Data of aSAH patients were gathered from the Long-term Prognosis of Emergency Aneurysmal Subarachnoid Hemorrhage (LongTEAM) Registry study (Registration No. NCT04785976). This observational cohort investigations was conducted at Beijing Tiantan Hospital in China from January 2015 to September 2022. All patients received a brain computed tomography (CT) scan, CT angiography (CTA) or digital subtraction angiography (DSA) to confirm the diagnosis of aSAH. The criteria for inclusion were: (1) older than 18 years; (2) presence of single aneurysm; (3) treated with surgical clipping or endovascular coiling within 72 h of onset; (4) admission through the emergency department; (5) completing follow-up. The criteria for exclusion included: (1) a prior history of SAH or other neurosurgical disease; (2) previous craniotomy or intracranial vascular interventions; (3) physical disability; (4) absence of data, laboratory tests, and radiological information; (5) patients with active or chronic inflammatory disease; (6) receipt of immunosuppressive therapy in the 6 months before admission. Informed consent was secured from patients or their authorized representatives prior to enrollment.

Demographic data encompassed age, gender, current smoking, current drinking, diabetes, hypertension, history of heart disease, history of antiplatelet, history of anticoagulant. The preoperative clinical status included the World Federation of Neurosurgical Societies (WFNS) grade, modified Fisher Scale (mFS) grade, Graeb score, Subarachnoid Hemorrhage Early Brain Edema Score (SEBES), Glasgow Coma Scale (GCS), Hunt Hess score, intraventricular hemorrhage (IVH), preoperative hydrocephalus, loss of conscious. Max diameter of aneurysm and treatment modality were also collected. The postoperative complications included major adverse cardiovascular events (MACE), delayed cerebral ischemia (DCI), postoperative intracranial infection, postoperative stress ulcer bleeding, anemia, pneumonia, and deep vein thrombosis (DVT). The detailed diagnostic criteria for these postoperative complications were shown in Supplementary Table 1. Fasting blood samples were collected within 24 h of admission after aSAH onset. The laboratory test contained glucose, Crea, aspartate aminotransferase (AST), alanine aminotransferase (ALT), total protein (TP), albumin, globulin, white blood cell (Wbc), monocyte, neutrophil, red blood cell (Rbc), hemoglobin, platelet and lymphocyte. The normal tests value range were shown in Supplementary Table 2. Some inflammation and nutritional markers, such as neutrophil to lymphocyte rate (NLR), systemic inflammation response index (SIRI), and prognostic nutritional index (PNI) were also collected. The following equations were applied to calculate these markers: HALP = Albumin (g/L) * Hemoglobin (g/L)* Lymphocyte (10⁹/L)/Platelet (10⁹/L), NLR = neutrophils/lymphocytes, SIRI = neutrophils × monocytes/lymphocytes, and PNI = Albumin (g/L) + Lymphocyte (10⁹/L) * 5 (7, 8).

The modified Rankin Scale (mRS) is a widely recognized neurological outcome assessment tool that evaluates functional outcomes in patients. This scale assigns scores ranging from 0, indicating the absence of any symptoms, to 6, which signifies the death of the patient. In this investigation, mRS was utilized to evaluate the clinical outcome of patients. At 90 days after discharge, patients would receive follow up via telephone or an outpatient appointment. Unfavorable outcome was defined as an mRS score from 3 to 6.

Normally distributed variables were presented as the means ± standard deviation (SD), while variables with skewed distribution were expressed as medians (25–75th percentile). Based on the tertiles of HALP, aSAH patients were divided into 3 groups (HALP < 30.286, 30.286 ≤ HALP ≤ 43.201, and HALP > 43.201). The Cochran–Armitage test was applied to assess the trends in categorical data, whereas one-way ANOVA was employed for continuous data. The relationship between HALP and clinical outcome in aSAH patients was analyzed through multivariate logistic regression analysis. Three adjusted models were established to exclude the effects of confounding factors. In the crude model, age and sex were adjusted as confounders. The minimally adjusted model included crude model, GCS, WFNS, Hunt Hess score, loss of consciousness, pneumonia, and MACE. The fully adjusted model accounted for minimally adjusted model, Wbc, monocyte, glucose, and Crea. Additionally, we also established restricted cubic spline (RCS) to evaluate the dose–response relationships between HALP and prognosis of aSAH. The number of knots were determined by the lowest Akaike information criterion (AIC) value. Receiver operating characteristic (ROC) analysis was conducted to assess the predictive capability of HALP for clinical outcome. Net reclassification improvement (NRI) and integrated discrimination improvement (IDI) were employed to assess the incremental predictive value of HALP in conjunction with conventional risk factors. Delong test was used to analyze the difference of predictive ability between HALP, NLR, SIRI, and PNI. The robustness of correlation between HALP and functional outcome of aSAH was assessed through subgroup analysis. Finally, we also explored the relationship between HALP and postoperative complications. This study calculated the sample size based on Events Per Variable (EPV), a widely recognized metric in statistical analysis (9). The detailed method was presented in Supplementary material. All statistical analyses were performed using R version 4.4.0 Statistical Software.

