Albumin level as an independent predictive factor for adverse outcomes in COVID-19 patients: a retrospective cohort study

Introduction: Research on the association between albumin (ALB) level and clinical outcomes of coronavirus disease 2019 (COVID-19) are limited. This study aimed to investigate the relationship between albumin level at the time of admission and adverse outcomes in patients with COVID-19. Methodology: This was a retrospective cohort study with 199 COVID-19 patients from five designated hospitals in Fujian Province who were enrolled between 22 January and 27 February, 2020. Clinical characteristics and laboratory values at the time of admission were collected. Adverse outcomes were defined as meeting at least one of the following criteria: development of acute respiratory distress syndrome (ARDS), respiratory failure, shock, multiple organ failure (MOF), intensive care unit (ICU) admission and in-hospital mortality event. The univariate and multivariate linear regression models and generalized additive models (GAM) were used to analyze the relationship between ALB and adverse outcomes. Results: A non-linear relationship with an inflection point of 32.6g/L was detected between ALB and adverse outcomes after adjusting for potential confounders. The odds ratio and the confidence intervals on the left and right sides of the inflection point were 0.204 (0.061-0.681) and 0.908 (0.686-1.203), respectively. This suggested that ALB was negatively correlated with adverse outcomes when ALB was less than 32.6 g/L, and for every 1 unit increase in ALB, the risk of adverse outcomes was reduced by 79.6%. Conclusions: The relationship between ALB and adverse outcomes of COVID-19 is non-linear. ALB level is an independent predictive factor for adverse outcomes in COVID-19 patients.


Introduction
Coronavirus disease 2019 (COVID-19) is a global pandemic and has been declared a public health emergency of international concern by the World Health Organization (WHO) [1]. Severe acute respiratory syndrome coronavirus 2 (SARS-COV-2) can affect multiple organs, leading to adverse outcomes, such as cerebrovascular events [2,3], myocardial infarction [4], pulmonary embolism [5], acute liver injury, and death [6,7]. In order to improve the management of patients, it is critical to identify predictors of adverse outcomes.
Albumin is a protein synthesized in the liver, and performs a variety of physiological roles in the body, such as providing oncotic pressure, binding and transporting materials, and maintaining acid base balance [8,9]. When the body is in critical condition, inflammatory mediators reduce albumin synthesis to preferentially synthesize other proteins required for acute phase response. Additionally, these mediators increase vascular permeability and cause albumin to escape into interstitial space and lead to hypoalbuminemia [10]. Previous studies have suggested an association between hypoalbuminemia and poor outcomes in critical patients [11], but the relationship between the albumin level and catastrophic events in COVID-19 remains unclear.
We explored the relationship between initial admission albumin concentration and adverse outcomes in COVID-19 patients, with the hypothesis that albumin level can be an independent predictive factor for adverse outcomes in COVID-19 patients. This can help clinicians identify high-risk patients early and take timely preventive measures.

Study design and participants
A total of 199 COVID-19 patients (age range: 16-93 years) who were admitted to five hospitals in the Fujian Province, including Fuzhou, Zhangzhou, Xiamen, Putian and Quanzhou, from 22 January to 27 February, 2020, were consecutively enrolled for this retrospective study. The clinical outcomes, discharge from hospital or death, were recorded up to March 3, 2020. The hospitals are designated for COVID-19 treatment by the government. All the patients were diagnosed with COVID-19 through real-time reverse transcriptase-polymerase chain reaction (RT-PCR) assay, according to the Guideline for Diagnosis and Treatment for Novel Coronavirus Pneumonia released by the National Health Commission of China (5th edition) [12]. This study was approved by the ethical committee in Zhongshan Hospital (Xiamen), Fudan University (B2020-003). The requirement for informed consent was waived because the data were urgently collected and analyzed anonymously.

Data collection
A team of professional physicians retrospectively reviewed the electronic medical records, comprising of clinical notes and laboratory values for our cohort of patients. We recorded the demographic data including age, gender, body mass index (BMI), and smoking habit, the comorbidities including hypertension, diabetes, cardiovascular disease, chronic kidney disease, pulmonary disease, tumor, and chronic liver disease, and the symptoms. In the case of laboratory values, initial values on the day of admission were collected. The outcomes were recorded, including development of acute respiratory distress syndrome (ARDS), respiratory failure, shock, multiple organ failure (MOF), intensive care unit (ICU) admissions, inhospital mortality events, and total adverse outcomes (sum of the adverse outcomes detailed above).

