Dogwood Center, Princeton, NJ
Copyright 1996: Dawn Day and Reuben Cohen


April 1996

RACE AND THE SPREAD OF HIV/AIDS RELATED TO INJECTION DRUG USE


Dawn Day and Reuben Cohen

Consider this statement:  In 1994, there were over 14,400 new injection-related AIDS cases among blacks and 7,200 new injection-related AIDS cases among whites.

A person reading this statement might come to a couple conclusions:

    • the injection-related AIDS epidemic is claiming a disproportionate number of victims among blacks.

    • blacks inject drugs more than whites.

In other words, these figures seem to be describing what individual whites and blacks are doing; they do not necessarily raise any question in the reader’s mind about the social context in which blacks are becoming infected with HIV/AIDS in such disproportionately large numbers.

However, suppose we change the statement to read:  Among person who inject drugs, African Americans are four times (or five times or eight times) as likely as whites to be diagnosed with AIDS.

A person reading this statement may well begin to ask why.  The behavior – injecting drugs – seems to be the same; why is the outcome so different?  

So in order to raise the question of systemic as well as individual problems, we need to look at the spread of injection-related HIV/AIDS in the context of who is using drugs.

We can explore the question of the relationship between HIV/AIDS, injection drug use, and race using data collected in quite different ways.  We can look at:

    • the number of new AIDS cases in relation to the number of persons who inject drugs. 

    • the prevalence of HIV among needle users in drug treatment centers across the United States.

The only measure of the HIV/AIDS epidemic among injecting drug users that has some claim to representing the United States as a whole is the number of new AIDS cases related to estimates of the number of persons who inject drugs.  It is to that measure that we now turn.


New Injection-Related AIDS Cases and Persons Who Inject Drugs
New AIDS Cases, by race
Each year the Centers for Disease Control and Prevention (CDC) publishes a count of the number of new AIDS cases for the previous year. That a patient has AIDS is determined by the attending physician using the uniform surveillance case definition developed by the CDC.  The physician is responsible for reporting the case to the local health authorities.

Although the completeness of reporting of diagnosed AIDS cases to state and local health departments varies by geographic region and patient population, studies conducted by state and local health departments indicate that reporting of AIDS cases in most areas of the United States is more than 85 percent complete.(1)  In other words, we can reasonably expect that the count of new injection-related AIDS cases, as reported by the CDC in any given year, is about 10 percent below what it would be in a world of perfect information.

For a substantial number of the cases (25 percent in the case of blacks and 10 percent in the case of whites in 1994), the exposure category is unknown. (2)   This means that the count of new injection-related AIDS cases as reported by CDC, particularly for blacks, is very likely too low.

The CDC count of new injection-related AIDS cases each year can be seen as undercounts, but they are the best count we have of persons infected with HIV/AIDS at a particular point in time.  This is because HIV/AIDS in its early stages has no distinctive symptoms. By the time people have AIDS, they are having one or more a series of identifiable conditions; they know they are sick, and their doctors know they are sick.

In 1994, the counts of new injection-related AIDS cases were as follows:(3)
            Black   14,440
            White     7,168

Our best information is that the people who got AIDS in 1994 probably became infected with HIV sometime during 1984-86.  And in the best of all possible worlds, we would look at 1994 AIDS cases compared to the number of persons who were injecting drugs in 1984-86.  Unfortunately good injection drug use data are not available for that period.  So we do the next best thing.  We use the years 1991-1994, an interval during which there is a reasonable amount of good information.

An index of injection drug use by race
Our inquiry into the racial composition of the group of persons who inject drugs starts with the estimates of the number of persons living in households who injected drugs as reported in the National Household Survey on Drug Abuse during the period 1991-1994. Then we make an adjustment to represent the number of injecting drug users among the nonhousehold population.  The nonhousehold adjustment is based on a 1991 survey of drug use among the nonhousehold population in the D.C. metropolitan area, plus relevant U.S. Census data.  These data have their limits, but the resulting estimates, stated as indices, are quite usable and useful for comparing white and black drug use. See the Technical Note for the details of this analysis.

