TRENDS AND PATTERNS OF ILLICIT DRUG USE IN THE USA:
IMPLICATIONS FOR POLICY

Lana D. Harrison, Center for Drug and Alcohol Studies, University of Delaware, USA

THE NATIONAL HOUSEHOLD SURVEY ON DRUG ABUSE (NHSDA)

The most complete source of information on the prevalence of drug use and related behaviours in the USA is the National Household Survey on Drug Abuse (NHSDA) . The survey has been conducted at 1 to 3 -year intervals of a representative sample of the population aged 12 and over. The 1971 and 1972 surveys, conducted under the auspices of the National Commission on Marihuana and Drug Abuse, primarily addressed marijuana use. The responsibility for the conduct of the survey was shifted to the National Institute of Drug Abuse (NIDA) at its formation in 1974, and then to the newly created Substance Abuse and Mental Health Services Administration (SAMHSA) at its establishment in October 1992. The survey currently yields prevalence estimates on the following classifications of drugs: alcohol, tobacco, cannabis, cocaine, inhalants, hallucinogens, heroin and the non medical use of psychotherapeutic agents (stimulants, sedatives, tranquillisers and analgesics). The survey includes the US population aged 12 and over.

The NHSDA derives its estimates from a national probability sample of households in the co terminous USA (Alaska and Hawaii were added to the sampling frame in 1991) . The household population includes more than 98% of the US population. Transient populations such as the homeless and people living in institutionalised group quarters such as military barracks and prisons have historically been excluded. Starting in 1991, non-institutional group quarters were added to the sampling frame so that people living in rooming houses and college dormitories were included. Youth have traditionally been oversampled, and black people and Hispanics have been oversampled since 1985 to allow for increased precision in the estimates of drug use among these population subgroups.

The size of the survey has grown with each survey administration; however, the largest increase in the sample size occurred between 1990 and 1991, when the number of respondents increased from 9259 to 32 628. The sample size was increased by legislation so as to increase the precision of estimates for rarely used drugs such as heroin and cocaine, and to increase the precision of drug use prevalence rates for specified demographic subgroups. The survey has been administered annually since 1990, with sample sizes close to 30 000.

As Figure 1 shows, in 1993 37.2% of the population aged 12 and over reported at least one occasion of illicit drug use in their lifetime, with 11.8% reporting some use in the past year and 5.6% in the past month. The most frequently used illicit drug was cannabis, with33.7% lifetime prevalence,9.0% past year prevalence and 4.3% past month prevalence. The non-medical use of a psychotherapeutic drugs was reported by 11.1 % of the population (lifetime prevalence). About 3.8% had used a psychotherapeutic drug for other than medical reasons in the past year, and 1.3% reported use in the past month. The lifetime prevalence rate for cocaine was 11.3%, with annual prevalence at 2.2% and prevalence in the past month at 0.6%. Alcohol and tobacco were used at much higher rates than any of the illicit drugs. About 83.6% of the population reported drinking alcohol in their lifetime; 66.5% had drunk in the past year and 49.6% in the past month. About 71.2% of the population had smoked cigarettes in their life time, with 29.4% reporting smoking in the past year and 24.2% in the past month. (See SAMHSA (1993) for a more complete description of the NHSDA and survey results. )

As drug use is highly correlated with age, the NHSDA provides estimates of drug use by age ranges for youth (ages 12-17), young adults (ages 18-25), middle adults (ages 26-34) and older adults (ages 35 and over). The highest prevalence rates are found among young adults, followed by middle adults. Youth report higher rates of illicit drug use than older adults, but older adults report greater alcohol and tobacco use than youth. The trends in drug use across the age groups are generally similar.(1) (The increase in lifetime prevalence rates observed among older adults results from the movement of birth cohorts with substantial drug use experience into the older age group, rather than an increase in the number of new users among older age groups. )

Higher rates of illicit drug use are found in the larger metropolises and their suburbs. Overall rates of illicit drug use were highest in the West. Other subgroup differences show males reporting higher usage rates for most illicit drugs than females. The unemployed report higher illicit drug use rates than the employed. College graduates report the highest rates of lifetime drug use, but the lowest rates of past month (or current) drug use.

The NHSDA is not a longitudinal study. However, there has been sufficient continuity in design and methodological procedures to allow computation of trend estimates from the repeated cross-sectional surveys. Overall rates of illicit drug use peaked in the USA sometime around 1979, and have since been decreasing. This may seem counterintuitive, considering the attention the drug problem has received in recent years. However, the Monitoring the Future (MTF) study, a nationally representative survey of high school seniors and young adults, also shows a peaking in rates of drug use about the same time (Johnston et al.,1994; University of Michigan, 1994). The MTF study has annually surveyed a rep representative sample of the nation's high school seniors since 1975. High school seniors were chosen as the focus of study because high school marks the transition from youth to adulthood. A subset of each graduating class has been followed up, creating a panel study of young adults ranging in age into their 30s.

Figure 2 shows that, among young adults, alcohol cannabis, cocaine, and the summary measure of an illicit drug use all peaked around 1979. Although the rates of use are lower for other age groups, the trend are very similar, although the steepest increases an decreases occurred among young adults. This suggests that the downward trend in drug use is secular occurring across the entire age spectrum simultaneously.