Statistical analysis
Adverse outcomes were defined as meeting at least one of the following criteria: development of ARDS, respiratory failure, shock, MOF, ICU admissions and in-hospital mortality events. Statistical analysis was performed in five steps. First, we analyzed the baseline characteristics of participants in accordance with the following principles and we grouped albumin (ALB) in tertiles: (1) continuous variables were expressed as the means ± standard deviations (normal distribution) or medians (interquartile) (skewed distribution), and categorical variables were expressed as frequency (percentage); and (2) the one-way ANOVA (normal distribution), Kruskal-Wallis H (skewed distribution) test and Chi square test (categorical variables) were used to analyze any significant differences between the means and proportions of the groups. Second, we used univariate and multivariate Cox regression analyses to assess relationships between ALB and adverse outcomes risk. The covariances, including age, gender, body mass index (BMI), current smoking, hypertension, cardiovascular disease, pulmonary disease, tumor, chronic liver disease, D-dimer, creatinine, creatine kinase (CK), leukocytes, neutrophil, lactate dehydrogenase (LDH), blood urea nitrogen (BUN), and fibrinogen, when added to this model changed the matched odds ratio by at least 10% and were adjusted. Third, generalized additive models (GAM) were used to identify non-linear relationships because ALB was a continuous variable. If a non-linear correlation was observed, a two piecewise linear regression model was used to calculate the threshold effect of ALB on adverse outcomes in terms of the smoothing plot. When the ratio between adverse outcomes and ALB appeared obvious in a smoothed curve, the recursive method automatically calculates the inflection point, where the maximum model likelihood will be used. The statistical analysis of this part was carried out using "mgcv" [13] package of R software. Fourth, subgroup analysis of the association between ALB and adverse outcomes was performed using stratified linear regression models. The modifications and interactions of subgroups were examined by likelihood ratio tests. All the analyses were performed with the statistical software package R (http://www.R-project.org, The R Foundation) and EmpowerStats (http://www.empowerstats.com, X&Y Solutions, Inc., Boston, MA). p values less than 0.05 (two-sided) were considered statistically significant.

Results
The average age of the cohort was 46.3 ± 16.4 years, and approximately 52.8% of the patients were male. We compared the baseline demographic, clinical, and biochemical characteristics of the patients by tertiles of ALB (Table 1). Patients in the lowest tertile were older, and more likely to have pulmonary disease compared to the subjects in the highest tertile of ALB. Participants with chronic liver disease were mainly in the middle tertile of ALB. Patients in the highest tertile group had significantly higher lymphocyte and hemoglobin, and lower prothrombin time (PT) and D-dimer compared to the lowest tertile group.
The univariate logistic regression models between baseline variables and adverse outcomes are presented in Table 2. The results of univariate analysis showed that age, hypertension, cardiovascular disease, pulmonary disease, tumor, fatigue, leukocytes, neutrophil, aspartate transferase (AST), LDH, creatinine, BUN, CK, fibrinogen and D-dimer were positively correlated with the risk of adverse outcomes. The ALB (OR = 0.850, 95% CI, 0.776 -0.931, p = 0.00045) was negatively correlated with the risk of adverse outcomes.
Univariate linear regression models were used to evaluate the associations between ALB and adverse outcomes. We have presented the results of multivariate Cox regression analyses in Table 3. After adjusting potential confounders, we found that ALB was negatively correlated with adverse outcomes (OR = 0.70, 95% CI, 0.56 -0.88, p = 0.002). We also analyzed ALB as a categorical variable (tertiles) for the purpose of sensitivity analysis, and found the same trend (p = 0.0303).  In the present study, we analyzed the non-linear relationship between ALB and adverse outcomes because ALB is a continuous variable (Figure 1). We found that the relationship between ALB and adverse outcomes was non-linear (after adjusting for age, gender, BMI, smoking status, hypertension, cardiovascular disease, pulmonary disease, tumor, chronic liver disease, D-dimer, creatinine, CK, leukocytes, neutrophil, LDH, BUN and fibrinogen). By using a two-piecewise linear regression model, we calculated the inflection point as 32.6. On the right of the inflection point, the odds ratio, 95% CI and p value were 0.908, 0.686 to 1.203 and 0.5032, respectively. However, we observed a negative relationship between ALB and adverse outcomes on the left side of the inflection point (OR = 0.204, 95% CI, 0.061 -0.681, p = 0.0097) ( Table 4). This suggests that when ALB was less than 32.6 g/L, for every 1 unit increase in ALB, the risk of adverse outcomes was reduced by 79.6%.
The results of subgroup analysis indicated that interaction effect of age was significant (p = 0.0113), while the test for interactions had no significant differences for gender, BMI, hypertension, diabetes, pulmonary disease and chronic liver disease (p values for interactions were > 0.05; cardiovascular disease, chronic kidney disease and tumor were not included in subgroup analysis because the sample size in the subgroup was less than 10) (Figure 2).

