Journal of Clinical Epidemiology
Volume 60, Issue 11 , Page 1203, November 2007

Erratum for “A new preference-based analysis for randomized trials can estimate treatment acceptability and effect in complaint patients” [J Clin Epidemiol 59 (2006) 685–696]

  • S.D. Walter

      Affiliations

    • Department of Clinical Epidemiology and Biostatistics, McMaster University, HSC-2C16, 1200 Main St West, Hamilton, Ontario, L8N 3Z5 Canada
    • ,
  • Gordon Guyatt

      Affiliations

    • Department of Clinical Epidemiology and Biostatistics, McMaster University, HSC-2C16, 1200 Main St West, Hamilton, Ontario, L8N 3Z5 Canada
    • Department of Medicine, McMaster University, 1200 Main St West, Hamilton, Ontario, Canada
    • ,
  • Victor M. Montori

      Affiliations

    • Department of Medicine, Mayo Clinic College of Medicine, Mayo E17-96, 200 First St SW, Rochester, MN, 55905-0001, USA
    • ,
  • R. Cook

      Affiliations

    • Department of Statistics and Actuarial Science, University of Waterloo, 200 University Ave, Waterloo, Ontario, N2L 3G1 Canada
    • ,
  • K. Prasad

      Affiliations

    • Department of Neurology, All India Institute of Medical Sciences, Ansari Nagar, New Delhi, PIN-110029, India

published online 08 October 2007.

Article Outline

 

In the above-mentioned article, in Table 8 (p. 692), the symbols p4 and p5 are reversed. The proportion of patients insisting on surgery should be p5 (and has the value as shown, 0.045), while the proportion of patients insisting on medical care should be p4 (value as shown, 0.003). These symbols are described correctly in the last paragraph of page 692, and the remainder of the calculations is unaffected.

Table 8. Example 4: Strokea rates in the NASCET studyb of carotid stenosis, by assigned and actual treatments
a) Outcome rates, by assigned and actual treatments
Assigned treatment
Medical careSurgery
FreqRate (%)FreqRate (%)
Actual treatmentMedical care61/31619.31/1100.0
Surgery1/156.726/3278.0
Total62/33118.727/3288.2
b) Estimated preference distribution
Compliersp1 = 0.952
Surgery-insistersp5 = 0.045
Medical-insistersp4 = 0.003
c) Estimated treatment effects by preference-based and traditional analyses
AnalysisEffect measure
RRRD
ITT0.4390.105
Per protocol0.4120.114
As-treated0.4040.117
Preference-based, compliers0.4210.110

aFatal and nonfatal ipsilateral strokes.

bData based on data in [9]. Ignores three incorrectly randomized patients included in ITT analysis. Six patients in the group randomized to medical care had strokes and subsequent surgery: these strokes are assumed to have been ipsilateral, and are therefore attributed as study events in the medical care group. Assumes one outcome event in the patient who crossed over from medical care to surgery. Other data are based on values reported in NASCET ITT and per protocol analyses.

The corrected Table 8 appears below.

PII: S0895-4356(07)00358-7

doi:10.1016/j.jclinepi.2007.09.005

Refers to article:

  • A new preference-based analysis for randomized trials can estimate treatment acceptability and effect in compliant patients , 27 March 2006

    S.D. Walter, Gordon Guyatt, Victor M. Montori, R. Cook, K. Prasad
    Journal of Clinical Epidemiology July 2006 (Vol. 59, Issue 7, Pages 685-696)

Journal of Clinical Epidemiology
Volume 60, Issue 11 , Page 1203, November 2007

Article Tools