ARTICLES nature publishing group

Vaporization as a Smokeless Cannabis Delivery

System: A Pilot Study
DI Abrams1,2,3, HP Vizoso1,3, SB Shade1,3, C Jay4,5, ME Kelly1,2,3 and NL Benowitz3,6

Although cannabis may have potential therapeutic value, inhalation of a combustion product is an undesirable delivery
system. The aim of the study was to investigate vaporization using the Volcanos device as an alternative means
of delivery of inhaled Cannabis sativa. Eighteen healthy inpatient subjects enrolled to compare the delivery of
cannabinoids by vaporization to marijuana smoked in a standard cigarette. One strength (1.7, 3.4, or 6.8%
tetrahydrocannabinol (THC)) and delivery system was randomly assigned for each of the 6 study days. Plasma
concentrations of D-9-THC, expired carbon monoxide (CO), physiologic and neuropsychologic effects were the main
outcome measures. Peak plasma concentrations and 6-h area under the plasma concentration–time curve of THC
were similar. CO levels were reduced with vaporization. No adverse events occurred. Vaporization of cannabis is a
safe and effective mode of delivery of THC. Further trials of clinical effectiveness of cannabis could utilize vaporization
as a smokeless delivery system.

The Institute of Medicine (10 m) report on Marijuana as Cannabis vaporization is a technology for delivering
Medicine published in 1999 concluded that ‘‘scientific data inhaled tetrahydrocannabinol (THC) and other cannabinoids
indicate the potential therapeutic value of cannabinoid drugs, while reducing toxic byproducts of smoked cannabis primarily
primarily THC, for pain relief, control of nausea and caused by combustion.2,3 By heating cannabis to a tempera-
vomiting, appetite stimulation; smoked marijuana, however ture between 180 and 2001C, it is possible to vaporize the
is a crude THC delivery system that also delivers harmful cannabinoids that reside on the trichomes on the surface of
substances’’.1 The report recommended that clinical trials of cannabis flowers and leaves, while avoiding combustion
cannabinoid drugs for symptom management should be (which occurs at 2301C and above) and attendant smoke
conducted with the goal of developing rapid onset, reliable, toxins. Vaporization is a relatively new technology. Various
and safe delivery systems. While acknowledging therapeutic vaporizer designs are currently under development. The
potential, the IOM report stressed that cannabis is not a feasibility of vaporization of THC has been demonstrated in
completely benign substance, but a powerful drug with a a series of laboratory studies involving different vaporizer
variety of effects, but ‘‘except for the harms associated with designs.2 An electric vaporizer was shown to release substantial
smoking, the adverse effects are within the range of those amounts of the THC while producing no measurable amounts
tolerated for other medications.’’ The report comments that of the benzene, toluene, and naphthalene, which are generated
‘‘because of the health risks associated with smoking, smoked when marijuana is smoked. Reductions in carbon monoxide
cannabis should generally not be recommended for long- (CO) and tar generation were also observed under vaporiza-
term medical use. Nonetheless, for certain patients, such as tion compared to smoking. Although no measurements were
the terminally ill or those with debilitating symptoms, the made of other smoke toxins, it is quite possible that the
long-term risks are not of great concern.’’ The Institute of vaporizer eliminated or substantially reduced the polycyclic
Medicine sends a clear message suggesting that smoking is aromatic hydrocarbons and other combustion-generated
not a desirable delivery system for the potential therapeutic toxins commonly found in cannabis smoke, as they form at
effects of cannabis. the higher temperatures of pyrolysis.

1
Community Consortium, Positive Health Program, San Francisco General Hospital, San Francisco, California, USA; 2Division of Hematology-Oncology,
San Francisco General Hospital, San Francisco, California, USA; 3Department of Medicine, University of California, San Francisco, California, USA; 4Division of Neurology,
San Francisco General Hospital, San Francisco, California, USA; 5Department of Neurology, University of California, San Francisco, California, USA; 6Division of Clinical
Pharmacology and Experimental Therapeutics, University of California, San Francisco, California, USA. Correspondence: DI Abrams ([email protected])
Received 29 September 2006; accepted 25 February 2007; published online 11 April 2007. doi:10.1038/sj.clpt.6100200

