therefore, the
study’s findings may not be generalizable to all women of
reproductive age in the selected countries. Our secondary
analyses suggested that empowerment measures were
similar among women who attended ANC compared to
those who did not, with the exception of Zimbabwe and
Burkina Faso where women with higher empowerment
scores were less likely to attend ANC. Further qualitative
research would help to elucidate the relationship between
male involvement in pregnancy and women’s empowerment
among this cohort. It is also possible that ANC
accompaniment measures may have varied if asked of
women themselves rather than their male partners.
The study was also limited by the cross-sectional nature
of DHS data. The cross-sectional measures did not capture
the dynamism of empowerment or women’s or couple’s
prior or accumulated experiences over time. In addition,
causal associations between women’s empowerment and
male accompaniment to ANC cannot be inferred. It is
equally conceivable that women’s interpretation and
response to the empowerment questions varied across
settings, and some aspects of women’s empowerment
may not have been captured by the indicators assessed
in the DHS. Ultimately, this may have decreased the
predictive value of the construct. Finally, use of a summative
index has been challenged by some researchers
for discounting item-level distinctions. We considered
the summative index a more appropriate approach in
examining the broader role of women’s empowerment,
particularly given the concerns regarding the adequacy
of any single item in an African context. Despite these
limitations, the study’s strengths are the use of multiple
empowerment indicators and dimensions, use of geographically
diverse and representative samples, and inclusion of
measures compatible with the existing literature