With the index for white injecting drug users set equal to 100, the available data suggest that the index for blacks is close to 33.  To allow for changes over time and other uncertainties in available data, we feel that a range of 50 to 25 would almost certainly contain the true index number for blacks.  Thus we conclude that there are least twice as many whites as blacks who inject drugs, and the true ratio may be as high as four times as many whites as blacks.

New AIDS cases and injection drug use by race
Using these index numbers, we can set up the formula for calculating the relationship of interest:

            14,440 black AIDS cases/ black index number
            7,168 white AIDS cases/ 100 (white index number)

When we insert the black index number of 50 into the formula, we get the result that among persons who inject drugs, blacks are about four times as likely as whites to get injection-related AIDS.  When we insert the black index number of 25 into the formula, the estimate is that blacks are eight times as likely as their white counterparts to get injection-related AIDS.

In other words, whether we pick the lower or higher index number for black injection drug use, the conclusion is the same: among persons who inject drugs, blacks are several times more likely than whites to get AIDS.

New AIDS cases and an index based on estimates of injection drug use are not the only way to examine the spread of HIV/AIDS among persons who inject drugs. 


HIV Among Persons in Drug Treatment
 We now turn to a data set that provides information on HIV (not AIDS) status and looks at persons entering drug treatment in selected sites (not the entire population of persons who inject drugs).

The CDC collects data on the HIV status of persons who enter participating drug treatment centers and who report injecting illicit drugs during the previous year.  HIV status is determined by blood samples.

In 1991 and 1992, the data reported here were collected in 35 metropolitan areas and in 78 treatment programs that submitted at least 100 blood specimens in 1991-1992.  Over half of the drug treatment centers offered methadone maintenance or methadone detoxification; programs at other centers included drug-free treatment, cocaine treatment, or therapeutic community programs.

The prevalence of HIV among persons in these programs (seroprevalence), ranged by center, from less than one percent to almost 53 percent, with a median of 8 percent. 

The median seroprevalence rates by race were:
            Black               18.4 percent    
            White                 3.8 percent

Medians can be difficult to interpret, but our confidence in using the national medians to compare black and white HIV seroprevalence is buttressed by the fact that HIV seroprevalence was higher among blacks in 50 of the 56 clinics that collected data from both whites and blacks.(4)

Doing the arithmetic, we find that, among injecting drug users who were in drug treatment in 1991-92,  blacks were almost five times as likely as whites to be HIV positive.

Like the NHSDA survey data, these data are not exactly what we would wish.  The data are carefully collected and analyzed but no claim is made that the group of treatment programs included in the study represents the entire population of injection drug users in treatment.  And of course, persons in drug treatment definitely do not represent all persons who inject drugs. 

So we have now looked at two entirely different data sets and come to the same conclusion.  Blacks who inject drugs are much more likely to have HIV/AIDS than their white counterparts who are doing exactly the same thing.  Separately the numbers from the different data sets are suggestive.  Together they form a consistent and compelling pattern.

 

TECHNICAL NOTE: DEVELOPMENT OF AN INDEX OF NEEDLE USE BY RACE

Each year the U.S. Government publishes estimates of the number of persons who inject drugs.  The information on needle use is collected through a self-administered questionnaire completed in the course of a personal interview.  The interviewer does not see the questionnaire the respondent completes.  These data collection procedures are designed to make the promise of confidentiality believable.  We do not know how successful they are in eliciting accurate information.  But we do know that this survey, the National Household Survey on Drug Abuse (NHSDA), yields higher estimates of drug use in the United States population than any of the other national surveys funded from time to time by the federal government that use other data collection procedures.

