Hi again, I'm back. Before you jump into the data and our code for the evaluation of a policy change on the Island of Sicily in Italy, I'm going to describe the background and study design for you, again, using the language of evaluation research design that we've covered together. Sicily is one of Italy's 20 regions, and the largest island. I've never been there, but I do hear it's beautiful. On January 1st, 2005, the Government of Sicily implemented a new tobacco control policy that placed a ban on smoking in public buildings and spaces. This type of smoking ban has been implemented in many jurisdictions, and several evaluations have shown an almost immediate effect on heart attacks, and other acute coronary events, or ACEs. Now, there's actually a biological mechanism at play here by which exposure to secondhand smoke can trigger an acute coronary event in someone with severe heart disease. Sicily wanted to evaluate its new smoking ban in multiple ways, but also to see if this type of immediate health outcome was observed on the island after the policy change. To do this, an interrupted time series evaluation design was used. Monthly data on ACEs from all hospitals in Sicily were analyzed from January 2002 through November 2006. Now, remember that a study design is a time series if there are multiple observations both before and after an intervention is implemented, so we can see if there's a change in the intercept or the slope of a trend line that happens at the same time as the intervention is introduced. Researchers took advantage of 59 months of data. I'm not sure why they did not use data from December of 2006 to make it 60 months, or 5 full years of data. But we have 59 months of data, and the unit of analysis for the data is months. Not full years, or not even quarters, they used month as the time and the time series. This is important because smoking, like almost all of other human behavior, has a seasonality to it. In fact, people tend to smoke less in summer months, or hotter months, than when it's colder. Using monthly data gives more observations, and also helps control for fluctuations due to the season of the year. The counterfactual in this study design is the expected trend line after 2005, when the new policy was introduced. Using data from before the policy change, from 2002 through 2005, an expected trend line is forecasted into the future. The actual or observed trend line from the actual data, from 2005 and 2006, is then compared with the expected or projected trend line. These studies we're often comparing what we might expect if a trend line continues uninterrupted, with what was observed after an interruption or a policy change. It's this comparison of the expected, or counterfactual, with what was actually observed, that allows us to measure the impact of the smoking policy change. The picture of what the study design looks like, is somewhat what I've pictured here. Now, I'm not writing out all the 59 observation points. That is too many O's for one slide, or for any one to be able to read and digest. But I hope that you see in this study design, there are three years of monthly observations, or 36 months of data before the X, or the policy change, and then there are 23 more months, or almost two years of monthly data, to observe after the policy change. In this design, the O's, or monthly observations, are the rates of hospitalization in Sicily due to acute coronary events. That's the study design. Have fun learning more about the data and the analysis.