Health Care Use by Frequent Marijuana Smokers
Who Do Not Smoke Tobacco
MICHAEL R. POLEN, MA, Portland, Oregon; STEPHEN SIDNEY, MD, MPH; IRENE S. TEKAWA, MS;
MARIANNE SADLER, MPH; andGARY D. FRIEDMAN, MD, MS, Oakland, California
Even though marijuana smoke contains carcinogens and more tar than tobacco smoke and marijuana intoxica-
tion has been implicated as a risk factor for injuries, relatively little epidemiologic evidence has identified mari-
juana use as a risk factor for ill health. This study is the first to examine the health effects of smoking marijuana
by comparing the medical experience of "daily" marijuana smokers who never smoked tobacco (n = 452) with a
demographically similar group of nonsmokers of either substance (n = 450). Marijuana smoking status was de-
termined during multiphasic health checkups at Kaiser Permanente medical centers between July 1979 and De-
cember 1985. Medical records were reviewed for as long as 2 years after the checkups. Frequent marijuana
smokers had small increased risks of outpatient visits for respiratory illnesses (relative risk [RR] = 1.19; 95% con-
fidence interval [Cl] = 1.01, 1.41), injuries (RR = 1.32; Cl = 1.10, 1.57), and other types of illnesses (RR = 1.09;
Cl = 1.02, 1.16) compared with nonsmokers; their risk of being admitted to a hospital was elevated but not sta-
tistically significant (RR = 1.51; Cl =0.93, 2.46). Analyses were adjusted for sex, age, race, education, marital
status, and alcohol consumption. Daily marijuana smoking, even in the absence of tobacco, appeared to be asso-
ciated with an elevated risk of health care use for various health problems.
(Polen MR, Sidney S, Tekawa IS, Sadler M, Friedman GD: Health care use by frequent marijuana smokers who do not smoke tobacco.
West J Med 1993; 158:596-601)
arijuana is the most widely used illicit recreational
drug in the United States. Marijuana smoke con-
tains carcinogens and more tar than tobacco smoke, and
the psychoactive effects ofmarijuana ingestion have been
implicated as a risk factor for injuries. Yet, surprisingly
little epidemiologic evidence is available concerning the
possible effects of smoking marijuana on respiratory
health, injury risk, or other health conditions.*
As a pilot project for a larger study of the health con-
sequences of smoking marijuana, we compared the med-
ical experience of persons who self-reported frequent
marijuana smoking but who never smoked tobacco with
that of a demographically matched group who reported
never smoking tobacco or marijuana. To our knowledge,
this study is the first to review the medical records of a
large number of non-tobacco smoking marijuana users.
Subjects and Methods
Study subjects were selected from members of the
Kaiser Permanente Medical Care Program who had at
least one multiphasic health checkup (MHC) at the Oak-
land or San Francisco, California, medical centers be-
*See also the editorial by D. P. Tashkin, MD, "Is Frequent Marijuana Smoking
Harmful to Health?" on pages 635-637 of this issue.
tween July 1979 and December 1985. The multiphasic
health checkup was completed by about 30,300 persons
per year during the study period; the examination was dis-
continued at San Francisco in 1980. TheMHC population
has been described previously and has provided the data
for numerous studies of smoking and drinking habits."2
The Kaiser Permanente membership is demographically
and socially heterogeneous, reflecting the diversity of the
San Francisco Bay Area population, but impoverished
and undereducated persons are underrepresented.3 Per-
sons who take the MHC tend to be more health conscious
and better educated than Kaiser Permanente members in
The MHC examinees were given several question-
naires documenting demographic and social characteris-
tics, medical history, health habits, the use of tobacco and
marijuana (questionnaire administered from July 1979 to
1986), and the use of alcohol (questionnaire administered
from 1978 to 1985). About 86% of the examinees com-
pleted the tobacco or marijuana use questionnaire, and
80% completed the alcohol use questionnaire.
Definitions ofStudy Groups
About 14,600 respondents to the tobacco-marijuana
survey during the study period reported smoking mari-
From the Kaiser Permanente Center for Health Research, Portland, Oregon (Mr Polen), and the Division of Research, Kaiser Permanente Medical Group, Oak-
land, Califomia (Dr Sidney, Ms Tekawa, Ms Sadler, and Dr Friedman).
This work was funded in part by the Kaiser Foundation Research Institute and by grant No. RO1 DA 06609 from the National Institute on Drug Abuse.
Reprint requests to Michael R. Polen, MA, Kaiser Permanente Center for Health Research, 3800 N Kaiser Center Dr, Portland, OR 97227-1098.
