Efficacy of bariatric surgery in COVID-19 patients: An updated systematic review and meta-analysis

Introduction Obesity is linked with poor outcomes in patients with SARS-CoV-2 infection. In patients with BMI > 35 kg/m2, increased rates of hospital and subsequently ICU admissions have been noted. Bariatric surgery resulting in sustained weight loss is hypothesized to decrease the morbid outcomes associated with COVID. In this review, we update the evidence on the topic. Methods An extensive literature search was conducted of electronic databases. Screening of the articles based on the eligibility criteria was followed by relevant data extraction. In addition to articles used in a previous meta-analysis, relevant databases were searched to filter for any new articles. Initially, two independent reviewers screened Pubmed and the Cochrane database followed by a thorough search of additional databases such as Google scholar and Medrxiv. Articles were first screened using title and abstract, followed by a full text read. Duplicates, meta-analysis, letter to the editors, and commentaries were excluded. No language restrictions were applied. Results A total of nine articles with a population of 1,130,341 were entered into RevMan. Patients with bariatric surgery displayed significantly decreased hospitalization (OR: 0.52, 95% CI [0.45, 0.61]), were less likely to be admitted to the ICU (OR: 0.44, 95% CI [0.29, 0.67]), and had reduced overall mortality (OR: 0.42, 95% CI [0.25, 0.70]). Conclusion Surgically induced weight loss is beneficial in reducing morbidity and mortality of COVID-19.

mitigating factor and a treatment tool to enhance the quality of life in patients suffering from  Extensive evidence is now being generated indicating the protective effects of bariatric surgery on clinical outcomes related to severe COVID-19. A recent meta-analysis published by Aminian et al. validated this rationale by concluding that prior bariatric surgery was associated with a lower rate of mortality and hospital admission in patients in obese patients infected with COVID-19 (6). However, since COVID-19 is an ongoing pandemic, an increasing level of research is continuously being published on the subject. We aim to conduct an updated systematic review and literature search on the association between prior bariatric surgery and the clinical outcomes of COVID-19 infection. This will further help to promote the narrative of an ongoing double pandemic, i.e., COVID-19 and obesity, and subsequently alert public health authorities to resume and enhance access to bariatric surgery during the current pandemic.

METHODS
This meta-analysis is reported according to the Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) guidelines for updated meta-analysis (7) and follows the structure laid out by the Cochrane Collaboration (8).

Data Sources and Search Strategy
A detailed literature search of PubMed and Cochrane Central was conducted from its inception up until January 2022. Google scholar and Medrxiv.org were utilized as an additional search engines to screen any non-peer reviewed articles. The search strategy used in the electronic databases included keywords ("COVID-19 OR "coronavirus" OR "SARS-Cov-2") and ("bariatric" OR "RYGB" OR "gastric bypass" OR "sleeve"). The complete search strategy is available in the supplementary material (S1). No restrictions on time, language, study design, and sample size were applied. Editorials and bibliographies of relevant review articles and unpublished databases were manually reviewed, ensuring none of the studies from white and grey literature were omitted.

Study Selection
Articles retrieved from the literature search were transferred to the Endnote Reference Library (Version X7.5; Clarivate Analytics, Philadelphia, Pennsylvania), where the duplicates were identified and removed. Two independent reviewers (N.H and A.M) used a two-phase blinded selection process. Articles were separately assessed according to their title and abstract details followed by a thorough full-text read. Selected articles were verified against the predefined inclusion criteria which were later matched and the duplicates were removed. In case of any conflict, a third reviewer (E.W) was consulted. Studies irrespective of language, containing fulltext, and those dating to January 2022 were included. Meta-analyses and/or systematic reviews, letters to the editors (LTEs), and perspectives were excluded. Studies with patients with the following characteristics were included in the study: a) COVID-19 positive and b) with or without bariatric surgery. The primary end-points of the study were: mortality, hospital admission, and severe COVID-19 infection (ICU admission and mechanical ventilation combined).