Both the NHSDA and MTF studies show that marijuana use peaked in 1979, and that the use of most psychotherapeutic drugs peaked in the early 1980s. Cocaine use peaked a few years earlier in the NHSDA among youth and young adults than among high school seniors in the MTF study. The same trends in cigarette smoking are found among youth in the NHSDA and among high school seniors in the MTF, with use levelling around 1979. Cigarette use has continually decreased among adults (18 years and over) since about 1979, but cigarette smoking has been relatively unchanged among youth (12-17 years old) since about 1979. Alcohol use peaked among all age groups around 1979. However, the 1993 and 1994 MTF survey results showed a reversal in the long-term downward trends for cannabis, hallucinogens, stimulants and tranquillisers (University of Michigan, 1994). The 1994 NHSDA results have not yet been released, but a small increase in annual marijuana use between 1992 and 1993 was noted, but only among youth. No other significant changes in prevalence rates occurred for the use of other drugs in any age group. It should also be mentioned that the NHSDA finds that frequent cocaine use, as measured by weekly use of the drug in the past year, shows little change since 1985. The number of weekly cannabis users fell between 1985 and 1991, although 1992 and 1993 rates were very similar to those found in 1991.

Data from panel studies of representative samples from each high school graduating class since 1976 find that the general pattern among young people in the USA is for active drug use to increase for the first few years out of high school, and then decrease. There are a few exceptions, however. Current and daily alcohol use, as well as binge drinking, all increase up to age 21. However, the only measure to decrease thereafter is binge drinking. Annual cocaine use also remains stable, rather than dropping, after age 21. And in general, lighter cigarette smokers become heavier smokers (O'Malley et al., 1984).

The predominate usage patterns for the illicit drugs is infrequent. Most who have used in their lifetime have not used in the past year, and most who have used in the past year have not used in the past month. Take cocaine as a case in point. A question in the NHSDA asks respondents how many times they have used cocaine in their life. In 1992,89.1 % have 'never used' and 6.4% have used fewer than 10 times in their lives. Only 1.7% have used 100 or more times. About three out of five who have ever used cocaine have not used in the last 3 years. Among past year users and past month users, the most frequent number of times used is once or twice. About one in seven monthly cocaine users were daily users, and 7.5% of past year cocaine users indicated that they had used cocaine daily in the past year. Although less is known about crack users, a similar pattern emerges. Most who have used in their lifetime used three or more years ago. Most are not currently using.

LIMITATIONS OF THE GENERAL POPULATION STUDIES

A limitation of the NHSDA and MTF studies referred to jointly in this article as the 'general population' surveys on drug use - is that subgroups with higher rates of drug use are omitted and/or under represented. MTF misses school dropouts and absentees. The NHSDA misses institutionalised members of the population living in group quarters such as those in prisons, jail, hospitals and active military personnel. It also misses those with no fixed address such as the homeless and transients, although shelters for the homeless have been included since 1991.

Good estimates of drug use are not available among all omitted subgroups, but there are epidemiological surveys of drug use among some. Jail and prison inmates (Bureau of Justice Statistics, 1991, 1993; Wish,1991) exhibit much higher rates of drug use than found among the general population. A study of the homeless in the Washington DC area in 1989 also found much higher rates of illicit drug and alcohol use among the homeless and those who are precariously housed. However, as the homeless are such a small part of the population, their higher drug usage rates do not boost the overall drug prevalence rates significantly (NIDA, 1993). Studies indicate that high school dropouts (Mensch and Kandel, 1988) exhibit higher rates of drug use than high school seniors. However, periodic worldwide surveys of US military personnel, conducted by the Department of Defense since 1980, show lower overall rates of illicit drug use among military personnel than their contemporaries. Further, drug use has decreased significantly in the military since 198 (Bray et al., 1992). The NHSDA provides comprehensive information on drug use among most population subgroups. As the subgroups excluded represent a relatively small proportion of the overall population, even the higher rates of drug use among them will not greatly increase drug use prevalence rates. The NHSDA also includes high school dropouts and 'criminals', except for those who are incarcerated for more than several weeks. However these subgroups may be under represented as a result of higher rates of non response. A recent series of methodological studies aimed at evaluating the quality of the data derived from the NHSDA suggest that non-response is not a large problem in the NHSDA (see Harrison,1993).

The results from the NHSDA and MTF survey are also doubted because they are based on self-report methods. But the accumulated scientific evidence suggests that the surveys fairly accurately measure drug use (Harrison, 1995). The surveys are state of-the-art using a number of methods to assuage respondents' concerns about anonymity and confidentiality. Most importantly, questions on drug use are self-administered so respondents are not required t speak their answers aloud to interviewers in either o the general population surveys.

The critics are not convinced and point to other data the interpretation of which leads to different conclusions. The general population survey's credibility were undermined in the late 1980s as the discrepancy widened between the survey findings and other measures of drug use, including public perception of the problem. The general population surveys pointed to small numbers of people with severe drug problems, whereas the public perceived that the problem was worsening (Reuter,1992, p.23).