ABBREVIATIONS USED IN TEXT
CI = confidence interval
MHC = multiphasic health checkup
RR = relative risk
juana more than six times in their lifetimes and that they
currently smoked marijuana. Among these persons, a
group of 746 marijuana smokers reported smoking mari-
juana almost every day and never smoking tobacco. We
defined the first MHC at which the tobacco-marijuana
survey was completed as the index MHC for the mari-
juana smoking group. We selected a comparison group of
nonsmokers from respondents who reported never smok-
ing marijuana or tobacco. For each marijuana smoking
subject, we attempted to match a nonsmoker by sex, age
(birth year), race (Asian, African American, white), and
index MHC date (within a month). We initially matched
controls with 709 marijuana smokers. To control for vary-
ing lengths of Kaiser Permanente membership, we se-
lected for study those pairs in which both members were
enrolled in Kaiser for either at least one or at least two
years after the index MHC. This process left 486 pairs for
whom medical charts were reviewed (65% of the 746
original marijuana smokers). An additional 70 persons
were excluded from this report because ofthe unavailabil-
ity of an alcohol questionnaire after December 1985. The
final sample thus consisted of452 marijuana smokers and
The index MHC questionnaire provided information
on sex, age, race, education, and marital status; self-re-
ported health status (a checklist of34 medical conditions);
the existence of serious illnesses; hospital admissions in
the past year; and the presence of problems with drugs.
Data on marijuana and tobacco use, duration ofmarijuana
use, and number of days ill with a cold, flu, or sore throat
were obtained from the supplemental tobacco-marijuana
survey. We used a drinking status measure derived in pre-
vious studies from items regarding quantity and fre-
quency of drinking on the alcohol survey.6 Six drinking
categories were defined for this study: abstainer, ex-
drinker, occasional (<1 drink per month), less than one
drink per day but more than one per month, one to two
drinks per day, and three or more drinks per day. The
small number of ex-drinkers was combined with abstain-
ers in the main analyses.
Kaiser Permanente outpatient medical records were
reviewed by two trained medical records technicians who
were unaware of subjects' marijuana smoking status. The
medical records ofeach pair of subjects were reviewed for
the same follow-up period, which was either one or two
years after the index MHC date, depending on the length
of membership in the Kaiser program. Following a writ-
ten protocol, records reviewers identified all office visits
to a physician or nurse practitioner. Prenatal visits were
excluded, as were telephone contacts, letters, and visits
for procedures only, such as ultrasonograms. As many as
three illnesses were coded as reasons for the visit, using
the International Classification of Diseases, 9th Revision,
codes.7 Each visit was assigned to one of threecategories
for this report: respiratory conditions (codes 460 to 519),
injuries and poisonings (codes 800 to 999), or other dis-
eases and conditions (all other codes). If more than one,
the reason listed first was used. Inpatient data were ob-
tained from Kaiser Permanente computer-based hospital
admission files. All hospital admissions were included.
Because of the small number of hospital admissions dur-
ing the follow-up period, we combined all diagnoses in
Differences in the distribution of baseline characteris-
tics were evaluated with x2 tests. Poisson regression was
used to estimate the relative risk for the marijuana smok-
ing group versus the nonsmoking group of outpatient vis-
its or hospital admissions. The matched-pairs design was
dropped in the analysis, and the variables used for match-
ing were instead controlled statistically to avoid dropping
both members of a pair when only one member was
missing data on a covariate. Covariates in each regres-
sion model included alcohol consumption, age, sex, race,
educational level, and marital status. Terms for marijuana
exposure interacting with drinking status were introduced
to examine possible interaction effects. The SAS statisti-
cal analysis package was used for all analyses.8
Characteristics of the marijuana smokers and the non-
smokers at the time of the index MHC are compared in
Table 1. Marijuana smokers had a lower educational level
and were less likely to be married than nonsmoking sub-
jects. Alcohol drinking levels differed dramatically be-
tween the two groups, with marijuana smokers being
more likely to drink alcohol and (among current drinkers)
to drink more heavily. Marijuana smokers reported more
days ill with a cold, flu, or sore throat in the past year and
were more likely than the nonsmoking group to report
having a serious problem with drugs.
A total of 6,088 visits were recorded, including 3,206
among the marijuana smoking group and 2,882 among
the nonsmoking group. At least one outpatient visit for
respiratory problems was made by 36% of the marijuana
smokers versus 33% of the nonsmokers, 39% versus 28%
made at least one visit for injury, and 94% versus 93%
made at least one visit for other reasons. After adjustment
for covariates, the marijuana smoking group showed
small but statistically significant increased risks of outpa-
tient visits for all three categories of conditions (Table 2).
