Kathleen E Renquist

A Decade of Change in Obesity Surgery

Background: The International (formerly National) Bariatric Surgery Registry began collecting data in January 1986. The aim of this study was to examine changes in the practice of surgical treatment of severe obesity that occurred during the decade of 1986 through 1995, as observed in the IBSR data. Methods: All data submitted to the IBSR during the decade were transferred to the IBM mainframe computer for analysis. Characteristics of operative type populations were compared over time using analysis of variance (ANOVA) for age, body mass index (BMI), operative weight and Chi-square (χ2) test for gender. Results: There has been a steady increase over the decade in mean patient weight. The operations used have changed from predominantly ‘simple’ operations to more frequent use of ‘complex’ operations. Within the categories of ‘simple’ and ‘complex’, an increase in the variety of operations occurred. As a group, patients with ‘simple’ operations have been heavier, more often male and public pay patients than those who have undergone ‘complex’ operations. One year weight loss was greater for Roux-en-Y gastric bypass (RGB) than vertical banded gastroplasty (VBG), but follow-up rates were too low to study the relative merits of the operations used. The reported incidence of operative mortality and serious complications (leak with peritonitis, abscess and pulmonary embolism) remained low. Conclusions: These observations and their implications can be summarized in three statements which relate to action for improved patient care in the beginning of the new century: (1) increasing weight of candidates for surgical treatment during this decade indicates the need for earlier use of operative treatment before irreversible complications of obesity can develop; (2) low risk of obesity surgery, decreasing postoperative hospital stay, and early weight control support the continued and increased use of surgical treatment; (3) continued widespread use of both ‘simple’ and ‘complex’ operations with increased modifications of standard RGB and VBG procedures emphasizes the need for standardized long-term data and analyses regarding both weight control and postoperative side-effects.

Perioperative risks and safety of surgery for severe obesity.

The National Bariatric Surgery Registry (NBSR) results reflect low perioperative risk for obesity surgery. Five deaths occurred within 40 d of operation in 5178 patients (0.1%). A subset of 3174 patients with complete information for complication and postoperative hospital stay was further studied. Females comprised 87% of the data set. Median values were determined for age, 37 y (18-70 y); operative weight, 121 kg (77-288 kg); and operative body mass index (BMI), 44 kg/m2 (29-91 kg/m2). Patients with no complications (89.7%) were reported to have a median postoperative stay of 4 d (2-23 d). The most severe complications were deep venous thrombosis (0.3%) and gastrointestinal leak (0.6%), with median postoperative hospital stay of 12 d (ranges 2-27 and 4-59 d, respectively). The most frequent complication reported was respiratory (4.5%), with median postoperative stay of 6 d (3-34 d). Median postoperative hospital stay for wound infection (1.6%) was 5 d.

Gallbladder Management in Obesity Surgery

Background: In the 1980s, some surgeons recommended routine cholecystectomy for patients undergoing bariatric surgery.This was based on the high prevalence of gallstones in the obese and concern that rapid weight loss would increase the risk of gallbladder disease. Others recommended waiting for a lower weight and a definite need.With increasing prevalence and severity of obesity and increased use of gastric reduction surgery for weight control, it seemed appropriate to review the current standard of care for cholecystectomy.A survey was also made of ursodeoxycholic acid usage for prevention of gallstone formation. Methods: Data collected from active contributors for the 28th Report of the International Bariatric Surgery Registry (IBSR) were examined. Two questionnaires were also sent to members of the American Society for Bariatric Surgery (ASBS). The first (Q1) asked about the indications for cholecystectomy. The second (Q2) asked about ursodeoxycholic acid usage for prevention of gallstone formation during rapid weight loss following surgical treatment of obesity. Results:There has been an increase in concurrent cholecystectomy during the last 15 years. Some of this is due to a shift from simple gastric restrictive operations to gastric bypass with gastric restriction. When the most extensive bypass of intestine is used, as in distal Roux-en-Y gastric bypass (RYGBPX) or biliopancreatic diversion with a duodenal switch (BPD-DS), all patients were reported to have undergone cholecystectomy. Only 30% of surgeons performing standard Roux-en-Y gastric bypass (RYGBP) remove normal-appearing gallbladders. Ursodeoxycholic acid is used to prevent gallstone formation in one-third of patients when a normal-appearing gallbladder is left in place. Conclusions: Prophylactic cholecystectomy is left to the discretion of the surgeon when RYGBP is used. There has been an increase in cholecystectomy and malabsorptive operations during the last 15 years.When most of the small bowel is bypassed, all remaining gallbladders are removed. For patients with simple restriction operations, normal-appearing gallbladders are usually left in place. Urso-deoxycholic acid during rapid weight loss for prevention of gallstone formation is used in one-third of patients with remaining gallbladders.

