Sublethal toxicities of 2,4-dinitrophenol as inferred from online self-reports.

INTRODUCTION: 2,4-dinitrophenol (DNP) is a mitochondrial toxin sometimes used as a weight loss agent. Reports of fatalities from DNP have been increasing since 2000, suggesting an increase in use. Our understanding of DNP toxicity in humans comes from reports to Poison Control and postmortem analyses, sources that are biased to more extreme presentations. This leads to a gap in our knowledge about the adverse effects of DNP at nonlethal doses. Here we investigate the doses and effects of DNP as reported online. METHODS: We analyzed publicly available Internet posts that we collected from 2017-2019. The posts came from anonymous users or users who voluntarily self-identified. We collected data from websites whose terms of use allow for the secondary analysis of data that their users agree to make public. We used natural language processing techniques that we had previously developed to extract doses, effects, and substances mentioned in each post. RESULTS: We collected 1,630 posts across 5 online forums and the Reddit forum r/DNP. The posts were from 1,234 unique usernames. The most commonly reported doses were between 150 to 300 mg each day followed by 300 to 450 mg each day. At those doses, the most reported adverse effects were profuse sweating and fatigue. Reports of thermoregulatory (sweating, feeling hot flashes or flushed), fatigue-related, and neurologically related symptoms were statistically significantly more frequent at reported daily doses greater than 150 mg than doses below 150 mg (post-hoc χ2-test with Bonferroni correction). The effects were judged as plausible by two board-certified medical toxicologists. Triiodothyronine, clenbuterol, testosterone, and trenbolone, an androgenic anabolic steroid were the most significantly co-mentioned substances. CONCLUSIONS: Fatigue, increased body temperature, and paresthesias from DNP are reported more frequently at doses greater than 150 mg each day than at doses less than 150 mg each day. Online discussions of DNP frequently mention androgenic anabolic steroids and other weight loss agents.

Fostering the Next Generation of Researchers: a Sustainable Mentoring Program for Early Career Toxicologists in Scientific Abstract Review.

The presentation of abstracts at scientific meetings is an important step in the dissemination of scientific discovery. Most scientific meetings recruit volunteer experts to evaluate and score submitted abstracts to determine which ones qualify for presentation. Reviewing an abstract is an important service to one's specialty, but there is typically no formal training or required instruction during medical toxicology fellowship on scientific abstract scoring. In order to provide structured training in abstract review, the American College of Medical Toxicology (ACMT) Research Committee launched the Annual Scientific Meeting (ASM) Abstract Review Mentor program in 2021. The goals of this program were to train fellows how to score scientific abstracts and provide them with new mentor connections to toxicologists outside of their training program. After evaluating 3 years of data from participating fellows-in-training and faculty mentors, we conclude that ACMT's Abstract Review Mentor program was successful in training future reviewers and fostering external mentorship relationships. All participants reported their experience in this program will change how they submit future abstracts to scientific meetings, help their future service as an abstract reviewer, and motivate their involvement in other specialty-related research activities. Implementing an abstract review training program is sustainable and a vital strategy for enhancing the dissemination of scientific discovery and training the next generation of medical toxicology researchers.

COVID-19 Therapeutics: Use, Mechanism of Action, and Toxicity (Vaccines, Monoclonal Antibodies, and Immunotherapeutics).

SARS-CoV-2 emerged in December 2019 and led to the COVID-19 pandemic. Efforts to develop therapeutics have led to innovations such as mRNA vaccines and oral antivirals. Here we provide a narrative review of the biologic therapeutics used or proposed to treat COVID-19 during the last 3 years. This paper, along with its companion that covers xenobiotics and alternative remedies, is an update to our 2020 paper. Monoclonal antibodies prevent progression to severe disease, are not equally effective across variants, and are associated with minimal and self-limited reactions. Convalescent plasma has side effects like monoclonal antibodies, but with more infusion reactions and less efficacy. Vaccines prevent progression for a larger part of the population. DNA and mRNA vaccines are more effective than protein or inactivated virus vaccines. After mRNA vaccines, young men are more likely to have myocarditis in the subsequent 7 days. After DNA vaccines, those aged 30-50 are very slightly more likely to have thrombotic disease. To all vaccines we discuss, women are slightly more likely to have an anaphylactic reaction than men, but the absolute risk is small.

