Abstract
Objective:
The main goal of this analysis was to determine if type 2 diabetes and HbA1c predict all-cause 30-day hospital readmission post-metabolic and bariatric surgery (MBS). It was hypothesized that a diagnosis of type 2 diabetes or high HbA1c values would predict all-cause hospital readmission rates post-MBS.
Methods:
A retrospective analysis from the 2015–2018 Metabolic and Bariatric Surgery Accreditation and Quality Improvement Program (MBSAQIP) cohort was completed (n= 744,776); 30,972 participants were readmitted during the 30-day post-MBS.
Results:
Mean age of the MBSAQIP sample was 45.1 (±11.5) years, majority female (80.7%) and non-Hispanic white (59.4%). All-cause hospital readmission rate was 4.2% and increased by 10% in those with uncontrolled type 2 diabetes (HbA1c >7.5% [>58 mmol/mol]); after adjustment, diabetes was not associated with increased re-admission. Uncontrolled type 2 diabetes, type 2 diabetes, and prediabetes resulted in less weight loss 30-day post-MBS.
Conclusion:
Our results based on a national MBS cohort showed that uncontrolled type 2 diabetes is associated with a greater likelihood of all-cause hospital readmission and reduced weight loss 30-day post-MBS. Both, type 2 diabetes and prediabetes were also associated with decreased weight loss 30-day post-MBS. Our findings highlight the need to classify and optimize glycemic control prior to MBS.
Keywords: MBS, diabetes, HbA1c, readmission, fasting glucose
Introduction
The prevalence of obesity continues to increase, particularly within the highest body mass index (BMI) categories (1). Metabolic and bariatric surgery (MBS) is a safe and effective treatment option for severe obesity, obese type 2 diabetes, and other cardiometabolic risk factors (2, 3). Severe obesity and type 2 diabetes, which are both associated with increased morbidity, mortality and decreased quality of life, are the most common indications for MBS. MBS induces type 2 diabetes remission in the majority of patients at two years, and a third of patients at 15 years, which is associated with fewer type 2 diabetes-related complications than usual care (4, 5).
It is estimated that one out of every four MBS patients will be re-admitted to the hospital within two years post-MBS (6). Previous studies have shown that 2-year hospital readmission rates vary by ethnicity. Non-Hispanic blacks (NHB) (OR: 1.56, 95 % CI:1.34–1.81) and Hispanics (OR: 1.29, 95 % CI:1.05–1.58) have the highest risk of readmissions post-MBS in comparison to non-Hispanic whites (NHW) (7). Type 2 diabetes has been shown to be a significant predictor of readmission, serious adverse events, and mortality post-MBS (7–13). A recent study demonstrated that patients with type 2 diabetes who had undergone MBS had up to 15-fold higher readmission rate compared to those without diabetes (14). Perioperative hyperglycemia has also shown to be independently associated with higher risk of post-operative infections, hospital readmission and greater length of stay following MBS (15). Understanding the impact of pre-MBS glycemic control on the risk of post-MBS adverse events would help inform best practice guidelines (16).
Patients with type 2 diabetes also experience suboptimal weight loss compared to patients without diabetes following MBS (17–21). Moreover, poor glycemic control is also associated with less weight loss post-MBS (22). In single-site studies, people with HbA1c values >8.0% (>64 mmol/mol) do not appear to be as successful with weight loss post-MBS (22) when compared with those who have better glycemic control (23). The main goal of this study was to evaluate if a diagnosis of type 2 diabetes and/or high pre-MBS HbA1c levels are predictors of all-cause 30-day hospital readmission rates in a nationally representative sample of MBS centers. A secondary goal was to examine ethnic group differences in all-cause post-MBS readmission rates and outcomes. It was hypothesized that type 2 diabetes or higher HbA1c values would predict all-cause hospital readmission post-MBS. Variations by ethnic group were also explored.
Methods
Study Design.
A prospective (30-day) cohort design was used for the MBSAQIP data collection (24). Data was used in accordance with the terms agreed upon receipt. A retrospective analysis was conducted. The UT Health institutional review board considers a retrospective analysis of public, anonymized data the MBSAQIP dataset exempt from review. STROBE cohort checklist was used in the creation of this manuscript.
Participants.
The Participant Use Data File (PUF) contains 173 HIPAA-compliant variables on 204,837 cases submitted from 854 centers in 2018; 200,374 from 832 centers in 2017; 186,772 from 791 centers in 2016; and 168,903 from 742 centers in 2015. Inclusion criteria were limited to patients aged above 19 and less than 70 years (19 < age ≤70). Cases with missing age (n= 233) were excluded from our analysis. The final analytical sample utilized for the present study included 744,776 adult patients (Figure S1) from which 30,972 were readmitted post-MBS.
Variables.
The primary dependent variable of this analysis was all-cause 30-day hospital readmission after MBS, while the primary independent variables were the presence of type 2 diabetes prior to surgery and pre-MBS HbA1c levels. Four HbA1c groups were created using the following cut points: (1) HbA1c <5.7% (<39 mmol/mol); (2) HbA1c 5.7–6.4% (39–46 mmol/mol); (3) HbA1c 6.5–7.5% (48–58 mmol/mol); and (4) HbA1c >7.5% (>58 mmol/mol). Patients classified as having type 2 diabetes included both patients who reported to their surgeon pre-operatively that they had been diagnosed with diabetes and those with an HbA1c ≥6.5% (≥48 mmol/mol) in order to capture individuals with undiagnosed disease. An HbA1c value between 5.7–6.4% was categorized as prediabetes, and an HbA1c value >7.5% (>58 mmol/mol) was categorized as uncontrolled type 2 diabetes. HbA1c was measured 25±23 days prior to MBS. Selected covariates included age, ethnicity, pre-operative BMI, HbA1c measurement timepoint (in days), and other comorbidities including hypertension, hyperlipidemia, gastroesophageal reflux disease (GERD), and obstructive sleep apnea (OSA). A secondary outcome was percentage of total body weight loss (%TBWL), which was calculated by using the following formula:
Data Source.
