The Association between Cannabis Product Characteristics and Symptom Relief

Abstract

Federal barriers and logistical challenges have hindered measurement of the real time effects from the types of cannabis products used medically by millions of patients in vivo. Between 06/06/2016 and 03/05/2018, 3,341 people completed 19,910 self- administrated cannabis sessions using the mobile device software, ReleafApp to record: type of cannabis product (dried whole natural Cannabis flower, concentrate, edible, tincture, topical), combustion method (joint, pipe, vaporization), Cannabis subspecies (C. indica and C. sativa), and major cannabinoid contents (tetrahydrocannabinol, THC; and cannabidiol, CBD), along with real-time ratings of health symptom severity levels, prior-to and immediately following administration, and reported side effects. A fixed effects panel regression approach was used to model the within-user effects of different product characteristics. Patients showed an average symptom improvement of 3.5 (SD = 2.6) on an 11-point scale across the 27 measured symptom categories. Dried flower was the most commonly used product and generally associated with greater symptom relief than other types of products. Across product characteristics, only higher THC levels were independently associated with greater symptom relief and prevalence of positive and negative side effects. In contrast, CBD potency levels were generally not associated with significant symptom changes or experienced side effects.

Full text

1
SCIENTIFIC REPORTS | (2019) 9:2712 | https://doi.org/10.1038/s41598-019-39462-1
www.nature.com/scientificreports
The Association between Cannabis
Product Characteristics and
Symptom Relief
Sarah S. Stith, Jacob M. Vigil, Franco Brockelman, Keenan Keeling & Branden Hall
Federal barriers and logistical challenges have hindered measurement of the real time eects from
the types of cannabis products used medically by millions of patients in vivo. Between 06/06/2016 and
03/05/2018, 3,341 people completed 19,910 self- administrated cannabis sessions using the mobile
device software, ReleafApp to record: type of cannabis product (dried whole natural Cannabis ower,
concentrate, edible, tincture, topical), combustion method (joint, pipe, vaporization), Cannabis
subspecies (C. indica and C. sativa), and major cannabinoid contents (tetrahydrocannabinol, THC;
and cannabidiol, CBD), along with real-time ratings of health symptom severity levels, prior-to and
immediately following administration, and reported side eects. A xed eects panel regression
approach was used to model the within-user eects of dierent product characteristics. Patients
showed an average symptom improvement of 3.5 (SD = 2.6) on an 11-point scale across the 27
measured symptom categories. Dried ower was the most commonly used product and generally
associated with greater symptom relief than other types of products. Across product characteristics,
only higher THC levels were independently associated with greater symptom relief and prevalence of
positive and negative side eects. In contrast, CBD potency levels were generally not associated with
signicant symptom changes or experienced side eects.
Medical cannabis markets are currently being ooded with thousands of cannabis strains with unique cannabi-
noid proles1, novel, uninvestigated cannabis-derived formulates and products with little to no clinical references
or formal guidance on how fundamental characteristics of the products themselves may aect pharmacodynam-
ics2,3. Federal laws have all but prohibited the use of prospective, pragmatic, naturalistic studies with random
treatment assignment for measuring the eects of cannabis consumed in vivo. What little clinical research does
exist is mostly limited to randomized controlled trials (RCTs) using synthetic cannabinoids or low quality and
potency cannabis obtained from the federal government that is unrepresentative of the cannabis products used by
millions of people every day4,5. Contributing to further confusion are historically contradictory messages coming
from the scientic community on the true risks and benets of cannabis consumption. For example, whereas
cannabis was once oen and sometimes still is described as component cause of schizophrenia6,7, several studies
now suggest the use of medical cannabis as an eective alternative therapy to antipsychotics and for treating
schizophrenia more generally8–11. Contradictory eects are oen attributed to the distinction between what has
been historically interpreted as cannabis’ harmful, psychoactive cannabinoid, tetrahydrocannabinol (THC), oen
described as providing the ‘high’ eects versus the therapeutic, non-psychoactive potential (sometimes described
as a ‘miracle cure’ in the popular media) of cannabidiol (CBD)12. In actuality, few large-scale investigations to
date have measured the relative eects of THC and CBD consumption in real-time under naturalistic conditions
among people diagnosed with schizophrenia or any other user group.