COVID-19 Therapeutics: Use, Mechanism of Action, and Toxicity (Xenobiotics).

SARS-CoV-2 emerged in 2019 and led to the COVID-19 pandemic. Efforts to develop therapeutics against SARS-Cov-2 led to both new treatments and attempts to repurpose existing medications. Here, we provide a narrative review of the xenobiotics and alternative remedies used or proposed to treat COVID-19. Most repositioned xenobiotics have had neither the feared toxicity nor the anticipated efficacy. Repurposed viral replication inhibitors are not efficacious and frequently associated with nausea, vomiting, and diarrhea. Antiviral medications designed specifically against SARS-CoV-2 may prevent progression to severe disease in at-risk individuals and appear to have a wide therapeutic index. Colloidal silver, zinc, and ivermectin have no demonstrated efficacy. Ivermectin has a wide therapeutic index but is not efficacious and acquiring it from veterinary sources poses additional danger. Chloroquine has a narrow therapeutic index and no efficacy. A companion review covers vaccines, monoclonal antibodies, and immunotherapies. Together, these two reviews form an update to our 2020 review.

Translator use not associated with longer time to pain medication in initial evaluation of low-severity geriatric trauma.

OBJECTIVES: Determine whether geriatric victims of blunt trauma who preferred to communicate in a language other than English waited longer for pain medication or received more imaging studies than English-speaking patients with the same age and injuries. Secondary outcomes were the type of medication administered and number of imaging studies. METHODS: We conducted a retrospective analysis of all trauma activations to a single academic urban medical center from January 2019 to October 2019. We included all hemodynamically stable patients older than 65, with head or torso trauma after a low energy injury, and on at least one medication that was an anti-coagulant, anti-platelet, or chemotherapeutic. RESULTS: We identified 1,153 unique patients (17, 379 radiologic studies) performed from January 2019 to October 2019, with a median of 5 (4-6) radiologic studies per patient. We excluded 419 patients for whom the language used was not reported (n = 7), no imaging was not reported (n = 16), or no medication was recorded as given (n = 409), leaving 734 patients for further analysis. Of those 734 patients, 460 preferred to communicate in English, 84 in Mandarin Chinese, 64 in Spanish, 37 in Cantonese Chinese, and 35 in Korean, and 29 in Russian. Across all languages patient age and Injury Severity Score (ISS) were comparable. Those who preferred to communicate in Spanish, Russian, or Korean were more likely to be female than those who preferred English, Mandarin, or Cantonese, but this tendency was not statistically significant (χ2-test; p = 0.051, 0.15 after Bonferroni correction). We did not find a statistically significant association between preferred language and time to medication, fraction of opioids used as first-line pain medication, or number of imaging studies performed. Across all patients, the most common medications administered were acetaminophen (524/734, 71%), any opioid (111/734, 15%), followed by local infiltration or nerve block with lidocaine (49/734, 6.7%). CONCLUSIONS: A retrospective analysis of patients with low-risk blunt trauma found no relationship between preferred language, time to pain medication, use of opioids or number of imaging studies.

Efficacy of lipid emulsion therapy in treating cardiotoxicity from diphenhydramine ingestion: a review and analysis of case reports.

