Abstract
IMPORTANCE
Estimates of weight regain following bariatric surgery vary widely.
OBJECTIVE
Describe weight regain following post-Roux-en-Y gastric bypass (RYGB) weight nadir, and compare common weight regain measures for association with clinical outcomes.
SETTING
Ten hospitals in six US cities: Portland, OR, Seattle, WA, Fargo, ND, Pittsburgh, PA, New York, NY and Greenville, NC.
DESIGN AND PARTICIPANTS
Adults undergoing bariatric surgical procedures entered a prospective cohort study between February 2006-February 2009 (N=2458) and completed pre-surgery, six month and annual assessments for up to seven years through January 2015.
Participants who underwent RYGB and were followed ≥5 years with ≥5 weight measurements were included (83%; 1406 of 1703, excluding deceased/reversals).
EXPOSURE
Weight regain assessed by five continuous measures (kg, body mass index [BMI; kg/m2], % pre-surgery weight, % nadir weight and % maximum weight loss) and eight dichotomous measures (per established thresholds) were compared in relation to clinical outcomes based on statistical significance, magnitude of association and model fit.
MAIN OUTCOME MEASURES
Progression of diabetes, hyperlipidemia, and hypertension, and declines in physical and mental health-related quality of life, and satisfaction with surgery.
RESULTS
Medians (25th–75thpercentiles) are reported. Pre-surgery age was 47 years (38–55) and BMI was 46.3 (42.3–51.8). Most participants were female (80.3%) and white (84.9%). Follow-up was 6.6 (5.9–7.0) years. Maximum weight loss was 37.4% (31.6%−43.3%) of pre-surgery weight, occurring 2.0 (1.0–3.2) years post-surgery. The rate of weight regain was highest in the first year following weight nadir, but regain continued across follow-up, ranging from 9.5% (4.7%−17.2%) to 26.8% (16.7%−41.5%) of maximum weight lost, one to five years post-nadir. The % participants who regained weight depended on threshold (e.g., five years post-nadir, 43.6% regained ≥5 BMI points, 50.2% regained ≥15% of nadir weight and 67.3% regained ≥20% of maximum weight lost). Percentage of maximum weight lost vs. other continuous weight regain measures, had the strongest associations with, and best model fits for, all outcomes except hyperlipidemia, which had a slightly stronger association with regain in BMI. Of dichotomous measures, ≥20% of maximum weight lost performed best with all outcomes except hyperlipidemia and satisfaction (≥10 kg and ≥25% of maximum weight lost were superior, respectively).
CONCLUSIONS
Among a large cohort of adults who underwent RYGB, weight regain quantified as percentage of maximum weight lost performed better for association with most clinical outcomes than the alternatives examined. These findings may inform standardizing measurement of weight regain in studies of bariatric surgery.
Keywords: bariatric surgery, Roux-en-Y gastric bypass, obese, weight nadir, weight gain, weight recidivism, weight maintenance
INTRODUCTION
Although weight loss patterns vary, bariatric surgery results in substantial and durable weight reduction for the majority of patients, making it the most effective treatment for severe obesity1–3. Still, like all weight loss interventions, weight regain, which may have deleterious effects on weight-related comorbidities1,4,5, health-related quality of life (HRQoL)6–8, patient satisfaction with surgery7, and health care costs9, is a concern following surgical treatment of obesity10,11.
Numerous studies have been conducted to understand the extent of weight regain following bariatric surgery. However, reported weight regain varies widely across studies10,11. Small samples and lack of generalizability of case series, use of clinical records or patient recall to estimate nadir weight, loss to follow-up, and variable follow-up time likely contribute to the inconsistency in the literature10,11, as does the lack of a standard measure of weight regain7,11,12. For example, weight regain has been calculated as absolute change in weight or body mass index (BMI), and as a percentage of pre-surgery weight, nadir weight, or maximum weight lost7,10,11. Furthermore, thresholds are commonly used to indicate clinically meaningful weight regain (e.g., ≥10%, ≥20% and ≥25% of maximum weight lost), without demonstrating the biological relevance or function of the specified threshold.7,11,12
The purpose of this study was to inform standardizing measurement of weight regain in studies of bariatric surgery by 1) describing weight regain following weight nadir assessed by five continuous and eight dichotomous measures in a large geographically diverse cohort of adults who underwent Roux-en-Y gastric bypass (RYGB) who were followed for five-seven years1; and 2) comparing the performance of these weight regain measures for association with six clinical outcomes (concurrent progression of diabetes, hyperlipidemia and hypertension, and decline in physical HRQoL, mental HRQoL and patient satisfaction with surgery).
MATERIALS AND METHODS
Design and Subjects
The Longitudinal Assessment of Bariatric Surgery-2 (LABS-2) study was a prospective cohort study of adults at least 18 years old undergoing a first bariatric surgical procedure between March 2006 and April 2009 performed by a participating surgeon at ten hospitals from six clinical centers throughout the United States (Figure 1)13. The institutional review board at each center approved the protocol, available online14, and participants gave written informed consent.
Standardized research assessments were conducted by research investigators and their research staff within 30 days pre-surgery, at 6 months post-surgery and annually post-surgery for six or seven years through January, 2015. Assessments were primarily conducted in-person, with the exception of the 6-month and 6 year assessments, which were brief and completed by telephone or mail. Depending on when participants underwent bariatric surgery, they had a maximum of eight or nine assessments over six or seven years, respectively. This analysis included participants who underwent RYGB, did not die or undergo a reversal or revision to a new bariatric procedure within five years, and met the criteria for nadir weight determination described below (Figure 1; N=1406).
Weight assessment.
During in-person assessments, weight was measured on a standard scale (Tanita Body Composition Analyzer, modelTBF-310). If this per-protocol weight was not obtained, weight measured by research or medical personnel on a non-study scale was utilized. If neither was available, a participant’s self-report of current weight based on any scale was used. In a prior LABS-2 study self-reported weights gathered during a phone contact that occurred within 30 days prior to in-person measured weights were evaluated15. Differences between measured and self-reported weights in this cohort averaged ≤1 kilogram and did not differ significantly by measured BMI or degree of postoperative weight change15.
Nadir weight determination.
The lowest weight among participants whose weight was measured at five or more assessments, at least one of which occurred during or after the five year assessment, was classified as nadir weight if weight was not missing at the assessments due immediately prior to and immediately following it. If the lowest weight was measured during the final year of study data collection, only weight at the assessment due immediately prior to it was required.
Weight loss and weight regain.
