The Economic Costs of Alcohol and Drug Abuse in the United States - 1992
Health Care Costs - Continued
4.3 Medical Consequences
The cost of alcohol abuse-caused illness treated in hospitals in 1992 is estimated to be about $5.9 billion. Costs for hospital treatment of drug abuse-caused illness are estimated at only $572 million because of a lack of data to attribute specific health consequences to drug abuse.
4.3.1 Hospital Costs
Alcohol and drug abuse problems have been targeted as an important and growing source of hospital-based medical care and expenditures. There are primarily three ways in which alcohol and drug abuse are seen in hospitals:
- Direct and specific consequences of alcohol or drug abuse, such as psychoses, withdrawal, or poisoning, may necessitate hospitalization.
- Alcohol or drug abuse may contribute to other illnesses or injuries that require treatment.
- Alcohol or drug abuse may complicate treatment of other illnesses or injuries, perhaps resulting in longer lengths of stay in the hospital.
Each of these results in distinctly different costs. To the degree that data are available for each, each is counted in this study.
This report relies on the two estimation methodologies employed in the prior studies of the impact of alcohol and drug abuse on hospital care. The first approach is termed the diagnosed population (DP) approach used by Rice et al. (1990), and the second is the illness-specific (IS) approach used in the two previous Research Triangle Institute studies (Cruze et al. 1981; Harwood et al. 1984). Both approaches count hospitalization days that result directly and specifically from alcohol and drug abuse - that is, they count all days for which the primary diagnosis was an alcohol or drug abuse-specific diagnosis (e.g., psychosis, withdrawal, and poisoning). Where the two approaches differ is in their estimation of indirect hospital days attributable to alcohol and drug abuse.
- Diagnosed-Population Approach. The diagnosed population (DP) approach suggests that alcohol and drug abuse problems complicate treatment for certain conditions, resulting in longer lengths of stay for persons with a secondary alcohol or drug abuse disorder than for persons with no secondary alcohol or drug abuse disorder.
- Illness-Specific Approach. The illness-specific (IS) approach is predicated on the evidence that certain health conditions (e.g., forms of cirrhosis) are caused by alcohol and drug abuse, and accordingly, some proportion of the total hospitalizations for conditions that are partially caused by alcohol and drug abuse ought to be attributed to alcohol and drug abuse problems.
Both approaches offer insights and have limitations regarding alcohol and drug abuse-related hospitalizations. The IS approach relies on prior research that has established a causal link between alcohol or drug abuse and illnesses and injuries. To date, the most extensive research available estimates the percentage of mortalities attributable to alcohol by diagnostic category (Stinson et al. 1993). For example, 75 percent of mortalities from malignant neoplasms of the esophagus are attributable to alcohol abuse. Applying the mortality data to morbidity allows us to allocate a share of all health expenditures for each relevant diagnostic category. By inference then, 75 percent of hospitalizations for malignant neoplasms of the esophagus also may be allocated to alcohol abuse. Using mortality-based attribution factors to estimate the extent of morbidity attributable to alcohol consumption generates a conservative bias. This approach will systematically exclude any nonfatal conditions caused by alcohol and will tend to understate the incidence of alcohol-caused conditions that are not always fatal. An additional source of conservative bias associated with the IS approach is that it is limited to the causal research available - the role of alcohol in mortality has been well studied, but the role of drug abuse has not.
The DP approach includes all diagnostic categories and is not constrained by availability of estimates concerning the degree to which alcohol and drug abuse causes other health problems. However, several health conditions have been found to be at least partially caused by alcohol and drug abuse, such that a fraction of all treatment days for those conditions can be attributed to alcohol and drug abuse. The DP approach only calculates the additional days associated with a co-occurring alcohol or drug abuse disorder.
The results of the DP approach are largely complementary to the IS approach, however, and this report combines both approaches in the final estimate of costs. Figure 4.1 demonstrates this logic in graphic form. The vertical axis shows length of stay; the horizontal axis shows the number of discharges. The first two columns, treatment for alcohol and drug problems and illnesses specifically caused by alcohol or drug abuse, are fully counted in the IS and DP approaches and are counted in this report (accounting for a little more than 0.5 million discharges). The third column includes disorders partially caused by alcohol or drug abuse (again, about 0.5 million discharges), for which the IS approach counts the proportion of discharges and days (slashes on the left side of the column). Under the DP approach, additional days are calculated for all discharges where alcohol or drug abuse is listed as a secondary diagnosis. This includes more than 2 million discharges, of which 0.5 million are the same as those used in the IS approach. The final column, on the right, represents another 32 million discharges in 1992, none of whom were hospitalized for conditions related to alcohol or drug abuse.