INTRODUCTION: Lipid emulsion therapy (LET) has been most thoroughly studied to reverse local anesthetic systemic toxicity (LAST). Case reports suggest that LET can successfully rescue cardiovascular collapse from bupropion, amitriptyline, and propranolol. The efficacy of LET against refractory hypotension and dysrhythmias from diphenhydramine, a commonly ingested lipophilic cardiotoxic agent, is less well described. OBJECTIVE: Summarize the evidence that LET rescues cardiac ion channel blockade (QRS, QTc widening) or hypotension attributable to diphenhydramine overdose. METHODS: We searched MEDLINE, EMBASE, and Google Scholar for English-language full-length case reports of diphenhydramine (DPH) intoxication in patients 17 years of age or older. We extracted data with a PRISMA-compliant protocol, dividing the case reports into two groups, one that received LET and one that did not. We performed a pooled analysis to compare the change in mean arterial pressure (MAP), QRS duration, and QTc duration between the two groups. RESULTS: We identified 23 reports (25 patients). Lipid emulsion therapy (LET) was used in 6 cases because the patient suffered from hypotension refractory to traditional resuscitation. Those who received LET and those who did not were comparable in age, gender, amount ingested, and frequency of seizures. The mean arterial pressure (MAP) decreased by 4.5 ± 11.5 mm Hg in those who did not receive LET compared to an increase in MAP 37 ± 17.5 mm Hg in those who did receive LET. The QRS narrowed by 29 ± 33.9 ms (no LET group) vs 68 ± 49.5 ms (LET group) and QTc by 168.5 ± 126.75 ms (no LET group) vs 134 ± 88 ms (LET group). All values are expressed as median ± interquartile range. One out of the 6 patients who received LET died after withdrawal of care. In the group that did not receive LET 4 out of 19 died and 3 had no outcome reported. DISCUSSION: LET may improve MAP in patients with hypotension refractory to vasopressors due to diphenhydramine toxicity. We found no significant effect of LET on QRS or QTc duration. These results are limited by a small sample size, reporting bias of case reports, incomplete data, and heterogeneity. CONCLUSION: An analysis of pooled case reports suggests that LET may rescue hypotension when other methods have failed in patients with hypotension attributable to diphenhydramine overdose.

Diagnosis of Acute Poisoning using explainable artificial intelligence.

INTRODUCTION: Medical toxicology is the clinical specialty that treats the toxic effects of substances, for example, an overdose, a medication error, or a scorpion sting. The volume of toxicological knowledge and research has, as with other medical specialties, outstripped the ability of the individual clinician to entirely master and stay current with it. The application of machine learning/artificial intelligence (ML/AI) techniques to medical toxicology is challenging because initial treatment decisions are often based on a few pieces of textual data and rely heavily on experience and prior knowledge. ML/AI techniques, moreover, often do not represent knowledge in a way that is transparent for the physician, raising barriers to usability. Logic-based systems are more transparent approaches, but often generalize poorly and require expert curation to implement and maintain. METHODS: We constructed a probabilistic logic network to model how a toxicologist recognizes a toxidrome, using only physical exam findings. Our approach transparently mimics the knowledge representation and decision-making of practicing clinicians. We created a library of 300 synthetic cases of varying clinical complexity. Each case contained 5 physical exam findings drawn from a mixture of 1 or 2 toxidromes. We used this library to evaluate the performance of our probabilistic logic network, dubbed Tak, against 2 medical toxicologists, a decision tree model, as well as its ability to recover the actual diagnosis. RESULTS: The inter-rater reliability between Tak and the consensus of human raters was κ = 0.8432 for straightforward cases, 0.4396 for moderately complex cases, and 0.3331 for challenging cases. The inter-rater reliability between the decision tree classifier and the consensus of human raters was, κ = 0.2522 for straightforward cases, 0.1963 for moderately complex cases and 0.0331 for challenging cases. CONCLUSIONS: The software, dubbed Tak, performs comparably to humans on straightforward cases and intermediate difficulty cases, but is outperformed by humans on challenging clinical cases. Tak outperforms a decision tree classifier at all levels of difficulty. Our results are a proof-of-concept that, in a restricted domain, probabilistic logic networks can perform medical reasoning comparably to humans.

Geospatial correlation between COVID-19 health misinformation and poisoning with household cleaners in the Greater Boston Area.

