COVID-19 and its Severity in Bariatric Surgery-Operated Patients
See Commentary, pg. 18.
Abstract
Objective
Obesity is a major risk factor for severe forms of coronavirus disease (COVID-19), but little is known about the post–bariatric surgery (BS) setting. The prevalence of likely COVID-19 and its risk factors in patients followed up after BS was assessed.
Methods
A total of 738 patients who underwent BS and were followed up at a university medical center were surveyed. A retrospective comparison of characteristics at baseline, 1 year after BS, and at the time of lockdown was performed between patients with COVID-19–likely events (CL) based on a combination of reported symptoms and those for whom COVID-19 was unlikely.
Results
CL occurred in 62 (8.4%) patients, among whom 4 (6.4%) had a severe form requiring hospitalization and 1 (1.6%) died. The CL group had a higher proportion of persistent type 2 diabetes (T2D) at last follow-up (36.2% vs. 20.3%, P = 0.01). BMI at the time of lockdown was lower in the CL group (30.2 ± 5.1 vs. 32.8 ± 6.5 kg/m2; P < 0.01) with higher percent weight loss since BS in the CL group. Severe forms of COVID-19 requiring hospitalization were associated with persistent T2D at the last follow-up visit.
Conclusions
In BS patients, CL were associated with persistent T2D and lower BMI.
Study importance
What is already known?
- ► Obesity is a major risk factor for severe forms of coronavirus disease (COVID-19) and related mortality.
- ► Bariatric surgery leads to successful long-term weight loss and improvement of comorbidities that may decrease the negative outcomes of COVID-19.
What does this study add?
- ► The prevalence of COVID-19–likely events in bariatric surgery–operated patients is similar to what is observed in the general population.
- ► Remission of type 2 diabetes after bariatric surgery is associated with a lower risk for COVID-19.
- ► A higher surgery-induced weight loss and lower weight and BMI post surgery are associated with COVID-19–likely events.
How might these results change the direction of research or the focus of clinical practice?
- ► Patients after bariatric surgery, in the case of nonremission of type 2 diabetes, should be considered at higher risk for COVID-19.
- ► Further research is needed to clarify the links between weight loss, malnutrition, and COVID-19.
Introduction
Obesity is a recognized major risk factor for severe forms of coronavirus disease (COVID-19) and related mortality, independently of obesity-associated comorbidities ((1-4)). Although several hypotheses have been suggested to explain why, much remains to be understood. Beyond the increased prevalence of type 2 diabetes (T2D), cardiovascular disease, alterations in respiratory function, and increased risk of pulmonary embolism, the obesity-related low-grade inflammation could be an additional mechanism ((5)).
Bariatric surgery (BS) is increasingly performed in persons with severe obesity, especially in France, which ranks third worldwide regarding the number of patients operated on ((6)). However, very little is known on COVID-19 in post-BS patients. On the one hand, beneficial effects induced by BS could decrease patient’s risk for severe COVID-19. Indeed, BS leads to successful long-term weight loss, metabolic improvement (including T2D remission) ((7)), improvement in sleep apnea syndrome ((8)), and decreased low-grade systemic inflammation ((9)). On the other hand, BS can lead to malnutrition and to vitamin deficiencies such as vitamin D, of which a deficit was suggested to enhance the severity of COVID-19 ((10)). Though recent guidelines were proposed to help prioritizing which patients could undergo bariatric and metabolic surgery during the COVID-19 pandemic ((11)), whether patients who already had BS display a particular susceptibility to COVID-19 risk and severity needs urgent evaluation.
We conducted a retrospective observational study in a cohort of patients who had BS to estimate the prevalence of COVID-19 and evaluate factors associated with COVID-19 incidence and severity in this setting.
Methods
Study population
The study is based on our ongoing BS cohort, “BARICAN,” of 937 patients followed up at the Nutrition Department of Pitié-Salpêtrière University Hospital (Paris, France). Patients were operated on in the Surgery Departments of Hôtel-Dieu, Ambroise Paré, and Pitié-Salpêtrière hospitals in Paris between 2004 and 2020. Ethical approval was obtained from the French Research Ethics Committee of CPP Ile de France-1 N°13533 and by the “Commission nationale de l’informatique et des libertés” No. 1222666. Informed written consent was obtained from all participants.
