DOES MARIJUANA USE CAUSE LONG-TERM COGNITIVE DEFICITS?

To the Editor: Dr Solowij and colleagues [1] concluded that their findings
"confirm that long-term heavy cannabis users show impairments in memory and
attention that endure beyond the period of intoxication." In his
accompanying Editorial, Dr Pope [2] pointed out that this study could not
establish a causal relationship between use of marijuana and later declines
in cognitive performance. Neither Solowij et al nor Pope, however, referred
to laboratory studies designed to assess causality, such as ours, which
evaluated the effects of acute marijuana administration on complex
cognitive performance in regular marijuana smokers. [3] Unlike the subjects
of Solowij et al, these individuals were not seeking treatment and had
heavier marijuana use, averaging 24 marijuana cigarettes per week.
Participants smoked a single marijuana cigarette during 3 separate
outpatient sessions containing varying amounts of tetrahydrocannabinol,
which had minimal effects on cognitive functioning. Chait [4] reported
similar findings.

Both of these laboratory studies found minimal cognitive deficits after
marijuana administration in experienced users and suggest that recent
marijuana use is a minimal confounder in experienced marijuana users. Data
from well-controlled laboratory studies in combination with data from
retrospective studies can ultimately provide a more comprehensive view of
marijuana-related effects on human cognitive performance.

Erik W. Gunderson, MD; Suzanne K. Vosburg, PhD; Carl L. Hart, PhD;
Department of Psychiatry, Division on Substance Abuse, New York State
Psychiatric Institute and College of Physicians and Surgeons of Columbia
University, New York

1. Solowij N, Stephens RS, Roffman RA, et al, for the Marijuana Treatment
Project Research Group. Cognitive functioning of long-term heavy cannabis
users seeking treatment. JAMA. 2002;287:1123-1131. (
http://www.mapinc.org/drugnews/v02/n395/a10.html )

2. Pope HG Jr. Cannabis, cognition, and residual confounding. JAMA.
2002;287:1172-1174 ( http://www.mapinc.org/drugnews/v02/n396/a01.html )

3. Hart CL, van Gorp W, Haney M, et al. Effects of acute smoked marijuana
on complex cognitive performance. Neuropsychopharmacology. 2001;25:757-765.

4. Chait LD. Subjective and behavioral effects of marijuana the morning
after smoking. Psychopharmacology (Berl). 1990;100:328-333.

To the Editor: While the study of Dr Solowij and colleagues [1]
demonstrates a significant difference in cognitive function between
long-term users and short-term/nonusers of marijuana, I am concerned about
possible selection biases. Because all the marijuana users in this trial
were actively seeking assistance with reduction or cessation, the sample
may be biased toward those individuals who feel they have a substance use
problem so severe it requires treatment.

Michael Watson, MC, USNR, Department of Family Practice, Naval Hospital,
Jacksonville, Fla

1. Solowij N, Stephens RS, Roffman RA, et al, for the Marijuana Treatment
Project Research Group. Cognitive functioning of long-term heavy cannabis
users seeking treatment. JAMA. 2002;287:1123-1131. (
http://www.mapinc.org/drugnews/v02/n395/a10.html )

To the Editor: I would like to point out 2 defects in the study of Dr
Solowij et al. First, it does not control for age-related differences in
cognitive function, which could potentially cause the differences between
the long- and shorter-term user groups. Second, cannabinoids are present
for many days after ingestion. Age-related differences in excretion may
also explain the apparent difference in function between long- and
shorter-term users of marijuana.

Julia R. Nyquist, MD; San Anselmo, Calif

In Reply: Dr Gunderson and colleagues refer to controlled laboratory
studies that found minimal cognitive deficits in experienced users after
acute marijuana administration. These findings support our interpretation
that the observed long-term effects were unlikely to be confounded by
residual effects of recent cannabis use. Our study was designed
specifically to investigate chronic effects, with years of use as our
variable of interest. By requiring several hours abstinence prior to
testing, we induced an unintoxicated cognitive state that long-term users
typically operate in for substantial periods in their daily life. We showed
that impairments were generally unrelated to withdrawal and recent use. We
concluded that a probable causal relationship exists because we controlled
for potential confounding factors.

