Exploring Twitter chatter to assess the type and availability of cannabis-related products in Thailand.

Cannabis-related tweets were collected between January and April 2022 to estimate the availability and characteristics of cannabis products advertised on Twitter amid the legalization of recreational cannabis in Thailand. The Twitter API was called using the tweepy Python library to collect cannabis-related tweets in the Thai language. A total of 185,558 unique tweets were collected over the duration of the data collection period based on 83 search terms. Twenty thousand random tweets were manually coded by four Thai native speakers to assess the volume and characteristics of tweets proposing cannabis. 72.6% of collected tweets from the 20,000 random samples were coded as relevant to the study. 54.6% of relevant tweets were advertising cannabis products, 29.8% were personal communications, and 15.6% were related to news or media content. Among the tweets that advertised cannabis products, 94.4% proposed cannabis flower, 2.4% cannabis edibles and 1.8% cannabis concentrates. Consumption of potent forms of cannabis such as cannabis edibles and concentrates increase the risk of harmful side-effects, especially in a population with limited knowledge about these products. Our findings call for additional monitoring efforts and for increasing the public awareness on potent cannabis products emerging in Thailand.

"Pressed OXY M30 Pills, Great Press, Potent, Fast Shipping!!!": Availability of Counterfeit and Pharmaceutical Oxycodone Pills on One Major Cryptomarket.

From 2018 to 2021, seizures of counterfeit oxycodone pills containing non-pharmaceutical fentanyl or other novel synthetic opioids increased significantly contributing to continuing increases in overdose mortality in Northern America. Evidence suggests that counterfeit pills are distributed through cryptomarkets. This article presents data regarding the availability and characteristics of oxycodone pills advertised on one major cryptomarket between January and March 2022. Collected data were processed using a dedicated Named Entity Recognition algorithm to identify oxycodone listings and categorized them as either counterfeit or pharmaceutical. Frequency of listings, average number of pills advertised, average prices per milligram, number of sales, and geographic indicators of shipment origin and destination were analyzed. In total, 2,665 listings were identified as oxycodone. 48.2% (1,285/2,665) of these listings were categorized as counterfeit oxycodone, advertising a total of 652,699 pills (93,242.7 pills per datapoint) offered at a lower price than pharmaceutical pills. Our data indicate the presence of a large volume of counterfeit oxycodone pills both in retail- and wholesale-level amounts mostly targeting US and Canadian customers. These exploratory findings call for more research to develop epidemiological surveillance systems to track counterfeit pill and other drug availability on the Dark web environment.

Exploring Televend, an innovative combination of cryptomarket and messaging app technologies for trading prohibited drugs.

BACKGROUND: Digital technologies continue to facilitate drug trading. Televend was an innovative combination of multiple digital technologies, with its backend hosted on the darknet, while purchases were made through the messaging app Telegram. Here, we provide an initial characterisation of this nascent market. METHODS: Televend and White House Market (WHM) were scraped (Jun-Jul 2021) and a global cross-sectional web survey of 15,513 drug buyers (Global Drug Survey; GDS) was conducted (Dec 2020-Mar 2021). RESULTS: Televend was 10% of the size of WHM, the largest drug cryptomarket (4515/44,830 listings per week). Both markets predominantly contained drug-related listings covering similar drug categories, with similar country of origin and destination. Very few GDS drug buyers reported use of Televend (0.73%). Most Televend buyers (68/114) reported buying cannabis, then cocaine (20), MDMA (17), and LSD (12). The Televend and darknet groups had similar demographic and drug use characteristics; whereas compared with app purchasers, older age increased the odds of Televend use (aRRR = 1.06, p < .001), identifying as a cisgender woman decreased the odds (aRRR=0.43, p = .004), while last-year use of a greater number of drug types (aRRR = 1.20, p < .001) and less frequent drug use (aRRR=0.998, p = .032) increased the odds of Televend purchase. CONCLUSIONS: While smaller, Televend was not noticeably different in its drug offerings to its largest cryptomarket competitor, and it attracted a cohort more similar to darknet than to app drug buyers. Future Televend-like markets may be attractive to people with less specialised technical knowledge who already routinely scroll through social media feeds.

