Testa M A, Simonson D C, Turner R R
Department of Biostatistics, Harvard School of Public Health, Boston, MA 02115, USA.
Diabetes Care. 1998 Dec;21 Suppl 3:C44-52. doi: 10.2337/diacare.21.3.c44.
Outcomes research is used increasingly for assessing the health economic benefits of new therapeutic programs and interventions. The measurement properties of the outcomes assessment tools are important. If overlooked, they can mislead health care administrators and caregivers regarding the importance and value of these programs and interventions. We reviewed the literature and conducted two analyses to determine the absolute, relative, and operative quality-of-life ranges for people with type 2 diabetes. Quality of life and fasting blood glucose and HbA1c concentrations were measured at baseline and at 4, 8, and 12 weeks of treatment in 569 men and women randomized to either glipizide gastrointestinal therapeutic system (GITS) or placebo in a double-blind, multicenter clinical trial. A subgroup of 290 patients completed a diabetes-specific health states questionnaire at endpoint (week 12 or early termination) rating 10 health-state descriptions on a health thermometer scale ranging from 0 (death) to 100 (full health). Health losses at the higher end of the scale had a greater negative utility than did comparable losses at lower health states, indicating patients' strong preferences for maintaining asymptomatic or mildly symptomatic conditions. Patients rated their current health state at 83.4 +/- 0.8% of full health and indicated that a loss of 27 points below this value would prevent them from living and working as they currently do. The calibration analysis applied to the quality-of-life scales suggested that the targeted range for clinical investigation and quality-of-care evaluation must be more narrowly focused. Effect sizes as seemingly small as 2% (0.25 responsiveness units) on the absolute scale can correspond to quality-of-life losses of 15-20% on the personal operative scale. Differences in glycemic control clearly affected quality of life. Those patients with the best HbA1c responses (decreasing 1.5% or more from baseline) versus those with the worst responses (increasing 1.5% or more from baseline) were separated by 0.6 responsiveness units for the overall quality-of-life summary measure. The calibration analysis suggested that this degree of better glycemic control provides a nearly 50% gain in quality of life according to personal expectations within the operative range. In conclusion, general measures of quality of life may be too crude and insensitive to capture the important gains in health outcomes due to new therapeutic interventions and programs in diabetes. Quality-of-care evaluations for diabetes are at risk of favoring inferior programs with lower costs simply because gains or losses in health outcomes go undetected.
结果研究越来越多地用于评估新治疗方案和干预措施的健康经济效益。结果评估工具的测量属性很重要。如果被忽视，它们可能会在这些方案和干预措施的重要性和价值方面误导医疗保健管理人员和护理人员。我们回顾了文献并进行了两项分析，以确定2型糖尿病患者的绝对、相对和实际生活质量范围。在一项双盲、多中心临床试验中，对569名随机分配接受格列吡嗪胃肠治疗系统（GITS）或安慰剂治疗的男性和女性，在基线以及治疗的第4、8和12周测量了生活质量、空腹血糖和糖化血红蛋白浓度。290名患者的一个亚组在终点（第12周或提前终止）完成了一份针对糖尿病的健康状况问卷，在从0（死亡）到100（完全健康）的健康温度计量表上对10种健康状况描述进行评分。量表高端的健康损失比低健康状态下的可比损失具有更大的负效用，这表明患者强烈倾向于维持无症状或轻度症状的状况。患者将他们当前的健康状态评为完全健康的83.4±0.8%，并表示低于该值27分的损失将使他们无法像目前这样生活和工作。应用于生活质量量表的校准分析表明，临床研究和医疗质量评估的目标范围必须更窄。在绝对量表上看似小至2%（0.25个反应单位）的效应大小，在个人实际量表上可能对应15 - 20%的生活质量损失。血糖控制的差异明显影响生活质量。糖化血红蛋白反应最佳（比基线降低1.5%或更多）的患者与反应最差（比基线升高1.5%或更多）的患者，在总体生活质量综合测量中相差0.6个反应单位。校准分析表明，根据个人在实际范围内的期望，这种程度的更好血糖控制可使生活质量提高近50%。总之，一般的生活质量测量可能过于粗略和不敏感，无法捕捉糖尿病新治疗干预措施和方案所带来的健康结果的重要改善。糖尿病的医疗质量评估有偏向成本较低但效果较差的方案的风险，仅仅是因为健康结果的改善或损失未被发现。
