摘要
目的结合肝炎复发次数实例,选取负二项回归模型拟合数据,为计数资料的正确分析提供理论依据。方法对山西省太原市传染病医院2007年6月至12月乙型肝炎、丙型肝炎患者复发情况的随访调查资料,通过SAS软件GENMOD(广义线性模型)过程拟合负二项回归模型,对肝炎复发的影响因素进行分析。结果年龄越大、受感染时间越长、治疗信心越缺乏、认为收入对治疗的影响越明显的肝炎患者复发次数越多。结论SAS/GENMOD过程拟合负二项回归模型,程序简便,结果输出规范,可提供常用的连接函数和概率分布,结果可更准确地解释肝炎复发次数的情况。
Objective We used a negative binomial regression model to explore its goodness of fit with the real-world relapses of hepatitis, aiming at providing scientific evidences for analysis of numeric count data. Methods The follow-up data of patients with hepatitis B and hepatitis C between June and December in 2007, delivered by a single physician from a Taiyuan infection hospital, was analyzed using the GENMOD （generalized linear model） process of SAS to fit the negative binomial regression and to look for influencing factors of hepatitis relapse. Results Relapses of hepatitis increased with advanced age, long duration of the disease, poor confidence with treatment and the strong belief that financial factors are related with outcomes. Couclusion The process of SAS/GENMOD based on fitness with negative binomial regresions provided simple and formatted outputs covering common link functions and probability distributions. The results may be useful to explain the relapses of hepatitis.
出处
《中国药物与临床》
CAS
2008年第10期782-784,共3页
Chinese Remedies & Clinics
基金
山西省自然科学基金资助项目(2007011087)
山西省科技攻关基金资助项目(051100-11)
关键词
负二项回归
线性模型
计数资料
SAS程序
Negative binomial regression
Linear models
Count data
SAS procedure
