A modified least-squares regression approach to the estimation of risk difference

被引:123
作者
Cheung, Yin Bun [1 ]
机构
[1] London Sch Hyg & Trop Med, Dept Epidemiol & Populat Hlth, MRC Trop Hlth Grp, London, England
关键词
binomial model; least-squares analysis; regression analysis; risk; robustness; variance-covariance matrix; variance heterogeneity;
D O I
10.1093/aje/kwm223
中图分类号
R1 [预防医学、卫生学];
学科分类号
1004 ; 120402 ;
摘要
Risk ratio and risk difference are parameters of interest in many medical studies. The risk ratio has a property that the value for the outcome Y = 0 is not the inverse of the risk ratio for the outcome Y = 1. This property makes risk ratios inappropriate in some situations. Estimation of risk difference often encounters the problem that the binomial regression model fails to converge. Recently discussed alternatives may have the same problem of nonconvergence or are difficult to implement. Here the author proposes a modified least-squares regression approach- unweighted least-squares regression with a Huber-White robust standard error-for estimation of risk differences. Four versions of the robust standard error are considered. The binomial, ordinary least-squares, and modified least-squares estimators are compared analytically in a simple situation of one exposure variable. Multivariable regression analyses are simulated to demonstrate the usefulness of the approach. For sample sizes of approximately 200 or less, a small-sample version of the robust standard error is recommended. The method is illustrated with data from a patient survey in which the binomial regression fails to converge in the analyses of four out of five outcome variables.
引用
收藏
页码:1337 / 1344
页数:8
相关论文
共 22 条
[1]   Parameter estimation and goodness-of-fit in log binomial regression [J].
Blizzard, L ;
Hosmer, DW .
BIOMETRICAL JOURNAL, 2006, 48 (01) :5-22
[2]   The equivalence and difference between the English and Chinese versions of two major, cancer-specific, health-related quality-of-life questionnaires [J].
Cheung, YB ;
Thumboo, J ;
Goh, C ;
Khoo, KS ;
Che, W ;
Wee, J .
CANCER, 2004, 101 (12) :2874-2880
[3]  
Cohen J., 1988, POWERSTATISTICALSCIE, DOI 10.4324/9780203771587
[4]  
Deddens JA, 2004, AM J EPIDEMIOL, V159, P213, DOI 10.1093/aje/kwh022
[5]  
Gould W, 1999, MAXIMUM LIKELIHOOD E
[6]   Model-based estimation of relative risks and other epidemiologic measures in studies of common outcomes and in case-control studies [J].
Greenland, S .
AMERICAN JOURNAL OF EPIDEMIOLOGY, 2004, 160 (04) :301-305
[7]   INTERPRETATION AND CHOICE OF EFFECT MEASURES IN EPIDEMIOLOGIC ANALYSES [J].
GREENLAND, S .
AMERICAN JOURNAL OF EPIDEMIOLOGY, 1987, 125 (05) :761-768
[8]  
Hardin JW., 2001, GEN LINEAR MODELS EX
[9]   Using heteroscedasticity consistent standard errors in the linear regression model [J].
Long, JS ;
Ervin, LH .
AMERICAN STATISTICIAN, 2000, 54 (03) :217-224
[10]   Do English and Chinese EQ-5D versions demonstrate measurement equivalence? An exploratory study [J].
Nan Luo ;
Ling-Huo Chew ;
Kok-Yong Fong ;
Dow-Rhoon Koh ;
Swee-Cheng Ng ;
Kam-Hon Yoon ;
Sheila Vasoo ;
Shu-Chuen Li ;
Julian Thumboo .
Health and Quality of Life Outcomes, 1 (1)