Mental Health Conditions Among Patients Seeking and Undergoing Bariatric Surgery: A Meta-analysis

Importance  Bariatric surgery is associated with sustained weight loss and improved physical health status for severely obese individuals. Mental health conditions may be common among patients seeking bariatric surgery; however, the prevalence of these conditions and whether they are associated with postoperative outcomes remains unknown.

Objective  To determine the prevalence of mental health conditions among bariatric surgery candidates and recipients, to evaluate the association between preoperative mental health conditions and health outcomes following bariatric surgery, and to evaluate the association between surgery and the clinical course of mental health conditions.

Data Sources  We searched PubMed, MEDLINE on OVID, and PsycINFO for studies published between January 1988 and November 2015. Study quality was assessed using an adapted tool for risk of bias; quality of evidence was rated based on GRADE (Grading of Recommendations Assessment, Development and Evaluation) criteria.

Findings  We identified 68 publications meeting inclusion criteria: 59 reporting the prevalence of preoperative mental health conditions (65 363 patients) and 27 reporting associations between preoperative mental health conditions and postoperative outcomes (50 182 patients). Among patients seeking and undergoing bariatric surgery, the most common mental health conditions, based on random-effects estimates of prevalence, were depression (19% [95% CI, 14%-25%]) and binge eating disorder (17% [95% CI, 13%-21%]). There was conflicting evidence regarding the association between preoperative mental health conditions and postoperative weight loss. Neither depression nor binge eating disorder was consistently associated with differences in weight outcomes. Bariatric surgery was, however, consistently associated with postoperative decreases in the prevalence of depression (7 studies; 8%-74% decrease) and the severity of depressive symptoms (6 studies; 40%-70% decrease).

Conclusions and Relevance  Mental health conditions are common among bariatric surgery patients—in particular, depression and binge eating disorder. There is inconsistent evidence regarding the association between preoperative mental health conditions and postoperative weight loss. Moderate-quality evidence supports an association between bariatric surgery and lower rates of depression postoperatively.

Bariatric surgery is an accepted method of promoting weight loss and improving obesity-related physical comorbidities in severely obese individuals.^1-4 Mental health conditions have received less attention. Approximately 19% of US residents have a mental health condition,⁵ including 8% with depression⁶ and 1% to 5% with binge eating disorder.^7,8 These conditions may be more common among patients seeking bariatric surgery than in the general population.^9-12 However, published prevalence estimates vary^13,14 and no systematic reviews describe the prevalence of conditions such as depression, substance abuse, and eating disorders.

We conducted a systematic review with 3 aims: (1) to estimate the prevalence of mental health conditions in patients seeking and undergoing bariatric surgery; (2) to evaluate the association between preoperative mental health conditions and weight loss after surgery; and (3) to evaluate the association between surgery and the clinical course of mental health conditions.

We convened an expert panel to define the scope of our review and develop the protocol.¹⁵ Our PROSPERO registration number is CRD42014008675. Based on input from our panel, we defined mental health conditions as depression, anxiety, posttraumatic stress disorder (PTSD), personality disorders, substance abuse disorders, suicidality or suicidal ideation, and eating disorders—primarily binge eating disorder. Eating behaviors not classified as disorders (eg, grazing and cognitive restraint) and cigarette smoking were outside our scope.

We used weight loss as our postoperative bariatric outcome. Mental health outcomes included suicide incidence and the prevalence, symptoms, and treatment of mental health conditions. We defined persons seeking or being evaluated for bariatric surgery as candidates to distinguish them from populations in which all patients received surgery (recipients).

We searched PubMed, MEDLINE on OVID, and PsycINFO for English-language studies published before November 23, 2015. Search terms included bariatric, obesity, the names of surgical procedures, psychiatric disorders, and eating disorders (eAppendix in the Supplement). We supplemented our search with: (1) citations from an earlier systematic review by Livhits and colleagues¹¹ (search dates 1988-2010); (2) studies recommended by our technical expert panel; and (3) reference mining of included articles.

