Keywords

22.1 Introduction

In December 2019, pneumonia cases with unknown etiology emerged in Wuhan, China (Lu et al.). Lower respiratory tract sampling and sequencing process could reveal a novel coronavirus from the family of betacoronavirus, which was subsequently named 2019 novel coronavirus (2019-nCoV) (Zhu et al. 2020). This family also included severe acute respiratory syndrome (SARS) and the Middle East respiratory syndrome (MERS), which have caused outbreaks over the last two decades (Ksiazek et al. 2003; de Wit et al. 2016; Jabbari et al. 2020). Coronavirus disease 2019 (COVID-19) has rapidly spread throughout the world, posing a global emergency (Hanaei and Rezaei 2020; Rezaei 2020a).

The Chinese Centers for Disease Control and Prevention (CDC) has suggested classifying COVID-19 patients according to their clinical state into three categories. Mild disease manifestations include no pneumonia or mild pneumonia and occur in 81% of cases. Severe disease characterized by dyspnea, respiratory frequency ≥ 30/min, blood oxygen saturation (SpO2) ≤ 93%, PaO2/FiO2 ratio < 300, and/or lung infiltrates >50% within 24–48 h is present in 14% of cases. Critical conditions accompanied by respiratory failure, septic shock, and/or multiple organ dysfunction or failure occur in 5% of cases (The Novel Coronavirus Pneumonia Emergency Response Epidemiology 2020). While most patients experience mild respiratory illness, some cases, especially the old population and patients with comorbid conditions, develop a severe form of disease requiring intensive management (Guan et al. 2020b; Lotfi and Rezaei 2020). Chinese CDC has reported an overall case fatality rate of 2.3%, with the majority of deceased patients being 60 years old and above and having preexisting comorbid conditions such as hypertension, cardiovascular disease, and diabetes (The Novel Coronavirus Pneumonia Emergency Response Epidemiology 2020; Shamshirian and Rezaei 2020). Accordingly, the resistance and susceptibility to disease are governed by genetic factors and nongenetic factors such as biological age, pregnancy, and chronic conditions, which affect the expression of the receptors which mediate virus entry, i.e., angiotensin-converting enzyme 2 (ACE2), and the function of the immune system (Yousefzadegan and Rezaei 2020; Ahmadi et al. 2020; Ahanchian et al. 2020; Darbeheshti and Rezaei 2020; Babaha and Rezaei 2020; Mirbeyk and Rezaei 2020).

With the increasing number of people infected with the virus, the burden of COVID-19 on healthcare systems has been overwhelming. Thus, early and efficient detection of high-risk patients is a priority for maximizing the utilization of limited resources, while current diagnostics are not as rapid and as reliable as the current pandemic condition needs (Basiri et al. 2020a). Furthermore, no clinically proven specific antiviral agent has been confirmed for COVID-19. However, approaches targeting the immune system and virus entry, computational methods of drug discovery, regenerative medicine, medical biotechnology, and picotechnology offer hope for the future amid the pandemic (Mohamed et al. 2020b; Rezaei 2020b; Rabiee et al. 2020; Basiri et al. 2020b; Sharifkashani et al. 2020; Lotfi et al. 2020; Mansourabadi et al. 2020; Pashaei and Rezaei 2020; Fathi and Rezaei 2020; Jahanshahlu and Rezaei 2020b). Thus supportive management, including oxygen therapy and conservative fluid management, along with anti-inflammatory and antiviral agents, remains to be the most important management strategy (Cascella et al. 2020; Saghazadeh and Rezaei 2020b). Dynamic changes of laboratory parameters, as well as clinical and imaging findings of patients infected with COVID-19, have been depicted in the literature (Zhou et al. 2020a; Liu et al. 2020a). Results are, however, not consistent. The present chapter aims to identify laboratory and clinical correlates of adverse outcomes in patients with COVID-19. For this purpose, a systematic evaluation of literature was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA).

22.2 Methods

22.2.1 Search Strategy

The published literature was searched from December 2019 to April 20, 2020. Databases including EMBASE, MEDLINE via PubMed, and Scopus were searched for observational studies of COVID-19. The keywords for search strategy included “Wuhan seafood market pneumonia virus,” “COVID19,” “2019-nCoV,” “novel coronavirus,” “coronavirus disease-19,” “severe acute respiratory syndrome coronavirus 2,” “SARS-CoV-2,” “characteristic,” “finding,” “lab,” “imaging,” “comorbidit*,” “sever*,” “critical,” “hospital*,” “ICU,” “non-ICU,” “death,” “mortality,” and “surviv*.” Detailed search strategies are presented in Table 22.1.