is is the rst study to measure how fundamental characteristics of cannabis products consumed in vivo aect
immediate symptom relief and experienced side eects. We operationalize our research question using a mobile
device soware application (app). Although hundreds of cannabis-themed soware apps are available for public
use13, the ReleafApp educational soware14 is the rst app designed specically to record how the route of admin-
istration, combustion method, cannabis subspecies, and major cannabinoid contents are associated with real-time
measurements of symptom severity levels, prior to and immediately following administration of cannabis, and the
manifestation of myriad possible side eects. Despite recent advocacy for the benets of CBD over THC, the vast
majority of observational studies showing an association between patient-managed cannabis use and improvements
University of New Mexico, The Department of Psychology, Albuquerque, USA. Correspondence and requests for
materials should be addressed to J.M.V. (email: vigilj@unm.edu)
Received: 4 May 2018
Accepted: 22 January 2019
Published: xx xx xxxx
OPEN
Content courtesy of Springer Nature, terms of use apply. Rights reserved

2
SCIENTIFIC REPORTS | (2019) 9:2712 | https://doi.org/10.1038/s41598-019-39462-1
www.nature.com/scientificreports
www.nature.com/scientificreports/
in symptoms related to, for example, chronic pain15, multiple sclerosis and Parkinson’s disease16, post-traumatic
stress disorder17, and schizophrenia18 relied on public or commercially available cannabis that has been hybridized
for high THC and low CBD contents, thus suggesting that THC may be an important determinant of user outcomes.
Findings from this study are expected to contribute to guidelines for safe and eective cannabis consumption19,20,
which until now have been limited to anecdotal or retrospective reporting and ungeneralizable experiments.
Methods
Study Design. Institutional Review Board approval was obtained from the University of New Mexico for this
study and methods were performed in accordance with the approved guidelines. e preexisting anonymized data
were obtained with user informed consent through the owner of the ReleafApp, MoreBetter, Ltd., and subject to an
investigator condentiality agreement. e ReleafApp patient education and cannabis treatment management tool was
designed to track patient sessions and real-time cannabis use experiences in order to optimize the therapeutic eects
of consuming cannabis, while minimizing negative side eects. ReleafApp users voluntarily download the application
and enter information on the product they intend to consume, including type of product (whole natural dried ower,
concentrate, edible, tincture, and topical); when applicable, combustion method (joint, dry or water pipe, and vape);
plant subspecies (C. indica, C. sativa, or hybrid); and THC and CBD potency levels (percentage of total weight)21,22.
Testing of the potencies of both cannabinoids is almost universally required under U.S. medical marijuana laws and
generally reported on product labels. THC and CBD levels were capped at 35% for ower due to biological limitations
on how much THC and CBD a plant can contain. Prior to beginning a session, the patient is required to enter a neg-
ative health symptom, selected from 27 possible symptom categories, for which they are attempting to use cannabis
therapeutically. (A list of symptom categories and frequencies is available in Supplementary TableS1.)
Aer entering a symptom, patients are prompted to record a starting symptom level on a visual analogue scale
from 0 (no detectable symptom level) to10 (severe). e patient then taps the prompt on the screen to begin the
session. From that time until the patient closes the session, they can enter multiple symptom severity levels as
frequently as desired. For the current analyses, we include in our sample only patients entering starting symptoms
greater than 0 and recording at least one symptom level within 90 minutes of starting the session; we use the last
symptom level recorded within that timeframe as the ending symptom level. Our nal sample includes 19,910
sessions and 3,341 patients who recorded at least one product characteristic in the ReleafApp between 06/06/2016
and 03/05/2018. Because the entry of product characteristics is voluntary, the sample sizes used in our analyses
vary depending on which product characteristics are included. Whole natural dried Cannabis ower and con-
centrates made from the ower are the most common types of products and most likely to contain information
on the full spectrum of product characteristics. Panels A through E of Table1 show descriptive statistics for the
product characteristics recorded by the ReleafApp; specically, product type (Panel A), ower and concentrate
combustion method (Panel B), subspecies (Panel C), THC potency (Panel D), and CBD potency (Panel E).
Study Outcomes. Our main outcome is changes in symptom severity following cannabis consumption (ending
symptom level minus starting symptom level) as shown in Table1, Panel F. During a session, the patient also can report
side eects, including 12 negative side eects, 19 positive side eects, and 11 context-specic side eects, crowd-sourced
from users, dispensaries, beta testers, and app developers (Supplementary TableS2). Patients can select as many side
eects as they like, and at least one side eect was reported in 78% of sessions in our sample. We use as outcome varia-
bles whether a patient reported any side eect by category and the percent of the number of side eects available in each
of our three respective categories that a patient selected (Table1, Panel G). e most commonly reported negative side
eects are Dry Mouth (26%) and feeling Foggy (23%), the most frequent positive side eects are Relaxed (63%) and
Peaceful (54%), and the most common context-specic side eects are feeling High (37%) and irsty (27%).
Statistical Analysis. Our basic statistical model uses a least squares panel regression approach with repeated
observations (sessions) at the patient level to analyze how product characteristics aect symptom relief and side
eects. We include patient-specic xed eects to control for time-invariant user characteristics, in order to
compare how changing a product characteristic aects symptom relief and side eects for a given user rather than
comparing outcomes across users who may dier in many ways, including which products they choose to con-
sume. Because higher starting symptom levels are associated with greater symptom relief (r = 0.60), we control
for the starting symptom level in all of our regressions. Standard errors are clustered at the user level to control
for heteroskedasticity and arbitrary correlation. Because the number of observations varies substantially by prod-
uct category, we run regressions separately for each product characteristic category (product type, combustion
method, subspecies, and THC and CBD levels) as well as regressions with all product characteristics included.
