IMPORTANCE The association between industry payments to physicians and prescribing rates of the brand-name medications that are being promoted is controversial. In the United States, industry payment data and Medicare prescribing records recently became publicly available. OBJECTIVE To study the association between physicians' receipt of industry-sponsored meals, which account for roughly 80% of the total number of industry payments, and rates of prescribing the promoted drug to Medicare beneficiaries. DESIGN, SETTING, AND PARTICIPANTS Cross-sectional analysis of industry payment data from the federal Open Payments Program for August 1 through December 31, 2013, and prescribing data for individual physicians from Medicare Part D, for all of 2013. Participants were physicians who wrote Medicare prescriptions in any of 4 drug classes: statins, cardioselective β-blockers, angiotensin-converting enzyme inhibitors and angiotensin-receptor blockers (ACE inhibitors and ARBs), and selective serotonin and serotonin-norepinephrine reuptake inhibitors (SSRIs and SNRIs). We identified physicians who received industry-sponsored meals promoting the most-prescribed brand-name drug in each class (rosuvastatin, nebivolol, olmesartan, and desvenlafaxine, respectively). Data analysis was performed from August 20, 2015, to December 15, 2015. EXPOSURES Receipt of an industry-sponsored meal promoting the drug of interest. MAIN OUTCOMES AND MEASURES Prescribing rates of promoted drugs compared with alternatives in the same class, after adjustment for physician prescribing volume, demographic characteristics, specialty, and practice setting. RESULTS A total of 279 669 physicians received 63 524 payments associated with the 4 target drugs. Ninety-five percent of payments were meals, with a mean value of less than $20. Rosuvastatin represented 8.8% (SD, 9.9%) of statin prescriptions; nebivolol represented 3.3% (7.4%) of cardioselective β-blocker prescriptions; olmesartan represented 1.6% (3.9%) of ACE inhibitor and ARB prescriptions; and desvenlafaxine represented 0.6% (2.6%) of SSRI and SNRI prescriptions. Physicians who received a single meal promoting the drug of interest had higher rates of prescribing rosuvastatin over other statins (odds ratio [OR], 1.18; 95% CI, 1.17-1.18), nebivolol over other β-blockers (OR, 1.70; 95% CI, 1.69-1.72), olmesartan over other ACE inhibitors and ARBs (OR, 1.52; 95% CI, 1.51-1.53), and desvenlafaxine over other SSRIs and SNRIs (OR, 2.18; 95% CI, 2.13-2.23). Receipt of additional meals and receipt of meals costing more than $20 were associated with higher relative prescribing rates. CONCLUSIONS AND RELEVANCE Receipt of industry-sponsored meals was associated with an increased rate of prescribing the brand-name medication that was being promoted. The findings represent an association, not a cause-and-effect relationship.

Key Points Question Can a prediction model for mortality in the intensive care unit be improved by using more laboratory values, vital signs, and clinical text in electronic health records? Findings In this cohort study of 101 196 patients in the intensive care unit, a machine learning–based model using all available measurements of vital signs and laboratory values, plus clinical text, exhibited good calibration and discrimination in predicting in-hospital mortality, yielding an area under the receiver operating characteristic curve of 0.922. Meaning Applying methods from machine learning and natural language processing to information already routinely collected in electronic health records, including laboratory test results, vital signs, and clinical free-text notes, significantly improves a prediction model for mortality in the intensive care unit compared with approaches that use only the most abnormal vital sign and laboratory values.

Importance Commercial virtual visits are an increasingly popular model of care for the management of common, acute illnesses. In commercial virtual visits, patients access a website to be connected synchronously—via videoconference, telephone, or webchat—to a physician with whom they have no prior relationship. There has been no assessment of whether the care delivered through those websites is similar, or whether quality varies among the sites. Objective To assess the variation in quality of care among virtual visit companies. Design We performed an audit study using trained standardized patients. Setting The standardized patients presented to commercial virtual visit companies with six common, acute illnesses (ankle pain, streptococcal pharyngitis, viral pharyngitis, acute rhinosinusitis, low back pain, and recurrent urinary tract infection). Participants The eight commercial virtual visit websites with the highest web traffic. Main Outcome Measures The primary outcomes were completeness of histories and physical examinations, naming the correct diagnosis (versus an incorrect diagnosis or not naming any diagnosis), and adherence to guidelines of key management decisions. Results Standardized patients completed 599 commercial virtual visits from May 2013 to July 2014. Histories and physical examinations were complete in 69.6% (95% confidence interval [CI], 67.7%-71.6%) of virtual visits, diagnoses were correctly named in 76.5% (CI, 72.9%-79.9%), and key management decisions were adherent to guidelines in 54.3% (CI, 50.2%-58.3%). Rates of guideline-adherent care ranged from 34.4% to 66.1% across the eight websites. Variation across websites was significantly greater for viral pharyngitis and acute rhinosinusitis (12.8-82.1%) than for streptococcal pharyngitis and low back pain (74.6-96.5%) or ankle pain and recurrent urinary tract infection (3.4-40.4%). There was no statistically significant variation in guideline adherence by mode of communication (video vs. telephone vs. webchat). Conclusions We found significant variation in quality among companies providing virtual visits for management of common acute illnesses. There was more variation in performance for some conditions than for others, but there was no variation by mode of communication.