The impact of marijuana use on glucose, insulin, and insulin resistance among us adults
The Impact of Marijuana Use on Glucose, Insulin, andInsulin Resistance among US AdultsElizabeth A. Penner, BS,,Hannah Buettner, Murray A. Mittleman, MD, ,aUniversity of Nebraska College of Medicine, Omaha; bDepartment of Epidemiology, Harvard School of Public Health, Boston, Mass;cCardiovascular Epidemiology Research Unit, Department of Medicine, Beth Israel Deaconess Medical Center, Boston, Mass.
BACKGROUND: There are limited data regarding the relationship between cannabinoids and metabolicprocesses. Epidemiologic studies have found lower prevalence rates of obesity and diabetes mellitus inmarijuana users compared with people who have never used marijuana, suggesting a relationship betweencannabinoids and peripheral metabolic processes. To date, no study has investigated the relationshipbetween marijuana use and fasting insulin, glucose, and insulin resistance. METHODS: We included 4657 adult men and women from the National Health and Nutrition ExaminationSurvey from 2005 to 2010. Marijuana use was assessed by self-report in a private room. Fasting insulin andglucose were measured via blood samples after a 9-hour fast, and homeostasis model assessment of insulinresistance (HOMA-IR) was calculated to evaluate insulin resistance. Associations were estimated usingmultiple linear regression, accounting for survey design and adjusting for potential confounders. RESULTS: Of the participants in our study sample, 579 were current marijuana users and 1975 were pastusers. In multivariable adjusted models, current marijuana use was associated with 16% lower fastinginsulin levels (95% confidence interval [CI], À26, À6) and 17% lower HOMA-IR (95% CI, À27, À6). Wefound significant associations between marijuana use and smaller waist circumferences. Among currentusers, we found no significant dose-response. CONCLUSIONS: We found that marijuana use was associated with lower levels of fasting insulin andHOMA-IR, and smaller waist circumference. Ó 2013 Elsevier Inc. All rights reserved. The American Journal of Medicine (2013) -, ---
KEYWORDS: Glucose; Insulin; Insulin resistance; Marijuana use
Marijuana is the most commonly used illicit drug in the
marijuana in 19 states and the District of Columbia, physicians
United States, and use is increasing. The 2010 National
will increasingly encounter marijuana use among their patient
Survey on Drug Use and Health reported that between 2007
populations.Marijuana use is associated with an acute
and 2010, the prevalence of marijuana use among persons aged
increase in caloric intake,and people who smoke marijuana
12 years and older increased from 5.8% to 6.9%, meaning there
have higher average caloric intake levels than nonuser
are an estimated 17.4 million current users of marijuana.
Despite these associations with increased caloric intake,
Approximately 4.6 million of these users smoked marijuana
marijuana use has been associated with lower body mass
daily or almost dailyWith the recent legalization of recrea-
index (Band a lower prevalence of and dia-
tional marijuana in 2 states and the legalization of medical
betes mellitus.The mechanisms underlying this paradoxhave not been determined, and the impact of regular mari-juana use on insulin resistance and cardiometabolic risk
Funding: None. Conflict of Interest: None.
factors remains unknown. In this study of 4657 participants
Authorship: All authors had access to the data and played a role in
in the National Health and Nutrition Examination Survey
(NHANES) from 2005 to 2010, we examined the associa-
EAP and HB are joint first authors.
tions between habitual marijuana use and measures of
Requests for reprints should be addressed to Murray A. Mittleman, MD,
fasting glucose and insulin levels, insulin resistance, and
DrPH, 375 Longwood Ave, Boston, MA 02215.
components of the metabolic syndrome.
0002-9343/$ -see front matter Ó 2013 Elsevier Inc. All rights reserved.
The American Journal of Medicine, Vol -, No -, - 2013
corrected because of bias from quality controls (SolomonPark Research Laboratories, Kirkland, Wash), using the
following formula: corrected HDL-C ¼ [(Solomon Park
The NHANES is a cross-sectional, continuous survey
assigned HDL-C value) Â (participant HDL-C)]/(quality
administered annually by the National Center for Health
control HDL-C value associated with participant sample)].
Statistidata are released in 2-year increments. The
From 2007 to 2010, HDL-C testing was performed at the
ics, Inc, South San Francisco, Calif).
