EPICURERATES, this path string must include
the trailing . Thus, the subcommand is RPATH EPICURERATES. The environment variable ERATES
can be set to the path for the rate files. Purpose
Associates external rates with the cells in an event-time table.
Syntax
GETRATES CAUSE code AS vname
{AGE vname} {YEAR vname} {SEX vname}
{STRATA vname}
{CMBSEX} {USE M | F} {CMBSTRATA} {USE 1 | 2}
{FILE1 fname} {FILE2 fname]}{ RPATH pstr}
{USMORT | USINC | JPNMORT | DMKINC}
{LISTPARM} {LISTRATES} {LISTCAUSE} @
Arguments and Subcommands
CAUSE code
Indicates the sequence number for the cause of death of interest in the rate file(s). The value of code must be a positive integer. You can use the LISTCAUSE subcommand to obtain a list of all cause codes in the rate files included with EPICURE.
AS vname
Specifies which variable will contain the expected rates. This subcommand is required.
AGE vname
Specifies the age variable. The default name for this variable is age. An age variable must be present. This variable should be coded in the same units as the age category cutpoints in the rate file(s).
YEAR vname
Specifies the time period variable. The default name for this variable is year. A year variable must be present. This variable should be coded in the same units as the time category cutpoints in the rate file(s).
SEX vname
Specifies the sex indicator variable. The default name for this variable is sex. This variable must be coded as 0 and 1 or 1 and 2 for males and females, respectively. The sex variable is not necessary if the rate file contains only one set of rates, the USE subcommand is used to select the male or female rates from the rate file(s), or the CMBSEX subcommand is used to combine the two sets of rates.
STRATA vname
Specifies the strata index variable. There is no default strata variable name. This variable name must be specified if there are two rate files and the CMBSTRATA option is not used. The strata index variable should be coded as 0 and 1 or 1 and 2.
CMBSEX
Instructs the program to combine male and female rates using population weights in the rate file(s). This subcommand is useful when the data contain no information about sex but the rate files contain both male and female rates.
USE c
Indicates which of the two sets of rates in the rate files is to be used. If the argument, c, is M, only the male (first set of) rates are used. If the argument is F, only the female (second set of) rates are used. This is useful in a study of a sex-specific disease.
CMBSTRATA
Instructs the program to combine the rates in two strata-specific rate files using population weights in the files. This subcommand is useful when the rates are stratified, but the data come from a population composed of members from both populations in the proportions indicated by the weights in the rate files.
USE n
When one is using more than one rate file, this subcommand indicates which of the two files is to be used. If the argument, n, is 1, only the first file is used, while if the argument is 2, only the second file is used.
FILE1 fname
Specifies the name of the first rate file. There must be at least one rate file. Up to two rate files can be used. If two files are used, the files will contain rates for different strata or subpopulations, for example, whites and non-whites. At least one rate file must be specified. As described below, the USMORT, USINC, JPNMORT, and DNMMORT subcommands can be used to request the standard EPITOME/EPICURE rate files. If two files are specified and the CMBSTRATA option is not chosen, the STRATA subcommand must be used to indicate the name of the variable which indicates which rate file to use for each cell.
FILE2 fname
Specifies the name of the second rate file. When two files are used, the files will contain rates for different strata or subpopulations, for example, whites and non-whites. At least one rate file must be specified. As described below, the USMORT, USINC, JPNMORT, and DMKINC subcommands can be used to request the standard EPITOME/EPICURE rate files. If two files are specified and neither the CMBSTRATA subcommand nor the (stratum) USE subcommand are chosen, the STRATA subcommand must be used to indicate the name of the variable which indicates which rate file to use for each cell.
RPATH pstr
Specifies a string as a prefix for rate file names. This string
should generally be a path name. The string is appended exactly as specified so
it must be terminated with a proper separator. This means that if the rate files
are in a directory called EPICURERATES, this path string must include
the trailing . Thus, the subcommand is RPATH EPICURERATES. The environment variable ERATES
can be set to the path for the rate files.
USMORT
Use the United States cancer mortality rate files for whites (file 1) and non-whites (file 2) provided with the EPICURE package. Unless changed by other requests, for example, a STRATA subcommand, the rate files will be merged.
USINC
Use the Connecticut tumor registry cancer incidence rate files. Since this is a single rate file, requests related to stratification are ignored.
JPNMORT
Use the Japanese cancer mortality rate files. Since this is a single rate file, requests related to stratification are ignored.
DMKINC
Use the Danish cancer incidence rate files. Since this is a single rate file, requests related to stratification are ignored.
LISTPARM
Requests a listing of the current GETRATES parameter settings.
LISTRATE
Requests a listing of the contents of the rate file by the cause of interest.
LISTCAUSE
Requests a listing of the cause codes in the current rate file. If this option is requested, the only output is a list of the cause codes.