"Etazene, safer than heroin and fentanyl": Non-fentanyl novel synthetic opioid listings on one darknet market.

BACKGROUND: Novel synthetic opioids are fueling the overdose deaths epidemic in North America.Recently, non-fentanyl novel synthetic opioids have emerged in forensic toxicological results. Cryptomarkets have become important platforms of distribution for illicit substances. This article presents the data concerning the availability trends of novel non-fentanyl synthetic opioids listed on one cryptomarket. METHODS: Listings from the EmpireMarket cryptomarket "Opiates" section were collected between June 2020 and August 2020. Collected data were processed using eDarkTrends Named Entity Recognition algorithm to identify novel synthetic opioids, and to analyze their availability trends in terms of frequency of listings, available average weights, average prices, quantity sold, and geographic indicators of shipment origin and destination information. RESULTS: 35,196 opioid-related listings were collected through 12 crawling sessions. 17 nonfentanyl novel synthetic opioids were identified in 2.9 % of the collected listings for an average of 9.2 kg of substance available at each data point. 587 items advertised as non-fentanyl novel synthetic opioids were sold on EmpireMarket for a total weight of between 858 g and 2.7 kg during the study period. 45.5 % of these listings were advertised as shipped from China. CONCLUSIONS: Fourteen of the 17 non-fentanyl novel synthetic opioids were identified for the first time on one large cryptomarket suggesting a shift in terms of novel non-fentanyl synthetic opioids availability. This increased heterogeneity of available novel synthetic opioids could reduce the efficiency of existing overdose prevention strategies. Identification of new opioids underpins the value of cryptomarket data for early warning systems of emerging substance use trends.

Listed for sale: Analyzing data on fentanyl, fentanyl analogs and other novel synthetic opioids on one cryptomarket.

BACKGROUND: The United States is facing a "triple wave" epidemic fueled by novel synthetic opioids. Cryptomarkets, anonymous marketplaces located on the deep web, play an increasingly important role in the distribution of illicit substances. This article presents the data collected and processed by the eDarkTrends platform concerning the availability trends of novel synthetic opioids listed on one cryptomarket. METHODS: Listings from the DreamMarket cryptomarket "Opioids" and "Research Chemicals" sections were collected between March 2018 and January 2019. Collected data were processed using eDarkTrends Named Entity Recognition algorithm to identify opioid drugs, and to analyze their availability trends in terms of frequency of listings, available average weights, average prices, and geographic indicators of shipment origin and destination information. RESULTS: 95,011 opioid-related listings were collected through 26 crawling sessions. 33 novel synthetic opioids were identified in 3.3 % of the collected listings. 44.7 % of these listings advertised fentanyl (pharmaceutical and non-pharmaceutical) or fentanyl analogs for an average of 2.8 kgs per crawl. "Synthetic heroin" accounted for 33.2 % of novel synthetic opioid listings for an average 1.1 kgs per crawl with 97.7 % of listings advertised as shipped from Canada. Other novel synthetic opioids (e.g., U-47,700, AP-237) represented 22 % of these listings for an average of 6.1 kgs per crawl with 97.2 % of listings advertised as shipped from China. CONCLUSIONS: Our data indicate consistent availability of a wide variety of novel synthetic opioids both in retail and wholesale-level amounts. Identification of new substances highlights the value of cryptomarket data for early warning systems of emerging substance use trends.

Global trends, local harms: availability of fentanyl-type drugs on the dark web and accidental overdoses in Ohio.

