Introduction

Spontaneous intracerebral hemorrhage is a devastating subtype of stroke with high rates of morbidity, disability and mortality [1]. In 2019, there were 3.41 million new intracerebral hemorrhage cases, accounting for 27.9% of all new strokes, and 2.89 million stroke deaths [2]. As known, the risk of intracerebral hemorrhage increases significantly with age. Despite improvements in public blood pressure regulation, intracerebral hemorrhage still remains a global public health concern as the population ages. [3]. Early neurological deterioration (END) is a common complication observed in patients with intracerebral hemorrhage and is typically defined as a ≥ 4-point increase in the National Institutes of Health Stroke Scale (NIHSS) score or a ≥ 2-point decrease in the Glasgow Coma Scale (GCS) score or death from baseline to 24 h [4]. Early neurological deterioration occurs in 7.47–40.99% of individuals after intracerebral hemorrhage [5] and is related to poor functional outcomes, increased mortality, and a greater risk of disability [6, 7]. Early identification of early neurological deterioration thus becomes essential for improving the outcomes of patients after intracerebral hemorrhage. However, early neurological deterioration is an underlying process that clinicians cannot see directly since continuous neuroimaging monitoring is difficult [8]. Identifying predictors of early neurological deterioration is thus crucial for detecting high-risk individuals and optimizing patient care and outcomes.

Several studies have demonstrated relationships between early neurological deterioration and sociodemographic variables, clinical characteristics, and neuroimaging parameters. In terms of sociodemographic factors, older age and male sex were found to be independent predictors of early neurological deterioration [6, 9]. For clinical characteristics, the NIHSS score on admission, anticoagulant medication, systolic blood pressure, plasma adrenomedullin, serum myeloperoxidase and other characteristics were found to be predictive of early neurological deterioration [10,11,12,13]. For neuroimaging parameters, hematoma volume, intraventricular hemorrhage, hematoma expansion, and intraventricular extension were reported to predict the occurrence of early neurological deterioration [4, 11]. However, most of those previous studies had small sample sizes and found contentious predictors of early neurological deterioration in patients with intracerebral hemorrhage.

A systematic review and meta-analysis can provide a comprehensive and up-to-date synthesis of the current evidence, assisting in the identification of predictors of early neurological deterioration in patients with intracerebral hemorrhage. Only one meta-analysis, which included 14 studies and 2,088 patients, has addressed the predictors of early neurological deterioration in patients with intracerebral hemorrhage [14]. This meta-analysis included studies with varying definitions of early neurological deterioration. For example, Delgado, Alvarez-Sabín [15] defined early neurological deterioration as an increase in the NIHSS score of  ≥ 4 points within the first 48 h. Sykora, Diedler [16] used the definition of a 4-point increase in the NIHSSS score within the first 72 h. In addition, this meta-analysis performed a meta-analysis synthesis by merging the results from both multivariate and univariate logistic regression analyses, which may have overlooked significant confounders. The considerations stated above likely lead to partiality. Furthermore, numerous new studies providing additional predictors have emerged since the last meta-analysis. As a result, it is critical to undertake an updated systematic review that incorporates predictors derived only using multivariable logistic regression analysis with a similar definition of early neurological deterioration to reduce bias.

Therefore, this study aimed to conduct a systematic review and meta-analysis to identify the predictors of early neurological deterioration in patients with intracerebral hemorrhage. The findings of this study could provide an evidence-based foundation for healthcare providers to identify high-risk patients for early neurological deterioration and optimize management strategies.

Methods

This review was conducted following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses 2020 (PRISMA 2020) declaration. This review's comprehensive protocol was registered with PROSPERO under the registration number CRD42023484527.

