00_MacrosFormats.sas

**************************
Ryan Quan
Data Transformations
Children Center's Data
2014-06-10
**************************
This is an analysis of the effect of prenatal mercury exposure
on childhood obesity using the SAS dataset
"*********" from the Children's Center Data. 

Please run this .sas program first in order to load the
necessary macros and formats used in this analysis.

***********************
FORMATS
**********************;
proc format;
	value hgtertiles
	0 = "First (<1.5ug/L)"
	1 = "Second (1.5-3.7ug/L)"
	2 = "Third (>3.7ug/L)";
	value income
	1 =	"<10,000"
	2 =	"10,000 - 20,000"
	3 =	"20,001 - 30,000"
	4 =	"30,001 - 40,000"
	5 =	"40,001 - 50,000"
	6 =	"50,001 - 60,000"
	7 =	"60,001 - 70,000"
	8 =	"70,001 - 80,000"
	9 =	"80,001 - 90,000"
	10 = ">90,000"
	88,99,. = "Unknown";
	value ethnicity
	3 =	"Dominican/Domin American"
	5 =	"African American";
	value gender
	1 = "Female"
	2 = "Male";
	value yesno
	1 = "Yes"
	2 = "No"
	9,99,8,. = "Unknown";
run;
proc format;
	value hgepa
	0 = "low"
	1 = "high";
run;
proc format;
	value hgtertiles
	0 = "First (<1.5ug/L)"
	1 = "Second (1.5-3.7ug/L)"
	2 = "Third Tertile(>3.7ug/L";
run;
proc format;
	value bmi
	2 = "obese"
	1 = "overweight"
	0 = "normal"
	-1 = "thin"
	-2 = "severely thin";
run;


***********************
MACROS
**********************;

/*Mean*/
%macro mean(var=);
	proc means data=mercury N mean std min max nmiss;
		var &var;
	run;
%mend mean;


/*Freq*/
%macro freq(var=);
	proc freq data=mercury;
		table &var;
	run;
%mend freq;


/*Tabulate*/
%macro tab(data=);
	proc tabulate data = &data missing;
		class newgendr A18_0 B02D_0 A17_0 A08_0 HG_CORD_tertiles;
		tables newgendr A18_0 B02D_0 A17_0 A08_0 HG_CORD_tertiles, n pctn;
		format newgendr gender. A18_0 ethnicity. B02D_0 yesno. A17_0 yesno. A08_0 income. HG_CORD_tertiles hgtertiles.;
	run;
	data mercury_cont;
		set &data;
		if birthwt > 0;
		if A03_0 >=0;
	run;
	proc tabulate data = mercury_cont;
		var birthwt A03_0;
		tables birthwt A03_0, mean std;
	run;
%mend tab;



/*Tabulate w/ BMI Categories*/
%macro tabBMI_cat(data=, BMI_cat =);
	proc tabulate data = &data missing;
		class newgendr A18_0 B02D_0 A17_0 A08_0 HG_CORD_tertiles &BMI_cat;
		tables newgendr A18_0 B02D_0 A17_0 A08_0 HG_CORD_tertiles, &BMI_cat*(n pctn);
		format newgendr gender. A18_0 ethnicity. B02D_0 yesno. A17_0 yesno. A08_0 income. HG_CORD_tertiles hgtertiles.;
	run;
	data mercury_cont;
		set &data;
		if birthwt > 0;
		if A03_0 >=0;
	run;
	proc tabulate data = mercury_cont;
		class &BMI_cat;
		var birthwt A03_0;
		tables birthwt A03_0, &BMI_cat*(mean std);
	run;
%mend tabBMI_cat;



/*Mean + Geomean*/
%macro geomean (data=, class=);
	proc means data = &data;
		class &class;
		var HG_CORD;
	run;
	proc means data = &data;
		class &class;
		var LOGHG_CORD;
		output out = gm mean = mean_log std = std_log lclm=lci uclm=uci replace;
	run;
	data gm_HG_CORD;
		set gm;
		geomean = exp(mean_log);
		std = exp(std_log);
		lcgm = exp(lci);
		ucgm = exp(uci);
	run;
	proc print data = gm_HG_CORD;
	run;
%mend;



