This paper proposes that the problem of estimating drug abuse prevalence may be approached effectively through inexpensive means and without the need of experts. A situation is hypothesized in the city of "Metropolis" whereby a junior staffer is able to make reasonable and compelling estimates of the problem using information gathered from various city agencies and knowledgeable individuals over a two-week period. Each day's information-gathering is described, with the thought processes that sort out and coalesce the findings. In the matter of weeks the information is used to justify expenditures for treatment and prevention programs.
The Twelve Step method is an effective way for substance abusers to help each other begin and sustain a life of recovery. Likewise, "street-based" AIDS outreach education and bleach distribution is an effective way to slow the spread of HIV among the intravenous drug using (IVDU) community. These approaches are attractive because they are very cheap, and they don't require Ph.D.'s or M.D.'s or other highly-trained professionals. That's important in an era where there simply aren't enough dollars or experts to sustain the fight against the problems of substance abuse.
This paper proposes that the problem of estimating drug abuse prevalence can also be approached cheaply and without the need of experts with advanced degrees. It asks you, the reader, to imagine that you are a junior staffer at the Planning Department of the city of "Metropolis". You have only a Bachelor's degree and no special training in epidemiology, but you are known as a person of imagination and energy. Your boss asks you, "If I gave you two weeks and five hundred dollars cash for incidental expenses, do you think you could come up with some estimates of how many drug abusers we've got in Metropolis?" Impulsively, you answered "yes", and now you have to begin your task:
Next you lay out your plan for looking at drug use in Metropolis. The most important things to get clear to yourself are, "what drugs am I interested in looking at?" and, "how will I define the various levels of use-- occasional, weekly, daily, and so on?" You realize that the number of "users" may be many times more than the number of "abusers"-- especially for marijuana and cocaine, less so for heroin. You also realize that many, if not most, of the people in the illicit drug scene are users of more than one drug.
Now that you have a basic framework for what you're looking for, you begin making phone calls to the local agencies whose help you'll need. You introduce yourself, describe your purpose, and line up appointments over the next several days. In a couple of cases, you find that you have to rely on a prominent friend to assure the cooperation of wary directors.
In the afternoon, you go down to the main library and dig out the latest census estimates of Metropolis' population, broken down by age, sex, race, and urban/suburban districts. In short order, you gain an understanding of the population makeup of your community. While at the library, you also locate and copy a couple of interesting articles from the Journal of Psychoactive Drugs: "The Prevalence of Drug Use in San Francisco in 1987" by John Newmeyer (in the April-June 1988 issue) and "A Spreadsheet for AIDS: Estimating Heterosexual Injection Drug User Population Size from AIDS Statistics in San Francisco" by Michael Aldrich et al (in the July-September 1990 issue). You check out a textbook on statistical epidemiology to use as a reference in the coming days.
The staff of the medical examiner's office point out that, in many cases, the cause of death could not be ascertained-- it may have been suicide, or homicide, or accidental overdose, or some other kind of accident, or simply a natural cause unrelated to any drug use. You reply that the uncertainty about cause is OK, because you regard this group as simply a sample of Metropolis people who had used heroin, or cocaine, or whatever.
While waiting for someone from the busy intake staff to get a moment to comply with your request, you review the summary stats from Detox Central. The staff researcher has prepared tables showing the characteristics of the clientele, in temrs of primary drug of abuse, age, sex, and race. The cross-tabulations are really interesting-- you realize that the heroin abusers of your city are demographically rather different from the cocaine abusers, and very different from the methamphetamine abusers. There are also tables showing responses to the intake questionnaires-- how much drugs the clients reported using, what year they first used their primary drug, how they claimed to be supporting their drug habits, and how many children they had had.
After lunch, the intake staff worker completes his review of Detox Central's charts. From his notes on your 3 x 5 cards, you find, among other things, that 9 of the 60 1989 decedents who had heroin metabolites in their bodies also had been clients of Detox Central during 1988. The staff worker tells you, however, that about 10% of their clientele uses false names. You say, "Hmm-- I think my best guess is that if all your clients had used their real names, we'd have had 10 'matches' rather than just 9."
You put pencil to paper and do a simple calculation: If 10 out of 60 decedents were also seen at Detox Central in 1988, and if 2,000 heroin abuser clients had been seen altogether in that year, the best guess is that 6 x 2,000 or 12,000 people abused heroin that year. You are pleased-- you have made your first stab at estimating the number of heroin users in Metropolis. And Detox Central's data is going to allow you to make similar estimates based on their cocaine and "speed" clients, since these drugs, too, were frequently found in 1989's dead bodies.
Your midday hours are spent with a couple of street outreach workers, who have busied themselves in seeking out Metropolis IV drug users to talk to them about the risk of AIDS and the use of bleach to disinfect hypodermic equipment. It doesn't take you long to realize that you can't get much closer to the "front line" than this. The outreach workers are seeing people that drug programs, or the police, or the emergency rooms, haven't had any contact with as yet. This gives you an idea. You request that the two workers ask a simple question of their street clients: "Were you treated by Detox Central last year?" You make an appointment to see them next Thursday.
