Division of Cancer Control and Population Sciences, National Cancer Institute, Rockville, Maryland.
Office of Strategic Planning, Legislation, and Scientific Policy, National Institute of Minority Health and Health Disparities, Bethesda, Maryland.
Cancer Epidemiol Biomarkers Prev. 2017 Nov;26(11):1611-1618. doi: 10.1158/1055-9965.EPI-17-0213. Epub 2017 Sep 8.
Using the National Health Interview Survey (NHIS), we examined the effect of question wording on estimates of past-year mammography among racially/ethnically diverse women ages 40-49 and 50-74 without a history of breast cancer. Data from one-part ("Have you had a mammogram during the past 12 months?") and two-part ("Have you ever had a mammogram"; "When did you have your most recent mammogram?") mammography history questions administered in the 2008, 2011, and 2013 NHIS were analyzed. χ tests provided estimates of changes in mammography when question wording was either the same (two-part question) or differed (two-part question followed by one-part question) in the two survey years compared. Crosstabulations and regression models assessed the type, extent, and correlates of inconsistent responses to the two questions in 2013. Reports of past-year mammography were slightly higher in years when the one-part question was asked than when the two-part question was asked. Nearly 10% of women provided inconsistent responses to the two questions asked in 2013. Black women ages 50 to 74 [adjusted OR (aOR), 1.50; 95% confidence interval (CI), 1.16-1.93] and women ages 40-49 in poor health (aOR, 2.22; 95% CI, 1.09-4.52) had higher odds of inconsistent responses; women without a usual source of care had lower odds (40-49: aOR, 0.42; 95% CI, 0.21-0.85; 50-74: aOR, 0.42; 95% CI, 0.24-0.74). Self-reports of mammography are sensitive to question wording. Researchers should use equivalent questions that have been designed to minimize response biases such as telescoping and social desirability. Trend analyses relying on differently worded questions may be misleading and conceal disparities. .
利用国家健康访谈调查（NHIS），我们研究了问题措辞对种族/民族多样化的 40-49 岁和 50-74 岁、无乳腺癌病史的女性过去一年乳房 X 光检查率的影响。分析了 2008 年、2011 年和 2013 年 NHIS 中使用的一份式（“在过去 12 个月中，您进行过乳房 X 光检查吗？”）和两份式（“您是否进行过乳房 X 光检查？”“您最近一次进行乳房 X 光检查是什么时候？”）乳房 X 光检查史问题的数据。 χ 检验提供了在两个调查年份中问题措辞相同时（两份式问题）或不同时（两份式问题后接一份式问题）时乳房 X 光检查率变化的估计。交叉表和回归模型评估了 2013 年对两个问题不一致回答的类型、程度和相关性。在询问一份式问题的年份，过去一年进行乳房 X 光检查的报告略高于询问两份式问题的年份。大约 10%的女性对 2013 年询问的两个问题给出了不一致的回答。50-74 岁的黑人女性（调整后的比值比[aOR]，1.50；95%置信区间[CI]，1.16-1.93）和健康状况较差的 40-49 岁女性（aOR，2.22；95% CI，1.09-4.52），出现不一致回答的可能性更高；没有常规医疗服务来源的女性可能性更低（40-49 岁：aOR，0.42；95% CI，0.21-0.85；50-74 岁：aOR，0.42；95% CI，0.24-0.74）。乳房 X 光检查的自我报告对问题措辞敏感。研究人员应使用经过设计的等效问题，以尽量减少响应偏差，如 telescoping 和社会期望。依赖措辞不同的问题的趋势分析可能具有误导性，并掩盖差异。