THE DRUG ABUSE WARNING NETWORK

The Drug Abuse Warning Network (DAWN), which has tracked the number of drug related hospital emergency room episodes since the early 1970s, is one study that is frequently referred to as indicative of increasing drug usage rates. DAWN also collects information on drug-related deaths in a convenience sample of medical examiners (MEs) in 27 large metropolitan areas. The emergency room component was originally designed as a national probability sample, but, over the years, the representativeness of the sample deteriorated as hospitals dropped out and replacement hospitals were recruited non-randomly. However, the DAWN study was redesigned in 1989, with a new national panel of approximately 600 hospital emergency rooms. The sample was also designed to provide self-representing estimates of drug related emergency room episodes for 27 major metropolitan areas in the co-terminous USA.

Data are gathered by DAWN staff reviewing hospital emergency room (ER) medical records for drug mentions. The patient - and/or those accompanying them - frequently self-report whether the use of any drug(s) precipitated their visit. Medical tests may also be used to determine a drug-related ER incident. In the ME component of the DAWN study, toxicology reports are available in virtually all cases to deter mine drug-related events. Each drug mention is recorded separately, as well as in combination with other drugs (except alcohol, which is always reported in combination with other drugs). Each drug mentioned contributes to the overall trends for the individual drugs, even though many episodes involve multiple drug mentions.

Estimates have recently been computed using back weighting procedures to produce nationally representative estimates of drug-related ER episodes since 1978. Cocaine-related ER episodes increased dramatically between 1985 and 1989, and, as Figure 3 shows, reached their highpoint in 1993, in contrast to the decreasing trend in cocaine use found in the general population surveys. Heroin use is very rare in the general population, making the compilation of trend estimates in the general population surveys difficult. However, the available information suggests that heroin use peaked in the late 1970s, about the same time as cannabis and a number of other drugs. In contrast, heroin-related ER visits increased nearly fourfold between 1978 and 1989. Dropping in 1990, the numbers rebounded to their highest level in 1993. The derivation of estimates for cannabis related ER visits before 1988 has not yet been completed. However, cannabis-related ER episodes rose to their highest level in 1993 since 1988. Marijuana is often mentioned in combination with other drugs, particularly alcohol and cocaine (SAMHSA,1994).

Data on drug-related deaths are collected from the MEs' office in 27 major metropolitan areas. Since 1985, the data show an increase in cocaine-related deaths from 717 to 3910 in 1992, and for heroin/morphine-related deaths from 1433 to 3805. There is more stability in the number of marijuana-related deaths, ranging from a low of 105 to a high of 457 between 1985 and 1992.Therefore, the ME data parallel the ER data for cocaine, and heroin/morphine, but diverge for marijuana.

Nevertheless, the results of the DAWN study are not necessarily incongruent with those of the general population surveys. One possible explanation is that changes in drug usage patterns have rendered them more problematic. For example, although infrequent users may have quit using (resulting in decreasing prevalence rates), those who do use may have become heavier users - ingesting more of the drug per occasion. The trend towards increasing polydrug use and/or changes in patterns of simultaneous drug use could also be a factor in the escalating trends apparent in Figure 3 for the individual drugs. Changes in access to health care could also influence the trends. Individuals without health care frequently use emergency rooms for their medical care. As drug use tends to be higher among lower socioeconomic groups, their increasing numbers could affect drug-related emergency episodes. For cocaine, in particular, a trend towards more dangerous routes of administration, especially freebasing and smoking crack cocaine, could explain some of the increase in cocaine related ER visits and deaths.

Another possible explanation is that an increase in the purity of drugs has induced the increase in both ER visits and deaths. According to the Drug Enforcement Administration (DEA), the average purity of retail cocaine increased from 50-60% in 1985 to 72% in 1987. A slight drop in average purity to 66% was recorded in 1989. Purity dropped to 54% in 1990, but increased to 59% in 1991 and 64% in 1992 (National Narcotics Intelligence Consumers Committee-NNICC,1993). The tetrahydrocannabinol (THC) content of commercial-grade marijuana increased from 0.82 in 1980 to 3.12 in 1985. Only slight increases have been recorded since, with a 3.68 THC content in nationwide samples of commercial grade marijuana. The average purity of heroin for retail heroin was 37% in 1992, much higher than the average purity of 26.6% in 1991. It is also much high er than the 7.0% average purity of a decade ago (NNICC, 1993). Therefore the trends in purity for cocaine appear generally to have been slightly up and down, whereas those for marijuana are stable to increasing having increased the most between 1980 and 1985, and the purity of heroin continues to increase. This does not closely parallel the trends observed in ER episodes and deaths attributed to the individual drugs, so the purity explanation cannot explain all the variation in these types of episodes.

In an analysis of frequent cocaine users and their use of treatment, Gfroerer and Brodsky (1993) suggest that DAWN is not a good measure solely of frequent cocaine use. ER patients represent only a small percentage of frequent cocaine users. They point out there is little reason for the trends in the overall prevalence of cocaine to mirror those of cocaine related ER visits. DAWN is not a prevalence study it is an incident-based study measuring health-related consequences of drug use. It records number of incidents (ER visits and deaths), not the percentage of some group reporting drug use experiences. Although there is undoubtedly a connection between health consequences and prevalence rates, they are not the same things.

DRUG USE AND CRIMINALITY

There are several sources of information available about drug use among prison and jail inmates. The Bureau of Justice Statistics (BJS) conducts periodic nationally representative surveys of state prison (1974, 1979, 1986, 1991) and jail inmates (1983, 1989) . These surveys show much higher rates of alcohol and illicit drug use than in the general population. Half of the prison inmates in 1991 and 44% of the (convicted) jail inmates in 1989 reported use of an illicit drug in the month before committing the offence leading to their incarceration (see BJS,1988, 1991,1993).

