ORIGINAL CONTRIBUTION

Cognitive Functioning of Long-term

Heavy Cannabis Users Seeking Treatment
Nadia Solowij, PhD Context Cognitive impairments are associated with long-term cannabis use, but the
Robert S. Stephens, PhD parameters of use that contribute to impairments and the nature and endurance of
cognitive dysfunction remain uncertain.
Roger A. Roffman, DSW
Objective To examine the effects of duration of cannabis use on specific areas of
Thomas Babor, PhD, MPH cognitive functioning among users seeking treatment for cannabis dependence.
Ronald Kadden, PhD Design, Setting, and Participants Multisite retrospective cross-sectional neuro-
Michael Miller, PhD psychological study conducted in the United States (Seattle, Wash; Farmington, Conn;
and Miami, Fla) between 1997 and 2000 among 102 near-daily cannabis users (51
Kenneth Christiansen, PsyD long-term users: mean, 23.9 years of use; 51 shorter-term users: mean, 10.2 years of
Bonnie McRee, MPH use) compared with 33 nonuser controls.
Janice Vendetti, MPH Main Outcome Measures Measures from 9 standard neuropsychological tests that
assessed attention, memory, and executive functioning, and were administered prior
for the Marijuana Treatment Project to entry to a treatment program and following a median 17-hour abstinence.
Research Group
Results Long-term cannabis users performed significantly less well than shorter-

I
N THE CURRENT CLIMATE OF DEBATE term users and controls on tests of memory and attention. On the Rey Auditory Ver-
about marijuana laws and interest in bal Learning Test, long-term users recalled significantly fewer words than either shorter-
marijuana as medicine,1 one issue re- term users (P=.001) or controls (P=.005); there was no difference between shorter-
term users and controls. Long-term users showed impaired learning (P=.007), retention
mains unresolved: Does heavy, fre- (P = .003), and retrieval (P = .002) compared with controls. Both user groups per-
quent, or prolonged use of cannabis lead formed poorly on a time estimation task (P⬍.001 vs controls). Performance measures
to a deterioration in cognitive function often correlated significantly with the duration of cannabis use, being worse with in-
that persists well beyond any period of creasing years of use, but were unrelated to withdrawal symptoms and persisted after
acute intoxication? Is the functioning of controlling for recent cannabis use and other drug use.
the brain altered in the long term? With Conclusions These results confirm that long-term heavy cannabis users show im-
over 7 million people using cannabis pairments in memory and attention that endure beyond the period of intoxication and
weekly or more often in the United States worsen with increasing years of regular cannabis use.
alone2 and the potential for increased JAMA. 2002;287:1123-1131 [Link]
physician recommendations for select pa-
tients to use cannabis therapeutically,1 in cannabis users in the unintoxicated ever, the parameters of use that are
answers to these questions are of signifi- state (and in children exposed to can- associated with short- or long-lasting cog-
cant public health concern.3,4 Scientific nabis in utero7) in controlled studies us- nitive and brain dysfunction have not
evidence from past research clearly ing brain event-related potential tech-
Author Affiliations: National Drug and Alcohol Re-
showed that gross impairment related to niques6,8-10 and neuropsychological search Centre, University of New South Wales, Syd-
chronic cannabis use did not occur but assessments11-15 including complex tasks. ney, and Department of Psychology, University of Wol-
longong, Wollongong (Dr Solowij), New South Wales,
was inconclusive with regard to the pres- Brain imaging studies of cannabis us- Australia; Department of Psychology, Virginia Poly-
ence of more specific deficits.5,6 Recent ers have demonstrated altered func- technic Institute and State University, Blacksburg, Va
(Dr Stephens); Innovative Programs Research Group,
studies with improved methods have tion, blood flow, and metabolism in pre- School of Social Work, University of Washington, Se-
demonstrated changes in cognition and frontal and cerebellar regions.16-19 Studies attle (Dr Roffman); Department of Community Medi-
failing to detect cognitive decline asso- cine (Dr Babor and Mss McRee and Vendetti) and De-
brain function associated with long- partment of Psychiatry (Dr Kadden), University of
term or frequent use of cannabis. Spe- ciated with cannabis use20 may reflect in- Connecticut Health Center, Farmington; and The Vil-
cific impairments of attention, memory, sufficient heavy or chronic use of can- lage South Inc, Miami, Fla (Drs Miller and Christiansen).
Other Members of the Marijuana Treatment Project
and executive function have been found nabis in the sample or the use of Research Group are listed at the end of this article.
insensitive assessment instruments. Im- Corresponding Author and Reprints: Nadia Solowij,
PhD, National Drug and Alcohol Research Centre, Uni-
pairments appear to increase with dura- versity of New South Wales, Sydney, NSW 2052, Aus-
For editorial comment see p 1172.
tion and frequency of cannabis use; how- tralia (e-mail: [Link]@[Link]).

©2002 American Medical Association. All rights reserved. (Reprinted) JAMA, March 6, 2002—Vol 287, No. 9 1123

Downloaded From: [Link] by a Purdue University User on 05/21/2015

 COGNITIVE FUNCTIONING OF LONG-TERM CANNABIS USERS

been fully elucidated. The attribution of use) were controlled through match- regard to group assignment. Self-
deficits to lingering acute effects, drug ing or statistical methods. The sample reported drug and alcohol use were as-
residues, abstinence effects, or lasting size required for this study was deter- sessed by the Addiction Severity In-
changes caused by chronic use contin- mined by estimating a 94% chance of dex,26 a separate structured interview,
ues to be debated.5,6 Animal research sug- detecting a moderate effect size of 0.5 and the Time Line Follow Back proce-
gests an important role for the cannabi- SD units at a 2-tailed ␣ of .05. dure.27,28 The Structured Clinical Inter-
noid receptor in regulating the neural view for Diagnostic and Statistical Manual
activity critical for memory process- Recruitment Procedure of Mental Disorders, 4th Edition (DSM-
ing.21-24 Long-term use of cannabis may and Assessment of Drug Use IV) Axis I Disorders (SCID)29 assessed
result in altered functioning of the can- Sixty-five of the 102 cannabis users were cannabis dependence. Duration of regu-
nabinoid receptor and its associated neu- delayed-treatment participants from the lar (at least twice per month) cannabis
romodulator systems. Marijuana Treatment Project, a mul- use was an averaged composite mea-
This study investigated the nature of tisite US study (Seattle, Wash; Farming- sure derived from the Addiction Sever-
cognitive impairments associated with ton, Conn; and Miami, Fla) conducted ity Index, SCID, and the structured in-
long-term cannabis use employing data between 1997 and 2000 of the effective- terview. Current frequency of cannabis
collected from a large clinical trial of ness of brief treatments for cannabis de- use was calculated from the Time Line
chronic users seeking treatment for can- pendence.25 The remainder were re- Follow Back procedure.
nabis dependence. The study com- c r u it e d through the Marijuana
pared 102 cannabis users assessed prior Treatment Project specifically for this Inclusion/Exclusion Criteria
to treatment on carefully selected neu- study. Participants provided written in- Cannabis users were included if they
ropsychological tests with 33 nonuser formed consent as approved by the eth- had used cannabis regularly for at least
controls. The parameters of cannabis ics committees of the participating in- 3 years, were currently using at least
use that contribute to impairment were stitutions and were paid $75 for once a week, were seeking treatment to
examined. It was hypothesized that per- completing the cognitive assessments. assist them to cease or reduce their use
formance would deteriorate as the num- Controls (n=33) were recruited from the of cannabis, and were willing to par-
ber of years of regular use increased. general population through media ad- ticipate in the treatment program of-
vertisements at only 1 site. The con- fered. Participants were excluded if they
METHODS trols were told that the researchers were had ever had a serious illness or injury
Design studying the effects of exposure to drugs that may have affected the brain, any
A multisite, retrospective, cross- and alcohol on cognitive functioning, psychotic disorder, met a current
sectional comparison-group design was and that at present only individuals at DSM-IV diagnosis of dependence on any
used to compare (1) long-term users the lighter end of the spectrum of drug other drug or alcohol, or had a poor
with a mean of 23.9 years of regular can- experience were required. The aim was command of the English language.
nabis use; (2) shorter-term users with to minimize cannabis use among con-
a mean of 10.2 years of regular use; and trols while approximating the other Sample Characteristics
(3) nonusers of cannabis. Key con- characteristics of the cannabis-using TABLE 1 provides demographic infor-
founding variables (age, IQ, other drug sample. Assessors were not blinded with mation and cannabis use parameters.

