Search

Study Objective: Preexisting conditions have been found to be an independent predictor of mortality after trauma. However, no consensus has been reached as to what indicator of preexisting condition status should be used, and the contribution of preexisting conditions to mortality prediction models is unclear. This study aims to identify the most accurate way to model preexisting condition status to predict inhospital trauma mortality and to evaluate the potential gain of adding preexisting condition status to a standard trauma mortality prediction model., Methods: The study comprised all patients from the trauma registries of 4 Level I trauma centers. Information provided by individual preexisting conditions was compared to 3 commonly used summary measures: (1) absence/presence of any preexisting condition, (2) number of preexisting conditions, and (3) Charlson Comorbidity Index. The impact of adding preexisting condition status to 2 baseline risk models, the current standard Trauma and Injury Severity Score model and an improved model based on nonparametric transformations of quantitative variables, was evaluated by the area under the receiver operating characteristics curve., Results: Discrimination for predicting mortality in the improved model was as follows: baseline risk model: area under the receiver operating characteristics curve=0.935; baseline risk model+individually modeled preexisting conditions: area under the receiver operating characteristics curve=0.941; baseline risk model+presence of any preexisting condition: area under the receiver operating characteristics curve=0.937; baseline risk model+number of preexisting conditions: area under the receiver operating characteristics curve=0.939; baseline risk model+Charlson Comorbidity Index: area under the receiver operating characteristics curve=0.938., Conclusion: Preexisting condition status is an independent predictor of mortality from trauma that provides a modest improvement in mortality prediction. The total number of preexisting conditions is a good summary measure of preexisting condition status. The Charlson Comorbidity Index is no better than the total number of preexisting conditions and is therefore not recommended for use in trauma mortality modeling.

Background: Anatomic injury severity scores can be grouped into two classes; consensus-derived and data-derived. The former, including the Injury Severity Score (ISS), the New Injury Severity Score (NISS), and the Anatomic Profile Score (APS), are based on the severity score of the Abbreviated Injury Scale (AIS), assigned by clinical experts. The latter, including the International Classification of Disease Injury Severity Score (ICISS) and the Trauma Registry Abbreviated Injury Scale Score (TRAIS) are based on survival probabilities calculated in large trauma databases. We aimed to compare the predictive accuracy of consensus-derived and data-derived severity scores when considered alone and in combination with age and physiologic status., Methods: Analyses were based on 25,111 patients from the trauma registries of the four Level I trauma centers in the province of Quebec, Canada, abstracted between April 1998 and March 2005. The predictive validity of each severity score was evaluated in logistic regression models predicting hospital mortality using measures of discrimination (Area Under the Receiver Operating Characteristics curve [AUC]) and calibration (Hosmer-Lemeshow statistic [HL])., Results: Data-derived scores had consistently better predictive accuracy than consensus-derived scores in univariate models (p < 0.0001) but very little difference between scores was observed in models including information on age and physiologic status. The difference in AUC between the least accurate severity score (ISS) and the most accurate severity score (TRAIS) was 15% in anatomic-only models but fell to 2% in models including age and physiologic status., Conclusions: Data-derived scores provide more accurate mortality prediction than consensus-derived scores do when only anatomic injury severity is considered but offer little advantage if age and physiologic status are taken into account. This may be because of the fact that data-derived scores are not an independent measure of anatomic injury severity.

Background: The International Classification of Disease Injury Severity Score (ICISS) and the Trauma Registry Abbreviated Injury Scale Score (TRAIS) are trauma injury severity scores based on probabilities of survival. They are widely used in logistic regression models as raw probability scores to predict the logit of mortality. The aim of this study was to evaluate whether these severity indicators would offer a more accurate prediction of mortality if they were used with a logit transformation., Methods: Analyses were based on 25,111 patients from the trauma registries of the four Level I trauma centers in the province of Quebec, Canada, abstracted between 1998 and 2005. The ICISS and TRAIS were calculated using survival proportions from the National Trauma Data Bank. The performance of the ICISS and TRAIS in their widely used form, proportions varying from 0 to 1, was compared with a logit transformation of the scores in logistic regression models predicting in-hospital mortality. Calibration was assessed with the Hosmer-Lemeshow statistic., Results: Neither the ICISS nor the TRAIS had a linear relation with the logit of mortality. A logit transformation of these scores led to a near-linear association and consequently improved model calibration. The Hosmer-Lemeshow statistic was 68 (35-192) and 69 (41-120) with the logit transformation compared with 272 (227-339) and 204 (166-266) with no transformation, for the ICISS and TRAIS, respectively., Conclusions: In logistic regression models predicting mortality, the ICISS and TRAIS should be used with a logit transformation. This study has direct implications for improving the validity of analyses requiring control for injury severity case mix.

