A Generalized Estimating Equations Approach for Spatially Correlated Binary Data: Applications to the Analysis of Neuroimaging Data

Biometrics ◽  
1995 ◽  
Vol 51 (2) ◽  
pp. 627 ◽  
Author(s):  
Paul S. Albert ◽  
Lisa M. McShane
Keyword(s):  
Generalized Estimating Equations ◽  
Binary Data ◽  
Estimating Equations ◽  
Correlated Binary Data ◽  
Spatially Correlated ◽  
Neuroimaging Data ◽  
Generalized Estimating

scholarly journals Time effects of bacterial vaginosis on infant morbidities in Kenya assessed using modified skewed generalized estimating equations

2020 ◽  
Author(s):  
Ngugi Mwenda ◽  
Ruth Nduati ◽  
Mathew Kosgey ◽  
Gregory Kerich
Keyword(s):  
Bacterial Vaginosis ◽  
Generalized Estimating Equations ◽  
Binary Data ◽  
Estimating Equations ◽  
Health Conditions ◽  
Positive Coefficient ◽  
Time Effects ◽  
Infant Morbidity ◽  
Multiple Health ◽  
Generalized Estimating

Abstract Infant morbidity and mortality are indicators used globally as measures of a country’s health status. Among the 8 millennium development goals (MDGs), this study aimed to address goal four (MDG 4) on the reduction of child mortality and six (MDG 6) on combating HIV and other diseases. We assessed different health conditions caused by bacterial vaginosis (BV) that could have life-long effects among infants. We aimed to address the time effects of BV on the long-term cause of infants' morbidities when asymmetry is assumed. We analyzed infant data from HIV-positive mothers with known BV status from a randomized controlled trial study conducted in Nairobi, Kenya. We aimed to investigate the effect of BV on infant morbidity with time from birth up to the age of 6 months. We derived a score for morbidity incidences depending on illnesses reported in the register during scheduled visits only. By adjusting for the mother’s BV status, child’s HIV status, sex, feeding status, and weight for age, we used two approaches for analysis. We considered and fitted the traditional generalized estimating (GEE) equations and our proposed skewed generalized estimating equations (SGEE). Overall, we included information on 327 infants. One thousand nine hundred sixty-two repeated measurements were available for analysis. Among the 327 mothers, 148 (45%) tested positive for BV, while 179 (55%) tested negative. We found that BV, gender, and time were associated with multiple health conditions in infants. Infants of women who tested positive for BV, at month 1, had 4.46 higher odds of various health conditions compared to infants of mothers who tested negative. The effects of BV tended to decrease with time, and at 5 months of age, children in the BV group had 1.10 times the odds of experiencing morbidity incidence. In the SGEE model, BV was statistically significant at the 0.05 level with a positive coefficient, indicating that children in the BV group had a higher probability of experiencing multiple morbidities. BV is a significant predictor of infant morbidity because its effects on exposed infants could persist over time. In contrast, the traditional GEE results showed an insignificant positive coefficient. The results indicate the need to factor in the skewness during analysis in case of data transformation, especially when converting from continuous to binary data for parsimony and straightforward interpretation of the effects of covariates. Maternal BV status was positively associated with morbidity incidences, which highlights the need for early intervention for infected women. Accelerated programs promoting access to BV treatment with proper infant handling practices that better deal with emerging multiple health conditions in infants may prove useful in reducing the incidence of infant morbidity in Kenya. Emphasis on care to promote better health for infants during growth is necessary to achieve the MDGs.

scholarly journals Penalized generalized estimating equations for relative risk regression with applications to brain lesion data

2021 ◽  
Author(s):  
Petya Kindalova ◽  
Michele Veldsman ◽  
Thomas E Nichols ◽  
Ioannis Kosmidis
Keyword(s):  
Relative Risk ◽  
Generalized Estimating Equations ◽  
Binary Data ◽  
Large Scale ◽  
Brain Lesion ◽  
Estimating Equations ◽  
Variance Function ◽  
Parameter Estimates ◽  
Data Set ◽  
Generalized Estimating

Motivated by a brain lesion application, we introduce penalized generalized estimating equations for relative risk regression for modelling correlated binary data. Brain lesions can have varying incidence across the brain and result in both rare and high incidence outcomes. As a result, odds ratios estimated from generalized estimating equations with logistic regression structures are not necessarily directly interpretable as relative risks. On the other hand, use of log-link regression structures with the binomial variance function may lead to estimation instabilities when event probabilities are close to 1. To circumvent such issues, we use generalized estimating equations with log-link regression structures with identity variance function and unknown dispersion parameter. Even in this setting, parameter estimates can be infinite, which we address by penalizing the generalized estimating functions with the gradient of the Jeffreys prior. Our findings from extensive simulation studies show significant improvement over the standard log-link generalized estimating equations by providing finite estimates and achieving convergence when boundary estimates occur. The real data application on UK Biobank brain lesion maps further reveals the instabilities of the standard log-link generalized estimating equations for a large-scale data set and demonstrates the clear interpretation of relative risk in clinical applications.

Sign in / Sign up

Export Citation Format

Share Document