The dimensional structure of the MacNew Health Related Quality of Life questionnaire: A Mokken Scale Analysis O. Friedrich a, ⁎, J. Sipötz a,b , W. Benzer c , E. Kunschitz a,b , S. Höfer d a Karl Landsteiner Institute for Scientific Research in Clinical Cardiology, Vienna, Austria b Department of Cardiology, Hanusch Krankenhaus, Vienna, Austria c Department of Interventional Cardiology, Academic Hospital, Feldkirch, Austria d Department of Medical Psychology, Innsbruck Medical University, Austria abstract article info Article history: Received 9 February 2015 Received in revised form 16 April 2015 Accepted 17 April 2015 Keywords: MacNew Mokken Scale Analysis Health related quality of Life Coronary Artery Disease Objective: The MacNew Health related Quality of Life Questionnaire is a widely used instrument for the assess- ment of health related quality of life in cardiac patients. The study addresses for the first time the dimensional structure of the MacNew with Mokken Scale Analysis (MSA). Methods: Separate exploratory MSA of the MacNew was conducted in a large Spanish (n = 1012) and a medium sized Austrian sample (n = 262) of patients with Coronary Artery Disease (CAD) after Percutaneous Coronary Intervention (PCI). The results of both samples were summarized in a synthesis model. Confirmatory MSA and Confirmatory Factor Analysis (CFA) were used to evaluate the model. Results: The synthesis model comprises 21 items forming a unidimensional sum scale of moderate strength. On the level of subdomains we define two strong unidimensional subscales (restriction: 6 items, and emotional: 10 items) and two smaller item sets (symptoms: 2 items and social: 3 items). 5 items were excluded due to low scalability in both samples. Conclusion: Our results generally support the use of the MacNew Global score, with the limitation, that five items may be questionable with regard to scalability. On the level of unidimensional subscales MSA suggests to differ- entiate between a six-item restriction scale and a ten-item emotional scale. The study demonstrates that Mokken Scale Analysis complements the results of factor analysis and can contribute to a more comprehensive under- standing of the dimensional structure of Health-related Quality of Life questionnaires. © 2015 Elsevier Inc. All rights reserved. Introduction The MacNew questionnaire is a widely used instrument for disease specific measurement of Health Related Quality of Life (HRQL) in cardiac patients. The questionnaire is originated in the Quality of Life after Myo- cardial Infarction (QLMI) interview, introduced by Oldridge et al. [1,2] in order to investigate the effects of cardiac rehabilitation after myocardial infarction in anxious and/or depressed patients. The QLMI instrument was generated by a clinimetric approach through interviews with phy- sicians, nurses, health care professionals and patients with myocardial infarction leading to a set of 26 items covering five domains (symptoms, restrictions, confidence, self-esteem and emotional functioning). Based on the QLMI Lim et al. [3] and Valenti et al. [4] developed the patient self-administered MacNew questionnaire. The MacNew con- tains 27 questions, 24 items of the QLMI and three newly added items, rated by a seven point Likert scale. The MacNew measures HRQL in three domains (emotional, physical, and social). A global scale can be calculated by summing up all items [4]. The MacNew has been made available in over 80 countries and validated in over 16 languages [5] in patients with a variety of cardiac conditions ranging from myocardial infarction to angina, acute coronary syndrome, heart failure and in pa- tients with a pacemaker and is recommended as a core heart disease quality of life instrument [6]. The three-dimensional factor structure of the MacNew, originally proposed by Valenti et al. [4] for the Australian English version on the basis of Principal Component Analysis (PCA) is characterized by some ambiguity. About half of the items (12/26 — one item was excluded from analysis) showed substantial cross-loadings and were therefore allocated to more than one subscale. The three-factor solution was generally accepted for different language versions of the MacNew even if the results of factor analysis differed sometimes partly, especially in smaller samples and in particular with regard to items of the physical and social subscales [6]. In order to analyse HRQL on the level of the subdomains a unidimensional factor structure would be highly prefera- ble. However, given the nature of HRQL (being multidimensional and Journal of Psychosomatic Research 79 (2015) 43–48 ⁎ Corresponding author at: Karl Landsteiner Institute for Scientific Research in Clinical Cardiology, Hanusch Krankenhaus, Heinrich-Collin-Str. 30, A-1120 Vienna, Austria. Tel.: +43 1 680 212 3203; fax: +43 1 910 21 85219. E-mail address: o.friedrich@gmx.at (O. Friedrich). http://dx.doi.org/10.1016/j.jpsychores.2015.04.007 0022-3999/© 2015 Elsevier Inc. All rights reserved. Contents lists available at ScienceDirect Journal of Psychosomatic Research