I am trying to code an interrupted time series for antenatal attendance pre and post lockdown. I have two questions:
1. I want to work out immediate post lockdown (that occurred 10 months into the trial) and trend to the end of the trial - how would I code for both of those please? I have used MLETAS and set a dummy time variable but not sure how to set a time trend for immediate vs trend over the whole time.
2. My data is in number of events, would you recommend I report change in number of events or calculate a change in rate?
My readout looks like this:
Negative binomial regression Number of obs = 18
Wald chi2(1) = 13.04
Dispersion = mean Prob > chi2 = 0.0003
Log pseudolikelihood = -127.74359 Pseudo R2 = 0.0153

------------------------------------------------------------------------------
| Robust
iron | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
lockdown | -.459981 .1273829 -3.61 0.000 -.7096469 -.2103151
_cons | 6.772079 .0427427 158.44 0.000 6.688304 6.855853
-------------+----------------------------------------------------------------
/lnalpha | -1.546523 .8651169 -3.242121 .1490748
-------------+----------------------------------------------------------------
alpha | .2129872 .1842588 .0390809 1.16076
------------------------------------------------------------------------------

.