Table 3.

Mechanism-outcome realist review findings, by health system case and intervention

Reported Intervention Outcome of Interest	Mechanism Findings
Geisinger
Intervention 1: ProvenHealthNavigator (PHN) Advanced patient-centered medical home (PCMH) model including embedded RN case managers specially trained in population health management and employed by the Geisinger Health Plan.
Significant reductions in both admissions (18% less) and readmissions (36% less) [17] with suggestion of dose-dependent response [21] in Medicaid populations as well as significant prevention of end stage chronic disease indicators for elderly patients [18].	Proactive identification of at-risk individuals by RN case managers using claims-based intelligence followed by review with PCP allowed for enhanced ability to proactively address acute exacerbations and chronic care needs, particularly in elderly populations [17, 18, 20, 21].
Significant improvements in patient perception of care: PHN patients twice as likely to report noticing differences in care, care coordination and service, and believe that the quality of care is difference and improved. Additionally, significantly 12.4% less like to report using the ED as a usual point of care [20].	RN case manager time was dedicated to care coordination – direct phone lines for patients, in-person development of care plans with highest-risk patients, and close follow up post discharge [17, 20].
Intervention 2: Geisinger Monitoring Program (GMP) Post-discharge patients received automated phone calls that collected individualized information using EHR-integrated automated Interactive Voice Response (IVR) technology, with areas of concern or non-compliance requiring follow up pushed to the embedded PC RN case manager in real time via EHR. Four-week program; not intended to replace traditional contact.
Compared to case management alone without the IVR tool, there was a 44% intent-to-treat reduction in likelihood for readmissions, driven by second year of study (“incremental benefit”).	Efficient extension of case manager capacity; design based on failure of a manual readmissions prevention program of similar scale (~ 30 m in manual program vs 2-3 m in GMP).
Only 4% failed to participate in the full program compared to large drop-out rate seen in other telemonitoring program studies.	Attention to optimal balance of frequency of automated contact. Patients receive calls 1x per week and programming to retry calls at set intervals.
Intervention 3: Comprehensive Care Clinic (CCC) Augmentation of PCMH model offering intensive primary care for Adolescents and Young Adults with Special Care and Health Needs (AYASCHN). For patients meeting complexity criteria, primary care provided by multidisciplinary care team including an internal medicine-pediatric physician, advanced practice practitioner, pharmacist, and embedded RN case manager.
Per member per month reduction of ~ 78% reduction for acute hospital admissions (P = 0.053) and ~ 60.3% in ED visits (P = 0.017)	Detailed patient/family self-management education surrounding exacerbations and how to respond provided, paired with availability of same day services, may have helped alleviate seeking emergency care unnecessarily
Per member per month total cost reduction of 28% ($3931 observed vs. $5451 expected; P = 0.028). Inpatient cost appears to drive the reduction as the most significant source (P = 0.028).	Care plans developed with family in joint appointments with RN case manager and dually trained internal medicine-pediatrics physician, as well as required laboratory, screening and other monitoring for medication optimization, may have helped to alleviate common gaps AYASCHN face when transitioning into regular adult primary care.
Kaiser Permanente
Intervention 1: “5-Element Transitional bundle” Multi-element standardized discharge protocol to reduce all-cause readmissions, designed using outcomes data and patient-identified post-discharge needs, with iterative improvements.
Significant decreases in readmissions rates up to 12.1% (P < 0.0001) and risk ratios from 0.72 to 0.66 (0.0001) moving from 75th to 90th percentile nationally.	Greater patient connection in-patient providers appears to be the major drivers of change, including the creation of a 24/7 “post-hospital hotline” with triage nurses able to page hospitalists as needed (required re-brokering of physician contracts to extend of hospitalist oversight for 48 hours post-discharge), and 48-hour follow up for all patients or 30-day case management for high risk patient by transition RNs.
HCAHPS discharge instruction scores moved from 50th percentile to 90th percentile nationally (P < 0.0001). Over an 8 year period mean time to first appointment post discharge reduced by 4.6 days (P < 0.0001)	Clear, discharge summaries with single post-hospital line listed to call for any reason and all follow up tests and appointments listed – key element in transitional bundle required ambulatory care appointments made prior to discharge made this feasible.
Intervention 2: Nurse Knowledge Exchange Plus (NKEPlus) Set of minimum specifications and acronyms designed to improve nurse handoffs and patient engagement at shift change.