Data Extraction and Quality Assessment
Following variables of interest were extracted from each study on a standard excel sheet: study population, study design, sample size, number of patients in each group (bariatric and non-bariatric), general patient characteristics of each group (age, gender, and race), comorbidities present amongst patients at baseline (hypertension, diabetes mellitus, hyperlipidemia, renal disease, and respiratory illness) and primary end-points including mortality, hospital admission, ICU admission, and mechanical ventilation. Two independent reviewers (E.W and A.M) conducted Quality Assessment to determine the risk of bias in each study. Newcastle-Ottawa scale (9) was employed for the pooled retrospective cohort studies. Studies were gauged on the selection, comparability, and outcomes of the representative cohorts. Those displaying most representativeness of the community, had least attrition loss, longer follow-up duration, and a standardized method to assess the outcome were attributed as having lower risk of bias. Any discrepancies found were resolved by consensus and discussion. (Table S1) Additionally, GRADE (10), an assessment tool to gauge the quality of systematic reviews/meta-analyses was employed. Studies included in the analysis are evaluated on the basis of risk of bias, inconsistency, indirectness, imprecision, and probability of a bias in publication. A detailed summary is provided in Table 2.

Statistical Analysis
This analysis was performed using Review Manager (RevMan) Version 5.4 Cochrane Collaboration. Generic invariance and the random-effects model were employed to derive odds ratios (ORs) and their corresponding 95% confidence intervals (CIs) for the pooled outcomes.
Forest plots were used to represent each outcome. Higgins I 2 statistics were utilized to assess the heterogeneity across the pooled studies. The value of I2=25%-50% was considered mild, 50%-75% moderate, and >75% severe heterogeneity. A p-value <0.05 was considered significant.
Since there were less than 10 studies, the funnel plot could not be used to evaluate the publication bias.

Study selection and characteristics
Our search of electronic databases yielded a total of 275 articles. After an initial screening of abstracts, 93 studies were assessed for detailed evaluation. Of these studies, a total of 9 studies (5,10-17) satisfied our eligibility criteria and were included in the quantitative synthesis, as shown in Figure 1. The remaining studies shortlisted for full-length assessment were excluded on being either single-arm (n=38) or having incomplete data (n=34). All nine studies included in the meta-analysis were retrospective cohorts with a total patient population of n=1,130,341. Most studies were peer reviewed except for Bramante et al. (12), were propensity matched cohorts, conducted analysis between 2020-2021, and contained patients tested positive for COVID through PCR. However, a few didn't specify the screening method and included COVID susceptible cases as well (10)(11). Most were conducted in the US, however, a few were based on the Iranian (10) and French (11) databases. Except for Purdy and Bramante et.al (5,13), studies specified the nature of the surgical technique, with majority including Roux-en-Y gastric bypass (11)(12)(14)(15)(16)(17)(18), sleeve gastrectomy (11)(12)(14)(15)(16)(17)(18), and gastric banding (12,14,17). Among the included studies, females were the predominant sex in both bariatric and non-bariatric groups, with diabetes and hypertension being the common comorbidities. Characteristics of all included studies are present in Table. 1.

Mortality (Figure 2)
A random-effect analysis of eight studies reporting mortality was performed by pooling odds ratios (ORs) from dichotomous data. Compared with non-bariatric, the bariatric surgery group

Sensitivity Analysis
A leave-one-out sensitivity analysis was carried out to measure the degree of disproportionate effects produced by one single study. Purdy et al. (5) was identified as the study heavily influencing mortality and severe COVID-19 infection outcomes due to differences in the number and demographics of BMS and non-BMS groups. Despite the large population size, the methodology remains substandard due to the lack of individual patient data and multivariate analysis. Removal of Purdy et al. resulted in a decrease in heterogeneity from 73% to 0% in mortality and from 83% to 0% in severe COVID-19 infection. Both forest plots are present in supplementary figures 2 and 3.

