Post-mortem computed tomography in the assessment of fatal child abuse

Abstract

Introduction: The main purpose of forensic pathology is to determine the cause and manner of death of deceased persons, evaluate the fatality of injuries sustained, and methodically document, analyse and synthesise findings in a comprehensible way for presentation in a court of law. Worldwide, there has been a decline in autopsy rates due to cultural and religious objection to the conventional autopsy. Clinicians had to find alternative ways to establish the cause of death. The increased application of post-mortem imaging in forensic pathology has resulted in researchers exploring the suitability of both post-mortem Computed Tomography (PMCT) and post-mortem Magnetic Resonance Imaging (PMMRI) to establish the cause of death, determine injuries sustained and whether this imaging modality can replace the autopsy in certain cases. The main objectives of this research study were: to compare the degree of concordance between PMCT and the forensic autopsy in terms of the spectrum and anatomical location of injuries identified; to analyse the injury categories not readily diagnosed by both examinations; to ascertain whether PMCT can accurately establish the cause of death; and to explore whether selected forensic cases can undergo PMCT in the absence of the forensic autopsy. Methods: Thirty children, all younger than 18 years with a history of suspected fatal child abuse or sudden unexplained death, underwent a whole-body PMCT examination using a 64 slice Computed Tomography (CT) scanner, followed by a forensic autopsy. All PMCT studies were reviewed independently by two paediatric radiologists (hereafter termed PMCT reviewers). Both reviewers were blinded to the forensic autopsy findings. The forensic autopsies were conducted by experienced forensic pathologists who were blinded to the PMCT findings. The radiology and forensic pathology reports were analysed using the International Classifications of Disease 10 codes for the respective injury types identified. Injury types identified were divided into five anatomical regions: head and neck; thorax; abdomino-pelvic cavity; spinal column and spinal cord; extremities and one miscellaneous group and comparisons made. Injury types were further grouped into nine common categories and also analysed. Percent agreement was used to establish the degree of concordance between PMCT and the forensic autopsy for injury types diagnosed. The Cohen-Kappa statistic was used to measure interrater reliability between PMCT reviewers and the forensic autopsy to assign a cause of death. Results: The forensic autopsy recorded 348 findings (mostly injuries) compared to 241 with PMCT resulting in a 69% agreement. The total combined number of findings observed by both the forensic autopsy and PMCT was 437. PMCT had an 86% agreement (n = 99/115) with the forensic autopsy for injuries within the head and neck, 60% for the thorax (n = 93/156), 53% abdomen and pelvis (n = 17/32), 93% extremities (n = 14/15) and 32% in the miscellaneous category (n = 7/22). The forensic autopsy had a 73% agreement (n = 8/11) with PMCT for spinal column and spinal cord injuries. PMCT had a total of 196 (45%) discrepant findings compared to 89 (20%) for the forensic autopsy. Discrepant findings were those injuries identified by one examination and not the other. Despite the high number of discrepant findings for PMCT, this modality was able to identify 89 additional injuries not recorded by the forensic autopsy. When analysing injury categories, the forensic autopsy identified more haemorrhagic, hollow organ, large blood vessel, muscle, soft tissue and solid organ injuries compared to PMCT. Conversely PMCT identified more gas collections and skeletal injuries compared to the forensic autopsy. There was a good-to-perfect concordance between the forensic autopsy and PMCT for diagnosis of intracerebral haemorrhages (71%), lung collapse (83%), subarachnoid haemorrhages (SAH) (86%), haemothoraces (92%), rib fractures (97%), diastasis of skull bones (100%) and haemoperitoneum (100%). The forensic autopsy diagnosed more brain contusions (n = 0/8) brain laceration (n = 2/11) subdural haemorrhage (SDH) (n = 1/6) blood and gastric content aspirations (n = 1/11) lung haemorrhages (n = 1/14) lung parenchymal injuries (n = 6/27) cardiac injuries (n = 2/15) and abdominal organ injuries (n = 2/14). With PMCT more facial bone (n = 0/7) skull fractures (n = 26/38) (particularly orbital fractures) intraventricular haemorrhages (n = 2/8) brain compression/swelling (n = 3/7) pneumocephali (n = 0/13) and pneumothoraces (n = 3/20) were diagnosed. PMCT showed perfect agreement (100%) with the forensic autopsy for assigning a cause of death for blunt force head injuries and very good agreement (91%) for gunshot injuries. PMCT further had good agreement (75%) with the forensic autopsy for establishing a cause of death in two fatal physical abuse cases. The mean percent agreement of PMCT for assigning the correct cause of death for all 25 unnatural deaths in this study was 80%, suggesting a very good agreement with the forensic autopsy. There was 0% agreement between the forensic autopsy and PMCT for establishing the cause of death in five natural deaths. An incorrect cause of death was assigned with PMCT in one case. The Cohen Kappa statistic measuring interrater reliability for the cause of death rating assigned by the forensic autopsy and PMCT reviewers 1 and 2 was k = 0,624 (95% Confidence interval: 0.45 – 0.80; p = 0.00) and k = 0,582 (95% Confidence interval 0.41 - 0.76, p = 0.00) respectively. This implied a substantial and moderate level of agreement between the forensic autopsy and Reviewers 1 and 2 respectively. The interrater reliability between Reviewers 1 and 2 was k = 0.905, (95% Confidence interval 0.78 – 1.00, p = 0.00) suggesting a near perfect agreement. Conclusion: This study confirmed that establishing the nature and anatomical location of injuries with PMCT for subjects, following traumatic deaths, is good but remains poor for natural deaths or unnatural deaths where no clear physical injuries are evident. PMCT is unable to assess organ pallor, measure free fluid; nor diagnose skin and mucosal injuries or establish organ weight. PMCT was unable to identify all entrance and exit gunshot wounds. Based on the number of discrepant findings for PMCT, and clinical and physical assessments not possible with this imaging modality, the findings of this study did not support the notion that PMCT can replace the forensic autopsy. PMCT should serve as an adjunct examination to forensic autopsies of children whose demise was due to suspected fatal abuse, irrespective of the initial manner of death. PMCT must therefore be routinely employed as a supplementary examination to the forensic autopsies in the assessment of suspected fatal child abuse.