Abstract
目的
探讨红细胞分布宽度/红细胞压积 (RDW/HCT) 比值与脓毒症患者28 d全因死亡率的相关性。
方法
使用MIMIC-IV数据库进行回顾性分析,并利用eICU-CRD数据库进行验证。分别从上述数据库中纳入27 335例及17 406例成年脓毒症患者。根据RDW/HCT比值四分位数分组,主要结局指标为28 d全因死亡率。采用Kaplan-Meier法评估各组生存率,利用多变量Cox回归模型和限制性立方样条(RCS)分析RDW/HCT比值与28 d死亡风险的相关性,并进行亚组分析。
结果
MIMIC-IV队列和eICU-CRD队列脓毒症患者28 d死亡率分别为19.3%与16.0%。在MIMIC-IV队列中,Kaplan-Meier(K-M)生存曲线显示:RDW/HCT比值升高与28 d死亡风险增加相关(P<0.0001)。多变量Cox回归分析显示:RDW/HCT比值升高是28 d死亡的独立危险因素(HR=1.89,95% CI:1.52~2.36,P<0.001)。RCS分析显示RDW/HCT比值与28 d死亡率呈非线性“U”型关系。亚组分析未发现显著交互作用。eICU-CRD验证队列的结果趋势与MIMIC-IV数据库一致。
结论
RDW/HCT比值升高与脓毒症患者28 d全因死亡率增加相关。
Keywords: 脓毒症, 红细胞分布宽度/红细胞压积比值, 28 d全因死亡率, MIMIC-IV/eICU-CRD数据库
Abstract
Objective
To investigate the association between the red blood cell distribution width to hematocrit (RDW/HCT) ratio and 28-day all-cause mortality in patients with sepsis.
Methods
A retrospective cohort study was conducted using the MIMIC-IV database (27 335 adult patients), and the eICU-CRD database was used for external validation (17 406 adult patients). The patients were grouped by RDW/HCT ratio quartiles. The primary outcome was 28-day all-cause mortality. Kaplan-Meier survival curves were used to assess the survival rates. Multivariable Cox regression and restricted cubic spline (RCS) analyses were performed to examine the association between RDW/HCT ratio and patient mortality. Subgroup analyses were conducted to assess the robustness of the findings.
Results
The 28-day mortality rates were 19.3% in the MIMIC-IV cohort and 16.0% in the eICU-CRD cohort. In the MIMIC-IV cohort, Kaplan-Meier analysis showed that a higher RDW/HCT ratio was associated with an increased risk of 28-day mortality in septic patients (log-rank P<0.0001). Multivariable Cox regression analysis suggested that an elevated RDW/HCT ratio was an independent risk factor for 28-day mortality in septic patients (HR=1.89, 95% CI: 1.52-2.36, P<0.001). RCS analysis showed a non-linear, U-shaped relationship between the RDW/HCT ratio and 28-day mortality of septic patients. No significant interactions were observed in subgroup analyses. The trends in the eICU-CRD validation cohort were consistent with those observed in the MIMIC-IV cohort. Conclusion Elevation of the RDW/HCT ratio is associated with an increased 28-day all-cause mortality rate in patients with sepsis.
Keywords: Sepsis, Red blood cell distribution width/hematocrit, 28-day all-cause mortality, MIMIC-IV/ eICU-CRD database
脓毒症是由感染引发的,以机体免疫反应失调为特征的一类疾病。脓毒症相关死亡约占全球总死亡人数的20%[1, 2]。早期识别高危患者并实施针对性干预,对于降低病死率具有重要的临床意义。目前,降钙素原(PCT)、C反应蛋白(CRP)和中性粒细胞表面抗原CD64等生物标志物已被证实可预测脓毒症早期预后[3, 4],但其在临床应用中仍存在局限性。如PCT在非感染性炎症时亦可升高,特异性不足;CRP的敏感性和特异性均较差,难以准确区分脓毒症与其他炎症疾病;CD64虽敏感性和特异性均较高,但检测成本高、需特殊设备,在临床难以广泛应用[5, 6]。因此,寻找一种新生物标志物,既具有较好的敏感性和特异性,又易于临床推广,将有助于早期预测脓毒症相关死亡。
近年来,红细胞分布宽度 (RDW)的升高被证实是脓毒症死亡的独立危险因素[7-9]。RDW反映红细胞体积异质性[10],其升高提示红细胞稳态失调,可能与端粒缩短、氧化应激、炎症、以及促红细胞生成素异常导致的生成受损或代谢与存活异常有关。同时,红细胞压积 (HCT) 作为浓缩红细胞相对于全血红细胞体积的指标,既往研究显示低HCT是脓毒症30 d内死亡率增加的独立危险因素[11]。然而,RDW或HCT水平易受营养状况、液体复苏以及红细胞输注等因素的干扰[12-14]。
既往研究发现复合指标在预测脓毒症预后较单一指标稳定性更好。如一项研究发现乳酸/血细胞比值可预测脓毒症死亡[15],但未对混杂因素进行校正。也有研究发现中性粒细胞/淋巴细胞比值以及淋巴细胞/单核细胞比值可用于预测脓毒症预后,但研究样本量小,仅47例[16]。相比之下,RDW/HCT可能更能综合反映生理状况,降低单项指标的个体变异干扰。原因在于:脓毒症相关的全身炎症反应和氧化应激可抑制骨髓造血功能、缩短红细胞寿命,促进未成熟红细胞进入外周循环,导致RDW升高[8];另一方面血液稀释、贫血以及红细胞破坏增加可致HCT下降,降低机体携氧能力,加重组织缺氧和多器官功能损伤,影响患者预后[11]。基于此,RDW/HCT 比值的变化可综合反映脓毒症患者红细胞生成、炎症反应和氧化应激等病理生理过程。然而,目前仅有慢性阻塞性肺部疾病急性加重期的队列研究报道RDW/HCT升高可预测28 d全因死亡[17],该比值在预测脓毒症预后中的作用尚不明确。
据此,本研究拟基于美国重症监护医学信息数据IV 3.1(MIMIC-IV)数据库,采用回顾性队列研究的方法,利用Kaplan-Meier(K-M)生存曲线、多变量Cox回归和限制性立方样条回归(RCS)分析RDW/HCT比值与脓毒症患者28 d全因死亡率的相关性,并通过亚组分析评估结果的稳健性。同时,基于电子重症监护病房协作研究数据库2.0(eICU-CRD)进行外部验证,以期为脓毒症早期死亡识别提供简单有效的评估指标。
1. 资料和方法
1.1. 数据来源
本研究采用回顾性队列研究方法,从MIMIC-IV 3.1纳入27 335例成年脓毒症患者。验证队列为电子重症监护病房协作研究数据库eICU-CRD,该数据库收集了2014~2015年来自美国208家医院的患者信息,且与MIMIC-IV由不同医院组成,彼此相互独立。本研究从eICU-CRD 2.0共纳入17 406例成年脓毒症患者。两数据库均已对患者信息进行去隐私化处理,符合医学伦理学标准。研究已获得MIMIC-IV 3.1及eICU-CRD 2.0数据库使用授权 (证书编号:63806772)。纳入标准 :符合脓毒症3.0诊断标准且首次入重症监护病房(ICU)的患者。排除标准:年龄<18岁;入住ICU(<24 h);RDW或HCT缺失。
