The Safety and Efficacy of Bariatric Surgery: The Longitudinal Assessment of Bariatric Surgery (LABS)

Rationale for Research Designs in Bariatric Surgery

The 1991 NIH Consensus Conference generated several important and unanswered questions. The LABS Steering Committee initially considered several potential study designs to address such questions and determined that, within the constraints of available time and resources, an observational study was the most efficient means to test a multitude of important hypotheses about bariatric procedures. Hence, LABS includes at its core, a multi-center, observational database including prospectively collected detailed patient and operative characteristics; short and longer-term clinical outcomes; and behavioral and health system outcomes. Because relatively little is known about the factors involved in adverse and favorable outcomes in bariatric surgery, a meticulously designed and implemented database that includes substantial data in a variety of content areas was considered necessary to test a multitude of hypotheses. Although randomized clinical trials provide higher order evidence of efficacy than do observational studies, the number of questions that can be addressed by a clinical trial is more limited.

In addition, the context of this decision to favor an observational study over a randomized controlled trial (RCT) was guided by the fact that research in the field of bariatric surgery has been associated with several challenges in the study of surgical versus non-surgical treatment for severe obesity; feasibility, lack of comparable outcomes, and ethical issues. Past RCTs have compared procedures (e.g., vertical banded gastroplasty versus gastric bypass ¹²) or surgical techniques (e.g., laparoscopic versus open gastric bypass ¹³). However, there are considerable ethical and compliance barriers to a RCT comparing operative and non-operative approaches to obesity treatment. In 1995, the Swedish Obesity Subjects study (SOS), the largest prospective study to date of bariatric surgery, was unable to obtain Institutional Review Board approval for a RCT of bariatric surgery versus medical care due to the known risks and unclear expected benefits of surgical therapy ¹⁴. Randomizing patients who do not have access to bariatric surgery for financial reasons was also considered by the LABS Steering Committee. This option was ruled out as the research funds that would be required to pay for both the surgical and medical interventions were not available and the financial incentive of a potentially cost-free operation was judged to represent excessive coercion. Furthermore, in its recent Evidence Report on Pharmacological and Surgical Treatment of Obesity, the Agency for Health Research and Quality states that the data were considered to be so conclusive regarding the superiority of surgical therapy to existing pharmaceutical and diet therapies in patients with a body mass index (BMI) greater than 40 kg/m² that, in the absence of significant advances in medical therapies, comparative studies were not considered to be warranted ¹⁵. Given the limited weight loss associated with even the best studies of non-surgical approaches, it would be difficult to recruit extremely obese subjects willing to be randomized and to have equipoise on the part of clinicians ¹⁶.

The use of a non-randomized comparator group including people who did not have bariatric surgery was also considered. However, several limitations of that approach make it difficult to interpret results. These include differences between those who do and who do not go to surgery, including physical and mental health issues, differences in socioeconomic status, and underlying physician or patient motivation all of which may affect results and limit the ability to generalize the findings.

Given these difficulties and finite resources, the LABS consortium determined that high priority scientific questions could be most efficiently addressed by creating an extensive database to test and to explore hypotheses related to bariatric surgical outcomes. The limitations of such an observational approach were also considered. Optimally, data should be collected prospectively and in sufficient detail to measure known and currently unknown factors related to outcomes. LABS investigators attempted to standardize definitions of data items and data collection procedures. Common protocols were designed to include specific data collection time points, data collection instruments, and methods for computerizing data collection, entry, and analysis. Manuals of operations and procedures were created that defined each data element and provided instructions for collection. Data collectors, including study coordinators and surgeons, underwent training and certification with respect to protocols and data collection definitions prior to collecting any data for LABS. Data elements included breaking down each of the surgical procedures into its component parts (“surgical procedural components”, e.g. length of alimentary limb, pouch size) and structuring a measuring scheme for each. Extensive quality control procedures were put into place and procedures to identify and correct faulty data implemented. Frequent contact between DCC and Clinical Center staffs, including site visits, help to assure complete and accurate data collection. Operations memos to clarify or modify procedures are distributed to all relevant study personnel to maintain consistency in data collection across sites.