Factors Influencing Survival following Surgical Treatment of Obesity

Background: The prevalence of obesity in the United States and the surgical treatment of obesity have increased since 1999. An important measure of outcome following surgical treatment is survival. Methods: This study began with data prospectively collected from Jan 1, 1986 to Dec 31, 1999 by 55 data collection sites, representing 77 surgeons who used standardized data collection software developed by the International Bariatric Surgery Registry (IBSR). A subset of 18,972 subjects was submitted to the National Death Index (NDI) for search of death occurring from Jan 1, 1986 to Dec 31, 2001. The univariate survival analysis included Kaplan-Meier plots and log-rank tests. Cox proportional-hazards (PH) frailty model was used to identify risk factors and estimate hazard ratios in a multi-factor survival analysis. Covariates included gender, operative age, body mass index, operation category (simple and complex), operation year, diabetes, smoking and hypertension as recorded prior to operation. Results: Deaths were found for 3.45% of the patients (654/18,972). Average follow-up was 8.3 years. Age, gender, BMI, history of smoking, diabetes, and hypertension were significant predictors of survival. Operation category (P=0.13) and operation year (P=0.89) were not significant predictors of survival. Conclusion: Simple and complex operations were equally effective in keeping patients alive in this cohort of patients operated on for severe obesity from 1986 to 1999. Young, female, non-smoking patients with low BMI at operation and no history of diabetes or hypertension had the longest survival. Longer follow-up for death is needed before any recommendations can be made for operation category based on survival.

The Effect of Follow-up on Reporting Success for Obesity Surgery

Background: Much is written about the importance of follow-up in determining the effect of surgical treatment for obesity upon weight loss. When patients are lost to follow-up, it has been suggested that these patients should be considered as failures. This study was undertaken to determine the effect of incorporating patients not followed in a definition of success for weight loss at one year. Methods: Data from 34 surgical practices were used to study the effect of using two different denominators, patients followed (DF) or patients eligible (DE), to define success. The numerator used in both methods was the number of patients with ≤50% EW at 1 year. Results: One-year follow-up was 61% (5091/8356). Success was 67% (3423/5091) when calculated using the denominator patients followed (DF). No correlation was found between success and follow-up when data within each surgical practice were averaged and used in a correlation analysis. Conclusions: This study does not support the thesis that patients who fail to return for follow-up should be considered as having failed in weight control.

A Study on Predicting Weight Loss Following Surgical Treatment for Obesity

To date, longitudinal weight loss analyses (curve-fitting) have been complicated by non-linear weight patterns, incomplete follow-up, and varied follow-up times. Therefore, the cross-sectional design (one time point survey) was chosen to study predictors of weight loss at yearly postoperative time intervals (± 6 months). Mean values for the initial cohort of 7,540 patients were: age 37.4 years (± 9.4), weight 124.0 kg (± 25.5), height 165.4 cm (± 8.5). Females comprised 87.7% of the data set. Followup was 62.5% at year one, 32.0% at year 2, 20.6% at year 3, and 15.4% at year 4. The multiple regression model used included 20 explanatory variables and was performed separately for four yearly time points. Only operative weight, initial visit height, age, and operative type were consistent predictors of weight loss (p < 0.05) at all time points examined and accounted for 40-50% of weight loss variation.

Pay Status as a Predictor of Outcome in Surgical Treatment of Obesity.

