Staying on target: Maintaining a balanced resuscitation during damage-control resuscitation improves survival

BACKGROUND

Damage-control resuscitation (DCR) improves survival in severely bleeding patients. However, deviating from balanced transfusion ratios during a resuscitation may limit this benefit. We hypothesized that maintaining a balanced resuscitation during DCR is independently associated with improved survival.

METHODS

This was a secondary analysis of the Prospective Observational Multicenter Major Trauma Transfusion (PROMMTT) study. Patients receiving >3 U of packed red blood cells (PRBCs) during any 1-hour period over the first 6 hours and surviving beyond 30 minutes were included. Linear regression assessed the effect of percent time in a high-ratio range on 24-hour survival. We identified an optimal ratio and percent of time above the target ratio threshold by Youden's index. We compared patients with a 6-hour ratio above the target and above the percent time threshold (on-target) with all others (off-target). Kaplan-Meier analysis assessed the combined effect of blood product ratio and percent time over the target ratio on 24-hour and 30-day survival. Multivariable logistic regression identified factors independently associated with 24-hour and 30-day survival.

RESULTS

Of 1,245 PROMMTT patients, 524 met the inclusion criteria. Optimal targets were plasma/PRBC and platelet/PRBC of 0.75 (3:4) and ≥40% time spent over this threshold. For plasma/PRBC, on-target (n = 213) versus off-target (n = 311) patients were younger (median, 31 years; interquartile range, [22-50] vs. 40 [25-54]; p = 0.002) with similar injury burdens and presenting physiology. Similar patterns were observed for platelet/PRBC on-target (n = 116) and off-target (n = 408) patients. After adjusting for differences, on-target plasma/PRBC patients had significantly improved 24-hour (odds ratio, 2.25; 95% confidence interval, 1.20-4.23) and 30-day (odds ratio, 1.97; 95% confidence interval, 1.14-3.41) survival, while on-target platelet/PRBC patients did not.

CONCLUSION

Maintaining a high ratio of plasma/PRBC during DCR is independently associated with improved survival. Performance improvement efforts and prospective studies should capture time spent in a high-ratio range.

LEVEL OF EVIDENCE

Epidemiologic/prognostic study, level II; Therapeutic, level IV.

The impact of missing trauma data on predicting massive transfusion

BACKGROUND

Missing data are inherent in clinical research and may be especially problematic for trauma studies. This study describes a sensitivity analysis to evaluate the impact of missing data on clinical risk prediction algorithms. Three blood transfusion prediction models were evaluated using an observational trauma data set with valid missing data.

METHODS

The PRospective Observational Multicenter Major Trauma Transfusion (PROMMTT) study included patients requiring one or more unit of red blood cells at 10 participating US Level I trauma centers from July 2009 to October 2010. Physiologic, laboratory, and treatment data were collected prospectively up to 24 hours after hospital admission. Subjects who received 10 or more units of red blood cells within 24 hours of admission were classified as massive transfusion (MT) patients. Correct classification percentages for three MT prediction models were evaluated using complete case analysis and multiple imputation. A sensitivity analysis for missing data was conducted to determine the upper and lower bounds for correct classification percentages.

RESULTS

PROMMTT study enrolled 1,245 subjects. MT was received by 297 patients (24%). Missing percentage ranged from 2.2% (heart rate) to 45% (respiratory rate). Proportions of complete cases used in the MT prediction models ranged from 41% to 88%. All models demonstrated similar correct classification percentages using complete case analysis and multiple imputation. In the sensitivity analysis, correct classification upper-lower bound ranges per model were 4%, 10%, and 12%. Predictive accuracy for all models using PROMMTT data was lower than reported in the original data sets.

CONCLUSION

Evaluating the accuracy clinical prediction models with missing data can be misleading, especially with many predictor variables and moderate levels of missingness per variable. The proposed sensitivity analysis describes the influence of missing data on risk prediction algorithms. Reporting upper-lower bounds for percent correct classification may be more informative than multiple imputation, which provided similar results to complete case analysis in this study.

Waiver of consent in noninterventional, observational emergency research: the PROMMTT experience

BACKGROUND

In the PRospective Observational Multicenter Major Trauma Transfusion (PROMMTT) study, waiver of consent was used because previous literature reported low response rates and subsequent bias. The goal of this article was to examine the rationale and tradeoffs of using waiver of consent in PROMMTT.

METHODS

PROMMTT enrolled trauma patients receiving at least 1 U of red blood cells within 6 hours after admission at 10 US Level 1 trauma centers. Local institutional review boards (IRBs) from all sites approved the study. Site 8 was required by their IRB to attempt consent but was allowed to retain data on patients unable to be consented.

