This method is based on the fact that most health data moves in a straight line over time, especially over a 5 year period. Even though the data may vary over 5 years, it moves in a straight line trend. Health data may often jump up in one year, only to go back to its historical trend the next. 
It is usually meaningless to analyze data where less than 20 events occur over the study period. An event is a death, an occurrence of a disease, etc. This is called the 'problem of small numbers'. If you have very few events, a change of one or two can have a big impact of the rate, and can lead to unjustified conclusions. For example. If 100 people die from diabetes in your community each year, 2 more dieing is only a 2 % change, which is not a big deal. If you have 10 people die of diabetes and then 2 more die, this is a 20% change, and could lead to the conclusion that there
is a diabetes 'epidemic'. This data is meaningless. People in a community often express concern to health officials because of an outbreak of a disease in their community; this usually involves small numbers of diseases and normal variations. Using 5 years of data will usually eliminate the 'problem of small numbers'. For more information see:
http://www.health.state.pa.us/stats/techassist/flow1a.htm