From January 2015 to September 2022, a total of 1,268 aSAH patients were included in the LongTEAM registry study. 106 patients were lost to follow up; 74 patients with a prior history of neurosurgical condition, and 340 patients with a missing laboratory test. After excluding 520 ineligible aSAH patients, 748 patients were enrolled in this study. The flow chart of patients enrollment was exhibited in Supplementary Figure 1. Among these patients, 210 aSAH patients (28.07%) had unfavorable functional outcomes.

The baseline characteristics analysis was shown in Table 1. 320 (42.78%) patients were male, 67 (8.96%) patients with Hunt Hess score 4–5, and 394 (52.67%) patients received surgical clipping. According to the baseline HALP levels (HALP < 30.286, 30.286 ≤ HALP ≤ 43.201, and HALP > 43.201), we categorized aSAH patients into three groups. Compared to patients with high HALP level, those with lower HALP levels tended to be females, with a greater percentage of WFNS grade 4–5 cases, with a greater percentage of Hunt Hess grade 4–5 cases, with a higher incidence of preoperative loss of consciousness, had a higher prevalence of postoperative MACE and pneumonia complications.

Additionally, patients with higher HALP were more likely to have lower levels of glucose, Crea, Wbc, monocyte, NLR, and SIRI, but higher levels of hemoglobin and lymphocyte. Furthermore, patients with high HALP levels demonstrated more favorable clinical outcomes.

The relationship between baseline HALP score levels and the risk of unfavorable functional outcome was presented in Table 2. In the fully adjusted model, the risk of unfavorable functional outcome decreased with each increment in HALP score levels Odds Ratio: 0.095, 95% Confidence Interval: 0.056–0.163, p value < 0.001. After adjusting for age, gender, loss of consciousness, pneumonia, preoperative GCS, MACE, preoperative WFNS, Wbc, preoperative Hunt Hess score, monocyte, glucose, and Crea, the RCS analysis showed there was an intriguing L-shaped relationship between HALP score levels and prognosis of aSAH (p for non-linear < 0.001), suggesting that as HALP score values increased, there was a progressive decrease in the risk of poor functional outcome (shown in Figure 1).

The ROC curve indicated that the area under the curve (AUC) was 0.795. The optimal cut-off value was determined by the maximal Youden index (sensitivity + specificity−1). In this research, the optimal HALP cut-off value as a predictor was 30.456 (sensitivity: 79.6%, specificity: 67.6%). The result of ROC analysis identified that HALP had significant potential in predicting unfavorable functional outcomes in patients with aSAH (Table 3).

To compare the predictive ability of HALP with other single inflammatory or nutritional indicators, we analyzed the ROC curve of NLR, SIRI, and PNI, as shown in Figure 2. The Delong test revealed that HALP score demonstrated a significantly greater predictive capability compared to the other three indicators (Supplementary Table 3). Furthermore, the NRI and IDI were calculated to determine whether adding the HALP score to TAPS mode enhances risk prediction for adverse prognosis in patients with aSAH. As presented in Table 4, the inclusion of the HALP score in the TAPS model (which comprises age, WFNS, mFS, Graeb score, white blood cell count, and surgical clipping) resulted in a notable enhancement in risk reclassification for unfavorable prognosis (NRI: 48.40%, p < 0.001; IDI: 19.09%, p < 0.001).

Table 5 shows the results of subgroup analysis for HALP. Based on the cut-off value of HALP score, we divided patients into low-HALP group (HALP < 30.456) and high-HALP group (HALP ≥ 30.456). The results of the analysis indicated that there were no notable interactions observed between the HALP score and the predetermined factors that were specified in the study. All p-values associated with these interactions were greater than 0.05, suggesting that any potential relationships between the HALP score and the identified factors were not statistically significant. This finding implied that the HALP score’s effects remain consistent across the various factors considered, reinforcing the robustness of the results presented in this research.

Among the 748 patients, pneumonia was identified as the most prevalent postoperative complication, occurring in 278 patients (37.16%), followed by anemia, which occurred in 273 patients (36.50%). After adjusting for age, gender, GCS, WFNS, Hunt Hess score, loss of consciousness, Wbc, monocyte, glucose, and Crea, multivariate logistic regression analysis revealed that patients in the highest HALP groups (>43.201) exhibited the lowest risk of pneumonia (OR: 0.586, 95% CI: 0.388–0.887, p = 0.012, as illustrated in Table 6). The RCS curve showed a significant linear relationship between pneumonia and HALP levels (p for non-linear = 0.445, p for overall = 0.004, as illustrated in Figure 3).