WHAT HAPPENED TO THE "CRACK BABIES"?

Gary A. Emmett, MD

A few years ago the newspapers and medical journals were full of apocalyptic articles predicting that urban school districts would be totally overwhelmed with crack babies born to women caught up in the epidemic of the late 1980s and early 1990s. These children were going to be worse than any previous set of children of addicts and would be the Attention Deficit/Hyperactivity Disease children from hell. It is now 1997 and our urban schools are overwhelmed by middle class flight and the social disasters of our inner cities, but there is no overwhelming set of crack babies in the early primary grades. What happened?

Primarily, despite what the newspapers and medical journals said in the nineties, there is no specific "Crack Baby" syndrome corresponding to the Fetal Alcohol Syndrome (which is having the severe effects on many Native American reservation schools in the western US) or the untreated tertiary syphilis syndrome that was a problem in our schools 70 years ago. Crack cocaine does not produce a syndrome: a clearly recognizable series of abnormalities that is consistent from patient to patient. Exposing fetuses to cocaine may or may not have lasting consequences, but current research demonstrates that by the time the child reaches age five, the effect of the disastrous social situation that many crack cocaine users share with other economically deprived children washes out any measurable effect of the cocaine itself on these children's school performance.

Dangers of Maternal Cocaine Use

There are dangers to the mother and child if the mother is using crack cocaine. (Pharmacologically, powder should have similar effects, but there aren't good human studies.)

Women who use crack cocaine are more than twice as likely to deliver a low birth weight infant (very early, very wasted or both) than socially matched controls. Users are three times as likely to have placental abruptions or deliver an infant with a low 5-minute Apgar score (a test of whether the child needs to go to the intensive care nursery), but these two outcomes may be due to the use of other drugs in conjunction with cocaine. (Sprauve, et al, 1997; Regalado et al, 1996; Eyler et al, 1995) After the child is delivered, even when adjusting for the fact that many crack cocaine users are multi-drug users, the babies of crack users are less autonomically stable (they have trouble regulating temperature, breathing, heart rate, blood pressure), have poorer fine and gross motor function soon after birth and have more abnormal reflex reactions. (Richardson et al, 1996, Bender et al, 1995) The more cocaine used, the worse the short-term neurological outcome. (Martin et al, 1996) Crack babies also are much more expensive for society, with markedly increased medical costs compared to controls. (Behnke et al, 1997) But these effects do not constitute a syndrome because they seem to gradually disappear and are not easily recognizable by age 3 years.

After the child leaves the hospital the situation does not improve. Crack using mothers are not very loving or giving mothers in general (Hawley et al, 1995) and as a consequence their children do not learn these interpersonal skills very well. They may have permanent change in their brain wave patterns with unknown long-term results. (Prichep et al, 1995) Because cocaine is a powerful vasoconstrictor, vascular accidents may occur in utero, leaving poorly vascularized or even missing portions of the gastro-intestinal tract or, more commonly, leaving similar areas in the brain. Poorly vascularized areas leave leukomalacia (softening and scarring of the white matter of the brain) especially in the basal ganglia or around the ventricles and totally unvascularized areas leave porencephalic cysts (literally holes in the brains) (Cohen et al, 1994). My clinical experience suggests that these lesions do not disappear and may result in any outcome from no apparent damage to permanent and severe motor or developmental delay (in plainer English, cerebral palsy and/or mental retardation).

Learning Problems

There are difficult-to-educate young students, but they may not be primarily the result of crack cocaine. In a long term study out of the poorest areas of Philadelphia, Dr. Hallam Hurt has shown that the emotional impoverishment of the homes where women take crack cocaine has more significance on school performance by school entry than does the use of cocaine during the pregnancy (at least for those children who survived the neonatal period) (Hurt 1997b, 1997c). We have known for a few years that the mothers who took crack cocaine were demographically different than other mothers delivering at the same hospitals in that they were older; more likely to use nicotine, alcohol and other street drugs; less likely to be married and more likely to be black (Richardson and Day, 1994), so Dr. Hurt controlled for those factors. Her preliminary work (Hurt, Brodsky et al 1995; Hurt, Malmud et al 1997) showed that play at 18, 24 and 30 months was not different in amount or kind for children exposed to crack cocaine versus controls matched for income, parental education, and family situation. What is very important is that although the developmental quotient of the cocaine and matched controls did not differ from each other over time, their development steadily fell compared to an age matched middle class group (Hurt, Brodsky et al 1995). By age 5 the developmental quotient of both the crack cocaine exposed children and the matched controlled group was almost 1 standard deviation below the middle-class group. Those children in both study groups whose developmental quotients were greater than 90 (less than 25% of the total) differed from their less developed agemates only in having more enriching home experiences and caretakers who were more involved in their lives (Hurt, 1997a).

