Fiabilité et reproductibilité de la classification de l’American Association for the Surgery of Trauma et impact sur le suivi radiologique des traumatismes rénaux

25 janvier 2018

Auteurs : Q.-B. Phan, E. Mourey, L. Estivalet, B. Delattre, F. Bardet, O. Chevallier, D. Louis, L.S. Aho, R. Loffroy, L. Cormier
Référence : Prog Urol, 2018, 1, 28, 12-17




 




Introduction


Surgical management with a high risk of nephrectomy was previously the main management of renal trauma [1]. In order to avoid renal unit loss and to preserve renal function, conservative management has risen. It was allowed by improvements of interventional radiology, endourology and Computed Tomography (CT) scan.


CT scan is the gold standard for initial evaluation of renal injury in stable [2, 3] or even unstable patients [4]. It allows an accurate assessment of the renal injury type and extension. The American Association for the Surgery of Trauma (AAST) Organ Injury Scale (OIS) is the most used classification to sort kidney injuries (Appendix 1) [5]. This 5-grade scale is a predictive way to assess morbidity, mortality, need for surgery and for nephrectomy with increasing grade [6, 7, 8].


Renal trauma management depends on CT scan findings. Indeed, AAST scale has a central place in modern management algorithms. According to international guidelines, high-grade injuries should be treated, if possible, conservatively and should have a control CT scan 48-96h after the trauma to minimize the risk of missed complications [2, 3].


However, using AAST OIS could be difficult since CT scan analysis is subjective [6, 9]. For example, it could be hard to distinguish a grade V injury from a kidney with multiples grade III or IV injuries (Figure 1).


Figure 1
Figure 1. 

Multiples grade III and IV injuries or grade V injury?




We hypothesize that the AAST grading for renal trauma has significant interpersonal variability. A rating mistake could drive to an inadequate radiologic monitoring and potential morbidity. The principal aim of the study was to assess the subjectivity of AAST OIS rating and the impact on radiologic monitoring.


Material and methods


After obtaining approval from the Institutional Review Board, we studied a retrospective cohort of patients with a blunt renal injury initially evaluated by CT scan, admitted to our university hospital between 2010 and 2015.


To obtain a homogeneous group, we excluded penetrating renal injuries. CT scan imaging (arterial and venous phases, delayed imaging at the discretion of the radiologist) was obtained after injection of intravenous contrast material using a power injector.


Patients were identified through the administrative data and our database. We analyzed the patient's demographics (age, sex, body mass index), trauma mechanism (vehicle crash, pedestrian accident, fall or assault), renal and associated injuries, and radiologic monitoring during hospital stay. A follow-up CT scan was performed in case of clinical signs of complications (e.g., fever, worsening flank pain, ongoing bleeding). If the patient was asymptomatic, a control CT scan was realized five days after the trauma irrespective of renal injury grade. We also analyzed the emergency interventions at admission and the urological complications during the hospital stay.


We used AAST OIS to evaluate the renal injury severity. The initial CT scan was retrospectively analyzed independently by three raters: a resident urologist, a senior urologist skilled in renal pathology and a senior radiologist. The two senior physicians were blind to the clinical data. Each case of grading disagreement was analyzed collegially by the three initial physicians to find a consensus and obtain a final AAST rating.


For statistical analysis, we analyzed the AAST 5 levels grading, then we classified the population into two groups according to the renal injury severity: low-grade group (grades I to III) and high-grade group (grades IV and V) [9, 10, 11, 12, 13] in order to evaluate the radiologic follow-up.


Statistical analysis was performed using Epi Info â„¢ 6.0 (Centers of Disease Control). The reproducibility of AAST scaling was evaluated by Cohen's Kappa coefficient with confidence interval (CI) to measure the inter-rater agreement. The coefficient value is included between 0 and 1. The nearer to 1 is the Kappa coefficient, the higher is the agreement as shown in Appendix 2 [14].


Results


After data collection, 123 patients with blunt renal injury diagnosis were included. We excluded 26 patients for administrative data mistakes in diagnosis code or initial CT scan unavailable. We analyzed 97 patients who presented 101 blunt renal injury. Demographic data, trauma mechanism and emergency interventions are described in Table 1.


The AAST scaling for each of the 3 raters is presented in Table 2. The final scaling found respectively 58.4% and 41.6% of low and high-grade renal injury. The Kappa coefficient was 0.36 [0.24-0.47] between the three raters. The distribution of each rater's AAST scaling has been compared to the final scaling through the Cohen's Kappa coefficient (Table 2).