It should be noted that the experienced cannabis users in these laboratory
studies did not approach the long-term durations reported by participants
in our study, and that the potency of cannabis smoked in the community is
generally greater than that administered in the laboratory. Nevertheless,
we agree that controlled laboratory studies provide a valuable complement
to naturalistic studies like ours. Just as acute effects of cannabis differ
in experienced vs naive subjects, long-term effects vary with the frequency
and duration of cannabis use. This and the mechanisms involved in the
development of tolerance to the acute effects of cannabinoids on cognition
are complex issues that require further research.

Dr Watson expresses concern about selection bias. Although the participants
in this study were seeking treatment, their impairments were related
specifically to the number of years that cannabis had been used,
replicating our previous findings in cannabis users not seeking treatment.
[1] Thus, regardless of treatment seeking, there is good evidence for a
neurobiological explanation underlying cognitive impairments that develop
over many years of exposure to cannabis.

Dr Nyquist claims that there was a lack of control for age differences
between groups. We included age as a covariate in analyses where it
correlated with test performance and we performed semipartial correlations
to examine the unique contributions of age and duration of cannabis use to
the variance in cognitive test performance (reported in Table 4). Because
age and duration of cannabis use are so inextricably linked, isolation of
effects associated with years of cannabis use relies on statistical control
methods; our results showed a greater unique contribution from the years of
cannabis use. We ensured that the control group did not differ in age from
the overall cannabis user sample prior to their division into long- and
shorter-term user groups. Our previous studies1 have shown cognitive
impairments in long-term cannabis users compared with age-matched controls.
We are unaware of any literature showing age-related differences in
excretion of cannabinoid metabolites.

Nadia Solowij, PhD; National Drug and Alcohol Research Centre, University
of New South Wales, Sydney, Australia; Department of Psychology, University
of Wollongong, Wollongong, Australia

Thomas Babor, PhD, MPH; Department of Community Medicine, University of
Connecticut Health Center, Farmington

Robert Stephens, PhD; Department of Psychology, Virginia Polytechnic
Institute and State University, Blacksburg

Roger A. Roffman, DSW; Innovative Programs Research Group, School of Social
Work, University of Washington, Seattle, for the Marijuana Treatment
Project Research Group

1. Solowij N. Cannabis and Cognitive Functioning. Cambridge, United
Kingdom: Cambridge University Press; 1998.

To the Editor: In his Editorial accompanying our paper on cognitive
functioning in long-term heavy cannabis users, [1] Dr Pope [2] makes
inferences that question the validity of our findings. We point out that
the possible confounding factors that Pope alludes to were in fact
controlled in our study. We also wish to clarify other issues that he
brings up.

First, we did not claim that the cognitive impairments associated with
long-term heavy cannabis use in our study were irreversible; we only showed
no performance differences between those abstaining for less than or more
than 17 hours (range, up to 240 hours). A reversible deficit need not
necessarily be due to a residue of cannabinoids or to withdrawal, which our
data did not support. These impairments could be interpreted in terms of a
gradual adaptation of the nervous system to prolonged exposure to exogenous
cannabinoids, possibly resulting in altered functioning of the endogenous
cannabinoid or other neuromodulator systems. After prolonged abstinence,
these systems may well return to healthy function. Future analyses from
this study will investigate recovery of function in the same sample 4
months after cessation or reduction of cannabis use.

Second, Pope suggests that our results may have been influenced by residual
confounding. Our screening of participants was very thorough: there was no
greater incidence among the cannabis users of head injury, concussion,
hospitalization, treatment seeking for psychological or emotional problems,
or use of prescription medications. Data (not reported in the article)
showed no association between performance on the cognitive tests and
psychological distress as measured by the Beck Depression Inventory,
State-Trait Anxiety Inventory, and Brief Symptom Inventory, on which
shorter-term users generally had the highest scores yet did not differ from
controls in cognitive performance.

Similarly, no site differences were found in either sociodemographics or
cognitive test performance. The sex ratio did not differ between any of our
groups but since Pope and Yurgelun-Todd had previously reported sex
differences in cognitive effects of cannabis, [3] we also investigated
these and found none. Contrary to Pope's assertion that the results may be
explained by differences in prior abuse of other substances, we showed that
significant memory impairment was evident in the long-term users after
excluding participants with previous histories of other substance use. We
also reported analyses that countered the hypothesis that these effects
might be due to age or to recent use of cannabis.