For inclusion, each study had to address at least 1 specific aim (ie, prevalence, associations, or both), report findings from primary research, and report data for adults (≥18 years) with a body mass index (BMI [calculated as weight in kilograms divided by height in meters squared]) of at least 35. We required all mental health diagnoses to be made preoperatively and excluded studies that asked patients to recall their preoperative health status.

We required studies that reported prevalence data to measure mental health conditions using a formal method (such as a validated instrument or the Diagnostic and Statistical Manual of Mental Health Disorders-5-SCID (Fifth Edition, Structured Clinical Interview). Studies using a diagnostic scale to make diagnoses had to report a threshold value and the corresponding prevalence estimate. For example, we included studies stating, “15% of candidates had major depressive disorder based on a Beck Depression Inventory (BDI) threshold of 20,” but not studies reporting, “the mean BDI score was 25.”

Association studies were also required to make the diagnosis of mental health conditions preoperatively, but these studies were allowed to report either prevalence or nonprevalence measurements of postoperative health status as long as they were formally assessed. For example, we included association studies reporting a change in mean BDI after surgery even if a threshold for the diagnosis of depression was not stated.

Two independent reviewers assessed each study for inclusion and extracted data including country, patient composition (candidates vs recipients), method of recruitment, sample demographics, details of the operation, mental health conditions assessed, methods of assessment, and whether the surgical team was blinded to the assessment. This final item was added based on evidence that prevalence estimates may be higher if patients are told their response will not be shared with the surgical team.¹⁶ Discrepancies were reconciled through discussion.

We estimated the prevalence of each mental health condition using a random-effects meta-analysis for pooled binomial data.¹⁷ We performed formal tests of heterogeneity using the I² statistic.¹⁸ We also explored for potential heterogeneity by stratifying our analyses on variables potentially related to either true differences in prevalence or the risk of bias, including study population (candidates vs recipients), study design (randomized vs observational, consecutive vs nonconsecutive sampling), and, for depression, whether the diagnosis was made using a scale or clinical interview. We compared estimates between subgroups using meta-regression¹⁹; finding no statistically significant differences, we present only unstratified data. All analyses were conducted using Stata 12.1.

If studies reported both lifetime and current prevalence, we used current values; if the timeline was not specified, we assumed data pertained to current prevalence. If studies used multiple methods to diagnose a mental health condition, our study psychiatrist selected the most relevant measure. We compared summary estimates for each condition with published estimates for the US population and considered a relative increase of 25% or greater to be clinically relevant.

For association studies, we classified articles as those measuring bariatric outcomes vs mental health outcomes. Results were further grouped by the outcome being measured. Given the variability in study design and outcome measurement, association data were reported narratively.

A preliminary search returned a large number of small, single-institution observational studies. To assess their value, we first restricted eligibility to studies with more than 500 participants, calculated weighted median estimates, and then compared our results with estimates calculated under a relaxed criterion (≥ 200 participants). Because no differences were found, we concluded that further relaxing the threshold to include studies with sample sizes of less than 200 would be very unlikely to influence our results. We, therefore, restricted our final analyses to: (1) randomized controlled trials (RCTs); (2) multi-institutional observational studies; (3) single-institution studies with random or consecutive sampling and a sample size of at least 200; and (4) single-institution studies with nonconsecutive sampling and a sample size of at least 500. Even with these restrictions, our sample included 70% of published patient data.

We assessed the quality of included studies by adapting existing quality assessment tools.^20,21 We selected 6 items related to risk of bias (study design, sampling strategy, follow-up rate at primary end point, criterion standard for diagnosis, whether assessments were kept confidential, and controlling for confounders) and 1 item each on generalizability (multisite vs single site) and statistical power (sample size).

