Relationship between women’s characteristics and continuum of care for maternal health in Kenya: Complex survey analysis using structural equation modeling

ABSTRACT

The objective of this study was to understand and estimate the complex relationships in the continuum of care for maternal health to provide information to improve maternal and newborn health outcomes. Women (n = 4,082) aged 15–49 years in the 2008/2009 Kenya Demographic and Health Survey data were used to explore the complex relationships in the continuum of care for maternal health (i.e., before, during, and after delivery) using structural equation modeling. Results showed that the use of antenatal care was significantly positively related to the use of delivery care (β = 0.06; adjusted odds ratio [AOR] = 1.06; 95% confidence interval [CI]: 1.02–1.10) but not postnatal care, while delivery care was associated with postnatal care (β = 0.68; AOR = 1.97; 95% CI: 1.75–2.22). Socioeconomic status was significantly related to all elements in the continuum of care for maternal health; barriers to delivery of care and personal characteristics were only associated with the use of delivery care (β = 0.34; AOR = 1.40; 95% CI: 1.30–1.52) and postnatal care (β = 0.03; AOR = 1.03; 95% CI: 1.01–1.05), respectively. The three periods of maternal health care were related to each other. Developing a referral system of continuity of care is critical in the Sustainable Development Goals era.

KEYWORDS:

• Antenatal care
• continuum of care
• delivery care
• maternal health care utilization
• postnatal care
• structural equation modeling

Funding

The authors are indebted to National Yang-Ming University for their financial support. The funders had no role in designing the study, analysis and interpretation, preparation of the manuscript, and decision to publish.

Supplemental Material

Supplemental materials are available for this article at http://dx.doi.org/10.1080/03630242.2016.1222327

Supplemental materials include the following: Supplemental Table 1: Reliability analysis of measurement variables in the five latent constructs (n = 4,082); and Supplemental Table 2: The correlation and covariance matrices of the latent constructs.

Additional information

Funding

The authors are indebted to National Yang-Ming University for their financial support. The funders had no role in designing the study, analysis and interpretation, preparation of the manuscript, and decision to publish.