The key idea is what makes a raw comparison of death rates between two countries unfair, and what standardisation fixes.
Mortality is strongly tied to age: older people die more often. If one country has many elderly citizens and the other has a young population, the older-skewed country will show a higher crude death rate even if, age for age, the risk is identical. That difference in the age make-up of the two populations is the confounder.
Direct standardisation solves this by taking the age-specific death rates from each country and applying them to one shared standard population structure. The resulting age-adjusted rates strip out the influence of differing age profiles, so any remaining gap reflects real differences in mortality rather than demographics.
It is not about causes of death, nor about reconciling numerators or denominators in isolation; it is specifically about evening out the age distribution.
\[\boxed{\text{Age distribution}}\]