Objective
Using augmented intelligence clinical decision tools and a risk score-guided multidisciplinary team-based care process (MTCP), this study evaluated the MTCP for heart failure (HF) patients’ 30-day readmission and 30-day mortality across 20 Intermountain Healthcare hospitals.
Background
HF inpatient care and 30-day post-discharge management require quality improvement to impact patient health, optimize utilization, and avoid readmissions.
Methods
HF inpatients (N = 6182) were studied from January 2013 to November 2016. In February 2014, patients began receiving care via the MTCP based on a phased implementation in which the 8 largest Intermountain hospitals (accounting for 89.8% of HF inpatients) were crossed over sequentially in a stepped manner from control to MTCP over 2.5 years. After implementation, patient risk scores were calculated within 24 hours of admission and delivered electronically to clinicians. High-risk patients received MTCP care (n = 1221), while lower-risk patients received standard HF care (n = 1220). Controls had their readmission and mortality scores calculated retrospectively (high risk: n = 1791; lower risk: n = 1950).
Results
High-risk MTCP recipients had 21% lower 30-day readmission compared to high-risk controls (adjusted P = .013, HR = 0.79, CI = 0.66, 0.95) and 52% lower 30-day mortality (adjusted P < .001, HR = 0.48, CI = 0.33, 0.69). Lower-risk patients did not experience increased readmission (adjusted HR = 0.88, P = .19) or mortality (adjusted HR = 0.88, P = .61). Some utilization was higher, such as prescription of home health, for MTCP recipients, with no changes in length of stay or overall costs.
Conclusions
A risk score-guided MTCP was associated with lower 30-day readmission and 30-day mortality in high-risk HF inpatients. Further evaluation of this clinical management approach is required.
The high frequency of healthcare encounters for patients with heart failure (HF) is due in part to the increase in HF incidence and prevalence and is also related to the complexity of HF clinical management. Further, penalties for excessive readmissions persist through the Hospital Readmissions Reduction Program (HRRP) in the United States, which has been driving resources toward reducing HF patients’ 30-day readmissions. Unfortunately, evidence suggests that success in readmission reduction may have resulted at the expense of worse patient health (ie, higher mortality rates). While likely not the intended outcome of a program based on financial penalties, the patient-centered focus of medical culture requires improvement in both readmission and mortality outcomes for the well-being of patients and the sustainability of the healthcare system.
While reducing financial costs and improving outcomes are both major quality objectives today, precision medicine using practical and existing information about patients also provides the opportunity to reduce the temporal costs to clinicians. Previously, feasible clinical decision tools were created to rapidly augment clinician knowledge by providing early and valuable patient health information without requiring clinician involvement in the risk stratification process. This directed targeting of care facilitated personalization by combining risk scores with a clinician-derived risk-guided care process. The purpose of this study was to evaluate the impact of a healthcare system-wide readmission and mortality risk reduction program among HF patients admitted for inpatient care.
Methods
Objectives and hypothesis
The objective of this study was to evaluate the phased implementation of a risk-stratified multidisciplinary team-based care process (MTCP) among HF inpatients across Intermountain Healthcare. The MTCP consists of precision medicine care pathways used to individualize care based on each patient’s clinical risk score results. The primary study aim was to evaluate the hypothesis that MTCP use is associated with a lower risk of major adverse clinical outcomes at 30 days after discharge from an index HF inpatient admission. Co-primary endpoints were 30-day readmission and 30-day mortality. Secondary aims included evaluating whether the MTCP is associated with in-hospital outcomes, healthcare utilization, costs, or clinical decisions. Finally, in the control group a prospective validation was performed of the inpatient HF (iHF) risk score as a predictor of 30-day readmission and the Intermountain Mortality Risk Score (IMRS) as a predictor of 30-day mortality.
Study population
During January 1, 2013, through November 30, 2016, HF inpatients were studied at 20 Intermountain Healthcare hospitals in a clinical quality improvement implementation of the MTCP. The research evaluation of this clinical project was performed based on the principles of the Declaration of Helsinki and approval was received from the Intermountain Healthcare Institutional Review Board for the evaluation of patient health information under a waiver of consent. No extramural funding was used to support this work. The authors are solely responsible for the design and conduct of this study, all study analyses, the drafting and editing of the manuscript, and its final contents. Two other Intermountain hospitals were not studied because they either do not provide HF care (ie, The Orthopedic Specialty Hospital) or do not admit adults (ie, Primary Children’s Hospital).