Remarks
This command is used to add population (expected) rates to an event-time table. Although not required, the table is usually created by DATAB. By adding expected rates to an event-time table, AMFIT can be used to carry out Standard Incidence Rates/Standard Mortality Rate (SIR/SMR) analyses which generalize those available in programs such as Monson (1974).
This command uses rate files with a special format (described in detail in EPICURE Rate File Format). A rate file contains age- and time-specific external (population) rates and population sizes for one or more causes, identified by integer codes (such as three digit International Cause of Death (ICD) codes). The number of age- and time-categories and the category bounds are determined from information in the rate file header. A single rate file may contain two sets of rates, which we (somewhat arbitrarily) call sex-specific rates. In this case, the first set of rates is called (again somewhat arbitrarily) male rates and the second set, female rates. It is possible to use two rate files at once, which allows a limited degree of stratification. If two rate files are used, the files must have the same structure, that is, the same number of "sexes" and the same number, coding, and order of causes.
The user must specify the age and time variables that determine the appropriate rate in a cause-specific rate table. It is also necessary to specify the variable that contains the expected rates. If sex-specific rates are to be added to the table, the variable which determines the appropriate sex-specific rate for a cell must be specified. If stratified rates are used, (for example, rates based upon ethnic group), a stratum index variable must be specified. The sex and stratum index variables must be coded as 0 and 1 or 1 and 2. (This can be done using transformations.)
The USE subcommands specify a subset of sex or stratum rates to update the table. The CMBSEX and CMBSTRATA commands are used to instruct the program to use average rates, computed using population weights in the rate files, to update the table. If the rate file(s) contain sex-specific rates but a sex index variable is not (explicitly or implicitly) specified and the USE subcommand is not used, sex-specific rates will be averaged using the population weight information in the rate files. Similarly, if stratum-specific rate files are specified without the specification of a stratum index variable or a USE command, the stratum-specific rates will be averaged using the population weight information in the rate files.
Expected rates are added by finding the first age- time- category (in the appropriate sex and stratum rate table) which includes the age and time indicator values for the cell. The inclusion test is carried out using the default DATAB inclusion rule, (that is, greater than or equal to the lower bound and strictly less than the upper bound). If an age, time, or index variable is out of range, the expected rate variable is set to the internal EPICURE missing value code.
In some analyses, one might be interested in the expected numbers of cases. These can be computed as the product of the number of person-years and the (appropriately scaled) rate. The rate file header contains information on the rate scaling factor, for example, 100,000 for rates per 100,000 person-years or 1,000 for rates given per 1,000 person-years. In order to use the rates, it is necessary to know this scaling factor. To aid in this rescaling, the rate scaling factor is stored in a named constant called #_RTFACT. The rates can be rescaled quite easily using this constant. For example, to get the rates in the event-count table as per 1,000 person years, use the following transformation (following the GETRATES command):
TRAN erate = erate*1000/#
rtfact @
Although you can provide your own rate files, various mortality and cancer incidence rate files are distributed with EPICURE. The USMORT, USINC, JPMORT, and DMKINC commands can be used to set up the standard rate files. For the United States cancer mortality files, the standard setup assigns the sex-specific rates for whites as file 1 and the sex-specific non-white rates as file 2. The race-specific rates are merged unless a stratum variable is specified.
Examples
a) To append sex-specific Japanese leukemia mortality rates (assumed to be cause number 15 in the rate file) to the user's data. All items necessary for addition of the rates are specified explicitly.
GETRATES CAUSE 15 AS lkrate FILE1
\EPICURE\RATES\JPNMORT.OXR
SEX sex AGE age YEAR year @
b) To repeat the last example, omitting those commands that are the same as the defaults:
GETRATES CAUSE 15 AS lkrate JPNMORT @
c) To repeat the previous example using an alternate path for the rate file, to rescale the rates, and to compute the expected number of cases in each cell:
GETRATES CAUSE 15 AS lkrate RPATH I:\ RATES\ JPNMORT @
TRAN lkrate =
lkrate/#rtfact; lkexp = lkrate
pyr @
d) Use the U.S.
mortality files to add population (white and nonwhite) lung cancer rates
(assumed to be the 10
set of rates in the rate file) to
an event-time table which is not stratified on race. The rates will be
averaged.
GETRATES CAUSE 10 AS lgrate USMORT SEX sx
AGE attage YEAR time @
e) Use the U.S. mortality files to add population (white and nonwhite)lung cancer rates to an event-time table stratified on race. The rates will be averaged.
GETRATES CAUSE 10 AS lgrate USMORT SEX sx
AGE attage YEAR time STRATA race @
f) Use the U.S. mortality files to add cervical cancer rates (assumed tobe cause 22) for black females to an event-time table when the table contains no sex or race indicators.
GETRATES CAUSE 22 AS cvrate USMORT
USE F USE 2 @