How payers could unlock value by improving transitions of care

| Article

Transitions of care (ToC)—the movement of patients from one care setting to another, such as discharges from hospitals to the home, nursing facilities, and other settings—are a ubiquitous part of healthcare systems worldwide. Each transition represents an opportunity for healthcare stakeholders (providers, payers, health services, and technology companies,1 alongside public and community resources) to improve patient care. Transitions could be viewed as critical inflection points, or “fresh starts,”2 for patients and as powerful opportunities for providers and payers to deepen and sustain patient engagement.

Preventing avoidable hospital readmissions (defined as repeat admissions, typically within 30 days of an initial admission) are an important aspect of ToC management. Such prevention improves a range of healthcare outcomes, from patient experience (such as less discomfort from an additional inpatient stay) and care quality (for example, avoiding exacerbation of the illness) to affordability with respect to patient out-of-pocket costs and direct costs (see sidebar, “Readmissions by the numbers”).

Although readmission challenges reflect real fractures within US healthcare, the landscape is changing, creating new opportunities for payers and other healthcare stakeholders to improve ToC. The US Department of Health and Human Services has supported the postacute-care ecosystem through data standardization, interoperability, and updated postacute-care payment models for facilities. In parallel, innovations in health services are leading to improved data sharing, easier remote-patient monitoring, and better patient engagement.

This article explores the role of data and analytics in preventing readmissions and details three areas of focus for ToC improvement: engaging patients early and digitally, establishing real-time decision support infrastructure, and facilitating “last mile” care delivery and digital patient engagement.

Data and analytics underpin the effort to reduce readmissions

Avoiding preventable readmissions is, in many ways, the linchpin for successful transitions. The readmission opportunity is underscored by a telling fact that emerged from our analysis: patients with very similar profiles in terms of episodic condition, health risk scores, and social circumstances routinely have different readmission outcomes.

To illustrate, consider two patients, Jack and Fred (Exhibit 1). Both are 68 years old, were originally admitted to a hospital for heart failure with a major complication or comorbidity (MCC), and had comparable patient risk scores based on chronic conditions and demographic factors. Using a machine learning (ML) model that controls for more than 1,500 episodic, clinical, and other (such as social determinant of health [SDoH]) features, we find that on average, each would be predicted to have about a 23 percent likelihood of a 30-day readmission in absence of primary care follow-up.

A machine learning model suggests that following up with a primary-care physician substantially reduces the likelihood of readmission.

However, Fred sees a primary-care physician (PCP) within seven days of discharge, which reduces his risk of readmission to 15 percent. Jack does not, resulting in a difference of eight percentage points in their respective risks.3

Tech-enabled ecosystems are at the crux of ToC activities, and payers could play a critical role in orchestrating such ecosystems with their partners. From an infrastructure perspective, provider–payer data interoperability is accelerating data sharing, including with admission, transfer, and discharge (ADT) systems—an opportunity payers could invest in today. From a health services perspective, digital-focused companies providing solutions ranging from virtual-care delivery (such as specialty consults) to care coordination solutions (such as digital pharmacy solutions) and member engagement applications (such as remote monitoring or self-help apps) are eager for payer partnerships. From a data intelligence perspective, ML models have become a powerful and widely adopted capability for payers and their partners to predict readmission risk and design intervention strategies that have been demonstrated to work.

Integrated care providers, including ChenMed and Kaiser Permanente, are demonstrating what tech-enabled ToC ecosystems could look like, with lessons for payers. Kaiser Permanente’s Northwest US branch, for example, reduced its all-cause readmission rate by about one-third (from 14.0–15.0 percent to 9.5–10.0 percent) by introducing a postdischarge bundle of activities. The bundle included stratifying patients into risk tiers at the time of admission, standardizing discharge summaries and including relevant information for patients and providers (such as medication changes and test results), coordinating among doctors on accurate medication lists and quantities, ensuring a nurse calls the patient within 48 to 72 hours to remind patients of must-dos, helping schedule a PCP visit within seven days, arranging mobile-health partners for high-risk patients, and setting up a 24/7 hotline for patients to ask questions and get help with emergencies.4

Three areas of focus for improvement

Payers play a critical role in shaping ToC outcomes through program designs and partnerships with providers. Core programs include value-based reimbursements aligned to 30-day readmissions and other outcomes, utilization management programs for appropriate length of stays, and case management supporting patient discharge, typically for high-risk patients.