Search strategy

A comprehensive literature search was conducted in electronic databases, including PubMed, Embase, Web of Science, the Cochrane Library, and CINAHL, from inception to November 18, 2023. The search strategy was developed by combining the following free words and MeSH terms: “cerebral hemorrhage”, “cerebrum hemorrhage*”, “cerebral parenchymal hemorrhage*”, “intracerebral hemorrhage*”, “cerebral hemorrhage*”, “cerebral brain hemorrhage*”, “neurological worsening”, “neurological deterioration”, “neurologic worsening”, and “neurologic deterioration”. The exact search algorithms used in each electronic database are displayed in Supplementary Material 1. The literature search was restricted to human species and publications in English. Furthermore, reference lists were examined to identify additional references from the included articles as well as from previous reviews or meta-analyses.

Study selection

After removing duplicated articles with Endnote software, two reviewers independently reviewed the study titles and abstracts for eligibility using the inclusion and exclusion criteria. Once a study was deemed eligible by any reviewer, the full text was retrieved. The full texts of the retrieved publications were independently assessed by two reviewers for final inclusion. Any discrepancies were discussed to reach a consensus by two reviewers, with the participation of a third reviewer if necessary.

The inclusion criteria following PECOs principles (P: participants; E: exposures; C: comparisons; O: outcomes; s: study design) were as follows: (1) P: patients diagnosed with spontaneous intracerebral hemorrhage by CT/MRI; (2) E (exposures): study identifying at least one risk factor for early neurological deterioration using multivariate logistic regression; (3) O (outcomes): early neurological deterioration defined as an increase in the NIHSS score, a decrease in the GCS scores, or death from baseline to 24 h; and (4) S (study design): prospective or retrospective cohort studies. If numerous studies addressed the same cohort, only the most recent study with the largest sample size was included. The exclusion criteria were as follows: (1) younger than 18 years; (2) not an original study (review, conference abstract, or case report); (3) no interested outcomes; (4) data cannot be extracted; and (5) low-quality studies.

Data extraction

The data of the included studies were extracted by two reviewers independently. To reach a consensus, the two reviewers discussed any disagreements, and if needed, consulted a third reviewer. The following information was extracted from the included studies and coded: first author, publication year, location, study design, sample size, age, sex, presence or absence of early neurological deterioration, and predictors identified with p values, 95% confidence intervals (CIs), odds ratios (ORs), risk ratios (RRs), or hazard ratios (HRs). The extracted data were stored in Microsoft Excel.

Quality appraisal

The Newcastle–Ottawa Scale (NOS) was used to assess the quality of the included studies [17]. The NOS for cohort studies consists of three dimensions, selection, comparability, and outcome, with a total of eight items. The maximum NOS score for a cohort study is nine points, which can be divided into three categories: low quality (≤ four points), moderate quality (five or six points), and high quality (≥ seven points)[18]. The quality appraisal procedure was independently conducted by two reviewers. Any disagreements were resolved through consultation with a third reviewer.

Data analysis

The meta-analysis in this study was conducted using the STATA 17.0 software package (Stata Corp LP, College Station, TX). The pooled prevalence and 95% confidence intervals (CIs) for early neurological deterioration were estimated. To assess predictors of early neurological deterioration, odds ratios (ORs) and their 95% CIs were synthesized. Heterogeneity was evaluated by the I2 value and Q statistics. If there was low statistical heterogeneity (p > 0.10 and I2 ≤ 50%), the fixed effects model was employed for meta-analysis; otherwise, the random effects model was utilized. A sensitivity analysis was conducted to examine the robustness of the overall findings by omitting the included studies one by one. Funnel plots and Egger's test were used to assess publication bias [19, 20]. If there was any publication bias, a trim-and-fill analysis was carried out [21]. Subgroup analysis was employed to examine variations in the prevalence and predictors of early neurological deterioration in subgroups categorized by study location and the baseline used to define early neurological deterioration.

Results

Study process

Figure 1 illustrates the flow diagram of the study screening process. The initial search yielded 3,051 documents from databases, 918 of which were duplicates. After evaluating the studies using the inclusion and exclusion criteria, a total of 32 studies were ultimately included.

