You have to be careful relying on isolated stats taken out of context.
You also have to be careful not to confuse correlation with causation. It is possible that there could be a correlation between low levels of blood alcohol and driving risk without any causal link. e.g. if driving is most dangerous at certain times of day for other reasons (roads busier, darkness, tiredness, etc), and if drivers are more likely to have non-zero blood alcohol at the same times of day then there will be a correlation between blood alcohol and risk even if the alcohol isn't causing the risk.
Not that i approve of driving with any intoxicant in your system.
You also have to be careful not to confuse correlation with causation. It is possible that there could be a correlation between low levels of blood alcohol and driving risk without any causal link. e.g. if driving is most dangerous at certain times of day for other reasons (roads busier, darkness, tiredness, etc), and if drivers are more likely to have non-zero blood alcohol at the same times of day then there will be a correlation between blood alcohol and risk even if the alcohol isn't causing the risk.
Not that i approve of driving with any intoxicant in your system.