The Abbreviated Injury Scale (AIS) is an anatomic scale devised and owned by the Association for the Advancement of Automotive Medicine (AAAM). The scale, first introduced in 1971 and periodically revised, groups injuries by body region (eg, head, thorax, extremity, etc) and rates specific injuries according to a 6-point score: 1 = minor, 2 = moderate, 3 = serious, 4 = severe, 5 = critical, 6 = fatal or untreatable. The AIS forms the basis of the Injury Severity Score (see below). Organized into 9 “chapters” or body regions (head, face, neck, thorax, abdomen/pelvis, spine, upper extremities, lower extremities, burns/other trauma), the “AIS dictionary” gives structured injury descriptions for each region. Importantly, the AIS does not include complications (eg, pneumonia) or other outcomes (eg, death, hearing loss, etc) as injuries. Not until 1990 was any account for age or size included in the AIS descriptors when “Age <15” was added as a descriptor to some injuries.
The AIS is a proprietary injury measurement tool, and while it forms the basis of the ISS, trauma surgeons are taught and use the OIS clinically. The OIS was developed when in 1987, the AAST appointed a committee of trauma surgeons tasked to develop a comprehensive scale of specific organ injuries. These organ-specific scales grade injuries, as with the AIS, as 1 = minimal, 2 = mild, 3 = moderate, 4 = severe, 5 = massive, and 6 = lethal. The main difference between the OIS and the AIS was an attempt to describe a magnitude of anatomic disruption, excluding any estimate of blood loss or effect of intervention. In children, the OIS is most useful as a guide to protocol-driven, non operative management of solid organ injuries, particularly spleen, liver, and kidney. While the OIS does predict some outcome, these ordinal scales are non monotonic: As the renal OIS demonstrates, there is a much larger difference in risk of nephrectomy between grades 3 and 4 than between grades 2 and 3 (Fig. 77-9).
The OIS is only an ordinal scale, not an interval scale. The risk of nephrectomy increases sharply above grade 3.
The Anatomic Profile (AP) was introduced in 1990 in response to perceived weaknesses in the AIS and, more particularly, in the ISS (described below), but is still derived from the AIS. In general, the AP uses a 4-component model emphasizing the brain/spinal cord, thorax/neck, and all other remaining injuries (with the fourth component a summary of minor injuries, typically neglected as having little predictive power). The AP is the square root of the sum of the squares of these components, similar to (but not exactly the same as) the square root of the ISS, and was posited as being analogous to the resultant or norm of several vectors. The AP is a score, and carries predictive information that forecasts mortality when mapped to a logistic regression model of survival probability. It is rarely used in pediatric trauma, but does form an essential component of another composite score, A Severity Characterization of Trauma (ASCOT) (see below).
ICD-9 Codes and ICD-9 Derived Scores
The International Classification of Diseases, version 9, is essentially a nominal scale, with codes that roughly map to OIS and AIS grades (see Table 77-3). The American College of Surgeons now requires International Classification of diseases ninth Edition (ICD-9-CM) codes for injury coding in the National Trauma Databank, offering a chance to develop new injury scoring methods from large computerized database. The ICD-9 Injury Severity Score (ICISS) was developed by first calculating survival risk ratios (SRR) for every injury type described in the ICD-9. Each individual SRR is calculated by dividing the number of survivors by the total number of patients with a particular ICD-9-CM. The ICISS for any individual patient is then calculated as the product of all survival risk ratios for an individual patient's traumatic ICD-9 codes (SRR1 × SRR2 × SRR3 × ⋯ × SRRlast injury). This score is reported to yield better survival prediction than the ISS, while avoiding the numerical curiosities of the ISS.
