0
We're unable to sign you in at this time. Please try again in a few minutes.
Retry
We were able to sign you in, but your subscription(s) could not be found. Please try again in a few minutes.
Retry
There may be a problem with your account. Please contact the AMA Service Center to resolve this issue.
Contact the AMA Service Center:
Telephone: 1 (800) 262-2350 or 1 (312) 670-7827  *   Email: subscriptions@jamanetwork.com
Error Message ......
Original Investigation |

A Simple Prediction Rule for All-Cause Mortality in a Cohort Eligible for Bariatric Surgery FREE

Raj S. Padwal, MD, MSc1,2; Scott W. Klarenbach, MD, MSc1; Xiaoming Wang, PhD1; Arya M. Sharma, MD, PhD1,2; Shahzeer Karmali, MD3; Daniel W. Birch, MD, MSc3; Sumit R. Majumdar, MD, MPH1,2
[+] Author Affiliations
1Department of Medicine, University of Alberta, Edmonton, Alberta, Canada
2Alberta Diabetes Institute, Edmonton, Alberta, Canada
3Department of Surgery and Center for the Advancement of Minimally Invasive Surgery, Royal Alexandra Hospital, Edmonton, Alberta, Canada
JAMA Surg. 2013;148(12):1109-1115. doi:10.1001/jamasurg.2013.3953.
Text Size: A A A
Published online

Importance  Current eligibility criteria for bariatric surgery use arbitrarily chosen body mass index (BMI) (calculated as weight in kilograms divided by height in meters squared) thresholds, an approach that has been criticized as arbitrary and lacking evidence.

Objectives  To verify the importance of BMI as a mortality predictor, to identify other important mortality predictors, and to construct a mortality prediction rule in a population eligible for bariatric surgery.

Design  We studied individuals from a population-representative register who met contemporary eligibility criteria for bariatric surgery (BMI, ≥35.0 alone or 30.0-34.9 with an obesity-related comorbidity) from January 1, 1988, through December 31, 1998. We used binary logistic regression to construct a parsimonious model and a clinical prediction rule for 10-year all-cause mortality.

Setting  The United Kingdom General Practice Research Database, a population-representative primary care registry.

Participants  Fifteen thousand three hundred ninety-four patients aged 18 to 65 years.

Exposure  Bariatric surgery.

Main Outcome and Measure  Ten-year all-cause mortality.

Results  Mean (SD) age was 46.9 (11.9) years, BMI was 36.2 (5.5), and 63.2% of the patients were women. All-cause mortality was 2.1%, and mean follow-up duration was 9.9 years. The final model, which included age (odds ratio, 1.09 per year [95% CI, 1.07-1.10]), type 2 diabetes mellitus (2.25 [1.76-2.87]), current smoking (1.62 [1.28-2.06]), and male sex (1.50 [1.20-1.87]), had a C statistic of 0.768. Although BMI significantly predicted mortality (odds ratio, 1.03 per unit [95% CI, 1.01-1.05]), it did not improve model discrimination or calibration. We divided clinical prediction rule scoring into 4 tiers. All-cause mortality was 0.2% in tier 1, 0.9% in tier 2, 2.0% in tier 3, and 5.2% in tier 4.

Conclusions and Relevance  All-cause 10-year mortality in obese individuals eligible for bariatric surgery can be estimated using a simple 4-variable prediction rule based on age, sex, smoking, and diabetes mellitus. Body mass index was not an important mortality predictor. Further work is needed to define low, moderate, and high absolute risk thresholds and to provide external validation.

Figures in this Article

Global uptake of bariatric surgery, considered the most effective treatment currently available for severe or medically complicated obesity, has increased 8-fold during the past decade to 350 000 procedures annually.1 Surgery is associated with 24% (95% CI, 0.01-0.41) reductions in 10-year all-cause mortality, substantial increases in type 2 diabetes mellitus remission rates, and improvements in other obesity-related comorbidities and quality of life.24

Eligibility criteria for bariatric surgery, originally proposed during a 1991 US National Institutes of Health (NIH) consensus conference,5 endorse surgery for patients with treatment-refractory obesity measured by a body mass index (BMI) (calculated as weight in kilograms divided by height in meters squared) of 35.0 to 39.9 with a major obesity-related comorbidity or by a BMI of 40.0 or greater. New evidence supporting the use of surgery in patients with a BMI less than 35.0 has prompted expansion of these criteria.6,7 Recently, the US Food and Drug Administration approved gastric banding (but not other procedure types) for patients with a BMI greater than 30.0 and a major obesity-related comorbidity.8 Similarly, the International Diabetes Federation9 expanded eligibility for all bariatric procedures to patients with type 2 diabetes mellitus and class 1 obesity (BMI, 30.0-34.9).

Therefore, although individuals with medically complicated class 1 obesity constitute the broadest group currently considered eligible for bariatric surgery, potential eligibility (which determines access and reimbursement) varies widely according to BMI, comorbidity presence, and procedure type. This variation means that a prioritization or triage system is implicitly being used to determine who does and does not receive surgery among those individuals generally considered eligible.1012 Current criteria are structured such that a 5-unit increment in BMI (ie, 35.0-40.0) obviates any further requirement for obesity-related comorbidity. In addition, under the original NIH criteria, different obesity-related comorbidities are considered to be of equal importance for eligibility determination. Conversely, recent reductions in the BMI thresholds for type 2 diabetes mellitus imply that this comorbidity should be considered more important than others. Other than data demonstrating that type 2 diabetes mellitus is highly responsive to surgery,13 to our knowledge, no other published data validate these concepts. In fact, the choice of BMI thresholds has been criticized as arbitrary and lacking evidence.11,14

The objective of this study was to identify mortality predictors within a sample of patients eligible for bariatric surgery that was derived from a large population-representative primary care registry. Examining the importance of BMI as a mortality predictor was an explicit aim. Specifically, we sought to construct a parsimonious clinical prediction rule for all-cause mortality composed of commonly available, clinically relevant predictor variables. We believed that such a rule would help to identify and summarize key prognostic indicators and calculate individual mortality risk.

