Testing the Dual-State-Process assumption in the preventive care services use

• Main Author: Dimitris Zavras
• Format: Article
• Language: English
• Published: Edizioni FS 2020-03-01
• Series: Journal of Health and Social Sciences
• Subjects: dual-state-process assumption; economics; greece; preventive health services; sampling zeros; statistics; structural zeros; vuong test; zero-inflated negative binomial model.
• Online Access: https://journalhss.com/publications/2020-51/

Introduction: Health services use is often measured as a count variable that is characterized by an excess of zeros. Zeros are generally considered to be generated from a dual-state process, i.e., sampling zeros concern a group of at-risk individuals, while structural zeros concern a group of not-at-risk individuals. However, in several studies, especially those regarding preventive services use, the dual-state-process assumption is questionable. In this sense, the objective of this paper is to investigate whether the dual-state-process assumption holds in the case of preventive services use. Methods: For the purpose of this study, we analyzed data from a Panhellenic cross-sectional survey that was conducted in 2017. The survey used stratified random sampling, and the sample selection strata were defined by age, gender, urbanity status of permanent residence and prefecture. The sample consisted of 2003 adults. A computer-assisted telephone interviewing method was used for the data collection. Since the outcome variable was the number of times that preventive health services were used, the analysis was based on the comparison between a zero-inflated negative binomial model and a standard negative binomial model through the corrected Vuong test. Several health, socioeconomic, demographic and structural factors of the Greek health care system were used as independent variables. Results: According to the analysis, the dual-state-process assumption does not hold in the case of preventive services use and since the need for receiving preventive care exists in most age and gender groups, this is probably due to the fact that preventive services use is infrequent, meaning that the majority of zeros are sampling zeros. Discussion and Conclusion: The results highlight the need for testing the assumption if zero-inflated count models are to be used.