Online Patient-Provider Cannabis Consultations

Abstract

Cannabis has been legalized, decriminalized, or medicalized in over half the U.S. states. With restrictions on cannabis research, accepted standards to guide clinical practice are lacking. Analyzing online communications through a digital health platform, we characterized patient questions about cannabis use and provider responses. Coded for content were 4,579 questions posted anonymously online between March 2011 through January 2017, and the responses from 1,439 U.S. licensed clinicians. Provider responses to medical cannabis use questions were coded for sentiment: “negative”, “positive”, and “mixed.” Responses could be “thanked” by patients and receive “agrees” from providers. The most frequent themes were detection of cannabis use (25.3%), health harms (19.9%), co-use with other substances (9.1%), and medical use (8.2%). The 425 medical cannabis use questions most frequently related to treatment of mental illness (20.3%), pain (20.0%), and cancer care (6.7%). The 762 provider responses regarding medical cannabis use were coded for sentiment as 59.5% negative, 28.6% mixed, and 11.8% positive. Provider sentiment was most positive regarding cannabis use for palliative care and most negative for treating respiratory conditions, poor appetite, and mental illness. The proportion of positive sentiment responses increased from 17.6% to 32.4%. Provider responses coded as negative sentiment received more provider “Agrees” (mean rank=280) than those coded as positive (mean rank=215), beta coefficient=0.33; 95% CI: 0.05, 0.62; p=.02. Cannabis use is a health topic of public interest. Variability in provider responses reflects the need for more research and consensus building to inform evidence-based clinical guidelines for cannabis use in medicine.

Introduction

Since 2002, adult use of cannabis has been increasing.¹ In 2017, 26 million Americans reported cannabis use in the past 30 days, with prevalence of 22% among adults aged 18-to-25 and 8% among adults 26 years and older.¹

Currently, 34 states, the District of Columbia, Guam, Puerto Rico, and the U.S. Virgin Islands have legalized cannabis for the treatment of medical conditions.² Fourteen states and the District of Columbia have also legalized cannabis for recreational use. Another 12 states permit the use of products with low-tetrahydrocannabinol (THC)/high cannabidiol (CBD) for medical reasons.² THC is the primary psychoactive ingredient of cannabis. CBD is another ingredient of cannabis but does not produce the same effects as THC.³

At the federal level, cannabis remains illegal. The U.S. Federal Drug Enforcement Agency (DEA) classifies cannabis as a Schedule I substance, which is reserved for drugs with a high potential of abuse and no accepted medical value. Schedule I requires the highest level of DEA permission and background checks to study cannabis and its constituents.^4,5 Furthermore, U.S. researchers can only study the cannabis grown on a government-contracted farm at the University of Mississippi, which, relative to what patients can obtain at state-legalized dispensaries, is restricted in potency and form of delivery (i.e., flower only).

To consolidate research findings toward a consensus, in 2017, the National Academies of Sciences, Engineering, and Medicine (NAS) published a comprehensive in-depth review of 10,000 studies investigating what could conclusively be said about the health effects of cannabis.⁶ The report found strong evidence from randomized control trials to support the conclusions that cannabis or its constituents (i.e., cannabinoids) are effective for treating chronic pain; as antiemetics in the treatment of chemotherapy-induced nausea and vomiting; and for improving patient-reported multiple sclerosis spasticity symptoms. Other research has found an anxiolytic-like effect of the cannabinoid cannabidiol (CBD) in patients with social anxiety disorder.⁷ There also is moderate evidence for cannabinoids, mainly nabiximols, in improving short-term sleep outcomes in those with chronic medical conditions (e.g., fibromyalgia).⁸

States that have legalized cannabis for medical purposes list qualifying conditions, which do not necessarily map to the evidence-base. California, the first state to legalize medical cannabis in 1996, references medical cannabis for the treatment of “cancer, anorexia, AIDS, chronic pain, spasticity, glaucoma, arthritis, migraine, or any other illness for which marijuana provides relief.”⁹ Some states, including New York and New Mexico, list opioid use disorder as a qualifying condition.^10–12

