Dr. Andrew Schechtmann recently took me to task over the statistics used to gain approval for a new indication for the human papilloma virus vaccine. And rightly so, as I agree that relative risk (which is the most commonly advertised result) only tells part of the story. In fact, we really should try to look for the absolute risk reduction which can then be inverted into the number-needed-to-treat to see if the intervention really makes sense.
So let me attempt to make amends by discussing a freshly published retrospective cohort study of 75,761 vaccinees compared to 227,283 unvaccinated age-matched cases (all immunocompetent adults >60yo) in which there was a statistically significant 45% reduction in herpes zoster (shingles) diagnoses in those who received the vaccine (consistent w/previous randomized double blind placebo controlled trials) compared to those who did not. Sure, this is impressive but consider that the rate of infection was 6.4 per 1,000 person-years in those vaccinated compared to 13.0 per 1,000 person-years in those not vaccinated. The absolute risk reduction is therefore 6.6 per 1,000. Take the reciprocal of 6.6/1,000 and you get an NNT of 152. In other words, you'd need to vaccinate just 152 patients to prevent one herpes zoster diagnosis.
Moreover, you're talking about a common disease entity (zoster/shingles) in a common group of patients (immunocompetent elderly) as compared to a relatively rare disease (anal cancer) in a much smaller group of patients (men who have sex with men). Therefore, you can see the greater imperative here for us to vaccinate our elderly from a painful & potentially chronically disabling disease (postherpetic neuralgia).