Overview of LABS’ Goals

The goals of the LABS study are to assess the risks and health benefits associated with bariatric surgery and to identify aspects of the procedures as well as patient characteristics that are associated with optimal outcomes. To achieve these goals, LABS investigators defined a range of several relevant outcome domains in bariatric surgery. Whenever possible, LABS includes objective measures of patient status and co-morbid disease burden. When objective measures of disease (e.g. – 24 hour pH testing for gastroesophageal reflux) are not feasible, validated and standardized data collection instruments are used, if available. Investigators sought to identify existing data collection instruments which are psychometrically sound. When validated data collection instruments were not available, LABS investigators created new instruments appropriate for patients undergoing bariatric surgery or adapted questionnaires from other clinical studies.

The LABS study is organized into three phases, termed respectively LABS-1, LABS-2, and LABS-3. LABS-1 includes all patients who are at least 18 years of age and who undergo bariatric surgery at participating centers by LABS-certified surgeons. The primary goal of LABS-1 is the evaluation of the short-term safety of bariatric surgery. The primary endpoints include important adverse outcomes, such as death and percutaneous or operative re-intervention, that occur within 30 days of operation. A limited dataset of patient and operative characteristics is gathered to describe the frequency of these events in different subgroups and to assess the relationship between adverse outcomes and patient and operative characteristics. The sample size for LABS-1 was selected to be able to obtain precise estimates of safety events (e.g. 30-day mortality), and to have at least 80% power to identify a two-fold (or greater) increase in the risk of 30-day mortality, re-interventions, or deep vein thrombosis/pulmonary embolism between important subgroups (e.g. men vs. women, BMI under 50 kg/m² vs. BMI at least 50 kg/m²). (LABS-1 forms)

The primary goal of LABS-2 is to evaluate the longer-term safety and efficacy of bariatric surgery and to more comprehensively evaluate patient characteristics as they relate to short- and intermediate-term outcomes. The sample size for LABS-2, approximately 2400 patients, was determined to be able to detect “small” effect sizes ¹⁷ for continuous outcomes (e.g. change in excess body weight), and at least a two-fold increase or decrease in incidence or prevalence for categorical efficacy outcomes (e.g. resolution of diabetes). In as much as LABS was designed to address multiple hypotheses, many of which are exploratory, adjustments for multiple comparisons did not enter into the sample size calculation. When appropriate for particular analyses, such adjustments will be considered and implemented. The sample size, which provides adequate statistical power to detect differences in occurrence of some uncommon events in relatively small subgroups, will be adequate to allow such adjustment when events are more common or subgroups are larger. To address the aims of LABS-2, the extensive collection of demographic, anthropometric, clinical, behavioral, surgical, and post-surgical care variables will be used to determine their associations with outcomes. Data will be collected prior to surgery, during the surgery, and post-surgery (30 days, 6-month, 1-year, 2-year follow-up visits). LABS-3 will involve further subsets, determined in composition and size by the hypotheses underlying the mechanisms to be studied, of patients from the LABS-2 cohort. One LABS-3 study measures psychosocial and behavioral aspects of obesity, such as quality of life whereas a second LABS-3 study examines the mechanisms underlying diabetes resolution. A proposed third LABS-3 study will assess changes in health economic outcomes and resource utilization. Other aspects of the pathophysiology of obesity and obesity-related diseases will be studied, either through additional sub-studies or by separately funded Ancillary Studies approved by the LABS Ancillary Studies Subcommittee (Funded LABS Ancillaries & ASC guidelines).

Approach to Define the Operative Procedures

While there are several different types of bariatric procedures, the Roux-en-Y gastric bypass (RYGB) is the most commonly performed in the US ¹⁸. The restrictive laparoscopic adjustable gastric band (LAGB) is increasingly used in the US ¹⁹. The biliopancreatic diversion (BPD), with or without the duodenal switch (BPD DS), has also grown in use. Finally, some LABS centers perform a planned two-stage procedure in very high risk, extremely obese patients in which a BPD or BPD DS follows an initial gastric sleeve. There is widespread variability among surgeons in the ways RYGB and BPD are performed. Different approaches to the size of the gastric pouch, length of the alimentary limb, method of anastomotic construction, and extent of nerve interruption of the stomach may or may not have important roles in postoperative outcomes. Thus, any data-gathering project in bariatric surgery that aims to link procedural characteristics with outcomes must acknowledge such variability and either standardize the intervention or measure it with enough detail to assess and analyze the variation. The LABS study elected the latter approach, and will measure each aspect of this variation in detail.

