Olson S H
Memorial Sloan-Kettering Cancer Center, Box 44, 1275 York Avenue, New York, NY 10021, USA.
Am J Epidemiol. 2001 Sep 15;154(6):574-81. doi: 10.1093/aje/154.6.574.
There is concern that participation in case-control studies has declined. To address this question, the author and colleagues reviewed data from 82 US and Canadian case-control studies published in epidemiologic journals during two periods: 1988-1990 and 1997-January 1999. The median year of data collection, which ranged from 1972 to 1996, was the primary independent variable. Reported response among both cases and controls remained constant over this time period. The regression coefficients (beta) were small: For each year, there was a change of 0.15 percentage points for cases (p = 0.56) and -0.16 percentage points for controls (p = 0.54). Possible confounders included the location where the study had been conducted and, for cases, the disease under study (cancer vs. others). After adjustment for these factors in case groups, there was still no association between year of data collection and response: For each year, there was a change of -0.20 percentage points (p = 0.43). After adjustment of results for study location among controls, there was a moderate decline over time: For each year, there was a change of -0.44 percentage points (p = 0.12). Overall, reported response in more recent studies was similar to that in earlier studies; for control groups, this may reflect changes in locations where the studies were conducted.
有人担心参与病例对照研究的情况有所下降。为了解决这个问题，作者及其同事回顾了82项美国和加拿大病例对照研究的数据，这些研究发表在流行病学杂志的两个时期：1988 - 1990年和1997年 - 1999年1月。数据收集的中位年份（范围从1972年到1996年）是主要的自变量。在这个时间段内，病例组和对照组报告的应答率保持不变。回归系数（β）很小：每年，病例组的变化为0.15个百分点（p = 0.56），对照组为 - 0.16个百分点（p = 0.54）。可能的混杂因素包括研究开展的地点，对于病例组而言，还包括所研究的疾病（癌症与其他疾病）。在病例组中对这些因素进行调整后，数据收集年份与应答率之间仍然没有关联：每年的变化为 - 0.20个百分点（p = 0.43）。在对照组中对研究地点的结果进行调整后，随着时间推移有适度下降：每年的变化为 - 0.44个百分点（p = 0.12）。总体而言，近期研究报告的应答率与早期研究相似；对于对照组来说，这可能反映了研究开展地点的变化。