Despite its associations with increased
mass index and prevalence of diabetes.
Blood Pressure, Body Mass Index, and Waist Circum-
ference. All measurements were collected during the
Have you ever, even once, smoked marijuana or
physical examination in mobile examination centers,
hashish?” (yes, no, refused, don’t know);
according to standard NHANES protocol.Blood pressure
How long has it been since you last used marijuana or
estimates were calculated by averaging 3 blood pressure
hashish?” (answers were given as number of days, weeks,
readings. BMI was calculated as weight in kilograms
divided by the square of height in meters.
During the past 30 days, on how many days did you use
Responses to these questions were used to classify
Characterization of Sociodemographics and
participants as never users (never smoked marijuana, n ¼
2103); past users (smoked marijuana at least once but not in
Participants reported age, sex, race/ethnicity, education
the past 30 days, n ¼ 1975); and current users (smoked
level, income, marital status, tobacco use, physical activity
marijuana at least once in the prior 30 days, n ¼ 579).
level, and alcohol use. Race/ethnicity was classified asHispanic, non-Hispanic white, non-Hispanic black, orother. We classified education level as less than high
school, high school or equivalent, or some college. Income
Insulin, Glucose, Homeostatic Model Assessment Insulin
was categorized as less than $20,000, $20,000 to $44,999,
Resistance Score, High-Density Lipoprotein Cholesterol,
$45,000 to $74,999, and greater than or equal to $75,000.
Hemoglobin A1c, and Triglycerides. Participants pro-
Participants were classified into 3 groups of tobacco ciga-
vided blood samples in the morning after a 9-hour fast.
rette exposure: current user of tobacco cigarettes, past user
The homeostasis model assessment of insulin resistance
of tobacco cigarettes, and lifetime nonsmoker (defined as
(HOMA-IR), a measure of insulin resistance, was calculated
<100 cigarettes in lifetime). Physical activity was classified
as fasting serum insulin (mU/mL) Â fasting plasma glucose
as active, defined as report of any regular moderate or
vigorous physical activity, or inactive, defined as report of
From 2005 to 2006, high-density lipoprotein cholesterol
no regular moderate or vigorous physical activity. Alcohol
(HDL-C) testing was performed at Johns Hopkins Univer-
use was classified as nondrinkers, less than or equal to 1
sity, using the Hitachi 717 and Hitachi 912 (Roche Diag-
drink per week, 1 to 14 drinks per week, or more than 14
nostics, Indianapolis, Ind). In this cycle, values were
Characteristics (%) of Participants from the National Health and Nutrition Examination Survey (n ¼ 4657), 2005 to 2010*
Analyses were weighted to reflect national population estimates.
multiple linear regression models with BMI, logarithmic
All analyses were weighted to adjust for the complex
fasting insulin levels, fasting glucose levels, logarithmic
sampling design of the NHANES. We used chi-square tests
to compare baseline characteristics across never, former, and
levels, HDL-C levels, systolic blood pressure, diastolic
blood pressure, and waist circumference as continuous
Because data on income were missing in 306 participants
outcomes. We first examined models adjusted for age
(7%) in the study population, we used Markov chain Monte
and sex, and then performed multivariable regressions
Carlo multiple imputation to simulate 5 complete datasets.
accounting for all of the following covariates, which were
All statistical analyses were performed in each dataset. The
specified a priori as potential confounders: age, sex, race/
results were then averaged using the mi estimate command
ethnicity, education level, income, marital status, tobacco
in STATA, and P values and confidence intervals (CIs)
use, physical activity level, and alcohol use. Because BMI
incorporating the uncertainty in the imputed estimates were
may mediate the association between marijuana use and our
We then compared the imputed and observed
study outcomes, we examined the impact of further adjust-
values to assess the reasonableness of the imputation model.
ing for BMI in multivariable models.
Insulin, HOMA-IR, and triglycerides were skewed and
We examined whether there was a nonlinear association
log-transformed to approximate normality. We fit separate
between frequency of marijuana use and logarithmic fasting
The American Journal of Medicine, Vol -, No -, - 2013
Mean Values (Standard Errors) of Fasting Insulin and Glucose According to Average Marijuana Use Among Participants from the
National Health and Nutrition Examination Survey, 2005 to 2010
BMI ¼ body mass index; DBP ¼ diastolic blood pressure; HbA1c ¼ hemoglobin A1c; HDL-C ¼ high-density lipoprotein cholesterol; HOMA-IR ¼
homeostasis model assessment of insulin resistance; SBP ¼ systolic blood pressure.
