Search

Objectives To investigate the clinical effectiveness of treatment with hyperbaric oxygen for neonates with hypoxic-ischaemic encephalopathy. This treatment is frequently used in China but much less often in the West. Data sources Western (Cochrane controlled trials register and database of systematic reviews, Medline, Embase, CINAHL, and HealthSTAR) and Chinese (China Hospital Digital Library, Chinese Medical Journal Network) databases and hand search of Chinese journals. No language restrictions. Review methods Randomised or quasi-randomised controlled trials of treatment with hyperbaric oxygen compared with "usual care" in term neonates with hypoxic-ischaemic encephalopathy. Outcomes included mortality and long term neurological sequelae. Standardised forms were used to extract and compare data. Criteria of York Centre for Reviews and Dissemination were used to assess quality. Analysis was mainly qualitative but included meta-analysis. Results 20 trials were found, mainly from Chinese sources. The reporting quality of trials was poor by Western (CONSORT) standards. Treatment with hyperbaric oxygen had better outcomes dian the comparator in almost all trials. The odds ratios of the meta-analyses were 0.26 (95% confidence interval 0.14 to 0.46) for mortality and 0.41 (0.27 to 0.61) for neurological sequelae. Conclusion Treatment with hyperbaric oxygen possibly reduces mortality and neurological sequelae in term neonates with hypoxic-ischaemic encephalopathy. Because of the poor quality of reporting in all trials and the possibility of publication bias, an adequately powered, high quality randomised controlled trial is needed to investigate these findings. The Chinese medical literature may be a rich source of evidence to inform clinical practice and other systematic reviews. [ABSTRACT FROM AUTHOR]

Currently systematic reviews focus on diagnosis or effectiveness of treatment. It is the understanding of disease aetiology that underpins medical education, practice and research. Whether an association meets causal criteria is usually assessed qualitatively. However, this can also be examined through evidence synthesis and systematic reviews to evaluate disease causation and mechanisms are much needed. It is important in such a review to specify the questions to be addressed regarding causal criteria such as strength, consistency, temporality, specificity, biological gradient, plausibility and experimental evidence. The next step is to conduct a thorough literature search to identify the relevant studies and to assess them for their quality, particularly in relation to the risk of bias, ascertainment of exposures and ascertainment of outcomes. Data synthesis can then examine if the observed associations in collated studies are consistent, strong and temporal using techniques such as meta-analysis, testing for heterogeneity and meta-regression. Biological plausibility and coherence with existing theories can also be examined systematically through an assessment of the basic scientific literature. Experimental evidence might also be collated and synthesised to determine if removal of a causal agent alters the outcome. Through these steps a systematic review can help to establish whether an association is causal or not. [ABSTRACT FROM AUTHOR]

When generating guidelines, quality of evidence is frequently reported in tabulated form capturing several domains, for example, study design, risk of bias and heterogeneity. Increasingly, this is done using the Grading of Recommendations Assessment, Development and Evaluation approach. As assimilating large amount of tabulated data across several comparisons and outcomes spread over many pages (sometimes hundreds) is not easy, there is a need to present evidence summaries in a more effective way. A graphic display plotting the several domains used in evidence grading on equiangular spokes starting from the same point, the data length of each spoke proportional to the magnitude of the quality, succinctly captures tabulated information. These plots allow easy identification of deficiencies, outliers and similarities in evidence quality for individual and multiple comparisons and outcomes, paving the way for their routine use alongside tabulated information. [ABSTRACT FROM AUTHOR]

The article presents a study that determines the properties of published risk models and scores for predicting type 2 diabetes in adults. The study describes 10 mechanisms by which measuring the risk of diabetes might improve outcomes. It reveals that not all risk models or scores had robust statistical properties and had been externally validated on a different population.