3.2 Characteristics
The 55 included studies comprised 48 cross-sectional studies [5–9, 11, 12, 14, 15, 17, 18, 20–23, 26–58] (Table 1), 6 cohort studies [3, 10, 13, 16, 24, 25], and 1 case-control study [19] (Table 2), including 25,939 patients ranging in age from 18 to 94 years across Italy, CHN, UK, US, Spain, France, Greece, Austria, Singapore, India, Germany, Turkey, Switzerland, Sweden, and Portugal. The meta-analysis results of the research indicators are shown in Table 3. Among them, 48 cross-sectional studies had average AHRQ scale scores > 7, and 6 cohort studies and 1 case-control study had average NOS scores > 7, implying that these articles were all of high quality.
Table 1
Studies characteristics and quality. (Cross-sectional study).
Study | Published time | Country | Study design | Patients | Age | Gender(male) | AHRQ grade |
Amit, M. et al. | 2020 | Israel | Cross-sectional study | 156 | 72(60–82) | 108/48 | 9 |
Bonazzetti, C et al. | 2020 | Italy | Cross-sectional study | 89 | 61.5(53.1–68.7) | 69/20 | 8 |
Fu, Guoping et al. | 2020 | CHN | Cross-sectional study | 51 | 60.94 ± 14.87(25–87) | 27/14 | 8 |
Giacobbe, D. R et al. | 2020 | Italy | Cross-sectional study | 78 | 66 IQR 57–70 | 60/18 | 9 |
Anwar, Asad et al. | 2021 | UK | Cross-sectional study | 44 | 17–77 M59.5(IQR 50.5–64.5)/21–80 M59(IQR 49-67.5) | 34/10 | 8 |
Bardi, Tommaso et al. | 2021 | Spain | Cross-sectional study | 140 | 61(57–67) | 108/32 | 8 |
d’Humières, C et al. | 2021 | FRANCE | Cross-sectional study | 197 | 59(50–68) | 148/49 | 8 |
Dupuis, Claire et al. | 2021 | FRANCE | Cross-sectional study | 303 | 61(53–70) | 239/64 | 9 |
Grasselli, G. et al. | 2021 | Italy | Cross-sectional study | 774 | 62 (54–68) | 597/177 | 9 |
Karruli, A. et al. | 2021 | Italy | Cross-sectional study | 32 | 68 [55.25–75] | 23/9 | 9 |
Kokkoris, S. et al. | 2021 | Greece | Cross-sectional study | 50 | Median age 64 | 36/14 | 8 |
Llitjos, Jean-Francois et al. | 2021 | FRANCE | Cross-sectional study | 176 | 63 (55–73) | 134/42 | 8 |
Ong, C. C. H. et al. | 2021 | Singapore | Cross-sectional study | 71 | 52(39–66) | 59/12 | 9 |
Ramos, Rafael et al. | 2021 | Spain | Cross-sectional study | 213 | 61(52–71) | 110/103 | 8 |
Roedl, Kevin et al. | 2021 | Germany | Cross-sectional study | 223 | 69 (58-77.5) | 163/60 | 8 |
Rollas, Kazim et al. | 2021 | Turkey | Cross-sectional study | 38 | NR | NR | 9 |
Søgaard, K. K et al. | 2021 | Switzerland | Cross-sectional study | 41 | 64.8(54.7–72.1) | 31/10 | 9 |
Suarez-de-la-Rica, A. et al. | 2021 | Spain | Cross-sectional study | 107 | 62.2 ± 10.6 | 76/31 | 8 |
Yakar, Mehmet Nuri et al. | 2021 | Turkey | Cross-sectional study | 249 | 71(61–80) | 172/77 | 9 |
Yao, Ren-qi et al. | 2021 | CHN | Cross-sectional study | 35 | 64(59–67) | 25/10 | 8 |
Zamora-Cintas, M. I. et al. | 2021 | Spain | Cross-sectional study | 54 | NR | NR | 8 |
Zhang, J. et al. | 2021 | CHN | Cross-sectional study | 32 | 63.34 ± 12.48 | 20/12 | 9 |
Ahlstrom, Bjorn et al. | 2022 | Swedish | Cross-sectional study | 7382 | 63 (53–72) | 5191/2191 | 9 |
Brücker, W. et al. | 2022 | Germany | Cross-sectional study | 61 | 66.4 ± 13.3 | 34/27 | 9 |
