Proactive Detection (PROTECT) and Safety Planning to Shorten Emergency Department Stays for Psychiatric Patients

Produce creative effective simple solutions (PROCESS), a leadership framework that supports complex change within mental health care services, was initiated at PAH. The 3×3 overlapping phases of the model comprise envision (enquire, explore, empathy), enable (empower, exchange, evolve), and enact (execute, evaluate, embed) (Figure 1; see also an online supplement to this article). PROCESS was coproduced in the highly successful proactive management of integrated services and environments (PROMISE) initiative to reduce coercion in care (17).

We adopted the five-step (report daily, reflect weekly, review monthly, rethink quarterly, refresh annually) proactive governance of recovery settings and services (PROGRESS) governance framework (see the online supplement), which has previously been used in PROMISE to enhance patient and staff experiences in mental health care (17) and to roll out a restorative culture in line with the Zero Suicide approach (18).

The innovative safety-planning intervention from PROTECT (13) to mitigate risk for suicide and self-harm in the first hour, 2 days, and 7 days after presentation (1-2-7) was rolled out (see the online supplement). Ten conversational prompts were used to sequentially explore and mitigate suicide and self-harm risk (19).

The foundation of the intervention at PAH was laid in the last quarter of 2019. The implementation plan was finalized in November 2019 after the annual Refresh event (see the online supplement), and the 1-2-7 safety-planning initiative was rolled out at PAH in December 2019. Care as usual continued at the NIS. The ED mental health (EDMH) departments of both hospitals had a similar size, had a similar staffing profile, operated 24 hours per day and 7 days per week, and served local metropolitan catchment populations (with similar demographic characteristics and diversity within the same city). Both PAH and the NIS were within the same hospital and health service and thus were similarly affected by systemwide developments (detailed in the Strengths and Limitations section below). Compared with the rest of the world, the impact of the COVID-19 pandemic was minimal in Queensland, Australia, in 2020, and its impact was similar in both sites.

Anonymized, 11-month preintervention and 13-month postimplementation data were extracted from the journey board, an electronic system that tracks patients as they transition through different teams and settings within the care organization. These data included the total number of mental health patients attending the ED, monthly ALOS, number of patients whose ALOS exceeded 12 hours and 24 hours, and the proportion of monthly conversion of mental health presentations to inpatient psychiatric admissions and ACT referrals for both sites. This study was approved by the Metro South Human Research Ethics Committee.

We conducted t test and regression analyses to estimate the impact of the safety-planning intervention on selected outcomes. Specifically, t tests were used to estimate changes in outcomes postimplementation (December 2019–December 2020) compared with preintervention (January 2019–November 2019). We also tested the difference in outcomes between the intervention site and NIS. A generalized linear model was used to estimate the impacts of the intervention with a difference-in-differences approach. A Poisson link distribution was used for count outcomes (e.g., number of patients), and Gaussian distribution was used to analyze continuous outcomes (e.g., the ALOS). Robust standard errors were used to mitigate the effects of heteroscedasticity (in which the variance of the residuals may not be constant).

The intervention impact was represented as the parameter of the interaction between the intervention site as a binary variable (PAH=1, NIS=0) and intervention time as a binary variable (postimplementation=1, preintervention=0). This parameter was the change in outcome at the intervention site (compared with NIS) after intervention rollout (compared with the preintervention period). A p≤0.05 was used to determine the statistical significance of estimates. Each presentation to the ED for patients who presented multiple times was considered an individual presentation. We analyzed the data with Stata, version 14.

Because of the multifactorial nature of ALOS and destinations for onward referrals, our results must be interpreted with caution. The intervention was complex, and although safety planning was the primary practice change, the leadership approach and supporting governance processes also underwent significant alterations. A difference-in-differences approach is a robust analytical tool, but any cause-and-effect relationship should be considered speculative. Although safer care is the primary objective of a clinical culture that promotes collaborative safety planning, we hypothesized that it would also reduce ALOS for patients attending the ED with mental health challenges. Those scheduled to be discharged back home could leave sooner, because safety planning supported the earlier formulation of an accurate opinion regarding unaddressed risks in the community. This observation was captured in the significant decrease in the proportion of patients with an ALOS >24 hours: 24% preintervention to 6% postimplementation. Not having to provide monitoring for those waiting in the ED allowed staff more time to evaluate and treat the significantly higher number of patients in 2020: a monthly mean of 563 patients compared with 517 in 2019. In the monthly ED PROGRESS review and Mortality and Morbidity meetings, staff also reported feeling more confident in discharging people at an earlier stage. This rising staff confidence was evidenced by a month-by-month steady decline in ED ALOS from 18.5 hours in November 2019 to 5.5 hours in December 2020.

Rigorous safety planning also decreased the conversion rate of ED psychiatric presentations to psychiatric admissions (23% preintervention vs. 18% postimplementation). Inpatient staff reported at the monthly ward PROGRESS reviews how the revamped leadership (i.e., appointment of a new consultant [T.T.], full-time team leader [T.M.], clinical nurse consultant, resident medical officer, and transition coordinator) brought clarity of thought to clinical decision making through an enhanced admissions checklist. On the checklist, the purpose of psychiatric admission for each patient was clearly defined, and, whenever possible, community alternatives were identified. Many patients, who in 2019 would have been admitted to inpatient care, received follow-up care at home with ACT support, the least restrictive setting possible for a psychiatric patient’s recovery. The resulting bed availability allowed the leadership to address a preexisting narrative of keeping people out of the hospital. This narrative was changed to a timely and proportionate response to bring people into the hospital at an earlier stage of deterioration so that their stay in the most restrictive environment would be as short as possible and unnecessary long stays could be avoided.