1.2. 数据收集
使用PostgresSQL(v16.3.2)和Navicate Premium(v16)提取以下数据:一般资料:年龄、性别、种族;合并症:心梗、充血性心力衰竭、脑血管疾病、恶性肿瘤、糖尿病、慢性肾脏疾病、急性肾损伤;入住ICU 24 h之内的实验室化验结果:RDW、HCT、血小板(PLT)、红细胞(RBC)、白细胞(WBC)、凝血酶原时间(PT)、部分促凝血酶原时间(PTT)、国际标准化比值(INR)、血清肌酐(Scr)、血尿素氮(BUN)、血糖以及电解质(钾、钠);生命体征:心率、平均动脉压、呼吸频率;临床治疗措施:血管活性药物、激素类药物、连续性肾脏替代治疗(CRRT),机械通气;入住ICU 24 h之内的评估:查尔森合并症指数(CCI)、序贯器官衰竭评分(SOFA)、格拉斯哥昏迷评分法(GCS)。
1.3. 结局指标
28 d全因死亡率:定义为从入ICU当日起28 d内的任何原因死亡。
1.4. 统计学分析
本研究采用R 4.3.3 软件行统计分析。连续变量的正态性采用 Shapiro-Wilk检验进行评估,并结合直方图及Q–Q图对判断数据分布,符合正态分布的连续性变量用均数±标准差表示,不符合正态分布的连续性变量用中位数(四分位数间距)表示。分类变量以频率(百分比)表示。基线资料组间比较采用单因素方差分析、K-W检验和卡方检验。对缺失比例小于20%的变量采用多重插补法,通过多重插补链式方程(MICE)生成5个插补数据集,经一致性评价后,选取其中一套插补数据进行后续分析。本研究遵循加强流行病学中观察性研究报告质量声明(STROBE)声明, 构建多因素Cox比例风险模型分析RDW/HCT比值对预后的独立影响。模型1:未校正任何混杂因素;模型2: 校正年龄、性别、种族;MIMIC-IV的模型3:在模型2基础上加入CCI、SOFA、GCS、RBC、WBC、PT、PTT、INR、Scr、BUN、血糖、血钠、血钾、既往病史(心梗、脑血管疾病、充血性心力衰竭、恶性肿瘤、慢性肾脏疾病、急性肾损伤)、心率、平均动脉压、呼吸频率、血管活性药物、激素类药物、CRRT;eICU-CRD的模型3校正因素同MIMIC-IV的模型3的校正因素,除外糖尿病和CRRT。
通过RCS分析,设置4个节点(第5、35、65和95百分位),构建样条函数,拟合RDW/HCT与死亡率的关系。若存在非线性关联,通过递归法及最大似然法确定拐点,并利用似然比检验对比分段Cox回归和普通Cox回归模型的拟合优度,检验是否存在阈值或饱和效应,P<0.05说明存在显著效应,两者关系为非线性关系。
利用亚组分析探讨不同亚组中连续RDW/HCT与28 d死亡率之间的关系。亚组分析采用分层Cox回归,校正混杂因素同模型3,使用似然比检验进行交互作用评价,并绘制亚组森林图。所有检验均为双侧检验,P<0.05被认为差异具有统计学意义。
2. 结果
2.1. 纳入人群总体特征
本研究纳入MIMIC-IV 3.1中的27 335例脓毒症患者,中位年龄为68岁,其中58.1%为男性;28 d内死亡率为19.3% (n=5276)。RDW/HCT中位数为0.46 (IQR: 0.38~0.58)。根据RDW/HCT比值的四分位数将患者分为4组:Q1(RDW/HCT≤0.384;n=6834)、Q2(0.384<RDW/HCT≤0.463;n=6834)、Q3(0.463<RDW/HCT≤0.576;n=6833)和Q4(RDW/HCT>0.576;n=6834)。4组患者28 d内的死亡率分别是16.8%、16.6%、18.0%、25.9%。
脓毒症患者基线特征分析显示(表1)。RDW/HCT比值较高组死亡率高于较低组,差异有统计学意义(P<0.001)。与RDW/HCT比值较低组相比,比值较高组患者危重程度更重(SOFA评分,GCS评分),合并心衰、恶性肿瘤、糖尿病、肾病的比例较高,PT、PTT、INR、Scr、BUN较高,但HCT、RBC、WBC、血糖水平较低。
表1.
RDW/HCT4组脓毒症患者入ICU 24 h内基线资料比较
Tab.1 Comparison of baseline characteristics of septic patients in 4 RDW/HCT quadrants within 24 h of ICU admission
| MIMIC-IV cohort |
Overall (n=27 335) |
Q1 (≤0.384; n=6834) |
Q2 (>0.384, ≤0.463; n=6834) |
Q3 (>0.463, ≤0.576; n=6833) |
Q4 (>0.576; n=6834) |
Statistics | P |
|---|---|---|---|---|---|---|---|
| Age [year, Median (Q1, Q3)] | 68 (57, 79) | 65(53, 76) | 69 (58, 80) | 70 (60, 80) | 68 (57, 79) | χ²=364.77 | <0.001 |
| Male [n (%)] | 15 891 (58.1) | 4646 (68.00) | 3933 (57.6) | 3710 (54.3) | 3602 (52.7) | χ²=397.43 | <0.001 |
| Race [n (%)] | |||||||
| White | 17 952 (65.7) | 4228 (61.9) | 4569 (66.9) | 4668 (68.3) | 4487 (65.7) | χ²=69.33 | <0.001 |
| Other | 9383 (34.3) | 2606 (38.1) | 2265 (33.1) | 2165 (31.7) | 2347 (34.3) | ||
| Score system | |||||||
| CCI [Median (Q1, Q3)] | 5 (3, 7) | 4 (2, 6) | 5 (3, 7) | 5.0 (4, 8) | 6 (4, 8) | χ²= 1 715.28 | <0.001 |
| SOFA [Median (Q1, Q3)] | 5(3, 8) | 4 (3, 7) | 5 (3, 7) | 5.0 (4, 8) | 6 (4, 9) | χ²=943.45 | <0.001 |
| GCS [Median (Q1, Q3)] | 15 (13, 15) | 15 (13, 15) | 15 (13, 15) | 15 (14, 15) | 15 (13, 15) | χ²=15.14 | 0.002 |
| Laboratory tests | |||||||
| RDW [%, Median (Q1, Q3)] | 14.6 (13.5, 16.2) | 13.4 (12.8, 14.1) | 14.1 (13.4, 15.0) | 15.0 (14.0, 16.2) | 17.2 (15.6, 19.1) | χ²=11 563.01 | <0.001 |
| HCT [%, Median (Q1, Q3)] | 31.8 (27.2, 36.7) | 39.7 (37.1, 42.9) | 33.6 (31.7, 35.8) | 29.3 (27.2, 31.6) | 24.7 (22.4, 27.2) | χ²=20 177.78 | <0.001 |
| PLT [K/μL, Median (Q1, Q3)] | 183 (128, 252) | 204 (156, 260) | 186 (136, 251) | 178 (125, 255) | 152 (93, 239) | χ²=897.56 | <0.001 |
| RBC [K/μL, Median (Q1, Q3)] | 3.5 (3.0, 4.1) | 4.3 (4.0, 4.7) | 3.7 (3.4, 4.0) | 3.2 (2.9, 3.5) | 2.7 (2.4, 3.1) | χ²=16 582.42 | <0.001 |
| WBC [K/μL, Median (Q1, Q3)] | 11.5 (8.0, 16.0) | 12.3 (8.9, 16.6) | 11.7 (8.4, 16.0) | 11.4 (7.9, 15.8) | 10.4 (6.7, 15.7) | χ²=288.16 | <0.001 |
| PT [s, Median (Q1, Q3)] | 14.6 (12.8, 17.3) | 13.3 (12.1, 15.2) | 14.2 (12.7, 16.5) | 15.0 (13.3, 17.5) | 16.1 (13.9, 20.1) | χ²=2 526.17 | <0.001 |