As America's opioid crisis has become an "epidemic of epidemics," Ohio has been identified as one of the high burden states regarding fentanyl-related overdose mortality. This study aims to examine changes in the availability of fentanyl, fentanyl analogs, and other non-pharmaceutical opioids on cryptomarkets and assess relationship with the trends in unintentional overdoses in Ohio to provide timely information for epidemiologic surveillance. Cryptomarket data were collected at two distinct periods of time: (1) Agora data covered June 2014-September 2015 and were obtained from Grams archive; (2) Dream Market data from March-April 2018 were extracted using a dedicated crawler. A Named Entity Recognition algorithm was developed to identify and categorize the type of fentanyl and other synthetic opioids advertised on cryptomarkets. Time-lagged correlations were used to assess the relationship between the fentanyl, fentanyl analog and other synthetic opioid-related ads from cryptomarkets and overdose data from the Cincinnati Fire Department Emergency Responses and Montgomery County Coroner's Office. Analysis from the cryptomarket data reveals increases in fentanyl-like drugs and changes in the types of fentanyl analogues and other synthetic opioids advertised in 2015 and 2018 with potent substances like carfentanil available during the second period. The time-lagged correlation was the largest when comparing Agora data to Cincinnati Emergency Responses 1 month later 0.84 (95% CI 0.45, 0.96). The time-lagged correlation between Agora data and Montgomery County drug overdoses was the largest when comparing synthetic opioid-related Agora ads to Montgomery County overdose deaths 7 months later 0.78 (95% CI 0.47, 0.92). Further investigations are required to establish the relationship between cryptomarket availability and unintentional overdose trends related to specific fentanyl analogs and/or other illicit synthetic opioids.

A Twitter-based survey on marijuana concentrate use.

AIMS: The purpose of this paper is to analyze characteristics of marijuana concentrate users, describe patterns and reasons of use, and identify factors associated with daily use of concentrates among U.S.-based cannabis users recruited via a Twitter-based online survey. METHODS: An anonymous Web-based survey was conducted in June 2017 with 687 U.S.-based cannabis users recruited via Twitter-based ads. The survey included questions about state of residence, socio-demographic characteristics, and cannabis use including marijuana concentrates. Multiple logistic regression analyses were conducted to identify characteristics associated with lifetime and daily use of marijuana concentrates. RESULTS: Almost 60% of respondents were male, 86% were white, and the mean age was 43.0 years. About 48% reported marijuana concentrate use. After adjusting for multiple testing, significant predictors of concentrate use included: living in "recreational" (AOR = 2.04; adj. p = .042) or "medical, less restrictive" (AOR = 1.74; adj. p = .030) states, being younger (AOR = 0.97, adj. p = .008), and daily herbal cannabis use (AOR = 2.57, adj. p = .008). Out of 329 marijuana concentrate users, about 13% (n = 44) reported daily/near daily use. Significant predictors of daily concentrate use included: living in recreational states (AOR = 3.59, adj. p = .020) and using concentrates for therapeutic purposes (AOR = 4.34, adj. p = .020). CONCLUSIONS: Living in states with more liberal marijuana policies is associated with greater likelihood of marijuana concentrate use and with more frequent use. Characteristics of daily users, in particular, patterns of therapeutic use warrant further research with community-recruited samples.

"You got to love rosin: Solventless dabs, pure, clean, natural medicine." Exploring Twitter data on emerging trends in Rosin Tech marijuana concentrates.

BACKGROUND: "Rosin tech" is an emerging solventless method consisting in applying moderate heat and constant pressure on marijuana flowers to prepare marijuana concentrates referred to as "rosin." This paper explores rosin concentrate-related Twitter data to describe tweet content and analyze differences in rosin-related tweeting across states with varying cannabis legal statuses. METHOD: English language tweets were collected between March 15, 2015 and April 17, 2017, using Twitter API. U.S. geolocated unique (no retweets) tweets were manually coded to evaluate the content of rosin-related tweets. Adjusted proportions of Twitter users and personal communication tweets per state related to rosin concentrates were calculated. A permutation test was used to analyze differences in normalized proportions between U.S. states with different cannabis legal statuses. RESULTS: eDrugTrends collected 8389 tweets mentioning rosin concentrates/technique. 4164 tweets (49.6% of total sample) posted by 1264 unique users had identifiable state-level geolocation. Content analysis of 2010 non-retweeted tweets revealed a high proportion of media-related tweets (44.2%) promoting rosin as a safer and solventless production method. Tweet-volume-adjusted percentages of geolocated Twitter users and personal communication tweets about rosin were respectively up to seven and sixteen times higher between states allowing recreational use of cannabis and states where cannabis is illegal. CONCLUSION: Our results indicate that there are higher proportions of personal communication tweets and Twitter users tweeting about rosin in U.S. states where cannabis is legalized. Rosin concentrates are advertised as a safer, more natural form of concentrates, but more research on this emerging form of marijuana concentrate is needed.

