Avoidable hospitalization, also called potentially preventable hospitalization, has been used extensively as an indicator of the accessibility and overall effectiveness of the primary care system. The indicator is regularly used by governments to provide an evidence-based basis for targeted interventions designed to control medical care expenditures. Previous research has found that a universal health insurance system is not necessarily associated with low variability in avoidable hospitalization risk among geographic areas or sub-populations. Taiwan has implemented the National Health Insurance program for more than 15 years, but the referral system is still not adequate. Thus, here raises a question: Does the implementation of Taiwan’s National Health Insurance program help reduce avoidable hospitalization rates, longitudinally speaking? Another related question is: What is the trend of primary care manpower in Taiwan? Lastly, are the aforementioned two trends correlated? Hence, the three-year research proposal aims to probe the note-above questions. The research methods are illustrated as follows: The First Year: (a). The research objective is to assess the trend of avoidable hospitalization rates during the 15-year period (1996-2010) in Taiwan. Avoidable hospitalizations will be identified based on the definitions proposed by the Institute of Medicine, and Prevention Quality Indicators (PQIs) and Pediatric Quality Indicators by the Agency for Healthcare Research and Quality in the US. ICD-9-CM codes will be used to help identify patient diagnosis information from the database mentioned below. (b). The data source is the Longitudinal Health Insurance Database (LHID 2005) from the National Health Research Institute. (c). For trend analysis, two kinds of statistical analyses will be performed -- (i). A random intercept logistic regression model will be used to assess whether the subject has an avoidable hospitalization (a binary variable), while controlling for age, sex, region and random subject effect. (ii). Joinpoint regression analysis will be conducted to analyze the trend in the age-adjusted avoidable hospitalization rates (a continuous variable), and determine if those changes are statistically significant. The Second Year: (a). The research objective is to perform trend analysis of primary care manpower during the 15-year period (1996-2010) in Taiwan. (b).The main data sources are the Longitudinal Health Insurance Database (LHID 2005), “Current Situation of Medical Facilities, Medical Personnel, and Medical Services” released yearly by Department of Health, and “Statistics Yearbook of Practicing Physicians and Health Care Organizations in Taiwan” by Taiwan Medical Association. (c). Both techniques of a random intercept logistic regression model and joinpoint regression analysis will be carried out for trend analysis. The Third Year: (a). The research objective is to explore whether there is correlation between the aforementioned two time series. (b). The data source is the analytical results from previous two years. (c). Generally speaking, the objective is to estimate the cross-correlation function (CCF) between pairs of time series. Consequently, the Box-Jenkins approach which fits the time series to an autoregressive integrated moving average (ARIMA) model will be conducted. Given the elevated risk of adverse health events and higher costs associated with hospitalization, increased attentions and efforts from policymakers, clinical practitioners and hospital administrators to reducing avoidable hospitalizations is clearly warranted. It is hoped that the current research proposal could identify areas where investment might help reduce the rate of avoidable hospitalizations and consequently improve the quality of health care in the future.
|Effective start/end date||8/1/11 → 7/31/12|
- Avoidable hospitalization
- Primary care manpower
- Trend analysis
- National Health Insurance
Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.