RESULTS

Of 121 subjects enrolled at Site 8, 55 consents were obtained (46%), and no patient or legally authorized representative refused to give consent. Of the patients, 36 (30%) died, and 6 (5%) were discharged before consent could be attempted. Consent was attempted but not possible among 24 patients (20%). Of the 10 clinical sites, 6 of the local IRBs approved collection of residual blood samples, 1 had previous approval to collect timed blood samples under a separate protocol, and 3 reported that their local IRBs would not approve collection of residual blood under a waiver of consent.

CONCLUSION

Waiver of consent was used in PROMMTT because of the potential adverse impact of consent refusals; however, there were no refusals. If the IRB for Site 8 had required withdrawal of patients unable to consent and destruction of their data, a serious bias would likely have been introduced. Other tradeoffs included a reduction in sites participating in residual blood collection and a smaller than expected amount of residual blood collected among sites operating under a waiver of consent. Noninterventional emergency research studies should consider these potential tradeoffs carefully before deciding whether waiver of consent would best achieve the goals of a study.

Time-dependent prediction and evaluation of variable importance using superlearning in high-dimensional clinical data

BACKGROUND

Prediction of outcome after injury is fraught with uncertainty and statistically beset by misspecified models. Single-time point regression only gives prediction and inference at one time, of dubious value for continuous prediction of ongoing bleeding. New statistical machine learning techniques such as SuperLearner (SL) exist to make superior prediction at iterative time points while evaluating the changing relative importance of each measured variable on an outcome. This then can provide continuously changing prediction of outcome and evaluation of which clinical variables likely drive a particular outcome.

METHODS

PROMMTT data were evaluated using both naive (standard stepwise logistic regression) and SL techniques to develop a time-dependent prediction of future mortality within discrete time intervals. We avoided both underfitting and overfitting using cross validation to select an optimal combination of predictors among candidate predictors/machine learning algorithms. SL was also used to produce interval-specific robust measures of variable importance measures (VIM resulting in an ordered list of variables, by time point) that have the strongest impact on future mortality.

RESULTS

Nine hundred eighty patients had complete clinical and outcome data and were included in the analysis. The prediction of ongoing transfusion with SL was superior to the naive approach for all time intervals (correlations of cross-validated predictions with the outcome were 0.819, 0.789, 0.792 for time intervals 30-90, 90-180, 180-360, >360 minutes). The estimated VIM of mortality also changed significantly at each time point.

CONCLUSION

The SL technique for prediction of outcome from a complex dynamic multivariate data set is superior at each time interval to standard models. In addition, the SL VIM at each time point provides insight into the time-specific drivers of future outcome, patient trajectory, and targets for clinical intervention. Thus, this automated approach mimics clinical practice, changing form and content through time to optimize the accuracy of the prognosis based on the evolving trajectory of the patient.

Defining when to initiate massive transfusion: a validation study of individual massive transfusion triggers in PROMMTT patients

BACKGROUND

Early predictors of massive transfusion (MT) would prevent undertriage of patients likely to require MT. This study validates triggers using the Prospective Observational Multicenter Major Trauma Transfusion (PROMMTT) study.

METHODS

All enrolled patients in PROMMTT were analyzed. The initial emergency department value for each trigger (international normalized ratio [INR], systolic blood pressure, hemoglobin, base deficit, positive result for Focused Assessment for the Sonography of Trauma examination, heart rate, temperature, and penetrating injury mechanism) was compared for patients receiving MT (≥ 10 U of packed red blood cells in 24 hours) versus no MT. Adjusted odds ratios (ORs) for MT are reported using multiple logistic regression. If all triggers were known, a Massive Transfusion Score (MTS) was created, with 1 point assigned for each met trigger.

RESULTS

A total of 1,245 patients were prospectively enrolled with 297 receiving an MT. Data were available for all triggers in 66% of the patients including 67% of the MTs (199 of 297). INR was known in 87% (1,081 of 1,245). All triggers except penetrating injury mechanism and heart rate were valid individual predictors of MT, with INR as the most predictive (adjusted OR, 2.5; 95% confidence interval, 1.7-3.7). For those with all triggers known, a positive INR trigger was seen in 49% receiving MT. Patients with an MTS of less than 2 were unlikely to receive MT (negative predictive value, 89%). If any two triggers were present (MTS ≥ 2), sensitivity for predicting MT was 85%. MT was present in 33% with an MTS of 2 greater compared with 11% of those with MTS of less than 2 (OR, 3.9; 95% confidence interval, 2.6-5.8; p < 0.0005).

CONCLUSION

Parameters that can be obtained early in the initial emergency department evaluation are valid predictors for determining the likelihood of MT.

LEVEL OF EVIDENCE

Diagnostic, level II.