Secondary data talk 2010

Editor's Notes

• #14: Shock index is defined as (HR) / Systolic Blood pressure (SBP) It has been proposed as a a triage tool for both emergency department and disaster settings In fact, the Colorado department of public health uses shock index as part of a destination triage criteria for an emergency flu pandemic disaster plan for adults. It has also been used a monitoring tool for therapeutic efficacy in the adult population There are no established pediatric normal values.
• #15: In the adult population, an elevated shock index is defined as > 0.9. It has been associated with blood loss, admissions, intensive care interventions, and ultimately poor outcome. Additionally, both experimental and clinical studies have shown that shock index has an inverse linear relationship with left ventricular function during acute circulatory failure. In other words as the shock index goes up, the left ventricular function deteriorates. There is only one pediatric study of shock index. It examined ill children who were transported to a tertiary care hospital. In this study, there was a positive association with mortality and, a reduction in shock index was associated with improved outcome.
• #16: The objective of our study was to evaluate shock index as a predictor for admission in an emergency department population of children Si was evaluated independent of HR and SBP
• #17: The objective of our study was to evaluate shock index as a predictor for admission in an emergency department population of children Si was evaluated independent of HR and SBP
• #18: Two data sources were analyzed in our study. A healthy population dataset; The National Health and Nutrition Examination Survey (NHANES) was studied from 1999-2006 An emergency department dataset; The National Hospital Ambulatory Medical Care Survey (NHAMCS ED) dataset was studied from 2004-2006..
• #19: Again since there are no known normal values for the pediatric cohort, we tabulated norms off the “normal healthy population”, otherwise known as NHANES. This data will be presented tomorrow in the poster session. After deriving the age and gender specific percentiles for SI, HR, and SBP, we applied these predictors to the ED population which was used to address the study question. The NHANES or normal population study did not measure blood pressure in children less than 8 years old, therefore our study of the ED population was limited to patients age 8-21 years.
• #20: Again since there are no known normal values for the pediatric cohort, we tabulated norms off the “normal healthy population”, otherwise known as NHANES. This data will be presented tomorrow in the poster session. After deriving the age and gender specific percentiles for SI, HR, and SBP, we applied these predictors to the ED population which was used to address the study question. The NHANES or normal population study did not measure blood pressure in children less than 8 years old, therefore our study of the ED population was limited to patients age 8-21 years.
• #21: Again since there are no known normal values for the pediatric cohort, we tabulated norms off the “normal healthy population”, otherwise known as NHANES. This data will be presented tomorrow in the poster session. After deriving the age and gender specific percentiles for SI, HR, and SBP, we applied these predictors to the ED population which was used to address the study question. The NHANES or normal population study did not measure blood pressure in children less than 8 years old, therefore our study of the ED population was limited to patients age 8-21 years.
• #25: The majority of the healthy population or normative data will be presented tomorrow. This shows the shock index percentiles by age. The sample size is > 13, 000, this represents 57.2 million patients with weighted data. The white horizontal line indicates a SI of 0.9.
• #28: Again since there are no known normal values for the pediatric cohort, we tabulated norms off the “normal healthy population”, otherwise known as NHANES. After deriving the age and gender specific percentiles for SI, HR, and SBP, we applied these predictors to the ED population which was used to address the study question. The NHANES or normal population study did not measure blood pressure in children less than 8 years old, therefore our study of the ED population was limited to patients age 8-21 years.
• #30: Ideally, we would have used “shock” as our outcome measure. However, in the NHAMCS ED dataset, which includes 7 years of nationally representative data, there were too few patients with the shock-related indicators we considered. So, we decided to use “admission” as our outcome, as it was the most common outcome measure that was indicative of higher acuity illness in the dataset.
• #31: There were insufficient data to evaluate shock or a surrogate shock marker as an outcome in the ED population. Thus, we chose “admission” as our outcome variable.
• #32: Logistic regression was used to model the association between predictor variables and the outcome, admission There were two approaches for the primary predictor variable: SI > 95th SI > .9 Admission, as the outcome, was chosen. There were insufficient data to evaluate shock or a surrogate shock marker as an outcome in this dataset.
• #33: Cut-point selection for percentiles was based on frequency distribution in the ED pediatric population. The 95th percentile for SI and HR were used. The 25th percentile for SBP was used. The absolute cut-point of 0.9 was based on adult literature. Younger children will have SI > 0.9, however this dataset only had the ability to evaluate patients greater than 8 years.
• #34: Our methods for the logistic regression used two models which were identical except that model 1 used SI >95% for age and gender and model 2 used a cut-point set at 0.9.
• #35: Other independent variables were HR > 95 % for age and gender, SBP < 25th % for age and gender, Age, Gender Race Ethnicity and Payor type colinearity between variables was assessed
• #36: THE emergency department data or NHAMCS ED dataset had over 18,000 ED visits This represents 58.9 million patients with weighted data. The patients were ages 8-21 years. 4 % were admitted
• #37: 14% had a SI > 95 percentile for age and gender 19% had a SI > 0.9 29% had a HR > 95 percentile 6% had a SBP < 25 percentile < 1% had SI > 95% in the context of normal HR (< 95%) and SBP (> 25%)
• #38: In bivariate chi-square analyses, SI was associated with admission (p < 0.0001) This was true for both SI > 95th % and SI > 0.9
• #39: In bivaraite analyses, both SI cutoffs were associated with Admission. The SI >95% is represented in yellow and SI > 0.9 is represented in oranage
• #40: The unadjusted odds ratios are 2.97 for the SI cutoff of the 95th percentile, and 2.63 for the SI cutoff of 0.9. Both were significant.
• #41: In model 1, we used logistic regression to analyze the association between our outcome, admission, and the shock index cutoff defined by the 95th percentile for age and gender. Our primary independent variable, SI greater than 95th percentile, was associated with the outcome, with an odds ratio of 1.54.
• #42: In model 2, we used logistic regression to analyze the association between our outcome, admission, and the shock index cutoff of 0.9. Our primary independent variable, SI greater than 0.9, was associated with the outcome, with an odds ratio of 1.50.
• #43: The limitations of our study were that: No children under 8 years were evaluated The sample had insufficient numbers either to analyze the population who had abnormal SI in the context of normal HR and sBP, or to use “shock” as an outcome. Admission is based on provider variability as well as patient severity No ability to assess unscheduled return visits