How do practice nurses and healthcare assistants deliver a behaviour change intervention to lower cholesterol and cardiovascular disease risk factors in people with severe mental illnesses in primary care (Primrose)?

Problem

People with severe mental illnesses (SMI) such as schizophrenia and bipolar have an increased risk of mortality from cardiovascular disease (CVD). We tested a primary care nurse/healthcare assistant led behavioural intervention (Primrose) in a cluster randomised controlled trial with 326 patients across 76 GP practices in England. We found no difference in total cholesterol between the intervention and routine care groups at 12 month follow up. It is not known how well practice nurses/healthcare assistants deliver behaviour change interventions. We carried out a fidelity assessment to determine the extent to which behaviour change techniques (BCTs) were used as intended within the intervention.

Approach

We aimed to evaluate the delivery of a primary care led behavioural intervention (Primrose) consisting of eight BCTs to help people with SMI lower their CVD risk. BCTs included behavioural goal setting, action planning, monitoring progress, involving supportive others, providing positive feedback, recording behaviour, coping with setbacks and forming habits. Practice nurses and healthcare assistants audio-recorded Primrose intervention appointments with recruited patients. A random 20% sample of audio recordings (n=87/431 comprising of 14/72 first appointments and 73/359 subsequent appointments) were transcribed and analysed to assess whether (1) BCTs of the Primrose intervention were delivered, (2) appropriate behavioural goals were set, (3) certain BCTs were delivered more than others and (4) whether there was a difference in fidelity of delivery between practice nurses and healthcare assistants. Transcripts were coded by two researchers using a fidelity checklist designed by the research team. Fidelity was assessed by examining the proportion of behavior change components specified in the protocol that were delivered in individual appointments. Inter-rater reliability was assessed using percentage agreement and Cohens kappa.

Findings

67.7% of BCTs were delivered as intended with the highest delivery of ‘reviewing progress with goals’ (90.2%) and the lowest delivery of ‘forming habits (47.8%)’. Patients set a goal in 13 out of the 14 first appointments analysed; however only 8/14 (57.1%) patients set a behavioural goal associated with lowering cholesterol, and no patients addressed statin adherence or initiation. Practice nurses had significantly higher fidelity, delivering more BCTs to protocol than healthcare assistants (79.5% vs 64.3%) (t = 2.23, p = 0.037). Inter-rater reliability for coding was high (86% with a Cohens kappa of 0.668).

Consequences

This study presents a rigorous approach to assessing fidelity to an evidence-based behaviour change intervention delivered in primary care. The moderate fidelity to intervention delivery indicates that training programmes should focus more on BCTs with lower fidelity (e.g. habit formation) and emphasise the importance of statin initiation where indicated. Tailoring the intervention to suit the skills of different primary care staff may also lead to improved fidelity and clinical outcomes.