Effect of distance and level of provision on various aspects of ANC use in rural Zambia (n = 24051).

110, 23 and 23 missing/unknown values for the three outcomes respectively, counted as later than first trimester, less than four visits, and no good quality.2Evidence against linearity over categories: p = 0.36, p = 0.74 and p = 0.18, respectively.3Confounders included in the adjusted models:First trimester visit: mother's education, household wealth, modern fertility attitude of mother, literacy level of mother and average women's media use in the cluster.4+ ANC visits: mother's education, household wealth, wantedness of pregnancy, modern fertility attitudes of mother and of husband, ethnic group, religion, violence experience, family composition and average men's opinion on women's autonomy in the cluster.Good quality ANC: mother's education, household wealth, ethnic group, modern fertility attitudes of mother, average women's fertility attitudes in the cluster and average women's exposure to health information in the cluster.