On bivariate Poisson regression models

Almuhayfith, Fatimah E.; Alzaid, Abdulhamid A.; Omair, Maha A.

doi:10.1016/j.jksus.2015.09.003

Cited by 10 publications

(5 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We use NLMIXED procedure in SAS software to estimate our models. 15 16 Note that we have provided access to both the data and the SAS code for this analysis. 17 18…”

Section: Methodsmentioning

confidence: 99%

Compress the curve: a cross-sectional study of variations in COVID-19 infections across California nursing homes

2021

View full text Add to dashboard Cite

ObjectiveNursing homes’ residents and staff constitute the largest proportion of the fatalities associated with COVID-19 epidemic. Although there is a significant variation in COVID-19 outbreaks among the US nursing homes, we still do not know why such outbreaks are larger and more likely in some nursing homes than others. This research aims to understand why some nursing homes are more susceptible to larger COVID-19 outbreaks.DesignObservational study of all nursing homes in the state of California until 1 May 2020.SettingThe state of California.Participants713 long-term care facilities in the state of California that participate in public reporting of COVID-19 infections as of 1 May 2020 and their infections data could be matched with data on ratings and governance features of nursing homes provided by Centers for Medicare & Medicaid Services (CMS).Main outcome measureThe number of reported COVID-19 infections among staff and residents.ResultsStudy sample included 713 nursing homes. The size of outbreaks among residents in for-profit nursing homes is 12.7 times larger than their non-profit counterparts (log count=2.54; 95% CI, 1.97 to 3.11; p<0.001). Higher ratings in CMS-reported health inspections are associated with lower number of infections among both staff (log count=−0.19; 95% CI, −0.37 to −0.01; p=0.05) and residents (log count=−0.20; 95% CI, −0.27 to −0.14; p<0.001). Nursing homes with higher discrepancy between their CMS-reported and self-reported ratings have higher number of infections among their staff (log count=0.41; 95% CI, 0.31 to 0.51; p<0.001) and residents (log count=0.13; 95% CI, 0.08 to 0.18; p<0.001).ConclusionsThe size of COVID-19 outbreaks in nursing homes is associated with their ratings and governance features. To prepare for the possible next waves of COVID-19 epidemic, policy makers should use these insights to identify the nursing homes who are more likely to experience large outbreaks.

show abstract

“…We use NLMIXED procedure in SAS software to estimate our models. 15 16 Note that we have provided access to both the data and the SAS code for this analysis. 17 18…”

Section: Methodsmentioning

confidence: 99%

Compress the curve: a cross-sectional study of variations in COVID-19 infections across California nursing homes

2021

View full text Add to dashboard Cite

show abstract

“…In the bivariate Poisson analysis, the magnitude and significance of the associations between each of the independent variables and the dependent variable were initially verified [ 57 ]. The results of the bivariate analysis are presented as the crude incidence rate ratio, 95% confidence interval, regression coefficient, standard error, and p-value.…”

Section: Methodsmentioning

confidence: 99%

Zika virus infection and microcephaly: spatial analysis and socio-environmental determinants in a region of high Aedes aegypti infestation in the Central-West Region of Brazil