Weights measured during or within six months following a pregnancy were excluded. Weight loss from surgery was calculated in kilograms (kg), BMI (kg/m2), and percentage of pre-surgery weight16. Weight regain from nadir weight was calculated in kg17, BMI18, percentage of pre-surgery weight19, percentage of weight nadir20–23, and percentage of maximum weight lost24,25. Formulas are provided in supplemental material (eAppendix 1). In addition, eight common thresholds for clinically meaningful weight regain (i.e., ≥10 kg11, ≥5 BMI points 11, ≥10% of pre-surgery weight19, ≥10%20 and ≥15%26 of nadir weight, and ≥10%27, ≥20%24 and ≥25%28 of maximum weight lost) were applied. Some previously reported weight regain thresholds (e.g., “any” weight regain, regain to a BMI >35 after successful loss) were not utilized due to their lack of plausible biological relevance7,16.
Clinical (non-weight) outcomes of bariatric surgery.
Using laboratory tests, physical measures, and questionnaires described below, six outcomes were calculated for each assessment following weight nadir, with status at weight nadir as the reference. Outcome definitions were developed a priori specifically for this analysis after the completion of the LABS-2 study, using standard thresholds when available to determine progression of disease or minimal clinically important decline.
Glycated hemoglobin (HbA1c) and low-density lipoprotein (LDL) were measured by a central laboratory. Blood pressure was measured by research staff. These data were not collected at the year 6 assessment. Medication use was self-reported. HRQoL was measured with the Medical Outcomes Study 36-item Short-Form Health Survey (SF- 36). The physical component summary (PCS) and mental component summary (MCS) scores are commonly used in reports of bariatric surgery and sensitive to weight loss and gain29. Norm-based methods transform the scores to a mean of 50 and standard deviations of 10 in the general U.S. population (range 0–100)30. Lower scores imply more disability/worse function. In January 2010 an item was added to the annual assessment to assess participants’ satisfaction with the results of their first bariatric surgery on a 7 point scale from 1, “very satisfied,” to 7, “very dissatisfied.”
Progression of diabetes was defined as: 1) a change from not taking diabetes medication to taking diabetes medication; 2) a change from not taking insulin to taking insulin; or 3) an increase of HbA1c by at least 0.5% to at least 5.7%31,32. Among women reporting polycystic ovarian syndrome, the first criteria required use of a diabetes medication other than metformin. Progression of hyperlipidemia was defined as: 1) a change from not taking lipid lowering medication to taking lipid lowering medication; or 2) an increase of LDL by at least 10 mg/dL to at least 100 mg/dL33. Progression of hypertension was defined as: 1) a change from not taking hypertension medication to taking hypertension medication; 2) an increase of systolic blood pressure (SBP) by at least 5 mm Hg to at least 120 mm Hg; or 3) an increase of diastolic blood pressure (DBP) by at least 5 mm Hg to at least 80 mm Hg or greater34. A clinically important decline in physical and mental HRQoL was defined as a decrease ≥5 points in the PCS and MCS scores, respectively35. A clinically important decline in satisfaction with surgery was defined as an increase by at least 1 point to a rating of at least 3 (i.e., “somewhat satisfied” – “very dissatisfied”).
Sociodemographics.
Sociodemographics, which are reported as descriptors of the analysis sample, were self-reported. Participants were asked whether they were married or “living as married.” Race was set to missing for participants who did not self-report their race as one or more of the investigator- defined categories (i.e., white/Caucasian, black/African-American, Asian, American Indian/Alaska Native, Native Hawaiian/other Pacific Islander).
Analysis.
Analyses were conducted using SAS version 9.4 (SAS Institute, Cary, NC, USA). All reported P values are two-sided; P values less than 0.05 are considered to be statistically significant. Data were assumed to be missing at random. Mixed models described below controlled for site, and pre-surgery age and smoking status, which were associated with missing follow-up data1, as fixed effects. The assumption of weight, comorbidity and SF-36 data missing at random in the LABS-2 cohort was evaluated previously and there was no indication that data were not missing at random1,36.
Descriptive statistics were used to summarize pre-surgery characteristics of participants. Frequencies and percentages are reported for categorical data. Medians, 25th and 75th percentiles are reported for continuous data. The Pearson chi square test and the Fisher’s Exact test for categorical variables, the Cochran-Armitage test for ordinal variables, and the Wilcoxon rank sum test for continuous variables were used to assess differences in distributions of pre-surgery characteristics of LABS-2 participants in the analysis sample to those who were excluded due to missing data. A difference in weight lost, measured as percentage of pre-surgery weight, between these groups was assessed using a linear mixed model (LMM) fit using maximum likelihood with a person-level random intercept, with a group indicator (i.e., included or excluded) as a discrete fixed effect, and time since surgery as a continuous fixed effect.
Descriptive statistics were used to summarize post-surgery weight loss and weight regain. Participants who reached weight nadir during the final year of data collection (n=119) or were pregnant at nonmissing assessments following weight nadir (n=1) were excluded from evaluation of weight regain.
LMMs with a person-level random intercept were used to model continuous weight regain measures by time since weight nadir, and test associations with the continuous fixed effect of time since weight nadir. Likewise, Poisson mixed models with robust error variance (PMM) were used to model dichotomous weight regain measures by time since weight nadir. PMM were also used to test associations with the continuous fixed effect of time since weight nadir and the six clinical outcomes. For both LMM and PMM linear and quadratic terms for time since weight nadir were considered; the quadratic term was retained if significant.
A series of PMMs were used to test and estimate associations between each weight regain measure with six outcomes: progression of diabetes, hyperlipidemia and hypertension, and decline in physical HRQoL, mental HRQoL and satisfaction. The relevant corresponding measure(s) (i.e., hemoglobin A1c, low-density lipoprotein, systolic blood pressure and diastolic blood pressure, the Physical HRQoL, the Mental HRQoL, or satisfaction rating) at time of weight nadir, and time since weight nadir were entered as continuous fixed effects. The adjusted relative risk (RR) with 95% confidence interval (CI), statistical significance (P) and Bayesian Information Criterion (BIC) from each model are reported. Because the magnitude of a RR estimate for a continuous variable varies by the unit selected for presentation, a standard unit, the sample’s median weight regain across post-nadir assessments, was applied.
Comparing the performance of weight regain measures.
The significance of P-values, RR estimates and BICs were compared to evaluate the significance and magnitude of associations and model fits, respectively, first among continuous weight regain measures, then among dichotomous measures. Given the models for each outcome were identical except for the weight regain measure, a difference of >2 in BIC is considered to be good evidence that the weight regain measure with the lower BIC fits better37. Weight regain measures were considered to perform similarly if the difference in BICs was ≤2, or the model fit was better with one measure but the magnitude of significant associations was larger with another. The significance of P-values and BICs were also compared between each continuous weight regain measure with its dichotomous counterpart for indication of a dose-response or threshold effect. Finally, for each outcome, the BIC of the best continuous measure was compared to the best dichotomous measure to determine which had the better fit. RR estimates of continuous vs. dichotomous measures were not compared since the RR for a continuous variable differs by the selected unit.
