This study recruited patients from our prospectively maintained patients-with-ICH database (RIS-MIS-ICH ClinicalTrials.gov Identifier: NCT03862729) between January 2015 and October 2019. The criteria for enrollment were as follows: (1) ICH diagnosed by emergent CT or computed tomographic angiography (CTA) within 24 h; (2) no cerebral hernia at admission and hematoma volume more than 20 ml; (3) no obstructive hydrocephalus caused by IVH; (4) patients with GCS score > 8 at admission; (5) initially treated by conservative approaches and no emergency surgical intervention was arranged. Ethical approval was obtained through the relevant ethics committee of the First Affiliated Hospital of Fujian Medical University (Ethical Approval Number: MRCTA, ECFAH of FMU [2018] 082-1). Additionally, this study also followed the relevant Chinese laws, regulations, and guidelines, as well as international laws and regulations.

The patients in this study were treated according to two guidelines, namely Chinese multidisciplinary expert consensus: Diagnosis and Treatment of Spontaneous Cerebral Hemorrhage (2015) and Guidelines for the Management of Spontaneous Intracerebral Hemorrhage: A Guideline for Healthcare Professionals From the American Heart Association/American Stroke Association (2015) (12). For patients with ICH along with hematoma volume more than 20 ml, with neurological dysfunction but without cerebral hernia, conservative treatment and minimally invasive surgery (MIS) were recommended as the first-line treatment. The consciousness and neurological functions of patients treated by conservative therapy were closely observed. If the GCS decreased 2 or more scores or the functional defect worsened, conversion to surgical treatment was recommended. CT scanning was performed in the conservative treatment group at the 6, 12, 24, 48, and 72 h after stroke.

All the data retrieved from our prospective ICH patient registration database including: (1) personal information: gender, age, etc.; (2) history of present illness: time of onset, initial symptoms, pre-hospital management; (3) past medical history: hypertension, diabetes, anticoagulants, antiplatelet drugs, antihypertensive drugs, and other drugs; (4) physical examination: vital sign (blood pressure/respiration rates/temperature/heart rates), neurological examination (pupillary reflex, GCS, the strength of contra-lesional extremities, etc.); (5) laboratory assay: coagulation function, blood routine test, etc.; (6) radiological imaging: (a) initial CT: hematoma location, depth of hematoma, whether with IVH or not, hematoma volume, PHE, hematoma shape and density categorical scales score (13), black hole sign, island sign, swirl sign, blood-fluid level in a hematoma (2, 3). Additionally, hematoma shape and density categorical scales scores include 2 novel 5-point categorical scales, ranging from Category 1 (most regular shape and most homogeneous density) to Category 5 (most irregular shape and most heterogeneous density) (13); (b) repeat CT: expansion of hematoma and/or PHE, whether cerebral infarction occurred, whether hydrocephalus occurred; (7) post-hospitalization assessments: neurological examination daily, and mainly focused on GCS score and functional deficits; (8) follow-up data: all patients were followed up every 6 months after a stroke at the outpatient department or by the phone call. This study focused on mRS score, all-cause mortality, and follow-up events (re-hemorrhage, cerebral infarction, seizures, etc.) 1 year after stroke. Thus, HE was defined as a 33% increase in hematoma volume or an absolute increase of 6 ml in a repeat CT scan. Similarly, expansion of PHE was defined as an increase in the ratio of PHE volume to hematoma volume or an absolute increase of 5 ml in the repeat CT scan. The early deterioration was defined as a sudden decline of GCS more than 2 or conversion to surgical treatment within 72 h after admission. The poor prognosis was defined as mRS ≥ 4 at 1 year after stroke.

For data collection, data management, and quality control, refer to the registration information of the RIS-MIS-ICH study on the ClinicalTrials.gov website, which is as follows: https://www.clinicaltrials.gov/ct2/show/NCT03862729?term=fuxin$+$lin&draw=2&rank=2. The project-related data was collected through the electronic data registry system (Real Data Medical Research Inc.). (eRDDM syestem, Ningbo, Zhejiang Province, China) And the radiological image information and follow-up data were determined by two neuroradiologists with consensus.

Demographic and clinical characteristics were descriptively summarized as mean (SD) for continuous variables and as number (%) for categorical variables. The chi-square test and Fisher's exact test were used for categorical variables, while Student's t-test was used for continuous variables. Only the variables with a P < 0.1 on univariate analysis were enrolled in multivariate analysis. The independent risk factors of early deterioration and poor prognosis 1 year after stroke were determined by multivariate analysis, respectively. The corresponding scores were assigned based on the weights of each independent risk factor. Accordingly, a prediction model of early deterioration and prognosis 1 year after stroke was established. The receiver operating characteristic (ROC) curves were drawn and analyzed, to find the cut-off point and calculate the AUC. P < 0.05 was considered significant.

A total of 632 patients with ICH were screened (Supplementary Figure 1), of which only 376 patients were within 24 h from the onset to the initial CT. The results showed that 158 patients were without cerebral hernia at admission and with a hematoma volume of more than 20 ml. Among them, 35 patients chose surgical treatment as the primary treatment modality, while other 123 patients preferred the initial conservative treatment. However, only 86 (70%) patients from the initial conservative treatment group had complete follow-up data.