None of these conclusions is a surprise to anyone providing primary pediatric care to the poor. Pediatricians know that school performance is tied primarily to the quality of home personal interactions and the amount of structured learning given by the parents, not genetics.

Causes of Learning Problems

Why then the myth of the crack baby invasion of our schools? We see what crack cocaine does to our patients and to our cities, so we expect it, even want it, to be evil. When Dr. Hurt presented her papers at the national pediatric research meetings this year she was attacked as if she was promoting cocaine because the results were not bad enough for the audience. The medical establishment expected cocaine to cause lasting harm, because other drugs of abuse do so. Fetal alcohol syndrome is clearly associated with drinking alcohol (although the cofactors causing the syndrome are not clear) and causes microcephaly, unusual facial configuration and severe learning disorders. Maternal opiate use can lead to neonatal addiction; the resulting withdrawal syndrome can result in death if not treated. To hear that cocaine did not cause as much long-term damage as nicotine was a surprise.

There is a widespread belief that the mother taking crack cocaine is raising children not prepared to cope with the world and this generates a strong desire to punish that mother. Last October, South Carolina's Supreme Court upheld a law that a mother who delivers an infant with cocaine in its urine has committed a felony and should go to jail. Multiple studies show that foster care may actually be worse for the child's eventual outcome than being in the care of a crack-using mother. The most successful intervention has involved intensive teaching of mothering skills combined with drug rehabilitation and the prospect of imprisonment if the course was not finished.

Newborns born to cocaine-using mothers do have clear neurological problems, and these problems may persist. But in the poorest, least educated segment of our population the poor development caused by these children's social situation is more devastating than the crack exposure. To find the effects of in utero crack exposure on the nervous system of older children, if there is one, a study must be done among rich, educated mothers who took cocaine during their pregnancies, and that study has not been done.

In short, cocaine use by a pregnant women may cause premature birth, poor fetal weight gain, and vascular accidents in utero, but the poor school performance of crack babies is more closely tied to the environment than to the crack they have been exposed to. The best policy for crack babies is teaching their mothers how to be better mothers. The question is: Do we know how?

Behnke M. Eyler FD. Conlon M. Casanova OQ. Woods NS. How fetal cocaine exposure increases neonatal hospital costs. Pediatrics. 99(2):204-8. 1997 Feb.

Bender SL. Word CO. Diclemente RJ. Crittenden MR. Persaud NA. Ponton LE. The developmental implications of prenatal and /or postnatal crack cocaine exposure in preschool children: a preliminary report. Journal of Development & Behavioral Pediatrics. 16(6):418-30. 1995 Dec.

Cohen HL. Sloves JH. Laungani S. Glass L. DeMarinis P. Neurosonographic findings in full-term infants born to maternal cocaine abusers: visualization of subependymal and periventricular cysts. Journal of Clinical Ultrasound. 22(5):327-33. 1994 Jun.

Eyler FD, Behnke M. Prenatal cocaine exposure. Consequences for child and family. Journal of the Florida Medical Association. 82(9):603-5. 1995 Sep.

Hawley TL. Halle TG. Drasin RE. Thomas NG. Children of addicted mothers: effects of the crack epidemic on the caregiving environment and the development of pre-schoolers. American Journal of Orthopsychiatry. 65(3):364-79. 1995 Jul.

Hurt H. (1997a) Inner-city achievers: Who are they? Abstract presented at American Pediatric Society/ Society for Pediatric Research meetings 1997.

Hurt H. (1997b) Inner-city children with in utero cocaine exposure (COC) do not differ from Controls (CON) in problem solving ability. Abstract presented at American Pediatric Society/ Society for Pediatric Research meetings 1997.