Sensitivity analysis.
While the likelihood-based approaches used to analyze the data described above produce consistent estimates of relative risks and other parameters among the analysis sample, the extent of missing data in this study was not trivial. Therefore, a sensitivity analysis using multiple imputation was conducted to address the potential impact of excluding participants from the analysis sample due to inadequate weight measurement. For the 1703 eligible participants (eFigure 1), twenty imputed datasets were generated in which missing weights, missing data components of the other clinical outcomes (e.g. HbA1c and diabetes medication use), and select missing variables measured pre-surgery (race, diabetes status and smoking status) were imputed based on full conditional specification (also known as chained equation or sequential regression approach). Satisfaction with surgery was excluded from the sensitivity analysis since this measure was missing at all assessments prior to 2010, when it was added to the study protocol. For imputing binary variables, logistic regression with logit link was used, for categorical variables logistic regression with generalized logit link was used, and for continuous variables, linear regression models were used. Time points at which the observations were not expected by design (e.g., HbA1c was not collected at the 6 year assessment) were not imputed. Details of the models used for imputation are provided in the supplemental material (eAppendix 2). For each imputed dataset, the same analyses were conducted to report weight regain by time since weight nadir and to obtain the relative risks and corresponding standard errors for each weight regain measure and clinical outcome. Of 1703 eligible participants, a minimum of 137 and maximum of 158 participants who reached weight nadir during the final year of data collection in the 20 imputed datasets were excluded. The estimates were averaged over the 20 imputations and the standard errors were adjusted using Rubin’s formula38 to produce a Wald-type CI for the relative risks after multiple imputation.
Results:
Requirements for weight nadir determination were met by 1406 of 1703 (82.6%) participants who underwent RYGB, who did not undergo a reversal or revision to another bariatric procedure or die prior to the 5 year assessment (eFigure 1). Among the full sample, the median age was 47 years (25th–75th%ile, 38–55), 80.3% were female, 84.9% were white and median BMI was 46.3 (25th–75th%ile, 42.3–51.8). Additional pre-surgery characteristics of the full analysis sample and post-nadir subsample are reported in Table 1.
Table 1.
Demographic and clinical characteristics of adults prior to Roux-en-Y Gastric Bypass in the full analysis sample and in the post-nadir subsample
| Full sample n (%) of 1406b |
Post-nadir subsamplea n (%) of 1286b |
|||
|---|---|---|---|---|
| Age, years | ||||
| Median (25th −75th %-ile) | 47 | (38, 55) | 46 | (38, 55) |
| Female | 1129 | (80.3) | 1030 | (80.1) |
| Race | (n=1394) | (n=1276) | ||
| White | 1193 | (85.6) | 1092 | (85.6) |
| Black | 150 | (10.8) | 137 | (10.7) |
| Asian | 3 | (0.2) | 3 | (0.2) |
| American Indian/Alaska Native | 14 | (1.0) | 13 | (1.0) |
| Native Hawaiian/other Pacific Islander |
3 | (0.2) | 3 | (0.2) |
| Two or more | 31 | (2.2) | 28 | (2.2) |
| Hispanic/Latino ethnicity | 61 | (4.3) | 58 | (4.5) |
| Married or “living as married” | 818 | (58.2) | 751 | (58.4) |
| Education | (n=1307) | (n=1196) | ||
| High school or less | 305 | (23.3) | 277 | (23.2) |
| Some college | 559 | (42.8) | 517 | (43.2) |
| College degree | 443 | (33.9) | 402 | (33.6) |
| Unemployed | 50 | (3.6) | 46 | (3.6) |
| Household income, US $ | (n=1271) | (n=1166) | ||
| Less than 25,000 | 243 | (19.1) | 212 | (18.2) |
| 25,000–49,999 | 358 | (28.2) | 333 | (28.6) |
| 50,000–74,999 | 300 | (23.6) | 277 | (23.8) |
| 75,000–99,999 | 194 | (15.3) | 181 | (15.5) |
| ≥100,000 | 176 | (13.8) | 163 | (14.0) |
| Current/recent smoker, No./total (%) | 183/1404 | (13.0) | 156/1284 | (12.1) |
| Weight, kg | ||||
| Median (25th −75th %-ile) | 286.0 | (256.0, 328.0) | 285.0 | (256.0, 325.0) |
| Body mass index, kg/m2 | ||||
| Median (25th −75th %-ile) | 46.3 | (42.3, 51.8) | 46.2 | (42.2, 51.7) |
| Diabetes, No./total (%) | 502/1383 | (36.3) | 438/1221 | (35.9) |
| Hyperlipidemia, No./total (%) | 525/1219 | (43.1) | 399/1035 | (38.6) |
| Hypertension, No./total (%) | 972/1383 | (70.3) | 869/1244 | (69.9) |
| SF-36 Physical component scorec | (n=1287) | (n=1183) | ||
| Median (25th −75th %-ile) | 35.1 | (26.7, 43.9) | 35.3 | (27.0, 44.2) |
| SF-36 Mental component scorec | (n=1287) | (n=1183) | ||
| Median (25th −75th %-ile) | 51.6 | (42.8, 57.2) | 52.0 | (43.4, 57.4) |
Abbreviations: SF-36, Short-Form 36-item Health Survey
Participants who reached weight nadir during the final year of data collection (n=119) or were pregnant at nonmissing assessments following weight nadir (n=1) were excluded from evaluation of weight regain by time since weight nadir.
Unless otherwise indicated.
Norm-based methods transform the SF-36 scores to a mean of 50 and standard deviations of 10 in the general U.S. population (range 0–100). Lower scores imply more disability/worse function.
Those who were excluded from both analysis samples due to missing data were younger (median age 39 vs. 47 years; P <0.001) and more likely to smoke (20.3% vs. 13.0%; P<0.001); a full comparison of pre-surgery characteristics between those included vs. excluded from the analysis sample is available in supplemental material (eTable 1). Across follow-up there was not a significant difference in weight loss between these groups (supplemental material, eTable 2).
Weight assessment in the analysis sample.
Median follow-up was 6.6 years (25th–75th%ile, 5.9–7.0) post-RYGB. Reflecting the strict inclusion criteria, weight was obtained in almost all surviving non-pregnant participants at each follow-up assessment; specifically, in 98.0% (1375/1403) at 6 months, 98.4% (1376/1399) at year-1, 98.4% (1361/1383) at year-2, 96.3% (1341/1392) at year-3, 95.7% (1327/1387) at year-4, 95.9% (1337/1394) at year 5, 94.5% (1313/1390) at year-6, and 92.6% (875/945) at year-7, excluding those due for a year-7 assessment prior to data collection ending. Of 11,811 weight measures across all assessments, 68.5% were measured per-protocol, 6.9% were determined by research or medical personnel on a non-study scale, and 24.6% were self-reported.