Table 1. Demographic and Cannabis Use Details of the Sample*

Cannabis Users

All Shorter-term Users Long-term Users Controls

No. 102 51 51 33
Sex, male (%) 75 (74) 36 (71) 39 (76) 22 (67)
Age, mean (SD) [range], y 35.4 (8.6) [19-55] 28.7 (5.5) [19-45]† 42.1 (5.2) [34-55]† 34.8 (11.1) [19-65]
Education, mean (SD) [range], y‡ 14.3 (2.3) [10-22] 14.1 (2.5) [10-22] 14.5 (2.0) [11-20] 14.8 (1.8) [12-18]
Full-scale IQ, mean (SD) [range]§ 105.4 (6.7) [87.4-118.5]㛳 105.1 (7.4) [87.4-118.5] 105.7 (5.9) [92.7-118.3] 107.9 (4.7) [94.5-117.2]
Duration of use, mean (SD) [range], y¶ 17.1 (7.9) [2.7-31.7] 10.2 (3.8) [2.7-17.0] 23.9 (4.1) [17.3-31.7] ...
Frequency of use, median (range), d/mo# 27.9 (3.5-30) 28.3 (5.2-30) 27.4 (3.5-30) ...
*Ellipses indicate not applicable.
†Significantly different from controls at P⬍.001.
‡The self-reported number of years of formal education completed.
§Estimated from a combination of North American Adult Reading Test, Wide Range Achievement Test−Revised reading subtest scores, and the Barona Index.
㛳Significantly different from controls at P⬍.05.
¶The number of years that cannabis had been used since regular use commenced (at least twice per month), composite from self-report, Addiction Severity Index, and Structured
Clinical Interview Axis 1 Disorders assessments.
#The median number of days per month that cannabis was used at entry to the study, based on average use per 30 days from self-reported use during the past 14 weeks.

1124 JAMA, March 6, 2002—Vol 287, No. 9 (Reprinted) ©2002 American Medical Association. All rights reserved.

Downloaded From: [Link] by a Purdue University User on 05/21/2015

 COGNITIVE FUNCTIONING OF LONG-TERM CANNABIS USERS

The user group was split at the median Participants were required to ab- cocaine (n=24), amphetamines (n=11),
for duration of cannabis use to enable stain from cannabis for at least 12 hours hallucinogens (n=17), and sedatives/
comparisons of long-term users, shorter- prior to testing and to provide 2 urine hypnotics or minor tranquilizers (n=7).
term users, and controls. No meaning- samples (1 the night before testing, an- Current use of other drugs was less than
ful division of groups could be achieved other during the test session). The me- once a month or not at all for 93.1% of
on the basis of frequency of cannabis use, dian self-reported time since last use of the sample. More than half of the con-
which was almost daily for the major- cannabis was 17 hours (range, 7-240 trols (51.5%) had never tried any other
ity of the sample. Sex distribution and hours); this did not differ between long- drug and the remainder had only tried
years of education did not differ be- and shorter-term users. At the time of other drugs experimentally. “Pure
tween groups. The majority of users testing, 70% of the sample reported that sample” analyses excluded all partici-
(68.6%) and controls (63.6%) were they were not experiencing any dis- pants with histories of regular or heavy
white. Overall, users and controls did not comfort after abstaining from canna- use of alcohol or other drugs.
differ in age, but long-term users were bis. Twice as many shorter-term users
significantly older and shorter-term us- than long-term users (P = .03) re- Neuropsychological Tests
ers were significantly younger than con- ported mild withdrawal symptoms such and Procedures
trols (P⬍.001). Premorbid intelligence as cravings, irritability, depression, Nine neuropsychological tests were
was estimated by several methods and anxiety, sleep, or appetite distur- administered in the order listed in
averaged: the Wide Range Achieve- bances. In 78.3% of cases, creatinine- TABLE 2,38-46 along with the 2 tests used
ment Test—Revised reading subtest normalized urinary cannabinoid me- to assess premorbid IQ. 30-32 A 10-
(WRAT-R READ)30,31; the North Ameri- tabolite (THC-COOH) levels on the day minute rest break was given after the Rey
can Adult Reading Test (NAART)32; and of testing were less than or equivalent Auditory Verbal Learning Test (RAVLT)
the Barona Index.33 The mean esti- to those from the night before.34-37 Ab- Recognition test. Tests were adminis-
mated full-scale IQ (FSIQ) did not dif- stinence from cannabis was supported tered by trained assistants and took ap-
fer between the 3 groups based on du- by significant correlations between the proximately 2 hours to complete. Qual-
ration of cannabis use. The majority of level of normalized urinary cannabi- ity assurance procedures were adopted
the sample (82.4% long-term, 88.2% noid metabolite on the day of testing to ensure that procedures were stan-
shorter-term users) reported experienc- and the self-reported time since last use dardized at each site with ongoing su-
ing problems with memory, attention, (bivariate correlation coefficient pervision and review of audiotaped
or concentration, which they attrib- [r], − 0.46; P⬍.001), and the quantity assessments by centralized staff through-
uted to their use of cannabis. used on the last occasion divided by the out the course of the study.
time since last use (r, 0.39; P⬍.001).
Cannabis Use, Required The effects of these measures of re- Data Analysis
Abstinence, and Urinalysis cent use were examined in relation to Each cognitive test was analysed us-
Users first tried cannabis at a mean age test performance. “Pure sample” analy- ing SPSS version 10.0 (SPSS Institute,
of 15.3 (SD, 2.6) years with regular use ses excluded users with higher metabo- Chicago, Ill) with analysis of covari-
(at least twice a month) commencing at lites in the second urine sample. No ance (ANCOVA) for normally distrib-
age 17.5 (SD, 3.2) years. Cannabis had cannabinoid metabolites were de- uted variables or nonparametric tests
been used on a median 29 of the past 30 tected in the urine of the control par- of group differences for skewed data.
days (range, 1-30). Almost the entire ticipants. The FSIQ and age were included as co-
sample (98%) met the DSM-IV criteria variates in analyses where they corre-
for cannabis dependence. The median Other Drug Use lated with test performance. All par-
amount of cannabis smoked per week No other drug metabolites were de- ticipants were initially included in
was 1 quarter of an ounce (range, 0.01- tected in any urine sample. Tobacco and analysis, with the overall cannabis user
2.00 oz) with 2 average-sized joints typi- alcohol use was minimal. Alcohol was sample first compared with the con-
cally smoked per day (range, 0.12-20.00). consumed on a median of 3.4 and 1.7 trol group (evaluated at P⬍.05), fol-
None of these cannabis-use parameters days per month among users and con- lowed by comparisons on the basis of
differed between the long- and shorter- trols, respectively. Almost one third of duration of cannabis use (long- vs
term user groups. Twenty-two controls users and 46.8% of controls drank less shorter-term users vs controls, evalu-
had either never tried cannabis or used than once a month or not at all. Forty- ated at P⬍.01). For 2-way interac-
it 10 or fewer times in their lives and 11 eight percent of the cannabis users had tions, the Greenhouse-Geisser method
had used cannabis weekly to monthly only tried drugs other than cannabis a was used to adjust the df where appro-
while at school or college between 4 and few times or never; 52% had used other priate and for multiple comparisons, a
30 years ago. Controls with a history of drugs socially/recreationally primar- Bonferroni adjustment controlled for
cannabis use were excluded from “pure ily during high school and college. Past type I error. Analysis of covariance was
sample” analyses. histories of regular drug use included repeated on a purer sample that strictly
©2002 American Medical Association. All rights reserved. (Reprinted) JAMA, March 6, 2002—Vol 287, No. 9 1125

Downloaded From: [Link] by a Purdue University User on 05/21/2015

 COGNITIVE FUNCTIONING OF LONG-TERM CANNABIS USERS

ference list presented only once; P=.08).

