eAppendix. Search Strategies
Customize your JAMA Network experience by selecting one or more topics from the list below.
Wasson LT, Cusmano A, Meli L, et al. Association Between Learning Environment Interventions and Medical Student Well-beingA Systematic Review. JAMA. 2016;316(21):2237–2252. doi:10.1001/jama.2016.17573
Copyright 2016 American Medical Association. All Rights Reserved.
What undergraduate medical education learning environment interventions are associated with improved emotional well-being among medical students?
In a systematic review of the medical literature, only 28 articles described empirically evaluated interventions and only 3 included randomization, so methodologic rigor was limited. However, some data support preclinical pass/fail grading, mental health programs, wellness programs, mentoring programs, curricular restructuring, and multicomponent program reform.
There is limited evidence to support learning environment interventions for improvement of emotional well-being among medical students. High-quality research is needed.
Concerns exist about the current quality of undergraduate medical education and its effect on students’ well-being.
To identify best practices for undergraduate medical education learning environment interventions that are associated with improved emotional well-being of students.
Learning environment interventions were identified by searching the biomedical electronic databases Ovid MEDLINE, EMBASE, the Cochrane Library, and ERIC from database inception dates to October 2016. Studies examined any intervention designed to promote medical students’ emotional well-being in the setting of a US academic medical school, with an outcome defined as students’ reports of well-being as assessed by surveys, semistructured interviews, or other quantitative methods.
Data Extraction and Synthesis
Two investigators independently reviewed abstracts and full-text articles. Data were extracted into tables to summarize results. Study quality was assessed by the Medical Education Research Study Quality Instrument (MERQSI), which has a possible range of 5 to 18; higher scores indicate higher design and methods quality and a score of 14 or higher indicates a high-quality study.
Twenty-eight articles including at least 8224 participants met eligibility criteria. Study designs included single-group cross-sectional or posttest only (n = 10), single-group pretest/posttest (n = 2), nonrandomized 2-group (n = 13), and randomized clinical trial (n = 3); 89.2% were conducted at a single site, and the mean MERSQI score for all studies was 10.3 (SD, 2.11; range, 5-13). Studies encompassed a variety of interventions, including those focused on pass/fail grading systems (n = 3; mean MERSQI score, 12.0), mental health programs (n = 4; mean MERSQI score, 11.9), mind-body skills programs (n = 7; mean MERSQI score, 11.3), curriculum structure (n = 3; mean MERSQI score, 9.5), multicomponent program reform (n = 5; mean MERSQI score, 9.4), wellness programs (n = 4; mean MERSQI score, 9.0), and advising/mentoring programs (n = 3; mean MERSQI score, 8.2).
Conclusions and Relevance
In this systematic review, limited evidence suggested that some specific learning environment interventions were associated with improved emotional well-being among medical students. However, the overall quality of the evidence was low, highlighting the need for high-quality medical education research.
Medical schools strive to educate knowledgeable, caring, and professional physicians and pay particular attention to opportunities for improving the undergraduate medical education (UME) learning environment as they realize its influence on the education of future physicians.1
Quiz Ref IDA critical element of the learning environment is its effect on student well-being. Although matriculating US medical students begin training with significantly lower rates of depression and burnout and report better mental and emotional quality of life than other college-educated young adults,2 their reported well-being decreases during the UME years. The reported rate of moderate to severe depression is approximately 14% and of burnout symptoms is 52%—higher than reported by other graduate students or population control samples.3,4 Studies indicate that up to 11% of medical students report suicidal ideation.5
The Association of American Medical Colleges includes in its vision for improving medical education “the health and well-being of learners.”6 This systematic review evaluated the association between UME learning environment interventions and the emotional well-being of students.
Potentially relevant articles were identified (Figure) by searching the biomedical electronic databases Ovid MEDLINE, EMBASE, the Cochrane Library, and ERIC from database inception dates to October 2016 (eAppendix in the Supplement). Additional records were identified by scanning the reference lists of relevant studies and reviews published between May 2011 and October 2016 and by using the “similar articles” feature in PubMed and the “cited reference search” in Web of Science. We searched for gray literature (“that which is produced on all levels of government, academics, business and industry in print and electronic formats, but which is not controlled by commercial publishers”)7 through ongoing trial registries, academic dissertations, and websites of relevant organizations (eg, Association of American Medical Colleges) (eAppendix in the Supplement).
