| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Ann Thorac Surg 2003;75:1066-1067
© 2003 The Society of Thoracic Surgeons
a Centre for Neuroscience, The University of Melboume, c/o 51 Leicester St3053 Carlton, Victoria, Australia
b School of Psychology, La Trobe University, c/o 51 Leicester St3053 Carlton, Victoria, Australia
c Centre for Anaesthesia and Cognitive Function, St. Vincent’s Hospital, Victoria St3050 Fitzroy, Victoria, Australia
e-mail: alex{at}neuro.mhri.edu.au
To the Editor:
We thank Dr Whitaker for his comments on our recent review of the statistical methods used in studies of postoperative cognitive decline (POCD) in patients who have undergone coronary artery surgical procedures [1]. He highlights the important point that the interpretation of statistical tests used to investigate POCD is constrained by both the design and the objectives of the study. Most of the studies cited in our review reduced the standard deviation (SD) index to a binary impairment/no impairment outcome. Importantly, these studies were mostly observational in nature and were designed primarily to determine the incidence of POCD. The use of a binary outcome measure is therefore understandable, although as suggested, alternative approaches exist. As Whitaker states, the SD index can also be treated as a continuous variable, more commonly known as a z score. Such an interpretation of the SD index is more appropriate when the study objective is not to determine incidence of POCD, but rather to test a specific hypothesis such as the effects of a putative neuroprotective agent on the severity of POCD [2] or to determine the association between a genetic risk factor and the severity of POCD [4].
We concur completely that the use of continuous measures provides greater statistical power and allows more sophisticated analyses than the use of binary outcome measures in studies designed to determine the nature and the severity of POCD. In fact, we [4] recently demonstrated that application of a binary clinical criterion for "cognitive impairment" in older people on a single occasion resulted in a 50% false-positive classification rate. This false-positive rate was reduced dramatically when the same criterion was applied serially and only those individuals who consistently met the criterion were rated as impaired. Further, when serial data collected in individuals were considered on a continuous scale rather than as a binary outcome (below or above the cutoff score), greater sensitivity to subsequent cognitive impairment was observed.
Like Whitaker, we believe the use of continuous data is crucial to understanding the nature and the severity of POCD in patients who have undergone coronary artery surgical procedures. We also believe that observational studies designed to estimate the incidence of POCD using binary classification criteria are likely to contain high rates of false-positive classifications [3].
References
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ANN THORAC SURG | ASIAN CARDIOVASC THORAC ANN | EUR J CARDIOTHORAC SURG |
| J THORAC CARDIOVASC SURG | ICVTS | ALL CTSNet JOURNALS |