Hurt H. (1997c) A prospective comparison of children with in utero cocaine exposure (COC) and Controls (CON) on the Battelle Developmental Inventory. Abstract presented at American Pediatric Society/ Society for Pediatric Research meetings 1997.

Hurt H. Brodsky NL. Betancourt L. Braitman LE. Malmud E. Giannetta J. Cocaine-exposed children: follow-up through 30 months. Developmental and Behavioral Pediatrics. 16(1):29-35. 1995 Feb.

Hurt H. Malmud E. Betancourt L. Brodsky NL. Giannetta J. A prospective evaluation of early language development in children with in utero cocaine exposure and in control subjects. The Journal of Pediatrics. 130(2):310-2. 1997 Feb.

Martin JC. Barr HM. Martin DC. Streissguth AP. Neonatal neurobehavioral outcome following prenatal exposure to cocaine. Neurotoxicology & Teratology. 18(6):617-25. 1996 Nov-Dec.

Prichep LS. Kowalik SC. Alper K. de Jesus C. Quantitative EEG characteristics of children exposed in utero to cocaine. Clinical Electroencephalography. 26(3):166-72. 1995 Jul.

Regalado MG. Schechtman VL. Del Angel AP. Bean XD. Cardiac and respiratory patterns during sleep in cocaine-exposed neonates. Early Human Development. 44(3):187-200. 1996 May.

Richardson GA. Day NL. Detrimental effects of postnatal cocaine exposure: illusion or reality? Journal of the American Academy of Child & Adolescent Psychiatry. 33(1):28-34. 1994 Jan.

Richardson GA. Hamel SC. Goldschmidt L. Day NL. The effects of prenatal cocaine use on neonatal neurobehavioral status. Neurotoxicology & Teratology. 18(5):519-28. 1996 Sep-Oct.

Spauve ME. Kindsay MK. Herbert. Graves W. Adverse perinatal outcome in parturients who use crack cocaine. Obstetrics & Gynecology. 89(5 pt 1):674-8. May 1997.

THE DISTRIBUTIVE POLITICS OF DRUG POLICY

David Boyum

Almost all students of American drug policy believe that the current regime is excessively punitive. Even John DiIulio, who enjoys the most hawkish reputation among well-known scholars, has argued that we sentence too many non-violent drug offenders to long prison terms.

When analysts claim that American drug policy is unduly punitive, they generally mean one of two things. The more modest assertion is that, compared to alternative policies, drug enforcement is not cost-effective at the margin. Some argue that we would get more for our criminal justice dollar if we incarcerated fewer drug offenders and more non-drug lawbreakers; others maintain that a dollar spent on treatment programs reduces drug use more than a dollar spent on enforcement. The bolder position is that, at the margin, aggressive drug enforcement and the punishment of drug offenders now do more harm than good. Good here is usually defined in terms of reduced drug use, while harm includes public spending on police, courts, and prisons, as well as the crime, violence, disorder, and corruption associated with illicit drug markets.

But if analysts have explained what they see as wrong with current policy, they haven't explained why our political representatives and public officials have given us an ineffective policy. And their failure to do so is not a side issue. For the worth of a policy recommendation is not simply a function of the public benefit that would accrue if the policy were adopted; one has to weigh the political feasibility of the policy and a realistic forecast of how its actual implementation would differ from its initial design. It is not hard to demonstrate how some completely-redesigned-from-the-ground-up, impervious-to-special-interests drug treatment system would be a vast improvement over what we have now. Then again, if automobile manufacturers didn't have to worry about building and selling cars, designing them would also be easy. But just as you can't design next year's model unless you understand how this year's is built, you can't plan tomorrow's public policy if you don't comprehend the politics of today.

Drug Policy Politics

As a first step towards a political understanding of drug policy, take what is probably the most common policy advice of academic experts: that we should spend less on enforcement and more on treatment. A standard starting point in political analysis is to assume that all government action produces winners and losers, and that the relative political strength of each is an important explanation and predictor of political outcomes. Who, then, stands to gain and lose by shifting public resources from enforcement to treatment?

