Bender B G, Iklé D N, DuHamel T, Tinkelman D
Department of Pediatrics, National Jewish Center for Immunology and Respiratory Medicine, Denver, CO 80206, USA.
J Allergy Clin Immunol. 1997 Feb;99(2):197-203. doi: 10.1016/s0091-6749(97)70096-4.
Prevention of study patient attrition and assessment of its impact on outcome data are problems that receive little attention despite their obvious importance in asthma research.
The medical, demographic, and psychologic characteristics of asthmatic children and adults who dropped out of a yearlong medication trial were assessed to determine whether this group differed from those who completed the study, potentially introducing bias into the data set and interfering with completion of the study's objectives.
Profiles of 362 adult and pediatric asthmatic patient dropouts from the multicenter trial were contrasted with profiles of those who completed the study. Despite a 1-month prerandomization screening, 24% of patients failed to complete the trial for varied reasons, which largely included noncompliance and treatment dissatisfaction.
Although attrition rates were equal among adults and children, dropout-completer differentiation was not. Adult completers did not differ from dropouts in any variables. However, pediatric dropouts were more likely than completers to be female (67% and 36%, p = 0.008) and to have more reactive airways (PD20, 2.29 +/- 1.32 and 5.2 +/- 1.23, p = 0.05), to have reduced scores on tests of intelligence (Full Scale IQ, 102.2 +/- 2.6 and 112.5 +/- 1.6, p = 0.002) and problem solving (Wisconsin Card Sorting Test Error Scores, 39.8 +/- 4.1 and 29.1 +/- 2.0, p = 0.01), and to have increased behavioral problems (Child Behavior Checklist Total Problem Score, 60.7 +/- 2.5 and 53.6 +/- 1.1, p = 0.003).
These findings demonstrate the potential of patient attrition to bias outcome in clinical trials and underscore the necessity of: (1) preventing its occurrence, (2) correctly assessing its causes, and (3) determining its ultimate impact on study results. Strategies for each of these three tasks should be implemented at the study's initial planning stages.
尽管研究对象流失的预防及其对结果数据的影响在哮喘研究中具有明显的重要性，但这些问题很少受到关注。
对退出为期一年药物试验的哮喘儿童和成人的医学、人口统计学和心理特征进行评估，以确定该组与完成研究的人群是否存在差异，这可能会给数据集引入偏差并干扰研究目标的完成。
将多中心试验中362名成人和儿童哮喘患者退出者的资料与完成研究的患者资料进行对比。尽管在随机分组前进行了1个月的筛查，但仍有24%的患者因各种原因未能完成试验，主要原因包括不依从和对治疗不满意。
虽然成人和儿童的流失率相同，但退出者与完成者的差异并不相同。成人完成者在任何变量上与退出者均无差异。然而，儿童退出者比完成者更有可能为女性（分别为67%和36%，p = 0.008），气道反应性更高（PD20，2.29±1.32和5.2±1.23，p = 0.05），智力测试得分降低（全量表智商，102.2±2.6和112.5±1.6，p = 0.002），解决问题能力降低（威斯康星卡片分类测试错误得分，39.8±4.1和29.1±2.0，p = 0.01），行为问题增加（儿童行为检查表总问题得分，60.7±2.5和53.6±1.1，p = 0.003）。
这些发现证明了研究对象流失在临床试验中使结果产生偏差的可能性，并强调了以下几点的必要性：（1）防止其发生；（2）正确评估其原因；（3）确定其对研究结果的最终影响。这三项任务的策略应在研究的初始规划阶段实施。
