Showing papers on "Poisson regression published in 2021"

Zero-Inflated Generalized Poisson Regression Model with an Application to Domestic Violence Data

Abstract: The generalized Poisson regression model has been used to model dispersed count data. It is a good competitor to the negative binomial re- gression model when the count data is over-dispersed. Zero-inflated Poisson and zero-inflated negative binomial regression models have been proposed for the situations where the data generating process results into too many zeros. In this paper, we propose a zero-inflated generalized Poisson (ZIGP) regression model to model domestic violence data with too many zeros. Es- timation of the model parameters using the method of maximum likelihood is provided. A score test is presented to test whether the number of zeros is too large for the generalized Poisson model to adequately fit the domestic violence data.

Racial/Ethnic Heterogeneity and Rural-Urban Disparity of COVID-19 Case Fatality Ratio in the USA: a Negative Binomial and GIS-Based Analysis.

Abstract: The 2019 coronavirus disease (COVID-19) has exacerbated inequality in the United States of America (USA). Black, indigenous, and people of color (BIPOC) are disproportionately affected by the pandemic. This study examines determinants of COVID-19 case fatality ratio (CFR) based on publicly sourced data from January 1 to December 18, 2020, and sociodemographic and rural-urban continuum data from the US Census Bureau. Nonspatial negative binomial Poisson regression and geographically weighted Poisson regression were applied to estimate the global and local relationships between the CFR and predictors-rural-urban continuum, political inclination, and race/ethnicity in 2407 rural counties. The mean COVID-19 CFR among rural counties was 1.79 (standard deviation (SD) = 1.07; 95% CI 1.73-1.84) higher than the total US counties (M = 1.69, SD = 1.18; 95% CI: 1.65-1.73). Based on the global NB model, CFR was positively associated with counties classified as "completely rural" (incidence rate ratio (IRR) = 1.24; 95% CI: 1.12-1.39) and "mostly rural" (IRR = 1.26; 95% CI: 1.15-1.38) relative to "mostly urban" counties. Nonspatial regression indicates that COVID-19 CFR increases by a factor of 8.62, 5.87, 2.61, and 1.36 for one unit increase in county-level percent Blacks, Hispanics, American Indians, and Asian/Pacific Islanders, respectively. Local spatial regression shows CFR was significantly higher in rural counties with a higher share of BIPOC in the Northeast and Midwest regions, and political inclination predicted COVID-19 CFR in rural counties in the Midwest region. In conclusion, spatial and racial/ethnic disparities exist for COVID-19 CFR across the US rural counties, and findings from this study have implications for public health.

Too many zeros and/or highly skewed? A tutorial on modelling health behaviour as count data with Poisson and negative binomial regression.

Abstract: Background: Dependent variables in health psychology are often counts, for example, of a behaviour or number of engagements with an intervention. These counts can be very strongly skewed, and/or co...

Long-term Dietary Flavonoid Intake and Subjective Cognitive Decline in US Men and Women.

Abstract: Objective To prospectively examine the associations between long-term dietary flavonoids and subjective cognitive decline (SCD). Methods We followed 49,493 women from the Nurses9 Health Study (NHS) (1984–2006) and 27,842 men from the Health Professionals Follow-Up Study (HPFS) (1986–2002). Poisson regression was used to evaluate the associations between dietary flavonoids (flavonols, flavones, flavanones, flavan-3-ols, anthocyanins, polymeric flavonoids, and proanthocyanidins) and subsequent SCD. For the NHS, long-term average dietary intake was calculated from 7 repeated semiquantitative food frequency questionnaires (SFFQs), and SCD was assessed in 2012 and 2014. For the HPFS, average dietary intake was calculated from 5 repeated SFFQs, and SCD was assessed in 2008 and 2012. Results Higher intake of total flavonoids was associated with lower odds of SCD after adjustment for age, total energy intake, major nondietary factors, and specific dietary factors. In a comparison of the highest vs the lowest quintiles of total flavonoid intake, the pooled multivariable-adjusted odds ratio (OR) of 3-unit increments in SCD was 0.81 (95% confidence interval [CI] 0.76, 0.89). In the pooled results, the strongest associations were observed for flavones (OR 0.62 [95% CI 0.57, 0.68]), flavanones (0.64 [0.58, 0.68)]), and anthocyanins (0.76 [0.72, 0.84]) (p trend Conclusion Our findings support a benefit of higher flavonoid intakes for maintaining cognitive function in US men and women.

