Djalalov Sandjar, Rabeneck Linda, Tomlinson George, Bremner Karen E, Hilsden Robert, Hoch Jeffrey S
Centre for Excellence in Economic Analysis Research, Li Ka Shing Knowledge Institute, St. Michael's Hospital, Toronto, ON, Canada (SD, JSH)
Pharmacoeconomics Research Unit, Cancer Care Ontario, Toronto, ON, Canada (SD, LR, JSH)
Med Decis Making. 2014 Aug;34(6):809-18. doi: 10.1177/0272989X14536779. Epub 2014 Jun 5.
To perform a systematic review of utility weights for colorectal cancer (CRC) health states reported in the scientific literature and to determine the effects of disease factors, patient characteristics, and utility methods on utility values.
We identified 26 articles written in English and published from January 1980 to January 2013, providing 351 unique utilities for CRC health states elicited from 6546 unique respondents. The CRC utility data were analyzed using linear mixed-effects models with CRC type, stage, time to or from initial care, utility measurement instrument, and administration method as independent variables.
In the base case model, the estimated utility for a patient with stage I to III CRC more than 1 year after surgery, rated using a self-administered time tradeoff instrument, was 0.90. Stage, time to or from initial care, and utility measurement instrument were associated with statistically significant utility differences ranging from -0.19 to 0.02. Utilities for patients with stage IV cancer were 0.19 lower (P < 0.001) than for those with stage I to III cancer. Utilities elicited at more than 1 year after surgery were 0.05 higher than those elicited at 3 months after surgery (P = 0.008). Estimates of differences between utility measurement instruments were sensitive to how repeated scores in the same patient group were treated, and other findings were sensitive to how the disease stage was modeled and method of administration.
Variations in reported utilities are associated with factors such as cancer stage, time to or from initial care, and utility measurement instrument. More research is needed to study why apparently similar patients report different quality of life.
对科学文献中报道的结直肠癌（CRC）健康状态的效用权重进行系统评价，并确定疾病因素、患者特征和效用方法对效用值的影响。
我们检索了1980年1月至2013年1月期间发表的26篇英文文章，这些文章提供了来自6546名独特受访者的351个针对CRC健康状态的独特效用值。使用线性混合效应模型对CRC效用数据进行分析，将CRC类型、分期、初次治疗时间、效用测量工具和给药方法作为自变量。
在基础病例模型中，使用自我管理的时间权衡工具对术后1年以上的I至III期CRC患者的估计效用值为0.90。分期、初次治疗时间和效用测量工具与具有统计学意义的效用差异相关，差异范围为-0.19至0.02。IV期癌症患者的效用值比I至III期癌症患者低0.19（P < 0.001）。术后1年以上得出的效用值比术后3个月得出的效用值高0.05（P = 0.008）。效用测量工具之间差异的估计对同一患者组中重复得分的处理方式敏感，其他结果对疾病分期的建模方式和给药方法敏感。
报告的效用值差异与癌症分期、初次治疗时间和效用测量工具等因素有关。需要更多研究来探讨为什么明显相似的患者报告的生活质量不同。