As noted earlier, the chief benefit of vigorous drug enforcement is reduced drug use. The principal costs are public expenditures and the negative side effects of black markets. (Some would add to the debit side of the ledger the individual, family, and neighborhood effects of locking up large numbers of young minority men.) Now, as a thought experiment, consider how these benefits and costs stack up from the perspective of two types of communities: middle-class suburban towns and poor inner-city neighborhoods.

If we use drug tests given in suburban and inner-city hospitals as an indicator, cocaine and heroin use is rare in the suburbs, but not in inner cities, where it is at least ten times as common. And if we limit our focus to chronic abuse of these substances--that is, to those most in need of treatment services--the difference in prevalence would surely be even greater. Drugs can be purchased in the suburbs, but not easily (which is why New Jersey license plates are seen at drive-through drug markets in Manhattan's Washington Heights), and the arrest and imprisonment of a suburban resident on drug charges is a newsworthy event. Almost unheard of is the kind of open drug selling that plagues inner-city neighborhoods.

Because poor urban communities bear more than their fair share of most social problems, it is reasonable to ask whether drug enforcement is to blame for the concentration of drug problems in inner cities. The probable answer is that inner cities would fare worse than suburbs under any drug policy, but that prohibition and zealous enforcement have magnified the difference. As evidence of this proposition, note that while alcohol abuse and alcohol-related violence are more common in poor neighborhoods, the distribution of these licit drug problems looks egalitarian when set against the distribution of illicit drug problems. This is hardly surprising: because harsh punishment of drug dealing makes it a risky and disreputable activity, we would expect drug markets to locate in those communities whose social structure is least resistant to them, and in which there are young men whose futures are sufficiently bleak (or whose discount rates are sufficiently high) to make drug dealing look like an opportunity. Much the same could be said about the likely effect of enforcement on the social distribution of drug abuse.

Treatment Serves the Poor

What all of this suggests is that an expansion in publicly funded drug treatment at the expense of drug enforcement is a bad deal for the middle class--at least when judged from the viewpoint of narrow self-interest. Drug enforcement offers the suburbanites considerable protection from drug abuse, and they suffer few of the unintended side effects. And to the extent that the middle class could benefit from improved access to treatment services, the key issues are private health insurance coverage and job protection for those who acknowledge addiction. Publicly supported treatment programs mainly serve poor addicts, who would clearly gain from additional funding. So too would the families and criminal victims of poor addicts, who are also predominantly poor. Inner city residents might benefit as well from a reduction in drug enforcement, although that would probably depend on which aspects of enforcement were relaxed.

While this analysis is far from a comprehensive account of why our drug policy so favors enforcement over treatment, it certainly makes the enforcement bent more understandable. It also, depending on one's perspective, makes current policy look more or less justifiable. That the overwhelming majority of Americans would find themselves ill-served by a shift in public resources from drug enforcement to drug treatment is not something that drug policy critics have told us, nor is it a small consideration in a democratic political system. On the other hand, if one believes that public policy should give disproportionate weight to the welfare of the least fortunate, our present drug policy is hard to defend.

IN WHAT SENSE (IF ANY) IS MARIJUANA A GATEWAY DRUG?

Robert MacCoun

The notion that cannabis use is a "gateway" to hard drug use is a time-honored staple of the drug policy debate. Given the relative subtlety of marijuana's direct harms, the gateway notion has served as a major rationale for sustaining (or escalating) cannabis prohibition since the mid-1960s. For at least as long, doves have dismissed the evidence for the gateway as spurious, trivial in magnitude, or both. The lack of agreement may have less to do with self-serving interpretations of the data than with considerable confusion about the meaning of the gateway concept. In this brief essay, I suggest that there are at least 7 coherent (though not necessarily credible) interpretations of the evidence; some support hawkish views, some support quite dovish alternatives, and one even suggests the whole issue might have little policy relevance.

Version 1: The First Step. The most simplistic version is that (a) almost all heroin and cocaine users first used cannabis, and so (b) cannabis must be a stepping stone toward hard drug use. The evidence for proposition (a) is overwhelming, at least in this country and many other Western nations. But of course the inference to (b) is a non sequitur; temporal precedence is necessary but not sufficient to establish causality. Thus, doves frequently dismiss the gateway notion as a classic "post hoc, ergo propter hoc" fallacy: "Yes, but before cannabis, they started with milk; is milk a gateway substance?" This retort demolishes Version 1, but leaves the other interpretations untouched. (I'll examine a more sophisticated rebuttal, in which alcohol replaces milk, later.)