Caiazzo, L.et al. | 2022 | Italy | Cross-sectional study | 89 | 68.1 ± 9.3 | 66/23 | 8 |
Cidade, Jose Pedro et al. | 2022 | Portugal | Cross-sectional study | 118 | 63.3 ± 13.1 | 87/31 | 8 |
Ćurčić, M. et al. | 2022 | Croatia | Cross-sectional study | 692 | NR | NR | 8 |
da Costa, R. L. et al. | 2022 | Brazil | Cross-sectional study | 191 | 69.66 ± 16.13 | 116/75 | 8 |
De Bruyn, A. et al. | 2022 | Belgium. | Cross-sectional study | 94 | 69.65 ± 11.29 | 55/39 | 9 |
DeVoe, C. et al. | 2022 | US | Cross-sectional study | 126 | 58.1 ± 17.9 | 85/41 | 8 |
Erbay, Kubra et al. | 2022 | Turkey | Cross-sectional study | 85 | 67.23 ± 13.05 | 54/31 | 8 |
Kozlowski, Bartosz et al. | 2022 | Poland | Cross-sectional study | 172 | 67.76 ± 11.16 | 112/60 | 8 |
Kurt, Ahmet Furkan et al. | 2022 | Turkey | Cross-sectional study | 470 | 66 ± 14.87 | 301/169 | 9 |
Lepape, Alain et al. | 2022 | FRANCE | Cross-sectional study | 4465 | 63.30 ± 11.68 | 3132/1333 | 8 |
Mantzarlis, K. et al. | 2022 | Greece | Cross-sectional study | 84 | 68.85 ± 12.17 | 56/28 | 9 |
Mustafa, Z. U. et al. | 2022 | Pakistan | Cross-sectional study | 636 | NR | 398/238 | 7 |
Pandey, M. et al. | 2022 | UK | Cross-sectional study | 299 | NR | 101/198 | 9 |
Roda, Silvia et al. | 2022 | Italy | Cross-sectional study | 22 | 61.36 ± 10.30 | 20/2 | 8 |
Routsi, C. et al. | 2022 | Greece | Cross-sectional study | 600 | NR | NR | 8 |
Russo, A. et al. | 2022 | Italy | Cross-sectional study | 32 | 62.50 ± 10.99 | 21/11 | 8 |
Seitz, T. et al. | 2022 | Austria | Cross-sectional study | 117 | 57.2 ± 11.9 | 72/45 | 8 |
Torrecillas, Miriam et al. | 2022 | Spain | Cross-sectional study | 220 | 63.65 ± 12.69 | 169/51 | 8 |
Alenazi, T. A. et al. | 2023 | Saudi Arabia | Cross-sectional study | 118 | 60.97 ± 16.32 | 74/43 | 8 |
Alessandri, F. et al. | 2023 | Italy | Cross-sectional study | 138 | 62.20 ± 15.36 | 97/41 | 9 |
Bedenić, B. et al. | 2023 | Croatia | Cross-sectional study | 118 | 71 years (range 25–94) | 78/40 | 8 |
Bonazzetti, C. t al. | 2023 | Italy | Cross-sectional study | 537 | 64.65 ± 11.15 | 402/135 | 9 |
Guanche Garcell, H. et al. | 2023 | Cuban | Cross-sectional study | 130 | NR | NR | 8 |
Taysi, M. R. et al. | 2023 | Turkey | Cross-sectional study | 205 | 68.4 ± 13.1 | 119/86 | 8 |
AHRQ: Agency for Healthcare Research and Quality; NR: not reported. |
Table 2
Studies characteristics and quality. (Cohort study and Case-control study)
Study | Published time | Country | Study design | Patients | Age | Gender (male) | NOS grade |
Cataldo, M. A et al. | 2020 | Italy | Cohort study | 57 | 62 ± 13 | 41/16 | 7 |
Garcia, Pedro David Wendel et al | 2020 | European | Cohort study | 639 | 63 (53–71) | 480/159 | 8 |
Zhang, H. et al. | 2020 | CHN | Cohort study | 38 | 64.76 ± 13.76 | 32/6 | 7 |
Massart, N. et al. | 2021 | France,Switzerland,Belgium | Cohort study | 4010 | NR | NR | 8 |
Palanisamy, N. et al. | 2021 | India | Cohort study | 750 | 60 ± 17.71 | 562/188 | 8 |
Bartoszewicz, M. et al. | 2023 | Poland | Cohort study | 201 | 66.1 ± 12.1 | 114/87 | 8 |
Dupper, A. C. et al. | 2022 | US | Case-control study | 96 | 64.91 ± 9.51 | 57/39 | 8 |
NOS: Newcastle-Ottawa Scale; NR: not reported. |
Table 3
Outcomes of meta-analysis.
