Page 97 - Journal of Special Operations Medicine - Summer 2015

Figure 1 Injury scoring processes. physicians with clinical responsibility for patients would
differ from JTTR entries resulting from chart review.

Methods
Setting
The Joint Force Medical Group Hospital at Camp Bas-
tion was a UK-led, coalition medical facility in Hel-
mand Province in southern Afghanistan, which provided
medical support to counterinsurgency operations by the
International Security Assistance Force. Military wound-
ing patterns dominated admissions to this facility, with
75% of admissions with NISS ≥16 injured by explosive
Note: AIS(M), Abbreviated Injury Score (Military version); ISS, Injury mechanisms and 23% by gunshot; 2% were injured by
Severity Score; NISS, New Injury Severity Score.
the blunt trauma more typical of civilian injuries. 19
the relationship between these scoring systems. NISS has
been shown to be a more accurate predictor of mortal- The UK JTTR
ity and morbidity in various clinical settings, 9–13 and it is The UK JTTR is a prospectively gathered database of
suggested that NISS reflects military injury burden more all patients (regardless of affiliation) who trigger activa-
accurately than ISS, especially if derived from AIS(M). tion of the trauma team at a deployed UK field hospital
14
20
and of all UK Service personnel evacuated to the UK.
Previous studies have demonstrated large variation in Trauma nurse coordinators (TNCs) record and code in-
the number of injuries identified per case in civilian juries after treatment at the deployed facility, based on
15
trauma. Compared with a reference set of cases, up chart review and imaging. ISS and NISS are calculated
to 31% of injuries may be missed by trained coders. from AIS(M).
16
Agreement for specific codes assigned to individual inju-
ries may be as low as 39%. 16,17 This variation may lead Study Design
to limited reproducibility of estimation of overall injury In this preliminary, prospective, observational study,
burden, with one study having found the probability of physicians coded a patient’s injuries immediately after
any two raters assigning the same ISS being as low as treating them. Rather than investigating possible differ-
28% (rising to only 51% agreement if cases are placed ences in medical and nursing interpretation of written
in bands of severity). However, variation in coding of patient records, this design allowed the authors to in-
18
specific injuries may not compromise the reproducibil- vestigate whether the definitive JTTR record accurately
ity of overall measures of trauma burden: in a study of recorded the injuries identified by the physicians.
six trained raters, despite only 39% agreement for al-
located AIS codes, the intraclass correlation coefficient Selection Criteria
(ICC) for ISS was “almost perfect (ICC 0.90).” In con- Patients were included in the study if they triggered
17
trast, a study of 10 raters found that the limits of agree- trauma team activation (any casualty assigned the high-
ment for every pairing of raters exceeded “clinically est triage category or for whom team activation was ini-
acceptable” bounds (defined by the authors as ±9 ISS tiated by any member of staff). Cases were excluded if
or NISS units). Despite these limitations, ISS remains they had not suffered injury (e.g., a patient allocated a
16
the most commonly reported measure of injury sever- high triage category due to a medical emergency).
ity in the trauma literature, and AIS(M), ISS, and NISS
are routinely recorded in the JTTR. Ultimately, all three Registration and Approval
measures depend on correct identification and coding This study was authorized by and registered with the
of individual injuries. The perspective of a surgeon who Royal Centre for Defence Medicine (RCDM/Res/au-
has seen an injury firsthand and that of a rater scoring dit/1036/12/0264) and with the US Army Joint Combat
from chart review may differ. This could lead to differ- Casualty Research Team as a performance improvement
ent scores being assigned to the same injury. project for which institutional review board approval
was not required. We adhered to the Guidelines for Re-
The aim of this study was to compare AIS(M) scores and porting Reliability and Agreement Studies. 21
derivatives assigned contemporaneously by the treat-
ing clinicians at a UK deployed military medical facility Statistics
against entries for the same trauma casualties recorded in The ISS and NISS generated by each method were found
the JTTR. Based on anecdotal and personal experience of to be asymmetrically distributed around their medians
working with JTTR data, we hypothesized that scoring by by using the Shapiro–Wilk test. Therefore, medians were

Variability in Injury Severity Scoring After Combat Trauma 87