Table 22.1 Search strategy
Full size table

22.2.2 Study Selection

The studies identified were imported into an EndNote library. After duplicate removal, the two groups (three reviewers in each group) performed the screening titles and abstracts for inclusion and then reviewing the full text of potentially relevant articles. Any questions or conflicts were resolved through discussion with the members of the other group. Concerning studies with overlapping samples, the study that had the largest sample size was included in the final analysis. Studies that i. included subjects with confirmed COVID-19 based on laboratory testing; ii. evaluated individual laboratory characteristics, symptoms, and comorbidities in predicting severe COVID-19 infection or mortality caused by COVID-19 infection; and iii. were written in English were eligible to be included in the systematic review. If the criteria for disease severity were not determined, then the study was excluded (Table 22.2). Also, types of articles other than original observational studies, e.g., opinion, editorial, case report, book chapter, review, and meta-analysis, studies confined to the pediatric population or pregnant women, and studies rated as low-quality evidence were not included.

Table 22.2 The definition of outcome and population
Full size table

22.2.3 Data Extraction and Quality Assessment

We used the Newcastle-Ottawa scale (NOS) and its modified version for the assessment of the quality of retrospective cohort studies and cross-sectional studies. This scale gives the maximum score of 9 points for the least risk of bias in four domains: the selection of study groups (4 points), comparability of groups (2 points), and ascertainment of exposure and outcomes (3 points). Age and sex were defined as the most relevant covariate for comparability. For the cross-sectional studies, we used the modified NOS, which gives a maximum of 10 scores for the least risk of bias in three domains: the selection of study groups (5 points), comparability of groups (2 points), and ascertainment of exposure and outcomes (3 points).

Three authors independently extracted the data by a predesigned spreadsheet in Excel, including the author’s name, publication year, setting (country/hospital name), study design, sample size, clinical and laboratory and comorbidities, and severity criteria.

22.2.4 Data Synthesis

All between-group meta-analyses were performed using Comprehensive Meta-Analysis. For the continuous type of outcome, we entered the number of participants in two groups and the mean and standard deviation (SD) of laboratory data. We used methods described in Luo et al. (2018) and Wan et al. (2014) to estimate mean and SD based on median (IQR). A meta-analysis was conducted for data on laboratory and clinical characteristics, comorbidities, and vital status when there were at least three between-group comparisons for each variable. Because studies used different measurement scales, we employed the standardized mean difference (SMD) as effect size (ES) measure. For categorical data, we used the dichotomous method to calculate the odds ratio (OR) with its 95% confidence interval as ES measure. As explained by Higgins and Green et al., the ES of 0.2, 0.5, and 0.8 represent small, moderate, and large effect estimates, respectively (Higgins and Wells 2011).

We investigated heterogeneity across studies by Cochran’s Q test, which is calculated by the weighted sum of squared differences between individual study ES estimates. A P-value of 0.10 or less indicates the presence of heterogeneity. The I2 index was used to get a more precise estimate of heterogeneity. If the I2 is less than 40%, the heterogeneity is considered to be not significant. The fixed-effects model is the preferred method of meta-analysis. However, if I2 estimates exceeded 40%, the random-effects approach was applied as the meta-analysis model. We based our analysis on the different outcome stratification including critical/ICU vs noncritical/non-ICU, severe/critical vs non-severe, and survivors vs non-survivors. Subgroup analysis was performed to explore the potential sources for heterogeneity. We also applied univariate meta-regression analyses to investigate the effect of potential moderators, e.g., sample size, the difference in the mean age of two groups (years), and percentage of female patients on the effect size and heterogeneity. Finally, to identify studies exerting excessive influence on the results or causing high heterogeneity among studies, we conducted a sensitivity analysis by systematically excluding each study at a time and then rerunning the analysis to assess the change in ES.

