Background
Perinatal mortality is a statistic that aggregates fetal
deaths, commonly known as stillbirths, and early neonatal deaths.
The emotional devastation of these events for women and families
is compounded by socio-cultural beliefs and practices obscuring
their existence in some societies and national health statistics.
Community health workers engaged in pregnancy tracking and vital
events documentation are in a position to capture these events if
they are trained in correct event classification and supported in
the data collection process.
In Malawi, the Institute for International Programs
supported community health workers, known as Health Surveillance
Assistants (HSAs), in the monthly documentation of pregnancies,
births, and deaths to estimate annual rates of under-five
mortality. We implemented a data quality documentation training
and data editor verification process to evaluate documentation of
vital events, namely adverse pregnancy outcomes. We also
implemented a cluster randomized mHealth intervention designed to
improve Real-time Mortality Monitoring HSA documentation of
pregnancies and pregnancy outcomes. We present the results of
these two data quality evaluations in this dissertation.
Methods and findings
To evaluate the documentation of adverse pregnancy
outcomes— induced abortions, miscarriages, stillbirths, and early
neonatal deaths— we compared HSA reported adverse pregnancy
iii
outcomes against results from the data editor verification
process. Classification of early pregnancy loss was poor in both
districts, despite improving slightly in the post-training
period. To evaluate the effectiveness of the mHealth intervention
in improving documentation of pregnancy outcomes, we matched
reported pregnancies with live births and adverse pregnancy
outcomes. The mHealth interventions improved documentation of
matched pregnancies in both treatment groups relative to the
baseline period, yet improvements were not noted between groups
during the intervention period.
Conclusions
Until a fully functional civil registration and vital
statistic system is implemented in Malawi, HSAs are a potential
source of perinatal mortality data but need substantial support
to meet high data quality standards. This level of support was
not achieved through the data quality training or mHealth
intervention. Further research should be conducted to identify
the most effective trainings and mHealth applications to support
community health worker documentation of pregnancies and
pregnancy outcomes, with a strong focus on adverse pregnancy
outcomes.