Assessing outcomes of large-scale public health interventions in the absence of baseline data using a mixture of Cox and binomial regressions