Working paper 02 — Preliminary analyses complete
Variance, Drivers, and Timeliness of Facility-Based Obstetric Care in Kakamega County, Kenya
When a mother delivers in a Kenyan health facility, what determines whether she gets good, timely care — the facility she chose, the provider she drew, or the moment she arrived?
Austin Tucker · Harvard University · 2026
1,742
deliveries observed
10
facilities
−2.4pp
post-handoff quality
01 — Why it matters
Getting mothers to deliver in facilities was the great maternal health push of the last two decades — and it worked. But facility delivery only saves lives if the care inside is good. Understanding where quality varies (between facilities? between providers? delivery to delivery?) tells us whether to invest in infrastructure, training, staffing, or care processes.
02 — How I answer it
Using direct clinical observations of 1,742 deliveries across 10 facilities from Kenya's Service Delivery Redesign, I decompose the variance in care quality and timeliness across facility, provider, and delivery levels, estimate peer effects among providers working together, and measure what happens to care quality when a delivery is handed off between providers.
03 — What I find
Quality variance concentrates at the facility and provider levels — but timeliness variance lives at the delivery level. Providers exert measurable peer effects on each other (~0.06 after case-mix and leave-one-out corrections), and care quality drops 2.4 percentage points after a shift handoff.
04 — What it means for policy
Quality improvement and timeliness improvement are different problems. Quality tracks who and where; timeliness tracks when and what else is happening — so staffing-level and workflow interventions, not just training, are on the critical path.