We assessed publication bias visually by the degree of funnel plot asymmetry and used Begg-Mazumdar Kendall’s tau (Begg and Mazumdar 1994) and Egger bias test to quantify it (Duval and Tweedie 2000). A P-value of 0.10 or less suggests the presence of publication bias across studies. The trim and fill method was used to adjust effect sizes in cases that there was evidence of publication bias (Duval and Tweedie 2000).

22.3 Results

22.3.1 Characteristics of Selected Studies

The database search resulted in 461 records. After the removal of duplicates (n = 60) and retrieval of additional 14 studies through hand search of additional references, the title/abstract of 415 search results was screened for potential eligibility. Three hundred forty papers were excluded through screening. Seventy-five full-text articles were reviewed in detail, and 32 met the inclusion criteria. Moreover, an updated search was conducted and resulted in the inclusion of 22 studies. In total, 54 studies were included in the meta-analysis. Figure 22.1 provides an overview of study selection for systematic review and meta-analysis according to PRISMA guidelines.

Fig. 22.1
figure 1

PRISMA flow diagram of study selection

Full size image

22.3.2 Study and Patient Characteristics

As summarized in Table 22.3, all included studies were from China, except two studies, which were from Singapore (Young et al. 2020; Fan et al. 2020). All studies were designed as retrospective cohort studies except two with a prospective cohort design (Liu et al. 2020b; Momtazmanesh et al. 2020) and one which was a cross-sectional study (Zhang et al. 2020b). Tables 22.4 and 22.5 show the results of the quality assessment for studies included in the data synthesis.

Table 22.3 Characteristics of studies included in the systematic review
Full size table
Table 22.4 The results of quality assessment for cohort studies included in the systematic review
Full size table
Table 22.5 The results of quality assessment for a cross-sectional study included in the systematic review
Full size table

There were 16 studies that compared critical or severe vs non-severe patients (Zhao et al. 2020; Zhang et al. 2020a; Guan et al. 2020b; Li et al. 2020a; Zhou et al. 2020b; Cai et al.; Tian et al. 2020; Liu et al. 2020b, d; Wu et al. 2020; Feng et al. 2020c; Wan et al. , 2020; Xu et al. 2020; ; Han et al. 2020b; Wang et al. 2020d); 3 studies compared 3 groups of patients, namely, critical, severe, and non-severe (Feng et al. 2020b; Liu et al. 2020c; Han et al. 2020a); 20 studies compared severe vs non-severe patients (Qu et al. 2020; Zhang et al. 2020b, c; Wang et al. 2020e; Cao 2020; Gao et al. 2020; Chen et al. 2020a, e; Liu et al. 2020a; lei and Jian-ya 2020; Lu et al. 2020; Chu et al. 2020; Ji et al. 2020; Xie et al. 2020; Zheng et al. 2020a; c; Li et al. 2020c; Gong et al. 2020; Young et al. 2020; Qin et al. 2020; Nie et al. 2020); 6 studies compared critical/ICU vs noncritical/non-ICU patients (Wang et al. 2020a; Fan et al. 2020; Du et al. 2020; Yan et al. 2020; Huang et al. 2020; Chen et al. 2020b); and 9 studies compared survivor vs non-survivor patients (Tang et al. 2020; Yang et al. 2020; Zhou et al. 2020a; Wu et al. 2020; Momtazmanesh et al. 2020; Chen et al. 2020c, d; Li et al. 2020c; Wang et al. 2020c; Cao et al. 2020). Two studies were not entered into analysis because of the shared samples (Chen et al. 2020c; Huang et al. 2020). A total of 8603 patients were included in the meta-analysis. Tables 22.6, 22.7, 22.8, 22.9, 22.10, 22.11, and 22.12 provide a summary of the results of meta-analyses, subgroup analyses, and meta-regressions.

