PT  - JOURNAL ARTICLE
AU  - Thomas A Willis
AU  - Suzanne Hartley
AU  - Liz Glidewell
AU  - Amanda Farrin
AU  - Michelle Collinson
AU  - Michael Holland
AU  - Paul Carder
AU  - Cheryl Hunter
AU  - David Meads
AU  - Claire Hulme
AU  - Robbie Foy
TI  - A cluster-randomised evaluation of an adaptable implementation strategy targeting ‘high impact’ indicators in primary care
AID  - 10.3399/bjgp18X696917
DP  - 2018 Jun 01
TA  - British Journal of General Practice
PG  - bjgp18X696917
VI  - 68
IP  - suppl 1
4099  - http://bjgp.org/content/68/suppl_1/bjgp18X696917.short
4100  - http://bjgp.org/content/68/suppl_1/bjgp18X696917.full
SO  - Br J Gen Pract2018 Jun 01; 68
AB  - Background Recognised gaps between evidence and practice in primary care present particular implementation challenges when addressing multiple priorities.Aim To evaluate the effectiveness of a multifaceted, adaptable implementation package targeting four different ‘high impact’ indicators.Method We undertook two parallel, pragmatic cluster randomised trials using balanced incomplete block designs with parallel process evaluation. General practices in West Yorkshire, UK, were recruited using an ‘opt out’ process. The adaptable implementation package included audit and feedback, educational outreach visits and computerised support with embedded behaviour change techniques tailored to each indicator. Practices were randomised to packages targeting either type 2 diabetes control or risky prescribing of non-steroidal anti-inflammatory drugs, or packages targeting either anticoagulation in atrial fibrillation or blood pressure control in patients at high risk of cardiovascular events. Respective primary endpoints comprised: achievement of all recommended levels of haemoglobin A1c, blood pressure and cholesterol; risky prescribing levels; anticoagulation prescribing; and achievement of recommended blood pressure levels. Outcomes at 11 months used routinely collected data.Results 178 out of 244 eligible practices participated. The implementation package reduced risky prescribing (odds ratio 0.82; 97.5% confidence interval 0.67 to 0.99). There was no effect on other primary endpoints.Conclusion This highly pragmatic, robust evaluation suggests the value of targeting risky prescribing, given predictable population reductions in avoidable morbidity, deaths and hospital admissions. However, in broad terms, an adapted ‘one-size-fits-all’ approach did not consistently work, with no improvement for other targeted indicators.