Shojaee Sajad, Zayeri Farid, Nasserinejad Maryam, Ghasemzadeh Ali, Sadat Beheshti Shirazi Saeedeh, Khodadoostan Mahsa
Gastroenterology and Liver Diseases Research Center, Research Institute for Gastroenterology and Liver Diseases, Shahid Beheshti University of Medical Sciences, Tehran, Iran.
Department of Biostatistics, Faculty of Paramedical Sciences, Shahid Beheshti University of Medical Science, Tehran, Iran.
Gastroenterol Hepatol Bed Bench. 2019;12(Suppl1):S136-S144.
The aim of this research was to estimate the changing rate of odds ratio (OR) by varying degrees of hepatitis B virus (HBV) underreporting.
Data registering is usually associated with extensive errors such as misclassification, under-reporting, missing data due to lack of co-operation, error prone factors, and in medical studies, inadequate diagnosis of physicians or low accuracy of laboratory tests. In the present study, which discuss the actual impact of vaccination on HBV prevention, exposure and response were prone to various errors. Furthermore, some people in the community are possibly infected to the virus while were not reported in the count of patients with HBV infection.
This was a case control study. Cases included patients with HBV referring to the gastroenterology and liver disease research center. The control group included patients without HBV who underwent a fatty liver test at Taleghani hospital laboratory. Bayesian approach and Gibbs sampling algorithm were used to estimate OR.
According to results, misclassification rate was mild in raw data, but with an increase in degree of underreporting for 50 and 500 of unreported cases, OR increased by about half and more than double, respectively, while sensitivity diminished strikingly.
Our analysis asserted that knowing the degree of underreporting is essential to accurately calculate OR and sensitivity. In addition, despite varying OR in different samples, overall the results were similar according to the pattern of exposure and response association.
本研究的目的是评估在不同程度的乙型肝炎病毒（HBV）漏报情况下优势比（OR）的变化率。
数据记录通常伴随着大量错误，如错误分类、漏报、因缺乏合作导致的数据缺失、易出错因素，以及在医学研究中，医生诊断不足或实验室检测准确性低。在本研究中，讨论了疫苗接种对HBV预防的实际影响，暴露和反应容易出现各种错误。此外，社区中的一些人可能感染了病毒，但在HBV感染患者计数中未被报告。
这是一项病例对照研究。病例包括转诊至胃肠病学和肝病研究中心的HBV患者。对照组包括在塔莱加尼医院实验室进行脂肪肝检测的无HBV患者。采用贝叶斯方法和吉布斯抽样算法估计OR。
根据结果，原始数据中的错误分类率较轻，但随着未报告病例数分别为50和500时漏报程度的增加，OR分别增加了约一半和两倍多，而敏感性显著降低。
我们的分析断言，了解漏报程度对于准确计算OR和敏感性至关重要。此外，尽管不同样本中的OR有所不同，但总体而言，根据暴露和反应关联模式，结果是相似的。