Table 22.6 Summary of the results of meta-analyses of clinical and laboratory characteristics of patients with COVID-19: patients who survived versus patients who died
Full size table
Table 22.7 Summary of the results of meta-analyses of laboratory measures in patients with COVID-19: patients with a severe form of COVID-19 versus patients with a non-severe form of COVID-19
Full size table
Table 22.8 Summary of the results of meta-analyses of clinical characteristics and comorbidities in patients with COVID-19: patients with a severe form of COVID-19 versus patients with a non-severe form of COVID-19
Full size table
Table 22.9 Summary of the results of meta-analyses of clinical and laboratory characteristics of patients with COVID-19: patients who either were in critical condition or required ICU admission versus patients who neither were in critical condition nor required ICU admission
Full size table
Table 22.10 Summary of the results of meta-analyses of subgroups of patients with a severe and non-severe form of COVID-19
Full size table
Table 22.11 Adjusted ES using trim and fill methods
Full size table
Table 22.12 Summary of the results of meta-regressions
Full size table

22.3.3 Laboratory Findings

22.3.3.1 Severity Analysis

Twenty-seven studies provided data on white blood cell (WBC) values for patients with severe (n = 1131) and non-severe (n = 3293) COVID-19. Meta-analysis showed that patients with non-severe outcome had lower WBC compared to patients with severe COVID-19 (SMD −0.40; 95% CI, −0.58, −0.21; P < 0.0001). There was no evidence of publication bias (Begg’s P = 0.77; Egger’s P = 0.10). Significant heterogeneity existed across studies (I2 = 84%), and therefore, the random-effects model was applied. Meta-regression analyses demonstrated no significant effect of percentages of female patients, the difference in the mean age, and sample size on the heterogeneity. Neither subgroup meta-analyses nor sensitivity analysis showed a significant effect of specific outcome or study on the effect size and related heterogeneity. Meta-analysis of data on four studies showed that critical COVID-19 patients (n = 131) had higher WBC than noncritical COVID-19 patients (n = 348) (SMD −0.58; 95% CI, −0.79, −0.37; P < 0.0001). There was no significant heterogeneity and evidence of publication bias as well.

Pooled data from 22 studies, including 2215 non-severe and 1109 severe COVID-19 patients, demonstrated that polymorphonuclear (PMN) cell counts were increased in severe compared to non-severe patients (SMD −0.72; 95% CI, −0.88, −0.57; P < 0.0001). There was significant heterogeneity among studies (I2 = 64%). No evidence of publication bias was observed (Begg’s P = 0.498; Egger’s P = 0.496). Meta-analysis for data derived from four studies comparing critical and noncritical COVID-19 patients showed a similar result with the SMD of −0.74 (95% CI, −1.10 to −0.39; P < 0.0001). Critical patients had higher PMN compared to the noncritical group.

Patients with severe outcomes (n = 1890) had lower lymphocyte count compared to non-severe patients (n = 4035) (SMD 0.68; 95% CI, 0.53, 0.84; P < 0.0001). Heterogeneity was high (I2 = 80%). Also, critical patients had lower lymphocyte counts than noncritical patients.

Patients with severe disease (n = 2288) had significantly lower platelets than non-severe cases (n = 627) with an SMD of 0.21 (95% CI, 0.06, 0.35; P = 0.004). Moderate heterogeneity was present (I2 = 50.5%). No evidence of publication bias was observed (Begg’s P = 0.707; Egger’s P = 0.241).

Eight studies documented CD4 counts for 506 non-severe and 587 severe patients. Because of evidence of publication bias (Begg’s P = 0.009), the ES was adjusted using trim and fill method, resulting in an ES of 0.80 (95% CI, −0.06, 1.62), which showed no significant difference between severe and non-severe patients in regard to CD4 counts. Heterogeneity was high (I2 = 96%).

Total bilirubin (TB) was investigated in 15 studies, with a total of 1528 non-severe and 650 severe individuals. Severe patients had higher TB compared to the non-severe group (SMD −0.33; 95% CI, −0.46, −0.21, P < 0.0001), and no substantial heterogeneity was seen (I2 = 31%).

Alanine transaminase (ALT) was documented in 23 studies accounting for a total of 1875 non-severe and 775 severe patients. ALT levels were higher in severe patients than non-severe patients (SMD −0.52; 95% CI, −0.77, −0.27; P < 0.0001). There was evidence of publication bias (Begg’s P = 0.009). The ES was adjusted using the trim and fill method and estimated to be about −0.56 with 95% CI, −0.79, −0.32. There was high heterogeneity across studies (I2 = 85.5%). A sensitivity analysis showed that I2 dropped to 49% with the associated ES of −0.36 (95% CI, −0.50, −0.22) when the study (Zheng et al. 2020a) was excluded. Similarly, AST levels were higher in severe patients compared to non-severe patients (SMD −0.79; 95% CI, −1.00, −0.58; P < 0.0001). Heterogeneity was high (I2 = 74.5%). Critical patients had higher AST and ALT levels compared to noncritical patients (SMD −0.78; 95% CI, −1.27, −0.29; and SMD −0.45; 95% CI, −0.68, −0.22), respectively.

