Reducing Delay From Referral to Admission at a U.S. First-Episode Psychosis Service: A Quality Improvement Initiative

The STEP program has delivered a model of specialty team–based comprehensive care, that is, coordinated specialty care (CSC) (4), since 2006. In 2014, after completing a pragmatic randomized trial that established the effectiveness of its CSC service (5), STEP reorganized around a population health framework (6). Any individual age 16–35 years having a schizophrenia spectrum diagnosis within the past 3 years and residing in the 10-town catchment area was eligible to access the STEP clinic. STEP’s early intervention service was organized into two teams, overseen by the program director (V.H.S.). An early detection team, led by an outreach and assessment coordinator (P.M.), curated the referral-to-admission process and then handed over care to the CSC team (led successively by J.M.P. and S.I.). In 2015, STEP launched Mindmap to proactively recruit individuals with first-episode psychosis into care, with an overall aim to reduce DUP in the community (5). Mindmap included three interleaved components to comprehensively target sources of delay across regional pathways to care: public education (using social and mass media), outreach and detailing of referral sources (by dedicated staff), and RAS (by the clinical service).

Referrals were accepted from both affected individuals and their families and from a wide range of stakeholders both inside (e.g., hospitals, primary care, and outpatient behavioral health clinics) and outside (e.g., college staff and community nonprofits) the health care sector. A brief structured questionnaire was used to establish a person’s eligibility for the program.

We followed a two-phase model for QI (7). (Details of the model are available in an online supplement to this column.) A weekly meeting hosted by the program director that included members of the early detection team and CSC leadership allowed coordination of these efforts and served as the locus of QI activities for RAS. In the first QI phase, we set an aim (i.e., reduce delay from referral to admission), established a measure (i.e., delay to admission [DTA]), and identified a repertoire of change tactics.

The time from referral to STEP admission (i.e., the DTA) was chosen as the target measure for this QI initiative and defined as the interval (in days) between confirmation of a person’s eligibility for treatment and consent to enter services for first-episode psychosis. We did not use the date of first inquiry, because many calls to our clinic were for unrelated diagnoses or services (8), and eligibility determination was typically completed within 24 hours, leaving little room for further improvement. The target length of time of DTA was set at ≤7 days. This target reflected a pragmatic tradeoff between the wish to engage patients as rapidly as possible and pragmatic realities of and constraints on workflows. Previous experience (5) with inpatient referrals indicated that although eligibility and procedures for enrollment in STEP could be completed soon after admission to a hospital unit, a few intervening days were often necessary for inpatient teams to sufficiently stabilize patients’ symptoms and enable meaningful voluntary participation of the patients in aftercare planning. For outpatient referrals, some time had to be allowed for arranging travel of a patient to the clinic when eligibility could not be determined remotely.

Historical referral data (5) informed anticipatory tactics for key causes of delay. The team reviewed processes for receiving referrals, eligibility determination, baseline assessment, and consent for care. This review elicited multiple suggestions for improvement that were linked to several hypothesized causes of DTA. These causes were assigned categories (which were not mutually exclusive) to organize the selection of improvement efforts in preparation for the second phase and included the following causes of delay. Patient-related causes included, for example, lack of insight, severe disorganization or cognitive impairment, medication-related sedation, and legal involvement. Family-related causes consisted of, for instance, ambivalence about the value of psychiatric treatment, reluctance to accept the diagnosis, or pressure from the patient to decline care. Referrer-related causes included limited awareness of STEP or how to optimally connect with STEP from a hospital-based or community setting. Finally, service-related causes of delay included, for example, procedures to determine eligibility and unavailability of staff to obtain consent from a reluctant patient.

The second phase involved recurrent plan-do-study-act (PDSA) cycles (a figure detailing this process is available in the online supplement) that implemented shifting combinations of the tactics envisioned above but also newly generated strategies in response to expected and emerging sources of DTA. Consistent with QI methodology (9), DTA was reviewed at the weekly meeting, and these joint reviews were used to evaluate progress toward the goal of DTA ≤7 days. For example, detailed discussions of outliers (unusually long or short DTAs) helped reveal novel or inadequately addressed sources of delay or successes that could be further leveraged in subsequent PDSA cycles. Process changes that emerged as stably successful and feasible to sustain became incorporated into the standard operating procedures of the early intervention team.

One example of a successful PDSA cycle (available in the online supplement) targeted long-standing delays in handoffs between inpatient psychiatric units and STEP. When referrals were made close to the day of planned discharge, it was often difficult to confirm eligibility for the program, and patients were sometimes significantly delayed or even lost to follow-up at STEP. With a focus on reducing delays to STEP from specific inpatient units (plan), the tactic implemented was for the outreach coordinator to periodically visit the units that made the most referrals (do). After several weeks of implementation, this approach reduced DTAs in several cases by both incentivizing early referral calls from inpatient staff and reducing no-shows at the first outpatient follow-up visit at STEP (study). Weekly visits by the outreach coordinator to these and other acute units subsequently became part of the standard operating procedures for the service (act). Multiple analogous PDSA cycles were implemented during the 4-year study period, and a few more examples are summarized below.