Nearly a third (31%) of prison inmates, in 1991, were under the influence of drugs at the time that they committed the offences leading to their incarceration, and 17% reported committing the offences to get money to buy drugs. Just over a quarter (27%) of jail inmates in 1989 were under the influence of drugs when committing their offences, and 13% said that they committed their offences to obtain money for drugs. Those incarcerated for robbery, burglary and larceny were more likely to report committing their crimes to obtain money. These income-generating crimes are often substantially increased during addiction. Most inmates reported starting drug use before they were arrested the first time. However, regular use of major drugs (heroin, cocaine, PCP, LSD and methadone) typically began after their first arrest.

Substantial alcohol use was also reported by incarcerated populations. In fact, higher proportions report being under the influence of alcohol than any individual illicit drug when committing the offences leading to their incarceration. An estimated 18% of the 1991 prison population reported being under the influence of alcohol only, and 14% reported being under the influence of both alcohol and illicit drug(s).

Figure 4 shows the trends in drug use among prison inmates between 1974and 1991, and Figure 5 shows the trend between 1983 and 1989 for convicted jail inmates. The trends in drug use among prison and jail inmates generally mirror those found in the general population surveys with the exception of cocaine. Cannabis use has been dropping, and there have been dramatic reductions in heroin use, especially among prison inmates. However, contrary to the steep decreases in cocaine use recorded among the general population since the mid- 1980s, cocaine use continues to increase among inmates. Nearly a third of prison inmates in 1991 reported regular use of cocaine at sometime in their life compared to 22% in 1986. About 14% of prison inmates in 1991 reported committing their offences under the influence of cocaine, compared with 1 % of inmates in 1974,4.6% in 1979 and 10.7% in 1986.

People charged with drug offences accounted for 23% of all those held in local jails in 1989, in comparison with about 9% in 1983. The number of black inmates charged with drug offences increased from 25% to 48%, Hispanics from 20% to 25%, and women from 9% to 14%. The number of white, non Hispanic inmates charged with drug offences Secreasedfrom44% to 26% between 1983 and 1989. These changes occurred as the total number of inmates held in jails increased by 77% (BJS, 1991). Over the same period, the number of people held for federal drug offences increased by 328% in comparison to 47% for all other offences (BJS,1992).

Part of the increasing trend in cocaine use among incarcerees may result from the increasing trend indrug-related arrests, particularly those for cocaine and heroin. Between 1980 and 1992, the number of arrests for drug offences by State and local police nearly doubled from 580 901 to 1066 400 (Uniform Crime Reports,1993). Although the 1980 total was dominated by arrests for marijuana (70%) and possession offences (82%), by 1992, the distribution of heroin/cocaine related arrests (53%) exceeded the number for marijuana arrests (32.1 %), although distribution arrests accounted for about equivalent shares in 1980 (27%) and 1992 (27.2%) (see Maguire et al.,1993). Drug arrests began their rapid escalation in ernest after 1983. After falling between 1989 and 1990, they have since been relatively stable. The trend had been towards an increase in the numbers arrested for sales and distribution over the period, but 1992 signalled a bit of a reversal in the trend.

Not only are we arresting more drug offenders, more of them are being found guilty and a greater proportion of those found guilty are being sentenced to prison. Table 1 shows that over the period 1980-92, there was a 346% increase in the number of federal defendants sentenced to prison in US District Courts, compared with a 71% increase for non-drug offences (BJS, 1992). Almost half of all federal offenders sentenced to prison in 1990 were convicted of drug offences. Comparable data are not readily available for State and local courts who deal with the lion's share of defendants charged with drug offences. However, we know that state prison commitments rose from 29 per 100 drug felony arrests in 1986 to 40 in 1990 (Reuter,1993). Nearly half of the increase in prisoners since 1980 was linked to drug offenders entering prison. In 1992, prison commitments for drug offences reached 30% of all new commitments. About a quarter of the inmates in local jails in 1992 were held on drug charges. The average time served by offenders sentenced in state and local courts is less than that served by those sentenced in US District courts, but the average prison sentence for drug offences in US District Courts grew from just under 4 years in 1980 to about 7 years in 1992. Both state and federal courts numbers show significant increases in arrests, but even larger increases in jail and prison commitments.

The fact that we are arresting and convicting more drug law violators, and that those arrested for drug offences are the most likely to indicate that they were 'high' when committing the crime leading to their imprisonment, has an impact on the prevalence and trends in drug use among inmates. The trends are also influenced by increasing numbers of drug offenders receiving jail or prison sentences, as well as the increase in sentence lengths. A recently commissioned study by Attorney General Janet Reno concluded that more than one-fifth of the federal prison population consists of 'low-level' drug offenders, defined as people convicted of drug crimes who have no prior prison time, no current or prior violence in their records, and no involvement in sophisticated criminal activity (Criminal Justice Newsletter,1994).

