In the following slides the presentation follows the same graphic display.  The filled circle presents the point estimate of protection and 95% confidence intervals are shown as the horizontal line.  Observations to the right of the vertical line indicate protection, to the left harm, and overlapping indicates no statistical proof for either protection or harm.  The scale for protection is shown on a logarithmic scale to reflect that these are relative measures.

Five prospective clinical trials in the pre-chemotherapy era have studied the protective efficacy of BCG in the prevention of death from tuberculosis.  The point estimates for four of the trials are 80 per cent and more, but the 95% confidence intervals overlap zero in three of the five studies.  These studies were done in early years when statistical considerations were not accorded the priority as they do nowadays (all the confidence intervals were calculated retrospectively from the data).

To top