205470 Trend analysis of national emergency department visits: A comparison of multilevel models and weighted least square regression

Monday, November 9, 2009

Motao Zhu, MD, MS, PhD , Dept. of Community Medicine and Injury Control Research Center, West Virginia University, Morgantown, WV
Haitao Chu, MD, PhD , Division of Biostatistics, University of Minnesota, Minneapolis, MN
Mary Carter, PhD , Health Sciences Center, West Virginia University, Morgantown, WV
  
Background

The number of emergency department (ED) visits increased by 20% and the ED visit rate increased by 7% from 1995 to 2005 in the United States (US). The trend analysis is not straightforward for the national surveys with complex sampling designs.

Methods

We applied the fixed effects model, random effects model, and weighted least square regression to analyze the trend of ED visits from 1997 through 2006 in the US. A total of 311,288 ED visits from the National Hospital Ambulatory Medical Care Survey were analyzed. They were weighted to represent one billion visits.

Results

For the national ED visits, the fixed effects model and the weighted least square regression produced the identical estimate of an annual increase of 2.4 million (95% confidence interval CI: 0.3 to 4.4 million). The random effects model suggested that the number of visits over the years was homogeneous after taking the linear trend into consideration. For the national ED visit rate per 100 persons, the fixed effects model and weighted least square regression produced identical results. The estimate was 0.41 (95% CI: -0.30 to 1.11). The random effects model suggested the rate of ED visit over the years was homogeneous after taking the linear trend into consideration.

Conclusions

We found an annual increase of 2.4 million ED visits. The linear trend between the ED visit rate and year was not statistically significant. Fixed effects and random effects model are flexible models to estimate the trend, especially for surveys with complex sampling designs.

Learning Objectives:
1) Analyze the national trend of emergency department visits from 1997 through 2006 2) Compare the fixed effects model, random effects model, and weighted least square regression in trend analysis of a survey of complex sampling design

Keywords: Surveillance, Emergency Department/Room

Presenting author's disclosure statement:

Qualified on the content I am responsible for because: I designed and carried out the study.
Any relevant financial relationships? No

I agree to comply with the American Public Health Association Conflict of Interest and Commercial Support Guidelines, and to disclose to the participants any off-label or experimental uses of a commercial product or service discussed in my presentation.

See more of: ICEHS Special Session on Injury Research Methods
See more of: Injury Control and Emergency Health Services