In order to explore whether cannabis product characteristics dier across symptom categories in their associa-
tion with momentary symptom relief and side eect proles, we conduct sub-analyses with samples dened by the
three most frequently reported symptom categories: anxiety (16% of the sample), back pain (8%), and depression
(10%). Lastly, we explore whether side eect proles vary with product characteristics. We regress reports of any,
and the proportion of side eects selected from each side eect category (negative, positive, context-specic) on
our product characteristics using ordinary least squares for consistency across models, although our side eect
outcomes are constrained to {0,1} and [0,1]. All predicted outcomes from our regressions fall between zero and one.
Results
Table2 shows our results for the eects of product characteristics on patient symptom relief. All regressions
control for the starting symptom level, which renders the constant positive. e similar size of the coecients for
the starting symptom level and the constant mean that patients reporting symptom levels of 1 may experience
little symptom relief. However, even with a starting symptom level of just 2, users are predicted to experience
Content courtesy of Springer Nature, terms of use apply. Rights reserved

3
SCIENTIFIC REPORTS | (2019) 9:2712 | https://doi.org/10.1038/s41598-019-39462-1
www.nature.com/scientificreports
www.nature.com/scientificreports/
statistically signicant symptom relief. (e average user reports a starting symptom level of 6, SD = 2.2.) e
number of sessions, R-squared and number of users are also reported and vary across regressions, with each
regression designated by a column.
e rst column showing the eects of consuming the dierent product types on symptom relief relative to
the eects from ower suggests that ower provides more symptom relief than any other type of product. e
second column comparing C. indica and C. sativa plant sub-species to hybrid plants suggests that products from
pure indicas may increase symptom relief while sativa strains may decrease it. Combustion method does not
explain any dierence in symptom relief across products within users, as shown in Column [3]. Column [4] of
Table2 shows that whereas higher THC oers greater symptom relief, higher CBD oers no statistically signi-
cant benet. e last column of Table2 includes all product characteristics. Due to the inclusion of combustion
method, only ower and concentrate product types are included. e eect of THC becomes stronger and is the
only statistically signicant determinant of symptom relief among the product characteristics.
Figure1 shows changes in symptom severity by THC and CBD percentage category for dried natural ower,
the most popular and homogenous type of cannabis product in the sample, aer adjusting for the remainder
of the product characteristics. Regression coecients (controlling for the remaining product characteristics;
see Supplementary TableS3) showed that ower containing the middle (10–19%) and highest (20–35%) THC
potency levels was associated with greater symptom improvement than ower in the lowest THC potency cate-
gory (0–9%). As with the omnibus tests in Table2, variability in CBD levels in ower was not associated with dif-
ferences in symptom improvement. Figure2 shows adjusted changes in symptom severity by the THC percentage
Mean Std.
Dev. Minimum Maximum
Panel A: Product Type (19,910 sessions, 3,341 users)
Concentrate 0.17 0.38 0 1
Edible 0.05 0.21 0 1
Flower 0.74 0.44 0 1
Tincture 0.04 0.20 0 1
Top ica l 0.00 0.06 0 1
Panel B: Subspecies (17,197 sessions, 2,996 users)
Hybrid 0.48 0.50 0 1
C. indica 0.30 0.46 0 1
C. sativa 0.22 0.41 0 1
Panel C: Combustion Method (16,902 sessions, 2,936)
Joint 0.13 0.33 0 1
Pipe 0.43 0.49 0 1
Vap e 0.45 0.50 0 1
Panel D: THC (6,958 sessions, 1,260 users)
% THC 28.3 22.8 0 100
THC < 10% 0.16 0.36 0 1
THC 10–19% 0.29 0.45 0 1
THC 20–34% 0.36 0.48 0 1
THC 35%+0.20 0.40 0 1
Panel E: CBD (5,400 sessions, 1,123 users)
% CBD 11.6 16.0 0 100
CBD < 1% 0.25 0.43 0 1
CBD 1–9% 0.33 0.47 0 1
CBD 10–34% 0.34 0.47 0 1
CBD 35%+0.08 0.28 0 1
Panel F: Outcome and Control Variables (19,910 sessions, 3,341 users)
Symptom Change −3.5 2.6 −10 9
Starting Symptom Level 6.0 2.2 1 10
Ending Symptom Level 2.4 2.2 0 10
Panel G: Side Eects (15,617 sessions, 2,757 users)
Any Negative Side Eect 0.62 0.48 0.00 1.00
% of Negative Side Eects 0.12 0.14 0.00 0.92
Any Positive Side Eect 0.95 0.23 0.00 1.00
% of Positive Side Eects 0.25 0.18 0.00 1.00
Any Context-Specic Side Eect 0.78 0.41 0.00 1.00
% of Context-Specic Side Eects 0.22 0.20 0.00 1.00
Table 1. Descriptive Statistics. Notes: Nineteen positive, twelve negative, and eleven context-specic side eects
were available for selection.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

4
SCIENTIFIC REPORTS | (2019) 9:2712 | https://doi.org/10.1038/s41598-019-39462-1
www.nature.com/scientificreports
www.nature.com/scientificreports/
category of ower run separately for the three most common medical cannabis patient conditions, anxiety, back
pain, and depression. Regression coecients controlling for the remaining product characteristics showed that
only users treating depression showed greater symptom improvement from using ower in the middle and high-
est THC potency categories relative to the least potent ower. In contrast, variability in THC was not associated
with statistically signicant dierences in symptom relief for back pain or anxiety beyond its overall eect on
mean symptom improvement levels (see Supplemental TableS3).