Risk factors | No. of studies | Heterogeneity Analysis | Statistical model | statistical method | Effect estimate | P |
I² | P | (95%CI) |
Hypertension | 10 | 70.4% | 0.000 | Random-effects | OR | 1.30(0.92,1.83) | 0.131 |
Chronic pulmonary disease | 11 | 23.4% | 0.221 | Fixed-effects | OR | 1.07(0.90,1.29) | 0.443 |
Diabetes | 12 | 50.2% | 0.024 | Random-effects | OR | 1.34(1.04,1.73) | 0.022* |
Gender | 14 | 0.0% | 0.059 | Fixed-effects | OR | 1.28(1.10,1.50) | 0.006* |
Liver disease | 6 | 2.3% | 0.402 | Fixed-effects | OR | 0.86(0.47,1.58) | 0.635 |
Immunosuppressive diseases | 5 | 29.9% | 0.222 | Fixed-effects | OR | 1.11(0.88,1.40) | 0.375 |
Chronic kidney disease | 6 | 0.0% | 0.751 | Fixed-effects | OR | 1.20(0.78,1.84) | 0.411 |
Heart disease | 10 | 0.0% | 0.550 | Fixed-effects | OR | 1.00(0.85,1.17) | 0.957 |
Tocilizumab | 9 | 34.3% | 0.144 | Fixed-effects | OR | 1.04(0.74,1.46) | 0.815 |
Tumors | 9 | 10.2% | 0.350 | Fixed-effects | OR | 1.04(0.78,1.37) | 0.807 |
ECMO | 4 | 74.1% | 0.009 | Random-effects | OR | 2.70(1.17,6.26) | 0.020* |
Tracheal intubation | 4 | 67.8% | 0.025 | Random-effects | OR | 8.68(4.68,16.08) | < 0.001* |
Mechanical ventilation | 2 | 0.0% | 0.385 | Fixed-effects | OR | 22.00(3.77,128.328) | 0.001* |
Methylprednisolone | 2 | 13.5% | 0.282 | Fixed-effects | OR | 2.24(1.24,4.04) | 0.008* |
Methylprednisolone + Tocilizumab | 2 | 71.0% | 0.063 | Random-effects | OR | 4.54(1.09,18.88) | 0.037* |
Steroids | 3 | 87.6% | 0.000 | Random-effects | OR | 1.17(0.15,9.23) | 0.882 |
Remdesivir | 2 | 54.4% | 0.139 | Random-effects | OR | 0.80(0.14,4.41) | 0.794 |
Renal replacement therapy | 2 | 97.9% | 0.000 | Random-effects | OR | 0.86(0.11,6.57) | 0.882 |
Central venous catheterization | 2 | 0.0% | 0.559 | Fixed-effects | OR | 9.33(3.06,28.43) | < 0.001* |
Length of stay in ICUs | 8 | 0.0% | 0.712 | Fixed-effects | WMD | 10.37(9.29,11.44) | < 0.001* |
SAPS II score | 2 | 58.3% | 0.122 | Random-effects | WMD | 6.43(0.23,12.63) | 0.042* |
WMD: weight mean difference; OR: odds ratio; CI: confidence interval; *:P < 0.05. |
3.3 Result synthesis
3.3.1 Patient-related factors
Fourteen studies [8–21] explored the correlation between gender and BSI in COVID-19 patients in ICUs. A pooled analysis showed that male COVID-19 patients in ICUs were 28% more likely to develop BSI (OR = 1.28, 95% CI: 1.10–1.50, P = 0.006, I2 = 0.0%). (Fig. 2)
Two studies [10, 15] analyzed the correlation between SAPS II scores and ICU-BSI in COVID-19 patients. Meta-analysis showed that higher SAPS II scores were positively correlated with an increased incidence of BSI in COVID-19 patients in ICUs (WMD = 6.43, 95% CI: 0.23–12.63, P = 0.042, I2 = 58.3%). (Fig. 2)
Twelve studies [8–13, 15–18, 20, 21] investigated the correlation between DM and BSI in COVID-19 patients in ICUs. Most of these articles did not indicate whether DM was a risk factor for BSI. Our pooled analysis unraveled that DM increased the occurrence of BSI in COVID-19 patients in ICUs by 34% (OR = 1.34, 95% CI: 1.04–1.73, P = 0.022, I2 = 50.2%). (Fig. 2)
There were conflicting results about the association between hypertension and BSI in COVID-19 patients in ICUs. Ten studies [8, 10–13, 16–20] were involved with mixed results. Meta-analysis demonstrated no correlation between hypertension and BSI in COVID-19 patients in ICUs (OR = 1.30, 95% CI:0.92–1.83, P = 0.131, I2 = 70.4%). (Fig. 2)
Chronic pulmonary disease
Because COVID-19 mainly attacked the respiratory system, we extensively investigated the correlation between chronic pulmonary disease and BSI in COVID-19 patients in ICUs through 11 studies [8–12, 15–17, 19–21]. Meta-analysis showed no correlation between chronic pulmonary disease and BSI in COVID-19 patients in ICUs (OR = 1.07, 95% CI: 0.90–1.29, P = 0.443, I2 = 23.4%). (Fig. 2)