We rated the quality of evidence for each finding based on criteria established by the GRADE (Grading of Recommendations Assessment, Development and Evaluation) Working Group.²² Quality was classified as high, moderate, low, or very low based on study design, magnitude of the association, and consistency of the results across studies. All classifications were determined by consensus among the research team.

Our literature search yielded 2191 titles. To this, we added 79 citations from Livhits and colleagues¹¹ and 15 references identified by our expert panel. Of these, 567 articles addressed our specific aims (Figure). After screening, 68 articles met inclusion criteria (59 regarding prevalence of mental health conditions; 27 regarding associations between preoperative mental health conditions and postoperative outcomes). Eighteen of these articles contributed to both analyses.

We identified 59 publications reporting prevalence data from 52 studies (eTable and eReferences in the Supplement). The most commonly studied condition was depression (34 studies) followed by binge eating disorder (25), and anxiety (22). In general, 70% to 80% of patients were women, the mean age was 40 to 50 years, and the mean BMI was 45 to 50.

Random-effects pooled estimates for the prevalence of each mental health condition are presented in Table 1. Conditions ranged in prevalence from any mood disorder (23% [95% CI, 15%-31%]) to psychosis (1% [95% CI, 0%-1%]). The 3 most common individual diagnoses were depression (19% [95% CI, 14%-25%]), binge eating disorder (17% [95% CI, 13%-21%]), and anxiety (12% [95% CI, 6%-20%]). Prevalence estimates were 9% (95% CI, 5%-13%) for previous suicidal ideation, 3% (95% CI, 1%-4%) for substance abuse disorders, and 1% (95% CI, 1%-2%) for PTSD. I² statistics were high for all estimates, ranging from 57% to 99%.

The quality of evidence was moderate regarding an increased prevalence of mental health conditions among bariatric patients based on the number of studies and the moderate-to-large magnitude of the prevalence difference. At 23%, the prevalence of any mood disorder among bariatric patients was greater than the US population estimate of 10%.²³ The prevalence of depression (19%) and binge eating disorder (17%) was also greater among bariatric patients than in the general US population (8% for depression; 1%-5% for binge eating disorder).^6-8

We identified 27 publications from 26 studies reporting data on associations: 6 reporting on bariatric outcomes^{24,26,31,35,45-47}; 17 reporting on mental health outcomes^{3,14,25,27-30,32-34,36,37,39,40,42-44}; and 3 studies reporting on both.^38,41,48 Patient demographics were similar to those reported in prevalence studies. Table 2 summarizes each study; quality assessments are listed in Table 3. Of the 26 studies, 18 were prospective (including 3 RCTs) and 16 involved consecutive sampling. Follow-up rates were high (>80% at the primary end point) in 9 of the 18 prospective studies, moderate (60%-80%) in 7, and low (<60%) in 2. Fourteen of the 26 studies controlled for confounders^{3,14,24,26,29-35,38,40,41,47} using statistical methods or design; 3 studies used both matching and multivariable regression.^3,32,34

There was inconsistent evidence regarding the association between preoperative mental health conditions and postoperative weight loss. Five studies reported on depression. The first found no difference in weight loss at 1 year but less weight loss at 4 years for patients with preoperative depression or anxiety (7.9- vs 12.5-unit decrease in BMI) (P = .047).²⁴ Four other studies found no difference in postoperative weight loss between patients with and without preoperative depression. The first found no association between preoperative depression and weight loss at 6 months (mean difference in excess BMI loss, −0.6% [95% CI, −3.8% to 2.6%]) or 1 year after surgery (−1.9% [95% CI, −4.0% to 7.8%]).⁴⁵ The second found no association between BDI scores and weight loss but reported a reduced risk of significant weight regain (≥15% of total weight loss) in patients with higher preoperative BDI scores (odds ratio, 0.94 per unit increase [95% CI, 0.91-0.98]).³⁵ Although retrospective, this study controlled for 14 preoperative and postoperative confounders. The third reported no difference in percent BMI loss between patients with and without depression (difference at 6 months, 27.8% vs 27.8% [P = .95]; at 12 months, 35.8% vs 36.5% [P = .45]; and at 24 months, 38.3% vs 37.4% [P = .52]).⁴⁸ This study had a high follow-up rate at 6 months (85%) and at 12 months (80%) but low follow-up at 24 months (47%). The fourth study reported no difference in excess weight loss at 1 year between patients with depression (43.4%) and those without depression (43.9%; P = nonsignificant [exact value was not provided]) but had a low follow-up rate (53%) and did not control for cofounders.⁴⁶ One study found no association between a current mood disorder and weight loss 2 or 3 years after surgery, but this study did not specifically report on depression.⁴¹