| PTT [s, Median (Q1, Q3)] | 31.1 (27.3, 38.3) | 29.9 (26.6, 35.9) | 30.6 (27.0, 36.4) | 31.4 (27.5, 38.3) | 33.2 (28.4, 42.1) | χ²=513.23 | <0.001 |
| INR [Median (Q1, Q3)] | 1.3 (1.2, 1.6) | 1.2 (1.1, 1.4) | 1.3 (1.1, 1.5) | 1.4 (1.2, 1.6) | 1.5 (1.3, 1.8) | χ²=2 371.37 | <0.001 |
| Scr [mg/dL, Median (Q1, Q3)] | 1.0 (0.8, 1.6) | 1.0 (0.8, 1.3) | 1.0 (0.7, 1.4) | 1.0 (0.7, 1.7) | 1.2 (0.8, 2.2) | χ²=479.71 | <0.001 |
| BUN [mg/dL, Median(Q1, Q3)] | 21 (14, 35) | 18 (13, 26) | 19 (13, 31) | 22 (14, 38) | 27 (16, 47) | χ²=1 129.84 | <0.001 |
| Glucose [mg/dL, Median (Q1, Q3)] | 129 (106, 166) | 135 (111, 177) | 130 (107, 166) | 126 (104, 162) | 124 (101, 160) | χ²=295.51 | <0.001 |
| Potassium [mmol/L, Median (Q1, Q3)] | 4.2 (3.8, 4.6) | 4.1 (3.8, 4.6) | 4.1 (3.7, 4.6) | 4.2 (3.8, 4.6) | 4.2 (3.7, 4.7) | χ²=18.36 | <0.001 |
| Sodium [mmol/L, Median (Q1, Q3)] | 139 (136, 141) | 139 (136, 141) | 139 (136, 141) | 139 (136, 141) | 138 (134, 141) | χ²=184.94 | <0.001 |
| Commorbidities | |||||||
| Myocardial infarct [n (%)] | 4876 (17.8) | 1221 (17.9) | 1218 (17.8) | 1302 (19.0) | 1135 (16.6) | χ²=13.96 | 0.003 |
| Heart failure [n (%)] | 8249 (30.2) | 1615 (23.6) | 2000 (29.3) | 2298 (33.6) | 2336 (34.2) | χ²=232.35 | <0.001 |
| Cerebrovascular disease [n (%)] | 4201 (15.4) | 1447 (21.2) | 1070 (15.7) | 865 (12.7) | 819 (12.0) | χ²=276.24 | <0.001 |
| Cancer [n (%)] | 3753 (13.7) | 588 (8.6) | 693 (10.1) | 979 (14.3) | 1493 (21.9) | χ²=608.11 | <0.001 |
| Diabetes [n (%)] | 8499 (31.1) | 1653 (24.2) | 2099 (30.7) | 2436 (35.7) | 2311 (33.8) | χ²=242.45 | <0.001 |
| Renal disease [n (%)] | 6170 (22.6) | 834 (12.2) | 1292 (18.9) | 1843 (27.0) | 2201 (32.2) | χ²=911.60 | <0.001 |
| AKI [n (%)] | 22 195 (81.2) | 5596 (81.9) | 5430 (79.5) | 5499 (80.5) | 5670 (83.0) | χ²=32.04 | <0.001 |
| Vital signs | |||||||
| Heart rate [min-1, Median (Q1, Q3)] | 88 (77, 103) | 88 (76, 103) | 87(76, 102) | 86 (76, 101) | 90 (79, 105) | χ²=98.09 | <0.001 |
| MAP [mmHg, Median (Q1, Q3)] | 81 (70, 93) | 86 (75, 99) | 81 (71, 93) | 79 (69, 90) | 77 (67, 89) | χ²=1040.97 | <0.001 |
| RR [min-1, Median (Q1, Q3)] | 19 (15, 23) | 19 (16, 24) | 18(15, 23) | 18 (15, 22) | 19 (16, 24) | χ²=142.05 | <0.001 |
| Interventions | |||||||
| Vasopressor [n (%)] | 13 623 (49.8) | 3075 (45.0) | 3432 (50.2) | 3640 (53.3) | 3476 (50.9) | χ²=99.58 | <0.001 |
| Corticosteroids [n (%)] | 6207 (22.7) | 1721 (25.2) | 1475 (21.6) | 1322 (19.4) | 1689 (24.7) | χ²=88.43 | <0.001 |
| CRRT [n (%)] | 1898 (6.9) | 318 (4.7) | 361 (5.3) | 412 (6.0) | 807 (11.8) | χ²=343.84 | <0.001 |
| Ventilator [n (%)] | 24 071 (88.1) | 6077 (88.9) | 6068 (88.8) | 6070 (88.8) | 5856 (85.7) | χ²=343.84 | <0.001 |
| 28-day mortality [n (%)] | 5276 (19.3) | 1145 (16.8) | 1134 (16.6) | 1228 (18.0) | 1769 (25.9) | χ²=258.58 | <0.001 |
CCI:Charlson comorbidity index; SOFA: Sequential organ failure assessment; GCS: Glasgow coma scale; RDW: Red blood cell distribution width;HCT: Hematocrit; PLT: Platelet; RBC: Red blood cell; WBC: White blood cell;PT:Prothrombin time;PTT:Partial thromboplastin time;INR:International normalized ratio;Scr:Serum creatinine;BUN:Blood urea nitrogen;AKI:Acute kidney injury;MAP:Mean arterial pressure;CRRT:Continuous renal replacement therapy.
2.2. K-M生存曲线分析
K-M生存曲线(图1)显示,4组脓毒症患者的28 d生存率有统计学差异(P<0.001)。与RDW/HCT比值较低组相比(Q1、Q2),比值较高组(Q3、Q4)患者的生存率显著降低,其中Q4组生存率最低.
图1.

RDW/HCT4组与脓毒症患者28 d死亡率的K-M生存分析曲线
Fig.1 Kaplan-Meier survival curves for 28-day mortality in patients with sepsis in the 4 RDW/HCT groups.
2.3. 多因素Cox回归分析
根据28 d预后将患者分为死亡组与未死亡组。与未死亡组患者相比,死亡组患者入ICU时SOFA评分更高,RDW/HCT更高(表2)。两组在年龄、性别、种族、疾病危重程度(SOFA评分,GCS评分)、RDW、RBC、WBC、PT、PTT、INR、Scr、BUN、血糖、钾、钠、合并心梗、脑血管疾病、充血性心力衰竭、恶性肿瘤、肾脏疾病等疾病的比例,以及心率、呼吸频率、血压、使用血管活性药物、激素类药物、CRRT的比例均有显著差异(P<0.05,表2)。
表2.