In Table3, we present the relationship between product characteristics and side eect proles. We restrict
our analysis to only concentrates and ower to capture the relationship between dierent combustion methods
and side eects. While these methods do not explain any statistically signicant variation in symptom relief, they
may be associated with dierent side eect proles. We run our full model using any and the percent of each
category of side eects selected by the user. Concentrates showed a weaker association with positive side eects,
but do not appear to dier from ower in their association with negative or context-specic side eect report-
ing. Indica-based products are associated with a greater likelihood of reporting negative side eects and some
evidence of fewer positive and more context-specic side eects, relative to hybrid- and sativa-based products.
Outcome = Symptom Change (Ending - Starting Symptom)
(1) (2) (3) (4) (5)
Panel A: Product Type, omitted category = ower
Concentrate 0.194** 0.080
(0.076) (0.212)
Edible 0.340***
(0.105)
Tincture 0.498***
(0.117)
Top ica l −0.216
(0.355)
Panel B: Subspecies, omitted category = hybrid
C. indica −0.103** −0.057
(0.041) (0.083)
C. sativa 0.096*0.195
(0.055) (0.121)
Panel C: Combustion Method, omitted category = joint
Pipe −0.061 −0.004
(0.084) (0.194)
Vap e 0.051 0.051
(0.092) (0.210)
Panel D: THC and CBD, omitted categories = THC < 10% and CBD < 1%
THC 10–19% −0.215*−0.220*
(0.117) (0.134)
THC 20–34% −0.235** −0.315***
(0.112) (0.121)
THC 35%+−0.252** −0.342**
(0.126) (0.166)
CBD 1–9% −0.089 −0.026
(0.121) (0.126)
CBD 10–34% 0.038 0.079
(0.105) (0.092)
CBD 35%+−0.222 −0.241
(0.209) (0.227)
Starting Symptom Level −0.704*** −0.721*** −0.726*** −0.709*** −0.717***
(0.028) (0.031) (0.032) (0.056) (0.063)
Constant 0.588*** 0.691*** 0.719*** 0.909*** 0.850**
(0.165) (0.184) (0.212) (0.340) (0.403)
Number of sessions 19,910 17,197 16,898 4,439 3,869
R-squared 0.330 0.340 0.338 0.337 0.346
Number of users 3,341 2,996 2,936 900 787
Table 2. Eects of Product Characteristics on Symptom Relief. Notes: Regressions control for individual user
xed eects. Concentrate is relative to Flower, C. indica and C. sativa are relative to Hybrid, THC categories are
relative to THC 0–9%, CBD categories are relative to CBD 0%, and Pipe and Vape are relative to Joint. Standard
errors are clustered at the user level (shown in parentheses). ***p < 0.01, **p < 0.05, *p < 0.1.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

5
SCIENTIFIC REPORTS | (2019) 9:2712 | https://doi.org/10.1038/s41598-019-39462-1
www.nature.com/scientificreports
www.nature.com/scientificreports/
Higher THC generally is associated with increased reporting of all three types of side eects. Just as with symp-
tom relief, CBD appears to have little eect on side eect reporting of any kind. ere is some evidence that
vaping is associated with decreased reporting of negative side eects relative to smoking joints. Lastly, in order to
show that a handful of very frequent ReleafApp users are not driving the results, we replicated our main results
from Table2 including only individuals who completed ten sessions or fewer, and separately, including only the
rst ve sessions for all users. THC again is the greatest predictor of symptom relief among the product charac-
teristics. (See Supplemental TableS4 for detailed results).