*Means for insulin, HOMA-IR, and triglycerides are geometric.
insulin and HOMA-IR among current users of marijuana by
relationship and no evidence for a U- or J-shaped curve. We
including the difference between median intake and reported
did not find any significant associations between marijuana
intake and the square of this value as continuous terms in
use and triglyceride levels, systolic blood pressure, or dia-
our multivariable regression model. Because people with
diabetes mellitus may alter their marijuana use habits, we
In an analysis adjusting for BMI, a potential mediator
also performed a sensitivity analysis excluding participants
of the associations between marijuana use and the car-
with diabetes mellitus. All analyses were conducted using
diometabolic outcomes, the associations between current
STATA 12 (StataCorp LP, College Station, Tex).
marijuana use and fasting levels of insulin, HOMA-IR, andwaist circumference were attenuated, but remained statisti-cally significant (). In addition, the results were not
materially different in analyses that excluded participants
Of the 4657 NHANES participants in our study sample, 579
(representing 12.2%) were current users of marijuana and1975 (representing 47.7%) had used marijuana at least oncein their lifetime, but not in the past 30 days. Compared with
lifetime nonusers, participants who reported marijuana use
In this large, cross-sectional study, we found that subjects
in the past month tended to be male, younger, and current
who reported using marijuana in the past month had lower
levels of fasting insulin and HOMA-IR, as well as smaller
In unadjusted analyses, past and current marijuana use
waist circumference and higher levels of HDL-C. These
were associated with lower levels of fasting insulin, glucose,
associations were attenuated among those who reported
HOMA-IR, BMI, and hemoglobin A1c (Current
using marijuana at least once, but not in the past 30 days,
marijuana use also was found to be inversely associated with
suggesting that the impact of marijuana use on insulin and
waist circumference. Models adjusted for age and sex
insulin resistance exists during periods of recent use.
demonstrated statistically significant associations between
There have been discrepant findings on the relationship
past and current use of marijuana with lower levels of
between marijuana use and BMI. A study of young adults
fasting insulin, glucose, HOMA-IR, and BMI. Also, current
examining associations between marijuana use and cardio-
use was associated with higher HDL-C levels and lower
vascular risk factors reported no significant trend between
waist circumference (In multivariable-adjusted
marijuana use and whereas analyses of 2 large
models, the associations of current marijuana use with lower
nationally representative surveys found lower BMI and
levels of fasting insulin and HOMA-IR, as well as with
decreased prevalence of obesity.Few studies have
higher HDL-C levels and lower waist circumference,
explored possible underlying explanations for these associ-
remained statistically significant (Compared with
ations. However, a recent analysis using NHANES III data
participants reporting never having used marijuana in their
showed that marijuana users had a lower prevalence of
lifetimes, current use was associated with 16% lower fasting
diabetes mellitus compared with nonusers;similar results
insulin levels (95% CI, À26 to À6), 17% lower HOMA-IR
have been found with administration of cannabidiol in
(95% CI, À27 to À6), and 1.63 mg/dL higher HDL-C levels
a mouse model.In the present study, we demonstrate
(95% CI, 0.23-3.04) in multivariable adjusted models.
a significant association between current marijuana use and
Among current users, we found no significant dose-response
lower levels of fasting insulin and insulin resistance in
Adjusted Mean/Percent Differences in Measures of Carbohydrate Metabolism and Body Mass Index According to Marijuana Use Among Participants From the National Health and
Nutrition Examination Survey, 2005 to 2010
1.37 (À0.01, 2.74) À0.65 (À1.13, À0.18) À0.49 (À1.68, 0.70)
À1.08 (À2.39, 0.24) À0.36 (À1.28, 0.56)
À1.54 (À2.24, À0.83) À3.50 (À5.31, À1.69)
0.15 (À1.19, 1.49) À0.08 (À0.63, 0.47)
À1.04 (À2.55, 0.47) À0.01 (À1.06, 1.04)
À2.8% (À4.6%, 10.7%) 1.22 (À0.25, 2.70)
Multivariable adjusted, excludingdiabetic persons
0.49 (À0.87, 1.84) À0.15 (À0.71, 0.41)
BMI ¼ body mass index; HOMA-IR ¼ homeostasis model assessment of insulin resistance. *Insulin, HOMA-IR, and triglycerides were log-transformed. †Adjusted for age, sex, race/ethnicity, education level, income, marital status, tobacco use, alcohol use, and physical activity.