The NAS 2017 report also summarized evidence of the adverse health effects of cannabis use, identifying substantial evidence of worsened respiratory symptoms with long-term cannabis smoking; lower birth weight of offspring following prenatal exposure; and the development of schizophrenia or other psychoses, with the highest risk among the most frequent users.⁶ Prior reports on the adverse health effects of cannabis use identified reduced overall cortical gray matter volumes, increased impulsivity, increased executive dysfunction with adolescent cannabis use onset;¹³ and indicators of hepatotoxicity with chronic cannabis use.¹⁴

Patients are more likely to initiate a discussion about medical cannabis for treatments than their provider.¹⁵ However, lacking federal or state regulation of product contents, including THC levels; knowledge of possible medication interactions; or an understanding of health effects with maintenance use, many clinicians and health care systems are wary to recommend cannabis even when legal.^16–18 The few studies that have examined providers’ views on medical cannabis have had low response rates and relied on self-reported beliefs rather than observations of actual practices.^19–21

With the expanding legalization of cannabis use, more patients are likely to seek direction from clinicians. Online platforms provide an opportunity for patients to seek medical advice, in some cases with anonymity. Digital health communication platforms increase accessibility, and when catalogued and analyzed for research, can provide novel insights.

Analyzing patient-provider online communications through an anonymous digital health platform, we aimed to characterize patients’ questions about cannabis use and analyze provider sentiment in their responses to patients’ questions concerning medical cannabis use. The digital health platform allows for patients visiting the site to indicate satisfaction or “Thanks” for a provider response and for other providers to indicate their endorsement of a provider’s response through an “Agrees” function.²² It was hypothesized that provider sentiment toward medical cannabis use would vary by the health condition of interest and would be more positive for those conditions with greater evidence; would be associated with the status of cannabis legalization in the state in which the provider was licensed; and would become more positive over time due to greater research evidence, broader state legalization, and social normalization. Patient “Thanks” and provider “Agrees” by provider sentiment also were examined.

Methods

Study Design and Setting:

This research is an exploratory analysis of patient questions concerning cannabis use and the corresponding responses from U.S. licensed clinicians. Patient questions and provider responses were posted online between March 2011 through January 2017 on the HealthTap digital health service. HealthTap has a repository of anonymous patient questions on all types of health-related topics with responses from 140,000 U.S.-licensed clinicians. Providers find out about HealthTap through a variety of sources (e.g., press coverage, word-of-mouth, email outreach), can sign-up for free, choose how actively they participate, and can search and respond to questions on topics where they have experience and expertise. Questions are also directed to providers based on their specialty. This study had no direct contact or engagement with participants, as the HealthTap data existed and are publicly available. Hence, the study qualified for an exemption from Stanford University’s Institutional Review Board.

Data Identification and Coding:

To identify patient questions of interest, the study team created a list of cannabis-related keywords (e.g., marijuana, cannabis, THC, CBD, weed), which was utilized by a HealthTap collaborator to identify and extract patient questions concerning cannabis and the corresponding provider responses. Cleaning, coding, and analyses were conducted from February through December 2018. Initial data cleaning was performed to remove irrelevant questions. For coding of themes, a grounded theory approach was utilized.²³ A defined framework was iteratively developed through team consultation and coding of the corpus (Table 1). The coding schema incorporated current literature topics (e.g., medical uses, health effects, use in pregnancy) and emergent themes (e.g., secondhand exposure). Patients’ questions were coded fully, including any reported reasons for cannabis use and specific questions they had of the provider. Questions and responses often addressed multiple topics; hence, they could be coded for more than one theme, with a maximum of three.

Table 1.

Frequency of Cannabis Keywords in 4,579 Patients’ Questions Concerning Cannabis

Keywords	Frequency n (%)
Marijuana	1848	(40.4)
Medical marijuana	157
Marijuana abuse	11
Weed	1554	(33.9)
THC	483	(10.6)
Cannabis	354	(7.7)
Cannabis oil	16
Pot	250	(5.5)
Vap* (i.e., vaping, vape, vapor)	53	(1.2)
Joint	50	(1.1)
Ganja	31	(0.7)
Blunt	21	(0.5)
Hash	20	(0.4)
Hashish	8
Bong	20	(0.4)
CBD	18	(0.4)
Cannabinoid	7	(0.2)
Kush	4	(0.1)
Spliff	2	(<0.1)
Sativa	2	(<0.1)
Grass	1	(<0.1)

Note: Medical marijuana and marijuana abuse are subcategories included in the counts for the broader term marijuana; similarly, for cannabis oil as a subcategory for cannabis and hashish for hash. Patient questions could contain more than one marijuana related keyword.

