Premium
Psychometric properties of a Swedish version of the Consequences of Screening – Breast Cancer questionnaire
Author(s) -
Bolejko Anetta,
Brodersen John,
Zackrisson Sophia,
WannHansson Christine,
Hagell Peter
Publication year - 2014
Publication title -
journal of advanced nursing
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.948
H-Index - 155
eISSN - 1365-2648
pISSN - 0309-2402
DOI - 10.1111/jan.12385
Subject(s) - rasch model , psychosocial , breast cancer , clinical psychology , cronbach's alpha , psychometrics , medicine , anxiety , breast cancer screening , reliability (semiconductor) , lung cancer screening , lung cancer , mammography , cancer , psychology , psychiatry , oncology , developmental psychology , power (physics) , physics , quantum mechanics
Abstract Aim To evaluate the psychometric properties of a questionnaire addressing psychosocial consequences of false‐positive mammographic screening. Background The Consequences of Screening ‐ Breast Cancer and Lung Cancer questionnaires target psychosocial consequences of false‐positive cancer screening. The Consequences of Screening ‐ Breast Cancer questionnaire and ten items not considered lung cancer specific from the Lung Cancer questionnaire have been adapted for use in mammographic screening in Sweden, but remain psychometrically untested. Design Instrument development paper with psychometric cross‐sectional and test–retest design. Methods Twelve scales of a Swedish questionnaire version were tested by the Rasch model and traditional psychometric methods. Women with false‐positive (Group I, n = 640) and negative (Group II , n = 802) screening mammography responded to the study questionnaire and the Nottingham Health Profile during 2009–2011. Results Iterative analyses resulted in nine scales demonstrating Rasch model fit, but all scales exhibited poor targeting with relatively large floor effects. Corrected item‐total correlations exceeded the recommended criterion. Score differences between Groups I and II and correlations with Nottingham Health Profile sections followed an expected pattern. Cronbach's α and test–retest reliability was acceptable for group‐level assessments for ten and seven scales, respectively. Conclusions Five scales (Sense of dejection, Anxiety, Behavioural, Sleep and Existential values) of the Swedish questionnaire version demonstrated the best psychometric properties. Other scales should be used more cautiously. Although filling an important gap, causes of concern were identified across scales. The questionnaire should therefore be considered for group‐level assessments rather than for measurement of individual degrees of psychosocial consequences.