Quality Assessment
From the nine retrospective cohort studies included in this meta-analysis, one study, Ianelli et al. (11), was classified as of "Fair" quality, whereas the rest were of "Good" quality. Detailed quality assessment is included in the supplementary material (Table S2). GRADE rated all three outcomes as having moderate to critical importance and attributed the evidence as highly certain.

DISCUSSION:
This study is an updated meta-analysis that further confirms the protective function of bariatric surgery in patients infected with SARS-COV2, as stated by Aminian et al. (6). However, our meta-analysis consists of a larger population (n= 1,130,341 vs. 9022) and additionally evaluates the severity of infection. Patients admitted to the ICU and/or those who underwent mechanical ventilation were labeled with severe COVID-19 infection. Bariatric surgery was found to be unanimously beneficial; linked with decreased rates of hospitalization, less severe infection, and a more significant mortality benefit.
In patients infected with SARS-COV2, obesity has been attributed as the single most prognosis determining factor. According to a meta-analysis, individuals with obesity have an almost 50% higher risk of contracting SARS-COV2, with every patient getting hospitalized and every 3 in 4 at risk for ICU admission. Consequently, dismal outcomes such as the need for mechanical ventilation and ultimately death were observed in every 4 and 3 out of 6 individuals, respectively (19). Moreover, intensive care patients, those having a BMI of >35 kg/m 2 are reported to be at a greater risk for disease progression (3). Hypothetically, weight loss measures in general may ameliorate the severity of the disease and fasten recovery though surgical measures supersede pharmacological treatment.
Patients with bariatric surgery are 44% less likely to witness a severe event such as a mechanical event and/or ICU admission. The mortality rate in the surgical group is 42% less than in the nonsurgical group, which is higher (22 %) than that reported by the previous meta-analysis (6).
Similarly, the rate of inpatient admissions in the surgical group was lower as compared to the previous meta-analysis [ Furthermore, additional studies stratifying the outcomes of bariatric surgery on the basis of demographics and metabolic profile must be conducted to get a more holistic view.
Like any other meta-analysis, this too has its inherent limitations. Firstly, as very few studies have reported their metabolic profile, subgroup analysis could not be performed. Secondly, although the majority of the data being propensity-matched eliminates the risk of heterogeneity I2, findings by Purdy et al. (5) render the data inhomogeneous (severe COVID-19 infection I2 =83%, mortality= 73%). Other than that, the implicit bias of selective outcomes and data reporting cannot be gauged and eliminated as such. One example is the study by the University of Minnesota (13), which did not contribute to any heterogeneity, is not peer-reviewed, and hence, the findings may introduce bias. As seen in our analysis, females were more likely to undergo surgery than their male counterparts. However, multivariate regression revealed that the male gender presents with more severe disease, leading to increased morbid outcomes (39). Due to the lack of individual outcomes in both genders, subgroup analysis could not be performed.
Hence, the surgical benefit in COVID-19 patients could not be stratified based on gender.
Although a few propensity-matched studies did not match BMI, race, and age, this did not affect the results as these factors were not included in our subgroup analysis. However, future metaanalyses may need to be wary of these confounders. Furthermore, the temporal relationship between surgery and contracting SARSCOV2 could not be evaluated due to the lack of data. In addition, assessment made by GRADE is rendered subjective, and recommendations made on the evidence provided should be dealt with caution. Although the findings of this meta-analysis lend credibility to the previous, future meta-analyses must evaluate for the missing endpoints.

Conclusion:
Bariatric surgery assists in alleviating the disease progression in patients with SARS-COV2.
Decreased hospitalization, reduced ICU admissions, and lower mortality rates are positive outcomes in moderate to severe obesity patients. Future studies are needed to evaluate the endpoints points, highlight the confounders, and devise strategies for continual surgical access during pandemics.
Funding: This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Disclosure:
The authors report no proprietary or commercial interest in any product mentioned or concept discussed in this article.