Table 2

Adjusted and unadjusted analyses—a simple example from a hypothetical trial of a treatment for acute stroke where, for some reason, a greater number of mild patients was randomised to the treatment being tested

All patientsPatients with moderate strokePatients with mild stroke
Calculating relative risks of being dead and dependent:
Overall unadjusted treatment effect is 250/1000÷420/1000  =  0.60
Treatment effect for those with moderate stroke is 100/250÷360/750 = 0.83
Treatment effect for those with mild stroke is 150/750÷60/250 = 0.83
So an overall adjusted treatment effect would be 0.83.
Dead or dependent25042067010036046015060210
Alive and independent7505801330150390540600190790
Total 10001000200025075010007502501000