.-
help for ^aflogit^                                         (STB-42: sbe21)
.-

Attributable fraction from logistic regression
----------------------------------------------

	^aflogit^ [term [term ...]] [weight] [^if^ exp] [^in^ range]
		[^,^ ^cc^ ^ref^erence^(^term ^=^ # [term ^=^ # ...]^)^ ^l^evel^(^#^)^ ]

^fweight^s and ^aweight^s are allowed; see help @weights@.

^aflogit^ can only be run after unconditional logistic or poisson regression, see
help @logistic@, @logit@, @blogit@, @glm@ or @poisson@.


Description
-----------

^aflogit^ estimates adjusted measures of the population attributable
fraction from the current logistic or poisson regression model. Without 
arguments, ^aflogit^ reports attributable fractions for all terms in the
model which are positively associated with the outcome.

Standard errors and confidence intervals are based on asymptotic
approximations, see Greenland and Drescher (1993).

The weights, ^if^, and ^in^ clauses used in the last regression model (if any)
are picked up automatically except after @logit@, @blogit@ and @poisson@
in which case the user should specify the weights and selection used.


Options
-------

^cc^ indicates that the data come from an unmatched case-control study.
    The default is to assume the data come from a cross-sectional or
    cohort study. It is important to set this according to the type of
    data you have since the calculations (and hence estimates) are 
    different in the two cases.

^reference^ determines the reference level (the non-exposure category)
    for each term in the model. By default this is zero for every term.
    For continuous variables it may be desirable to set the reference 
    level to a non-zero value (e.g. the mean for a variable such as
    blood pressure).

^level(^#^)^ specifies the significance level, in percent, for confidence
    intervals of the coefficients; see help @level@.


Examples
--------

   . ^logistic mdrtb age sex hiv prevtb bornabr^
   . ^aflogit hiv prevtb bornabr^
   . ^aflogit age hiv, ref(age=16)^
   . ^aflogit^

   . ^xi: logit cc i.alcohol i.smoke*i.agegp [fw=n]^
   . ^aflogit Ialcohol* [fw=n], cc^


Reference
---------

Greenland S. and K. Drescher. 1993. Maximum likelihood estimation of the
attributable fraction from logistic models. Biometrics 49: 865-872.


Author
------

      Tony Brady
      PHLS Statistics Unit
      U.K.
      email: tbrady@@phls.ac.uk


Also see
--------

    STB:  STB-42 sbe21
On-line:  help for @blogit@, @cc@, @glm@, @logistic@, @logit@, @level@