Discussion
Given just how common it is for cannabis patients to try dierent types of products and methods of administra-
tion23,24, it is surprising how few previous investigations have examined which fundamental characteristics of the
products consumed in vivo by millions of people daily are associated with real-time patient outcomes and expe-
rienced side eects. While RCTs may be the ‘gold standard’ for measuring the pharmacodynamics of synthetic,
standardized (usually particular symptom focused) medications, they are poorly suited for understanding the
eects of a medication with substantial heterogeneity in product characteristics and consumption methods across
the estimated 2.2 million state-legal medical cannabis patients in the United States25. Our observational study
using mobile app technology was designed to measure these eects in real-time among a large sample of patients
using cannabis for treating their medical symptoms under naturalistic conditions. On average, responders expe-
rienced signicant improvements across the 27 health symptom categories measured. Dried, whole natural ower
was associated with greater symptom relief than the use of other types of products (i.e., concentrates, edibles, tinc-
tures, and topicals). However, and despite the fact that dierent routes of administration deliver variable amounts
of cannabinoid contents and have dierent metabolomics26–31, we did not nd variation in symptom relief with
use of pipes, joints, or vaporization combustion devices. Products made from pure C. indica strains were more
eective than products made from C. sativa, matching patient-reported preferences for the former for treating
conditions such as pain and insomnia32,33. However, once we controlled for cannabinoid contents, none of the
other product characteristics predicted variability in symptom levels. Only THC potency levels showed inde-
pendent associations with symptom relief and experiences of both positive and negative side eects, with higher
Figure 1. Adjusted Change in Symptom Severity by THC and CBD Percentage Category in Flower.
Figure 2. Adjusted Change in Symptom Severity by THC Percentage Category in Flower & Symptom Type.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

6
SCIENTIFIC REPORTS | (2019) 9:2712 | https://doi.org/10.1038/s41598-019-39462-1
www.nature.com/scientificreports
www.nature.com/scientificreports/
levels resulting in larger eects. In contrast, we did not observe an independent link between CBD levels and any
of the omnibus symptom eects measured in the current study across nearly 20,000 user sessions.
Variability in cannabinoid proles may partially explain inconsistent ndings in the literature on, for exam-
ple, the benets of using cannabis for treating chronic neuropathic pain, with eectiveness observed in some
studies4,34, but not others35. Similarly, while many patient groups (e.g., sleep–disturbed medical cannabis users)
have reported a preference for high CBD concentrates36, we did not observe any patient outcomes varying by
CBD potency levels alone. One possibility is that many of the CBD potency levels displayed on labels of the
products consumed in the study were inaccurate (e.g., inated), as is currently common in the medical cannabis
industry37. Alternatively, it is possible that CBD has more latent eects than THC (e.g., expanding beyond the
90 minute observation window), has an impact on symptoms infrequently reported in our data, or that CBD’s
eects may not lend themselves to perceptual detection and subjective reporting. e phytocannabinoid family
of CBDs are known to dier from other cannabinoids such as THC in several ways, including having no anity
to CB1 receptors, serving as an antagonist to GPR55 receptors and as an inverse modulator of the eects of THC
and perhaps the endocannabinoid system more generally, as well as functioning as an immuno-suppressant and
anti-inammatory agent38,39. us, it is possible that while CBD may operate inconspicuously to improve certain
health outcomes, the adjunctive consumption of THC is needed to consciously experience or be aware of such
eects.
Notwithstanding the innovative nature and potential implications of the study’s ndings, our observational,
quasi-eld experiment had unavoidable limitations, including the lack of a control group, e.g., non-cannabis
users with similar symptoms, salient characteristics, past experiences, and voluntary reporting, which could lead
to either: a) overestimation of the eectiveness of product characteristics if users who have negative experiences
with cannabis are more likely to drop out of the sample by choosing not to use the ReleafApp; or b) underestima-
tion of cannabis’ eectiveness if users fail to use the ReleafApp due to already being satised with their product
Variables
(1) (2) (3) (4) (5) (6)
Negative % of
Negative Positive % of Positive Context-
Specic % of Context-
Specic
Concentrate −0.024 0.008 −0.097** −0.090** −0.052 −0.001
(0.080) (0.028) (0.044) (0.037) (0.101) (0.062)
C. indica 0.079** 0.015*0.011 −0.033** 0.030 0.045***
(0.035) (0.009) (0.017) (0.016) (0.026) (0.010)
C. sativa 0.003 −0.002 −0.000 −0.022 −0.041 −0.043**
(0.045) (0.008) (0.012) (0.016) (0.033) (0.019)