Kappa coefficient was 0.5 between senior radiologist and senior urologist (CI=[0.40-0.59]); 0.71 between senior urologist and resident urologist (CI=[0.61-0.82]); 0.35 between senior radiologist and resident urologist (CI=[0.26-0.44]).


In severity subgroup analysis, the agreement is higher (Table 3). Kappa coefficient was 0.59 [0.48-0.71] between the three raters; 0.8 between senior radiologist and senior urologist (CI=[0.59-1]); 0.79 between senior urologist and resident urologist (CI=[0.60-0.99]); 0.70 between senior radiologist and resident urologist (CI=[0.50-0.91]).


There was a disagreement in 50 cases (49.5%) between the senior urologist's and the senior radiologist's ratings. The AAST scaling gap was mainly of 1 grade in 82%, 2 grades in 12% and at least 3 grades in 6%. The differences concerned 31 low grade and 19 high-grade injuries. Those disagreements brought to a severity subgroup change in 34% (n =17) in comparison to final scaling and drove to theoretical modification of radiologic monitoring.


Among the 17 cases of severity subgroup change, 4 patients had a urological complication of their trauma. A patient presented a nephritic colic due a ureteral blood clot treated by retrograde ureteral stenting on day 1. Two patients underwent an ongoing blood loss with necessity of angioembolisation (respectively on day 2 and day 17). The last patient presented an unexplained fever on day 4 due to a missed diagnosed injury of collecting system (grade IV injury initially rated grade III) managed non-operatively.


Discussion


CT scan is the gold standard for initial radiographic evaluation of renal trauma [2, 3]. The AAST OIS is the most used classification to assess kidney injury [5] and has a central place in modern management algorithms. Despite the very clear items of the AAST OIS, the grading of renal injuries is subjective and varies among radiologists and urologists as shown by the variation of Cohen's Kappa agreement coefficient between the raters.


Some authors distinguish 2 severity groups: low- and high-grade injury. This difference is associated with an increased morbidity, mortality and need for surgery or nephrectomy in high-grade injuries group. The classification of grade III injuries is still discussed. Some authors consider grade III injuries in this group [15, 16, 17, 18]. However, our high-grade group included only grade IV and V injuries according to recent literature [9, 10, 11, 12, 13].


Accuracy of AAST scaling is fundamental since it has an important prognostic value [6, 7, 8]. AAST scaling influences renal injury management since only high-grade trauma should have a radiologic monitoring [2, 3]. In our cohort, 2 cases of AAST rating discrepancies with severity group change presented a complication potentially missed without a control CT scan. The main risk of an injury-rating mistake is a downstaging. Indeed, it can increase morbidity by missed complications [19, 20] as illustrate in our case of undiagnosed urinoma. An accurate scaling could help to reduce non-necessary radiologic monitoring during the patient stay and avoid potential CT scan complications such as contrast nephropathy, increased radiation and its risk of radio-induced cancers [21, 22].


The heterogeneity of CT scan acquisition (time or protocol) could make the AAST scaling difficult since some CT scan were realized in local hospitals around our institution. For example, delayed imaging was inconstant in our study (50% of CT scans) or in the literature, although it allows a more accurate anatomic grading of renal injuries and a potential targeted radiologic monitoring [23]. In fact, without delayed imaging, a grade III injury can be liken to a grade IV injury as observed in our cohort.


The strength of our study was firstly its triple independent AAST rating with two senior raters blind to the clinical data and secondly the analyze of renal injury scale subjectivity. To our knowledge, only one study to date performed a double grading of the renal injuries [13] but none has quantitatively assessed the AAST OIS subjectivity.


The agreement between the 3 rater's classifications was assessed by the Cohen's Kappa coefficient. It should be read with caution: usually a subjective variable is associated to a lower Kappa coefficient than with an objective variable [14]. It could explain the outcomes of our study quite disappointing with an agreement considered as fair to moderate. The physician experience probably influences the AAST rating quality, since the resident urologist obtained the lowest Kappa coefficient. Since the urologist was skilled in renal pathology, the difference between the senior radiologist and the senior urologist could be explained by subjectivity of CT scan reading and AAST scale interpretation. Despite the very clear items of the AAST OIS, its use could be difficult especially for high-grade injuries. Indeed, this group is heterogeneous: it includes both vascular and parenchymal injuries. The term of shattered kidney is also subjective and could correspond to multiple grade III or IV injuries [6, 9] as shown in Figure 1. The study results had to be considered with caution since the small number of reviewers for the imaging. It is difficult to differentiate intra- and inter-specialty discrepancies and to conclude a definite subjectivity of the AAST OIS. An additional study with more CT scan raters would be helpful.