Table 2. Neuropsychological Tests Administered and Cognitive Functions Assessed*
The proportion of subjects with a
Neuropsychological Test Cognitive Functions Assessed
very poor learning ability (acquisition
Wide Range Achievement Test−Revised reading Premorbid IQ
subtest (WRAT-R READ)30,31 ⬍3 words over 5 trials) was greater
Speed of Comprehension (SOC) test (Speed and Rate of verbal information processing among long-term users (13.7%) than
Capacity of Language Processing [SCOLP])38 controls (0%) (P=.007) but not shorter-
Rey Auditory Verbal Learning Test (RAVLT)39-41 Memory span, verbal learning and retrieval term users (5.9%). The proportion of
efficiency, susceptibility to interference
long-term users recalling fewer than 10
Stroop Test42 with additional interference Attention, cognitive flexibility, inhibition
condition43 of distractor stimuli, suppression of words on trial V (27.5%) was more than
habitual response among shorter-term users (8.5%) or
Wisconsin Card Sorting Test (WCST), Problem solving, conceptual ability controls (3.0%) (P = .002). Signifi-
computerized44
cantly more long-term users (23.5%)
Alphabet Task† Cognitive flexibility, executive function
lost 3 or more words over the 20-
Omitted Numbers‡ Working memory
minute delay between trials VI and VII
Time Estimation and Production§ Temporal judgment
than shorter-term users (4.3%) or con-
RAVLT 20-minute delay trial (VII), Recognition test Long-term retention, recognition memory
trols (3.0%) (P =.003). Long-term us-
North American Adult Reading Test (NAART)32 Premorbid IQ
ers showed a smaller primacy effect in
Auditory Consonant Trigrams (Brown-Peterson)45 Short-term retention under distractor
conditions the serial position curve than either
Paced Auditory Serial Addition Test (PASAT)46 Information processing, working memory, other group (P=.02). Groups did not
divided and sustained attention differ in the recency effect or in words
*Tests were administered in the order listed. recalled from the middle of the list.
†Composed of timed loud, silent, and alternating recital trials.
‡Recognition of omitted item from a jumbled aural list of numbers from 1 to 10 (10 trials). Users overall and long-term users rec-
§Composed of 3 trials: unwarned estimation of time to complete the preceding task (mean, 3 minutes 18 seconds)
(Time Estimation A); time production (1 minute 40 seconds); and warned passive estimation (2 minutes) (Time Esti- ognized fewer words than controls from
mation B). list A (overall, P=.03; long-term, P=.01)
and list B (overall, P = .01; long-term,
excluded those participants with ei- curves of shorter-term users and con- P=.04) but long-term users did not dif-
ther a history of other drug use or pos- trols were similar but long-term users fer from shorter-term users. More than
sible recent use of cannabis prior to test- showed a learning curve with a less half of the long-term users (55%) had a
ing. Semipartial correlations examined steep gradient and long-term users re- recognition score for list A of 12 or less
the unique contributions of FSIQ, age, called fewer words on every trial. The compared with 28% of shorter-term us-
duration of cannabis use, and recency sum of words recalled across all trials ers and 21% of controls (P=.002). Long-
of cannabis use to the variance in cog- I through VII inclusive of trial B (re- term users misassigned more words (me-
nitive test performance. ferred to here as RAVLT sum; range, 37- dian, 2) than shorter-term users and
114) correlated significantly and in- controls (each median, 0) (P⬍.001). A
RESULTS versely with the duration of cannabis greater proportion of long-term users
Results from the 9 neuropsychologi- use after controlling for age and FSIQ (13.7%) compared with shorter-term us-
cal tests are shown in TABLE 3 for can- (partial r, − 0.23; P = .01). When ana- ers (6.4%) and controls (0%) actually
nabis users overall, for groups based on lysed by ANCOVA, there was a signifi- identified fewer words on recognition
duration of cannabis use, and for con- cant effect of group (F 2,127 = 8.36; than they had just prior during recall on
trols. Effect sizes are calculated be- P⬍.001) whereby long-term users re- trial VII (P=.02). Long-term users’ per-
tween long-term users and controls us- called significantly fewer words than ei- formance was significantly poorer than
ing the SD of the controls. ther shorter-term users (95% confi- published norms47 for the general popu-
dence interval [CI] for difference, 3.84- lation on most measures from the
Speed of Comprehension 19.18; P=.001) or controls (95% CI for RAVLT.
Cannabis user groups did not differ difference, 2.83-19.93; P=.005) with no
from controls in the number of items difference between shorter-term users Stroop Test
completed (range, 23-100) but users and controls. When all trials were in- Cannabis users did not differ signifi-
overall made more errors (P = .03) cluded in a repeated measures AN- cantly from controls after inclusion of
(range, 0-5). These results suggest that COVA, a significant interaction be- covariates in any condition or on in-
cannabis users are more likely to sac- tween group and trial (F14,889 = 2.84; terference scores. While there were no
rifice accuracy for speed. P = .007) suggested that long-term us- performance differences between Color-
ers recalled fewer words than shorter- Word (CW) and Color-Read (CR) in
Rey Auditory Verbal Learning Test term users or controls on every trial the control group, performance on CR
Mean words recalled on each trial are (P⬍.05 for each comparison) except the was, however, poorer than on CW in
depicted in the FIGURE. The learning first, with a trend on trial B (the inter- both long- (P⬍.001) and shorter-term
1126 JAMA, March 6, 2002—Vol 287, No. 9 (Reprinted) ©2002 American Medical Association. All rights reserved.

Downloaded From: [Link] by a Purdue University User on 05/21/2015

 COGNITIVE FUNCTIONING OF LONG-TERM CANNABIS USERS

users (P=.03). Color-Read was the ad- results suggest that cannabis users are Test (WCST) measure but a trend on one:
ditional interference condition de- vulnerable to task complexity with long-term users failed to maintain the set
signed to increase demands on execu- increasing demands creating more more often than shorter-term users
tive function.43 There was an inverse sources of interference that adversely (P=.05) or controls (P=.07). Research
relationship between duration of can- affect performance. suggests that this measure best repre-
nabis use and number of items com- sents attentional dysfunction.39 There was
pleted on CR (partial r,−0.27; P=.003) Wisconsin Card Sorting Test no evidence of impaired performance
and CW (partial r,−0.27; P= .004) af- There were no significant group differ- with increasing years of cannabis use after
ter controlling for age and FSIQ. These ences on any Wisconsin Card Sorting controlling for covariates.