Six studies [8, 9, 12, 15, 16, 21] investigated the correlation between liver disease and ICU-BSI in COVID-19 patients. Meta-analysis showed no correlation between liver disease and BSI in COVID-19 patients in ICUs (OR = 0.86, 95% CI: 0.47–1.58, P = 0.635, I2 = 2.25%). (Fig. 3)
Seven studies [9–11, 15, 19–21] investigating the association between chronic kidney disease and BSI in COVID-19 patients in ICUs were included. One article was excluded by sensitivity analysis [10] and therefore six articles were included in the meta-analysis. It showed no correlation between chronic kidney disease and BSI in COVID-19 patients in ICUs (OR = 1.20, 95% CI: 0.78–1.84, P = 0.411, I2 = 0.0%). (Fig. 3)
Ten studies [9–11, 13, 15, 16, 19–21] investigating the correlation between heart disease and BSI among COVID-19 patients in ICUs were included. Meta-analysis showed no correlation between heart disease and BSI in COVID-19 patients in ICUs (OR = 1.00, 95% CI: 0.85-1.17P = 0.957, I2 = 0.0%). (Fig. 3)
Immunosuppressive diseases
All five studies [9, 15, 16, 19, 21] showed no correlation between immunosuppression and ICU-BSI in COVID-19 patients. Meta-analysis also showed no correlation between immunosuppression and BSI in COVID-19 patients in ICUs (OR = 1.11, 95% CI: 0.88–1.40, P = 0.375, I2 = 29.9%). (Fig. 3)
Nine studies [8–11, 17, 19–21] investigating the correlation between tumors and BSI in COVID-19 patients in ICUs were included. Meta-analysis showed no correlation between tumors and BSI in COVID-19 patients in ICUs (OR = 1.04, 95% CI: 0.78–1.37, P = 0.807, I2 = 10.2%). (Fig. 3)
3.3.2 Treatment-related factors
Four studies [10, 15, 16, 21] were included to investigate the association between tracheal intubation and ICU-BSI in COVID-19 patients. Meta-analysis revealed that tracheal intubation increased the risk of BSI in COVID-19 patients in ICUs by nearly 9-fold (OR = 8.68, 95% CI: 4.68–16.08, P < 0.001, I2 = 67.8%). (Fig. 4)
The correlation between mechanical ventilation and BSI in COVID-19 patients in ICUs was investigated by three studies [14, 16, 17]. Since no heterogeneity was found (P = 0.147, I2 = 47.9%), a fixed-effects model was adopted and unraveled marked differences (OR = 4.98, 95% CI: 2.73–9.08, P < 0.001). After sensitivity analysis, the heterogeneity was greatly reduced (P = 0.385, I2 = 0.0%) when the study of Palanisamy,N et al. [16] was excluded. The main source of heterogeneity might be the large sample size of their study, which tended to lead to unstable results compared to other studies with small sample sizes. Thus, this study was excluded because it led to a significant bias. The pooled analysis after exclusion using a fixed-effect model (OR = 22.00, 95% CI: 3.77-128.328, p < 0.001) showed statistically significant differences. The meta-analysis showcased that mechanical ventilation increased the risk of BSI by 22 times in COVID-19 patients in ICUs. The excluded study by Palanisamy,N et al. also showed that mechanical ventilation could increase the risk of BSI by 4-fold, in agreement with our results.(Fig. 4)
Many critically ill patients have used ECMO for supportive care. Including four studies [10, 12, 18, 55], we explored the correlation between ECMO and BSI among COVID-19 patients in ICUs. Meta-analysis manifested that ECMO increased the risk of BSI in COVID-19 patients in ICUs by nearly three times (OR = 2.70, 95% CI: 1.17–6.26, P = 0.020, I2 = 74.1%). (Fig. 4)
Central venous catheterization (CVC)
Two studies[11, 18] investigated the correlation between CVC and ICU-BSI in COVID-19 patients. Meta-analysis showed that CVC increase the Catheter-related BSI (OR = 9.33, 95% CI: 3.06–28.43, P < 0.001, I2 = 0.0%). (Fig. 4)
Renal replacement therapy (RRT)
Two studies [10, 12] investigating the correlation between RRT and BSI in COVID-19 patients in ICU were included. Meta-analysis showed no correlation between BSI and RRT in COVID-19 patients in ICUs (OR = 0.86, 95% CI: 0.11–6.57, P = 0.882, I2 = 97.9%). (Fig. 4)
