Defining Dysfunction: Group Means Versus Incidence Analysis—A Statement of Consensus

This statement is intended to amplify and extend the previously published Consensus Statement [1Murkin JM Newman SP Stump DA Blumenthal JA Statement of consensus on assessment of neurobehavioral outcomes after cardiac surgery.Ann Thorac Surg. 1995; 59: 1289-1295Abstract Full Text PDF PubMed Scopus (523) Google Scholar]in which issues related to the timing and nature of cognitive tests to be employed in cardiac surgical patients were discussed and ratified. At that meeting a broad spectrum of expert investigators convened to establish a working consensus on methodologic issues related to neurobehavioral dysfunction occurring in the postoperative period after cardiac operations. The current Statement refers specifically to the cognitive tests (Trails A, Trails B, Grooved Pegboard, and Rey Auditory Verbal Learning) as listed in the 1995 Consensus Statement [1Murkin JM Newman SP Stump DA Blumenthal JA Statement of consensus on assessment of neurobehavioral outcomes after cardiac surgery.Ann Thorac Surg. 1995; 59: 1289-1295Abstract Full Text PDF PubMed Scopus (523) Google Scholar]. It was precirculated among all current conference registrants as well as available participants in the previous Consensus meeting to solicit their input and to encourage the distribution and further discussion of this document among interested individuals at their own institutions. An open forum meeting was then held at the conclusion of the Outcomes ’97 meeting, during which all items in this Consensus Statement were discussed point by point until general consensus was achieved. 1.Individual change scores represent the most sensitive means of detecting clinical factors affecting postoperative neurobehavioral change in cardiac surgical patients. There is a need to identify individuals experiencing postoperative neurobehavioral change within various study groups to identify potential underlying etiologic mechanisms. This is further impetus for the development of a minimum decrement in performance, identified by change score results, and definable as the threshold level for cognitive change. This threshold should be set to combine sufficient sensitivity with a high degree of specificity, to avoid false positive outcomes, and can best be achieved by knowing how much change might be expected in an appropriate control group [2Mahanna EP Blumenthal JA White WD et al.Cognitive decline following cardiac surgery a comparison of criteria for defining neuropsychological dysfunction.Ann Thorac Surg. 1996; 61: 1342-1347Abstract Full Text PDF PubMed Scopus (238) Google Scholar, 3Stump DA Selection and clinical significance of neuropsychological tests.Ann Thorac Surg. 1995; 59: 1340-1344Abstract Full Text PDF PubMed Scopus (101) Google Scholar].2.Group mean score is influenced by the variable performance on test scores of the study group as a whole. In a cardiac surgical population, certain individuals will be expected to score at an improved level as a consequence of previous exposure to the testing techniques. Individuals experiencing postoperative neurobehavioral dysfunction often will score less well than their baseline measure. As a consequence, a group mean score will reflect both the individuals whose performance is enhanced as a consequence of practice and those whose performance is compromised due to postoperative neurobehavioral dysfunction. Accordingly the net result may be no significant group differences from baseline score. Therefore, the group mean score does not necessarily detect a significant decline in a subset of individuals in the group.3.Consistency of setting for cognitive testing is a prerequisite for valid data collection. Cognitive test performance can be affected by a variety of extraneous factors to influence test scores. The test-retest variability attributable to changes in test setting and examiner is not known but is likely to be test-specific. To optimize reproducible results, subjects should ideally be tested by appropriately trained individuals in a quiet, neutral environment designed to minimize distractions and interruptions. For follow-up testing a similar setting and time of day, as well as the same tester, should ideally be employed. Significant variability in test setting between follow-ups will likely reduce the probability of detecting a change between follow-ups.4.Scores on repeat cognitive testing may be influenced by the “practice effect.” Performance of the various tests employed in a series of cognitive tests is influenced by previous experience and familiarity with tests of a similar nature, known as the “practice effect.” As such there is a small but quantifiable improvement in performance associated with repeat administration of cognitive tests independent of the use of parallel forms [4Newman SP Analysis and interpretation of neuropsychologic tests in cardiac surgery.Ann Thorac Surg. 1995; 59: 1351-1355Abstract Full Text PDF PubMed Scopus (78) Google Scholar]. The magnitude of the practice effect is test-specific and depends on a number of factors, in particular age and education. The failure to correct for the practice effect reduces the likelihood of detecting a change between follow-ups. Torkel Åberg, MD (University Hospital of Northern Sweden, Umea, Sweden)Joe Arrowsmith, MD (UCL Hospitals, London, UK)Denise Barbut, MD (Cornell University, New York, New York)Michael A. Borger, MD (University of Toronto, Toronto, Canada)Louis M. Borowicz, PhD (Johns Hopkins University, Baltimore, Maryland)Robert S. Brooker, MD (Bowman Gray School of Medicine, Winston-Salem, North Carolina)Chris Brown-Mahoney, PhD (University of Minnesota, Minneapolis, Minnesota)David J. Chambers, PhD (St Thomas’ Hospital, London, UK)Laurie K. Davies, MD (University of Florida, Gainesville, Florida)Jan Diephuis, MD (Academic Hospital of Utrecht, Utrecht, the Netherlands)Michael Erb, MD (University of Tübingen, Tübingen, Germany)Kazuhiko Hanzawa, MD, PhD (Niigata University School of Medicine, Niigata City, Japan)Eric J. Heyer, MD, PhD (Columbia-Presbyterian Medical Center, New York, New York)Daniel G. Knauf, MD (University of Florida College of Medicine, Gainesville, Florida)Stephen A. Mills, MD (High Point, North Carolina)Mark F. Newman, MD (Duke University, Durham, North Carolina)Arno Nierich, MD (Academic Hospital of Utrecht, Utrecht, the Netherlands)Wilfred B. Pugsley, MD (Middlesex Hospital, London, UK)Gary Roach, MD (Kaiser Permanante, San Francisco, California)Bertil Rosberg, MD, PhD (University Hospital of Malmo, Malmo, Sweden)Ola A. Selnes, PhD (Johns Hopkins University, Baltimore, Maryland)Kyosti A. Sotaniemi, MD (University of Oulu, Oulu, Finland)Marc St-Amand, MD (University of Western Ontario, London, Canada)Graham E. Venn, FRCS (St.