Search findings and study characteristics
The initial search revealed 3,071 studies, which narrowed to 2,151 studies after duplicate removal. After screening, 1,578 studies were removed and 573 articles were assessed with full-text review. After removing 493 non-relevant studies, our systematic review included a total of 80 studies (Additional file 3: Figure E1). The full list of the included studies is available in Additional file 4 [19–98].
The 80 studies included 43,248 patients, with a median of 130 patients per study (interquartile range 73–317). The studies were conducted in Asia (n = 48), Europe (n = 22), and North America (n = 10). The countries included China (n = 43), the United States of America (n = 10), Italy (n = 9), Spain (n = 5), Iran (n = 3), the United Kingdom (n = 3), South Korea (n = 1), India (n = 1), Denmark (n = 1), France (n = 1), Greece (n = 1), Norway (n = 1), and Poland (n = 1). The study outcomes were death (n = 41, 51.3%), admission to the ICU (n = 15, 18.8%), admission to the ICU or death (n = 9, 11.3%), and critical type of COVID-19 (n = 15, 18.8%). A median of 23.7% of patients had suffered the critical outcome in the overall population and was highest in studies from North America (34.8%), followed by Europe (26.5%) and Asia (17.3%) (P < 0.001). Most of the studies were retrospective observational studies (n = 68, 85.0%). Of the 80 studies, 50 (62.5%) did not specify any restriction in patient selection, while 13 studies included patients with certain comorbidities, seven with certain COVID-19 severity, and four with computed tomography findings (Table 1).
Table 1
Characteristics of studies included in analysis
Variables | Total | Asia | Europe | North America | P |
N = 80 | n = 48 | n = 22 | n = 10 |
Number of patients | 130 (73– 317) | 136 (99–323) | 114 (36–233) | 104 (72– 1000) | 0.328 |
Study outcome | | | | | < 0.001 |
| Death | 41 (51.3) | 24 (50.0) | 15 (68.2) | 2 (20.0) | |
| Admission to the ICU | 15 (18.8) | 6 (12.5) | 5 (22.7) | 4 (40.0) | |
| Admission to the ICU or death | 9 (11.3) | 3 (6.3) | 2 (9.1) | 4 (40.0) | |
| Critical type COVID-19* | 15 (18.8) | 15 (31.3) | 0 (0.0) | 0 (0.0) | |
Proportion of patients with critical outcome, % | 23.7 (14.8– 34.4) | 17.3 (13.6–27.9) | 26.4 (19.0–45.8) | 34.8 (30.0– 42.7) | < 0.001 |
Country | | | | | NA |
| China | 43 (53.8) | 43 (89.6) | 0 | 0 | |
| United States of America | 10 (12.5) | 0 | 0 | 10 (100.0) | |
| Italy | 9 (11.3) | 0 | 9 (40.9) | 0 | |
| Spain | 5 (6.3) | 0 | 5 (22.7) | 0 | |
| Iran | 3 (3.8) | 3 (6.3) | 0 | 0 | |
| United Kingdom | 3 (3.8) | 0 | 3 (13.6) | 0 | |
| South Korea | 1 (1.3) | 1 (2.1) | 0 | 0 | |
| India | 1 (1.3) | 1 (2.1) | 0 | 0 | |
| Denmark | 1 (1.3) | 0 | 1 (4.6) | 0 | |
| France | 1 (1.3) | 0 | 1 (4.6) | 0 | |
| Greece | 1 (1.3) | 0 | 1 (4.6) | 0 | |
| Norway | 1 (1.3) | 0 | 1 (4.6) | 0 | |
| Poland | 1 (1.3) | 0 | 1 (4.6) | 0 | |
Study design | | | | | 0.004 |
| Retrospective observational | 68 (85.0) | 45 (93.8) | 14 (63.6) | 9 (90.9) | |
| Prospective cohort | 12 (15.0) | 3 (6.3) | 8 (36.4) | 1 (10.0) | |
