Although most published studies refer to obesity as one of the independent predictors of disease severity and worse outcome in hospitalized COVID-19 patients (1–4), the results regarding mortality are still inconsistent. While some meta-analysis authors observed no significant relationship between obesity and COVID-19 mortality (5, 6), others emphasize such a relationship exists only in younger patients and those with fewer comorbidities (7, 8).

Potential mechanisms by which obesity adversely affects the course of SARS-CoV-2 infection are chronic inflammation and immune response dysregulation, endothelial dysfunction, increased thrombogenic potential, endocrine dysfunction, and the simultaneous presence of other known risk factors (such as cardiovascular disease, metabolic syndrome, and diabetes mellitus) (2, 4, 9).

Given that most of these pathophysiological mechanisms are the effect of adipose tissue (dominantly visceral), the main limitation of published studies is that body composition was assessed solely based on body mass index (BMI), without insight into the presence and distribution of adipose tissue. Moreover, several studies in which abdominal adipose tissue had been assessed using CT scan emphasized the importance of visceral adipose tissue, rather than subcutaneous, on COVID-19 severity and worse outcome (10).

For the reasons stated, it is valuable to examine the impact of body and visceral fat on the course and outcome of the novel coronavirus infection and their correlation with other significant predictors of disease severity, primarily inflammatory biomarkers. In addition, bioelectrical impedance analysis (BIA) measurements could be more precise than BMI in predicting the risk of mortality and worse outcome in hospitalized COVID-19 patients. To the best of the authors’ knowledge, this is the first study regarding BIA measurements and COVID-19 outcomes.

The research was conducted on 216 patients with COVID-19 consecutively hospitalized at our Center between October and December 2021, in a period of the presumed predominance of SARS-CoV-2 delta variant in our country. The majority of hospitalized COVID-19 patients in our cohort had disturbed body composition, with only two out of ten hospitalized patients having normal BMI and % BF, and 14.4% of patients having normal VF levels. The shown disturbances of body composition in hospitalized COVID-19 patients are not unexpected. First, obesity is a globally raging pandemic whose consequences are also noticeable in Serbia. According to a WHO report from 2013, 58.6% of the adult population in Serbia were overweight or obese, and, according to a model at the time, the predicament was that the obesity prevalence in 2020 would be 44% in adult men and 31% in adult women (15). Second, several studies have demonstrated that obesity is a significant risk factor for hospital admission (3, 6, 16), therefore it is somewhat expected to have a high prevalence of obesity among hospitalized COVID-19 patients. We must note that only one patient (man, 70 years old) was underweight according to BMI and % BF; therefore, he was neglected in further analysis. Although some studies have shown an increased risk for death and the need for mechanical ventilation in underweight patients (17).

Regarding age categories, patients younger than 65 years had a higher prevalence of obesity according to BMI and % BF measurements. In contrast, older patients had significantly higher VF levels. We accentuate that the high VF levels are associated with numerous health disorders and general mortality (18, 19), along with worse outcomes and death in patients with COVID-19 (10, 20–23). Therefore, excessive VF levels could be one of the links associated with increased mortality and severity in older patients with COVID-19, among others (7, 16, 24, 25). Moreover, some studies advocate obesity as a risk factor for COVID-19 mortality and severity dominantly for younger patients, with weaker or no impact at all in older patients (5, 26, 27). Perhaps this could be a misconclusion, for cited studies have used BMI alone as a tool for accessing obesity, neglecting the significance of VF (7, 16, 24, 25).

In the initial analysis, only % BF and VF level were significant predictors of mortality, while BMI, although borderline, lacked statistical significance (Table 3). However, after adjusting OR for age, sex, days from disease onset, and obesity-related comorbidities (diabetes mellitus), all three anthropometric methods were significant predictors of mortality, with both % BF and VF having higher aOR values compared to BMI (aOR 3.353, aOR 3.05, and aOR 2.387, respectively). Results regarding ICU admission and experiencing either primary end-point were more concordant, where all three anthropometric measurement methods had significant predictive importance, with % BF (aOR 7.411 and 6.085, respectively) and VF (aOR 3.424 and 3.208, respectively) again having higher aOR values compared to BMI (aOR 3.113 and 2.769, respectively).

We must note that comorbidities selection for OR adjustment was arbitrary, and diabetes mellitus was chosen as a known obesity-related condition. Furthermore, a different selection of “adjusting” variables in a model (in addition to age and gender) did not significantly alter the aOR and “p” values of either anthropometric measurement. In a sensitivity analysis (Supplementary Table 3), conducted by implementing all socio-demographic and medical history data in a model, all three anthropometric measurements remained significant predictors of primary end-points, with % BF having an increase of aOR at the expense of a wider confidence interval range. In addition to anthropometric parameters, the model showed a significant predictive value of the Charlson comorbidity index for mortality and female sex for ICU admission.

