Association Between Physician Burnout and Patient Safety, Professionalism, and Patient Satisfaction: A Systematic Review and Meta-analysis

Key Points

Question  Is physician burnout associated with low-quality, unsafe patient care?

Findings  This meta-analysis of 47 studies on 42 473 physicians found that burnout is associated with 2-fold increased odds for unsafe care, unprofessional behaviors, and low patient satisfaction. The depersonalization dimension of burnout had the strongest links with these outcomes; the association between unprofessionalism and burnout was particularly high across studies of early-career physicians.

Meaning  Physician burnout is associated with suboptimal patient care and professional inefficiencies; health care organizations have a duty to jointly improve these core and complementary facets of their function.

Importance  Physician burnout has taken the form of an epidemic that may affect core domains of health care delivery, including patient safety, quality of care, and patient satisfaction. However, this evidence has not been systematically quantified.

Objective  To examine whether physician burnout is associated with an increased risk of patient safety incidents, suboptimal care outcomes due to low professionalism, and lower patient satisfaction.

Data Sources  MEDLINE, EMBASE, PsycInfo, and CINAHL databases were searched until October 22, 2017, using combinations of the key terms physicians, burnout, and patient care. Detailed standardized searches with no language restriction were undertaken. The reference lists of eligible studies and other relevant systematic reviews were hand-searched.

Study Selection  Quantitative observational studies.

Data Extraction and Synthesis  Two independent reviewers were involved. The main meta-analysis was followed by subgroup and sensitivity analyses. All analyses were performed using random-effects models. Formal tests for heterogeneity (I²) and publication bias were performed.

Main Outcomes and Measures  The core outcomes were the quantitative associations between burnout and patient safety, professionalism, and patient satisfaction reported as odds ratios (ORs) with their 95% CIs.

Results  Of the 5234 records identified, 47 studies on 42 473 physicians (25 059 [59.0%] men; median age, 38 years [range, 27-53 years]) were included in the meta-analysis. Physician burnout was associated with an increased risk of patient safety incidents (OR, 1.96; 95% CI, 1.59-2.40), poorer quality of care due to low professionalism (OR, 2.31; 95% CI, 1.87-2.85), and reduced patient satisfaction (OR, 2.28; 95% CI, 1.42-3.68). The heterogeneity was high and the study quality was low to moderate. The links between burnout and low professionalism were larger in residents and early-career (≤5 years post residency) physicians compared with middle- and late-career physicians (Cohen Q = 7.27; P = .003). The reporting method of patient safety incidents and professionalism (physician-reported vs system-recorded) significantly influenced the main results (Cohen Q = 8.14; P = .007).

Conclusions and Relevance  This meta-analysis provides evidence that physician burnout may jeopardize patient care; reversal of this risk has to be viewed as a fundamental health care policy goal across the globe. Health care organizations are encouraged to invest in efforts to improve physician wellness, particularly for early-career physicians. The methods of recording patient care quality and safety outcomes require improvements to concisely capture the outcome of burnout on the performance of health care organizations.

The view that physician wellness is an indicator of the quality of health care organizations is not new—the concept was introduced decades ago and has since gained increasing support.^1-4 The most well-known inverse metric of physician wellness is burnout, defined as a response to prolonged exposure to occupational stress encompassing feelings of emotional exhaustion, depersonalization, and reduced professional efficacy.⁵ There is evidence that the prevalence of burnout in physicians is high and that its result on the personal lives of physicians is profound.⁶ The 2017 Medscape Physician Lifestyle Report suggests that 50% of physicians in the United States report signs of burnout, representing a rise of 4% within a year.⁷ Burnout is associated with increased risk for cardiovascular disease and shorter life expectancy, problematic alcohol use, broken relationships, depression, and suicide.^8,9

Despite consistent findings regarding the high prevalence of burnout and the detrimental personal consequences for physicians, research evidence about the outcome of physician burnout on the quality of care delivered to patients is less definitive. A number of empirical studies have found that physicians with burnout are more likely to be involved in patient safety incidents,⁸ fail on critical aspects of professionalism that determine the quality of patient care (eg, adherence to treatment guidelines, quality of communication, and empathy), and receive lower patient satisfaction ratings.¹⁰ Moreover, 2 recent systematic reviews have associated high burnout in health care professionals with the receipt of less-safe patient care.^11,12 However, these reviews have significant limitations. One included heterogeneous samples of health care professionals rather than physicians in particular, making quantification of these links using meta-analysis risky¹²; the second focused on a limited number of studies.¹¹ Both systematic reviews failed to explore complementary dimensions of patient safety, such as suboptimal care outcomes resulting from low professionalism and patient satisfaction, and neither used meta-analysis to quantify the strength of the associations.¹¹