The American College of Surgeons (ACS) and the American Society for Metabolic and Bariatric Surgery (ASMBS) merged their programs in 2012 to form the Metabolic and Bariatric Surgery Accreditation and Quality Improvement Program (MBSAQIP)(25). The merged 2015–2018 MBSAQIP PUF were used for this analysis; they include MBS patients who received their clinical care at an accredited center. Perioperative care is standardized across centers ensuring reliable data (26).
Sample Size.
The MBSAQIP database PUF contains a total sample size of 760,886 participants from years 2015–2018, inclusive. After excluding 857 due to missing data, a sample of 760,029 participants was left. Additional participants were excluded due to age <19 years (n= 3566), age >70 (n=11454), or missing age (n=233). Final analyses were performed on 744,776 participants.
Statistical Analysis.
Descriptive analysis was performed for baseline characteristics including age, sex, BMI (highest and closest to MBS), 30-day weight loss, procedure type, HbA1c values, and relevant comorbidities. A comparison of patient characteristics between readmitted and non-readmitted patients, as well as ethnicity, was performed using t-tests or one-way ANOVA for continuous variables and chi-square tests for categorical variables. All-cause readmissions were analyzed only if they occurred in the indexed hospitals. All-causes of 30-day readmissions were stratified by ethnicity and included abdominal pain, bile reflux gastritis, chest pain, gallstone disease, nausea, vomiting, etc. as well as more severe complications such as anastomotic ulcer, bleeding, cerebrovascular accident, cardiac event, gastrointestinal perforation, wound infection/evisceration, among others. Crude odds ratios and adjusted odds ratios were calculated for all-cause 30-day hospital readmissions with type 2 diabetes diagnosis (control, prediabetes, and type 2 diabetes) and HbA1c levels (HbA1c ≤ 7.5% [≤ 58 mmol/mol] and >7.5% [>58 mmol/mol]) as separate independent variables. Adjusted logistic regression controlled for age, ethnicity, pre-operative BMI, HbA1c measurement timepoint, and other comorbidities including hypertension, hyperlipidemia, GERD, and OSA. %TBWL between four ethnicities was compared by Analysis of Variance (ANOVA) test. One-way ANOVA and two-sample t-tests were performed to detect any differences of %TBWL within ethnicities by type 2 diabetes status and HbA1c groups dataset respectively. The assumptions of ANOVA were not violated in our analysis. If the results of an ANOVA tests were significant, pairwise t-tests with Bonferroni correction were conducted to find the differences between any two groups.
Sensitivity analysis was conducted to detect the impact of missing data on the primary outcome. We first created a logistic regression model controlling for all covariates listed above with HbA1c (missing vs. non-missing) as the primary independent variable, and also performed a chi-square test to directly compare all-cause readmission rates between patients with and without HbA1c data. All statistical analyses were performed using SAS v9.4 (SAS Institute, Cary, NC). Type one error was maintained at 5%.
Results
The analytical population (N= 744,776) was classified into three groups as follows: 1) those without evidence of type 2 diabetes or pre-diabetes (controls; n= 521,548), from which 13% (n= 73,378) had an HbA1c measurement available for analysis. Of this group, 69% (n= 50,565) had normal HbA1c values (<5.7% [<39 mmol/mol]); 2) from the total type 2 diabetes group (n= 192,251) 93.8% reported a pre-MBS diabetes diagnosis and 6.2% were undiagnosed but with an HbA1C ≥6.5% [48 mmol/mol]); from this total diabetes group 29.2% (n= 56,170) had a pre-MBS HbA1c measurement with 41.4% of them (n= 23,265) labeled as uncontrolled diabetes (i.e. HbA1c > 7.5% (>58 mmol/mol); and 3) those without type 2 diabetes diagnosis but with an HbA1c in the pre-diabetes range (prediabetes, N= 30,977). Of note, 18.5% (n= 137,708) of the total sample had an HbA1c measurement available for analysis. The mean age in the control, prediabetes and diabetes groups was 43.5 (SD 11.4), 46.0 (SD 11.1), and 49.1 (SD 11.0), respectively ( p<0.001), the mean pre-MBS BMI was 44.2 (SD 8.4), 45.5 (SD 8.1), and 45.1 (SD 8.4) kg/m2, respectively (p<0.001), and the percentage of females was 83.2% (n= 434,156), 73.8% (n= 141,966), and 79.8% (n= 24,716) respectively (p<0.001). Laparoscopic sleeve gastrectomy and laparoscopic Roux-en-Y gastric bypass were performed in 85% of the patients. More than half (59%) of the participants were NHW, 16% were NHB, 9% were Hispanic, and 15% from other ethnicities. Mean HbA1c was for the entire cohort was 6.6 ± 2.1% (49 mmol/mol) (Table 1).
Table 1.
Pre- and Post-operative characteristics of selected population (MBSAQIP, 2015–2018).
