Maternal and Neonatal Health Intervention Coverage and Measurement in Urban Kenya

Thumbnail Image


StOngeAhmad_umd_0117E_23342.pdf (1.66 MB)
No. of downloads:

Publication or External Link





High coverage of recommended maternal and neonatal health (MNH) interventions can reduce neonatal mortality in low- and middle-income countries. In Kenya, however, despite overall gains in MNH intervention coverage and neonatal survival, progress in urban areas has slowed. The neonatal mortality rate in Nairobi is 24% higher than rural areas, reversing a longstanding urban advantage in MNH. This multiple method dissertation draws on both quantitative and qualitative data to examine the role of five MNH interventions in explaining this urban disadvantage—health facility delivery (HFD), skilled birth attendance (SBA), early postnatal care (ePNC), skilled postnatal care (sPNC), and early breastfeeding initiation (eBFI). First, using data on mother-child pairs from the Kenya Demographic and Health Survey (KDHS) (n=1,058) this dissertation examines intra-urban variation in coverage, as high urban mortality rates may be explained by differences in coverage in rapidly expanding and impoverished urban informal settlements (IS) as compared to formal settlements (FS). Urban places of residence were operationalized using household- and neighborhood-level characteristics to differentiate informal settlements from formal ones. IS were defined as KDHS clusters where >50% of households lacked >=2 of 4 planned community features (access to improved water and sanitation, durable housing materials, and sufficient living area). Urban categories included: IS in cities, IS in towns, and FS in cities and towns. No differences were observed between settlement type and coverage of any intervention except ePNC. Neonates in IS in towns were less likely to receive ePNC compared to those in FS, suggesting that quality-adjusted measures of intervention coverage might offer evidence of intra-urban variation. Second, using semi-structured qualitative interviews with mothers (n=20) and providers (n=12) in two IS, this dissertation addresses concerns with the measurement of these five interventions within household-based surveys. Specifically, it explores the potential influence of social desirability bias and comprehension of questions and key terms on the validity of mothers’ responses to survey questions that capture coverage. Data were analyzed using template and thematic analysis. Most mothers (1) associate payment with private facilities, (2) generalize the term “hospital” to other facility types, (3) generalize the terms “doctor” and “nurse” to other facility-based employees, (4) determine provider cadre based on role, uniform color, and gender, (5) understand the meaning of ‘assisting with delivery’ and ‘postnatal care check’ as intended, (6) report the timing of a range of postnatal checks, and (7) did not provide concrete descriptions of how they identify time to initiation of breastfeeding or first postnatal check. Women in these communities might: (1) indicate they delivered at a private facility to appear to be wealthier than their peers, (2) indicate they had a SBA if their true attendant was a TBA, and (3) over-report breastfeeding. Findings from this dissertation suggest that the urban disadvantage in neonatal mortality may not be explained by differences in intervention coverage within growing IS communities and may be more complex. Poor quality of facility-based care and low validity of data measuring intervention coverage in household surveys may partly explain these findings. Linking household surveys with facility-based data, using quality-adjusted measures to examine intra-urban variation in coverage, and further exploring of the role of desirability bias and comprehension in survey data across diverse settings is recommended moving forward.