Estimates of absolute benefit or risk from meta-analysis --------------------------------------------------------
^metannt , measure(^or|rr|rd^) size(^#^) confint(^#,#^) baseline(^#[,#...#]^)^ >
Description -----------
This program is intended to aid interpretation of meta-analyses of binary data by presenting the effect sizes in absolute terms. Both the number needed to treat (NNT) and the number of events avoided (or added) per 1000 are presented.
The ^Number Needed to Treat (NNT)^ is the number of individuals required to experience the intervention in order to expect there to be one additional event to be observed. Assuming the event is undesireable, this is termed the "number needed to treat to benefit" (NNTB). If the intervention arm experiences more events, this is commonly refered to as the "number needed to treat to harm" (NNTH).
^metannt^ calculates this by deriving an effect size (e.g. a risk ratio), applying it to a population with a given event prevalance, and from this deriving a projected event rate if the population were to receive the intervention. The NNT is equal to 1/(control group event rate - treatment group event rate).
The ^number of avoided or excess events (respectively) per 1000 population^ is the difference between the two event rates multiplied by 1000. The intervention group event rate is calculated in the same manner as with the NNT. Optionally a confidence interval is also presented, using the confidence limits of the effect size applied to the control group event rate.
Options for ^metannt^ -------------------
^baseline()^ specifies the baseline (ie. control group) event rates, from which the NNT and number of events avoided are computed from. More than one baseline event rate may be specified.
^measure(^rr|or|rd^)^ specifies whether the estimated effect size pooled is a risk ratio, odds ratio or (absolute) risk difference.
^size(^#^)^ denotes the size of the effect
^confint(^#,#^)^ denotes the confidence interval around the effect size.
By default the saved results from the r() macros within @metan@ are used to derive the control group event rate, the effect measure, the effect size and its confidence interval. However, unless a recent version of metan (1.71 onwards) was used immediately beforehand, the first three options must be specified. The confidence interval is optional.
Examples --------
. ^metannt , measure(rr) size(0.2) baseline(0.1,0.2,0.3)^
. ^metan tdeath tnodeath cdeath cnodeath, or^ . ^metannt , baseline(0.15)^
Authors -------
Michael J Bradburn, Jonathan J Deeks, Douglas G Altman. Centre for Statistics in Medicine, Institute of Health Sciences, Old Road, Headington, Oxford OX3 7LF, UK.
email mike.bradburn@@cancer.org.uk
Also see --------
On-line: help for @metan@ (if installed), @meta@ (if installed)