One team member coded all patient questions. A random sample of 5% of the questions and provider answers were independently double coded by a research associate, and percentage of agreement was determined to be 85.2% for coding of patient questions (kappa statistic [κ]=.82); 89.7% for medical cannabis treatment categories (κ=.88); and 85.0% for provider sentiment (κ=.74). Coding for themes (questions and provider responses to treatment codes) was conducted in ATLAS.ti Qualitative Data Analysis version 7.

For patient questions concerning medical cannabis use, a second level of coding for the health condition of interest (e.g., cancer, pain; Table 2) and for provider sentiment (i.e., negative, positive, or mixed; Table 3) was conducted. For example, if a question mentioned being pregnant and asked about use of cannabis for help with appetite it was coded in the main coding of themes for pregnancy and treatment and then within treatment it was coded under poor appetite. Provider responses were coded as “negative” if they contained mainly negative sentiment toward cannabis use, “mixed” if they referenced positive and negative attributes of medical cannabis use, and “positive” if they mainly contained positive sentiment toward cannabis use. Lastly, we created a variable indicating whether cannabis was legalized in some form in the state in which the provider was licensed by the date of the provider response.

Table 2.

Patient Questions Related to Cannabis: Themes, Coding Frequencies (N=5,211 Codes), and Example Quotes

Theme	Questions Related to:	Frequency n (%)	Example quotes
Detection	Detection and elimination of cannabis chemicals (e.g., THC, CBD)	1317 (25.3)	“Will hemp oil capsules make you test positive for marijuana on a drug screening?” “Is cranberry juice good for getting rid of weed from our system?”
Health Harms	Health harms of cannabis from coughing and cognition to disease states and death	1035 (19.9)	“Can smoking marijuana permanently alter one’s cognition, and school performance?” “Can weed cause heart attacks?”
Drug Co-Use	The effects of the co-use of cannabis with illicit drugs, alcohol, tobacco (blunt or blunt chasers), and/or recommended medications	473 (9.1)	“So what will happen if someone takes ecstasy and smokes marijuana afterward?” “What happens if I smoke marijuana while taking amox-clav 875 mg for an infection in my hand?”
Treatment	Use of cannabis to treat physical or psychological conditions	425 (8.2)	“I have glaucoma. Is it okay to take medical marijuana?” “Will marijuana treat nausea in chemo and HIV patients?”
Medical Interactions	Cannabis use in interaction with medical procedures (e.g., surgery, blood tests), healing, or with a medical condition or disease state (e.g., head injury)	422 (8.1)	“Can weed cause a false positive TB test?” “Could marijuana prevent my eye infected with viral conjunctivitis from healing?”
Pregnancy	The effects of cannabis use on fertility, pregnancy, and in breastfeeding	367 (7.0)	“Can smoking marijuana affect female fertility?” “How are male fertility and marijuana related?” “How long does marijuana take to leave your breast milk?”
Addiction	Cannabis habituation and addiction	263 (5.1)	“If I smoke weed every day, am I psychologically dependent on marijuana?” “How can I stop smoking marijuana?” “Could I have withdrawal symptoms when I quit smoking weed?”
Secondhand Exposure	Contact highs and harms of secondhand cannabis smoke and dermal (e.g., oils) exposure to self or others	197 (3.8)	“Can you get high from secondhand marijuana smoke?” “Will second-hand marijuana smoke hurt my baby?”
Minor	Cannabis use in minors or questions related to parenting	122 (2.3)	“What is a reasonable punishment for catching my child smoke weed?” “My friend has a child, but smokes weed sometimes. If he goes to doctor to get blood work and tests positive for weed, would the doc call children services?”
Harm Reduction	Harms of cannabis relative to tobacco and combustion vs. vaping or edible cannabis and relative to alcohol and illicit drugs	122 (2.3)	“What do you consider worse marijuana abuse or alcohol?” “Is eating marijuana edibles healthier than smoking?” “Are portable vaporizers safer to use to smoke marijuana?”
Legality	The legality of cannabis use	78 (1.5)	“In which states is marijuana legal?” “Is it illegal to have THC in your system?”
Dosing	Cannabis dosage	22 (0.4)	“Can you overdose on marijuana?”
Other	Cannabis’s intoxication effects and short-term side effects (e.g., red eyes) and other general and miscellaneous questions	368 (7.0)	“What is a side effect of marijuana?” “Can cannabis oil get you high?”