INTRODUCTION: Calls to poison control about exposure to household cleaners have increased during the COVID-19 pandemic. This dynamic may reflect increased exposure from public health efforts as well as health misinformation shared on social media. METHODS: We analyzed the dynamics of calls to the Regional Center for Poison Control and Prevention serving Massachusetts and Rhode Island (MARI PCC) and tweets discussing treating COVID-19 with house cleaners from January 20, 2020 to May 5, 2020. We obtained publicly available tweets discussing the use of household cleaners to "cure COVID" from the same time period with geographic co-ordinates indicating that they were emitted from the Greater Boston Area. RESULTS: Our main finding is that public health efforts were followed by a sustained increase in calls after March 15, 2020 (10 ± 2 calls per day before to 15 ± 2.5 after) while misinformation on social media was associated with intermittent spikes in calls. Overall, calls significantly increased during the study period by 34% as compared to the previous 8 years, mostly reporting unintentional ingestions with no serious effects. The daily volume of tweets and retweets was significantly correlated with daily call rates to MARI PCC for the surrounding 7-10 days. CONCLUSIONS: Health misinformation on social media about using household cleaning agents to treat COVID-19 and public health efforts lead to different dynamics in PCC calls. Public health efforts were followed by a sustained increase in calls after March 15, 2020 while misinformation on social media was followed by intermittent spikes in calls. This analysis is the first to link the geospatial dynamics of social media and public health interventions to poison center calls about exposure to household cleaners.

COVID-19: Therapeutics and Their Toxicities.

SARS-CoV-2 is a novel coronavirus that emerged in 2019 and is causing the COVID-19 pandemic. There is no current standard of care. Clinicians need to be mindful of the toxicity of a wide variety of possibly unfamiliar substances being tested or repurposed to treat COVID-19. The United States Food and Drug Administration (FDA) has provided emergency authorization for the use of chloroquine and hydroxychloroquine. These two medications may precipitate ventricular dysrhythmias, necessitating cardiac and electrolyte monitoring, and in severe cases, treatment with epinephrine and high-doses of diazepam. Recombinant protein therapeutics may cause serum sickness or immune complex deposition. Nucleic acid vaccines may introduce mutations into the human genome. ACE inhibitors and ibuprofen have been suggested to exacerbate the pathogenesis of COVID-19. Here, we review the use, mechanism of action, and toxicity of proposed COVID-19 therapeutics.

The Role and Promise of Artificial Intelligence in Medical Toxicology.

Artificial intelligence (AI) refers to machines or software that process information and interact with the world as understanding beings. Examples of AI in medicine include the automated reading of chest X-rays and the detection of heart dysrhythmias from wearables. A key promise of AI is its potential to apply logical reasoning at the scale of data too vast for the human mind to comprehend. This scaling up of logical reasoning may allow clinicians to bring the entire breadth of current medical knowledge to bear on each patient in real time. It may also unearth otherwise unreachable knowledge in the attempt to integrate knowledge and research across disciplines. In this review, we discuss two complementary aspects of artificial intelligence: deep learning and knowledge representation. Deep learning recognizes and predicts patterns. Knowledge representation structures and interprets those patterns or predictions. We frame this review around how deep learning and knowledge representation might expand the reach of Poison Control Centers and enhance syndromic surveillance from social media.

Incidence of Pediatric Cannabis Exposure Among Children and Teenagers Aged 0 to 19 Years Before and After Medical Marijuana Legalization in Massachusetts.

