As the tables show, initial analysis using regression methods did not show any statistically significant results between the intervention and control group. As the intervention was conducted over 15 months, with 9 months of preparation, it was decided to use data from a period of 36 months. The 36 months was divided as 9 months of pre intervention, 15 months of intervention and 12 months of post intervention period. This 36 months longitudinal data was then analysed using a different statistical model the Generalised Estimating Equation to understand the effect of time and intervention.
The data on HoNOS for the participants from 2012 to end of 2014 was arranged in 3 monthly periods. A HoNOS measure is taken every 3 months for all patients as a 3 monthly review, which is the logic in dividing the time period in 3 month intervals. If there are more than one measure in each of this period, the first entry on these 3 monthly periods were taken for each of the participants. For some, there were missing data in some of these periods.
Each of these periods are named a “wave” for classification purpose. The 3 monthly periods also fall under different stages of intervention. Following table shows how the time periods, waves and stages of intervention are connected.
// See the table below