28 d未死亡组和死亡组的脓毒症患者入ICU 24 h内基线资料比较
Tab.2 Comparison of baseline characteristics between 28-day survivors and non-survivors with sepsis within 24 h of ICU admission
| MIMIC-IV Cohort | Survivor (n=22 059) | Non-survivor (n=5276) | Statistics | P |
|---|---|---|---|---|
| Age [year, Median (Q1, Q3)] | 67 (56, 77) | 73.00 (62, 83) | Z=-24.67 | <0.001 |
| Male [n (%)] | 12 943 (58.7) | 2948 (55.9) | χ²=13.70 | <0.001 |
| Race [(n%)] | ||||
| White | 14 781 (67.0) | 3171 (60.1) | χ²=90.03 | <0.001 |
| Other | 7278 (33.0) | 2105 (39.9) | ||
| Score system | ||||
| CCI [Median (Q1, Q3)] | 5 (3, 7) | 7 (5, 9) | Z=-Inf | <0.001 |
| SOFA [Median (Q1, Q3)] | 5 (3 7) | 7 (4, 10) | Z=-32.18 | <0.001 |
| GCS [Median (Q1, Q3)] | 15 (14, 15) | 15(12, 15) | Z=-7.35 | <0.001 |
| Laboratory tests | ||||
| RDW [%, Median (Q1, Q3)] | 14.4 (13.4, 15.9) | 15.5 (14.1, 17.6) | Z=-32.37 | <0.001 |
| PLT [K/μL, Median (Q1, Q3)] | 183 (129, 250) | 184 (118, 262) | Z=-1.57 | 0.117 |
| HCT [%, Median (Q1, Q3)] | 31.8 (27.3, 36.6) | 31.9 (27.0, 37.0) | Z=-1.03 | 0.305 |
| RDW/HCT [Median (Q1, Q3)] | 0.46 (0.38, 0.56) | 0.49 (0.40, 0.63) | Z=-14.40 | <0.001 |
| RBC [K/μL, Median (Q1, Q3)] | 3.5 (3.0, 4.1) | 3.4 (2.9, 4.1) | Z=-4.74 | <0.001 |
| WBC [K/μL, Median (Q1, Q3)] | 11.3 (7.9, 15.6) | 12.5 (8.5, 18.0) | Z=-11.84 | <0.001 |
| PT [s, Median (Q1, Q3)] | 14.4 (12.7, 16.8) | 15.4 (13.2, 20.3) | Z=-19.55 | <0.001 |
| PTT [s, Median (Q1, Q3)] | 30.8 (27.2, 37.2) | 33.1 (28.1, 43.9) | Z=-16.45 | <0.001 |
| INR [Median (Q1, Q3)] | 1.3 (1.2, 1.5) | 1.4 (1.2, 1.9) | Z=-20.58 | <0.001 |
| Scr [mg/dL, Median (Q1, Q3)] | 1.0 (0.7, 1.5) | 1.3 (0.9, 2.2) | Z=-25.52 | <0.001 |
| BUN [mg/dL, Median (Q1, Q3)] | 19 (13, 31) | 30 (18, 49) | Z=-35.21 | <0.001 |
| Glucose [mg/dL, Median (Q1, Q3)] | 127 (106, 163) | 137 (107, 184) | Z=-9.92 | <0.001 |
| Potassium [mmol/L, Median (Q1, Q3)] | 4.1 (3.7, 4.6) | 4.2 (3.8, 4.8) | Z=-9.00 | <0.001 |
| Sodium [mmol/L, Median (Q1, Q3)] | 139 (136, 141) | 138 (135, 142) | Z=-3.80 | <0.001 |
| Commorbidities | ||||
| Myocardial infarct [n (%)] | 3744 (17.0) | 1132 (21.5) | χ²=58.38 | <0.001 |
| Cerebrovascular disease [n (%)] | 3104 (14.1) | 1097 (20.8) | χ²=147.86 | <0.001 |
| Heart failure [n (%)] | 6306 (28.6) | 1943 (36.8) | χ²=137.20 | <0.001 |
| Cancer [n (%)] | 2626 (11.9) | 1127 (21.4) | χ²=321.44 | <0.001 |
| Diabetes [n (%)] | 6851 (31.1) | 1648 (31.2) | χ²=0.06 | 0.802 |
| Renal disease [n (%)] | 4682 (21.2) | 1488 (28.2) | χ²=118.63 | <0.001 |
| AKI [n (%)] | 17 309 (78.5) | 4886 (92.6) | χ²=557.65 | <0.001 |
| Vital Signs | ||||
| Heart rate [min-1, Median (Q1, Q3)] | 87 (76, 102) | 93 (79, 108) | Z=-14.77 | <0.001 |
| MAP [mmHg, Median (Q1, Q3)] | 81 (71, 93) | 80 (69, 93) | Z=-4.22 | <0.001 |
| RR [min-1, Median (Q1, Q3)] | 18 (15, 22) | 21 (17, 25) | Z=-27.32 | <0.001 |
| Interventions | ||||
| Vasopressor [n (%)] | 10 514 (47.7) | 3109 (58.9) | χ²=216.09 | <0.001 |
| Corticosteroids [n (%)] | 4637 (21.0) | 1570 (29.8) | χ²=185.16 | <0.001 |
| CRRT [n (%)] | 1047 (4.8) | 851 (16.1) | χ²=853.85 | <0.001 |
| Ventilator [n (%)] | 19 412 (88.0) | 4659 (88.3) | χ²=0.38 | 0.539 |
Cox比例风险回归分析显示(表3),RDW/HCT作为连续变量与28 d死亡风险呈正相关。模型1中,HR=3.29,95% CI:2.87~3.78,P<0.001;模型2中,HR=3.39,95% CI:2.94-3.91,P<0.001;模型3中,HR=1.89,95% CI:1.52~2.36,P<0.001。进一步以Q2组作为参照组。结果显示,Q4组在所有模型中均显示与28 d死亡风险显著正相关(P<0.001)。其中,模型3显示,与Q2组相比,Q4组死亡风险增加31%(HR=1.31,95% CI:1.19~1.44,P<0.001),但Q1与Q3组与Q2组相比,死亡风险无统计学意义。趋势性检验提示RDW/HCT与28 d死亡率之间可能存在非线性关系(P<0.001)。
表3.
RDW/HCT与脓毒症患者28 d死亡风险关系的多变量Cox回归分析
Tab.3 Multivariable Cox regression analysis of the association between RDW/HCT ratio and 28-day mortality risk in patients with sepsis
| 28-day mortality | HR (95% Cl) P value | ||
|---|---|---|---|
| Model1a | Model2b | Model3c | |
| MIMIC-IV Cohort | |||
| Continuous variables | 3.29 (2.87, 3.78)<0.001 | 3.39 (2.94, 3.91)<0.001 | 1.89 (1.52, 2.36)<0.001 |
| Q2 | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) |
| Q1 | 1.02 (0.94, 1.10) 0.687 | 1.09 (1.01, 1.19) 0.031 | 1.03 (0.94, 1.13) 0.525 |
| Q3 | 1.09 (1.01, 1.18) 0.043 | 1.07 (0.99, 1.16) 0.110 | 1.07 (0.98, 1.16) 0.119 |
| Q4 | 1.63 (1.51, 1.75)<0.001 | 1.63 (1.51, 1.76)<0.001 | 1.31 (1.19, 1.44)<0.001 |
| P for trend | <0.001 | <0.001 | <0.001 |
| eICU-CRD Cohort | |||
| Continuous variables | 3.10 (2.45-3.94) <0.001 | 3.57 (2.79-4.55) <0.001 | 0.93 (0.67-1.30) 0.681 |
| Q2 | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) |
| Q1 | 0.96 (0.85-1.09) 0.520 | 1.02 (0.90-1.15) 0.801 | 1.23 (1.08-1.40) 0.002 |
| Q3 | 1.09 (0.98 -1.21) 0.123 | 1.08 (0.97-1.20) 0.188 | 0.88 (0.79-0.99) 0.030 |
| Q4 | 1.34 (1.21-1.48)<0.001 | 1.40 (1.26-1.55)<0.001 | 0.97 (0.87-1.09) 0.660 |
| P for trend | <0.001 | <0.001 | 0.015 |
a: Unadjusted; b: Adjusted for age,sex,and race; c: Fully adjusted for age, sex, race, score system, RBC, WBC, PT, PTT, INR, Scr,BUN, Glucose, potassium, sodium, myocardial infarct, cerebrovascular disease, heart failure, malignantcancer, renal disease, AKI,vital signs, vasopressor, corticosteroids, CRRT. In the eICU-CRD cohort, CRRT and diabetes were not adjusted for; other covariates were the same as in MIMIC-IV.
