OBJECTIVESAlthough proficient systems-based practice is a foundational skill for physicians, how best to teach it has not been well established. An elective course for fourth-year medical students wherein participants had an immersive experience with multiple interprofessional staff was created and analyzed. The authors hypothesized that participating students and interprofessional staff would show gains in systems-based knowledge and interprofessional communication.METHODSThe course was a 2-week elective experience for fourth-year medical students at the Larner College of Medicine at the University of Vermont, Burlington, VT, USA. Participants integrated into a variety of interprofessional, non-physician, and administrative roles within the hospital system. Pre- and post-elective systems-based knowledge and interprofessional communication were assessed. Participating interprofessional staff were also surveyed on their experiencesRESULTSFrom 2019 through 2022, 14 students participated in the elective, all of whom provided data. All participating students showed a quantitative improvement in systems-based knowledge and qualitatively commented on the high value of the elective in furthering their understanding of interdisciplinary care and communication. Of the 22 participating interprofessional staff surveyed, 17 responded (response rate 77%), and data showed high satisfaction with the experience and that having students learn more about their jobs improved their own job satisfaction.CONCLUSIONSAn immersive, hands-on experience with interprofessional colleagues showed dual benefits for both students and staff alike. Such an elective experience is scalable to other institutions nationally and should become a standard part of medical student curricula.

Abstract Introduction Clinical departments at academic medical centers strive to deliver clinical care, provide education and training, support faculty development, and promote scholarship. These departments have experienced increasing demands to improve the quality, safety, and value of care delivery. However, many academic departments lack a sufficient number of clinical faculty members with expertise in improvement science to lead initiatives, teach, and generate scholarship. In this article, we describe the structure, activities, and early outcomes of a program within an academic department of medicine to promote scholarly improvement work. Methods The Department of Medicine at the University of Vermont Medical Center launched a Quality Program with three primary goals: (a) improve care delivery, (b) provide education and training, and (c) promote scholarship in improvement science. The program serves as a resource center for students, trainees and faculty, offering education and training, analytic support, consultation in design and methodology, and project management. It strives to integrate education, research, and care delivery to learn, apply evidence and improve health care. Results Over the first 3 years of full implementation, the Quality Program supported an average of 123 projects annually, including prospective clinical quality improvement initiatives, retrospective assessment of clinical programs and practices, and curriculum development and evaluation. The projects have yielded a total of 127 scholarly products, defined as peer‐reviewed publications and abstracts, posters, and oral presentations at local, regional, and national conferences. Conclusions The Quality Program may serve as a practical model for promoting care delivery improvement, training, and scholarship in improvement science while advancing the goals of a learning health system at the level of an academic clinical department. Dedicated resources within such departments offer the potential to enhance care delivery while promoting academic success for faculty and trainees in improvement science.

Acute respiratory distress syndrome (ARDS), originally described in 1967, affects more than 3 million individuals each year throughout the world and accounts for approximately 10% of all admissions to the intensive care unit. Despite substantial progress in defining the epidemiology and pathogenesis of the syndrome, there is no specific treatment and mortality rates remain high. Barriers to finding specific therapeutic interventions include the inability to predict who will get ARDS, inadequate definitions and specific diagnostic markers, the heterogeneity of the patient population, complexities of the pathogenesis, and the impact of clinical care. Measurements of biomarkers have identified these barriers as well as contributed to the current understanding of the disease. The COVID-19 pandemic resulted in a dramatic increase in patients with ARDS, driving an urgent need to understand the pathogenesis and develop and implement therapeutic interventions. Past studies of biomarkers in ARDS can provide insight that could help to meet those needs more rapidly.

Background: Pharmacokinetics (PK) of pharmaceuticals and pharmaconutrients are poorly understood in critically ill patients, and dosing is often based on healthy subject data. This might be particularly problematic with enteral medications due to metabolic abnormalities and impaired gastrointestinal tract absorption common in critically ill patients. Utilizing enteral fish oil, this study was undertaken to better understand and define PK of enteral omega-3 fatty acids (eicospentaenoic acid [EPA] and docosahexaenoic acid [DHA]) in critically ill patients with severe sepsis. Materials and methods: Healthy volunteers (n = 15) and mechanically ventilated (MV) adults with severe sepsis (n = 10) were recruited and received 9.75 g EPA and 6.75 g DHA daily in two divided enteral doses of fish oil for 7 days. Volunteers continued their normal diet without other sources of fish oil, and sepsis patients received standard enteral feeding. Blood was collected at frequent intervals during the 14-day study period. Peripheral blood mononuclear cells (PMBCs) and neutrophils were isolated and analyzed for membrane fatty acid (FA) content. Mixed linear models and t-tests were used to analyze changes in FA levels over time and FA levels at individual time points, respectively. PK parameters were obtained based on single compartment models of EPA and DHA kinetics. Results: Healthy volunteers were 41.1 +/- 10.3 years; 67% were women. In patients with severe sepsis (55.6 +/- 13.4 years, 50% women), acute physiologic and chronic health evaluation (APACHE) II score was 27.2 +/- 8.8 at ICU admission and median MV duration was 10.5 days. Serum EPA and DHA were significantly lower in sepsis vs. healthy subjects over time. PBMC EPA concentrations were generally not different between groups over time, while PBMC DHA was higher in sepsis patients. Neutrophil EPA and DHA concentrations were similar between groups. The half-life of EPA in serum and neutrophils was significantly shorter in sepsis patients, whereas other half-life parameters did not vary significantly between healthy volunteers and sepsis patients. Conclusions: While incorporation of n-3 FAs into PBMC and neutrophil membranes was relatively similar between healthy volunteers and sepsis patients receiving identical high doses of fish oil for one week, serum EPA and DHA were significantly lower in sepsis patients. These findings imply that serum concentrations and EPA and DHA may not be the dominant driver of leukocyte membrane incorporation of EPA and DHA. Furthermore, lower serum EPA and DHA concentrations suggest that either these n-3 FAs were being metabolized rapidly in sepsis patients or that absorption of enteral medications and pharmaconutrients, including fish oil, may be impaired in sepsis patients. If enteral absorption is impaired, doses of enteral medications administered to critically ill patients may be suboptimal. (C) 2019 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved.

In this issue of the Journal (pp. 840–847), Vranas and colleagues examine gender differences in authorship of critical care literature (1). They describe several findings of gender disparity in authorship, including low rates of first and senior authorship, a tendency toward publication in lower-impact journals than those in which male authors are published, and slow growth in female authorship over the last decade (1). These findings add to a growing list of examples of the persistence of gender disparity not only in critical care medicine but also in academic medicine as a whole (2, 3). The authors acknowledge several limitations, including potential confounding by journal and search term selection, gender misclassification, author bias, and lack of a denominator of female academic intensivists worldwide. Although recognizing these limitations, Vranas and colleagues report findings that echo trends reported across a multitude of medical specialties (4–9). The underrepresentation of female authors in academic medicine is of significant concern, given the importance of publication in both the growth of academic careers and furthering the science of medicine. This phenomenon has been described as “both a reflection and a cause of women’s under-representation and disadvantage in other areas of the scientific enterprise” (10), leading to a “vicious circle” of reduced research funding and publication rates (3). It is important to continue to investigate this trend, not only to avoid losing sight of this issue but also to help us get to the root cause and ultimately develop strategies to improve the presence of women in academic publishing. Vranas and colleagues keep this issue at the forefront with their present study while also being the first, to our knowledge, to describe this phenomenon in our own discipline. The authorship gender disparity in critical care literature may in part be explained by the continued male predominance of the specialty. Although the gender gap in medical school admissions and matriculations closed in the last two decades, many specialties remain predominantly male (11). A 2018 study found that only 37% of members of critical care professional societies were women. In addition, despite comprising about one-third of the critical care workforce internationally, women remain underrepresented in leadership positions, including society presidents, board and council members, and symposia chairs (12). Other reports have noted that although in some regions women may actually represent closer to 50% of practicing intensivists, they continue to be less likely to chair committees, serve on editorial and guideline boards, or hold academic leadership positions (13, 14). Authorship disparity not only correlates with but also may partially explain the paucity of women among academic leaders. Individuals in leadership positions often have strong research funding histories (4–7, 9). Fewer publications among women may in turn lead to lower rates of promotion and less likelihood of successfully securing funding, thus reducing advancement into leadership roles and again highlighting the vicious cycle associated with reduced publication and academic productivity (3). Although not specifically addressed in the report by Vranas and colleagues, studies of gender authorship disparity in other medical specialties have noted additional concerning trends. For example, orthopedic surgery literature has reported that although female authorship has increased substantially over the last several decades, women still publish fewer papers per author than their male counterparts do (15), and female authors were less likely than male authors to continue to publish 5 years after their original publication (15). These findings raise concern for reduced academic productivity and lower rates of retention in academic medicine among female researchers. It is yet to be seen whether this trend exists in academic intensive care medicine. Although the gender authorship disparity may be improving, Vranas and colleagues found slow rates of increase in overall, first, and senior female authorship (1) from 2008 to 2018. This finding has been echoed in other medical and surgical specialties and correlates with slow rates of growth of women in academic leadership positions (2, 4, 6, 10, 16). Given the close relationship between publication and academic advancement, these trends are likely to remain closely linked. To advance women in leadership roles and academia in general, it is crucial to increase female contributions to the body of scientific literature. The findings of this and other studies of female authorship disparity should not be viewed solely in a negative light. Although the rates of growth are slow in our specialty and many others, female authorship is growing, and the gender gap is narrowing. Recognizing a disparity is the first step toward devising strategies to improve it. Many studies, including the one presented by Vranas and colleagues, have delved further into the data to identify trends that may help craft a solution to increase female representation in academic authorship and leadership. This study and several others observed higher rates of female first authorship when a senior author is also female (5). These findings highlight the importance of strong female mentorship in the recruitment, promotion, and retention of young women in academic medicine (2, 10). Further studies are needed to determine the best strategies to promote successful female mentor–mentee relationships. In addition to identifying the need for strong mentorship, studies of gender publication trends have observed that male authors tend to use positive language more frequently than female authors in publications describing their research (17). This difference in rates of positive language use were greatest in higher-impact journals, and positive language correlated with increased citation of publications (17). These findings suggest that training female academicians to present their research using more affirmative language could result in higher rates of publication in increasingly prestigious journals and greater acknowledgment of their work via citations in other publications. Ongoing studies such as the one presented in this This article is open access and distributed under the terms of the Creative Commons Attribution Non-Commercial No Derivatives License 4.0 (http://creativecommons.org/licenses/by-nc-nd/4.0/). For commercial usage and reprints, please contact Diane Gern (dgern@thoracic.org).