Creatinine was increased in severe patients (n = 751) compared to non-severe patients (n = 2320) (SMD −0.21; 95% CI, −0.34, −0.08; P = 0.001). No evidence of publication bias was found (Begg’s P = 0.785; Egger’s P = 0.282). Moderate heterogeneity was seen (I2 = 52%). Furthermore, BUN was higher in severe patients than non-severe patients (SMD −0.55; 95% CI, −0.93, −0.17; P < 0.0001).

Severe patients showed no difference in potassium levels compared with non-severe patients (SMD = 0.18, p = 0.269). However, the exclusion of a study (Feng et al. 2020b) could resolve heterogeneity (I2 = 0.0), and the ES became significant (SMD 0.25; 95% CI, 0.12, 0.39; P < 0.0001). Sodium levels were not different between severe and non-severe patients, and sensitivity analysis showed no significant effect of any individual study.

Troponin I (TnI) was higher in severe patients (n = 199) than non-severe patients (n = 585) (SMD −0.98; 95% CI, −1.18, −0.77). Thirteen studies reported data on albumin levels, including 868 severe patients and 1719 non-severe patients. Severe patients had lower albumin levels (SMD 1.02; 95% CI, 0.81, 1.23; P < 0.0001). Also, severe (n = 1170) patients had higher procalcitonin levels compared to non-severe patients (n = 707) (SMD −0.81; 95% CI, −1.07, −0.55; P < 0.0001).

A meta-analysis of 11 studies providing data on prothrombin time (PT) showed that patients with severe disease were not different from non-severe patients (SMD −0.30; 95% CI, −0.64, 0.03; P = 0.074). Also, PTT showed no significant difference between the two groups (SMD = 0.05; 95% CI, −0.29 to 0.39; P = 0.772). Critical patients had higher PT compared to noncritical patients (SMD −0.27; 95% CI, −0.50, −0.46, P = 0.019), though PTT was not different between two critical and noncritical groups (SMD 0.11; 95% CI, −0.12, 0.34; P = 0.347).

Seventeen studies reported data on LDH levels, and the meta-analysis revealed that severe patients had higher LDH levels than non-severe patients (SMD −1.33; 95% CI, −1.54, −1.17; P < 0.0001). There was moderate heterogeneity (I2 = 68.41). Also, severe patients had significantly higher D-dimer levels compared to non-severe patients (SMD −0.93; 95% CI, −1.12, −0.74; P < 0.0001). Heterogeneity among studies was relatively high (I2 = 72.77%), but meta-regression analysis could not find any significant potential moderators.

22.3.3.2 Survival Analysis

A total of five studies reported the white blood cell count concerning survival outcomes. Non-survivor patients (n = 707) had significantly higher WBC than survivors (n = 365) with an ES of −1.01 (95% CI, −1.33, −0.69). No evidence of publication bias was observed (Begg’s P = 0.327; Egger’s P = 0.356). Heterogeneity was high (I2 = 79%). Sensitivity analysis was performed, leaving out a study that only included patients with ARDS (acute respiratory syndrome) (Wu et al. 2020), and as a result, heterogeneity was abolished, and ES became −1.18 (95% CI, −1.33, −1.03).

Analysis of four studies with a total number of 570 survivors and 311 non-survivors revealed significantly higher PMN counts in non-survivor patients compared to survivors (SMD −1.05; 95% CI, −1.53, −0.58; P < 0.0001). Heterogeneity was high (I2 = 87.76%). When removing the abovementioned study (Wu et al. 2020), heterogeneity dropped to 55%, and the ES remained significant −1.32 (SMD −1.32; 95% CI, −1.59, −1.06).

Patients who survived (n = 727) had significantly higher lymphocyte than non-survivors (n = 397) (SMD 0.97; 95% CI, 0.52, 1.42). Significant publication bias was observed (Begg’s P = 0.024 Egger’s P = 0.043). The trim and fill method was used to adjust the ES (SMD 1.20; 95% CI, 1.06, 1.34). No individual study showed a significant influence on the ES.

