Effects of Interdisciplinary Team Care Interventions on General Medical Wards: A Systematic Review

Importance  Improving the quality of health care for general medical patients is a priority, but the organization of general medical ward care receives less scrutiny than the management of specific diseases. Optimizing teams’ performance improves patient outcomes in other settings, and interdisciplinary practice is a major target for improvement efforts. However, the effect of interdisciplinary team interventions on general medical ward care has not been systematically reviewed.

Objectives  To describe the range of objective patient outcomes used in studies of general medical ward interdisciplinary team care, and to evaluate the performance of interdisciplinary interventions against them.

Evidence Review  We searched EMBASE, MEDLINE, and PsycINFO from January 1, 1998, through December 31, 2013, for interdisciplinary team care interventions in adult general medical wards using an objective patient outcome measure. Reference lists of included articles were also searched. The last search was conducted on January 29, 2014, and the narrative and statistical analysis was conducted through December 1, 2014. Study quality was assessed using the Cochrane Effective Practice and Organization of Care group’s tool.

Findings  Thirty of 6934 articles met the selection criteria. The studies included 66 548 patients, with a mean age of 63 years. Nineteen of 30 (63%) studies reported length of stay, readmission, or mortality rate as their primary outcome, or did not specify the primacy of their outcomes. The most commonly reported objective patient outcomes were length of stay (23 of 30 [77%]), complications of care (10 of 30 [33%]), in-hospital mortality rate (8 of 30 [27%]), and 30-day readmission rate (8 of 30 [27%]). Of 23 interventions, 16 (70%) had no effect on length of stay, 12 of 15 (80%) did not reduce readmissions, and 14 of 15 (93%) did not affect mortality. Five of 10 (50%) interventions reduced complications of care. In an exploratory quantitative analysis, the interventions did not consistently reduce the relative risk of early readmission or early mortality, or the weighted mean difference in length of stay. All studies had a medium or high risk of bias.

Conclusions and Relevance  Current evidence suggests that interdisciplinary team care interventions on general medical wards have little effect on traditional measures of health care quality. Complications of care or preventable adverse events may merit inclusion as quality indicators for general medical wards. Future study should clarify how best to implement interdisciplinary team care interventions and establish quality metrics that are credible to both health care professionals and patients in this setting.

Understanding how to better organize health care for general medical patients is an international priority for safety improvement.^1,2 Despite this need, we know relatively little about how to improve processes and outcomes for these patients. Missed opportunities to provide necessary health care are more common³ than technical procedural failings.⁴ More than any other health care setting, general medical wards generate the errors that lead to preventable deaths⁴; yet, the fundamental organization of ward-based care receives less attention than the management of specific diseases. Systems of general medical ward care are usually unchanged across successive generations of health care professionals, in contrast to the dynamic developments in the treatment of individual pathological conditions.

We identified 5 paths by which organizations may seek to improve the quality and safety of health care on the general medical ward: staffing levels and team composition; staff communication and collaboration; standardization of the processes of health care; early treatment of deteriorating patients; and changes in the local safety climate.⁵ We focus here on interdisciplinary communication and collaboration. Physicians are increasingly required to formalize their interdisciplinary practice, which is a measure widely believed to improve efficiency; yet, to our knowledge, a systematic assessment of interdisciplinary care specific to the general medical ward setting has not been published.

Defining the measures on which to base that assessment is difficult; there is no consensus as to the metrics that best reflect the quality of general medical ward care. Quality can relate to structures, processes, or outcomes.⁶ Interpretation of outcome data can be contentious, although debate largely focuses on the inappropriate use of outcomes (rather than processes) to rank institutions’ comparative performance.⁷ In research, outcome data are of more use⁷; they are also directly meaningful for patients and potentially influenced by interdisciplinary interventions. Interdisciplinary operating room safety checklists, for example, reduce surgical morbidity and possibly mortality,^8-10 and complex cultural interventions in the intensive care unit (ICU) reduce catheter-associated bloodstream infections.^11,12 We therefore conducted a systematic review to describe the range of objective patient outcomes used in studies of the general medical ward environment, and to evaluate the performance of interdisciplinary team care interventions against them. Interdisciplinary team care can be defined as the structured working practices that dictate which different health care practitioners interact together to contribute to patient care, as well as when and how they do so. This definition incorporates staff who are routinely expected to attend to the patient (eg, nurses and ward physicians) as well as those who intermittently provide specialist review and advice.