"Retweet to Pass the Blunt": Analyzing Geographic and Content Features of Cannabis-Related Tweeting Across the United States.

OBJECTIVE: Twitter data offer new possibilities for tracking health-related communications. This study is among the first to apply advanced information processing to identify geographic and content features of cannabis-related tweeting in the United States. METHOD: Tweets were collected using streaming Application Programming Interface (March-May 2016) and were processed by eDrugTrends to identify geolocation and classify content by source (personal communication, media, retail) and sentiment (positive, negative, neutral). States were grouped by cannabis legalization policies into "recreational," "medical, less restrictive," "medical, more restrictive," and "illegal." Permutation tests were performed to analyze differences among four groups in adjusted percentages of all tweets, unique users, personal communications only, and positive-to-negative sentiment ratios. RESULTS: About 30% of all 13,233,837 cannabis-related tweets had identifiable state-level geo-information. Among geolocated tweets, 76.2% were personal communications, 21.1% media, and 2.7% retail. About 71% of personal communication tweets expressed positive sentiment toward cannabis; 16% expressed negative sentiment. States in the recreational group had significantly greater average adjusted percentage of cannabis tweets (3.01%) compared with other groups. For personal communication tweets only, the recreational group (2.47%) was significantly greater than the medical, more restrictive (1.84%) and illegal (1.85%) groups. Similarly, the recreational group had significantly greater average positive-to-negative sentiment ratio (4.64) compared with the medical, more restrictive (4.15) and illegal (4.19) groups. Average adjusted percentages of unique users showed similar differences between recreational and other groups. CONCLUSIONS: States with less restrictive policies displayed greater cannabis-related tweeting and conveyed more positive sentiment. The study demonstrates the potential of Twitter data to become a valuable indicator of drug-related communications in the context of varying policy environments.

Characterizing marijuana concentrate users: A web-based survey.

AIMS: The study seeks to characterize marijuana concentrate users, describe reasons and patterns of use, perceived risk, and identify predictors of daily/near daily use. METHODS: An anonymous web-based survey was conducted (April-June 2016) with 673 US-based cannabis users recruited via the Bluelight.org web-forum and included questions about marijuana concentrate use, other drugs, and socio-demographics. Multivariable logistic regression analyses were conducted to identify characteristics associated with greater odds of lifetime and daily use of marijuana concentrates. RESULTS: About 66% of respondents reported marijuana concentrate use. The sample was 76% male, and 87% white. Marijuana concentrate use was viewed as riskier than flower cannabis. Greater odds of marijuana concentrate use was associated with living in states with "recreational" (AOR=4.91; p=0.001) or "medical, less restrictive" marijuana policies (AOR=1.87; p=0.014), being male (AOR=2.21, p=0.002), younger (AOR=0.95, p<0.001), number of other drugs used (AOR=1.23, p<0.001), daily herbal cannabis use (AOR=4.28, p<0.001), and lower perceived risk of cannabis use (AOR=0.96, p=0.043). About 13% of marijuana concentrate users reported daily/near daily use. Greater odds of daily concentrate use was associated with being male (AOR=9.29, p=0.033), using concentrates for therapeutic purposes (AOR=7.61, p=0.001), using vape pens for marijuana concentrate administration (AOR=4.58, p=0.007), and lower perceived risk of marijuana concentrate use (AOR=0.92, p=0.017). CONCLUSIONS: Marijuana concentrate use was more common among male, younger and more experienced users, and those living in states with more liberal marijuana policies. Characteristics of daily users, in particular patterns of therapeutic use and utilization of different vaporization devices, warrant further research with community-recruited samples.

Increases in synthetic cannabinoids-related harms: Results from a longitudinal web-based content analysis.