Data Sources

The University of Alberta research ethics board approved the study protocol. Informed consent to conduct this secondary analysis of existing registry data was not required. We analyzed data from the UK General Practice Research Database (GPRD) (recently amalgamated into the Clinical Practice Research Datalink), the world’s largest, most comprehensive computerized database of anonymized, linked longitudinal medical records.15,16 The GPRD contains population-representative data from more than 600 UK primary care practices (6% of the UK population).15,16 The United Kingdom has a comprehensive national health care system, 98% of UK residents are registered in a general practice, and all care is centralized within the general practice record.17,18

The GPRD contains the complete primary care record, including patient demographics (eg, age and sex) and health behaviors (eg, alcohol intake and smoking), physiological and laboratory data (blood pressure, BMI, blood glucose levels, and cholesterol level profile), clinician-assigned diagnoses and disease registries for hypertension and type 2 diabetes mellitus, and outpatient prescription medications. Medical diagnoses are coded using the Oxford Medical Information System and Read codes.15,16,19

Validation of GPRD data is performed on an ongoing basis in accordance with standardized guidelines that certify certain practices as up-to-standard or of research quality. More than 350 validation studies have been performed, and more than 700 peer-reviewed studies have been published.20 A systematic review of more than 200 studies found GPRD medical diagnostic coding to be highly accurate (a median of 89% of medical diagnoses confirmed as accurate).20 Drug exposure data have been similarly validated and found to be of equally high quality.16,21 Active smoking data are 80% accurate compared with population-based survey data.22 Although BMI is measured annually in only one-half to two-thirds of GPRD patients, 99.2% of obese patients have at least 1 BMI measurement recorded during a 10-year span.23 Only 0.6% of BMI values in the GPRD have been found to be inconsistent or implausible.23

Study Population

We studied adults (aged 18-65 years) registered with GPRD up-to-standard practices from January 1, 1988, through December 31, 1998, and considered eligible for bariatric surgery. An age cutoff of 65 years was chosen because very few bariatric procedures are performed in patients older than 65 years.24 For the primary analysis, in accordance with contemporary recommendations, we considered patients with a BMI of 35.0 or greater or a BMI of 30.0 to 34.9 with an obesity-related comorbidity to be eligible for surgery.8,9,13 Hypertension, dyslipidemia, heart failure, type 2 diabetes mellitus, sleep apnea, osteoarthritis, coronary artery disease, and cerebrovascular disease (CVD) were obesity-related comorbidities used to define eligibility.25

The index date, defined as the date of initial cohort entry, corresponded to the first time a surgery-eligible BMI was recorded. We excluded patients with any of the following contraindications to surgery, which were defined using medical (Read) codes and/or by medication use (within the year before the index date): pregnancy, suicidality, psychosis, severe (medication-refractory) depression, bulimia, active cancer, and prior bariatric surgery. We also performed a subgroup analysis in patients meeting the original NIH eligibility criteria for surgery. To evaluate the potential impact of unrecognized serious baseline illness, we also performed a sensitivity analysis after excluding early deaths (mortality within 6 months of the index date).

Primary Outcome

The primary outcome, 10-year all-cause mortality, was ascertained by linking with the mortality records of the UK Office of National Statistics. These mortality records undergo frequent and extensive assessment to verify accuracy and validity; registration of all deaths in the United Kingdom is required by law.26 Each patient was followed up until death or for 10 years. A 10-year time horizon was chosen because it is considered clinically relevant and is widely used for nonbariatric risk prediction.27

Chronic Health Behaviors and Obesity-Related Comorbidities

We focused on clinically important, relatively common chronic comorbidities that had well-documented associations with obesity or all-cause mortality and are readily identified in most clinical settings.28,29 In addition to age, sex, and BMI, we modeled the impact of health behaviors (smoking and alcohol intake) and chronic medical conditions (hypertension, type 2 diabetes mellitus, dyslipidemia, stable depression, sleep apnea, heart failure, coronary artery disease, CVD, and gastroesophageal reflux disease). Comorbidities, smoking, and alcohol use were identified using disease registries or medical (Read) codes and/or by use of medication.15,16,19 Comorbidities were considered to be present at baseline if identified at any point before or within 3 months of the index date.

Statistical Analysis

We first examined descriptive variables and calculated means, medians, and proportions. Mortality was compared across BMI classes within age strata using Fisher exact tests. We originally intended to perform Cox proportional hazards regression to model time to death. Because the proportional hazards assumption was violated, we used binary logistic regression instead to predict 10-year all-cause mortality.

Similar modeling methods were used for the primary analysis (performed first), subgroup analysis, and sensitivity analyses. Age and BMI were modeled as continuous variables, and the remaining predictors were modeled as dichotomous variables. All variables were included in the initial model. Next, nonsignificant (P ≥ .05) and then significant (P < .05) predictors were removed manually in stepwise fashion. We examined the effect of variable exclusion on the Akaike information criterion (to assess relative goodness of fit) and model C statistic (to assess discrimination).30,31 Variables with a minimal effect on both analyses were excluded to create a parsimonious final model.