Compared to non-survivors (n = 376), survivors had significantly higher platelet counts (SMD 0.45; 95% CI, 0.19, 0.70; P = 0.001). Heterogeneity was relatively high (I2 = 67.56%). When the study (Yang et al. 2020) was removed, there was no heterogeneity across studies further (I2 = 0%), while the pooled effect remained significant.

Non-survivor patients showed higher ALT (SMD −0.36, P < 0.0001) and total bilirubin (SMD = −0.79, p < 0.0001) than survivors. Albumin levels were measured in five studies. Survivor patients showed higher albumin levels compared to non-survivors (SMD = 0.88, p < 0.0001). However, heterogeneity was high (I2 = 87.95). Sensitivity analysis showed a robust ES. Additionally, with the exclusion of two studies (Wu et al. 2020; Momtazmanesh et al. 2020), I2 reduced to 58.41% (Wu et al. 2020; Momtazmanesh et al. 2020).

IL-6 (SMD −1.22; P < 0.0001) and procalcitonin (SMD −1.05; P < 0.0001) levels were significantly higher in non-survivors than survivors. For both analyses, there was no evidence of publication bias and significant heterogeneity.

A meta-analysis of data from five studies showed no difference between non-survivors (n = 343) and survivors (n = 590) regarding creatinine levels (SMD −0.34; 95% CI, −0.77, 0.09; P = 0.123). However, there was high heterogeneity (I2 = 87.55). Sensitivity analysis excluding two studies reduced heterogeneity to I2 = 0%, though the ES remained not significant (SMD 0.016; 95% CI, −0.18, 0.21; P = 0.876).

PT was significantly higher in non-survivors than survivors (SMD −0.78; 95% CI, −1.89, −0.37; P < 0.0001). However, there was no difference regarding PTT between survivors and non-survivors (SMD 0.26; 95% CI, −0.62, 0.41; P = 0.695).

22.3.4 Symptoms and Vital Status

22.3.4.1 Severity Analysis

Compared to severe patients, non-severe patients were less likely to have chills (OR 0.62; 95% CI, 0.42, 0.93), sputum production (OR 0.66; 95% CI, 0.54 to 0.79), dyspnea (OR 0.22; 95% CI, 0.14, 0.33), hemoptysis (OR 0.25; 95% CI, 0.11, 0.55), fatigue (OR 0.71; 95% CI, 0.52, 0.97), and abdominal pain (OR 0.49; 95% CI, 0.24, 0.98). Patients with severe outcomes had significantly higher fever than non-severe patients (SMD −0.38; 95% CI, −0.67, −0.09; P = 0.01).

22.3.4.2 Survival Analysis

The only symptoms that showed a difference between survivors and non-survivors were dyspnea and fatigue, which were less likely to occur in survivors (OR 0.34; 95% CI 0.25, 0.46; and OR 0.75; 95% CI 0.57, 0.90).

22.3.5 Comorbidities and Demographic Characteristics

22.3.5.1 Severity Analysis

Severe patients (n = 4006) were significantly older than non-severe patients (n = 1493) (SMD −0.59; 95% CI, −0.76, −0.41). A meta-analysis of data derived from nine studies showed that non-severe patients were less likely to be smokers than severe patients (OR 0.53; 95% CI, 0.38, 0.73). Male patients had a higher risk of developing the severe disease compared to females (OR 1.32; 95% CI, 1.17, 1.49). Non-severe patients were less likely to have diabetes (OR 0.36; 95% CI, 0.26 to 0.52), chronic heart disease (OR 0.34; 95% CI, 0.25, 0.45), COPD (OR 0.20; 95% CI, 0.12, 0.33), cerebrovascular disease (OR 0.30; 95% CI, 0.18, 0.50), hypertension (OR 0.40; 95% CI, 0.30, 0.52), CKD (OR 0.35; 95% CI, 0.18, 0.68), and malignancy (OR 0.42; 95% CI, 0.24, 0.72) than severe patients. For all analyses, there was low to moderate heterogeneity, with I2 ranging from 0 to 47%.

22.3.5.2 Survival Analysis

A meta-analysis of data from eight studies demonstrated that non-survivors (n = 435) were older than survivors (n = 974) with the SMD of −1.07 (95% CI, −1.35, −0.78; P < 0.0001). Male patients were not different from females concerning survival (p = 0.066; I2 = 51.46).