Five studies (2 consecutive, 3 nonconsecutive) provided mixed results regarding the association between preoperative binge eating disorder and postoperative weight loss. One consecutive-sample study found that patients with preoperative binge eating disorder lost more weight after surgery (0.19 percentage point–greater decrease in BMI; P = .014), while the other consecutive-sample study found no difference (Cohen d = 0.24; P = .29).^26,31 The first monitored patients with a current diagnosis of binge eating disorder for 4 years; the second included patients with a lifetime diagnosis of binge eating disorder and followed up patients up to 3 years after surgery. All 3 nonconsecutive sample studies, including an RCT comparing 2 bariatric surgery procedures, found no difference in weight loss between patients with and without preoperative binge eating disorder.^38,41,47 One study⁴⁷ involved patients with a lifetime diagnosis and followed-up patients for 1 year; the other 2 studies involved patients with a current diagnosis of binge eating disorder and analyzed patients twice: at 1 year and then either 3 or 5 years after surgery.

The quality of evidence was very low for all associations between preoperative mental health conditions and postoperative weight loss, with the exception of depression, for which the quality of evidence was upgraded to low because all 5 studies reported no difference in postoperative weight loss at 1 year.

Bariatric surgery was associated with lower rates and fewer symptoms of multiple mental health conditions, particularly depression. Depression improved following bariatric surgery in 11 of the 12 studies^{14,25,27,28,33,36,39,42-44,48} including 2 RCTs that evaluated behavioral health interventions after surgery. This included a reduction in both depression prevalence^{14,25,27,33,41,43,48} and the frequency and severity of depressive symptoms.^{14,28,36,39,42,44} Ten studies were prospective (5 with high follow-up, 5 with moderate) and 3 controlled for confounding using multivariable regression.

Two RCTs evaluating behavioral health interventions after bariatric surgery found improvements in depression for both the intervention group and the usual-care group. One study found lower rates of depression at 6 months and 1 year after surgery for participants in a comprehensive behavioral-motivational program and controls, with slightly larger improvements in the intervention group.²⁷ The second found lower Hospital Anxiety and Depression Scale scores at 1 year for both groups of an intervention assessing preoperative cognitive behavioral therapy.⁴²

Eight prospective^{14,33,36,39,41,43,44,48} and 2 retrospective cohort studies^25,28 found improvements in depression: 5 reporting on prevalence^{25,33,41,43,48}; 4 on symptoms^28,36,39,44; and 1 on both.¹⁴ One large prospective study tracking 3045 bariatric surgery patients in the United Kingdom reported a reduction in depression prevalence during the first 3 years after surgery (odds ratio, 0.82 for year 1 [95% CI, 0.78-0.87]; odds ratio, 0.83 for year 2 [95% CI, 0.76-0.90]; and odds ratio, 0.87 for year 3 [95% CI, 0.78-0.97]).⁴³ However, this reduction in prevalence did not persist after 3 years. A second prospective study found a decrease in depression prevalence from 18% preoperatively to 6% at 2 years.³³ A third reported a decrease in depression prevalence from baseline (45%) to 6 months after surgery (12%) and also to 12 months after surgery (13%). A slight increase was observed between 12 and 24 months but prevalences remained below preoperative values.⁴⁸ A fourth study included more than 5500 US bariatric surgery patients and reported a 49% (95% CI, 41%-57%) reduction in depression prevalence within 4 months after surgery; however, this study did not control for confounders.²⁵ One study of 165 patients participating in a Longitudinal Assessment of Bariatric Surgery (LABS) consortium substudy found no change in the prevalence of depression over time (7.3% at baseline, 7.1% at 2 years, and 8.5% at 3 years).⁴¹