Although the placement of the LAGB is more standardized, the methods and strategies for post-procedural band adjustment also vary considerably between practitioners. This is particularly important for post-procedural interventions in the LAGB where it has been demonstrated that adjustments (rather than the band alone) are the key to effective weight loss ²⁰. This may also be relevant for RYGB and BPD, where behavioral and dietary interventions before and after the procedure may have important effects on the outcome. If unmeasured, the impact of these pre-, peri-, and post-procedural components of bariatric surgery might inappropriately be ascribed to the procedure itself. LABS will not standardize surgical procedures or pre- or post operative care, per se, but rather will attempt to measure and then describe this variability through strictly defined and enforced/audited data collection and analysis.

Approach to Assessing Outcome Domains in Bariatric Surgery

The short- and longer-term risks of bariatric surgery differ with aspects of the procedure ²¹ and in relation to patient characteristics, such as age, sex, and co-morbidities ^9,22,23. A complete assessment of bariatric surgery involves ascertaining its impact on patient health, well-being, quality of life, and health care utilization. For descriptive purposes, LABS investigators have clustered outcomes into “safety” and “non-safety” domains, The terminology involved in describing untoward events has been problematic within the field of bariatric surgery. Anastomotic complications are of primary interest in the RYGB and BPD procedures. However, there are no universally agreed-upon definitions ²⁴. To assess which anastomotic complications clinicians consider to be clinically significant, LABS investigators developed a measurement scheme to record all POR and details of the presumed and confirmed reasons. For standardization, LABS has an Adjudication Committee that will review and classify all deaths and those surgical re-operations and unplanned post-discharge anticoagulation therapies for which the reason could not be confirmed.

The limited dataset of LABS-1 was designed to describe the frequency of interventions or hospitalizations occurring within 30 days after bariatric surgery and factors associated with those events and with death. The duration of post-procedure surveillance for important adverse outcomes determines the number and types of events identified. Administrative claims demonstrate that data analyses looking only at in-hospital mortality miss approximately half of the deaths within 30-days of gastric bypass ²². The appropriate time window to describe a post-surgical event remains unclear. By convention, events occurring within 30-days after operation are considered surgically related.

The LABS-2 dataset was designed to measure and evaluate longer-term outcomes and to evaluate the efficacy of bariatric surgery. Key groups of “non-safety” outcomes for LABS-2 include (but are not limited to) weight loss, changes in body composition, functional impairment, psychosocial function (including quality of life), cardiovascular, metabolic, pulmonary, renal, musculoskeletal, urogynecologic, reproductive, and gastrointestinal outcomes.

Risk stratification

Risk stratification is a useful tool to put safety measures into context. Currently, there is no evidence-based scheme for risk stratification in bariatric surgery. Acquiring the broad evidence base to support the development of such a risk-stratification tool is one of the goals of LABS. Administrative claims and case series data suggest that older patients and males have higher risk of perioperative death than younger patients and females, respectively ^23,25. However, little is known about other patient or procedural factors that relate to outcomes. LABS-1 and LABS-2 will study the extent to which patient demographic characteristics, BMI, co-morbid conditions, medication use and physiologic status relate to these outcomes. From this evidence-base of associated factors, LABS investigators plan to provide the evidence and elements for a meaningful risk stratification/adjustment strategy. Since many poor surgical outcomes are relatively infrequent events, a large study cohort is necessary to define a risk stratification scheme.

LABS-1 is testing the following risk assessment hypotheses: that men have higher rates of POR and 30-day mortality rate than women, that greater BMI is associated with higher rates of VTE, POR and 30-day death, and that wound infection is more common among open than laparoscopic procedures. Furthermore, the severity of various co-morbidities, collected on all participants in LABS-1, will be examined for association with poor outcomes (LABS-1 Pre-Operative form). LABS-2 includes more detailed measures of severity than LABS-1 and analyses will help determine if they are more predictive of events than the less detailed measures used in LABS-1. Other hypotheses to be tested in LABS-2 include whether other, more detailed, patient factors and characteristics of the surgical procedure (“surgical procedural components”) are related to outcome.