572 VOLUME 82 NUMBER 5 | NOVEMBER 2007 | www.nature.com/cpt

 ARTICLES

A recent evaluation of the Volcanos vaporizer device used

herbal cannabis or pure cannabinoid ethanolic solution
preparations to test the efficacy and reproducibility of THC
delivery into the balloon receptacle.4 Cannabinoids were
measured in the THC-containing materials before and after
vaporization, and in the vapor that was generated by the
device and collected within the balloon. The results validated
the Volcanos vaporizer as an efficient and reproducible
mode of delivery of D-9-THC. On average, 54% of the
applied dose of THC was recovered in the balloon receptacle.
This study investigated vaporization using the Volcanos
device compared to smoked cannabis. This is the first
pharmacokinetic and pharmacodynamic evaluation conducted
in humans to determine whether the Volcanos may be an
appropriate system for use in clinical effectiveness studies.

RESULTS
Baseline characteristics of study subjects
A total of 68 patients were screened for eligibility between
August 2004 and May 2005. Of these, 47 were not enrolled
(33 patients were unavailable to commit to a 6-day
hospitalization, 10 patients were excluded as a result of their
medical history or concurrent illness, and four patients were
excluded because of active substance abuse). Twenty-one
patients were randomly assigned; however, three patients did
not complete the intervention of the study phase (one patient
for non-adherence to the General Clinical Research Center
(GCRC) rules of comportment, one patient for acute
influenza, and one patient withdrew consent), leaving 18
total patients for analysis.
Participants were predominately men (83%), Caucasian
(72%), with some college education (94%). All of the partici-
pants were active marijuana users (median 5–6, range 3–10
marijuana cigarettes in the past 30 days). None had used the
Volcanos device, although one participant had previously
experienced vaporized marijuana using a similar device.

Primary outcome measure

The mean and 95% confidence intervals (CIs) for the plasma
concentrations of THC at each time point for each strength Figure 1 Plasma THC using vaporizer and smoked cannabis by THC
of THC using both vaporization and smoking are presented strength (mean and 90% CI).
in Figure 1. The vaporizer resulted in higher plasma
concentrations of THC compared to smoked marijuana at statistically significant (ratio: 0.87; 95% CI: 0.84, 0.90;
30 and 60 min at each strength (Table 1). The two modalities Po0.001 per 1% increase in THC strength) and did not
were not significantly different from one another at any of the appear to differ between vaporization and smoking (ratio for
three strengths in the 6-h area under the plasma THC interaction: 0.92; 95% CI: 0.79, 1.05; P ¼ 0.25) in a mixed
concentration–time curve (AUC), or for the peak THC model which included fixed effects for randomization, a
plasma concentrations measured at 2 min. linear term for THC strength, and a term for the interaction
There was evidence of decreasing bioavailability and/or between these effects.
titration of THC intake with increasing strength of THC. The There was also evidence of titration of intake of THC with
plasma THC AUC derived from the vaporizer normalized for increasing THC strength based on puffing behavior. The
the THC strength was highest at 1.7% THC (27.1 ng h/ml/%) number of puffs taken using smoked marijuana remained
and was progressively lower at higher THC strengths (3.4% stable with increasing strength THC (mean puffs, 95% CI: 6.1
THC: 20.5 ng h/ml/% and 6.8% THC: 14.3 ng h/ml/%; (4.8, 7.3), 5.9 (4.9, 6.8), and 6.4 (5.3, 7.6) for 1.7, 3.4, and
Table 1), suggesting higher bioavailability and/or more 6.8% THC, respectively; mixed model analysis ratio: 1.01;
intensive puffing at lower THC potency. This decline was 95% CI: 0.96, 1.05; P ¼ 0.81). The number of puffs taken

CLINICAL PHARMACOLOGY & THERAPEUTICS | VOLUME 82 NUMBER 5 | NOVEMBER 2007 573

ARTICLES

Table 1 THC pharmacokinetics for vaporized cannabis and ratio of vaporized vs smoked cannabisa,b
Vaporizer Vaporizer/smoked ratio

