Crude and adjusted associations (ORs and 95% CI) between (A) early neonatal mortality and (B) facility delivery and distance to delivery services and level of care.

asample sizes are reduced due to missing values of some confounding variables.
bconfounding variables: meńs opinion on female autonomy in cluster, ethnicity, partneŕs occupation, partneŕs education, womeńs media use in cluster, education, wanted pregnancy, siblings under 7 years old in household, estimate of newborn size (by mother), meńs media use in cluster, womeńs mobility autonomy in cluster, language, womeńs financial autonomy in cluster, multiple pregnancy, occupation, marital status, age at birth, modern attitudes, meńs modern attitudes in cluster, exposure to health programmes in the media, media use, sex of index child.
cconfounding variables: partneŕs education in years, relationship autonomy, partneŕs occupation, media use, womeńs financial autonomy in cluster, wealth, womeńs relationship autonomy in cluster, modern attitudes, newborn size estimate (by mother), marital status, occupation, household composition and siblings under 7 years old, education, literacy, womeńs mobility autonomy in cluster.
dconfounding variables: wealth, womeńs relationship autonomy in cluster, partneŕs education in years, education in years, partneŕs occupation, meńs opinion on female autonomy in cluster, womeńs modern attitudes in cluster, womeńs financial autonomy in cluster, womeńs autonomy to seek health care in cluster.
econfounding variables: womeńs relationship autonomy in cluster, meńs modern attitudes in cluster, language, wealth, womeńs autonomy to seek health care in cluster, meńs opinion on female autonomy in cluster.