How Many Daily Heroin Users Are There in the U.S.? Somewhere Between 60,000 and 1 Million. Maybe.

According to the National Survey on Drug Use and Health (NSDUH), about 620,000 Americans used heroin in 2010. But according to a new report commissioned by the Office of National Drug Control Policy, something like 1.5 million Americans were "chronic heroin users" that year. That group includes anyone who has consumed heroin on four or more days in the previous month.

This dramatic discrepancy results from the report's attempt to count heroin users missed by NSDUH, whether because they did not respond honestly, because they did not respond at all, or because they were not part of the household population sampled by the survey. To adjust for such undercounting, Beau Kilmer and eight other drug policy analysts at the RAND Corporation rely mainly on data from the Arrestee Drug Abuse Monitoring Program (ADAM), which includes urinalysis as well as a survey. Because it focuses on arrestees, ADAM is more likely than NSDUH to identify heavy drug users. But its sample, unlike NSDUH's, is not nationally representative, so Kilmer and his colleagues must perform a series of calculations to convert ADAM numbers into total male arrestees testing postive for heroin and divide those users by frequency of use. Then they add estimates for men who were not arrested as well as for women and teenagers, based partly on NSDUH and information about overdoses, emergency room episodes, and treatment admissions.

The RAND researchers convert their estimate of heroin users into estimates of total consumption and spending. They use similar methods to estimate the size of the cocaine, methamphetamine, and marijuana markets. In the case of marijuana, a drug for which NSDUH seems to be a better guide, Kilmer et al. rely mainly on numbers from that survey, inflated by 25 percent and supplemented by ADAM data for respondents with criminal records. It is an impressive, headache-inducing feat, but one that is subject to "great uncertainty" because of the assumptions involved and the limits of the data, as the authors repeatedly acknowledge. "In many cases," they say, "the extent of the uncertainty cannot be bounded or quantified." They do not really know, for example, "the extent to which one can trust arrestees' self-reports about their spending on illegal drugs" or "how to extrapolate just ten urban areas' arrest records to the country as a whole." 

The RAND estimate for frequent heroin users is much higher than the one reported by NSDUH. In a recent USA Today op-ed piece, Kilmer and one of his collaborators, Jonathan Caulkins, say "estimates from the 2010 NSDUH suggest there were only about 60,000 daily and near daily heroin users"—i.e., people who used heroin on 21 or more days in the previous month. They argue that "the real number is closer to 1 million."

Yet whether you look at NSDUH data or at the RAND estimates, the trend from 2000 through 2010 looks similar. "Heroin consumption remained fairly stable throughout the decade," Kilmer et al. write, "although there is some evidence of an increase in the later years." They note that "there was a steady increase in the amount of heroin seized within the United States and at the southwest border from 2007 through 2010," but they caution that seizure levels may reflect enforcement efforts and traffickers' tactics rather than consumption. The RAND estimates of heroin consumption indicate "essentially no change" from 2000 to 2010. The report does not cover the two subsequent years, when NSDUH reported an increase in the number of past-month heroin users, from 239,000 in 2010 to 335,000 in 2012.

The trends for the other drugs are also similar to what the NSDUH data suggest. "Multiple indicators are consistent with an increasing trend in meth consumption over the first half of the decade and a subsequent decline through 2008," Kilmer et al. write. "From 2002 to 2010, the amount of marijuana consumed in the United States likely increased by about 40 percent while the amount of cocaine consumed in the United States decreased by about 50 percent." Throughout the period covered by the report, estimated annual spending on the four drugs in the United States totaled about $100 billion, although the breakdown changed substantially. "In 2000," say the RAND researchers, "much more money was spent on cocaine than marijuana; in 2010 the opposite was true."