If you make a bunch of assumptions about the distribution, you don't need a large sample size to make decent estimations about the population. Of course, the only way that works is with random or stratified random sampling. With reporting bias, I'm not sure any country has enough data to make statistically confident statements with any precision.

Folks with a stronger statistics background are welcome to correct me.