Among 123 patients who received initial conservative treatment, the average age was 58.1 ± 13 years old, including 93 men (75.6%) and 30 women (24.4%). The average admission GCS score was 11.7 ± 2.6, and the average time from onset to the initial CT scan was 11.5 ± 8.3 h. The initial CT showed that 89 patients (72.4%) had basal ganglia hemorrhage (deep), 38 patients (30.9%) had hematoma less than 1 cm from the cortex (superficial), and 46 patients (37.4%) had IVH. The average hematoma volume was 37.7 ± 14.6 ml, and the average PHE volume was 14.9 ± 14.4 ml. According to the previously published literature (13), the results of hematoma shape categorical scale scores were as follows: 21 (17.1%) scored 1, 44 (35.8%) scored 2, 35 (28.5%) scored 3, 17 (13.8%) scored 4, and 6 (4.9%) scored 5. The hematoma density categorical scale score of 1 to 4 were 56 (45.5%), 42 (34.1), 18 (14.6%), and 7 (5.7%), respectively. Based on morphologic features of hematoma on initial CT, the number of patients with black hole sign, island sign, swirl sign, the blood-fluid level was 36 (29.3%), 61 (49.6%), 20 (16.3%), and 7 (5.7%), respectively. During 72 h after admission, 62 (50.4%) were converted to surgical treatment, 67 (54.5%) suffered from early deterioration. The average GCS score at discharge and mRS score at discharge was 12.6 ± 3.3 and 3.3 ± 1.2, respectively. Among the 86 (70%) patients with complete follow-up data at 1 year, 27 (31.4%) had seizures, 3 (3.5%) had re-hemorrhage, and 1 (1.2%) had cerebral infarction. The 1-year-follow-up mRS score showed as follows: 15 (17.4%) scored 1, 22 (25.6%) scored 2, 9 (10.5%) scored 3, 27 (31.4%) scored 4, 6 (7%) scored 5, and 7 (8.1%) scored 6. Therefore, 40 patients (46.5%) had poor prognosis (mRS ≤ 4) (Table 1).

The univariate analysis indicated that admission GCS score (OR (odds ratio), 1.35; 95% CI,1.16-1.58; P < 0.001), time from onset to initial CT (OR, 1.04; 95% CI, 0.1–1.09; P = 0.061), hematoma volume (OR, 0.91; 95% CI, 0.88–0.95; P < 0.001), volume of hematoma and PHE (OR, 0.97; 95% CI, 0.95–0.98; P < 0.001), hematoma shape categorical scale score (OR, 0.66; 95% CI, 0.46–0.94; P = 0.021), black hole sign (OR, 2.05; 95% CI,0.91-4.6; P = 0.083), island sign (OR, 2.84; 95% CI, 1.36–5.92; P = 0.005) were the risk factors for early deterioration of patients with ICH. Then, these factors were included in the Logistic multiple regression model. The results showed that admission GCS score (OR, 1.43; 95% CI, 1.18–1.74; P < 0.001) and hematoma volume (OR, 0.9; 95% CI, 0.84–0.97; P = 0.003) were the independent risk factors (Table 2). Given the statistical analysis results and facilitating clinical practice, 13 < GCS ≤ 15 was assigned a score of 1, 10 < GCS ≤ 13 was assigned a score of 2, and 8 ≤ GCS ≤ 10 was assigned a score of 3. Additionally, hematoma volume ≤ 30 ml was assigned a score of 1, and volume of hematoma > 30 ml was assigned a score of 2. A scoring model for early deterioration was constructed to predict disease progression of patients with ICH at the early stage (Table 3). ROC curve analysis indicated that the prediction model is strongly associated with the early deterioration of patients with ICH, with an AUC of 0.778 and a cut-off point of 3.5 (Supplementary Figure 2). In brief, for the patients with ICH with a score of 2 (13.3%) or 3 (36.4%), the risk of deterioration during hospitalization was low, but for patients with a score of 4 (67.6%) or 5 (86.7%), the risk of deterioration during hospitalization was higher (Figure 1).

The univariate logistic regression analysis indicated that with a history of hypertension (OR, 0.36; 95% CI, 0.13–1; P = 0.049), admission GCS score (OR, 0.8; 95% CI, 0.67–0.95; P = 0.012), hematoma location (OR, 0.07; 95% CI, 0.02–0.32; P = 0.001), depth of hematoma (OR, 0.28; 95% CI, 0.1–0.79; P = 0.016), hematoma volume (OR, 1.04; 95% CI, 1–1.08; P = 0.032), island sign (OR, 0.35; 95% CI, 0.15–0.85; P = 0.019) were the risk factors for the poor prognosis at 1 year after stroke. Then, these factors above were included in the multiple logistic regression model, and the results showed that hematoma location (OR, 0.027; 95% CI, 0.004–0.17; P < 0.001) and hematoma volume (OR, 1.09; 95% CI, 1.03-1.15; P <0.001) were independent risk factors. The island sign (OR, 0.5; 95% CI, 0.16–1.57; P = 0.051) had a trend to significant (Table 4). Based on the statistical analysis results and to facilitate clinical use, hematoma volume ≤ 25 ml was assigned a score of 1, and hematoma volume > 25 ml was assigned a score of 2. The lobar hematoma was assigned a score of 1, but basal ganglia hematoma was assigned a score of 3. Additionally, the appearance of the island sign on the initial CT scan was assigned a score of 1. Then, a scoring model for predicting the prognosis 1 year after stroke was constructed (Table 3). ROC curve analysis indicated that the prediction model was strongly associated with the prognosis of patients with ICH 1 year after stroke, with an AUC of.792 and a cut-off point of 4.5 (Supplementary Figure 3). In brief, for the patients with a score of 2 to 4, the risk of poor prognosis in 1 year after stroke was low, but for the patients with a score of 5 (29.2%) or 6 (52.8%), the risk of poor prognosis in 1 year after stroke was higher (Figure 2). To facilitate the clinical evaluation, the various types of prognostic scores were summarized in Figure 3.