Sex-specific hemoglobin thresholds for oxygen saturation: A non-linear regression analysis based on Tibetan inpatients

  • Zhuoma GONGJUE 1 ,
  • Yiting MAO 1 ,
  • Puzhen DAWA 1 ,
  • Ciren LABA 1 ,
  • Qi YAN , 1, 2, *
Expand
  • 1. Department of Anesthesiology, People's Hospital of Xizang Autonomous Region, Lhasa 850002, China
  • 2. Department of Anesthesiology, Peking University People's Hospital, Beijing 100044, China
YAN Qi, e-mail,

Received date: 2025-08-12

  Online published: 2026-01-05

Supported by

Peking University People's Hospital Research and Development Funds(RDJP2022-68)

Peking University People's Hospital Research and Development Funds(RDE2022-21)

Beijing Resident Standardized Training Quality Improvement Project(2023011)

Xizang Autonomous Region Natural Science Foundation Group Medical Aid Project(XZ2024ZR-ZY013(Z))

Copyright

All rights reserved. Unauthorized reproduction is prohibited.

Abstract

Objective: To investigate the complex association between hemoglobin levels and resting pulse oxygen saturation (SpO2) among surgical inpatients in a high-altitude environment, and to define precise, gender-specific physiological adaptation thresholds for hemoglobin, thereby providing evidence-based reference values for health management strategies in high-altitude populations. Methods: This cross-sectional study enrolled adult inpatients from People's Hospital of Xizang Autonomous Region (altitude: 3 650 m) between January 2023 and October 2024. To rigorously evaluate the relationship between hemoglobin and resting SpO2, multivariate linear regression analysis was performed. Furthermore, restricted cubic spline models and likelihood ratio tests were utilized to explore potential non-linear threshold effects. Models were adjusted for potential confounding factors, including age, body mass index, hypertension, diabetes, smoking, alcoholism, and comorbid pulmonary diseases. Results: A total of 3 083 inpatients were enrolled in the final analysis, comprising 1 450 males and 1 633 females. The restricted cubic spline analysis revealed a significant non-linear relationship between hemoglobin levels and SpO2 for the total population (non-linear test, P=0.006), indicating a distinct turning point in the dose-response curve. Two-piecewise linear regression models subsequently identified distinct inflection points for each sex. For female inpatients, the hemoglobin threshold was determined to be 15.482 g/dL; Above this level, SpO2 significantly decreased as hemoglobin increased (β=-0.477, 95%CI: -0.760 to -0.193, P=0.001). Similarly, for male inpatients, the threshold was identified at 17.288 g/dL; Exceeding this value resulted in a significant inverse correlation between hemoglobin and SpO2 (β=-0.344, 95%CI: -0.550 to -0.138, P=0.001). Conclusion: This study establishes specific hemoglobin thresholds for oxygenation status in hospitalized patients at high altitude: 15.482 g/dL for females and 17.288 g/dL for males. Beyond these cut-off values, further increases in hemoglobin are associated with a deterioration in SpO2, suggesting a transition from physiological adaptation to maladaptive hemoconcentration. These findings highlight the necessity of sex-stratified hemoglobin monitoring and potential preoperative optimization in high-altitude health management to prevent hypoxia-related complications.

Cite this article

Zhuoma GONGJUE , Yiting MAO , Puzhen DAWA , Ciren LABA , Qi YAN . Sex-specific hemoglobin thresholds for oxygen saturation: A non-linear regression analysis based on Tibetan inpatients[J]. Journal of Peking University(Health Sciences), 2026 , 58(1) : 196 -200 . DOI: 10.19723/j.issn.1671-167X.2026.01.026

高原环境由于大气压力降低和氧分压减少[1],需要机体进行复杂的适应性反应[2-3],以确保向机体提供足够的氧气。血红蛋白的适应性改变在高海拔生理反应中处于核心地位[4],在低压低氧环境中,为了维持正常的动脉氧含量,机体通过增加血红蛋白浓度进行代偿。这种适应性反应一方面源于血浆容量的快速减少(血液浓缩),另一方面依赖于缺氧诱导因子(hypoxia inducible factor,HIF),特别是HIF-2α介导的红细胞生成增加[4]。脾脏收缩也可短期提升血红蛋白浓度,在缺氧应激期间增加氧气输送[5]。脉搏氧饱和度(pulse oxygen saturation,SpO2)是一种广泛应用的非侵入性指标,可以方便且及时地评估外周氧饱和度,使用该指标监测氧合状态可以为高原环境中的健康管理提供可靠的数据。
虽然增加血红蛋白浓度在缺氧条件下可有效增强氧气运输,但过度的适应性反应,特别是病理性红细胞增多,会带来更加不利的后果[4],尤其在慢性高原病中,明显的血红蛋白水平提升会使血液黏度增加,从而加重血栓及心血管风险。但目前尚无研究分析最佳的血红蛋白适应水平,也未见关于高原居住者是否存在性别特异的最佳血红蛋白适应值的相关报道。既往有研究从理论计算的角度出发,发现血红蛋白水平达到14.7 g/dL时,静脉血氧分压达到最大值[6];也有学者提出,对于高海拔居住者,血红蛋白水平在18~21 g/dL可能是生理适应向病理损伤转化的过渡区间[7]。本研究旨在通过横断面研究,采用SpO2评估氧合状态,划定关于血红蛋白的有益生理适应与有害过度适应之间的阈值,希望能为高海拔健康管理提供依据。