/*Regression Models*/
%macro regmod (data = , depvar =);

/* Sort to get correct reference categories*/
proc sort data = &data;
	by descending HG_CORD_TERTILES A17_0 descending A18_0 descending newgendr;
run;

/* Using Tertiles of Cord Blood Mercury*/

/*Univariate*/
proc glm data = &data order = data plots = (diagnostics);
	class HG_CORD_TERTILES;
	model &depvar = HG_CORD_TERTILES / solution;
run;

/*Simple Multivariate*/
proc glm data = &data order = data plots = (diagnostics);
	class HG_CORD_TERTILES newgendr A18_0;
	model &depvar = HG_CORD_TERTILES birthwt A18_0 newgendr / solution;
	format HG_CORD_TERTILES hgtertiles. newgendr gender. A18_0 ethnicity.;
run;

/*Big Multivariate*/
proc glm data = &data order = data plots = (diagnostics);
	class HG_CORD_TERTILES newgendr A18_0 A17_0;
	model &depvar = HG_CORD_TERTILES birthwt A18_0 A17_0 newgendr A03_0 A08_0 / solution;
	format HG_CORD_TERTILES hgtertiles. A18_0 ethnicity. A17_0 yesno. newgendr gender.;
run;

/*Using Log of Cord Blood Mercury*/

/*Univariate*/
proc glm data = &data order = data plots = (diagnostics);
	model &depvar = LOGHG_CORD / solution;
run;

/*Simple Multivariate*/
proc glm data = &data order = data plots = (diagnostics);
	class newgendr A18_0;
	model &depvar = LOGHG_CORD birthwt A18_0 newgendr / solution;
	format newgendr gender. A18_0 ethnicity.;
run;

/*Big Multivariate*/
proc glm data = &data order = data plots = (diagnostics);
	class newgendr A18_0 A17_0;
	model &depvar = LOGHG_CORD birthwt A18_0 A17_0 newgendr A03_0 A08_0 / solution;
	format A18_0 ethnicity. A17_0 yesno. newgendr gender.;
run;
%mend;

01_SummaryStatistics.sas

**************************
Ryan Quan
Summary Statistics
Children Center's Data
2014-06-01
**************************
This is an analysis of the effect of prenatal mercury exposure
on childhood obesity using the SAS dataset
"*******" from the Children's Center Data. 

PREREQUSITE: 00_MasterProgram.sas

GOAL: Running PROC MEANS and PROC FREQ for summary statistics 
of our variables of interest.

**********************
READING IN MERCURY DATA
**********************;
libname rq "&root\Data\";
data mercury;
	set rq.AR_JG_2014002a;
run;

***********************
SUMMARY STATISTICS
***********************;


***********************
EXPOSURE VARIABLES
***********************;
ODS RTF FILE = "&root\Results\01_Mercury_PROCMEANS.rtf";
ODS NOPROCTITLE;

%mean(var = HG_BLOOD)	/*Maternal Blood Hg*/
%mean(var = HGMFLAG)	/*CDC Flag Mother Hg NEW 2005*/
%mean(var = HG_CORD)	/*Cord Blood Hg*/
%mean(var = HGCFLAG)	/*CDC Flag Cord Hg NEW 2005*/
%mean(var = PB_FLAG)	/*Lead Flag Cord*/
/****************************
OUTCOME VARIABLES 5 Years Old
*****************************/

%mean (var = BMI60)		/*Body Mass index value*/
%mean (var = BMIZ60)	/*Z-score BMI for age*/

/****************************
OUTCOME VARIABLES 7 Years Old
*****************************/
%mean (var = BMI84)		/*Body Mass index value*/
%mean (var = BMIZ84)	/*Z-score BMI for age*/

%mean (var = bcacfatp7)	/*Fat % child analyzer*/
%mean (var = bcacffm7)	/*Fat-Free Mass*/
%mean (var = BCACTBW7)	/*Total Body Weight*/
%mean (var = BCACIWB7)	/*Whole body impedence*/
%mean (var = BCWC_C7)	/*Waist circumference in cm*/