Patients were included (N = 6182) who were adults admitted for inpatient care and, based on a previously-described diagnostic algorithm, had a high probability of receiving the primary discharge diagnosis of HF. All HF patients in the 20 hospitals were considered controls if they were admitted between January 1, 2013 and February 2, 2014. To minimize confounding and selection biases, a staged cross-over of hospitals into the MTCP intervention was performed. On February 3, 2014, the MTCP and clinical decision tool risk scores were implemented as a pilot at McKay-Dee Hospital, a regional referral hospital north of Salt Lake City that accounts for 13.7% of HF inpatients at Intermountain. In February 2015 through August 2016, seven other hospitals that account for 76.1% of HF inpatients at Intermountain were crossed over in staged steps (as described in Supplemental Figure S1A) into MTCP utilization. The stepped approach exposed more hospitals (or clusters) to the intervention about every 3 months, although the timing of the steps was not equally spaced (see Supplemental Figure S1A). This phased approach helps to control issues of a before-after design because the intervention is implemented at different times in different hospitals and, thus, controls and intervention patients overlap temporally and confounding due to secular trends in care is ameliorated. Using a stepped design does not refer to whether randomization was used, and herein the hospital implementation order was based on clinical exigencies (ie, not cluster randomized).
Multidisciplinary team-based care process
The risk score-guided precision medicine MTCP was developed by HF cardiologists and advanced practice clinicians at Intermountain ( Figure 1 ). Team member duties are provided in Table I and were focused primarily on the inpatient period. HF patients receiving care in hospitals where the MTCP was implemented were identified and evaluated within 24 hours of admission via an electronic risk calculator embedded in the electronic health record, as previously described. Patient risk scores (ie, iHF 30-day readmission score and IMRS 30-day mortality score) were delivered to clinicians via that application and the clinical team utilized them to assess patient needs and assign a care pathway. The iHF and IMRS scores were previously developed and validated using the complete blood count and basic metabolic profile and, for iHF, B-type natriuretic peptide. The scores were calculated by a clinical decision support application within the electronic health record utilizing the most recent laboratory measurements that were available, which usually were performed within 24 hours of hospital admission as standard of care laboratory testing, but if unavailable used the last measured values up to 6 months prior to admission. These scores were based on a pragmatic parsimonious modeling approach in which not all possible factors that are independently predictive are necessary to reliably predict risk, as quantitatively demonstrated previously. Risk score categories of high, moderate, and low for 30-day readmission or mortality were derived in prior analyses. An electronic calculator can be accessed through the Intermountain website for calculating risk scores and determining the risk group.
| Team Member | Clinical Pathway Responsibilities for High-Risk Patients |
|---|---|
| Bedside Nurse | Use HF nursing care plan Start HF education early, use teach-back, involve caregivers, and reinforce concepts Utilize MAWDS HF diary and video, fluid tracker, and white board to highlight daily educational goals Identify barriers to learning: eg, cognitive and language barriers |
| Charge Nurse | Ensure daily risk list is posted Assist bedside nurses with high risk activities, as needed |
| Pharmacist | Perform medication reconciliation and enhanced medication education Review HF medication indications with prescribing provider Identify barriers (eg, history of non-adherence, financial, and etc.) Refer to patient assistance program for financial assistance, when applicable Individualize educational session with the patient and patient’s caregiver prior to discharge |
| Social Worker | Perform psycho-social assessment within 24 hours of admission Engage social services early Suggest automatic home health referral |
| Dietician | Provide enhanced dietary education to patient/family about dietary considerations for HF and other comorbid conditions |
| Care Coordinator | Discuss plan of care, educational needs, and discharge needs Confirm transportation is available at discharge Communicate the plan of care to the patient, patient’s caregiver, and IDT members Assess and assist with advanced directives Plan homecare, when ordered Secure follow-up clinic appointment within 3–7 days after discharge (Site specific) Perform follow-up phone call 48–72 hours after discharge with MAWDS follow-up, HF care focus, determine if patient is following the discharge plan of care, and identify any other issues Refer patient and their caregiver to HF classes and resources |
| HF Clinic Nurse | (Site specific) Perform follow-up phone call to patient within 48–72 hours after discharge |
| Providers (MD, APC) | Order homecare evaluation for medication and HF management Consider consults for palliative care or for advanced therapies Use electronic discharge tool (a decision support tool that nudges providers using choice architecture, for example: default discharge medications based on admission and in-hospital prescriptions and on formulary considerations) Transmit plan electronically to outpatient provider Provide clear hand-over communication to skilled nursing facility, as applicable |
