CANDAT.gz is a gzipped version of the full Ohio county-level cancer
mortality data (ie, not yet aggregated over age brackets), originally
compiled by Owen Devine.  Best reference for this dataset might be:

Xia, H. and Carlin, B.P. (1998),
``Spatio-temporal models with errors in covariates:  
Mapping Ohio lung cancer mortality,''
{\em Statistics in Medicine}, {\bf 17}, 2025--2043.

Amongst other things, this paper includes the U.S. Standard million
age distribution (1970), which can be used to age-standardize the rates.

**********

Description of ASCII file for Ohio county-level cancer mortality data

ASCII II comma deliminated file
Data dictionary for cancer deaths in Ohio 1968-1988.



field 1 = 5 digit county FIPS code

field 2 = cancer site               Revision of the International
                                   Classification of Diseases
                                        8th            9th
                                   (1968-1978)       (1979-1988) 
      
           4=trachea,bronchus,lung     162                162

           7=leukemia              204-207        202.4,203.1
                                                    204-208

field 3 = year of death

field 4 = sex
           1=male
           2=female

field 5 = age at death
          1 = 0-4             7 = 45-54           
          2 = 5-9             8 = 55-64
          3 = 10-14           9 = 65-74
          4 = 15-24          10 = 75-84 
          5 = 25-34          11 = 85 +  
          6 = 35-44

field 6 = race
          1 = white
          2 = non-white

field 7 = total number of deaths

field 8 = total population

Note: Death counts and population estimates are from the NCHS Compressed
      Mortality File.