Most (87%) of the simulated absolute risk differences were [less than or equal to] 0, indicating that vaccination is (slightly) protective against GBS.
Beyond these benefits, the tendency of influenza vaccination to reduce a person's overall risk of acquiring GBS under many conditions (although the absolute risk differences are extremely small) should strengthen confidence in the safety of influenza vaccination and allow health professionals to better put the risk of GBS in context when communicating risks and benefits to potential vaccinees.
In the earlier study, only 1 man (in the placebo group) got CRPS, for an absolute risk difference
of 0.08 (95% CI, -0.07 to 0.27).
In terms of intermediate outcomes, there was no clinically significant difference among the groups in need for additional agents to control blood pressure or in adverse effects, except for increased rate of angioedema in patients on lisinopril (absolute risk difference
[ARD] vs chlorthalidone=0.3%, NNH=333) arid increased rates of elevated fasting blood glucose and potassium supplementation in patients on chlorthalidone (ARD vs lisinopril=3.5% and 6% with NNH=29 and 16, respectively).
Men in the radical prostatectomy group had lower prostate cancer mortality than patients in the watchful waiting group at 8 years (absolute risk difference
[ARD]=6.6; 95% confidence interval [CI], 2.1-11.1; number needed to treat [NNT]=15).
In trials comparing amoxicillin with other antibiotics, there were no differences in cure or failure rate (absolute risk difference
for clinical cure 3.2%; 95% CI, -1.5% to 7.8%).
Absolute risk differences
for perinatal depression ranged from a 1% increased reduction in controls to a 32% increased reduction among women who received counseling.
For each education group, we obtained risk ratios (RR) and absolute risk differences
(RD) for the association between period and stillbirth, adjusted for maternal age, parity, marital status, and mother tongue.
Absolute risk differences in the probability of the outcomes between the preschool and non-preschool group were calculated from the fully adjusted Model 2 (using the generalised linear model results only) by setting all predictor variables at their mean values (StataCorp, 2007).
Table 2 shows the effect estimates and Table 3 converts these into adjusted absolute risk differences.
There was minimal change with the addition of child social disadvantage variables (Model 2) with a fully adjusted estimate of prevalence ratio of 0.75, 95% confidence interval 0.52-1.06, and absolute risk differences of -2.1% (Table 3).
Adding childhood socio-economic indicators did not change the effect size or precision further (prevalence ratio 0.72, 95% confidence interval 0.40-1.31, absolute risk differences -1.1%, third column).