Weight loss.
Median time to weight nadir was 2.0 years (25th–75th%ile, 1.0–3.2). However, 8.5% (n=119) of participants’ nadir weight was reached in the final year of study follow-up (i.e., year 6 or 7); the cumulative incidence of time to maximum weight loss is provided in supplemental material (eFigure 2). Median (25th–75th%ile) maximum weight loss was 37.4% (31.6–43.3) of pre-surgery weight. Median (25th–75th%ile) BMI nadir was 28.8 (25.7–33.1). By participants’ last assessment, when median (25th–75th%ile) weight loss was 28.0% (20.6–35.6), median (25th–75th%ile) BMI was 33.2 (29.7–38.5). Additional measures of maximum weight loss and weight loss at participants’ last assessment (≥5 years) are provided in supplemental material (eTable 3).
Weight regain from nadir.
Weight regain was a quadratic function of time since weight nadir (P<.01 in all weight regain models), such that the rate of weight regain was largest in the first year after weight nadir and decreased over time (supplemental material, eTable 4; eAppendix 3). For example, weight regain, reported as median (25th–75th %ile), was 9.5% (4.7%−17.2%), 22.5% (12.9%−34.5%) and 26.8% (16.7%−41.5%) of maximum weight lost, 1, 3 and 5 years following weight nadir, respectively. Additional weight regain measures, by time since weight nadir, are reported in Table 2. At each time point, the upper quartile of participants gained at least 2.5 times more than the bottom quartile, for all continuous weight regain measures, indicating substantial variability in the magnitude of weight regain. Percentages of participants with clinically meaningful weight regain varied by measure. For example, 5 years post-nadir, 43.6% regained ≥5 BMI points, 50.2% regained ≥15% of nadir weight and 86.5% regained ≥10% of maximum weight lost (Table 2). In the sensitivity analysis among the larger RYGB sample using multiple imputation, descriptive statistics of weight regain by time since weight nadir were similar (e.g., 5 years post-nadir, 45.6% vs. 43.6% regained ≥5 BMI points, 52.3% vs 50.2% regained ≥15% of nadir weight and 87.3% vs 86.5% regained ≥10% of maximum weight lost) (supplemental material eTable 5).
Table 2.
Observed weight regain by time since post-surgery weight nadir among adults who underwent Roux-en-Y gastric bypass.
| Time since weight nadira |
||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| 1 Year | 2 Years | 3 Years | 4 Years | 5 Years | ||||||
| Samplea, No. | 1286 | 1286 | 1286 | 1286 | 1286 | |||||
| Pregnant | 21 | 10 | 14 | 14 | 10 | |||||
| Deceased | 0 | 4 | 5 | 7 | 12 | |||||
| Data collection ended | 0 | 61 | 125 | 198 | 327 | |||||
| Missing weight | 0 | 78 | 85 | 101 | 166 | |||||
| Weight measured | 1265 | 1133 | 1057 | 966 | 771 | |||||
| Timing, median (25 th −75th %-ile) | ||||||||||
| Years since initial surgery | 3.0 | (2.1–3.9) | 3.8 | (3.1–4.3) | 4.6 | (4.0–5.2) | 5.3 | (5.0–6.0) | 6.0 | (5.8–6.6) |
| Weight regain, median (25 th −75th %-ile) | ||||||||||
| Weight, kg | 4.5 | (2.3–8.2) | 8.2 | (5.0–13.2) | 10.4 | (5.9–16.3) | 11.8 | (7.3–18.6) | 12.7 | (7.3–19.5) |
| BMI, kg/m² | 1.6 | (0.8–2.8) | 3.0 | (1.7–4.6) | 3.7 | (2.2–5.8) | 4.2 | (2.6–6.7) | 4.5 | (2.7–6.8) |
| % of pre-surgery weight | 3.5 | (1.7–6.1) | 6.5 | (3.8–9.7) | 8.3 | (4.8–12.2) | 8.9 | (5.7–14.0) | 9.7 | (6.0–14.4) |
| % of nadir weight | 5.7 | (2.7–9.6) | 10.1 | (6.0–16.1) | 12.9 | (7.5–19.4) | 14.2 | (8.6–22.3) | 15.0 | (9.2–23.2) |
| % of max weight lost | 9.5 | (4.7–17.2) | 17.8 | (10.2–27.3) | 22.5 | (12.9–34.5) | 24.6 | (16.1– 39.4) |
26.8 | (16.7–41.5) |
| Clinically important weight regainb, No. (%) | ||||||||||
| ≥10 kg | 205 | (16.2) | 441 | (38.9) | 542 | (51.3) | 551 | (57.0) | 474 | (61.5) |
| ≥ 5 BMI points | 95 | (7.5) | 235 | (20.7) | 354 | (33.5) | 379 | (39.2) | 336 | (43.6) |
| ≥10% of pre-surgery weight |
100 | (7.9) | 267 | (23.6) | 392 | (37.1) | 420 | (43.5) | 376 | (48.8) |
| ≥10% of nadir weight | 297 | (23.5) | 576 | (50.8) | 676 | (64.0) | 669 | (69.3) | 559 | (72.5) |
| ≥15% of nadir weight | 125 | (9.9) | 325 | (28.7) | 422 | (39.9) | 453 | (46.9) | 387 | (50.2) |
| ≥10% of max weight lost | 604 | (47.8) | 859 | (75.8) | 880 | (83.3) | 839 | (86.9) | 667 | (86.5) |
| ≥20% of max weight lost | 235 | (18.6) | 492 | (43.4) | 599 | (56.7) | 612 | (63.4) | 519 | (67.3) |
| ≥25% of max weight lost | 148 | (11.7) | 340 | (30.0) | 465 | (44.0) | 476 | (49.3) | 427 | (55.4) |
Abbreviations: BMI, body mass index; max, maximum.
1286 participants contribute data to at least one column of the table. Participants were ineligible for inclusion at a specific time point in relation to weight nadir if pregnant, deceased or data collection had ended prior to the time point.
Progression of comorbidities and declines in HRQoL and satisfaction.