| Variables | Total (n= 744,776) | Readmitted (n = 30,978) | Non-Readmitted (n = 713,798) | P-valuee | NHW (n= 442,271) | NHB (n= 120,893) | Hispanic (n= 67,430) | Others (n= 114,092) | P-valuec |
|---|---|---|---|---|---|---|---|---|---|
| Agea,y,mean (SD) | 45.1 (11.5) | 45.8 (11.7) | 45.0 (11.5) | <0.001 | 46.3 (11.6) | 43.6 (10.7) | 41.6 (10.9) | 43.8 (11.5) | <0.001 |
|
| |||||||||
| Sex, f, n (%) | 600,838 (80.7) | 25,539 (82.4) | 575,299 (80.6) | <0.001 | 349,364 (78.9) | 105,669 (87.4) | 54,280 (80.5) | 91,525 (80.22) | <0.001 |
|
| |||||||||
| Pre-op BMI, mean (SD) | 44.5 (8.4) | 44.4 (9.6) | 44.5 (8.3) | 0.209 | 44.1 (8.3) | 46.3 (8.9) | 44.1 (8.1) | 44.2 (8.2) | <0.001 |
|
| |||||||||
| Highest BMI, mean (SD) | 46.5 (8.9) | 46.6 (10.0) | 46.4 (8.8) | 0.014 | 46.2 (8.8) | 48.1 (9.3) | 45.8 (8.5) | 45.9 (8.6) | <0.001 |
|
| |||||||||
| 30-day TBWL, % (SE) | 15.1 (0.05) | 12.8 (0.26) | 15.1 (0.05) | <0.001 | 14.2 (0.06) | 13.2 (0.13) | 11.9 (0.17) | 22.2 (0.16) | <0.001 |
|
| |||||||||
| Procedure, n (%) | <0.001 | <0.001 | |||||||
| LSG | 457,698 (61.5) | 13,850 (44.7) | 443,848 (62.2) | 262,102 (59.3) | 81,605 (67.5) | 43,386 (64.3) | 70,605 (61.9) | ||
| LRYGB | 177,344 (23.8) | 10,885 (35.1) | 166,459 (23.3) | 110,020 (24.9) | 23,867 (19.74) | 16,218 (24.1) | 27,239 (23.9) | ||
| Other | 109,734 (14.7) | 6,243 (20.2) | 103,491 (14.5) | 70,149 (15.86) | 15,511 (12.8) | 7,826 (11.6) | 16,248 (14.2) | ||
|
| |||||||||
| Comorbidity,n(%) | |||||||||
| Hypertension | 429,505 (57.7) | 19,229 (62.1) | 410,276 (57.5) | <0.001 | 258,648 (58.5) | 75,801 (62.7) | 34,276 (50.8) | 60,780 (53.3) | <0.001 |
| Hyperlipidemia | 168,089 (22.6) | 8,024 (25.9) | 160,065 (22.4) | <0.001 | 110,294 (24.9) | 22,951 (19.0) | 12,259 (18.2) | 22,585 (19.8) | <0.001 |
| GERD | 238,005 (32.0) | 12,666 (40.9) | 225,339 (31.6) | <0.001 | 157,928 (35.7) | 32,087 (26.5) | 17,344 (25.7) | 30,646 (26.9) | <0.001 |
| Sleep apnea | 261,378 (35.1) | 11,551 (37.3) | 249,827 (35.0) | <0.001 | 165,577 (37.4) | 39,849 (33.0) | 19,448 (28.8) | 36,504 (32.0) | <0.001 |
| Previous diabetes | 180,415 (24.2) | 8,528 (27.6) | 171,867 (24.1) | <0.001 | 105,877 (23.9) | 30,168 (25.0) | 17,165 (25.5) | 27,205 (23.8) | <0.001 |
|
| |||||||||
| HbA1cb,c, mean (SD) | 6.6 (2.1) | 6.6 (2.0) | 6.5 (2.1) | 0.092 | 6.5 (2.1) | 6.7 (2.0) | 6.6 (2.1) | 6.5 (2.0) | <0.001 |
| <5.7%, n (%) | 54,992 (39.9) | 2,101 (38.5) | 52,891 (40.0) | <0.001 | 34,363 (42.4) | 7,241 (31.2) | 5,153 (39.5) | 8,235 (40.3) | <0.001 |
| 5.7–6.4%, n (%) | 41,962 (30.5) | 1,529 (28.0) | 40,433 (30.6) | 23,092 (28.5) | 8,463 (36.5) | 4,011 (30.8) | 6,396 (31.3) | ||
| 6.5–7.5%, n (%) | 17,489 (12.7) | 784 (14.4) | 16,705 (12.6) | 10,176 (12.6) | 3,185 (13.7) | 1,608 (12.3) | 2,520 (12.3) | ||
| >7.5%, n (%) | 23,265 (16.9) | 1,048 (19.2) | 22,217 (16.8) | 13,434 (16.6) | 4,285 (18.5) | 2,268 (17.4) | 3,278 (16.0) | ||
|
| |||||||||
| Ethnicity, n (%) | <0.001 | ||||||||
| Hispanic | 67,430 (9.1) | 2,603 (8.4) | 64,827 (9.1) | ||||||
| NHW | 442,271 (59.4) | 17,766 (57.4) | 424,505 (59.5) | ||||||
| NHB | 120,893 (16.2) | 6,363 (20.5) | 114,620 (16.0) | ||||||
| Other | 114,092 (15.3) | 4,246 (13.7) | 109,846 (15.4) | ||||||
TBWL: total body/percent weight loss; LSG: laparoscopic sleeve gastrectomy; LRYGB: laparoscopic Roux-en-Y gastric bypass; BMI: body mass index; GERD: gastroesophageal reflux disease
19 < age ≤ 70
A total of 137,708 patients reported pre-operative HbA1c
HbA1c < 5.7% (<39 mmol/mol), HbA1c 5.7–6.4% (39–46 mmol/mol), HbA1c 6.5–7.5% (48–58 mmol/mol), and >7.5% (>58 mmol/mol)
Two-sample t-test or One-way ANOVA for continuous variables and chi-square test for categorical variables
After excluding 6 patients with missing data, a total of 30,972 participants or 4.2% of the total sample were readmitted within 30 days of MBS. In comparison to people who were not readmitted, those who were readmitted were older, had a higher pre-MBS maximum BMI and greater prevalence of comorbidities including hypertension, hyperlipidemia, pre-MBS, type 2 diabetes, GERD, and OSA (Table 1). Patients with type 2 diabetes (either previously diagnosed or with HbA1c values ≥6.5% [≥48 mmol/mol]) were 17% more likely to be readmitted than controls (OR=1.17, 95% CI 1.14–1.20). Interestingly, patients with prediabetes were 12% less likely to experience all-cause readmission within 30 days compared to controls (OR 0.88, 95% CI 0.83–0.94). Adjusted odds ratios moderated the overall effect of type 2 diabetes on all-cause readmission risk (aOR=0.96, 95% CI 0.90–1.03). Uncontrolled type 2 diabetes (HbA1c >7.5% [58 mmol/mol]) resulted in an 8% higher risk for all-cause readmission that increased to a 10% risk after adjusted analysis (aOR 1.10, 95% CI 1.01–1.19) in comparison to patients with controlled diabetes (HbA1C ≤7.5% [≤58 mmol/mol]) (Table 2).
Table 2.
