Department of Population Medicine, Harvard Pilgrim Health Care Institute and Harvard Medical School, Boston, Massachusetts.
Division of Cancer Epidemiology and Genetics, NCI, NIH, Bethesda, Maryland.
Cancer Epidemiol Biomarkers Prev. 2022 Oct 4;31(10):1890-1895. doi: 10.1158/1055-9965.EPI-21-1451.
Evaluations of cancer etiology and safety and effectiveness of cancer treatments are predicated on large numbers of patients with sufficient baseline and follow-up data. To assess feasibility of FDA's Sentinel System's electronic healthcare data for surveillance of malignancy onset and examination of product safety, this study examined patterns of enrollment surrounding new-onset cancers.
Using a retrospective cohort of patients based on administrative claims, we identified incident events of 19 cancers among 292.5 million health plan members from January 2000 to February 2020 using International Classification of Diseases (ICD) diagnosis codes. Annual incident cases were stratified by sex, age, medical and drug coverage, and insurer type. Descriptive statistics were calculated for observable time prior to and following diagnosis.
We identified 10,697,573 incident cancer events among members with medical coverage. When drug coverage was additionally required, number of incident cancers was reduced by 41%. Medicare data contributed 61% of cases, with similar duration trends as other insurers. Mean duration of follow-up prior to diagnosis ranged from 4.0 to 4.6 years, whereas follow-up post diagnosis ranged from 1.1 to 3.3 years. Approximately a third (36.1%) had at least 2 years both prior to and following diagnosis.
The FDA Sentinel System's electronic healthcare data may be useful for characterizing relatively short latency cancer risk, examining cancer drug utilization and safety after diagnosis, and conducting surveillance for acute adverse events among patients with cancers.
A national distributed system with electronic health data, the Sentinel system provides opportunity for rapid pharmacoepidemiologic assessments relevant in oncology.
癌症病因学和治疗安全性及有效性的评估取决于大量具有足够基线和随访数据的患者。为评估 FDA 哨兵系统电子医疗数据在监测恶性肿瘤发病和检查产品安全性方面的可行性，本研究检查了新发癌症周围的入组模式。
我们使用基于行政索赔的患者回顾性队列，使用国际疾病分类（ICD）诊断代码，在 2000 年 1 月至 2020 年 2 月期间，从 2.925 亿健康计划成员中确定了 19 种癌症的首发事件。根据性别、年龄、医疗和药物覆盖范围以及保险公司类型对每年的首发病例进行分层。对诊断前和诊断后的可观察时间进行描述性统计。
我们在有医疗覆盖的成员中发现了 10697573 例新发癌症事件。当另外需要药物覆盖时，新发癌症的数量减少了 41%。医疗保险数据贡献了 61%的病例，其趋势与其他保险公司相似。诊断前的平均随访时间为 4.0 至 4.6 年，而诊断后的平均随访时间为 1.1 至 3.3 年。大约三分之一（36.1%）在诊断前后均至少有 2 年的随访时间。
FDA 哨兵系统的电子医疗数据可能有助于描述相对较短潜伏期的癌症风险，检查诊断后癌症药物的利用和安全性，并对癌症患者的急性不良事件进行监测。
一个具有电子健康数据的全国分布式系统，哨兵系统为快速开展肿瘤学相关药物流行病学评估提供了机会。