In the first year following weight nadir, 9.9%, 25.9%, and 46.2% of participants experienced progression of diabetes, hyperlipidemia and hypertension, respectively. A clinically important decline in the physical HRQoL and the mental HRQoL was experienced by 20.2% and 27.7% of participants, respectively, and 12.4% experienced a decline in satisfaction. The prevalence of clinical outcomes increased linearly with time since weight nadir (i.e., the linear term, but not quadratic term, for time was significant in all models; supplemental material, eTable 6). Five years following weight nadir, prevalences were 35.3% (diabetes), 68.4% (hyperlipidemia), 71.5% (hypertension), 42.0% (physical HRQoL), 32.8% (mental HRQoL), and 27.6% (satisfaction).
Comparing the performance of weight regain measures.
Table 3 provides the relative risk (RR), statistical significance (P) and model fit (BIC) for each measure of weight regain for progression of diabetes, and clinically important declines in the physical HRQoL, the mental HRQoL, and satisfaction, respectively. These measures for progression of hyperlipidemia and hypertension are provided in supplemental material (eTable 7). All five continuous weight regain measures were significantly related to progression of diabetes, progression of hypertension, decline in the physical HRQoL, and decline in satisfaction. For these outcomes, percentage of maximum weight lost had the highest RR point estimates per median weight regain and the lowest BICs, although model fits with kg and percentage of pre-surgery weight were similar (i.e., BICs within 2) to percentage of maximum weight lost for satisfaction. Based on RRs and BICs, percentage of maximum weight lost was also better than or similar to other weight regain measures for decline in the mental HRQoL (along with percentage of pre-surgery weight). Only two continuous weight regain measures, increase in kg and BMI, were significantly associated with progression of hyperlipidemia.
Table 3.
Associations between common measures of weight regain and concurrent declines in select health outcomes among adults who underwent Roux-en-Y gastric bypass.
| Progression of diabetesa n=689 |
Clinically important decline in physical HRQoLb n=903 |
Clinically important decline in mental HRQoLb n=903 |
Clinically important decline in satisfactionc n=272 |
|||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| RRd | (95% CI) | P value | BICe | RRd | (95% CI) | P value | BICe | RRd | (95% CI) | P value | BICe | RRd | (95% CI) | P value | BICe | |
| Weight regain, continuousf | ||||||||||||||||
| Weight in kg, per 9.1 kg | 1.40 | (1.18–1.67) | <.001 | 1543.2 | 1.21 | (1.12–1.31) | <.001 | 3541.9 | 1.08 | (1.00–1.17) | 0.050 | 3739.2 | 1.41 | (1.14–1.74) | 0.002 | 584.4 |
| BMI, per 3.2 kg/m² | 1.41 | (1.19–1.68) | <.001 | 1543.2 | 1.21 | (1.12–1.31) | <.001 | 3542.8 | 1.09 | (1.01–1.18) | 0.03 | 3738.5 | 1.44 | (1.16–1.78) | <.001 | 583.3 |
| % pre-surgery weight, per 6.9% | 1.45 | (1.22–1.73) | <.001 | 1541.9 | 1.22 | (1.12–1.32) | <.001 | 3544.1 | 1.11 | (1.02–1.20) | 0.02 | 3737.7 | 1.45 | (1.12–1.87) | 0.005 | 585.5 |
| % nadir weight, per 11.0% | 1.36 | (1.15–1.60) | <.001 | 1545.5 | 1.16 | (1.07–1.25) | <.001 | 3551.2 | 1.08 | (1.00–1.17) | 0.04 | 3738.8 | 1.35 | (1.07–1.70) | 0.011 | 586.9 |
| % max weight lost, per 18.9% | 1.51 | (1.27–1.78) | <.001 | 1535.8 | 1.28 | (1.18–1.38) | <.001 | 3532.8 | 1.11 | (1.02–1.20) | 0.02 | 3737.6 | 1.56 | (1.19–2.04) | 0.001 | 582.6 |
| Clinically important weight regaing | ||||||||||||||||
| ≥10 kg | 1.56 | (1.15–2.11) | 0.004 | 1549.4 | 1.39 | (1.19–1.63) | <.001 | 3548.0 | 1.12 | (0.96–1.30) | 0.15 | 3741.9 | 1.91 | (1.21–3.01) | 0.006 | 584.3 |
| ≥ 5 BMI points | 1.66 | (1.19–2.30) | 0.002 | 1548.6 | 1.40 | (1.18–1.65) | <.001 | 3550.2 | 1.23 | (1.04–1.45) | 0.014 | 3737.3 | 1.84 | (1.13–3.01) | 0.02 | 586.9 |
| ≥10% of pre-surgery weight | 1.63 | (1.19–2.25) | 0.003 | 1549.5 | 1.45 | (1.23–1.71) | <.001 | 3544.4 | 1.17 | (1.00–1.38) | 0.054 | 3739.6 | 1.73 | (1.05–2.84) | 0.03 | 587.1 |
| ≥10% of nadir weight | 1.37 | (1.02–1.84) | 0.03 | 1553.2 | 1.36 | (1.15–1.59) | <.001 | 3552.0 | 1.11 | (0.95–1.29) | 0.19 | 3741.0 | 1.82 | (1.14–2.91) | 0.013 | 586.5 |
| ≥15% of nadir weight | 1.41 | (1.03–1.92) | 0.03 | 1554.0 | 1.36 | (1.16–1.59) | <.001 | 3550.3 | 1.19 | (1.01–1.39) | 0.03 | 3739.6 | 1.51 | (0.93–2.46) | 0.09 | 589.8 |
| ≥10% of max weight lost | 1.42 | (1.02–1.98) | 0.04 | 1554.0 | 1.42 | (1.17–1.72) | <.001 | 3550.4 | 1.09 | (0.92–1.29) | 0.31 | 3742.2 | 2.54 | (1.34–4.82) | 0.005 | 582.9 |
| ≥20% of max weight lost | 1.64 | (1.22–2.19) | <.001 | 1547.2 | 1.55 | (1.33–1.82) | <.001 | 3535.5 | 1.23 | (1.06–1.43) | 0.008 | 3735.8 | 2.33 | (1.50–3.63) | <.001 | 579.4 |
| ≥25% of max weight lost | 1.64 | (1.21–2.20) | 0.001 | 1547.3 | 1.43 | (1.22–1.68) | <.001 | 3546.5 | 1.16 | (0.99–1.36) | 0.06 | 3739.9 | 2.22 | (1.44–3.42) | <.001 | 575.9 |
Abbreviations: BIC=Bayesian Information Criterion; HRQoL=health-related quality of life; max=maximum; RR= Relative Risk.
A change from not taking diabetes medication to taking diabetes medication, change from not taking insulin to taking insulin, or an increase of glycated hemoglobin (HbA1c) by at least 0.5% plus a post-weight nadir value of 5.7% or greater.
A decrease of ≥5 points on the Short-Form 36-item Physical Component Summary (PCS) or Mental Component Summary (MCS) score, respectively.
