190 



PAR for cigarettes and lung cancer is 87% indicating that 87% of lung cancers could 

 have been prevented if there were no cigarette smoking. Based on extensive research, 

 we assigned a PAR for each of the 72 substance-abuse related diseases (which are listed 

 in the Appendix). With the help of a medical records coder, we then identified the 

 diagnostic codes associated with these diseases. For each Medicaid discharge that 

 involved any of these primary diagnoses, we multiplied the associated PAR for that 

 disease by the total number of Medicaid days reported for that diagnosis to determine the 

 days attributable to substance abuse. 



Two health problems, AIDS and birth complications proved particularly difficult with 

 respect to estimating their costs resulting from substance abuse. For example, 

 determining AIDS days was difficult, given that an ATDS-related condition (such as 

 pneumocystosis) is often the primary diagnosis and AIDS is only listed secondarily. In 

 fact, only 10,0(X) Medicaid discharges had AIDS as the primary diagnosis, clearly an 

 underestimate. To further complicate matters, not all cases that have AIDS as a 

 secondary diagnosis are hospitalized due to AIDS: someone may be hospitalized for an 

 appendicitis and only coincidentally have AIDS. Thus, these hospital days could not be 

 attributed to AIDS or substance abuse. To get a more precise estimate of AIDS -related 

 hospital days, we identified the primary diagnoses for all Medicaid discharges that had 



measures than PARs, including estimates from large surveys and from medical experts. For example, in 

 the case of AIDS, we used 1992 Center for Disease Control (CDC) surveillance data to estimate the 

 percentage of these cases that were caused by intravenous drug use (IVDU). This surveillance data does 

 not establish causality, it merely categorizes new cases by the risk groups they fall into. In 1992. 55% 

 of new pediatric AIDS cases, and 33% of adult cases fell into the IVDU risk group. We applied these 

 percentages to total reported Medicaid AIDS days to estimate those that were substance abuse-related. 



-15- 