Abbreviations: THC, tetrahydrocannabinol; CBD, cannabidiol

Table 3.

Coding of Patient Questions (N=464) Related to Medical Cannabis Use by Health Condition, Frequencies, and Example Quote

Condition	Frequency n (%)	Example quote
Mental illness	94 (20.3)	“Can weed help with bipolar disorder?”
Paina	93 (20.0)	“Will medical marijuana help for chronic back pain?” “Can marijuana help my migraine headaches?”
General medical helpfulness	40 (8.6)	“Is smoking weed occasionally beneficial for health?” “What’s better for your health medical marijuana or prescribed pills?”
Cancer, including lung cancer	31 (6.7)	“Can marijuana help with colon cancer?”
Sleepa	31 (6.7)	“Does medical marijuana help with insomnia?”
Eye conditions	30 (6.5)	“Can marijuana be used to treat glaucoma?”
Conditions of the central nervous systema	28 (6.0)	“Can medical marijuana help seizures?” “Where can I find some medical cannabis for my multiple sclerosis?”
Conditions of the bone, joints, skin, and other tissues	23 (5.0)	“Can I try cannabis for rheumatoid arthritis in hand?” “Can medicinal marijuana help with symptoms of fibromyalgia?”
Respiratory conditions (e.g., asthma, bronchitis, COPD)	17 (3.7)	“Does smoking weed help people with COPD?”
Stomach and digestive conditions	15 (3.2)	“Is marijuana an effective treatment for gastroparesis?”
Nauseaa	14 (3.0)	“Does smoking marijuana get rid of nausea?”
Medical cannabis cards and state laws	14 (3.0)	“What diseases qualify for a medical marijuana card?”
Endocrine disorders	12 (2.6)	“Can medical marijuana treat adrenal or endo disease?” “Does type 2 diabetes and diabetic retinopathy qualify you for a medical marijuana?”
Poor appetite	7 (1.5)	“I was diagnosed with cachexia and I smoked some marijuana. Would that help appetite?”
Palliative care	4 (0.9)	“Are there any legal drugs that do what marijuana does in palliative care?”
Other specific conditions	11 (2.4)	“Can cannabis help to stimulate hair growth due to alopecia” “Can medical marijuana help heal a foot infection?”

Note:

^aIndicates condition with evidence for therapeutic benefits for cannabis or its components according to the NAS 2017 report.

Analyses:

Simple counts summarized patient cannabis-related questions over time. Frequencies were calculated to summarize the cannabis-related word tag identifiers in patients’ questions and the themes of the patient questions. Mean ranks and medians were calculated to summarize patient “Thanks” and provider “Agrees” for the provider responses to questions regarding medical use of cannabis. A multinomial generalized linear mixed model (GLMM) with a logit link^24,25 was used to estimate associations for response date (year, entered as a quadratic variable) and state cannabis legalization where the provider was licensed (yes/no) on sentiment (negative vs mixed vs positive). Separate GLMMs with a Poisson distribution and log link^24,25 examined the associations of response sentiment with “Thanks” and “Agrees.” These analyses using GLMM were adjusted for the clustering by provider, as providers could answer more than one patient question. A p value of .05 from 2-tailed tests of GLMMs were considered statistically significant. Analyses were conducted using IBM SPSS version 25.

