Development and validation of the self-efficacy for medical communication scale.

Abstract

12124 Background: Most studies of clinician-patient communication use scales created ad hoc with unknown validity. To provide a standard measure for future studies, we developed and validated a new scale of clinician-reported skills in communicating difficult news: the Self-Efficacy for Medical Communication (SEMC) scale. Methods: Using evidence-based scale development guidelines, we created 16 items sampling a range of communication skills, including “Disclose difficult news in manageable chunks, so the patient is not overwhelmed,” and “Determine how to present information based on the patient’s emotional state.” Items are rated on Likert scales from 1 ( cannot do at all) to 10 ( highly certain can do). We constructed two forms—one assessing communication with patients and one with family—using identical items but replacing “patient” with “family/caregiver.” We examined the convergent and discriminant validity of the SEMC (correlations with similar and dissimilar measures) as well as its reliability and factor structure. A total of 221 clinicians working in oncology settings (physicians, nurses, medical students) completed measures online. Convergent measures included medical communication items from past studies; the Self-Perceived Communication Competence Scale to measure communication ability outside the medical realm; and the General Self-Efficacy and Occupational Self-Efficacy scales to measure overall self-efficacy/confidence. Discriminant measures included the Ten Item Personality Inventory to measure personality factors; the Maslach Burnout Inventory to measure job burnout; and the Satisfaction with Life Scale to measure well-being. Finally, the Marlowe-Crowne Social Desirability (MCSD) scale measured motivation to “look good” in responding to survey questions. Results: Mean scores were similar for the patient (126.36) and family (127.09) forms (max score 160), both with excellent reliability ( alphas =.94,.96, respectively). Because these forms were almost perfectly correlated ( r =.95, p <.001), we used only the patient form in subsequent analyses. Factor analysis demonstrated that the SEMC measures a unitary construct ( eigenvalue = 9.0). Its mean correlation was higher with convergent ( r =.46) than discriminant measures ( r =.22), supporting its validity. Moreover, its correlation with the MCSD was small ( r =.28) and no larger than between the MCSD and other measures, indicating minimal social-desirability effects. Finally, no differences emerged for gender or profession; higher scores did correlate with age ( r =.29, p <.001) and years working in oncology ( r =.18, p =.01). Conclusions: Our findings support the SEMC’s validity and reliability. Scores on the patient and family forms were similar, indicating that either may be used. The SEMC provides a useful tool for measuring clinician-rated communication skills in future research, ultimately allowing standardization across studies.