However, in our experience, payers could be doing even more to improve patient outcomes and reduce preventable admissions through a tech-enabled patient engagement strategy. As discussed in previous McKinsey articles,5The role of personalization in the care journey: An example of patient engagement to reduce readmissions,” McKinsey, August 5, 2021; and Jenny Cordina, Greg Gilbert, Nevada Griffin, and Rohit Kumar, “Next-generation member engagement during the care journey,” McKinsey, July 23, 2019. payers are still in the early stages of adopting the kinds of consumer engagement strategies common to B2C companies, including AI-based targeting and agile consumer engagement campaigns.6 We have identified three areas that payers can consider focusing on for improvement7:

1. Engaging patients early and digitally

The first few days of discharge provide a critical opportunity for care coordinators to engage with patients on activities including scheduling PCP or specialist follow-up visits, ensuring adherence to taking medications, identifying discharge destination preferences, and addressing potential barriers to healthcare access. Recent Centers for Medicare & Medicaid Services (CMS) interoperability rules8 are improving ADT data connectivity and accelerating healthcare information exchange offerings, creating new opportunities for payers to use real-time data for early patient engagement.

Advanced-analytics models predicting patient readmission risk and guiding intervention strategies are another critical technology opportunity. Payers are well aware of the advantage of high-touch patient engagement pre- and postdischarge, but they could take a more nuanced approach to implementing patient interventions based on risk level.

Claims-based ML models are increasingly generating accurate predictions for intervention planning based on the characteristics of the patient and relevant episode. McKinsey analysis that used an ML model with Medicare fee-for-service (FFS) data from CMS’s Limited Data Set (LDS) files showed promising results (Exhibit 2). For the 5 percent of patients predicted to have the highest risk of readmission if no action is taken, about 75 percent were readmitted; for the top 15 percent of highest-risk patients, roughly 55 percent were readmitted.

Using machine learning modeling, health plans could more accurately predict readmission rates for patients with congestive heart failure.

By examining a diverse set of member data—including information from electronic health record feeds, diagnostic data (such as labs, genetics, pathology, and imaging), and patient-centered data collected through wearables, patient-reported outcomes, and at-home monitoring devices—payers could obtain valuable insights to improve healthcare outcomes.

Other tech-driven workflow enhancements could further improve ToC, including provider-level analytics that help prioritize in-person resources for providers with the highest volume of readmissions and diagnosis-related, group-level analytics that identify anomalies in the length of stay for postacute care.

Additionally, participants in the health services ecosystem could help payers move toward 100 percent early engagement of members. Payers could partner with companies with ToC patient-facing platforms to facilitate video and digital connections with patients in the hospital or shortly after discharge, and to engage members with interactive voice response (IVR) and text messaging.

Overall, tech-enabled data sharing, readmission analytics, workflow modules, and health services partnerships provide the basis for high-ROI ToC program design. Some payers are taking the lead in prioritizing deployment of high-touch, in-person resources. For example, a New York–based health plan conducted a pilot program at a target hospital that resulted in a 21 percent reduction in readmissions of its dually eligible Medicare and Medicaid members.9 It arranged for nurse practitioners to visit members in the hospital and collaborate with the hospital team. Representatives called members with reminders to set up follow-up visits and fill medications. And nurses conducted home visits with continued medication reconciliation, family education, and coordination with the care team.

2. Establishing real-time decision support infrastructure

The provider’s recommendation on a patient’s discharge destination (such as a skilled-nursing facility [SNF] or at-home health) is among the most important aspects of a ToC.

Indeed, in our analysis of predictive factors for readmission for Medicare patients, discharge destination ranked second (behind patient health risk profile) in predictive power. For individuals with similar characteristics—DRG, health risk profiles, SDoH factors, and interventions (such as PCP follow-ups and medication reconciliation)—the discharge destination choice influenced probability of readmission, on average, by more than 30 percent in either direction, depending on case particulars.

Based on interviews with physicians and discharge managers and readmission analytics, in many situations there is no unambiguously “right” discharge destination. For example, Medicare patients with heart failure and shock with major complications or comorbidities are discharged to SNFs about 20 to 25 percent of the time, and to home health 25 to 35 percent of the time. Interviewees shared that providers often err on the side of higher-intensity discharge sites and facilities with which they have established relationships.

