ANC8 analysis result

Published

December 8, 2025

Preface

Research plan

Introduction

Research objectives:

  • Comparing mortality rate between ANC 8 recommendation and ANC 4 recommendation

  • Impact of contextual provider quality on the results. Hypothesize that ANC 8 only effective among regions with sufficient quality.

Methods

Data source

DHS from 2016: after WHO recommendation issued

Study population and sample size

42 countries

XXX regions

Outcomes

  • Neonatal mortality (i.e., deaths within the first month)

  • Infant mortality (i.e., deaths within one year since birth)

  • Under-five child mortality (i.e., deaths within the first five years)

ANC

3 groups

  • ANC < 4

  • ANC 4-7

  • ANC ≥ 8

Contextual provider quality (PQ) indicators

Derived from ecological linear models (district/province level): Mortality ~ composite index of health provider quality (equipment, delivery facilities, skilled birth attendance, etc)

  • Residuals < 0 → higher provider quality

  • Residuals > 0 → lower provider quality

Covariates

Child level

  • Child’s sex (male, female)

  • birth order (1, 2~3, 4~5, ≥6)

  • birth interval (first birth, <24, 24~47, ≥48 months)

  • delivery mode (normal, caesarean)

  • Institutional delivery

Mother level

  • mother’s age

  • marital status (currently married or living together, never/formerly married)

  • education (none, primary, secondary, higher, college)

  • age at marriage (<18, ≥18 years)

  • Household level

  • SES (PCA of many items)

  • Source of drinking water: safe for water piped into dwelling or yard/plot, public tap/standpipe, tube well or borehole, protected well or spring, rain water, and bottled water, and unsafe otherwise.

  • The sanitation facility was defined as improved if households had access to flush to piped sewer system, septic tank, or pit latrine, ventilated improved pit latrine, pit latrine with slab, and composting toilet, and unimproved otherwise.

  • Solid fuels were used for cooking (yes, no) was considered as a crude measure of household air quality.

  • Household place of residence (urban, rural)

Data analysis

Descriptive statistics

% Neonatal deaths (%) Infant deaths (%) Under-five deaths (%) by participant’s characteristics

Country specific analysis

Multilevel analysis - Model 1: only intercept

  • Model 2: With all covariates

  • Model 3: With ANC

  • Model 4: With contextual PQ

  • Model 5: With interaction between ANC and PQ

Pooled analysis

Using meta-analysis with random effect model to pool results from individual countries

  • Pool all countries

  • Pool by regions

Meta regression to explain heterogeneity if exists

Sensitivity analysis

  • Remove those with ANC > 12 (or > 15) to avoid reverse causality (women with known fetal abnormality tend to go for ANC more frequently → high number of ANC → more mortality)

  • Contextual provider quality (PQ) indicators using different composite indices