Selection of Working Correlation Structure and Best Model in GEE Analyses of Longitudinal Data

Cui, James JiSheng; Qian, Guoqi

doi:10.1080/03610910701539617

Cited by 128 publications

(100 citation statements)

References 21 publications

Supporting

Mentioning

100

Contrasting

Order By: Relevance

“…On the other hand, the modified QIC and the original QIC can be used not only for the selection of the correlation structures but also for covariate selection in the GEE method. The applicability of the original QIC as a criterion for selecting the covariate was evaluated by Pan (2001a) and Cui and Qian (2007). The performance of the modified QIC for model selection should be investigated in future researches.…”

Section: Discussionmentioning

confidence: 99%

Modifications of QIC and CIC for Selecting a Working Correlation Structure in the Generalized Estimating Equation Method

2011

View full text Add to dashboard Cite

The generalized estimating equation (GEE) method is a popular method for analyzing longitudinal data. An inappropriate specification of the working correlation structure reduces the efficiency of the GEE estimation. Pan (2001a) and Hin and Wang (2009) proposed a quasi-likelihood under the independence model criterion (QIC) and a correlation information criterion (CIC) for selecting a proper working correlation structure, respectively. In this study, we proposed modifications to the QIC and CIC using the variance estimators of the GEE with improved small-sample properties. In a simulation study, the performance of the modified QIC and CIC was better than that of the original QIC and CIC. The modified methods were illustrated using the data for an air pollution study.

show abstract

Section: Discussionmentioning

confidence: 99%

Modifications of QIC and CIC for Selecting a Working Correlation Structure in the Generalized Estimating Equation Method

2011

View full text Add to dashboard Cite

show abstract

“…Linearity of numerical variables was also tested. The best-fit model was selected according to the smaller quasi-likelihood under the independence model criterion value [21] . Adjusted odds ratio (ORa) and 95% confidence interval (CI) were calculated for each of the predictors.…”

Section: Discussionmentioning

confidence: 99%

Hypnotics and Triazolobenzodiazepines - Best Predictors of High-Dose Benzodiazepine Use: Results from the Luxembourg National Health Insurance Registry

Cloos¹,

Bocquet

Rolland-Portal³

et al. 2015

Psychother Psychosom

View full text Add to dashboard Cite

Background: Benzodiazepines are not all the same concerning their risk of high-dose use. Methods: We studied benzodiazepine use from the Luxembourg national records of all insured. We calculated the 12-year prevalence from 1995 to 2007. Benzodiazepine users were divided into 3 groups, short-term with no longer than 3-month intake, intermediate with multiple administration with at least a 1-year interruption, and continuous who never stopped. A high-dose user (HDU) was defined as a patient who received a higher dose than the yearly maximum usual therapeutic dose. Results: An average of 16.0% of the adult insured population received at least 1 benzodiazepine annually, 42.9% were older than 50, 55.9% were women, and 5.4% were HDUs. We found that 32.6% were short-term users, 49.0% intermediate and 18.4% continuous. Compared to diazepam, hypnotics had higher risks for high-dose use in at least 1 age group at first-benzodiazepine intake, the risks being greater in elderly subjects and women, the highest risks being with triazolam (adjusted odds ratio = 215.85; 95% confidence interval = 133.75-348.35) in the 69- to 105-year-old group at first-benzodiazepine intake. Anxiolytics had a low risk except for alprazolam and prazepam in the 69- to 105-year-old group at first-benzodiazepine intake, clonazepam and clobazam had the lowest risk in 18- to 43-year-olds at first-benzodiazepine intake. Alprazolam had dispensed volumes increased by threefold over the 12-year period. Conclusion: All hypnotics had higher risks for high-dose use compared to diazepam in continuous users. Two anxiolytics, clonazepam and clobazam, had the lowest risks. Hypnotics and the triazolobenzodiazepines alprazolam and triazolam were most problematic. Elderly subjects and women are at greater risks.

show abstract

“…Parsimonious models were selected based on the lowest QIC score (quasi-likelihood under the independence model criteria). 33,34 Final models were selected based on the lowest QIC scores and a meaningful pattern of results across the models, as described below.…”

Section: Distress Psychological Distressmentioning

confidence: 99%

The Effect of Social Networks and Social Support on Mental Health Services Use, Following a Life Event, among the Baltimore Epidemiologic Catchment Area Cohort

Maulik

Eaton

Bradshaw

2010

J Behav Health Serv Res

View full text Add to dashboard Cite

The study examined the association between life events and mental health services use, accounting for social networks and social support. Main and stress-buffering effects were estimated using longitudinal data from the Baltimore Epidemiologic Catchment Area cohort (1,920 participants in 1993-1996, of whom 1,071 were re-interviewed in 2004-2005). Following a life event, the odds of using general medical services were increased by almost 50% when there was increased social support from spouse/partner (referral function). The odds of using mental health services within general health setup were reduced by 60% when there was increased support from relatives (stress-reduction function). Increased social support from friends and relatives was associated with a 40-60% decreased odds of using specialty psychiatric services after experiencing different life events (stress-reduction function). Overall, social support rather than social networks were more strongly associated with increased mental health service use following a life event. The implications for service delivery and program development are discussed.

show abstract

Selection of Working Correlation Structure and Best Model in GEE Analyses of Longitudinal Data

Cited by 128 publications

References 21 publications

Modifications of QIC and CIC for Selecting a Working Correlation Structure in the Generalized Estimating Equation Method

Modifications of QIC and CIC for Selecting a Working Correlation Structure in the Generalized Estimating Equation Method

Hypnotics and Triazolobenzodiazepines - Best Predictors of High-Dose Benzodiazepine Use: Results from the Luxembourg National Health Insurance Registry

The Effect of Social Networks and Social Support on Mental Health Services Use, Following a Life Event, among the Baltimore Epidemiologic Catchment Area Cohort

Contact Info

Product

Resources

About