Purpose: Bronchoalveolar fluid (BALF) and plasma biomarkers are often endpoints in early phase randomized trials (RCTs) in acute respiratory distress syndrome (ARDS). With ARDS mortality decreasing, we analyzed baseline biomarkers in samples from contemporary ARDS patients participating in a prior RCT and compared these to historical controls.Materials and methods: Ninety ARDS adult patients enrolled in the parent trial. BALF and blood were collected at baseline, day 41, and day 81. Interleukins-8/-6/-1/-1 receptor antagonist/-10; granulocyte colony stimulating factor; monocyte chemotactic protein-1; tumour necrosis factor-; surfactant protein-D; von Willebrand factor; leukotriene B-4; receptor for advanced glycosylation end products; soluble Fas ligand; and neutrophil counts were measured.Results: Compared to historical measurements, our values were generally substantially lower, despite our participants being similar to historical controls. For example, our BALF IL-8 and plasma IL-6 were notably lower than in a 1999 RCT of low tidal volume ventilation and a 2007 biomarker study, respectively.Conclusions: Baseline biomarker levels in current ARDS patients are substantially lower than 6-20years before collection of these samples. These findings, whether from ICU care changes resulting in less inflammation or from variation in assay techniques over time, have important implications for design of future RCTs with biomarkers as endpoints.

Four decades ago, U.S. life expectancy was within the same range as other high-income peer countries. However, during the past decades, the United States has fared worse in many key health domains resulting in shorter life expectancy and poorer health—a health disadvantage. The National Heart, Lung, and Blood Institute convened a panel of national and international health experts and stakeholders for a Think Tank meeting to explore the U.S. health disadvantage and to seek specific recommendations for implementation research opportunities for heart, lung, blood, and sleep disorders. Recommendations for National Heart, Lung, and Blood Institute consideration were made in several areas including understanding the drivers of the disadvantage, identifying potential solutions, creating strategic partnerships with common goals, and finally enhancing and fostering a research workforce for implementation research. Key recommendations included exploring why the United States is doing better for health indicators in a few areas compared with peer countries; targeting populations across the entire socioeconomic spectrum with interventions at all levels in order to prevent missing a substantial proportion of the disadvantage; assuring partnership have high-level goals that can create systemic change through collective impact; and finally, increasing opportunities for implementation research training to meet the current needs. Connecting with the research community at large and building on ongoing research efforts will be an important strategy. Broad partnerships and collaboration across the social, political, economic, and private sectors and all civil society will be critical—not only for implementation research but also for implementing the findings to have the desired population impact. Developing the relevant knowledge to tackle the U.S. health disadvantage is the necessary first step to improve U.S. health outcomes.

RationaleTwo distinct acute respiratory distress syndrome (ARDS) subphenotypes have been identified using data obtained at time of enrolment in clinical trials; it remains unknown if these subphenotypes are durable over time.ObjectiveTo determine the stability of ARDS subphenotypes over time.MethodsSecondary analysis of data from two randomised controlled trials in ARDS, the ARMA trial of lung protective ventilation (n=473; patients randomised to low tidal volumes only) and the ALVEOLI trial of low versus high positive end-expiratory pressure (n=549). Latent class analysis (LCA) and latent transition analysis (LTA) were applied to data from day 0 and day 3, independent of clinical outcomes.Measurements and main resultsIn ALVEOLI, LCA indicated strong evidence of two ARDS latent classes at days 0 and 3; in ARMA, evidence of two classes was stronger at day 0 than at day 3. The clinical and biological features of these two classes were similar to those in our prior work and were largely stable over time, though class 2 demonstrated evidence of progressive organ failures by day 3, compared with class 1. In both LCA and LTA models, the majority of patients (>94%) stayed in the same class from day 0 to day 3. Clinical outcomes were statistically significantly worse in class 2 than class 1 and were more strongly associated with day 3 class assignment.ConclusionsARDS subphenotypes are largely stable over the first 3 days of enrolment in two ARDS Network trials, suggesting that subphenotype identification may be feasible in the context of clinical trials.

Four decades ago, U.S. life expectancy was within the same range as other high-income peer countries. However, during the past decades, the United States has fared worse in many key health domains resulting in shorter life expectancy and poorer health—a health disadvantage. The National Heart, Lung, and Blood Institute convened a panel of national and international health experts and stakeholders for a Think Tank meeting to explore the U.S. health disadvantage and to seek specific recommendations for implementation research opportunities for heart, lung, blood, and sleep disorders. Recommendations for National Heart, Lung, and Blood Institute consideration were made in several areas including understanding the drivers of the disadvantage, identifying potential solutions, creating strategic partnerships with common goals, and finally enhancing and fostering a research workforce for implementation research. Key recommendations included exploring why the United States is doing better for health indicators in a few areas compared with peer countries; targeting populations across the entire socioeconomic spectrum with interventions at all levels in order to prevent missing a substantial proportion of the disadvantage; assuring partnership have high-level goals that can create systemic change through collective impact; and finally, increasing opportunities for implementation research training to meet the current needs. Connecting with the research community at large and building on ongoing research efforts will be an important strategy. Broad partnerships and collaboration across the social, political, economic, and private sectors and all civil society will be critical—not only for implementation research but also for implementing the findings to have the desired population impact. Developing the relevant knowledge to tackle the U.S. health disadvantage is the necessary first step to improve U.S. health outcomes.

