Background to the Study
In 2003 NICE published Clinical Guideline 3 which reviewed the use of routine pre-operative tests prior to routine surgery. Prior to the guideline preparation a systematic review by Munro, Booth and Nicholl was undertaken on behalf of the HTA programme in 1997. The guideline development group undertook their own review of the literature. These two reviews defined and updated the purpose of pre-operative testing of apparently healthy patients.
Of the evidence base used to produce the guideline over 50% was graded as amber i.e. the benefit of the test was unknown. Therefore, despite the existence of some primary research, the evidence on which to base pre-operative testing protocols was inconclusive. Alongside this there has been an increasing awareness of the possibility of subjecting patients to unnecessary tests, and of the issues involved in dealing with the results of tests that may alarm patients but have little clinical significance.
The aims of this study
- undertake a systematic review of the literature of the clinical effectiveness of routine testing of full blood count (FBC), electrolytes and renal function (U&E), and pulmonary function (PFT) as part of the pre-operative assessment procedures for patients classified as American Association of Anaesthesiologist (ASA) grade 1 and 2 who are undergoing minor or intermediate procedures.
- Evaluate the cost effectiveness of mandating or withdrawing each of these tests from routine pre-operative assessment for patients ASA grade 1 and 2 and minor and intermediate surgery
- Compare the evidence with the recommendations in the NICE Guidance (2003) and observed practice in NHS hospitals
- Identify using modelling techniques the expected value of information (EVI) whether there is value in the NHS in commissioning further primary research into the routine use of FBC, U&Es and PFTs in this patient population.
The searches will not be restricted by date of language.
Inclusion and exclusion criteria
- Adult patients classified as ASA grade 1 or 2 undergoing minor (grade 1) or intermediate (grade 2) surgery (including elective general surgery, day surgery, and minor orthopaedic procedures) as classified by the CCSD Schedule of Procedures 2005.1 Where possible, to subdivide these into the following subgroups:
Apparently healthy patients with no clinical indication for testing FBC, U&Es and PFTs
Patients with common comorbidities (e.g. respiratory disease, renal disease)
Patients receiving treatments likely to alter test results (e.g. diuretics).
It was originally planned to limit the population to adults aged 16 to 60. However, because of the paucity of relevant studies which met this inclusion criterion, the population was later extended to include all adult patients.
- Routine preoperative testing of:
Full blood count (FBC) (including haemoglobin concentration, haematocrit, platelet count, and white blood cell count)
Electrolytes and renal function (U&E) (including sodium, potassium, urea, and creatinine)
Pulmonary function (PFT) (including some or all of spirometry, blood gas analysis, measurement of respiratory mechanics, measurement of transfer function, and exercise testing of respiratory system)
- No routine preoperative testing
- Abnormal test results
- Changes in management following abnormal test results in patients whose preoperative clinical examinations were normal
- Adverse events possibly related to the test result
- Adverse events probably or possibly caused by the process of testing
- All-cause mortality
- Any country
- 1980 onwards
- Controlled non-randomised studies (eg cohort studies)
- Case–control studies
- Case series
- Case reports
- Systematic reviews
- Economic evaluations
The following publication types will be excluded from the review:
- Animal models
- Narrative reviews, editorials, opinions.
The references identified by the electronic literature searches will be sifted in three stages. Screening for relevance first by title and then by abstract. Those papers which seem from their abstract (or if there is no abstract available) to be relevant will be retrieved and read in full. At each step, studies which do not satisfy the inclusion criteria will be excluded.
Data extraction strategy
A customised data extraction form will be based on those proposed by the NHS Centre for Reviews and Dissemination.8 Where possible, data will be extracted by one reviewer, and thoroughly checked by a second reviewer; any disagreements will be resolved by discussion.
Where available, data relating to the following outcomes will be extracted:
- all-cause mortality
- significant abnormal test findings
- change of management
- length of hospital stay
- adverse effects probably or possibly related to the test result
- adverse events probably or possibly caused by the process of testing.
Quality assessment strategy
We propose to assess to use criteria based on those proposed by the NHS Centre for Reviews and Dissemination8 (see Appendix 2) to assess the methodological quality of randomised trials which meet the inclusion criteria.
We will assess the methodological and reporting quality of case series studies which meet the inclusion criteria using a customised quality tool combining generic criteria proposed by the NHS Centre for Reviews and Dissemination8 and Chambers et al.10 with review-specific criteria, as follows:
- Generic criteria:
Were patients recruited prospectively?
Were patients recruited consecutively?
Were at least 90% of those included at baseline followed up (prospective studies only)?
Was loss to follow-up reported or explained (prospective studies only)?
Was follow-up long enough for important events to occur?
Were outcomes assessed using objective criteria or was blinding used?
