Table 1 Data-driven Quality Improvement in Primary Care (DQIP)

Trial: The DQIP trial was a cluster randomised, stepped wedge trial in 33 practices from one Scottish health board (NHS Tayside) which aimed to reduce high-risk prescribing of nonsteroidal anti-inflammatory drugs (NSAIDs) and/or selected antiplatelet agents. All practices received the intervention but were randomised to one of 10 different start dates.

Intervention: The DQIP intervention comprised education, an informatics tool and a financial incentive. In DQIP, the researchers delivered education and training to the general practices (clusters) with high fidelity (of form and function); the general practices were then free to organise themselves as they saw fit to deliver the intervention to patients (fidelity of function, but variation in form). For the process evaluation design, the DQIP intervention was conceptualised as two interventions; the intervention delivered to clusters (intervention 1), and the intervention delivered to patients (intervention 2) as there may be different processes working at the cluster and individual patient level. Data were collected on intervention 1 in all practices and in a purposive sample for intervention 2, using a comparative case study design for both interventions.

Purpose: The DQIP mixed method process evaluation aimed to qualitatively explore how patients and practices responded to the intervention and quantitatively examine how a change in high risk prescribing was associated with practice characteristics and implementation processes.

Qualitative design: A mixed method multiple comparative case study with general practices as the units of analysis.

Quantitative design: Prespecified analysis to explore associations between practice characteristics, implementation processes and change in prescribing.

Sample: One practice was sampled per cohort of the trial. Ten general practices were purposively sampled based on their initial adoption of the intervention, four practices which rapidly implemented the intervention and six who were initial implementation failures.

Data collection: Routine data collected during the trial were used to inform the quantitative part of the process evaluation, and to mirror the trial design, a case was selected for qualitative data collection from each cohort in the trial.

Conceptual and theoretical framework: Framework for process evaluation design [45] and Normalisation Process Theory [34].

Analysis: An in-depth description of each case was constructed detailing the practice characteristics and perceptions of all staff who participated in the interviews, with additional data from the educational outreach observation and informal interviews. Deductive theoretical analysis using the Normalisation Process Theory was also conducted. The Framework technique of matrixes facilitated detailed exploration by theoretical construct theme, practice and cross- and within-case comparisons. Thematic and theoretical saturation was achieved.

What did the case study process evaluation design reveal about context? Case study design illustrated that to achieve effective implementation agreement that the topic (NSAIDs and antiplatelets) was important among all clinical staff was fundamental. In addition, that practices made plans early in the process to implement the intervention including responsibility for the work and regularly evaluated their progress. Also, how practices internally organised to do the work varied illustrating this was not important for effective evaluation. Case study design was important for illustrating that implementation failure occurred at different stages depending on the practice culture and context, illuminating the differences in organisational processes and the contextual and organisational factors which impacted on effective implementation. The case study’s holistic approach to understanding how the context and culture within each practice influenced processes was key to explaining whether the intervention worked or not. Also, practice context was not fixed, so most practices adapted and were able to deliver some elements.

DQIP case study design model