The magnitude of missing data (MD) in trauma registries (TR) stands out as a main limitation for inter-institutional benchmarking. In Argentina, Fundación Trauma developed in 2009 a TR as part of an inclusive trauma program, which is currently active in 11 hospitals from Buenos Aires province.
The objective of this communication is to describe MD patterns in systolic arterial pressure (SAP), respiratory rate (RR) and Glasgow coma scale (GCS) at admission with data from TR and the strategies for dealing with this issue. Materials and methods: Observational study using de-identified data from patients prospectively included in the TR. TR includes codification tools for AIS, ICD, RTS, ISS, and TRISS.
Results: 1 2675 c ases w ere i dentified, o ut o f w hich 11886 were complete. After excluding: (1) Patients <16 years old, (2) cases not evaluable by ISS and (3) cases transferred out, 6050 remained the subject of interest.
Systolic arterial pressure was the variable with less MD (20.6%) and RR the one with the highest proportion of MD (31.6%). In 13% the three variables were incomplete and in 62% there was no MD. MD proportion was higher for women, subjects > 65y, patients with limb injuries, patients with an ISS <16 and with assault as injury mechanism (all p < 0.001).
We found no pattern linking MD and hospitals relating to the volume of admissions or injury severity. Little’s MCAR test was performed on MD with a p-value < 0.001, so multiple imputation was not a choice for further analysis.
Logistic regression analysis showed that to have at least one injury in any of the three cavities (thorax, head or abdomen) and being injured as a motorcyclist was associated to a lower proportion of MD.
Conclusion: The main finding from the analysis of this sample was the huge heterogeneity of MD between hospitals, which oscillated between 4% and 70% for RR (the variable with the highest proportion of MD) showing no identifiable profile. We believe that beyond the identification of a pattern that could adjust the whole sample, the main contribution of this investigation is the identification of each hospital’s pattern of MD to help in the development of a target-directed approach for the reduction of MD.
Keywords: Hospital mortality, Injury severity score, Registries, Trauma, Wounds, and injuries.