To be statistically significant something has to relative to a larger number of the same. So, if we were told that 50,000 people died this winter then that could only be considered against the total population because we have no other data. To be statistically significant the percentage needs to be at least 1% and above. From the above fictional example 50,000 is roughly 0.0007% of the population (65million) and would be considered as insignicant.
Next you need to define your data to get a number that relates to something much more specific. So far this winter I heard a quote that 39 people had died from flu. Of those most had died from the H1N1 virus and just 3 from other strains of flu.
So we could take the 36 and compare that to the previous year(s) to see if it is significant, and we could also compare the three to the total of 39. In the latter case we have a 7% (approx) result and therefore that is statistically significant for compiling data and future measurements, BUT only as a percentage of the total number of all cases of flu.
Now take the 36. Let us assume that during the same period last year 50 people died from the H1N1 virus. So this year the result of 72% is significant and any report would say that this year the number of deaths from the H1N1 is lower and only 72% against the previous year. That is a result which is statistically significant. But taking that figure against the total population it would be such a small percentage that it would be statistically INsignificant.
So, to be statistically significant you first need to be comparing like with like and then the result has to be of a size that makes it significant against the total in that particular group. I hope this will help you in understanding your question.