Discussion
We used generalized linear model (GLM) and GAM models to elucidate the relationship between ALB and adverse outcomes among participants. As is shown in multivariate Cox regression analyses, ALB was negatively correlated with the risk of adverse outcomes. This relationship seems to diminish with age. The same trend was observed when we analyzed ALB as a categorical variable. However, the results obtained from GAM and two-piecewise linear regression model showed that the relationship between ALB and adverse outcomes was non-linear, and the correlations between ALB and adverse outcomes were different on the left and right sides of the inflection point (ALB = 32.6). ALB, as assessed at baseline, was not statistically significant on the right side of the inflection point, but it was negatively associated with adverse outcomes on the left of the inflection point. It means that ALB is negatively correlated with adverse outcomes when  The area between two dotted lines represents 95% CI. Each point indicates the magnitude of the ALB and is connected to form a continuous line. The magnitude of the ALB was not correlated with adverse outcomes when it was > 32.6 g/L. Conversely, the magnitude of the ALB showed a significant correlation with adverse outcomes when it was ≤ 32.6 g/L. The risk of adverse outcomes decreased as the ALB increased. ALB: Albumin. ALB is less than 32.6 g/L, and for every 1 unit increase in ALB, the risk of adverse outcomes was reduced by 79.6%. Several studies have indicated a decrease in serum albumin concentrations in patients affected by COVID-19 [14,15]. Hypoalbuminemia has a negative impact on clinical outcomes in COVID-19 patients [16][17][18][19]. Hypoalbuminemia has been identified as an independent predictor of mortality in COVID-19 patients [17]. However, our study differs from the previous studies in that our patients were from five different hospitals. Furthermore, our study has a few strengths. First, we not only used GLM to evaluate the linear relationship between ALB and adverse outcomes, but also used GAM to clarify their non-linear relationship. GAM has advantages in analyzing nonlinear relations; it can handle non-parametric smoothing and will fit a regression spline to the data. The use of GAM helped us better discover the real relationship between exposures and outcomes. Second, this study is an observational study, including unavoidable potential confounders; therefore, we used strict statistical adjustment to minimize residual confounding. Third, we had the positive finding that when ALB was less than 32.6 g/L (32.6, per 1 change in the text), for every 1 unit increase in ALB, the risk of adverse outcomes was reduced by 79.6%. The clinical value of this finding is that the association of ALB and adverse outcomes can only be observed when ALB levels do not reach a certain threshold (ALB = 32.6g/L), which can guide clinical practice directly.
The relationship between hypoalbuminemia and more adverse outcomes may have several explanations. First, as an anti-inflammatory and antioxidant protein, albumin plays a crucial role in scavenging free oxygen radicals, which can cause tissue ischemia-reperfusion injury and even an intense systemic inflammatory response [20,21]. Previous studies indicated that albumin concentrations were inversely correlated with white blood cell, neutrophil-to-lymphocyte ratio, Creactive protein and interleukin-6, which suggested that hypoalbuminemia might be due to the systemic inflammatory state in COVID-19 [17,22]. It is well known that inflammation may be responsible for the extravasation of serum albumin into the interstitial space due to an expanded capillary vascular permeability, with an increased volume distribution of albumin [10]. In this context, the role of albumin in scavenging oxygen free radicals is not potent enough to protect against the cytokine storm and the ensuing organ failure. Besides, albumin not only has anticoagulant properties, but also inhibits oxidative stress-related coagulation and platelet activation [23,24]. Therefore, the negative impact of hypoalbuminemia on coagulation activation may be associated with a higher risk of COVID-19 adverse outcomes. Hence, in addition to prior known biomarkers, such as procalcitonin, C-reactive protein, lymphocyte count, D-dimer, troponin I, aspartate transaminase and alanine transaminase, associated with severe COVID-19 [25], serum albumin level might help in prognostic risk stratification.
There are some limitations in our study. First, the study population may not be large enough and some bias may have occurred. Second, the subgroup analysis was not adjusted for potential confounding variables because of the limited number of positive events. Third, we showed only the predictive value of baseline albumin level for outcome of COVID-19, yet the changes in albumin level during the evolution of COVID-19 were not reflected in our data set, and whether the dynamic changes of albumin level are more predictive of adverse outcomes remains unknown. Fourth, this dataset only included the clinical data at the beginning of the pandemic, and the analysis period was short, which limits generalizations and further observations, especially since new strains emerged later on.

Conclusions
We demonstrated that the relationship between ALB and adverse outcomes of COVID-19 is non-linear. ALB is negatively correlated with adverse outcomes when ALB is less than 32.6 g/L. These findings further expand the potential role of ALB as a prognostic predictor in COVID-19, suggesting that for COVID-19 patients who had admission ALB lower than 32.6g/L, the risk of adverse outcomes was much higher, and they need more intensive and active treatment.