.-
help for ^strate^                                           (STB-40: ssa10)
.-

Tabulation of rates and SMR's for st data
-----------------------------------------

^strate^ [varlist] [^if^ exp] [^in^ range] [^using(^filename^)^]
      [^,^ ^cl^uster^(^variable^)^ ^fc^odes^(^rule^)^ ^g^raph ^j^ack ^l^evel^(^#^)^
      ^noli^st ^nom^iss ^now^hisker ^rep^lace ^sc^ale^(^#^)^ ^smr(^variable^)^ 
      ^graph_options^]


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

^strate^ tabulates the rate, formed from the number of failures divided
by the person-years, by different levels of one or more categorical 
explanatory variables (declared in the ^varlist^ of the command). Confidence 
intervals for the rate are also given.

By default, confidence intervals for rates are calculated using the 
quadratic approximation to the Poisson log likelihood for the log rate
parameter. However, for situations in which the Poisson assumption is 
questionable, jackknife confidence intervals may also be calculated. The 
jackknife option also allows for the case where there are multiple records
for the same "cluster" (usually subject).

The command also implements computation of SMR's, after merging the data 
with a suitable file of "standard" rates.

Optionally a summary dataset may be saved in a file. This contains counts of 
failures and person-time, rates (or SMR's), confidence limits, and all the 
categorical variables in the varlist. This could be used for further 
calculations, or simply to lay the table out nicely with the ^table^
command.


Options
-------

^cluster()^ defines a categorical variable which indicates clusters of data
to be used by the jackknife. If the jackknife option is selected and this
variable is not declared, it is taken as the ^id^ variable defined in the
^st^ data.

^fcodes()^ specifies a recode rule for the failure indicator in the ^st^
data. All codes matching the rule are treated as failures and all others 
as censoring. 

^graph^ requests a graph of rates against exposure category; graph options 
are allowed.

^jack^ specifies that jackknife confidence intervals are required.

^level(^#^)^ resets the level for the confidence intervals from the usual 
default. 

^nolist^ supresses the listed output. Only use this option if you are creating
a summary output file.

^nomiss^ restricts analysis to records with no missing values in any of
the explanatory variables. Otherwise missing values are simply treated as 
extra categories.

^nowhisker^ omits the confidence intervals from the graph.

^replace^ allows the summary data file to overwrite one of the same name.

^scale()^ defines the time units in which the rate will be reported. For 
example, if the ^st^ times are coded in days, setting scale to 365250
results in rates per 1000 person-years.

^smr()^ specifies a variable which holds, for each record, an appropriate 
reference rate. The program then calculates SMR's rather than rates. This 
option will usually be used after using ^stlexis^ to split the follow-up 
records by age bands (and possibly calendar periods) and adding the 
reference rates to the file using ^merge^


Examples
--------

^strate hieng, fcodes(410/414)^ - Rates of failure code 410-414 by ^loweng^ 
^dyrate hieng using pytable, noli^ - Summary dataset rather than output 
^strate hieng, jack level(90)^  - Rates with jackknife confidence intervals
^strate hieng, smr(ewrate)^     - SMR relative to E & W rates, by ^loweng^ 


Authors
-------

      David Clayton
      MRC Biostatistical Research Unit
      Cambridge
      email: david.clayton@@mrc-bsu.cam.ac.uk

      Michael Hills
      London School of Hygiene and Tropical Medicine (retired)
      email: mhills@@regress.demon.co.uk


Also see
--------

    STB:  STB-40 ssa10
On-line:  help for @stlexis@, @dyrate@