Figure 4.1 Graphic Depiction of Illness-Specific and Diagnosed-Population Approaches
AOD = alcohol or drug.
LOS = length of stay.
Note: All shaded areas are estimated in this study.
18.104.22.168 Analysis of Hospital Days Under Alternative Approaches
Analyses were conducted using the National Center for Health Statistics' 1990, 1991, and 1992 National Hospital Discharge Survey (NHDS), including data for newborns. The NHDS reports the number of inpatients discharged and days of care for short-stay, non-Federal hospitals by primary and secondary diagnosis as catalogued in the International Classification of Diseases (ICD). A complete list of the ICD version 9 (ICD-9) codes is included in Appendix A.
22.214.171.124.1 Hospital-Based Alcohol and Drug Abuse Treatment
Both the DP and IS approaches estimate the number of days of alcohol and/or drug abuse-specific treatment provided in hospital settings. It is estimated that in 1992, there were more than 524,000 discharges where the treatment provided was primarily for an alcohol or drug abuse diagnosis (see table 4.6). Almost 4.2 million days of care were delivered, of which 65 percent were for alcohol and 35 percent were for drug abuse diagnoses. An episode of hospital-based treatment averaged 8.4 days for primary alcohol diagnoses and 7.3 days for primary drug abuse diagnoses. Most of these hospital treatment episodes were probably short-term detoxification episodes, and a single alcohol or drug abuser could have had more than one, and possibly many, such episodes. The costs for hospital-based care were estimated separately by Harwood et al. (1994) and are discussed in section 4.1.
126.96.36.199.2 Treatment for Conditions Specifically Caused by Alcohol or Drug Abuse
Some illnesses are definitionally and specifically caused by alcohol or drug abuse (e.g., alcoholic cardiomyopathy), where the role of alcohol or drugs was defined by the diagnosticians at the time that a patient was treated. For these cases, all admissions and discharges are allocated to either alcohol or drug abuse as appropriate. It is estimated that there were about 50,000 discharges in 1992 for illnesses that are specifically caused by alcohol abuse, resulting in about 410,000 days of care. It is also estimated that there were fewer than 1,000 discharges for drug abuse disorders, resulting in about 6,000 days of care. These estimates are unreliable, however, because of the small cell sizes in the survey.
Medical treatment for fetal alcohol syndrome and other effects of drug exposure on the fetus or newborn are discussed in depth under special disease groups later in this chapter. As a result, values from table 4.7 for these diagnoses are not included in the total hospital costs associated with alcohol or drug abuse.
188.8.131.52.3 Additional Hospital Days Using Diagnosed-Population Approach
The number of additional hospital days that result from co-occurring alcohol and drug abuse disorders was calculated by comparing lengths of stay and applying the additional length of stay to the number of discharges found among persons with a co-occurring alcohol or drug abuse disorder. To accomplish this, we first excluded discharges for primary diagnoses of alcohol or drug abuse disorders (see the previous section for a discussion of treatment for alcohol or drug abuse problems in the hospital). Second, we segmented the remaining discharges into four groups: (1) patients who had no co-occurring alcohol or drug abuse disorder, (2) patients who had an alcohol but no drug disorder, (3) patients who had a drug but no alcohol disorder, and (4) patients who had both an alcohol and drug disorder.
The lengths of stay for persons with no alcohol or drug disorder diagnosis were subtracted from the lengths of stay for persons with alcohol, drug, or combined alcohol and drug disorders listed as a secondary diagnosis. The result of the subtraction is the estimated additional days per episode resulting from a co-occurring alcohol or drug disorder. The additional days per episode were then multiplied by the number of discharges for each group with a co-occurring alcohol or drug disorder to obtain total additional days of care.