The American Journal of Medicine, Vol -, No -, - 2013
multivariable adjusted analyses even after excluding
underestimation of drug use would likely yield results
participants with prevalent diabetes mellitus.
biased toward observing no association.
Particular focus has been given to the plant cannabinoid
It is possible that the inverse association in fasting insulin
(-)-trans-D9-tetrahydrocannabinol, which acts as a partial
levels and insulin resistance seen among current marijuana
agonist at both the cannabinoid type 1 and 2 receptors, and
users could be in part due to changes in use patterns among
cannabidiol, which has lower affinity for the cannabinoid
those with a diagnosis of diabetes (ie, those with diabetes
receptors but appears to antagonize both cannabinoid type 1
may have been told to cease smoking). However, in the
and In addition, it has been found that repeated
sensitivity analysis excluding those subjects with a diag-
administration of cannabinoids reduces cannabinoid type 1
nosis of diabetes mellitus, associations between marijuana
receptor density, producing a tolerance to its physiologic
use and insulin levels, HOMA-IR, waist circumference, and
Thus, a dose-response relationship may be ex-
HDL-C were similar and remained statistically significant.
pected; however, we did not find any evidence of this in thepresent study.
Although not completely elucidated, the mechanisms by
which cannabinoids affect peripheral metabolism via these
With the recent trends in legalization of marijuana in the
receptors have been studied extensively; the cannabinoid
United States, it is likely that physicians will increasingly
type 1 receptor antagonist, rimonabant, was found to
encounter patients who use marijuana and should there-
improve insulin sensitivity in wild-type mice, but not in
fore be aware of the effects it can have on common
adiponectin knockout mice, suggesting that adiponectin at
disease processes, such as diabetes mellitus. We found
least partially mediates the improvement in insulin sensi-
that current marijuana use is associated with lower levels
tivityadiponectin has been reported to improve insulin
of fasting insulin, lower HOMA-IR, and smaller waist
sensitivity.This rimonabant-induced improvement in
insulin resistance has been confirmed in human studies.Furthermore, in a randomized clinical trial, rimonabant was
significantly associated with an increase in plasma adipo-
1. Results from the 2010 National Survey on Drug Use and Health:
nectin levels, as well as weight loss and a reduction in waist
Volume I. Summary of National Findings. Rockville, MD: Substance
circumference.Cannabis itself, when administered to
Abuse and Mental Health Services Administration; 2011.
obese rats, was associated with weight reduction and an
2. State Medical Marijuana Laws. National Conference of State Legisla-
increase in the weight of pancreata, implying beta-cell
tures. Available at: . Accessed February 7, 2013.
protection.In addition, cannabinoid type 1 knockout mice
3. Foltin RW, Fischman MW, Byrne MF. Effects of smoked marijuana on
are resistant to diet-induced obesity, suggesting that the role
food intake and body weight of humans living in a residential labo-
of this receptor is central in the metabolic processes leading
to obesity.Given that 2 of the main active phytocanna-
4. Smit E, Crespo CJ. Dietary intake and nutritional status of US adult
binoids in marijuana, (-)-trans-D9-tetrahydrocannabinol and
marijuana users: results from the Third National Health and NutritionExamination Survey. Public Health Nutr. 2001;4:781-786.
cannabidiol, are classified as partial agonists and antago-
5. Rodondi N, Pletcher MJ, Liu K, Hulley SB, Sidney S. Marijuana use,
nists, respectively, and are thus capable of producing
diet, body mass index, and cardiovascular risk factors (from the
antagonistic effects at the cannabinoid receptors, it is
CARDIA study). Am J Cardiol. 2006;98:478-484.
possible that the associations observed in the aforemen-
6. Le Strat Y, Le Foll B. Obesity and cannabis use: results from 2
tioned studies, as well as in the present study, are due at
representative national surveys. Am J Epidemiol. 2011;174:929-933.
