MEASURES OF potential impact


The Number Needed to Treat


Risk difference can be used to compute number needed to treat or NNT - which represents the number of patients treated to prevent one outcome.


Example - study of patients with mild hypertension.

       Treatment of hypertension with b-antagonist/diuretic vs. placebo.



RD = 2.6% 1.4% = 1.2% or .012


How many hypertensive patients need to be treated to prevent 1 stroke?


NNT = 1 / .012 = 83.333


Results: treat 83 hypertensive individuals for 10 years, 1 stroke prevented.


6-2 Measures of Potential Impact (continued)


Potential Impact The Concept


v   The etiologic fraction: measure of impact when exposure is a potential cause of disease.

v   The prevented fraction: measure of impact when exposure prevents disease.


Etiologic Fraction [Population]


The etiologic fraction [population] answers the question: what proportion of new cases are attributable to the exposure?



The etiologic fraction [pop] of .35 tells us that 35% of all cases are due to continuing smoking.


In other words, if all patients had quit smoking after their heart attack, there would have been 35% fewer deaths deaths.



Etiologic Fraction for Person-Time Studies


ID = 40 / 104,820 = 38.2 per 100,000 p-yrs



Etiologic Fraction among the Exposed



53% of all deaths among continuing smokers were due to continued smoking.



6-3 Measures of Potential Impact (continued)


Prevented Fraction [Population]



(Note: RR should be 0.04 above)


55% of the total expected cases were prevented by immunization.


Using person-time data,


Prevented Fraction among the Exposed





The prevented fraction of .96 among the exposed tells us that 96% of total expected cases among those immunized were prevented by immunization.