Citation

Nagarsheth KH, Gandhi SS, Heidel RE, Kurek SJ, Angel C. J. Surg. Res. 2011; 171(1): 28-30.

Affiliation

University of Tennessee Medical Center, Knoxville, Tennessee.

Copyright

(Copyright © 2011, Elsevier Publishing)

DOI

10.1016/j.jss.2011.03.063

PMID

21514601

Abstract

BACKGROUND: Children under age 12 y represent 15% of all-terrain vehicle (ATV)-related deaths, and those under 16 y old represent >36% of deaths nationwide. In recent years, this has accounted for an increasing proportion of pediatric trauma victims and longer hospitalizations secondary to worsened injuries. We believe it is possible to create a simple mathematical model that can be used to predict hospital length of stay. METHODS: A retrospective review of the trauma registry was performed for all pediatric patients who were involved in ATV accidents from January 2000 to December 2009. Four hundred twenty pediatric patients were identified. A model to predict for total LOS in pediatric patients involved in ATV accidents was constructed. SPSS ver. 17 (SPSS Inc., Chicago, IL) was utilized to conduct all statistical analyses. Statistical significance of regression coefficients was assumed at a P < 0.05 level. RESULTS: We performed a hierarchical multiple regression analysis to build a model that would predict for total length of stay (LOS). A logarithmic transformation was employed on LOS as a dependent variable due to skewness. In Step 1, ISS accounted for a 25% increase in shared variance in LOS (P < 0.001). In Step 2, Glasgow Coma Score (GCS) accounted for 4.3% more variance in LOS (P < 0.001). Finally, in Step 3, the presence of a closed head injury further increased (3.6%) the amount of shared variance in the model (P < 0.001). The final model accounts for 32.9% of the shared variance in total LOS. When using the logarithmic transformation, the final model is: Total LOS = 1.00 + 0.05 injury severity score (ISS) - 0.06 (GCS) + 0.35 closed head injury (CHI). CONCLUSION: Based on our data and statistical analysis, we found it was possible to create a mathematical model that could predict hospital LOS in pediatric ATV accident victims.

Language: en