THC, % outcome measure Mean 95% CI Minimum Maximum Odds ratio 95% CI* P-value
1.7%
AUC0–6 46.00 34.89, 57.11 15.59 98.08 1.26 0.94, 1.68 0.12
Cmax (=C2) 68.95 46.99, 90.91 6.00 186.20 1.01 0.65, 1.58 0.97
C30 18.94 10.57, 27.32 4.90 79.90 1.95 1.37, 2.80 0.001
C60 7.56 6.02, 9.50 3.70 16.50 1.56 1.26, 1.93 0.001
C180 3.05 1.99. 4.00 0.10 9.40 1.31 0.83, 2.06 0.25
C360 1.87 0.97, 2.77 0.20 8.20 1.17 0.82, 1.66 0.38
Puffs 10.06 8.81, 11.30 7.00 17.00 1.71 1.47, 2.00 0.001
AUC/THC % 27.06 20.52, 33.60 9.17 57.69 1.26 0.94, 1.68 0.12

3.4%
AUC0–6 69.76 52.91, 86.62 22.30 140.44 0.99 0.81, 1.21 0.95
Cmax (=C2) 112.45 84.55, 140.65 36.70 201.10 1.07 0.64, 1.80 0.80
C30 23.04 17.74, 28.35 28.35 43.20 1.50 1.29, 1.73 0.001
C60 12.58 9.46, 15.70 3.30 24.20 1.41 1.11, 1.79 0.006
C180 4.14 3.05, 5.24 1.40 10.10 1.24 1.06, 1.46 0.008
C360 2.94 1.55, 4.34 0.60 12.90 1.34 1.03, 1.75 0.03
Puffs 9.17 8.23, 10.10 4.00 13.00 1.58 1.36, 1.84 0.001
AUC/THC % 20.52 15.56, 25.48 6.56 41.31 0.99 0.81, 1.21 0.95

6.8%
AUC0-6 96.79 67.51, 126.06 18.98 278.20 1.22 0.98, 1.54 0.08
Cmax (=C2) 187.12 100.65, 273.59 22.50 813.20 1.19 0.86, 1.65 0.30
C30 28.80 22.19, 35.41 9.20 50.00 1.45 1.16, 1.82 0.001
C60 15.99 12.41, 19.58 4.60 29.40 1.38 1.13, 1.69 0.002
C180 4.81 3.65, 5.96 1.10 9.20 1.15 0.88, 1.52 0.31
C360 2.99 0.79, 5.20 0 19.50 0.88 0.53, 1.45 0.62
Puffs 8.55 7.72, 9.40 5.00 11.00 1.43 1.11, 1.85 0.006
AUC/THC % 14.23 9.93, 18.54 2.79 40.91 1.22 0.98, 1.54 0.08
AUC, area under the curve; CI, confidence interval; THC, tetrahydrocannabinol. aAUCs in ng h/ml; Cmax values in ng/ml. bAnalysis conducted using mixed models to adjust for
day of observation.

using vaporized marijuana tended to decrease with increasing (Po0.001) at each THC strength. The increase in CO (AUC
strength of THC, but the trend was not significant (mean for CO) decreased during smoking (P ¼ 0.003 for trend), but
puffs, 95% CI: 10.1 (8.8, 11.3), 9.2 (8.2, 10.1), and 8.6 (7.7, not vaporization (P ¼ 0.25) with increasing THC strength. The
9.4) for 1.7, 3.4, and 6.8% THC, respectively; mixed model expired CO AUC per puff is an indicator of how much smoke
ratio: 0.97; 0.92, 1.01; P ¼ 0.17). is inhaled per puff for the smoked marijuana. The CO AUC
per puff decreased progressively (1.7% THC: [mean, 95% CI]:
Secondary outcome measures 2.8 (2.2, 3.3); 3.4% THC: 2.1 (1.1, 3.0); 6.8% THC: 1.2 (0.6,
The levels of exhaled CO increased very little after vaporiza- 1.9); Po0.001 for trend), consistent with taking smaller puffs
tion; mean ¼ 1.9 p.p.m.; 95% CI: 4.4, 0.6 for 1.7% THC; with increasing THC content in the marijuana.
mean ¼ 1.8 p.p.m.; 95% CI: 3.7, 0.7 for 3.4% THC; and
mean ¼ 0.5 p.p.m.; 95% CI: 1.9, 0.9 for 6.8% THC), Subjective and safety observations
whereas there was a substantial increase after smoking Self-reported high did not differ during vaporization com-
marijuana (mean ¼ 15.5 p.p.m.; 95% CI: 11.0, 20.1 for 1.7% pared to smoking overall (6-h AUC) or at any observation
THC; mean ¼ 11.9 p.p.m.; 95% CI: 6.8, 17.1 for 3.4% THC; after consumption of cannabis (Figure 3). Self-reported high
mean ¼ 7.0 p.p.m.; 95% CI: 4.0, 10.0 for 6.8% THC) did increase significantly during both vaporization and
(Figure 2). This difference was statistically significant smoking with increasing strength of THC (Po0.001).