Final model calibration was assessed using the Hosmer-Lemeshow goodness-of-fit test.30 To further confirm that the exclusion of each predictor had a minimal effect on model predictive ability, we calculated a net reclassification improvement (NRI) index (performed separately for each variable).32 First, 4 mortality strata were chosen, with an upper limit of 5% (<1.75%, 1.75%-3.49%, 3.50%-4.99%, ≥5.00%), which indicates high-risk status according to expert consensus.27 Second, patients were categorized into these strata according to their predicted risk in models constructed with and without the predictor variable (performed separately in patients with and without events). Third, the between-model net change in the proportion of patients appropriately reclassified (by adding the predictor in question) was tabulated.31,32 The significance level for the NRI analysis was set at .1 (with P > .1 indicating that exclusion of the predictor variable from the model was appropriate).

A bootstrap analysis using 1000 replicative cycles with replacement was performed to internally validate our findings and verify the magnitude, 95% confidence intervals, and statistical significance of the final β coefficient estimates.31 Clinical prediction rule scoring weights were created for each variable in the final model by dividing each β coefficient by the smallest model coefficient and rounding up to the nearest integer. We used commercially available statistical software (SAS, version 9.3 [SAS Institute Inc] and R, version 2.15.0 [http://cran.r-project.org/bin/windows/base/old/2.15.0/]) for the analyses.

Further sensitivity analyses were added after peer review. First, we examined the effect of adding quadratic terms for age and BMI (to assess for a nonlinear relationship with mortality) on the C statistic. Second, we constructed an accelerated failure time model for survival analysis, which does not assume proportional hazards, to assess whether the results were consistent with logistic regression.33

Baseline Characteristics

The initial sample size was 17 340 patients; of these, 1946 (11.2%) were excluded (Supplement [eFigure]). Included and excluded patients had similar mean (SD) age (46.9 [11.9] vs 47.3 [12.0] years), BMI (36.2 [5.5] vs 35.7 [5.7]), and sex distribution (63.2% vs 64.1% female). Baseline characteristics of the final cohort (n = 15 394) are summarized in Table 1. Mean (SD) duration of follow-up was 9.9 (0.6) years.

Rates and Correlates of Mortality

All-cause 10-year mortality was 2.1% (n = 330). Mortality within age strata did not differ significantly between BMI classes (P > .05) (Figure). Age, type 2 diabetes mellitus, male sex, smoking, BMI, coronary artery disease, and CVD were significant predictors of all-cause mortality, with age and diabetes having the largest impact (Table 2). Exclusion of other variables not significantly associated with all-cause mortality had minimal effect on the Akaike information criterion statistics (2908.6 before vs 2901.5 after exclusion) and C statistic (0.771 vs 0.768). Body mass index and CVD contributed minimally to the Akaike information criterion (2901.5 before vs 2910.2 after exclusion) and C statistic (0.768 vs 0.762) and were then excluded for the sake of parsimony. The NRI indexes confirmed that their exclusion was appropriate (NRI index for BMI, 2.0% [P = .15]; for coronary artery disease, 0.6% [P = .35]; and for CVD, 1.8% [P = .13]).

Place holder to copy figure label and caption
Figure.
Age-Stratified Mortality by Body Mass Index (BMI) Class

Class 1 indicates BMI (calculated as weight in kilograms divided by height in meters squared) of 30.0 to 34.9; class 2, BMI of 35.0 to 39.9; and class 3, BMI of at least 40.0.aAll patients had at least 1 obesity-related comorbidity.

Graphic Jump Location
Table Graphic Jump LocationTable 2.  Binary Logistic Regression Predicting All-Cause Mortality

Age (odds ratio, 1.09 per year [95% CI, 1.07-1.10]), type 2 diabetes mellitus (2.25 [1.76-2.87]), current smoking (1.62 [1.28-2.06]), and male sex (1.50 [1.20-1.87]) constituted the final model (Table 2), with P = .99 for overall goodness of fit. The bootstrap method analysis produced variance estimates and P values of identical magnitude to those found in the primary analysis.

Clinical Prediction Rule

The 4-variable clinical prediction rule scores ranged from 0 to 107. To simplify use of the prediction rule, scores were categorized into 4 tiers (Table 3). Mortality ranged from 0.2% for tier 1 to 5.2% for tier 4 (Table 3).

Sensitivity and Subgroup Analyses

First, results were unchanged after excluding early deaths (data not shown). Second, patients meeting NIH eligibility criteria (n = 5071) included 118 deaths (2.3%). Age (odds ratio, 1.08 [95% CI, 1.05-1.10]), type 2 diabetes mellitus (2.98 [2.02-4.34]), and CVD (2.82 [1.36-5.88]) were significant predictors of mortality (C statistic, 0.763). Body mass index was not a significant predictor of mortality, and exclusion of BMI again had no effect on model variables. The C statistic changed little after excluding CVD (0.762); however, the NRI index (7.3% [P = .03]) indicated that this variable should be retained in the model. Third, adding quadratic terms for BMI and age did not improve model C statistics (data not shown). Fourth, the accelerated failure time model results were identical to those of the primary analysis (Supplement [eTable]).

In a study of 15 394 patients who were eligible for bariatric surgery, all-cause 10-year mortality was 2.1%. The main drivers of mortality were age and type 2 diabetes mellitus. Although their association with the outcome was statistically significant, established CVD and BMI did not contribute materially to the mortality prediction in the primary analysis. Our results challenge conventional wisdom emphasizing the importance of BMI as a prognostic indicator for mortality in individuals eligible for bariatric surgery and, of equal import, the practice of using BMI as a surrogate for long-term prognosis.