Odds ratios for all-cause hospital readmission by diabetes diagnosis and HbA1c level (MBSAQIP, 2015–2018).
| Variable | Crude Odds Ratioc (95% CI) | Adjusted Odds Ratiod (95% CI) | ||||
|---|---|---|---|---|---|---|
| Total | NHW | NHB | Hispanic | Others | ||
| Type 2 diabetes a | ||||||
| No pre-diabetes or type 2 diabetes | 1.0 (ref) | |||||
| Pre-diabetes | 0.88* (0.83–0.94) | 0.84* (0.77–0.92) | 0.80* (0.70–0.90) | 1.03 (0.85–1.24) | 1.0 (0.86–1.17) | 0.85* (0.79–0.92) |
| Type 2 diabetes | 1.17* (1.14 – 1.20) | 1.17* (1.13 – 1.21) | 1.16* (1.10 – 1.23) | 1.10* (1.01 – 1.20) | 1.16* (1.09 −1.25) | 0.96 (0.90 −1.03) |
| HbA1c b | ||||||
|
≤7.5%
(≤58 mmol/mol) |
1.0 (ref) | |||||
| >7.5% (>58 mmol/mol) | 1.08 (0.99–1.17) | 1.09 (0.98 – 1.22) | 1.11 (0.93–1.33) | 1.11 (0.83–1.48) | 0.92 (0.74–1.15) | 1.10* (1.01 – 1.19) |
NHW: Non-Hispanic White, NHB: Non-Hispanic Black, HbA1c: Glycated hemoglobin
Significant differences in comparison to reference
No pre-diabetes or type 2 diabetes: patients without a diabetes diagnosis and an A1C <5.7% (n=521548); Pre-diabetes: patients without a diabetes diagnosis and an A1C 5.7–6.4% (n=30977); Type 2 diabetes: all patients with a prior diagnosis of diabetes or an A1C ≥ 6.5% (n=192251)
HbA1c>7.5%: patients with uncontrolled diabetes (n=23265); HbA1c≤7.5%: patients with controlled diabetes (n=32905)
crude logistic regression
Adjusted logistic regression controlling for age, race/ethnicity, pre-operative BMI, HbA1c measurement timepoint, other comorbidities including hypertension, hyperlipidemia, GERD, and sleep apnea
Ethnic group distribution for all-cause readmission within 30 days was 57% NHW, 21% NHB and 8% Hispanic, representing 4.0%, 5.3% and 3.9% of the NHW, NHB and Hispanic cohorts, respectively. All ethnic groups showed a higher risk of all-cause readmission due to type 2 diabetes ranging from 10–17%, with NHB having the highest all-cause readmission risk. The main two causes for 30-day post-MBS all-cause readmissions for all ethnicities were gastrointestinal/nutritional-related events (nausea, vomiting, fluid, electrolyte, nutritional depletion, etc.) and abdominal pain. The third cause of all-cause readmission was different by ethnicity. Anastomotic/staple line leak was the third most common cause of all-cause readmission in NHW, whereas pulmonary embolism and bleeding were the third most common causes for all-cause readmission in NHB and Hispanics, respectively. Overall, ethnic differences were seen for most all-cause readmission events (Table 3).
Table 3.
Causes of 30-day post-operative readmissions by ethnicity, (MBSAQIP 2015–2018).
| Variable, n (%) | Total | NHW | NHB | Hispanic | Others | P-valueb |
|---|---|---|---|---|---|---|
| Abdominal Pain | 4312 (13.92) | 2348 (7.58) | 879 (2.84) | 438 (1.41) | 647 (2.09) | <0.001 |
| Anastomotic Ulcer | 316 (1.02) | 194 (0.63) | 59 (0.19) | 25 (0.08) | 38 (0.12) | <0.001 |
| Anastomotic/Staple Line Leak | 1470 (4.75) | 994 (3.21) | 197 (0.64) | 108 (0.35) | 171 (0.55) | <0.001 |
| Band Erosion | 19 (0.06) | 17 (0.05) | 2 (0.01) | 0 (0) | 0 (0) | 0.006 |
| Band Slippage/Prolapse | 15 (0.05) | 11 (0.04) | 1 (0.00) | 0 (0) | 3 (0.01) | 0.267 |
| Bile Reflux Gastritis | 32 (0.10) | 19 (0.06) | 3 (0.01) | 2 (0.01) | 8 (0.03) | 0.310 |
| Bleeding | 1318 (4.26) | 827 (2.67) | 185 (0.60) | 114 (0.37) | 192 (0.62) | <0.001 |
| CVA | 58 (0.19) | 39 (0.13) | 12 (0.04) | 2 (0.01) | 5 (0.02) | <0.001 |
| Cardiac N.O.S. | 221 (0.71) | 149 (0.48) | 48 (0.15) | 7 (0.02) | 17 (0.05) | <0.001 |
| Chest Pain | 268 (0.87) | 127 (0.41) | 81 (0.26) | 20 (0.06) | 40 (0.13) | <0.001 |
| GI Perforation | 240 (0.77) | 155 (0.50) | 40 (0.13) | 15 (0.05) | 30 (0.10) | <0.001 |
| Gallstone Disease | 420 (1.36) | 232 (0.75) | 76 (0.25) | 50 (0.16) | 62 (0.20) | <0.001 |
| Gastric Distention | 35 (0.11) | 21 (0.07) | 5 (0.02) | 3 (0.01) | 6 (0.02) | 0.091 |
| Gastro-Gastric Fistula | 38 (0.12) | 28 (0.09) | 5 (0.02) | 1 (0.01) | 4 (0.01) | <0.001 |
| Incisional Hernia | 208 (0.67) | 128 (0.41) | 45 (0.15) | 15 (0.05) | 20 (0.06) | <0.001 |
| Infection/Fever | 812 (2.62) | 515 (1.66) | 92 (0.30) | 80 (0.26) | 125 (0.40) | <0.001 |
| Internal Hernia | 181 (0.58) | 107 (0.35) | 40 (0.13) | 14 (0.05) | 20 (0.06) | <0.001 |
| Intestinal Obstruction | 1372 (4.43) | 844 (2.73) | 243 (0.78) | 110 (0.36) | 175 (0.57) | <0.001 |