NHSDA Estimates of Needle Use
The NHSDA estimates of needle use vary substantially from year to year.  Some of the variation is due to normal sampling variability in sample surveys.  Some may be attributed to actual changes in behavior.  We also know that some year-to-year differences are the result of changes in methodology. There may be additional factors at work to create changes from year to year.  But our goal here is not to look at trends.  It is to create index numbers that can guide us as we look at relative needle use by blacks and whites.  And for this purpose, the NHSDA information on needle use can serve as a useful building block. 

There are two different kinds of information on needle use collected in the NHSDA surveys.  One is needle use in the past year (Table 1) and the other is ever used needles in one’s lifetime or lifetime needle use (Table 2).  

Table 1.  Creation of an index (white = 100) 
of persons (a) who have used needles in the past year,
based on the 4-year average of 1991-1994

 

White

Black

1991

815,000

154,000

1992

504,000

88,000

1993

320,000

133,000

1994 (b) 

298,000

106,000

4-year average

484,000

120,000

Past year index 
(white = 100)

100

25

 

Table 2.  Creation of an index (white = 100)
of persons (a) who have ever used needles 
(i.e. lifetime use), 
based on the 4-year average of 1991-1994

 

White

 Black

1991

2,729,000

559,000

1992

2,346,000

395,000

1993

2,079,000

504,000

1994 (b) 

2,075,000

386,000

4-year average

2,307,000

461,000

Ever used index 
(white = 100)

100

 20

Footnotes to Tables 1 and 2

(a) The surveys cover the civilian non-institutionalized population age 12 and over. The samples include persons living in some group quarters, such as civilians living on military installations, college dormitories, and homeless shelters. The samples do not include the homeless not in shelters.

(b) In 1994, NHSDA used two different questionnaires to ask about needle use. The numbers reported here are the average of the responses to the two different questionnaires used that year. 

Sources for Tables 1 and 2: National Household Survey on Drug Abuse: Population Estimates. Rockville Md: National Institute on Drug Abuse and Substance Abuse and Mental Health Services Administration. 1991 (Revised 1992) p. 1,2 and 106; 1992 p. 106; 1993 p. 104; and 1994 p. 107, A-16

 

For each set of data separately, we have first taken an average of the data for the four years from 1991 to 1994.  The average over four years is important for our purposes because it washes out the year-to-year variation over this period.  We then created an index for each set of data, setting the average for whites = 100.  In the case of needle use in the past year, the index number for blacks is 25; in the case of lifetime needle use, the index number for blacks is 20.

The choice was made to continue the analysis using the index based in needle use in the past year because (a) lifetime needle use seemed to go too far back in time; (b) needle use in the past year seemed to be the more conservative choice because it showed a slightly smaller gap between whites and blacks, and finally (c) the comparative data on needle use for whites and blacks in the D.C. metropolitan area drug use study was available only for past year needle use. (The importance of this point will become apparent in the discussion that follows.)

We have chosen 1991 as the starting point for our index because it was in 1991 that the definition of the population to be surveyed was changed to include persons living in college dorms, rooming houses and homeless shelters.  Even with this expanded definition, the NHSDA survey sample still did not include persons in prisons or the homeless living in the streets, two groups that could be expected to include substantial numbers of persons who inject drugs.  It is to data that will give us some understanding of needle use among these groups that we now turn. 

Injecting Drug Use Among the Incarcerated and Homeless in the D.C. Metropolitan Area
In 1991, the U.S. Government commissioned a study in the District of Columbia metropolitan area  to examine drug use among persons not living in households. Although it is not ideal for our purposes, the D.C. metro area drug study is the only well designed, area probability study done in the United States that has examined injecting drug use among the nonhousehold population such as the street homeless and imprisoned by race. And thus we will use the information from the D.C. study, in conjunction with 1990 U.S. Census data, to create crude estimates of needle use in the nonhousehold population, by race for the United States as a whole.