Studies had to have examined the outcomes associated with any intervention aiming to promote students’ emotional well-being in the setting of an academic US medical school. The well-being outcome had to be obtained through surveys, semistructured interviews, or other quantitative methods. Open-ended response formats were excluded because their methodologic quality could not be appraised with the instrument used in this review. Medical education interventions measured with open-ended responses have been reviewed and appraised elsewhere.8,9
Study quality was assessed using the Medical Education Research Study Quality Instrument (MERSQI), which was developed to appraise the methodologic quality of quantitative medical education research.10 MERSQI scores have been positively correlated with editorial decisions to publish and with the presence of external funding for the research conducted.10 The instrument is based on 10 design and methods criteria: study design, number of institutions studied, response rate, data type, internal structure, content validity, criterion validity, appropriateness of data analysis, complexity of analysis, and outcome level. These criteria form 6 domains, each with a maximum score of 3 and a minimum of 0 or 1, that sum to produce a total score that ranges from 5 to 18.
The MERSQI was preferred to the Newcastle-Ottawa Scale–Education (NOS-E), another assessment tool for medical education research quality, because it was found to have generally higher interrater reliability (0.68-0.89)11 than the NOS-E. This may be due to its more objective assessments of design strengths and weaknesses, although it omits items on the comparability of groups and blinding.11 Although there are no defined cutoff values differentiating high-quality from low-quality study methods, 1 study used a MERSQI score of 14.0 or higher as an a priori cutoff of high quality.12
Two review authors (L.M. and L.F.) independently scanned the title or abstract of all search results to determine which studies required further assessment, investigated all potentially relevant articles as full text, selected studies to include in this review, assigned a MERSQI score for each, and calculated a mean quality score across studies. Data disagreements were resolved by consultation with the third and fourth review authors (L.T.W. and I.L.). The original intention noted in the study protocol was to conduct a meta-analysis, but because of the considerable variation in the interventions, study designs, and outcomes, we did not pool the studies quantitatively, as they were judged to not be combinable.69
The literature search yielded 4207 publications, of which 28 met the eligibility criteria for this systematic review (Figure). Publications were excluded if they were irrelevant or did not meet the inclusion criteria; for example, we excluded publications that focused on medical residents rather than medical students, measured academic rather than well-being outcomes, or contained interventions not focused on the learning environment. The studies included at least 8224 student participants (1 study did not report a sample size) and encompassed a variety of designs, including single-group cross-sectional or posttest only (n = 10), single-group pretest/posttest (n = 2), nonrandomized 2-group (n = 13), and randomized clinical trial (RCT; n = 3) designs; 89.2% were conducted at a single site. They had a wide range of approaches to improving students’ well-being that are categorized and described below (pass-fail grading systems [n = 3], mental health programs [n = 4], mind-body skills education/training [n = 7], curriculum structure [n = 3], multicomponent program reform [n = 5], wellness programs [n = 4], and group-based faculty advisor/mentor programs [n = 3]). Individual study results are described below and statistical details are provided for many key findings; additional results and methods are shown in Table 1 and Table 2. The included studies’ methodologic rigor varied, with MERSQI scores ranging from 5.0 to 13.0 (mean score, 10.3; SD, 2.11 [n = 28]). The mean MERSQI score in published medical education studies, as assessed in another review, was 10.0.10 The studies with the highest-quality methods crossed all types of interventions and all types of outcome measures. The highest-scored categories tested interventions involving pass/fail grading, mental health programs, and mind-body skills education/training.
Bloodgood et al13 (n = 281; MERSQI, 11.5) and Rohe et al15 (n = 81; MERSQI, 12.0) each described that a cohort of preclinical students graded according to a pass/fail grading system, compared with an earlier student cohort evaluated according to a 5-interval grading system (A/B/C/D/F), reported statistically significantly better well-being. They reported less anxiety, depression,13 and stress15 and better well-being13 and group cohesion scores at various study time points.15 These 2 studies differed, however, in the durability of improvements. Bloodgood et al13 found no difference at 2 years between the cohort of students with a 2-year pass/fail system compared with a cohort of students with a 5-interval system on measures of anxiety (General Well-Being Schedule [GWB]14 anxiety subscore [range, 3-28; lower scores indicate more severe distress]; mean, 14.08 vs 14.20; P = .86), depression (GWB14 depression subscore [range, 2-22; lower scores indicate more severe distress]; mean, 15.56 vs 15.35; P = .71), or well-being (GWB14 well-being subscore [range, 3-18; lower scores indicate more severe distress]; mean, 10.59 vs 10.40; P = .67). Rohe et al15 reported a persistent difference at 2 years between grading cohorts on a measure of stress (Perceived Stress Scale16 [range, 0-40; higher score indicates more stress]; mean, 15.8 [SD, 6.8] vs 20.5 [SD, 7.8]; P = .01) and speculated that this difference was due to continuing reports of elevated group cohesion (Perceived Cohesion Scale19 [range, 0-36; higher scores indicate more cohesion]; mean, 33.8 [SD, 8.0] vs 29.0 [SD, 9.9]; P = .02).