| LAGB - Port, Tubing or Band Problem | 26 (0.08) | 21 (0.07) | 0 (0) | 0 (0) | 5 (0.01) | 0.222 |
| Medication-Related | 86 (0.28) | 58 (0.19) | 10 (0.03) | 7 (0.02) | 11 (0.04) | <0.001 |
| Musculoskeletal Pain | 59 (0.19) | 39 (0.13) | 7 (0.02) | 5 (0.02) | 8 (0.03) | 0.003 |
| Myocardial Infarction | 65 (0.21) | 43 (0.14) | 11 (0.04) | 3 (0.01) | 8 (0.03) | <0.001 |
| Nausea and Vomiting, Fluid, Electrolyte, or Nutritional Depletion | 8509 (27.47) | 4374 (14.12) | 2257 (7.29) | 699 (2.26) | 1179 (3.81) | <0.001 |
| Nephrolithiasis | 119 (0.38) | 74 (0.24) | 11 (0.04) | 14 (0.05) | 20 (0.06) | 0.021 |
| Abdominal Sepsis | 893 (2.88) | 542 (1.75) | 163 (0.53) | 81 (0.26) | 107 (0.35) | <0.001 |
| Respiratory Failure | 219 (0.71) | 157 (0.51) | 21 (0.07) | 12 (0.04) | 29 (0.09) | <0.001 |
| Planned Surgery | 109 (0.35) | 67 (0.22) | 22 (0.07) | 5 (0.02) | 15 (0.05) | <0.001 |
| Pneumonia | 703 (2.27) | 455 (1.47) | 114 (0.37) | 42 (0.14) | 92 (0.30) | <0.001 |
| Psychiatric-Related | 44 (0.14) | 21 (0.07) | 9 (0.03) | 4 (0.01) | 10 (0.03) | 0.044 |
| Pulmonary Embolism | 758 (2.45) | 384 (1.24) | 251 (0.81) | 45 (0.15) | 78 (0.25) | <0.001 |
| Renal Insufficiency | 150 (0.48) | 81 (0.26) | 38 (0.12) | 12 (0.04) | 19 (0.06) | <0.001 |
| Shortness of Breath | 313 (1.01) | 186 (0.61) | 67 (0.22) | 20 (0.06) | 40 (0.13) | <0.001 |
| Strictures/Stomal Obstruction | 596 (1.92) | 303 (0.98) | 136 (0.44) | 68 (0.22) | 89 (0.29) | <0.001 |
| Vein Thrombosis Requiring Therapy | 1037 (3.35) | 554 (1.79) | 225 (0.73) | 98 (0.32) | 160 (0.52) | <0.001 |
| Wound Infection/Evisceration | 1043 (3.37) | 679 (2.19) | 137 (0.44) | 78 (0.25) | 149 (0.48) | <0.001 |
| Other | 4908 (15.85) | 2971 (9.59) | 870 (2.81) | 405 (1.31) | 662 (2.14) | <0.001 |
| Total a | 30972 (100.0) | 17764 (57.36) | 6362 (20.54) | 2602 (8.40) | 4244 (13.70) | <0.001 |
ICU: Intensive Care Unit; ED: Emergency Department Visits; UTI: Urinary Tract Infection; SSI: Surgical Site Infections, NHW: Non-Hispanic White, NHB: Non-Hispanic Black, CVA: Cerebrovascular accident; N.O.S: not other specified; CHF: chronic heart failure
Nmissing = 6
Chi-square or fisher’s exact test as appropriate
The average %TBWL one-month post-MBS was 15.1%. Overall, a significantly attenuated 30-day %TBWL was found in patients with type 2 diabetes (14.57 ± 0.10%) and prediabetes (13.48 ± 0.26%), versus controls (15.30 ± 0.07%), (p < 0.001). Regardless of type 2 diabetes status, “other ethnicities” recorded the greatest weight loss (see Table 1). In patients with type 2 diabetes, Hispanics, and NHB achieved less weight loss compared to NHW. Within-ethnicity analysis showed both NHW and NHB with prediabetes and type 2 diabetes independently showed to experience less weight loss than their respective controls (%TBWL 13.66 ± 0.13% vs 14.58 ± 0.08%; %TBWL 12.51 ± 0.24% vs 13.71± 0.16%, P<0.05, respectively). Interestingly, patients with prediabetes lost less weight than patients with type 2 diabetes, which remained significant after stratification by ethnicity with the exception of Hispanics. When comparing weight loss based on HbA1c values, we observed significantly less weight reduction in patients with uncontrolled type 2 diabetes (HbA1c values >7.5% [>58 mmol/mol]) versus their counterparts with controlled diabetes. In addition, among patients with HbA1c values >7.5% (>58 mmol/mol), NHB experienced less weight loss versus NHW. Within ethnicity analysis showed that NHB patients with HbA1c values >7.5% (>58 mmol/mol) had less weight loss versus NHB patients with controlled diabetes (Table S1).
Readmitted patients had a lower proportion of patients in the <5.7% (<39 mmol/mol) and 5.7–6.4% (39–46 mmol/mol) HbA1c categories and a higher proportion in both the 6.5–7.5% (48–58 mmol/mol), and >7.5% (>58 mmol/mol) categories in comparison to non-readmitted patients (Table 1). The percentage of patients reporting a pre-MBS diagnosis of type 2 diabetes are presented in Figure 1 stratified by available HbA1c results. Patients with higher HbA1c levels were more likely to have reported a diagnosis of type 2 diabetes pre-MBS; 8.1% within the HbA1c <5.7% (<39 mmol/mol) and 69.1% within the HbA1c >7.5% (>58 mmol/mol) categories. Similar to previously reported data, 29% (N=11,836) of individuals with an HbA1c ≥ 6.5% (48 mmol/mol) had undiagnosed diabetes (27).
Figure 1. Percentage of patients with type 2 diabetes by HbA1c category.
