Specialty practices use predictive analytics to enhance value-based care

In 2018, nearly three years after the passage of MACRA, the first wave of fee-for-value payment became tangible for physician practices nationwide. That’s when participants in alternative payment models and the Merit-based Incentive Payment System began receiving detailed performance results and notice of the accompanying financial implications. 

Specialty practices in these programs can draw lessons from the Oncology Care Model (OCM), which includes more than 3,000 providers representing almost 200 practices. Participating oncologists are required to deliver holistic patient care while generating sustainable cost savings.

Many participants have made transformative changes in clinical skills, processes and technologies to support delivery of higher-quality, more cost-efficient care. Despite their efforts and investments, the majority of practices have failed to provide sufficient savings to earn a performance-based payment from the Centers for Medicare & Medicaid Services (CMS). 

In the early phases of the program, practices focused on retrospective analysis of cost drivers and savings opportunities — namely ED visits, hospitalizations, end of life and high-cost drugs — and then developed programs to mitigate these costs. These included extended office hours, care management, palliative care and evidence-based care pathways. 

However, these activities are costly for practices to administer, and because the majority of costs are driven by a minority of patients, it will be increasingly important to focus practice resources and efforts on those patients at greatest risk.

To support this focus, practices must evolve from retrospective cost and quality-performance tracking and reporting to real-time patient risk stratification, predictive analytics and precision medicine based on heterogeneous source data. Getting there will require four steps.

First, broaden and harmonize data sources

Practices need unprecedented access to a broad array of data sources, including those that span multiple care settings. Necessary data elements may include:

• Clinical encounter notes
• Patient survey results
• Genomic data
• Lab results
• Pathology reports and imaging
• Payer claims data
• Prescriptions
• Pharmacy claims
• Hospital census records 

To create a patient- and provider-centric, longitudinal view that incorporates these diverse data types, the data will need to be syntactically scrubbed, identity-matched and semantically normalized.

OCM participants discovered early on that CMS claims data and retrospective performance reports were not timely enough to proactively address negative performance outliers, and therefore hindered their chances of earning a performance-based payment. To address this issue, a handful of practices are leveraging the data to which they already have access — including both clinical and financial systems — to develop analytics, gain real-time visibility into their quality and cost performance, and identify actionable insights.

Second, prioritize high-impact KPIs and apply predictive analytics to develop actionable insights unique to each patient

As the stakes get continually higher, practices will be compelled to prioritize interventions, focusing on patient behaviors that have a direct impact on value-based care performance. The emphasis should be on proactively identifying high-risk patients for whom enhanced services should be initiated.

Additionally, a more nuanced understanding of available treatment options and their impact on total cost, mortality and clinical effectiveness, tied to patients’ individual biogenomic profiles, becomes increasingly important to guide treatment decisions — especially as the cost of breakthrough novel therapies increases. 

To stay ahead of the performance curve, some OCM participants have begun adopting predictive analytics to prioritize high-risk patient interventions and inform treatment decisions, resulting in improved outcomes and reduced costs. These actionable insights span the patient journey and include:

• Predicting which patients are at risk of experiencing an acute adverse event or are most likely to return to the hospital after being discharged
• Tracking practice patterns, including high-cost resource utilization and treatment decisions, to proactively identify performance outliers
• Informing traditional economic assessments of drug selection with value-based insights, including impact on adverse events
• Identifying patients who would benefit from a referral to palliative care or hospice 

The insights are based on established algorithms and statistical models from literature and, increasingly, on machine learning and AI techniques. 

Third, apply these insights to treatment decisions and care management programs, and align actions with their associated value-based performance implications

The advent of value-based payment models has placed newfound importance on clinical performance by tying payment to practice-reported quality measures. Moreover, specialists increasingly are responsible for managing patient outcomes and costs across all conditions and care settings. 

By applying patient-specific, actionable insights to treatment decisions and associated care plans, specialists and their teams are better equipped to deliver patient-centered care while positively addressing the performance indicators that affect the financial success of a practice:

• Reducing avoidable adverse events
• Closing documentation gaps by enforcing new behaviors, manually curating unstructured data or gaining access to external systems (e.g., hospital admissions, discharge and transfer reports).
• Managing variations in care
• Driving and sustaining cost reductions

Early OCM performance results indicated that closing documentation gaps would be paramount. While managing resource use lowered total cost of care among their patient populations, most practices were still exceeding the CMS target price. They discovered that failure to fully identify comorbidities could lead to target pricing that was unrealistically low, and unfortunately most oncologists under-code comorbidities.

Moreover, closing documentation gaps, particularly for patients with comorbidities, enabled care teams to proactively manage these patients and ensure appropriate referrals. 

Fourth, establish a governance model that aligns clinical and financial leadership to manage value-based performance

In new value-based care models, specialists will be expected to assume a more active role to ensure clinical documentation aligns with quality-measure reporting requirements. For many practices, this change requires a cultural shift. 

To drive the necessary cross-team engagement, highest-performing practices are introducing dedicated clinical and value-based care leadership, varying degrees of financial incentives and cultural changes to encourage the use of evidence-based care pathways and improve documentation gaps. For example, some are incorporating relevant metrics into monthly provider performance reviews. Others have focused on top-down communications and a robust physician education program, including webinars and related resources. 

Foremost OCM practices have begun making quality measures and related insights visible to clinicians at the point of care in real time. In parallel, a majority of practices have augmented existing billing staff with value-based care experts, making them better equipped to proactively identify performance gaps and facilitate targeted clinician engagement.  

The keys to long-term success

As the OCM program passes its midpoint, practices recognize the need to leverage vast amounts of data to predict performance outcomes, model and manage anticipated revenue streams and collaboratively address negative performance outliers. 

They have also learned that, in addition to implementing transformational activities such as care coordination focused on high-risk patients, they need to introduce new tactics for closing documentation gaps, managing patients with comorbidities and optimizing use of novel therapies.

Looking ahead, success under value-based care will necessitate the convergence of precision medicine and predictive analytics. Practices will need to integrate evidence-based pathways directly into their clinical workflows and solicit data and insights from health plans and pharma to better understand the holistic value of novel therapies.