We searched the EMBASE, MEDLINE, and PsycINFO databases for English-language studies published from January 1, 1998, through December 31, 2013 (see eMethods in the Supplement for the full search strategy). We also manually searched the reference lists of included studies. Ethics approval was not required because the study was conducted using existing published data.

We included primary reports of interdisciplinary team care interventions on adult general medical wards in which the interventions were evaluated against an objective patient outcome (eTable 1 in the Supplement). Interventions that relied solely on a staff member taking on a dedicated coordinating or facilitating role (eg, case management) were excluded, as were interventions that targeted the continuation of care by a similar group during the following shift (eg, handoff processes). Interventions focused on the period of inpatient hospitalization. Randomized controlled trials (RCTs), studies with an appropriate comparison group, and interrupted time series were included. We used the criteria and terminology of the Cochrane Effective Practice and Organization of Care¹³ review group to define study type.

One investigator (S.P.) reviewed each title and abstract to determine the study’s eligibility. A second reviewer (R.D.) independently screened 15% of the titles and abstracts to assess the reliability of the inclusion and exclusion criteria. Studies that met the criteria were examined in full by both reviewers, and the final set of studies was confirmed after discussion with a third reviewer (N.S.).

Two reviewers (H.A. and S.P.) extracted data from each study. When required information was not initially available, authors were contacted for further data or clarification. Two reviewers (B.E.B. and S.P.) independently assessed the risk of bias for each study, with disagreements resolved by consensus. We used predefined criteria based on the Cochrane Effective Practice and Organization of Care group’s guidance,¹⁴ to rate studies as having low, medium, or high risk of bias. Other systematic reviews of heterogeneous health care interventions use a similar approach.¹⁵ Interventions were classified as low, medium, or high intensity. No single measure of intensity could adequately be applied to such heterogeneous studies, but factors taken into account included the frequency of the mandated use of the intervention; autonomy of the individuals on the interdisciplinary team to enact recommendations; resource provision; and follow-up.

We categorized interventions based on their focus on interdisciplinary team composition or practice, and subcategorized interdisciplinary composition interventions by the type of specialist input they delineated. We describe the studies’ outcome measures and the adjustment methods used to account for their patient case mix. For meaningful analysis, outcomes were grouped into early (occurring within 30 days of receiving the intervention) or late outcomes (31 days to 12 months). When appropriate, we assessed the correlation between the studies’ intensity ratings and the likelihood of a statistically significant effect on the given outcome (χ² analysis; differences were considered significant at P < .05). The studies were too heterogeneous for summary effect statistics to be wholly reliable. Instead, we present the available data as an exploratory analysis in weighted bubble charts, with pooled values and 95% CIs (Figures 1, 2, and 3).^16-36 The relative risk of outcomes or weighted mean difference (for length-of-stay data) was calculated with DerSimonian and Laird random-effects modeling. Between-study and within-study variances contributed to study weighting. Analyses were performed using Stata, version 12 (StataCorp LP).

A total of 6934 potentially relevant citations were identified. After excluding abstracts from conferences, there was 100% agreement between 2 reviewers (S.P. and R.D.) on the initial inclusion of a 15% sample of abstracts (1040 citations). After full text review, 30 studies met the inclusion criteria,^16-45 26 of which were identified from the initial database search and 4 from hand-searching reference lists (eFigure in the Supplement). These studies comprised 8 RCTs,^{16,17,36-39,41,43} 9 cluster RCTs,^{19,21,24,25,29,34,42,44,45} 8 nonrandomized cluster trials,^{22,23,26,30,31,33,35,40} 4 controlled before-after studies,^18,20,28,32 and 1 interrupted time series.²⁷

Study characteristics are summarized in the Table, with more complete descriptions in eTable 2 in the Supplement. Other than 1 example,²⁹ studies compared an interdisciplinary team care intervention with usual care. Interventions were grouped into 2 categories. The first group of interventions altered the composition of the interdisciplinary team who routinely attended to the patient, such as by requiring additional specialists or professionals to provide advice. The second group addressed team practice—the logistics of when, where, and how team members would work together.

The 30 studies included 66 548 patients with a mean age of 63 years and a variety of primary diagnoses. Six studies targeted patients with specific diagnoses (delirium,^16,17,33 community-acquired pneumonia,²⁹ acute stroke,³⁹ and advanced liver disease¹⁸), although they were treated on general medical rather than specialist wards. One study evaluated a service for patients taking anticoagulant medication.⁴⁴ Twenty-one studies were conducted in North America, 8 in other developed countries, and 1 in an upper-middle-income country. Trials were conducted in a range of settings, including community and urban teaching hospitals, quaternary academic facilities, and Veterans Affairs hospitals. Five were multicenter studies.