Log-Linear Bayesian Additive Regression Trees for Multinomial Logistic and Count Regression Models

Abstract: We introduce Bayesian additive regression trees (BART) for log-linear models including multinomial logistic regression and count regression with zero-inflation and overdispersion. BART has been app...

Multimorbidity and its associated factors among adults aged 50 and over: A cross-sectional study in 17 European countries

Abstract: Aims To estimate the prevalence of multimorbidity among European community-dwelling adults, as well as to analyse the association with gender, age, education, self-rated health, loneliness, quality of life, size of social network, Body Mass Index (BMI) and disability. Methods A cross-sectional study based on wave 6 (2015) of the Survey of Health, Ageing and Retirement in Europe (SHARE) was conducted, and community-dwelling participants aged 50+ (n = 63,844) from 17 European countries were selected. Multimorbidity was defined as presenting two or more health conditions. The independent variables were gender, age group, educational level, self-rated health, loneliness, size of network, quality of life, BMI and disability (1+ limitations of basic activities of daily living). Poisson regression models with robust variance were fit for bivariate and multivariate analysis. Results The prevalence of multimorbidity was 28.2% (confidence interval-CI 95%: 27.5.8-29.0) among men and 34.5% (CI95%: 34.1-35.4) among women. The most common health conditions were cardiometabolic and osteoarticular diseases in both genders, and emotional disorders in younger women. A large variability in the prevalence of multimorbidity in European countries was verified, even between countries of the same region. Conclusions Multimorbidity was associated with sociodemographic and physical characteristics, self-rated health, quality of life and loneliness.

Count Regression Models with an Application to Zoological Data Containing Structural Zeros

Abstract: Recently, count regression models have been used to model over- dispersed and zero-inflated count response variable that is affected by one or more covariates. Generalized Poisson (GP) and negative binomial (NB) regression models have been suggested to deal with over-dispersion. Zero- inflated count regression models such as the zero-inflated Poisson (ZIP), zero-inflated negative binomial (ZINB) and zero-inflated generalized Pois- son (ZIGP) regression models have been used to handle count data with many zeros. The aim of this study is to model the number of C. caretta hatchlings dying from exposure to the sun. We present an evaluation frame- work to the suitability of applying the Poisson, NB, GP, ZIP and ZIGP to zoological data set where the count data may exhibit evidence of many zeros and over-dispersion. Estimation of the model parameters using the method of maximum likelihood (ML) is provided. Based on the score test and the goodness of fit measure for zoological data, the GP regression model performs better than other count regression models.

A new estimator for the multicollinear Poisson regression model: simulation and application.

Abstract: The maximum likelihood estimator (MLE) suffers from the instability problem in the presence of multicollinearity for a Poisson regression model (PRM) In this study, we propose a new estimator with some biasing parameters to estimate the regression coefficients for the PRM when there is multicollinearity problem Some simulation experiments are conducted to compare the estimators' performance by using the mean squared error (MSE) criterion For illustration purposes, aircraft damage data has been analyzed The simulation results and the real-life application evidenced that the proposed estimator performs better than the rest of the estimators

Modified ridge-type for the Poisson regression model: simulation and application

Abstract: The Poisson regression model (PRM) is employed in modelling the relationship between a count variable (y) and one or more explanatory variables. The parameters of PRM are popularly estimated using ...

Long-term Health and Social Outcomes in Children and Adolescents Placed in Out-of-Home Care.

