Introduction

Intracerebral hemorrhage (ICH) score has been widely used as a consistent and reliable clinical grading scale for predicting mortality. However, ICH score had not been used to predict good outcome or significant disability for those who were alive. We intended to address whether any modifications would increase prediction accuracy for mortality as well as the extent of morbidity for those who survived.

Methods

We conducted a retrospective cohort study, involving all non-traumatic ICH patients admitted to our hospital between September 2018 and July 2020. All non-traumatic ICH patients who were admitted to the stroke unit and registered in our stroke database had their medical records, neuroimaging, and laboratory test results reviewed. Only patients with complete medical records and available CT imaging and laboratory test results were included in our study. Independent predictors of mortality (modified Rankin scale/mRS of 6) or good outcome vs. significant disability (mRS≤2 vs. mRS 3–5, respectively) were identified by logistic regression. A modified ICH (mICH) score was compared with the original ICH (oICH) score for its diagnostic performance (DP). Overall DPs were graded and ranked according to Youden Index (YI).

Results

As many as 311 patients were eligible with both 39.9% rate of 30-day mortality and good outcome. Factors independently associated with mortality were low GCS and high NIHSS on admission (P = 0.002, <0.001, respectively), and presence of respiratory failure (P < 0.001). Independent factors for good outcome were low NIHSS on admission and mass effect (midline shift > 5 mm) [both P < 0.001]. A modification of ICH score from the original was made by substituting GCS with NIHSS (0 –10 = 1; 11 – 20 = 2; >20 = 3), changing age cut-off point to > 55 years old (= 1), and adding respiratory failure (= 1), and mass effect (= 1). Overall, mICH scored better over oICH score with respect to sensitivity and had comparable specificity for both 30-day mortality and good outcome (sensitivity 80.6% vs. 50.8%; specificity 88.7% vs. 89.3%; YI 0.69 vs. 0.40, respectively) and good outcome (sensitivity 86.3% vs. 77.4%; specificity 74.6% vs. 77.8%; YI of 0.61 vs. 0.55, respectively). There was only one patient with oICH and none on mICH score of 0, who died and none survived with oICH and mICH score of ≥ 5 and ≥ 7, respectively. The proportion of 30-day mortality and good outcome increased in a more linear fashion with mICH score.

Conclusions

The mICH score was proven to be reliable and consistent as a risk grading assessment for non-traumatic ICH patients. The mICH was statistically superior to oICH score in predicting 30-day mortality and good outcome.

中文翻译：

---

改良的 ICH 评分在评估 30 天死亡率和非创伤性脑出血的良好结果方面优于原始 ICH 评分

介绍

脑出血（ICH）评分已被广泛用作预测死亡率的一致且可靠的临床分级量表。然而，ICH 评分并未用于预测存活者的良好结局或严重残疾。我们打算解决任何修改是否会提高死亡率的预测准确性以及幸存者的发病程度。

方法

我们进行了一项回顾性队列研究，纳入了 2018 年 9 月至 2020 年 7 月期间入住我院的所有非创伤性 ICH 患者。所有入住脑卒中病房并在我们的脑卒中数据库中登记的非创伤性 ICH 患者都有他们的病历、神经影像学，并且审查了实验室测试结果。我们的研究仅包括具有完整医疗记录和可用 CT 成像和实验室检查结果的患者。通过逻辑回归确定死亡率（改良 Rankin 量表/mRS 为 6）或良好结果与显着残疾（分别为 mRS≤2 与 mRS 3-5）的独立预测因子。将修改后的 ICH (mICH) 评分与原始 ICH (oICH) 评分的诊断性能 (DP) 进行比较。根据约登指数 (YI) 对整体 DP 进行分级和排名。

结果

多达 311 名患者符合条件，30 天死亡率为 39.9%，结果良好。与死亡率独立相关的因素是入院时低 GCS 和高 NIHSS（分别为P  = 0.002，<0.001），以及存在呼吸衰竭（P  <0.001）。获得良好结果的独立因素是入院时 NIHSS 低和占位效应（中线偏移 > 5 mm）[均为P < 0.001]。通过将 GCS 替换为 NIHSS (0 –10 = 1; 11 – 20 = 2; >20 = 3) 对原始 ICH 评分进行了修改，将年龄截止点更改为 > 55 岁 (= 1)，并添加呼吸衰竭（= 1）和质量效应（= 1）。总体而言，mICH 在敏感性方面得分优于 oICH 评分，并且在 30 天死亡率和良好结果方面具有相当的特异性（敏感性分别为 80.6% 和 50.8%；特异性 88.7% 和 89.3%；YI 分别为 0.69 和 0.40）和良好的结果（敏感性分别为 86.3% 对 77.4%；特异性 74.6% 对 77.8%；YI 分别为 0.61 对 0.55）。只有 1 例 oICH 患者，mICH 评分 0 分，oICH ≥ 5 和 mICH ≥ 7 分的患者死亡，无存活者。30 天死亡率和良好结局的比例随着 mICH 评分呈线性增加。

结论

mICH 评分被证明是可靠且一致的，可作为非创伤性 ICH 患者的风险分级评估。在预测 30 天死亡率和良好结果方面，mICH 在统计学上优于 oICH 评分。