THE DRUG USE FORECASTING (DUF) STUDY

Another study of drug use among a segment of the criminal population is the Drug Use Forecasting (DUF) study, which provides estimates of drug use among those primarily arrested for serious felony offences in 23 major cities in the USA. There should be a great deal of overlap in the NHSDA and DUF studies. Most of those arrested are subsequently released. Therefore, to the extent that arrestees are not held for a lengthy period of time, they should be included in the NHSDA. Harrison and Gfroerer (1992) demonstrate a great deal of congruity between NHSDA estimates of arrests and those obtained by the FBI, indicating that the NHSDA covers a high percentage of DUF arrestees.

The DUF study uses urinalysis, which detects drug use in the past 2-3 days fairly accurately, to measure drug use among arrestees. The previously mentioned surveys (NHSDA, MTF and DAWN) and BJS's inmate surveys all use self-report methods to measure drug use prevalence. The DUF project shows that, in 1992, the median value of males and females testing positive for at least one illicit drug in the combined DUF samples was 63% and 67.5%, respectively (see National InstituteofJustice,1994). Cocaine is the drug most frequently detected by urinalysis. In 1993, this number ranged from 19% « among male arrestees in Omaha to 66% in Manhattan. Fully 69% of the women in Cleveland tested t positive for cocaine, with a median of 46% across the combined city samples. Cannabis use was much lower with a range from 42% in Omaha and Indianapolis to 21 % in Manhattan among men, and from 25% in Indianapolis to 9% in Washington DC for women (with a median of 16.5%). Opiate positives ranged from 28% among men in Chicago to 1% in Fort Lauderdale. Among women, opiate positives range from 23% in Manhattan to 3% in Fort Lauderdale. Many arrestees tested positive for multiple drugs. All in all, the DUF study has demonstrated a very high rate of drug use among felony arrestees in major US cities, but there is not a readily discemible pattern. Slightly higher prevalence rates are consistently found in a few cities such as Manhattan, Chicago and San Diego, and lower rates in Portland and Omaha. Amphetamine use remained highest among arrestees in San Diego. There was little change in the numbers testing positive for at least one illicit drug among the combined city samples of males and females from 1988 to 1992, although there are some minor fluctuations (CESAR,1993). Cocaine use has increased in some cities and decreased in others, but the general pattern is one of stability. Following a slow, but steady decline, marijuana use began to rise in 1992 (NIJ, 1994). Opiate use has been stable to decreasing since 1988.

TABLE 1: Trends in arrest conviction and sentencing patterns for drug and non-drug offences in US district Courts 1980, 1985-92

No. of defendants in cases terminated in US District Courts

*Preliminary data for 1992 Source Bureau of Justice Statistics 1992

More so than any other piece of information, the DUF study has called into the question the validity of drug use estimates obtained using self report methods (see Harrison, 1995). DUF data show that only about half the felony arrestees who test positive for cocaine admit use in the last 3 days. Recent opiate use is reported by about 60% who test positive, and about 40% who test positive for amphetamines report recent use. Marijuana use was reported in the past month by about 55% of those who tested posizive. (Cannabis can be detected for up to 30 days in the urine of heavy users. )

Nevertheless, there are important nuances of the DUF study and data collection methods that must be factored into the interpretation of these data. Although the DUF study carries guarantees of anonymity and confidentiality, arrestees are interviewed in jail within 24 hours of their arrest - sometimes through cell bars. The study is not representative of arrestees, even in the cities from where the estimates are derived. The DUF samples are general Iy derived from a single central booking facility in the cities included in the study. Every city had more than one central booking facility. DUF arrestees generally constitute serious felony offenders who may not have been offered, or may not be able to post, bail. Inter views are conducted aloud, whereas studies show that self administered questionnaires produce higher drug use prevalence rates (Turner et al., 1992). Methodological studies also demonstrate that individuals are more willing to admit to more distant rather than more proximal drug use. Therefore, we would expect more underreporting of drug use in the last 3 days than in the last month. Although the results of the DUF study suggest that only about half of serious felony offenders are willing to report their recent drug use, these results are not inherently generalisable to the general population. Further, when considering the full range of information from the DUF study, the congruency between self-report and urinalysis reaches 70% and higher, depending on the particular drug. Therefore, the results of the DUF study should not be interpreted as damning of all self report statistics on drug use, especially those gathered in state of-the art general population surveys such as the NHSDA and MTF.

Reviewing the major surveys and incident-based studies on drug use in the USA demonstrates their conflicting results. Bearing in mind the limitations of each, perhaps the results are not so contradictory. Woven together, they paint a picture of drug use in US society. In the overview, illicit drug use is a relatively rare event in US society. Alcohol and tobacco use are much more prevalent. Mortality and morbidity data show illicit drugs to be only a moderately serious problem compared with alcohol and tobacco. However, illicit drug use is implicated in crime (but so too is alcohol), which has considerable ramifications for society.

Illicit drug and alcohol use are much more prevalent among populations involved with criminal justice. Many serious offenders use drugs heavily, and upwards of 13% report committing their crimes to obtain money to purchase drugs. Heroin, which is hardly measurable in the general population, has been decreasing among criminal justice populations, but still about 10% or more of serious offenders may have used recently. DAWN and treatment admission data suggest that the heroin-addicted population is an ageing cohort comprising primarily inner city minorities. The DUF study also shows that older arrestees are more likely to test positive for opiates. Cocaine is now the drug of choice among inmate populations, steadily increasing in prevalence since themid-1970s to the early 1990s.