Interaction between marijuana smoking and alcohol
consumption was significant in relation to visits for injury
and for other reasons but not in relation to visits for respi-
ratory problems. The percentage distributions of visits for
injury (none versus .1) by drinking level suggest that the
generally higher injury risk for the marijuana smoking
group compared with the nonsmoking group was reduced
in the heaviest drinking level (Table 3). For other dis-
THE WESTERN JOURNAL OF MEDICINE
HEALTH OF FREQUENT MARIJUANA SMOKERS
OF FREQUENT MARIJUANA
eases, the difference in risk was higher in the nondrinking
and heaviest drinking levels (Table 3).
The duration of marijuana smoking was associated in
different ways with the three categories of visits (Table 4).
The risk of respiratory visits was significantly elevated for
persons who had smoked marijuana for less than ten
years, but not for those who had smoked for ten years or
more. Among marijuana smokers, there was a negative
association between duration of smoking and risk of visits
for respiratory problems (P= .0002). For injury visits,
however, a longer duration of marijuana smoking was as-
sociated with a greater risk. Compared with the non-
smoker group, persons who had smoked marijuana for 15
years or more had twice the risk of visits for injury. A test
for linear trend was significant (P= .0001). For other
types of visits, only those who had smoked marijuana for
five to nine years had a significantly increased risk com-
pared with the nonsmoker group, and the other duration
categories did not follow a consistent pattern.
There were 86 hospital admissions in the two study
groups. The relative risk for the marijuana smoking group
compared with the nonsmoker group was elevated but not
statistically significant (relative risk [RR]=1.51; 95%
confidence interval [CI]=0.93, 2.46; P=.10). A duration
of marijuana use of less than five years was of borderline
significance (RR= 2.02; 95% CI=0.98, 4.15; P =.06).
TABLE 2.-Relative Risk (RR) ofOutpatient Medical Visits for Respiratory, Injury, and Other
Conditions byMarijuana Smokers (n=452) Versus Nonsmokers (n=450)*
Nonsmokers (reference) ..............
Marijuana smokers ..............
'Adjusted for sex, age, race, educational level, marital status, and alcohol consumption.
TABLE 1.-Self-reported Characteristics ofMarijuanaSmokers (n = 452)and Nonsmokers (n = 450)at
Index Multiphasic Health Checkup, July 1979 to 1985 (4b)
African American .............
High school or less ...........
Some college or vocational
College graduate or more .....
Missing data .................
Separated, divorced, widowed .
Never married ................
Missing data .................
Alcohol drinking status
Never drank ..................
Former drinker ...............
<1 drink/mo ...............
>1 drink/mo, <1/day .......
1-2 drinks/day .............
.3 drinks/day ..............
Current problem with drugs?
Number of health conditions (34 possible)
Missing data .................
Mean (SD) ...................
Daysill with cold, flu, sore throat in past year
Missing data .......
Serious illness in past year?
Missing data ...
Admitted to hospital in past year?
Missing data ...
Duration ofmarijuana use, yr
THE WESTERN JOURNALOF MEDICINE
There was no significant interaction between marijuana
smoking and alcohol use.