Heterogeneity in ARDS, whether it be in epidemiology, response to therapy, outcomes, or clinical practice, continues to both provide opportunities for and confound investigators and clinicians. The recent publication by Laffey et al1Laffey J.G. Madotto F. Bellani G. et al.Geo-economic variations in epidemiology, patterns of care, and outcomes in patients with acute respiratory distress syndrome: insights from the LUNG SAFE prospective cohort study.Lancet Respir Med. 2017; 5: 627-638Abstract Full Text Full Text PDF PubMed Scopus (79) Google Scholar in Lancet Respiratory Medicine, which demonstrates the significant geo-economic variations in patients with ARDS, adds to the complexity of the syndrome but also provides tremendous opportunities for new areas of investigation and improvement in clinical care. The Large Observational Study to Understand the Global Impact of Severe Acute Respiratory Failure (LUNG SAFE)2Bellani G. Laffey J.G. Pham T. et al.Epidemiology, patterns of care, and mortality for patients with acute respiratory distress syndrome in intensive care units in 50 countries.JAMA. 2016; 315: 788-800Crossref PubMed Scopus (2651) Google Scholar was a landmark prospective cohort study of 12,906 patients from 459 ICUs in 50 countries enrolled over a 4-week period. The study confirmed that the period prevalence of ARDS is high, accounting for 10.4% of ICU admissions, with a wide variation in the incidence of ARDS per ICU bed. It found that the syndrome was underrecognized, evidence-based strategies were underused, and mortality was high. Laffey et al1Laffey J.G. Madotto F. Bellani G. et al.Geo-economic variations in epidemiology, patterns of care, and outcomes in patients with acute respiratory distress syndrome: insights from the LUNG SAFE prospective cohort study.Lancet Respir Med. 2017; 5: 627-638Abstract Full Text Full Text PDF PubMed Scopus (79) Google Scholar analyzed a subset of 2,813 patients from LUNG SAFE diagnosed with ARDS according to the Berlin definition on study day 1 or 2 and grouped into three geo-economic regions: European high-income countries, high-income countries (rest of the world), and middle-income countries. Of note, there were no data from ICUs in low-income countries, and the majority of patients were in academic medical centers, although the investigators worked to be inclusive. In addition to finding that there were significant geographic and economic differences in the epidemiology, clinical care, and outcomes between the three groups, they also found an independent relationship between gross national income per person and mortality. The relationship between income and mortality was not explained by ICU structure or staffing. This opens the door to a host of questions and hypotheses to be asked and tested, which will include research in diverse areas (eg, genetics, basic mechanisms of injury, epidemiology, economics, and social science) and will require new collaborators, not only with unique expertise but also from unique geographic locations. This study by Laffey et al1Laffey J.G. Madotto F. Bellani G. et al.Geo-economic variations in epidemiology, patterns of care, and outcomes in patients with acute respiratory distress syndrome: insights from the LUNG SAFE prospective cohort study.Lancet Respir Med. 2017; 5: 627-638Abstract Full Text Full Text PDF PubMed Scopus (79) Google Scholar further highlights the lack of clinical recognition and the underuse of evidence-based clinical care reported in LUNG SAFE. Of all patients adjudicated by the study to have ARDS, the syndrome was recognized clinically in only 62% of the cases, with the lowest level of recognition in high-income/non-European counties (54%) and the highest recognition in middle-income countries (66%). Well-established best clinical practices were grossly underused. For example, in aggregate, protective ventilation was used in only 63% of the patients receiving mechanical ventilation, but of interest, there was no significant geographic variation. The use of adjunctive measures was low and highly variable. In patients with severe ARDS, prone positioning and neuromuscular blockade were used more often in European high-income countries than in the other regions. These are sobering results. Unfortunately, however, they are neither new nor surprising. In 2004, two separate groups3Young M.P. Manning H.L. Wilson D.L. et al.Ventilation of patients with acute lung injury and ARDS: has new evidence changed clinical practice?.Crit Care Med. 2004; 32: 1260-1265Crossref PubMed Scopus (176) Google Scholar, 4Rubenfeld G.D. Cooper C. Carter G. et al.Barriers to providing lung-protective ventilation to patients with acute lung injury.Crit Care Med. 2004; 32: 1289-1293Crossref PubMed Scopus (229) Google Scholar found that ARDS was underrecognized and low-tidal volume ventilation was underused, and they identified several potential barriers to the implementation of best practice. More recently in an editorial accompanying the original LUNG SAFE paper in the Journal of the American Medical Association,5Clarke B.J. Moss M. The acute respiratory distress syndrome. Dialing in the evidence?.JAMA. 2016; 315: 759-760Crossref PubMed Scopus (6) Google Scholar Clark and Moss highlight the critical need to implement evidence-based practice into clinical care. The data remind us of the philosophical question: If a tree falls in a forest and there is no one in the woods to hear it, does it make a sound? As clinicians and researchers, we frequently lament the lack of therapeutic intervention in ARDS, yet we do not use and implement the knowledge that we already have. If ARDS is not recognized, it cannot be treated, and if patients do not receive care supported by strong evidence, we will not have an impact on their mortality and other clinical outcomes. We need to implement that knowledge into practice. What can we learn from the middle-income countries, where clinical recognition of ARDS is 66%, that could increase recognition in high-income/rest of the world ICUs, where it is only 54%? How could clinical recognition be increased to 100%? How could the definition be modified or expanded to include all potential patients, even those in less resource-intensive environments in which patients cannot meet the current Berlin definition of ARDS because there is no chest radiography, readily available assessments of oxygenation, or even available ventilator support? What is the best way to disseminate new knowledge to impact outcomes? How can evidence-based practices be implemented universally to decrease mortality? How can we get more listeners into the woods? These and other research questions are ripe for investigation by implementation scientists. These questions highlight another critical area of opportunity: the need for broader diversity and inclusion in scientific collaborations and in expert and consensus panels to be able to understand and impact critical illness in all areas of the world. We know from the LUNG SAFE and other studies1Laffey J.G. Madotto F. Bellani G. et al.Geo-economic variations in epidemiology, patterns of care, and outcomes in patients with acute respiratory distress syndrome: insights from the LUNG SAFE prospective cohort study.Lancet Respir Med. 2017; 5: 627-638Abstract Full Text Full Text PDF PubMed Scopus (79) Google Scholar, 2Bellani G. Laffey J.G. Pham T. et al.Epidemiology, patterns of care, and mortality for patients with acute respiratory distress syndrome in intensive care units in 50 countries.JAMA. 2016; 315: 788-800Crossref PubMed Scopus (2651) Google Scholar, 6Adhikari N.K. Fowler R.A. Bhagwanjee S. Rubenfeld G.D. Critical care and the global burden of critical illness in adults.Lancet. 2010; 375: 1339-1346Abstract Full Text Full Text PDF PubMed Scopus (743) Google Scholar that critical illness is a global issue, low-income areas are disproportionately impacted by critical illness (sepsis),6Adhikari N.K. Fowler R.A. Bhagwanjee S. Rubenfeld G.D. Critical care and the global burden of critical illness in adults.Lancet. 2010; 375: 1339-1346Abstract Full Text Full Text PDF PubMed Scopus (743) Google Scholar, 7Beale R. Reinhart K. Brunkhorst F.M. et al.PROGRESS Advisory BoardPromoting Global Research Excellence in Severe Sepsis (PROGRESS): lessons from an international sepsis registry.Infection. 2009; 37: 222-232Crossref PubMed Scopus (117) Google Scholar mortality is impacted by racial and ethnic differences,8Erikson S.E. Shipak M.G. Martin G.S. et al.Racial and ethnic disparities in mortality from acute lung injury.Crit Care Med. 2009; 37: 1-6Crossref PubMed Scopus (95) Google Scholar and diversity in working groups adds value.9Mehta S, Burns KE, Machado FR, et al. Gender parity in critical care medicine [published online ahead of print February 27, 2017]. Am J Respir Crit Care Med. http://dx.doi.org/10.1164/rccm.201701-0076CP.Google Scholar Critical care has always been multidisciplinary, not only in the clinical setting but also in the research environment. The profound geo-economic variations in ARDS should encourage us to expand that diversity even further.

