Development and quality assessment of the psychometric properties of the Self-Efficacy in Lifestyle Counselling scale (SELC 20 + 20) using Rasch analysis.

ABSTRACT

BACKGROUND: Globally as well as in Sweden, diseases that are caused by unhealthy lifestyle habits are the most common causes of death and disability. Even though there are guidelines that oblige all health-care professionals to counsel patients about lifestyle, studies have shown that it is not prioritized within healthcare. One reason for this among nurses has been shown to be lack of confidence in knowledge and counselling skills. This study aimed to develop, and quality assess the psychometric properties of an instrument to measure self-efficacy in lifestyle counselling. METHODS: An instrument inspired by an American instrument, following Bandura's recommendations for development of self-efficacy measures, was developed according to Swedish national guidelines for disease-prevention. The instrument was revised after cognitive interviews with nursing students, university teachers within health sciences, and clinical experts, then administrated to 310 nursing students at different levels in their education. The instrument was tested with Rasch Measurement Theory, with focus on dimensionality, local dependency, targeting, reliability, response category functioning, Rasch model fit, and differential item functioning by age, gender, educational level and previous health care education. RESULTS: The development of the instrument resulted in 20 + 20 items, 20 items about self-efficacy in knowledge, and 20 items about self-efficacy in ability to counsel persons about their lifestyle. The analyses showed that knowledge and ability are two different, but related, constructs, where ability is more demanding than knowledge. The findings provide support (considering dimensionality and local dependency) for that all 20 items within the knowledge construct as well as the 20 items within the ability construct can be summed, achieving two separate but related total scores, where knowledge (reliability 0.81) is a prerequisite for ability (reliability 0.84). Items represented lower self-efficacy than reported by the respondents. Response categories functioned as expected, Rasch model fit was acceptable, and there was no differential item functioning. CONCLUSIONS: The SELC 20 + 20 was found to be easy to understand with an acceptable respondent burden and the instrument showed good measurement properties.