A previous study34 also found that BMI class was a poor mortality discriminator in a nationally representative US sample of obese patients (BMI, ≥30.0). These findings contrast with a wealth of prospective general-population cohort data demonstrating that a BMI greater than 30.0 increases mortality risk compared with a lower BMI.29,35 How should we reconcile these disparate findings? We believe that, in the previous studies, the differing baseline BMI levels in the comparison groups constitute a major contributor and, in aggregate, these findings suggest that once a certain degree of obesity is established, BMI ceases to be an important mortality predictor. The findings of a recent study of more than 660 000 individuals with a BMI greater than 18.5 support this view. In this analysis, BMI class, diabetes mellitus, and hypertension were significant mortality predictors.36

To our knowledge, only 1 other study has examined the utility of an alternative staging system to BMI for mortality prediction in individuals eligible for bariatric surgery.34 In that population-representative US sample (n = 546), the Edmonton Obesity Staging System, a 5-point ordinal score, predicted mortality independently of BMI.34 However, a clinical prediction rule was not developed. Furthermore, the Edmonton Obesity Staging System is of limited practical utility because it is complex, depends on data elements that are not measured routinely (eg, functional status), and arbitrarily considers different obesity-related comorbidities to be of equivalent prognostic value.

The prediction rule constructed in the present study enables individual risk calculation and facilitates prognostic comparisons between different individuals. Unlike other commonly used contemporary risk engines,27,37 this prediction rule is specific to a population eligible for bariatric surgery and was constructed from a cohort containing a substantial proportion (17.3%) of severely obese individuals (BMI, ≥40.0). The rule also differs from other risk engines in that it predicts all-cause mortality rather than cardiovascular mortality or cardiovascular events. As such, further work is needed to define low, moderate, and high absolute risk thresholds. Other researchers27 have defined a 10-year cardiovascular mortality risk of 5% as high and indicative of the need for treatment of cardiovascular risk factors. Although a similar threshold could be considered in individuals eligible for bariatric surgery, a 10-year morality rate of 5% is extremely high in relative terms considering the young age distribution of our cohort. Indeed, given a 10-year 1998 US general-population all-cause mortality base rate of 0.4% (among 20-year-old nonsmoking women),38 mortality increased 5-fold in tier 3 patients and more than 12-fold in tier 4 patients.

Individual mortality risk quantification might also be useful on a system management level to help inform decisions about allocating surgery where large demand-supply gaps exist. Worldwide, access to publicly funded bariatric surgery is limited and patients face protracted, multiyear wait times.10,12 No evidence-based allocation system currently exists, and a risk-treatment paradox has emerged (ie, compared with individuals eligible for surgery, surgical recipients are more likely to be at lower risk because they have a greater likelihood of being young, female, and free of comorbidity).10,12 Whether to incorporate individual risk quantification into allocation decisions and how best to do this and account for other important factors (eg, operative risk, expected surgery-related benefit, and nonmortality end points) remain unresolved and undefined.

Despite its novelty and strengths, this study has several limitations. First, we focused on all-cause mortality and did not examine other clinically important end points. Second, comorbidity prevalence was lower than expected prevalence estimates based on contemporaneous data.39 This unexpected finding could be related to systematic comorbidity underdiagnosis in this obese population and may have potentially biased odds ratio estimates toward the null. Third, we studied a sample from the United Kingdom eligible for bariatric surgery, which should be borne in mind when generalizing the results beyond the United Kingdom. Fourth, we used BMI levels to define cohort entry. Although more than 99% of obese GPRD patients have a BMI recorded at least once in 10 years, only one-half to two-thirds have a BMI recorded annually.23 Thus, cohort entry may have been discordant with obesity onset, which could reduce baseline BMI measurement accuracy and also possibly reduce follow-up duration. Notably, GPRD patients experience slow weight gain over time (mean BMI increase of 0.09 annually).23 Thus, even in the case of a multiyear temporal disconnect between obesity onset and BMI measurement, only a small mean error in baseline BMI would be expected.

In conclusion, our findings demonstrate that factors other than BMI are important in predicting the risk of death in patients eligible for bariatric surgery and that, of the obesity-related comorbidities, type 2 diabetes mellitus is the most important mortality predictor. Given that diabetes mellitus is highly amenable to surgical treatment,3 a strong case could be made for prioritizing it over BMI or other comorbidities. Regardless, our findings call into question current practices, such as the inordinate emphasis placed on BMI increments of 5.0 embedded in contemporary eligibility criteria. Once our findings are externally validated, we believe that a clinical prediction rule similar to ours should be adopted to optimize delivery of bariatric care.

Accepted for Publication: April 15, 2013.

Corresponding Author: Raj S. Padwal, MD, MSc, Department of Medicine, University of Alberta, 2F1.26 Walter C. Mackenzie Health Sciences Centre, 8440-112th St, Edmonton, AB T6G 2B7, Canada (rpadwal@ualberta.ca).

Published Online: October 16, 2013. doi:10.1001/jamasurg.2013.3953.

Author Contributions: Dr Padwal 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: Padwal, Klarenbach, Sharma, Majumdar.

Acquisition of data: Padwal.

Analysis and interpretation of data: Padwal, Wang, Sharma, Karmali, Birch, Majumdar.

Drafting of the manuscript: Padwal, Wang.

Critical revision of the manuscript for important intellectual content: Padwal, Klarenbach, Sharma, Karmali, Birch, Majumdar.

