Palmier James, Lanzrath Brian J
ExamOne, 10101 Renner Blvd, Lenexa, KS 66219, USA.
J Insur Med. 2012;43(3):162-8.
Identification of statistically significant laboratory and biometric predictors of cancer-related mortality among insured individuals.
Numerous clinical studies have identified correlations between various laboratory results and physical measurements and cancer risk, often of a uni-variate nature. A study of life insurance claims has permitted a broad multivariate analysis of laboratory and biometric risk factors for cancer mortality in an insured population.
Of the applicants with complete laboratory and physical measurement profiles, 1.25 million were available and followed for an average of 4.7 years. Dates of 518 life insurance claims resulting from cancer deaths were recorded, and the resulting data set analyzed by multivariate Cox Proportional Hazards regression to identify statistically significant predictors of cancer-related mortality among insured individuals.
Among demographic variables, cancer deaths were found to be strongly associated with age and tobacco use, but not with gender. Among serum and urine analytes, liver function tests (principally GGT and ALP), hypocholesterolemia, proteinurea, and low fructosamine were found to be independently predictive of cancer mortality. Among physical measurement variables, there was a positive relationship between cancer mortality risk and height and relatively weak relationships with pulse and blood pressure. Weight and body mass index (BMI) were not statistically significant covariates.
The findings highlight the potential value of laboratory analytes and biometric measurements to cancer risk assessment including low to low-normal values in analytes (particularly cholesterol and fructosamine) whose diagnostic value in clinical practice and underwriting may be advantageous.
确定参保人群中与癌症相关死亡率具有统计学显著意义的实验室检查指标和生物特征预测因素。
众多临床研究已确定各种实验室检查结果和身体测量指标与癌症风险之间存在相关性，且通常具有单变量性质。一项针对人寿保险理赔的研究允许对参保人群中癌症死亡率的实验室检查和生物特征风险因素进行广泛的多变量分析。
在拥有完整实验室检查和身体测量资料的申请人中，有125万人可供研究，并平均随访4.7年。记录了518起因癌症死亡导致的人寿保险理赔日期，并通过多变量Cox比例风险回归分析所得数据集，以确定参保人群中与癌症相关死亡率具有统计学显著意义的预测因素。
在人口统计学变量中，发现癌症死亡与年龄和吸烟密切相关，与性别无关。在血清和尿液分析物中，肝功能检查（主要是γ-谷氨酰转移酶和碱性磷酸酶）、低胆固醇血症、蛋白尿和低果糖胺被发现可独立预测癌症死亡率。在身体测量变量中，癌症死亡风险与身高呈正相关，与脉搏和血压的关系相对较弱。体重和体重指数（BMI）不是具有统计学显著意义的协变量。
这些发现突出了实验室分析物和生物特征测量在癌症风险评估中的潜在价值，包括分析物中低至低正常水平的值（特别是胆固醇和果糖胺），其在临床实践和承保中的诊断价值可能具有优势。