Fig. 1
figure 1

Flow diagram of study selection

Full size image

Study characteristics

Table 1 displays the characteristics of the included studies. Among the 32 studies, five (15.6%) were retrospective cohort studies and twenty-seven (84.4%) were prospective studies. Only three studies (9.4%) were performed in multicenter settings. The included studies involved 5,014 patients, with 1,097 in the early neurological deterioration group and 3,683 in the case–control group (one study did not provide detailed data). The prevalence of early neurological deterioration ranged from 7.47% to 40.99%.

Table 1 Characteristics of the included studies
Full size table

Quality assessment of the included studies

Thirty studies were of high quality, while two studies were of moderate quality according to the NOS. Supplementary Material 2 displays the detailed quality assessment results.

Prevalence of early neurological deterioration

The pooled prevalence in our meta-analysis involving 31 studies was 23% (95% CI 21–26%, p < 0.01), with high heterogeneity (I2 = 77.34%, p < 0.01). In subgroup analysis, the pooled prevalence showed no significant difference in subgroups stratified either by study location or by baseline used to define early neurological deterioration (Supplementary Figs. 1, 2).

Predictors of early neurological deterioration

Age

Only two studies provided data regarding the relationship between age and early neurological deterioration in patients with intracerebral hemorrhage, yielding no statistically significant association (OR = 1.00, 95% CI 0.98, 1.03, p = 0.88), with no heterogeneity (I2 = 0.05%, p = 0.72) (Supplementary Fig. 3).

NIHSS score

Our meta-analysis included seventeen studies that examined the relationship between NIHSS score at admission and early neurological deterioration in patients with intracerebral hemorrhage. The meta-analysis revealed a 1.24-fold risk for early neurological deterioration when the NIHSS score increased by one point (OR = 1.24, 95% CI 1.17, 1.30, p < 0.01), with high heterogeneity (I2 = 67.65%, p < 0.01) (Fig. 2). Subgroup analysis revealed a significant difference in the pooled effect size of the NIHSS score among subgroups stratified by the baseline used to define early neurological deterioration. The sensitivity analysis revealed no statistically significant changes in the pooled effect size by eliminating studies by turns (Supplementary Figs. 4, 5).

Fig. 2
figure 2

Odds ratio of NIHSS for early neurological deterioration

Full size image

Hematoma volume

Eighteen studies provided data examining the relationship between hematoma volume and early neurological deterioration, revealing a 7% increase in risk for early neurological deterioration with increasing hematoma volume (OR = 1.07, 95% CI 1.06, 1.09, p < 0.01), with high heterogeneity (I2 = 59.52%, p < 0.01) (Fig. 3). Subgroup analysis revealed that the pooled effect size significantly differed among in subgroups based on the baseline used in defining early neurological deterioration. No statistically significant differences were found among the subgroups based on study location. The sensitivity analysis revealed no statistically significant changes in the pooled effect size by eliminating studies by turns (Supplementary Figs. 6–8).

Fig. 3
figure 3

Odds ratio of hematoma volume for early neurological deterioration

Full size image

Intraventricular hemorrhage

Three studies provided data examining the relationship between intraventricular hemorrhage and early neurological deterioration. This meta-analysis found that the patients with intraventricular hemorrhage had a 3.5-fold greater risk of early neurological deterioration (OR = 3.50, 95% CI 1.64, 7.47, p < 0.01), with high heterogeneity (I2 = 54.87%, p = 0.11) (Fig. 4A). Subgroup analysis revealed no statistically significant differences in the pooled effect size. The sensitivity analysis indicated no statistically significant changes in the pooled effect size by eliminating studies by turns (Supplementary Figs. 9–11).