Importance: Pediatric health care contacts due to cannabis exposure increased in Colorado and Washington State after cannabis (marijuana) policies became more liberal, but evidence from other US states is limited. Objective: To document the incidence of pediatric cannabis exposure cases reported to the Regional Center for Poison Control and Prevention (RPC) before and after medical marijuana legalization (MML) in Massachusetts. Design, Setting, and Participants: Cross-sectional comparison of pediatric cannabis exposure cases 4 years before and after MML in Massachusetts. The exposure cases included those of 218 children and teenagers aged between 0 and 19 years, as reported to the RPC from 2009 to 2016. Census data were used to determine the incidence. Data analysis was performed from November 12, 2018, to July 20, 2019. Exposure: Cannabis products. Main Outcomes and Measures: Incidence of RPC-reported cannabis exposure cases, both single substance and polysubstance, for the age group of 0 to 19 years, and cannabis product type, coingestants, and clinical effects. Results: During the 8-year study period (2009-2016), the RPC received 218 calls involving cannabis exposure (98 single substance, 120 polysubstance) in children and teenagers aged 0 to 19 years, representing 0.15% of all RPC calls in that age group for that period. Of the total exposure cases, males accounted for 132 (60.6%) and females 86 (39.4%). The incidence of single-substance cannabis calls increased from 0.4 per 100 000 population before MML to 1.1 per 100 000 population after (incidence rate ratio, 2.4; 95% CI, 1.5-3.9), a 140% increase. The age group of 15 to 19 years had the highest frequency of RPC-reported cannabis exposures (178 calls [81.7%]). The proportion of all RPC calls due to single-substance cannabis exposure increased overall for all age groups from 29 before MML to 69 afterward. Exposure to edible products increased after MML for most age groups. Conclusions and Relevance: Pediatric cannabis exposure cases increased in Massachusetts after medical marijuana was legalized in 2012, despite using childproof packaging and warning labels. This study provides additional evidence suggesting that MML may be associated with an increase in cannabis exposure cases among very young children, and extends prior work showing that teenagers are also experiencing increased cannabis-related health system contacts via the RPC. Additional efforts are needed to keep higher-potency edible products and concentrated extracts from children and teenagers, especially considering the MML and retail cannabis sales in an increasing number of US states.

A Review of Natural Language Processing in Medical Education.

Natural language processing (NLP) aims to program machines to interpret human language as humans do. It could quantify aspects of medical education that were previously amenable only to qualitative methods. The application of NLP to medical education has been accelerating over the past several years. This article has three aims. First, we introduce the reader to NLP. Second, we discuss the potential of NLP to help integrate FOAM (Free Open Access Medical Education) resources with more traditional curricular elements. Finally, we present the results of a systematic review. We identified 30 articles indexed by PubMed as relating to medical education and NLP, 14 of which were of sufficient quality to include in this review. We close by discussing potential future work using NLP to advance the field of medical education in emergency medicine.

Candyflipping and Other Combinations: Identifying Drug-Drug Combinations from an Online Forum.

Novel psychoactive substances (NPS) refer to synthetic compounds or derivatives of more widely known substances of abuse that have emerged over the last two decades. Case reports suggest that users combine substances to achieve desired psychotropic experiences while reducing dysphoria and unpleasant somatic effects. However, the pattern of combining NPS has not been studied on a large scale. Here, we show that posts discussing NPS describe combining nootropics with sedative-hypnotics and stimulants with plant hallucinogens or psychiatric medications. Discussions that mention sedative-hypnotics most commonly also mention hallucinogens and stimulants. We analyzed 20 years of publicly available posts from Lycaeum, an Internet forum dedicated to sharing information about psychoactive substance use. We used techniques from natural language processing and machine learning to identify NPS and correlate patterns of co-mentions of substances across posts. We found that conversations mentioning synthetic hallucinogens tended to divide into those mentioning hallucinogens derived from amphetamine and those derived from ergot. Conversations that mentioned synthetic hallucinogens tended not to mention plant hallucinogens. Conversations that mention bath salts commonly mention sedative-hypnotics or nootropics while more canonical stimulants are discussed with plant hallucinogens and psychiatric medications. All types of substances are frequently compared to MDMA, DMT, cocaine, or atropine when trying to describe their effects. Our results provide the largest analysis to date of online descriptions of patterns of polysubstance use and further demonstrate the utility of social media in learning about trends in substance use. We anticipate this work to lead to a more detailed analysis of the knowledge contained online about the patterns of usage and effects of novel psychoactive substances.

Epidemiology from Tweets: Estimating Misuse of Prescription Opioids in the USA from Social Media.