et al. 2021

View full text Add to dashboard Cite

Background More than 5 years after the Zika virus (ZIKV) epidemic, Zika infection remains a major concern in regions with high Aedes infestation. The objectives of this study were (i) to identify clusters of ZIKV infection and microcephaly, and/or central nervous system (CNS) alterations associated with congenital infection during the epidemic peak in 2016 and subsequently, in 2017 and 2018; (ii) to measure the non-spatial correlation between ZIKV infection and microcephaly and/or CNS alterations associated with congenital infection; and (iii) to analyse the sociodemographic/economic, health, and environmental determinants associated with the incidence of ZIKV in a region of high infestation by Aedes aegypti in the Central-West Region of Brazil. Methods This ecological study analysed 246 municipalities in the state of Goiás (6.9 million inhabitants). The data were obtained from the Information System for Notifiable Diseases (ZIKV cases) and the Public Health Event Registry (microcephaly and/or CNS alterations associated with congenital infection). Incidence rates and prevalence of ZIKA infection were smoothed by an empirical Bayesian estimator (LEbayes), producing the local empirical Bayesian rate (LEBR). In the spatial analysis, ZIKV infection and microcephaly cases were georeferenced by the municipality of residence for 2016 and grouped for 2017 and 2018. Global Moran's I and the Hot Spot Analysis tool (Getis-Ord Gi* statistics) were used to analyse the spatial autocorrelation and clusters of ZIKV infection and microcephaly, respectively. A generalised linear model from the Poisson family was used to assess the association between ecological determinants and the smoothing incidence rate of ZIKV infection. Results A total of 9892 cases of acute ZIKV infection and 121 cases of microcephaly were confirmed. The mean LEBR of the ZIKV infection in the 246 municipalities was 22.3 cases/100,000 inhabitants in 2016, and 10.3 cases/100,000 inhabitants in 2017 and 2018. The LEBR of the prevalence rate of microcephaly and/or CNS alterations associated with congenital infection was 7 cases/10,000 live births in 2016 and 2 cases/10,000 live births during 2017–2018. Hotspots of ZIKV infection and microcephaly cases were identified in the capital and neighbouring municipalities in 2016, with new clusters in the following years. In a multiple regression Poisson analysis, ZIKV infection was associated with higher population density, the incidence of dengue, Aedes larvae infestation index, and average rainfall. The important determinant of ZIKV infection incidence reduction was the increase in households attended by endemic disease control agents. Conclusions Our analyses were able to capture, in a more granular way, aspects that make it possible to inform public managers of the sentinel areas identified in the post-epidemic hotspots.

show abstract

“…These models provide sufficient flexibility by allowing two correlated response variables that different predictors influence. Besides, a bivariate model is more useful for inference and prediction purposes because it allows us to correctly determine the dependencies between two dependent variables [ 23 ]. In recent years, different models were developed that each has some advantages and disadvantages and may present different results in different situations [ 24 ].…”

Section: Backgroundsmentioning

confidence: 99%

“…The bivariate Poisson model is the most extensively used, although it has the same constraints as the univariate Poisson model, and the variance estimates of the model are influenced when there is over-dispersion or under-dispersion in the data. Furthermore, the presence of negative correlation and heterogeneity in variance decreases the model's efficiency [ 23 ]. Bivariate negative binomial (BNN) regression, Dirichlet negative multinomial (DNM) regression model, and diagonal inflated bivariate Poisson (DIBP) regression are introduced in the literature for the solution of these issues and have better performance to describe bivariate responses that are interdependent and also have the feature of hyper diffraction [ 25 , 26 ].…”

Section: Backgroundsmentioning

confidence: 99%

Analysis of injuries and deaths from road traffic accidents in Iran: bivariate regression approach

et al. 2022

View full text Add to dashboard Cite

Backgrounds This study aims to estimate and compare the parameters of some univariate and bivariate count models to identify the factors affecting the number of mortality and the number of injured in road accidents. Methods The accident data used in this study are related to Kermanshah province in march2020 to march2021. Accidents areas were divided into 125 areas based on density characteristics. In a one-year period, 3090 accidents happened on the suburban roads of Kermanshah province, which resulted in 398 deaths and 4805 injuries. Accident information, including longitude and latitude of accident location, type of accident (fatal and injury), number of deaths, number of injuries, accident type, the reason of the accident, and the kind of accident were all included as population-level variables in the regression models. We investigated four frequently used bivariate count regression models for accident data in the literature. Results In bivariate analysis, except for the DNM model, there is a reasonable decrease in the AIC measures of the saturated model compared to the reduced model for the other three models. For the injury models, MSE is lowest, respectively for DIBP (137.87), BNB (289.46), BP (412.36) and DNM (3640.89) models. These results are also established for death models. But, in univariate analysis, only injury models almost present reasonable results. Conclusions Our findings show that the IDBP model is better suitable for evaluating accident datasets than other models. Motorcycle accidents, pedestrian accidents, left turn deviance, and dangerous speeding were all significant variables in the IDBP death model, and these parameters were linked to accident mortality.

show abstract

On bivariate Poisson regression models

Cited by 10 publications

References 23 publications

Compress the curve: a cross-sectional study of variations in COVID-19 infections across California nursing homes

Compress the curve: a cross-sectional study of variations in COVID-19 infections across California nursing homes

Zika virus infection and microcephaly: spatial analysis and socio-environmental determinants in a region of high Aedes aegypti infestation in the Central-West Region of Brazil

Analysis of injuries and deaths from road traffic accidents in Iran: bivariate regression approach

Contact Info

Product

Resources

About