Reed et al20 (n = 2056; MERSQI, 12.5) compared well-being among students at different medical schools with grading systems that were categorized as either having 3 or more intervals (eg, honors/pass/fail) or pass/fail and found that systems with 3 or more intervals were associated with statistically significantly more stress (β = 1.91; 95% CI, 1.05-2.78; P < .001) and burnout (odds ratio, 1.58; 95% CI, 1.24 to 2.01; P < .001), and a higher likelihood of considering withdrawing from medical school (odds ratio, 1.91; 95% CI, 1.30-2.80; P = .001).
Thompson et al24 (n = 120; MERSQI, 11.5) evaluated a multipronged program aimed at reducing mental health stigma and making services more accessible. The study found that significantly smaller proportions of the student cohort exposed to the program compared with the prior student cohort reported symptoms of mild or probable depression (14/58 [24.1%] vs 26/44 [59.1%]; P < .01) and suicidal ideation (1/33 [3.0%] vs 13/43 [30.2%]; P < .001).26 Seritan et al29 (number of participants not reported; MERSQI, 11.5) examined a different multipronged mental health/wellness program offering prevention, support, and enhanced clinical services, which was associated with improved student ratings of personal counseling, mental health, and stress management services.29 Percentages of self-referral to mental health services increased from a baseline rate of 50% to a postintervention rate of 91%. For both findings, statistical significance was not reported.29
Two studies evaluated programs consisting of education and a web-based mental health screening survey to facilitate students’ use of mental health services. Downs et al26 (n = 1008; MERSQI, 13.0) described a program that was associated with an increase in mental health service utilization and a non–statistically significant decrease in assessed suicide risk during the 4 years, perhaps due to low screening rates (34%). Moutier et al31 (n = 498; MERSQI, 11.5) reported that that 11% of medical students exposed to another educational program were referred to a mental health care professional, although no comparison was provided and the screening rate was also low (27%).
Two RCTs evaluated mind-body programs. Erogul et al32 (n = 58; MERSQI, 12.0) found that students randomized to attend a mindfulness program reported a significant reduction in stress after intervention (Perceived Stress Scale16; mean change, 3.63; 95% CI, 0.37-6.89; P = .03) but not at 6-month follow-up (mean change, 2.91; 95% CI, −0.37 to 6.19; P = .08). However, students in the mind-body program reported a significant increase in self-compassion that persisted at 6-month follow-up (Self-Compassion Scale33 [range, 0-5; higher score indicates more self-compassion]; mean change, 0.56; 95% CI, 0.25-0.87; P = .001).32 In the study by Holtzworth-Munroe et al34 (n = 40; MERSQI, 10.0), students randomized to a mind-body program were reported to have significantly more awareness of tension (F5,18 = 37.16; P < .001), better ability to deal with school stress (F5,18 = 5.05; P < .04), and less test anxiety at 10-week follow-up (F1,22 = 10.42; P < .005).
Three studies evaluated mind-body programs using a pretest/posttest design with nonrandomized control groups. Kraemer et al36 (n = 52; MERSQI, 12.0) found that students undergoing mind-body skills training reported significantly improved distress tolerance (Distress Tolerance Scale G37 [range, 1-5; higher scores indicate higher levels of distress tolerance]; mean change, 0.53; 95% CI, 0.92-0.14; P = .01); no difference was found for the control group. Rosenzweig et al39 (n = 302; MERSQI, 11.0) described a mindfulness-based stress reduction program associated with significant improvements in total mood disturbance (Profile of Mood States18 [range, 0-200; higher scores indicate higher mood disturbance]; intervention group pretest mean, 38.7 [SD, 33.3] vs posttest mean, 31.8 [SD, 33.89]; P = .05; control group pretest mean, 28.0 [SD, 31.2] vs posttest mean, 38.6 [SD, 32.8]; P < .001; interaction P < .001). Finkelstein et al40 (n = 72; MERSQI, 11.0) found a significant group × time interaction association with improved anxiety (F1,2 = 3.95; P < .05).
