Evaluating the impact of a very low-cost intervention to increase practices' engagement with data and change prescribing behaviour: a randomized trial in English primary care. (30th March 2021)

• Record Type: Journal Article
• Authors: Curtis, Helen J; Bacon, Seb; Croker, Richard; Walker, Alex J; Perera, Rafael; Hallsworth, Michael; Harper, Hugo; Mahtani, Kamal R; Heneghan, Carl; Goldacre, Ben
• Is Part Of: Family practice. Volume 38:Number 4(2021)
• Page Start: 373
• Page End: 380
• Publication Date: 2021-03-30
• Subjects: anti-bacterial agents -- antimicrobial stewardship -- feedback -- general practice -- information dissemination -- information-seeking behaviour; Primary care (Medicine) -- Periodicals; Clinical medicine -- Periodicals; 616.005
• DOI: 10.1093/fampra/cmaa128
• Languages: English
• ISSNs: 0263-2136

Abstract

Abstract: Background: Unsolicited feedback can solicit changes in prescribing. Objectives: Determine whether a low-cost intervention increases clinicians' engagement with data, and changes prescribing; with or without behavioural science techniques. Methods: Randomized trial (ISRCTN86418238). The highest prescribing practices in England for broad-spectrum antibiotics were allocated to: feedback with behavioural impact optimization; plain feedback; or no intervention. Feedback was sent monthly for 3 months by letter, fax and email. Each included a link to a prescribing dashboard. The primary outcomes were dashboard usage and change in prescribing. Results: A total of 1401 practices were randomized: 356 behavioural optimization, 347 plain feedback, and 698 control. For the primary engagement outcome, more intervention practices had their dashboards viewed compared with controls [65.7% versus 55.9%; RD 9.8%, 95% confidence intervals (CIs): 4.76% to 14.9%, P < 0.001]. More plain feedback practices had their dashboard viewed than behavioural feedback practices (69.1% versus 62.4%); but not meeting the P < 0.05 threshold (6.8%, 95% CI: −0.19% to 13.8%, P = 0.069). For the primary prescribing outcome, intervention practices possibly reduced broad-spectrum prescribing to a greater extent than controls (1.42% versus 1.12%); but again not meeting the P < 0.05 threshold (coefficient −0.31%, CI: −0.7% to 0.1%, P = 0.104). The behavioural impact group reduced broad-spectrum prescribing to a greater extent than plain feedback practices (1.63% versus 1.20%; coefficient 0.41%, CI: 0.007% to 0.8%, P = 0.046). No harms were detected. Conclusions: Unsolicited feedback increased practices' engagement with data, with possible slightly reduced antibiotic prescribing ( P = 0.104). Behavioural science techniques gave greater prescribing effects. The modest effects on prescribing may reflect saturation from similar initiatives on antibiotic prescribing. Clinical Trial Registration: ISRCTN86418238.