Pipe −0.109 −0.022 −0.046 0.002 −0.041 −0.020
(0.077) (0.024) (0.060) (0.021) (0.045) (0.035)
Vap e −0.168** −0.037 −0.044 −0.004 −0.050 −0.047
(0.076) (0.024) (0.049) (0.024) (0.059) (0.031)
THC 10–14% 0.113*** 0.013 0.017 0.055*** 0.178*** 0.066***
(0.043) (0.010) (0.018) (0.015) (0.050) (0.020)
THC 15–34% 0.090*0.016 0.013 0.069*** 0.203*** 0.085***
(0.053) (0.013) (0.019) (0.018) (0.051) (0.021)
THC 35%+0.198** 0.050** 0.065*0.118*** 0.182** 0.095*
(0.091) (0.025) (0.036) (0.033) (0.072) (0.052)
CBD 1–9% 0.030 −0.007 0.000 −0.051*** −0.007 −0.009
(0.042) (0.014) (0.015) (0.018) (0.038) (0.024)
CBD 10–34% 0.027 −0.012 0.011 −0.032*−0.013 0.005
(0.029) (0.008) (0.013) (0.018) (0.047) (0.028)
CBD 35%+0.042 −0.017 0.016 −0.033 0.093 −0.027
(0.062) (0.022) (0.024) (0.022) (0.070) (0.053)
Starting Symptom Level 0.005 0.003** 0.001 −0.003 0.007 0.003
(0.007) (0.001) (0.002) (0.002) (0.005) (0.002)
Constant 0.566*** 0.108*** 0.988*** 0.288*** 0.629*** 0.160***
(0.081) (0.023) (0.048) (0.029) (0.077) (0.039)
Observations 3,220 3,220 3,220 3,220 3,220 3,220
R-squared 0.016 0.015 0.009 0.047 0.025 0.052
N Users 665 665 665 665 665 665
Table 3. Relative Associations of Product Characteristics with Side Eects. Notes: Regressions control for
individual user xed eects. e Concentrate is relative to Flower, C. indica and C. sativa are relative to Hybrid,
THC categories are relative to THC between 0 and 10%, and CBD categories are relative to 0% CBD, and Pipe
and Vape are relative to Joint. Standard errors are clustered at the user level (shown in parentheses). ***p < 0.01,
**p < 0.05, *p < 0.1.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

7
SCIENTIFIC REPORTS | (2019) 9:2712 | https://doi.org/10.1038/s41598-019-39462-1
www.nature.com/scientificreports
www.nature.com/scientificreports/
choices and their eects. It is also important to note that the patient-reported outcomes were not cross-referenced
with clinical assessments. As with any observational study there is the potential confound of a placebo eect,
and given that cannabis products advertised as containing higher THC contents are generally more expensive
to consumers, they may be subject to a buyer’s justication eect (magnied appreciation to justify an added
expense of purchase). Another limitation is that the ReleafApp may be better suited to tracking the more imme-
diate responses of users of concentrates, ower, and to some extent, tinctures versus the longer to peak eects of
edibles and topicals. It is also possible that people who choose not to use the ReleafApp have dierent experiences
with product characteristics than those who do use the app. Another limitation was the inability to distinguish
subtleties across product types, such as pipes, which can vary in material construction and potential chemical
reactions (e.g., hydrolysis via water pipes). Finally, we anticipate that greater nuances exist in the eects of product
characteristics, and particularly, cannabinoid contents across symptom categories, but the current study is lim-
ited by our sample size within each respondent subgroup. Future research will capitalize on our ever increasing
sample size to analyze the pharmacodynamic interactions of major cannabinoids and other organic compounds
including terpenoids, as well as the harm of cannabis production practices. For example, the use of solvents to
extract cannabinoids for making concentrates used in making non-ower products (e.g., edibles, tinctures) may
place patients at risk for respiratory and cardiovascular problems40 and be a cause of increased emergency cases
of Cannabinoid Hyperemesis Syndrome41.
In conclusion, rapid increases in the popularity of medical cannabis and the associated increase in the number
of patients highlight the urgency of investigating and directing eective usage. Cannabis use carries the risk of
addiction and short-term impairments in cognitive and behavioral functioning, including the potential for safety
issues in the workplace or while driving. However, with preliminary evidence that cannabis may treat an even
wider range of conditions than those tracked in this study, including cancer42,43, it is imperative that the scientic
community develop innovative strategies such as the use of mobile technology for measuring the multidimen-
sional relationships among cannabis product characteristics, patient health conditions, perceived symptom relief,
and side eect manifestation.
Data Availability
e data that support the ndings of this study are available from MoreBetter Ltd. but restrictions apply to the
availability of these data, which were used under license for the current study, and so are not publicly available.
Data are however available from the authors upon reasonable request and with permission of MoreBetter Ltd.
References
1. Piper, B. J. et al. Substitution of medical cannabis for pharmaceutical agents for pain, anxiety, and sleep. J Psychopharmacol. 31,
569–575 (2017).
2. Hill, . P. & Weiss, . D. Minimal physical health ris associated with long-term cannabis use—but buyer beware. JAMA. 315,
2338–2339 (2016).
3. ubin, . Medical marijuana is legal in most states, but physicians have little evidence to guide them. JAMA. 317, 1611–1613 (2017).
