Association between multimorbidity, intensive care unit admission, and death in patients with COVID-19 in Brazil: a cross-section study, 2020

ABSTRACT BACKGROUND: Multimorbidity can influence intensive care unit (ICU) admissions and deaths due to coronavirus disease (COVID-19). OBJECTIVE: To analyze the association between multimorbidity, ICU admissions, and deaths due to COVID-19 in Brazil. DESIGN AND SETTING: This cross-sectional study was conducted using data from patients with severe acute respiratory syndrome (SARS) due to COVID-19 recorded in the Influenza Epidemiological Surveillance Information System (SIVEP-Gripe) in 2020. METHODS: Descriptive and stratified analyses of multimorbidity were performed based on sociodemographic, ventilatory support, and diagnostic variables. Poisson regression was used to estimate the prevalence ratios. RESULTS: We identified 671,593 cases of SARS caused by COVID-19, of which 62.4% had at least one morbidity. Multimorbidity was associated with male sex, age 60–70 and ≥ 80 years, brown and black skin color, elementary education and high school, ventilatory support, and altered radiologic exams. Moreover, all regions of the country and altered computed tomography due to COVID-19 or other diseases were associated with death; only the northeast region and higher education were associated with ICU admission. CONCLUSION: Our results showed an association between multimorbidity, ICU admission, and death in COVID-19 patients in Brazil.


INTRODUCTION
The coronavirus disease , caused by severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2), rapidly spread worldwide, causing approximately 185 million cases and more than 4 million deaths between December 31, 2019, and June 30, 2021. 1 In Brazil, COVID-19 cases that progress to severe acute respiratory syndrome (SARS), leading to hospitalizations and deaths, are monitored using clinical samples analyzed in reference laboratories. Case notification is mandatory, and records are stored in the Influenza Epidemiological Surveillance Information System (SIVEP-Gripe) from the SARS Surveillance network, initially implemented to monitor the influenza epidemic in 2000. 2 Since the emergence of COVID-19, scientific literature has addressed the virological characteristics of SARS-Cov-2 and clinical complications arising from its infection in different populations. Although severity is high in older individuals and males, some studies have shown a relationship between COVID-19 and pre-existing morbidities 3,4 (e.g., cardiovascular diseases), [5][6][7] which are associated with increased intensive care unit (ICU) admissions and deaths.
Studies have also shown an association between morbidity and COVID-19; however, only a few have investigated multimorbidity (i.e., the co-occurrence of two or more chronic diseases for a specific period 8 ) as a factor predisposing patients to ICU admission and death. 5,9 Brazil had the highest number of COVID-19 cases in Latin America and currently it also has a high prevalence of diabetes, hypertension, and cardiovascular diseases. [5][6][7] Therefore, studies on association between multimorbidity, and ICU admissions and deaths due to COVID-19, are needed to provide basic knowledge for more complex studies establishing multicausality. [5][6][7] Therefore, this study aimed to analyze the association between multimorbidity, ICU admission, and death due to COVID- 19

Variables
Outcome variables were ICU admission (yes or no), and death due to COVID-19, which were based on the progression of cases to "yes" (death due to  or "no" (cure or death due to other causes) answers.
The independent variable, multimorbidity, was addressed using • Computed tomography (CT) was categorized as negative or positive for COVID-19 or other diseases. We did not assess the "not performed" category for radiologic examinations and CT.

Statistical analysis
R software 4.0.4 (R Foundation, Vienna, Austria) 11 was used to analyze the data. The absolute and relative frequencies were calculated for each morbidity and outcome.
We calculated the number of morbidities and estimated the prevalence (P%), prevalence ratio (PR), and 95% confidence interval (95% CI) for ICU admission and death due to COVID-19.
The association between multimorbidity and outcomes was investigated using raw (number of morbidities) and stratified (multimorbidity) analyses, according to sociodemographic, ventilatory support, and diagnostic variables.
Hierarchical adjusted analysis, associated multimorbidity and sociodemographic, ventilatory support, and diagnostic variables, with ICU admission and death. Three blocks were considered: country region, sociodemographic, and support and diagnostic variables.
Poisson model with robust variance was used to estimate PR and 95% CI since outcomes of interest had prevalence of > 10%. 12 We selected variables using bivariate analysis between outcomes and region, sociodemographic, ventilatory support, and diagnosis variables; P ≤ 0.20 was set as cutoff point for initial model selection.
The model was adjusted to retain variables with the lowest Akaike information criterion values and theoretical criteria. We then assessed the influential point (i.e., absolute value of standardized errors > 3) and collinearity between predictor variables (i.e., positive variables with values > 10). The Hosmer-Lemeshow test determined the goodness of fit of the final model, considering a good fit when P ≥ 0.05.