Statistical analysis: Padwal, Wang, Majumdar.

Obtained funding: Padwal, Majumdar.

Administrative, technical, and material support: Padwal, Sharma, Karmali, Birch.

Study supervision: Padwal.

Conflict of Interest Disclosures: Drs Padwal, Klarenbach, Sharma, and Majumdar are supported by an alternative funding plan from the Government of Alberta and the University of Alberta. Dr Sharma is supported by an Alberta Health Services Chair in Obesity Research and Management. Drs Klarenbach and Majumdar hold salary support awards from Alberta Heritage Foundation for Medical Research and Alberta Innovates-Health Solutions. Dr Majumdar holds the Endowed Chair in Patient Health Management of the Faculties of Medicine and Dentistry and Pharmacy and Pharmaceutical Sciences, University of Alberta. Dr Karmali has received honoraria from and has consulted for Ethicon Endo-Surgery, Covidien, and Gore (bariatric surgery equipment manufacturers). Dr Birch has received research funding from Johnson and Johnson and Ethicon Endo-Surgery and has consulted for Johnson and Johnson, Ethicon Endo-Surgery, Covidien, Bard, Baxter, and Olympus. No other disclosures were reported.

Funding/Support: This study was supported by research grants 106572 and 86642 from the Canadian Institutes of Health Research.

Role of the Sponsor: The Canadian Institutes of Health Research 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: This study is based in part on data from the GPRD obtained under license from the UK Medicines and Healthcare Products Regulatory Agency. However, the interpretation and conclusions contained in this study are those of the authors alone. 

Previous Presentation: This study was presented at the Third Annual Canadian Obesity Summit; May 3, 2013; Vancouver, British Columbia, Canada.

