Estimation in the zero-inflated bivariate Poisson model, with an application to health-care utilization data

Data on the demand for medical care is usually measured by a number of different counts. These count data are most often correlated and subject to high proportions of zeros. However, excess zeros and the dependence between these data can jointly affect several utilization measures.In this paper, the zero-inflated bivariate Poisson regression model (ZIBP) was used to analyze health-care utilization data. First, the asymptotic properties of the maximum likelihood estimator (MLE) of this model were investigated theoretically.Then, a simulation study is conducted to evaluate the behaviour of the estimator in finite samples. Finally, an application of the ZIBP model to health care demand data is provided by way of illustration.

Download Full-text

Estimation in the zero-inflated bivariate Poisson model, with an application to health-care utilization data

Afrika Statistika ◽

10.16929/as/2021.2767.183 ◽

2021 ◽

Vol 16 (2) ◽

pp. 2767-2788

Author(s):

Konan Jean Geoffroy Kouakou ◽

Ouagninia Hili ◽

Jean-Etienne Ouindllassida Dupuy

Keyword(s):

Health Care ◽

Health Care Utilization ◽

Asymptotic Properties ◽

Poisson Model ◽

Care Utilization ◽

Likelihood Estimator ◽

Health Care Demand ◽

Excess Zeros ◽

Finite Samples ◽

Utilization Data

Data on the demand for medical care is usually measured by a number of different counts. These count data are most often correlated and subject to high proportions of zeros. However, excess zeros and the dependence between these data can jointly affect several utilization measures.In this paper, the zero-inflated bivariate Poisson regression model (ZIBP) was used to analyze health-care utilization data. First, the asymptotic properties of the maximum likelihood estimator (MLE) of this model were investigated theoretically. Then, a simulation study is conducted to evaluate the behaviour of the estimator in finite samples. Finally, an application of the ZIBP model to health care demand data is provided by way of illustration.

Download Full-text

Health status and health care utilization after discharge from geriatric in-hospital stay – description of a register-based study

BMC Health Services Research ◽

10.1186/s12913-021-06751-3 ◽

2021 ◽

Vol 21 (1) ◽

Author(s):

E. Rydwik ◽

R. Lindqvist ◽

C. Willers ◽

L. Carlsson ◽

G. H. Nilsson ◽

...

Keyword(s):

Primary Care ◽

Health Care ◽

Health Status ◽

Health Care Utilization ◽

Care Utilization ◽

Future Studies ◽

Healthcare Data ◽

Utilization Data ◽

Hospital Stays ◽

Socioeconomic Data

Abstract Background This study is the first part of a register-based research program with the overall aim to increase the knowledge of the health status among geriatric patients and to identify risk factors for readmission in this population. The aim of this study was two-fold: 1) to evaluate the validity of the study cohorts in terms of health care utilization in relation to regional cohorts; 2) to describe the study cohorts in terms of health status and health care utilization after discharge. Methods The project consist of two cohorts with data from patient records of geriatric in-hospital stays, health care utilization data from Stockholm Regional Healthcare Data Warehouse 6 months after discharge, socioeconomic data from Statistics Sweden. The 2012 cohort include 6710 patients and the 2016 cohort, 8091 patients; 64% are women, mean age is 84 (SD 8). Results Mean days to first visit in primary care was 12 (23) and 10 (19) in the 2012 and 2016 cohort, respectively. Readmissions to hospital was 38% in 2012 and 39% in 2016. The validity of the study cohorts was evaluated by comparing them with regional cohorts. The study cohorts were comparable in most cases but there were some significant differences between the study cohorts and the regional cohorts, especially regarding amount and type of primary care. Conclusion The study cohorts seem valid in terms of health care utilization compared to the regional cohorts regarding hospital care, but less so regarding primary care. This will be considered in the analyses and when interpreting data in future studies based on these study cohorts. Future studies will explore factors associated with health status and re-admissions in a population with multi-morbidity and disability.

Download Full-text