Evaluating Counseling for Choice in Malawi: A Client-Centered Approach to Contraceptive Counseling

Abbreviations: aOR, adjusted odds ratio; CI, confidence interval; ICC, intraclass correlation; OR, odds ratio; PCCC, person-centered contraceptive counseling.

↵^a Estimated from multilevel mixed effects model with ordinal or binary outcomes. Unadjusted OR estimated from bivariate models.

↵^b Estimated from models that include demographic variables (age, marital status, and education), a categorical variable for number of living children (defined as none, 1–2, 3–4, or 5 or more), and variables for method chosen at provider visit (including no method chosen) and type of facility (public or private). All adjusted models comprised of 1,179 observations.

↵^c Estimated from unadjusted model with random intercepts for health facility but without fixed effects predictors.

↵^d Estimated variance for the random effects at the health facility level; estimate has not been exponentiated.

^e Significant at P<.01.

^f Significant at P<.05.

Abbreviations: aOR, adjusted odds ratio; CI, confidence interval; ICC, intraclass correlation; OR, odds ratio.

↵^a Estimated from multilevel mixed effects models including facility-level random effects. “Gave enough information” and “confidence to use chosen method” were modeled as ordinal variables. The “provider addressed all concerns” outcome was modeled as a binary variable. Odds ratios estimated from multilevel mixed effects model with ordinal or binary outcomes. Unadjusted odds ratios estimated from bivariate models.

↵^b Estimated from models that include demographic variables (age, marital status, and education), a categorical variable for number of living children (defined as none, 1–2, 3–4, or 5 or more), and variables for method chosen at provider visit and type of facility (public or private).

↵^c Estimated from unadjusted model with random intercepts for health facility but without fixed effects predictors.

↵^d Estimated variance for the random effects at the health facility level in the adjusted model; estimate has not been exponentiated.

↵^e Number of observations in the adjusted model; complete case analysis.

↵^f Significant at P<.01.

↵^g Significant at P<.05.

↵^h For modeled estimates, we use a binary version of the indicator that combines "no" and "don't know" responses in a single category (versus “yes”).

↵ⁱ Asked only of women who received injectables, oral contraceptive pills, condoms, or emergency contraceptive pills.