Five studies reported on changes in depressive symptoms. The first reported a 63% reduction in self-reported depressive symptoms at 12 months.⁴⁴ The second found a 28% to 54% decrease in BDI subscale scores at 12 months.²⁸ The third reported lower mean BDI scores at 1, 2, 3, and 4 years after LABG.³⁶ The fourth reported lower mean BDI scores at 1, 2, 3, 4, and 5 years after LABG but focused on patients with a BMI of 30 to 40.³⁹ The final study from the LABS consortium, a high-quality study based on its high follow-up rates (83% at 6 months, 79% at 1 year) and adjustment for confounders via multivariable regression, found a reduction in depression incidence and depression symptoms at 3-year follow-up. There was no change in hospitalizations for depression between 0 and 2 years postoperatively; however, hospitalization rates increased during the third year: 0.9% at baseline vs 1.7% in year 3 (P = .03).¹⁴ The clinical significance of this late increase in hospitalization remains unknown.

All 3 studies reporting on changes in the prevalence or severity of binge eating disorder found improvement 2 years after surgery.^38,39,41 However, rates increased slightly when measured at later time points and returned to baseline in 1 of the 2 studies reporting on prevalence.⁴¹

Evidence regarding bariatric surgery and alcohol abuse was mixed. In 1 study, rates of alcohol consumption and abuse increased significantly during the second year after surgery, but only for patients undergoing Roux-en-Y gastric bypass (RYGB) as opposed to laparoscopic adjustable gastric band (LAGB).²⁹ The Swedish Obese Subjects study, a high-quality study based on its high follow-up rate (87% at 2 years) and 2 methods of controlling for confounders, found similar increases in consumption and self-reported alcohol abuse among vertical-banded gastroplasty patients, but did not report longitudinal data to determine year-to-year changes.³⁴ Neither study found a difference in alcohol consumption or abuse among LAGB patients, consistent with a third study focusing only on LAGB patients.³³

Three studies reported on the incidence of suicide after bariatric surgery. The first found higher rates for patients undergoing bariatric surgery in Pennsylvania over a 10-year period (13.7 per 10 000 men and 5.2 per 10 000 women) than among an age- and sex-matched US population (2.4 per 10 000 men and 0.4 per 10 000 women).³⁰ The second reported a higher rate of suicide among patients undergoing RYGB than among matched controls; however, this difference was not statistically significant (2.6 vs 0.9 per 10 000 person-years, P = .22).³ The third compared patients undergoing RYGB to individuals who sought but did not undergo bariatric surgery and found no difference in suicide rates between the groups.³⁷ The 3 studies relied on retrospective data and none accounted for potential differences in the prevalence of mental health conditions.

One additional study used administrative data to compare the incidence of self-harm emergencies before and after bariatric surgery in Ontario, Canada. The authors found an increased rate of emergency department visits for self-harm postoperatively (rate ratio, 1.54 [95% CI, 1.03-2.30]), but none of these visits resulted in death and the authors were unable to distinguish suicide attempts from unintentional self-harm injuries.⁴⁰

There was no clear evidence regarding postoperative changes in either bipolar disorder or PTSD. One study reported no change in the rate of bipolar disorder after surgery,³³ while another found no association between surgery and time to psychiatric hospitalization or the risk-adjusted use of outpatient psychiatric services for bipolar disorder.³² The only study on PTSD did not perform statistical tests.³³

The quality of evidence was moderate that bariatric surgery is associated with lower postoperative rates of depression, fewer symptoms of depression, and decreased usage of antidepressant therapies, at least during the first 3 years after surgery. The quality of evidence was very low for all other associations. Table 4 summarizes the quality of evidence for our findings.