Restriction in patient selection | 0.856 |
| None | 50 (62.5) | 30 (62.5) | 12 (54.6) | 8 (80.0) | |
| Certain comorbidity | 13 (16.3) | 6 (12.5) | 5 (22.7) | 2 (20.0) | |
| Certain severity | 7 (8.8) | 5 (10.4) | 2 (9.1) | 0 (0.0) | |
| Patients with CT results | 4 (5.0) | 3 (6.3) | 1 (4.6) | 0 (0.0) | |
| Certain age group | 2 (2.5) | 2 (4.2) | 0 (0.0) | 0 (0.0) | |
| Certain symptom | 1 (1.3) | 1 (2.1) | 0 (0.0) | 0 (0.0) | |
| Multiple PCR tests | 1 (1.3) | 1 (2.1) | 0 (0.0) | 0 (0.0) | |
| Hospitalized via ER | 1 (1.3) | 0 (0.0) | 1 (4.6) | 0 (0.0) | |
| Certain race | 1 (1.3) | 0 (0.0) | 1 (4.6) | 0 (0.0) | |
Numbers are presented as number (percentage) or median (interquartile range). P-values are calculated from chi-square test, Fisher’s exact test, or Kruskal–Wallis test. |
*Critical type COVID-19 refers to disease extent with respiratory failure, septic shock, and/or multiple organ dysfunction |
Abbreviations: ICU, intensive care unit; NA, not applicable; CT, computed tomography; PCR, polymerase chain reaction; ER, emergency room. |
Baseline patient characteristics and symptoms
In the overall population, a proportion of 0.57 [0.55–0.59] were male, with a median age of 66.00 [61.14–68.85] years. The common underlying medical conditions were hypertension (pooled proportion 0.41 [0.35–0.47]), smoking history (pooled proportion 0.23 [0.19–0.27]), and diabetes (pooled proportion 0.21 [0.17–0.25]). The common symptoms were fever (pooled proportion 0.79 [0.70–0.86]), cough (pooled proportion 0.65 [0.60–0.70]), and anorexia (pooled proportion 0.58 [0.43–0.72]) (Table 2).
Table 2
Summary of patient characteristics according to the continents of the studies performed.
Variables | Total | n | Asia | n | Europe | n | North America | n |
Male sex | 0.57 (0.55–0.59) | 51 | 0.52 (0.50–0.55) | 24 | 0.64 (0.61–0.68) | 19 | 0.57 (0.53–0.60) | 8 |
Age, years | 66.00 (63.14–68.85) | 77 | 58.98 (54.93–63.02) | 46 | 74.71 (70.38–79.04) | 21 | 61.53 (50.53–72.54) | 10 |
Ethnicity | | | | | | | | |
| Non-Hispanic White | 0.30 (0.13–0.47) | 4 | – | 0 | – | 0 | 0.30 (0.13–0.47) | 4 |
| Hispanic | 0.27 (0.24–0.29) | 4 | – | 0 | – | 0 | 0.27 (0.24–0.29) | 4 |
| Non-Hispanic black | 0.15 (0.10–0.19) | 4 | – | 0 | – | 0 | 0.15 (0.10–0.19) | 4 |
| Asian | 0.06 (0.02–0.10) | 4 | – | 0 | – | 0 | 0.06 (0.02–0.10) | 4 |
| Unknown/Others | 0.21 (0.00–0.42) | 4 | – | 0 | – | 0 | 0.21 (0.00–0.42) | 4 |
Underlying medical condition | | | | | | | |
| Hypertension | 0.41 (0.35–0.47) | 45 | 0.27 (0.23–0.31) | 21 | 0.51 (0.44–0.59) | 16 | 0.62 (0.58–0.66) | 8 |
| Smoking history | 0.23 (0.19–0.27) | 23 | 0.13 (0.04–0.26) | 8 | 0.29 (0.21–0.39) | 8 | 0.30 (0.25–0.35) | 7 |
| Diabetes | 0.21 (0.17–0.25) | 46 | 0.15 (0.12–0.17) | 20 | 0.21 (0.17–0.24) | 18 | 0.38 (0.32–0.43) | 8 |
| Cardiac disease | 0.18 (0.15–0.22) | 46 | 0.13 (0.08–0.20) | 21 | 0.24 (0.18–0.31) | 17 | 0.20 (0.16–0.24) | 8 |
| Renal disease | 0.12 (0.09–0.15) | 33 | 0.04 (0.02–0.06) | 11 | 0.18 (0.12–0.25) | 14 | 0.21 (0.16–0.27) | 8 |
| Malignancy | 0.10 (0.08–0.12) | 38 | 0.03 (0.02–0.04) | 16 | 0.23 (0.16–0.32) | 15 | 0.11 (0.08–0.15) | 7 |
