The author, an editor, or a technical staff member, depending on the bulletin’s resources and situation, can carry out a literature search. The aim is to gather the most reliable information needed to answer the questions raised during the previous step.
The following is an overview of the main types of sources that can be consulted, depending on the type of article being prepared, with a few examples in each category (see the appendix at the end of the manual for links to electronic sources):
• Basic textbooks and formularies - the most recent edition, either in print or an electronic version - remain an essential information source. Examples are Martindale, the British National Formulary, Meyler’s Side Effects of Drugs, some WHO publications, such as the WHO Model Formulary etc.
• Systematic reviews produced by independent sources, for example, those available from the Cochrane Library, or from Clinical Evidence, or the Database of Abstracts of Reviews of Effects (DARE);
• Guidelines issued by medico-economic evaluation bodies like the National Institute for Clinical Excellence (NICE), the Scottish Intercollegiate Guidelines Network (SIGN), the Canadian Centre of Health Technology Assessment (CCOHTA), the National Guidelines Clearinghouse (NGC), USA, etc.
• Health care evaluation agencies’ recommendations, such as those of the International Network of Health Technology Assessment Agencies (INAHTA), or of national agencies.
• Published articles in scientific and medical journals. These are now easily retrievable from computerised international databases such as Medline or Embase. It is important to distinguish between original publications of randomised clinical trials, meta-analyses, reviews, etc. and to look carefully at the methods, study design and editorial procedures used by authors. It is often difficult for beginners to evaluate the reliability of results of published studies. It may be helpful to attend a training session in critical appraisal of journal articles, for example at an ISDB summer school or workshop. Some bulletins have published methodology articles like “Reading Between the Lines of Clinical Trials”.6 The book How to Read a Paper by Trisha Greenhalgh is also extremely useful.7 See also Chapter 8 for a discussion of analysis of published clinical trial results.
• Unpublished reports of clinical trial results. These may be obtained from the industry (through the on-line register of some companies, even if incomplete or biased, or on request), through regulatory agencies or from clinical experts. It is important to distinguish between full trial reports and abstracts, as the latter are generally of much lower quality. There is some controversy among bulletins over the use of unpublished data. Some avoid them because the reports of results have not been peer reviewed, because they may not be easily accessible to readers, or because they give too little information on the methodology to enable one to interpret the results (this is particularly true with abstracts and congress communications). Others have no systematic confidence in peer reviews. They use unpublished reports because in certain areas few or no published reports exist. This is the case for some new drugs or suspected side effects. It is also useful to make unpublished data visible in order to clarify the existing state of knowledge or underline inconsistencies.
• Databases dealing specifically with drug side effects such as Reactions, Current Problems, etc. Even if some are not independent, most databases of adverse drug reactions are worth looking at: the evidence is often scarce and cross-checking to verify the reliability of reports is important.
• Databases on patients’ experience, such as the Database of Personal Experiences of Health and Illness (DIPEx).
• Promotional material and other unreliable information. It is very useful to find out what types of misleading claims are being made about a topic or a specific drug, in order to address these claims in the article.
• Evaluation reports published by drug regulatory agencies, like those provided by the U.S. Food and Drug Administration (FDA) or other national agencies, or the European Medicines Agency (EMEA). Although they may not be complete and much of their content comes from pharmaceutical companies, they are useful if cross-checked.
• National drug utilisation data, or consumption or reimbursement data, when available.
• Drug prices.
Internet literature searches and the use of computerised databases are efficient means to search for documentation, provided of course that Internet access is available, and that the person carrying out a search has the necessary training to select relevant and appropriate references. It is easy to carry out a search, but selection of references requires more editorial and clinical experience.
If you are starting a new bulletin, it is essential to know that good quality, successful articles can be based on a few well-selected information sources and that quantity does not mean quality. The credibility of a drug bulletin does not depend on the length of the bibliographies at the end of its articles. The references should be valid, relevant and up-to-date, avoiding speculation and opinions not based on evidence.
In some cases you may face a lack of evidence even after a thorough search for documentation. Do not hide this from the reader. You can explain what questions remain unanswered and the degree of uncertainty remaining. The article can conclude with recommendations for the most reasonable therapeutic attitude, given what is and is not currently known, and express a willingness to change if new evidence becomes available.