574 VOLUME 82 NUMBER 5 | NOVEMBER 2007 | www.nature.com/cpt

 ARTICLES

Figure 3 Self-reported ‘‘high’’ at each time point for each mode of

administration and THC concentration (mean and 95% CI).

compared to smoking marijuana cigarettes. We found that

THC levels were generally similar over 6 h for the two types of
Figure 2 Expired CO at each time point for each mode of administration
delivery. The vaporizer was associated with higher plasma
and THC strength (mean and 95% CI).
THC concentrations at 30 min and 1 h compared to smoking
at each THC strength, suggesting that absorption was faster
Although blinded with regard to dose, eight participants with the vaporizer.
selected the day they received 3.4% THC (seven vaporized, Bioequivalence criteria developed for drugs require that
one smoked) as their most preferred treatment day; four the CIs for the ratios of AUC for the test and reference
participants selected the day they received 6.8% THC via products be between 80 and 125% to be judged bioequiva-
vaporization, and six participants had no treatment day lent.5 Using these criteria, we were not able to establish the
preference. Overall, vaporization was the preferred method of bioequivalence of vaporization and smoking of marijuana. A
administration by 14 participants, smoking was preferred by much larger study would be needed to establish bioequiva-
two, and two reported no preference. During the course of lence in this setting.
the study, no adverse events were reported. Of interest was that the systemic dose of THC, as
estimated by the plasma AUC, normalized for the THC
DISCUSSION content of the cannabis, varied with THC strength. The dose
Our study provides novel data on the absorption of THC of THC normalized for concentration of THC in the cannabis
from marijuana inhaled via the Volcanos vaporizer system was greater at lower compared to higher THC strengths, both