FOR RELATED ARTICLE SEE PAGE 755The first report of ARDS was by Ashbaugh et al1Ashbaugh D.G. Bigelow D.B. Petty T.L. Levine B.E. Acute respiratory distress in adults.Lancet. 1967; 2: 319-323Abstract PubMed Google Scholar in 1967. The 12 patients in that initial report all had acute respiratory distress in common, but the underlying conditions were variable and included trauma, pneumonia, and pancreatitis. Since that time, clinicians and scientists have debated as to whether all patients with ARDS were “the same” and therefore should be “lumped together”2Petty T.L. Editorial: the adult respiratory distress syndrome (confessions of a lumper).Am Rev Respir Dis. 1975; 111: 713-715PubMed Google Scholar or they were different and should be “split”3Murray J.F. The adult respiratory distress syndrome (may it rest in peace).Am Rev Respir Dis. 1975; 111: 716-718PubMed Google Scholar into distinct subgroups. Significant advances have been made by using each of those philosophies. FOR RELATED ARTICLE SEE PAGE 755 Those of us involved in the early translational and clinical studies became increasingly frustrated by the lack of consistency in results and the failure of therapeutic interventions that seemed so promising in animal models. In general, we were “lumpers,” but we all likely realized that ARDS was more complex. We both remember our own aha moments when transitioning from discussions of data from a lipopolysaccharide (LPS)-induced ARDS animal model to the collection of ICU samples from patients at risk for ARDS. Early animal models often followed the “lumper” strategy by examining young, generally healthy male animals and data from both direct (ie, hyperoxia, intratracheal LPS administration) and indirect (ie, cecal ligation/puncture, IV LPS administration) injury aggregated to discern the pathogenesis of the syndrome. However, patients in the ICUs were vastly different from the animals being studied. They had a variety of risk factors (including sepsis and trauma), and they were not all young, previously healthy, male, or white. In addition, they had various organ failures, were mechanically ventilated, and taking multiple medications. The ability to be a “lumper” became increasingly untenable. Initially, the challenge of heterogeneity was daunting, but over time it became clear that it provided an opportunity to better elucidate the pathogenesis of ARDS and for the potential to develop targeted therapeutic interventions. Definitions for both ARDS and underlying risk factors (ie, sepsis) evolved4Fowler A.A. Hamman R.F. Good J.T. et al.Adult respiratory distress syndrome: risk with common predispositions.Ann Intern Med. 1983; 98: 593-597Crossref PubMed Scopus (584) Google Scholar, 5Abraham E. Dinarello C.A. Matthay M.A. et al.Consensus conference definitions for sepsis, septic shock, acute lung injury and ARDS—time for re-evaluation.Crit Care Med. 2000; 28: 232-235Crossref PubMed Scopus (304) Google Scholar, 6Ranieri V.M. Rubenfeld G.D. Thompson B.T. et al.Acute respiratory distress syndrome: the Berlin definition.JAMA. 2012; 307: 2526-2533Crossref PubMed Scopus (6577) Google Scholar and encompassed not only etiologies such as sepsis and trauma but the division of patients into those sustaining direct vs indirect lung injury. The impact of preexisting and comorbid conditions (including age, sex, ethnicity, diabetes, alcohol, obesity, kidney injury, smoking, and genetics), as well as features of clinical care such as mechanical ventilation, fluid management, and sedation, was identified. Furthermore, the site of measurements (ie, BAL vs blood) and the assay methods used to measure myriad biomarkers were also recognized as important contributors to the variations seen in the results.7Parsons P.E. Moore F.A. Moore E.E. et al.Studies on the role of tumor necrosis factor in ARDS: the lack of an independent standard.Am Rev Respir Dis. 1992; 146: 694-700Crossref PubMed Scopus (80) Google Scholar The initial translational and clinical studies were small and often included patients from only a single center, making it difficult to both identify and characterize the many heterogeneous patient populations that were included within the syndrome. However, large, well-characterized cohorts (including ones from the National Heart, Lung, and Blood Institute’s ARDS Network, the Mayo Clinic, and Vanderbilt) have been created over the last two decades, significantly increasing the power to both detect and analyze differences, identifying new and unique clinical phenotypes and biomarkers. The article by Luo et al8Luo L. Shaver C.M. Zhao Z. et al.Clinical predictors of hospital mortality differ between direct and indirect acute respiratory distress syndrome.Chest. 2017; 151: 755-763Google Scholar in this issue of CHEST is an important addition to the ARDS story. Although previous studies have clearly shown that the pathogenesis of direct and indirect ARDS is different,9Calfee C.S. Janz D.R. Bernard G.R. et al.Distinct molecular phenotypes of direct vs indirect ARDS in a single-center and multicenter studies.Chest. 2015; 147: 1539-1548Abstract Full Text Full Text PDF PubMed Scopus (229) Google Scholar it was not known if the clinical characteristics and predictors of mortality differed between these groups. Taking advantage of the opportunities provided by the large number of carefully phenotyped patients enrolled in the Validation of Biomarkers for Acute Lung Injury Diagnosis (VALID) study, the authors evaluated the clinical characteristics and predictors of mortality in patients with ARDS from direct insults vs those with ARDS from indirect insults. They determined that the overall mortality rate was not different between the two groups, but the clinical characteristics associated with mortality did vary. In patients with direct injury, age, lung injury score, and nonpulmonary organ failure were all associated with an increased mortality, whereas diabetes was associated with a decrease in hospital mortality. In contrast, in indirect ARDS, only an increase in organ failure was associated with an increase in mortality. Moreover, factors previously associated with mortality from ARDS in general were found only to associate with mortality from direct ARDS, suggesting that many of our established assumptions about ARDS may, in fact, be as heterogeneous as those afflicted. As Luo et al8Luo L. Shaver C.M. Zhao Z. et al.Clinical predictors of hospital mortality differ between direct and indirect acute respiratory distress syndrome.Chest. 2017; 151: 755-763Google Scholar clearly point out, there are limitations to this study, including the lack of inclusion of patients with common risk factors such as trauma, the difference in size of the direct (250 patients) vs indirect (167 patients) injury cohorts, and the retrospective nature of the study, which prevented assessment of causality (ie, the impact of diabetes on mortality). Furthermore, the American European Consensus Definition of ARDS was used in this study and not the Berlin definition,6Ranieri V.M. Rubenfeld G.D. Thompson B.T. et al.Acute respiratory distress syndrome: the Berlin definition.JAMA. 2012; 307: 2526-2533Crossref PubMed Scopus (6577) Google Scholar which is more commonly used currently. Despite those limitations, the results are important. They provide further evidence that there are distinct patient populations who all share the same diagnosis (ARDS) but vary in clinical characteristics. Whenever a new, potentially distinctive, cohort has been identified, it has led to studies providing unique insights into the pathogenesis of the syndrome, identification of novel biomarkers, and new analytic approaches such as latent class analysis.10Calfee C.S. Delucchi K. Parsons P.E. et al.Latent class analysis of ARDS subphenotypes: analysis of data from two randomized controlled trials.Lancet Respir Med. 2014; 2: 611-620Abstract Full Text Full Text PDF PubMed Scopus (716) Google Scholar Each of these approaches moves us closer to the goals of improved diagnostics, targeted therapeutic interventions, and a more personalized approach to care. Clinical Predictors of Hospital Mortality Differ Between Direct and Indirect ARDSCHESTVol. 151Issue 4PreviewDirect (pulmonary) and indirect (extrapulmonary) ARDS are distinct syndromes with important pathophysiologic differences. The goal of this study was to determine whether clinical characteristics and predictors of mortality differ between direct or indirect ARDS. Full-Text PDF

IMPORTANCE The role of cytomegalovirus (CMV) reactivation in mediating adverse clinical outcomes in nonimmunosuppressed adults with critical illness is unknown.OBJECTIVE To determine whether ganciclovir prophylaxis reduces plasma interleukin 6 (IL-6) levels in CMV-seropositive adults who are critically ill.DESIGN, SETTING, AND PARTICIPANTS Double-blind, placebo-controlled, randomized clinical trial (conducted March 10, 2011-April 29, 2016) with a follow-up of 180 days (November 10, 2016) that included 160 CMV-seropositive adults with either sepsis or trauma and respiratory failure at 14 university intensive care units (ICUs) across the United States.INTERVENTIONS Patients were randomized (1: 1) to receive either intravenous ganciclovir (5 mg/kg twice daily for 5 days), followed by either intravenous ganciclovir or oral valganciclovir once daily until hospital discharge (n = 84) or to receive matching placebo (n = 76).MAIN OUTCOMES AND MEASURES The primary outcome was change in IL-6 level from day 1 to 14. Secondary outcomes were incidence of CMV reactivation in plasma, mechanical ventilation days, incidence of secondary bacteremia or fungemia, ICU length of stay, mortality, and ventilator-free days (VFDs) at 28 days.RESULTS Among 160 randomized patients (mean age, 57 years; women, 43%), 156 received 1 or more dose(s) of treatment, and 132 (85%) completed the study. The mean between-group change in IL-6 level was not significantly different. Among secondary outcomes, CMV reactivation in plasma was significantly lower in the ganciclovir group. The ganciclovir group had more VFDs in both the intention-to-treat population and in the prespecified sepsis subgroup. There were no significant between-group differences in other secondary outcomes.[GRAPHICS]CONCLUSIONS AND RELEVANCE Among CMV-seropositive adults with critical illness due to sepsis or trauma, ganciclovir did not reduce IL-6 levels and the current study does not support routine clinical use of ganciclovir as a prophylactic agent in patients with sepsis. Additional research is necessary to determine the clinical efficacy and safety of CMV suppression in this setting.