/****************************
OUTCOME VARIABLES 9 Years Old
*****************************/
%mean (var = BMI9)		/*Body Mass index value*/
%mean (var = BMIZ9)		/*Z-score BMI for age*/

%mean (var = bcacfatp9)	/*Fat % child analyzer*/
%mean (var = bcacffm9)	/*Fat-Free Mass*/
%mean (var = BCACTBW9)	/*Total Body Weight*/
%mean (var = BCACIWB9)	/*Whole body impedence*/
%mean (var = BCWC_C9)	/*Waist circumference in cm*/

/****************************
OUTCOME VARIABLES Tanner Stage
*****************************/
%mean (var = BMI_T)		/*Body Mass index value*/
%mean (var = BMIZ_T)	/*Z-score BMI for age*/

%mean (var = TANWC1)	/*Waist Circumference*/
%mean (var = TANWC2)
%mean (Var = TANWC3);

ODS RTF close;


***********************
SUMMARY STATISTICS
***********************
Running PROC FREQ for summary
statistics of our variables of interest.


***********************
EXPOSURE VARIABLES
***********************;
ODS RTF FILE = "&root\Results\01_Mercury_PROCFREQ.rtf";
ODS NOPROCTITLE;

%freq(var = HG_BLOOD)	/*Maternal Blood Hg*/
%freq(var = HGMFLAG)	/*CDC Flag Mother Hg NEW 2005*/
%freq(var = HG_CORD)	/*Cord Blood Hg*/
%freq(var = HGCFLAG)	/*CDC Flag Cord Hg NEW 2005*/
%freq(var = PB_FLAG)	/*Lead Flag Cord*/
/****************************
OUTCOME VARIABLES 5 Years Old
*****************************/

%freq (var = BMI60)		/*Body Mass index value*/
%freq (var = BMIZ60)	/*Z-score BMI for age*/

/****************************
OUTCOME VARIABLES 7 Years Old
*****************************/
%freq (var = BMI84)		/*Body Mass index value*/
%freq (var = BMIZ84)	/*Z-score BMI for age*/

%freq (var = bcacfatp7)	/*Fat % child analyzer*/
%freq (var = bcacffm7)	/*Fat-Free Mass*/
%freq (var = BCACTBW7)	/*Total Body Weight*/
%freq (var = BCACIWB7)	/*Whole body impedence*/
%freq (var = BCWC_C7)	/*Waist circumference in cm*/

/****************************
OUTCOME VARIABLES 9 Years Old
*****************************/
%freq (var = BMI9)		/*Body Mass index value*/
%freq (var = BMIZ9)	/*Z-score BMI for age*/

%freq (var = bcacfatp9)	/*Fat % child analyzer*/
%freq (var = bcacffm9)	/*Fat-Free Mass*/
%freq (var = BCACTBW9)	/*Total Body Weight*/
%freq (var = BCACIWB9)	/*Whole body impedence*/
%freq (var = BCWC_C9)	/*Waist circumference in cm*/

/****************************
OUTCOME VARIABLES Tanner Stage
*****************************/
%freq (var = BMI_T)		/*Body Mass index value*/
%freq (var = BMIZ_T)	/*Z-score BMI for age*/

%freq (var = TANWC1)	/*Waist Circumference*/
%freq (var = TANWC2)
%freq (Var = TANWC3);

ODS RTF close;

02_DroppingObservations.sas

**************************
Ryan Quan
Dropping Observations
Children Center"s Data
2014-06-09
**************************
This is an analysis of the effect of prenatal mercury exposure
on childhood obesity using the SAS dataset
"AR_JG_2014002a.sas7bdat" from the Children"s Center Data. 