Fig. 4
figure 4

A Odds ratio of intraventricular hemorrhage for early neurological deterioration; B Odds ratio of intraventricular extension for early neurological deterioration

Full size image

Intraventricular extension

Three studies provided data examining the relationship between intraventricular extension and early neurological deterioration. The meta-analysis found a 3.95-fold increase in the risk of early neurological deterioration in patients with intraventricular extension (OR = 3.95, 95% CI 1.96, 7.99, p < 0.01), with no heterogeneity (I2 = 0, p = 0.69) (Fig. 4B). Subgroup analysis revealed no statistically significant differences in the pooled effect size. The sensitivity analysis indicated no statistically significant changes in the pooled effect size by eliminating studies by turns (Supplementary Figs. 12–14).

Hematoma expansion

Four studies provided data examining the relationship between hematoma expansion and early neurological deterioration. The meta-analysis found an 8.77-fold increase in the risk of early neurological deterioration in patients with hematoma expansion (OR = 9.77, 95%CI 4.43, 17.40, p < 0.01), with high heterogeneity (I2 = 50.03%, p = 0.11) (Fig. 5A). Subgroup analysis found no statistically significant differences in the pooled effect size. The sensitivity analysis indicated no statistically significant changes in the pooled effect size by eliminating studies by turns (Supplementary Figs. 15–17).

Fig. 5
figure 5

A Odds ratio of hematoma expansion for early neurological deterioration; B Odds ratio computed tomographic angiography for early neurological deterioration

Full size image

Spot sign under computed tomographic angiography (CTA)

Only two studies provided data examining the relationship between spot sign and early neurological deterioration. The meta-analysis found that patients with spot sign under CTA had a 5.77-fold greater risk of early neurological deterioration (OR = 5.77, 95%CI 1.53, 20.23, p = 0.01), with high heterogeneity (I2 = 79.37%, p = 0.03) (Fig. 5B).

Publication bias

Supplementary Fig. 18 displays the funnel plot, which indicates significant publication bias. Additionally, Egger’s test also revealed significant publication bias (p < 0.001). The trim-and-fill analysis revealed that there were no statistically significant changes in the pooled effect size after seven studies were imputed.

Discussion

This meta-analysis conducted a comprehensive search of multiple databases to identify relevant studies. Ultimately, data from a total of 32 studies and 5,014 patients were included. The pooled prevalence of early neurological deterioration was determined to be 23%. Six variables, including the NIHSS score at admission, hematoma volume, intraventricular hemorrhage, intraventricular extension, hematoma expansion, and spot sign under CTA, were found to be significantly associated with early neurological deterioration in patients with intracerebral hemorrhage.

This meta-analysis revealed that the initial NIHSS score and hematoma volume, which are usually thought to be strong predictive variables for mortality and functional outcomes [22], were also predictors of early neurological deterioration in patients with intracerebral hemorrhage. Specifically, patients with higher NIHSS scores and larger hematoma volumes were found to be at a higher risk of early neurological deterioration. These findings support previous models proposed by Law [7] and He [6], which included both the NIHSS score and hematoma volume as predictors of neurological deterioration in patients with intracerebral hemorrhage. However, given that most intracerebral hemorrhage patients have depressed consciousness, routine evaluation of the NIHSS score may be limited. To expedite assessment and anticipate early neurological deterioration in the future, it may be advantageous to employ alternate severity tools, such as the ICH score and GCS [23, 24], which are widely used and validated. Notably, subgroup analysis revealed differences in the pooled effect sizes of the NIHSS score and hematoma volume among subgroups stratified by the baseline used to define early neurological deterioration. As a result, future research must define early neurological deterioration in precise terms.

This meta-analysis revealed that the presence of intraventricular hemorrhage and intraventricular extension were also predictive factors for early neurological deterioration. Intraventricular hemorrhage and intraventricular extension following intracerebral hemorrhage are hypothesized to be related to inflammation, hydrocephalus, and the neurotoxic effects of blood products on the diencephalon, thus contributing to early neurological deterioration [7, 25, 26]. These findings emphasize the necessity of early neuroimaging scans with CT/MRI in patients with intracerebral hemorrhage, as well as the need for a follow-up scan within 24 h for patients who have stable examination and consciousness [27].