BACKGROUND: The misuse of prescription opioids (MUPO) is a leading public health concern. Social media are playing an expanded role in public health research, but there are few methods for estimating established epidemiological metrics from social media. The purpose of this study was to demonstrate that the geographic variation of social media posts mentioning prescription opioid misuse strongly correlates with government estimates of MUPO in the last month. METHODS: We wrote software to acquire publicly available tweets from Twitter from 2012 to 2014 that contained at least one keyword related to prescription opioid use (n = 3,611,528). A medical toxicologist and emergency physician curated the list of keywords. We used the semantic distance (SemD) to automatically quantify the similarity of meaning between tweets and identify tweets that mentioned MUPO. We defined the SemD between two words as the shortest distance between the two corresponding word-centroids. Each word-centroid represented all recognized meanings of a word. We validated this automatic identification with manual curation. We used Twitter metadata to estimate the location of each tweet. We compared our estimated geographic distribution with the 2013-2015 National Surveys on Drug Usage and Health (NSDUH). RESULTS: Tweets that mentioned MUPO formed a distinct cluster far away from semantically unrelated tweets. The state-by-state correlation between Twitter and NSDUH was highly significant across all NSDUH survey years. The correlation was strongest between Twitter and NSDUH data from those aged 18-25 (r = 0.94, p < 0.01 for 2012; r = 0.94, p < 0.01 for 2013; r = 0.71, p = 0.02 for 2014). The correlation was driven by discussions of opioid use, even after controlling for geographic variation in Twitter usage. CONCLUSIONS: Mentions of MUPO on Twitter correlate strongly with state-by-state NSDUH estimates of MUPO. We have also demonstrated that a natural language processing can be used to analyze social media to provide insights for syndromic toxicosurveillance.

Differences in Self-expression Reflect Formal Evaluation in a Fourth-year Emergency Medicine Clerkship.

INTRODUCTION: Medical schools have begun to incorporate self-reflection exercises into their curricula, with the belief that these exercises help students master the material more deeply and perform better. Reflection may be a potential learning tool for emergency medicine (EM), but there are few data supporting this hypothesis. The authors evaluated the relationship between a linguistic marker of the degree of reflection after a student's shift in an emergency department and that student's clerkship performance. METHODS: The authors conducted a retrospective case series by analyzing the performance and reflective statements of 116 students from a single medical school who participated in a required EM clerkship at one or two of four clinical sites from 2013-14. After each shift, an attending emergency physician evaluated the student according to the RIME (Reporter-Interpreter-Manager-Educator) scheme. The authors developed software to extract the text from those comments, remove uninformative words and standardize the remaining words. The authors determined the most common words and two-word phrases that students used to describe their shift. The correlation between students' final clerkship grades and the fraction of student comments with at least one content word was analyzed. RESULTS: Of the 145 possible students, 116 were included for analysis. The other 29 were excluded as they were visiting students who did not receive a final numeric grade. The correlation between final grade and the number of completed self-reflections was 0.32. The correlation between final grade and the average number of words in each self-reflection was 0.21. The first correlation is significantly greater than 0 (p=0.03, t-test), but the second correlation is not (p=0.16, t-test). The median final grade of those who wrote reflections on more than half of their shifts was significantly greater than those who wrote reflections half of the time, 83.675 versus 79.23 (p=0.05, 2-sample Kolmogorov-Smirnov test). CONCLUSION: Students who reflected more frequently received a higher grade in an EM clerkship for fourth-year medical students. The number of words in each reflection was not significantly correlated with grade performance. The most common words and phrases students wrote were associated with learning and managing patients.

"Irony" of managing refractory anemia with transfusional support in hemophagocytic lymphohistiocytosis.

Hemophagocytic lymphohistiocytosis (HLH) is a life-threatening derangement of the immune system in which host macrophages phagocytose the patient's own blood cells. Herein, we present the case of a patient with HLH and associated refractory anemia who developed rapid iron deposition in the liver after transfusion of sixteen units of packed red blood cells (RBCs). Before transfusion, neither a liver biopsy nor computed tomography scan demonstrated iron deposition in the organ parenchyma. After receiving sixteen units of packed RBCs, liver iron concentration rose to 6.7 mg/g dry weight, which is highly unusual in other diseases requiring transfusional support.

Operator experience determines performance in a simulated computer-based brain tumor resection task.

