Calculated Risk: How an Algorithm Can Improve Care for High-Risk Patients

This blog is part of a series written by Sampath Narayanan, CIO of ConcertoHealth, on the work his team is doing to improve population health outcomes.

ConcertoHealth originated as a primary care medical group focused on managing dual-eligible Medicare/Medicaid beneficiaries. According to CMS, this population costs the US healthcare system more than $306 billion annually. As a risk-bearing provider partner to health plans, ConcertoHealth has developed expertise in managing America’s highest risk patients — those who are frail, elderly or diagnosed with multiple chronic conditions — so that they can receive the right care at the right place when they need it most.

To best focus resources, ConcertoHealth pays special attention to the subset of patients considered at most risk for worsening health due to an increased prevalence of chronic conditions and multiple comorbidities. Fifty-two percent of ConcertoHealth patients have three or more chronic conditions. Those patients that are Medicaid recipients also face additional social challenges, including inconsistent access to a phone or transportation, and oftentimes exhibit low medication adherence. These social obstacles only compound, resulting in higher readmission rates as compared to other populations with similar diagnoses.

Risk Stratification

With economic responsibility for the complete care of more than 22,000 patients, ConcertoHealth first tackles the management of this population by stratifying patients across risk bands. This is a complex process informed by multiple sources of data, bound by varying levels of state and federal regulations and health plan contract stipulations, and guided by regional macro-level health trends. Risk stratification is necessary to ensure that all patients — particularly those who are at highest risk for worsening health — are identified in a timely manner, so that they can receive focused care.

Risk calibration is dependent upon the maturity (i.e., volume and quality) of the data available for each patient. To mitigate the risk of gaps in data dependency, ConcertoHealth aggregates a variety of clinical, social, behavioral, claims and risk-related data around patients from multiple sources, including payers, laboratories, pharmacy benefits managers, health information exchanges and CMS.

Using this dataset, ConcertoHealth has developed a risk stratification model to best identify each patient’s level of risk: high, rising, medium or low. Depending on the maturity, relevance, and up-to-date data for a specific patient, the stratification model employs one or more of 10 algorithms, in a hierarchical order, to identify each patient’s risk level. These algorithms include both proprietary logic and well-known industry-standard models, including the Charlson Comorbidity Index and the Johns Hopkins ACG methodology, that weigh each patient’s demographics, chronic conditions, hospital and skilled nursing facility (SNF) activity, and enrollment in government health programs, as well as ConcertoHealth’s own clinical judgment.

Model of Care in Action

ConcertoHealth has designed a member-centric care model that uses risk status to inform the appropriate level of patient outreach. The care model prescribes the cadence of patient outreach, including health risk assessments, care plan updates, medication reconciliation, and follow-ups to emergency department (ED) activity. Patients determined to be at high risk or rising risk require more frequent interaction and more intensive intervention; therefore, care managers engage these patients more often to monitor their health status, manage gaps in care, and coordinate the interdisciplinary care team.

With variances across health plan partners, benefit structures, and risk bands, our care management and clinical teams have been challenged to scale their outreach efforts. This led to the development of Patient3D, a proprietary population health analytics platform that leverages a rules engine to create prioritized patient work queues.

ConcertoHealth has also developed medical economics dashboards and operational reports. These analytical outputs allow us to monitor key performance indicators, including: ED visits per thousand, acute discharges per thousand, average length of stay and 30-day readmissions/returns to acute care rates. Based on the trends and inferences, ConcertoHealth delivers proactive clinical interventions to address quality care gaps and improve health outcomes.

“Hotspotters” algorithm

In conjunction with these datasets, we apply a proprietary “hotspotters” algorithm to surface high-risk patients based on longitudinal health history, ED visit patterns, outstanding quality care gaps, cost of claims, and other data markers. Resulting patient lists include high utilizers who might experience difficulty in accessing care, or who lack engagement with healthcare providers; in turn, they represent a significant, disproportionate share of medical expense. By identifying and prioritizing these patients, ConcertoHealth’s Patient Engagement Team can intervene, provide social services and logistical support, and even deploy field-based primary care services where and when they are most needed.

A recent example of a “hotspotter” patient is a 56-year-old female from Washington who is homeless. In the last year, she visited the ED more than 50 times, with a history of heart disease, chronic obstructive pulmonary disease, rheumatoid arthritis, and previous methamphetamine use. When she presented at the ED after an apparent fall, Patient3D triggered the Patient Engagement Team social worker to visit her and arrange same-day transportation to a primary care provider. Her anxiety about her own health — coupled with her son’s poor health — resulted in continued ED spikes, despite housing finally being secured for both individuals. The turnaround moment only came when ConcertoHealth found a way to establish primary care for her son as well. Through both primary care and social worker management, ConcertoHealth is able to closely monitor their housing status, reinforce their care plans, educate them on warning signs, and reduce their ED usage.

To proactively manage this high-risk population, ConcertoHealth recently expanded its field-based care teams in both Washington and Michigan, deploying physicians, nurse practitioners, medical assistants, pharmacists, and patient engagement specialists to meet patients where they need help the most. ConcertoHealth has equipped these field-based care teams with mobility solutions, including tablets and mobile software that allows them to provide and document care, irrespective of whether the patient is in the hospital, SNF, or home. Patient3D, enhanced with a “hotspotter” flag for qualifying patients, continues to serve as the work queue and source-of-truth for care teams to prioritize interventions.