PO.PS01.10 · 人群科学
急诊科（ED）参与癌症诊断与Medicare人群中不同癌种生存率之间的关联
Association between emergency department (ED) involvement in cancer diagnosis and survival across cancer types in the Medicare population
作者与单位 Authors & Affiliations
摘要 Abstract
中文摘要
背景：相当一部分癌症是在急诊就诊后被诊断出来的，这一途径始终与确诊时分期较晚、生存率较差及患者体验较差相关联。我们旨在量化Medicare人群中不同癌种里ED参与癌症诊断与总生存期之间的关联。
方法：我们对2010年至2020年间确诊为浸润性癌症、随访死亡情况至2020年12月31日的SEER-Medicare受益人开展了一项回顾性队列研究，将最早的癌症相关理赔记录定义为索引日期。我们排除了在索引日期前不足连续12个月及索引日期后不足1个月拥有Medicare A/B/D部分连续参保，或在一年内有既往癌症诊断的受益人。在索引日期前30天内有急诊就诊的患者被归类为其诊断存在ED参与。总生存期采用17个独立的多变量Cox比例风险模型进行评估，每个模型对应一个根据SEER主要癌症部位并具有可用AJCC分期数据所定义的癌症分组（乳腺、前列腺、肺、结直肠、膀胱/尿路上皮、子宫、肾脏、淋巴瘤、胰腺、口腔、肝脏、胃、卵巢、食管、宫颈、肛门）。协变量包括年龄、性别、种族/族裔、双重资格状态、Charlson合并症指数、诊断年份及肿瘤分期。本分析未对某些潜在混杂因素进行校正，包括症状持续时间、体能状态及治疗可及性或起始情况的差异，这些在数据集中未被记录。
结果：在818,120名新诊断癌症的受益人（平均年龄74.4岁）中，26.4%的诊断存在ED参与。在所有癌种中，ED参与的比例随分期升高而增加（I期：13.8%，II期：26.3%，III期：23.7%，IV期：47.3%），并在不同癌种间有所不同，从乳腺癌的6.1%到胰腺癌的53.4%不等。研究期间，存在ED参与的患者死亡率为44.4%，而无ED参与者为19%，46%的癌症死亡发生在ED参与之后。ED参与在几乎所有癌种中均与显著更高的死亡率相关，宫颈癌除外（HR 0.81；95% CI 0.28-2.37），范围从胰腺癌（1.61；1.54-1.67）和结直肠癌（1.88；1.84-1.93）到食管癌（3.86；2.25-6.62）和膀胱癌（4.08；3.23-5.18）。
结论：ED参与与癌症患者总体死亡率的相当一部分相关，并且在校正了包括社会经济状况、合并症及诊断时癌症分期等患者因素后，仍是死亡率的一个强有力的独立预测因素。这些发现凸显了推动更早、非急诊诊断途径及结构化随访策略的必要性。
查看英文原文 English abstract
Background : A substantial proportion of cancers are diagnosed following an emergency presentation, a pathway consistently associated with advanced stage at diagnosis, poorer survival, and worse patient experience. We aimed to quantify the association between ED involvement in the diagnosis of cancer and overall survival across different cancer types in the Medicare population.
Methods : We conducted a retrospective cohort study of SEER-Medicare beneficiaries diagnosed with invasive cancer between 2010 and 2020 and followed for mortality until 12/31/2020, defining the earliest cancer-related claim as the index date. We excluded beneficiaries without at least 12 months of continuous Medicare Parts A/B/D enrollment before and 1 month after the index date or with a prior cancer diagnosis within one year. Patients with an ED visit within 30 days before the index date were classified as having ED involvement in their diagnosis. Overall survival was evaluated using 17 independent multivariate Cox proportional hazards models, each corresponding to a cancer grouping defined according to SEER major cancer sites with available AJCC staging data (breast, prostate, lung, colorectal, bladder/urothelial tract, uterine, kidney, lymphoma, pancreatic, oral, liver, stomach, ovarian, esophageal, cervical, anal). Covariates included age, sex, race/ethnicity, dual eligibility status, Charlson Comorbidity Index, year of diagnosis, and tumor stage. This analysis did not adjust for some potential confounders including differences in duration of symptoms, performance status, and therapy access or initiation which were not captured in the dataset.
Results : Among 818,120 beneficiaries newly diagnosed with cancer (mean age 74.4 years), 26.4% had ED involvement in their diagnosis. The proportion with ED involvement increased with advancing stage across all cancer types (stage I: 13.8%, II: 26.3%, III: 23.7%, IV: 47.3%) and varied across cancer types, ranging from 6.1% for breast cancer to 53.4% for pancreatic cancer. Mortality during the study period was 44.4% among patients with ED involvement versus 19% among those without, with 46% of cancer deaths following ED involvement. ED involvement was associated with significantly higher mortality across nearly all cancer types, except cervical cancer (HR 0.81; 95% CI 0.28-2.37), ranging from pancreatic (1.61; 1.54-1.67) and colorectal (1.88; 1.84-1.93) to esophageal (3.86; 2.25-6.62) and bladder (4.08; 3.23-5.18).
Conclusion : ED involvement was associated with a significant fraction of overall mortality in patients with cancer and was a strong independent predictor of mortality after adjustment for patient factors including socioeconomics, comorbidities, and cancer stage at diagnosis. These findings highlight the need for strategies that promote earlier, non-emergency diagnostic pathways and structured follow-up.
利益披露 Disclosure
S. Heitkamp,
Grail, Inc. Other, Employee of Avalere Health, which received funding from GRAIL for this research.
E. Olsen, None.
A. Tafazzoli,
Grail, Inc. Employment, Stock.
B. Houpt,
Grail, Inc Other, Employee of Avalere Health, which received funding from GRAIL for this research.
O. Hunt,
Grail, Inc. Other, Employee of Avalere Health, which received funding from GRAIL for this research.
A. R. Kansal,
Grail, Inc. Employment, Stock.
A. M. Fendrick,
AbbVie ).
Centivo ).
EmblemHealth ).
Elektra ).
Exact Sciences ).
Grail, Inc ).
Health at Scale Technologies Stock, ).
HealthCorum ).
Johnson & Johnson ).
Medtronic ).
MedZed ).
Merck ).
Mother Goose Health Stock, ).
Phathom Pharmaceuticals ).
Proton Intelligence ).
Sempre Health Stock, ).
Sera Prognostics ).
Silver Fern Healthcare Stock, ).
Teladoc Health ).
E. Klein,
Grail, Inc Employment, Stock.