15283-B-Blokker

88 Chapter 5 the pathological processes causing death, and to allow meaningful comparison of our study with the literature. Grouping of major diagnoses had to be done retrospectively, because only then the consensus COD was known. Another limitation of this study is that only few of the 2197 patients who died during the study period were autopsied, and even fewer underwent MIA. This was inevitable, because MIA was only possible if next-of-kin after having given consent to CA, which was the case in only 13.4% of deceased, also consented to MIA. Furthermore, a number of consented MIA could not be executed for logistic reasons. However, the COD established in this sample, most often cardiovascular or pulmonary pathology, are a fair representation of the findings at routine autopsies in Western countries with high- income economies. 142 Since pathological/histological examination of the brain was consented in a relatively small proportion of MIA (24/99) and CA (38/99) cases, and only 20 underwent both biopsies and brain autopsy, we decided to include all histologically examined cases in our comparison of MIA and CA. The analysis of diagnoses on the brain showed no discrepancies as to COD, more diagnoses with MIA (based on imaging), and more major diagnoses with CA (based on histology). These results emphasize the significance of pathological/histological examination of the brain, in agreement with the large contribution of histological examination to diagnoses established with MIA in general. Finally, this being a single center study, the conclusions may not be generally applicable. The high performance of MIA, agreeing with consensus COD in 96/99 cases, should not be surprising, as imaging provides a substitute for gross examination at CA, and excellent guidance for sampling for histology and cytology. Technically the present MIA was improved compared to our previous study, in which biopsies were fewer, and either taken randomly, or guided by ultrasonography. 57 Apart from the current study, our previous one is the only reporting on the diagnostic performance of MIA in patients who died in-hospital using the combination of PMCT and PMMR, and image-guided biopsies. Therefore, our results cannot easily be compared to the literature. In a cohort of 182 cases, Roberts et al. 102 compared PMCT and PMMR to CA in coroner’s cases, and found an agreement for COD of 70% (95% CI: 62.6;76.4). Most often missed were ischemic heart disease, pulmonary embolism, pneumonia, and intra-abdominal lesions. Westphal et al. 104 , investigating the feasibility of PMCT only in 29 cases, reported an accuracy for COD of 68% and a positive predictive value of 75%. In agreement with these studies, we found that PMCT and PMMR alone could not reliably diagnose common COD such as pneumonia, myocardial infarction, peripheral pulmonary emboli, gastrointestinal ischemia and sepsis without biopsy confirmation.