Eightstudies were included [12–14, 17, 18, 20, 21, 53], all of which showed a strong correlation between the length of stay in ICUs and the occurrence of BSI in COVID-19 patients in ICUs. A meta-analysis showed that the longer the ICU stay, the higher the risk of BSI in COVID-19 patients in ICU (WMD = 10.37, 95% CI:9.29–11.44, P < 0.001, I2 = 0.0%). (Fig. 4)
3.3.3 Medication-related factors
The correlation between Tocilizumab and BSI in COVID-19 patients was investigated in 10 studies [8–12, 16, 17, 20, 21, 55]. One article was excluded by sensitivity analysis [21] and therefore nine articles were enrolled in the meta-analysis. There was no correlation between Tocilizumab and BSI in COVID-19 patients in ICUs (OR = 1.04, 95% CI: 0.74–1.46, P = 0.815, I2 = 34.3%). However, this may explain why Tocilizumab is widely used for severe and critically ill COVID-19 patients in ICUs under the guidance of guidelines. (Fig. 5)
Two studies [8, 12] investigated the association between Methylprednisolone and ICU-BSI in COVID-19 patients. Meta-analysis signified that Methylprednisolone was linked with BSI in COVID-19 patients in ICUs (OR = 2.24, 95% CI: 1.24–4.04, P = 0.008, I2 = 13.5%). Meanwhile, we found that the combination of Methylprednisolone and Tocilizumab was associated with the occurrence of BSI in COVID-19 patients in ICUs (OR = 4.54, 95% CI: 1.09–18.88, P = 0.037, I2 = 71%). This may imply that Methylprednisolone is widely used for treatment and that the combination of Methylprednisolone and Tocilizumab may be used for COVID-19 patients in ICUs with more critical conditions. (Fig. 5)
The correlation between steroid use and ICU-BSI in COVID-19 patients was investigated in 3 studies [9, 17, 21]. Meta-analysis showed no correlation between steroid use and BSI in COVID-19 patients in ICUs (OR = 1.17, 95% CI: 0.15–9.23, P = 0.882, I2 = 87.6%). This may be related to the fact that steroids are widely used as they are believed to improve the recovery of patients. (Fig. 5)
The correlation between Remdesivir and ICU-BSI in COVID-19 patients was investigated in 2 studies [9, 21]. Meta-analysis showed no correlation between Remdesivir and BSI in COVID-19 patients in ICUs (OR = 0.80, 95% CI: 0.14–4.41, P = 0.794, I2 = 54.4%). This may be related to the fact that Remdesivir is considered a potent drug for the treatment of COVID-19, with significant efficacy, and therefore is more widely used for severe and critically ill patients in ICUs. (Fig. 5)
3.4 Sensitivity analysis
The stability of the results of the remaining articles was estimated by excluding each article in turn. Sensitivity analyses for gender, SAPS II score, DM, hypertension, chronic pulmonary disease, liver disease, heart disease, immunosuppressive disease, tumor, tracheal intubation, ECMO, CVC, RRT, length of stay in ICUs, and the use of Methylprednisolone, Steroids, and Remdesivir revealed that the results were relatively stable. In the sensitivity analysis of mechanical ventilation, the study by Palanisamy,N et al. [16] greatly impacted the results, so the results were pooled after the exclusion of that article, and the results were more stable. Similarly, in the sensitivity study of chronic kidney disease, it was found that Massart,N et al. [10] greatly influenced the results. After excluding the article and re-combining the results, the results were more stable. In the sensitivity study on the use of Tocilizumab, the study by Bonazzetti,C et al. [21] greatly influenced the results. The results were more stable when the article was excluded, and the results were re-combined. The sensitivity analyses of other factors implied stable and insignificant changes, so these studies were retained. (Figure S1.S2.S3)