source: main/adopters/nm/trunk/src/main/backend_qModules/3.0/_EPHT/mortmc/NM_RateAA10K_GE35.def @ 24491

#NM_RateAA10K_GE35.def
#Calculates directly age-adjusted rates per 10,000 for 
#  records that have been filtered by IBIS-Q (data=tmp)
#Uses the IBIS-Q data_frame for cross1 and cross2
#Uses an "agepop_frame" so all agegrp18 (agepop elsewhere)
#  groups are always included/represented  (weights will sum to 1.0)
#NM_ in filename = applies cell suppression rule
#Outputs redflag var based on RSE
#
f type special  
######################################### 
--------BoNdArY-------- 
1 script 
OPTIONS MPRINT MLOGIC SYMBOLGEN NONUMBER NODATE linesize=175 PAGESIZE=4000;
OPTION SPOOL;

        %LET MULTIPLIER=10000;
        %LET STDVAR=agepop;
        %LET NCAT=11;
 ************************** 1. TMP ******************************;
 * The dataset 'tmp' is the numerator dataset that has been read ;
 * in already by ibis-q. Any filters have already been applied.  ;
 * The proc summary counts deaths by agepop, cross1, and cross2. ;
 * The variable "x" must be in the dataset, it is set equal to 1.;
 ****************************************************************;
        proc summary data=tmp; 
                var x;  
                class &STDVAR.  %cross1%
                ?cross2? %cross2%
                ;
                output out=tmp sum=count;
        proc sort data=tmp; 
          by %cross1%  %cross2% &STDVAR. ;
          run;
        proc print data=tmp noobs; 
                title1 '---------------------------------'; 
                title2 '1. TMP - numerator dataset'; 
                run;

 ********************** 2. %data_frame% *************************;
 * %data_frame% is a dataset created by IBIS-Q. It consists of   ;
 * %cross1% and %cross2% (if the user specified %cross2%) and a  ;
 * variable named "count" that has been set to "0". The results  ;
 * of the proc summary must be merged with the %data_frame%      ;
 * dataset.                                                      ;
 ****************************************************************;
        data agepop_frame;
                retain count 0;
                do 
                        &STDVAR. = 1 to &NCAT. by 1 ;
                        output;
                end;
        run;
        *************************************************************************************;
        * If user selects specific dimension values, e.g., years, the total dimension value  ;
        * will not be included in the data frame. This creates a dimension value for the     ;
        * total row. But if user does not select specific values (e.g., selects a year,      ;
        * then crosses by sex), there will be a total row for the cross1 and cross2 values.  ;
        * So in a later step, the new_frame dataset is sorted with a "nodupkey" option to    ;
        * remove the additional dimension total values.                                      ;
        *************************************************************************************;
        data crosstotalframe;
                retain count 0;
                do
                        %cross1%=.;
                        ?cross2? %cross2%=.;
                        output;
                end;
        run;
        data df_%cross1%%cross2%;
                set df_%cross1%%cross2% crosstotalframe;
        run;
        proc print data=df_%cross1%%cross2% noobs; title2 'cross1cross2 frame with total row'; run;
        proc print data=agepop_frame noobs; title2 'agepop frame'; run;
        proc sql;
                create table new_frame as
                select df_%cross1%%cross2%.*, agepop_frame.*
                from df_%cross1%%cross2%, agepop_frame ;
        quit;
        proc sort data=new_frame nodupkey;
                by %cross1%
                   ?cross2? %cross2%
                   &STDVAR. ;
                run;
        proc print data=new_frame noobs; title2 'new_frame, sorted with nodupkey'; run;
        proc sort data=tmp; by %cross1% %cross2% &STDVAR. ; 
          run;

        proc sort data=tmp out=sorted; 
          by %cross1% %cross2% &STDVAR; run;
        data newtmp;
        merge new_frame sorted;  * tmp, must list frame dataset first, then sorted tmp;
                by %cross1% %cross2% &STDVAR.;
                drop _TYPE_ _FREQ_ ;
        run;

        proc print data=newtmp noobs; 
                title2 '2. NEWTMP, after frame merged with tmp'; 
                run;
        data tmp;
                set newtmp;
        run;