The results replicate findings from our earlier studies that used different
cognitive tests and measures of brain electrical activity [4] to show that
cognitive impairments worsen with the number of years of cannabis use. Few
studies have investigated the effects of duration of cannabis use. Of
course, there may be unknown influences affecting associations of this kind
but the evidence from our study supports the most parsimonious conclusion
that it is the years of cannabis use that produces the impairment.

Nadia Solowij, PhD; National Drug and Alcohol Research Centre, University
of New South Wales, Sydney, Australia; Department of Psychology, University
of Wollongong, Wollongong, Australia

Robert Stephens, PhD; Department of Psychology, Virginia Polytechnic
Institute and State University, Blacksburg

Roger A. Roffman, DSW; Innovative Programs Research Group, School of Social
Work, University of Washington, Seattle

Thomas Babor, PhD, MPH; Department of Community Medicine, University of
Connecticut Health Center, Farmington for the Marijuana Treatment Project
Research Group

1. Solowij N, Stephens RS, Roffman RA, et al, for the Marijuana Treatment
Project Research Group. Cognitive functioning of long-term heavy cannabis
users seeking treatment. JAMA. 2002;287:1123-1131.

2. Pope HG Jr. Cannabis, cognition, and residual confounding. JAMA. (
http://www.mapinc.org/drugnews/v02/n396/a01.html )

3. Pope HG Jr, Yurgelun-Todd D. The residual cognitive effects of heavy
marijuana use in college students. JAMA. 1996;275:521-527.

4. Solowij N. Cannabis and Cognitive Functioning. Cambridge, United
Kingdom: Cambridge University Press; 1998.

In Reply: Dr Solowij and colleagues provide reassurance regarding their
extensive efforts to control for possible confounds. Our similar study [1]
of equally long-term cannabis users agrees with theirs in finding cognitive
impairment hours to days after discontinuing cannabis. However, our studies
still disagree on one important point: Solowij et al found increasing
cognitive impairment with increasing duration of cannabis use, whereas we
did not.

I still believe that the most parsimonious explanation for this discrepancy
is residual confounding, either from inadequate adjustment for measured
confounders or (perhaps more likely) from the presence of unmeasured
confounders. This is because both studies depend heavily on the assumption
that, after appropriate statistical adjustments, longer- and
shorter-duration cannabis users are comparable on all factors, other than
the amount of exposure, that would influence the outcome. [2, 3] Such
comparability may be almost impossible to achieve in a retrospective study,
particularly since preexposure cognitive function and latent vulnerability
to neuropsychiatric disorders (either unexpressed or only partially
expressed) may predispose to duration of cannabis use and may influence
outcome.

Even in seemingly well-matched groups, minor confounders can substantially
alter estimated effects. One cannot exclude the possibility that among
cannabis users spontaneously seeking psychiatric treatment for their drug
use, subtle neuropsychiatric factors, not induced by cannabis, may affect
cognitive performance despite the best efforts to control for such factors.
The most that can be concluded is that the effect sizes observed in our 2
studies are simultaneously consistent either with no duration-associated
deficits at all (all observed differences being due to residual
confounding) or with a substantial association of possible clinical
importance. Therefore, I stand by my conclusion that we must live with
uncertainly.

Harrison G. Pope, Jr, MD; Biological Psychiatry Laboratory, McLean
Hospital, Harvard Medical School, Belmont, Mass

1. Pope HG Jr, Gruber AJ, Hudson JI, Huestis MA, Yurgelun-Todd D.
Neuropsychological performance in long-term cannabis users. Arch Gen
Psychiatry. 2001;58:909-915.

2. Greenland S, Robins JM. Identifiability, exchangeability, and
epidemiologic confounding. Int J Epidemiol. 1986;15:412-418.

3. Little RJ, Rubin DB. Causal effects in clinical and epidemiologic
studies via potential outcomes: concepts and analytical approaches. Annu
Rev Public Health. 2000;21:121-145.