DISCUSSION

American drug policy has traditionally been heavily dependent on the criminal law. The punitive approach is presumed to reduce drug use and abuse by making drugs more expensive and/or less accessible. It is also supposed to increase disapproval of drugs and lead to desistance, as well as to prevent youth from initiating. Incapacitating sellers and users should also reduce overall prevalence rates (Reuter, 1992). The federal drug control budget increased from$1.5 million in 1980 to $6.7 million inconstant dollars in 1990. It exceeded $10.98 billion in 1991, representing an almost sevenfold increase in 10 years. The budget request for 1996 is 14.6 billion, representing a 33% increase since 1991. Most of this money has gone to enforcement or supply side reduction programmes, with estimates of 80% and higher if both State and Federal expenditures are considered (Reuter, 1992). Supply side reduction includes activities such as interdiction, law enforcement and international supply reduction efforts. The remaining 20% went to demand side reduction efforts such as education, rehabilitation and research programmes. By far the largest increase in the budget has been for domestic law enforcement. The question is: Has this expenditure of funds and the increasing reliance on criminal justice policies helped to reduce drug use and its consequences for US society?

There are limited data available to respond to several of the premises on which current drug policy is based, but they suggest that drug policy is not having the desired effect. Availability and prices appear to have been largely unchanged. For example, marijuana availability has remained relatively stable with 80% or more of high school seniors in the MTF study saying it is 'very easy' to get since 1975. Cocaine has increased in availability since about 1986, although use has been dropping over the same interval among high school seniors. The NNICC report shows that the average price of most drugs has not increased in recent years, with the exception of cannabis. The overall rates of drug use have been decreasing in the general population; however, data suggest that concerns about the physically harmful effects of drugs has precipitated the decrease (Bachman et al.,1990). In the second half of the 1980s aggressive arrest policies led to large increases in the number of incarcerations. Reuter (1992) points out that taking both the increasing population of prisoners and the proportional increases in their size in jails and prisons, an estimated 450 000 additional drug users were removed from the population who might be involved in drug sales or regular use. However, the DUF study shows little changes in the trends in drug use among arrestees, so it appears that the stiffer sanctions are not impacting on these users. The BJS inmate studies show the use of drugs such as cannabis and heroin decreasing just like in the general population, but cocaine use has not decreased. Further, there has been no concurrent decrease in ER visits, the purity of drugs, and even the prevalence of weekly cocaine use in the NHSDA has been stagnant since 1985.

It appears that the biggest declines in drug use have been among casual users, those least likely to show up in jails or ERs. And although we have been able to bring about a considerable reduction in the prevalence of illicit drug use for society as a whole, the numbers of cocaine-related ER visits and deaths have been increasing since at least the late 1970s. Heroin related ER episodes increased fivefold between 1978 and 1993. These increases may be related to a number of factors, especially an ageing cohort of users. Reuter (1992) suggests that an increasing share of the drug-abusing population are inner city minority populations. The poorer users are more likely to be criminally active and their criminal activity may be exacerbated by drug use. It is likely that occasional users account for a small share of total consumption.

THE OFFICE OF NATIONAL DRUG CONTROL POLICY (ONDCP)

The 1988 Anti Drug Omnibus Control Act created the ONDCP - the Drug Czar's Office - in 1989. The role of ONDCP was to coordinate the drug related activities of a number of federal agencies so as, among other things, to inform policy making. Section 1005 of the Anti-Drug Abuse Act of 1988 required ONDCP to establish a strategy with 'comprehensive, research-based, long-range goals for reducing drug use in the United States', along with 'short term measurable objectives which the Director determines may be realistically achieved in the two year period beginning on the date of the submission of the Strategy'. The goals were shaped by the availability of credible data sources to a considerable extent. Perhaps more important goals exist but were not specified because of the difficulty in measuring outcomes. The established goals rely heavily on the NHSDA and, to a lesser extent, the MTF and DAWN studies. Of the 10 goals in the first and second strategies, six use the NHSDA, one uses the MTF study and one uses the DAWN study. The remaining two, which have to do with drugs entering the country and domestic marijuana production, have not been evaluated because no acceptable measures are available.

Since the 1989 and 1990 strategies were printed within 4 months of one another, their 2 year goals should be evaluated based on data from the 1990 and 1992 NHSDAs. We find the following:

1. The 10% reduction in past month illicit drug use in strategy 1 and the 15% reduction in strategy 2 were both met between 1990 and 1992.

2. The 10% reduction in the number of adolescents reporting any illegal drug use in the past month in strategy 1 was met, but the 15% in strategy 2 was not met.

3. The 10% reduction in the number of occasional cocaine users in strategy 1 and the 15% reduction in strategy 2 were both met between 1990 and 1992.

4. The goal of a 50% reduction in the numbers reporting weekly cocaine use in strategy 1 and the 60% goal in strategy 2 were not met

5. The 20% reduction in the number of adolescents reporting past month cocaine use in strategy 1 and the 30% reduction in strategy 2 were not met.

Neither of the DAWN based goals of a 10% reduction in ER visits for illicit drugs in strategy 1 between 1989 and 1991 and a 15% reduction between 1990 and 1992 in strategy 2 were met, nor were the MTF based goal of a 10% reduction in the number of high school seniors who disapprove of illegal drug use in strategy 1 or a 20% reduction in strategy 2 met. All in all, a fairly disappointing scorecard.