Two studies evaluating medical student mind-body programs with a pretest/posttest design without a control group also reported associations with significant improvements in well-being. Greeson et al44 (n = 44; MERSQI, 11.5) reported improved stress (Perceived Stress Scale16; pretest mean, 29.73 [SD, 9.61]; posttest mean, 20.25 [SD, 9.03]; P < .001; d = 1.38) and mindfulness (pretest mean, 29.24 [SD, 5.54]; posttest mean, 33.88 [SD, 6.13]; P < .001; d = 0.92). Bond et al46 (n = 27; MERSQI, 11.5) reported improved self-regulation (Self-Regulation Questionnaire49 [range, 1-5; higher score indicates more self-regulation]; mean change, 0.13 [SD, 0.20]; P = .003; d = −0.41) and self-compassion (Self-Compassion Scale33; mean change, 0.28 [SD, 0.61]; P = .04; d = −0.55).
Elements of curriculum structure targeted by studies identified in this review were varied. Reed et al20 (n = 2056; MERSQI, 12.5) compared elements of curriculum structure at different medical schools. Students who reported more clinical contact hours were significantly less likely to report serious thoughts of dropping out (odds ratio, 0.96; 95% CI, 0.93-1.00; P = .03). Although the number of tests was not associated with any difference in well-being, spending more time taking tests was associated with significantly higher perceived stress (β = 0.29; 95% CI, 0.10-0.84; P = .003) and lower mental quality of life (β = 2.79; 95% CI, 4.09-1.50; P < .001).20
Camp et al51 (n = 275; MERSQI, 12.0) found that students in a new problem-based learning curriculum, compared with a lecture-based one, had similar reports of depression with covariate adjustment. A prematriculation summer enrichment program for medicine and nonscience undergraduate majors from underrepresented groups described reports of gaining confidence, making friends, and perceiving an easier transition to medical school (n = 92; MERSQI, 7.0).53
Vanderbilt University restructured its medical school learning environment, which, after multiple iterations, ultimately took the form of “learning communities” or colleges within the school. These intentionally developed groups of faculty and students work together longitudinally, with functions that include mentoring, wellness programming (including mind-body skill training, career advising, and personal and professional development), and formal medical humanities coursework. Several different studies evaluated the multicomponent program at various stages of its development and implementation. Drolet and Rodgers54 (n = 116; MERSQI, 6.5) evaluated the faculty advisor/mentor program after the addition of several components and found that 95% of students reported a positive experience with the wellness program. Fleming et al55 (n = 245; MERSQI, 6.0) assessed the association of the most recent program iteration, including colleges, and found that more than 91% of students reported that colleges contributed at least somewhat meaningfully to their medical school experience. Real et al56 (n = 450; MERSQI, 10.5) reported that students credited the program in general (and more specifically, faculty mentors), the student-led programming committee, and annual retreats with lowering reported rates of burnout.
The St Louis University School of Medicine also undertook multicomponent program reform that was introduced in phases to preclinical students: (1 component) pass/fail grading for preclinical courses, reduced preclinical contact hours, extended electives, and learning communities; (2 components) addition of mind-body skills training; and (3 components) addition of anatomy course reform. As reported in a study by Slavin et al58 (n = 890; MERSQI, 12.0), phase 1 was significantly associated with improved depression, stress, and cohesion by the end of the second year of UME. The 2-component phase was associated with significantly improved anxiety, stress, and cohesion by the end of the first year of UME; depression was reported to be improved by the end of the second year of UME.58 The 3-component phase was associated with statistically significant improvements in all measures of well-being by the end of the first year, persisting through the second year of UME.58
Strayhorn59 (n = 478; MERSQI, 12.0) compared one school’s curriculum changes with a comparison school’s curriculum and found significant time × school interactions that favored the changes with regard to reported stressors (F1467 = 6.41; P = .01), mental well-being (F1460 = 9.32; P = .002), and social well-being (F1466 = 5.37; P = .02).