BS, baseline and follow-up, and clinical and biological data collection
Detailed patients’ clinical, biological, and anthropometric characteristics were obtained before surgery (baseline) and 12 months after BS. T2D status was defined as proposed by the American Diabetes Association ((12)) at baseline, at 12 months, and at the last follow-up before the survey (i.e., last known T2D status). Body composition was assessed by whole-body fan-beam dual-energy x-ray absorptiometry scan (Hologic Inc., Waltham, MA, USA) ((13)). Blood samples were collected after a 12-hour overnight fast. Insulin resistance was assessed by the homeostatic model assessment of insulin resistance ((14)).
Data collection during lockdown
Lockdown in France extended from March 10 to May 11, 2020. All patients received a phone call between April 27, 2020, and May 27, 2020, and were asked to contribute to a survey by answering a set of standardized questions about COVID-19, medical events, and changes in lifestyle during lockdown. If the patient was unable to answer to the survey (e.g., hospitalization, death), information was gathered from relatives. A patient was considered lost to follow-up if no answer was obtained after three calls. COVID-19 events were considered likely (CL) in patients who had experienced an episode of anosmia/dysgueusia or the association of fever (self-measured temperature ≥ 37.8°C) and one symptom among cough, dyspnea, nose discharge/swelling, or odynophagia since February 2020. The combination of such symptoms has been considered as most relevant for the CL status in recent reports ((15, 16)). Patients hospitalized for COVID-19 or with a positive polymerase chain reaction test for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) were also considered as CL. Self-measured weight before the lockdown period was recorded. Patients’ smoking and vitamin supplementation status were reported. Recent exposure defined by a close contact with a person with confirmed Sars-CoV-2 infection was noted.
Statistical analysis
Continuous variables were expressed as mean (SD). Categorical variables were expressed as absolute values and percentages. Continuous variables with a nonparametric distribution were log-transformed before analysis. Student’s t tests for continuous variables and Pearson’s χ2 test or Fisher’s exact test for categorical variables were used to compare characteristics of patients according to their COVID-19–likely status (likely [CL] or unlikely infection [CU]). For significant results, a multivariate logistic regression was performed, adjusted on the time elapsed since surgery and last known T2D status. Significance was set at P < 0.05. Analyses were conducted using R studio software version 1.2.1335 (The R Foundation for Statistical Computing, Vienna, Austria).
Results
Baseline and 12-month characteristics
From the 937 operated patients followed up, we could collect recent information for 738 (78.8%) of them. Patients were middle-aged and mostly female, and 44.0% had T2D before surgery (Table 1). The main procedures performed were Roux-en-Y gastric bypass (54.4%) and sleeve gastrectomy (45.0%). Only four (0.4%) had gastric banding. The mean time elapsed between surgery and the time of the call was 3.7 (2.7) years, and the proportion of patients within their first year of follow-up (i.e., weight-loss phase) did not differ between CL and CU groups (Table 2). Mean weight loss 1 year after surgery was close to 30% of preoperative weight, and the proportion of T2D was halved 1 year after surgery (Table 1).