Although epidemiologic studies concerning the health
consequences of smoking marijuana are relatively few, a
long-term use of marijuana has been implicated as a pos-
sible hazard to mental, pulmonary, immune, and repro-
ductive functioning,9 and marijuana intoxication has been
implicated as a risk factor for accidents and injuries.'0 Be-
cause marijuana is usually ingested by smoking, possible
respiratory effects have naturally been the most fre-
Compared with tobacco smokers, marijuana smokers
typically smoke fewer cigarettes per day but consume
more of the cigarette, inhale longer and more deeply, and
retain the smoke longer in the lungs." This probably ex-
plains why the respiratory deposition of tar and adsorp-
tion of carbon monoxide in experimental subjects were
four and five times higher, respectively, after smoking
marijuana than after smoking tobacco.'2 Because of the
high prevalence of cigarette smoking among marijuana
smokers, however, it has been difficult to assess the inde-
pendent association between marijuana smoking and res-
piratory disease, including lung cancer.'3 Marijuana use
has been linked to respiratory problems."4-'7 In one study,
heavy smokers of marijuana, whether or not they also
smoked tobacco, reported more chronic bronchitis symp-
toms and more acute bronchitis episodes than nonsmokers
of either marijuana or tobacco.'5 Heavy marijuana smok-
ing, with or without concomitant tobacco smoking, ap-
pears to affect large (but not small) airway function
adversely and to produce histologic lesions in the air-
ways.'5 "il" A positive association between smoking "non-
tobacco cigarettes" (presumed to be marijuana) and res-
piratory symptoms in smokers and nonsmokers of to-
bacco was reported in a population survey in Tucson,
Arizona. Lung function was decreased among male-but
not female-smokers of nontobacco cigarettes more than
among tobacco smokers."9
Marijuana has repeatedly been found to be the second
most common drug, after alcohol, present in the blood of
nonfatally and fatally injured persons,A'21 although the role
marijuana may play in injury-producing events remains
uncertain.22 Laboratory studies have shown decreased
driving-related skills after smoking marijuana.'0 In one
experimental study, driving performance declined sub-
stantially after ingesting marijuana and alcohol together,
but did not decline after taking either substance alone.23
In a Swedish study following a cohort of 45,540 male
military conscripts for 15 years, heavier cannabis users
had a nearly three times greater risk of death than
nonusers, but the association did not remain statistically
significant after control for the use ofalcohol, other drugs,
and social background variables.' In another study of the
same Swedish cohort, heavier cannabis users had an ele-
vated risk of schizophrenia compared with nonusers, even
after adjusting for psychosocial covariates.25
As the first evidence based on medical records of non-
tobacco smoking, daily marijuana users, our results make
a unique contribution to the growing research on the
harmful health effects of marijuana. Our finding of an in-
creased risk of respiratory-related outpatient visits ex-
pands the evidence suggesting that frequent marijuana
smoking may increase the risk of respiratory illness in-
dependent of tobacco smoking. Marijuana smokers in our
TABLE 3.-Percentage Making 1 orMore Outpatient Visits for Injury and
Other (Nonrespiratory) Causes byMarijuana andAlcohol Use
Never or former ...........
<1 drink/day ..............
1-2 drinks/day ............
.3 drinks/day .............
TABLE 4.-Relotive Risk (RR) ofOutpatient Medical Visits forRespiratory, Injury, and
OtherConditions byDuration ofMarijuana Smoking (n = 895)'
Duration ofMonjuona Exposure, yr
Cl = confidence interval
'Adjusted for sex, age, race, educational level, marital status, and alcohol consumption.
HEALTH OF FREQUENT MARIJUANA SMOKERS
study also reported a higher prevalence of upper respira-
tory tract infections compared with nonsmokers. On the
other hand, the duration of marijuana smoking appeared
to be inversely related to the risk of outpatient visits for
respiratory problems. This result was contrary to our ex-
pectation and remains an issue for future research. In our
data, long-term marijuana smokers may be the "sur-
vivors" of a selection process in which persons who expe-
rienced respiratory symptoms were more likely to quit
smoking marijuana early in the process.
We also found increased risks of injury-related and
other (nonrespiratory, noninjury) outpatient visits among
marijuana smokers, suggesting that marijuana use may
have many adverse health effects. The complex interac-
tion between marijuana and alcohol use in relation to the
risk of medical care use in both of these areas underscores
the important role of alcohol in combination with mari-
juana.2627 The duration of marijuana use appeared to be
positively related to the risk of making injury-related vis-
its. As with respiratory-related visits, this result was unex-
pected. It is not clear why a longer use ofmarijuana would
be associated with a greater injury risk and a lower respi-
ratory risk, but these results deserve additional study. The
duration of marijuana use was not related to the risk of
other types of outpatient visits.
Marijuana smokers in our study also tended to have an
elevated risk of being admitted to a hospital. Although the
association was not statistically significant, it was consis-
tent with our results regarding medical office visits and
thus supports the hypothesis that marijuana smoking is as-
sociated with adverse health effects.
Several caveats should be noted. We had no data re-
garding the use of other drugs such as cocaine; if associ-
ated with marijuana use, they may account for the
observed differences. Only 6% of the marijuana smokers,
however, reported having serious problems with drugs at
their index MHC.
Another potential problem is our reliance on self-
reporting. The use of illegal or socially undesirable sub-
stances such as tobacco, alcohol, marijuana, and other
drugs may well be underreported.28 We are unable to as-
sess this possible bias. Our study subjects, however, com-
pleted their questionnaires in a health care setting in
which confidentiality was assured.
