Improving Patient Flow and Access to Care at an Academically Affiliated Urban Safety-Net Hospital

Both the adult and adolescent units had separate teams addressing the challenge of reducing LOS and improving patient flow. The team caring for adolescents consisted of all the physicians practicing on the unit, the medical director, the nurse manager, the psychologist, the occupational therapist, and the head of adolescent social work. The team caring for adults consisted of the medical director, the nurse manager, the lead psychologist, and the head of adult social work. Teams were chosen by the medical director to represent all disciplines present on each unit. Because many more physicians were present on the adult unit than on the adolescent unit, it was not possible to have all of them present for large group discussions. The medical director therefore met with the physicians on the adult unit separately. A review of how patient flow affects access to care was reviewed with each team.

Members of each discipline discussed barriers to patient flow, and their representative reported to the large multidisciplinary team on each unit. The medical director of both units facilitated the discussions. Identified barriers included delays in discharge planning, delays in obtaining court-ordered medications (only on the adult unit), a staffing model that precluded weekend discharges, and social disparities prevalent among patients with severe and persistent mental illness. Physicians showed considerable responsibility in assisting patients and frustration in trying to address problems that could not be resolved by hospitalization, such as insufficient housing and poverty. A wide variation in LOS of patients also occurred among individual physicians. Discussions with individual physicians on both ends of this high variability highlighted significant philosophical differences in the goals of hospitalization and an absence of knowledge regarding how their individual practice compared with community norms.

The variability in perspectives highlighted a larger problem endemic to psychiatric hospitalization, namely, the lack of standardized guidelines. The medical director met with hospital administration to review identified barriers to timely admission and discharge. Baseline data were then reviewed with both the adult and adolescent care teams. The barriers discussed were addressed by having each team outline an individual workflow targeting more proactive discharge planning, increasing weekend discharges, addressing delays in obtaining court-ordered medication, and developing consensus guidelines on the end point of a hospitalization.

Consistent with national guidelines, patient flow was measured by actual LOS and the LOS index (LOSI) (4). The LOSI is a ratio reflecting an individual’s LOS relative to a reference sample (i.e., individual LOS:expected LOS), with LOSI values <1 indicating that the individual’s LOS is lower than that of the reference sample and values >1 indicating that the individual’s LOS is greater than that of the reference. We categorized our patients on the basis of All Patients Refined–Diagnostic Refined Groups (APR-DRGs) assessments and compared their APR-DRG metrics with those of national averages. APR-DRGs can be used to group patients on the basis of diagnosis, risk for death, and severity of illness (5). Using APR-DRGs allows one to control for medical complexity when comparing patient groups. We provided physicians with feedback on their individual performance relative to national norms. The LOS in the adult unit in the 2016–2018 period ranged from 13.6 to 16.8 days (LOSI 1.38–1.63), and the adolescent unit LOS in the same period ranged from 7.5 to 9.4 days (LOSI 1.23–1.53). The average LOS for individual physicians ranged from roughly 7 days to >15 days.

On the adult unit, social workers changed their schedules to allow for weekend coverage in order to begin discharge planning with patients earlier in their hospitalization. On the adolescent unit, discharge planning historically did not begin until several days into hospitalization at the first family meeting. This workflow was changed such that discussions of follow-up care began on the initial phone contact with the family on hospital day 1. Family meetings were scheduled earlier in hospitalization to expedite therapeutic work with the family. Unit programming was altered to increase the number of multifamily group therapy sessions fourfold and to boost the number of group sessions for patients twofold. The expanded programming and added family work allowed for more intensive treatment, facilitating earlier comfort with transitioning out of the hospital.

Efforts to increase weekend discharges specifically targeted the adult unit. Historically, our weekend on-call attending physician covered both units as well as the consultation liaison and addiction medicine services. Staffing models were changed so that the on-call attending physician covered only the adult and adolescent units, allowing time for weekend discharges and follow-ups for medication titration. Staffing constraints did not allow for the consistent presence of a psychiatrist for adolescents and a social worker on weekends. Therefore, increasing weekend discharges on the adolescent unit posed quality and safety concerns and was not attempted.

