Ojha Rohit P, Offutt-Powell Tabatha N, Evans Eva L, Singh Karan P
Division of Population Sciences, Dana-Farber Cancer Institute, 450 Brookline Avenue, Boston, MA 02115, USA. rohit
Arch Environ Occup Health. 2011;66(4):241-4. doi: 10.1080/19338244.2010.539641.
The objective of this study was to determine the proportion of ecologic studies published during a 20-year period regarding environmental exposures and cancer in which correlation coefficients or coefficients of determination were used as a measure of association. The authors performed a descriptive analysis of published literature by conducting a systematic review of PubMed to identify eligible ecologic studies published between 1991 and 2010. The reported measure of association was extracted for all eligible studies. During the 20-year study period, 35/105 (33%, 95% confidence limits [CL]: 25%, 43%) ecologic studies used correlation coefficients or coefficients of determination as a measure of association. These results indicate that the use of correlation coefficients and coefficients of determination as measures of association in ecologic studies of environmental exposures and cancer is relatively common, despite extensive literature discouraging their interpretation as valid measures of association.
本研究的目的是确定在20年期间发表的关于环境暴露与癌症的生态学研究中,使用相关系数或决定系数作为关联度衡量指标的研究比例。作者通过对PubMed进行系统综述,对已发表文献进行描述性分析,以识别1991年至2010年间发表的符合条件的生态学研究。提取所有符合条件研究中报告的关联度衡量指标。在20年的研究期间,35/105项(33%,95%置信区间[CL]:25%,43%)生态学研究使用相关系数或决定系数作为关联度衡量指标。这些结果表明,在环境暴露与癌症的生态学研究中,使用相关系数和决定系数作为关联度衡量指标相对普遍,尽管有大量文献不鼓励将其解释为有效的关联度衡量指标。