||Research to provide better quality evidence/recommendations [on preoperative testing]:- Implementation of the recommendations of this guideline together with improved information technology to support routine data collection have the potential to create large datasets that can provide answers to many of the descriptive questions that remained unanswered by the systematic review of the literature. [NICE recommends] that careful consideration is given to the possibility of establishing standardised minimum datasets across Trusts. These datasets would serve two functions, namely assessing compliance with this guideline (for example, with respect to documentation of the reasons for ordering a test) and providing data describing testing practice and the frequency of health outcomes. Summary statistics from the datasets could provide some of the parameter estimates required for a detailed economic model. One of the most striking findings of the systematic review was that there are no published studies comparing the effectiveness of alternative strategies of preoperative testing. In principle, better information technology to support routine data collection could also provide the infrastructure for a large cluster-randomised controlled trial, eg by randomising NHS Trusts that carry out elective surgery to different testing strategies. Such a trial would undoubtedly give the best quality evidence to inform the guideline. However, a trial of this kind would require unprecedented collaboration and organisation across the NHS and [NICE believes] that it is not feasible. It should be noted that the NHS R&D Health Technology Assessment prioritised research to evaluate the effectiveness of preoperative testing in high risks groups but did not commission any project. A trial of the use of appropriately trained nurses compared with pre-registration house officers for preoperative assessment was commissioned but did not vary guidance about the criteria for testing. An alternative approach is to model the costs and consequences of carrying out preoperative tests in different patient populations‚?¶ There is a need to develop, with the help of relevant clinicians, specific models to create a better representation of the complexities of decision making for each preoperative test and also to collect data to estimate the key parameters (ie the frequencies, health outcomes and costsof different consequences).