2.4. RCS分析和饱和阈值效应分析
进一步行RCS分析,显示RDW/HCT比值和患者28 d死亡风险呈非线性“U型”关系(P<0.001,图2)。阈值效应分析显示,28 d死亡风险的拐点为0.35(表4)。RDW/HCT<0.35,比值与28 d死亡率呈显著负相关(HR=0.022,95% CI:0.003~0.162,P<0.001);RDW/HCT≥0.35,28 d死亡风险随比值升高而显著增高(HR=2.034,95% CI:1.628~2.541,P<0.001)。进一步比较标准线性模型与分段线性模型的拟合优度,对数似然比检验证实分段线性模型显著优于标准线性模型(对数似然比P<0.001),提示阈值效应的存在。
图2.

RDW/HCT与脓毒症患者28 d死亡风险的RCS分析
Fig.2 Restricted cubic spline (RCS) analysis of the association between RDW/HCT ratio and 28-day mortality risk in patients with sepsis.
表4.
RDW/HCT与脓毒症患者28 d死亡率的饱和阈值效应分析
Tab.4 Threshold effect analysis of the association between RDW/HCT ratio and 28-day mortality in patients with sepsis
| 28-day mortality | MIMIC-IV Cohort | P | eICU-CRD Cohort | P |
|---|---|---|---|---|
| HR (95% CI) | HR (95% CI) | |||
| Standard linear model | 2.04 (1.64-2.55) | <0.001 | 1.18(0.85-1.63) | 0.32 |
| Two-piecewise linear model | ||||
| Inflection point | 0.35 | 0.48 | ||
| RDW/HCT <Inflection point | 0.022 (0.003-0.162) | <0.001 | 0.031(0.011-0.089) | <0.001 |
| RDW/HCT≥Inflection point | 2.034 (1.628-2.541) | <0.001 | 1.969(1.402-2.766) | <0.001 |
| P for Log-likelihood ratio | <0.001 | <0.001 |
2.5. RDW/HCT的亚组分析结果
进一步分析不同亚组之间RDW/HCT与脓毒症患者28 d死亡率之间的关系。根据性别、年龄、心肌梗死、充血性心力衰竭、慢性肾脏疾病、恶性肿瘤、是否使用激素类药物进行亚组分析,结果显示:RDW/HCT在各亚组之间均无显著交互作用(P>0.05),RDW/HCT与脓毒症患者28 d死亡率之间的关系具有稳健性。
图3.

RDW/HCT与脓毒症患者28 d死亡风险亚组分析森林图
Fig.3 Forest plot of subgroup analysis for the association between RDW/HCT ratio and 28-day mortality in patients with sepsis.
2.6. eICU-CRD外部验证
2.6.1. eICU-CRD人群总体特征
本研究在eICU-CRD数据库纳入17 406例脓毒症患者,中位年龄67岁,48.1%为男性;28 d内死亡率为16.0%(2785例)。RDW/HCT的中位数为0.48 (IQR:0.40~0.58)。根据RDW/HCT比值4分位数分组: Q1(RDW/HCT≤0.400;n=3813)、Q2(0.400<RDW/HCT≤0.480;n=4551)、Q3(0.480<RDW/HCT≤0.580;n=4499)和Q4 (RDW/HCT>0.580;n=4543)。四组患者28 d内的死亡率分别是11.1%、13.2%、16.4%和22.5%。表5列出了4组脓毒症患者的基线特征。RDW/HCT比值较高组死亡率显著高于比值较低组(P<0.001)。与RDW/HCT比值较低组相比,比值较高组的患者SOFA评分更高,合并恶性肿瘤及慢性肾脏疾病的比例较高,RDW、WBC、PT、PTT、INR、Scr、BUN、血钾浓度较高,但HCT、RBC、血糖水平较低。
表5.
RDW/HCT4组脓毒症患者入ICU 24 h内基线资料比较
Tab.5 Comparison of baseline characteristics of septic patients in the 4 RDW/HCT quadrants with 24 h of ICU admission
| eICU-CRD cohort |
Overall (n=17 406) |
Q1 (≤0.400; n=3813) |
Q2 (>0.400, ≤0.480; n=4551) |
Q3 (>0.480 ≤0.580; n=4499) |
Q4 (>0.580; n=4543) |
Statistics | P |
|---|---|---|---|---|---|---|---|
| Age [year, Median (Q1, Q3)] | 67 (55, 78) | 64 (52,76) | 68 (56, 80) | 69 (58, 79) | 66 (55, 76) | χ²=144.83 | <0.001 |
| Male [n (%)] | 8380 (48.1) | 1525 (40.0) | 2267 (49.8) | 2356 (52.4) | 2232 (49.1) | χ²=140.42 | <0.001 |
| Race [n (%)] | χ²=103.42 | <0.001 | |||||
| White | 13 481 (77.5) | 3105 (81.4) | 3622 (79.6) | 3444 (76.6) | 3310 (72.9) | ||
| Other | 3925 (22.6) | 708 (18.6) | 929 (20.4) | 1055 (23.5) | 1233 (27.1) | ||
| Score system | |||||||
| SOFA [Median (Q1, Q3)] | 5 (3, 8) | 4 (2, 7) | 4 (2, 7) | 5 (3, 8) | 6 (3, 9) | χ²=526.51 | <0.001 |
| GCS [Median (Q1, Q3)] | 15 (12, 15) | 15 (12, 15) | 15 (12, 15) | 15 (12, 15) | 15 (11, 15) | χ²=39.98 | <0.001 |
| Laboratory tests | |||||||
| RDW [%, Median (Q1, Q3)] | 16.0 (14.6, 18.0) | 14.2 (13.5, 15.0) | 15.2 (14.4, 16.2) | 16.6 (15.5, 17.8) | 19.3 (17.6, 21.6) | χ²=9 526.80 | <0.001 |
| HCT [%, Median (Q1, Q3)] | 33.6 (30.1, 37.6) | 40.3 (37.8, 43.2) | 35.1(33.1, 37.5) | 31.9 (29.8, 34.1) | 28.8 (26.7, 31.2) | χ²=10 381.12 | <0.001 |
| RBC [K/μL, Median (Q1, Q3)] | 3.7 (3.3, 4.2) | 4.4 (4.1, 4.7) | 3.9 (3.6, 4.2) | 3.5 (3.3, 3.9) | 3.22 (2.9, 3.6) | χ²=7 222.93 | <0.001 |
| WBC [K/μL, Median (Q1, Q3)] | 15.5 (10.8, 21.9) | 15.4 (11.2, 20.7) | 15.5 (11.1, 21.5) | 15.6 (10.6, 22.4) | 15.8 (10.2, 23.1) | χ²=1.44 | 0.696 |
| PT [s, Median (Q1, Q3)] | 16.9 (14.2, 23.9) | 15.6 (13.5, 20.7) | 16.3 (13.9, 22.3) | 17.2 (14.6, 24.7) | 18.3 (15.1, 26.9) | χ²=434.96 | <0.001 |
| PTT [s, Median (Q1, Q3)] | 37.7 (31.5, 53.0) | 34.9 (30.0, 47.1) | 36.3 (31.0, 50.0) | 38.8 (32.0, 54.8) | 40.6 (33.3, 57.2) | χ²=340.80 | <0.001 |
| INR [Median (Q1, Q3)] | 1.4 (1.2, 2.2) | 1.3 (1.1, 1.8) | 1.4 (1.2, 2.0) | 1.5(1.2, 2.3) | 1.6 (1.3, 2.5) | χ²=559.61 | <0.001 |
| Scr [mg/dL, Median (Q1, Q3)] | 1.4 (0.9, 2.6) | 1.1 (0.8, 1.8) | 1.3 (0.9, 2.3) | 1.6 (0.9, 2.9) | 1.8 (1.0, 3.3) | χ²=599.78 | <0.001 |
| BUN [mg/dL, Median (Q1, Q3)] | 33 (20, 55) | 25 (16, 41) | 30 (19, 50) | 37 (22, 59) | 41 (24, 65) | χ²=749.79 | <0.001 |
| Glucose [mg/dL, Median (Q1, Q3)] | 204 (150, 280) | 203 (150, 285) | 203 (148, 279) | 205 (150, 281) | 203 (152, 276) | χ²=2.50 | <0.001 |
| Potassium [mmol/L, Median (Q1, Q3)] | 4.5 (4.1, 5.0) | 4.4 (4.1, 4.8) | 4.4 (4.1, 4.9) | 4.5 (4.1, 5.0) | 4.6 (4.2, 5.1) | χ²=191.13 | <0.001 |
| Sodium [mmol/L, Median (Q1, Q3)] | 142(139, 146) | 142 (139, 145) | 142 (139, 146) | 142 (139, 146) | 142 (139, 146) | χ²=16.86 | <0.001 |
| Commorbidities | |||||||
