Implementing actionable population-based feedback in UK primary care: a longitudinal optimisation study using Clinical Performance Feedback Intervention Theory
Problem
Electronic health record (EHR) data are often used to measure the quality of primary care provided to patients, reported to health professionals as ‘quality indicators’ or ‘performance measures’ in electronic feedback systems (e.g. audit tools). Such feedback varies widely in its effectiveness at improving patient care, and there is a lack of clarity regarding the mechanisms by which it works. We developed Clinical Performance Feedback Intervention Theory (CP-FIT) to address this gap (SAPC abstract ID 788), and used it to inform the design of a novel electronic feedback system that suggests actions to clinicians for patients with long-term conditions (LTCs) in primary care including hypertension and atrial fibrillation (the Performance Improvement plaN GeneratoR [PINGR]). Our aims were to: 1) optimise PINGR prior to its wider implementation; 2) empirically test and refine CP-FIT to derive wider learning for electronic feedback systems.
Approach
We recruited 15 GP practices in Salford to use PINGR between November 2016 and June 2017 inclusive. Quantitative data were collected remotely on how health professionals used the system, and its potential effects on patient care. Analyses included visual summaries, descriptive statistics, process mining, and comparisons with patients not viewed in the software. Qualitative data were collected from semi-structured interviews guided by Normalisation Process Theory and field notes. Framework Analysis was informed by CP-FIT.
Findings
Forty-eight users (16 GPs, 6 nurses, 5 pharmacists and 21 non-clinicians) participated for a median of 4 months (range 1–8). Thirty-eight were interviewed at baseline, with twenty-five interviewed a second time. PINGR was used on 227 separate occasions to facilitate the care of 765 patients. Practices adopted the system in different ways, though largely used it to take patient-level rather than the organisation-level action. Patients viewed in the system were 1.6 (95% CI 1.5-1.7) times more likely to improve in at least one quality indicator compared with those not viewed. Barriers and facilitators to the system’s success were explained by CP-FIT in terms of: the resources available to use it; its perceived advantages; how compatible it was with pre-existing beliefs and ways of working; the credibility of its data; the complexity of the clinical problems it highlighted; and the ability to act on its recommendations.
Consequences
Using PINGR may lead to improved care for patients with LTCs, though requires further testing in a larger scale evaluation. CP-FIT successfully modelled and explained how electronic feedback could be optimised in practice. Our results add to CP-FIT by providing examples of how: particular phenomena may occur and inter-relate in practice; additional concepts lie outside the model; and how novel feedback designs may address some of its recommendations. These findings have implications for the design and implementation of clinical performance feedback interventions and policy initiatives in primary care.