We discovered that most referrals with ambiguous information about a person’s eligibility for STEP originated in outpatient services. The sources of ambiguity included lack of knowledge within these services about STEP’s eligibility criteria or implicit requests for a second opinion on a challenging case. This delay in determination of eligibility created a backlog of partially processed and time-consuming referrals for the outreach coordinator. To accelerate the eligibility determination (plan), separate procedures (do) were implemented for referrals from outpatient providers, wherein eligibility was rapidly clarified by telephone, and potentially eligible persons were invited for an in-person assessment if the referrer could not provide this information. Persons deemed eligible were admitted to STEP on the same day. In contrast, for referrals from inpatient services, for which STEP eligibility could usually be determined through telephone communication with clinical staff, the first visit (often on the unit) was used for completing STEP admission. This change did reduce the amount of staff time spent on outpatient referrals and reduced DTAs for patients determined to be eligible, without increasing the number of those lost to recruitment (study). This split procedure was thus retained (act).

In a few outlier cases, the staff responsible for early detection of psychosis speculated that a longer-than-usual DTA might have been avoided with more flexible offers to meet patients outside the building housing the CSC service. Some patients had revealed that they found it intimidating at first to enter our facility (which has a large sign identifying it as a “mental health center” and security guards in the lobby), although this concern quickly abated over subsequent visits. To reduce patients’ reluctance to visit services (plan), several tactics were deployed (do), including repeated telephone conversations to explicitly elicit and address specific fears (e.g., about being involuntarily admitted to a psychiatric unit at the visit) and offers to meet at a nearby café, where the outreach coordinator could more fully assess a patient’s concerns and offer to accompany the patient into the building at a future appointment. Although the use of such strategies was rare and was used mostly for referrals of outpatients (whom the outreach staff had not met on a hospital or emergency department unit), it elicited valuable information about patients’ concerns that besides reducing DTA also informed CSC clinicians’ subsequent efforts to retain the patient in care (study). Hence, the practice was continued on an ad hoc basis, for example, in cases of repeated no-shows for eligibility determination or CSC appointments (act).

Several individuals were found not to respond to voice mail, or they had provided a number with a potentially nonprivate voice mail. To reduce this delay in connecting with a potentially eligible patient while preserving their privacy (plan), the tactic of querying new patients for several cellphone numbers enabled staff to send texts to these phone numbers to confirm appointment times (do). Concerns about breaching patient privacy were mitigated by using standard language that did not identify the clinic but just provided the appointment date and time. This approach dramatically reduced DTA due to missed intake appointments (study) and became more routine practice for scheduling CSC visits (act).

RAS was implemented over a 4-year period and enabled staff to apply formal QI methods to refine intake procedures at STEP’s CSC service and to exceed the benchmark for reducing DTA. The baseline median DTA of 13.5 days in the year before the QI campaign significantly and sustainably fell to 3 days during the RAS intervention (Mann-Whitney test, U=2,744.5, Z=4.385, p<0.001). (A run chart is available in the online supplement.) Also, only eight (33%) of 24 admissions during the year preceding the RAS intervention initially met the 7-day standard, significantly rising to 104 (71%) of 147 admissions over the 4 years of the RAS intervention (p<0.001, chi-square test). Ongoing review of delays refractory to our QI approach and where our standard was not met revealed themes that will serve as targets for future cycles of improvement. In each case of delay, close liaison with the referral source and patient was attempted to facilitate eventual engagement in care.

This column describes a replicable and sustainable strategy for CSC services to reduce DUP. Specifically, RAS targeted causes of delay that were proximal to clinical workflows and did not require the additional resources necessary to influence more distal or community-based segments of the pathway to care. Also, the QI methodology is accessible to those without formal research experience and is available via an online curriculum (7) designed for workplace-based learning.

We were unable to find other studies that used formal QI to reduce DTA at CSC services, but findings from a similar study confirmed that delay from referral to admission significantly contributes to overall DUP (10). Our demonstration of a way to reduce DTA offers a tool for such CSC services to target an important segment of this delay.

Several limitations of this study can help contextualize these results. RAS was one of three components of a comprehensive campaign for early detection and treatment of psychosis. We therefore cannot exclude the possible confounding effects of public education, professional outreach, and detailing efforts that may have been delivered to patients and care providers who were more likely to rapidly engage with our CSC service. Also, the methodology of QI involving the use of combinations of interventions across several PDSA cycles reduced our confidence in attributing outcomes to specific strategies. However, the comparison with the historical STEP data and the fact that no other cointerventions, besides those related to the QI campaign, were implemented reduce the likelihood that unknown confounders influenced the improvements obtained with this QI process.

In this study, we tested and successfully implemented a QI process to reduce unnecessary delay in providing specific treatment for patients with first-episode psychosis by using a sustainable approach for reducing a segment of DUP. Our findings indicate that this approach can be integrated into regular clinical workflows.