Patients who survived were less likely to have diabetes (OR 0.49, 0.36 to 0.67), chronic heart disease (OR 0.31; 95% CI, 0.13–0.72), COPD (OR 0.13; 95% CI, 0.052–0.34), cerebrovascular disease (OR 0.125; 95% CI, 0.034–0.46), and CKD (OR 0.095; 95% CI, 0.023–0.39). For all analyses, there was low to moderate heterogeneity, with I2 ranging from 0 to 61%.

22.4 Discussion

Hundreds of thousands of deaths due to COVID-19 during only a few months have placed a heavy burden on the shoulder of each medical researcher by explicitly proving that our knowledge, at the present condition of the pandemic, can seldom offer a useful product to humanity unless all data are collected from other disciplines and then it would be entirely fair to arrive at conclusions from the accumulation of this data (Momtazmanesh et al. 2020; Mohamed et al. 2020a; Rzymski et al. 2020; Moradian et al. 2020; Kafieh et al. 2020). For example, medical doctors initially described COVID-19 as a disease of the respiratory system. It did not take long when a wave of studies reporting anosmia in patients with COVID-19 emerged, putting forward the neurotropism of COVID-19 (Yazdanpanah et al. 2020b; Jahanshahlu and Rezaei 2020a; Saleki et al. 2020). The next wave was actually of two sources of studies in parallel. One was studies that reported stroke events in young patients with COVID-19, and the other was studies that provided evidence of a high incidence of thromboembolic events in patients with COVID-19. Therefore, the second wave strongly confirmed that the involvement of the central nervous system might be why patients with COVID-19 die from acute respiratory failure in a few days as well as suggested coagulopathies as another significant contributor to death from COVID-19. In the meantime, many disciplinary perspectives on the immunology of COVID-19 come into the competition, and the perspective which is already dominant deals inflammatory cells and molecules widely distributed throughout the critical organs to mediate COVID-19 pathogenesis (Pourahmad et al. 2020; Bahrami et al. 2020; Yazdanpanah et al. 2020a; Mansourabadi et al. 2020; Saghazadeh and Rezaei 2020a; Pashaei and Rezaei 2020; Nasab et al. 2020; Saghazadeh and Rezaei 2020b; Rokni et al. 2020). As a result, taking a look at the current evidence links us to a variety of biomarkers and clinical features that might predict the prognosis of COVID-19. The present chapter aimed to offer the most reliable predictors, and for this, we performed a systematic synthesis of the data.

We have stratified patients according to their diverse outcomes: survivor vs non-survivor, critical vs noncritical, and severe vs non-severe. The analysis by severity outcome demonstrated that higher levels of WBC with PMN dominancy might be related to disease severity and subsequent mortality. In the analysis of the WBC in severe patients, a relatively high amount of heterogeneity was observed. Meta-regression of the sample size could only explain 28% of that heterogeneity, and the two other moderators, including the difference in age between two groups and percentages of female patients, could not explain heterogeneity significantly.

We could not find a significant difference regarding CD4+ and CD8+ cell counts between severe and non-severe patients. However, previous investigations have shown that the depletion of T cells occurs during the COVID-19 course (Diao et al. 2020; Fathi and Rezaei 2020). The study (Zheng et al. 2020b) has shown extensive lymphocyte subset reduction in the blood of patients infected with COVID-19 compared to patients with other types of pneumonia. CD45+ lymphocytes, CD3+ lymphocytes, CD4+ T cells, CD8+ T cells, and CD19+ B cells were significantly lower in patients with COVID-19 than those in non-COVID-19-infected pneumonia patients with the same radiological stage.

In line with previous studies, thrombocytopenia was found to be associated with the severity, mortality, and critical outcomes of COVID-19 (Lippi et al. 2020). However, the most remarkable effect size was observed for comparison between survivors and non-survivor patients, which showed moderate ES. Previous literature recognized thrombocytopenia as an indicator that occurs in critically ill patients and is associated with mortality (Williamson et al. 2013). Therefore, COVID-19 can be more challenging in patients whom themselves suffer from immune thrombocytopenia (Sahu et al. 2020).

