Each year across the world 1 billion persons are medically treated for their nonfatal injuries and almost 5 million trauma deaths are observed. The increasing risks and new hazards present in society today make this major global public health problem one of the most challenging for both prevention and care improvement.
The development and implementation of injury control and trauma care policy measures is vital and primarily moves forward through an evidence based approach, with surveillance and research data supporting decisions. But injuries can vary widely, both by external cause and by nature and severity of the lesions, and decision support by studies with the highest levels of evidence is therefore impossible across the whole spectrum of problems within this large and heterogeneous field.
When high level evidence studies are unavailable or unachievable, expert opinion derived by consensus methods may help guide decisions. One of these consensus methods, the Delphi method, is increasingly applied to support decisions in public health and clinical medicine.
This method originated in the 1960s at the RAND Corporation and one of the first applications aimed to forecast the effects of science and technology on society. The objective of most Delphi applications is the reliable and creative exploration of ideas or the production of suitable information for decision making. The method can be used when the primary source of information sought is informed judgement. The method is suited to situations of uncertainty in both the nature of the problem and the possible measures to address it effectively, and where existing information on the problem is not available or too costly to provide.
The Delphi method is a systematic approach to collect expert opinions without face-to-face interactions, which contains three essential elements [4]. First of all, anonymity: the respondents should not be aware of each other’s participation. Secondly, there should be a process of iteration and feedback: subsequent rounds are held with summaries of individual and group results. Finally, a statistical summary of group answers should be used, such as the median (the 50th percentile value of opinions), the Inter Quartile Deviation (IQD, i.e. the distance between the 25th and 75th value of opinions) and/or the percentage of respondents with similar opinions.
The Delphi method is not beyond criticism however, and potential – but addressable- problems concern the selection of panelists, the sample size and the design of the study and the questionnaires. Panel selection should follow a procedure governed by explicit criteria and sample sizes should be guided by the homogeneity of a group. In a homogeneous group of experts good results can be obtained in small panels (n = 10–15), but the size of the sample should be considerably larger when the opinion of various stakeholder groups must be taken into account. The incorporation of adequate iteration and feedback procedures in at least 2 rounds is essential in the study design and the avoidance of ambiguity should and can be carefully managed in questionnaire development.
If the aforementioned conditions are met Delphi studies can provide valuable information. They can be very helpful in formulating (international) consensus definitions. They can identify best practices in treatment procedures and provide clinical guidelines or audit filters for specific care domains, including trauma care. In the Netherlands, a Delphi study (n = 141) resulted in consensus by a representative national expert panel on a set of 21 guidelines describing the optimal process of care in the Netherlands for severely injured trauma patients. In Ghana, a Delphi study (n = 20) identified consensus on 22 trauma care and referral-specific audit filters for Low-and Middle Income Countries. In a previous issue of Injury, Vassallo et al. report on a Delphi study that identified consensus on life saving interventions after a major Incident. This is a an example of a well-conducted and well-reported Delphi study, but with the limitation that respondents were only recruited from the UK and South Africa. For these two countries, this study has provided a definition of a civilian priority one patient in case of a major incident and a list of lifesaving interventions that have been agreed by consensus. The authors acknowledge that while many of the results are internationally relevant and transferable, there may be some interventions that would not reflect practice in other resource poor environments. Delphi studies will never provide definite evidence of the way we should go and will also never lead to full consensus among all those involved within and between professions. But they can support decisions based on informed judgement and provoke further discussions at the hospital, regional, national and international level. This will stimulate further progress in injury control and trauma care and may help to reduce the unacceptable global burden of injury.