 ************************* 3. POP *******************************;
 * Sum the pop counts by age group, cross1 and cross2.           ; 
 * Read proc summary output, recode the agepop total records,    ;
 * cross1 and cross2 totals to -1.                               ;
 ****************************************************************;
        data poptmp; format  &STDVAR  %popcross1% %popcross2% 8.; set poptmp; run;  
        proc summary data=poptmp; 
                var popcount; 
                class  &STDVAR   %popcross1% %popcross2%;
                output out=pop   (drop=_TYPE_ _FREQ_)
                sum=popcount; 
        run; 
        data pop;
                set pop; 
                proc print data=pop noobs;  
                   title2 '3. POP - denominator dataset'; 
        run; 
  
 *********************** 4. NUMBERS *****************************;
 * Join tmp, pop.                                                ;
 * -1 values are left of out NUMBERS because stdpop doesn't have ;
 * any -1 values.                                                ;
 ****************************************************************;
        proc sql; 
                create table numbers as 
                select tmp.*, pop.* 
                from tmp left join pop
                on tmp.&STDVAR.=pop.&STDVAR. 
                ?popcross1? and tmp.%cross1%=pop.%popcross1%  
                ?popcross2? and tmp.%cross2%=pop.%popcross2% 
                ; 
        quit; 
        data numbers;
                set numbers;
                if &STDVAR. ^=.; *get rid of total rows to stop doubling from pop proc summary;
                if %cross1%=. then %cross1%=-1;
                ?cross2? if %cross2%=. then %cross2%=-1; 
        proc print data=numbers noobs; 
                title2 '4. NUMBERS';
                title3 ' ';
        run; 

 ************************ 5. STDWGT *****************************;
 * Select the appropriate records from the stdwgts  datafile.    ;
 * All weights are for U.S. 2000 standard population (per NCHS   ;
 * Statnote20). 'stdvar' values are as follows:                  ;
 *      agepop - the standard 11 age group weights                 ;
 *      agepopGE35 - agepop wgts for age>=35                       ;
 *      AADist10 - Statnote20 dist #10, for BRFSS                  ;
 *      AADist10GE40 - AADist10 for ages >=40                      ;
 *      AADist10GE50 - AADist10 for ages >=50                      ;
 *      AgeGrp18 - 5-year age groups, for cancer data              ;
 *      AgeGrp18LT50 - AgeGrp18 for age <50                        ;
 *      AgeGrp18GE50 - AgeGrp18 for age >=50                       ;
 *      AgeGrp18LT15 - AgeGrp18 for age <15                        ;
 *      AgeGrp18LT20 - AgeGrp18 for age <20                        ;
 ****************************************************************;
        data stdwgt;
                informat stdwgt 7.6;
                set pop.stdwgts;
                *pop libname is set in .cfg file with the saspop statement;
                if (stdvar='agepopGE35'); *stdvar name is case-sensitive;
                keep &STDVAR.  stdwgt;
        run; 
        proc print data=stdwgt noobs;  
                title2 '5. STDWGT- std. pop. weights for direct age-adjustment';
        run; 
 ******************** 6. NUMBERS2 *******************************;
 * Adds the standard population weights to the numerator and     ;
 * denominator table. This is done as a separate sql, left join  ;
 * to preserve all the -1 values.                                ;
 ****************************************************************;
        proc sql; 
                 create table numbers2 as 
                 select numbers.*, stdwgt.stdwgt
                 from numbers left join stdwgt
                 on numbers.&STDVAR.=stdwgt.&STDVAR. 
         quit; 
        proc print data=numbers2 noobs; 
                title2 '6. NUMBERS2- Standard POP wgts merged onto dataset'; 
        run; 

 *********************** 7. AGESPECIFIC *************************;
 * Calculate age-specific rates & cross-products                 ;
 ****************************************************************;
        data agespecific;
                set numbers2;
                r=(count/popcount);             *age-specific proportion;
                ratewgt=r*stdwgt;               *weighted age-specific proportion;
                * Montana method... rate_var=count*(stdwgt/popcount)**2;  * New formula from doh.wa.gov;
                rate_var=(stdwgt**2)*((r*(1-r))/popcount);

        proc print data=agespecific noobs;  
                title2 '7. AGESPECIFIC - combines numer and denom data, includes R (age-specific proportion),'; 
                title3 'RATEWGT (weighted age-specific rate), and RATE_VAR (rate variance) vars'; 
        run;
   
