Fig. 4

The predicted probability of reaching each cascade step by age as a continuous variable. Predicted probabilities were average adjusted predictions obtained from covariate-adjusted logistic regressions of diabetes care indicators on individuals’ socio-demographic characteristics (age, household wealth quintile, education, marital status, sex) and district-level fixed effects. We used restricted cubic splines with five knots for the continuous variable age. The knots were placed at the 5th, 27.5th, 50th, 72.5th, and 95th percentiles. Nineteen thousand four hundred fifty-three individuals with diabetes were included in this figure; 10,504 were “aware”, 8269 “treated”, and 5329 “controlled”