Of the 20 studies that addressed interdisciplinary team composition, 15 investigated proactive, unsolicited consultations from a specialist or specialist team.^{16,17,19-21,29,33-38,41,43,44} Consulting professionals specialized in geriatrics,^16,17,33,37 infectious diseases,^{19,21,29,34-36} intravenous therapy,³⁸ stroke,³⁹ pharmacotherapy,^41,43,44 or psychiatry.²⁰ Four studies assessed the effect of embedding additional health care professionals in rounding teams, incorporating pharmacists^40,42 or medical librarians,³⁰ or supervising medical subspecialists.¹⁸

Ten studies investigated interdisciplinary team practice, often using interdisciplinary rounds^22,24,25,45 or a package of interventions including interdisciplinary rounds.^23,26,28 Two studies assessed team localization,^31,32 with medical and nursing staff copositioned in the same geographic area in the hospital. One study described an interdisciplinary teamwork and communication program.²⁷

Of the 30 studies, 19 (63%) reported length of stay, readmission, or mortality rate as their primary outcome measure or did not specify the primacy of their outcome measures.^{17,18,20-24,28-33,36-39,41,43} The most commonly reported objective patient outcome measures were length of stay (23 of 30 [77%]), complications of care (10 of 30 [33%]), in-hospital mortality rate (8 of 30 [27%]), and 30-day readmission rate (8 of 30 [27%]). Seven of 9 studies (78%) that reported mortality rate after discharge used time points later than 30 days.^{16,17,23,37,39,41,43} Seven of 30 studies (23%) reported requirements for institutional care either at discharge or subsequently.^{16,17,23,26,33,37,39} One of 30 studies (3%) reported patients’ use of community health care services following discharge,⁴³ and 1 of 30 studies (3%) reported objective assessments of functional status.³⁷ Studies reported a mean of 3.1 objective patient outcome measures.

Several studies did not seek improvements in these outcomes per se, targeting a different aspect of care. Interventions involving infectious disease specialists, for example, were often restrictive in nature, advocating narrower-spectrum antibiotic choices or oral rather than intravenous therapy.^19,29,34,36 Those studies sought only to show that their intervention had no detrimental effect on patient outcomes, aiming instead to demonstrate improvements in processes of care.^19,29,36

No study had a low risk of bias; 7 (23%) had a medium risk of bias,^{16,17,27,28,36,39,43} and 23 (77%) had a high risk of bias.^{18-26,29-35,37,38,40-42,44,45} Eight studies (27%) involved low-intensity interventions^{17,30,34-36,39,41,43}; the remainder were of medium or high intensity.

Of the 30 studies, 16 (53%) did not adjust their results for their patient case mix.^{16,17,19-21,23,26,30,34-36,39-41,43,44} Eleven of 30 studies (37%) used patient characteristics as independent variables in a multivariate analysis to isolate the effect of their intervention.^{18,22,24,25,27,28,31,37,38,42,45} Two of 30 studies (7%) adjusted results for predicted outcomes based on diagnosis-related groups,^32,33 and 1 of 30 studies (3%) used a disease-specific severity index.²⁹ Only 7 of 30 studies (23%) accounted for autocorrelated outcomes, in which patients treated by the same physician (or team) are more likely to have similar results than those who are not.

Of 10 interdisciplinary team care interventions, 5 (50%) reduced complications of care.^{21,29,38,40,45} Three of those 5 harnessed input from infectious diseases specialists^21,29 or specialist nurses in reducing infective complications of peripheral intravenous catheters.³⁸ Both studies that reported preventable adverse events suggested improvement, either through embedding pharmacists in the clinical team⁴⁰ or formalizing interdisciplinary rounds.⁴⁵ Four of the 5 interventions that showed reduced complications of care were rated high intensity,^21,38,40,45 but there was no apparent relationship between high-intensity categorization and reduced complications (χ² = 0.476; P = .49). Zero of 7 interventions successfully reduced the need for institutional care after discharge; 4 of those 7 were rated high intensity.^16,23,26,33

Saint et al²⁸ documented significant secular reductions in length of stay. In this context, 16 of 23 interventions (70%) had no effect on the length of stay^{16,17,19,21,23-29,31,35,36,40,43}; 5 of 23 (22%) reduced the length of stay^{18,20,22,33,34}; and 1 (4%) may have increased the length of stay.⁴² A study by Esparza et al³⁰ that associated clinical librarian input with increased length of stay was too heavily confounded by patient group differences to support this conclusion, as acknowledged by its authors. Four of 5 effective interventions (80%) incorporated subspecialist input to the primary general medical ward team.^18,20,33,34 Again, 4 of 5 effective interventions were of high intensity,^18,20,22,33 without a significant overall correlation between intensity and effect on length of stay (χ² = 1.433; P = .49).