We have previously reported that obesity attenuates pulmonary inflammation in both patients with acute respiratory distress syndrome (ARDS) and in mouse models of the disease. We hypothesized that obesity-associated hyperleptinemia, and not body mass per se, drives attenuation of the pulmonary inflammatory response and that this e_ect could also impair the host response to pneumonia. We examined the correlation between circulating leptin levels and risk, severity, and outcome of pneumonia in 2 patient cohorts (NHANES III and ARDSNet-ALVEOLI) and in mouse models of diet-induced obesity and lean hyperleptinemia. Plasma leptin levels in ambulatory subjects (NHANES) correlated positively with annual risk of respiratory infection independent of BMI. In patients with severe pneumonia resulting in ARDS (ARDSNet-ALVEOLI), plasma leptin levels were found to correlate positively with subsequent mortality. In obese mice with pneumonia, plasma leptin levels were associated with pneumonia severity, and in obese mice with sterile lung injury, leptin levels were inversely related to bronchoalveolar lavage neutrophilia, as well as to plasma IL-6 and G-CSF levels. These results were recapitulated in lean mice with experimentally induced hyperleptinemia. Our findings suggest that the association between obesity and elevated risk of pulmonary infection may be driven by hyperleptinemia.

Academic clinical departments have the opportunity and responsibility to improve the quality and value of care and patient safety by supporting effective quality improvement activities. The pressure to provide high-value care while further developing academic programs has increased the complexity of decision making and change management in academic health systems. Overcoming these challenges will require faculty engagement and leadership; however, most academic departments do not have a sufficient number of individuals with expertise and experience in quality improvement and patient safety (QI/PS). Accordingly, the authors of this article advocate for a targeted and proactive approach to developing faculty working in QI/PS. They propose a strategy predicated on the identification of QI/PS as a strategic priority for academic departments, the creation of enabling resources in QI/PS, and the expansion of rigorous training programs in change management and in improvement and implementation sciences. Professional organizations, health systems, medical schools, and academic departments should recognize successful QI/PS work with awards and promotions. Individual faculty members should expand their collaborative networks, consider the generalizability and scholarly impact of their efforts when designing QI/PS initiatives, and benchmark the outcomes of their performance. Appointments and promotions committees should work proactively with department and QI/PS leaders to ensure that outstanding achievement in QI/PS is defined and recognized. As with the development of physician-investigators and clinician-educators, departments and health systems need a comprehensive approach to support and recognize the contributions of faculty working in QI/PS to meet the considerable needs and opportunities in health care.

Articles Subphenotypes in acute respiratory distress syndrome: latent class analysis of data from two randomised controlled trials Carolyn S Calfee, Kevin Delucchi, Polly E Parsons, B Taylor Thompson, Lorraine B Ware, Michael A Matthay, and the NHLBI ARDS Network Summary Background Subphenotypes have been identifi ed within heterogeneous diseases such as asthma and breast cancer, with important therapeutic implications. We assessed whether subphenotypes exist within acute respiratory distress syndrome (ARDS), another heterogeneous disorder. Methods We used data from two ARDS randomised controlled trials (ARMA trial and ALVEOLI trial), sponsored by the National Heart, Lung, and Blood Institute. We applied latent class modelling to identify subphenotypes using clinical and biological data. We modelled data from both studies independently. We then tested the association of subphenotypes with clinical outcomes in both cohorts and with the response to positive end-expiratory pressure (PEEP) in the ALVEOLI cohort. Findings We analysed data for 1022 patients: 473 in the ARMA cohort and 549 in the ALVEOLI cohort. Independent latent class models indicated that a two-class (ie, two subphenotype) model was the best fi t for both cohorts. In both cohorts, we identifi ed a hyperinfl ammatory subphenotype (phenotype 2) that was characterised by higher plasma concentrations of infl ammatory biomarkers, a higher prevalence of vasopressor use, lower serum bicarbonate concentrations, and a higher prevalence of sepsis than phenotype 1. Participants in phenotype 2 had higher mortality and fewer ventilator-free days and organ failure-free days in both cohorts than did those in phenotype 1 (p<0·007 for all). In the ALVEOLI cohort, the eff ects of ventilation strategy (high PEEP vs low PEEP) on mortality, ventilator-free days and organ failure-free days diff ered by phenotype (p=0·049 for mortality, p=0·018 for ventilator-free days, p=0·003 for organ-failure-free days). Interpretation We have identifi ed two subphenotypes within ARDS, one of which is categorised by more severe infl ammation, shock, and metabolic acidosis and by worse clinical outcomes. Response to treatment in a randomised trial of PEEP strategies diff ered on the basis of subphenotype. Identifi cation of ARDS subphenotypes might be useful in selecting patients for future clinical trials. Funding National Institutes of Health. Introduction Acute respiratory distress syndrome (ARDS) is a heterogeneous syndrome fi rst identifi ed in 1967 and defi ned by the clinical criteria of bilateral pulmonary opacities on chest radiograph, arterial hypoxaemia (partial pressure of arterial oxygen [PaO 2 ] to fraction of inspired oxygen [FiO 2 ] ratio <300), and exclusion of cardiac failure as the primary cause of the syndrome. 1–3 This defi nition was derived empirically on the basis of clinical experience, with the hypothesis that it would identify patients with non-cardiogenic pulmonary oedema, characterised by increased protein permeability of the alveolar–capillary membrane. Since the time of the original identifi cation of ARDS, and increasingly during the past two decades, there has been recognition of the clinical and biological heterogeneity within ARDS; 4,5 this hetero geneity might refl ect our incomplete under- standing of the biology of ARDS and probably contributes to the poor track record of phase 2 and 3 trials of new treatments for patients with ARDS. 6 As a result, some investigators have proposed subdividing ARDS on the basis of clinical risk factors, or by direct versus indirect cause of lung injury. However, no consensus exists on the appropriate approach to reduce ARDS heterogeneity. By contrast with ARDS, research in airways disease and cancer has made substantial progress towards identifying subphenotypes of disease, with important therapeutic implications. For example, subphenotypes based on the presence or absence of Th2-dependent infl ammation have been identifi ed in asthma, with important mechanistic and therapeutic implications. 7 This insight has led to new targeted treatments, such as a monoclonal antibody to interleukin-13, which is especially eff ective in individuals with Th2-predominant infl ammation. 8 Despite widespread recognition of the heterogeneity within common critical illness syndromes such as sepsis and ARDS, and some evidence suggesting that subphenotypes might exist within severe sepsis, 6,9,10 little data are available for whether such subphenotypes exist in ARDS. Latent class analysis is a well-validated statistical technique that uses mixture modelling to fi nd the best- fi tting model for a set of data, based on the hypothesis that the data contain several unobserved groups or www.thelancet.com/respiratory Published online May 20, 2014 http://dx.doi.org/10.1016/S2213-2600(14)70097-9 Lancet Respir Med 2014 Published Online May 20, 2014 http://dx.doi.org/10.1016/ S2213-2600(14)70097-9 See Online/Comment http://dx.doi.org/10.1016/ S2213-2600(14)70116-X Departments of Medicine and Anesthesia, Division of Pulmonary and Critical Care Medicine (C S Calfee MD, Prof M A Matthay MD) and Department of Psychiatry (Prof K Delucchi PhD), University of California San Francisco, San Francisco, CA, USA; Department of Medicine, Division of Pulmonary and Critical Care Medicine, University of Vermont, Burlington, VT, USA (Prof P E Parsons MD); Department of Medicine, Pulmonary and Critical Care Medicine Unit, Massachusetts General Hospital, Boston, MA, USA (Prof B T Thompson MD); Biostatistics Unit, Massachusetts General Hospital, Boston, MA, USA (Prof B T Thompson); Department of Medicine, Division of Allergy, Pulmonary and Critical Care, Vanderbilt, University, Nashville, TN, USA (Prof L B Ware MD); and Cardiovascular Research Institute, San Francisco, CA, USA (Prof M A Matthay) Correspondence to: Dr Carolyn S Calfee, University of California, San Francisco, Pulmonary and Critical Care Division, 505 Parnassus Avenue, Box 0111, San Francisco, CA 94143-0111, USA carolyn.calfee@ucsf.edu

Background Subphenotypes have been identified within heterogeneous diseases such as asthma and breast cancer, with important therapeutic implications. We assessed whether subphenotypes exist within acute respiratory distress syndrome (ARDS), another heterogeneous disorder.Methods We used data from two ARDS randomised controlled trials (ARMA trial and ALVEOLI trial), sponsored by the National Heart, Lung, and Blood Institute. We applied latent class modelling to identify subphenotypes using clinical and biological data. We modelled data from both studies independently. We then tested the association of subphenotypes with clinical outcomes in both cohorts and with the response to positive end-expiratory pressure (PEEP) in the ALVEOLI cohort.Findings We analysed data for 1022 patients: 473 in the ARMA cohort and 549 in the ALVEOLI cohort. Independent latent class models indicated that a two-class (ie, two subphenotype) model was the best fit for both cohorts. In both cohorts, we identified a hyperinflammatory subphenotype (phenotype 2) that was characterised by higher plasma concentrations of inflammatory biomarkers, a higher prevalence of vasopressor use, lower serum bicarbonate concentrations, and a higher prevalence of sepsis than phenotype 1. Participants in phenotype 2 had higher mortality and fewer ventilator-free days and organ failure-free days in both cohorts than did those in phenotype 1 (p<0.007 for all). In the ALVEOLI cohort, the effects of ventilation strategy (high PEEP vs low PEEP) on mortality, ventilator-free days and organ failure-free days differed by phenotype (p=0.049 for mortality, p=0.018 for ventilator-free days, p=0.003 for organ-failure-free days).Interpretation We have identified two subphenotypes within ARDS, one of which is categorised by more severe inflammation, shock, and metabolic acidosis and by worse clinical outcomes. Response to treatment in a randomised trial of PEEP strategies differed on the basis of subphenotype. Identification of ARDS subphenotypes might be useful in selecting patients for future clinical trials.