1 资料与方法

1.1 研究设计与伦理

本研究为横断面研究,经西藏自治区人民医院伦理委员会批准(ME-TBHP-25-KJ-101),免除患者知情同意。研究遵循加强流行病学观察性研究报告(strengthening the reporting of observational studies in epidemiology,STROBE)指南。

1.2 研究对象

收集2023年1月至2024年10月在西藏自治区人民医院(海拔3 650 m)外科住院拟行择期非心胸外科手术的年龄≥18岁藏族患者。排除标准:病历中缺少体重指数(body mass index,BMI)记录或BMI值不符合生理可能范围(>50 kg/m2或 < 10 kg/m2)[8];记录的静息SpO2 < 60%者(视为伪值,予以剔除)。

1.3 研究变量

1.3.1 主要暴露变量

本研究的主要暴露变量为血红蛋白水平(单位为g/dL),通过查阅病历资料,记录手术前30 d内最近一次检测值[9-10]

1.3.2 主要结局指标

本研究的主要结局指标为静息SpO2,采用鱼跃指夹式脉搏血氧仪(型号YX301)监测;通过查阅电子病历记录,记录与检测血红蛋白同一天的SpO2值。

1.3.3 协变量

协变量主要为人口学资料和病史,包括年龄、性别、高血压病史(是/否)、吸烟史(从不吸烟、现在/既往吸烟)、术前合并肺部疾病(是/否)。若术前胸部X线片提示渗出、纤维化、结节、肺气肿、胸腔积液、肺炎、肺间质改变等则判定为合并肺部疾病。

1.3.4 数据核查

为确保数据准确性,由一名不参与患者临床护理且接受过科研培训的住院医师作为独立观察员,将数据集与电子出院记录进行核对验证。

1.4 统计学方法

使用R软件(4.2.1版)进行统计分析。使用描述性统计呈现数据,计量资料经Shapiro-Wilk检验评估正态性,符合正态分布者以均数±标准差表示,非正态分布者以中位数(P25, P75)表示,计数资料以频数(百分比)表示。组间比较时,分类变量采用卡方检验,正态分布计量资料多组间比较采用单因素方差分析,非正态分布计量资料多组间比较采用Kruskal-Wallis H检验。
由于男性、女性血红蛋白水平本身存在的差异,本文对男性、女性分别进行分析。采用多因素线性回归模型分析血红蛋白水平与静息SpO2的关联,计算β值及其95%置信区间。协变量选择基于既往文献、临床经验。为探索性别特异的血红蛋白水平与静息SpO2间潜在的非线性剂量-反应关系,采用含四个节点(分别位于第5、35、65和95百分位数)的限制性立方样条模型[9]。节点设置遵循Harrell推荐方案[11],在优化模型灵活性的同时最小化过拟合风险。通过最大对数似然估计法进行分段线性多因素回归以确定关键血红蛋白阈值,并采用似然比检验评估阈值上下关联性差异。双侧P < 0.05为差异有统计学意义。

2 结果

2.1 研究对象基线特征

本研究共纳入藏族患者3 083例(表 1),包括男性1 450例、女性1 633例,男性和女性平均年龄分别为(47.3±14.1)岁和(43.0±14.7)岁,男性吸烟率和饮酒率均显著高于女性(P < 0.001),女性高血压患病率及BMI高于男性。男性和女性静息SpO2中位数均为90.0%(88.0%,91.0%),组间差异无统计学意义。
表1 藏族住院患者基线特征