Buckley and McAninch proposed a revision of AAST injury scaling in order to standardize the renal injury management [9]. Grades I to III remain the same as in the original classification. The new grade IV includes segmental vascular injuries and all renal collecting system injuries. The new grade V includes only main renal pedicle injury. This revised classification is based on the life threatening of main vascular injuries with a higher exploration rate and a lower renal salvage rate but this classification has not been validated yet.


In the same way, Altmann et al. suggested to create a grade VI for main pedicle injuries or renal injuries in hemodynamically unstable patients since their management is specific [24].


Some authors suggested a substratification of intermediate injuries (grade III and IV) according to the initial CT scan outcomes: perirenal hematoma size, intravascular contrast extravasation and renal laceration location [12, 16]. They calculated a Renal Trauma Risk Score (RTRS) to distinguish grade IVa injury likely to be managed conservatively and grade IVb injury likely to need an intervention (angiographic embolization or surgery) [12].


Our study has a selection bias due to its retrospective nature and its dependence on administrative data quality. The triple scaling was realized retrospectively for study purposes since AAST scaling was usually not mentioned in CT scan report at the time of admission though it should be the standard.


Our single center study also generate a local selection bias. High-grade injuries could be over-represented in our cohort, since our hospital is the regional reference center for severe polytrauma patients and emergency interventional radiology. However, our AAST grading is similar to other French studies [18, 25]. The comparison to North American studies is difficult since they include much more penetrating injuries which are quite uncommon in Europe [7, 8, 13].


In the international guidelines, only high-grade injuries should have a control CT scan 48-96h after the trauma [2, 3]. The impact of renal injury grading disagreement on radiologic monitoring was difficult to assess in our study, as we could not evaluate the frequency of missed complications since our institution protocol used to recommend so far a control CT scan five days after trauma irrespective of renal injury grade. However, we noted that AAST rating disagreements brought to a severity group change in 34% (n =17) and therefore to a potential change of radiologic monitoring. We strongly suggest a close collaboration between radiologist and urologist for AAST grading in order to minimize the grading mistakes and to realize an appropriate management and follow-up of renal injuries and a decrease in unnecessary radiological exams.


Conclusion


Our study demonstrates that AAST grading of renal injuries is subjective and suffers from inter-rater reliability. Our study strongly suggests the need of an accurate and collegial grading between radiologists and urologists. The optimal staging of renal trauma may lead to reduce the morbidity of unseen complications and the iatrogenic risk of useless imaging due to a staging mistake.


Disclosure of interest


The authors declare that they have no competing interest.



Appendix 1. AAST OIS classification [5]





Grade Description
- Contusion: Microscopic or gross hematuria, normal urological studies
- Subcapsular hematoma, nonexpanding without parenchymal
laceration 
II  - Nonexpanding perirenal hematoma confined to renal Retroperitoneum
- Laceration < 1.0 cm parenchymal depth of renal cortex without
urinary extravasation 
III  Laceration > 1.0 cm parenchymal depth of renal cortex, without
Collecting system rupture or urinary extravasation 
IV  - Parenchymal laceration extending through the renal cortex, medulla, and collecting system
- Vascular Main renal artery or vein injury with contained hemorrhage 
- Completely shattered kidney
- Vascular Avulsion of renal hilum which devascularizes kidney 





Appendix 2. Interpretation of Cohen's Kappa coefficient value





Kappa value Agreement
No agreement 
0.01-0.2  None to slight 
0.21-0.40  Fair 
0.41-0.60  Moderate 
0.61-0.80  Substantial 
0.81-0.99  Almost perfect 
Perfect 






Table 1 - Demographic data, trauma mechanism and emergency treatment.
  Total n (%) 
n   97 
Age (years)   34.5 (2-86) 
BMI (kg/m 2 )   23.1 (14-42) 
Sex    
Men  71 (73%) 
Women  26 (27%) 
Mechanism    
Vehicle crash  56 (58%) 
Car  17 (18%) 
Two-wheeler  36 (37%) 
Pedestrian accident  3 (3%) 
Fall  31 (32%) 
Assault  10 (10%) 
Emergency procedure    
Renal exploration 
Nephrectomy 
Endovascular treatment  10 (10%) 
Proximal embolization  3 (3%) 



Légende :
BMI: body mass index.