Outcome Domains in Bariatric Surgery

The outcome domains studied in LABS-2 and described below are listed in Table 1. This comprehensive evaluation of outcomes resulted from the consortium’s collaborative efforts in structuring a thorough, yet feasible and standardized, database. Table 2 outlines the standard forms and measures used in LABS-2 to assess each of these domains as well as the contact time points at which they will be administered. Even more detailed information on each of these outcome domains along with the definitions and specific measures used can be found on the links within Table 2 or, if applicable, the text of this manuscript.

Weight Loss and Body Composition

The primary intent of bariatric procedures is to induce weight loss by limiting intake and to promote behavioral changes in overall energy balance that result in significant and sustained decreases in weight. LABS-2 will measure weight at each of the annual follow-up visits using a standard scale (Tanita model #TBF-310H01A). The LABS investigators hypothesize that men will experience greater weight loss than women and that there is a direct relationship between physical activity and weight loss at follow-up intervals. We also hypothesize that diabetic patients will lose less weight and that a longer length of bypassed limb in gastric bypass surgery is associated with greater weight loss maintenance ^26,27. (LABS Physical Measurements Protocol)

Diabetes Mellitus and Insulin Resistance

Type 2 diabetes mellitus (T2DM), the metabolic syndrome, and insulin resistance syndrome (IRS) are common metabolic co-morbidities of obesity. Many case series have demonstrated significant and sustained improvements in these parameters after weight loss procedures ⁶ but measures of these parameters in large cohorts have been limited. LABS-2 will evaluate the longer-term efficacy of bariatric surgery with respect to T2DM based on a clinical history of medication use and on serial measurements of fasting blood glucose, and HbA1c. Assessing efficacy for preventing or resolving the metabolic syndrome and IRS will be based on fasting glucose, insulin, lipoprotein profiles, resting blood pressure, and waist circumference. It is hypothesized that improvement in T2DM, metabolic syndrome, and IRS will be related to degree of weight loss, degree of loss of fat mass, and level of physical activity at follow up ^28,29.

Cardiovascular and Pulmonary Disease

Obesity is a major risk factor for cardiovascular diseases (CVD) and obstructive sleep apnea and has been increasingly recognized in patients with extreme obesity ^6,30-35 The prevalence of sleep apnea and changes from baseline status will be assessed by self-report through the use of the Berlin Sleep Questionnaire ³⁶ and by the reported use of positive airway pressure devices. It is hypothesized that weight loss and reductions in neck circumference will be associated with improvements in sleep apnea. To assess the efficacy of bariatric surgery to reduce risk for CVD, LABS-2 will measure C-reactive protein, lipoprotein profiles, resting blood pressure, waist circumference, and clinical history of medication use. It is hypothesized that improvement in CVD risk factors will be related to the magnitude of weight loss, loss of fat mass, and lower BMI post-surgery ³⁷. Furthermore, change in cardiac function will be measured by the time to complete a 400-meter corridor walk. It is hypothesized that this will be related to age, BMI, sex, and other factors.

Renal Disease

Obesity causes and complicates diabetes and hypertension, the two most common causes of kidney failure³⁸. In addition, there are several mechanisms by which obesity may independently and negatively impact renal function such as by adipogenic hormones that may have direct injurious actions on the kidney ³⁹. However, bariatric surgery itself has been associated with progressive renal disease through a variety of mechanisms ⁴⁰, and may also contribute to the development of renal calculi. For these reasons LABS-2 will measure renal function by serum creatinine and cystatin and urinary albumin and creatinine, and will assess the prevalence of diagnosed nephrolithiasis at baseline and follow-up. It is hypothesized that albuminuria will diminish after successful bariatric surgery and that renal function as measured by serum creatinine will remain stable after successful surgery.