| Baseline (before surgery) | P value | 12 months after surgery | P value | |||||
|---|---|---|---|---|---|---|---|---|
| All, N = 738 | COVID-19 unlikely, n = 676 | COVID-19 likely, n = 62 | All, N = 553 | COVID-19 unlikely, n = 501 | COVID-19 likely, n = 52 | |||
| Age (y) | 50.0 (12.3) | 50.1 (12.3) | 49.6 (12.9) | 0.77 | 51.0 (12.3) | 51.0 (12.2) | 50.7 (12.5) | 0.86 |
| Sex (male), n (%) | 160 (21.7) | 148 (21.9) | 12 (19.4) | 0.76 | 114 (20.6) | 105 (21.0) | 9 (17.3) | 0.66 |
| Caucasian, n (%) | 611 (82.8) | 560 (82.8) | 51 (82.3) | 1.00 | 443 (80.1) | 401 (80.0) | 42 (80.8) | 1.00 |
| Weight (kg) | 123.6 (21.4) | 123.8 (21.5) | 121.7 (20.6) | 0.47 | 88.8 (19.6) | 89.1 (19.7) | 85.8 (17.5) | 0.21 |
| Height (cm) | 166.7 (8.6) | 166.6 (8.6) | 167.4 (9.3) | 0.51 | 166.5 (8.6) | 166.5 (8.6) | 167.0 (9.3) | 0.72 |
| BMI (kg/m2) | 44.7 (6.6) | 44.8 (6.5) | 44.2 (7.2) | 0.57 | 32.2 (6.3) | 32.2 (6.2) | 31.4 (7.4) | 0.44 |
| Body fat (%) | 48.1 (5.1) | 48.0 (5.0) | 48.5 (6.1) | 0.57 | 38.7 (6.7) | 38.7 (6.6) | 38.3 (7.5) | 0.73 |
| Type of surgery, n (%) | 1.00 | 1.00 | ||||||
| Gastric banding | 4 (0.5) | 4 (0.6) | 0 (0.0) | 3 (0.5) | 3 (0.6) | 0 (0.0) | ||
| RYGB | 399 (54.4) | 365 (54.4) | 34 (54.8) | 317 (57.4) | 287 (57.4) | 30 (57.7) | ||
| Sleeve gastrectomy | 330 (45.0) | 302 (45.0) | 28 (45.2) | 232 (42.0) | 210 (42.0) | 22 (42.3) | ||
| Weight loss (%) | - | - | - | 28.4 (8.4) | 28.3 (8.3) | 29.0 (9.8) | 0.64 | |
| Excess body weight loss (%) | 68.6 (23.8) | 68.1 (23.4) | 72.6 (26.7) | 0.25 | ||||
| Obstructive sleep apnea syndrome, n (%) | 512 (73.2) | 472 (73.8) | 40 (67.8) | 0.40 | 201 (40.0) | 184 (40.5) | 17 (34.7) | 0.52 |
| Hypertension, n (%) | 367 (53.0) | 338 (53.1) | 29 (50.9) | 0.85 | 220 (39.9) | 198 (39.6) | 22 (43.1) | 0.73 |
| Type 2 diabetes, n (%) | 308 (44.0) | 279 (43.6) | 29 (48.3) | 0.57 | 116 (21.0) | 97 (19.4) | 19 (36.5) | 0.01 |
| Glycemia (mmol/L) | 6.2 (2.1) | 6.2 (2.1) | 6.7 (2.6) | 0.14 | 5.1 (1.2) | 5.0 (1.2) | 5.2 (1.2) | 0.58 |
| HbA1c (%) | 5.7 (0.8) | 5.7 (0.7) | 6.0 (1.1) | 0.10 | 5.7 (0.8) | 5.7 (0.7) | 6.0 (1.1) | 0.04 |
| HOMA-IR | 5.3 (3.7) | 5.3 (3.7) | 5.7 (4.0) | 0.43 | 1.9 (1.2) | 1.9 (1.2) | 2.0 (1.3) | 0.57 |
| IL6 (pg/mL) | 6.2 (20.7) | 6.2 (21.5) | 6.1 (7.3) | 0.22 | 5.0 (18.3) | 4.8 (18.1) | 7.9 (19.9) | 0.09 |
| Vitamin D (ng/mL) | 22.0 (10.0) | 22.0 (10.2) | 22.6 (8.3) | 0.65 | 31.1 (9.3) | 31.0 (9.2) | 32.0 (10.1) | 0.49 |
| Vitamin B12 (pmol/L) | 321 (143) | 322 (137) | 306 (197) | 0.17 | 283 (149) | 284 (141) | 273 (212) | 0.23 |
- Results expressed as mean (SD) for continuous data and n (%) for categorical data. P values result from Student t test for continuous data and χ2 or Fisher exact test for categorical data between the two groups. HbA1c, glycated hemoglobin A1c; HOMA-IR, homeostatic model assessment of insulin resistance; IL6, interleukin 6; RYGB, Roux-en-Y gastric bypass.