Our meta-analysis also found that hematoma expansion is associated with an increased risk of early neurological deterioration. Hematoma expansion has been attributed to continued bleeding caused by original etiologies or subsequent vascular injury in 30–40% of intracerebral hemorrhage patients within 6 h of onset [28, 29]. Changes in hemodynamics may contribute to early neurological deterioration. Rodriguez-Luna, Coscojuela [30] also reported that the rate of hematoma expansion was a predictor of early neurological impairment. These findings highlight the clinical significance of predicting, monitoring, and stopping hematoma expansion in the early stage of intracerebral hemorrhage. The non-contrast computed tomography (NCCT) markers, such as heterogeneous densities within the hematoma or irregularities at its margins, have been reported as reliable predictors of hematoma expansion [27]. These markers help to improve the triage strategy and monitoring density in clinical practice.

In our meta-analysis, the spot sign observed by computed tomographic angiography was also revealed to be a predictor of early neurological deterioration. The computed tomographic angiography spot sign denotes contrast leakage into the hematoma. Previous research has revealed that the spot sign on computed tomographic angiography is related to persistent bleeding or rebleeding caused by coagulopathy within hematomas, systemic coagulopathy, or a reaction to rapid consumption of hemostatic substances [31, 32]. Continuous bleeding increases the risk of hematoma expansion and intraventricular hemorrhage [31, 33, 34], both of which can result in early neurological deterioration. This result, however, should be regarded with caution because this meta-analysis included data from only two studies with 385 individuals. More large-scale studies are needed in the future to investigate the predictive value of the spot sign for early neurological deterioration.

Our meta-analysis did not find a statistically significant association between age and early neurological deterioration in patients with intracerebral hemorrhage, in contrast to common perceptions [35]. More studies on the association between age and early neurological deterioration are needed. In fact, numerous variables, such as plasma leptin, visfatin, osteopontin, adrenomedullin, low-density lipoprotein cholesterol (LDL-C) levels, and anticoagulation treatment (Table 1), were identified as predictors of early neurological deterioration. These variables could not be synthesized in this meta-analysis because each variable was reported in only one study. However, these findings still offer new insights for healthcare providers to predict early neurological deterioration in patients with intracerebral hemorrhage and to guide further studies.

There are strengths and limitations in this study. This review was conducted using systematic and rigorous approaches. Only high-quality studies adjusted for relevant confounders and identified predictors using multivariate logistic regression analyses were included. Subgroup analysis and sensitivity analysis were used to address heterogeneity and increase the robustness of the pooled results. This meta-analysis has several limitations. First, several factors, such as plasma leptin, visfatin, and osteopontin levels were not further analyzed since they were reported in a small number of studies. Second, this study excluded unpublished articles and studies that did not employ multivariate logistic regression analyses, which may have introduced publication bias, especially influencing the pooled prevalence of early neurological deterioration. Third, the publication language was limited to English, which limited the comprehensiveness of the included studies. As a result, future research should overcome these limitations and examine predictors of early neurological deterioration in a more comprehensive manner.

Conclusion

This systematic review and meta-analysis found that the pooled prevalence of early neurological deterioration was 23%. Such a high prevalence emphasizes the importance of early neurological deterioration in patients with intracerebral hemorrhage. This meta-analysis also found that the initial NIHSS score, hematoma volume, hematoma expansion, intraventricular hemorrhage, intraventricular extension, and spot sign were all predictors of early neurological deterioration. By identifying these factors, healthcare providers may be better prepared to identify and manage high-risk patients, thereby improving outcomes. However, it is important to take publication bias into consideration when interpreting the findings. Further research and validation of these predictors in clinical practice are needed to fully integrate these findings into therapeutic practice.