In a self-hypnosis training RCT, Whitehouse et al62 (n = 35; MERSQI, 12.0) reported significant improvements in anxiety (Brief Symptom Inventory70 [range, 20-80; higher scores indicate higher anxiety]; orientation mean, 59.23 [SD, 9.41]; late semester mean, 56.31 [SD, 9.29]; examination stressor mean, 58.59 [SD, 10.43]; recovery mean, 52.64 [SD, 9.66]; interaction F3,96 = 2.96; P < .05). A cross-sectional survey (n = 26; MERSQI, 9.0) about access to student support groups reported that a majority of students felt less lonely and unique with their problems.63 An evaluation of a wellness elective (n = 66; MERSQI, 7.0) reported that only a minority of students agreed or strongly agreed that it altered their report of the importance of well-being or permission for self-care or provided coping strategies (no significance values reported).64 Kushner et al65 (n = 343; MERSQI, 8.0) evaluated a wellness course that included a section on behavior change plans; of the 9 students who set mental/emotional health goals, 6 reported achieving their goals (no significance values reported).
Three studies evaluated small group–based faculty advisor/mentor programs that were formally integrated into the academic curriculum. Sastre et al66 (n = 318; MERSQI, 9.5) evaluated a program in which competitively selected faculty had protected time for advising groups of students. Compared with students with traditional one-on-one volunteer faculty advisors, intervention students were significantly more likely to report that they agreed or strongly agreed that they were satisfied with how faculty advisors promoted wellness (72% vs 27%; P < .001) and that they agreed or strongly agreed that they would feel comfortable discussing their personal stress (62% vs 24%; P < .001) or mental health (51% vs 27%; P < .001) with their advisor.66 Coates et al67 (n = 100; MERSQI, 8.0) reported that fourth-year medical students involved in an intervention said they felt connected with faculty and with classmates (no significance values reported).
The evaluation of a program exclusively for first-year students by Ficklin et al68 (n = 151; MERSQI, 7.0) reported that students stated they were better acquainted with their peers, became close with some classmates, and were helped with anxiety related to starting medical school as a result of the program, but there was no comparison group and no significance values were reported.
This systematic review identified hundreds of articles on the UME learning environment, but only a small subset contained empirically evaluated interventions. No studies included in this systematic review met a quality cutoff of 14.0.12 Improving the content and context of the delivery of UME will benefit from studies with rigorous design, objective data collection, and appropriate intervention comparators, as used in other scientific and educational fields. Despite these limitations in the evidence, there are a number of key findings from this review that may be relevant for US medical schools.
Quiz Ref IDFirst, implementation of a preclinical pass/fail grading system should be considered. All of the studies reviewed here show that a preclinical pass/fail grading system improves medical student well-being. The duration of benefit can be finite, with any positive effect perhaps more likely to persist in the context of good medical school class cohesion.15 It is also important to consider educational repercussions of changing grading systems to ensure that rigorous mastery of educational material and professional preparedness is balanced with student well-being. Two studies in this review addressed this concern by showing that pass/fail grading systems can be associated with improved well-being without any significant change in course test scores, including United States Medical Licensing Examination Step 1 and 2 scores and subsequent postresidency specialty board certification scores.13,15 This is consistent with other literature exclusively focused on academic outcomes of pass/fail grading.71-73 According to the 2014-2015 Liaison Committee on Medical Education Annual Medical School Questionnaire, 87 of the 144 participating schools used pass/fail grading systems for at least some portion of the preclinical courses.74
Second, the accessibility and quality of mental health programs for medical students, as well as any stigma associated with these programs, should be taken into account.75 Students with mental health problems may be undertreated; in one study, fewer than half of the students who reported having contemplated suicide during medical school received counseling for their depression.76 Addressing mental health conditions with a formal program that includes treatment services is essential, and a multipronged program aimed at improving awareness, reducing stigma, and improving access to mental health care professionals seems to be an efficacious approach and is associated with lower depression and suicidal ideation rates.24
Quiz Ref IDThere are specific components of mental health programs that can be critical to improving students’ well-being. Barriers to medical students’ mental health treatment reported elsewhere include concern about stigma and lack of confidentiality, such as fear of documentation in the academic record and evaluators’ knowledge of student mental health conditions with subsequent career implications.4,47,48,76,77 Medical students reported preferring help from a mental health specialist, family, or friends rather than medical school personnel47 and reported preferring accessing mental health services through a location other than the office of student affairs.48 In other studies, students have reported concerns about time, convenience of office hours, location, and financial costs.4,47,48,77 Although these are small studies of implementation issues, they are worth considering for the introduction of student mental health programs.
