Advice on how to write a systematic review. JM Wardlaw. 14thJanuary 2010 1 How to write a paper – systematic reviewNote please see “Advice on how to write a paper” for information on general aspects of manuscript preparation, which should be read before you start to write up your systematic review. A systematic review is a piece of research – an experiment - in its own right that you do to establish whether a hypothesis is correct or not. Therefore the first step is to prepare a protocol describing the aims and objectives of the experiment and the methods. The protocol will help to keep you on the right track as you go through the literature. Systematic review methodology can be applied to any type of literature – epidemiological, randomised trials, observational studies, diagnostic tests, etc – the principals are the same although the search strategy, study assessment criteria, data extraction, and statistical analysis methods differ in some cases. Checklists for assessing the quality of papers in each type of review are given on the Equator Network (www.equator-network.org/).Hypothesis – test a is better than test b; or treatment a is better than treatment b; or animals with feature a have a shorter lifespan than animals with feature b.Aim – to determine whether, and in what circumstances, treatment a is better than treatment b in acute ischaemic strokeObjectives – to identify all randomised controlled trials testing treatment A versus B in patients with acute ischaemic stroke; to assess the quality of those trials; to determine the total number of patients included in such trials to date, and their characteristics; to determine whether treatment a works better than conventional treatment; to identify gaps in knowledge where new trials are needed, etcMethod –• what information are you looking for – the type of study (techniques, what patients, etc) – be very specific and make sure that the type of data that you are looking for will be relevant toyour research question and that you have not “drifted” off the topic• how and where to look – search strategy; time span; what literature databases; what languages; what journals to hand search; searching of reference lists in review articles, etc. • inclusion and exclusion criteria – what criteria will a paper have to meet for it to be included, and vice versa excluded?• what information to extract – minimum criteria for inclusion; quality assessment of studies; actual results; raw data and data presented as associations or correlations or odds ratios, etc • design your data extraction form – do this in sections so that you can do an early evaluation of a paper for key points on which it would either be included or not, so that you do not waste alarge amount of time on papers that are unlikely to provide useful data. Sometimes you can decide whether to include a paper very quickly, but often you are faced with a large pile of literature that you have to whittle down as quickly as possible, therefore you have to identify the promising papers quickly and put them to one side for later more detailed evaluation – some will later turn out not to be includable, but at this stage it is better to be over inclusive than to risk missing key papers through overzealous rejection before you have had a chance to go through them in detail.• primary and other outcomes – primary outcomes, secondary outcomes and subgroup analyses need to be pre-specified (just like for a primary experiment) - older people; peoplewith a particular characteristic, etc.• how are you going to present the data: summary tables, odds ratios in forest plots, etc • how to analyse – this will depend on the type of review you are doing but in the case of diagnostic tests is would include sensitivity; specificity; likelihood ratios; receiver-operatorcharacteristic curves; in the case of treatment reviews, it would be odds ratios; in the case of epidemiological or observational studies it might be standardised or weighted or normalised mean difference plus 95% CI.1,2 As you are doing the work, you MUST keep an accurate record of all the titles, abstracts and papers that you screened so that you can produce an accurate flow diagram to show how many potentially relevant papers you found at each stage of the search and evaluation process and how many were rejected at each stage.
本文来源：https://www.2haoxitong.net/k/doc/30ed2408581b6bd97f19ea0c.html
