ISS is a summary of the overall severity of anatomical injury for each patient. [15] It has an ordinal scale from 1-75 and is derived from the AIS severity scores for each injury. The main outcome was in hospital mortality. We also explored the association between size of bleeding and evacuation of haematoma. Analysis Deciding which variables should be considered confounders and which should be considered mediators that are on the causal pathway between bleeding Inhibitors,research,lifescience,medical and outcome requires a conceptual

framework. We could consider as confounders all variables shown to be associated with poor prognosis in TBI, such as age, severity of the TBI (as defined by GCS) and other CT scan abnormalities. However, some of these variables (e.g. brain swelling) might be on the causal pathway between bleeding and patient outcome, and others (e.g. GCS) might be Inhibitors,research,lifescience,medical a consequence of intracranial bleeding, though not on the causal pathway. Adjusting for these variables would attenuate a true association between bleeding and outcome. Because of the uncertainty in determining

which factors are confounders and which are on the causal Inhibitors,research,lifescience,medical pathway, we analysed the data from two conceptual frameworks, in the hope that the two different analyses would provide a better understanding of the association between IB and outcome. The first includes all potential confounders, the second, excludes those variables that could be on the causal pathway between IB and patient outcome. Each exposure and confounding variable was entered into a multivariable logistic ABT-263 datasheet regression analysis Inhibitors,research,lifescience,medical to analyse its relationship with outcome. A first analysis considered no bleeding as

the baseline category and mortality and haematoma evacuation as outcomes. Because we were interested in quantifying the mortality risk associated with large, as opposed to small IB, we also conducted a second analysis evaluating the effect of IB size on mortality using small IB as the baseline. To determine the functional form of the predictors age and GCS in the model, fractional polynomials, quadratic and cubic spline and Lowess smoothing were explored. Inhibitors,research,lifescience,medical Results Between 2001 and 2008 15,754 adult patients meeting study inclusion criteria for TBI presented to TARN hospitals and were submitted. In 1792 patients, the GCS on arrival at the first hospital was missing. The remaining 13,962 were used for this study. isothipendyl Table ​Table11 describe the characteristics of the study population. Almost three quarters of the patients were male. The median age was 41 years old, the median GCS was 13 and the median ISS was 18. The commonest mechanism of injury was road traffic crashes and in-hospital mortality was 22%. A total of 6445 patients (46%) had some type of IB (EPH, SDH, IPH or SAH). Of these patients 2,922 (45%) had one type of IB, 1018 (16%) had two types of IB, 1619 (25%) had three types of IB and 886 (14%) had four types IB. Table 1 Demographics SDH was the most common type, present in 30% of the patients.