Healey JS, Connolly SJ, Gold MR, Israel CW et. al. Subclinical atrial fibrillation and the risk of stroke. N Engl J Med. 2012 Jan 12; 366: 120-9.
This is a large, multi-center study performed in 23 countries with two parts. The first part was an observational study that included 2580 patients with a newly implanted pacemaker or defibrillator (2451 with pacemaker and 129 with ICD). Patients were eligible for the study if they were 65 or older, had a history of hypertension requiring medical therapy, and had undergone their first implantation of a St. Jude pacemaker (for sinus-node or atrioventricular-node disease) or ICD (for any indication) within the preceding 8 weeks before enrollment. Patients were excluded for any history of atrial fibrillation (AF) or atrial flutter lasting more than 5 minutes or for any previous treatment with a vitamin K antagonist for any reason. After 3 months, the patients’ devices were interrogated to determine if a subclinical atrial tachyarrhythmia (SAT) had occurred (defined as atrial rate 190 beats or more per minute for longer than 6 minutes). The patients were additionally followed for a mean of 2.5 years for the primary outcome of ischemic stroke or systemic embolism. Secondary outcomes were vascular death, myocardial infarction, stroke from any cause, and atrial tachyarrhythmias documented by surface electrocardiography. The second part of the study was a randomized controlled trial including the patients in this study with implanted pacemakers. Patients were randomized at their 3-month visit to have continuous atrial overdrive pacing (at a rate slightly higher than the patient’s intrinsic sinus rhythm) programmed as either “on” or “off”. The primary outcome of this portion of the study was symptomatic or asymptomatic atrial tachyarrhythmia lasting over 6 minutes and documented by surface electrocardiographic recording.
At the 3-month visit, at least one tachyarrhythmia was detected in 251 patients (10.1%), with clinical atrial tachyarrhythmias in only 7 patients. During the follow-up period, 11 of the 261 patients (4.2%) with SATs before 3 months had an ischemic stroke or systemic embolism (rate of 1.69% per year), as compared with 40 of the 2319 patients with no detection of SATs (1.7%, rate of 0.69% per year) (hazard ratio 2.4; 95% CI, 1.28 to 4.85; p=0.007). This risk was not significantly changed after adjusting for baseline risk factors for stroke (p=0.008). Of the 51 patients with stroke or systemic embolism, 11 had SATs by 3 months and none had clinical AF by 3 months (Population attributable risk of ischemic stroke or systemic embolism associated with SAT of 13%). There was no association between SATs and any of the secondary outcomes. The intervention of atrial overdrive pacing did not have a significant effect on the development of clinical AF. However, the rate of development of clinical AF was low in both groups and the study may not have had the power to detect this difference if it does exist.
This study was well done and it indicates that SATs occurred frequently in patients with pacemakers and a history of hypertension but no prior diagnosis of AF. The occurrence of SATs was associated with a significant increased risk of subsequent stroke.

- Jake Gessin

Roberts IS, et al. Post-mortem imaging as an alternative to autopsy in the diagnosis of adult deaths: a validation study. Lancet. 2012 Jan 14;379:136-42. 
We all know that all death begins in Radiology, but does it also end in Radiology?This study by Ian SD Roberts (not the same guy from CRASH-2), et al is a UK study performed at two centers looking at the accuracy of post-mortem imaging as an alternative to autopsy for determining cause of death. The idea of post-mortem imaging has arisen out of multiple issues: the high rate of autopsies in the UK (22% of all deaths), a national audit revealing poor quality autopsies and inadequate autopsy reports, as well as opposition from various groups after public organ retention scandals in the late 1990s and religious groups (especially in Jewish and Muslim communities) who oppose autopsy. In some locations in the UK, post-mortem imaging reports are reviewed by the coroner and autopsy is not performed in as many as 90% of cases with a definite radiographic cause of death (COD).
This study was a validation of smaller studies which showed important weaknesses in post-mortem imaging, especially inability to detect arterial occlusion and to differentiate between pulmonary edema and pneumonia. This study was a modified case control where the first coroner’s case of the day on pre-selected days was enrolled in the trial, the bodies were imaged per study protocol with CT and MRI, then underwent a routine autopsy.  Exclusion criteria were severe obesity (precluding imaging) or inability to obtain consent for imaging. Overall 208 cases were enrolled, 16 were rejected from the final dataset (15 due to incomplete imaging or scanner malfunction; 1 had been embalmed prior to imaging) and another 10 were used for training purposes to familiarize the radiologists with post-mortem imaging (as they had no prior post-mortem experience), leaving 182 included cases.
The paper lists 5 outcome measures, but does not identify a primary outcome measure. However from the discussion and results it would seem that the primary outcome measure was evaluating the accuracy of post-mortem imaging in diagnosis of cause of death in adults. Prominent ‘secondary’ outcomes include whether radiologists can accurately identify which cases might be diagnosed with post-mortem imaging and therefore would not need full autopsy, the relative accuracy of CT and MRI in detecting post-mortem pathologies and cause of death, inter observer variation in radiological cause of death, and whether accuracy is improved by use of specialist cardiac and neuroradiologists.
The enrolled cases underwent full body CT and MRI scanning before undergoing conventional autopsy. Multiple different radiologists reviewed these images. Each whole body CT and MRI was reviewed by 2 different radiologists (4 total) who were blinded to autopsy results. They each created an independent report, including a cause of death (COD). Then the two CT radiologist discussed their independent findings and created a consensus report, the same was done for the MRI reports. Finally, all four radiologists discussed their findings and created another consensus report. Additionally, 2 neuro-radiologists reviewed brain the CTs and MRIs only (all 182 cases), creating a consensus report on intracranial pathology (without COD). Finally different cardiac radiologists reviewed the cardiac CT and MRI (1 radiologist report per modality for the first 100 cases), respectively, taking into account the images of the rest of the body created their own reports (with COD) for the CT and MRI, then created a joint consensus report. In total, 11 separate reports were created for the first 100 cases, and 8 reports for the remaining 82 cases (no specialist cardiac reports). In each report, with the exception of the neuro-imaging, the radiologist was asked to comment on COD as well as their confidence in the COD. Specifically, the radiologists stated that their confidence in the COD was definite, probable, possible or uncertain.
When these reports were compared with the autopsy results, the rate of major discrepancy for COD was staggering. There were major discrepancies between imaging and autopsy for MRI (43% with major discrepancies; p=0.0046), CT (32%) and consensus (30%, p values not given for CT and consensus, but CIs overlap with MRI). Similarly, there were high rates of major discrepancy for the COD in cases where the radiologist(s) reported a definite cause of death and indicated “no autopsy necessary” (34%, 42%, 48% for CT, MRI and consensus report respectively). Notably, when cardiac radiologists provided a “definite COD” (36 of 97 cases), there were only 14% major discrepancies when compared with autopsy. The most common major discrepancies were in PE (10/10 missed by radiology), CAD (12/86), pneumonia (9/32 ), and bowel infarction (4/6). This seems to suggest that post-mortem imaging performs poorly when trying to identify thromboembolic disease, vascular occlusion and pneumonia (vs. cardiogenic pulmonary edema).  Finally, they note that over the duration of the study, the radiologists’ rate of major discrepancy did not decrease with more experience, despite cases being reviewed in batches so the radiologists could see the autopsy results of previous batch of cases before starting on a new batch of cases, though this data was not provided.
Given these results, it does not appear that post-mortem imaging will replace conventional autopsy anytime soon. However, these authors used major discrepancy in COD on autopsy as the gold standard, not determination of whether the patient died of natural causes, which may be a more realistic goal, especially in many of our coroner’s cases. Furthermore, the authors note that imaging may be better at detecting fractures or other signs of suspicious death. Further limitations to this study were the relatively small number of cases <200 and lack of experience of the radiologists at post-mortem imaging. While we in the ED are involved in arguably more “coroner’s cases” than most practitioners, this study is interesting, though not really practice changing. We will continue to make coroner’s reports, though there may come a time in the future we will leave that patient who codes in CT in the scanner after we call the code so they can get a post-mortem scan as part of the coroner’s clearance.

