Design Principles for Incident Reporting Systems in General Practice: Recurring Recommendations, Rarely...

A small set of design principles-anonymity, simplicity, voluntary participation, timely feedback and a non-punitive culture-is recommended for incident reporting systems in primary care with a consistency that is itself notable. Whether those principles have been tested has attracted far less attention than whether they have been named.

To establish what evidence supports the design principles conventionally recommended for incident reporting systems in primary care, and to report, as a proof of concept, the deployment of an instrument built on those principles among Italian general practitioners.

A two-part study comprising an evidence synthesis followed by an observational pilot. Part 1 was a structured evidence map based on a systematic search of PubMed, Scopus and the Cochrane Library, with backward and forward citation searching of 14 seed publications (PROSPERO CRD420251080728), reported according to PRISMA 2020, with duplicate screening. Each included source was coded against each principle by level of evidence, from not addressed through presupposed, recommended, contested and empirically examined to directly tested, the last requiring the principle to be compared or manipulated and related to a measured reporting outcome. Part 2 was a prospective descriptive deployment of a voluntary web-based reporting instrument among eight general practitioners in Apulia, Southern Italy.

A total of 202 sources were included, spanning 1992-2026. Sixteen compare conditions, four of them randomised trials and four with a parallel comparator group. Three sources compare a design characteristic against an alternative and relate the comparison to a measured reporting outcome, and their findings diverge. Confidential reporting yielded more codable reports than anonymous reporting, an observational comparison in which reporters chose the mode. A dictation system produced no significant difference in the number of reports submitted compared with paper, and users preferred paper. A standardised form may slightly increase reporting, an estimate from a single low-certainty study drawn from pharmacovigilance rather than incident reporting. Voluntary participation, timely feedback and a non-punitive culture have not been compared against an alternative in any source. None of them isolates a single design characteristic: two randomised trials intervene on safety culture rather than on the design of the reporting system, and the remainder test multicomponent packages from which no single principle is separable. During the deployment, which ran from February to June 2025, 10 reports were submitted involving 8 practitioners; the number of distinct reporters could not be established. Five of ten concerned aggression towards the practitioner or practice staff and six involved communication failures.

Within the corpus identified by this search, the design principles conventionally recommended for incident reporting systems in general practice rest on professional consensus rather than demonstrated effect. The field has conducted controlled studies, but has directed them at safety culture or at multicomponent interventions, so that after three decades the comparative evidence on any individual characteristic is sparse, heterogeneous and outcome-dependent. A proof-of-concept system built on these principles generated reports concerning aggression, communication and organisational friction more often than clinical error, indicating that these practitioners perceive workforce conditions as safety-relevant.

PROSPERO (CRD420251080728).