Buchwald  H, Oien  DM.  Metabolic/bariatric surgery worldwide 2008. Obes Surg. 2009;19(12):1605-1611.
PubMed   |  Link to Article
Sjöström  L, Narbro  K, Sjöström  CD,  et al; Swedish Obese Subjects Study.  Effects of bariatric surgery on mortality in Swedish obese subjects. N Engl J Med. 2007;357(8):741-752.
PubMed   |  Link to Article
Carlsson  LMS, Peltonen  M, Ahlin  S,  et al.  Bariatric surgery and prevention of type 2 diabetes in Swedish obese subjects. N Engl J Med. 2012;367(8):695-704.
PubMed   |  Link to Article
Padwal  R, Klarenbach  S, Wiebe  N,  et al.  Bariatric surgery: a systematic review of the clinical and economic evidence. J Gen Intern Med. 2011;26(10):1183-1194.
PubMed   |  Link to Article
US Department of Health and Human Services, National Institutes of Health. Gastrointestinal surgery for severe obesity. NIH Consensus Statement. http://www.consensus.nih.gov/1991/1991gisurgeryobesity084html.htm. Accessed September 18, 2012.
Dixon  JB, O’Brien  PE, Playfair  J,  et al.  Adjustable gastric banding and conventional therapy for type 2 diabetes: a randomized controlled trial. JAMA. 2008;299(3):316-323.
PubMed   |  Link to Article
O’Brien  PE, Dixon  JB, Laurie  C,  et al.  Treatment of mild to moderate obesity with laparoscopic adjustable gastric banding or an intensive medical program: a randomized trial. Ann Intern Med. 2006;144(9):625-633.
PubMed   |  Link to Article
US Food and Drug Administration. Medical devices: gastric banding. http://www.fda.gov/MedicalDevices/ProductsandMedicalProcedures/ObesityDevices/ucm350132.htm. Accessed August 1, 2012.
Dixon  JB, Zimmet  P, Alberti  KG, Rubino  F; International Diabetes Federation Taskforce on Epidemiology and Prevention.  Bariatric surgery: an IDF statement for obese type 2 diabetes. Surg Obes Relat Dis. 2011;7(4):433-447.
PubMed   |  Link to Article
Padwal  RS, Sharma  AM.  Treating severe obesity: morbid weights and morbid waits. CMAJ. 2009;181(11):777-778.
PubMed   |  Link to Article
Pories  WJ, Dohm  LG, Mansfield  CJ.  Beyond the BMI: the search for better guidelines for bariatric surgery. Obesity (Silver Spring). 2010;18(5):865-871.
PubMed   |  Link to Article
Flum  DR, Khan  TV, Dellinger  EP.  Toward the rational and equitable use of bariatric surgery. JAMA. 2007;298(12):1442-1444.
PubMed   |  Link to Article
Dixon  JB, le Roux  CW, Rubino  F, Zimmet  P.  Bariatric surgery for type 2 diabetes. Lancet. 2012;379(9833):2300-2311.
PubMed   |  Link to Article
Livingston  EH.  Inadequacy of BMI as an indicator for bariatric surgery. JAMA. 2012;307(1):88-89.
PubMed   |  Link to Article
Welcome to the Clinical Practice Research Datalink. www.cprd.com/intro.asp. Accessed October 2, 2012.
Jick  SS, Kaye  JA, Vasilakis-Scaramozza  C,  et al.  Validity of the General Practice Research Database. Pharmacotherapy. 2003;23(5):686-689.
PubMed   |  Link to Article
Mulnier  HE, Seaman  HE, Raleigh  VS, Soedamah-Muthu  SS, Colhoun  HM, Lawrenson  RA.  Mortality in people with type 2 diabetes in the UK. Diabet Med. 2006;23(5):516-521.
PubMed   |  Link to Article
Tannen  RL, Weiner  MG, Xie  D.  Use of primary care electronic medical record database in drug efficacy research on cardiovascular outcomes: comparison of database and randomised controlled trial findings. BMJ. 2009;338:b81. doi:10.1136/bmj.b81.
PubMed   |  Link to Article
Lawrenson  R, Williams  T, Farmer  R.  Clinical information for research: the use of general practice databases. J Public Health Med. 1999;21(3):299-304.
PubMed   |  Link to Article
Herrett  E, Thomas  SL, Schoonen  WM, Smeeth  L, Hall  AJ.  Validation and validity of diagnoses in the General Practice Research Database: a systematic review. Br J Clin Pharmacol. 2010;69(1):4-14.
PubMed   |  Link to Article
Azoulay  L, Schneider-Lindner  V, Dell’aniello  S, Filion  KB, Suissa  S.  Thiazolidinediones and the risk of incident strokes in patients with type 2 diabetes: a nested case-control study. Pharmacoepidemiol Drug Saf. 2010;19(4):343-350.
PubMed
Lewis  JD, Brensinger  C.  Agreement between GPRD smoking data: a survey of general practitioners and a population-based survey. Pharmacoepidemiol Drug Saf. 2004;13(7):437-441.
PubMed   |  Link to Article
Booth  HP, Prevost  AT, Gulliford  MC.  Epidemiology of clinical body mass index recording in an obese population in primary care: a cohort study. J Public Health (Oxf). 2013;35(1):67-74.
PubMed   |  Link to Article
Livingston  EH.  Procedure incidence and in-hospital complication rates of bariatric surgery in the United States. Am J Surg. 2004;188(2):105-110.
PubMed   |  Link to Article
National Heart, Lung, and Blood Institute, National Institutes of Health. Clinical Guidelines on the Identification, Evaluation, and Treatment of Overweight and Obesity in Adults: the Evidence Report. Bethesda, MD: National Institutes of Health; 1998. NIH publication 98-4083.
Office for National Statistics. Mortality statistics: quality and methodology information. 2012. http://ons.gov.uk/ons/index.html. Accessed February 26, 2013.
Perk  J, De Backer  G, Gohlke  H,  et al; European Association for Cardiovascular Prevention & Rehabilitation (EACPR); ESC Committee for Practice Guidelines (CPG).  European guidelines on cardiovascular disease prevention in clinical practice (version 2012): the Fifth Joint Task Force of the European Society of Cardiology and Other Societies on Cardiovascular Disease Prevention in Clinical Practice (constituted by representatives of nine societies and by invited experts). Eur Heart J. 2012;33(13):1635-1701.
PubMed   |  Link to Article
Hensrud  DD, Klein  S.  Extreme obesity: a new medical crisis in the United States. Mayo Clin Proc. 2006;81(10)(suppl):S5-S10.
PubMed   |  Link to Article
Bray GA, Pi-Sunyer X, Mulder JE. Health hazards associated with obesity. http://uptodate.com/contents/health-hazards-associated-with-obesity-in-adults. Accessed September 29, 2012.
Steyerberg  EW, Vickers  AJ, Cook  NR,  et al.  Assessing the performance of prediction models: a framework for traditional and novel measures. Epidemiology. 2010;21(1):128-138.
PubMed   |  Link to Article
Cook  NR.  Statistical evaluation of prognostic versus diagnostic models: beyond the ROC curve. Clin Chem. 2008;54(1):17-23.
PubMed   |  Link to Article
Pencina  MJ, D’Agostino  RBS  Sr, D’Agostino  RB  Jr, Vasan  RS.  Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Stat Med. 2008;27(2):157-172, 207-212.
PubMed   |  Link to Article
Robertson  HT, de los Campos  G, Allison  DB.  Turning the analysis of obesity-mortality associations upside down: modeling years of life lost through conditional distributions. Obesity (Silver Spring). 2013;21(2):398-404.
PubMed   |  Link to Article
Padwal  RS, Pajewski  NM, Allison  DB, Sharma  AM.  Using the Edmonton obesity staging system to predict mortality in a population-representative cohort of people with overweight and obesity. CMAJ. 2011;183(14):E1059-E1066. doi:10.1503/cmaj.110387.
PubMed   |  Link to Article
Berrington de Gonzalez  A, Hartge  P, Cerhan  JR,  et al.  Body-mass index and mortality among 1.46 million white adults. N Engl J Med. 2010;363(23):2211-2219.
PubMed   |  Link to Article
Livingston  EH, Ko  CY.  Effect of diabetes and hypertension on obesity-related mortality. Surgery. 2005;137(1):16-25.
PubMed   |  Link to Article
Wilson  PW, D’Agostino  RB, Levy  D, Belanger  AM, Silbershatz  H, Kannel  WB.  Prediction of coronary heart disease using risk factor categories. Circulation. 1998;97(18):1837-1847.
PubMed   |  Link to Article
Woloshin  S, Schwartz  LM, Welch  HG.  Risk charts: putting cancer in context. J Natl Cancer Inst. 2002;94(11):799-804.
PubMed   |  Link to Article
Must  A, Spadano  J, Coakley  EH, Field  AE, Colditz  G, Dietz  WH.  The disease burden associated with overweight and obesity. JAMA. 1999;282(16):1523-1529.
PubMed   |  Link to Article

Figures

Place holder to copy figure label and caption
Figure.
Age-Stratified Mortality by Body Mass Index (BMI) Class

Class 1 indicates BMI (calculated as weight in kilograms divided by height in meters squared) of 30.0 to 34.9; class 2, BMI of 35.0 to 39.9; and class 3, BMI of at least 40.0.aAll patients had at least 1 obesity-related comorbidity.