Quiz Ref IDThird, introducing wellness programs that teach mind-body–based stress-reduction skills should be considered. The majority of studies in this category, including 2 RCTs, indicate that such programs are associated with reduced stress, anxiety, better mood, and higher distress tolerance. This association was found even when skills were taught in condensed workshops lasting only 4 weeks,44 which is an important factor because programs must balance benefit derived from wellness programs with time investment.
Quiz Ref IDFourth, implementation of formal faculty advisor/mentor programs based in small groups and linked with curricular content should be examined. All 3 studies in this review that evaluated faulty advisor/mentor programs were highly regarded by students as a method of promoting wellness, although only 1 study tested for statistical significance.66 However, it is important that mentors do not grade students to keep their role as advisors separate from assessment to foster open communication.68 A small group–based mentoring model—rather than a one-on-one mentoring system—reduces the number of required faculty mentoring positions, allowing medical schools to have competitive selection for a subset of excellent faculty and may even enable financial support for this function.68 Outstanding faculty mentors are critical to the success of any mentoring program because they both relay explicit academic knowledge and exemplify implicit knowledge on professionalism, ethics, and values—the “hidden curriculum.”78
Fifth, the curriculum should be structured to balance clinical and nonclinical learning environments. Medical students report less burnout and stress when clinical time is increased.20 Many recent changes to curriculum have decreased clinical learning exposures, so consideration of where this movement can be reversed will be useful.
Sixth, comprehensive reform of the learning environment that incorporates many of these interventions is likely required. A detailed evaluation of the sequential implementation phases indicates that there may have been synergies among program components that were associated with improvements in medical student well-being.58
This study has a number of limitations. First, the primary studies varied widely in design, intervention content, and outcomes collected, precluding meta-analytic pooling. Second, the scope of the review was restricted to studies evaluating the quantitative effect of learning environment interventions on medical student well-being, although there are other aspects of the learning environment that deserve attention in a comprehensive redesign of the learning offered to medical students. Third, qualitative research was not included in this systematic review. Fourth, there are concerns about the ethics of randomization of education research.79,80 Historically, research conducted in established educational settings and involving normal educational practices were considered exempt from institutional review board oversight.81 However, issues of coercion and lack of informed consent about randomization of medical students when conducting learning environment interventions tests have recently been raised.82,83 These issues are complex and include whether there is a research component to the investigation of the education practice; whether there is an intent to publish; whether empirically established practices already exist; and whether the investigator has a hierarchical relationship to the participants, such as that held by a clerkship director or faculty advisor. Guidance is provided elsewhere for future UME educators to decide when and under what circumstances randomization is ethical and practical for learning environment interventions.82,83
In this systematic review, limited evidence suggested that some specific learning environment interventions were associated with improved medical student emotional well-being. However, the overall quality of these studies was low, highlighting the need for high-quality medical education research.
Corresponding Author: Karina W. Davidson, PhD, MASc, Center for Behavioral Cardiovascular Health, Columbia University Medical Center, 622 W 168th St, PH 9-314, New York, NY 10032 (email@example.com).
Author Contributions: Drs Davidson and Wasson had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Wasson, Cusmano, Meli, Louh, Young, Davidson.
Acquisition, analysis, or interpretation of data: Wasson, Cusmano, Meli, Louh, Falzon, Hampsey, Shaffer.
Drafting of the manuscript: Wasson, Cusmano, Meli, Louh, Falzon, Hampsey, Davidson.
Critical revision of the manuscript for important intellectual content: Wasson, Meli, Louh, Young, Shaffer, Davidson.
Obtained funding: Davidson.
Administrative, technical, or material support: Cusmano, Meli, Falzon, Hampsey, Davidson.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.
Funding/Support: Dr Davidson was supported by research grant K24 HL084034 from the National Heart, Lung, and Blood Institute. Dr Wasson was supported by research grant K08 HS024598 from the Agency for Healthcare Research and Quality. Ms Falzon was supported by research grant U24AG052175 and contract S15-0142, both from the National Institutes of Health, and by Columbia University Medical Center.
Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or decision to submit the manuscript for publication.