Abstract: Importance Children who are placed in out-of-home care may have poorer outcomes in adulthood, on average, compared with their peers, but the direction and magnitude of these associations need clarification. Objective To estimate associations between being placed in out-of-home care in childhood and adolescence and subsequent risks of experiencing a wide range of social and health outcomes in adulthood following comprehensive adjustments for preplacement factors. Design, setting, and participants This cohort and cosibling study of all children born in Finland between 1986 and 2000 (N = 855 622) monitored each person from their 15th birthday either until the end of the study period (December 2018) or until they migrated, died, or experienced the outcome of interest. Cox and Poisson regression models were used to estimate associations with adjustment for measured confounders (from linked population registers) and unmeasured familial confounders (using sibling comparisons). Data were analyzed from October 2020 to August 2021. Exposures Placement in out-of-home care up to age 15 years. Main outcomes and measures Through national population, patient, prescription drug, cause of death, and crime registers, 16 specific outcomes were identified across the following categories: psychiatric disorders; low socioeconomic status; injuries and experiencing violence; and antisocial behaviors, suicidality, and premature mortality. Results A total of 30 127 individuals (3.4%) were identified who had been placed in out-of-home care for a median (interquartile range) period of 1.3 (0.2-5.1) years and 2 (1-3) placement episodes before age 15 years. Compared with their siblings, individuals who had been placed in out-of-home care were 1.4 to 5 times more likely to experience adverse outcomes in adulthood (adjusted hazard ratio [aHR] for those with a fall-related injury, 1.40; 95% CI, 1.25-1.57 and aHR for those with an unintentional poisoning injury, 4.79; 95% CI, 3.56-6.43, respectively). The highest relative risks were observed for those with violent crime arrests (aHR, 4.16; 95% CI, 3.74-4.62; cumulative incidence, 24.6% in individuals who had been placed in out-of-home care vs 5.1% in those who had not), substance misuse (aHR, 4.75; 95% CI, 4.25-5.30; cumulative incidence, 23.2% vs 4.6%), and unintentional poisoning injury (aHR 4.79; 95% CI, 3.56-6.43; cumulative incidence, 3.1% vs 0.6%). Additional adjustments for perinatal factors, childhood behavioral problems, and traumatic injuries, including experiencing violence, did not materially change the findings. Conclusions and relevance Out-of-home care placement was associated with a wide range of adverse outcomes in adulthood, which persisted following adjustments for measured preplacement factors and unmeasured familial factors.

Estimating the change in soccer's home advantage during the Covid-19 pandemic using bivariate Poisson regression.

Abstract: In wake of the Covid-19 pandemic, 2019-2020 soccer seasons across the world were postponed and eventually made up during the summer months of 2020. Researchers from a variety of disciplines jumped at the opportunity to compare the rescheduled games, played in front of empty stadia, to previous games, played in front of fans. To date, most of this post-Covid soccer research has used linear regression models, or versions thereof, to estimate potential changes to the home advantage. However, we argue that leveraging the Poisson distribution would be more appropriate and use simulations to show that bivariate Poisson regression (Karlis and Ntzoufras in J R Stat Soc Ser D Stat 52(3):381-393, 2003) reduces absolute bias when estimating the home advantage benefit in a single season of soccer games, relative to linear regression, by almost 85%. Next, with data from 17 professional soccer leagues, we extend bivariate Poisson models estimate the change in home advantage due to games being played without fans. In contrast to current research that suggests a drop in the home advantage, our findings are mixed; in some leagues, evidence points to a decrease, while in others, the home advantage may have risen. Altogether, this suggests a more complex causal mechanism for the impact of fans on sporting events.

Attempted suicide rates before and during the COVID-19 pandemic: interrupted time series analysis of a nationally representative sample.

