it's using age-group specific data.
More importantly, their rate estimate does not taken into account obesity. US is at ~36%; India is at ~4%. As per Global Diabetes Atlas, estimated (not just diagnosed) prevalence of DM for US is ~9.3% and for India, ~5.5%. The translation of IFR needs to take into account of these factors, but they don't.
Per capita, India has more diabetes than much more wealthy countries.
This is the kind if a country it is. People may be dying from obeseti in the city, and starvation in the village just 100km apart.