Results

Sample description:

A search of 23 unique keywords pertaining to cannabis identified a total of 18,447 questions. Removing duplicated questions due to multiple keywords or more than one provider responding, the dataset was reduced to a unique set of 5,160 questions. Removing questions unrelated to cannabis, the final sample was 4,579 questions. As indicated in Table 1, the most frequent keywords in patient questions were “Marijuana” (40.4%); “Weed” (33.9%); “THC” (10.6%); “Cannabis” (7.7%); and “Pot” (5.5%); < 1% included “CBD.”

The final response set was 6,891 responses from 1,439 clinicians licensed in 50 states and the District of Columbia. Most frequently represented provider specialties were psychiatry (12.7%), family medicine (11.3%), internal medicine (8.5%), pediatrics (4.9%), clinical psychology (3.5%), obstetrics and gynecology (3.0%), dentistry (2.9%), ophthalmology (2.4%), pain management (2.3%), neurology (2.2%), addiction (1.9%), general practice (1.7%), and general surgery (1.6%). Provider responses to patient questions were posted from March 15, 2011 through the dataset extraction period of January 16, 2017. During this time, the response frequency rose to a peak in early 2013 (Figure 1). The majority of cannabis-related questions in the dataset were answered from mid-2012 through mid-2015.

Figure 1.

Volume of Questions Over Time (N=4579 questions), 2011 to 2017

Patient question themes:

The 4,579 patient questions received 5,211 content codes. Most frequently asked were questions regarding detection and elimination of cannabis or cannabis chemicals (e.g., THC, cannabis oil, CBD; 25.3%); followed by negative health effects, including neurocognitive adverse effects (19.9%); co-use with illicit drugs, alcohol, tobacco, or prescribed medications (9.1%); medical cannabis use (8.2%); use with existing health conditions or medical procedures (8.1%); effects on conception, pregnancy, and breastfeeding (7.0%); addiction (5.1%); secondhand exposure, including around youth (3.8%); use by minors (2.3%); use as a harm reduction alternative (2.3%); legality of use (1.5%); and dosing (0.4%) (Table 2).

A total of 425 patient questions concerned medical cannabis use. One question could focus on multiple medical uses, which led to a total of 464 treatment codes, in order of frequency: mental illness (20.3%); pain (20.0%); cancer, including lung cancer (6.7%); sleep issues (6.7%); eye conditions (6.5%); conditions of the central nervous system (6.0%); conditions of the bone, joints, skin and other tissues (5.0%); respiratory conditions (e.g., COPD, asthma) (3.7%); stomach and digestive conditions (3.2%); nausea (3.0%); endocrine disorders (2.6%); poor appetite concerns (1.5%); and palliative care (0.9%). Another 8.6% of questions concerned the general medical helpfulness of cannabis, 3.0% medical cannabis cards and state laws, and the remaining 2.4% a myriad of low frequency (< 2 questions) on specific health conditions coded collectively as “Other” (Table 3).

Provider response sentiment regarding medical cannabis use:

A total of 762 provider responses to patient questions regarding medical use of cannabis were posted online by 358 clinicians licensed in 46 states and the District of Columbia. Responses per provider ranged from 1 to 58, with a mode and median of 1. Coding indicated provider response sentiment toward medical cannabis use was 59.6% negative, 28.6% mixed, and 11.8% positive (Table 4). (See Table 5 for frequency of provider specialty by sentiment.) From 2011 through 2016, the main years of data collection, there was a significant change over time in negative sentiment (beta coefficients [ß]=−0.19; 95% CI: −0.33, −0.05; p=.008) and mixed sentiment (ß=−0.17; 95% CI: −0.31, −0.02; p=.02) compared to positive sentiment. The proportion of negative sentiment responses decreased from 58.8% in 2011 to 37.8% in 2016; mixed sentiment increased from 23.5% to 29.7%; and positive sentiment increased from 17.6% to 32.4%. Across all years, negative sentiment was greatest in response to questions regarding cannabis treatment for respiratory conditions (89.3%), poor appetite (80.0%), and mental illness (76.3%). In contrast, positive sentiment was greatest in response to questions regarding palliative care (44.4%), though positive sentiment responses were still in the minority (Table 4).

Table 4.