Finally, our study was intended to serve as a pilot "hy-
pothesis-generating" study of marijuana smoking and
health. Restricting subjects to nonsmokers of tobacco
allowed us to efficiently assess an association between
marijuana use and respiratory illnesses independent of to-
bacco use but at the cost of studying an unrepresenta-
tive sample of marijuana smokers, many of whom smoke
In summary, daily marijuana smoking appears to be
associated with respiratory conditions even among per-
sons who never smoked tobacco. This association is con-
sistent with a possible independent deleterious effect of
marijuana smoking on respiratory health. Frequent mari-
juana use also appears to be intimately linked to alcohol
consumption as a risk factor for injury-related and other
medical care. These data are limited, but a larger study in
progress (S.S., principal investigator) should elucidate the
joint roles of tobacco and marijuana in relation to respira-
tory health and of alcohol and marijuana in relation to
traumatic injuries and other illnesses. This study will in-
clude never users, formner users, and current users of mar-
ijuana, tobacco, and alcohol and will examine medical
experience and mortality over a longer period.
Physicians and other primary care professionals may
wish to consider counseling patients who frequently
smoke marijuana about the potentially widespread harm
its use may entail. In our study, physicians recorded mari-
juana use in the medical records of only 3% of the mari-
juana smokers, all ofwhom smoked daily or almost daily.
As evidence of the health consequences ofsmoking mari-
juana accumulates, physicians should take note of this
possibly important health behavior.
This work was assisted by Susan Reinheimer, Patricia O'Rourke,
Harald Kipp, Merrill Jackson, and Diana Holt.
1. Friedmnan GD, Seltzer CC, Siegelaub AB, Feldman R, Collen MF: Smoking
among white, black, and yellow men and women. Am JEpidemiol 1972; 96:23-35
2. Friedman GD, Sidney S, Polen MR: Smoking habits among multiphasicex-
aminees, 1979 to 1984. West M Med 1986; 145:651-656
3. Krieger N: Overcoming the absence of socioeconomic data in medical
records: Validation and application of a census-based methodology. Am J Public
Health 1992; 92:703-7 10
4. Friedman G: Effects of MHTS on patients,In Collen MF(Ed): Multiphasic
Health Testing Services. New York, NY, John Wiley & Sons, 1978, pp531-549
5. Hiatt RA, Friedman GD: Characteristics ofpatients referred for treatment of
end-stage renal disease in a defined population. Am J Public Health 1982; 72:829-
6. Armstrong MA, Klatsky A: Alcohol use and later hospitalization experi-
ence. Med Care 1989; 27:1099-1108
7. Commission on Professional and Hospital Activities: Intemational Classifi-
cation of Diseases, 9th Revision-Clinical Modification. Ann Arbor, Mich, World
Health Organization, 1978
8. SAS/STAT User's Guide, Version 6,4th Edition, Vol 2. Cary, NC, SAS In-
9. Drug Abuse and Drug Abuse Research: The Third Triennial Report to Con-
gress from the Secretary, Department of Health and Human Services. Rockville,
Md, National Institute on Drug Abuse, 1991
10. Moskowitz H: Adverse effects of alcohol and other drugs on human per-
formance. Alcohol Health Res World 1985;9:11-15
1 1. Schwartz RH: Marijuana: An overview. Pediatr Clin North Am 1987;
12. Wu TC, Tashkin DP, Djahed B, Rose JE: Pulmonary hazards of smoking
marijuana as compared with tobacco. N Engl J Med 1988; 318:347-351
13. Institute of Medicine: Marijuana and Health. Washinoton. DC, National
Academy Press, 1982
14. Tashkin DP: Pulmonary complications of smoked substance abuse,lIt Ad-
diction Medicine (Special Issue). West J Med 1990; 152:525-530
15. Tashkin DP, Fligiel S, Wu TC, et al: Effects of habitual use of marijuana
and/or cocaine on the lung, In Chiang CN, Hawks RL(Eds): Research Findings on
Smoking of Abused Substances-NIDA Research Monograph 99. Rockville, Md.
National Institute on Drug Abuse, US Department of Health and Human Services
publication No. (ADM) 90-1690, 1990
16. Tashkin DP, Coulson AH, Clark VA, et al: Respiratory symptoms and
lung function in habitual heavy smokers ofmarijuana alone, smokers ofmarijuana
and tobacco, smokers of tobacco alone, and nonsmokers. Amn Rev Respir Dis
1987; 135: 209-216
17. Tashkin DP, Shapiro BJ, Lee YE, Harper C: Subacute effects of heavy
marijuana smoking on pulmonary function in healthy men. N EngI J Med 1976:
18. Gong H Jr, Fligiel S, Tashkin DP, Barbers RG: Tracheobronchial changes
in habitual, heavy smokers ofmarijuana with and without tobacco. Am Rev Respir
Dis 1987: 136:142-149