We first excluded patients who underwent emergency surgery after admission, as surgery may impact the progression of hematoma. Based on this criterion, 113 patients who underwent emergency surgery after admission were excluded from the initial cohort of 865 patients with CC, leaving a final sample size of 248 patients for analysis. Among them, 96 patients with severe liver, spleen, and kidney organ damage on admission were excluded, along with 80 patients with incomplete clinical and imaging data, 26 patients aged below 18 years or pregnant, and 11 patients with a history of liver or kidney dysfunction, malignant tumors, or other hematological disorders. A total of 248 patients were enrolled in the study, including 142 patients in the nonprogress group and 106 patients in the progress group (Table 1). The average age of both groups was 58 years, with 154 male patients (62.1%). The mean age of the progress group (62.5 years) was significantly higher than that of the nonprogress group (57.5 years; P = 0.014). However, there were no significant differences between the nonprogress and progress groups in terms of GCS score on admission, hypertension, diabetes, neutrophil count, platelet levels, dNLR, prothrombin time, APTT, and creatinine levels (P > 0.05). Regarding the history of antiplatelet/anticoagulant drug use, the proportion of patients in the progress group was significantly higher than that in the nonprogress group (P = 0.004), indicating the potential impact of antiplatelet/anticoagulant drugs on the hematoma progression of CC in patients. It is worth noting that there were 15 surgical patients in the progress group, while there were no surgical patients in the nonprogress group. This is because in our center, conservative treatment is adopted for all patients with CC if the follow-up head CT shows no progression of the hematoma. When patients exhibit hematoma progression, such as a significant increase in hematoma volume or indications of elevated intracranial pressure leading to cerebral herniation, we promptly initiate emergency surgical treatment. The surgical procedures encompass contusion removal, intracranial pressure monitoring probe insertion, and decompressive craniectomy. Additionally, significant statistical differences were observed between the two groups in terms of NLR, SII, MLR, PLR, and D-dimer (P < 0.05). In terms of imaging findings, compared with the nonprogress group, the progress group had a higher proportion of simultaneous subdural hematoma (SDH) and epidural hematoma (EDH), as well as a greater number of cases with bilateral hematomas (SDH or EDH) present simultaneously, showing statistically significant differences (P < 0.005). As shown in Fig. 3, MLR levels were correlated with white blood cell count (r = 0.292, P < 0.001), neutrophil (r = 0.366, P < 0.001), and platelet (r =  − 0.227, P < 0.001), among other clinical factors.

Scatter plot of Spearman correlation analysis between MLR and white blood cells, platelets, and neutrophils. MLR, monocyte-to-lymphocyte ratio

As depicted in Fig. 4, the levels of MLR were significantly elevated in the progress group of patients compared with the nonprogress group (P < 0.001). Patients were divided into three subgroups based on their MLR levels: tertiles 1 (< 0.52), tertiles 2 (0.52–0.91), and tertiles 3 (≥ 0.91). The proportion of patients with high MLR levels who experienced progression of hematoma was much higher than that of patients with medium MLR levels and low MLR levels, while the proportion of patients with medium MLR levels experiencing progression was lower than that of patients with low MLR levels (67.5% vs. 26.8% vs. 33.7%, respectively; P < 0.001; Fig. 5). As shown in Table 2, compared with patients with medium MLR levels or low MLR levels, patients with high MLR levels of CC showed no significant differences in age, initial GCS score, hypertension, diabetes, history of antiplatelet/anticoagulant use, surgery, white blood cell count, activated APTT, creatinine, type of intracranial hematoma (SDH and EDH), and hematoma laterality (left and right; P > 0.05). However, they exhibited higher levels of neutrophils, lymphocytes, platelets, NLR, SII, PLR, derived dNLR, monocytes, and D-dimer (P < 0.05).

Violin plot of MLR levels in the progress and nonprogress groups. MLR, monocyte-to-lymphocyte ratio

Incidence rates of the progress and nonprogress groups categorized by MLR levels. MLR, monocyte-to-lymphocyte ratio

Table 3 displays the results of the multivariable logistic regression analysis on the factors influencing the hematoma progression of CC in patients. Initially, we conducted univariate analysis on all factors influencing the progression of hematoma, and then included all factors with a P value < 0.05, as well as the GCS score on admission that could potentially affect progression, in the multivariable analysis. We observed that although the univariate analysis showed that the GCS score on admission was not a significant risk factor for progression (P = 0.083), after adjusting for confounding variables, the results indicated that higher GCS scores on admission were associated with a reduced risk of hematoma progression in patients with CC (Odds Ratio [OR] 0.838, 95% Confidence Interval [CI] 0.729–00.964, P = 0.012). Furthermore, if patients had a history of anticoagulant/antiplatelet use on admission, it may indicate a higher risk of hematoma progression, which warrants attention from clinicians (OR 1.338, 95% CI 1.118–01.971, P = 0.044). Similarly, a high level of MLR was independently associated with the occurrence of hematoma progression in patients with CC (OR 3.546, 95% CI 1.187–010.597, P = 0.023). Moreover, higher levels of white blood cells on admission in patients also indicated a higher likelihood of hematoma progression (OR 1.157, 95% CI 1.013–01.321, P = 0.031) (Fig. 6).

Forest plot demonstrating factors influencing the hematoma progression of cerebral contusion in multivariable logistic regression analysis. CI, confidence interval, dNLR, derived neutrophil-to-lymphocyte ratio, EDH, epidural hematoma, GCS, Glasgow Coma Scale, MLR, monocyte-to-lymphocyte ratio, NRL, neutrophil-to-lymphocyte ratio, PLR, platelet-to-lymphocyte ratio, SDH, subdural hematoma, SII, systemic immune-inflammation index

Based on the results of the aforementioned multivariable logistic regression analysis, we constructed three prediction models using ROC curves. As shown in Fig. 7, model 1 included two factors: GCS score on admission and anticoagulant/antiplatelet therapy, with an AUC of 0.597. Model 2 added white blood cell to the previous two factors, resulting in GCS score on admission + anticoagulant/antiplatelet therapy + white blood cell, and the predictive model had an AUC of 0.624. When MLR was included as a factor in model 2, the predictive accuracy of the model significantly improved (AUC = 0.754).

Models for predicting the hematoma progression of cerebral contusion based on logistic regression analysis. Model 1: GCS score on admission + anticoagulant/antiplatelet therapy. Model 2: GCS score on admission + anticoagulant/antiplatelet therapy + white blood cell. Model 3: GCS score on admission + anticoagulant/antiplatelet therapy + white blood cell + MLR. AUC, area under the curve, GCS, Glasgow Coma Scale, MLR, monocyte-to-lymphocyte ratio