Problem Statement: Accurate data on coverage of health interventions is required for
continued reduction in child mortality. This study assessed the validity of the standard
indicator on care-seeking for child illness and methods for linking household and
provider data to improve measures of child health coverage.
Methods: The study was conducted in Southern Province, Zambia. Children <5 years
were given cards with barcodes. Healthcare providers tracked sick children brought for
care by scanning barcodes and distributing tokens. Provider preparedness to manage child
illness was assessed using a tool based on the Service Availability and Readiness
Assessment. We conducted a household survey on care-seeking for child illness in the
preceding two weeks. We compared maternal-reported and provider-documented careseeking
events. We linked household data on source of care with provider preparedness
data to estimate coverage using exact source care and measures of geographic proximity,
with data on all providers and only health facilities.
Results: Most children sought care from government facilities or community-based
agents (CBAs). We found high sensitivity (0.95, 95% CI: 0.88-0.98) and reasonable
specificity (0.74, 95% CI: 0.65-0.81) of maternal report of care-seeking for child illness.
Coverage of appropriate management of child illness, calculated using a measure based
on provider preparedness, was 55% (95% CI: 51%-58%) overall. Exact-match linking
was effective at this small scale. Most ecological linking methods produced similar
coverage estimates. Use of facility-only data reduced coverage estimates in the rural area
because CBAs were a common skilled source of care.
iii
Conclusions: Maternal report is a valid measure of source of care for child illness in
settings where utilization of public sector providers is high. With reliable household data
on source of care, exact-match linking may be a feasible method for producing more
informative estimates of coverage of appropriate management of child illness. Ecological
linking with data on a sample of all skilled providers may be as effective as exact-match
linking in areas with low variation in preparedness within a provider category or minimal
provider bypassing. More studies are needed at larger scale and a more diverse provider
landscape to further evaluate and guide linking methodology.