| Myocardial infarct [n (%)] | 1546 (8.9) | 341 (8.9) | 398 (8.8) | 447 (9.9) | 360 (7.9) | χ²=11.44 | 0.010 |
| Heart failure [n (%)] | 1787 (10.3) | 360 (9.4) | 469 (10.3) | 482 (10.7) | 476 (10.5) | χ²=4.02 | 0.259 |
| Cerebrovascular disease [n (%)] | 554 (3.2) | 141 (3.7) | 146 (3.2) | 123 (2.7) | 144 (3.2) | χ²=6.24 | 0.101 |
| Malignant cancer [n (%)] | 5203 (29.9) | 859 (22.5) | 1156 (25.4) | 1417 (31.5) | 1771 (39.0) | χ²=327.15 | <0.001 |
| Renal disease [n (%)] | 1665 (9.6) | 218 (5.7) | 406 (8.9) | 506 (11.3) | 535 (11.8) | χ²=107.83 | <0.001 |
| AKI [n (%)] | 14 061 (80.8) | 2935 (77.0) | 3649 (80.2) | 3736 (83.0) | 3741 (82.4) | χ²=58.63 | <0.001 |
| Vital Signs | |||||||
| Heart rate [min-1, Median (Q1, Q3)] | 97 (82, 111) | 98 (83, 112) | 96 (82, 111) | 96 (81, 111) | 97 (82, 112) | χ²=17.16# | <0.001 |
| MAP [mmHg, Median (Q1, Q3)] | 75 (64, 88) | 80 (68, 95) | 75 (64, 89) | 73 (63, 86) | 71 (61, 84) | χ²=477.10# | <0.001 |
| RR [min-1, Median (Q1, Q3)] | 21 (18, 26) | 22 (18, 26) | 21 (18, 26) | 21 (17, 26) | 21 (17, 26) | χ²=17.80# | <0.001 |
| Interventions | |||||||
| Vasopressor [n (%)] | 4371 (25.1) | 736 (19.3) | 1051 (23.1) | 1163 (25.9) | 1421 (31.3) | χ²=171.47 | <0.001 |
| Corticosteroids[n (%)] | 1485 (8.5) | 377 (9.9) | 384 (8.4) | 349 (7.8) | 375 (8.3) | χ²=12.93 | 0.005 |
| 28-day mortality[n (%)] | 2785 (16.0) | 422 (11.1) | 602 (13.2) | 738 (16.4) | 1023 (22.5) | χ²=239.20 | <0.001 |
2.6.2. Cox分析
进一步根据患者是否死亡分为两组。与未死亡组患者相比,死亡组入住ICU时SOFA评分及RDW/HCT比值更高(表6)。两组年龄、种族、危重程度(SOFA评分,GCS评分)、RDW、HCT、RBC、WBC、PT、PTT、INR、Scr、BUN、钾、钠,合并心梗、脑血管疾病、心力衰竭、恶性肿瘤、肾脏疾病等疾病的比例,以及心率、呼吸频率、血压、血管活性药物、糖皮质激素的比例均有显著差异(均P<0.05)。
表6.
28 d未死亡组和死亡组的脓毒症患者入ICU 24 h内基线资料比较
Tab.6 Comparison of baseline characteristics between 28-day survivors and non-survivors with sepsis within 24 h ICU admission
| eICU-CRD cohort | Survivor (n=14 621) | Non-survivor (n=2785) | Statistics | P |
|---|---|---|---|---|
| Age [year, Median (Q1, Q3)] | 66 (54, 77) | 72 (61, 81) | Z=-16.12 | <0.001 |
| Male [n (%)] | 7057 (48.3) | 1323 (47.5) | χ²=0.54 | 0.461 |
| Race [n (%)] | χ²=5.18 | 0.023 | ||
| White | 11 278 (77.1) | 2203 (79.1) | ||
| Other | 3343 (22.9) | 582 (20.9) | ||
| Score system | ||||
| SOFA [Median (Q1, Q3)] | 4 (2, 7) | 8 (5, 11) | Z=-Inf | <0.001 |
| GCS [Median (Q1, Q3)] | 15(13, 15) | 12 (8, 15) | Z=-33.23 | <0.001 |
| Laboratory tests | ||||
| RDW [%, Median (Q1, Q3)] | 15.8 (14.5, 17.8) | 17.0 (15.5, 19.3) | Z=-22.11 | <0.001 |
| HCT [%, Median (Q1, Q3)] | 33.7 (30.2, 37.6) | 33.1 (29.2, 38.0) | Z=-4.12 | <0.001 |
| RDW/HCT [Median (Q1, Q3)] | 0.47 (0.40, 0.57) | 0.52 (0.43, 0.63) | Z=-15.90 | <0.001 |
| RBC [K/μL, Median (Q1, Q3)] | 3.8 (3.3, 4.2) | 3.6 (3.2, 4.2) | Z=-8.12 | <0.001 |
| WBC [K/μL, Median (Q1, Q3)] | 15.0 (10.6, 21.0) | 19.0 (12.3, 27.3) | Z=-17.32 | <0.001 |
| PT [s, Median (Q1, Q3)] | 16.4 (14.0, 22.5) | 20.6 (15.8, 31.2) | Z=-22.47 | <0.001 |
| PTT [s, Median (Q1, Q3)] | 37.0 (31.0, 50.0) | 43.1(34.5, 69.4) | Z=-18.58 | <0.001 |
| INR [Median (Q1, Q3)] | 1.4 (1.2, 2.0) | 1.8 (1.4, 3.1) | Z=-24.69 | <0.001 |
| Scr [mg/dL, Median (Q1, Q3)] | 1.3 (0.8, 2.3) | 2.3 (1.3, 3.5) | Z=-27.55 | <0.001 |
| BUN [mg/dL, Median (Q1, Q3)] | 30 (18, 51) | 49 (31, 72) | Z=-29.60 | <0.001 |
| Glucose [mg/dL, Median (Q1, Q3)] | 203 (149, 280) | 208 (153, 279) | Z=-1.71 | 0.087 |
| Potassium [mmol/L, Median (Q1, Q3)] | 4.4 (4.1, 4.9) | 4.8 (4.2, 5.4) | Z=-21.53 | <0.001 |
| Sodium [mmol/L, Median (Q1, Q3)] | 142 (139, 145) | 143 (139, 148) | Z=-9.24 | <0.001 |
| Commorbidities | ||||
| Myocardial infarct [n (%)] | 1169 (8.0) | 377 (13.5) | χ²=88.76 | <0.001 |
| Cerebrovascular disease [n (%)] | 405 (2.8) | 149 (5.4) | χ²=50.54 | <0.001 |
| Heart failure [n (%)] | 1403 (9.6) | 384 (13.8) | χ²=44.63 | <0.001 |
| Malignant cancer [n (%)] | 4092 (28.0) | 1111 (39.9) | χ²=158.22 | <0.001 |
| Renal disease [n (%)] | 1318 (9.0) | 347 (12.5) | χ²=32.10 | <0.001 |
| AKI [n (%)] | 11 650 (79.7) | 2411 (86.6) | χ²=71.56 | <0.001 |
| Vital signs | ||||
| Heart rate [min-1, Median (Q1, Q3)] | 96 (82, 111) | 100 (84, 115) | Z=-6.77 | <0.001 |
| MAP [mmHg, Median (Q1, Q3)] | 75 (64, 89) | 71 (60, 85) | Z=-10.82 | <0.001 |
| RR [min-1, Median (Q1, Q3)] | 21 (17, 26) | 22 (18, 28) | Z=-9.18 | <0.001 |
| Interventions | ||||
| Vasopressor [n (%)] | 3286 (22.5) | 1085 (39.0) | χ²=338.02 | <0.001 |
| Corticosteroids [n (%)] | 1370 (9.4) | 115 (4.1) | χ²=82.34 | <0.001 |
在eICU-CRD队列(表3),模型1显示:RDW/HCT作为连续变量与28 d死亡率呈显著正相关(HR=3.10, 95% CI:2.45~3.94,P<0.001),但在模型2和模型3中,多变量充分校正后,两者无显著相关性。
根据RDW/HCT比值分为4组,模型3发现:与Q2组相比,Q1组28 d死亡风险显著升高 (HR=1.23,95% CI:1.08~1.40,P=0.002),Q3组风险显著降低(HR=0.88, 95% CI:0.79~0.99,P=0.030),Q4组则无显著性,提示RDW/HCT与28 d死亡率之间可能存在非线性关系。
2.6.3. K-M生存曲线分析
K-M生存曲线(图4)显示,RDW/HCT比值与28 d的生存率差异有统计学意义(P<0.001)。低水平组(Q1、Q2)生存率高于高水平组(Q3、Q4)。
图4.

