Issues in multiple imputation of missing data for large general practice clinical databases

被引:87
作者
Marston, Louise [1 ]
Carpenter, James R. [2 ]
Walters, Kate R. [1 ]
Morris, Richard W. [1 ]
Nazareth, Irwin [1 ,3 ]
Petersen, Irene [1 ]
机构
[1] UCL, Dept Primary Care & Populat Hlth, London NW3 2PF, England
[2] London Sch Hyg & Trop Med, Med Stat Unit, London WC1E 7HT, England
[3] MRC Gen Practice Res Framework, London NW1 2ND, England
基金
英国经济与社会研究理事会; 英国医学研究理事会;
关键词
clinical databases; missing data; multiple imputation; primary care databases; CORONARY-HEART-DISEASE; MYOCARDIAL-INFARCTION; BLOOD-PRESSURE; MORTALITY; RISK; METAANALYSIS; SMOKING; COHORT; ADULTS; WOMEN;
D O I
10.1002/pds.1934
中图分类号
R1 [预防医学、卫生学];
学科分类号
1004 ; 120402 ;
摘要
Purpose Missing data are a substantial problem in clinical databases. This paper aims to examine patterns of missing data in a primary care database, compare this to nationally representative datasets and explore the use of multiple imputation (MI) for these data. Methods The patterns and extent of missing health indicators in a UK primary care database (THIN) were quantified using 488 384 patients aged 16 or over in their first year after registration with a GP from 354 General Practices. MI models were developed and the resulting data compared to that from nationally representative datasets (14 142 participants aged 16 or over from the Health Survey for England 2006 (HSE) and 4 252 men from the British Regional Heart Study (BRHS)). Results Between 22% (smoking) and 38% (height) of health indicator data were missing in newly registered patients, 2004-2006. Distributions of height, weight and blood pressure were comparable to HSE and BRHS, but alcohol and smoking were not. After MI the percentage of smokers and non-drinkers was higher in THIN than the comparison datasets, while the percentage of ex-smokers and heavy drinkers was lower. Height, weight and blood pressure remained similar to the comparison datasets. Conclusions Given available data, the results are consistent with smoking and alcohol data missing not at random whereas height, weight and blood pressure missing at random. Further research is required on suitable imputation methods for smoking and alcohol in such databases. Copyright (C) 2010 John Wiley & Sons, Ltd.
引用
收藏
页码:618 / 626
页数:9
相关论文
共 35 条
[1]  
[Anonymous], DRINKING ADULTS BEHA
[2]  
[Anonymous], 1988, HLTH DEPRIVATION INE
[3]  
[Anonymous], 2007, Stata statistical software
[4]  
Bourke Alison, 2004, Inform Prim Care, V12, P171
[5]  
Carpenter J., 2008, MISSING DATA
[6]  
Carpenter J, 2007, Missing data in randomised controlled trials: a practical guide
[7]   Low identification of alcohol use disorders in general practice in England [J].
Cheeta, Survjit ;
Drummond, Colin ;
Oyefeso, Adenekan ;
Phillips, Tom ;
Deluca, Paolo ;
Perryman, Katherine ;
Coulton, Simon .
ADDICTION, 2008, 103 (05) :766-773
[8]   Contribution of changes in incidence and mortality to trends in the prevalence of coronary heart disease in the UK: 1996-2005 [J].
Davies, Alisha Ruth ;
Smeeth, Liam ;
Grundy, Emily Marjatta Dorothea .
EUROPEAN HEART JOURNAL, 2007, 28 (17) :2142-2147
[9]   Lifestyle variables and the risk of myocardial infarction in the General Practice Research Database [J].
Delaney J.A.C. ;
Daskalopoulou S.S. ;
Brophy J.M. ;
Steele R.J. ;
Opatrny L. ;
Suissa S. .
BMC Cardiovascular Disorders, 7 (1)
[10]   Validating the effects of drug treatment on blood pressure in the General Practice Research Database [J].
Delaney, Joseph A. C. ;
Moodie, Erica E. M. ;
Suissa, Sarny .
PHARMACOEPIDEMIOLOGY AND DRUG SAFETY, 2008, 17 (06) :535-545