Acute-phase reactants, including ESR, CRP, ferritin, and procalcitonin, were significantly increased in patients with severe disease compared to patients with the non-severe disease and in patients who died compared to patients who survived. In contrast, negative phase reactants such as albumin, on the other hand, were decreased in patients with severe disease and in patients who died.

Liver enzymes were increased in severe, critical, and deceased patients compared with their counterparts without adverse outcomes. Given the relatively low prevalence of chronic liver disease before infection with COVID-19, studies suggest that increased liver enzymes, as seen in severe patients, might result from a dysregulated immune response against the virus rather than underlying liver comorbidity (Mantovani et al. 2020). Studies postulating the same hypothesis emphasize an inflammatory storm having played a role in immune dysregulation. Other studies believe that respiratory distress syndrome-induced hypoxia and subsequent hepatic ischemia and hypoxia-reperfusion dysfunction might contribute to liver dysfunction in COVID-19 (Feng et al. 2020a).

Our findings pointed to a higher level of TnI in both severe patients and non-survivors. A cohort of 416 hospitalized patients with COVID-19 established that 19.7% of these patients had a cardiac injury and showed an independent correlation between cardiac injury and a higher risk of mortality in patients with COVID-19 (Shi et al. 2020). Several mechanisms could be responsible for cardiac injury, including increased myocardial demand relative to the blood supply and hypoxia, systemic inflammation derived from cytokines, and direct cardiac damage due to COVID-19 (Guo et al. 2020).

On the other hand, there was not a significant difference regarding creatinine neither between severe and non-severe patients nor between survivors and non-survivors. BUN, on the other hand, was significantly increased in severe patients compared to the non-severe group. The authors in Wang et al. (2020b) evaluated the kidney function of patients with COVID-19 over 4 weeks and found that patients with COVID-19 experienced no acute renal injury and only showed a mildly elevated BUN and creatinine. However, there is controversy around acute kidney injury in severe patients.

Electrolytes, including sodium and potassium, showed no significant difference between severe and non-severe patients. However, there was high heterogeneity in studies reporting potassium, and during sensitivity analysis, a significant ES was observed indicating hypokalemia in severe cases. Studies report hypokalemia to be present in a majority of critically ill patients. It can be attributed to increased urine potassium excretion due to the renin-angiotensin system (RAS) disturbances and enhanced delivery of sodium and water to distal tubules (Li et al. 2020b).

Also, PT and PTT in severe patients were not significantly different from non-severe patients. However, PT was higher in critical patients and non-survivors compared to noncritical and survivors, respectively. It might be due to the measurement time of coagulation parameters, which mostly was at the time of admission, while it would undergo significant change over the disease course. D-dimer, on the other hand, was significantly increased in patients with poor outcome.

Patients with poor outcomes, e.g., severe disease and mortality, were more likely to have COPD, CHD, DM, HTN, CKD, cerebrovascular disease, and underlying malignancy compared to their counterparts. A nationwide analysis of 1590 patients in China reported that after adjusting for age and smoking status, COPD, diabetes, hypertension, and malignancy predicted increased admission to ICU, need for invasive ventilation, and death rate (Guan et al. 2020a).

Finally, patients with dyspnea at baseline were about four times more likely to progress to severe or critical condition. Also, hemoptysis increased the likelihood of developing a severe disease by about four times.

22.5 Conclusion

The current pandemic of COVID-19 has affected the healthcare system profoundly (Moazzami et al. 2020), and this effect can take a long time to compensate for it, and as the risk of reinfection is possible (Jabbari and Rezaei 2020), we must work at practice management. Early and efficient detection of high-risk patients is a priority for maximizing the utilization of limited resources. The present chapter aimed to offer the most reliable predictors, and for this, we performed a systematic synthesis of the data. Meta-analyses identified several laboratory parameters (WBC, PMN, IL-6, total bilirubin, ALT, creatinine, troponin, procalcitonin, LDH, and D-dimer), comorbidities (diabetes, chronic heart disease, COPD, cerebrovascular disease, hypertension, CKD, and malignancy), and symptoms (dyspnea and hemoptysis) that might help the prediction of severe, critical, and lethal phenotypes of COVID-19. These parameters correlate with the immune system function, inflammation, coagulation, and liver and kidney function, supporting the view that COVID-19 is a multisystem disorder.