Nieri Michele, Clauser Carlo, Franceschi Debora, Pagliaro Umberto, Saletta Daniele, Pini-Prato Giovanpaolo
Dental School, University of Florence, Florence, Italy.
Clin Oral Implants Res. 2007 Aug;18(4):419-31. doi: 10.1111/j.1600-0501.2007.01350.x. Epub 2007 May 21.
The aim of the present study was to investigate the relationships among reported methodological, statistical, clinical and paratextual variables of randomized clinical trials (RCTs) in implant therapy, and their influence on subsequent research.
The material consisted of the RCTs in implant therapy published through the end of the year 2000. Methodological, statistical, clinical and paratextual features of the articles were assessed and recorded. The perceived clinical relevance was subjectively evaluated by an experienced clinician on anonymous abstracts. The impact on research was measured by the number of citations found in the Science Citation Index. A new statistical technique (Structural learning of Bayesian Networks) was used to assess the relationships among the considered variables.
Descriptive statistics revealed that the reported methodology and statistics of RCTs in implant therapy were defective. Follow-up of the studies was generally short. The perceived clinical relevance appeared to be associated with the objectives of the studies and with the number of published images in the original articles. The impact on research was related to the nationality of the involved institutions and to the number of published images.
RCTs in implant therapy (until 2000) show important methodological and statistical flaws and may not be appropriate for guiding clinicians in their practice. The methodological and statistical quality of the studies did not appear to affect their impact on practice and research. Bayesian Networks suggest new and unexpected relationships among the methodological, statistical, clinical and paratextual features of RCTs.
本研究旨在调查种植治疗随机临床试验（RCT）中所报告的方法学、统计学、临床和辅助文本变量之间的关系，以及它们对后续研究的影响。
材料包括截至2000年底发表的种植治疗RCT。对文章的方法学、统计学、临床和辅助文本特征进行评估并记录。由一位经验丰富的临床医生根据匿名摘要主观评估感知到的临床相关性。通过科学引文索引中发现的引用次数来衡量对研究的影响。使用一种新的统计技术（贝叶斯网络的结构学习）来评估所考虑变量之间的关系。
描述性统计显示，种植治疗RCT所报告的方法学和统计学存在缺陷。研究的随访通常较短。感知到的临床相关性似乎与研究目的以及原始文章中发表的图像数量有关。对研究的影响与参与机构的国籍以及发表的图像数量有关。
种植治疗的RCT（截至2000年）显示出重要的方法学和统计学缺陷，可能不适用于指导临床医生的实践。研究的方法学和统计学质量似乎并未影响其对实践和研究的影响。贝叶斯网络揭示了RCT的方法学、统计学、临床和辅助文本特征之间新的意外关系。