Table 3. Neuropsychological Test Results

Cannabis Users* P Value for Comparisons

Shorter-term Long-term Shorter-term Long-term Shorter- vs

All Users Users Controls Effect All vs Users vs Users vs Long-term
Test (n = 102) (n = 51) (n = 51) (n = 33) Size Controls Controls Controls Users
SCOLP-SOC,
median (range)†
Correct 10 (3-18) 11 (6-18) 10 (3-17) 10 (6-15) ... .06 .07 .10 .65
Errors 1 (0-8) 1 (0-8) 1 (0-6) 0 (0-3) ... .03 .05 .05 .99
RAVLT, mean (SD)‡
Trial I 6.3 (1.9) 6.5 (1.9) 6.1 (1.9) 7.0 (1.9) 0.47 .12 ⬎.99 .15 .59
Trial II 9.3 (2.7) 9.9 (2.6) 8.5 (2.5)* 9.9 (2.3) 0.61 .27 ⬎.99 .05 .004
Trial III 10.8 (2.5) 11.5 (2.3) 10.1 (2.6)* 11.4 (2.2) 0.59 .37 ⬎.99 .07 .003
Trial IV 11.5 (2.3) 12.1 (2.2) 10.9 (2.4)* 12.4 (2.2) 0.68 .10 ⬎.99 .02 .01
Trial V 12.2 (2.3) 12.7 (2.1) 11.5 (2.4)* 12.9 (1.6) 0.88 .19 ⬎.99 .03 .005
Trial B 6.0 (2.3) 6.5 (2.4) 5.5 (2.2) 6.9 (2.5) 0.56 .18 ⬎.99 .05 .07
Trial VI 10.0 (3.0) 10.9 (2.8) 9.2 (3.1)* 11.4 (2.2) 1.00 .07 ⬎.99 .005 .002
Trial VII 9.8 (3.5)* 11.1 (3.1) 8.5 (3.5)* 11.0 (2.7) 0.93 .13 ⬎.99 .004 ⬍.001
RAVLTsum 75.6 (17.2)* 81.4 (15.8) 70.3 (16.8)* 82.9 (14.8) 0.85 .14 ⬎.99 .005 ⬍.001
Recog_A 12.1 (3.1)* 13.1 (2.3) 11.1 (3.4)* 13.3 (1.7) 1.29 .03 ⬎.99 .01 .14
Recog_B 6.1 (3.7)* 7.2 (3.7) 5.0 (3.5)* 8.2 (3.2) 1.00 .01 ⬎.99 .04 .26
Stroop, mean (SD)§
Word 101.3 (15.1) 100.2 (16.4) 102.2 (14.0) 107.0 (15.9) 0.30 .13 .99 .34 ⬎.99
Color 75.6 (12.2) 75.8 (13.4) 75.4 (11.1) 74.5 (13.3) 0.07 .50 ⬎.99 ⬎.99 ⬎.99
Color-Word 45.4 (9.2)* 46.8 (9.4) 44.0 (8.8) 44.4 (10.2) 0.04 .25 .55 ⬎.99 ⬎.99
Color-Read 40.1 (7.9)* 42.2 (9.0) 37.7 (6.2) 41.4 (7.9) 0.47 .92 ⬎.99 ⬎.99 .42
WCST, median (range)㛳
Errors 28 (16-81) 27 (16-77) 29 (17-81) 30 (15-78) ... .67 .29 .77 .08
Perseverative 16 (7-49) 16 (7-45) 15 (8-49) 14 (8-63) ... .95 .55 .49 .17
responses
% Concept 72.7 (12.5-86.7) 73.1 (14.8-85.9) 71.9 (12.5-86.7) 71.9 (16.4-88.3) ... .82 .38 .64 .12
Trials 13 (10-75) 13 (11-64) 12 (10-75) 13 (10-101) ... .75 .90 .67 .74
Failures 1 (0-7) 1 (0-5) 2 (0-7) 1 (0-4) ... .42 .74 .07 .05
Alphabet Task,
median (range)¶
Alternating 18.6 (10.4-52.4)* 16.8 (10.4-37.2) 19.8 (13.0-52.4) 17.5 (11.5-33.2) ... .31 .91 .08 .07
Difference 11.3 (4.8-39.9)* 10.3 (5.0-29.6) 12.3 (4.8-39.9) 8.8 (3.2-24.7) ... .14 .54 .04 .09
Omitted Numbers, 6.7 (1.8) 7.0 (1.8) 6.4 (1.9) 6.3 (2.1) 0.05 .13 .64 .53 ⬎.99
mean (SD)#
Time Estimation, s**
Unwarned Task A, −64.4 (53.5)* −61.3 (54.1)* −67.7 (53.2)* −7.6 (88.6) 0.68 ⬍.001 ⬍.001 .01 ⬎.99
mean (SD)
Warned Task B, −1.0 (−55 to 85) −1.5 (−55 to 85) −0.5 (−40 to 75) 5 (−70 to 102) ... .65 .67 .70 .96
median (range)
Time Production, −15.6 (24.6) −14.9 (25.7) −16.4 (23.6) −19.0 (26.2) 0.10 .24 ⬎.99 .34 .79
mean (SD)
Auditory Consonant
Trigrams, mean (SD),
seconds of delay††
9 11.4 (2.7)* 12.1 (2.3) 10.7 (2.9)* 12.9 (1.9) 1.16 .03 ⬎.99 .002 .007
18 11.3 (2.5) 11.7 (2.0) 10.9 (3.0) 11.5 (2.8) 0.21 .79 .92 ⬎.99 .23
36 10.9 (2.9) 11.2 (2.9) 10.7 (2.9) 11.0 (2.7) 0.11 .84 ⬎.99 ⬎.99 .94
(continued)

©2002 American Medical Association. All rights reserved. (Reprinted) JAMA, March 6, 2002—Vol 287, No. 9 1127

Downloaded From: [Link] by a Purdue University User on 05/21/2015

 COGNITIVE FUNCTIONING OF LONG-TERM CANNABIS USERS

Table 3. Neuropsychological Test Results (cont)

Cannabis Users P Value for Comparisons

Shorter-term Long-term Shorter-term Long-term Shorter- vs

All Users Users Controls Effect All vs Users vs Users vs Long-term
Test (n = 102) (n = 51) (n = 51) (n = 33) Size Controls Controls Controls Users
PASAT, median (range)‡‡
PR Trial 1 2.56 (2.40-4.56) 2.51 (2.40-3.60) 2.58 (2.40-4.56)* 2.56 (2.40-3.55) ... .92 .16 .23 .007
PR Trial 2 2.23 (2.00-3.53) 2.18 (2.00-2.83) 2.28 (2.00-3.53) 2.24 (2.00-3.05) ... .43 .84 .10 .03
PR Trial 3 1.79 (1.60-3.56) 1.78 (1.60-3.56) 1.81 (1.60-2.52) 1.76 (1.60-2.63) ... .20 .49 .11 .35
PR Trial 4 1.36 (1.20-2.31) 1.35 (1.20-1.80) 1.38 (1.20-2.31) 1.33 (1.20-2.00) ... .25 .54 .15 .22
Total PR 8.07 (7.32-10.81) 7.90 (7.36-9.95) 8.21 (7.32-10.81) 7.85 (7.37-10.53) ... .32 .87 .05 .02
Total attempted 142.5 (74-191) 146.0 (96-191) 139.0 (74-188) 145.5 (110-190) ... .75 .99 .56 .45
Total correct 127.0 (26.5) 132.0 (26.8) 121.9 (25.4) 131.6 (28.9) 0.34 .74 ⬎.99 .84 .22
% Correct 64.8 (13.5) 67.3 (13.7) 62.2 (13.0) 67.1 (14.7) 0.33 .74 ⬎.99 .84 .22
Seconds 2.78 (1.88-5.88) 2.64 (1.88-4.67) 2.94 (2.09-5.88) 2.71 (1.89-4.54) ... .52 .97 .23 .14
*Significant correlation with the number of years of cannabis use after controlling for covariates (P⬍.05).
†SCOLP-SOC indicates Speed and Capacity of Language Processing-Speed of Comprehension test. Median scaled scores for the number of SOC test items completed in 2
minutes and errors (n = 96 users, n = 32 controls). Ellipses indicate not applicable.
‡RAVLT indicates Rey Auditory Verbal Learning Test. Mean words recalled on each trial of the RAVLT, total recalled across all trials (RAVLTsum), and median words recognized
from lists A and B (98-102 users varied on each trial).
§Mean items completed in 45 seconds for the Word, Color, and Color-Word conditions of the Stroop (age-corrected, n = 100 users) and the modified Color-Read condition (n = 101
users).
㛳WCST indicates Wisconsin Card Sorting Test. Median errors, perseverative responses, percentage of conceptual level responses, trials to complete the first category, and failures
to maintain the set (n = 101 users, n = 32 controls).
¶Median time to recite the alphabet alternating from loud to silent and median difference between the loud and alternating trials in seconds (n = 94 users, n = 29 controls).
#Mean number of correctly identified items from the Omitted Numbers task.
**Mean difference between actual time elapsed and estimated time in unwarned Time Estimation task A, negative scores indicate underestimation (n = 94 users, n = 31 controls),
median difference in warned Time Estimation task B (n = 98 users, n = 32 controls), and mean difference between time produced and time required in the Time Production task
(n = 102 users, n = 30 controls).
††Mean letters recalled with 9-, 18-, and 36-second delays (n = 31 controls).
‡‡Paced Auditory Serial Addition Test (PASAT). Median processing rates (PR) for PASAT trials 1-4, total PR, total attempted, total correct (mean [SD]), percentage correct (mean
[SD]), and seconds per correct response across the 4 trials (n = 98-100 users, n = 32 controls).