The safety measures and community flow–related changes are not reported in this article. In brief, they centered on proactive identification of patients whose condition had deteriorated in the community. The transition coordinator ensured that such patients got appropriate help and, if necessary, were admitted to an inpatient stay at an earlier stage, reducing patient suffering by offering an earlier intervention. Previously, by the time patients would come into the hospital, they would be so unwell that they would need longer lengths of stay, affecting bed availability. The backflow resulting from the previous approach would affect ED ALOS because patients awaiting inpatient admission would need to wait longer until a bed became available. The change in approach resulted in both shorter inpatient ALOS and increased bed availability.

The implementation of the 1-2-7 innovation was led by the consultant psychiatrist (T.T.) and team leader (T.M.) of the EDMH at PAH. This in-house ownership of the initiative and the support it provided by encouraging patients in positive risk-taking as part of safety planning was crucial to the initiative’s success. Because this initiative was new, the EDMH leadership also ran several safety protocols, including a daily audit of the appropriateness of all discharges, and provided valuable feedback and supervision. Although the principal focus of improvement actions was on the ED, the ALOS metrics cumulatively resulted from safety- and patient flow–related changes across the entire service. We believe that a singular focus on the ED would not have achieved the same results. A systemwide approach to patient flow that is underpinned in a care culture that strives to maintain the delicate balance of risk and recovery is needed (13, 20).

Our results provide conclusive evidence supporting the hypothesis that shifting care culture by using appropriate safety planning significantly reduces ED ALOS and increases diversions of patients to their communities. The therapeutic relationship serves as scaffolding (21, 22) within which some of the potential risks of this shift in culture can be anticipated and discussed in a supportive, solution-focused manner. Conceptual models like the Care Compass (5, 13, 19) were used to support staff in striking a balance between risk versus recovery and care versus control. The process of shared decision making and planning has required a shift in the mindset how staff work: from top to tap (23). Previously, staff would act from the top, feeling responsible for fixing problems; the new mindset of being on tap brought forth an enabling approach, with an emphasis on hope, which is an antidote against suicide and a vital ingredient for ongoing safety (24). Because of the mindset shift, the top 3 practice changes were prediction to prevention, past to future, and deficits to assets. These changes were supported by safety formulations using the DESPAIR (Diagnosis, Entrapment, Suicidal Thoughts and Plans, Past Attempts, Agitation, Risk Response) framework from PROTECT and clear and imminent risk formulations based on longitudinal risk (a person’s individual risk over time), unaddressed risk (unmitigated risk after safety planning), and cross-sectional risk (the person’s risk compared with risks for other patient populations such as inpatient, community, and primary care) (5, 13, 19) (see also the online supplement).

In this large study, we tracked the ALOS of 24,515 EDMH patient presentations across two major Australian metropolitan public-sector hospitals. We acknowledge that a double-blind randomized controlled trial is often recognized as a more rigorous tool to establish a relationship between cause and effect. This study design was not possible given the busy, noncontrollable hospital environment and the complex nature of the intervention; thus, a difference-in-differences approach was adopted to evaluate the intervention’s impact. Rather than compare outcomes before and after implementation of the intervention in one site, we opted for inclusion of comparator site data to mitigate the confounding effects of sitewide developments and challenges. These changes included a launch of a police and ambulance co-responder service, an investment in a flow crew to coordinate inpatient admissions, appointments of transition coordinators, weekend discharges, and variations in primary presenting problems because of the COVID-19 pandemic. Both PAH and the NIS were similarly affected by these changes. The advantage of mitigating confounders by choosing two study sites within the same organization was offset by some inevitable knowledge sharing between staff from the two sites that may have diluted the results of the difference-in-differences analysis. In addition, some differences at baseline, including PAH having higher ED ALOS and serving larger numbers of patients than the NIS, made it more challenging to roll out and embed safety planning in PAH. Also, some patients were unwilling to participate in safety planning, or it was not clinically indicated.

Because this study focused on patient flow metrics, patient and clinical variables were not analyzed. Daily intake reviews by a consultant psychiatrist were instituted for quality control along with monthly audits of 10 safety plans; however, conclusions about safer care cannot be drawn from this study’s results. This study was also limited by the fact that we could not differentiate the direct impact of PROCESS (leadership), PROGRESS (governance), and 1-2-7 (innovation) on ALOS. Similarly, the impact on ALOS of improvements in EDMH, which we report in this article, could not be differentiated from the impact of improvements in inpatient psychiatric or community care. We note that because this intervention is complex and spanned care systems, it was not possible to segregate the impacts of the improvements in each area on ED ALOS. Also, given the rare occurrence of suicide and its multifactorial nature, we did not report specific safety data in this study. Instead, to evaluate the impact of 1-2-7, we presented quantifiable clinical efficiency measures, such as ALOS and percentage conversion of mental health presentations to inpatient admissions. Reducing ALOS might have enhanced both patient and staff experiences; however, data for these metrics were not collected. We found that the intervention reduced ED ALOS by 7.7 hours, a reduction that is substantially larger than reductions reported for interventions in general EDs using lean management (25, 26). The reduction in ALOS and in the demand for beds could have been associated with proportional reductions in health care costs (27); however, costs of care were outside the scope of this study.