An increase of ≥1 on a 7 point scale to a rating of at least 3 (i.e., “somewhat satisfied” – “very dissatisfied”).
Adjusted for pre-surgery factors related to missing data (i.e., site, age and smoking status), MCS, PCS or HbA1c at time of weight loss nadir, respectively, and time since weight nadir as a continuous fixed effect (only linear term retained). Sample size for each model is based on n=1286 minus those with missing outcome data.
A difference of more than 2 is good evidence that the model with the lower BIC has better fit.
RR is reported per median weight regain across post-nadir time points.
Of dichotomous weight regain measures ≥20% of maximum weight lost had the highest RR point estimate and best model fit with decline in the physical HRQoL. It also performed better or similarly to other dichotomous weight regain measures, as determined by RR point estimates and BICs, for progression of diabetes (along with ≥5 BMI points and ≥25% of maximum weight loss), decline in the mental HRQoL (along with ≥5 BMI points), and decline in satisfaction (along with ≥10% of maximum weight loss and ≥25% of maximum weight loss). Additionally, ≥20% of maximum weight lost was the second best measure for hyperlipidemia (after ≥10 kg) and hypertension (after ≥10% of maximum weight lost).
In the sensitivity analysis using multiple imputation, RR point estimates were the same or higher than values from the primary analysis and the 95% CIs from the sensitivity versus primary analysis overlapped (supplemental material, eTable 8). Additionally, with the exception of progression of hyperlipidemia, the ordering of regain measures by RR point estimates in the sensitivity analysis was similar to the primary analysis; compared to other regain measures, % maximum weight loss, for continuous measures, and ≥20% maximum weight loss, for dichotomous measures, had the highest values (eTable 8).
In general, dichotomous measures of weight regain did not perform as well as their continuous counterparts, as indicated by their lack of statistical significance with outcomes (e.g., ≥5 BMI points with progression of hyperlipidemia) (Table 3) or inferior model fit with outcomes (e.g., all eight dichotomous measures with progression of diabetes) (Table 3; eTable 7). Of 48 comparisons, a dichotomous weight regain measure only resulted in a better model fit than its continuous counterpart in three comparisons (≥20% of maximum weight lost with hyperlipidemia; ≥20% and ≥25% of maximum weight lost with satisfaction), whereas continuous weight regain measures resulted in better model fits in 30 comparisons (supplemental material, eTable 9). When the best continuous measure was compared to the best dichotomous measure for each outcome (percentage of maximum weight lost vs. a dichotomous counterpart for all outcomes except hyperlipidemia), the continuous weight regain measure resulted in better model fits for 3 outcomes (diabetes, hypertension and PCS), a similar model fit for 2 outcomes (hyperlipidemia and MCS) and a worse model fit for 1 outcome (satisfaction) (supplemental material, eTable 9).
Discussion
Among a large geographically diverse cohort of adults who underwent RYGB, a comparison of the relative performance of common weight regain measures in relation to several important outcomes of bariatric surgery demonstrated that weight regain measured as percentage of maximum weight lost, generally performed better than the alternatives examined. These findings may inform standardizing measurement of weight regain in studies of bariatric surgery.
All continuous measures and most dichotomous measures of weight regain were associated with progression of diabetes and hypertension, and declines in physical HRQoL and satisfaction with surgery. Some weight regain measures were also weakly associated with progression of hyperlipidemia and decline in mental HRQoL. While the relative performance of weight regain measures varied across clinical outcome, in general, percentage of maximum weight lost performed best (i.e., had the strongest associations and best model fits with clinical outcomes) of all continuous measures examined, and ≥20% of maximum weight lost performed best among dichotomous measures. For diabetes, hypertension and physical HRQoL, the model with percentage of maximum weight lost fit better than the model with the best dichotomous measure, likely reflecting that information is lost when applying a threshold compared to a continuous measure that allows dose-repose assessment across the entire distribution. However, the best dichotomous measure had a similar model fit to the best continuous measure in relation to hyperlipidemia and mental HRQoL, and had a better model fit for satisfaction, indicating there may be a threshold effect for at least some outcomes.
While a few studies have reported weight regain using multiple measures22,27,39, only one prior study has compared weight regain measures in a sample of adults who have undergone bariatric surgery. Lauti et al. reported weight regain in 55 patients five years following gastric sleeve using a variety of measures, and evaluated associations between weight regain measures with satisfaction with surgery, and the Bariatric Analysis Reporting Outcome System (BAROS) score, which incorporates weight loss, changes in medical conditions, HRQoL and reoperations7. Two definitions of clinically meaningful weight regain, >5 BMI points and >25% of excess weight loss, but not a third, >10 kg, were associated with a lower odds of satisfaction with surgery7. In addition, all three dichotomous weight regain measures were inversely associated with the BAROS score. However, several continuous measures of weight regain, i.e., BMI, percentage of pre-surgery weight, percentage excess BMI, and percentage excess weight loss, explained more of the variance in the BAROS score than the dichotomous measures, suggesting a dose-response relationship with weight regain. While Lauti et al7 did not evaluate percentage of maximum weight lost, Jirapinyo et al.6 recently reported this measure was negatively and linearly associated with the Bariatric Quality of Life (BQL) Index score (β =−0.56; p=0.001). Thus, the current study’s findings, as well as those reported by Lauti et al.7 and Jirapinyo et al.6, support a dose-response relationship, i.e., the less weight regained the better, for at least some clinical outcomes (e.g., diabetes, hypertension and physical HRQoL).
The median timing of weight nadir in the current study (i.e., 2 years) was similar to the mean time reported in three other RYGB samples22,26,39. However, there was substantial variation in the timing of weight nadir within the LABS-2 cohort, with approximately 20% continuing to lose weight following their four year assessment, possibly due to complications (e.g., ulcer) or other problems. Likewise, there was substantial variability in the magnitude of weight regain, suggesting that patient-level factors play an important role in weight loss maintenance. Given the detrimental effects of weight regain on progression of comorbidities, and decline in HRQoL and satisfaction, coupled with the fact that the rate of weight regain is highest in the first year following weight nadir, early detection and treatment of weight regain may be important for maximizing the long-term benefits of RYGB. Future work is needed to develop tools for patients and providers to more easily recognize and understand the impact of weight regain, such as eAppendix 3.
Unlike previous studies in the bariatric literature, which generally report weight regain at one time point since surgery, this study reported weight regain over time as a function of time since weight nadir. This was helpful for 1) highlighting the substantial impact that timing has on estimates of weight regain, and 2) revealing that the rate of regain in highest in the first year following nadir but continues at a diminishing rate over time. However, the difference in time scales across studies (i.e., time since surgery vs. time since weight nadir) makes comparison of weight regain in the LABS-2 RYGB sample to previous studies difficult. Given LABS-2 had standard data collection with relatively frequent weight measurement and high retention among a large, geographically diverse sample, the weight regain statistics from this study may be more generalizable to clinical practice in the U.S. than prior reports7,10,20–27,39. Of note, while 17% of the RYGB cohort was excluded from the primary analysis due to missing weight data, a sensitivity analysis using multiple imputation confirmed that the primary results are representative of the larger RYGB cohort.