We found moderate-quality evidence that preoperative mental health conditions are common in patients seeking and receiving bariatric surgery. Meta-analysis of published data estimated that 23% of patients undergoing bariatric surgery reported a current mood disorder—most commonly depression—while 17% were diagnosed with an eating disorder. Both estimates are higher than published rates for the general US population, suggesting that special attention should be paid to these conditions among bariatric patients.^7,8,49-51 Other mental health conditions, such as psychosis, PTSD, and personality disorders are less common but may be more prominent in select subgroups such as US veterans.⁵²

We found no clear evidence that preoperative mental health conditions are associated with differential weight loss after surgery. Our results are consistent with a previous systematic review reporting no association between depression, anxiety, or binge eating disorder and postoperative weight loss.¹¹ In contrast to that review, we found no evidence supporting an association between personality disorders and poor weight loss.

We found some evidence to suggest that weight-loss surgery was associated with a reduction in the prevalence, frequency, and severity of depressive symptoms. However, our study cannot establish a causal relationship and several pathways may exist. First, weight loss may improve body image, self-worth, empowerment, and interpersonal relationships. Second, changes in digestion or intestinal absorption after surgery may alter the brain’s biochemical signaling. Third, patients whose depression may improve on its own may be more likely to undergo surgery. Although our results should not be interpreted as indicating that surgery is a treatment for depression, severely obese patients with depression may gain psychological benefits in addition to the physical benefits already associated with surgery.^53-56

The safety of patients with mental health conditions who undergo surgery appears to be complex. We found increased rates of alcohol abuse after surgery—particularly among RYGB patients—compared with similar populations treated nonoperatively. Our results were derived from 3 studies,^29,33,34 2 of which showed increased self-reported alcohol problems after RYGB^29,34; none of the 3 studies found increased rates after LAGB. These findings are consistent with results from another systematic review⁵⁷ and animal studies, which suggest that increased alcohol consumption after RYGB may be due to physiological changes following intestinal bypass that are not present after LAGB (eg, postprandial changes to intestinal hormone secretion).^58,59 One prospective cohort study suggested an increased rate of suicide after surgery³⁷ while another reported increased rates of self-harm emergencies.⁴⁰ Population-based estimates, including a recent meta-analysis by Peterhänsel and colleagues,⁶⁰ also demonstrate above-average rates of suicide among bariatric surgery patients but have been unable to disentangle the operation from comorbid mental health conditions. Further research is needed to assess whether bariatric patients are at higher risk for suicide and alcohol abuse as well as the appropriate mechanisms for postoperative monitoring in patients with a history of depression or substance abuse.

Our study has limitations. First, included studies varied in their use of scales, thresholds, and definitions of outcomes. The diagnosis of binge eating disorder, for example, underwent significant redefinition during our search period and did not exist as a psychiatric disorder until the DSM-5 was released in 2013.⁶¹ We report the scale or instrument used to make a diagnosis; for diagnoses made by interview, however, it was not always clear what criteria were used. Second, our results do not address the severity or chronicity of mental health conditions, which may be independently associated with postoperative bariatric or mental health outcomes. Third, bariatric candidates with severe mental illness are often screened out prior to referral for bariatric surgery and excluded from published studies. Our prevalence estimates, therefore, are likely to be conservative and our data on associations specific to patients who are referred to and considered for surgery. Fourth, our definition for a meaningful difference in the prevalence of mental health conditions was arbitrary, but our finding of a 100% increase in prevalence for several conditions seems to have face validity. Fifth, our sample size cutoffs were chosen to maximize the value of information for resources needed and our analyses suggest that including small, single-institution studies would not affect our results; however, we cannot exclude this possibility. Sixth, our results may not generalize since the majority of studies reported data from a single institution and, as a whole, capture the experiences of fewer than 200 hospitals and outpatient surgery centers performing bariatric surgery around the world. Seventh, we cannot exclude the possibility of publication bias.