Liver Function

Another problem of growing public health concern is the increased prevalence of non-alcoholic fatty liver disease (NAFLD) in obese populations ⁴¹ and the growing identification of NAFLD when evaluated by liver biopsy in patients undergoing a bariatric surgical procedure ⁴². There are limited data available defining the prevalence and severity of steatohepatitis (NASH), as assessed by intra-operative liver biopsy, in extremely obese persons undergoing bariatric surgery, although NASH is an increasingly recognized cause of “cryptogenic” cirrhosis ⁴³. It is hypothesized that the prevalence and severity of NASH is underestimated by traditional clinical measures and that liver disease severity will be correlated with short-term post-surgical morbidity. We also hypothesize that increased liver size at the time of operation is associated with a higher rate of failed laparoscopic approaches to bariatric procedures.

Behavioral/Psychosocial Factors

Evidence suggests that pre-existing psychological and behavioral factors may influence outcomes following bariatric surgery ⁴⁴. It is hypothesized that patients who have active, untreated substance abuse, binge eating, or depression at baseline will experience higher rates of post-surgical medical complications and less weight loss ^45,46. Conversely, those subjects who intentionally lose weight prior to operation will achieve greater weight loss in the short and longer term ⁴⁷. Behavioral measures will be assessed at both baseline and follow-up, and will include questions on pre-surgery weight loss practices and eating patterns (including binge eating and eating beyond satiation), tobacco and alcohol use, history of psychiatric disorders, and counselor/therapist contact. Depressive symptoms will be assessed by the Beck Depression Inventory, Version 1 ⁴⁸. Objective measures of physical activity will be assessed by the STEPWATCH 3 Step Activity Monitor (www.cymatech.com) at baseline and follow-up visits.

Musculoskeletal and Functional Status

Osteoarthritis, either caused or aggravated by obesity, is a major limiting comorbid condition among the population of patients undergoing bariatric surgery. Functional limitations due to back, hip, and knee joint degeneration is a leading cause of functional decline, use of durable medical goods (wheel chairs, walkers, electric scooters) and impaired quality of life. LABS investigators will test the hypothesis that functional status will improve with surgery and the extent of improvement will be associated with the degree of weight loss. In addition, we will investigate whether functional limitations prior to surgery are linked to poor outcomes following the bariatric procedure. Functional status will be assessed in LABS through a combination of self-reports (walking ability, use of assistance devices) as well as a timed corridor walk.

Gender Issues

Obesity affects all aspects of well-being including those that are gender specific. There are also potential gender differences in the longer-term efficacy of bariatric surgery. For example, obesity is a known risk factor for several health conditions specific to and prevalent among women such as menstrual abnormalities, infertility ⁴⁹, and urinary incontinence ⁵⁰. It is hypothesized that menstrual abnormalities, fertility, urinary incontinence, and symptoms of polycystic ovarian diseases will improve following bariatric surgery. These will be assessed by several questionnaires.

Nutrient Deficiencies

Another potential long-term risk of bariatric surgery is nutrient deficiency ⁵¹. LABS-2 will be able to investigate micro- and macro-nutrient deficiencies by surgical procedure by hypothesizing more frequent occurrences with malabsorptive procedures ^26,52 and by various components of surgery such as roux limb length and pouch size. Plasma and serum samples will be stored in a specimen repository for future analysis of macro and micronutrients.

Economic Impact

Workers who are obese have a high prevalence of work limitations ⁵³ and severe obesity increases the number of work loss days and is an important factor in the workplace ^54,55. The effects of weight loss surgery on productivity at work, absenteeism and presenteeism are not well studied. LABS will administer several validated questionnaires to assess this impact including the Work Productivity and Activity Impairment (WPAI) form V2.0 ⁵⁶. It is hypothesized that patients undergoing surgery will lose fewer days of work and that productivity at work will improve after surgery.

Biospecimens

Blood and urine specimens will be obtained from LABS-2 participants at baseline, and post-operatively at 6 months, 12 months and annually thereafter. Aliquots of whole blood, plasma, and serum will be banked in the NIDDK Biospecimens Repository for future investigations into factors such as changes in metabolic parameters and markers of risk. An additional quantity of whole blood will be drawn to be used for future DNA analysis, with appropriate consents. These specimens will be a major resource for funded, LABS-associated Ancillary Studies. Non-LABS investigators may also request access to these biospecimens through application via the LABS ancillary studies process.