The 1993 Interim Strategy represented a major departure from the prior strategies (ONDCP, 1993 ) It rejects the use of the drug war analogy. User accountability no longer forms the core of the drug programme, but rather helping hard core users. The Interim Strategy shifted the focus away from the 'easy' part of the drug problem- reducing casual intermittent use - to reducing drug use and its consequences by hard-core users. It proposes to mount an aggressive drug treatment strategy to reduce the number of hard-core users, and proposes to give al] drug users access to treatment service through Health Care Reform. It views the drug problem not in isolation, but as inextricably linked to other issues such as health, the economy, violence, and family and community stability. The new policy also recognises that efforts to stop the flow of narcotics into the USA have failed. The interdiction focus has moved towards a new campaign to persuade producing counties to shut off the flow and disrupt international trafficking syndicates. Changes in the USA's interdiction policy of recent years have been difficult to implement in practice because foreign aid was often tied to interdiction efforts. These changes in policy perhaps signal that policy-makers are paying more attention to research and proven results. However, before patting ourselves on the back, let's examine a study of 'policy-makers' and how they view research.

THE DC METROPOLITAN AREA OPINION LEADERS' STUDY

Many lessons and implications can be drawn from the various surveys and studies available on drug use, depending on the question of interest. However, policy makers rarely have the necessary knowledge to understand the strengths and limitations of the various surveys, let alone integrate the findings. Researchers should not expect them to. Too often, policy makers seem to focus on one particular piece of information, study or group of studies, without taking a fuller range of information into account. It is the responsibility of researchers to translate the results of their research into language that the policymakers understand. Forging a closer working relationship between policy and research would ensure that researchers understand the information policymakers require, and policy-makers understand the limitations of research.

A survey of Area Opinion Leaders was conducted in 1 991 by NIDA in the Washington DC Metropolitan Statistical Area (MSA - basically the city and the surrounding area which are economically interdependent) (see NIDA, 1993). The study sought to obtain the opinion of about 200 individuals who influenced policy decisions and, directly or indirectly, addressed the drug abuse problem. Initial respondents were selected, and snowball sampling methods were used in which the selected opinion leaders recommended others who met the particular characteristics specified for inclusion in the study. Lists of potential respondents were then developed and individuals were randomly selected from those lists. The sample was stratified by jurisdiction ( DC, Maryland or Virginia), type of responsibility (decision maker or expert) and substantive focus (education, criminal justice, public health and government). The final sample size was 162 for an 81% representation rate.

As the DC MSA is a unique area, the results of the Opinion Leaders study may not be generalisable to other areas. Nevertheless, some interesting results were obtained that have implications for researchers and others interested in how policy makers inform themselves. The opinion leaders were asked to name and assess the accuracy of their sources of information on drug abuse. Most opinion leaders mentioned more than one source of information, with the most common being direct contact with drugs users or drug programmes (86%), followed by the media (82%) and research on drug use (78%). A variety of other sources was mentioned to a lesser extent. When asked to assess the accuracy of information from the three sources, 53% of opinion leaders perceived their direct contacts with drug users or drug programmes as very accurate, compared with research which was seen as very accurate by 30%. The media were seen as the least accurate source of information with only 6% saying it was very accurate, although 72% said the media were a somewhat accurate source of information.

Of the opinion leaders who reported having used research, about a third (36%) said it was very helpful and 57% said it was somewhat helpful. The most frequent criticisms levelled against drug-related research (from a list) were that research does not ask the right questions to achieve useful results (50%), and that it focuses on inappropriate populations ( i.e. the general population who are not typically drug users) (49% ). An estimated 41% said that research is typically biased in its approach, 39% said that it presents contradictory results, and 39% said that it rarely yields useful recommendations. However, few agreed on the type of information they needed from research. Criminal justice and education officials were the most critical of research saying that it did not ask the right questions (58%). Criminal justice officials were most likely to criticise research for not focusing on the right populations (68%). Public health and government officials were less critical of research, but 44% said that it did not ask the right questions and focused on the wrong populations. Government officials were the least critical of drug research, although 44% said that it was biased in its approach. *

The opinion leader's interest in the practical utility of research results was manifested in their differential information needs by their professional responsibilities. Criminal justice officials and education officials were most often interested in descriptive epidemiological data such as the demographic characteristics of users. Public health officials were most often interested in the effectiveness of various prevention and treatment measures (41% ). Government officials ( 63.6% ) were most often interested in the methodological features of the research itself, such as the validity and reliability, as well as the replicability of the research. This interest may indicate their desire to evaluate the research results before using them to direct policy.

The opinion leaders expressed a need for greater practical utility and applicability of research results. Almost two in five opinion leaders criticised research for failing to make recommendations. They also expressed concern that hidden or hard-to-reach populations, who are the most likely to use drugs, are the least often studied. Obviously, there is a need for greater collaboration between the research and policy-making communities. Researchers need to think about the policy relevance of their research studies. Policy-makers need to make their needs for particular types of information known to the research community and funding agencies. Improving communication would go a long way towards improving the policy relevance of research, and providing more effective drug policy based on proven research results.

Lana D. Harrison, PhD, Center for Drug and Alcohol Studies, University of Delaware, 17 E. Main St, Newark, DE 19716, USA.