Sentiment Frequencies for Provider Responses to Patient Questions about Medical Cannabis Use

Treatment Code	Provider Responses N (%)	Sentimentc, n (%)
		Negative (n=505)	Mixed (n=242)	Positive (n=100)
Mental illness	190 (22.4)	145 (76.3)	40 (21.1)	5 (2.6)
Paina	169 (22.0)	87 (51.5)	56 (33.1)	26 (15.4)
General medical helpfulness	71 (8.4)	28 (39.4)	24 (33.8)	19 (26.8)
Sleepa	63 (7.4)	42 (66.7)	13 (20.6)	8 (12.7)
Eye conditions	58 (6.8)	40 (69.0)	12 (20.7)	6 (10.3)
Cancer	51 (6.0)	26 (51.0)	19 (37.3)	6 (11.8)
Conditions of the bone, joints, skin, & other tissues	48 (5.7)	22 (45.8)	17 (35.4)	9 (18.8)
Conditions of the central nervous systema	46 (5.4)	22 (47.8)	16 (34.8)	8 (17.4)
Respiratory conditions	28 (3.3)	25 (89.3)	3 (10.7)	0 (0.0)
Endocrine disorders	22 (2.6)	14 (63.6)	7 (31.8)	1 (4.5)
Medical cannabis cards and state law	21 (2.5)	6 (28.6)	13 (61.9)	2 (9.5)
Nauseaa	23 (2.7)	9 (39.1)	10 (43.5)	4 (17.4)
Stomach conditions	23 (2.7)	17 (73.9)	4 (17.4)	2 (8.7)
Poor appetite	10 (1.2)	8 (80.0)	2 (20.0)	0 (0.0)
Palliative care	9 (1.1)	1 (11.1)	4 (44.4)	4 (44.4)
Other specific conditions	15 (1.8)	13 (86.7)	2 (13.3)	0 (0.0)
Total	847b	59.6%	28.6%	11.8%

Notes:

^aIndicates condition with evidence for therapeutic benefits from cannabis or its components according to the NAS 2017 report.

^bThe 847 coded responses were based on the 762 provider responses that could be categorized into more than one category.

^cProvider responses were coded as “negative” if they contained mainly negative sentiment toward cannabis use (e.g., “No: There is no medical need for marijuana with eye diseases”), “mixed” if they referenced positive and negative attributes of medical cannabis use (e.g., “No: cannabis will not cure cancer but it may ease pain associated with cancer”), and “positive” if they mainly contained positive sentiment toward cannabis use (e.g., “Yes: Fibromyalgia is one of the disorders that can be treated with cannabis”).

Table 5.

Sentiment Frequencies for Unique Provider Responses to Patient Questions by Provider Specialty

Provider Specialty	Provider Responses N (%)	Sentiment, n (%)
		Negative (n=505)	Mixed (n=242)	Positive (n=100)
Psychiatry	218 (28.6)	148 (67.9)	62 (28.4)	8 (3.7)
Addiction Medicine	56 (7.3)	32 (57.1)	18 (32.1)	6 (10.7)
Pediatrics	44 (5.8)	32 (72.7)	9 (20.5)	3 (6.8)
Neurology	42 (5.5)	20 (47.6)	14 (33.3)	8 (19.0)
Internal Medicine	39 (5.1)	23 (59.0)	13 (33.3)	3 (7.7)
Ophthalmology	37 (4.9)	24 (64.9)	10 (27.0)	3 (8.1)
Clinical Psychology	35 (4.6)	28 (80.0)	5 (14.3)	2 (5.7)
Family Medicine	35 (4.6)	19 (54.3)	7 (20.0)	9 (25.7)
General Practice	35 (4.6)	8 (22.9)	11 (31.4)	16 (45.7)
Pain Management	33 (4.3)	15 (45.5)	15 (45.5)	3 (9.1)
Holistic Medicine	20 (2.6)	4 (20.0)	9 (45.0)	7 (35.0)
Rheumatology	15 (2.0)	10 (66.7)	4 (26.7)	1 (6.7)
Other Specialties	153 (20.1)	83 (54.2)	47 (30.7)	23 (15.0)
Total	762

Of the 358 clinicians responding to patient questions on medical cannabis use, 58.5% were licensed in states where cannabis use was legalized, 40.7% in states where cannabis use was not legalized, and 0.8% did not have their state listed. Providers’ response sentiment did not differ significantly by state legalization status, for negative (ß=0.41; 95% CI: −0.14, 0.95; p=.14) and mixed (ß=0.17; 95% CI: −0.38, 0.71; p=.55) compared to positive sentiment. For providers licensed in states where cannabis use was legalized versus was not legalized, response sentiment was 14% vs. 10% positive, 56% vs. 63% negative, and 31% vs. 27% mixed.