Limitations.
This study has several limitations. First, this study did not include the gastric sleeve procedure, which was relatively uncommon at the time study participants underwent surgery (2006–2009), but is now the most common procedure in the United States40. However, RYGB remains a common primary procedure, as well as a common revisional procedure for weight regain after gastric sleeve. A second limitation is that weight nadir could have occurred between research assessments. However, utilizing research assessments with standardized weight measurement at 6 months and annually post-surgery to estimate nadir weight is likely superior to patient recall years after weight nadir, or clinical records, which often have large time lapses after the first postoperative year. Third, comorbidities were not assessed at the 6 year minimal research assessment, which limited the sample size for analyses evaluating progression of diabetes, hyperlipidemia and hypertension during weight regain. Likewise, patient satisfaction with surgery was not assessed in the first few years of data collection, resulting in a smaller sample size for this outcome. Fourth, this study’s evaluation of weight regain measures did not include interpretation and ease of use by patients and clinicians.
Conclusion.
Among a large cohort of adults who underwent RYGB, weight regain quantified as percentage of maximum weight lost performed better for association with most clinical outcomes than the alternatives examined. These findings may inform standardizing measurement of weight regain in studies of bariatric surgery.
Supplementary Material
KEY POINTS.
Question:
Are there approaches to measuring weight regain following bariatric surgery that are more predictive of clinical outcomes than others?
Findings:
In this prospective cohort study of 1406 adults who underwent Roux-en-Y gastric bypass with median follow-up of 6.6 years, weight regain measured in % maximum weight lost (versus in kilograms, body mass index, % pre-surgery weight or % nadir weight), had the strongest associations with, and best model fits for declines in mental health, satisfaction with surgery, and most physical health measures.
Meaning:
These findings may inform standardizing measurement of weight regain in studies of bariatric surgery.
Acknowledgements
Access to Data: Dr. King and Ms. Hinerman had full access to all of the data and take responsibility for the integrity of the data and the accuracy of the data analysis.
This clinical study was a cooperative agreement funded by the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK). Grant numbers: DCC -U01 DK066557; Columbia - U01-DK66667 (in collaboration with Cornell University Medical Center CTRC, Grant UL1-RR024996); University of Washington - U01-DK66568 (in collaboration with CTRC, Grant M01RR-00037); Neuropsychiatric Research Institute - U01-DK66471; East Carolina University – U01-DK66526; University of Pittsburgh Medical Center – U01-DK66585 (in collaboration with CTRC, Grant UL1-RR024153); Oregon Health & Science University – U01-DK66555.
Authors have no support from any organization for the submitted work; no financial relationships with any organizations that might have an interest in the submitted work in the previous three years; and no other relationships or activities that could appear to have influenced the submitted work other than the following: Dr. Courcoulas reports grants from Covidien/Ethicon J&J, outside the submitted work.
Footnotes
Conflicts of Interest
All authors will complete the ICMJE Form for Disclosure of Potential Conflicts of Interest.
Conflicts of Interest Disclosures: Drs. King and Belle, and Ms. Hinerman have nothing to disclose
References
- 1.Courcoulas AP, King WC, Belle SH, et al. Seven-Year Weight Trajectories and Health Outcomes in the Longitudinal Assessment of Bariatric Surgery (LABS) Study. JAMA Surg 2018;153(5):427–434. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Puzziferri N, Roshek TB 3rd, Mayo HG, Gallagher R, Belle SH, Livingston EH. Long-term follow-up after bariatric surgery: a systematic review. JAMA 2014;312(9):934–942. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Sjostrom L Review of the key results from the Swedish Obese Subjects (SOS) trial - a prospective controlled intervention study of bariatric surgery. J Intern Med 2013;273(3):219–234. [DOI] [PubMed] [Google Scholar]
- 4.Brethauer SA, Aminian A, Romero-Talamas H, et al. Can diabetes be surgically cured? Long-term metabolic effects of bariatric surgery in obese patients with type 2 diabetes mellitus. Ann Surg 2013;258(4):628–636. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Sjostrom L, Lindroos A-K, Peltonen M, et al. Lifestyle, Diabetes, and Cardiovascular Risk Factors 10 Years after Bariatric Surgery. N Engl J Med 2004;351(26):2683–2693. [DOI] [PubMed] [Google Scholar]
- 6.Jirapinyo P, Abu Dayyeh BK, Thompson CC. Weight regain after Roux-en-Y gastric bypass has a large negative impact on the Bariatric Quality of Life Index. BMJ Open Gastroenterol 2017;4(1):e000153. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Lauti M, Lemanu D, Zeng ISL, Su’a B, Hill AG, MacCormick AD. Definition determines weight regain outcomes after sleeve gastrectomy. Surg Obes Relat Dis 2017;13(7): 1123–1129. [DOI] [PubMed] [Google Scholar]
- 8.Karlsson J, Taft C, Ryden A, Sjostrom L, Sullivan M. Ten-year trends in health-related quality of life after surgical and conventional treatment for severe obesity: the SOS intervention study. Int J Obes (Lond) 2007;31(8):1248–1261. [DOI] [PubMed] [Google Scholar]
- 9.Sheppard CE, Lester EL, Chuck AW, Birch DW, Karmali S, de Gara CJ. The economic impact of weight regain. Gastroenterol Res Pract 2013;2013:379564. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Karmali S, Brar B, Shi X, Sharma AM, de Gara C, Birch DW. Weight recidivism post-bariatric surgery: a systematic review. Obes Surg 2013;23(11):1922–1933. [DOI] [PubMed] [Google Scholar]
- 11.Lauti M, Kularatna M, Hill AG, MacCormick AD. Weight Regain Following Sleeve Gastrectomy-a Systematic Review. Obes Surg 2016;26(6):1326–1334. [DOI] [PubMed] [Google Scholar]
- 12.Nedelcu M, Khwaja HA, Rogula TG. Weight regain after bariatric surgery-how should it be defined? Surg Obes Relat Dis 2016;12(5):1129–1130. [DOI] [PubMed] [Google Scholar]
- 13.Belle SH, Berk PD, Courcoulas AP, et al. Safety and efficacy of bariatric surgery: Longitudinal Assessment of Bariatric Surgery. Surg Obes Relat Dis 2007;3(2):116–126. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.NIDDK: Longitudinal Assessment of Bariatric Surgery (LABS) https://www.niddkrepository.org/studies/labs/. Accessed September 22, 2017.
