Outcome measures

Each consultation recording was analysed, by listening to the audio-recordings, by two independent raters using two measures. One measure was the 12-item Observing Patient Involvement (OPTION-12) scale, which has good discriminative validity, concurrent validity, and interrater and intra-rater reliability [20, 21]. It contains 12 items scored on a five-point scale: (0) the behaviour was not observed; (1) a minimal attempt is made; (2) the behaviour is observed with a minimal skill level; (3) the behaviour is executed to a good standard; and (4) the behaviour is executed to a high standard. Total scores were re-scaled to 0–100. A second measure was 5 items (1 subscale) of the Assessing Communication about Evidence and Patient Preferences (ACEPP) tool. This was used as the OPTION scale does not specifically evaluate communication of the quantitative benefits and harms of the options. It has good reliability and has been used previously to assess evidence communication in consultations [22, 23]. The items rate clinicians’ performance in describing the benefits/harms in terms of patient outcomes, the likelihood of benefits/harms, and the evidence source. Items were scored as: the behaviour was not observed (0); behaviour was observed at a basic level (0.5); or observed to an extended level (1).

To establish scoring reliability, three of us (MB, EG, TH) independently rated an initial sample of recordings and responses were discussed until agreement was reached. Two of us (MB, EG) independently rated the remainder. Any rating discrepancies were resolved by a third person (TH). The two raters also extracted verbatim any mention of antibiotic benefits and harms.

Patients’ perceptions of their involvement in the decision-making process were measured using the CollaboRATE-5 scale (score range 0 to 5) [24, 25]. It asks three questions about what occurred in the consultation: 1) deliberation of the health issue, 2) exploration of patient preferences, and 3) integration of patient preferences [25]. The scale has demonstrated significant discriminative validity, excellent intra-rater reliability and concurrent validity with other measures of SDM [24].

Decisional conflict is a condition of uncertainty about options involving trade-offs and potential for regret. It was measured using the 10-item low literacy decisional conflict scale [26]. In this study, patients’ feelings conflict about whether they felt that their decision (using antibiotics or not) was the best for them was assessed. The scale has good validity and reliability [26]. The low-literacy version uses a question-and-answer format with three response options (yes, no, unsure), with scoring from 0 (low decisional conflict) to 100 (high decisional conflict) [27].

Patients’ confidence in decision-making was measured using four items from the decision self-efficacy scale [28], which has high internal consistency [29]. Scoring of each item is from 0 (not at all confident) to 100 (very confident).