Ethical aspects
This study used anonymous information from the public domain. Thus, authorization for data collection and approval by the research ethics committee were not required. had at least one morbidity, and 97.0% with up to three morbidities were hospitalized due to SARS. Table 1 shows the frequency distribution of morbidities according to ICU admission and mortality. The frequency of morbidities ranged from 34.1% (systemic arterial hypertension) to 47.1% (kidney diseases) in patients admitted to the ICU, and from 16.0%
We observed that 29.5% (57,331) of the patients admitted to the ICU and 25.2% (57,359) of the patients who died had no morbidities. The prevalence and prevalence ratio of ICU admissions and deaths increased with an increase in number of morbidities ( Table 2).    Table 3).
Regarding the associations between multimorbidity and outcomes according to support and diagnostic variables, the prevalence of ventilatory support was high in patients admitted to the ICU (52.7%) and those who died (51.1%). We also found that a high prevalence according to imaging tests; altered radiologic exams were associated with ICU admission (49.3%) and death (48.6%), while CT positivity for COVID-19 or other diseases was associated with ICU admission (50.3%) and death (41.5%) ( Table 4).
The hierarchical adjusted analysis ( We did not find associations between the three Brazilian regions and positive CT findings for COVID-19 or other diseases and ICU admission, or between higher education and death. Collinearity was not observed between variables. The most influential point was no lower than 0.005. Furthermore, the goodness-of-fit test indicated a good fit in both ICU admission (P = 0.358) and death (P = 0.105).

DISCUSSION
We aimed to analyze the association between multimorbidity, ICU admission, and death due to COVID-19 in Brazil. We found associations between multimorbidity, male sex, black skin color, ventilatory support, and altered radiologic exams.
The high percentage of morbidities in the studied population was expected and corroborated the literature 13 since individuals, with some morbidity and COVID-19, are more likely to be admitted to the ICU or they may expire.
The frequency of morbidities analyzed in this study (e.g., diabetes mellitus, systemic arterial hypertension, obesity, and cardiac diseases) was higher than those in the literature, 13,14 even compared to a study conducted in the Brazilian population. 15 We also obtained more robust results due to the size and national scope of the SIVEP-Gripe database, which is different from previous studies. [13][14][15] The simultaneous effects of morbidities explain the increase in hospitalizations and deaths due to COVID-19. Therefore, assessing conditions that led to ICU admission or death when two or three additional conditions (e.g., hyperglycemia, dyslipidemia, or arterial hypertension) were considered to classify this syndrome. 16,17 The P% and PR of ICU admission and death due to COVID-19 increased with increase in the number of morbidities. This result was expected; 17 however, the increase was significant in the presence of two or three morbidities. These data indicate a worse prognosis for patients with COVID-19 and multimorbidity, raising concerns for health services due to the high costs and increased demand of the health care personnel and technological support.
Analysis by age groups suggested that younger individuals were less affected by COVID-19 than adults and older individuals. 14,18 We also found associations between age group and ICU admission or death in patients with multimorbidity. Age is an essential factor to assess the time to ICU admission or death due to COVID-19. 19 Finally, the SARS notification form did not inform whether deaths were caused during the disease or later due to post-disease complications. Similarly, the length of stay in the ICU may be a relevant factor in the assessment of cases.
In this study we highlight the assessment performed with the morbidities and outcomes, since it may be more expressive when considering isolated, dyad, and triad morbidities.
From the present study, we concluded that the prevalence of ICU admission and death was high in patients with morbidities, and that the increment in number of morbidities increased the prevalence and prevalence ratio of outcomes. An association between multimorbidity and ICU admissions due to COVID-19 was observed when adjusted for male sex, black and brown skin colors, age between 18 and 40 years, patients with some degree of education, use of ventilatory support, and altered radiological examinations. Regarding deaths due to COVID-19, multimorbidity was associated with male sex, black and brown skin colors, age ≥ 60 years, ventilatory support, altered radiologic exams, and CT findings indicating COVID-19 or other diseases.
Our findings may help train healthcare personnel to offer specialized care to patients with morbidities and COVID-19.
Furthermore, we expect competent healthcare groups in the three spheres of the government to disseminate knowledge about multimorbidity and COVID-19 to reduce the spread of the disease and its impact on the healthcare system.