Table 77-3Injury Scaling Methods Comparison |Favorite Table|Download (.pdf) Table 77-3 Injury Scaling Methods Comparison
|Comparison of Spleen Injury Scales |
|Injury Description ||Injury Type ||OIS ||AIS-90 ||ICD-9 |
|Subcapsular, <10% of surface area ||Hematoma ||I ||2 || |
|Capsular tear, <1 cm of parenchymal depth ||Laceration ||I ||2 || |
Subcapsular, 10% to 50% of surface area
Intraparenchymal, <5 cm in diameter
|Hematoma ||II ||2 || |
|Capsular tear, 1 to 3 cm parenchymal depth that does not involve a trabecular vessel ||Laceration ||II ||2 || |
Subcapsular, >50% of surface area;
Ruptured supcapsular or parenchymal
Intraparenchymal ≥5 cm or expanding
|Hematoma ||III ||3 || |
|>3 cm parenchymal depth, or involving trabecular vessels ||Laceration ||III ||3 || |
|Laceration involving segmental or hilar vessels producing major devascularization (>25% of spleen) ||Laceration ||IV ||4 || |
|Completely shattered spleen ||Laceration ||V ||5 ||865.04 |
|Hilar vascular injury which devascularizes spleen ||Vascular ||V ||5 ||865.14 |
The ICISS appears to give good results in children (and functions well as a hospital benchmarking tool), but a simpler method (Single Worst Injury, SWI) predicts outcome by taking the survival risk ratio of the individual child's single worst (ICD-9 described) injury. Since release of the ICISS, the Osler group reported their updated method, the Trauma Mortality Prediction Model (TMPM-ICD9), based on a repeated probit regression algorithm to yield a probability of death. In particular, the TMPM-ICD9 maps the 5 worst injuries (or really, the severity measures of the ICD-9 codes of the 5 worst injuries) onto a standard cumulative distribution function. It has not yet been validated in pediatric patients, but promises accurate mortality prediction in a way that leverages billing data already being collected nationwide. Another interesting approach based on ICD-9 criteria is the Bayesian Logistic Injury Severity Score (BLISS). In general, it is expected that ICD-9 based systems will gain in importance, not only because they give monotonic predictions of mortality (unlike older methods like ISS; see below), but because the American College of Surgeons has adopted the ICD-9 coding as the National Trauma Data Standard (Table 77-3).
The ISS is one of the most commonly used measures of injury severity, despite its oddities. Originally conceived to describe blunt trauma (automatically giving it a theoretical advantage for use in children where blunt trauma is predominant), revisions to the AIS dictionary improved the performance of the ISS in penetrating trauma. The ISS is calculated as
where A, B, and C are the 3 highest AIS values from different body regions. In general, an ISS > 15 is considered “severe,” that is, associated with a high mortality risk. An AIS code of 6 for any body region automatically raises the ISS to the maximum value of 75. The ISS does not allow measurement of multiple injuries to any single region, and for this reason, the New Injury Severity Score (NISS) was introduced. The NISS is defined as the sum of the squares of the Abbreviated Injury Scale scores of each of a patient's 3 most severe AIS injuries even if they are in the same body region. However, the NISS does not appear to predict mortality much better than the ISS, and if anything violates assumptions of monotonicity even more than the ISS (see below).
Several authors have pointed out other flaws in the ISS. Chief among these is the nonmonotonic relationship of ISS scores to actual mortality. In part, this is because of the odd sum-of-squared-scores method that excludes a number of values (yielding not 75, but only 44 discrete values). Worse, the values are not evenly distributed across the scale (clustering most reachable values below 50, with just 6 above 50). Meanwhile, because the ISS allows some values to be reached in several ways, there is both information loss, and an implication that there is equal mortality for these combinations, something that has not been shown. For these reasons, the ISS is plainly not an interval measure, and really follows something less than an ordinal scale. Moreover, the ISS frequently overestimates the mortality risk in children. For example, a child with a pulmonary contusion (AIS 4) automatically has an ISS of at least 16 and a high implied mortality, yet this injury is often revealed on now-common CT scans, and children typically recover.