RDW/HCT四组数与脓毒症患者28 d死亡率的K-M生存分析曲线
Fig.4 Kaplan Meier survival curves for 28-day mortality in patients with sepsis in the 4 RDW/HCT groups.
2.6.4. RCS分析和饱和阈值效应分析
RCS分析显示RDW/HCT比值与28 d死亡率存在非线性关系(P<0.001),其趋势与MIMIC-IV队列一致,均呈“U型”关系。饱和阈值效应分析显示死亡风险拐点为0.48(表4),RDW/HCT<0.48时,比值与28 d死亡率呈显著负相关 (HR=0.031, 95% CI:0.011~0.089,P<0.001),RDW/HCT≥0.48,28 d死亡风险随比值升高而显著增加 (HR=1.969,95% CI:1.402~2.766,P<0.001),且分段线性模型显著优于标准线性模型(对数似然比P<0.001)。
图5.

RDW/HCT与脓毒症患者28 d死亡风险的RCS分析
Fig.5 RCS analysis of the association between RDW/HCT and 28-day mortality risk in patients with sepsis.
3. 讨论
本研究基于MIMIC-IV和eICU-CRD数据库的大样本分析发现,RDW/HCT比值与脓毒症患者28 d全因死亡呈非线性“U型”关系。且均存在显著的阈值效应。这一结果在不同数据库中表现出良好的稳健性。尽管两数据库的风险拐点存在差异(MIMIC-IV为0.35,eICU-CRD为0.48),这可能与不同来源患者的异质性、实验室检测方法的差异以及人群分布特征有关,但整体非线性趋势的高度一致性提示RDW/HCT作为脓毒症预后评估指标的临床价值。
相对于单独使用RDW或HCT,RDW/HCT比值综合了红细胞数量和红细胞体积的变化,能更全面地反映红细胞在生成与功能方面的双重异常。既往研究关于RDW/HCT比值与疾病预后的相关性研究较少。Li 等[18]研究发现RDW能较好预测危重心血管病人的全因死亡率。两者存在非线性关系,表现为死亡率随着RDW的降低而下降;直至拐点,死亡率随着RDW的升高而上升,随后进入平台期。也有研究发现HCT与急性缺血性脑卒中男性患者住院期间死亡率存在“U型”相关,过高的HCT值与死亡风险升高相关[19]。本研究为避免RDW或HCT受患者自身的慢性疾病、营养状况和输血补液的影响,采用RDW/HCT,发现在MIMIC-IV脓毒症人群中,当RDW/HCT比值≥0.35,RDW/HCT比值每增加1,脓毒症患者的28 d死亡风险增加2.034倍,这提示该复合指标识别高死亡风险的脓毒症人群具有较高的敏感性。
然而,RDW/HCT比值增高与脓毒症患者死亡风险增加的机制尚不清楚,可能与以下几个因素有关:氧供与微循环障碍:低HCT意味着血液中红细胞数量减少,血液携氧能力下降。Luo等[11]研究显示,与常规HCT组相比,低HCT组脓毒症患者的30 d死亡率增加了1.589倍(HR=1.589,95% CI:1.009~2.979)。在此基础上,RDW升高表明红细胞变异性增大,可能增加微循环中的血流阻力,影响组织灌注,从而加重器官缺氧与缺血性损伤[20]。红细胞稳态受损:RDW升高与红细胞生成紊乱相关,且与心血管疾病、慢性肺部疾病、急性脑梗死等多种疾病的预后有关[21-24]。红细胞数量与质量的异常可能反映了脓毒症引起的红细胞生成功能的衰退与寿命的缩短。研究表明,RDW对脓毒症死亡率的预测能力与SOFA评分和APACHE-II评分相当[25]。血流动力学不稳定:RDW升高和HCT下降可能反映了脓毒症患者血容量变化或贫血加重,引发血压不稳定、低血压和心输出量下降等,增加脓毒症死亡风险[26, 27]。炎症与氧化应激叠加:高RDW伴随炎症级联反应的放大与氧化应激的加剧,产生的活性氧还可致红细胞变形性降低;若HCT同时降低,将进一步加重机体缺氧状况,加速重要脏器衰竭[13, 28-30]。
本研究存在一定局限性。首先,作为回顾性研究,无法排除选择偏倚和数据收集过程中潜在的误差。其次,MIMIC-IV数据库与eICU-CRD的RDW/HCT的拐点存在差异,但整体趋势的一致性表明该非线性关系具有稳健性,未来需在更大规模、多中心及更具一致性的人群中开展前瞻性临床研究,明确其临床指导意义。
综上所述,RDW/HCT比值升高与脓毒症患者28 d全因死亡率有关。作为一种简便、高效的生物标志物,该指标有助于脓毒症患者早期死亡风险分层及预后评估。
基金资助
国家自然科学基金(82002084,81971806);四川省科技计划项目(2024NSFC1535,2023YFH0074)
Supported by National Natural Science Foundation of China (82002084,81971806).
