Woodard Stacy, Archer Linda, Zell Elizabeth, Ronveaux Olivier, Birmingham Maureen
Department of Biostatistic Quintiles Inc., Research Triangle Park, NC, USA.
Ann Epidemiol. 2007 Aug;17(8):628-33. doi: 10.1016/j.annepidem.2007.01.038. Epub 2007 Jun 6.
The goal of the Data Quality Audit (DQA) is to assess whether the Global Alliance for Vaccines and Immunization-funded countries are adequately reporting the number of diphtheria-tetanus-pertussis immunizations given, on which the "shares" are awarded. Given that this sampling design is a modified two-stage cluster sample (modified because a stratified, rather than a simple, random sample of health facilities is obtained from the selected clusters); the formula for the calculation of the standard error for the estimate is unknown. An approximated standard error has been proposed, and the first goal of this simulation is to assess the accuracy of the standard error. Results from the simulations based on hypothetical populations were found not to be representative of the actual DQAs that were conducted. Additional simulations were then conducted on the actual DQA data to better access the precision of the DQ with both the original and the increased sample sizes.
数据质量审计(DQA)的目标是评估由全球疫苗免疫联盟资助的国家是否充分报告了白喉-破伤风-百日咳免疫接种的数量,而“份额”正是基于这些数据授予的。鉴于这种抽样设计是一种改良的两阶段整群抽样(之所以说是改良,是因为从选定的群组中获得的是卫生设施的分层随机样本,而非简单随机样本);用于计算估计值标准误差的公式尚不清楚。有人提出了一个近似标准误差,本模拟的首要目标是评估该标准误差的准确性。基于假设总体进行的模拟结果被发现不能代表实际开展的DQA。随后对实际DQA数据进行了额外模拟,以更好地评估在原始样本量和增加样本量情况下DQA的精度。