In this systematic review, we examined whether physician burnout is associated with lower quality of patient care focusing on (1) patient safety incidents, (2) suboptimal care outcomes resulting from low professionalism, and (3) lower patient satisfaction. We also evaluated the influence of key sources of heterogeneity on these associations, including the health care setting in which physicians are working and the reporting method of patient care outcomes (physician reported, patient reported, or system recorded). This study is essential to acquire a holistic understanding of the association between physician burnout and health care service delivery and confirm the need for dynamic organization-wide resolutions to mitigate burnout.

This systematic review was conducted and reported in accordance with the Reporting Checklist for Meta-analyses of Observational Studies (MOOSE)¹³ and Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidance.¹⁴ The completed MOOSE checklist is available in eTable 1 in the Supplement.

MEDLINE, PsycInfo, EMBASE, and CINAHL were searched until October 22, 2017. The searches included combinations of 3 key blocks of terms (physicians, burnout, patient care) involving Medical Subject Headings terms and text words (eTable 2 in the Supplement). Relevant systematic reviews and the reference lists of the eligible studies were hand-searched; there were no language restrictions.

Physicians working in any health care setting were eligible for inclusion. Any quantitative study reporting data on the association between physician burnout and patient safety were eligible.

Burnout was the primary outcome evaluated with standardized measures, such as the Maslach Burnout Inventory (MBI) or equivalent. The MBI assesses the 3 dimensions of the burnout experience, including emotional exhaustion, depersonalization, and personal accomplishment, and produces separate scores for each dimension.¹⁵ We also included studies reporting measures of depression and emotional distress, as these are closely related to burnout, but these outcomes were analyzed separately.¹⁶

Patient safety incidents were defined as “any unintended events or hazardous conditions resulting from the process of care, rather than due to the patient's underlying disease, that led or could have led to unintended health consequences for the patient or health care processes associated with safety outcomes.”^17[p9] Examples of patient safety incidents are adverse events, adverse drug events, or other therapeutic and diagnostic incidents.

Professionalism operationalized was based on Stern’s 4 core principles: excellence, accountability, altruism, and humanism.¹⁸ As indicators of low professionalism, we included suboptimal adherence to treatment guidelines (eg, US Preventive Services Task Force guidelines on prescription of recommended treatments and medications, test-ordering practices, referrals to treatment or other services, and discharge), reduced professional integrity (eg, malpractice claims), poor communication practices (eg, provision of suboptimal information to patients), and low empathy. We viewed reduced professionalism as an indicator of suboptimal quality of care and a precursor of patient safety incidents¹⁹ because it involves some type of omission or commission error with potential to result in a patient safety incident. Patient satisfaction was based on patient-reported measures, such as satisfaction and perceived enablement scores.

Data to allow the computation of an effect size in each study were sought. We extracted these data from the published reports where available, and we contacted the lead authors of studies that did not report sufficient data to compute an effect size (ie, reported only P values).

Gray literature (eg, unpublished conference presentations, theses, government reports, and policy statements) was excluded. We also excluded studies that reported generic health outcomes, such as quality of life, overall well-being, or resilience.

The results of the searches were exported into EndNote (Clarivate Analytics). After removal of duplicates, a 2-stage selection process was followed. At stage 1, titles and abstracts of studies were screened for relevance. At stage 2, full texts of studies ranked as relevant in stage 1 were accessed and fully screened against the eligibility criteria. A standardized Excel data extraction spreadsheet (Microsoft Inc) was devised to facilitate the extraction of (1) descriptive data from the studies, including study characteristics (eg, design and setting), participant characteristics (eg, age, sex) and main outcome measures (physician burnout measure, indicators of suboptimal care), and (2) quantitative data for computing effect sizes in each study. The data extraction spreadsheet was piloted in 5 randomly selected studies before use. We used 3 widely used fundamental criteria adapted from guidance on the assessment of observational studies (cross-sectional and cohort studies)²⁰: (1) a response rate of 70% or greater at baseline (yes, 1; no/unclear, 0), (2) control for confounding factors in analysis (yes, 1; no/unclear, 0), and (3) study design (longitudinal, 1; cross-sectional, 0).