In the MTCP ( Figure 1 ), high-risk patients (ie, patients with a high iHF score and/or a high IMRS—just one of the two scores had to be high to assign a patient to the high-risk group) were cared for using a distinct care pathway than were the lower-risk patients (ie, lower-risk = patients with both iHF and IMRS scores in the moderate or low risk score categories). Briefly, those in the high-risk stratum received enhanced assessment from the care team ( Table I ) that lead to personalized changes in their inpatient care, higher intensity post-discharge follow-up, and a more precise post-discharge care plan, and they also received standard HF care. The MTCP was primarily focused on the inpatient period in which a daily care coordination meeting with each member of the care team correlated care by discussing each high-risk patient’s condition, needs, and care plan. After the meeting, the pharmacist and dietician would meet for a more in-depth discussion about medication and dietary needs of individual patients. Care managers worked with providers to determine the suitability of home care for each high-risk patient, and other enhanced care duties ( Table I ) had to be signed off by each team member on a spreadsheet in the unit when they had completed each duty. Through these approaches, care was systematized and standardized carefully, which did not occur in standard HF care. Further, one systematized approach that also was developed was a “weekend” care process since staffing was more limited on weekends and holidays; in this process, a designated team member (generally a charge nurse or pharmacist) would manage the process and send an email to the HF nurse to ensure that any weekend issues were communicated to the HF nurse who resumed MTCP management on Monday. Finally, for the MTCP a monthly meeting would review the function of the process and devise improvements that could break down barriers to precision care. Staff at each hospital was receptive and enthusiastic when the MTCP was taught and implemented at their facility during a site visit by the team that developed the program. The Intermountain system-wide cardiovascular clinical program leadership ensured that adoption of the MTCP became permanent by following up with staff at each facility and performing regular site visits.
Lower-risk patients and controls received standard HF care, but not the enhanced MTCP. Standard care utilized approaches known to improve mortality and hospitalizations among HF patients. These included that at discharge, a follow-up appointment for within 7 days was scheduled to ensure best care, and medication reconciliation was performed to ensure that appropriate medications were prescribed. After discharge, patients received a follow-up phone call within 72 hours from a nurse care coordinator (for MTCP recipients, the call was conducted at 24 hours) to answer questions, remind the patient to attend their follow-up appointment, and ensure that the patient remained in good health based on the Medication Activity Weight Diet Symptoms (MAWDS) patient education and self-management program. During the inpatient stay, standard education included having a floor nurse explain the MAWDS program and other HF education to the patient. Patients receiving MTCP-based care, though, received MAWDS and other HF education from an HF nurse or (in smaller hospitals) a designated floor nurse with enhanced training in MAWDS and HF teaching methods provided this enhanced education.
If a patient had lower-risk iHF and IMRS scores at the beginning of their inpatient stay but one or both of the risk scores rose into the high risk score category during the admission, the patient was moved into the high-risk care pathway for the remainder of the hospitalization. However, these transitions were not tracked systematically, so the analysis could not consider which patients moved to the higher risk pathway later in their hospitalization, but this should be a small proportion. The MTCP was applied to high-risk inpatients (n = 1221) first and lower-risk patients received standard HF inpatient care (n = 1220). For this study, control patients were retrospectively categorized as being high risk (n = 1791) or lower risk (n = 1950) at the time of admission.
Study outcomes
Because a decreased readmission rate may result due to higher mortality and a lower mortality may increase readmission, the study evaluated co-primary study outcomes of 30-day all-cause readmission and 30-day all-cause mortality. Readmissions were queried from the Intermountain electronic data warehouse that contains all electronic health record data from each hospital and clinic in the system. Because Intermountain provides healthcare services to approximately two thirds of the population in its catchment area, these electronic records have been found to identify >90% of re-hospitalizations. Further, any change in provider for subsequent hospitalization is unlikely to be based systematically on the risk scores used in this study because patients were not provided with the scores. Readmission constituted admission for any cause to any Intermountain hospital within 30 days after discharge from the index inpatient admission. These included admission to inpatient care from any location and any emergency department visits that did not include transfer to an inpatient bed. Mortality at 30 days was determined electronically from health system records, Utah death certificates, and national Social Security death records.
