A new direction toward dose response analysis employing benchmark dose (BMD) method

DOI

抄録

<p>A number of scientific approaches using dose-response experimental data have been used. As an alternative method to classical observational method including the use of no observable adverse effect level (NOAEL), the formality of benchmark dose (BMD) method has been elaborated. The BMD method determines the threshold dose by fitting various statistical models to the dose-response curve, which addresses the problems surrounding the use of NOAEL because it can account for the response data across different doses and can help in objectively calculating the point of departure. The benchmark dose lower bound (BMDL), which is the lower (one-sided) limit of the 95% confidence interval of BMD, can yield a point of departure that is comparable to that based on NOAEL. To employ the BMD method, it is critical to select the best performing BMDL by following the statistical fitting procedures of multiple mathematical models. Parameterized models only characterize reality, so multiple models (usually nine or more) are commonly fitted to the same experimental dataset. As a result, many BMDL values can act as the candidate of preferred reference dose. However, the reference dose should be the best performing BMDL and it must be selected, for example, as the one that gives the best fitting results. There are two additional issues in selecting or determining the BMDL. First, the BMD method uses a specified percentile point (e.g. 10% of the benchmark response, abbreviated as BMD10) as the threshold for the reference value, but the 10% percentile point is never strictly objective. Second, some fitted models (e.g. the Weibull model) yield different parameter estimates when restrictions to the range of parameters are imposed in advance of the inference procedure. In this symposium, technical issues and corresponding guidelines are comprehensively reviewed and presented.</p>

収録刊行物

詳細情報 詳細情報について

  • CRID
    1390567172568440192
  • NII論文ID
    130007898472
  • DOI
    10.14869/toxpt.47.1.0_s12-3
  • 本文言語コード
    ja
  • データソース種別
    • JaLC
    • CiNii Articles
  • 抄録ライセンスフラグ
    使用不可

問題の指摘

ページトップへ