- Bryce Pulliam

Inaba K, et al. Does size matter? A prospective analysis of 28-32 versus 36-40 French chest tube size in trauma. J Trauma. 2012 Feb;72(2):422-427.
Large bore chest tubes (CTs) have been recommended for the treatment of traumatic pneumothoraces (PTX) and hemothoraces (HTX). ATLS recommends a 36-Fr tube for HTX and 38-Fr tube for massive HTX. Theoretically, a larger opening facilitates the speed and completeness of the drainage of blood while mitigating the propensity of blood to clot in the tube. However, no data are available to support these practices. This single center prospective observational study at a Level 1 trauma center hypothesized that 1) there would be no difference in clinically relevant outcomes and 2) there would be an increase in pain in those who have placement of a large CT as measured by a VAS pain score within an hour of placement.  Small CTs are 28-32 Fr; large CTs are 36-40 Fr. Inclusion criteria: all patients with CT placed within first 12 hours of admission for chest injury from January 2007 to January 2010. Exclusion criteria: patients who died within 24 hours of CT insertion. CTs were placed by EM or surgery residents and supervised by an attending physician. Size of CT was at the discretion of the attending physician. Prophylactic IV antibiotics were given prior to insertion. 353 CTs were placed in 293 patients; 52.7% were small CTs, 47.3% were large CTs. For HTX, 52.3% were small CTs and 47.7% were large CTs. Large CTs were placed more often in patients with GCS ≤8, severe head injuries (AIS ≥3), SBP <90 mmHg, and ISS ≥25. No statistical differences were found between small and large CTs for chest injury type, initial CT output, duration of placement, and complication rates, including incidence of retained HTX and type and number of interventions required to treat the retained HTX (additional CT insertion, intrapleural thrombolysis, image guided drainage, VATS, thoracotomy). For PTX, 53.4% were small CTs and 46.6% were large CTs. No statistical differences were found between small and large CTs in incidence of unresolved PTX or reinsertion of CT for treatment of unresolved PTX. Overall, VAS score for small CTs was 6.0 ± 3.0 and 6.7 ± 3.0 for large CTs (p=0.237). The French scale for CT size measurement describes the outer diameter of the tube whereas the limitations of tube function relate to the inner diameter of the tube, which varies depending on tube wall thickness. Furthermore, flow is limited by the narrowest point in the tube system, including the standard connector used in the drainage system (regardless of CT size), number and size of side holes (differs with size and manufacturer), tube kinks, effect of wall suction, solid or semisolid material within pleural cavity, and blood and muscle obstructions. In conclusion, the authors state there is no difference in the rate of complications, needed interventions, or pain sustained in the placement of small versus large CTs. They also recommend further evaluation of percutaneously placed drainage systems (sub-20 Fr range) that have comparably favorable results to larger CTs placed in the open technique. Limitations of this study include observational nature of the study (rather than randomized), lack of blinding for selection of tube size, as well as inability to adjust for confounders of flow limitation as described above. While there is no difference in VAS pain score between small and large CTs, measuring this within an hour of placement rather than at the time of insertion may be a limitation. In addition, this measurement does not take into consideration sedation or analgesia used during the procedure, other distracting injuries, or skill and prior experience of the residents who placed the CTs.

- Vivienne Ng