BACKGROUND: Synthetic Cannabinoid Receptor Agonists (SCRA), also known as "K2" or "Spice," have drawn considerable attention due to their potential of abuse and harmful consequences. More research is needed to understand user experiences of SCRA-related effects. We use semi-automated information processing techniques through eDrugTrends platform to examine SCRA-related effects and their variations through a longitudinal content analysis of web-forum data. METHOD: English language posts from three drug-focused web-forums were extracted and analyzed between January 1st 2008 and September 30th 2015. Search terms are based on the Drug Use Ontology (DAO) created for this study (189 SCRA-related and 501 effect-related terms). EDrugTrends NLP-based text processing tools were used to extract posts mentioning SCRA and their effects. Generalized linear regression was used to fit restricted cubic spline functions of time to test whether the proportion of drug-related posts that mention SCRA (and no other drug) and the proportion of these "SCRA-only" posts that mention SCRA effects have changed over time, with an adjustment for multiple testing. RESULTS: 19,052 SCRA-related posts (Bluelight (n=2782), Forum A (n=3882), and Forum B (n=12,388)) posted by 2543 international users were extracted. The most frequently mentioned effects were "getting high" (44.0%), "hallucinations" (10.8%), and "anxiety" (10.2%). The frequency of SCRA-only posts declined steadily over the study period. The proportions of SCRA-only posts mentioning positive effects (e.g., "High" and "Euphoria") steadily decreased, while the proportions of SCRA-only posts mentioning negative effects (e.g., "Anxiety," 'Nausea," "Overdose") increased over the same period. CONCLUSION: This study's findings indicate that the proportion of negative effects mentioned in web forum posts and linked to SCRA has increased over time, suggesting that recent generations of SCRA generate more harms. This is also one of the first studies to conduct automated content analysis of web forum data related to illicit drug use.

"When 'Bad' is 'Good'": Identifying Personal Communication and Sentiment in Drug-Related Tweets.

BACKGROUND: To harness the full potential of social media for epidemiological surveillance of drug abuse trends, the field needs a greater level of automation in processing and analyzing social media content. OBJECTIVES: The objective of the study is to describe the development of supervised machine-learning techniques for the eDrugTrends platform to automatically classify tweets by type/source of communication (personal, official/media, retail) and sentiment (positive, negative, neutral) expressed in cannabis- and synthetic cannabinoid-related tweets. METHODS: Tweets were collected using Twitter streaming Application Programming Interface and filtered through the eDrugTrends platform using keywords related to cannabis, marijuana edibles, marijuana concentrates, and synthetic cannabinoids. After creating coding rules and assessing intercoder reliability, a manually labeled data set (N=4000) was developed by coding several batches of randomly selected subsets of tweets extracted from the pool of 15,623,869 collected by eDrugTrends (May-November 2015). Out of 4000 tweets, 25% (1000/4000) were used to build source classifiers and 75% (3000/4000) were used for sentiment classifiers. Logistic Regression (LR), Naive Bayes (NB), and Support Vector Machines (SVM) were used to train the classifiers. Source classification (n=1000) tested Approach 1 that used short URLs, and Approach 2 where URLs were expanded and included into the bag-of-words analysis. For sentiment classification, Approach 1 used all tweets, regardless of their source/type (n=3000), while Approach 2 applied sentiment classification to personal communication tweets only (2633/3000, 88%). Multiclass and binary classification tasks were examined, and machine-learning sentiment classifier performance was compared with Valence Aware Dictionary for sEntiment Reasoning (VADER), a lexicon and rule-based method. The performance of each classifier was assessed using 5-fold cross validation that calculated average F-scores. One-tailed t test was used to determine if differences in F-scores were statistically significant. RESULTS: In multiclass source classification, the use of expanded URLs did not contribute to significant improvement in classifier performance (0.7972 vs 0.8102 for SVM, P=.19). In binary classification, the identification of all source categories improved significantly when unshortened URLs were used, with personal communication tweets benefiting the most (0.8736 vs 0.8200, P<.001). In multiclass sentiment classification Approach 1, SVM (0.6723) performed similarly to NB (0.6683) and LR (0.6703). In Approach 2, SVM (0.7062) did not differ from NB (0.6980, P=.13) or LR (F=0.6931, P=.05), but it was over 40% more accurate than VADER (F=0.5030, P<.001). In multiclass task, improvements in sentiment classification (Approach 2 vs Approach 1) did not reach statistical significance (eg, SVM: 0.7062 vs 0.6723, P=.052). In binary sentiment classification (positive vs negative), Approach 2 (focus on personal communication tweets only) improved classification results, compared with Approach 1, for LR (0.8752 vs 0.8516, P=.04) and SVM (0.8800 vs 0.8557, P=.045). CONCLUSIONS: The study provides an example of the use of supervised machine learning methods to categorize cannabis- and synthetic cannabinoid-related tweets with fairly high accuracy. Use of these content analysis tools along with geographic identification capabilities developed by the eDrugTrends platform will provide powerful methods for tracking regional changes in user opinions related to cannabis and synthetic cannabinoids use over time and across different regions.

