1. Regarding combining the different years data together, the 2012-2013 and 2013-2014 can be combined together, because the two years data collection used the same sampling frame. When combing the two collections, the hospitals and health centers are exhausted, means every hospitals and health centers in Senegal should be in the data, except a few not respondent. For the health posts, combing the two collections does not need to adjust the sampling weight because the total sampling fraction is approximately the same.
2. But the 2014-2015 data cannot be combined with previous years data because the 2015 data used an updated sampling frame. On the other hand, I don't see the necessity to combine the data for trend analysis, especially for comparative analysis.
3. Cautions should be paid when comparing the survey results at regional level because the survey was not designed for this purpose, the regional level sample size is not powered for trend detection, except for hospitals and health centers for two-year combined (2012-2014 vs 2015-2016) because they represent each a census.

Hope this helps

Ruilin