Table 1 Baseline characteristics by sex in Tibetan inpatients

Variables Total (n=3 083) Male (n=1 450) Female (n=1 633) P
Age/years, $\bar x \pm s$ 45.0±14.6 47.3±14.1 43.0±14.7 < 0.001
BMI/(kg/m2), $\bar x \pm s$ 23.8±4.2 23.5±4.0 24.1±4.4 < 0.001
Hypertension, n(%) 275 (8.9) 107 (7.4) 168 (10.3) 0.005
Diabetes, n(%) 46 (1.5) 26 (1.8) 20 (1.2) 0.194
Smoking history, n(%) 441 (14.3) 419 (28.9) 22 (1.3) < 0.001
Alcoholism, n(%) 503 (16.3) 408 (28.1) 95 (5.8) < 0.001
Comorbid pulmonary disease, n(%) 1 505 (48.8) 729 (50.3) 776 (47.5) 0.127
Hemoglobin/(g/dL), $\bar x \pm s$ 15.5±2.5 16.8±2.1 14.4±2.2 < 0.001
SpO2 at rest/%, M (P25, P75) 90.0 (88.0, 91.0) 90.0 (88.0, 91.0) 90.0 (88.0, 91.0) 0.121
SpO2 at rest<88%, n(%) 615 (19.9) 268 (18.5) 347 (21.2) 0.055

BMI, body mass index; SpO2, pulse oxygen saturation.

2.2 女性血红蛋白水平与静息SpO2的关联分析

多因素线性回归分析结果见表 2。限制性立方样条分析结果表明,女性血红蛋白水平与静息SpO2存在非线性关联(图 1),提示女性血红蛋白水平与静息SpO2的关系存在阈值效应。分段线性多因素回归分析结果显示,当女性血红蛋白 < 15.482 g/dL时,静息SpO2无显著变化(β=-0.021,95%CI:-0.121~0.080,P=0.688);当女性血红蛋白≥15.482 g/dL时,静息SpO2随血红蛋白升高显著下降(β=-0.477,95%CI:-0.760~-0.193,P=0.001,表 3)。上述结果提示,在高原地区外科住院女性患者中可能存在一个最佳血红蛋白阈值,即15.482 g/dL,当血红蛋白在该水平附近时,用静息SpO2评价的氧合状态达到最佳。
表2 藏族住院患者血红蛋白与静息脉搏氧饱和度关联性的多因素线性回归分析(按性别分层)

Table 2 Multivariable linear regression analysis of the association between hemoglobin and pulse oxygen saturation at rest in Tibetan female and male inpatients

Items Crude coefficient (95%CI) Crude P value Adjusted coefficient (95%CI) Adjusted P value
Female -0.141 (-0.211 to -0.066) < 0.001 -0.081 (-0.149 to 0.004) 0.038
Male -0.184 (-0.255 to -0.113) < 0.001 -0.124(-0.202 to -0.047) 0.002
图1 藏族女性(A)和男性(B)住院患者血红蛋白与静息SpO2的非线性关联

Figure 1 Non-linear association between hemoglobin and SpO2 at rest in Tibetan female (A) and male (B) inpatients

Adjusted for age, body mass index, hypertension, diabetes, smoking history, alcoholism, comorbid pulmonary disease. The solid red curve depicts the fitted non-linear association, with the surrounding shaded area representing the 95% confidence interval. The lower panel displays the histogram of hemoglobin concentration distribution within the sample. SpO2, pulse oxygen saturation.

表3 藏族住院患者血红蛋白对静息脉搏氧饱和度影响的阈值效应分析

Table 3 Threshold effect analysis of hemoglobin on pulse oxygen saturation at rest in Tibetan inpatients

Gender Hemoglobin n Coefficient (95%CI) P value
Female < 15.482 g/dL 1 164 -0.021 (-0.121, 0.080) 0.688
≥15.482 g/dL 469 -0.477 (-0.760, -0.193) 0.001
Likelihood ratio test < 0.001
Male < 17.288 g/dL 904 0.032 (-0.099, 0.163) 0.634
≥17.288 g/dL 546 -0.344 (-0.550, -0.138) 0.001
Likelihood ratio test < 0.001

Adjusted for age, body mass index, hypertension, diabetes, smoking history, alcoholism, comorbid pulmonary disease. Coefficient represents change in pulse oxygen saturation per 1 g/dL hemoglobin increase.