Was an appropriate measure of variability reported?
- Review-specific criteria:
Were the patients’ age and ASA status adequately reported?
Was the operation type and/or risk classification adequately reported?
Were all operations elective?
Were all the tests conducted genuinely routine, or might some have been indicated?
Was a definition of normal or abnormal results provided?
Where appropriate, meta-analysis will be used to pool results, and summary statistics will be derived for each study and a weighted average of the summary statistics be computed across the studies.
We propose to survey the current protocol use in NHS Trusts in England and Wales to establish if the NICE Guidance is being adhered to. This will be carried out by sending paper questionnaires based on the Abacus study in 2005 to nurses involved in pre operative assessment care.
Analysis plan: final protocol
The objective of the study is the value of routinely testing FBC, U&E and PFT in patients with (1) no apparent clinical indication and (2) subgroups with common comorbidities.
The originally proposed approach to construct an economic model based on published literature has not proved possible due to the lack of published evidence on the effectiveness of these tests.
We have identified a large routine patient level data set on tests, surgical procedure and outcomes at Leeds Teaching Hospital Trust with an excess of 1m records. We propose to utilise econometric methods to address the following questions.
- Having adjusted for patient level characteristics and surgical intervention does not having the tests result in worse outcomes for patients without comorbidities?
- Having adjusted for patient level characteristics and surgical intervention does not having the tests result in worse outcomes for patients with comorbidities?
The subset of records for minor and intermediate risk surgical procedures have been identified on the basis of BUPA schedule of procedures. This is consistent with the methods used in our previous work on pre-operative assessment.
The outcomes to be used for these analyses are:
- Length of stay – continuous in days
- Readmission within 30 days
- Hospital mortality
For categorical variables (Readmission within 30 days and Hospital mortality) we will logistic regression models; whilst for the continuous variable (length of stay) we will estimate linear regression models.
The pre-specified conditioning variables in each analysis will be:
- Age –in years
- Sex (1 = female, 0 = male)
- socio-economic status – IMD04_decile
- Primary Diagnosis
- Secondary Diagnosis
- Surgical Procedure
The key explanatory variables will be:
- Full Blood Count;
- U&E; and
- Pulmonary function test
These will be entered as dummy variables. We will also enter joint dummy variables for each possible combination of these tests; e.g. FBC and U&E.
The models will be estimated separately for patients with and without comorbidities. Comorbidities will be modelled in two ways. First we will use the presence or absence of a secondary diagnosis as evidence of a comorbidity. Secondly, we will use whether patients had additional tests as a proxy for the presence of a comorbidity. We believe the second measure may be a more sensitive measure for the presence of a comorbidity, although obviously less specific than the recorded secondary diagnosis.
The outputs of these analyses will be six separate models assessing whether the absence of any combination of the three tests is associated with a difference in any of the three measures of outcome. The models will be assessed using standard statistical measures for goodness of fit, specification, and collinearity.
We will report the models in full and whether there is a statistically significant relationship between the absence of any combination of the three tests and length of stay, 30 day re-admission and hospital mortality.
Where the models report a significant relationship we will examine the costs incurred for the tests and the costs associated with the different outcomes in order to assess the likely value of the tests to the NHS. Given the extremely large number of observations available for these analyses, we judge that the absence of statistically significant relationship is sufficient to treat absence of evidence as evidence of absence.
Examining the value of more research
The estimated models can be used to explore the value of information associated with further research into these tests. The standard errors on the model parameters will provide a measure of the uncertainty associated with the relationship between the presence or absence of the test results and patient outcomes. Should we find evidence of a relationship between the routine tests and any of these outcomes, it will be appropriate to examine the value of undertaking further research, such as a prospective randomised controlled trial of these test, to inform future policy.
Using monte carlo simulation it is possible to simulate a distribution for the expected outcomes. The simulated distributions will describe the probability that the use of the routine tests are associated with difference in each of the outcomes and by extension the risk that using the central estimate to guide practice will lead to making the wrong decision. By attaching a value to each of the outcomes, e.g. the cost of a readmission or the value of a statistical life, it will be possible to attribute a value to reducing the risk of making the wrong decision through further research.
NIHR Journals Library, Southampton (UK)
Czoski-Murray C, Jones ML, McCabe C, et al. What is the Value of Routinely Testing Full Blood Count, Electrolytes and Urea, and Pulmonary Function Tests Before Elective Surgery in Patients with No Apparent Clinical Indication and in Subgroups of Patients with Common Comorbidities: A Systematic Review of the Clinical and Cost-Effective Literature. Southampton (UK): NIHR Journals Library; 2012 Dec. (Health Technology Assessment, No. 16.50.) Appendix 3, Final protocol.