 ************************ 8. AARATE ******************************;
 * Sum the age-specific rates into aa rates, and merge wgtmax.    ;    
 *****************************************************************;
        proc summary data=agespecific nway; 
                var ratewgt rate_var count popcount; 
                class %cross1% %cross2%; 
                output out=aarate (drop=_TYPE_ _FREQ_ )
                sum(ratewgt rate_var count popcount)=aarate aarate_var count popcount; 
        run;
        data aarate;
                set aarate;
                drop _TYPE_ _FREQ_ ;
                run;
                proc sort data=aarate; by %cross1% %cross2%; run;
        proc print data=aarate noobs; 
                title2 '8. AARATE - summed across weighted &STDVAR. cross-products to get AA rate and'; 
                title3 'rate variance within each cross-by var'; 
      run; 

 ******************* 9. TMP *************************************;
 * Almost final IBIS output dataset.                             ;
 * Make sure no summary rows are included. All -1 totals will be ;
 * coded back to '.' after this step.                            ;
 ****************************************************************;
        data rate;
                set aarate;
                if %cross1%=-1 then %cross1%=.;         *Recode the statewide totals.;
                ?cross2? if %cross2%=-1 then %cross2%=.;
                
                if (count>0) then do;
                                rateper=(aarate*&MULTIPLIER.);
                                rate_se=sqrt(aarate_var);
                                stderr=(rate_se*&MULTIPLIER.); 
                        end;
                if (count<=0) then do;
                                count=0;

                                rateper=0;
                                stderr=sqrt((3/popcount)*(1-(3/popcount))/popcount)*&MULTIPLIER.;
                        end;
                t1=(rateper-(1.96*stderr)); 
                if (t1<0) then t1=0; 
                LL=put(t1, 8.2);
                UL=put((rateper+(1.96*stderr)), 8.2);
                LL=compress(LL);
                UL=compress(UL);
                n=count;

 *********************** 10. Red Flag ****************************;
 * redflag is the statistical stability indicator based on the    ;
 * relative standard error (RSE, or coefficient of variation).    ;
 * Redflag values created here are converted to images or special ;
 * characters in IBIS-View application XSLTfiles, for instance:   ;
 * (xslt\html\query\module\result\ResultPage.xslt, ...Values.xslt ;
 *****************************************************************;
                if count>0 then do;
                        rse=(stderr/rateper);
                        redflag=put(' ', $12.); 
                        if rse>.3 then redflag=put('Unstable', $12.); 
                        if rse>.5 then redflag=put('Very Unstable', $12.);
                        if stderr=. then redflag=put('Unstable', $12.); 
                end;
                if count<=0 then redflag=put('Unstable', $12.);  *no variance, n=0, rse=div by zero;
        run;

 ************* 11. New Mexico Small Numbers Rule *******************;
 * Suppress cells if the numerator in (1 2 3) AND the denominator   ;
 * is less than 20. For Counts, must run the crude rate code to     ;
 * capture the denominator, but only output the N. ZW's program     ;
 * uses ".A" to identify cells for suppression. I have co-opted     ;
 * his method so I can use the NM logic for cell suppression instead;
 * of the standard IBIS logic. And I need to use ZW's program       ;
 * because it will suppress the table marginals that can be used to ;
 * calculate the suppressed cells. If this code is used, the .def   ;
 * file should have the "NM_" prefix. Needs suppressed_variabes     ;
 * code at the end of this file to work.                            ;
 *******************************************************************;
        data tmp;
                set rate;
                if ((0<n<4) and (popcount<5000)) then do;
                        n = .A;
                        rateper = .A;
                        LL = put('**', 8.0);
                        UL = put('**', 8.0);
                        redflag = put('Suppressed', $12.);
                end;
        proc print data=tmp noobs;  title2 '11. TMP - final dataset to pass to IBIS View app'; 
        run;

--------BoNdArY--------
f out_variable 
# definition for output file
f xml_out_map_file XMLRate10KNDLCLUCL.map
--------BoNdArY--------
f out_detail lbl_not_used__see_xml_out_map_file
 rateper 15.1
 n 15.0
 popcount 15.0
 LL 15.3
 UL 15.3
 redflag 15.0
--------BoNdArY--------
 ***************** 12. suppressed variables *************************;
 * ZW's CGI program must be told how many variables it will need to  ;
 * suppress and which ones they are. NOTE: If the SAS code, above, is;
 * commented out, these lines can be left in the .def file without   ;
 * causing any problems. They will only be used if the SAS code,     ;
 * above is active, OR if the small_num and small_pop parameters     ;
 * are active in the .CFG file, and with non-zero values.            ;
 ********************************************************************;
--------BoNdArY--------
1 suppressed_variables 2
 n
 rateper
--------BoNdArY--------