Table 2 - Inter-rater variability of AAST scaling.
AAST rater  AAST grade n (%) 
Agreement Kappa coefficient (CI) 
  II  III  IV   
Senior radiologist  7 (6.9%)  23 (22.8%)  35 (34.7%)  17 (16.8%)  19 (18.8%)  0.73 [0.63-0.83] 
Senior urologist  21 (20.8%)  14 (13.9%)  19 (18.8%)  37 (36.6%)  10 (9.9%)  0.55 [0.45-0.65] 
Resident urologist  22 (21.8%)  18 (17.8%)  30 (29.7%)  28 (27.7%)  3 (3.0%)  0.46 [0,36-0,56] 
Final scaling  12 (11.9%)   18 (17.8%)   29 (28.7%)   24 (23.8%)   18 (17.8%)   0.36 [0.25-0.47] 



Légende :
CI: confidence interval.



Table 3 - Inter-rater variability of AAST scaling in severity subgroup analyze.
AAST rater  AAST severity subgroup 
Agreement Kappa coefficient (CI) 
  Low grade  High grade   
Senior radiologist  65 (64.4%)  36 (35.6%)  0.85 [0.66-1,04] 
Senior urologist  54 (53.5%)  47 (46.5%)  0.78 [0.59-0.97] 
Resident urologist  70 (69.3%)  31 (30.7%)  0.72 [0.54-0.91] 
Final scaling  59 (58.4%)   42 (41.6%)   0.59 [0.48-71] 



Légende :
CI: confidence interval.



Grade  Description 
- Contusion: Microscopic or gross hematuria, normal urological studies
- Subcapsular hematoma, nonexpanding without parenchymal
laceration 
II  - Nonexpanding perirenal hematoma confined to renal Retroperitoneum
- Laceration < 1.0 cm parenchymal depth of renal cortex without
urinary extravasation 
III  Laceration > 1.0 cm parenchymal depth of renal cortex, without
Collecting system rupture or urinary extravasation 
IV  - Parenchymal laceration extending through the renal cortex, medulla, and collecting system
- Vascular Main renal artery or vein injury with contained hemorrhage 
- Completely shattered kidney
- Vascular Avulsion of renal hilum which devascularizes kidney 





Kappa value  Agreement 
No agreement 
0.01-0.2  None to slight 
0.21-0.40  Fair 
0.41-0.60  Moderate 
0.61-0.80  Substantial 
0.81-0.99  Almost perfect 
Perfect 