Abstract: BACKGROUND: To characterize the association between the protracted biopsychosocial coronavirus disease 2019 (COVID-19) pandemic exposures and incident suicide attempt rates. METHODS: Data were from a nationally representative cohort based on electronic health records from January 2013 to February 2021 (N = 852 233), with an interrupted time series study design. For the primary analysis, the effect of COVID-19 pandemic on incident suicide attempts warranting in-patient hospital treatment was quantified by fitting a Poisson regression and modeling the relative risk (RR) and the corresponding 95% confidence intervals (CIs). Scenarios were forecast to predict attempted suicide rates at 10 months after social mitigation strategies. Fourteen sensitivity analyses were performed to test the robustness of the results. RESULTS: Despite the increasing trend in the unexposed interval, the interval exposed to the COVID-19 pandemic was statistically significant (p < 0.001) associated with a reduced RR of incident attempted suicide (RR = 0.63, 95% CI 0.52-0.78). Consistent with the primary analysis, sensitivity analysis of sociodemographic groups and methodological factors were statistically significant (p < 0.05). No effect modification was identified for COVID-19 lockdown intervals or COVID-19 illness status. All three forecast scenarios at 10 months projected a suicide attempt rate increase from 12.49 (7.42-21.01) to 21.38 (12.71-35.99). CONCLUSIONS: The interval exposed to the protracted mass social trauma of the COVID-19 pandemic was associated with a lower suicide attempt rate compared to the unexposed interval. However, this trend is likely to reverse 10 months after lifting social mitigation policies, underscoring the need for enhanced implementation of public health policy for suicide prevention.

A new adjusted Liu estimator for the Poisson regression model

Abstract: The Poisson regression model (PRM) is usually applied in the situations when the dependent variable is in the form of count data. For estimating the unknown parameters of the PRM, maximum likelihood estimator (MLE) is commonly used. However, its performance is suspected when the regressors are multicollinear. The performance of MLE is not satisfactory in the presence of multicollinearity. To mitigate this problem, different biased estimators are discussed in the literature, that is, ridge and Liu. However, the drawback of using the traditional Liu estimator is that in most of the times, the shrinkage parameter d, attains a negative value which is the major disadvantage of traditional Liu estimator. So, to overcome this problem, we propose a new adjusted Poisson Liu estimator (APLE) for the PRM which is the robust solution to the problem of multicollinear explanatory variables. For assessment purpose, we perform a theoretical comparison with other competitive estimators. In addition, a Monte Carlo simulation study is conducted to show the superiority of the new estimator. At the end, two real life applications are also considered. From the findings of simulation study and two empirical applications, it is observed that the APLE is the most robust and consistent estimation method as compared to the MLE and other competitive estimators.

Renting-in cropland, machinery use intensity, and land productivity in rural China

Abstract: This study examines the impacts of renting-in cropland on machinery use intensity, utilizing an innovative endogenous-treatment Poisson regression (ETPR) model and survey data from wheat farmers in...

Fitting Parametric and Semi-parametric Conditional Poisson Regression Models with Cox's Partial Likelihood in Self-controlled Case Series and Matched Cohort Studies

Abstract: The self-controlled case series (SCCS) and the matched cohort are two frequently used study designs to adjust for known and unknown con- founding effects in epidemiological studies. Count data arising from these two designs may not be independent. While conditional Poisson regres- sion models have been used to take into account the dependence of such data, these models have not been available in some standard statistical soft- ware packages (e.g., SAS). This article demonstrates 1) the relationship of the likelihood function and parameter estimation between the conditional Poisson regression models and Cox's proportional hazard models in SCCS and matched cohort studies; 2) that it is possible to fit conditional Pois- son regression models with procedures (e.g., PHREG in SAS) using Cox's partial likelihood model. We tested both conditional Poisson likelihood and Cox's partial likelihood models on data from studies using either SCCS or a matched cohort design. For the SCCS study, we fitted both parametric and semi-parametric models to model age effects, and described a simple way to apply the parametric and complex semi-parametric analysis to case series data.