Of the studies that failed to demonstrate a statistically significant change in length of stay, 9 of 16 (56%) reported increased length of stay with their intervention.^{16,17,24-26,28,31,35,43} Singh et al³¹ explained how this result could be an unintended consequence of localization interventions. However, the exploratory quantitative summary did not detect a significant change in length of stay overall. The weighted mean difference in length of stay for interdisciplinary team composition interventions was 0.087 (95% CI, −0.083 to 0.257). For interdisciplinary team practice interventions, the weighted mean difference in length of stay was 0.001 (95% CI, −0.035 to 0.037) (Figure 1).

Twelve of 15 interventions (80%) did not reduce readmissions. One study (7%) reported readmissions at 7 days,²⁸ 8 studies (53%) reported readmissions at 30 days,^19,26-32 5 studies (33%) reported readmissions at points between 60 days and 6 months,^23,36,41-43 and 1 study (7%) reported readmissions at 1 year.³⁷ One study reported both 3- and 6-month readmissions.⁴²

One of 3 infectious disease interventions (33%) reduced 60-day readmissions related to relapsing infection³⁶; the other 2 similar interventions did not affect readmissions, albeit when measured at 30 days.^19,29 One of 4^40-43 pharmacy interventions reduced admissions (at 3 months), although the effect had dissipated by 6 months.⁴² Zero of 6 interdisciplinary team practice interventions reduced readmissions.^{23,26-28,31,32} One study suggested reduced readmissions after localizing teams and introducing service census limits, but the results did not meet the predefined statistical threshold allowing for multiple comparisons.³² Increased readmissions in the study by Esparza et al³⁰ were again attributed to confounding factors. Of the 2 interventions that reduced readmissions, one was of low intensity³⁶ and the other was of high intensity.⁴²An exploratory analysis showed interdisciplinary team care interventions had an inconsistent effect on early readmissions. Team composition interventions tended to increase early readmissions, albeit with important confounding factors in the included studies: weighted risk ratio (RR), 1.341 (95% CI, 1.120-1.607). Team practice interventions did not significantly reduce early readmissions: weighted RR, 0.995 (95% CI, 0.912-1.085) (Figure 2).

Last, 1 study reported contact with primary health care (ambulatory visits to medical specialists, family physicians, and the emergency department)⁴³ and 1 reported objective functional measures at 12 months.³⁷ The interdisciplinary team interventions had no effect on these outcomes.

Only 1 of 15 (7%) studies that reported mortality rate showed a significant effect at any time point²³ (Figure 3). Mudge et al²³ were the only group, to our knowledge, to describe a sustained increase in allied health care professional resourcing (physiotherapy, occupational therapy, social work, dietetics, and speech therapy) alongside an interdisciplinary structure to make use of those staff. The intervention was rated high intensity. However, the observed reduction in in-hospital mortality did not persist at 6 months.²³ Team practice interventions tended to reduce early mortality, although only 2 studies contributed data: weighted RR, 0.665 (95% CI, 0.449-0.986). Interdisciplinary team composition interventions did not significantly reduce early mortality: weighted RR, 0.925 (95% CI, 0.816-1.049) (Figure 3).

Studies of interdisciplinary team care interventions on general medical wards most commonly choose length of stay, complications, readmission, or mortality rates as their primary outcome measures. In this setting, we found that most interdisciplinary interventions do not affect these outcomes, although there is some evidence that improvements in interdisciplinary collaboration may reduce complications of care. Significant contemporaneous secular reductions in length of stay are reported,²⁸ which these interventions did not reduce further. Interdisciplinary interventions confined to the inpatient setting are unlikely to reduce readmissions, the need for institutional care after discharge, or mortality rates—although simultaneous investments in staffing and team structures may reduce mortality rates in the short term. Limited reports of success suggest that high-intensity delivery is necessary, but not sufficient, for an intervention to have the desired effect on any of these outcomes.