Ratings were not used to exclude articles prior to synthesis but to provide a context for assessing the validity of the findings (eg, sensitivity analyses). Screening, data extraction, and the critical appraisal were independently undertaken by 2 reviewers (M.P. and K.G.). The interrater agreement was high (κ coefficients, 0.91, 0.89, and 0.88, respectively). Any disagreements were resolved by discussion and the involvement of a third reviewer.

The primary outcome was the association of burnout (overall burnout, emotional exhaustion, depersonalization, and personal accomplishment) with suboptimal patient care indicators (patient safety incidents, professionalism, and patient satisfaction). Secondary outcomes were depression and emotional distress with suboptimal patient care. Odds ratios (ORs) together with 95% CIs were calculated for all primary and secondary outcomes in each study. Studies were eligible for inclusion in more than 1 analysis (eg, if they reported all 3 dimensions of burnout and/or >1 suboptimal patient care outcome), but none of the studies is represented twice in the same analysis to avoid double counting. Odds ratios were typically computed from dichotomous data (number/rates of safety incidents), but continuous data (ie, means) were also converted to ORs using appropriate methods proposed in the Cochrane Handbook.²¹ An OR greater than 1 indicates that burnout is associated with increased risk of suboptimal patient care outcomes, whereas an OR less than 1 indicates that burnout is associated with reduced risk for suboptimal patient care outcomes. Owing to high heterogeneity, random-effects models were applied to calculate pooled ORs in all analyses.^22,23

Heterogeneity was assessed using the I² statistic, with values of 25%, 50%, and 75% indicating low, moderate, and high heterogeneity, respectively.²⁴ A sensitivity analysis was performed to evaluate the stability of the results when only studies less susceptible to risk of bias were retained in the analysis. One prespecified subgroup analysis explored whether the main findings were influenced by the reporting method of patient care outcomes (physician reported, system based). We also conducted 2 post hoc subgroup analyses to examine whether the geographic region of the studies (US vs non-US studies) and the career stage of physicians (residents/early career vs middle/late career) influenced the main findings. We inspected the symmetry of the funnel plots and performed the Egger test to examine for publication bias.²⁵ All meta-analyses were performed in Stata, version 14 (StataCorp) using the metaan command.²⁶ Funnel plots were constructed using the metafunnel command,²⁷ and the Egger test was computed using the metabias command.²⁸

We identified 5234 records and, following the removal of duplicates, we screened 3554 titles and abstracts for eligibility in this review. After screening, 47 studies met our inclusion criteria.^{4,8,10,22,23,29-70} The flowchart of the study selection process is presented in Figure 1.

Descriptive details of the eligible studies are presented in the Table. Across all 47 studies, a pooled cohort of 42 473 physicians was formed. The median number of recruited physicians was 243 (range, 24-7926; 25 059 [59.0%] men). Median age of the physicians was 38 years (range, 27-53 years). Our pooled cohort consisted of physicians at different stages of their career; 21 studies were primarily based on residents and early-career (≤5 years post residency) physicians (44.7%) and 26 considered experienced physicians (55.3%). Thirty studies were based on hospital physicians (63.8%), 13 studies were based on primary care physicians (27.7%), and 4 were based on mixed samples of physicians across any health care setting (8.5%). Thirty-seven studies were cross-sectional (78.7%) and 10 were prospective cohort studies (21.3%). Twenty-three of the studies were conducted in the United States (48.9%), 15 in Europe (31.9%), and 9 elsewhere (19.1%).

All studies used validated measures of physician burnout. The Maslach Burnout Inventory (the original or revised iterations) was the most common measure of burnout (41 of 43 studies that reported data on burnout [87.2%]).⁵ Fourteen studies reported secondary measures of depression and emotional distress, which were analyzed separately. Twenty-one studies reported patient safety incidents, 28 reported indicators of low professionalism, and 7 studies reported measures of patient satisfaction. Nine studies reported more than 1 of these outcomes. Patient safety incidents and suboptimal patient care due to low professionalism were assessed based on physician self-reports across the majority of the studies (17 of 21 [81.0%] and 22 of 29 [75.9%] studies, respectively), whereas the remaining used patient record reviews and surveillance systems. Patient satisfaction was based on self-reports by patients.

Nineteen studies reported a response rate of 70% or greater at baseline (40.4% met criterion 1), 36 studies adjusted for confounders in the analyses (76.6% met criterion 2), and 10 studies were prospective cohorts (21.3% met criterion 3). In total, 20 (42.6%) studies met at least 2 of the 3 quality criteria, whereas only 5 studies (10.6%) met all 3 criteria. The results of the critical appraisal assessment are presented in eTable 3 in the Supplement.