Number above bars indicate percentage within each HbA1c category. Numbers within table represent actual number of patients within each subgroup. a HbA1c range 4.0–5.69%, b HbA1c range 7.6–20.0%
Pre- and post-MBS characteristics stratified by ethnicity revealed NHW were older, and had higher prevalence of hyperlipidemia, GERD, and OSA. However, NHW also had the highest percentage of LRYGB procedures and the highest representation within the HbA1c <5.7% (<39 mmol/mol) group compared to other ethnic groups. NHB had the highest proportion of female participants, prevalence of hypertension, BMI (maximal and pre-operative), HbA1c 6.5–7.5% (48–58 mmol/mol) and >7.5% (>58 mmol/mol), as well as the lowest proportion of patients with LRYGB. Lastly, Hispanic patients had the highest proportion of previously diagnosed type 2 diabetes and ‘other ethnicities’ had the greatest 30-day weight loss (Table 1) compared to their ethnic group counterparts.
Discussion
This nationally representative sample of MBS results demonstrated that uncontrolled type 2 diabetes (HbA1c >7.5%, [>58 mmol/mol]) is a significant predictor of all-cause hospital readmission and weight loss 30-day post MBS. Our study reported an all-cause 30-day hospital readmission of 4.2% which is comparable to previous studies (7, 9). Prior studies had shown type 2 diabetes and perioperative hyperglycemia to be important risk factors for early all-cause hospital readmission (7–9, 11, 13, 15). Only one other study (n= 1,718) measured pre-operative HbA1c levels and showed a higher likelihood of all-cause hospital readmission in patients with HbA1c >7.3% (>56 mmol/mol) (16), which corroborates our own findings of greater likelihood of re-admission among those with HbA1c > 7.5% (58 mmol/mol). HbA1c values in the prediabetes and type 2 diabetes ranges were also associated with decreased weight loss at 30 days post-MBS.
Although NHB comprised 16.2% of the total sample, the ethnic group represented 20.5% of the readmitted population, confirming previous reports showing that NHB have a higher all-cause hospital readmission rate after MBS (7, 9). Potential drivers of all-cause readmission in NHB individuals include the observed higher prevalence of hypertension, higher preoperative BMI, and high HbA1c values. Type 2 diabetes diagnosis showed the strongest effect on all-cause readmission in NHW. It is possible that within ethnicity analysis had higher statistical power to detect significant differences in NHW as they comprised the majority (60%) of the total sample and 57% of all-cause hospital readmissions.
In agreement with previous research that reported lower post-MBS weight loss at 1-year in people with type 2 diabetes (17–21), we observed a significant effect of prediabetes and type 2 diabetes on weight loss at 30 days after surgery. Our results show that patients with uncontrolled type 2 diabetes (HbA1c >7.5% [>58 mmol/mol]) lost less weight than their controlled counterparts. This concurs with previous research showing that HbA1c values > 8.0% (>64 mmol/mol) are associated with a decrease in weight loss of 6–18 kg at 18 months post-MBS in comparison to groups with HbA1c <6.5% and between 6.5 – 7.9% (22, 23). It could be hypothesized, based on recent research showing a beneficial effect of pre-MBS weight loss on glycemia (28), that inducing pre-MBS weight loss and improving pre-operative glycemic control could potentiate post-MBS weight loss and reduce all-cause hospital readmissions.
The lower weight loss in patients with prediabetes versus those with type 2 diabetes is an interesting observation. Again, we could hypothesize that the beta-cell dysfunction characterizing prediabetes and type 2 diabetes (29) worsens over time, resulting in decreased insulin production and secretion (30). While insulin promotes lipogenesis and inhibits lipolysis, advanced insulin resistance causes insulin to lose its antilipolytic effect (31), so that changes in weight loss by type 2 diabetes status may be related to the effects of chronic obesity and insulin resistance.
Our findings suggest that surgeons should measure HbA1c on all patients prior to MBS to stratify and mitigate the risk for all-cause post-MBS hospital readmission and to identify those who are at risk for suboptimal weight loss. Notably, only 18.5% of our total sample had an HbA1c measured before MBS and a full 29% of those who were tested had diabetes by HbA1c criteria but did not report it pre-operatively. A second recommendation would be to appropriately code and update the diagnoses of patients without a history of diabetes who are found to have HbA1c values >6.5% (48 mmol/mol) on pre-operative testing. These two recommendations would allow for more accurate classification of type 2 diabetes status within the national MBSAQIP database for future research.
Limitations
Some limitations of the current analysis should be mentioned. First, the limited availability of blood glucose measurements in the perioperative period prevents us from drawing a more complete representation of the patients’ glucose control. Second, a considerable number of patients had missing HbA1c data. However, the sample reporting HbA1c values was still large (n= 137,708) in comparison to the sample utilized in prior studies. Sensitivity analysis of missing versus non-missing data resulted in an absolute difference of 0.2%. Although statistically significant, this difference is small and is not likely to be of clinical relevance
Third, HbA1c values were not collected at similar times across the institutions. However, we controlled for HbA1c measure timepoint and found no significant difference by timepoint. Furthermore, because HbA1c values represent the glycemic exposure in the past 2–3 months it is less subject to change on a daily/weekly basis and may give a better overview of glycemic exposure. Lastly, some may argue that 30-day post-MBS weight loss is not clinically valuable even though results showed statistical significance. However, other research has shown that early (6-week) post-MBS weight loss predicts long-term weight loss (32). As observed, ethnicity lead to significant different in our outcome variables. Therefore, generalizability of our results to ethnicities others than the ones selected in the creation of this manuscript might not be applicable.
Conclusion
Our results here based on a national MBS dataset showed that uncontrolled type 2 diabetes (HbA1c >7.5%) is associated with a greater likelihood of all-cause hospital readmission and lower weight loss 30-day post MBS. Both type 2 diabetes and prediabetes were also associated with reduced weight loss 30 days post-MBS. Our findings highlight the need to properly classify individuals with diabetes and further understand the degree of glycemic control prior to MBS.
Supplementary Material
Study Importance Questions:
-
What is already known about this subject?
Diabetes is a predictor of all-cause hospital readmission and decreased weight loss after bariatric surgery.
Non-Hispanic Blacks and Hispanics have the highest rates of all-cause readmission after bariatric surgery.
-
What are the new findings in your manuscript?