References



Wessells H., Suh D., Porter J.R., Rivara F., MacKenzie E.J., Jurkovich G.J., et al. Renal injury and operative management in the United States: results of a population-based study J Trauma 2003 ;  54 : 423-430 [cross-ref]
Morey A.F., Brandes S., Dugi D.D., Armstrong J.H., Breyer B.N., Broghammer J.A., et al. Urotrauma: AUA guideline J Urol 2014 ;  192 : 327-335 [cross-ref]
D.J. Summerton, N. Djakovic, N.D. Kitrey, F.E. Kuehhas,. Urological Trauma. Uroweb. Available from: urological-trauma/.
Fu C.-Y., Yang S.-J., Liao C.-H., Lin B.-C., Kang S.-C., Wang S.-Y., et al. Hypotension does not always make computed tomography scans unfeasible in the management of blunt abdominal trauma patients Injury 2015 ;  46 : 29-34 [cross-ref]
Moore E.E., Shackford S.R., Pachter H.L., McAninch J.W., Browner B.D., Champion H.R., et al. Organ injury scaling: spleen, liver, and kidney J Trauma 1989 ;  29 : 1664-1666 [cross-ref]
Santucci R.A., McAninch J.W., Safir M., Mario L.A., Service S., Segal M.R. Validation of the American Association for the Surgery of Trauma organ injury severity scale for the kidney J Trauma 2001 ;  50 : 195-200 [cross-ref]
Shariat S.F., Roehrborn C.G., Karakiewicz P.I., Dhami G., Stage K.H. Evidence-based validation of the predictive value of the American Association for the Surgery of Trauma kidney injury scale J Trauma 2007 ;  62 : 933-939 [cross-ref]
Kuan J.K., Wright J.L., Nathens A.B., Rivara F.P., Wessells H.American Association for the Surgery of Trauma American Association for the Surgery of Trauma Organ Injury Scale for kidney injuries predicts nephrectomy, dialysis, and death in patients with blunt injury and nephrectomy for penetrating injuries J Trauma 2006 ;  60 : 351-356 [cross-ref]
Buckley J.C., McAninch J.W. Revision of current American Association for the Surgery of Trauma Renal Injury grading system J Trauma 2011 ;  70 : 35-37 [cross-ref]
Hotaling J.M., Sorensen M.D., Smith T.G., Rivara F.P., Wessells H., Voelzke B.B. Analysis of diagnostic angiography and angioembolization in the acute management of renal trauma using a national data set J Urol 2011 ;  185 : 1316-1320 [cross-ref]
Moudouni S.M., Hadj Slimen M., Manunta A., Patard J.J., Guiraud P.H., Guille F., et al. Management of major blunt renal lacerations: is a nonoperative approach indicated? Eur Urol 2001 ;  40 : 409-414 [cross-ref]
Dugi D.D., Morey A.F., Gupta A., Nuss G.R., Sheu G.L., Pruitt J.H. American Association for the Surgery of Trauma grade 4 renal injury substratification into grades 4a (low risk) and 4b (high risk) J Urol 2010 ;  183 : 592-597 [cross-ref]
Malcolm J.B., Derweesh I.H., Mehrazin R., DiBlasio C.J., Vance D.D., Joshi S., et al. Nonoperative management of blunt renal trauma: is routine early follow-up imaging necessary? BMC Urol 2008 ;  8 : 11 [cross-ref]
McHugh M.L. Interrater reliability: the kappa statistic Biochem Medica 2012 ;  22 : 276-282 [cross-ref]
Sarani B., Powell E., Taddeo J., Carr B., Patel A., Seamon M., et al. Contemporary comparison of surgical and interventional arteriography management of blunt renal injury J Vasc Interv Radiol 2011 ;  22 : 723-728 [cross-ref]
Nuss G.R., Morey A.F., Jenkins A.C., Pruitt J.H., Dugi D.D., Morse B., et al. Radiographic predictors of need for angiographic embolization after traumatic renal injury J Trauma 2009 ;  67 : 578-582 [cross-ref]
McGuire J., Bultitude M.F., Davis P., Koukounaras J., Royce P.L., Corcoran N.M. Predictors of outcome for blunt high-grade renal injury treated with conservative intent J Urol 2011 ;  185 : 187-191 [cross-ref]
Charbit J., Manzanera J., Millet I., Roustan J.-P., Chardon P., Taourel P., et al. What are the specific computed tomography scan criteria that can predict or exclude the need for renal angioembolization after high-grade renal trauma in a conservative management strategy? J Trauma 2011 ;  70 : 1219-1228
Mulligan J.M., Cagiannos I., Collins J.P., Millward S.F. Ureteropelvic junction disruption secondary to blunt trauma: excretory phase imaging (delayed films) should help prevent a missed diagnosis J Urol 1998 ;  159 : 67-70 [cross-ref]
Brown S.L., Hoffman D.M., Spirnak J.P. Limitations of routine spiral computerized tomography in the evaluation of blunt renal trauma J Urol 1998 ;  160 : 1979-1981 [cross-ref]
Brenner D.J., Hall E.J. Computed tomography - an increasing source of radiation exposure N Engl J Med 2007 ;  357 : 2277-2284 [cross-ref]
Maeder M., Klein M., Fehr T., Rickli H. Contrast nephropathy: review focusing on prevention J Am Coll Cardiol 2004 ;  44 : 1763-1771 [cross-ref]
Blankenship J.C., Gavant M.L., Cox C.E., Chauhan R.D., Gingrich J.R. Importance of delayed imaging for blunt renal trauma World J Surg 2001 ;  25 : 1561-1564 [cross-ref]
Altman A.L., Haas C., Dinchman K.H., Spirnak J.P. Selective nonoperative management of blunt grade 5 renal injury J Urol 2000 ;  164 : 27-31 [cross-ref]
Saidi A., Bocqueraz F., Descotes J.-L., Cadi P., Terrier N., Boillot B., et al. [Blunt kidney trauma: a ten-year experience] Prog Urol 2004 ;  14 : 1125-1131






© 2017 
Elsevier Masson SAS. Tous droits réservés.