Perspectives Viewpoints•The cost of training internal medicine residents in the United States has increased largely secondary to duty hour limits and new accreditation requirements.•The costs exceed previously published estimates and the average per resident outlay of federal graduate medical education support by $51,737 to $79,999 per year.•New models of graduate medical education funding should address the actual costs of training to adequately meet national physician workforce needs. •The cost of training internal medicine residents in the United States has increased largely secondary to duty hour limits and new accreditation requirements.•The costs exceed previously published estimates and the average per resident outlay of federal graduate medical education support by $51,737 to $79,999 per year.•New models of graduate medical education funding should address the actual costs of training to adequately meet national physician workforce needs. Departments of medicine train approximately 30% of the 115,000 residents and fellows in the United States.1Brotherton S.E. Etzel S.L. Graduate medical education, 2012-2013.JAMA. 2013; 310: 2328-2346Crossref PubMed Scopus (59) Google Scholar Yet, departments have few reliable external sources of funding for graduate medical education and are handicapped in negotiations with teaching hospitals (the recipients of state and federal graduate medical education funding) in part because of uncertainty of the actual costs of training.2Nasca T.J. Veloski J.J. Monnier J.A. et al.Minimum instructional and program-specific administrative costs of educating residents in internal medicine.Arch Intern Med. 2001; 161: 760-766Crossref PubMed Scopus (16) Google Scholar, 3Zeidel M.L. Kroboth F. McDermot S. et al.Estimating the cost to departments of medicine of training residents and fellows: a collaborative analysis.Am J Med. 2005; 118: 557-564Abstract Full Text Full Text PDF PubMed Scopus (22) Google Scholar Moreover, several government advisory committees are recommending cuts to graduate medical education funding.4Steinmann A.F. Threats to graduate medical education funding and the need for a rational approach: a statement from the Alliance for Academic Internal Medicine.Ann Intern Med. 2011; 155: 461-464Crossref PubMed Scopus (15) Google Scholar In fiscal year 2012, the Centers for Medicare & Medicaid Services (CMS), the major payer explicitly covering graduate medical education costs, spent $10.1 billion ($9.3 billion by Medicare and $0.8 billion by Medicaid) in support of graduate medical education.4Steinmann A.F. Threats to graduate medical education funding and the need for a rational approach: a statement from the Alliance for Academic Internal Medicine.Ann Intern Med. 2011; 155: 461-464Crossref PubMed Scopus (15) Google Scholar, 5U.S. Government Accountability Office Communication: Health care workforce: federally funded training programs in fiscal year 2012. Available at: http://www.gao.gov/assets/660/656960.pdf. Accessed January 29, 2014.Google Scholar Other federal agencies that contribute smaller amounts include the Department of Veterans Affairs, Department of Defense, Health Resources and Services Administration, and National Institutes of Health. The government's formula for graduate medical education support via the CMS was established by Congress in the 1980s and consists of direct medical education and indirect medical education payments to teaching hospitals. Direct medical education payments (∼$2.7 billion annually) cover Medicare's share of costs directly related to running graduate medical education programs (eg, trainee salaries, fringe benefits, faculty supervision/teaching time, administrative costs, facilities overhead, and malpractice insurance). These costs are influenced significantly by program and institutional requirements of the Accreditation Council for Graduate Medical Education (ACGME) and the fair market values of faculty salaries, among other things. Indirect medical education payments (∼$7.5 billion annually) were established to reimburse teaching hospitals for the perceived increased complexity of patient care that resulted in longer lengths of stay, additional testing, need for more advanced diagnostic and therapeutic modalities, and a higher case mix index.4Steinmann A.F. Threats to graduate medical education funding and the need for a rational approach: a statement from the Alliance for Academic Internal Medicine.Ann Intern Med. 2011; 155: 461-464Crossref PubMed Scopus (15) Google Scholar In 2001, Nasca et al2Nasca T.J. Veloski J.J. Monnier J.A. et al.Minimum instructional and program-specific administrative costs of educating residents in internal medicine.Arch Intern Med. 2001; 161: 760-766Crossref PubMed Scopus (16) Google Scholar estimated the annual minimum instructional and program-specific administrative costs of training internal medicine residents using the 1998 ACGME requirements. They demonstrated that the minimum instructional and program-specific administrative costs per resident (fixed and variable combined) were inversely proportional to program size with the per-resident costs in larger programs (126 residents in their model) being approximately 25% lower than in smaller programs (21 residents). Furthermore, the model by Nasca et al2Nasca T.J. Veloski J.J. Monnier J.A. et al.Minimum instructional and program-specific administrative costs of educating residents in internal medicine.Arch Intern Med. 2001; 161: 760-766Crossref PubMed Scopus (16) Google Scholar demonstrated that the per-resident minimum instructional and program-specific administrative costs in outpatient intensive programs were approximately 8% higher than in inpatient intensive programs. Overall, they calculated that the total per-resident minimum instructional and program-specific administrative costs (excluding resident salary and benefits) ranged from $26,197 (large, inpatient intensive programs) to $58,025 (small, outpatient intensive programs) annually. In a subsequent similar analysis, Zeidel et al3Zeidel M.L. Kroboth F. McDermot S. et al.Estimating the cost to departments of medicine of training residents and fellows: a collaborative analysis.Am J Med. 2005; 118: 557-564Abstract Full Text Full Text PDF PubMed Scopus (22) Google Scholar estimated that the annual per-resident cost to departments of medicine for training internal medicine residents in 2003 (excluding resident salary and benefits) was $35,164. However, a number of recent influences on training have affected these earlier cost estimates. Nuckols and Escarce6Nuckols T.K. Escarce J.J. Cost implications of ACGME's 2011 changes to resident duty hours and the training environment.J Gen Intern Med. 2012; 27: 241-249Crossref PubMed Scopus (31) Google Scholar estimated the additional costs of duty hour requirements to be approximately $1 billion for additional personnel costs. This added expense occurs at a time when the 2010 National Commission on Fiscal Responsibility and Reform and the Medicare Payment Advisory Commission (MedPAC) have recommended that indirect medical education payments to teaching hospitals can be reduced by approximately one half because the relevant differences between teaching and nonteaching hospitals have narrowed significantly. Steinman4Steinmann A.F. Threats to graduate medical education funding and the need for a rational approach: a statement from the Alliance for Academic Internal Medicine.Ann Intern Med. 2011; 155: 461-464Crossref PubMed Scopus (15) Google Scholar argued that the recommended cuts to indirect medical education would have a significantly negative impact on graduate medical education funding because all 2007 sources of local and federal government funding resulted in an estimated $104,717 per-resident payment in direct medical education despite his estimate that direct medical education costs per resident should have been $130,000. Therefore, the current indirect medical education offset may be critical as a cross-subsidy to make up the approximately $25,000 shortfall per resident. Furthermore, since 1998, the ACGME has increased the ambulatory requirement of internal medicine training, increased the level of required associate program director and core faculty support, and implemented the next accreditation system requiring increased administrative and faculty time not accounted for in the direct medical education payment per resident estimated. These factors prompted the Global Issues Subcommittee of the Accreditation Committee of the Alliance for Academic Internal Medicine (AAIM) to update and estimate the actual direct costs of training an internal medicine resident in 2013. In 2012-2013, the Global Issues Subcommittee reviewed published analyses of costs of training residents in internal medicine and other disciplines and embarked on an effort to update the model of costs published in 2001 by Nasca et al.2Nasca T.J. Veloski J.J. Monnier J.A. et al.Minimum instructional and program-specific administrative costs of educating residents in internal medicine.Arch Intern Med. 2001; 161: 760-766Crossref PubMed Scopus (16) Google Scholar The Nasca group used 1998 ACGME program requirements and defined different program types by percent of ambulatory training time (outpatient intensive, traditional, inpatient intensive) and sizes (21, 42, 84, and 126 total residents) to estimate the minimum instructional and program-specific administrative costs of educating internal medicine residents. Our group sought to estimate actual rather than minimum costs of training and updated