D. Bryant, P. Stratford, R. Marx, S. Walter, and G. Guyatt reply:
Dr. Lack1 and Professor Lingard2 are both quite right that there is good reason to think that patients' assessment of their status obtained on the day of surgery may be distorted by contextual factors. Indeed, it was concern about such effects that motivated our study. At the same time, it is possible that patients can, despite the stresses of the day, report their status accurately.
As always, empirical evidence trumps theoretical considerations. We demonstrate through three separately tested hypotheses that measures obtained on the day of surgery are similar to measures obtained four weeks preoperatively in this population. So, as it turns out, theoretical concerns have proved groundless for this particular population.
It is still possible that there is a small subpopulation—the very anxious, for instance—to whom our results do not apply. To test this interesting hypothesis that Professor Lingard has suggested, we repeated all analyses on the quartile of our population with the greatest emotional dysfunction on the Short Form-36 mental health domain on the day of surgery. The results in this population proved essentially the same as that for other patients. Thus, we find no evidence of a subpopulation to whom our overall results do not apply.
In her commentary, Professor Lingard2 expresses concern over our decision to report the Knee Injury and Osteoarthritis Outcome (KOOS) score as an overall aggregate score. As it turns out, the statement we provided to defend the decision proved open to misinterpretation. We stated that "each domain yielded similar results, and thus, for ease of reporting, we computed an overall aggregate score only …" By this statement, we did not mean that the scores on each domain were similar but rather that the results of the analyses for each domain were similar and that given that this study already reports the results of five questionnaires, to report similar results in all domains of each questionnaire might be unnecessarily burdensome to the reader. In fact, the mean score and standard deviation was 56.3 ± 18.5 for the symptom domain, 59.0 ± 19.0 for the pain domain, 68.9 ± 18.9 for the function domain, 35.7 ± 23.9 for the sport and recreation domain, and 29.8 ± 18.0 for the quality-of-life domain. These results are consistent with those reported in previous studies.
Dr. Lack1 and Professor Lingard2 are correct that, although these results may apply to other populations, we will not be confident until direct testing in these other populations takes place. Further experiments addressing the issue of day-of-surgery ratings will, therefore, be most welcome.
These letters originally appeared, in slightly different form, on . They are still available on the web site in conjunction with the article to which they refer.