The included studies took place in a variety of settings, from safety-net hospitals to large academic facilities, and across numerous countries with disparate health care systems. In contrast to many trials, the studies included here had relatively few exclusion criteria. These findings are therefore generally applicable to a broad range of medical inpatients internationally, although the evidence on which they draw is limited in both quantity and quality.

Our results broadly correlate with earlier work evaluating nontechnical interventions. A Cochrane review⁴⁶ of RCTs could only label collaborative interprofessional interventions as “promising” rather than “proven.” Multidisciplinary team training in obstetrics does not improve clinical outcomes,^47,48 nor did an interdisciplinary safety program for surgical ward teams.⁴⁹ Similarly, interventions with the primary aim of promoting a culture of patient safety do not have a clear effect on patient outcomes.⁵⁰ The evidence that interdisciplinary interventions can improve patient outcomes appears limited to the operating room and the ICU.^8-11 Outside these structured units, we may need a more considered approach to the assessment and improvement of interdisciplinary team care.

Every medical discipline now requires an updated set of core trial outcomes—a key data set to which each new study should contribute—in addition to its specific findings.⁵¹ Interdisciplinary care in internal medicine may need this set of outcomes more than most; it is striking that both the systems of care, and our favored parameters for evaluating them, have gone unchanged for decades.^28,51 Meaningful collaboration is increasingly seen as the bellwether for a high-performing ward, but our findings suggest that changes in interdisciplinary team care are not well captured by the objective patient measures currently used in this setting. There is little apparent consensus on the essential outcomes that general medical ward interventions should report, nor how to adjust them to reflect the ward’s case mix. Other authors have also suggested that the focus on length of stay, readmissions, and mortality rate misses the point.^52-54 Good interdisciplinary team care may prolong admissions, perhaps by fostering an environment in which the patient’s medical and social needs before discharge are better identified. Thirty-day readmission rates may be largely determined by the provision of services in the community,⁵³ and mortality rates correlate poorly with quality of care.^7,55,56

In contrast, a Delphi expert consensus⁵⁷ identified the 72-hour readmission rate as a valid quality indicator for hospitalized patients with complex needs. This rate was the only preferred patient outcome among several process measures⁵⁷; other consensus panels addressing complexity have focused solely on processes as quality indicators.⁵⁸ Our analysis suggests that complications of care or preventable adverse events may also have promise as clinically meaningful outcomes reflecting the quality of general medical ward care. These outcomes, together with selected process data, merit consideration in a core data set. Too few studies reported postdischarge use of outpatient services or objective measures of patients’ functional performance to judge whether these too may be useful. We believe that patients themselves should also be invited to tell health care professionals which objective outcomes they hold most important. Whether their individual experiences of health care correlate with its technical quality remains unclear⁵⁹; nonetheless, as a group, patients are too rarely asked to participate in defining what matters.^57,58,60

With relevant outcomes better established, future studies should evaluate how best to implement interdisciplinary team care interventions. Interdisciplinary care is not a panacea; understanding the preconditions for its success (eg, adequate staffing levels or strong leadership support) and whether it potentiates other safety initiatives should be a priority. New interventions should also capitalize on 3 important phenomena. First, interdisciplinary teams become more effective as their members develop personal autonomy to enact recommendations.⁶¹ Second, practical changes that facilitate better teamwork may be more effective than simply investing in teamwork training per se.^5,27 Third, clinical and managerial engagement are crucial if successful changes are to persist after a study, and authors should describe the implementation model accompanying any intervention alongside its objective evaluation.

To our knowledge, this review is the first systematic attempt to assess interventions targeted specifically at the organization of care on general medical wards. Our definition of interdisciplinary team care establishes a meaningful class of intervention. We identified a focused collection of work that is relevant to modern practice. Previous reviews of interdisciplinary interventions brought together distinctly different contexts—for example, internal medicine and orthopedic surgery⁶¹ or psychiatry and the ICU.⁶² However, general medical wards face different challenges than other areas of the hospital, including more frequent transitions of care, geographic team dispersion, and heterogeneous patient populations for whom there is no single best pathway for diagnosis or treatment.⁵ We believe these differences argue for a specific analysis of general medical wards rather than extrapolating from other clinical departments to this unique setting. Including a range of study designs, as we have done here, facilitates a deeper understanding of context-specific improvement strategies than reviews that rely on RCTs alone.^46,63 Last, our definition of intervention intensity mirrored an updated conceptual framework for understanding implementation fidelity.⁶⁴