Table 4.8 shows the number of patient discharges, days of care, and average length of stays, distributed by major diagnostic categories, for the four groups previously discussed. The vast majority of discharges (94.1 percent) did not have any secondary alcohol or drug abuse diagnosis, and 92.1 percent of the total days of care were delivered to persons with no secondary alcohol or drug abuse diagnosis. Out of 34.4 million discharges from hospitals, 2.0 million had a secondary (but no primary) diagnosis of alcohol and drug abuse. Of those, 1.1 million had a secondary alcohol abuse diagnosis, 770,000 had a secondary drug abuse diagnosis, and 140,000 had secondary diagnoses of both alcohol and drug abuse. Rice et al.'s (1990) previous estimations of hospital-related stays included mental illness (and alcohol and drug abuse) as a category for analysis of psychiatric and alcohol and drug abuse comorbidity. The present report analyzes costs from psychiatric comorbidity separately below, and this category is not included in the table.
As expected, lengths of stay were longer for patients with co-occurring alcohol and drug problems (7.7 days) compared with stays for other patients (5.6 days). In some disease categories, lengths of stay with a recorded co-occurring alcohol and drug abuse disorder were twice as long as those among noncomorbid patients. However, there were two categories for which length of stay was not longer among comorbid patients. First, lengths of stay for endocrine, nutritional, and metabolic disorders were slightly lower among comorbid than among noncomorbid patients. Second, lengths of stay for patients with injuries and poisoning were lower among comorbid patients than among noncomorbid patients. Additional research should examine whether injuries and poisonings among persons with an alcohol or drug abuse comorbidity are more serious and more likely to result in death, which would explain the shorter lengths of stay. Finally, comorbid patients with digestive problems accounted for more additional days of care (628,000) than any other category.
Because of the extremely large sample sizes, it is expected - but not established - that the substantial differences in length of stay between populations are statistically significant. Future assessments of the additional days of hospital care resulting from co-occurring alcohol and drug abuse disorders would benefit from a more comprehensive analysis that includes other factors that could explain variation in length of stay. For example, age, sex, and sociodemographic status may covary with alcohol and drug abuse and explain part of the additional length of stay observed here (e.g., see Bradley and Zarkin 1996).
184.108.40.206.4 Alcohol and Drug Abuse-Specific Hospital Episodes Using the Illness-Specific Approach
As discussed above, considerable research evidence has accumulated concerning the unique contribution of alcohol to illnesses and injuries that result in death (Stinson et al. 1993). For other conditions (e.g., cirrhosis), a proportion of the cases are caused by alcohol abuse. Research evidence is available regarding the proportion of mortalities related to alcohol consumption. It is assumed here that for any given illness or injury partially caused by alcohol, the proportion of hospitalizations caused by alcohol is equal to the proportion of deaths from that diagnosis that is caused by alcohol. This assumption will tend to understate the role of alcohol in causing hospitalizations because the data on the role of alcohol in mortality from various causes do not reflect alcohol's role in causing nonfatal conditions and understate the role of alcohol in causing diseases that are not always the ultimate cause of death. Despite this bias, the IS approach is the best available technique for estimating the role of alcohol in causing various illnesses and injuries. Unfortunately, similar evidence is not available for a variety of drug abuse-caused illnesses. Table 4.9 shows the number of discharges related to alcohol and drug abuse calculated using the IS approach.
Using the IS approach, it is estimated that more than 511,000 hospital discharges in 1992 were for illnesses or injuries caused by alcohol abuse, for a total of just under 4.0 million days of hospital care. Some conditions that are caused in part by alcohol use (e.g., 75 percent of malignant neoplasms of the esophagus) nonetheless had relatively few discharges. On the other hand, whereas only 10 percent of fractures were estimated to have been caused by alcohol use, the fact that there were many hospitalizations for fractures means that fractures accounted for almost 20 percent of the alcohol-related discharges and 20 percent of alcohol-related days of care.
220.127.116.11.5 Discussion of the Differences Between Diagnosed-Population and Illness-Specific Approaches
Both the DP and IS approaches include the days of care for the treatment of alcohol and drug abuse disorders as well as disorders that are specifically and directly caused by alcohol and drug abuse. In all, this represents 3.1 million days of care for alcohol abuse and about 1.5 million days of care for drug abuse.