Literature data agree that obesity, defined by BMI, is a significant predictor of disease severity (OR 1.47–5.47) (3–6, 28, 29), need for intensive care unit (OR 1.29–5.49) (3, 5, 6), and invasive mechanical ventilation (OR 1.2–6.01) (3, 5, 6, 17). However, the results regarding mortality are still inconsistent. Although some studies advocate obesity, defined by BMI, as a significant predictor of mortality, with OR ranging from 1.04 to 4.4, or even higher (1, 3, 4, 26, 30–32), others failed to show statistical significance or even showed negative predictive values (3, 5, 6, 16, 33, 34).

A relatively wide range of OR values in these studies, as well as lack of statistical significance for mortality, could be explained by different BMI cut-offs, sample size, diversity of study population (regarding diverse socio-demographic and comorbidity characteristics of the cohort, as well as different COVID-19 severity among patients), the predominance of different SARS-CoV-2 mutation variants, and other. Also, it is important to point out that all cited studies used BMI as the only measurement for defining obesity, possibly leading to misinterpretation of body composition by neglecting total body and visceral fat, especially in older and more comorbid patients (2). This is important because the majority of mechanisms by which obesity adversely affects the course of SARS-CoV-2 infection (chronic inflammation and immune response dysregulation, endothelial dysfunction and increased thrombogenic potential, endocrine dysfunction, etc.) are mostly effects of the adipose tissue (2, 4, 9). In addition, several studies in which abdominal adipose tissue had been evaluated using CT scan emphasized the importance of visceral adipose tissue on COVID-19 severity and mortality (10, 20–23). One of the pathophysiological explanations of this phenomenon lies in the fact that visceral adipose tissue, compared to subcutaneous, secretes 2–3 times higher concentrations of interleukin 6 (35), which is associated with the development of severe forms and fatal outcomes for patients with COVID-19 (1, 7). In our cohort, patients with excessive VF had significantly higher serum levels of CRP (p = 0.014) and IL-6 (p = 0.028) on admission compared to those with normal VF levels, possibly suggesting higher inflammation grade. We also note that patients with very high % BF had significantly higher values of LDH (p = 0.024), another notable COVID-19 predictor (1, 7), while no statistically significant relationship was found between BMI and any proinflammatory marker on admission. Finally, despite a good agreement between BMI and % BF in defining obesity, 24.5% of patients with very high % BF values were categorized as non-obese according to BMI. We accentuate the high incidence of mortality and ICU admission in this group of patients (24.8 and 55.6%, respectively).

All stated mechanisms could explain, at least partially, why BMI lacked statistical significance in terms of predicting mortality of patients with COVID-19 in cited studies. Also, stated mechanisms could explain why BIA measurements, both % BF and VF, had higher OR in predicting each primary end-point compared to BMI. Due to the relatively small sample size and other study limitations, perhaps the exact OR values for anthropometric measurements could not be generalized, particularly in terms of mortality. However, the results are suggestive of a link between obesity and COVID-19 severity and mortality.

Obesity is a globally raging pandemic that is, in addition to many other comorbidities and all-cause mortality, a significant predictor of COVID-19 severity and death. For that reason, intensive obesity prevention campaigns and programs should be one of the main focuses of healthcare systems worldwide.

Bioelectrical impedance analysis measurements could be a helpful tool in predicting COVID-19 severity and mortality on admission.

By having insight into the total body and visceral fat distribution, BIA measurements (both % BF and VF) were stronger predictors of each primary end-point (mortality and ICU admission) compared to BMI.

Our study had several limitations. First, COVID-19, like infection and inflammation, can impact body composition. To minimize that effect, we have measured patients in the initial days of hospitalization. Second, a substantial number of patients were excluded from the study, because of their inability to undergo BIA assessment (such as poor general condition, dementia, and lack of limbs). Third, we have included a relatively small number of patients for the generalization of the results.

Finally, although BIA measurements have satisfactory insight into total body fat and fat-free mass and are widely used for body composition assessment in the general population, this method has difficulty distinguishing visceral from abdominal fat, for which CT and MRI remain the gold standard (36, 37). Due to stated study limitations and presented results, the authors suggest and encourage continuing research on this issue.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by the University Clinical Center Kragujevac, Kragujevac, Serbia. The patients/participants provided their written informed consent to participate in this study.

DS, VZ, MiP, VM, and IC contributed to conception and design of the study. VZ, MaP, IC, and VM ensured quality of the research and performed final review of the manuscript. DS and MiP organized the database. SM and NZ performed the statistical analysis and contributed to presentation of results. DT, AD, MM, DB, AA, and AJ contributed to data collection and helped in writing chapters of the manuscript. DS and MiP wrote the first draft of the manuscript. All authors contributed to manuscript revision, read, and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.