PREREQUSITE: 00_MacrosFormats.sas

GOAL: To drop observations from our analysis that do not meet
the following inclusionary criteria:
	*poor sample quality
	*insufficient sample amount
	*below limit of detection
	*clotted samples (from both lead and mercury flags)
	*missing measurement for cord blood mercury

**********************
READING IN MERCURY DATA
**********************;
libname rq "&root\Data";
data mercury;
	set rq.AR_JG_2014002a;
run;


*****************************
Assigning and Dropping Variables
with Exclusionary Criteria
*****************************;

ODS RTF FILE = "&root\Results\02_Mercury_MISSING.rtf";
ODS NOPROCTITLE;
%mean(var = HG_CORD)	/*N=503*/

/* HG_CORD - Cord blood mercury
-9.00	Missing Sample
-8.00	Unsatisfactory Sample
-7.00	QNS
.07	< LOD (0.14)
.10	< LOD (0.2) */
data mercury;
	set mercury;
	if HG_CORD = -9.0 then delete;
run;
/*19 observations dropped*/

data mercury;
	set mercury;
	if HG_CORD = -8.0 then delete;
run;
/*1 observations dropped*/

data mercury;
	set mercury;
	if HG_CORD = -7.0 then delete;
run;
/*0 observation dropped*/

data mercury;
	set mercury;
	if HG_CORD < 0.07 & HG_CORD > . then delete;
run;
/* 0 observations dropped*/

/* note that if we used a LOD <.10, we would
have dropped 1 observation that fell in between
0.07 and 0.10. */

data mercury;
	set mercury;
	if HG_CORD = . then delete;
run;
/* 224 observations dropped*/

%mean(var = HG_CORD);	/*N = 483*/

/* HGCFLAG - CDC flag cord Hg NEW 2005
5	clotted
6	small clots
7	results repeated & confirmed
9	sample fine */
data mercury;
	set mercury;
	if HGCFLAG =5 then delete;
run;
/*N=6; 0 observations dropped*/

data mercury;
	set mercury;
	if HGCFLAG =6 then delete;
run; 
/*N=13; 0 observations dropped*/

%mean(var = HG_CORD); 	/*N = 483*/

*PB_FLAG	
1	< LOD (0.6 or 0.3)
2	unsatisfactory specimen
3	QNS
4	no flag
5	clotted
9	missing;
proc freq data = mercury;	/*no overlap in flags*/
	tables PB_FLAG*HGCFLAG/missing;
run;

data mercury;
	set mercury;
	if PB_FLAG = 2 then delete;
run; 
/*N=5; 2 observations dropped*/

data mercury;
	set mercury;
	if PB_FLAG = 5 then delete;
run;
/*N=34; 25 observations dropped*/	

%mean(var = HG_CORD);	/*N = 456*/

*********************************
Uncomment block below to exclude mothers
with invalid blood samples
*********************************;

/*
*HG_BLOOD - Mom blood mercury
-9.00	Missing Sample
-8.00	Unsatisfactory Sample
-7.00	QNS
.10	< LOD (0.2);
data mercury;
	set mercury;
	if HG_BLOOD =-9 then delete;
	if HG_BLOOD =-8 then delete;
	if HG_BLOOD =-7 then delete;
	if HG_BLOOD >=0 & HG_BLOOD <.10 then delete;
run;
%mean(var = HG_CORD)	*N = 454;

*HGMFLAG - CDC flag mother Hg NEW 2005
5	clotted
6	small clots
7	results repeated & confirmed
9	sample fine;
data mercury;
	set mercury;
	if HGMFLAG =5 then delete;
	if HGMFLAG =6 then delete;
run;
%mean(var = HG_CORD)	*N = 454;	
*/

******************************
SUMMARY VISUALIZATIONS
******************************;
proc univariate data = mercury;
	var HG_CORD;
	histogram;
run;

proc sgplot data = mercury;
	hbox HG_CORD;
run;

/*Save formatted dataset to library*/
data rq.mercury_nomissing;
	set mercury;
run;
ODS RTF close;

03_DataTransformations.sas

**************************
Ryan Quan
Data Transformations
Children Center's Data
2014-06-10
**************************
This is an analysis of the effect of prenatal mercury exposure
on childhood obesity using the SAS dataset
"********" from the Children's Center Data. 

PREREQUISITE: 00_MacrosFormats.sas

GOAL: To create new categorical/dichotomous
variables and log-transform the dependent variable
to create a dataset that will be ready for analysis.