| All, N = 738 | COVID-19 unlikely, n = 676 | COVID-19 likely, n = 62 | P a | Adjusted odds ratio (95% CI)b | Adjusted Pb | |
|---|---|---|---|---|---|---|
| Time since surgery (y) | 3.7 (2.7) | 3.7 (2.7) | 4.2 (2.3) | 0.01 | 1.02 (0.92-1.12) | 0.68 |
| Surgery < 1 year ago, n (%) | 81 (11.0) | 78 (11.5) | 3 (4.8) | 0.16 | 0.70 (0.11-2.68) | 0.65 |
| Last known T2D status | 141 (21.8) | 120 (20.3) | 21 (36.2) | 0.01 | 2.17 (1.20-3.86) | < 0.01 |
| Weight (kg) | 90.7 (20.4) | 91.2 (20.8) | 85.0 (14.6) | < 0.01 | 0.98 (0.97-0.99) | 0.05 |
| BMI (kg/m2) | 32.6 (6.5) | 32.8 (6.5) | 30.2 (5.1) | < 0.01 | 0.92 (0.87-0.97) | < 0.01 |
| Obesity (BMI > 30), n (%) | 424 (61.7) | 396 (63.1) | 28 (47.5) | 0.03 | 0.52 (0.30-0.92) | 0.02 |
| Weight loss (%) | 26.3 (10.9) | 26.0 (10.9) | 29.3 (10.7) | 0.03 | 1.03 (1.00-1.06) | 0.04 |
| Excess body weight loss (%) | 63.2 (27.5) | 62.4 (27.7) | 71.6 (24.9) | 0.01 | 1.01 (1.00-1.03) | 0.02 |
| Vitamin supplements, n (%) | 581 (81.0) | 535 (81.4) | 46 (76.7) | 0.47 | 0.70 (0.36-1.44) | 0.30 |
| Corticosteroid treatment at last visit, n (%) | 16 (2.5) | 15 (2.6) | 1 (1.8) | 1.00 | 0.66 (0.04-3.43) | 0.70 |
| Confirmed Sars-CoV-2 exposure, n (%) | 135 (18.9) | 108 (16.5) | 27 (45.8) | < 0.01 | 5.59 (3.04-10.3) | < 0.01 |
| Smoker, n (%) | 74 (10.5) | 66 (10.2) | 8 (13.6) | 0.55 | 1.64 (0.68-3.54) | 0.23 |
- Results expressed as mean (SD) for continuous data and n (%) for categorical data.
- a P values result from Student t test for continuous data and χ2 or Fisher exact test for categorical data between the two groups.
- b Logistic regression adjusted for time since surgery and last known T2D status.
Prevalence and factors associated with CL
After BS, 62 (8.4%) patients were categorized as CL. There was no difference regarding the type of surgery or baseline characteristics between CL and CU groups. However, time elapsed since surgery was significantly higher in CL (P = 0.01). One year after BS, the CL group had a higher proportion of patients with persistent T2D and higher HbA1c level (Table 1). Persistent T2D at the last follow-up visit before survey was also positively associated with CL (Table 2). BMI at the time of lockdown was lower in the CL versus CU group, with a higher percent weight loss and a lower prevalence of obesity (BMI > 30). Importantly, a multivariate analysis showed these differences were independent of the last known T2D status and the time since surgery (Table 2). CL patients reported more often a recent COVID-19 exposure (P < 0.01). There was no significant difference in the vitamin supplementation status between the two groups.
Characteristics of the patients with severe forms of COVID-19
Among the 62 CL cases, 4 (6.4%) required hospitalization and 1 (1.6%) death was reported. Mean age was 61.8 years. Importantly, the proportion of patients with persistent T2D was higher in patients requiring hospitalization than in those who did not (respectively, n = 4 [100%] vs. n = 17 [31.4%]; P = 0.014) (Table 3).