Where the IS and DP approaches differ, however, is in their respective assumptions about the contribution of alcohol or drug abuse to other illnesses and injuries treated in hospitals. The DP approach calculates that about 2.3 million additional days of care were brought about as a result of the co-occurrence of alcohol or drug disorders. The IS approach calculates that about 4.0 million days of care were brought about for the treatment of disorders that are partially caused by alcohol abuse (similar causal factors are not available for drug abuse).
Both approaches have substantial advantages and disadvantages. The IS approach relies on a rigorously conducted and reviewed body of research to determine the days of care attributed to alcohol abuse. All of the costs of care for patients whose diseases have been caused by alcohol or drug abuse are counted with the IS approach, whereas only the additional days of care associated with co-occurring alcohol and drug abuse disorders are counted with the DP approach. The DP approach therefore stands to underrepresent the cost of alcohol and drug abuse-induced illness by a wider margin than does the IS approach. However, there are methodological and other limits associated with applying "cause" as required by the IS approach (see discussion in chapter 3). Moreover, similar data on the causal role of drug abuse in morbidity are not yet available.
The DP approach does capture the full array of diagnostic categories, not just those for which specific research on alcohol and drug abuse causality has been reported. Additionally, the fundamental argument that alcohol and drug abuse complicates treatment of other conditions is compelling. The DP approach ought to be extended with multivariate analyses, which would allow for modeling of additional days caused by alcohol or drug abuse, holding constant the effects of other factors such as age, gender, and income. However, the DP approach relies heavily on care providers to accurately document secondary alcohol or drug abuse conditions. Moreover, the DP approach misses cases where alcohol or drug abuse was a cause but for which alcohol or drug abuse would never be diagnosed - such as an individual who does not abuse alcohol but who is injured in an alcohol-related crash.
18.104.22.168 Estimating Costs
The DP and IS approaches are not mutually exclusive and, in fact, are complementary. By removing the diagnoses that are caused by alcohol abuse and drug abuse from the DP approach, the resulting additional days can be directly added to the illness-specific days. The result is an estimated total of 10 million days of care provided in hospitals as a consequence of alcohol and/or drug abuse (see table 4.10). Of those days, about 5.3 million days are for the treatment of illnesses resulting from alcohol problems or for longer lengths of stay associated with co-occurring alcohol disorders. Approximately 0.5 million (512,000) days of care are associated with longer lengths of stay brought about by a co-occurring drug disorder, plus a small number of days attributed to drug-caused illnesses.
The average cost per day of hospital care in 1992 was estimated to be $1,012 for care provided in short-term, non-Federal hospitals. This includes the costs incurred by the hospital and by independent physicians. The hospital per diem, $816, is based on reported expenditures among American Hospital Association members for 1992 (American Hospital Association 1993). The cost per inpatient day for outside physician services is estimated to be $196 and is derived by dividing total physician office receipts for inpatient care by the total inpatient hospital days in the United States in 1992. Applying the average cost per day to the number of hospital days associated with alcohol or drug abuse results in an estimated $5.3 billion for alcohol problems and $518 million for drug abuse problems.
4.3.2 Veterans Affairs and Other Federal Hospitals
No data are available concerning treatment provided in Veterans Affairs and other Federal hospitals of the medical consequences of alcohol and drug abuse. It is therefore assumed that the relative proportion of costs incurred as a result of treating the health consequences of alcohol or drug abuse in non-Federal hospitals also applies to Veterans Affairs and other Federal hospitals. Veterans Affairs and other Federal hospital revenues compose 9.5 percent of total hospital revenues in the United States (Levit et al. 1996). Using this, we can estimate the amount spent in hospitals by multiplying the non-Federal hospital costs attributed to alcohol and drug abuse (5,777 days at $816 per day) by the appropriate factor. The result is an estimated cost of $495 in Federal, short-stay hospital costs, of which $451 million and $44 million are allocated to alcohol and drug abuse, respectively. Further research into the impact of alcohol on the Veterans Affairs and other Federal hospital systems is needed: It is altogether possible that there is a higher incidence of alcohol- and drug-related illness and injury seen in settings that treat veterans compared with the incidence in general acute care settings.