- 15.Christian NJ, King WC, Yanovski SZ, Courcoulas AP, Belle SH. Validity of self-reported weights following bariatric surgery. JAMA 2013;310(22):2454–2456. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Brethauer SA, Kim J, el Chaar M, et al. Standardized outcomes reporting in metabolic and bariatric surgery. Surg Obes Relat Dis 2015;11(3):489–506. [DOI] [PubMed] [Google Scholar]
- 17.Felsenreich DM, Langer FB, Kefurt R, et al. Weight loss, weight regain, and conversions to Roux-en-Y gastric bypass: 10-year results of laparoscopic sleeve gastrectomy. Surg Obes Relat Dis 2016;12(9):1655–1662. [DOI] [PubMed] [Google Scholar]
- 18.Christou NV, Look D, Maclean LD. Weight gain after short- and long-limb gastric bypass in patients followed for longer than 10 years. Ann Surg 2006;244(5):734–740. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Field AE, Wing RR, Manson JE, Spiegelman DL, Willett WC. Relationship of a large weight loss to long-term weight change among young and middle-aged US women. Int J Obes Relat Metab Disord 2001;25(8):1113–1121. [DOI] [PubMed] [Google Scholar]
- 20.da Silva FB, Gomes DL, de Carvalho KM. Poor diet quality and postoperative time are independent risk factors for weight regain after Roux-en-Y gastric bypass. Nutrition 2016;32(11–12):1250–1253. [DOI] [PubMed] [Google Scholar]
- 21.Monaco-Ferreira DV, Leandro-Merhi VA. Weight Regain 10 Years After Roux-en-Y Gastric Bypass. Obes Surg 2017;27(5):1137–1144. [DOI] [PubMed] [Google Scholar]
- 22.de Hollanda A, Ruiz T, Jimenez A, Flores L, Lacy A, Vidal J. Patterns of Weight Loss Response Following Gastric Bypass and Sleeve Gastrectomy. Obes Surg 2015;25(7):1177–1183. [DOI] [PubMed] [Google Scholar]
- 23.Livhits M, Mercado C, Yermilov I, et al. Patient behaviors associated with weight regain after laparoscopic gastric bypass. Obes Res Clin Pract 2011;5(3):e169–266. [DOI] [PubMed] [Google Scholar]
- 24.Yanos BR, Saules KK, Schuh LM, Sogg S. Predictors of Lowest Weight and Long-Term Weight Regain Among Roux-en-Y Gastric Bypass Patients. Obes Surg 2015;25(8):1364–1370. [DOI] [PubMed] [Google Scholar]
- 25.Abu Dayyeh BK, Lautz DB, Thompson CC. Gastrojejunal stoma diameter predicts weight regain after Roux-en-Y gastric bypass. Clin Gastroenterol Hepatol 2011;9(3):228–233. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Bastos EC, Barbosa EM, Soriano GM, dos Santos EA, Vasconcelos SM. Determinants of weight regain after bariatric surgery. Arq Bras Cir Dig 2013;26 Suppl 1:26–32. [DOI] [PubMed] [Google Scholar]
- 27.Roslin M, Damani T, Oren J, Andrews R, Yatco E, Shah P. Abnormal glucose tolerance testing following gastric bypass demonstrates reactive hypoglycemia. Surg Endosc 2011;25(6):1926–1932. [DOI] [PubMed] [Google Scholar]
- 28.Cooper TC, Simmons EB, Webb K, Burns JL, Kushner RF. Trends in Weight Regain Following Roux-en-Y Gastric Bypass (RYGB) Bariatric Surgery. Obes Surg 2015;25(8):1474–1481. [DOI] [PubMed] [Google Scholar]
- 29.Kolotkin RL, Andersen JR. A systematic review of reviews: exploring the relationship between obesity, weight loss and health-related quality of life. Clin Obes 2017;7(5):273–289. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Ware JE J, Kosinski M, Keller SK. SF-36® physical and mental health summary scales: a user’s manual. Boston, MA: The Health Institute; 1994. [Google Scholar]
- 31.Lenters-Westra E, Schindhelm RK, Bilo HJ, Groenier KH, Slingerland RJ. Differences in interpretation of haemoglobin A1c values among diabetes care professionals. Neth J Med 2014;72(9):462–466. [PubMed] [Google Scholar]
- 32.Lorenzo C, Wagenknecht LE, Hanley AJ, Rewers MJ, Karter AJ, Haffner SM. A1C between 5.7 and 6.4% as a marker for identifying pre-diabetes, insulin sensitivity and secretion, and cardiovascular risk factors: the Insulin Resistance Atherosclerosis Study (IRAS). Diabetes Care 2010;33(9):2104–2109. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Executive Summary of The Third Report of The National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, And Treatment of High Blood Cholesterol In Adults (Adult Treatment Panel III). JAMA 2001;285(19):2486–2497. [DOI] [PubMed] [Google Scholar]
- 34.Whelton PK, Carey RM, Aronow WS, et al. 2017 ACC/AHA/AAPA/ABC/ACPM/AGS/APhA/ASH/ASPC/NMA/PCNA Guideline for the Prevention, Detection, Evaluation, and Management of High Blood Pressure in Adults: Executive Summary: A Report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines. Hypertension 2018;71(6):1269–1324. [DOI] [PubMed] [Google Scholar]
- 35.Norman GR, Sloan JA, Wyrwich KW. Interpretation of changes in health-related quality of life: the remarkable universality of half a standard deviation. Med Care 2003;41(5):582–592. [DOI] [PubMed] [Google Scholar]
- 36.King WC, Chen JY, Belle SH, et al. Change in Pain and Physical Function Following Bariatric Surgery for Severe Obesity. JAMA 2016;315(13):1362–1371. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Kass RER, Adrian E. “Bayes Factors”. J Am Stat Assoc 1995;90 (430):773–795. [Google Scholar]
- 38.Rubin DB. Multiple Imputation for Nonresponse in Surveys New York: John Wiley and Sons; 2004. [Google Scholar]
- 39.Kofman MD, Lent MR, Swencionis C. Maladaptive eating patterns, quality of life, and weight outcomes following gastric bypass: results of an Internet survey. Obesity (Silver Spring) 2010;18(10):1938–1943. [DOI] [PubMed] [Google Scholar]
- 40.Ponce J, Nguyen NT, Hutter M, Sudan R, Morton JM. American Society for Metabolic and Bariatric Surgery estimation of bariatric surgery procedures in the United States, 2011–2014. Surg Obes Relat Dis 2015;11(6):1199–1200. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.