Guidelines from the American Society for Metabolic and Bariatric Surgery and the Department of Veterans Affairs/Department of Defense recommend routine preoperative health assessments, including a review of patients’ mental health conditions.^62,63 Other groups advocate for a more comprehensive, preoperative mental health examination in addition to the general evaluation currently performed by medical and surgical teams.^64,65 The results of our study do not defend or rebut such a recommendation.

Much of the difficulty in determining the effectiveness of preoperative mental health screening is due to the limitations of current screening strategies, which use a variety of scales and focus on mental health diagnoses rather than psychosocial factors. Previous reviews have suggested that self-esteem, mental image, cognitive function, temperament, support networks, and socioeconomic stability play major roles in determining outcomes after bariatric surgery.^10,66 Future studies would benefit from including these characteristics as well as having clear eligibility criteria, standardized instruments, regular measurement intervals, and transparency with respect to time-specific follow-up rates. By addressing these methodological issues, future work can help to identify the optimal strategy for evaluating patients’ mental health prior to bariatric surgery.

Mental health conditions are common among patients seeking and undergoing bariatric surgery, particularly depression and binge eating disorder. There is inconsistent evidence regarding the association between preoperative mental health conditions and postoperative weight loss. Moderate-quality evidence supports an association between bariatric surgery and lower rates of depression postoperatively.

Corresponding Author: Aaron J. Dawes, MD, Department of Surgery, David Geffen School of Medicine at UCLA, Ronald Reagan UCLA Medical Center, 757 Westwood Plaza, B711, Los Angeles, CA 90095 (adawes@mednet.ucla.edu).

Author Contributions: Drs Dawes and Shekelle had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Dawes, Maggard-Gibbons, Maher, Shekelle.

Acquisition, analysis, or interpretation of data: Dawes, Maggard-Gibbons, Maher, Booth, Miake-Lye, Beroes, Shekelle.

Drafting of the manuscript: Dawes, Maggard-Gibbons, Booth.

Critical revision of the manuscript for important intellectual content: Dawes, Maher, Booth, Shekelle.

Statistical analysis: Dawes, Booth.

Obtained funding: Shekelle.

Administrative, technical, or material support: Dawes, Miake-Lye, Beroes.

Study supervision: Maggard-Gibbons, Shekelle.

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: The publication is based on a systematic review conducted by the Evidence-based Synthesis program funded by the Department of Veterans Affairs. Dr Dawes was supported by the VA Office of Academic Affiliations through the VA/Robert Wood Johnson Clinical Scholars Program.

Role of the Funder/Sponsors: The Department of Veterans Affairs (VA) 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 findings and conclusions in this article are those of the authors, who are responsible for its contents. The findings and conclusions do not necessarily represent the views of the Department of Veterans Affairs or the US government. Therefore, no statement in this article should be construed as the official position of the Department of Veterans Affairs. Dr Livingston, JAMA Deputy Editor, was not involved in the editorial review of or decision to publish this article.

Additional Contributions: We would like to thank the members of our technical expert panel: David E. Arterburn, MD, MPH (Group Health Research Institute), Sarah P. Catanese, PhD (Northwestern University Feinberg School of Medicine), Edward H. Livingston, MD (JAMA), Matthew L. Maciejewski, PhD (Durham VA Medical Center), James E. Mitchell, MD (University of North Dakota School of Medicine and Health Sciences), John Morton, MD, MPH (Stanford University School of Medicine), Karen Oliver, PhD (Providence VA Medical Center), Thomas Rutledge, PhD (VA San Diego Healthcare System), and Bruce M. Wolfe, MD (Oregon Health and Sciences University). None of these individuals were compensated in association with their contributions to this article.