REFERENCES

Bachman JG, Johnston LD, O'Malley PM (1990). Explaining the recent decline in cocaine use among adults: Further evidence that perceived risks and disapproval lead to reduced drug use. Journal of Health and Social Behavior 3 1: 173-84.

Bray RM, Kroutil LA, Luckey JW, Wheeless SC, lannacchine VG, Anderson DW, Marsden BME, DuntemanGH (1992). Highlights 1992: Worldwide Survey of Substance Abuse and Health Behaviors among Military Personnel. Washington DC: Assistant Secretary for Defense, Department of Defense.

Bureau of Justice Statistics (BJS) (1988) . Drug Law Violators 198S86. Washington DC: US Department of Justice, BJS.

Bureau of Justice Statistics (1991) . Drugs and Jail Inmates. Washington DC: US Department of Justice, BJS.

Bureau of Justice Statistics (1992) . Federal Criminal Case Proœssing, 1982-91. Washington DC: US Department of Justice, BJS.

Bureau of Justice Statistics (1993). Survey of State Prison Inmates,1991. Washington DC: US Department of Justice, BJS.

CESAR (1993) . Criminal populations in U.S. continues to be plagued bydrug abuse. CESAR FAX 2(46): 22November.

Criminal Justice Newsletter (CJN ) (1994) .21 % of Federal pris oners found to be low level drug offenders. Crirninal Justice Newsletter25(3): 7.

Gfroerer JG, Brodsky MD (1993). Frequent cocaine users and their use oftreatment. American Journal of Public Health 83: 1149-54.

Harrison LD (1993) . The strengths and limitations of self report surveys in monitoring drug prevalence. Paper present ed at the 45th meeting of the American Society of Criminology, Phoenix, October 27-30.

Harrison LD ( 1995). The validity of self reported data on drug use.Journal of Drug lssues 25( 1): 91-112.

Harrison LD, Gfroerer J ( 1992) . The intersection of drug use and criminal behavior: Results from the National Household Survey on Drug Abuse. Crime and Delinquency 38: 422-43.

JohnstonLD, O'MalleyPM, BachmanJG ( 1 994).NationaLSurvey Results on Drug Use from Monitoring the Future Study, 1975-1 993. Washington DC: NIH Publication No. 943809.

Maguire K, PastoreAL, FlanaganTJ (1993). Bureau of Justice Statistics Sourcebook of Criminal Justice Statistics -1992. Washington DC: US Department of]ustice, BJS.

Mensch B, Kandel D (1988). Underreporting of substance use in a national longitudinal youth cohort: individual and interviewer effects. Public Opinion(2uarterly 52: 100-24.

National Institute of Justice (NIJ) (1994). Drug Use Forecasting in 1993: Annual Report on Adult Arrestees. National Institute of Justice, US GPO387 159/00045: November.

National Institute on Drug Abuse (NIDA) (1993) . Views of Area Opinion Leaders about Drug Abuse in the Washington, DC, Metropolitan Area: 1991. Rockville, MD: National Institutes of Health, NIDA.

National Narcotics Intelligence Consumers Committee (NNICC) (1993) . The NNICC Report 1992. Arlington, VA: Drug Enforcement Administration.

Office of National Drug Control Policy (ONDCP) (1989). Nanonal Drug Control Strategy. Washington DC: US Govemment Printing Office.

Office of National Drug Control Policy (1990) . National Drug Control Strategy. Washington DC: US Government Printing Office.

Office of National Drug Control Policy (1991). National Drug Control Strategy. Washington DC: US Government Printing Office.

Office of National Drug Control Policy (1992). National Drug Control Strategy. Washington DC: US Government Printing Office.

Office of National Drug Control Policy ( 1993 ) . Breaking the Cycle of Drug Abuse, 1993 Interim National Drug Control Strategy. Washington DC: US Government Printing Office.

Office of National Drug Control Policy (1995) . National Drug Control Strategy: Strengthening Communities' Response to Drugs and Crime. Washington DC: US Government Printing Office.

O'Malley PM, Bachman JG, Johnston LD (1984). Reliability and consistency in self-reports of drug use. Intemational Joumal of the Addictions 18: 805-24.

Reuter P (1992) . Hawks ascendant: The punitive trend of drug policy.Daedalus 121(3): 15-52.

Reuter P (1993) . Drug policy: Recent lessons . Congressional Testimony. Santa Monica: The Rand Corporation.

Substance Abuse Mental Health Service Administration (SAMHSA) (1993). National Household Survey on Drug Abuse: Main Findings 1991. Rockville, MD: DHHS Pub. No. (SMA)93-1980.

Substance Abuse Mental Health Service Administration ( 1994).Estimates from the Drug Abuse Waming Network. Rockville, MD: SAMHSA.

Turner CF, Lessler JT, Gfroerer JC (Eds) (1992) . Survey Measurement of Drug Use: Methodological Studies. Rockville, MD: DHHS Pub. No. (ADM)92-1929.

Uniform Crime Reports (1993). Crime in the United States, 1992.US Department of Justice, Federal Bureau of lnvestigation, Uniform Crime Reports.

University of Michigan (1994) . Press release on the 1993 Monitoring the future survey. News and Information Service.

Wish ED (1991) . U.S. drug policy in the 1990s: Insights from new data from arrestees. Intemational Joumal of the Addictions 25: 377-409.