Improving Morbidity During Post-Acute Care Transitions for Sepsis

BACKGROUND

Sepsis is a common and life-threatening condition defined by organ dysfunction due to a
dysregulated response to infection (Fleischmann, 2016). Aggressive early sepsis
identification and treatment initiatives have decreased hospital mortality for patients with
sepsis (Rhodes, 2017). As mortality rates have improved, there has been a growing recognition
of the downstream effects of sepsis for the approximately 14 million annual sepsis survivors
who encounter increased long-term mortality and morbidity across functional, cognitive, and
psychological domains (Iwashyna, 2010; Shah, 2013; Schuler, 2018; Borges, 2015; Annane, 2015;
Prescott, 2015). Currently, there is a disconnect between the post-acute care needs of sepsis
survivors and the resources available to these patients (Huang, 2016; Ortego, 2015; Winters,
2010; Nesseler, 2013). Inadequate post-sepsis care strategies are reflected by adverse
outcomes post-sepsis including increased mortality risk and strikingly high rates of
healthcare utilization, including a 90-day hospital readmission rate of 40% and over 3
billion dollars in preventable costs (Prescott, 2016; Goodwin, 2015; Prescott, 2014; Jones,
2015). To address the specific gaps in treatment for sepsis survivors, international experts
developed best-practice recommendations to guide delivery of post-sepsis care (Prescott,
2018). These best-practice recommendations are directed towards the specific challenges and
sequelae following a sepsis hospitalization and include: i) identification and treatment of
new physical, mental, and cognitive deficits; ii) review and adjustment of medications; iii)
surveillance of treatable conditions that commonly lead to poor outcomes; and iv) focus on
palliative care when appropriate. However, implementation of recommended post-sepsis
practices is hindered by a gap in understanding how to best integrate interventions into the
complex and fragmented post discharge setting (e.g., lack of provider time and patient
engagement, limited access to care management, and insufficient institutional support)
(Taylor, 2019; Brownson, 2012; Bodenheimer, 2008; Coleman, 2004).

RATIONALE

In randomized controlled trials (RCTs), including our previous work, successfully implemented
care transition programs using nurse navigators have been shown to reduce hospital
readmissions and costs. To better enhance transitions of care for the highest risk, complex
patients with suspected sepsis, we propose extending this evidence using a nurse-facilitated
care transition program for patients in the post-sepsis transition period to improve the
implementation of recommended care practices and bridge care gaps. This approach, called the
Sepsis Transition and Recovery (STAR) program, is the next step in the progression of our
group's work on improving discharge transitions and sepsis processes of care. A key aspect of
this initiative includes the ability to identify sepsis survivors at the greatest risk for
poor outcomes. For example, one-quarter of sepsis survivors account for three-quarters of
hospital readmissions and costs, indicating that identifying high-risk sepsis patients for
targeted facilitation of best-practice care could efficiently impact quality and cost.

Our STAR program uses near real-time risk modeling to identify high-risk patients and a
centrally located nurse, virtually connected to participating hospitals, to coordinate the
application of evidence-based recommendations for post-sepsis care, overcome barriers to
recommended care, and bridge gaps in service that can serve as points of failure for complex
patients. During their hospitalization, high-risk patients enter into a transition pathway
integrated within Atrium Health Hospital Medicine's Transition Services program and includes
the following core components: i) Introduction to STAR process prior to discharge (confirm
provider consults e.g., PT, ID, palliative); ii) Disease-specific education and discharge
"playbook"; iii) Virtual hospital follow-up evaluation within 48 hours including medication
reconciliation; iv) Second, post-acute virtual follow-up within 72 hours (symptom monitoring,
confirm provider follow-up); v) Weekly contact with care management team; vi) Referral to
provider follow-up (e.g., primary care provider, transition clinic) as appropriate; vii)
Coordinated transition to the next appropriate care location after 30 days from time of
discharge. The STAR navigator also meets weekly with the Medical Director of the Atrium
Health Transition Services program who provides additional clinical oversight of ongoing
cases.

The IMPACTS (Improving Morbidity during Post Acute Care Transitions for Sepsis) evaluation
will examine if implementation of the STAR program within a large healthcare system will
improve outcomes for high-risk sepsis patients. This randomized program evaluation is
designed to be a seamless part of routine care in a real-world setting to generate knowledge
of best practices for implementation and dissemination of post-sepsis transitions of care.

INVESTIGATIONAL PLAN

Overall Study Design

This real-world pragmatic randomized program evaluation will compare the effectiveness of the
Sepsis Transition And Recovery (STAR) program versus usual care on post-sepsis care and
patient outcomes. The STAR program is informed by existing evidence and designed using the
Chronic Care Model to increase best-practice adherence and care coordination, resulting in
improved transitions between hospitals and post-acute care during sepsis recovery.