We are not the first analysts to use the D.C. metropolitan area study in this way.  A major study by the National Research Council and Institute of Medicine on HIV and needle use recently used the D.C. metropolitan area study to suggest that the NHSDA estimates of needle use, if adjusted to include the nonhousehold population, would increase by about 50 percent.(5)

Tables 3a and 3b highlight some of the similarities and differences between the D.C. metropolitan area and the rest of the United States.

 

Table 3a.  Selected characteristics of the D.C. Metro Area,
large metro areas, and the total United States, 1991

Drug Use in the household population age 12+

Wash D.C.
Metro Area

Large Metro
Areas

Total U.S.

Past Year

Percent of population

     Any illicit drug

11.7

13.7

12.7

     Excluding marihuana

7.8

7.5

7.2

     Cocaine

3.6

3.4

3.0

           Crack

0.9

0.6

0.5

     Heroin

0.4

0.3

0.2

Lifetime

Percent of population

     Any illicit drug

39.9

39.9

37.0

     Excluding marihuana

22.7

21.2

20.4

     Cocaine

13.5

13.5

11.5

           Crack

2.2

2.3

1.9

     Heroin

1.7

1.5

1.3

 

Table 3b. Selected characteristics of the D.C. Metro Area
  and the total United States, 1991

 

Wash D.C.
Metro Area

Total U.S.
Household population age 12+ Percent distribution

Total

100

100

White

62

77

Black

27

11

Hispanic

5

8

Other

6

3

Household population age 12+

Number

Total

3,171,915

203,652,481

White

1,961,454

157,703,527

Black

861,899

23,207,480

Hispanic

64,775

16,377,383

Other

83,787

6,364,091

Sources for Tables 3a and 3b: 

Drug use for all areas and household population for the D.C. metro area:
The Washington, D.C. Metropolitan Area Drug Study: The Prevalence of Drug Use in the D.C. Metropolitan Area Household and Nonhousehold Populations: 1991. Technical Report #8; Rockville MD: National Institute on Drug Abuse; 1994 (p. 27,74f)

Household population for the United States:
National Household Survey on Drug Abuse: Population Estimates 1991 Revised 11/20/92. Rockville MD: National Institute on Drug Abuse; 1992. (p. 14)


Injecting Drug Use Among the Incarcerated and Homeless in the United States
Table 4 outlines the procedures used to create indices of needle use for the household and nonhousehold populations of the U.S., using two different definitions of the nonhousehold population.

Among the nonhousehold population, the D.C. metropolitan area drug study found that about 9 percent of whites and 11 percent of blacks had injected drugs in the past year.  Each of these figures is, of course, substantially higher than the corresponding figure for the household population.  In step 2 of both Procedures A and B, these percentages are used to estimate how many injecting drug users there are in the total U.S. nonhousehold population.

In the end, whether Procedure A or B is used makes very little difference.  In both cases, an index number set to 100 for whites results in an index number for blacks in the 33-34 range.(6)

Our objective has not been to produce precise estimates of the amount of injection drug use by whites and blacks.  Hard data that would permit such estimates are simply not available.  We have, however, used state-of-the-art statistical surveys to produce what we believe to be a reasonable index of the white-to-black ratio of injection drug use.  The pattern of evidence suggests that there are about three times as many whites as blacks who inject drugs (an index number of 100 compared to 33 or so).  To allow for changes over time and other uncertainties that remain in the state-of-the-art data, we further suggest that, compared with the index of 100 for whites, the true index for blacks almost certainly falls within the range of 50 to 25.  Thus we conclude that there are at least twice as many white injection drug users as there are black injection drug users in the United States, and that the true ratio may be as high as four times as many.