7. Rajavashisth TB, Shaheen M, Norris KC, et al. Decreased prevalence
least in part to this adiponectin-mediated mechanism.
of diabetes in marijuana users: cross-sectional data from the National
In our analyses, we presented alternative models,
Health and Nutrition Examination Survey (NHANES) III. BMJ Open.
controlling for BMI as a potential confounder of the rela-
tionship between marijuana use and the remainder of
8. National Health and Nutrition Examination Survey, Centers for Disease
the cardiometabolic parameters. We generated this model
Control and Prevention. Available at: Accessed September 10, 2012.
because of the potential for BMI to affect marijuana use and
9. Anthropometry Procedures Manual. National Health and Nutrition
independently affect the cardiometabolic parameters. On
Examination Survey. Atlanta, GA: Centers for Disease Control and
the other hand, BMI may be a mediator of the association
between marijuana use and the cardiometabolic outcomes,
10. Schafer JL. Analysis of Incomplete Multivariate Data, Vol. 72.
and thus was excluded from our primary multivariable
London, UK: Chapman & Hall/CRC; 1997.
11. Weiss L, Zeira M, Reich S, et al. Cannabidiol lowers incidence of
diabetes in non-obese diabetic mice. Autoimmunity. 2006;39:143-151.
12. Pertwee RG. The diverse CB1 and CB2 receptor pharmacology of
three plant cannabinoids: delta9-tetrahydrocannabinol, cannabidiol and
delta9-tetrahydrocannabivarin. Br J Pharmacol. 2008;153:199-215.
This was a cross-sectional study with all of the inherent
13. Petitet F, Jeantaud B, Reibaud M, Imperato A, Dubroeucq M-C.
Complex pharmacology of natural cannabinoids: evidence for partial
limitations of that study design. In addition, data on mari-
agonist activity of D9-tetrahydrocannabinol and antagonist activity of
juana use were self-reported and may be subject to under-
cannabidiol on rat brain cannabinoid receptors. Life Sci. 1998;63:
estimation or denial of illicit drug use.However,
14. Hirvonen J, Goodwin RS, Li CT, et al. Reversible and regionally
18. Després J-P, Golay A, Sjöström L. Effects of rimonabant on metabolic
selective downregulation of brain cannabinoid CB1 receptors in
risk factors in overweight patients with dyslipidemia. N Engl J Med.
chronic daily cannabis smokers. Mol Psychiatry. 2012;17:642-649.
15. Migrenne S, Lacombe A, Lefèvre A-L, et al. Adiponectin is required to
19. Levendal R, Schumann D, Donath M, Frost C. Cannabis exposure
mediate rimonabant-induced improvement of insulin sensitivity but not
associated with weight reduction and b-cell protection in an obese rat
body weight loss in diet-induced obese mice. Am J Physiol Regul
model. Phytomedicine. 2012;19:575-582.
Integr Comp Physiol. 2009;296:R929-R935.
20. Ravinet Trillou C, Delgorge C, Menet C, Arnone M, Soubrié P. CB1
16. Sowers JR. Endocrine functions of adipose tissue: focus on adipo-
cannabinoid receptor knockout in mice leads to leanness, resistance to
nectin. Clin Cornerstone. 2008;9:32-38.
diet-induced obesity and enhanced leptin sensitivity. Int J Obes (Lond).
17. Wierzbicki A, Pendleton S, McMahon Z, et al. Rimonabant improves
cholesterol, insulin resistance and markers of non-alcoholic fatty liver
21. Harrison ER, Haaga J, Richards T. Self-reported drug use data:
in morbidly obese patients: a retrospective cohort study. Int J Clin
what do they reveal? Am J Drug Alcohol Abuse. 1993;19:
Univ.-Prof. Doz. (ETH) Dr. Anton Amann “FEM_PERS bietet Schülerinnen und Schülern die phantastische Chance, mit WissenschafterInnen der Österreichischen Akademie der Wissenschaften aktiv an einem medizinischen Forschungsprojekt mitzuarbeiten. Somit werden durch Sparkling Science die Karrierewege in naturwissenschaftliche und medizinische Studien geebnet.” FEM_PERS Entwicklun
Stomachache Relief Abdominal pain has many causes. Only rarely is the cause serious. The general information section of Pediatric Planet has more information about when to be concerned. If your child’s physician has determined that there is not a serious cause of the abdominal pain there are a number of over-the-counter medications that may be helpful. The key to getting benefit is