ABSTRACT
PURPOSE: To comprehensively evaluate the incidences of abnormal neuroimaging findings in patients with COVID-19 via a systematic review and meta-analysis. METHOD: PubMed-MEDLINE and EMBASE were searched for original articles reporting imaging findings of the brain in adult patients with COVID-19 between January 1, 2020 and October 9, 2020. Abnormal neuroimaging findings were categorized as (1) cerebral microhemorrhages, (2) acute spontaneous intracranial hemorrhage (ICH), (3) acute to subacute infarcts, and (4) encephalitis or encephalopathy. Pooled incidences of neuroimaging findings were assessed using random-effects modeling. Between-study heterogeneity was explored by using the χ2 statistic for pooled incidences and the inconsistency index I2. The quality of the studies was evaluated using the Risk of Bias Assessment Tool for Nonrandomized Studies. Subgroup meta-regression analysis was performed to identify potential sources of heterogeneity. RESULTS: Twenty-one eligible papers, including 2125 patients, were identified. The pooled incidences of cerebral microhemorrhages, acute spontaneous ICH, acute/subacute infarcts, and encephalitis/encephalopathy were 6.9 % (95 % confidence interval [CI], 4.9 %-8.9 %), 5.4 % (95 % CI, 3.1 %-7.6 %), 24.0 % (95 % CI, 16.1 %-31.8 %), and 3.3 % (95 % CI, 1.9 %-4.7 %), respectively. Substantial heterogeneities were noted for all neuroimaging findings (I2 = 87 %-97 %). Significant publication biases were present in the pooled incidences. In the subgroup meta-regression analysis, patients with mean or median ages over 65 years showed a significantly lower incidence of encephalitis/encephalopathy (P < 0.001). Furthermore, studies reported that patients in ICU had significantly higher incidences of cerebral microhemorrhages (Pâ¯<â¯0.001) and encephalitis/encephalopathy (Pâ¯<â¯0.001). CONCLUSIONS: Considerable incidences of abnormal neuroimaging findings have been reported in patients with COVID-19. Acute to subacute cerebral infarction was the most prevalent neuroimaging finding.