2.3 男性血红蛋白水平与静息SpO2的关联分析

多因素线性回归分析结果见表 2。限制性立方样条分析结果揭示男性血红蛋白水平与静息SpO2存在非线性关联(图 1),提示男性血红蛋白水平与静息SpO2的关系存在阈值效应。分段线性多因素回归结果表明,当男性血红蛋白水平 < 17.288 g/dL时,静息SpO2无显著变化(β=-0.021,95%CI:-0.121~0.080,P=0.634);当男性血红蛋白水平≥17.288 g/dL时,静息SpO2随血红蛋白升高显著下降(β=-0.344,95%CI:-0.550~-0.138,P < 0.001,表 3)。上述结果提示,在高原地区外科住院男性患者中可能存在一个最佳血红蛋白阈值,即17.288 g/dL,当血红蛋白在该水平附近时,用静息SpO2评价的氧合状态达到最佳。

3 讨论

本研究深入探讨了高原地区居民血红蛋白水平与静息SpO2之间的关联,特别关注血红蛋白适应的益处与过度适应的潜在风险。本研究发现,男性患者和女性患者中血红蛋白水平均存在一个临界值,超过该值,静息SpO2会随着血红蛋白水平的增加而出现明显下降,提示过度提高血红蛋白水平可能不一定改善氧合状态。这与既往关于高原红细胞增多症的研究相符,该症通常定义为男性血红蛋白水平高于21 g/dL,女性高于19 g/dL[12-13]。本研究发现的血红蛋白阈值低于高原红细胞增多症的定义,提示在高原环境中血红蛋白水平的监测应更加严格,以防止健康风险。
既往研究多关注较低的血红蛋白水平对氧输送和临床结局的影响。例如,对心脏骤停患者的研究发现,血红蛋白低于9.9 g/dL与不良神经结局显著相关,提示低血红蛋白水平是氧输送不足和预后不良的重要风险因素[14]。部分研究对血红蛋白与血氧饱和度的关系进行了建模和回归分析。在单心室循环患儿的数学建模分析中,血红蛋白低于9 g/dL时,动脉和静脉血氧饱和度均难以维持在安全范围,需通过将血红蛋白升高至13 g/dL以上或增加心输出量来改善氧合[15]。在对先天性心脏病发绀患者的研究中,学者通过回归方程预测了不同血氧饱和度下的“最优”血红蛋白水平,发现二者在稳定患者中呈现较强的负相关关系,当血氧饱和度为90%时,最优血红蛋白水平位于17.5 g/dL,该值与本研究男性患者血红蛋白水平最佳阈值接近[16]。Villafuerte等[6]通过数学建模分析了安第斯高原男性在静息状态下的血红蛋白水平与氧输送关系,结果显示随着血红蛋白升高,静脉氧分压可在动脉氧分压仅适度下降的情况下得到提升和维持,但当血红蛋白达到14.7 g/dL时,静脉氧分压达到最大值,进一步升高血红蛋白并不能带来额外的氧输送优势。
目前,尚未见到有文献通过真实的患者数据得出血红蛋白增加到一定程度后对氧输送增加无益的界值,相关研究多为理论推导或针对特殊人群(心脏病、发绀患者)的间接推测。本研究填补了性别特异性血红蛋白最佳阈值研究的空白,为优化高原健康管理策略提供了新的视角。但本研究也存在一定的局限性。第一,研究对象为西藏三级医院的住院患者,结果可能不适用于健康人群或其他高原地区人群;医院海拔约3 650 m,未来的研究可纳入不同海拔人群,以验证阈值的普适性。第二,尽管校正了主要混杂变量,但未测量其他相关因素,如高原居住时长、体力活动强度及遗传适应特征(如EPAS1基因)等,这些也可能影响血红蛋白-氧饱和度的关联。第三,很多疾病会影响SpO2,如慢性阻塞性肺疾病、睡眠呼吸暂停低通气综合征等,但其诊断需要依赖于肺功能、多导睡眠图等特定检查,在高原医疗环境中难以获得;但由于本研究聚焦于外科术前择期手术患者,该人群通常处于相对稳定的临床状态,这在一定程度上可以减少因急性呼吸衰竭造成的严重偏倚,此外,本研究判定患者合并或未合并肺部疾病的依据是术前胸部X线片是否异常(包括渗出、纤维化、结节、肺气肿、肺炎、肺间质改变表现等多种表现),虽不能等同于特定疾病诊断,但仍能有效识别出存在潜在结构性肺病的患者。第四,SpO2监测受探头位置、末梢循环等干扰。第五,虽然发现性别特异性阈值,但未建立该阈值与临床结局(如器官功能障碍、活动耐量)的直接关联证据链。因此,未来还需进行更大规模的前瞻性队列研究,以观察这些阈值对长期健康结局的预测价值,并探讨其背后的因果关系与生理机制。
综上所述,本横断面研究表明,在高原外科住院患者中血红蛋白水平与静息SpO2之间存在性别特异性的关联阈值,这一发现为未来探索如何优化高原适应策略以改善人群健康提供了新的视角。