We will also split the dataset by age, retaining only observations
with anthropometric measures (BMI Z-scores).

**********************
READING IN MERCURY DATA
**********************;
libname rq "&root\Data";
data mercury;
	set rq.mercury_nomissing;
run;

*************************
CREATING CATEGORICAL VARIABLES
*************************;
ODS RTF FILE = "&root\Results\03_Mercury_CategoricalVars.rtf";
ODS NOPROCTITLE;

data mercury;	/*EPA Cut-offs for young children*/
	set mercury;
	if HG_CORD < 5.8 then HG_CORD_EPA = 0;
	if HG_CORD >= 5.8 then HG_CORD_EPA = 1;
	format HG_CORD_EPA hgepa.;
run;

%mean (var = HG_CORD_EPA);
%freq (var = HG_CORD_EPA);

proc univariate data = mercury;
	var HG_CORD;
	output out=HG_CORD_tertiles pctlpts=33 77 pctlpre=HG_CORDP;
run;
proc print data = HG_CORD_tertiles;
run;

data mercury;
	set mercury;
	if HG_CORD >= 0 & HG_CORD <1.5 then HG_CORD_tertiles = 0;
	if HG_CORD >= 1.5 & HG_CORD <3.7 then HG_CORD_tertiles = 1;
	if HG_CORD >= 3.7 then HG_CORD_tertiles = 2;
	format HG_CORD_tertiles hgtertiles.;
run;

%mean (var = HG_CORD_tertiles);
%freq (var = HG_CORD_tertiles);

data mercury;
	set mercury;
	array BMI{3} BMIZ60 BMIZ84 BMIZ9;
	array BMI_cat{3} BMIZ60_cat BMIZ84_cat BMIZ9_cat;
	do i = 1 to 3;
		BMI_cat{i} = .;
		if BMI{i} >= 2 then BMI_cat{i} = 2;
		if BMI{i} < 2 & BMI{i} >= 1 then BMI_cat{i} = 1;
		if BMI{i} < 1 & BMI{i} > -1 then BMI_cat{i} = 0;
		if BMI{i} <= -1 & BMI{i} > -2 then BMI_cat{i} = -1;
		if BMI{i} <= -2 & (BMI{i} > .) then BMI_cat{i} = -2;
	end;
run;

data mercury;
	set mercury;
		format BMIZ60_cat bmi.;
		format BMIZ84_cat bmi.;
		format BMIZ9_cat bmi.;
run;

%mean (var = BMIZ60_cat BMIZ84_cat BMIZ9_cat);
%freq (var = BMIZ60_cat BMIZ84_cat BMIZ9_cat);

ODS RTF close;

*************************************************
LOG TRANSFORMATION OF HG_CORD
*************************************************;
data mercury;
	set mercury;
	LOGHG_CORD = LOG(HG_CORD);
run;

*************************************************
Saving completed / subsetted datasets to library
************************************************;

/*N = 456*/
data rq.mercury_complete;
	set mercury;
run;

/*N = 321*/
data rq.mercury_05;
	set mercury;
	if BMIZ60 > .;
run;

/*N = 326*/
data rq.mercury_07;
	set mercury;
	if BMIZ84 > .;
run;

/*N = 176*/
data rq.mercury_09;
	set mercury;
	if BMIZ9 > .;
run;

04_CohortDescriptives.sas

**************************
Ryan Quan
Table 1
Children Center's Data
2014-06-10
**************************
This is an analysis of the effect of prenatal mercury exposure
on childhood obesity using the SAS dataset
"********" from the Children's Center Data. 

PREREQUISITES: 00_MacroFormats.sas, 03_DataTransformations.sas

GOAL: To find out the origin of each child from each follow-up
group. This is done to answer the following types of questions:
	* how many observations were followed up at ages 5, 7, and 9?
	* how many observations were followed up at ages 7, but not ages 5?
	* how many observations were followed up at ages 5 and 9, but not 7?
	