4.3.3 Outpatient Medical Care
Outpatient medical care for alcohol abuse-specific and alcohol abuse-related conditions cost approximately $1.75 billion during 1992.
Alcohol and drug abusers seek help from outpatient medical care providers for alcohol and drug abuse-specific and alcohol and drug abuse-related disorders. To estimate this, the IS approach used in the previous section was also applied to data on outpatient medical care utilization from the annual National Ambulatory Medical Care Survey. The DP approach that was also discussed in the previous section is not appropriate for this data source: NAMCS is, in effect, a survey of outpatient visits that are not longitudinally connected. It is impossible to determine whether persons with a comorbid alcohol and drug abuse problem use more or less outpatient services than persons with no comorbidity. Data for 1990 were analyzed, and the same causal factors employed in the hospital analysis above were used here. The findings are shown in table 4.11. Alcohol and drug abuse treatment service visits are shown, but their cost is included as part of the discussion on specialty alcohol and drug abuse treatment in section 4.2.
It is estimated that about 10.5 million visits were made to outpatient medical care providers for alcohol and drug abuse-specific or alcohol and drug abuse-caused problems, other than primary treatment for the alcohol and drug abuse itself. These visits were for alcohol-specific and alcohol-related disorders only. There were no ambulatory care visits in the data source for drug abuse-specific disorders (that is, disorders definitionally attributable to drug abuse). In addition, as noted earlier, there are no causal data with which to estimate drug abuse-caused disorders. Two-thirds of the outpatient medical services were provided by general practitioners, including family doctors and pediatricians, and about one-third were provided by "other" physicians, including specialists. We estimate the cost per physician visit to have been $166 in 1992. Altogether, the 10.5 million outpatient visits for consequences attributed to alcohol and drug abuse cost about $1.7 billion.
4.3.4 Nursing Homes
This study estimates that $623 million in nursing home costs can be attributed to alcohol abuse. Total nursing home expenditures in 1992 were $62.3 billion (NCHS 1995). Rice et al. (1990) attributed 2 percent of costs in 1985 to alcohol problems and none to drug abuse problems. Harwood et al. (1984) attributed 0.8 percent to alcohol and none to drugs. Rice's attribution factor included cases in which alcohol was a secondary as well as primary diagnosis. However, alcohol may not be the main factor underlying placement in nursing homes. The 1985 National Nursing Home Survey (NNHS) determined that 3.9 percent of nursing home residents have an alcohol or drug abuse problem (NCHS 1989, table A). Further analysis of NNHS suggests that 26 percent of the residents with an alcohol or drug abuse problem may in fact have a primary diagnosis of alcohol or drug abuse, which is 1.0 percent of the entire nursing home population. Applying this estimate of 1.0 percent to the total national expenditures on nursing homes yields $623 million for alcohol and drug abuse combined. All of this is allocated to alcohol abuse.
National expenditures on drugs and medical nondurables were $71.2 billion in 1992 (Levit et al. 1996). This study assumes that the proportion of these expenditures that went for conditions related to or caused by alcohol abuse is comparable to the proportion of inpatient hospital days for illnesses that are caused in whole or in part by alcohol abuse. This is the approach used in the studies by Cruze et al. (1981) and Harwood et al. (1984). Accordingly, 2.2 percent of the total national expenditure on drugs and medical nondurables, or about $1.57 billion, is allocated to alcohol abuse. Again, there is no equivalent basis for ascertaining a similar proportion for drug abuse; instead, pharmaceutical costs due to drug abuse are included in the discussion of specific diseases in this chapter.
4.3.6 Other Health Professionals
Total expenditures for other health professionals in the United States - such as dentists, midwives, and home health technicians - were $42.1 billion in 1992 (NCHS 1995). It is again assumed that 2.2 percent of this total, or $0.9 billion, was attributable to the medical consequences of alcohol. This may be a slight overestimate of the cost attributable to other health professionals, because some of the other health professionals include psychologists and social workers, whose costs have been estimated separately under the category for specialty alcohol and drug abuse treatment providers. However, psychologists and social workers represent only about $4.0 billion out of the $42.1 billion in expenditures for other health professionals (Harwood et al. 1994). Costs due to drug abuse are included in the discussion of specific diseases presented in this chapter.