利益冲突  所有作者均声明不存在利益冲突。

作者贡献声明  贡觉卓玛、毛奕汀:提出研究思路,收集、整理数据,论文草稿撰写;达娃普珍:收集、整理数据;拉巴次仁:审定论文;闫琦:提出研究思路,设计研究方案,撰写论文及审定论文。所有作者均参与论文修改,并对最终文稿进行审读和确认。

1
Forrer A , Gaisl T , Sevik A , et al. Partial pressure of arterial oxygen in healthy adults at high altitudes: A systematic review and meta-analysis[J]. JAMA Netw Open, 2023, 6 (6): e2318036.

DOI

2
Storz JF , Bautista NM . Altitude acclimatization, hemoglobin-oxygen affinity, and circulatory oxygen transport in hypoxia[J]. Mol Aspects Med, 2022, 84, 101052.

DOI

3
Parati G , Agostoni P , Basnyat B , et al. Clinical recommendations for high altitude exposure of individuals with pre-existing cardiovascular conditions: A joint statement by the European Society of Cardiology, the Council on Hypertension of the European Society of Cardiology, the European Society of Hypertension, the International Society of Mountain Medicine, the Italian Society of Hypertension and the Italian Society of Mountain Medicine[J]. Eur Heart J, 2018, 39 (17): 1546- 1554.

DOI

4
Villafuerte FC , Simonson TS , Bermudez D , et al. High-altitude erythrocytosis: Mechanisms of adaptive and maladaptive responses[J]. Physiology, 2022, 37 (4): 175- 186.

DOI

5
Schagatay E , Lunde A , Nilsson S , et al. Spleen contraction elevates hemoglobin concentration at high altitude during rest and exercise[J]. Eur J Appl Physiol, 2020, 120 (12): 2693- 2704.

DOI

6
Villafuerte FC , Cárdenas R , Monge-C C . Optimal hemoglobin concentration and high altitude: A theoretical approach for Andean men at rest[J]. J Appl Physiol (1985), 2004, 96 (5): 1581- 1588.

DOI

7
Reeves JT , Leon-Velarde F . Chronic mountain sickness: Recent studies of the relationship between hemoglobin concentration and oxygen transport[J]. High Alt Med Biol, 2004, 5 (2): 147- 155.

DOI

8
Kendale SM , Blitz JD . Increasing body mass index and the incidence of intraoperative hypoxemia[J]. J Clin Anesth, 2016, 33, 97- 104.

DOI

9
Zhao BC , Lei SH , Liu JM , et al. Sex-specific associations between preoperative hemoglobin and outcomes after major non-cardiac surgery: A retrospective cohort study[J]. Anesth Analg, 2023, 137 (5): 1019- 1028.

DOI

10
Ripoll JG , Smith MM , Hanson AC , et al. Sex-specific associations between preoperative anemia and postoperative clinical outcomes in patients undergoing cardiac surgery[J]. Anesth Analg, 2021, 132 (4): 1101- 1111.

DOI

11
Harrell FE . Regression modeling strategies: With applications to linear models, logistic and ordinal regression, and survival analysis[M]. Cham, Switzerland: Springer International Publishing, 2015.

12
Liu X , Yang C , Zhang X , et al. Association between hemoglobin concentration and hypertension risk in native Tibetans at high altitude[J]. J Clin Hypertens, 2024, 26 (1): 17- 23.

DOI

13
De Ferrari A , Miranda JJ , Gilman RH , et al. Prevalence, clinical profile, iron status, and subject-specific traits for excessive erythrocytosis in Andean adults living permanently at 3, 825 meters above sea level[J]. Chest, 2014, 146 (5): 1327- 1336.

DOI

14
Zama Cavicchi F , Iesu E , Franchi F , et al. Low hemoglobin and venous saturation levels are associated with poor neurological outcomes after cardiac arrest[J]. Resuscitation, 2020, 153, 202- 208.

DOI

15
Ahmed M , Acosta SI , Hoffman GM , et al. Mathematical analysis of hemoglobin target in univentricular parallel circulation[J]. J Thorac Cardiovasc Surg, 2023, 166 (1): 214- 220.

DOI

16
Broberg CS , Jayaweera AR , Diller GP , et al. Seeking optimal relation between oxygen saturation and hemoglobin concentration in adults with cyanosis from congenital heart disease[J]. Am J Cardiol, 2011, 107 (4): 595- 599.

DOI

Outlines

/