**********************
READING IN MERCURY DATA
**********************;
libname rq "&root\Data";
data mercury;
	set rq.mercury_complete;
run;
data mercury_05;
	set rq.mercury_05;
run;
data mercury_07;
	set rq.mercury_07;
run;
data mercury_09;
	set rq.mercury_09;
run;

ODS RTF FILE = "&root\Results\04_Mercury_Exclusion.rtf";

title "Follow-Up Numbers for Exclusionary Chart";

/*Children followed up at age 7 years*/
proc means data = mercury_07;
	var bmiz60;
	title "7Y followed up from 5Y";
run;
data mercury_00to07;
	set mercury_07;
	if bmiz60 =.;
run;

proc means data = mercury_00to07;
	var sid;
	title "7Y followed up First Time";
run;


/*Children followed up at age 9 Years */
proc means data = mercury_09;
	var bmiz84;
	title "9Y followed up from 7Y";
run;

data mercury_05to09;
	set mercury_09;
	if bmiz84 =.;
run;
proc means data = mercury_05to09;
	var bmiz60;
	title "9Y followed up from 5Y";
run;

data mercury_00to09;
	set mercury_09;
	if bmiz84 =. & bmiz60 =.;
run;
proc means data = mercury_00to09;
	var sid;
	title "9Y followed up First Time";
run;

ODS RTF close;

05_Table1_BMI.sas

**************************
Ryan Quan
Table 1 - BMI Descriptives
Children Center's Data
2014-06-16
**************************
This is an analysis of the effect of prenatal mercury exposure
on childhood obesity using the SAS dataset
"*********" from the Children's Center Data. 

PREREQUISITE: 00_MacrosFormats.sas, 03_DataTransformations

GOAL: To tabulate the counts/pct for categorical
variables and mean/std for continuous variables for the total
cohort and children with anthropometric measures at (5, 7, 9). 

**********************
READING IN MERCURY DATA
**********************;
libname rq "&root\Data";
data mercury;
	set rq.mercury_complete;
run;
data mercury_05;
	set rq.mercury_05;
run;
data mercury_07;
	set rq.mercury_07;
run;
data mercury_09;
	set rq.mercury_09;
run;
***********************
MACROS
**********************;

ODS RTF FILE = "&root\Results\05_Mercury_Table1_Complete.rtf";
title "Table 1 - Total Cohort";

proc means data = mercury;
	var bmiz60 bmiz84 bmiz9;
run;

%tab(data = mercury);

/*Annual Household Income at Child's Age*/
proc tabulate data = mercury;
	class A08_60 A08_84;
	tables A08_60 A08_84, n pctn;
	format A08_60 income. A08_84 income.;
	title "Annual Household Income at Child's Age";
run;

ODS RTF close;

ODS RTF FILE = "&root\Results\05_Mercury_Table1_05.rtf";
title "Table 1 - Child at Age 5";

%tab(data = mercury_05);
/*Age 5 does NOT have measurements other than BMI*/

proc means data = mercury_05;
	var HG_CORD bmi60 bmiz60;
run;

proc univariate data = mercury_05;
	var HG_CORD bmi60 bmiz60;
	histogram;
	qqplot;
run;

ODS RTF close;

ODS RTF FILE = "&root\Results\05_Mercury_Table1_07.rtf";
title "Table 1 - Child at Age 7";

%tab(data = mercury_07)

proc means data = mercury_07;
	var HG_CORD bmi84 bmiz84 bcacfatp7 bcacfatm7 bcacffm7 bcactbw7 bcaciwb7;
run;

/*missing value in bcwc_c7*/

data mercury_07i;
	set mercury_07;
	if bcwc_c7 > 0;
run;

proc means data = mercury_07i;
	var bcwc_c7;
run;

proc univariate data = mercury_07;
	var HG_CORD bcacfatp7 bcacfatm7 bcacffm7 bcactbw7 bcaciwb7;
	histogram;
	qqplot;
run;
proc univariate data = mercury_07i;
	var bcwc_c7;
	histogram;
	qqplot;
run;

ODS RTF close;

ODS RTF FILE = "&root\Results\05_Mercury_Table1_09.rtf";
title "Table 1 - Child at Age 9";