CLINICAL PHARMACOLOGY & THERAPEUTICS | VOLUME 82 NUMBER 5 | NOVEMBER 2007 575

ARTICLES

for vaporized and smoked cannabis. This observation Study medication. The National Institute on Drug Abuse provided
suggests either dose-dependant bioavailability or self-titra- pre-rolled cannabis cigarettes, weighing on average 0.9 g and
containing 1.7, 3.4, and 6.8% D-9-THC, respectively. The cigarettes
tion of THC intake. Self-titration of drug intake means that were kept in a locked and alarmed freezer until they were dispensed
smokers adapt their smoking behavior to obtain desired to a locked freezer in the San Francisco General Hospital General
levels of THC from the particular delivery system, taking Clinical Research Center where the in-patient study was conducted.
more puffs and/or inhaling more efficiently at lower The cigarettes were bisected; one half to be smoked and the contents
compared to higher THC strengths. Supporting the idea of of the other half to be vaporized. The half cigarettes were rehydrated
in a humidifier overnight before their use. Patients were housed in a
titration was the trend to take more puffs at lower THC room with a fan ventilating to the outside. Research staff monitored
concentrations of vaporized marijuana and the higher CO patients during smoking sessions, weighed the cannabis cigarettes
per puff at lower THC concentrations of smoked marijuana. immediately before and after they were administered to patients,
The phenomenon of self-titration of psychoactive drug intake and returned all leftover material to the pharmacy. To maximize
from an inhaled delivery system is well documented for standardization of inhaled doses, patients followed the Foltin
uniform puff procedure where inhalation for 5 s is followed by a
nicotine from cigarette smoking,6 but to our knowledge has 10 s breath hold, then exhalation; the entire process is repeated after
not been previously reported for marijuana. 45 s.7 Study participants smoked or vaporized cannabis once a day.
Whereas smoking marijuana increased CO levels as Subjects were instructed to continue puffing until they exhausted
expected for inhalation of a combustion product, there was smoke or vapor from the delivery device or until they had inhaled as
little if any increase in CO after inhalation of THC from the much as they could tolerate.
vaporizer. This indicates little or no exposure to gaseous The vaporizer device. The Volcanos vaporizer was obtained from
combustion toxins. Combustion products are harmful to Storz & Bickel GmbH & Company (Tuttlingen, Germany) and was
health and reflect a major concern about the use of marijuana employed according to the manual provided. The device works as a
cigarettes for medical therapy as expressed by the Institute of vaporizer that evaporates the active substances or aromas from plant
Medicine. Although we did not measure other combustion material by using a hot airflow (Figure 4). Cannabis placed in the
filling chamber is heated by the device to 1901C. The vaporized
products such as polycyclic aromatic hydrocarbons and compounds are collected in the inflatable, detachable bag fitted with
oxidant gases, the observation of little or no CO exposure a mouthpiece and a one-way valve that allows the vapor to remain in
suggests little or no exposure to these other compounds. The the balloon until inhalation. It required two to three balloon
vaporizer was well tolerated, with no reported adverse effects. inflations to vaporize each half cigarette. Subjects also followed the
Most subjects preferred the vaporizer compared to marijuana Foltin puff procedure when inhaling the vaporization product.
smoking, supporting its potential for medical therapy. Thus, Study design and procedures. The study was a 6-day ‘‘proof of
the Volcanos is an acceptable system and may provide a safer concept’’ pilot study to investigate the delivery of cannabinoids by
way to deliver THC than smoking marijuana cigarettes. way of vaporization of cannabis compared to cannabis smoked in a
In summary, we provide data indicating that the standard cigarette. The in-patient setting permitted us to measure
availability of THC delivered by the Volcanos vaporizer is plasma THC concentration over time and to rigorously assess the
primary and secondary outcome variables in a controlled clinical
comparable to that of marijuana cigarettes. Vaporization of environment.
marijuana does not result in exposure to combustion gases,
and therefore is expected to be much safer than smoking Screening visit. Once a subject for the protocol had been identified,
marijuana cigarettes. The vaporizer was well tolerated and details of the study were carefully discussed and the subject was
preferred by most subjects compared to marijuana cigarettes. asked to read and sign a consent form. Subjects were asked questions
about their medical history including psychiatric illness and
The Volcanos device is an effective and apparently safe substance abuse. Subjects were asked to abstain from smoking or
vehicle for THC delivery, and warrants further investigation ingesting cannabis 48 h before their hospitalization based on our
in clinical trials of cannabis for medicinal purposes. prior studies which indicated that after 24 h of abstinence, plasma
THC concentrations are sufficiently low so that the concentration-
time curve could be determined after the experimental exposure.8
METHODS
Study patients. Participants were healthy adults between the ages of GCRC in-patient hospitalization (days 1–6). Subjects inhaled three
21 and 45 years who were current cannabis users and had smoked strengths of cannabis (1.7, 3.4, and 6.8% THC) as smoked cigarettes
cannabis within the past 30 days but in an amount totaling less than and three as vaporized cannabis using the Volcanos device. Half of
10 cannabis cigarettes or the equivalent. Subjects with active substance one cigarette was inhaled via one of the two delivery systems on each
abuse (e.g., recurrent or continuous drug and/or alcohol use) or of the 6 in-patient GCRC days. The uniform puff procedure
diagnosed with marijuana dependence as defined in DSM-IV code no. described above was utilized to attempt to standardize inhalation.
304.30. were excluded. Subjects were required to abstain from Blood was drawn at 2, 30, 60, 180, and 360 min after smoking on
smoking cannabis for 48 h before their admission into the GCRC at each of the 6 inhalation days to measure the concentrations of THC.
San Francisco General Hospital (SFGH). The study was approved by Expired CO was measured using the Ecolyzers before inhalation,
the Institutional Review Board at the University of California San and 2, 30, 60, 180, and 360 min after inhalation.
Francisco, the Research Advisory Panel of California, the Drug Subjects rated the subjective ‘‘high’’ they experienced using a
Enforcement Administration, the Food and Drug Administration, 100 mm visual analog scale anchored by ‘‘none’’ and ‘‘highest ever’’.
and the National Institute on Drug Abuse. Written informed consent On day 5 before discharge, subjects were asked to choose which in-
was obtained from all patients. The trial was monitored by an patient day they preferred. Subjects were asked to rate their
independent Data Safety Monitoring Board (DSMB) established by preferences from 1 to 5 with 1 indicating very satisfied and 5
the University of California Center for Medicinal Cannabis Research. indicating very dissatisfied.