Patients with uncontrolled type 2 diabetes, type 2 diabetes, and even prediabetes experience less weight loss after MBS.
Our results stratified by ethnicity show non-Hispanic blacks to have the highest rates of all-cause readmission with both Hispanics and non-Hispanic blacks showing lower weight loss in comparison to non-Hispanic whites.
-
How might your results change the direction of research or the focus of clinical practice?
Our findings highlight the importance of aggressive glucose control prior to MBS to decrease all-cause hospital readmission and improve weight loss post-surgery.
Acknowledgements
The authors are thankful to all study participants and staff members that have participated in the obtention of the MBSAQIP dataset.
The present study was made possible thanks to a retrospective, secondary analysis of public, anonymized MBSAQIP data.
Funding: This review article was funded by the National Institutes of Health, National Institute on Minority Health and Health Disparities (grant #R01MD011686).
Footnotes
Disclosure: The authors declared no conflict of interest
REFERENCES:
- 1.Flegal KM, Kruszon-Moran D, Carroll MD, Fryar CD, Ogden CL. Trends in Obesity Among Adults in the United States, 2005 to 2014. JAMA. 2016;315(21):2284–91. Epub 2016/06/09. doi: 10.1001/jama.2016.6458. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Schauer PR, Kashyap SR, Wolski K, Brethauer SA, Kirwan JP, Pothier CE, et al. Bariatric surgery versus intensive medical therapy in obese patients with diabetes. N Engl J Med. 2012;366(17):1567–76. Epub 2012/03/28. doi: 10.1056/NEJMoa1200225. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Maggard MA, Shugarman LR, Suttorp M, Maglione M, Sugerman HJ, Livingston EH, et al. Meta-analysis: surgical treatment of obesity. Ann Intern Med. 2005;142(7):547–59. Epub 2005/04/06. doi: 10.7326/0003-4819-142-7-200504050-00013. [DOI] [PubMed] [Google Scholar]
- 4.Carlsson LM, Peltonen M, Ahlin S, Anveden A, Bouchard C, Carlsson B, et al. Bariatric surgery and prevention of type 2 diabetes in Swedish obese subjects. N Engl J Med. 2012;367(8):695–704. Epub 2012/08/24. doi: 10.1056/NEJMoa1112082. [DOI] [PubMed] [Google Scholar]
- 5.Sjostrom L, Peltonen M, Jacobson P, Ahlin S, Andersson-Assarsson J, Anveden A, et al. Association of bariatric surgery with long-term remission of type 2 diabetes and with microvascular and macrovascular complications. JAMA. 2014;311(22):2297–304. Epub 2014/06/11. doi: 10.1001/jama.2014.5988. [DOI] [PubMed] [Google Scholar]
- 6.Telem DA, Talamini M, Gesten F, Patterson W, Peoples B, Gracia G, et al. Hospital admissions greater than 30 days following bariatric surgery: patient and procedure matter. Surg Endosc. 2015;29(6):1310–5. Epub 2014/10/09. doi: 10.1007/s00464-014-3834-x. [DOI] [PubMed] [Google Scholar]
- 7.Khorgami Z, Andalib A, Aminian A, Kroh MD, Schauer PR, Brethauer SA. Predictors of readmission after laparoscopic gastric bypass and sleeve gastrectomy: a comparative analysis of ACS-NSQIP database. Surg Endosc. 2016;30(6):2342–50. Epub 2015/08/27. doi: 10.1007/s00464-015-4477-2. [DOI] [PubMed] [Google Scholar]
- 8.El Chaar M, Stoltzfus J, Gersin K, Thompson K. A novel risk prediction model for 30-day severe adverse events and readmissions following bariatric surgery based on the MBSAQIP database. Surg Obes Relat Dis. 2019;15(7):1138–45. Epub 2019/05/06. doi: 10.1016/j.soard.2019.03.005. [DOI] [PubMed] [Google Scholar]
- 9.Garg T, Rosas U, Rivas H, Azagury D, Morton JM. National prevalence, causes, and risk factors for bariatric surgery readmissions. Am J Surg. 2016;212(1):76–80. Epub 2016/05/03. doi: 10.1016/j.amjsurg.2016.01.023. [DOI] [PubMed] [Google Scholar]
- 10.Morgan DJ, Ho KM, Armstrong J, Baker S. Incidence and risk factors for intensive care unit admission after bariatric surgery: a multicentre population-based cohort study. Br J Anaesth. 2015;115(6):873–82. Epub 2015/11/20. doi: 10.1093/bja/aev364. [DOI] [PubMed] [Google Scholar]
- 11.Abraham CR, Werter CR, Ata A, Hazimeh YM, Shah US, Bhakta A, et al. Predictors of Hospital Readmission after Bariatric Surgery. J Am Coll Surg. 2015;221(1):220–7. Epub 2015/06/07. doi: 10.1016/j.jamcollsurg.2015.02.018. [DOI] [PubMed] [Google Scholar]
- 12.Leonard-Murali S, Nasser H, Ivanics T, Shakaroun D, Genaw J. Perioperative Outcomes of Roux-en-Y Gastric Bypass and Sleeve Gastrectomy in Patients with Diabetes Mellitus: an Analysis of the Metabolic and Bariatric Surgery Accreditation and Quality Improvement Program (MBSAQIP) Database. Obes Surg. 2020;30(1):111–8. Epub 2019/10/11. doi: 10.1007/s11695-019-04175-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Wise ES, Amateau SK, Ikramuddin S, Leslie DB. Prediction of thirty-day morbidity and mortality after laparoscopic sleeve gastrectomy: data from an artificial neural network. Surg Endosc. 2019. Epub 2019/10/02. doi: 10.1007/s00464-019-07130-0. [DOI] [PubMed] [Google Scholar]