Discussion
We surveyed a cohort of over 700 patients with a history of obesity followed up after BS to assess the prevalence of CL in this population and describe associated characteristics. The prevalence of CL was 8.4%, among which 6.4% were severe forms requiring hospitalization. The main difference between CL and CU patients was a higher BS-induced weight loss as well as lower weight and BMI at the time of lockdown for CL. Patients likely to have had COVID-19 also had a more often persistent T2D after surgery.
This is the first study to report and describe COVID-19 cases after BS, limiting the possibility to compare with results from other countries. However, at the time of our survey, a global estimation was performed in the greater Paris region (Ile de France), the area of residence of the majority of our patients, and reported an estimated prevalence of COVID-19 cases of 9.9% ((17)). This appears in the same magnitude as what we describe herein, with a similar proportion of patients requiring hospitalization.
T2D is recognized as a major risk factor for severe forms of COVID-19 ((18, 19)), and our study suggests that this remains true after BS. Patients with persistent T2D after BS were at higher risk for CL. Importantly, persistent T2D was associated with severe outcome of COVID-19.
One of the main differences observed in our likely infected patients was a lower BMI at the time of the survey, which is consistent with a more important post-BS percent weight loss, and this was independent of the duration of follow-up after BS. Malnutrition is known to be associated with the risk for viral pneumonia since the 1918 influenza pandemic ((20)), and it is possible that BS-induced malnutrition may contribute to a higher risk for infection with Sars-CoV-2. Here, we did not observe a difference in vitamin supplementation status between the two groups who also had similar vitamin levels in their 1-year follow-up. However, we cannot rule out differences in vitamin levels at the time of the infection.
Some limitations need to be mentioned. The definition of CL used here derives from a combination of self-reported symptoms with a good diagnostic performance but lacks specificity ((15)). Another limitation is that the time elapsed between surgery and our survey was heterogeneous and significantly different between the two groups, but importantly, adjustment for time since surgery did not alter our findings. Finally, potential confounders such as socioeconomic status, size of family, and working conditions could not be considered in this study.
Conclusion
Patients followed up after BS displayed rates of CL that appear in line with those of the general population. Persistent T2D and lower BMI after BS are associated with the risk and the severity of COVID-19. Further work is needed to assess in more detail lifestyle changes of patients after BS in the times of COVID-19 and the potential links between malnutrition and the risk for COVID-19.
Acknowledgments
We thank Valentine Lemoine for patient recruitment and inclusion. Data described in the manuscript and analytic code book will be made available upon request.
Funding agencies
Grant support in the field of bariatric surgery was obtained by Ministry of Health and Solidarity (Assistance Publique-Hôpitaux de Paris, to KC/PHRC Microbaria, to JAW/CRC Fibrota) and by the European Union (Metacardis to KC HEALTH-F4-2012-305312 to KC). The authors also thank the FORCE network for supporting obesity clinical research.
Disclosure
The authors declared no conflict of interest.
| COVID-19 patients requiring hospitalization, n = 4 | |
|---|---|
| Age (y) | 61.8 (4.08) |
| Sex (male), n (%) | 1 (25.0) |
| Type of surgery, n (%) | |
| RYGB | 2 (50.0) |
| Sleeve | 2 (50.0) |
| Characteristics at 12 months after surgery | |
| Obstructive sleep apnea syndrome, n (%) | 2 (50.0) |
| Hypertension, n (%) | 4 (100) |
| Type 2 diabetes, n (%) | 4 (100) |
| BMI (kg/m2) | 32.7 (4.60) |
| Body fat (%) | 36.4 (12.3) |
| Characteristics at the time of lockdown | |
| Time since surgery (y) | 5.05 (1.96) |
| Type 2 diabetes at last follow-up, n (%) | 4 (100) |
| Weight (kg) | 90.5 (7.78) |
| BMI (kg/m2) | 30.5 (2.63) |
| Weight loss (%) | 23.2 (0.27) |
| Excess body weight loss (%) | 63.6 (9.73) |
| Death, n (%) | 1 (25.0) |
- Results expressed as mean (SD) for continuous data and n (%) for categorical data.