This intervention only applied to the adult unit because adolescents very rarely require court-ordered medication. Historically, our physicians sought to obtain court-ordered medication in cases in which patients were refusing medication in the hospital, or when a patient had a history of noncompliance in the community. In Colorado, an outpatient provider working at a licensed facility must accept the court order for it to be enforced after a patient leaves the hospital. Obtaining a court order often takes 7–10 days. Although this step is necessary in returning someone to baseline functioning who is noncompliant with medication while an inpatient, it may not be necessary for patients who are compliant and do not have an outpatient provider who will accept the court order. For a variety of reasons, few community mental health centers will do this step, and prolonging hospitalizations to obtain court orders that will be dropped at discharge is a futile endeavor. Proactive identification of where the patient would receive follow-up care allowed us to avoid this outcome.

The social workers and physicians on both units noted that at times patients were kept in the hospital for social rather than medical reasons. Specific examples of this practice included attempting to find patients housing and giving them respite from difficult living situations. Hospital resources rarely address these types of problems. Understanding our role in the community and responsibility to prospective and current patients in regard to resource allocation allowed for a more objective lens through which to evaluate these challenges. When focusing on medical criteria for hospitalization, physicians could agree that the end point of a hospitalization for crisis stabilization should be when the patient is approaching normal functioning and has adequate outpatient follow-up. These philosophical discussions were accompanied by objective data regarding national norms. Individual physicians were given their LOS and LOSI data relative to national norms based on APR-DRGs.

The adolescent unit had 118 more admissions in 2019 than in 2018. The average daily census (ADC) decreased from 14.9 in 2018 to 12.9 in 2019. LOS for the adolescent unit decreased from 7.9 to 6.1 days, and the LOSI declined from 1.26 to 0.79. Variance in LOS among physicians was within 1.5 days.

The adult unit had 409 more admissions in 2019 than in 2018. The ADC increased from 37 in 2018 to 39.2 in 2019, which did not fully explain the increase in admissions. For reference, in 2017, the ADC was 38.5, although 472 fewer admissions occurred in 2017 than in 2019. On the adult unit, the LOS decreased from 14.5 to 7.5 days with a corresponding LOSI decline from 1.6 to 0.9. Weekend discharges increased by a factor of 10 on the adult unit. Variance in LOS among physicians was within 1.5 days.

Seven- and 30-day readmission rates to our facility did not change relative to historical cohorts. We could not examine readmissions to other facilities. Although physicians indicated that their pace of work was faster and that weekends were much busier, burnout rates as measured by the Mini Z survey remained <15% for 10 months postimplementation. This rate was not significantly different from preintervention rates and was lower than national averages (6). Patient satisfaction scores were compared 1-year pre- and postintervention, and raw scores for both units remained very similar. Scores specifically related to the discharge process remained the same or showed slight improvements. No suicides occurred within 30 days of discharge.

The changes outlined above yielded several unanticipated consequences. Admissions and discharges took more time for both physicians and nursing staff. This burden affected the adult unit much more than the adolescent unit because of their different sizes. Staffing was changed to allow for an extra charge nurse to assist on the adult unit. The way in which admissions were distributed to physicians was also changed to allow for a more even distribution of admissions among physicians. Because of the schedule changes to allow increased weekend coverage, social workers on both units experienced more frequent patient handoffs. This increase in handoffs resulted in less efficiency because of the constant need for social workers to familiarize themselves with new cases.

The components of our initiative significantly diverged from those in previous procedures to improve care and reduce LOS. Historically, medical needs were frequently enmeshed with nonmedical needs arising from social disparities, weekend workflows were provider centered as opposed to patient centered, goals of hospitalization were not standardized, and resource allocation and community needs, taking into account the numbers and needs of current and prospective patients, were not openly discussed. The considerable positive change in these diverse areas we achieved within 3 months of implementation of the initiative has been sustained. The ease with which our team adapted to the culture shift came as a surprise to the administration. Conversations with physicians revealed an appreciation for objective metrics of performance and a better understanding of how efficient patient flow reinforces the ethical pillars of autonomy and justice.

We conclude that a structured dialogue with frank discussions about expectations and shared values can foster rapid culture change at a community hospital. Ongoing attention and evaluation of intended and unintended consequences of process improvement are fundamental to successful implementation of improvement initiatives.