Because of resource limitations, STAR can only be made available to a limited number of
patients. To be objective in patient selection and allow for program evaluation, we will use
a data driven approach to identify patients as eligible for program referral. First, risk
modeling will identify patients as high risk for 30-day readmission or 30-day mortality
during the first day of the hospital admission. Then from this pool of high-risk patients, up
to 6 patients will be randomly selected each weekday to be referred to either receive usual
care or care delivered through the STAR program. The number of daily patients to be
randomized was selected to match targeted capacity for the STAR navigator and will be
reevaluated on a biweekly basis. Because variables that affect eligibility may change during
a hospital stay, initial eligibility will be re-confirmed at time of hospital discharge.
Specifically, patients who have had infection diagnosis ruled-out during their
hospitalization (i.e., rule-out documented in medical record) will be excluded for the
purposes of analysis. All remaining eligible patients at the time of discharge will be
included in analyses, which will be conducted using an intent-to-treat approach. Planned
enrollment is 708 patients (n=354 patients per study arm) and STAR program follow-up will be
completed 30 days after hospital discharge. Outcomes data will be tracked for 90 days and
captured from routinely collected data from the Atrium Health Enterprise Data Warehouse.
Given this evaluation protocol is part of a quality improvement intervention that relies on
using evidence-based interventions, only utilizes data collected as part of routine care, and
is minimal risk to patients, we requested that the institutional review board designate this
study as expedited research and grant a waiver of informed consent.

Primary Outcome Variable The primary outcome is a composite, dichotomous endpoint of
all-cause mortality or unplanned hospital readmission assessed 30 days post index hospital
discharge.

Secondary Outcome Variable(s)

1. all-cause 30- and 90-day mortality;

2. all-cause 30- and 90-day hospital readmission;

3. 30- and 90-day cause-specific hospital readmissions with primary diagnoses related to:
a) infection, b) chronic lung disease, c) heart failure, and d) acute kidney injury;

4. 30- and 90-day emergency department visits;

5. 30- and 90-day acute-care costs;

6. 30- and 90-day total healthcare costs (only in the subgroup of patients enrolled in a
Medicare Shared Savings Plan);

7. acute care-free days alive measured at 30 and 90 days, defined as the sum of days alive
without inpatient, observation, and emergency department encounters (rounded to full day
for any day with acute care utilization) during the interval after discharge.

Subject Selection Subject selection will occur via an automated query process for patient
list generation. Each weekday morning actively admitted patients at 3 study hospitals (i.e.,
Carolinas Medical Center, Carolinas Medical Center - Mercy, and Atrium Health Northeast) will
be identified from the electronic health record and Enterprise Data Warehouse and output into
daily eligibility lists based on the study's inclusion/exclusion criteria.

Randomization Eligible patients will be randomly allocated 1:1 to receive usual care or the
STAR program. Clinicians and patients are not blinded to treatment allocation. Due to
resource limitations, STAR will only be made available to a limited number of patients. Thus,
a constrained randomization scheme will be utilized to randomly allocate up to six eligible
patients to either STAR or usual care each weekday. The number of patients allocated to be
randomized daily was selected to match targeted capacity for the STAR navigators and will be
reevaluated on a biweekly basis and adjusted as needed.

Inclusion Criteria:

1. Admitted from the emergency department to inpatient or observation status at one of:
Carolinas Medical Center, Carolinas Medical Center - Mercy, or Atrium Health
Northeast;

2. ≥18 years of age upon admission;

3. oral/parenteral antibiotic or bacterial culture order within 24 hours of emergency
department presentation and

1. culture drawn first, antibiotics ordered within 48 hours or

2. antibiotics ordered first, culture ordered within 48 hours (adapted from criteria
applied in development of the Third International Consensus Definitions for
Sepsis and Septic Shock)

4. deemed as high-risk for 30-day readmission (i.e., ≥ 20%) or 30-day mortality (i.e., ≥
10%) using risk-scoring models

5. not discharged at the time of patient list generation

Exclusion Criteria:

1. prior randomization to either STAR or usual care study arms;

2. not a North Carolina resident or residence >2.5-hour drive time from treating
hospital;

3. the only antibiotic associated with patient is administered in the operating room as
this likely represents pre-operative infection prophylaxis and not presumed infection;

4. patients transferred from other acute care hospitals;

5. patients with a change in code status (i.e., do not resuscitate, do not intubate)
within 24 hours after admission due to the general assumption of increased risk of
exposure to less aggressive treatment;

6. patients with infection ruled out during the index hospitalization.