complete the alternating and loud tri- greater the delay interval the worse the
Figure. Mean Number of Words Recalled on
Each Trial of the Rey Auditory Verbal als, an index of interference and lack performance. In cannabis users, this gen-
Learning Test by Long- and Shorter-term of flexibility (partial r, 0.26; P =.006). eral pattern was apparent, though there
Cannabis Users and Controls was greater interference at the shorter-
Time Estimation Tasks delay interval than would be expected.
Controls Shorter-term Users Long-term Users
Cannabis users differed from controls
13 (P⬍.001) in Time Estimation Task A Paced Auditory Serial
where they estimated the time taken to Addition Test
No. of Words Recalled

11
complete the preceding (Omitted Num- Long-term users had slower processing
9 bers) task. Both long- and shorter-term rates than shorter-term users on trial 1
users underestimated the time by about (P=.007), with trends on trial 2 (P=.03)
7
one third of the actual time taken (64.4 and the total processing rate across all
5 seconds) and differed significantly from trials (P=.02). Group differences on all
controls (P = .01 and P⬍.001, respec- other measures failed to reach signifi-
I II III IV V B VI VII tively). Groups did not differ in the simple cance but the performance of the long-
Trials
and brief warned passive Time Estima- term users was poorer in comparison
Error bars represent SDs. tion Task B or Time Production, where with one set of norms49 but not an-
they could use strategies such as count- other.50
Alphabet Task and ing. Time estimation measures did not
Omitted Numbers correlate with duration of cannabis use. Pure Effects Attributable
Groups did not differ in the time taken to Cannabis Use and Effects
to complete any trial of the Alphabet Auditory Consonant Trigrams of Recent vs Chronic Use
Task or in the number of items correct Long-term users recalled significantly Excluding all participants with histo-
in the Omitted Numbers task. The log fewer items than shorter-term users ries of regular other drug or alcohol use,
time to complete the alternating trial of (P=.007), controls (P=.002), and pub- dependence or treatment, and controls
the Alphabet Task increased as a func- lished norms48 on only the 9-second delay with any history of regular cannabis use
tion of duration of cannabis use (par- condition. The number of items recalled within the past 20 years reduced the
tial r, 0.26; P =.006), as did the square did not correlate with duration of can- sample to 27 long-term users, 33 shorter-
root difference between times taken to nabis use. In the general population, the term users, and 26 controls. Despite the
1128 JAMA, March 6, 2002—Vol 287, No. 9 (Reprinted) ©2002 American Medical Association. All rights reserved.

Downloaded From: [Link] by a Purdue University User on 05/21/2015

 COGNITIVE FUNCTIONING OF LONG-TERM CANNABIS USERS

reduction in power to detect differ- covariates in ANCOVA did not change term users with a mean 24 years of regu-
ences between groups, there remained the significance of differences between lar cannabis use performed signifi-
a significant difference with ␣=.05 be- long- and shorter-term users. These re- cantly less well on tests of memory and
tween long-term users and controls on sults support the hypothesis that im- attention than nonuser controls and
RAVLTsum (P=.03), recognition of lists paired performance is not a conse- shorter-term users with a mean of 10
A (P = .004) and B (P = .01), and be- quence of recent use prior to testing or years’ use. The greatest impairment on
tween users overall and controls on the the extent of cannabinoid residues pre- almost every measure was from the
unwarned Time Estimation task (P=.02). sent. RAVLT, indicating a generalized
These results support the hypothesis that To explore further the influences of memory deficit with impaired learn-
impaired memory function and time es- duration of cannabis use and recency ing, retention, and retrieval. Long-
timation are specific to chronic use of of use, semipartial correlations were cal- term users recalled 2.5 fewer words than
cannabis. culated using the following predic- controls on the delayed recall trial
In a separate analysis, exclusion of us- tors: FSIQ, age, duration of cannabis where 49% of the long-term users’
ers whose urinary cannabinoid metabo- use, and hours since last use of canna- scores were more than 1 SD, and 21.6%
lite levels exceeded those from the night bis. As shown in TABLE 4, the unique were more than 2 SDs, below the con-
before testing by 50 ng/mg or more contribution of duration of cannabis use trol mean and normative data.47 A large
(n=18) still resulted in significant dif- to the variance of each test variable was proportion of long-term users’ recog-
ferences between long- and shorter- superior or at least equivalent to that nition scores were more than 1 SD
term users, and long-term users and con- of recency of use in all 6 test variables (51%) or 2 SDs (31.4%) below the con-
trols on RAVLT sum (P = .002 and that had significant contributions from trol mean and norms.47 Effect sizes for
P=.002, respectively), on recognition of at least 1 cannabis use parameter. Re- measures that differed significantly be-
lists A (P = .005 and P = .006) and B cent use contributed only to perfor- tween long-term users and controls
(P=.01 and P⬍.001), on the 9-second mance on the memory tests. The fact ranged from 0.56 to 1.29 across all tests,
delay of the Auditory Consonant Tri- that a minority of the sample, primar- indicating moderate to large effects.
grams test (P=.02 and P=.03), and us- ily shorter-term users, reported expe- These results do not indicate a se-
ers still differed from controls on time es- riencing mild withdrawal symptoms, vere memory problem but could nev-
timation (P=.005). When the sample was yet shorter-term users’ performance was ertheless translate into clinically sig-
split at the median for time since last use not impaired, supports the interpreta- nificant cognitive impairment and could
or level of urinary cannabinoid metabo- tion of the cognitive impairments ob- impact functioning in daily life. There
lite on the day of testing and analyzed by served as a long-term consequence of were significant differences between
ANCOVA, there were no differences on cannabis use and not a manifestation long-term users and controls on 6 of the
any measure between those who had of overtly experienced withdrawal. 9 tests administered and performance
used cannabis within the past 17 hours on 4 tests worsened as a function of in-
and those who had used cannabis 17 or COMMENT creasing years of cannabis use. De-
more hours ago, or those with high vs The results of this study have con- spite this and a range of up to 17 years
low levels of urinary metabolites and no firmed and extended previous find- of cannabis use in the shorter-term user
interactions with duration of cannabis ings of cognitive impairments among group, they differed significantly from
use. Including measures of recent use as chronic heavy cannabis users. Long- controls only on time estimation.

