If you wish to respond to a paper or other item already published in the BJA, please go to the abstract/full text version of that item and click on the link "E-Letters: Submit a response to the article".
Electronic Letters to:
|
|
E-letters: Electronic letters published:
-
![[Read E-letter]](/icons/misc/sectionDOWN.gif)
IL-18 and SC5b-9 for predicting neurocognitive dysfunction after cardiopulmonary bypass
- Rajappan-Nair A Kumar, Clare Cann, Judith Hall, Potteth Sudheer, Antony Wilkes (6 June 2007)
-
The perils of dichotomisation
- R Peter Alston (11 May 2007)
|
|
|||
|
Rajappan-Nair A Kumar Wales College of Medicine, Clare Cann, Judith Hall, Potteth Sudheer, Antony Wilkes
Send letter to journal:
|
We are grateful for the opportunity to respond to the comments made by Dr Alston about our study. We would like to point out that the use of one standard deviation (SD) in two or more of the nine neurocognitive tests to define neurocognitive dysfunction is not an arbitrary one. The 1 SD rule (one standard deviation decline on two or more tasks) and the 20- 20 rule (20% decline on 20% tasks) are the two most commonly applied means of examining postoperative cognitive dysfunction after CABG1. Conflicting reports exist in literature regarding the sensitivity and specificity of these two rules2,3. As regards the use of nine cognitive tests in our study, we chose to include the core tests from the recommendations of the “Statement of consensus on assessment of neurobehavioral outcomes after cardiac surgery”4. These were the same battery of tests used in our earlier study5. We would also like to point out that we did not make use of z scores in our previous study, as suggested by Dr. Alston. We do agree with the observation that a continuous variable such as a composite z score rather than a binary ‘impairment/no impairment’ outcome would have been more sensitive. We also accept that evidence has emerged since the completion of our study that there is a potential for false positives with the 1 SD rule6 and with increasing number of tests in the battery7. We feel that our study, inspite of any potential limitations, will stimulate interest in the role of IL-18, a key pro-inflammatory cytokine, in neurocognitive dysfunction following cardiac surgery. References: 1.Lewis MS, Maruff P, Silbert BS, Evered LA, Scott DA. The sensitivity and specificity of three common statistical rules for the classification of post-operative cognitive dysfunction following coronary artery bypass graft surgery Acta Anaesthesiol Scand 2006; 50:50-57 2.Mahanna EP, Blumenthal JA, White WD, et al. Defining neuropsychological. dysfunction after coronary artery bypass grafting. Ann Thorac Surg 1996; 61:1342– 1347 3.Rasmussen LS, Larsen K, Houx P, Berkav A, Tapp E, Zenz M. The assessment of postoperative cognitive function. Acta Anaesthesiol Scand 2001: 45: 275– 89. 4.Murkin JM, Newman SP, Stump DA, Blumenthal JA. Statement of consensus on assessment of neurobehavioral outcomes after cardiac surgery. Ann Thorac Surg 1995; 59: 1289-95. 5.Zamvar V, Williams D, Hall J, et al. Assessment of neurocognitive impairment after off-pump and on-pump techniques for coronary artery bypass graft surgery: prospective randomised controlled trial. BMJ 2002; 325: 1268-71 6.Raymond PD, Hinton-Bayre AD, Radel M, Ray MJ, Marsh NA. Assessment of statistical change criteria to define significant change in neuropsychological test performance following cardiac surgery. Eur J Cardiothorac Surg 2006; 29: 82-8. 7.Lewis MS, Maruff P, Silbert BS, Evered LA, Scott DA. Detection of Postoperative Cognitive Decline After Coronary Artery Bypass Graft Surgery is Affected by the Number of Neuropsychological Tests in the Assessment Battery Ann Thorac Surg 2006; 81: 2097 - 2104. Conflict of Interest:None declared |
|||
|
|
|||
|
R Peter Alston, anaesthetist Royal Infirmary of Edinburgh
Send letter to journal:
|
Dichotomisation of quatitative measures can result in loss of information about individual differences, loss of effect size and power in the case of bivariate relationships, spurious statistical significance and overestimation, and can potentially overlook nonlinear relationships, or less reliable measurements.1 Unfortunately, Kumar and colleagues have dichotomized the results of the cognitive tests using an arbitrary definition of one standard deviation. The cognitive measurements and the biological measures of inflammation are all continuous variables and as such, linear rather than logistic regression analysis should have been used. 2 Another limitation of their analytical approach is that they have used nine cognitive tests. This will lead to an increased possibility of statistical significance being found by chance when none exist. A robust approach as their group has done previously, would have been to create dimensionless z scores for each test and then summed them to give a single measure of cognition. 3,4 This could have then been used as the outcome variable in the linear regression analysis to determine any association with IL18 or SC5b-9. A lack of robust statistical methodology in research examining cognitive decrements associated with heart surgery has clouded the area. It behooves authors to use appropriate methods and reviewers and editors to reject papers that do not. References 1. MacCallum RC, Zhang S, Preacher KJ, Rucker DD. On the practice of dichotomization of quantitative variables. Psychol Methods 2002; 7: 19-40. 2. Altman DG, Royston P. The cost of dichotomising continuous variables. BMJ 2006; 332: 1080. 3. Zamvar V, Williams D, Hall J et al. Assessment of neurocognitive impairment after off-pump and on-pump techniques for coronary artery bypass graft surgery: prospective randomised controlled trial. BMJ 2002; 325: 1268. 4. Robson MJ, Alston RP, Deary IJ et al. Cognition after coronary artery surgery is not related to postoperative jugular bulb desaturation. Anesth Analg 2000; 91: 1317-26. Conflict of Interest:None declared |
|||