But in addition to temporal precedence, causal claims require evidence of correlation or statistical association. One way of thinking about statistical association is in terms of conditional probabilities, or relative risk. Is the probability of trying cocaine higher for those who have tried cannabis than for those who have not? Do cannabis triers face a higher statistical risk of hard drug addiction? Yes. The evidence for a correlation between cannabis use and hard drug use is also overwhelming (e.g., Kandel, Yamaguchi, and Chen, 1992). Here's where the "milk" argument falters; although all cocaine users (except a few lactose intolerants) have tried milk, there is surely no correlation between milk and cocaine use.

Version 2: The Spurious Correlation. Like temporal precedence, correlation is necessary but not sufficient to establish causality. It is plausible that the correlation between pot and cocaine (or heroin, or LSD) is spurious, reflecting some third factor that puts people at risk for use of both cannabis and hard drugs. Generating good candidates isn't hard; we've long known that behaviors like smoking, drinking, and illicit drug use tend to cluster in adolescence (e.g., Fergusson & Horwood, 1997), along with other problem behaviors like truancy, poor grades, fighting, property crime, and teenage pregnancy; in recent years we've discovered that depression belongs on the list. It is likely that for most problem youth, these are all manifestations of an unfortunate interaction of dispositional factors (poor self control and socialization skills, hyperactivity, and sometimes even neurological deficits) and bad situations (economic deprivation, limited or inappropriate adult supervision, neighborhood disorder). Investigators have put a lot more effort into establishing the temporal and correlational fingerprints of the gateway than into ruling out these alternative explanations.

Version 3: The Early Warning. Of course, even if the link between cannabis use and hard drugs isn't causal, early cannabis use might serve as an early warning signal. From this perspective, even if cannabis doesn't cause hard drug use, it reliably precedes it and predicts it. The data support this interpretation, but they also show that the diagnostic value is limited. As a signal, cannabis mostly generates false alarms. The majority of cannabis triers never try harder drugs, and of those who do, few become regular users, much less addicts. A brief adolescent phase of "acting out" in delinquency and casual drug experimentation is well within the range of statistical normality in the US and many other nations, and has been for generations. Among those who do proceed from pot to harder drug use (especially heavy use), one finds some statistical clustering of other problem indicators - a higher propensity toward criminality, joblessness, and so on. But adolescents who casually experiment with cannabis appear to function quite well with respect to schooling and mental health (Shedler & Block, 1990). Acknowledging these facts doesn't condone such experimentation; low risks in the aggregate can have big effects for some individuals, and minute harms at the individual level might sum up to important aggregate effects. But it does complicate one's interpretation of the cannabis signal. On the other hand, longitudinal research does indicate that some patterns of cannabis use - early onset and/or high frequency - are more reliable signals of subsequent problems with other drugs and other risky behaviors (e.g. Fergusson & Horwood, 1997).

Cannabis use is not without harm, especially for adolescents, but as a source of danger it is surely trumped by alcohol, tobacco, reckless driving, criminality, and unsafe sexual behavior. Unless of course it actually promotes hard drug addiction. But it is devilishly difficult to establish a causal role for cannabis; the surest test would be a controlled experiment, a prospect we are ethically precluded from pursuing. And even if one were to stipulate that cannabis plays a causal role in hard drug use, the policy implications are less clear than generally imagined; much depends on the nature of the linkage.

Version 4: The Trap. Version 4 is a "straw man" argument; I'm unaware of any hawks that seriously endorse it, but doves knock it down with great relish. The notion is that cannabis use inexorably (with high probability) yanks one into the trap of hard drug use. Thus one recent analysis argued that a gateway would exist if 75% or more of cannabis users went on to hard drugs. Since they didn't, and don't in any sample I'm aware of, one can argue that the gateway has been refuted. This purely deterministic notion of causality may work for billiard balls and other inanimate objects (at least above the subatomic level), but surely no one believes it applies to cause and effect in human actions. Thus, the fact that most cannabis triers never try hard drugs is not sufficient to exonerate cannabis from a causal role.