- 14.Ahmed A, AlBuraikan D, B AL, AlJohi W, Alanazi W, AlRasheed B. Readmissions and Emergency Department Visits after Bariatric Surgery at Saudi Arabian Hospital: The Rates, Reasons, and Risk Factors. Obes Facts. 2017;10(5):432–43. Epub 2017/10/11. doi: 10.1159/000456667. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Meister KM, Hufford T, Tu C, Khorgami Z, Schauer PR, Brethauer SA, et al. Clinical significance of perioperative hyperglycemia in bariatric surgery: evidence for better perioperative glucose management. Surg Obes Relat Dis. 2018;14(11):1725–31. Epub 2018/09/25. doi: 10.1016/j.soard.2018.07.028. [DOI] [PubMed] [Google Scholar]
- 16.Wysocki M, Waledziak M, Hady HR, Czerniawski M, Proczko-Stepaniak M, Szymanski M, et al. Type 2 Diabetes Mellitus and Preoperative HbA1c Level Have no Consequence on Outcomes after Laparoscopic Sleeve Gastrectomy-a Cohort Study. Obes Surg. 2019;29(9):2957–62. Epub 2019/05/16. doi: 10.1007/s11695-019-03936-y. [DOI] [PubMed] [Google Scholar]
- 17.Núñez-Núñez MA, León-Verdín MG, Muñoz-Montes N, Rodríguez-García J, Trujillo-Ortiz JA, Martínez-Cordero yC. Diabetes mellitus tipo 2 podría predecir una pérdida subóptima de peso después de una cirugía bariátrica. Nutr Hosp. 2018;35:1085–9. [DOI] [PubMed] [Google Scholar]
- 18.Campos GM, Rabl C, Mulligan K, Posselt A, Rogers SJ, Westphalen AC, et al. Factors associated with weight loss after gastric bypass. Arch Surg. 2008;143(9):877–83; discussion 84. Epub 2008/09/17. doi: 10.1001/archsurg.143.9.877. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Melton GB, Steele KE, Schweitzer MA, Lidor AO, Magnuson TH. Suboptimal weight loss after gastric bypass surgery: correlation of demographics, comorbidities, and insurance status with outcomes. J Gastrointest Surg. 2008;12(2):250–5. Epub 2007/12/12. doi: 10.1007/s11605-007-0427-1. [DOI] [PubMed] [Google Scholar]
- 20.Al-Khyatt W, Ryall R, Leeder P, Ahmed J, Awad S. Predictors of Inadequate Weight Loss After Laparoscopic Gastric Bypass for Morbid Obesity. Obes Surg. 2017;27(6):1446–52. Epub 2016/12/13. doi: 10.1007/s11695-016-2500-x. [DOI] [PubMed] [Google Scholar]
- 21.Cottam S, Cottam D, Cottam A, Zaveri H, Surve A, Richards C. The Use of Predictive Markers for the Development of a Model to Predict Weight Loss Following Vertical Sleeve Gastrectomy. Obes Surg. 2018;28(12):3769–74. Epub 2018/07/25. doi: 10.1007/s11695-018-3417-3. [DOI] [PubMed] [Google Scholar]
- 22.Perna M, Romagnuolo J, Morgan K, Byrne TK, Baker M. Preoperative hemoglobin A1c and postoperative glucose control in outcomes after gastric bypass for obesity. Surg Obes Relat Dis. 2012;8(6):685–90. Epub 2011/10/11. doi: 10.1016/j.soard.2011.08.002. [DOI] [PubMed] [Google Scholar]
- 23.Lee YC, Lee WJ, Lee TS, Lin YC, Wang W, Liew PL, et al. Prediction of successful weight reduction after bariatric surgery by data mining technologies. Obes Surg. 2007;17(9):1235–41. Epub 2007/12/13. doi: 10.1007/s11695-007-9322-9. [DOI] [PubMed] [Google Scholar]
- 24.Metabolic The and Bariatric Surgery Accreditation and Quality Improvement Program 2018 [June 2020]. Available from: https://www.facs.org/quality-programs/mbsaqip.
- 25.Surgery ACoSaASoMaB. Metabolic and Bariatric Surgery Accreditation and Quality Improvement Program 2012. Available from: https://www.facs.org/quality-programs/mbsaqip.
- 26.Surgery ACoSaASoMaB . User Guide for the 2018 Participant Use Data File (PUF): Metabolic and Bariatric Surgery Accreditation and Quality Improvement Program; 2019. Available from: https://www.facs.org/-/media/files/quality-programs/bariatric/mbsaqip_2018_puf_userguide.ashx.
- 27.Prevention USDoHaHSCfDCa. Estimates of Diabetes and its Burden in the United States. National Diabetes Statistics Report 2020 2020. p. https://www.cdc.gov/diabetes/pdfs/data/statistics/national-diabetes-statistics-report.pdf.
- 28.Sun Y, Liu B, Smith JK, Correia MLG, Jones DL, Zhu Z, et al. Association of Preoperative Body Weight and Weight Loss With Risk of Death After Bariatric Surgery. JAMA Netw Open. 2020;3(5):e204803. Epub 2020/05/15. doi: 10.1001/jamanetworkopen.2020.4803. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Cai X, Xia L, Pan Y, He D, Zhu H, Wei T, et al. Differential role of insulin resistance and beta-cell function in the development of prediabetes and diabetes in middle-aged and elderly Chinese population. Diabetol Metab Syndr. 2019;11:24. Epub 2019/03/16. doi: 10.1186/s13098-019-0418-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Butler AE, Janson J, Bonner-Weir S, Ritzel R, Rizza RA, Butler PC. Beta-cell deficit and increased beta-cell apoptosis in humans with type 2 diabetes. Diabetes. 2003;52(1):102–10. Epub 2002/12/28. doi: 10.2337/diabetes.52.1.102. [DOI] [PubMed] [Google Scholar]
- 31.Saponaro C, Gaggini M, Carli F, Gastaldelli A. The Subtle Balance between Lipolysis and Lipogenesis: A Critical Point in Metabolic Homeostasis. Nutrients. 2015;7(11):9453–74. Epub 2015/11/19. doi: 10.3390/nu7115475. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Manning S, Pucci A, Carter NC, Elkalaawy M, Querci G, Magno S, et al. Early postoperative weight loss predicts maximal weight loss after sleeve gastrectomy and Roux-en-Y gastric bypass. Surg Endosc. 2015;29(6):1484–91. Epub 2014/09/23. doi: 10.1007/s00464-014-3829-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.