the model by Nasca et al by incorporating current ACGME program requirements,7ACGME program requirements for graduate medical education in internal medicine. Effective July 1, 2013. Available at: http://acgme.org/acgmeweb/portals/0/PFassets/2013-PR-FAQ-PIF/140_internal_medicine_07012013.pdf. Accessed January 12, 2014.Google Scholar survey data from the Association of Program Directors in Internal Medicine (APDIM),8Association of Program Directors in Internal Medicine. APDIM Survey 2012. Available at: http://www.im.org/toolbox/surveys/APDIMSurveyData/Documents/2012_APDIM_Survey_summary_web.pdf. Accessed January 12, 2014.Google Scholar American Association of Medical Colleges (AAMC) salary data,9Survey of residents/fellows stipends and benefits report. October 2011. American Association of Medical Colleges. Available at: https://www.aamc.org/download/265452/data/2011stipendreport.pdf. Accessed January 12, 2014.Google Scholar, 10U.S. medical school faculty with an M.D. (or equivalent degree), median total compensation in $ thousands 2012. Available at: https://services.aamc.org/dsportal2/index.cfm?fuseaction=login.login&thread=jump.FSSREPORTS&appname=FSSREPORTS&frompermissionscheck=true. Accessed January 12, 2014.Google Scholar internal nonvalidated survey data of the AAIM Accreditation Committee membership, and the subcommittee's collective experience. The subcommittee was composed of program directors, vice chairs, and chairs of internal medicine from across the country and represented university and community teaching hospitals and various sized programs (39-169 residents). Our analysis enabled us to create an updated model of fixed and variable direct costs associated with meeting ACGME requirements of teaching and administration in 2013, as well as typical added costs associated with the administration of internal medicine programs. The administrative fixed costs of internal medicine training programs independent of type or size includes: (1) 5% department chair time; (2) 50% program director time; (3) 1 program administrator full time equivalent; (4) 1 secretary full-time equivalent; (5) 2 chief residents; (6) time required for 1 faculty member to attend the minimum number of required conference hours; (7) development costs of program director, program administrator, and chief residents; (8) faculty development; (9) evaluation management database; and (10) ACGME fees (Table 1).Table 1Fixed CostsCategoryCost ($)Chair salary + fringe @ 5% time27,495PD salary + fringe @ 50% time127,635Program administrator (n = 1)82,550Secretarial support (n = 1)44,450Chief residents (n = 2)115,000Core conference requirement28,651PD, administrator, and chief resident development8500Faculty development15,000Evaluation management database500ACGME fees5200Total454,981ACGME = Accreditation Council for Graduate Medical Education; PD = program director. Open table in a new tab ACGME = Accreditation Council for Graduate Medical Education; PD = program director. The costs attributable to teaching residents were calculated from the total estimated time faculty spend in teaching activities without other sources of compensation (ie, patient care revenue) and faculty pay scales by specialty. We assumed the following rotation types among all residencies: (1) inpatient general medicine; (2) critical care; (3) inpatient consultation; (4) night float; (5) emergency medicine; (6) subspecialty clinic; and (7) continuity clinic. By using AAIM Accreditation Committee membership survey data and our collective experience, we defined the ratio of house officers to faculty and hours per week of faculty time spent teaching (Table 2). We grouped faculty into the following types based on our rotation assumptions: (1) hospital internal medicine; (2) critical care; (3) continuity clinic–general internists; (4) ambulatory block–subspecialty practitioners; and (5) emergency medicine (Table 3). Median faculty salaries at the associate professor level as reported by the AAMC in 2011 were used to assign salary costs. The ambulatory block subspecialty practitioner salary was derived by averaging all AAMC reported categories of internal medicine subspecialty salaries excluding critical care, invasive cardiology, allergy immunology, and "other." As in the model by Nasca et al,2Nasca T.J. Veloski J.J. Monnier J.A. et al.Minimum instructional and program-specific administrative costs of educating residents in internal medicine.Arch Intern Med. 2001; 161: 760-766Crossref PubMed Scopus (16) Google Scholar we used a 13 four-week rotation schedule per year and defined outpatient intensive programs as 33% inpatient and 67% outpatient, inpatient intensive programs as 67% inpatient and 33% outpatient, and traditional as 60% inpatient and 40% outpatient. All models included 3 months of night float. The distribution of rotation months and the annual teaching costs per resident by rotation and program type are presented in Table 4.Table 2Basic Rotational AssumptionsRotationMedian No. of Residents Per Faculty (Range)Median No. of Teaching Hours Per Week (Range)Inpatient medicine3 (3-6)11.5 (5-30)Critical care4.5 (3-8)10 (4-40)Inpatient consultation1.5 (1-3)11 (4-20)Night float1.5 (1-6)5 (1-10)Emergency medicine2.5 (1-4)6 (0-10)Subspecialty clinic1.3 (1-4)6.5 (1-40)Continuity clinic4 (1-4)4 (1-40)∗Median teaching hours per 1 clinic session.∗ Median teaching hours per 1 clinic session. Open table in a new tab Table 3Median Faculty Salary RatesMedian Faculty Salary Rates∗As reported by the AAMC at associate professor level and not including benefits.Annual ($)Hourly ($)†A total of 40 hours/week for 52 weeks.Hospital internal medicine faculty195,00094Critical care and subspecialty211,000101Continuity clinic–general internists185,00089Ambulatory–subspecialists207,667100Emergency medicine252,000121∗ As reported by the AAMC at associate professor level and not including benefits.† A total of 40 hours/week for 52 weeks. Open table in a new tab Table 4Annual Costs of Teaching Per Resident by Program TypeRotationInpatient Intensive(mo)Inpatient Intensive Cost of Teaching Per Resident ($)Outpatient Intensive (mo)Outpatient Intensive Cost of Teaching Per Resident ($)Traditional (mo)Traditional Cost of Teaching Per Resident ($)Inpatient internal medicine1014,413686481014,413Critical care544893269343591Inpatient consultation926,66438888926,664Night float337603376033760Emergency medicine111611116111161Ambulatory block (subspecialist)817,7602044,400919,980Ambulatory continuity clinic130†Total number of sessions.11,570260†Total number of sessions.23,140130†Total number of sessions.11,570Vacation303030Total + 27% fringe39∗One month equals 4 weeks. Ambulatory continuity clinic not included in calculation of total months because it coincides with other rotations.101,36839117,71639103,047∗ One month equals 4 weeks. Ambulatory continuity clinic not included in calculation of total months because it coincides with other rotations.† Total number of sessions. Open table in a new tab Variable costs of training were those dependent on the size of the program. We examined 4 different sizes (24, 65, 120, and 160 residents) in each of the 3 program types because they have different associate program director and core faculty requirements. We attributed variable costs to the following: (1) 1% added time factor for program director for every 5 residents over 36 total; (2) required number of associate program director salaries and fringe at 20 hours per week; (3) associate program director development; (4) total per resident costs of teaching calculated in Table 4; (5) minimum core faculty numbers and time using median generalist salary in the 2011 AAMC report plus 5% administrative and development time; (6) added costs to observe/document next accreditation system milestones estimated at 10 hours of added faculty time per resident per year; (7) recruitment costs extrapolated from the APDIM survey ([mean spent/mean number of residents] × number of residents in program); (8) resident support estimated from the APDIM survey including the sum of mean scholarly stipend, educational stipend, other (simulation, objective structured clinical examination, advanced cardiac life support/basic life support certification), meals, graduation parties, social events, teaching awards divided by the mean number of residents multiplied by the program's number of residents; and (9) resident salaries and fringe calculated by using the mean of median postgraduate salaries from the 2011 AAMC resident stipend report and assuming equal numbers of residents at each level.9Survey of residents/fellows stipends and benefits report. October 2011. American Association of Medical Colleges. Available at: https://www.aamc.org/download/265452/data/2011stipendreport.pdf. Accessed January 12, 2014.Google Scholar The administrative fixed costs of training are presented in Table 1. A fringe rate of 27% is included in all salary-derived values. Basic rotational assumptions are shown in Table 2. Teaching time was defined as preparing for rounds, teaching, giving feedback, and completing evaluations including milestones. In general, there was agreement between the subcommittee's collective experience and the results on our internal survey; however, there was substantial variation between programs represented in the internal survey regarding the numbers of hours a faculty member spends in teaching. Median annual salaries are shown in Table 3. The annual cost per resident of teaching based on program type, rotation type, estimated teaching hours per year, ratio