576 VOLUME 82 NUMBER 5 | NOVEMBER 2007 | www.nature.com/cpt

 ARTICLES

Figure 4 Volcanos apparatus.

All adverse events were spontaneously reported by the subject or The primary outcome was the within-person ratio for the 6-h
observed by the study personnel and/or GCRC nursing staff, area under the curve (AUC0–6) for plasma concentration of THC,
documented along with any medical intervention, and evaluated comparing the vaporizer with smoking cannabis cigarettes. AUC0–6
according to standardized criteria in terms of severity, frequency, was computed using the linear trapezoidal method, assuming zero
duration, and relationship to study drug. Adverse events were THC concentration at baseline. This assumption was based on our
graded using the NIH Division of AIDS table for scoring severity of previous research that observed undetectable plasma concentration
adult adverse experiences.9 of THC 8 h after smoking in all subjects.8 For each mode of
administration and THC strength, we plotted the mean and 95% CIs
Randomization. The order of administration of the six combina- of the observed values at each time point. To assess the within-
tions of THC strength and delivery method for the 18 participants person ratio comparing vaporization to smoking, each outcome
was randomized in three 6  6 Latin squares. This ensured balance (AUC0–6, C2, C30, C60, C180, C360, number of puffs, AUC0–6 per THC
in the sense that each of the six combinations occurred exactly three percent, and AUC0–6 per puff) was log transformed for analysis
times on day 1, exactly three times on day 2, and so on. In addition, using mixed effects models. The overall effect of vaporization
the orders were restricted so that the two delivery methods for the compared to smoking for each parameter was assessed by fitting a
same strength always occurred on consecutive days. This was to fixed effect term for randomization (vaporization vs smoking),
prevent patients from developing an early preference for one controlling for strength of THC (indicators for 3.4% THC and 6.8%
delivery method if it was used with a higher strength cigarette than THC cannabis, relative to 1.7% THC cannabis). Each patient was
the other. Randomization was computer-generated, and study drug treated as a random effect. Another model was fit to assess THC
distribution was managed by a research pharmacist. Subjects and strength-specific effects of vaporization compared to smoking. This
study personnel were blinded to the THC strength. model included fitting additional fixed effects for the use of the
vaporizer at each strength of THC (vaporization at 1.7% THC,
Statistical analysis. The 18-patient target sample size was based on vaporization at 3.4% THC, and vaporization at 6.8% THC).
a standardized effect size to calculate sample size and power for the We also assessed the potential presence of order effects due to the
study. With a sample of 18 subjects, we had an 80% power to detect study day of observation, as well as potential practice effects due to
a true standardized effect size (E/S) of 0.70, using an a of 0.05, where additional experience using the vaporizer. To assess the presence of
E is the effect size and S is the standard deviation of the paired order effects, additional variables were added to both the overall and
differences.10,11 This calculation assumes use of a paired t-test using strength-specific models to assess whether day of observation
data at a single concentration of THC. impacted the outcomes, as well as whether there was a difference