4. National Academies of Sciences, Engineering, and Medicine; Health and Medicine Division; Board on Population Health and Public
Health Practice; Committee on the Health Eects of Marijuana: An Evidence eview and esearch Agenda. Washington (DC):
National Academies Press (US) (2017).
5. Stith, S. S. & Vigil, J. M. V. Federal barriers to Cannabis research. Science. 352, 1182 (2016).
6. Di Forti, M. et al. Proportion of patients in south London with rst-episode psychosis attributable to use of high potency cannabis:
A case-control study. e Lancet Psychiatry. 2, 233–238 (2015).
7. Gage, S. H. et al. Assessing causality in associations between cannabis use and schizophrenia ris: a two-sample Mendelian
randomization study. Psychological Medicine. 47, 971–980 (2017).
8. Fahoury, M. Could cannabidiol be used as an alternative to antipsychotics? Journal of Psychiatric Research. 80, 14–21 (2016).
9. Gururajan, A. & Malone, D. T. Does cannabidiol have a role in the treatment of schizophrenia? Schizophre nia Research. 176, 281–290
(206).
10. Iseger, T. A. & Bossong, M. G. A systematic review of the antipsychotic properties of cannabidiol in humans. Schizophrenia Research.
162, 153–161 (2015).
11. Christian, D. et al. Cannabis with high cannabidiol content is associated with fewer psychotic experiences. Schizophrenia Research.
130, 216–221 (2011).
12. National Institute of Drug Abuse. Marijuana as Medicine, https://www.drugabuse.gov/publications/drugfacts/marijuana-medicine
on (2018).
13. amo, D. E., Popova, L., Grana, ., Zhao, S. & Chavez, . Cannabis Mobile Apps: A Content Analysis. Eysenbach G, ed. JMIR
mHealth and uHealth. 3, e81 (2015).
14. eleafApp, https://releafapp.com/ (2018).
15. Vigil, J. M., Stith, S. S., Adams, I. M. & eeve, A. P. Associations between medical cannabis and prescription opioid use in chronic
pain patients: A preliminary cohort study. PLoS ONE. 12, e0187795 (2017).
16. indred, J. H. et al. Cannabis use in people with Parinson’s disease and Multiple Sclerosis: A web-based investigation.
Complementary erapies in Medicine. 33, 99–104 (2017).
17. Bonn-Miller, M. O., Babson, . A. & Vandrey, . Using cannabis to help you sleep: Heightened frequency of medical cannabis use
among those with PTSD. Drug and Alcohol Dependence. 136, 162–165 (2014).
18. Costain, W. F. e eects of cannabis abuse on the symptoms of schizophrenia: Patient perspectives. International Journal of Mental
Health Nursing. 17, 227–235 (2008).
19. MacCallum, C. A. & usso, E. B. Practical considerations in medical cannabis administration and dosing. European Journal of
Internal Medicine. 49, 12–19 (2018).
20. Fischer, B. et al. Lower-ris cannabis use guidelines: A comprehensive update of evidence and recommendations. American Journal
of Public Health. 107, e1–e12 (2017).
21. Stith, S. S., Vigil, J. M., Brocelman, F., eenan, . & Hall, B. Patient-reported symptom relief following medical cannabis
consumption. Frontiers in Pharmacology. 9, 916, https://www.frontiersin.org/articles/10.3389/fphar.2018.00916/full (2018).
22. Vigil, J. M. et al. Eectiveness of raw, natural medical Cannabis ower for treating insomnia under naturalistic conditions. Medicines.
5, 75. https://www.mdpi.com/2305-6320/5/3/75 (2018).
23. Blundell, M., Dargan, P. & Wood, D. A cloud on the horizon-a survey into the use of electronic vaping devices for recreational drug
and new psychoactive substance (NPS) administration. QJM. 111, 9–14 (2018).
Content courtesy of Springer Nature, terms of use apply. Rights reserved

8
SCIENTIFIC REPORTS | (2019) 9:2712 | https://doi.org/10.1038/s41598-019-39462-1
www.nature.com/scientificreports
www.nature.com/scientificreports/
24. Eggers, M. E. et al. Youth use of electronic vapor products and blunts for administering cannabis. Addictive Behaviors. 70, 79–82
(2017).
25. Marijuana Policy Project, https://www.mpp.org/issues/medical-marijuana/state-by-state-medical-marijuana-laws/medical-
marijuana-patient-numbers/ (2018).
26. Dinis-Oliveira, . J. Metabolomics of Δ9-tetrahydrocannabinol: implications in toxicity. Drug Metabolism Reviews. 48, 80–87
(2016).
27. Lefever, T. W. et al. Vaping synthetic cannabinoids: A novel preclinical model of e-cigarette use in mice. Substance Abuse: Research
and Treatment. 11, https://doi.org/10.1177/1178221817701739 (2017).