The pooled outcomes of the main analysis indicated that physician overall burnout is associated with twice the odds of involvement in patient safety incidents (OR, 1.96; 95% CI, 1.59-2.40; I² = 97.7%) (Figure 2). All dimensions of burnout were associated with significantly increased odds of involvement in patient safety incidents (emotional exhaustion: OR, 1.73; 95% CI, 1.43-2.08; I² = 97.3%; depersonalization: OR, 1.94; 95% CI, 1.29-2.90; I² = 99.3%; personal accomplishment: OR, 1.49; 95% CI, 1.23-1.81; I² = 96.4%). The heterogeneity across all analyses was moderate to high in most analyses as indicated by the I² values.

Symptoms of depression/emotional distress in physicians were associated with a 2-fold increased risk of involvement in patient safety incidents (OR, 2.38; 95% CI, 1.84-2.92; I² = 74%) (eFigure 1 in the Supplement).

Overall burnout in physicians was associated with twice the odds of exhibiting low professionalism (OR, 2.31; 95% CI, 1.87-2.85; I² = 89.5%) (Figure 3). Particularly, depersonalization was associated with a 3-fold increased risk for reporting low professionalism (OR, 3.00; 95% CI, 2.02-4.43; I² = 93.6%; P < .001). Emotional exhaustion and reduced personal accomplishment were associated with over 2.5-fold increased odds for low professionalism (OR, 2.71; 95% CI, 1.91-3.86; I² = 89.6%; OR, 2.49; 95% CI, 1.69-3.67; I² = 89.7%). Symptoms of depression or emotional distress were associated with 1.5 times increased risk for low professionalism (OR, 1.68; 95% CI, 1.44-1.92; I² = 61%) (eFigure 2 in the Supplement).

Overall burnout in physicians was associated with a 2-fold increased odds for low patient-reported satisfaction (OR, 2.28; 95% CI, 1.42-3.68; I² = 90.5%) (Figure 4). Particularly, depersonalization was associated with 4.5-fold increased odds for low patient-reported satisfaction (OR, 4.50; 95% CI, 2.34-8.64; I² = 91.6%). Personal accomplishment was also associated with over 2-fold increased odds for low patient-reported satisfaction (OR, 1.94; 95% CI, 1.25-3.01; I² = 72.2%), whereas emotional exhaustion was not significantly associated with patient-reported satisfaction (OR, 2.35; 95% CI, 0.83-6.64; I² = 96.6%).

No substantial funnel plot asymmetry was observed in the main analyses. The Egger test indicated that the results were not influenced by publication bias (Egger test P = .07) (eFigure 3 in the Supplement).

The pooled outcome sizes indicating an association derived by the studies with higher-quality scores (studies that met 2 of the 3 criteria) were similar to the pooled outcome sizes of the main analyses (overall burnout and safety incidents: OR, 1.93; 95% CI, 1.45-2.41; overall burnout and professionalism: OR, 2.32; 95% CI, 1.66-2.98).

Burnout was associated with twice the risk of physician-reported safety incidents and low professionalism (OR, 2.07; 95% CI, 2.03-2.11; I² = 65%; OR, 2.67; 95% CI, 2.19-3.15; I² = 56%, respectively), whereas the association between physician burnout and system-recorded safety incidents and low professionalism was statistically nonsignificant or marginally significant (OR, 1.00; 95% CI, 0.81-1.18; I² = 15%; OR, 1.15; 95% CI, 1.02-1.31; I² = 10%, respectively). Both subgroup differences were statistically significant (Cohen Q = 8.14 and 7.78; P = .007).

The pooled associations of physician burnout with patient safety incidents and low professionalism did not differ significantly across studies based on US physicians (OR, 1.69; 95% CI, 1.46-1.92; I² = 71%; OR, 2.02; 95% CI, 1.59-2.44; I² = 75%, respectively) and studies based on physicians in other countries (OR, 1.96; 95% CI, 1.62-2.30; I² = 82%; OR, 1.97; 95% CI, 1.57-2.38; I² = 87%, respectively). The Cohen Q tests for both analyses were statistically nonsignificant.