| Respiratory disease | 0.09 (0.07–0.11) | 44 | 0.04 (0.03–0.06) | 20 | 0.12 (0.10–0.15) | 17 | 0.17 (0.13–0.21) | 7 |
| Hepatic disease | 0.04 (0.02–0.06) | 18 | 0.06 (0.01–0.12) | 13 | 0.01 (0.01–0.01) | 3 | 0.01 (0.01–0.02) | 2 |
| Cerebral disease | 0.06 (0.05–0.08) | 22 | 0.05 (0.03–0.07) | 12 | 0.09 (0.06–0.11) | 7 | 0.06 (0.02–0.11) | 3 |
Symptoms | | | | | | | | |
| Fever | 0.79 (0.70–0.86) | 32 | 0.82 (0.76–0.88) | 17 | 0.80 (0.64–0.92) | 9 | 0.65 (0.42–0.85) | 6 |
| Cough | 0.65 (0.60–0.70) | 31 | 0.66 (0.59–0.73) | 17 | 0.60 (0.50–0.70) | 10 | 0.71 (0.63–0.79) | 4 |
| Anorexia | 0.58 (0.43–0.72) | 8 | 0.62 (0.47–0.75) | 7 | – | 0 | 0.31 (0.23–0.41) | 1 |
| Fatigue | 0.44 (0.32–0.55) | 17 | 0.50 (0.34–0.67) | 10 | 0.26 (0.17–0.37) | 3 | 0.39 (0.28–0.51) | 4 |
| Dyspnea | 0.43 (0.34–0.52) | 28 | 0.31 (0.21–0.42) | 14 | 0.50 (0.39–0.60) | 9 | 0.64 (0.57–0.70) | 5 |
| Sputum | 0.27 (0.20–0.35) | 16 | 0.35 (0.30–0.40) | 10 | 0.16 (0.08–0.27) | 3 | 0.15 (0.06–0.25) | 3 |
| Myalgia | 0.22 (0.17–0.27) | 22 | 0.22 (0.16–0.30) | 12 | 0.17 (0.09–0.28) | 6 | 0.30 (0.22–0.38) | 4 |
| Chest tightness | 0.21 (0.11–0.33) | 10 | 0.26 (0.13–0.41) | 7 | – | 0 | 0.12 (0.07–0.17) | 3 |
| Dizziness | 0.14 (0.04–0.28) | 5 | 0.13 (0.03–0.28) | 4 | – | 0 | 0.19 (0.04–0.46) | 1 |
| Diarrhea | 0.14 (0.10–0.19) | 25 | 0.11 (0.10–0.19) | 14 | 0.17 (0.09–0.28) | 6 | 0.23 (0.21–0.26) | 5 |
| Headache | 0.12 (0.08–0.16) | 19 | 0.10 (0.04–0.17) | 10 | 0.17 (0.05–0.32) | 4 | 0.12 (0.08–0.16) | 5 |
| Nausea | 0.12 (0.07–0.17) | 12 | 0.08 (0.04–0.13) | 5 | 0.09 (0.02–0.21) | 3 | 0.18 (0.16–0.20) | 4 |
| Sore throat | 0.09 (0.06–0.12) | 15 | 0.09 (0.05–0.14) | 8 | 0.09 (0.02–0.19) | 3 | 0.08 (0.06–0.10) | 4 |
| Vomiting | 0.08 (0.02–0.18) | 7 | 0.08 (0.01–0.22) | 5 | – | 0 | 0.09 (0.05–0.14) | 2 |
| Rhinorrhea | 0.08 (0.04–0.12) | 5 | 0.04 (0.02–0.07) | 2 | – | 0 | 0.10 (0.06–0.14) | 3 |
| Abdominal pain | 0.04 (0.01–0.09) | 5 | 0.04 (0.01–0.09) | 5 | – | 0 | – | 0 |
Numbers are presented as pooled value with 95% confidence intervals. Values represent proportions unless specified otherwise. |
Abbreviation: n, numbers of studies included in calculating the pooled value in the previous column |
Impacts of baseline demographics and underlying medical condition on the critical outcome
Of the 43,248 patients, 10,652 suffered the critical outcome. A meta-analysis of 51 studies revealed that male sex was associated with an increased risk of the critical outcome (pooled RR 1.26 [1.17–1.36], I2 = 36.7%). The results remained consistent in subgroup analyses of each continent: 24 studies from Asia showed a pooled RR of 1.42 [1.26–1.61] (I2 = 0.0%), 19 studies from Europe showed a pooled RR of 1.19 [1.02–1.40] (I2 = 48.2%), and 8 studies from North America showed a pooled RR of 1.23 [1.07–1.42] (I2 = 61.2%) (Additional file 3: Figure E2). Older age was consistently associated with an increased risk of the critical outcome across the three continents, with a pooled WMD of 10.61 [10.28–10.94] (I2 = 89.9%) (Additional file 3: Figure E3).