Version 5: The Tantalizer. Version 5 appears to be what most hawks actually have in mind. In this version, cannabis isn't a trap so much as a tease, tantalizing unwitting (and stoned) young minds with the prospect of experimentation with ever more intriguing varieties of intoxication. The idea is hardly preposterous on its face. Few go straight to Miro without passing through Magritte, and the same applies to complexity in music, food, wine, and literature. Still, plausibility isn't evidence; where's the beef? Until recently, one was hard pressed to find any expert endorsement of this version (see critiques by Kaplan, 1970; Kleiman, 1992). But much was made of two recent papers in Science (June 27, 1997) demonstrating that cannabis activates neurochemical processes that respond in qualitatively similar ways to cocaine, heroin, tobacco, and alcohol. Popular media coverage suggested a "smoking gun" - hard scientific evidence for a neurological basis for the gateway effect, an impression the source articles did little to discourage.

Naturally, few drew the equally plausible (or implausible) interpretation that alcohol is a gateway drug. In fact, alcohol is a bit of a red herring here. With respect to temporal precedence, the evidence that almost all hard drug users first tried alcohol is also overwhelming; moreover, for most of them, alcohol preceded cannabis. But experience with alcohol is so common in our culture that as a statistical correlate of hard drug use, having tried alcohol is more akin to having tried milk than having tried cannabis. (Prolonged heavy drinking is another matter altogether. Under any theory, the gateway notion is more credible when the gateway substance is used heavily and frequently, rather than lightly and sporadically.)

Arguably, these Science articles captured our attention because of the widespread belief that if scientists in lab coats can observe effects of cannabis on brain chemicals, than the locus of causation must be biological rather than psychological or sociological. I won't speak for sociologists, but every psychologist I know accepts the notion that neurochemical processes are involved in everything psychological, and everything sociological. (Your attitudes toward Prince and Prince Charles have neurochemical signatures, and some day we may know how to read them.) Identifying those neurochemical processes is of tremendous value, but establishing their existence hardly settles the issue of causality. Neither study actually examined the relationship between rats' cannabis consumption and their motivation to use harder drugs. And several critics have pointed out that rats won't even self-administer cannabis, raising questions about the applicability of the "rat model" to human behavior (see Coffin, 1997; also the exchange of letters in the August 8, 1997 issue of Science). This latter point is particularly troubling, because one's interpretation of these new studies depends on how one translates the rat doses to human equivalents--equivalents with respect to relevant psychological and physiological effects, and not just body weight. At one extreme, the rat doses might only compare to extremely heavy human consumption. If so, their generalizability to typical cannabis experience is rather remote. At the other extreme, the rat doses might be comparable to fairly typical human doses. If so, these new studies might actually imply that the neurochemical processes in question have a weaker hold on human behavior than many had assumed, since we know that dependence is relatively uncommon among typical human users (10% is a typical upper bound figure.).

Version 6: The Toe in the Water. Alternatively, a causal link between cannabis and later hard drug use might be more sociological than physiological (Kaplan, 1970; MacCoun, Reuter, & Schelling, 1996). For example, seemingly safe experiences with marijuana might reduce the perceived riskiness of harder drugs - the health risks, the legal risks, or both. Indeed, if one's experiences fail to confirm the dire predictions of prevention programs, one might discredit warning messages about the dangers of cocaine or heroin. If so, perhaps public information campaigns should draw a clearer distinction between marijuana and more dangerous drugs. Or experience with marijuana use without getting caught (as is true of the vast majority of such experiences) may challenge one's view of the omnipotence of law enforcement (see Parker & Grasmick, 1979). If so, a gateway might result from a reduction in the perceived legal risks of using harder drugs. One solution might be to drastically enhance the probability of detection of marijuana use (e.g., through mandatory drug testing); this approach is hard to justify absent more compelling risk evidence, and might well backfire politically. Alternatively, one might undermine the gateway by largely or completely depenalizing marijuana, to take it out of the realm of illicit behavior.

