265426 Multiple hospitalizations by unique individuals across 4 years of state-level data: Demographic profiles to identify areas of potential estimation bias

Sunday, October 28, 2012

Hyeong Jun Ahn, PhD , Biostatistics Core, University of Hawaii John A. Burns School of Medicine, Honolulu, HI
Tetine Sentell, PhD , Office of Public Health Studies, Univerisity of Hawaii at Manoa, Honolulu, HI
Jill Miyamura, PhD , Hawaii Health Information Corporation, Honolulu, HI
John J. Chen, PhD , Biostatistics Core, University of Hawaii, John A. Burns School of Medicine, Honolulu, HI
Background: State and national-level hospitalization data often do not identify unique individuals, and thus cannot account for multiple hospitalizations by the same individuals in estimations of rates and disparities. This study considered demographic profiles of individuals with multiple hospitalizations in a state-level data set to identify possible areas of estimation bias if multiple visits by these individuals are not considered in population-level analyses. Methods: Hawaii Health Information Corporation (HHIC) adult data from December 2006-December 2010 was used, which includes detailed discharge information for all hospitalizations in Hawaii. Chi-squared tests and multivariate logistic regressions compared individuals with multiple visits vs. a single visit in the study period. Transfers and those not from Hawaii and/or missing race/ethnic or payer data were excluded. Statistical significance is at the p<0.0001 level. Results: Out of 289,190 hospitalizations during 4 years, 168,143 were from unique individuals. 109,745 (65%) individuals had 1 visit, 32,659 (19.4%) had 2 visits, 12,242 (7.3%) had 3 visits, and 13,497 (8%) had 4+ visits. The highest number of hospitalizations by a unique individual was 55. Men (38.9%) were significantly more likely than women (32.1%) to have multiple hospitalizations, as were those who were older (45.1% if 65+ years, 32.5% if 40-64, and 24.2% if 18-39), and those on Medicare (46.7%) and Medicaid (34.6%) vs. private insurance (25.5%). Native Hawaiians had the highest percentage of multiple hospitalizations (36.8%), followed by Japanese (36.6%), White (34.3%), Chinese (33.9%) and Filipinos (31.4%). Those with multiple hospitalizations had significantly higher comorbidity scores (3.50) than those with one hospitalization (0.87). In multivariable adjusted analyses, Chinese (OR: 0.76; 95%CI:0.72-0.80), Filipino (OR:0.82; 95%CI:0.80-0.85), and Japanese adults (OR: 0.88; 95%CI:0.85-0.90) were significantly less likely than Whites to have multiple visits. Older age maintained significance in predicting multiple hospitalizations as did Medicaid (OR:1.61; 95%CI:1.56-1.67) and Medicare (OR: 1.86; 95%CI:1.78-1.95) vs. private insurance, and higher comorbidity scores (OR: 1.45; 95%CI:1.44-1.47). Gender and Hawaiian ethnicity were no longer significant in multivariable models. Conclusions: HHIC data show that ignoring multiple visits in population-level analyses will clearly violate the independence of data assumption behind many commonly used statistical models. Over 35% of the patients HHIC tracked over the 4 year period had more than one hospitalization with >15% having >3 hospitalizations. If the impact of multiple visits is not accounted for, estimation of health disparities would be severely biased as differences in this factor were seen across racial/ethnic, comorbidity status, payer groups, and across the lifespan.

Learning Areas:
Biostatistics, economics
Public health or related research

Learning Objectives:
1. List at least six demographic factors that are significantly associated with multiple hospitalizations in a state-level, all-payer hospitalization data set. 2. Explain the implications for these findings in our understanding of population-level hospitalizations rates, particularly possible biases in estimates of hospitalization disparities across ethnic groups and the lifespan.

Keywords: Health Care Utilization, Hospitals

Presenting author's disclosure statement:

Qualified on the content I am responsible for because: I helped to design, conduct, and interpret this study and it's results.
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.