PURPOSE: Develop measures to differentiate between experienced and inexperienced neurosurgeons in a virtual reality brain surgery simulator environment. METHODS: Medical students (n = 71) and neurosurgery residents (n = 12) completed four simulated Glioblastoma multiforme resections. Simulated surgeries took place over four days with intermittent spacing in between (average time between surgeries of 4.77 ± 0.73 days). The volume of tumor removed (cc), volume of healthy brain removed (cc), and instrument path length (mm) were recorded. Additionally, surgical effectiveness (% tumor removed divided by % healthy brain removed) and efficiency (% tumor removed divided by instrument movement in mm) were calculated. Performance was compared (1) between groups, and (2) for each participant over time to assess the learning curve. In addition, the effect of real-time instruction ("coaching") was assessed with a randomly selected group of medical students. RESULTS: Neurosurgery residents removed less healthy brain, were more effective in removing tumor and sparing healthy brain tissue, required less instrument movement, and were more efficient in removing tumor tissue than medical students. Medical students approached the resident level of performance over serial sessions. Coached medical students showed more conservative surgical behavior, removing both less tumor and less healthy brain. In sum, neurosurgery residents removed more tumor, removed less healthy brain, and required less instrument movement than medical students. Coaching modified medical student performance. CONCLUSIONS: Virtual Reality brain surgery can differentiate operators based on both recent and long-term experience and may be useful in the acquisition and assessment of neurosurgical skills. Coaching alters the learning curve of naïve inexperienced individuals.

Analysis of twitter users' sharing of official new york storm response messages.

BACKGROUND: Twitter is a social network where users read, send, and share snippets of text ("tweets"). Tweets can be disseminated through multiple means; on desktop computers, laptops, and mobile devices, over ethernet, Wi-Fi or cellular networks. This redundancy positions Twitter as a useful tool for disseminating information to the public during emergencies or disasters. Previous research on dissemination of information using Twitter has mostly investigated the characteristics of tweets that are most effective in raising consumer awareness about a new product or event. In particular, they describe characteristics that increase the chance the messages will be shared ("retweeted") by users. In comparison, little has been published on how information from municipal or state government agencies spreads on Twitter during emergency situations. Retweeting these messages is a way to enhance public awareness of potentially important instructions from public officials in a disaster. OBJECTIVE: The aim of this study is to (1) describe the tweets of select New York State and New York City agencies by public officials surrounding two notable recent winter storms that required a large-scale emergency response, and (2) identify the characteristics of the tweets of public officials that were most disseminated (retweeted). METHODS: For one week surrounding Superstorm Sandy (October 2012) and the winter blizzard Nemo (February 2013), we collected (1) tweets from the official accounts for six New York governmental agencies, and (2) all tweets containing the hashtags #sandy (or #nemo) and #nyc. From these data we calculated how many times a tweet was retweeted, controlling for differences in baseline activity in each account. We observed how many hashtags and links each tweet contained. We also calculated the lexical diversity of each tweet, a measure of the range of vocabulary used. RESULTS: During the Sandy storm, 3242 shared (retweeted) messages from public officials were collected. The lexical diversity of official tweets was similar (2.25-2.49) and well below the average for non-official tweets mentioning #sandy and #nyc (3.82). Most official tweets were with substantial retweets including a link for further reading. Of the 448 tweets analyzed from six official city and state Twitter accounts from the Nemo blizzard, 271 were related to the storm, and 174 had actionable information for the public. Actionable storm messages were retweeted approximately 24x per message, compared to 31x per message for general storm information. CONCLUSIONS: During two weather emergencies, New York public officials were able to convey storm-related information that was shared widely beyond existing follower bases, potentially improving situational awareness and disaster response. Official Sandy tweets, characterized by a lower lexical diversity score than other city- and Sandy-related tweets, were likely easier to understand, and often linked to further information and resources. Actionable information in the Nemo blizzard, such as specific instructions and cancellation notices, was not shared as often as more general warnings and "fun facts," suggesting agencies mix important instructions with more general news and trivia, as a way of reaching the broadest audience during a disaster.