%tab(data = mercury_09);

proc means data = mercury_09;
	var HG_CORD bcacfatp9 bcacfatm9 bcacffm9 bcactbw9 bcaciwb9;
run;

/*missing value in bcwc_c9*/

data mercury_09i;
	set mercury_09;
	if bcwc_c9 > 0;
run;

proc means data = mercury_09i;
	var bcwc_c9;
run;

proc univariate data = mercury_09;
	var HG_CORD bcacfatp9 bcacfatm9 bcacffm9 bcactbw9 bcaciwb9;
	histogram;
	qqplot;
run;
proc univariate data = mercury_09i;
	var bcwc_c9;
	histogram;
	qqplot;
run;

ODS RTF close;



ODS RTF FILE = "&root\Results\05_Mercury_Table1_BMIcat.rtf";
title "Descriptives by BMI Categories";
%tabBMI_cat(data = mercury_05, BMI_cat = BMIZ60_cat);
%tabBMI_cat(data = mercury_07, BMI_cat = BMIZ84_cat);
%tabBMI_cat(data = mercury_09, BMI_cat = BMIZ9_cat);
ODS RTF close;

06_Table2_Geomeans.sas

**************************
Ryan Quan
Table 2 - Geomeans
Children Center's Data
2014-06-27
**************************
This is an analysis of the effect of prenatal mercury exposure
on childhood obesity using the SAS dataset
"*********" from the Children's Center Data. 

PREREQUISITE: 00_MacrosFormats.sas, 03_DataTransformations.sas

GOAL: To report the arithmetic and geometric means for cord
blood mercury levels from the total cohort and follow-up groups.
We also stratify by gender, ethnicity, and public assistance.

**********************
READING IN MERCURY DATA
**********************;
libname rq "&root\Data";
data mercury;
	set rq.mercury_complete;
run;
data mercury_05;
	set rq.mercury_05;
run;
data mercury_07;
	set rq.mercury_07;
run;
data mercury_09;
	set rq.mercury_09;
run;

ODS RTF FILE = "&root\Results\06_Mercury_Geomeans.rtf";
title "Geometric Means for HG_CORD";

/*Total Cohort*/
%geomean(data = mercury);
%geomean(data = mercury, class = newgendr);	/*gender*/
%geomean(data = mercury, class = A18_0);	/*ethnicity*/
%geomean(data = mercury, class = A17_0);	/*public assistance*/

/*Age 5*/
%geomean(data = mercury_05)
%geomean(data = mercury_05, class = newgendr);	/*gender*/
%geomean(data = mercury_05, class = A18_0);		/*ethnicity*/
%geomean(data = mercury_05, class = A17_0);		/*public assistance*/
%geomean(data = mercury_05, class = BMIZ60_cat);/*BMI categorires*/

/*Age 7*/
%geomean(data = mercury_07)
%geomean(data = mercury_07, class = newgendr);	/*gender*/
%geomean(data = mercury_07, class = A18_0);		/*ethnicity*/
%geomean(data = mercury_07, class = A17_0);		/*public assistance*/
%geomean(data = mercury_07, class = BMIZ84_cat);/*BMI categorires*/

/*Age 9*/
%geomean(data = mercury_09)
%geomean(data = mercury_09, class = newgendr);	/*gender*/
%geomean(data = mercury_09, class = A18_0);		/*ethnicity*/
%geomean(data = mercury_09, class = A17_0);		/*public assistance*/
%geomean(data = mercury_09, class = BMIZ9_cat);	/*BMI categorires*/


ODS RTF close;

07_LinearModels.sas

**************************
Ryan Quan
Table 1
Children Center's Data
2014-06-29
**************************
This is an analysis of the effect of prenatal mercury exposure
on childhood obesity using the SAS dataset
"***********" from the Children's Center Data. 

PREREQUISITE: 00_MacrosFormats.sas, 03_DataTransformations.sas

GOAL: To construct the following linear models for follow-up groups
ages 5, 7, and 9:
	* CRUDE: 	BMI ~ HG_CORD
	* SIMPLE: 	BMI ~ HG_CORD + BIRTHWEIGHT + ETHNICITY + GENDER
	* BIG:		BMI ~ HG_CORD + BIRTHWEIGHT + ETHNICITY + GENDER 
						+ EDUCATION + PUBLIC ASSISTANCE + INCOME

We construct one set of models using the log-transformed HG_CORD and
another using HG_CORD_TERTILES.