Graphic Jump Location

Tables

Table Graphic Jump LocationTable 2.  Binary Logistic Regression Predicting All-Cause Mortality

References

Buchwald  H, Oien  DM.  Metabolic/bariatric surgery worldwide 2008. Obes Surg. 2009;19(12):1605-1611.
PubMed   |  Link to Article
Sjöström  L, Narbro  K, Sjöström  CD,  et al; Swedish Obese Subjects Study.  Effects of bariatric surgery on mortality in Swedish obese subjects. N Engl J Med. 2007;357(8):741-752.
PubMed   |  Link to Article
Carlsson  LMS, Peltonen  M, Ahlin  S,  et al.  Bariatric surgery and prevention of type 2 diabetes in Swedish obese subjects. N Engl J Med. 2012;367(8):695-704.
PubMed   |  Link to Article
Padwal  R, Klarenbach  S, Wiebe  N,  et al.  Bariatric surgery: a systematic review of the clinical and economic evidence. J Gen Intern Med. 2011;26(10):1183-1194.
PubMed   |  Link to Article
US Department of Health and Human Services, National Institutes of Health. Gastrointestinal surgery for severe obesity. NIH Consensus Statement. http://www.consensus.nih.gov/1991/1991gisurgeryobesity084html.htm. Accessed September 18, 2012.
Dixon  JB, O’Brien  PE, Playfair  J,  et al.  Adjustable gastric banding and conventional therapy for type 2 diabetes: a randomized controlled trial. JAMA. 2008;299(3):316-323.
PubMed   |  Link to Article
O’Brien  PE, Dixon  JB, Laurie  C,  et al.  Treatment of mild to moderate obesity with laparoscopic adjustable gastric banding or an intensive medical program: a randomized trial. Ann Intern Med. 2006;144(9):625-633.
PubMed   |  Link to Article
US Food and Drug Administration. Medical devices: gastric banding. http://www.fda.gov/MedicalDevices/ProductsandMedicalProcedures/ObesityDevices/ucm350132.htm. Accessed August 1, 2012.
Dixon  JB, Zimmet  P, Alberti  KG, Rubino  F; International Diabetes Federation Taskforce on Epidemiology and Prevention.  Bariatric surgery: an IDF statement for obese type 2 diabetes. Surg Obes Relat Dis. 2011;7(4):433-447.
PubMed   |  Link to Article
Padwal  RS, Sharma  AM.  Treating severe obesity: morbid weights and morbid waits. CMAJ. 2009;181(11):777-778.
PubMed   |  Link to Article
Pories  WJ, Dohm  LG, Mansfield  CJ.  Beyond the BMI: the search for better guidelines for bariatric surgery. Obesity (Silver Spring). 2010;18(5):865-871.
PubMed   |  Link to Article
Flum  DR, Khan  TV, Dellinger  EP.  Toward the rational and equitable use of bariatric surgery. JAMA. 2007;298(12):1442-1444.
PubMed   |  Link to Article
Dixon  JB, le Roux  CW, Rubino  F, Zimmet  P.  Bariatric surgery for type 2 diabetes. Lancet. 2012;379(9833):2300-2311.
PubMed   |  Link to Article
Livingston  EH.  Inadequacy of BMI as an indicator for bariatric surgery. JAMA. 2012;307(1):88-89.
PubMed   |  Link to Article
Welcome to the Clinical Practice Research Datalink. www.cprd.com/intro.asp. Accessed October 2, 2012.
Jick  SS, Kaye  JA, Vasilakis-Scaramozza  C,  et al.  Validity of the General Practice Research Database. Pharmacotherapy. 2003;23(5):686-689.
PubMed   |  Link to Article
Mulnier  HE, Seaman  HE, Raleigh  VS, Soedamah-Muthu  SS, Colhoun  HM, Lawrenson  RA.  Mortality in people with type 2 diabetes in the UK. Diabet Med. 2006;23(5):516-521.
PubMed   |  Link to Article
Tannen  RL, Weiner  MG, Xie  D.  Use of primary care electronic medical record database in drug efficacy research on cardiovascular outcomes: comparison of database and randomised controlled trial findings. BMJ. 2009;338:b81. doi:10.1136/bmj.b81.
PubMed   |  Link to Article
Lawrenson  R, Williams  T, Farmer  R.  Clinical information for research: the use of general practice databases. J Public Health Med. 1999;21(3):299-304.
PubMed   |  Link to Article
Herrett  E, Thomas  SL, Schoonen  WM, Smeeth  L, Hall  AJ.  Validation and validity of diagnoses in the General Practice Research Database: a systematic review. Br J Clin Pharmacol. 2010;69(1):4-14.
PubMed   |  Link to Article
Azoulay  L, Schneider-Lindner  V, Dell’aniello  S, Filion  KB, Suissa  S.  Thiazolidinediones and the risk of incident strokes in patients with type 2 diabetes: a nested case-control study. Pharmacoepidemiol Drug Saf. 2010;19(4):343-350.
PubMed
Lewis  JD, Brensinger  C.  Agreement between GPRD smoking data: a survey of general practitioners and a population-based survey. Pharmacoepidemiol Drug Saf. 2004;13(7):437-441.
PubMed   |  Link to Article
Booth  HP, Prevost  AT, Gulliford  MC.  Epidemiology of clinical body mass index recording in an obese population in primary care: a cohort study. J Public Health (Oxf). 2013;35(1):67-74.
PubMed   |  Link to Article
Livingston  EH.  Procedure incidence and in-hospital complication rates of bariatric surgery in the United States. Am J Surg. 2004;188(2):105-110.
PubMed   |  Link to Article
National Heart, Lung, and Blood Institute, National Institutes of Health. Clinical Guidelines on the Identification, Evaluation, and Treatment of Overweight and Obesity in Adults: the Evidence Report. Bethesda, MD: National Institutes of Health; 1998. NIH publication 98-4083.
Office for National Statistics. Mortality statistics: quality and methodology information. 2012. http://ons.gov.uk/ons/index.html. Accessed February 26, 2013.
Perk  J, De Backer  G, Gohlke  H,  et al; European Association for Cardiovascular Prevention & Rehabilitation (EACPR); ESC Committee for Practice Guidelines (CPG).  European guidelines on cardiovascular disease prevention in clinical practice (version 2012): the Fifth Joint Task Force of the European Society of Cardiology and Other Societies on Cardiovascular Disease Prevention in Clinical Practice (constituted by representatives of nine societies and by invited experts). Eur Heart J. 2012;33(13):1635-1701.
PubMed   |  Link to Article
Hensrud  DD, Klein  S.  Extreme obesity: a new medical crisis in the United States. Mayo Clin Proc. 2006;81(10)(suppl):S5-S10.
PubMed   |  Link to Article
Bray GA, Pi-Sunyer X, Mulder JE. Health hazards associated with obesity. http://uptodate.com/contents/health-hazards-associated-with-obesity-in-adults. Accessed September 29, 2012.
Steyerberg  EW, Vickers  AJ, Cook  NR,  et al.  Assessing the performance of prediction models: a framework for traditional and novel measures. Epidemiology. 2010;21(1):128-138.
PubMed   |  Link to Article
Cook  NR.  Statistical evaluation of prognostic versus diagnostic models: beyond the ROC curve. Clin Chem. 2008;54(1):17-23.
PubMed   |  Link to Article
Pencina  MJ, D’Agostino  RBS  Sr, D’Agostino  RB  Jr, Vasan  RS.  Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Stat Med. 2008;27(2):157-172, 207-212.
PubMed   |  Link to Article
Robertson  HT, de los Campos  G, Allison  DB.  Turning the analysis of obesity-mortality associations upside down: modeling years of life lost through conditional distributions. Obesity (Silver Spring). 2013;21(2):398-404.
PubMed   |  Link to Article
Padwal  RS, Pajewski  NM, Allison  DB, Sharma  AM.  Using the Edmonton obesity staging system to predict mortality in a population-representative cohort of people with overweight and obesity. CMAJ. 2011;183(14):E1059-E1066. doi:10.1503/cmaj.110387.
PubMed   |  Link to Article
Berrington de Gonzalez  A, Hartge  P, Cerhan  JR,  et al.  Body-mass index and mortality among 1.46 million white adults. N Engl J Med. 2010;363(23):2211-2219.
PubMed   |  Link to Article
Livingston  EH, Ko  CY.  Effect of diabetes and hypertension on obesity-related mortality. Surgery. 2005;137(1):16-25.
PubMed   |  Link to Article
Wilson  PW, D’Agostino  RB, Levy  D, Belanger  AM, Silbershatz  H, Kannel  WB.  Prediction of coronary heart disease using risk factor categories. Circulation. 1998;97(18):1837-1847.
PubMed   |  Link to Article
Woloshin  S, Schwartz  LM, Welch  HG.  Risk charts: putting cancer in context. J Natl Cancer Inst. 2002;94(11):799-804.
PubMed   |  Link to Article
Must  A, Spadano  J, Coakley  EH, Field  AE, Colditz  G, Dietz  WH.  The disease burden associated with overweight and obesity. JAMA. 1999;282(16):1523-1529.
PubMed   |  Link to Article