The pooled association of burnout with patient safety incidents did not differ significantly across studies based on residents and early-career physicians and studies based on middle- and late-career physicians (OR, 1.73; 95% CI, 1.46-2.00; I² = 79% vs OR, 1.87; 95% CI, 1.49-2.25; I² = 76% respectively; Cohen Q = 1.32; P = .17). However, the pooled association of burnout with low professionalism was significantly larger across studies based on residents and early-career physicians, compared with studies based on middle- and late-career physicians (OR, 3.39; 95% CI, 2.38-4.40; I² = 23% vs OR, 1.73, 95% CI, 1.46-2.01; I² = 67%, respectively; Cohen Q = 7.27; P = .003).

This systematic review and meta-analysis provides robust quantitative evidence that physician burnout is associated with suboptimal patient care in the process of health care service delivery. We found that physicians with burnout are twice as likely to be involved in patient safety incidents, twice as likely to deliver suboptimal care to patients owing to low professionalism, and 3 times more likely to receive low satisfaction ratings from patients. The depersonalization dimension of burnout appears to have the most adverse association with the quality and safety of patient care and with patient satisfaction. The association of burnout with low professionalism was particularly strong among studies based on residents and early-career physicians. The reporting method of patient safety incidents and professionalism had a significant influence on the results, suggesting that improved assessment standards for patient safety and professionalism are needed in the health care field.

Two previous systematic reviews have associated burnout in health care professionals with patient safety outcomes.^11,12 In the present review, we undertook a meta-analysis, enabling the quantification of these links and the exploration of key sources of heterogeneity among the studies. We focused on physicians but established links between burnout/stress and a wider range of patient care indicators, including patient safety incidents, low professionalism, and patient satisfaction. We chose to focus on physicians because the function of any health care system primarily relies on physicians, but evidence suggests that physicians are 2 times more likely to experience burnout than any other workers, including other health care professionals.^1,6,71 We thought it is critical, therefore, to better understand the association between physician burnout and patient safety, professionalism, and patient satisfaction. We chose to investigate a wider range of patient care indicators because, although professionalism and patient satisfaction are precursors of safety risks with potential to lead to active patient safety incidents,¹⁹ to our knowledge, previous research has not systematically reviewed the association between burnout/stress and these outcomes. Moreover, aspects of professionalism, such as poor empathy and suboptimal patient-physician rapport, could result in underinvestigated but important adversities for patients, such as psychological harm and an overall negative experience of health care.

We found that physician burnout is associated with a reduced efficiency of health care systems to deliver high-quality, safe care to patients. Preventable adverse events cost several billions of dollars to health care systems every year.⁷² Physician burnout therefore is costly for health care organizations and undermines a fundamental societal need for the receipt of safe care. Current interventions for improving health care quality and safety have mainly focused on identifying and monitoring vulnerable patients (eg, patients with complex health care needs) and occasionally vulnerable systems.^73,74 Our findings support the view that existing care quality and patient safety standards are incomplete; a core but neglected contributor is physician wellness.^1-4 This recommendation is in accordance with all well-recognized patient safety classification systems (eg, World Health Organization), which concur that there are 3 major contributory factors to patient safety incidents: patient, health care system, and clinician factors.

High depersonalization in physicians was particularly indicative that patient care could be at risk, as it had associations with both increased patient safety incidents and reduced professionalism. Depersonalization was also associated with lower patient satisfaction, suggesting that its results can be perceived by patients. These findings are consistent with existing evidence showing that depersonalization is related to low professionalism.^75,76 Depersonalization scores in physicians could be measured by health care organizations together with other well-established quality strategies to guide system-level interventions for improving quality of health care and patient safety.

Most of the studies relied on patient care outcomes that were self-reported by physicians. However, we failed to show significant links between physician burnout and patient safety outcomes recorded in the health care systems (eg, the health records of patients, surveillance). Concerns have frequently been raised about poor and inconsistent system recording of patient safety outcomes.⁷⁷ As such, our findings suggest that existing system-based assessment methods are incomplete and less sensitive to the full range of patient safety outcomes reported by physicians and patients. These uncaptured safety outcomes might include “near misses,” but may also concern incidents different in nature, such as psychological harm, that do not result in directly observable patient harm but may affect the physician-patient relationship and indirectly harm both parties.

Reporting systems for quality of care and patient safety outcomes require revision and better standardization across health care organizations. This standardization will enable larger and more rigorous studies of the association between physician burnout and key aspects of patient care that will be accessible at an organizational level and affect policy decisions. An alternative explanation for this finding is that physicians’ perceptions of safety are unreliable; however, this conclusion is not supported by previous research suggesting that staff-reported patient safety outcomes overlap with objective safety indicators.^78,79 That said, association between burnout and self-criticism on physicians’ reports and patient safety outcomes warrants further investigation.