Table 4. Predictor Correlations Between Hypothesized Predictors and Select Test Variables*
Full-Scale IQ Age Duration of Cannabis Use Recency of Cannabis Use†

Zero- P Zero- P Zero- P Zero- P

Test Variables Order Semipartial Value Order Semipartial Value Order Semipartial Value Order Semipartial Value
SCOLP-SOC, 0.55 0.52 ⬍.001 0.12 0.24 .005 0.01 −0.24 .005 0.11 0.05 .53
No. correct (冑—)
Stroop
Color-Word 0.27 0.29 .002 −0.26 0.09 .33 −0.34 −0.24 .01 0.09 0.05 .58
Color-Read 0.36 0.36 ⬍.001 −0.19 0.12 .17 −0.27 −0.24 .008 0.18 0.13 .15
RAVLT
RAVLTsum 0.21 0.22 .02 −0.31 0.06 .54 −0.37 −0.21 .02 0.25 0.21 .02
RAVLT recency 0.11 0.10 .32 −0.19 0.12 .20 −0.27 −0.22 .02 0.24 0.23 .02
Alphabet Task −0.21 −0.21 .03 0.12 −0.17 .09 0.21 0.25 .01 −0.04 −0.02 .81
Alternating, log
*SCOLP-SOC indicates Speed and Capacity of Language Processing-Speed of Comprehension; RAVLT, Rey Auditory Verbal Learning Test. P values are for semipartial correla-
tions. Test variables were significantly predicted by at least 1 cannabis use parameter.
†Defined as hours since last use of cannabis.

©2002 American Medical Association. All rights reserved. (Reprinted) JAMA, March 6, 2002—Vol 287, No. 9 1129

Downloaded From: [Link] by a Purdue University User on 05/21/2015

 COGNITIVE FUNCTIONING OF LONG-TERM CANNABIS USERS

Altered brain metabolism in shorter- support the assumption that the cogni- Analysis and interpretation of data: Solowij, Ste-
phens.
term users may be detected with sen- tive impairments observed in the long- Drafting of the manuscript: Solowij.
sitive techniques, such as functional term users were not preexisting but de- Critical revision of the manuscript for important in-
tellectual content: Solowij, Stephens, Roffman, Ba-
magnetic resonance imaging and posi- veloped as a result of their prolonged use bor, Kadden, Miller, Christiansen, McRee, Vendetti.
tron emission tomography, but the of cannabis. Impairment appeared un- Statistical expertise: Solowij, Stephens.
Obtained funding: Solowij, Stephens, Roffman, Ba-
clinical significance of such changes re- related to withdrawal phenomena. The bor, Kadden, Miller.
mains obscure. The strength of this cognitive functions assessed in this study Administrative, technical, or material support: Solowij,
study is in its assessment of overtly rel- are dependent on the intact function- Roffman, McRee, Vendetti.
Study supervision: Solowij, Stephens, Roffman, Ba-
evant cognitive processes; our results ing of the hippocampus, prefrontal cor- bor, Kadden, Miller, Christiansen, McRee, Vendetti.
suggest that shorter-term cannabis us- tex, and cerebellum,39,51-55 which are Other Members of the Marijuana Treatment Project
Research Group include Kathleen Carroll, PhD, Karen
ers are not impaired to an extent that dense with cannabinoid receptors.56 The Steinberg, PhD, (Coordinating Center, University of
would interfere with cognitive func- effects that exogenous cannabinoids ex- Connecticut Health Center, Department of Commu-
nity Medicine), Mark Litt, PhD (Farmington Clinical
tioning in their daily lives. The fact that ert on the cannabinoid receptor system Research Unit, University of Connecticut Health Cen-
the frequency of use was near daily and the role of endogenous cannabi- ter, Department of Psychiatry), Jean Donaldson, MA,
among long- and shorter-term users noids as suggested by animal re- and James Herrell, PhD (Center for Substance Abuse
Treatment).
suggests that the duration of cannabis search6,21-24 provide a credible neuro- Funding/Support: This study was conducted as part
use is a more salient contributor to the physiological explanation for the of the Marijuana Treatment Project, a Cooperative
Agreement for a multisite study of the effectiveness
development of cognitive impairment development of cognitive impairments of brief treatment for cannabis dependence. It was
than quantity or frequency of use. as the result of hypothesized long-term funded by the Substance Abuse and Mental Health
Services Administration, Center for Substance Abuse
While most cannabis users cease us- changes occurring over many years of Treatment (CSAT), US Department of Health and Hu-
ing in their mid-20s to late 20s, approxi- exposure to the drug. man Services (grants UR4 TI11270, UR4 TI11273, UR4
mately 20% continue to use through In conclusion, our results confirm TI11274, UR4TI11310). The research was con-
ducted in Farmington, Conn, Miami, Fla, and Seattle,
their 30s and beyond.2 This is the first that cognitive impairments develop as Wash, in cooperation with the following institutions:
study to our knowledge of a relatively a result of prolonged cannabis use, they University of Connecticut Health Center, The Village
South Inc, University of Washington, and Evergreen
large sample of long-term entrenched endure beyond the period of acute in- Treatment Services.
cannabis users seeking treatment. Con- toxication, and they worsen with in- Acknowledgment: We are grateful to Aimee Balmer-
Campbell, BA, Kara Brennan Dion, BA, David Du-
cern about perceived cognitive impair- creasing years of use. Impairments de- resky, MA, Dave Ghany, BA, Brian Glidden, BA, Cara
ment was one of many problems asso- velop gradually but may only become Gluskoter, MS, Cher Gunby, BA, Jennifer Haley, BA,
Heather Haynes, RN, Patricia Holkon, MA, Elise Ka-
ciated with cannabis use that led the clinically significant and detectable by bella, PhD, Priscilla Morse, MA, Joe Picciano, MS, Sam
users in this study to seek treatment. standard neuropsychological tests af- Schwartz, MSW, Megan Swan, MA, Debbie Tala-
This concern is unlikely to have biased ter 1 to 2 decades of cannabis use. Nev- mini, AS, and Anna Wolfe, BA, for input and assis-
tance with data collection and trial management, Pe-
the results of this study since a slightly ertheless, altered brain function with ter Caputi, BA, GradDip, for statistical advice, Brin
higher proportion of shorter-term vs subtle impairment has been shown to Grenyer, PhD, for comments on the manuscript, Eva
Congreve, DipLib, for library assistance, and to all par-
long-term users reported experiencing manifest earlier.6,8,9,11,17,18 It is also likely ticipants in this research.
cognitive problems, yet shorter-term us- that impairments would be greater
ers mostly did not differ from controls among comorbid substance-depen-
REFERENCES
on the cognitive tests. Nevertheless, it dent persons. The risk to most medi-
1. Watson SJ, Benson JA, Joy JE. Marijuana and medi-
is possible that long-term cannabis us- cal cannabis users is likely to be small, cine: assessing the science base: a summary of the 1999
ers in the community who are not seek- as long as they are not maintained at Institute of Medicine report. Arch Gen Psychiatry. 2000;
57:547-552.
ing treatment may not experience im- high doses for many years. For ha- 2. Substance Abuse and Mental Health Services Ad-
pairments to the same degree as those bitual users, the kinds of impairments ministration (SAMHSA). National Household Survey
assessed in this study. observed in this study have the poten- on Drug Abuse Series: H-6: Preliminary Results from
the 1997 National Household Survey on Drug Abuse.
While acknowledging the limita- tial to impact academic achievements, Rockville, Md: Office on Applied Studies; 1998. DHHS
tions of retrospective designs, if care- occupational proficiency, interper- publication SMA 98-3251.
3. Hall W, Babor TF. Cannabis use and public health:
fully controlled and analyzed, this ap- sonal relationships, and daily function- assessing the burden. Addiction. 2000;95:485-490.
proach is the most efficient way to ing. The extent to which these cogni- 4. Voelker R. “Decent research and closure” needed
on medical marijuana, says head of NIH panel. JAMA.
evaluate the long-term cognitive ef- tive impairments may recover following 1997;278:802.
fects of cannabis, given the costs and lo- cessation or reduction of cannabis use 5. Pope HG, Gruber AJ, Yurgelun-Todd D. The re-
gistical difficulties in using prospective will be addressed in a follow-up of this sidual neuropsychological effects of cannabis: the cur-
rent status of research. Drug Alcohol Depend. 1995;
research designs. The matching of sample subsequent to treatment for can- 38:25-34.
groups on measures of premorbid in- nabis dependence. 6. Solowij N. Cannabis and Cognitive Functioning.
Cambridge, England: Cambridge University Press;
tellectual functioning that are resilient 1998.
to brain damage, together with the ob- Author Contributions: Study concept and design: 7. Fried PA, Watkinson B, Gray R. Differential ef-
Solowij, Stephens, Roffman. fects on cognitive functioning in 9- to 12-year olds
served relationships between duration Acquisition of data: Stephens, Roffman, Kadden, Miller, prenatally exposed to cigarettes and marihuana. Neu-
of cannabis use and test performance, Christiansen, McRee, Vendetti. rotoxicol Teratol. 1998;20:293-306.