Mean HCAHPS score for 82 nursing units improved significantly by 3.3% (range 0.2-5.9% change per unit) over 4 years. Mean score for NKEPlus behaviors (based on four HCHAPS elements for RN-patient communication) for 60 units improved significantly by 4.7% (range 0.1-7.8% change per unit) over 4 years.	Nurses given protected time at shift change to conduct patient-engaged handoff at bedside via unit protocol changes (rounding on patient needs in the hour before shift change and non-nurse staff picking up extra responsibilities during shift change to minimize interruptions, as well as shift assignments limiting number of departing nurses incoming nurses heard from. Scale reflected at right supported by Innovation Consultancy which worked to ensure intervention was accepted, adapted, and sustained across many sites.
Intervention 3: Clinical Pharmacy Call Center (CPCC) Phone-based telepharmacy program developed to address high demand to regional call center related to drug therapies. 20-24 FTE personnel serve 400,000 patients by working screening and reconciliation for new and/or recently discharged patients, fulfilling refill requests, and spreading information related to guidance or plan changes to patients. CPCC also helps develop and advise on clinical guidelines, acting as a support to clinical staff with drug-related questions.
Approximately ~ 10 K annual new patient transitions, 30-40% of which are estimated to be Medicare beneficiaries. Random chart review showed a cost avoidance of $324 per member per year, representing a cost avoidance benefit to KPCR members of $16.2 million dollars and an estimated $4.3 million in avoided primary clinic costs.	CPCC clinical pharmacists work alongside regional call center physicians to make medication and lab orders for new patients awaiting first PCP appointment. This frees up PCP time to focus on other care needs and reduces need for multiple initial appointments.
SNF patients transitioned to home via CPCC clinical pharmacists were 78% less likely to die (95% CI, 0.06–0.88), 29% less likely to need an ED visit (95% CI, 0.36–1.39) and 17% more likely to follow up with primary care (95% CI, 0.99–1.37) versus usual care, which meant transition care being provided by primary care clinics alone. CPCC detected at least one potential drug problem in 90% of discharge summaries [26].	Working with Kaiser’s Chronic Care Coordination (CCC) Department, CPCC provided medication review, reconciliation, and counselling for SNF patients post-discharge in their home; researchers comment this potentially lower-stress setting may have allowed for increased retention versus clinical setting. Additionally, pharmacist to specifically instruct SNF patients or caregivers to bring the medications physically in front of them, which may not be realistic in clinical settings. Effects mediated by CCC coordinators who receive CPCC notes, provide additional services, and make recommendations to primary care providers [26].
Cleveland Clinic
Intervention 1: Connected Care SNF Cleveland Clinic physicians and advanced practice professionals visited patients discharged to SNFs 4-5 times per week and w/i 48 hours provided a comprehensive assessment and record review
Participating SNFs saw significant 6.8% absolute reduction in all-cause 30 day admissions versus usual care. Particular benefit to high-risk and medical (versus surgical) patients.	Meaningful, frequent face-to-face physician engagement with SNF patients not typically receiving continuing medical care after discharge to SNF. Emphasis on goals of care and palliative care specialist as part of connected care team ensure appropriate expectations and prevention of inappropriate readmissions.
Intervention 2: R.E.D.E. to Communicate: Foundations of Health Care Communication Relational skills course designed by and implemented in Cleveland Clinic. System-wide, physician targeted 8-hour block of training with small and large group skills practice.
For physicians who took part in the R.E.D.E. communication skills training, significant improvement to CGCAHPS measuring patient experience scores were higher than for controls (92.09 vs. 91.09, p < 0.03). Significant improvement in the post-course HCAHPS “Respect” domain means versus control groups (91.08 vs. 88.79 respectively, p = 0.02) [30].	Main mechanism appears to be improving physician self-efficacy and sense of worth. Significant improvements to empathy and burnout among physicians were seen alongside patient experience improvements, even after 3 months following the session, and pre-post surveys showing significant satisfaction with course [30]. R.E.D.E strategies being integrated directly into the medical interview likely reduced physician burden to find individual way to incorporate relational skills to patient interaction indirectly or as an additional point of care [29, 30].