Table 4.  Outline of the procedures used to create the indices of needle use for the household and nonhousehold populations on the United States, using two different definitions of the nonhousehold population.
  White Black
Procedure A    
Step 1 - Identify nonhousehold population A    
Correctional institutions (c) 417,000 476,000
Visible on the street (c) 15,000 16,000
Group homes (c) 30,000 15,000
Nonhousehold population A total 462,000 507,000
Step 2 - Estimation of the part of nonhousehold population A that injects drugs    
Nonhousehold population A total 462,000 507,000
Proportion of nonhousehold population A estimated to be injecting drug users (d) 0.093119 0.112044
Estimate of the number of persons in nonhousehold population A that injects drugs 43,021 56,807
Step 3 - Creation of nonhousehold population A index    
Estimate of the number of persons in nonhousehold population A that injects drugs 43,021 56,807
Needle use (4-year average from table 1) 484,000 120,000
Total 527,021 176,807
Nonhousehold population A index (White = 100) 100 34
Procedure B    
Step 1 - Identify nonhousehold population B    
Nonhousehold population A total 462,000 507,000
Other institutional quarters (c), (e) 307,000 98,000
Other noninstitutional quarters (c), (e) 351,000 66,000
Nonhousehold population B total 1,120,000 671,000
Step 2 - Estimation of the part of nonhousehold population B that injects drugs    
Nonhousehold population B total 1,120,000 671,000
Proportion of nonhousehold population B estimated to be injecting drug users (d) 0.093119 0.112044
Estimate of the number of persons in nonhousehold population B that injects drugs 104,293 75,182
Step 3 - Creation of nonhousehold population B index    
Estimate of the number of persons in nonhousehold population B that injects drugs 104,293 75,182
Needle use (4-year average from Table 1) 484,000 120,000
Total 588,293 195,182
Nonhousehold population B index (White = 100) 100 33

Footnotes to Table 4

(c)  1990 Census of Population. General Population Characteristics: United States. CP-1-1.  Washington, DC: Census and U.S. GPO; 1992. (p.64)

(d)  Prevalence of Drug Use in the D.C. Metropolitan Area Household and Nonhousehold Populations: 1991. Technical Report #8. Rockville, Md: National Institute on Drug Abuse; 1994. (Needle use data from Table 6.5, p.98 and undated revision.  Population data from Tables 5.1 and 5.2, pp.74 and 75)

(e)  The nonhousehold part of the D.C. metropolitan area drug study (MADS) does not include nursing homes and drug/alcohol abuse group homes, so Procedure A comes closer to duplicating MADS on a national scale than does Procedure B which includes these groups.

References

(1) Technical notes. HIV/AIDS Surveillance Report: U.S. HIV and AIDS cases reported through December 1994. Atlanta, GA: Centers for Disease Control and Prevention. 1994;6. (p.36)

(2) HIV/AIDS Surveillance Report: U.S. HIV and AIDS cases reported through December 1994. Atlanta, GA: Centers for Disease Control and Prevention. 1994;6. (p.11f)

(3) These figures are the sum of the following exposure groups among adults and adolescents: injecting drug use; men who have sex with men and inject drugs; and sex with an injecting drug user.

(4) National HIV Serosurveillance Summary: Results Through 1992. Vol. 3.  Atlanta, GA: Centers for Disease Control and Prevention; 1994. (p. 19f)

(5) Normand J, Vlahov D, Moses LE, eds. Preventing HIV Transmission: The Role of Sterile Needles and Bleach. Washington, DC: National Academy Press; 1995. (p 62f)

(6) We considered making some sort of additional adjustment for NHSDA nonrespondents, a group that accounts for far more people in the estimated people than do the nonhousehold part of the population.  However the proportions of nonrespondents were about the same for whites and blacks, so any overall adjustment for nonrespondents would not have changed our index numbers.  National Household Survey on Drug Abuse: Main Findings 1991. Rockville, MD: Substance Abuse and Mental Health Services Administration; 1993. (p.B-15)  Also, studies of item nonresponse have not indicated differences between whites and blacks in a way that would affect our analysis.  See Turner, CF, Lessler JT, Gfroerer, JC, eds. Survey Measurement of Drug Use: Methodological Studies. Rockville, MD: National Institute on Drug Abuse; 1992.