1130 JAMA, March 6, 2002—Vol 287, No. 9 (Reprinted) ©2002 American Medical Association. All rights reserved.

Downloaded From: [Link] by a Purdue University User on 05/21/2015

 COGNITIVE FUNCTIONING OF LONG-TERM CANNABIS USERS

8. Solowij N, Michie PT, Fox AM. Effects of long term 24. Bohme GA, Laville M, Ledent C, Parmentier M, Bury St Edmonds, England: Thames Valley Test Co;
cannabis use on selective attention: an event-related Imperato A. Enhanced long-term potentiation in mice 1992.
potential study. Pharmacol Biochem Behav. 1991;40: lacking cannabinoid CB1 receptors. Neuroscience. 39. Lezak MD. Neuropsychological Assessment. 3rd
683-688. 2000;95:5-7. ed. New York, NY: Oxford University Press; 1995.
9. Solowij N, Michie PT, Fox AM. Differential impair- 25. Marijuana Treatment Project Research Group. 40. Rey A. L’examen clinique en psychologie. Paris,
ments of selective attention due to frequency and du- Treating marijuana dependence: findings from a multi- France: Presse Universitaire de France; 1964.
ration of cannabis use. Biol Psychiatry. 1995;37:731- site study. Arch Gen Psychiatry. In press. 41. Spreen O, Strauss E. A Compendium of Neuro-
739. 26. McLellan AT, Kushner H, Metzger D, et al. The psychological Tests: Administration, Norms and Com-
10. Solowij N. Do cognitive impairments recover fol- Fifth Edition of the Addiction Severity Index. J Subst mentary. 2nd ed. New York, NY: Oxford University
lowing cessation of cannabis use? Life Sci. 1995;56: Abuse Treat. 1992;9:199-213. Press; 1998.
2119-2126. 27. Sobell LC, Sobell MB. Timeline follow-back: a tech- 42. Golden CJ. Stroop Color and Word Test: A Manual
11. Pope HG, Yurgelun-Todd D. The residual cogni- nique for assessing self reported alcohol consump- for Clinical and Experimental Uses. Wood Dale, Ill:
tive effects of heavy marijuana use in college stu- tion. In: Litten RZ, Allen JP, eds. Measuring Alcohol Stoelting; 1978.
dents. JAMA. 1996;275:521-527. Consumption: Psychological and Biochemical Meth- 43. Bohnen N, Jolles J, Twijnstra A. Modification of
12. Pope HG, Jacobs A, Mialet JP, Yurgelun-Todd D, ods. Totowa, NJ: Humana Press; 1992. the Stroop Color Word Test improves differentiation
Gruber S. Evidence for a sex-specific residual effect 28. Miller WR. Form 90: A Structured Assessment In- between patients with mild head injury and matched
of cannabis on visuospatial memory. Psychother Psy- terview for Drinking and Related Behaviors: Test controls. Clin Neuropsychol. 1992;6:178-184.
chosom. 1997;66:179-184. Manual: Project MATCH Monograph Series, Vol. 5. 44. The Wisconsin Card Sorting Test: Computer Ver-
13. Fletcher JM, Page JB, Francis DJ, et al. Cognitive Bethesda, Md: National Institute on Alcohol Abuse and sion. Alpharetta, Ga: Psychological Assessment Re-
correlates of long-term cannabis use in Costa Rican Alcoholism; 1996. NIH Publication 96-4004. sources Inc, CyberMetrics Testing Software Services;
men. Arch Gen Psychiatry. 1996;53:1051-1057. 29. First MB, Gibbon M, Spitzer RL, Williams JB. Us- 1989.
14. Elwan O, Hassan AAH, Naseer MA, et al. Brain er’s Guide for the Structured Clinical Interview for 45. Stuss DT, Stethem LL, Poirier CA. Comparison of
aging in a sample of normal Egyptians: cognition, edu- DSM-IV Axis I Disorders—Research Version. New three tests of attention and rapid information pro-
cation, addiction and smoking. J Neurol Sci. 1997; York, NY: Biometrics Research Department, New York cessing across six age groups. Clin Neuropsychol. 1987;
148:79-86. State Psychiatric Institute; 1996. 1:139-152.
15. Solowij N, Grenyer BFS, Peters R, Chesher G. Long 30. Jastak S, Wilkinson G. The Wide Range Achieve- 46. Levin HS, Mattis S, Ruff RM, et al. Neurobehav-
term cannabis use impairs memory processes and fron- ment Test: Manual of Instructions. Wilmington, Del: ioral outcome following minor head injury: a three-
tal lobe function. In: 1997 Symposium on the Can- Jastak Associates; 1984. center study. J Neurosurg. 1987;66:234-243.
nabinoids. Burlington, Vt: International Cannabinoid 31. Kareken DA, Gur RC, Saykin AJ. Reading on the 47. Geffen G, Moar KJ, O’Hanlon AP, Clark CR, Gef-
Research Society; 1997:84. Wide Range Achievement Test-Revised and parental fen LB. Performance measures of 16- to 86-year old
16. Loeber RT, Yurgelun-Todd DA. Human neuro- education as predictors of IQ: comparison with the Ba- males and females on the Auditory Verbal Learning
imaging of acute and chronic marijuana use: implica- rona formula. Arch Clin Neuropsychol. 1995;10:147- Test. Clin Neuropsychol. 1990;4:45-63.
tions for frontocerebellar dysfunction. Hum Psycho- 157. 48. Stuss DT, Stetham L, Pelchat G. Three tests of at-
pharmacol Clin Exp. 1999;14:291-301. 32. Blair JR, Spreen O. Predicting premorbid IQ: a re- tention and rapid information processing: an exten-
17. Block RI, O’Leary DS, Hichwa RD, et al. Cerebel- vision of the National Adult Reading Test. Clin Neu- sion. Clin Neuropsychol. 1988;2:246-250.
lar hypoactivity in frequent marijuana users. Neurore- ropsychol. 1989;3:129-136. 49. Roman DD, Edwall GE, Buchanan RJ, Patton JH.
port. 2000;11:749-753. 33. Barona A, Reynolds CR, Chastain R. A demographi- Extended norms for the Paced Auditory Serial Addi-
18. Block RI, O’Leary DS, Hichwa RD, et al. Effects cally based index of pre-morbid intelligence for the tion Task. Clin Neuropsychol. 1991;5:33-40.
of frequent marijuana use on memory-related re- WAIS-R. J Consult Clin Psychol. 1984;52:885-887. 50. Brittain JL, La Marche JA, Reeder KP, Roth DL,
gional cerebral blood flow. Pharmacol Biochem Be- 34. Bell R, Taylor EH, Ackerman B, Pappas AA. In- Boll TJ. Effects of age and IQ on Paced Auditory Se-
hav. In press. terpretation of urine quantitative 11-nor-delta-9- rial Addition Task (PASAT) performance. Clin Neu-
19. Block RI, O’Leary DS, Augustinack JC, et al. Ef- tetrahydrocannabinol-9-carboxylic acid to deter- ropsychol. 1991;5:163-175.
fects of frequent marijuana use on attention-related mine abstinence from marijuana smoking. Clin Toxicol. 51. Levy R, Goldman-Rakic PS. Segregation of work-
regional cerebral blood flow. Abstr Soc Neurosci. 2000; 1989;27:109-115. ing memory functions within the dorsolateral prefron-
26:2080. 35. Dackis CA, Pottash ALC, Annitto W, Gold MS. tal cortex. Exp Brain Res. 2000;133:23-32.
20. Lyketsos CG, Garrett E, Liang K-Y, Anthony JC. Can- Persistence of urinary marijuana levels after super- 52. Herman BP, Seidenberg M, Wyler A, et al. The
nabis use and cognitive decline in persons under 65 years vised abstinence. Am J Psychiatry. 1982;139:1196- effects of human hippocampal resection on the serial
of age. Am J Epidemiol. 1999;149:794-800. 1198. position curve. Cortex. 1996;32:323-334.
21. Hampson RE, Deadwyler SA. Cannabinoids, hip- 36. Ellis GM, Mann MA, Judson BA, Schramm NT, 53. Middleton FA, Strick PL. Anatomical evidence for
pocampal function and memory. Life Sci. 1999;65: Tashchian A. Excretion patterns of cannabinoid me- cerebellar and basal ganglia involvement in higher cog-
715-723. tabolites after last use in a group of chronic users. Clin nitive function. Science. 1994;266:458-461.
22. Hampson RE, Deadwyler SA. Cannabinoids re- Pharmacol Ther. 1985;38:572-578. 54. Schacter DL. Memory and awareness. Science.
veal the necessity of hippocampal neural encoding for 37. Fraser AD, Worth D. Urinary excretion profiles of 1998;280:59-60.
short-term memory in rats. J Neurosci. 2000;20:8932- 11-nor-9-carboxy-delta-9-tetrahydrocannabinol: a 55. Schmahmann JD, ed. The Cerebellum and Cog-
8942. delta-9-THCCOOH to creatinine ratio study. J Anal nition. San Diego, Calif: Academic Press; 1997.
23. Reibaud M, Obinu MC, Ledent C, et al. Enhance- Toxicol. 1999;23:531-534. 56. Herkenham M, Lynn AB, Little MD, et al. Can-
ment of memory in cannabinoid CB1 receptor knock- 38. Baddeley A, Emslie H, Nimmo Smith I. The Speed nabinoid receptor localization in brain. Proc Natl Acad
out mice. Eur J Pharmacol. 1999;379:R1-R2. and Capacity of Language-Processing Test Manual. Sci U S A. 1990;87:1932-1936.