Intervention 3: Mobile Stroke Treatment Unit (MSTU) MSTU staffed with RN, paramedic, EMT and crossed trained EMT-CT technologist. Vehicle is outfitted with CT scanner with output to neuroradiologist, point-of-care lab equipment, and A/V equipment for real-time instruction from Cleveland Clinic vascular neurologist (VN)
MSTU program patients received thrombolysis significantly sooner than control patients in two main ways: 38.5 minutes sooner from the alarm time and 26.5 minutes sooner from door. 25% of patient who received thrombolysis did so within recommended 60 minutes of symptom onset.	Dramatically and significantly sooner diagnostic service (CT scan) drove faster treatment times. Patient arrives to hospital and MSTU within similar timeframe, but researchers comment hospital environment brings delays to care delivery. Coordination with city EMS agencies for simultaneous dispatch ensured mechanism achieves greatest potential.
Mayo Clinic
Intervention 1: Teleneonatology program Mayo Clinic provides telemedicine consults to six community hospitals (level I and level II nurseries) throughout the region using A/V live-feedback instruction for high risk deliveries and/or newborn resuscitation. Goals include improve outcomes for neonates and reduce the need to transfer infants up 40-120mi away depending on site.
Neonates receiving teleneonatal consults were significantly more likely to undergo measurement of temperature, blood glucose and blood gas, and significantly more likely to undergo all three. In matched pairs analyses by blinded expert panels, telemedicine neonates were significantly more like to have been impacted positively within 1 h of birth and/or to have been provided with higher quality resuscitation [33].	Mixed method study into barriers and facilitators [34] revealed that reliable and easy to use IT infrastructure was important for this time-sensitive, live instruction intensive telemedicine intervention. Mayo’s Center for Connected Care worked closely to analyze, troubleshoot, and enhance IT capabilities and improve processes – such as switching from wireless to wired connections and allowing neonatologist control of camera angles [32].
Intervention 2: Integrated, Colocated Specialist (ICS) - Neurology Neurologist in co-located in primary care clinic at 0.6 FTE with partial allocation of 3 RNs and 3 MAs. The ICS-Neurology model goal was to increase collaboration with PCPs including physicians and APP. Curbside and electronic consultations were encouraged, and effect on face-to-face consultation with ICS-neurology, diagnostics and tertiary referral were measured.
Referrals for face-to-face neurology visits for patients reduced by 25%; Referrals to tertiary neurology reduced by 64%.	Collaboration was encouraged through alteration of the EHR such that the ordering of face-to-face neurology consultations were not permitted unless a curbside consultation (informal via phone or email) had been performed and recorded by PCP. As a result of this process feature, 33% of curbside consultations resulted in agreement with PCP plan and avoided the need for face-to-face consultation entirely.
Nearly one quarter of curbside consults resulted in escalation for the patient to be seen same day or sooner than next available appointment.	Encouragement of curbside consultation paired with the ICS-neurologist having approximately 50% unscheduled time meant that patients with urgent neurological symptoms identified via PCP- ICS Neurologist collaboration were able to access care sooner.
Intervention 3: MayoExpertAdvisor (MEA) Cardiovascular Risk Tool EHR-integrated clinical decision support tool for cardiovascular risk assessment and improved cholesterol management. Study provides simulation analysis to see effects on accuracy and time management among specialists, PCP and APP in routine care settings.
Clinicians who did not use MEA spent an average of 5 minutes and 8 seconds to determine the ASCVD score and determine a recommendation for patient care. With MEA, the clinicians spent a total of 1 minute and 31 seconds to calculate the ASCVD score and determine the recommendation, a reduced of 3 minutes and 38 sections. Chart review at the primary care practice indicates this equates to a time savings of 3 hours and 45 minutes per day, a half day of clinician time.	MEA saves clinician time by consolidating relevant information into one place. With MEA, providers did not have to move through and between multiple tabs to find relevant information. Further, MEA is not a separate application but conveniently integrated in the EHR via Mayo-develop EHR viewer called “Synthesis” that retrieves and presents data to clinicians in a more intuitive, easy-to-navigate format.
Clinicians without MEA had a 60.61% accuracy of ASCVD risk score calculation and 60.61% accuracy in selecting the guideline recommended treatment and there were significant differences in time to complete both tasks between physicians, NPs, Pas, and other clinical types. With MEA, clinicians had 100% accuracy in both and the time difference between providers normalized.	Mayo has put considerable resources into their IT infrastructure towards care standardization, ultimately allowing for the technology to support this and other similar interventions. Natural language processing (NLP) allowed auto-pulling from EHR is promoting ASCVD risk score accuracy – researchers comment that the most frequent errors that clinicians made in were related to not identifying the most recent data and inputting age, gender and smoking status incorrectly. Further, Mayo’s extensive development of clinical decision support drawing on Mayo physician-guided knowledge base [32] eliminated variation in recommendation within and between provider groups.