| Sumario: | Objectives: To characterise consumer referral pathways across publicly funded community mental health services, to identify delays and inefficient transition points and to determine specific event sequences preceding emergency department (ED) and inpatient admissions. Design: Retrospective analysis of linked administrative health records using process and sequence mining. Setting: Publicly funded community‐based mental health services (records linked across community, ED and inpatient settings); adults with at least one service contact in Western Australia between 2005 and 2022. Main Outcome Measures: Frequency and distribution of service events within referral pathways, pathway transition counts and median time intervals between service events and identification of sequential rules leading to ED or inpatient admission. Results: Altogether, 48,230 consumer referral pathways were constructed from 820,502 service event records (94% community, 3% ED and 3% inpatient). The most frequent initial transition from referral was to triage (37,953; 79%). Process mining revealed delays in subsequent transitions, with a median time (IQR) of 4 (0.9–39.7) days from triage‐to‐triage loops (21,831) and 3 (0.1–31.3) days from triage to ED presentation (9705). Sequence mining identified 15 rules; notably, one rule indicated that 59% of the 5228 pathways with three consecutive triage events were followed by an ED attendance. The pathways demonstrated high variability, with access to active care often following unpredictable service patterns. Conclusions: This study identified delays, inefficiencies and areas where integration can be improved. It also highlighted the importance of examining the consumers' pathways to and through care, which often begin well before they successfully engage with active care (treatment contacts following formal activation) in community mental health services. The heterogeneity of pathways implies that, for a substantial share of referrals, access to stable care is governed more by stochastic factors than by deliberate signposting or intentional recruitment.
|
|---|