Nonetheless, the heterogeneous studies described here could be categorized differently. We were unable to search for unpublished trials, precluding assessment of publication bias or selective reporting. Included studies used a variety of different designs; even allowing for this fact, most have severe methodological limitations that constrain the strength of our conclusions. The absence of any studies with a low risk of bias is noteworthy, and we included controlled before-after studies with only 1 intervention or control site, which are particularly prone to confounding.¹³ We aggregated results reported at different times; our distinction between early and late outcomes is arbitrary but mimics that used in clinical practice. Although we sought to contact study authors for clarification or additional data,^{21,23,26,28,33,40,43} only some replied.^21,23,28,43 The heterogeneity of the included studies limited quantitative analysis, and the search strategy may not have been fully sensitive to all relevant articles. Our search identified only the articles with objective patient outcomes, so we have not described process measures, subjective outcomes, or patient-reported outcomes; these too are valuable.

To date, most general medical ward interdisciplinary team care interventions have not improved the objective patient outcomes used to assess them. Outside the operating room and ICU, more sensitive measures are required to evaluate interdisciplinary care. In 1995, Campion^65(p1377) argued that “bedside consultations are not a form of technology, and we should not be as severe in judging them as we are in evaluating a surgical procedure.” We propose that equally careful consideration of interdisciplinary care is merited, although we need new objective measures of its effectiveness.

Accepted for Publication: February 27, 2015.

Corresponding Author: Samuel Pannick, MA, MBBS, MRCP, Imperial Patient Safety Translational Research Center, National Institute for Health Research and Imperial College London, Room 503, Wright Fleming Building, Norfolk Place, London W2 1PG, England (s.pannick@imperial.ac.uk).

Published Online: June 15, 2015. doi:10.1001/jamainternmed.2015.2421.

Author Contributions: Dr Pannick had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Pannick, Ashrafian, Beveridge, Wachter, Sevdalis.

Acquisition, analysis, or interpretation of data: Pannick, Davis, Ashrafian, Byrne, Athanasiou, Sevdalis.

Drafting of the manuscript: Pannick.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Pannick, Ashrafian, Athanasiou.

Obtained funding: Sevdalis.

Study supervision: Davis, Ashrafian, Beveridge, Wachter, Athanasiou, Sevdalis.

Conflict of Interest Disclosures: Dr Sevdalis reports delivering team-based assessment and training interventions to hospitals in England and the United States on a consultancy basis through London Safety and Training Solutions Ltd and, in the past, holding a consultancy agreement with Green Cross Medical Ltd for the development of training materials for cancer tumor boards. Dr Wachter reports serving as the immediate past-chair of the American Board of Internal Medicine (ABIM) (for which he received a stipend) and a current member of the ABIM Foundation board; receiving a contract to UCSF from the Agency for Healthcare Research and Quality for editing 2 patient-safety websites; receiving compensation from John Wiley and Sons for writing a blog; receiving royalties from Lippincott Williams & Wilkins and McGraw-Hill for writing/editing several books; receiving a stipend and stock options for serving on the Board of Directors of IPC Healthcare; serving on the scientific advisory boards for Amino.com, PatientSafe Solutions, CRISI, QPID, and EarlySense (for which he receives stock options); and holding the Benioff endowed chair in hospital medicine from Marc and Lynne Benioff. No other disclosures were reported.

Funding/Support: This study was supported by the National Institute for Health Research via the Imperial Patient Safety Translational Research Center (http://www.cpssq.org) (Drs Pannick, Davis, Byrne, and Sevdalis); from January 2015, the National Institute for Health Research Collaboration for Leadership in Applied Health Research and Care South London at King’s College Hospital National Health Service Foundation Trust (Dr Sevdalis); the West Middlesex University Hospital National Health Service Trust (Drs Pannick and Beveridge); and the Department of Surgery and Cancer at Imperial College London (Drs Ashrafian and Athanasiou).

Role of the Funder/Sponsor: The authors were fully independent of their funding institutions, which took no part in the design or conduct of the study; the collection, management, analysis, and interpretation of the data; the preparation, review, or approval of the manuscript; or the decision to submit the manuscript for publication.

Disclaimer: The views expressed are those of the authors and not necessarily those of the National Health Service, National Institute for Health Research, or Department of Health.