A Bayesian analysis of mortality outcomes in multicentre clinical trials in critical care.

BACKGROUND: Multicentre RCTs are widely used by critical care researchers to answer important clinical questions. However, few trials evaluating mortality outcomes report statistically significant results. We hypothesised that the low proportion of trials reporting statistically significant differences for mortality outcomes is plausibly explained by lower-than-expected effect sizes combined with a low proportion of participants who could realistically benefit from studied interventions. METHODS: We reviewed multicentre trials in critical care published over a 10-yr period in the New England Journal of Medicine, the Journal of the American Medical Association, and the Lancet. To test our hypothesis, we analysed the results using a Bayesian model to investigate the relationship between the proportion of effective interventions and the proportion of statistically significant results for prior distributions of effect size and trial participant susceptibility. RESULTS: Five of 54 trials (9.3%) reported a significant difference in mortality between the control and the intervention groups. The median expected and observed differences in absolute mortality were 8.0% and 2.0%, respectively. Our modelling shows that, across trials, a lower-than-expected effect size combined with a low proportion of potentially susceptible participants is consistent with the observed proportion of trials reporting significant differences even when most interventions are effective. CONCLUSIONS: When designing clinical trials, researchers most likely overestimate true population effect sizes for critical care interventions. Bayesian modelling demonstrates that that it is not necessarily the case that most studied interventions lack efficacy. In fact, it is plausible that many studied interventions have clinically important effects that are missed.

Framing of clinical information affects physicians' diagnostic accuracy.

BACKGROUND: Framing bias occurs when people make a decision based on the way the information is presented, as opposed to just on the facts themselves. How the diagnostician sees a problem may be strongly influenced by the way it is framed. Does framing bias result in clinically meaningful diagnostic error? METHODS: We created three hypothetical cases and asked consultants and registrars in Emergency Medicine and Internal Medicine to provide their differential diagnoses and investigations list. Two of the presentations were written two ways to frame the case towards or away from a particular diagnosis (Presentation 2 - pulmonary embolus (PE) and Presentation 3 - interstitial lung disease (ILD)) and these were randomly assigned to the participants. Both versions were however entirely identical in terms of the objective facts. Physician impressions and diagnostic plan were compared. A third presentation was identical for all and served as a control for clinician baseline 'risk-averseness'. RESULTS: There were significant differences in the differential diagnoses generated depending on the presentation's framing. PE and ILD were considered and investigated for the majority of the time when the presentation was framed towards these diagnoses, and the minority of the time when it was not. This finding was most striking in Presentation 2, where 100%versus50% of clinicians considered PE in their diagnosis when the presentation was framed towards PE. This result remained robust when undertaking stratified analysis and logistic regression to account for differences in seniority and baseline risk-averseness- neither of the latter variables had any effect on the result. CONCLUSION: We demonstrate a clinically meaningful effect of framing bias on diagnostic error. The strength of our study is focus on clinically meaningful outcomes: investigations ordered. This finding has implications for the way we conduct handovers and teach juniors to communicate clinical information.

Perioperative support of a patient with fat embolism syndrome with extracorporeal membraneoxygenation.

Fat embolism syndrome (FES) is a serious complication of trauma that can result in multiorgan failure, including the acute respiratory distress syndrome. Occasionally, the severity of respiratory failure associated with FES warrants support with venovenous extracorporeal membrane oxygenation (VV-ECMO), a therapy with widespread use but inconclusive evidence. Early definitive fracture fixation is the mainstay of preventing further fat embolism and ongoing organ dysfunction, but poses significant risks to the maintenance of the extracorporeal circuit. We describe a rare case of a patient who required VV-ECMO for respiratory support prior to fracture fixation. The risks of intraoperative fat embolisation causing sudden circuit failure were managed by having a spare circuit available outside the operating room with readiness for an emergency circuit change. Postoperative fat deposition in the oxygenator was managed by a circuit change. Our case is the first to describe preoperative initiation of VV-ECMO for FES and highlights why this therapy should not delay definitive fracture fixation and how it can be safely managed in this setting.

A comparative MRI study of cartilage damage in gout versus rheumatoid arthritis.

INTRODUCTION: Magnetic resonance imaging (MRI) is useful for detecting joint inflammation and damage in the inflammatory arthropathies. This study aimed to investigate MRI cartilage damage and its associations with joint inflammation in patients with gout compared with a group with rheumatoid arthritis (RA). METHODS: Forty patients with gout and 38 with seropositive RA underwent 3T-MRI of the wrist with assessment of cartilage damage at six carpal sites, using established scoring systems. Synovitis and bone oedema (BME) were graded according to Rheumatoid Arthritis MRI Scoring System criteria. Cartilage damage was compared between the groups adjusting for synovitis and disease duration using logistic regression analysis. RESULTS: Compared with RA, there were fewer sites of cartilage damage and lower total damage scores in the gout group (P = 0.02 and 0.003), adjusting for their longer disease duration and lesser degree of synovitis. Cartilage damage was strongly associated with synovitis in both conditions (R = 0.59, P < 0.0001 and R = 0.52, P = 0.0045 respectively) and highly correlated with BME in RA (R = 0.69, P < 0.0001) but not in gout (R = 0.095, P = 0.56). CONCLUSIONS: Cartilage damage is less severe in gout than in RA, with fewer sites affected and lower overall scores. It is associated with synovitis in both diseases, likely indicating an effect of pro-inflammatory cytokine production on cartilage integrity. However, the strong association between cartilage damage and BME observed in RA was not identified in gout. This emphasizes differences in the underlying pathophysiology of joint damage in these two conditions.

Methodological quality of systematic reviews in subfertility: a comparison of two different approaches.

BACKGROUND: Systematic reviews are used widely to guide health care decisions. Several tools have been created to assess systematic review quality. The measurement tool for assessing the methodological quality of systematic reviews known as the AMSTAR tool applies a yes/no score to eleven relevant domains of review methodology. This tool has been reworked so that each domain is scored based on a four point scale, producing R-AMSTAR. METHODS AND FINDINGS: We aimed to compare the AMSTAR and R-AMSTAR tools in assessing systematic reviews in the field of assisted reproduction for subfertility. All published systematic reviews on assisted reproductive technology, with the latest search for studies taking place from 2007-2011, were considered. Reviews that contained no included studies or considered diagnostic outcomes were excluded. Thirty each of Cochrane and non-Cochrane reviews were randomly selected from a search of relevant databases. Both tools were then applied to all sixty reviews. The results were converted to percentage scores and all reviews graded and ranked based on this. AMSTAR produced a much wider variation in percentage scores and achieved higher inter-rater reliability than R-AMSTAR according to kappa statistics. The average rating for Cochrane reviews was consistent between the two tools (88.3% for R-AMSTAR versus 83.6% for AMSTAR) but inconsistent for non-Cochrane reviews (63.9% R-AMSTAR vs. 38.5% AMSTAR). In comparing the rankings generated between the two tools Cochrane reviews changed an average of 4.2 places, compared to 2.9 for non-Cochrane. CONCLUSION: R-AMSTAR provided greater guidance in the assessment of domains and produced quantitative results. However, there were many problems with the construction of its criteria and AMSTAR was much easier to apply consistently. We recommend that AMSTAR incorporates the findings of this study and produces additional guidance for its application in order to improve its reliability and usefulness.