We also provide a correlation matrix for mom blood mercury and
cord blood mercury and for cord blood lead and cord blood mercury.


**********************
READING IN MERCURY DATA
**********************;
libname rq "&root\Data";
data mercury;
	set rq.mercury_complete;
run;
data mercury_05;
	set rq.mercury_05;
run;
data mercury_07;
	set rq.mercury_07;
run;
data mercury_09;
	set rq.mercury_09;
run;



ODS RTF FILE = "&root\Results\07_Mercury_HGBloodCorr.rtf";
title "Correlation between Mother's Blood Mercury and Cord Blood Mercury";
data mercury;
	set mercury;
	if HG_BLOOD =-9 then delete;
	if HG_BLOOD =-8 then delete;
	if HG_BLOOD =-7 then delete;
	if HG_BLOOD >=0 & HG_BLOOD <.10 then delete;
run;

proc corr data = mercury;
	var HG_BLOOD HG_CORD;
run;
ODS RTF close;

ODS RTF FILE = "&root\Results\07_Mercury_HGLEADCorr.rtf";
title "Correlation between Cord Blood Mercury & Cord Blood Lead";
proc corr data = mercury;
	var HG_CORD LEAD BMIZ60 BMIZ84 BMIZ9;
run;
ODS RTF close;

ODS RTF FILE = "&root\Results\07_LinearModel_Age5.rtf";
title "Linear Models for Age 5 Follow-Ups";
%regmod (data = mercury_05, depvar = BMIZ60);
ODS RTF close;

ODS RTF FILE = "&root\Results\07_LinearModel_Age7.rtf";
title "Linear Models for Age 7 Follow-Ups";
%regmod (data = mercury_07, depvar = BMIZ84);
ODS RTF close;

ODS RTF FILE = "&root\Results\07_LinearModel_Age9.rtf";
title "Linear Models for Age 9 Follow-Ups";
%regmod (data = mercury_09, depvar = BMIZ9);
ODS RTF close;

MasterProgram.sas

*********************************************************
*********************************************************
PROGRAM NAME: MasterProgram.sas
DATE CREATED: June 29, 2014
AUTHOR: Ryan Quan
PURPOSE: Runs the programs for Mercury/Obesity Analysis
INPUTS:	00_MacrosFormats.sas
		01_SummarySatistics.sas
		02_DroppingObservations.sas
		03_DataTransformations.sas
		04_CohortDescriptives.sas
		05_Table1_BMI.sas
		06_Table2_Geomeans.sas
		07_LinearModels.sas
OUTPUTS: See individual SAS programs for outputs
*********************************************************
*********************************************************

Before we begin, please set your working directory by changing
the filepath below as defined by the macro "root":

*********************************************************
				SET WORKING DIRECTORY
*********************************************************;
%let root= \\psf\Dropbox\Mercury_Obesity;


*********************************************************
					DESCRIPTION
*********************************************************

To run this program, please create the following folders in
your working directory:

	* Data
	* Results
	* Programs

For this program to run, the .sas programs listed above in the "INPUTS" 
section must be placed in your "Programs" folder. The dataset entitled
"AR_JG_2014002a.sas7bdat" must also be placed in your "Data" folder.

All dataset outputs will be created in the "Data" folder.

All ODS outputs will be be created in the "Results" folder.

*********************************************************
					PROGRAMS
*********************************************************;

%include "&root\Programs\00_MacrosFormats.sas";
%include "&root\Programs\01_SummarySatistics.sas";
%include "&root\Programs\02_DroppingObservations.sas";
%include "&root\Programs\03_DataTransformations.sas";
%include "&root\Programs\04_CohortDescriptives.sas";
%include "&root\Programs\05_Table1_BMI.sas";
%include "&root\Programs\06_Table2_Geomeans.sas";
%include "&root\Programs\07_LinearModels.sas";