Another finding is that studies based on resident and early-career physicians reported stronger links between burnout and low professionalism compared with studies based on middle- and late-career physicians. It is likely that burnout signs among residents and early-career physicians have detrimental associations with their work satisfaction, professional values, and integrity.^80-82 Health care organizations have a duty to support physicians in the demanding transition from training to professional life. Residents will be responsible for the health care delivery for over 2 decades in the future. Investments in their wellness and professional values, which are largely shaped during early-career years, are perhaps the most efficient strategy for building organizational immunity against workforce shortages and patient harm/mistrust.

This study has both strengths and limitations. We undertook a rigorous quantitative assessment of the association between burnout and patient care quality and safety in a pooled sample of more than 42 000 physicians. Meta-analysis allowed us to compare the results across individual studies, examine the consistency of associations, and explore variables that might account for inconsistency.

However, there are also limitations. A wide range of outcomes was included in this review, and some outcomes pooled together in the same subcategory exhibited substantial variation (eg, professionalism). Similarly, although we focused on physicians, this is a broad research population of health professionals working in various health care settings and specialties. We accounted for the large heterogeneity by applying random-effects models to adjust for study-level variations and by undertaking subgroup analyses to explore key factors that may account for variation. We only explored the outcome of basic sources of heterogeneity because multiple subgroup analyses inflate the probability of finding false results.⁸³ We excluded gray literature because the quality of research contained in the gray literature is generally lower and more difficult to combine with research contained in peer-reviewed journal articles.⁸⁴ The visual inspection of the funnel plot and Egger test did not identify evidence of publication bias in any of our analyses, which supports our decision. However, we cannot fully eliminate the possibility that the exclusion of gray literature has introduced undetected selection bias. Finally, the design of the original studies (mostly cross-sectional) imposes limits on our ability to establish causal links between physician burnout and patient safety, professionalism, and patient satisfaction and the mechanisms that underpin these links.

The primary conclusion of this review is that physician burnout might jeopardize patient care. Physician wellness and quality of patient care are critical and complementary dimensions of health care organization efficiency. Investments in organizational strategies to jointly monitor and improve physician wellness and patient care outcomes are needed. Interventions aimed at improving the culture of health care organizations as well as interventions focused on individual physicians but supported and funded by health care organizations are beneficial.^2,85,86 They should therefore be evaluated at scale and implemented.

Accepted for Publication: June 15, 2018.

Published Online: September 4, 2018. doi:10.1001/jamainternmed.2018.3713

Retraction: This article was retracted on May 18, 2020.

Correction: This article was corrected online March 4, 2019, to correct data-entry errors in the Table, typographical errors in Figure 2, and a reversed description in Figure 4 for favors low and favors high patient satisfaction.

Corresponding Author: Maria Panagioti, PhD, National Institute for Health Research (NIHR) School for Primary Care Research, NIHR Greater Manchester Patient Safety Translational Research Centre, Manchester Academic Health Science Centre, University of Manchester, Williamson Building, Oxford Road, Manchester M13 9PL, United Kingdom (maria.panagioti@manchester.ac.uk).

Author Contributions: Dr Panagioti 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: Panagioti, Geraghty, Panagopoulou, Chew-Graham, Peters, Riley, Esmail.

Acquisition, analysis, or interpretation of data: Panagioti, Geraghty, Johnson, Zhou, Hodkinson.

Drafting of the manuscript: Panagioti, Hodkinson.

Critical revision of the manuscript for important intellectual content: Panagioti, Geraghty, Johnson, Zhou, Panagopoulou, Chew-Graham, Peters, Hodkinson, Riley, Esmail.

Statistical analysis: Panagioti, Hodkinson.

Obtained funding: Panagioti.

Administrative, technical, or material support: Geraghty, Johnson, Zhou.

Supervision: Panagioti, Panagopoulou.

Conflict of Interest Disclosures: None reported.

Funding/Support: This study was funded by the United Kingdom National Institute of Health Research (NIHR) School for Primary Care Research (project 298) and the NIHR Greater Manchester Patient Safety Translational Research Centre funded Dr Panagioti’s time contributed to this project. The research team members were independent from the funding agencies.

Role of the Funder/Sponsor: The funding organizations 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; and decision to submit the manuscript for publication.

Disclaimer: The views expressed in this publication are those of the authors and not necessarily those of the National Health Service, the NIHR, or the Department of Health.