Version 7: The Foot in the Door. Finally, cannabis experience might cause hard drug experience indirectly, by bringing causal experimenters into contact with hard drug sellers. Indeed, sociologist Herman Cohen's (1972) articulation of this theory was instrumental in persuading the Dutch to permit low level cannabis sales in coffeeshops and nightclubs; the Dutch argue that this approach separates the soft and hard drug markets, weakening the gateway. In Amsterdam, as in the US, almost all hard drug users have used cannabis, but the vast majority of cannabis users have not used hard drugs (Cohen & Sas, 1997; MacCoun & Reuter, 1997). Only 22% of Amsterdam residents aged 12 and over who have ever used cannabis have also used cocaine. This compares to a figure of 33% for the United States. For heroin, the corresponding figures are 4% for Amsterdam and 3% for the U.S.--statistically identical. Thus, though the Dutch have failed to eliminate the statistical association between cannabis and hard drug use, it is possible that they have weakened it, at least for cocaine. But since the alleged gateway is a function of both the number of people who've tried marijuana and the probability of cocaine use given marijuana use, any increase in the former component (the prevalence of marijuana use) might offset reductions in the latter component (the probability of moving on to cocaine use). Unfortunately, there are reasons to believe the Dutch commerialization of cannabis may have had such an effect (see MacCoun & Reuter, 1997), although the net result of the two components is difficult to estimate.

Conclusion. Given our current state of knowledge, one can coherently argue that (a) the gateway is a myth--it doesn't exist; (b) the gateway is very real and it shows why we must sustain or strengthen our ban on marijuana, or (c) the gateway is very real and it shows why we should depenalize or even legalize marijuana. It all comes down to how one is defining the term "gateway," and which interpretation of the available evidence one favors. But this isn't an argument for nihilism, radical deconstructivism, or the like. Rather, it suggests the need for more focused studies of potential gateway mechanisms. For example, animal experiments might directly test the effects of cannabis exposure (frequency, duration, setting) to the propensity to self-administer harder drugs, or the willingness to work (bar pressing) to earn doses of heroin or cocaine. Existing panel survey datasets might permit more careful statistical control for background factors common to both marijuana use and harder drug use. New longitudinal studies might focus in on marijuana experience and its effects on the transition to hard drug use, examining the relative influence of changes in peer associations, drug dealing sources, and perceived health risks and legal sanctioning risks. In the absence of better causal evidence, a strong allegiance to any particular gateway theory would seem to reflect ideology or politics rather than science.

Cohen, H. (1972). Multiple drug use considered in the light of the stepping-stone hypothesis. The International Journal of the Addictions, 7, 27-55.

Cohen, P. D. A., & Sas, A. (1996). Cannabis Use, A Stepping Stone to Other Drugs? The Case of Amsterdam. (Department of Human Geography, University of Amsterdam, 1996). See http://www.frw.uva.nlcedro/ library/ASC95/ ASC95.html.43.

Coffin, P. O. (1997, August 9). I smell a rat. Slate.

Ferguson, D., & Horwood, K. (1997). Early onset cannabis use and psychosocial adjustment in young adults. Addiction, 92, 279-296.

Kandel, D. B., Yamaguchi, K., & Chen, K. (1992). Stages of Progression in Drug Involvement from Adolescence to Adulthood: Further Evidence for the Gateway Theory. J. Studies on Alcohol, 53(5), pp. 447-457.

Kaplan, J. (1970). Marijuana--The New Prohibition (New York, World Publishing Co.).

Kleiman, M. (1992). Against Excess: Drug Policy for Results (New York, Basic Books).

MacCoun, R., & Reuter, P. (1997). Interpreting Dutch cannabis policy: Reasoning by analogy in the legalization debate. Science, 278, 47-52.

MacCoun, R., Reuter, P., & Schelling, T. (1996). Assessing alternative drug control regimes. Journal of Policy Analysis and Management, 15, 1-23.

Parker, J., & Grasmick, H. (1979). The effect of actual crimes and arrests on people's perceptions of the certainty of arrest: An exploratory study of an untested proposition in deterrence theory. Criminology, 17, 366-379.

Shedler, J., & Block, J. (1990). Adolescent drug use and psychological health. American Psychologist, 45, 612-630.

Drug Policy Analysis Bulletin

c/o Federation of American Scientists

307 Massachusetts Avenue, NE

Washington, DC 20002