CLINICAL PHARMACOLOGY & THERAPEUTICS | VOLUME 82 NUMBER 5 | NOVEMBER 2007 577

ARTICLES

in measurements taken on the first day of the study compared to Center for Medicinal Cannabis Research for her invaluable assistance with
other study days. In these models, day of observation was treated regulatory affairs, data quality management, and interaction with the Data
as a linear variable with and without an additional indicator Safety Monitoring Board. This work was supported by the University
variable for the first study day. Similarly, to assess the presence of of California Center for Medicinal Cannabis Research and NIH Grant
practice effects, additional variables were added to both the overall 5-MO1-RR00083.
and strength-specific models to assess whether previous use of
the vaporizer impacted the outcomes. These models included either
a linear variable for how many days the participant had used CONFLICT OF INTEREST
the vaporizer or separate indicator variables for each day of The authors declared no conflict of interest.
vaporizer use.
To explore possible evidence of titration of THC intake and dose- & 2007 American Society for Clinical Pharmacology and Therapeutics
dependent changes in bioavailability, we created additional mixed 1. Joy, J.E., Watson, S.J., Benson, J.A., (eds). & Institute of Medicine.
models for number of puffs and AUC0–6 per THC percent, which Marijuana and medicine: Assessing the science base (National Academy
included fixed effects, as above, for randomization (vaporization vs Press, Washington, 1999).
smoking), as well as linear terms for strength of THC, and the 2. Gieringer, D. Cannabis vaporization: a promising strategy for smoke
interaction between randomization and strength of THC. As above, harm reduction. J. Cannabis Ther. 1, 3–4 (2001).
these models included a random effect for each patient. These 3. Russo, E. An interview with Markus Storz: June 19, 2002. J. Cannabis
models assess not only whether the ratio of the number of puffs or Ther. 3, 67–78 (2003).
the AUC per THC percent differs during vaporization and smoking 4. Hazekamp, A., Ruhaak, R., Zuurman, L., van Gerven, J. & Verpoorte, R.
Evaluation of a vaporizing device (Volcanos) for the pulmonary
but also whether the ratio increases or decreases with increasing
administration of tetrahydrocannabinol. J. Pharm. Sci. 95, 1308–1317
strength of cannabis, and whether this increase or decrease differs (2006).
during vaporization compared to smoking. 5. Center for Drug Evaluation and Research. Guidance for industry:
We compared the observed values for expired CO and self- bioavailability and bioequivalence studies for orally administered
reported high using similar methods. We plotted the mean and drug products—general considerations. Revision 1, March 2003.
95% CIs of response measures at each time point for each mode of Available at www.fda.gov/cder/guidance/index.htm (2006). Accessed
administration and THC strength. We also fit mixed models for the 29 August 2006.
6-h AUC for expired CO and self-reported high, as described above, 6. Benowitz, N.L. Compensatory smoking of low yield cigarettes. In: Risks
to compare within-person effects using vaporization and smoking. Associated with Smoking Low Machine-Measured Yields of Tar and
Nicotine (eds. Shopland, DR., Burns, DM., Benowitz, NL. & Amacher,
For 6-h AUC for CO, we fit models for the within-person arithmetic
RH.). NCI Smoking and Tobacco Control Monograph No. 13. 39–64
difference in effects, because we were unable to fit models for the (NIH, National Cancer Institute, Bethesda, MD, 2001). NIH Publication
ratio of effects for 6-h AUC for CO due to the presence of many no. 02-5074.
negative values (and therefore non-valid log transformation of these 7. Foltin, R., Fischman, M. & Byrne, M. Effects of smoked marijuana on
values) during vaporization. For 6-h AUC for self-reported high, we food intake and body weight of humans living in a residential
fit models for the within-person ratios in effects, as above. laboratory. Appetite 25, 577–582 (1988).
All analyses were conducted using SAS 8.2. 8. Abrams, D.I. et al. Short-term effects of cannabinoids in patients with
HIV-1 infection. Ann. Int. Med. 139, 258–266 (2003).
9. NIH Division of AIDS Table for Grading Severity of Adult Adverse
ACKNOWLEDGMENTS
Experiences, August 1992. Available at http://rcc.tech-res-intl.com/
We are grateful to our study participants; the General Clinical Research tox_tables.htm. Accessed 28 August 2006.
Center nursing staff for their meticulous adherence to protocol; Sheila 10. Dupont, W.D. & Plummer, W.D. Power and sample size calculations: a
Huang, PharmD; Peter Bacchetti, PhD, at the UCSF Department of review and computer program. Control. Clin. Trials 11, 116–128 (1990).
Biostatistics and Epidemiology for his assistance in creating the 11. Hulley, S.B. & Cummings, S.R. Designing Clinical Research (Williams
randomization scheme; and Heather Bentley at the University of California and Wilkins, Baltimore, 1988).

578 VOLUME 82 NUMBER 5 | NOVEMBER 2007 | www.nature.com/cpt