of house officers to faculty, and estimated hourly salary of faculty by discipline are reported in Table 4. Inpatient intensive programs were calculated to have the lowest annual per-resident teaching costs at $101,368 (including fringe). Outpatient intensive programs were calculated as having the highest annual per-resident teaching costs at $117,716, and traditional programs were similar to the costs of inpatient intensive programs at $103,047. The total fixed and variable cost calculations for the different program types and adjusted for size are presented in Table 5 (including a fringe benefit rate of 27% on salary-derived values). Fixed costs for all residencies independent of size or model are shown in Table 1 and total $454,981. As expected, variable costs increase as program size increases, in large part because of increased core faculty and associate program director numbers, teaching time, resident salary support, and administrative costs. Variable costs were highest in the outpatient intensive programs (with a per-resident cost range of $198,085 in programs with 160 residents to $224,668 in programs with 24 residents) and were lowest in inpatient intensive programs (with per-resident cost ranges of $181,737 in larger programs to $208,320 in programs with 24 residents). The increased per-resident cost of outpatient intensive programs can be explained by the smaller house officer to faculty ratio seen on ambulatory rotations rather than the team-based teaching that occurs on inpatient services.Table 5Total Fixed and Variable Residency Costs in Dollars by Type of Medicine ResidencyInpatient IntensiveOutpatient IntensiveTraditionalNumber of residents246512016024651201602465120160PD time0740221,44231,6530740221,44231,6530740221,44231,653APD salary and fringe111,000222,000444,000555,000111,000222,000444,000555,000111,000222,000444,000555,000APD development150030006000750015003000600075001500300060007500Rotations/ clinical teaching2,432,8326,588,92012,164,16016,725,7202,825,1847,651,54014,125,92019,423,1402,473,1286,698,05512,365,64017,002,755Minimum core faculty time370,286462,858740,5731,018,287370,286462,858740,5731,018,287370,286462,858740,5731,018,287NAS milestones21,13457,239105,672140,89621,13457,239105,672140,89621,13457,239105,672140,896Recruitment costs10,37328,09551,86869,15810,37328,09551,86869,15810,37328,09551,86869,158Resident support∗Simulation laboratories, graduation parties, libraries, meals, research support, and board review preparation.59,926162,299299,629411,99059,926162,299299,629411,99059,926162,299299,629411,990Resident salaries1,537,6564,164,4857,688,28010,571,3851,537,6564,164,4857,688,28010,571,3851,537,6564,164,4857,688,28010,571,385Subtotal4,544,70711,696,29821,521,62429,531,5894,937,05912,758,91823,483,38432,229,0094,585,00311,805,43321,723,10429,808,624Fixed costs454,981454,981454,981454,981454,981454,981454,981454,981454,981454,981454,981454,981Total4,999,68812,151,27921,976,60529,986,5705,392,04013,213,89923,938,36532,683,9905,039,98412,260,41422,178,08530,263,605Cost per resident208,320186,943183,138181,737224,668203,291199,486198,085209,999188,622184,817183,416APD = associate program director; NAS = Next Accreditation System; PD = program director.∗ Simulation laboratories, graduation parties, libraries, meals, research support, and board review preparation. Open table in a new tab APD = associate program director; NAS = Next Accreditation System; PD = program director. Our analysis was undertaken to provide reasonable estimates of the typical costs of training internal medicine residents in the current ACGME regulatory environment in the United States. As in the model by Nasca et al2Nasca T.J. Veloski J.J. Monnier J.A. et al.Minimum instructional and program-specific administrative costs of educating residents in internal medicine.Arch Intern Med. 2001; 161: 760-766Crossref PubMed Scopus (16) Google Scholar and a 2013 analysis by the RAND Corporation for MedPAC,11Wynn BO, Smalley R, Cordasco KM. Does it cost more to train residents or to replace them? A look at the costs and benefits of operating graduate medical education programs. RAND Corporation. Available at: http://www.rand.org/pubs/research_reports/RR324.html. Accessed December 15, 2014.Google Scholar we found economies of scale in that smaller outpatient intensive programs have higher per resident costs than larger inpatient intensive programs. Furthermore, we found that our estimated actual costs, which ranged from $181,737 per internal medicine resident in large inpatient intensive programs to $209,999 per internal medicine resident in small outpatient intensive programs, far exceeded that of Steinman's4Steinmann A.F. Threats to graduate medical education funding and the need for a rational approach: a statement from the Alliance for Academic Internal Medicine.Ann Intern Med. 2011; 155: 461-464Crossref PubMed Scopus (15) Google Scholar estimation of $130,000 per US resident. Our average cost estimate range of $183,138 to $199,486 per internal medicine resident for programs of 120 residents also exceeded the estimates in the RAND Corporation11Wynn BO, Smalley R, Cordasco KM. Does it cost more to train residents or to replace them? A look at the costs and benefits of operating graduate medical education programs. RAND Corporation. Available at: http://www.rand.org/pubs/research_reports/RR324.html. Accessed December 15, 2014.Google Scholar report for MedPAC. In that report, using 2008 cost reports and after which new program requirements and the next accreditation system were implemented, the average direct graduate medical education cost per resident for hospitals with 100 or more residents was $140,331 and for hospitals in which less than 25% or more than 50% of trainees are in primary care specialties ranged from $153,162 to $161,779. In 2012, the CMS contributed $2.67 billion in direct medical education to support the 115,000 trainees, averaging $23,217 in direct medical education support per resident, implying that the approximately $160,000 per resident shortfall (compared with our average per resident cost estimate) is made up by cross-subsidy with indirect medical education dollars and subsidy from hospitals, faculty practice plans, state support, or philanthropy. If indirect medical education dollars were directed entirely to resident support in 2012, the average per resident amount would be $64,783, which by itself leaves a per resident shortfall approximating $120,000. Recent recommendations by MedPAC and the 2010 National Commission on Fiscal Responsibility and Reform suggested major reductions in indirect medical education support to teaching hospitals12Ward R.C. Mainiero M.B. Graduate medical education in the era of health care reform.J Am Coll Radiol. 2013; 10: 708-712Abstract Full Text Full Text PDF PubMed Scopus (9) Google Scholar, 13Rye B. Assessing the impact of potential cuts in Medicare doctor-training subsidies.Bloomberg Government Study. February 28, 2012; (Available at:) (Accessed December 15, 2014)http://about.bgov.com/bgov/files/2012/03/ryestudy.pdfGoogle Scholar because the original assumptions for its creation may no longer be valid. Furthermore, significant reductions in hospital revenue from private and federal payers may result from provisions of the Patient Protection and Affordable Care Act of 2010 and the American Taxpayer Relief Act of 2013. Our analysis has some intrinsic weaknesses, including that it applies only to training in internal medicine, and estimates of compensation for faculty vary widely and ours are benchmarked to only 2 sources, the AAMC and APDIM surveys. We used an across-the-board fringe benefit rate of 27% when calculating salary numbers in our model, which may not represent the actual costs of faculty support in a given institution. Furthermore, each program has a different pattern of time commitment for faculty members and may or may not subsidize every minute that a faculty member devotes to teaching. Finally, it is often difficult for hospitals to accurately delineate all the non-care-related costs of medical education (eg, recruitment, faculty and resident development, retreats), and to approximate these hard-to-quantitate costs as accurately as possible, we sought information from the APDIM and a survey of the members of the AAIM who are directly involved in medicine programs. It is difficult for program directors, department chairs, and deans to accurately determine the total costs of training medicine residents, and we attempted to make a reasonable estimation based on available data and our collective experience. Our analysis that the cost of training residents in internal medicine in the current regulatory environment exceeds former estimates suggests that an increase in the average outlay of current federal support for graduate medical education of $51,737 to $79,999 may be needed by institutions to adequately support resident training. Currently, hospitals, sponsoring institutions, and departments of medicine are likely bearing the growing costs of training. New reimbursement models, quality initiatives, value-based purchasing, and national healthcare reform may reduce the margins that enable institutions and departments to absorb increased direct costs of training residents, and they may be compelled to consider reducing positions or closing programs. New models of graduate medical education funding should be sought to address the actual costs of training residents and to ensure that the number of residency training positions meets the needs of the nation.