"Those edibles hit hard": Exploration of Twitter data on cannabis edibles in the U.S.

AIMS: Several states in the U.S. have legalized cannabis for recreational or medical uses. In this context, cannabis edibles have drawn considerable attention after adverse effects were reported. This paper investigates Twitter users' perceptions concerning edibles and evaluates the association edibles-related tweeting activity and local cannabis legislation. METHODS: Tweets were collected between May 1 and July 31, 2015, using Twitter API and filtered through the eDrugTrends/Twitris platform. A random sample of geolocated tweets was manually coded to evaluate Twitter users' perceptions regarding edibles. Raw state proportions of Twitter users mentioning edibles were ajusted relative to the total number of Twitter users per state. Differences in adjusted proportions of Twitter users mentioning edibles between states with different cannabis legislation status were assesed via a permutation test. RESULTS: We collected 100,182 tweets mentioning cannabis edibles with 26.9% (n=26,975) containing state-level geolocation. Adjusted percentages of geolocated Twitter users posting about edibles were significantly greater in states that allow recreational and/or medical use of cannabis. The differences were statistically significant. Overall, cannabis edibles were generally positively perceived among Twitter users despite some negative tweets expressing the unreliability of edible consumption linked to variability in effect intensity and duration. CONCLUSION: Our findings suggest that Twitter data analysis is an important tool for epidemiological monitoring of emerging drug use practices and trends. Results tend to indicate greater tweeting activity about cannabis edibles in states where medical THC and/or recreational use are legal. Although the majority of tweets conveyed positive attitudes about cannabis edibles, analysis of experiences expressed in negative tweets confirms the potential adverse effects of edibles and calls for educating edibles-naïve users, improving edibles labeling, and testing their THC content.

"Time for dabs": Analyzing Twitter data on marijuana concentrates across the U.S.

AIMS: Media reports suggest increasing popularity of marijuana concentrates ("dabs"; "earwax"; "budder"; "shatter; "butane hash oil") that are typically vaporized and inhaled via a bong, vaporizer or electronic cigarette. However, data on the epidemiology of marijuana concentrate use remain limited. This study aims to explore Twitter data on marijuana concentrate use in the U.S. and identify differences across regions of the country with varying cannabis legalization policies. METHODS: Tweets were collected between October 20 and December 20, 2014, using Twitter's streaming API. Twitter data filtering framework was available through the eDrugTrends platform. Raw and adjusted percentages of dabs-related tweets per state were calculated. A permutation test was used to examine differences in the adjusted percentages of dabs-related tweets among U.S. states with different cannabis legalization policies. RESULTS: eDrugTrends collected a total of 125,255 tweets. Almost 22% (n=27,018) of these tweets contained identifiable state-level geolocation information. Dabs-related tweet volume for each state was adjusted using a general sample of tweets to account for different levels of overall tweeting activity for each state. Adjusted percentages of dabs-related tweets were highest in states that allowed recreational and/or medicinal cannabis use and lowest in states that have not passed medical cannabis use laws. The differences were statistically significant. CONCLUSIONS: Twitter data suggest greater popularity of dabs in the states that legalized recreational and/or medical use of cannabis. The study provides new information on the epidemiology of marijuana concentrate use and contributes to the emerging field of social media analysis for drug abuse research.