Secondary study outcomes included differences in readmission and mortality for lower-risk patients in the MTCP era compared to controls, for high-risk HF patients with preserved left ventricular ejection fraction [HFpEF, left ventricular ejection fraction ≥50% ], and for high-risk HF with reduced ejection fraction (HFrEF, ejection fraction <50%). Further, among high-risk and lower-risk patients separately, other secondary outcomes included differences between MTCP and control groups for patient discharge disposition/home care orders, inpatient length of stay, and total variable cost, plus the separated costs for the index encounter and the 30 days following discharge. For cost estimates, US government penalties for excessive readmissions were not included in study calculations and costs were normalized for inflation to 2013 based on Intermountain cost data.
Statistical considerations
Baseline characteristics were summarized overall for all patients together as percentages of discrete variables or measures of central tendency for continuous variables (ie, mean with standard deviation), with similar evaluation for those in the intervention and control groups. The chi-square test, Student’s T-test, or analysis of variance was used to evaluate differences in baseline variables between intervention and control. In addition to the MTCP intervention and control group designations, other study variables included demographics, clinical measurements, laboratory testing results, prior healthcare utilization, Charlson Comorbidity Index and component comorbidities, medication history, and previous advanced heart failure interventions. In addition to the variables listed in Table II and Supplemental Table S1, study variables included all parameters from the complete blood count and comprehensive metabolic profile.
| Characteristic | Overall | Intervention | Control | P |
|---|---|---|---|---|
| Sample Size | N = 6182 | n = 2441 | n = 3741 | —– |
| Demographics | ||||
| Age (years) | 71.1 ± 14.7 | 69.9 ± 15.0 | 71.9 ± 14.4 | <.001 |
| Sex (female) | 45.5% | 45.3% | 45.6% | .80 |
| Body Mass Index (kg/m 2 ) | 31.7 ± 9.9 | 31.9 ± 10.3 | 31.6 ± 9.6 | .25 |
| Race | ||||
| Caucasian | 90.2% | 89.8% | 90.5% | .51 |
| Asian | 0.8% | 0.8% | 0.8% | |
| Pacific Islander | 2.7% | 2.8% | 2.6% | |
| Native American | 0.9% | 0.9% | 0.9% | |
| African American | 2.0% | 2.5% | 1.7% | |
| Hispanic | 0.02% | 0% | 0.03% | |
| Unknown | 3.4% | 3.2% | 3.4% | |
| Marital Status | ||||
| Married | 51.3% | 50.7% | 51.8% | <.001 |
| Single/Never Married | 12.6% | 14.1% | 11.6% | |
| Divorced | 11.6% | 12.7% | 10.9% | |
| Widowed | 23.1% | 20.9% | 24.5% | |
| Unknown | 1.4% | 1.5% | 1.3% | |
| Selected Clinical/Laboratory Results and Risk Scores at Admission | ||||
| Systolic Blood Pressure (mmHg) | 136 ± 26 | 135 ± 26 | 136 ± 26 | .005 |
| Ejection Fraction (%) (n = 3514) | 45.1 ± 17.7 | 43.3 ± 17.9 | 45.7 ± 17.6 | <.001 |
| B-Type Natriuretic Peptide (pg/mL) (n = 3285) | 1129 ± 1059 | 1170 ± 1095 | 1105 ± 1037 | .09 |
| Hemoglobin (g/dL) | 12.3 ± 2.4 | 12.2 ± 2.4 | 12.3 ± 2.3 | .13 |
| Red Cell Distribution Width (%) | 15.8 ± 2.2% | 15.8 ± 2.2% | 15.8 ± 2.1% | .56 |
| White Blood Cell Count (K/μL) | 8.8 ± 6.6 | 9.0 ± 7.7 | 8.7 ± 5.7 | .08 |
| Sodium (mmol/L) | 138.2 ± 4.8 | 137.8 ± 4.5 | 138.4 ± 5.1 | <.001 |
| Creatinine (mg/dL) | 1.56 ± 1.30 | 1.63 ± 1.39 | 1.52 ± 1.24 | .002 |
| Intermountain Risk Score (IMRS) 30-day Mortality Category | ||||
| High ⁎ | 37.8% | 38.5% | 37.4% | .62 |
| Moderate ⁎ | 44.3% | 44.0% | 44.5% | |
| Low ⁎ | 17.9% | 17.4% | 18.1% | |
| Inpatient Heart Failure (iHF) 30-day Readmission Risk Score Category | ||||
| High ⁎ | 22.9% | 24.6% | 21.8% | .003 |
| Moderate ⁎ | 38.1% | 38.8% | 37.6% | |
| Low ⁎ | 39.0% | 36.6% | 40.7% | |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