Real-time decision support in the form of postacute-care performance analytics could give discharge planners much-needed visibility into network and claims data that only payers have access to today. These insights would go beyond patient DRG and acuity and include historical readmission data by facility for similar case types, SDoH data (such as proximity to patient home), indicators of whether the specific provider partners with ancillary care delivery services (such as PCP networks specializing in postacute-care follow-ups) or uses an interoperable data platform, and other standard information including the CMS Star rating system for nursing facilities.

Payers can create phone- or web-based real-time decision support tools to help providers decide which type of postacute-care facility is most appropriate for the patient. As illustrated in Exhibit 3, the tool could share information about in-network postacute-care facilities, including CMS-published metrics of quality, outcomes, and typical bed utilization as well as supplemental information about patient access to a facility (such as distance to home) and patient clinical history. The tool could also provide readmission rate statistics associated with different postacute levels of care and facilities. Such information sharing can work in concert with a preferred or tier-based postacute-care network strategy.

A decision support tool could help providers choose the optimal postacute-care facility for each patient.

More broadly, payers could create incentives for data sharing through preferred, or tiered, partner decisions. One regional payer recently set up an analytics team to estimate how SNF selection affects the likelihood of various adverse events (such as readmission and mortality) and launched conversations with multiple major health systems and a PCP network on preferred SNF networks, which has increased the volume of patients discharged to high-performing SNFs.

3. Facilitating last-mile care delivery and digital patient engagement

Ensuring that a patient adheres to the provider’s recommendations following discharge is among the biggest challenges of ToC, because returning to one’s original social setting, outside the clinical environment, can induce lapses that result in readmissions. Educating patients in the discharge process through simplified instructions in the patient’s first language and having conversations about postacute-care options, medications, and physician follow-ups are important; however, based on our interviews with physicians, patient compliance is often impeded by information overload and day-to-day distractions.

Through digital partnerships, payers could make it easier for patients to complete critical activities by providing varying levels of direct support, depending on patient need.

To help circumvent these challenges, payers could drive a technology strategy premised on a proactive, at-your-front-door approach to patient adherence through last-mile care delivery and digital patient engagement.

Through digital partnerships, payers could make it easier for patients to complete critical activities by providing varying levels of direct support, depending on patient need. To support medication compliance, payers could partner with digital pharmaceutical companies to arrange virtual consults (for medication reconciliation and education, for example), mail-order deliveries of medications, and “no cost” smart pill pack or automatic-dispensing solutions. For provider follow-up appointments, payers could schedule appointments on behalf of patients, auto-populate paperwork, and partner with on-demand transportation companies (such as Uber or Lyft) to facilitate travel. For the highest-risk patients, payers could partner with supplemental home health providers and supply patients with preloaded tablets with virtual health applications, treatment information, and care navigation resources.

To maximize ROI, payers could establish higher-touch facilitation for patients at higher risk of readmission and for patients for whom the intervention impact is greatest. For example, PCP and specialist follow-ups can potentially have more impact for patients with psychoses and heart failure than for those with renal failure and septicemia.

Digital technologies and agile patient engagement models have the potential to dramatically improve payer (and provider) patient engagement reach rates, response rates, and behavior-targeting rates. Patient engagement vendors are increasingly moving beyond sociodemographic, behavioral, and attitudinal characteristics and toward deep psychographic profiling that also takes into account values, principles, beliefs, and emotions to classify patients and optimize content messaging accordingly, based on important word triggers.

These technologies are available to in-house payer and provider care management teams through patient engagement workflow tools, and they can be combined with the collection and storage of campaign data and metrics to enable optimized patient engagement. Simple A/B tests can generate powerful insights. One health plan doubled its patient response rate by using personalized IVR and allowing its name to appear in caller ID on the recipient’s phone. A health system used automation technology to adjust the urgency of text messages to patients depending on the number of days since discharge and based on patient response. Payers can couple these patient engagement approaches with remote patient-monitoring technologies that trigger patient outreach when metrics cross certain thresholds (such as weight gain in a short period of time).

To maximize ROI, payers could establish higher-touch facilitation for patients at higher risk of readmission and for patients for whom the intervention impact is greatest.

Today, health plan leaders have a powerful opportunity to reimagine their approaches to ToC, leveraging the improved health data ecosystem and available digital patient engagement technologies to reduce preventable readmissions. We believe that the patient engagement approach is at the core of the substantial variation in readmission outcomes across payers and providers. The first step for leaders is to establish a clear view of the potential opportunity and then to make an enterprise-wide commitment to transforming ToC into a core lever for enterprise and product-specific goals with respect to affordability, patient satisfaction, and growth.

Explore a career with us