Actually it explains 1% of the impact, since the R-squared will be 0.01. But there's a huge amount of random noise in whether people get depressed, which we would always struggle to explain. A measure of the effect size is more interesting than a measure of R^2.