Correspondence

CME
Also Meets CME requirements for:
Browse CME for all U.S. States
Accreditation Information
The American Medical Association is accredited by the Accreditation Council for Continuing Medical Education to provide continuing medical education for physicians. The AMA designates this journal-based CME activity for a maximum of 1 AMA PRA Category 1 CreditTM per course. Physicians should claim only the credit commensurate with the extent of their participation in the activity. Physicians who complete the CME course and score at least 80% correct on the quiz are eligible for AMA PRA Category 1 CreditTM.
Note: You must get at least of the answers correct to pass this quiz.
Please click the checkbox indicating that you have read the full article in order to submit your answers.
Your answers have been saved for later.
You have not filled in all the answers to complete this quiz
The following questions were not answered:
Sorry, you have unsuccessfully completed this CME quiz with a score of
The following questions were not answered correctly:
Commitment to Change (optional):
Indicate what change(s) you will implement in your practice, if any, based on this CME course.
Your quiz results:
The filled radio buttons indicate your responses. The preferred responses are highlighted
For CME Course: A Proposed Model for Initial Assessment and Management of Acute Heart Failure Syndromes
Indicate what changes(s) you will implement in your practice, if any, based on this CME course.
Submit a Comment

Multimedia

Supplement.

eFigure. Cohort assembly

eTable. Sensitivity analysis: accelerated failure time model results

Supplemental Content

Some tools below are only available to our subscribers or users with an online account.

Web of Science® Times Cited: 4

Related Content

Customize your page view by dragging & repositioning the boxes below.

Articles Related By Topic
Related Collections
PubMed Articles
JAMAevidence.com

The Rational Clinical Examination EDUCATION GUIDES
Abdominal Aortic Aneurysm