Background: The Revised Trauma Score (RTS) calculated with Major Outcome Trauma Study weights (MTOS-RTS) is currently the standard physiologic severity score in trauma research and quality control. It is often confused with the Triage-RTS (T-RTS), a version that is easier to calculate but only intended for clinical triage., Objectives: To compare the accuracy of the MTOS-RTS to the RTS calculated with weights derived from the study population (POP-RTS) and the T-RTS, for predicting mortality in a trauma population., Methods: The study population consists of 22,388 patients, drawn from the trauma registries of three Level I trauma centers. The predictive accuracy of the MTOS-RTS, POP-RTS, and the T-RTS were compared using measures of discrimination and model fit from logistic regression models., Results: The MTOS-RTS, the POP-RTS, and the T-RTS had the same discrimination (Area under the Receiver Operating Curve [AUC] = 0.841). The POP-RTS and the T-RTS had a slightly better model fit than the MTOS-RTS (AIC = 8010, 8010, and 8067, respectively). The T-RTS had equal discrimination and equal or better model fit than the MTOS-RTS in the whole sample, in each of the three trauma centers and in the population of patients with severe head trauma. The T-RTS was also equivalent to the POP-RTS in all of these population sub-groups., Conclusions: The T-RTS could replace the MTOS-RTS as the standard physiologic severity score for trauma outcome prediction. The advantages of using the T-RTS over the MTOS-RTS are ease of calculation, the need for only one measure for triage and mortality prediction purposes and universal adaptation to a broad range of trauma populations.

Background: To validate the predictive value of the Glasgow Coma Score (GCS) and find the best way to model the score in a logistic regression model predicting mortality., Methods: Analyses were based on 20,494 patients from the trauma registries of three urban Level I trauma centers in the province of Quebec, Canada. The predictive value of the GCS and its components was evaluated in logistic regression models predicting in-hospital mortality with measures of discrimination and calibration. The performance of the GCS with no transformation and as an ordered categorical variable was compared with two transformation techniques: fractional polynomials and spline regression., Results: The GCS had excellent discrimination (area under Receiving Operator Characteristic Curve=0.833 95% confidence interval=0.820-0.846) but fairly poor calibration (Pearson's Chi-squared statistic=122 on 11 df). The eye component added no predictive information to the verbal and motor components in the whole sample but was important in certain sub-populations. Using the three components separately, rather than the sum, did not improve the predictive model. Fractional polynomial transformation of the GCS improved calibration and spline regression performed even better. GCS modeled as an ordered categorical variable performed badly both in terms of discrimination and calibration., Conclusions: The GCS in its present form is an efficient predictor of in-hospital mortality, which could benefit from statistical transformation in logistic regression models when the accuracy of estimated probabilities of mortality is important. The common use of GCS categories for modeling mortality leads to loss of information and should be discarded.

Background: To validate the accuracy of the Revised Trauma Score (RTS) and its components for predicting in-hospital mortality., Methods: Analyses were based on 22,388 patients from the trauma registries of three urban Level I trauma centers in the province of Quebec, Canada. The accuracy of RTS coded variables for the Glasgow Coma Score (GCSc), Systolic Blood Pressure (SBPc), and Respiratory Rate (RRc) for predicting mortality was evaluated in logistic regression models with measures of discrimination and model fit and compared with Fractional Polynomial (FP) transformations of each component., Results: RTS coded variables were associated with sparse data distributions and did not accurately represent the relation of GCS, SBP, and RR to mortality. FP models were always associated with significantly better discrimination (all p < 0.00001) and model fit. Survival probability estimates generated by the model with FP transformations were significantly different to those generated by the model with RTS-coded variables., Conclusions: The RTS in its present form does not accurately describe the relation of GCS, SBP, and RR to mortality. FP transformation would improve the accuracy of predicted survival probabilities used for performance evaluation and may improve control of confounding caused by of physiologic severity case mix in trauma research.

