Social media is being increasingly utilized by learners in community and academic practice. Among emergency medicine (EM) podcasts, the most highly utilized is EM:Rap. It currently has nearly 20,000 listeners, a significant proportion of the estimated 45,000 practicing emergency physicians in the United States. This study seeks to identify whether there is any association between topic discussion on EM:Rap and the adoption of that topic in clinical practice. For the pilot study, we measured the impact of EM:Rap discussion on the diagnosis of (1) cannabinoid hyperemesis syndrome (CHS), and the performance of emergency department (ED) cardioversion for atrial fibrillation or atrial flutter (EDCAF), as measured on prevalence in the National Emergency Department Sample (NEDS), a nationally representative data set of ED visits. This study was an observational case/control study of ED visits between 2008 and 2013. To analyze CHS, cases were defined as ED visits with nausea or vomiting and a code consistent with cannabis abuse in patients over age 17, based on ICD-9 codes. A time period was defined from 18 months prior to the first report in the medical literature to 18 months after the EM:Rap podcast was released. The primary analytic was the proportion of patients with a diagnosis of nausea or vomiting who also had a diagnosis of cannabis use. We utilized modified sharp regression discontinuity design, accounting for weighting. These data were examined graphically in each time period. Next, a multivariable generalized estimating equations regression model was built using a logit link function and an exchangeable correlation matrix for patient-level data considering only cases of nausea or vomiting from the data set clustered on the treating hospital, including an interaction term between time and time epoch as our primary analytic. To analyze EDCAF, we conducted a similar analysis of adults with a diagnosis of atrial fibrillation, and the dependent variable was a procedure code consistent with cardioversion. We performed an unweighted analysis of 180 million ED visits from the period of February 2008 - January 2013. Over time, there was a linear increase in the rate of CHS diagnosis that did not change significantly in the period after it was discussed on EM:Rap (p=0.2091). Similarly, we found no significant difference in the rate of EDCAF performance in the epoch after it was discussed on EM:Rap (p=0.2900). Despite its significant penetration among ED providers, discussion of a diagnosis (CHS) or a new treatment option (EDCAF) on EM:Rap did not result in any measurable aggregate change in practice, according to a nationally representative sample of claims data. We hypothesize these findings could reflect that topics in social media reflect changes that are already occurring in practice, but that they are not causal in speeding adoption of new diagnostic or treatment alternatives. These topics may be more reflective of current ED practice patterns, rather than being a driver or change in practice.
Read full abstract