The original objective of the study reported here was to review the literature on the clinical effectiveness and cost-effectiveness of the routine pre-operative use of three diagnostic tests – FBC, U&E and PFT – in the context of minor and intermediate surgery for otherwise healthy patients, and to synthesise the evidence identified in the context of a de novo cost-effectiveness model.

A comprehensive and systematic search of both the effectiveness and cost-effectiveness literature identified a large number of potentially relevant studies. However, when these studies were subjected to detailed review and quality assessment it became clear that the literature provides no evidence on the effectiveness and cost-effectiveness of these specific tests in the specific patient groups in the context of the UK NHS.

The limitations of the published clinical effectiveness literature – from the perspective of this study – included but were not limited to:

  • inadequate reporting of the surgery that patients were being prepared for
  • inadequate reporting of the specific tests undertaken and the results of the individual tests
  • inadequate reporting of the clinical response to test results; and
  • inadequate reporting of the outcomes of the surgery.

These limitations were by and large shared by the published cost-effectiveness literature. In addition, there were almost no studies from the UK NHS, which meant that the estimates of the resource use and cost reported in the identified papers were unlikely to be relevant to the NHS. The studies also failed to report disaggregated information on resource utilisation and cost and focussed on short-term clinical outcomes rather than health outcomes.

The cost-effectiveness literature that was identified did not look at the longer-term outcomes attributable to the use or non-use of these pre-operative tests (i.e. it focused on the difference in the incidence of perioperative complications and the costs associated with these). This is perhaps attributable to the fact that the studies were generally small and investigators quite possibly did not have the resources necessary to undertake longer-term follow-up. It might also be because the relationships between perioperative complications and longer-term health outcomes are insufficiently understood to allow the construction of models to predict these longer-term consequences in the absence of data.

Whatever the reason for the lack of longer-term health outcome data for these pre-operative tests, the literature does not support any robust conclusions about the value of the routine use of these pre-operative tests compared with alternative uses of the limited health-care resources.

In addition to the literature reviews, we repeated a survey of current practice commissioned by NICE as part of their guideline review process in 2005. The results indicate that the degree of uptake of the NICE guidance on pre-operative testing has increased substantially since the original study. The responses suggest that routine pre-operative testing in minor surgery in patients aged < 40 years has all but disappeared from the NHS.7

The results of the survey of practice could not be directly verified by this study. However, owing to the lack of published evidence we undertook an additional piece of work, analysing routine testing data from one large teaching hospital trust. The results of this analysis are discussed in more detail below, but they are consistent with the results of the survey and thus may represent a weak validation of these survey results.

The analysis of routine testing and surgical outcome data was not part of the original proposal. However, given the lack of published evidence on the clinical effectiveness and/or cost-effectiveness of these tests, and the importance of the question given the high volumes of preoperative testing across the NHS as a whole, we deemed it important to exhaust all reasonable avenues of enquiry in pursuit of relevant evidence.

We were fortunate that Leeds Teaching Hospitals Trust had maintained a database of all tests ordered that could be linked, at the individual patient level, to a number of measures of outcome of surgery. This provided a substantial number of observations on which we could estimate regression models. Although the details of the analysis are reported elsewhere, it is worth reiterating that the essence of the work was to estimate the relationship between utilisation of any of the three tests in the pre-operative assessment and the outcomes of surgery in a cohort of otherwise healthy patients undergoing minor or intermediate surgical procedures. It should be noted that we had to approximate the ASA grade 1 and ASA grade 2 score retrospectively. This is not the same as having an original anaesthetist’s score, which further increases the uncertainty in the interpretation of the results, as the case mix of the patients in the sample may be more or less diffuse than in the study scope.

If the tests were being used routinely and they were having a positive impact on outcomes, we would expect to see that patients who received the tests were likely to have shorter lengths of stay and more likely to be discharged from hospital by 30 days. The modelled relationships were exactly the opposite of what was expected. Many patients did not undergo any of these tests and those who did were more likely to have longer lengths of stay and less likely to be discharged by 30 days post operation.

In constructing a decision-analytic model for the cost-effectiveness of these tests it became clear that a number of key determinants of the value of these tests were dependent on the specific cause of the abnormal test result. There are multiple potential causes for abnormal tests results for all three tests. The appropriate clinical response, its resource implication and the expected outcomes of the treatment and hence the potential cost-effectiveness of the test are all dependent on the underlying cause. Constructing models for each possible abnormal test/cause combination was outside the scope of this project. However, any future work examining the cost-effectiveness of these tests in pre-operative assessment will have to frame the decision problem in this context if each parameter in the decision problem is going to be clearly specified.

The most defensible conclusion to be drawn from this study is that there is insufficient evidence to support the utilisation of these three tests as part of the routine pre-operative assessment in otherwise healthy patients undergoing minor and intermediate surgery. The survey and analysis of routine data from the Leeds Teaching Hospitals NHS Trust indicate that the time of universal utilisation of these tests in pre-operative assessment may indeed have passed. However, concerns over response rates and the risks of generalising from data on a single trust make this conclusion tentative.

This study raises the question of how to proceed in an evidence-based decision-making context when there is effectively no evidence related to the decision problem. We had originally proposed to address weaknesses in the evidence by using expert elicitation. However, when it became clear that virtually all of the decision parameters in the decision problem would require expert elicitation, the appropriateness of this strategy became questionable. Challenges associated with establishing who would be the appropriate experts for different parameters in the decision model, how to ensure the representativeness of the sample, and synthesising the evidence provided by different experts on different parameters meant that wholesale elicitation was methodologically questionable and pragmatically beyond the resources of this project.

We considered that establishing a representative sample of experts for the elicitation would be essential if the results of the analysis were to be credible to the medical and decision-making community. However, it would be equally problematic as we are not aware of methods for establishing that the relatively small samples of experts that would be feasible within project resources are representative of such a large community of practitioners. For these reasons, formal elicitation of expert opinion does not appear to offer an analytical solution to the health-care decision-maker’s dilemma of how to make an evidence-based decision in the absence of evidence.

Recommendations for further research

The total expenditure on pre-operative tests across the NHS remains significant. Given the almost complete absence of published evidence on the clinical effectiveness, safety and cost-effectiveness of routine use of these tests in uncomplicated patients undergoing ASA grade 1 and 2 procedures, any well-designed research would add to the current state of knowledge. However, to recommend specific research questions it would be necessary for us to have a view as to the value of additional information to decision-makers in the NHS. To assess the likely value of such research it would be necessary to have a robust assessment of the current scale of the routine use of these tests in the patient/procedure combinations of interest.

The low response rate to our survey, despite significant efforts at follow-up, suggests that this type of survey will not be a satisfactory strategy for scoping the scale of the research opportunity. A systematic identification of routine test databases held by NHS trusts is necessary to establish the feasibility of undertaking a multicentre version of the routine data analysis that we report for Leeds Teaching Hospitals Trust.

If feasible, this would allow the identification of the scale of the use of these tests in practice and the degree to which they are being used in otherwise healthy patients, rather than in response to a specific clinical indication. Only once this information is available will it be possible to establish whether or not any further research in this area is required and, if so, which research questions have the greatest potential value to the NHS.