Background: The goal of this study was to evaluate the impact of different trauma registry exclusion criteria on the assessment of trauma populations and outcome., Methods: All patients admitted to a Canadian regional trauma center from April 1, 1993 to March 31, 2002 with a diagnosis of trauma (ICD-9 codes 800 to 959) were reviewed. TOTAL included everyone. REGISTRY included only patients meeting any of four criteria: death during hospital stay, transfer received from another hospital, admission to the intensive care unit, or hospital stay of 3 days or more. NOHIP excluded patients with isolated hip fracture. REG/NOHIP combined both. ISS12 and ISS15 excluded patients with ISS <12 and 15, respectively., Results: There were 6,839 trauma patients. The percentage of excluded patients by group was: REGISTRY, 21.2%; NOHIP, 14.7%; REG/NOHIP, 34.9%; ISS12, 75%; and ISS15, 80.3%. Median length of stay was 7 days. Exclusions represented a total number of hospitalization days varying from 1.9% to 65.5% of TOTAL. Mortality was 6.9% for TOTAL, 8.6% for REGISTRY (p < 0.001), 5.7% for NOHIP (p = 0.009), 7.5% for REG/NOHIP (p=NS), 16.1% for ISS12 (p < 0.001), and 20.4% for ISS15 (p < 0.001). In groups with exclusions, transfer to long-term care varied from 0.14% to 23.5% in the excluded patients. For rehabilitation, these percentages varied from 0.14% to 17.6%., Conclusions: Registry exclusion criteria significantly alter the apparent severity of injury and resource utilization. The use of divergent exclusion criteria in the analysis of trauma registry data may be misleading.

Background: To investigate whether multiple imputation (MI) of missing Glasgow Coma Scale (GCS) values generates more accurate GCS frequency distributions and less biased parameter estimates in logistic regression models predicting mortality than the standard procedure of excluding observations with missing GCS values., Methods: The study population consisted of 5,065 patients with complete GCS information from the trauma registry of a Level 1 trauma center. Missing GCS values were imposed on the data set, and the performance of MI (extrapolating missing GCS from a data prediction model) and of deleting all data observations with missing GCS (list-wise deletion) were evaluated. GCS and Trauma and Injury Severity Score (TRISS) frequency distributions and parameter estimates were compared with true values from the original data set., Results: GCS and TRISS frequency values generated by MI were much more accurate than those generated by list-wise deletion. GCS and TRISS parameter estimates generated by MI all had acceptable bias and coverage rates when compared with true values. List-wise deletion provided biased parameter estimates for the GCS, the Revised Trauma Score, and the Injury Severity Score., Conclusion: MI is a valid solution to the problem of missing GCS data in trauma research. It allows the conservation of precious data observations and leads to unbiased estimates in consequent analyses. Analyses, which exclude observations with missing GCS data, provide biased results.

Hypothesis: A second injury in the same body region is associated with lower mortality than a second injury in a different body region, independently of injury severity and body region., Methods: The population consisted of 15,200 patients with two or more injuries from level I trauma centers in Quebec. The mortality odds ratio of having a same-region second injury (SR) as opposed to a different-region second injury (DR) was assessed., Results: Patients with a SR had 43% lower odds of mortality when compared to patients with a DR., Conclusion: A second injury in the same body region is associated with lower mortality than a second injury in a different body region.

The movement towards trauma care regionalization in Québec was initiated in 1990, with formal designation of three level I trauma centers in 1993. The purpose of this study is to evaluate the impact of trauma center designation on mortality. The study design is that of a two-cohort study, one assembled during 1987 when designation was not in effect, and the other during the first 5 months of designation. The study focuses on patients that fulfilled the following criteria: i) arrived alive at the hospital, and ii) were admitted. The outcome measures are adjusted mortality, and excess mortality as measured by the TRISS methodology. A total of 158 patients treated in 1987, and 288 treated in 1993, were identified. The mean age of the patients treated in 1993 was significantly higher (40.0, +/- 18.1), when compared with the 1987 group (30.9 +/- 18.1; p < 0.001). Patients in the 1987 cohort had a significantly higher proportion of injuries caused by stabbing (p = 0.02), and a significantly lower proportion caused by falls (p = 0.003). The 1987 cohort had a higher rate of abdominal injuries (p = 0.0001), and external injuries (p = 0.0001), and a significantly lower rate of head or neck injuries (p = 0.003), and injuries to the extremities (p = 0.0001). The mean Injury Severity Score (ISS) for the 1987 cohort was 14.96 (+/- 12.36), and 15.49 (+/- 11.61) in 1993 (p = 0.65). The crude mortality rate was 20% for 1987, and 10% for 1993. The crude odds ratio for mortality in 1987 was 2.10 with 95% confidence intervals between 1.22 and 3.62 (p = 0.006).(ABSTRACT TRUNCATED AT 250 WORDS)