参考文献
- 1. Rudd KE, Johnson SC, Agesa KM, et al. Global, regional, and national sepsis incidence and mortality, 1990-2017: analysis for the Global Burden of Disease Study[J]. Lancet, 2020, 395(10219): 200-11. doi: 10.1016/s0140-6736(19)32989-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Wiersinga WJ, van der Poll T. Immunopathophysiology of human sepsis[J]. eBioMedicine, 2022, 86: 104363. doi: 10.1016/j.ebiom.2022.104363 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Pierrakos C, Velissaris D, Bisdorff M, et al. Biomarkers of sepsis: time for a reappraisal[J]. Crit Care, 2020, 24(1): 287. doi: 10.1186/s13054-020-02993-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Icardi M, Erickson Y, Kilborn S, et al. CD64 index provides simple and predictive testing for detection and monitoring of sepsis and bacterial infection in hospital patients[J]. J Clin Microbiol, 2009, 47(12): 3914-9. doi: 10.1128/JCM.00628-09 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Pierrakos C, Vincent JL. Sepsis biomarkers: a review[J]. Crit Care, 2010, 14(1): R15. doi: 10.1186/cc8872 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Barichello T, Generoso JS, Singer M, et al. Biomarkers for sepsis: more than just fever and leukocytosis-a narrative review[J]. Crit Care, 2022, 26(1): 14. doi: 10.1186/s13054-021-03862-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Wu HS, Liao BL, Cao TS, et al. Diagnostic value of RDW for the prediction of mortality in adult sepsis patients: a systematic review and meta-analysis[J]. Front Immunol, 2022, 13: 997853. doi: 10.3389/fimmu.2022.997853 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Kim CH, Park JT, Kim EJ, et al. An increase in red blood cell distribution width from baseline predicts mortality in patients with severe sepsis or septic shock[J]. Crit Care, 2013, 17(6): R282. doi: 10.1186/cc13145 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Wu YC, Chen HH, Chao WC. Association between red blood cell distribution width and 30-day mortality in critically ill septic patients: a propensity score-matched study[J]. J Intensive Care, 2024, 12(1): 34. doi: 10.1186/s40560-024-00747-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Hong J, Hu X, Liu WY, et al. Impact of red cell distribution width and red cell distribution width/albumin ratio on all-cause mortality in patients with type 2 diabetes and foot ulcers: a retrospective cohort study[J]. Cardiovasc Diabetol, 2022, 21(1): 91. doi: 10.1186/s12933-022-01534-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Luo MD, Chen Y, Cheng Y, et al. Association between hematocrit and the 30-day mortality of patients with sepsis: a retrospective analysis based on the large-scale clinical database MIMIC-IV[J]. PLoS One, 2022, 17(3): e0265758. doi: 10.1371/journal.pone.0265758 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Wan JH, Kuang MB, Xiong SX, et al. Development and validation of dynamic clinical subphenotypes for acute pancreatitis using hematocrit and blood urea nitrogen trajectories in intensive care unit: a multicenter retrospective cohort study[J]. J Adv Res, 2026, S2090-1232( 26)00004-4. doi: 10.1016/j.jare.2026.01.004 [DOI] [PubMed] [Google Scholar]
- 13. Janz DR, Bastarache JA, Peterson JF, et al. Association between cell-free hemoglobin, acetaminophen, and mortality in patients with sepsis: an observational study[J]. Crit Care Med, 2013, 41(3): 784-90. doi: 10.1097/CCM.0b013e3182741a54 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Salvagno GL, Sanchis-Gomar F, Picanza A, et al. Red blood cell distribution width: a simple parameter with multiple clinical applications[J]. Crit Rev Clin Lab Sci, 2015, 52(2): 86-105. doi: 10.3109/10408363.2014.992064 [DOI] [PubMed] [Google Scholar]
- 15. Duan WT, Yang F, Ling H, et al. Association between lactate to hematocrit ratio and 30-day all-cause mortality in patients with sepsis: a retrospective analysis of the Medical Information Mart for Intensive Care IV database[J]. Front Med, 2024, 11: 1422883. doi: 10.3389/fmed.2024.1422883 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. 朱义玲, 崔娟娟, 李燕双, 等. 外周血中性粒细胞与淋巴细胞比值和淋巴细胞与单核细胞比值在脓毒症患者诊断和预后中的价值[J]. 中国感染与化疗杂志, 2025, 25(2): 149-54. doi: 10.16718/j.1009-7708.2025.02.005 [DOI] [Google Scholar]
- 17. Long ZW, Zeng QY, Ou YE, et al. Red blood cell distribution width/hematocrit ratio: a new predictor of 28 days all-cause mortality of AECOPD patients in ICU[J]. Int J Chronic Obstr Pulm Dis, 2024, 19: 2497-516. doi: 10.2147/COPD.S492049 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Li S, Zhang W, Liang X. Red blood cell distribution width and mortality risk in critically ill cardiovascular patients[J]. Heliyon, 2023, 9(11): e22225. doi: 10.1016/j.heliyon.2023.e22225 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Sico JJ, Myers LJ, Fenton BJ, et al. Association between admission haematocrit and mortality among men with acute ischaemic stroke[J]. Stroke Vasc Neurol, 2018, 3(3): 160-8. doi: 10.1136/svn-2018-000149 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Hou HF, Sun T, Li C, et al. An overall and dose-response meta-analysis of red blood cell distribution width and CVD outcomes[J]. Sci Rep, 2017, 7: 43420. doi: 10.1038/srep43420 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Tian Q, Mitchell BA, Moaddel R, et al. Metabolomic markers mediate erythrocyte anisocytosis in older adults: Results from three independent aging cohorts[J]. J Intern Med, 2023, 293(5): 589-99. doi: 10.1111/joim.13612 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Zhu MP, Dai LS, Wan L, et al. Dynamic increase of red cell distribution width predicts increased risk of 30-day readmission in patients with acute exacerbation of chronic obstructive pulmonary disease[J]. Int J Chron Obstruct Pulmon Dis, 2021, 16: 393-400. doi: 10.2147/COPD.S291833 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Lan W, Liu E, Sun D, et al. Red cell distribution in critically ill patients with chronic obstructive pulmonary disease[J]. Pulmonology, 2024, 30(1): 34-42. doi: 10.1016/j.pulmoe.2022.04.001 [DOI] [PubMed] [Google Scholar]
- 24. Kim J, Kim YD, Song TJ, et al. Red blood cell distribution width is associated with poor clinical outcome in acute cerebral infarction[J]. Thromb Haemost, 2012, 108(2): 349-56. doi: 10.1160/TH12-03-0165 [DOI] [PubMed] [Google Scholar]
- 25. Xu WG, Huo JY, Chen GJ, et al. Association between red blood cell distribution width to albumin ratio and prognosis of patients with sepsis: a retrospective cohort study[J]. Front Nutr, 2022, 9: 1019502. doi: 10.3389/fnut.2022.1019502 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Hernandez G, Ospina-Tascón GA, Kattan E, et al. Personalized hemodynamic resuscitation targeting capillary refill time in early septic shock: the ANDROMEDA-SHOCK-2 randomized clinical trial[J]. JAMA, 2025, 334(22): 1988-99. doi: 10.1001/jama.2025.20402 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Hariri G, Joffre J, Leblanc G, et al. Narrative review: clinical assessment of peripheral tissue perfusion in septic shock[J]. Ann Intensive Care, 2019, 9(1): 37. doi: 10.1186/s13613-019-0511-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Ma SM, Lin L, Zheng SW, et al. Associations between the red blood cell distribution width and 30-day mortality in critically ill patients with delirium: a retrospective study using the MIMIC-IV database[J]. Front Aging Neurosci, 2025, 17: 1599858. doi: 10.3389/fnagi.2025.1599858 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Aird WC. The hematologic system as a marker of organ dysfunction in sepsis[J]. Mayo Clin Proc, 2003, 78(7): 869-81. doi: 10.4065/78.7.869 [DOI] [PubMed] [Google Scholar]
- 30. Vella R, Panci D, Carini F, et al. Cytokines in sepsis: a critical review of the literature on systemic inflammation and multiple organ dysfunction[J]. Front Immunol, 2025, 16: 1682306. doi: 10.3389/fimmu.2025.1682306 [DOI] [PMC free article] [PubMed] [Google Scholar]
