************************************************************************************************;
*   Program:		maternal_comorbidity_index_code.sas					;
*																								;
*   Purpose:		To compute the maternal comorbidity index based on ICD-9 codes and age	;
*																								;
*   Description:	This code takes in a list of ICD-9 codes and ages and computes a materal;
*					comorbidity index for each individual			;													
*												;
*   Citation:		Pending									;
* 																								;
*																								;
*	definitions:	ICD9file = input datset listing each ICD-9 code for each individual	;
*																								;
*					agefile = input dataset with listing each for each individual;
*																								; 
*                 	ICD9variable = variable name for ICD-9 codes in ICD9file input file;
*																								;
************************************************************************************************;

data age;
	set agefile;

	if age>44 					then weight = 3;
	if 40<=age<=44 					then weight = 2;
	if 35=<age<40 					then weight = 1;

	keep uniqueidentifier age weight;

run;

data conditions;
    set ICDfile(rename=(ICD9variable = id));

	length disease $25;

	disease = 'nopoints';

	if substr(ICD,1,4) = '6425' or substr(ICD,1,4) = '6426'
								then disease = 'severe_pec';

	if ICD = '42822' or ICD = '42823' or ICD = '42832' or 
	ICD = '42833' or ICD = '42842' or ICD = '42843'
								then disease = 'chronic_CHF';

	if '745' <= substr(ICD,1,3) <= '746' or '7470' <= 
	substr(ICD,1,4) <= '7474' or substr(ICD,1,4) = '6485'
								then disease = 'cong_heart';

	if substr(ICD,1,4) = '4160' or substr(ICD,1,4) = '4168' 
	or substr(ICD,1,4) = '4169'
								then disease = 'pulm_HTN';

	if '412' <= substr(ICD,1,3) <= '414'
								then disease = 'chronic_IHD';

	if substr(ICD,1,4) = '2824' or substr(ICD,1,4) = '2826'
								then disease = 'sickle';

	if substr(ICD,1,4) = 'V272' or substr(ICD,1,4) = 'V273'
	or substr(ICD,1,4) = 'V274' or substr(ICD,1,4) = 'V275' 
	or substr(ICD,1,4) = 'V276' or substr(ICD,1,4)= 'V277' 
	or substr(ICD,1,4) = 'V278' or substr(ICD,1,3) = '651'
								then disease = 'multiples';

	if '394' <= substr(ICD,1,3) <= '397' or 
	substr(ICD,1,3) = '424'
								then disease = 'valve';

	if substr(ICD,1,4) = '7100' 
								then disease = 'sle';

	if substr(ICD,1,3) = '042' or substr(ICD,1,3) = 'V08'
								then disease = 'hiv';

	if substr(ICD,1,4) = '6424' 
								then disease = 'mild_PE';

	if '3052' <= substr(ICD,1,4) <= '3059' or 
	substr(ICD,1,3) = '304' or substr(ICD,1,4) = '6483' or 
	substr(ICD,1,4) = '3050'
								then disease = 'drugs';

	if substr(ICD,1,4) = '6410' or substr(ICD,1,4) = '6411'
								then disease = 'previa';

	if '581' <= substr(ICD,1,3) <= '583' or 
	substr(ICD,1,3) = '585' or substr(ICD,1,3) = '587' or
	substr(ICD,1,3) = '588' or substr(ICD,1,4) = '6462'
								then disease = 'chronic_RENAL';

	if '401' <= substr(ICD,1,3) <= '405' or 
	'6420' <= substr(ICD,1,4) <= '6422' 
								then disease = 'pre_HTN';

	if substr(ICD,1,4) = '6542'
								then disease = 'prev_CS';

	if substr(ICD,1,4) = '6423'
								then disease = 'gest_HTN';

	if substr(ICD,1,3) = '291' or substr(ICD,1,3) = '303' or 
	substr(ICD,1,4) = '3050'
								then disease = 'alcohol';

	if substr(ICD,1,3) = '493'
								then disease = 'asthma';

	if substr(ICD,1,3) = '250' or substr(ICD,1,4) = '6480'
								then disease = 'pre_DM';

	if substr(ICD,1,4) = '6427'
								then disease = 'mild_PE_pre_HTN';

	if disease ^= 'nopoints';

run;

proc sort nodupkey data = conditions;
	by uniqueidentifier disease;
run;

*Applying the weights;
data conditionweights;
	set conditions;

	weight = 0;

	if disease = 'severe_pec'			then weight = 5;

	if disease = 'chronic_CHF'			then weight = 5;

	if disease = 'cong_heart'			then weight = 4;

	if disease = 'pulm_HTN'				then weight = 4;

	if disease = 'chronic_IHD'			then weight = 3;

	if disease = 'sickle'				then weight = 3;

	if disease = 'multiples'			then weight = 2;

	if disease = 'valve'				then weight = 2;

	if disease = 'sle'				then weight = 2;

	if disease = 'hiv'				then weight = 2;

	if disease = 'mild_PE'				then weight = 2;

	if disease = 'drugs'				then weight = 2;

	if disease = 'previa'				then weight = 2;

	if disease = 'chronic_RENAL'		        then weight = 1;

	if disease = 'pre_HTN'				then weight = 1;

	if disease = 'prev_CS'				then weight = 1;

	if disease = 'gest_HTN'				then weight = 1;

	if disease = 'alcohol'				then weight = 1;

	if disease = 'asthma'				then weight = 1;

	if disease = 'pre_DM'				then weight = 1;

	if disease = 'mild_PE_pre_HTN'		        then weight = 2;

	keep uniqueidentifier disease weight;

run;


*Establishing preeclampsia and hypertension hierarchies and assigning weights for both
mild preeclampsia and pre-hypertension for ICD code 642.7x;

data conditionweights_pec;
	set conditionweights;

	if disease = 'mild_PE_pre_HTN' then disease = 'severe_pec';
	if disease = 'mild_PE' then disease = 'severe_pec';
	if disease = 'gest_HTN' then disease = 'severe_pec';

run;

proc sort data = conditionweights_pec;
	by uniqueidentifier disease descending  weight;
run;

proc sort nodupkey data = conditionweights_pec;
	by uniqueidentifier disease;
run;


data conditionweights_htn;
	set conditionweights;

	if disease = 'mild_PE_pre_HTN' then weight = 1;
	if disease = 'mild_PE_pre_HTN' then disease = 'pre_HTN';
	if disease ne 'pre_HTN' then delete;

run;

proc sort nodupkey data = conditionweights_htn;
	by uniqueidentifier disease;
run;

data conditionweights2;
	set conditionweights_pec conditionweights_htn;

	if disease = 'severe_pec' and weight = 1 then disease = 'pre_HTN';

run;

*Summing the weights;

data weights;
	set conditionweights2 age;
run;

proc sort nodupkey data = weights;
	by uniqueidentifier disease;
run;

data maternalcomorbidityindex;
	set weights(keep = uniqueidentifier weight);

	by uniqueidentifier;

	if first.uniqueidentifier then comorbidityindex = 0;

	comorbidityindex + weight;

	if last.uniqueidentifier then output;

	keep uniqueidentifier comorbidityindex;

run;