Underlying medical condition including cerebral disease (pooled RR 2.12 [1.67–2.70], I2 = 82.8%), hepatic disease (pooled RR 1.84 [1.22–2.77], I2 = 64.5%), cardiac disease (pooled RR 1.80 [1.60–2.03], I2 = 74.7%), renal disease (pooled RR 1.76 [1.53–2.04], I2 = 78.0%), hypertension (pooled RR 1.70 [1.49–1.93], I2 = 71.2%), malignancy (pooled RR 1.64 [1.46–1.83], I2 = 51.6%), respiratory disease (pooled RR 1.55 [1.36–1.78], I2 = 64.9%), diabetes (pooled RR 1.54 [1.40–1.69], I2 = 58.0%), and smoking history (pooled RR 1.23 [1.18–1.28], I2 = 0.1%) were risk factors for the critical outcome. Subgroup analyses across the three continents showed largely similar results; however, several differences were noted. First, the presence of respiratory disease was associated with a higher risk of the critical outcome in studies from Asia (pooled RR 2.16 [1.60–2.92], I2 = 57.8%) and Europe (pooled RR 1.50 [1.32–1.69], I2 = 16.6%), but not North America (pooled RR 1.07 [0.96–1.19], I2 = 0.0%). Second, the presence of hepatic disease was associated with a higher risk of the critical outcome from Europe (pooled RR 1.34 [1.15–1.56], I2 = 0.0%), but not from Asia (pooled RR 1.94 [0.90–4.16], I2 = 68.1%) and North America (pooled RR 0.97 [0.22–4.25], I2 = 39.4%) (Fig. 1).
Associations between patient symptoms and the critical outcome
The results of the meta-analysis showed that dyspnea (pooled RR 2.90 [2.10–4.03], I2 = 85.7%), anorexia (pooled RR 2.07 [1.21–3.52], I2 = 69.8%), dizziness (pooled RR 2.06 [1.39–3.06], I2 = 5.1%), and fatigue (pooled RR 1.43 [1.08–1.89], I2 = 62.8%) were significantly associated with the critical outcome. In the subgroup analysis, dyspnea was the only symptom that was consistently associated with the poor outcome in Asia (pooled RR 5.60 [3.24–9.65], I2 = 84.0%), Europe (pooled RR 1.45 [1.10–1.91], I2 = 28.7%), and North America (pooled RR 1.52 [1.27–1.81], I2 = 0.0%). Vomiting (pooled RR 2.43 [1.60–3.69], I2 = 0.0%), anorexia (pooled RR 2.38 [1.45–3.91], I2 = 49.5%), dizziness (pooled RR 2.23 [1.51–3.28], I2 = 0.0%), and fatigue (pooled RR 1.92 [1.23–3.02], I2 = 72.3%) were significantly associated with the critical outcome in studies from Asia, but not from Europe and North America (Fig. 2).
Associations between laboratory findings and the critical outcome
Higher levels of lactate dehydrogenase (pooled SMD 1.13 [1.06–1.20], I2 = 91.0%), blood urea nitrogen (pooled SMD 1.07 [0.97–1.17], I2 = 93.6%), neutrophil count (pooled SMD [0.83–0.99], I2 = 92.7%), white blood cell count (pooled SMD 0.72 [0.66–0.78], I2 = 92.6%), creatine kinase (pooled SMD 0.62 [0.53–0.72], I2 = 89.5%), AST (pooled SMD 0.61 [0.56–0.66], I2 = 87.4%), creatinine (pooled SMD 0.47 [0.42–0.52], I2 = 84.3%), prothrombin time (pooled SMD 0.43 [0.32–0.53], I2 = 32.2%), total bilirubin (pooled SMD 0.39 [0.30–0.48], I2 = 54.8%), and ALT (pooled SMD 0.20 [0.15–0.25], I2 = 30.2%) were associated with the critical outcome. In contrast, levels of hemoglobin (pooled SMD − 0.16 [-0.25–-0.08], I2 = 67.8%), platelet count (pooled SMD − 0.24 [-0.31–-0.17], I2 = 72.5%), and lymphocyte count (pooled SMD − 0.43 [-0.48–-0.38], I2 = 76.8%) were inversely related to the critical outcome.
