In order to appraise studies which are identified, and to extract preliminary data for a qualitative or quantitative synthesis (when appropriate), the systematic reviewer often has to depend on information available from published reports of the relevant studies. However, these published reports rarely provide sufficient evidence necessary to complete a thorough and unbiased systematic review[2]. In particular, allocation concealment and other aspects of study design have been poorly described in reports of randomised trials[3][4][5], and selective reporting of outcomes may result in biased availability of data to the meta-analyst[6]. Some important statistics may not be reported, for example standard deviations of continuous data (particularly changes from baseline), intra-class correlation coefficients for cluster-randomised trials and total number of patients both randomised and analysed. Further problems associated with extracting data from published reports are connected with inconsistencies between the analysis of different studies such as transformation of raw data, use of intention-to-treat analyses and varying approaches to outcome data such as ordinal and survival data.

To obtain missing information, systematic reviewers might approach the investigators of the studies they have identified. Some reviewers put considerable effort into tracking down investigators and data. The appropriate amount of effort will depend on the importance of the missing information to the validity of the systematic review. Contact details of a primary author may be obtained from most published papers, and in recent publications an email address and/or a telephone number may also be available. However, it is unclear which of various possible methods of first contact is most effective, both at establishing whether the contact details are correct and at obtaining information on the study.

The principle of contacting investigators is encouraged within many groups of the Cochrane Collaboration. The Cochrane Reviewers' Handbook[7] recommends researchers be contacted with a view to identifying additional trials. This can be a useful process[8], though no specific guidelines are provided in the Handbook for collecting additional data from trialists apart from in the context of individual patient data reviews in which the trialists are contacted for the raw data from their studies[9]. Chalmers et al. discuss issues of bias which may arise from including unpublished data, but conclude that these are unlikely to be more significant than biases associated with selective reporting. They also list a number of publications in which contacting investigators appeared to be beneficial[10].

In May 1999 we surveyed the 44 then-existing Cochrane Review Groups to determine what recommendations they provide to their reviewers on contacting trialists. Thirty-one Groups (70%) responded and we obtained four example letters and three generic letters that were supplied to reviewers. Two were simple letters that included a request for explicit data or information; two would be accompanied by a table to be completed; two would be accompanied by a form to be completed and the last appended a form which had been partially completed by the reviewer with data extracted from published reports, in which missing or unclear information was highlighted. Four letters provided some information on the Cochrane Collaboration, three in an introductory paragraph and one in the form of enclosed literature. An additional example letter was identified[10] with which summaries of extracted data were sent, along with a request that missing items be supplied.

Methods for approaching primary investigators for specific information have not been studied in detail, though there is a wealth of literature on methods for improving survey response. A Cochrane review has been initialed to systematically review studies of methods to influence the response rates to surveys[11]. Existing meta-analyses of controlled studies have indicated that preliminary notification[12], follow-up[13], monetary or non-monetary incentives[14] and appeals for help in a covering letter[15] can significantly improve response to surveys. Follow-up using special mail services or telephone has been advocated[16], as have numerous other techniques including pre-paid postage using stamps[17] and obtaining prior agreement to participate[18] (the so-called 'foot-in-the-door' method). These results have not all been replicated in eliciting responses from clinicians. Randomised trials have demonstrated that general practitioners can respond to monetary incentives [19][20] and free pencils[21] and that advance telephone calls can influence the response of gynaecologists[22]. However, pre-notification has not been found to affect physicians[23]. A qualitative study of reasons why general practitioners fail to respond highlighted that recipients might be too busy and that questionnaires might be lost or discarded[24].

Whether findings from survey research are relevant to requesting information from the principal investigators of trials to be included in Cochrane reviews is unclear. The principal barrier may be failure to make contact with the investigators rather than poor response to a request that is actually received. First Contact assesses whether implementing a selection of appropriate techniques can affect the collection of missing information for systematic reviews.

Aim of the trial
We aim to evaluate whether response and information retrieval rates from investigators, when contacted with a request for information to contribute to a systematic review, can be affected by how they are approached. We plan to do this by conducting a randomised trial of a standard letter compared with a more intensive approach involving advance warning, additional information, follow-up and other techniques aimed at improving response.

Terminology
Some care is appropriate in the terminology used to describe the people involved in First Contact since trialists and participants in the traditional sense might be seen as, conversely, participants and trialists in the context of this trial. Therefore, in this protocol, the following terms are used.

Systematic reviewer, or reviewer
A collaborator, or participant, in First Contact who is conducting a systematic review and is willing to randomise methods of obtaining information from the studies he or she identifies, (analogous to a centre in a multi-centre clinical trial).

Primary investigator, or investigator
The author or contact person of a published study who is to be approached by the systematic reviewer, (analogous to a patient in a multi-centre clinical trial since they will be randomised to receive one intervention).