28. Manwell, L. A. et al. A vapourized Δ9-tetrahydrocannabinol (Δ9-THC) delivery system part I: Development and validation of a
pulmonary cannabinoid route of exposure for experimental pharmacology studies in rodents. Journal of Pharmacological and
Toxicological Methods. 70, 120–127 (2014).
29. Nguyen, J. D. et al. Inhaled delivery of Δ9-tetrahydrocannabinol (THC) to rats by e-cigarette vapor technology. Neuropharmacology.
109, 112–120 (2016).
30. Vandrey, . et al. Pharmacoinetic profile of oral cannabis in humans: Blood and oral fluid disposition and relation to
pharmacodynamic outcomes. Journal of Analytical Toxicology. 41, 83–99 (2017).
31. Varlet, V. et al. Drug vaping applied to cannabis: Is “Cannavaping” a therapeutic alternative to marijuana? Scientic Reports. 6, 25599,
https://www.nature.com/articles/srep25599 (2016).
32. Cohen, N. L., Heinz, A. J., Ilgen, M. & Bonn-Miller, M. O. Pain, Cannabis Species, and Cannabis Use Disorders. Journal of Studies on
Alcohol and Drugs. 77, 515–520 (2016).
33. Pearce, D., Mitsouras, . & Irizarry, . Discriminating the Eects of Cannabis sativa and Cannabis indica: A Web Survey of Medical
Cannabis Users. Journal of Alternative & Complementary Medicine. 20, 787–791 (2014).
34. Andreae, M. H. et al. Inhaled cannabis for chronic neuropathic pain: an individual patient data meta-analysis. e Journal of Pain.
16, 1221–1232 (2015).
35. Müce, M., Phillips, T., adbruch, L., Petze, F. & Häuser, W. Cannabis-based medicines for chronic neuropathic pain in adults.
Cochrane Database of Systematic Reviews. 3, Art. No.: CD012182. https://doi.org/10.1002/14651858.CD012182.pub2 (2018).
36. Belendiu, . A., Babson, . A., Vandrey, . & Bonn-Miller, M. O. Cannabis species and cannabinoid concentration preference
among sleep-disturbed medicinal cannabis users. Addictive Behaviors. 50, 178–181 (2015).
37. Bonn-Miller, M. O. et al. Labeling accuracy of cannabidiol extracts sold online. JAMA. 318, 1708–1709 (2017).
38. Burstein, S. Cannabidiol (CBD) and its analogs: a review of their eects on inammation. Bioorganic & Medicinal Chemistry. 23,
1377–1385 (2017).
39. McPartland, J. M., Duncan, M., Di Marzo, V. & Pertwee, . G. Are cannabidiol and Δ9-tetrahydrocannabivarin negative modulators
of the endocannabinoid system? A systematic review. British Journal of Pharmacology. 172, 737–753 (2015).
40. Piano, M. Cannabis smoing and cardiovascular health: It’s complicated. Clin. Pharmacol. er. 102, 191–193 (2017).
41. ichards, J. . Cannabinoid hyperemesis syndrome: Pathophysiology and treatment in the emergency department. Journal of
Emergency Medicine. 54, 354–363 (2018).
42. Fowler, C. Delta9-tetrahydrocannabinol and cannabidiol as potential curative agents for cancer: A critical examination of the
preclinical literature. Clin. Pharmacol. er. 97, 587–596 (2015).
43. Śledzińsi, P., Zeyland, J., Słomsi, . & Nowa, A. e current state and future perspectives of cannabinoids in cancer biology.
Cancer Medicine. 7, 765–775 (2018).
Acknowledgements
is research was funded in part by donations to the University of New Mexico Medical Cannabis Research Fund
(mcrf.unm.edu). All authors had access to the data in the study and take responsibility for the integrity of the data
and the accuracy of the data analyses.
Author Contributions
J.M.V. and S.S.S. conceived the study. F.B., K.K., B.H. independently designed and developed the ReleafApp and
server infrastructure as part of their eort to help create an education tool for medical cannabis patients. S.S.S.
conducted the analyses. J.M.V. and S.S.S. draed the manuscript. All authors contributed substantially to its
intellectual content and revision.
Additional Information
Supplementary information accompanies this paper at https://doi.org/10.1038/s41598-019-39462-1.
Competing Interests: e authors are associated with the University of New Mexico Medical Cannabis
Research Fund, which was designed to support the costs of research investigating the safety and eectiveness of
medical cannabis. F.B., K.K. and B.H. are employed by MoreBetter Ltd. e authors have no other conicts of
interest.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional aliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Cre-
ative Commons license, and indicate if changes were made. e images or other third party material in this
article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons license and your intended use is not per-
mitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the
copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
© e Author(s) 2019
Content courtesy of Springer Nature, terms of use apply. Rights reserved

1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:
use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at
onlineservice@springernature.com