Chen Chuqian, Zhang Robert Jiqi
Department of Medical Humanities (C.C.), School of Humanities, Southeast University, Nanjing, China.
School of Psychology (R.J.Z.), Nanjing Normal University, Nanjing, China.
J Pain Symptom Manage. 2025 Oct;70(4):379-391.e3. doi: 10.1016/j.jpainsymman.2025.07.009. Epub 2025 Jul 15.
Understanding older adults' preferences for end-of-life care (EoLC) is vital for respecting their wishes and informing effective service planning and policy development. Previous research has examined factors influencing different dimensions of EoLC preferences separately, but few studies have explored these dimensions as interconnected patterns and viewed older adults as heterogeneous using a person-centered approach.
This study aims to: 1) identify heterogeneous latent patterns across seven dimensions of EoLC preferences among Chinese older adults; 2) describe and explain these patterns; and 3) predict membership within these patterns.
Survey data from 646 urban-dwelling older adults aged 60 and above across 26 provincial-level administrative divisions in Mainland China were analyzed. EoLC preferences regarding willingness to know diagnosis, willingness to know prognosis, decision-maker, treatment goals, place of care, caregiver, and setting advance directives were assessed alongside demographics, resources, knowledge and attitudes, and caregiving/bereavement experiences. Latent class analysis (LCA), 3-step regressions, and Catboost machine learning models were employed to identify subgroups, examine between-group differences, and predict subgroup membership, respectively.
LCA identified three latent patterns: "low self-determination, quality-goal, family-oriented care" (9.1%), "high self-determination, quality-goal, family-oriented care" (54.0%), and "high self-determination, quantity-goal, professional-oriented care" (36.9%). Significant between-group differences were found in education, marital status, living arrangements, family income, social support, EoLC knowledge, general trust, and professional-patient trust. Machine learning models revealed that high general trust predicts membership in the high self-determination, quality-goal, family-oriented care group, while low filial piety expectations predict membership in the high self-determination, quantity-goal, professional-oriented care group.
Among Chinese older adults, three EoLC preference patterns were found, which were characterized by low family connections, low trust in professionals combined with adequate resources, and extensive knowledge, respectively. High general trust and low filial piety expectations were key predictors for two of the three patterns.
了解老年人对临终关怀（EoLC）的偏好对于尊重他们的意愿以及为有效的服务规划和政策制定提供信息至关重要。以往的研究分别考察了影响临终关怀偏好不同维度的因素，但很少有研究将这些维度作为相互关联的模式进行探索，也很少有人以以人为本的方法将老年人视为异质群体。
本研究旨在：1）识别中国老年人在临终关怀偏好的七个维度上的异质潜在模式；2）描述并解释这些模式；3）预测这些模式中的成员归属。
对来自中国大陆26个省级行政区的646名60岁及以上的城市老年人的调查数据进行了分析。评估了关于了解诊断意愿、了解预后意愿、决策者、治疗目标、护理地点、护理者以及制定预先指示的临终关怀偏好，同时还收集了人口统计学、资源、知识和态度以及护理/丧亲经历等信息。分别采用潜在类别分析（LCA）、三步回归和Catboost机器学习模型来识别亚组、检验组间差异以及预测亚组成员归属。
LCA识别出三种潜在模式：“低自主决定权、质量目标、家庭导向型护理”（9.1%）、“高自主决定权、质量目标、家庭导向型护理”（54.0%）和“高自主决定权、数量目标、专业导向型护理”（36.9%）。在教育程度、婚姻状况、居住安排、家庭收入、社会支持、临终关怀知识、一般信任和专业-患者信任方面发现了显著的组间差异。机器学习模型显示，高一般信任预测属于高自主决定权、质量目标、家庭导向型护理组，而低孝顺期望预测属于高自主决定权、数量目标、专业导向型护理组。
在中国老年人中，发现了三种临终关怀偏好模式，其特点分别是家庭联系少、对专业人员信任低但资源充足以及知识丰富。高一般信任和低孝顺期望是这三种模式中两种模式的关键预测因素。