“Agrees” and “Thanks”:

The 762 provider responses concerning medical cannabis use received 852 patient “Thanks” and 1,288 provider “Agrees.” The number of “Thanks” per response ranged from 0 to 31, with a mode and median of 1; 48.0% of provider responses received at least one patient “Thanks”. “Thanks” did not differ significantly by provider response sentiment (negative [ß=0.07; 95% CI: −0.40, 0.53; p=.78] versus positive; mixed [ß=0.001; 95% CI: −0.39, 0.39; p=.998] versus positive). Mean rank for “Thanks” was 183.0, 180.5, and 192.8 for negative, mixed, and positive sentiment, respectively. The number of provider “Agrees” per response ranged from 0 to 9, with a mode of 1 and median of 2; 68.4% received at least 1 “Agree.” Provider responses that were negative toward medical cannabis use received significantly more provider “Agrees” (ß=0.33; 95% CI: 0.05, 0.62; p=.02) than provider responses that were positive. “Agrees” did not differ significantly by mixed sentiment (ß=0.10; 95% CI: −0.19, 0.37; p=.51) versus positive. Mean rank for “Agrees” was 280.2, 234.4, and 215.3 for negative, mixed, and positive sentiment, respectively.

Discussion

In a database of online patient–provider consultations from a publicly available digital health service, keyword searches identified nearly 5,000 unique patient questions related to cannabis use, confirming public interest. Keywords of marijuana, weed, THC, and cannabis, were most commonly used by patients in reference to the product, and future health care research will benefit from focusing on these terms. The number of patient questions rose to a peak in early 2013 with rising interest likely reflecting public discussion and increased state legalization in our study period from 2011 through 2016, and then declining numbers likely due to the database becoming saturated over time or as public information and understanding grew. The questions and responses are maintained on the site for others to search, reducing the need to ask the same questions overtime. Our patient-response data were posted up until January 2017, and it is likely that we captured the leading contemporary question themes concerning cannabis use in the U.S.

The detection and elimination of cannabis was the topic of most frequent interest, accounting for 1 in 4 questions, asked often in relation to employers. Cannabis is federally scheduled as a Class I substance, continues to be illegal for medical use in 16 states,² and employment is commonly contingent on the results of negative drug screens, which include cannabis.²⁶ One in five patient questions was related to cannabis as a catalyst for negative health conditions, such as heart attacks, depression, and cancer. One in nine patient questions concerned possible medication, alcohol, tobacco, or illicit drug interactions with cannabis. The questions were highly variable in reference to the medication or substance of concern and deserve further investigation as research into drug interactions with cannabis is sparse-to-nonexistent. That 7.0% of questions were related to the effects of cannabis on fertility and pregnancy is consistent with recent reports that women are increasingly using cannabis during child-bearing age and into their pregnancies;^27–29 the latter, in many cases to manage morning sickness.^30,31 Future research, resources, and attention should focus, in particular, on the effects of cannabis use in pregnancy and the education of gynecologic and obstetrics professionals. There has been a great deal of discussion about the use of medical cannabis to treat opioid use disorder or as an alternative pain medication for anything for which opioids are used.^32,33 While 20% of patients asked about cannabis as a treatment for pain, in this study, no patient asked specifically about use of cannabis to replace opioids.

