Institute of Occupational Medicine, Research Avenue North, Riccarton, Edinburgh, EH14 4AP, United Kingdom.
The Royal Zoological Society of Scotland RZSS Edinburgh Zoo, Edinburgh, EH12 6TS United Kingdom.
Ann Work Expo Health. 2017 Oct 1;61(8):939-953. doi: 10.1093/annweh/wxx074.
When applying simple screening (Tier 1) tools to estimate exposure to chemicals in a given exposure situation under the Registration, Evaluation, Authorisation and restriction of CHemicals Regulation 2006 (REACH), users must select from several possible input parameters. Previous studies have suggested that results from exposure assessments using expert judgement and from the use of modelling tools can vary considerably between assessors. This study aimed to investigate the between-user reliability of Tier 1 tools. A remote-completion exercise and in person workshop were used to identify and evaluate tool parameters and factors such as user demographics that may be potentially associated with between-user variability. Participants (N = 146) generated dermal and inhalation exposure estimates (N = 4066) from specified workplace descriptions ('exposure situations') and Tier 1 tool combinations (N = 20). Interactions between users, tools, and situations were investigated and described. Systematic variation associated with individual users was minor compared with random between-user variation. Although variation was observed between choices made for the majority of input parameters, differing choices of Process Category ('PROC') code/activity descriptor and dustiness level impacted most on the resultant exposure estimates. Exposure estimates ranging over several orders of magnitude were generated for the same exposure situation by different tool users. Such unpredictable between-user variation will reduce consistency within REACH processes and could result in under-estimation or overestimation of exposure, risking worker ill-health or the implementation of unnecessary risk controls, respectively. Implementation of additional support and quality control systems for all tool users is needed to reduce between-assessor variation and so ensure both the protection of worker health and avoidance of unnecessary business risk management expenditure.
当在 2006 年化学品注册、评估、授权和限制法规（REACH）下应用简单筛选（第 1 层）工具来估计特定暴露情况下的化学品暴露时，用户必须从几个可能的输入参数中进行选择。先前的研究表明，使用专家判断和建模工具进行暴露评估的结果在评估者之间可能有很大差异。本研究旨在调查第 1 层工具的用户间可靠性。远程完成练习和现场研讨会用于确定和评估工具参数以及可能与用户间变异性相关的用户人口统计学等因素。参与者（N=146）根据指定的工作场所描述（“暴露情况”）和第 1 层工具组合（N=20）生成皮肤和吸入暴露估计值（N=4066）。研究并描述了用户、工具和情况之间的相互作用。与个体用户相关的系统变化与随机用户间变化相比微不足道。尽管观察到大多数输入参数的选择存在差异，但过程类别（“PROC”）代码/活动描述符和粉尘水平的不同选择对最终暴露估计值的影响最大。不同工具用户对同一暴露情况生成了几个数量级的暴露估计值。这种不可预测的用户间变化将降低 REACH 流程的一致性，并可能导致暴露被低估或高估，从而分别导致工人健康受损或不必要的风险控制措施的实施。需要为所有工具用户实施额外的支持和质量控制系统，以减少评估者之间的差异，从而确保工人的健康得到保护并避免不必要的业务风险管理支出。