While many findings were consistent across the three continents, platelet count and hemoglobin levels showed different results. While platelet count was inversely associated with the critical outcome in studies from Asia (pooled SMD − 0.447 [-0.539–-0.354], I2 = 62.7%), this association was not observed in studies from Europe (pooled SMD 0.084 [-0.045–0.213], I2 = 39.5%) or North America (pooled SMD 0.076 [-0.170–0.322], I2 = 16.2%). In contrast, while lower hemoglobin levels were associated with the critical outcome in studies from Europe (pooled SMD − 0.389 [-0.569–-0.209], I2 = 34.5%) and North America (pooled SMD − 0.385 [-0.668–-0.101], I2 = 57.6%), this association was not identified in studies from Asia (pooled SMD − 0.061 [-0.163–0.041], I2 = 72.0%) (Fig. 3).
Impact of ethnicity on the critical outcome
Out of a total of 80 studies, only four studies from North America reported patients’ ethnicity according to the four categories (non-Hispanic white, non-Hispanic black, Hispanic, and Asian). The pooled proportions of each ethnicity were non-Hispanic white (0.30 [0.13–0.47]), Hispanic (0.27 [0.24–0.29]), non-Hispanic black (0.15 [0.10–0.19]), Asian (0.06 [0.02–0.10]), and others/unknown (0.21 [0.00–0.42]) (Table 2).
Compared with non-Hispanic white, Hispanic ethnicity (pooled RR 0.83 [0.71–.96], I2 = 8.4%) was associated with a lower risk of the critical outcome, while non-Hispanic black (pooled RR 0.84 [0.67–1.06], I2 = 28.3%) and Asian ethnicity (pooled RR 1.33 [0.86–2.06], I2 = 51.8%) was not (Additional file 3: Figure E4). Publication biases were not observed in these analyses (Egger’s P = 0.388, 0.282, and 0.557, respectively)
Assessment of study quality and publication bias
While most studies at least partly met the quality standards of each area, several studies did not. The two studies did not represent the population of interest (study participation), two studies did not adequately measure the prognostic factor of interest (prognostic factor measurement), and nine studies did not account for important potential confounders (confounding measurement and account). (Additional file 3: Table E2).
Most of the variables did not show any publication bias; however, male sex (P = 0.042), underlying diabetes (P = 0.001), malignancy (P = 0.020), cerebral disease (P = 0.042), symptoms of dyspnea (P = 0.019), and vomiting (P = 0.045) revealed significant publication bias. Laboratory findings of lymphocyte count (P = 0.003), ALT (P = 0.012), and AST (P = 0.023) also revealed publication biases (Additional file 3: Table E3).
Sensitivity analysis
A sensitivity analysis was performed in 17 studies without any restriction in patient selection, outcome confined to death, and at least partly achieving every standard of the six areas of potential study biases. The results were largely consistent with the main analyses. Male sex (pooled RR 1.17 [1.02–1.34], I2 = 44.8%) and older age (pooled WMD 11.94 [11.56–12.33], I2 = 79.0%) were risk factors for death. Most of the underlying comorbidities remained significant risk factors for death except for hepatic disease (pooled RR 1.99 [1.00–3.98], I2 = 52.0%). The pooled RRs for the comorbidities were: hypertension, 2.76 [1.95–3.90], I2 = 65.1%; cerebral disease, 2.72 [1.41–5.26], I2 = 87.2%; cardiac disease, 2.45 [1.97–3.05], I2 = 82.3%; respiratory disease, 2.11 [1.52–2.92], I2 = 56.1%; malignancy, 1.75 [1.37–2.23], I2 = 70.4%; renal disease, 1.62 [1.55–1.70], I2 = 0.0%; and diabetes, 1.42 [1.21–1.67], I2 = 58.8%. Symptoms of dyspnea (pooled RR 2.86 [1.35–6.07], I2 = 82.9%), fatigue (pooled RR 1.62 [1.05–2.51], I2 = 0.0%), and anorexia (pooled RR 2.16 [1.02–4.61], I2 = 5.8%) were associated with higher risk of death. The extent of heterogeneity in these categorical variables was largely reduced when further analyses were performed according to each continent. The association between laboratory findings and death was consistent with the main analyses, except that lower monocyte counts were associated with a higher risk of death (pooled SMD − 0.33 [-0.57–-0.09], I2 = 92.2%). The results of the sensitivity analysis are summarized in Additional file 3: Figure E5, and the details are described in Additional file 3: Table E4.