Responses yield insight into providers’ opinions of the medical uses of cannabis. Provider sentiment toward medical cannabis use tended to be negative overall though varied by medical concern of interest. Most negative was sentiment regarding cannabis use for the treatment of respiratory disorders, mental illness, and poor appetite. Sustained use of inhaled combusted products will cause respiratory harm, and use of cannabis in other forms (e.g., oils, wax, edibles) is not supported as a treatment for respiratory conditions. For mental illness, rather than a treatment, the evidence implicates cannabis use in the development of psychosis, particularly with longer duration of use and use of high-potency cannabis.³⁴ While posttraumatic stress disorder (PTSD) is listed as a qualifying condition for medical cannabis use in 28 states,³⁵ the NAS 2017 report found limited evidence that cannabis or cannabinoids are effective in improving PTSD symptoms.⁶ Similarly, while poor appetite is listed as a qualifying condition for cannabis use in some states (e.g., California), according to the NAS 2017 report there is “limited evidence that cannabis and oral cannabinoids are effective in increasing appetite and decreasing weight loss associated with HIV/AIDS,” and “insufficient evidence to support or refute the conclusion that cannabinoids are an effective treatment for cancer-associated anorexia-cachexia syndrome and anorexia nervosa.”⁶

In contrast, there was greater clinician support for cannabis use in palliative care. Similarly, in a national survey of hospice professionals, 91% reported support for cannabis use in hospice patients.³⁶ Among pediatric oncology providers in Illinois, Massachusetts, and Washington, 89% were supportive of medical cannabis for palliative care.³⁷ With greater study warranted, it appears clinicians may view medical cannabis as an alternative medicine approach when established treatments fail to provide adequate relief.

Overall, variability was observed in provider sentiment, suggesting the need for consensus building in the field, the development of clinical practice guidelines, and dissemination through medical education (i.e., instruction on medical cannabis and cannabinoid research). Of 101 medical school deans surveyed nationally, 66.7% reported that their graduates were not at all prepared to recommend cannabis.³⁸ A review of the Association of American Medical Colleges curriculum database found only 9% of medical schools teach anything about medical uses of cannabis.³⁸ In a survey of practicing clinicians in Washington State, most reported low knowledge and comfort level recommending medical cannabis.³⁹

The proportion of positive sentiment increased from 2011 to 2016, though the numbers were too small to test statistically by treatment code. A possible catalyst for the shift may be the U.S. Department of Justice’s decision in late 2013 to no longer challenge state cannabis law; publicized legalization of medical cannabis use in New York and Minnesota; and/or the passing of a senate spending bill in 2014 that blocked the Justice Department from funding enforcement of federal cannabis law in states where cannabis had been legalized.^2,40

Limitations:

A novel summary of online patient-provider communications regarding cannabis, the current study has limitations. Patients’ questions were posted anonymously, which limits understanding of the data origins. People could have posted multiple cannabis-related questions, within or across themes. Given the anonymous platform provided to patients, we were unable to control for dependency in the dataset of question. The providers were identified and their responses were posted publicly, which may have constrained their advice. Purely observational, this study did not assess patient experience, potential use of cannabis, or past medical history, information sources guiding provider opinions, or the overall patient-provider experience. The features for “Thanks” and “Agrees” could be influenced by factors other than content, such as patient gratitude that they received a response or how long an answer was on the site, the latter was controlled for in analyses. The political and controversial nature of discussions regarding cannabis, its health effects, and its legalization may create self-selection by providers who are adamant proponents or opponents of cannabis. Although this study’s exploratory findings generally cannot provide strong inferential evidence, they do illustrate a well-reasoned set of themes and inferences to inform future hypotheses. Lastly, though not an in-person clinical interaction, virtual medical contacts are increasingly common, and patients may be more open in their communications on cannabis in a virtual context.

Conclusion:

In the U.S., cannabis use is on the rise and practicing clinicians are likely to encounter a growing number of questions from their patients. While most provider comments online tended to view medical cannabis use negatively, the proportion of positive sentiment increased over time. Variability in provider responses suggest the need for more research, consensus building, and provider education to inform best practices and evidence-based guidelines for clinical care.

Highlights.

A majority of provider responses were negative in sentiment toward cannabis use.

Variability in sentiment was observed by question theme and over time.

Provider advice tended to discourage cannabis use for the medical conditions queried.

The proportion of positive sentiment increased over time.

The proportion of positive sentiment was greatest for palliative care.