Background: Higher complication rates and lower success in surgery for severe obesity have been reported for patients with government pay status. We examined the effect of pay status upon outcome in surgical treatment of obesity. Methods: This was an observational study from an aggregate data set of individual patient information. Government pay status (G) was defined as full or partial medical care payment through Medicare, Medicaid, or Veterans Administration. Payment entirely by private insurance was defined as private (P). Operations were classified as either simple (S, gastric restriction) or complex (C, gastric restriction with small bowel bypass). Two measures of outcome, perioperative complication rate and weight loss success (≤50% excess weight), were examined to determine pay status effect. Results: More G than P patients were treated with simple procedures (79% vs 51%, p < 0.05). Perioperative complication rates were more common for G than P patients (14.4% vs 9.1%, p < 0.05). One-year weight loss success was higher for P than G, regardless of operation type. Conclusion: Pay status should be included in characterization of patient groups and in the analysis of results when effectiveness of surgical treatment for severe obesity is reported.

Calculating Follow-up Rates

Follow-up rates are presented as a percentage but the method of calculation is seldom discussed. Determining a follow-up rate begins with identifying the numerator and denominator used in the calculation. Four methods of calculating follow-up rate after surgical treatment for weight reduction were studied. Method 1 follow-up rate, 91.2%, was calculated using patients with at least one follow-up contact as the numerator (6169) and the number of primary operations performed as the denominator (6764). Method 2 calculated patient contact on or beyond a specific time point: 65.9% for ≥ 6 months (4232/6424), ≥ 1 year (y) = 52.1% (3111/5977), ≥2 y = 29.8% (1455/4890), ≥3 y = 18.8% (710/3784), ≥4 y = 12.5% (331/2643). Follow-up calculated using method 3 determined patient contact within specific time points: 0-1 y = 89.7% (6064/6764), 1-2 y = 45.8% (2739/5977), 2-3 y = 23.0% (1124/4890), 3-4 y = 13.9% (526/3784), 4-5 y = 11.3% (298/2643). Method 4 studied patient contact within time points according to ASBS guidelines. Rates for method 4 were: 0-6 months = 58.8% (11,938/20,292), 6-12 months = 36.7% (4717/12,848), 1-2 y = 45.8% (2739/5977), 2-3 y = 23.0% (1124/4890), 3-4 y = 13.9% (526/3784). Results using these four methods of calculation emphasize the need to standardize follow-up rate before operative comparisons can be made.

Weight Loss Curve Analysis

The purpose of this study was to develop a mathematical model to describe weight loss trend over time and to determine differences, if any, among various weight loss trends. Weight change was studied following Roux-en-Y gastric bypass (RGB) and vertical banded gastroplasty (VBG). The total number of patients analysed was 3172, 63.5% VBG and 36.5% RGB. Median age (years) for VBG was 36 (range 18-70) and RGB 38 (range 18-66). Median operative body mass index (BMI) (kg/m2) for VBG was 45.0 (range 30.1-90.5) and RGB 43.3 (range 30.4-80.6). Females represented 87% of the VBG patients and 89% of RGB patients. Obesity category percentages for each operative type were: super--VBG 20%, RGB 14%; morbid--VBG 78%, RGB 80%; obese--VBG 2%, RGB 6%. The development of a mathematical model to study weight loss over time was a two-stage process. First, weight loss trend was determined as evidenced by individual patient BMI values plotted over time. It was observed that plots represented quadratic curves, so the model BMI = B0 + B1(time) + B2(time)2 was used. Second, the estimated betas (B0, B1, B2) were treated as response variables to estimate weight loss trend for the combination effect of three independent variables: gender (male and female), obesity category (super, morbid, obese) and operative type (VBG and RGB). It was found that the three independent variables simultaneously interact with patient weight loss (p = 0.0001).

Predictors of Two Obesity Complications: diabetes and hypertension

Gender, age, waist-hip ratio (WHR) and body mass index (BMI) have been reported to herald two complications of obesity: diabetes or hypertension. Most literature is based on patient populations with mean BMIs ranging from 22 to 35 kg/m2. This study population of severely obese patients selected for surgical treatment of obesity had a mean age of 37 ± 9.4 years, median WHR of 0.9 (0.4-2.1) and median BMI of 44.0 (29-89) kg/m2. It was found that age, WHR and BMI were significant predictors for diabetes or hypertension. Gender was not found to be a significant predictor for either diabetes or hypertension. A mathematical model was used to calculate the predictive probability of developing diabetes or hypertension using age, BMI and WHR.