Modeling the Impact of Weather Conditions on Pedestrian Injury Counts Using LASSO-Based Poisson Model

Abstract: Statistical models for measuring the impact of adverse weather conditions on pedestrian injuries are of great importance for enhancing road safety measures. The development of these models in the presence of high collinearity among the weather conditions poses a real challenge in practice. The collinearity among these conditions may result in underestimation of the regression coefficients of the regression model, and hence inconsistency regarding the impact of the weather conditions on the pedestrian injuries counts. This paper presents a methodology through which the penalization-based regression is applied to model the impact of weather conditions on pedestrian injury in the presence of a high level of collinearity among these conditions. More specifically, the methodology integrates both the least absolute shrinkage squared operator (Lasso) with the cross-validation approach. The statistical performance of the proposed methodology is assessed through an analytical comparison involving the standard Poisson regression, Poisson generalized linear model (Poisson-GzLM), and Ridge penalized regression model. The mean squared error (MSE) was used as a criterion of comparison. In terms of the MSE, the Lasso-based Poisson generalized linear model (Lasso-GzLM) revealed an advantage over the other regression models. Moreover, the study revealed that weather conditions involved in this study are of insignificant impact on pedestrian injury counts.

Mortality in hospitalized COVID-19 patients was associated with the COVID-19 admission rate during the first year of the pandemic in Sweden.

Abstract: Studies from the first pandemic wave found associations between COVID-19 hospital load and mortality. Here, we aimed to study if mortality of hospitalized COVID-19 patients was associated with the ...

Making Tweedie’s compound Poisson model more accessible

Abstract: The most commonly used regression model in general insurance pricing is the compound Poisson model with gamma claim sizes. There are two different parametrizations for this model: the Poisson-gamma parametrization and Tweedie’s compound Poisson parametrization. Insurance industry typically prefers the Poisson-gamma parametrization. We review both parametrizations, provide new results that help to lower computational costs for Tweedie’s compound Poisson parameter estimation within generalized linear models, and we provide evidence supporting the industry preference for the Poisson-gamma parametrization.

Effect of multimorbidity on utilisation and out-of-pocket expenditure in Indonesia: quantile regression analysis.

Abstract: Multimorbidity (the presence of two or more non-communicable diseases) is a major growing challenge for many low-income and middle-income countries (LMICs). Yet, its effects on health care costs and financial burden for patients have not been adequately studied. This study investigates the effect of multimorbidity across the different percentiles of healthcare utilisation and out-of-pocket expenditure (OOPE). We conducted a secondary data analysis of the 2014/2015 Indonesian Family Life Survey (IFLS-5), which included 13,798 respondents aged ≥40 years. Poisson regression was used to assess the association between sociodemographic characteristics and the total number of non-communicable diseases (NCDs), while multivariate logistic regression and quantile regression analysis was used to estimate the associations between multimorbidity, health service use and OOPE. Overall, 20.8% of total participants had two or more NCDs in 2014/2015. The number of NCDs was associated with higher healthcare utilisation (coefficient 0.11, 95% CI 0.07–0.14 for outpatient care and coefficient 0.09 (95% CI 0.02–0.16 for inpatient care) and higher four-weekly OOPE (coefficient 27.0, 95% CI 11.4–42.7). The quantile regression results indicated that the marginal effect of having three or more NCDs on the absolute amount of four-weekly OOPE was smaller for the lower percentiles (at the 25th percentile, coefficient 1.0, 95% CI 0.5–1.5) but more pronounced for the higher percentile of out-of-pocket spending distribution (at the 90th percentile, coefficient 31.0, 95% CI 15.9–46.2). Multimorbidity is positively correlated with health service utilisation and OOPE and has a significant effect, especially among those in the upper tail of the utilisation/costs distribution. Health financing strategies are urgently required to meet the needs of patients with multimorbidity, particularly for vulnerable groups that have a higher level of health care utilisation.