©2002 American Medical Association. All rights reserved. (Reprinted) JAMA, March 6, 2002—Vol 287, No. 9 1131

Downloaded From: [Link] by a Purdue University User on 05/21/2015

 LETTERS

Table. Relative Risk of Parkinson Disease Among Patients With Polio Compared With a Nonexposed Cohort Matched by Age and Sex
Persons, No. Observed Parkinson Cases, No.

Patients Age-/Sex-Matched Patients Age-/Sex-Matched

Patient Group With Polio Cohort With Polio Cohort RR (95% CI)*
Total 5421 21626 29 50 2.3 (1.4-3.6)
Paralytic 2003 7979 13 23 2.2 (1.1-4.3)
Nonparalytic 2335 9317 7 14 1.9 (0.7-4.7)
Primary lymphocytic meningitis 592 2367 7 7 4.1 (1.4-11.9)
Suspected polio 491 1963 2 6 1.3 (0.2-5.7)
*Relative risk (RR) calculated according to person-years at risk. CI indicates confidence interval.

CI, 0.7-4.7) and in patients only suspected of having poliomy- served PD risk is not particular to the poliovirus but also ap-
elitis (RR, 1.3; 95% CI, 0.2-5.7). plies to other viruses infecting the central nervous system.
Comment. Although it has long been hypothesized that po- Nete Munk Nielsen, MD, PhD
liomyelitis is associated with an increased risk of PD,4 to our Klaus Rostgaard, MSc
knowledge this has never been empirically demonstrated. The Henrik Hjalgrim, MD, PhD
observed increased PD risk does not necessarily imply that po- Peter Aaby, MSc, DMSc
liovirus is directly implicated in PD pathogenesis. Rather, we Department of Epidemiology Research
speculate that by reducing the number of neurons essential to Danish Epidemiology Science Centre
Statens Serum Institut
normal neuronal functions, the virally induced damage may Copenhagen, Denmark
enhance the effect of normal age-related neuronal degenera-
Dorthe Askgaard, MD
tion and thus precipitate PD.1,4 Department of Infectious Diseases M
We acknowledge possible limitations of our data. Because National University Hospital
patients with poliomyelitis may be admitted to hospitals or may Copenhagen
attend outpatient clinics more often than other persons, de- Funding/Support: This study was supported by the Danish Medical Research Coun-
tection bias could arise. However, in Denmark, the diagnosis, cil, the Danish Development Research Council, the Danish National Research Foun-
evaluation, and treatment of PD normally take place at neu- dation, the Wedell-Wedellsborg’s Foundation, and the National Polio Society (PTU).
Acknowledgment: We are grateful to the staff at Copenhagen City Archives, who
rological departments or neurological outpatient clinics. More- helped us identify the polio patients’ records.
over, although patients with PD initially may consult private 1. Calne DB, Eisen A, McGeer E, Spencer P. Alzheimer’s disease, Parkinson’s dis-
neurologists or general practitioners, the vast majority of pa- ease, and motoneurone disease: abiotrophic interaction between ageing and en-
tients with PD will at some point undergo clinical evaluation vironment? Lancet. 1986;2:1067-1070.
2. Scott WK, Nance MA, Watts RL, et al. Complete genomic screen in Parkinson
or hospitalization at specialized hospital departments because disease: evidence for multiple genes. JAMA. 2001;286:2239-2244.
of the complexity of the disease. Therefore, we think that most 3. Bodian D. Histopathologic basis of clinical findings in poliomyelitis. Am J Med.
1949;6:563-577.
Danish patients with PD would be registered in the NHDR, and 4. Vincent FM, Myers WG. Poliomyelitis and parkinsonism. N Engl J Med. 1978;
we consider detection bias to be an unlikely explanation for 298:688-689.
5. Munk NM, Wohlfahrt J, Aaby P, et al. Cancer risk in a cohort of polio patients.
our findings. Int J Cancer. 2001;92:605-608.
Patients with polio may present a wide range of neurologi- 6. Lassen HCA. Sammenlignende undersøgelser over primær serøs meningitis og
paralytisk poliomyelitis [Comparative studies of primary lymphocytic menigitis and
cal symptoms, which could cause diagnostic ambiguity. If di- paralytic polio]. Ugeskr Læg. 1939;3:73-80.
agnostic misclassification would explain our observations we
would have expected the risk of PD to be particularly in-
creased in patients with paralytic polio. However, an in- CORRECTION
creased risk of PD was also observed in patients with nonpara-
lytic polio. Moreover, the likely inclusion of patients with Incorrect Measure: In the Original Contribution entitled “Cognitive Functioning
of Long-term Heavy Cannabis Users Seeking Treatment” published in the March
nonpolio virus-related meningitis in the group of patients with 6, 2002, issue of THE JOURNAL (2002;287:1123-1131), the legend for the Figure
primary lymphocytic meningitis may indicate that the ob- should indicate that error bars represent SEM, not SD.

©2002 American Medical Association. All rights reserved. (Reprinted) JAMA, April 3, 2002—Vol 287, No. 13 1651

Downloaded From: [Link] by a Purdue University User on 05/21/2015