/************************************************************************/
/*									                  */
/*	Title:		trauma_estimate.do     					*/
/*	Author:		S. Goble, Statitician NTDB  				*/
/*	Project: 	National Sample Project (NSP) 				*/
/*												*/
/*	Purpose: Create statistical estimates for head injuries  		*/
/*  		 by analyzing the weighted data taking into account 		*/
/*			 the sample design.	  		   			*/
/*												*/
/*	Input data:	1.	The final weights and Strata indicators 		*/
/* 					for each incident 				*/
/*				Name: Weights 						*/
/*				Variables needed:		Name: 			*/
/*					Incident ID 		INC_KEY 		*/
/*					Facility ID 		FAC_KEY 		*/
/*					Strata 			STRATA		*/
/*					Weights			WEIGHTS		*/
/*												*/
/*			   2.	Information pertaining to a diagnosis 		*/
/*					made about the trauma incident.		*/
/*				Name: Diagnosis		 				*/
/*				Variables needed: 		Name: 		*/
/*					Incident ID 		INC_KEY 		*/
/*					Diagnosis code	      DCODE			*/
/*												*/
/*			   3.	Includes information about the patient 		*/
/*					and incident demographics.			*/
/*				Name: Demo		 					*/
/*				Variables needed: 	Name: 			*/
/*					Incident ID 	INC_KEY 			*/
/*					Age			AGE				*/
/*					Gender		GENDER	   		*/
/*					Race			RACE				*/
/*												*/
/*			   4.	Includes information pertaining to the outcome 	*/
/*					of the trauma incident.		   		*/
/*				Name: Outcome		 				*/
/*				Variables needed: 		Name: 		*/
/*					Incident ID 		INC_KEY 		*/
/*					Discharge status		DISSTATUS		*/
/*					Hospital lengt of stay	LOS			*/
/*					ICU length of stay	ICU day		*/
/*												*/
/*	Output: Frequency estimate of gender, race and discharge status	*/
/*			Mean estimate of Age, LOS and ICU days			*/
/*												*/ 
/*	Created: April, 2007								*/
/*												*/
/*	**There are 8 weighting strata that are combinations of 		*/
/*	4 Census regions and 2 designated levels of trauma care 		*/
/*	level I or level II. (non-ntdb data not available for 2003 data)	*/
/************************************************************************/

clear
set memory 700000

* Change the following:  'D:\data\NSP\Files_sent_out\Data_Sets' to  '\yourpathname\'; 
/*folder for saving input and output Data_Sets*/

* WEIGHTS 
insheet using D:\data\NSP\Files_sent_out\Data_Sets\Weights.csv 
save D:\data\NSP\Temp\Weights.dta, replace
clear

* VALID TRAUMA RECORDS EXCLUDING HIP FRACTURES  
insheet using D:\data\NSP\Files_sent_out\Data_Sets\Diagnos.csv  
* KEEP VALID TRAUMA D-CODE
drop if dcode<800 
drop if dcode>=960 
drop if dcode>=905 & dcode<910 
drop if dcode>=910 & dcode<925
drop if dcode>=930 & dcode<940 
* EXCLUDE HIP-FRACTURES
drop if dcode>=820 & dcode<821
sort inc_key
* KEEP ONE RECORD PER INCIDENT
by inc_key: gen idkey=1 if _n==1
keep if idkey==1
save D:\data\NSP\Temp\Diagnos.dta, replace
clear 

* DEMOGRAPHICS 
insheet using D:\data\NSP\Files_sent_out\Data_Sets\Demo.csv 
sort inc_key
save D:\data\NSP\Temp\Demo.dta, replace
clear

* OUTCOME
insheet using D:\data\NSP\Files_sent_out\Data_Sets\Outcome.csv 
sort inc_key
save D:\data\NSP\Temp\Outcome.dta, replace
clear

*MERGE FILES
use D:\data\NSP\Temp\Weights.dta
sort inc_key

** KEEP ONLY RECORDS WITH WEIGHT AND VALID TRAUMA CODE
merge inc_key using D:\data\NSP\Temp\Diagnos.dta
keep if _merge==3   
drop _merge 

sort inc_key
merge inc_key using D:\data\NSP\Temp\Demo.dta
keep if _merge==3
drop _merge

sort inc_key
merge inc_key using D:\data\NSP\Temp\Outcome.dta
keep if _merge==3
drop _merge

* REFORMAT FILES FOR PROPORTIONAL ESTIMATES
gen dead=0
replace dead=1 if disstatus=="Dead"
replace dead=. if disstatus==""

gen racecat=0
replace racecat=1 if race=="White, not of Hispanic Origin"
replace racecat=2 if race=="Black"
replace racecat=3 if race=="Asian or Pacific Islander"
replace racecat=4 if race=="Hispanic"
replace racecat=5 if race=="Native American or Alaskan Nati"
replace racecat=6 if race=="Other"
replace racecat=. if race==""

gen male=0
replace male=1 if gender=="Male"
replace male=. if gender==""

drop dcodedescr
save D:\data\NSP\Temp\analyze.dta, replace

* STATISTICAL ANALYSES 
svyset fac_key [pweight=weights], strata(strata)
*ESTIMATING MEAN OF AGE LOS AND ICUDAYS
svy: mean age
svyset fac_key [pweight=weights], strata(strata)
svy: mean los 
svyset fac_key [pweight=weights], strata(strata)
svy: mean icudays

svyset fac_key [pweight=weights], strata(strata)
svy: prop dead 
svyset fac_key [pweight=weights], strata(strata)
svy: prop male
svyset fac_key [pweight=weights], strata(strata)
svy: prop racecat