You turn to your notes from last Monday's review of local census data. The estimate for 1988 was that the black male population of Metropolis, aged 21 to 45, was 27,500. If the professor had a good random sample of this population, that means that .08 x 27,500, or about 2,200 Metropolis residents, were young black men with a recent history of heroin use. But your notes from Detox Central inform you that 20% of their heroin-user clientele were black males in the 21-45 age range. So, an estimate of the overall heroin population is given by 2,200 divided by .20, or 11,000. That's not far off from your earlier estimates.
You are gratified to find that the sociologist has also inquired about cocaine and PCP use. This enables you to use this "synthetic" method for these drugs as well, because you also have the necessary data from Detox Central as to what proportion of the abuser client groups are young black men. But you are concerned about the many sources of potential error in this method, such as the veracity of the black men's responses to the professor's survey questions, and the possibility that his sampling might have missed large numbers of homeless or transient men.
You are disappointed to find that Metropolis is not included in the DAWN survey of drug mentions from U.S. emergency rooms. Your calls to these ER facilities have not been helpful, either. However, an obstetrician from the big downtown public hospital calls you with an intriguing report. She has a keen interest in addicted babies, and she has enumerated 90 cases of infants born in 1988 with heroin dependency. She estimates that there were probably an equal number in Metropolis that she didn't locate-- that is, there were about 180 babies born that year to heroin-dependent mothers.
You realize that you have a good idea about the birth rate of heroin-dependent females, once again thanks to the Detox Central statistics. Detox Central routinely asks their entering clients about their history of childbearing. From this, you are able to deduce that the female heroin-dependent clientele has a birth rate of about 60 per 1,000 women per year. If there were 180 babies in 1988, that means that there were about 180/.06, or 3,000 heroin-dependent women. But the stats from Detox Central, the methadone programs, and the medical examiner all suggest that something like one out of three heroin abusers in Metropolis is female. So, you figure that there must be about 9,000 heroin-dependent people in the city.
The afternoon is more fruitful. You interview a number of officers on the topic of crimes against property. They provide you with statistics on the amount of reported property crime, the estimated value of the property stolen, and the probable cash value of the stolen property to the thief. They also make some educated guesses as to the "multiple" that should be put on these figures, to account for unreported thefts. Finally, they have some good notions of what proportion of criminals of this sort are supporting drug habits. From this you are able to calculate reasonable upper-bound and lower-bound estimates of the dollar value from property crimes that is applied to drug habits. You compare this to the Detox Central data on the annual cost of the typical heroin habit, and find to your surprise that all the property crimes of Metropolis can support no more than 2,500 or so heroin habits at a time. Since at least one out of five Metropolis heroin users self-report as supporting their habits mainly by crimes against property, your upper-limit estimate of heroin abusers now becomes 2,500 divided by .2, or 12,500.
Later in the day you talk with an epidemiologist who has been tracking the AIDS epidemic in Metropolis. From him you get the count of reported cases of AIDS ascribed to heterosexual drug users for each year up through 1989. He also has data from the anonymous HIV antibody testing sites of the city; these show that the annual caseloads of heterosexual IVDUs who came to be tested turned out to have a 2% seropositive rate in 1985, a 4% rate in 1986, a 7% rate in 1987, and a 12% rate in 1988. Using some recent estimates of the rate of progression of HIV+ persons to AIDS diagnoses, you use your desk computer to apply the Aldrich "back calculation method" to see what estimate of the underlying IVDU population at risk for AIDS is most consistent with Metropolis' seroconversion and AIDS diagnosis data. You conclude that a prevalence figure of 11,000 gives the best fit.
After writing sixteen pages on your word processor, you are done. You then write the Executive Summary, taking care to think about the frame of mind of mayors, legislators, and other top-level decision makers. You are convinced that cost-benefit analyses will show that most kinds of treatment or HIV-prevention efforts among drug users are worth the money invested. You want to be sure that some kind of estimate of the size of the at-risk groups are made, so that the cost-benefit analyses, and the policymaking based thereon, can proceed.
Late in the afternoon, you finish the draft of your report and circulate copies to your colleagues and your interviewees for their comments. At the end of the day, when going over your expenses, you realize that much of the $500 cash you were given was spent in gourmet lunches and dinners to thank the people for their generous help in digging through their files. You reflect that you probably could have gotten by with less than $500, but you needed every hour of two work weeks to complete 55 phone calls, 23 site visits, 200 pages of reading, lots of chart reviews, and a great deal of reflection, analysis, and writing-up. Only a tiny amount of your time was spent in actually making the statistical estimates-- twenty times as much time was needed to get the bits of data needed from all the different sources. You wonder if this input of effort would be typical for a review of prevalence indicators for other cities of one million.
Some Weeks Later