The Power of Data: Avoiding Hospitalizations for People with Late-Stage Chronic Kidney Disease

Keeping people living with late-stage chronic kidney disease or kidney failure out of the hospital is one of the most impactful ways to lower the total cost of care. That can be especially challenging given the many comorbidities faced by this complex patient population, including hypertension and diabetes. Despite these challenges, Interwell Health’s holistic care model, combined with predictive analytics, is driving improved health outcomes, avoiding hospitalizations, and reducing costs.

Interwell’s hospitalization rate is almost 20 percent below the national average.¹ This includes an average of all 125,000 patients under management, inclusive of all stages (CKD 3-5 and ESKD), risk factors, geographies, comorbidities, or level of engagement. This national scale is unprecedented in the industry, demonstrating the total population health impact of a comprehensive approach to care management and the advantages of having the largest network of 1,700 nephrologists nationwide.

Interwell also delivers results with a strong regional breadth of value-based care operations across many areas of the country. In partnership with a regional health plan over the past two years, Interwell reduced hospitalizations among members with late-stage chronic kidney disease (CKD) by 25 percent, while driving hospitalizations down by 30 percent for members with end-stage kidney disease (ESKD). The work to reduce hospitalizations lowered the cost of care in the late-stage CKD and ESKD populations by 13 percent for our payer partner.

These shared savings are being re-invested into programs that will further optimize the company’s care model, creating additional machine learning models to drive even greater reductions in hospitalizations.

Improving care through population health

Interwell has put together leading experts across the company to expand an investment in population health, with teams established to improve patient care in four areas: performance management, risk stratification, health equity, and program accreditation and delegation.

“Population Health is really about translating data to new insights that can ultimately be used to drive change and make a difference in our patient’s lives,” said Corrin Arone, vice president of population health and value-based care.

Because reducing hospitalizations is one of the biggest keys to driving down total cost of care, the population health team develops insights that are handed off to program managers who share specific action plans with physician practices and the clinical team. The population health team reviews whether those action plans have the desired impact.

“We are helping physicians identify opportunities and implement new interventions, unique to their patient populations, that can bring the greatest improvements to their patient’s health and total cost of care,” said Arone. “We can make specific, targeted recommendations based on information produced from numerous and complex data sources, and in a way that could not otherwise be done manually.”

The National Committee for Quality Assurance (NCQA) has awarded accreditation to Interwell’s Population Health Program. This accreditation is a testament to the company’s commitment to enhancing and updating its care model by focusing on performance analysis and continuous quality improvement. The insights developed by the population health team also feed new innovations at Interwell, including within its proprietary care management platform to integrate data for timely interventions.

“The Population Health Management programs moves us in greater alignment with the increased focus on person centered population health management,” said Margaret E. O’Kane, President, NCQA. “Not only does it add value to existing quality improvement efforts; it also demonstrates an organization’s highest level of commitment to improving the quality of care that meets people’s needs.”

More than half of all employees at Interwell are members of the care team directly engaging patients each day, including nurses, care coordinators, social workers, and dietitians. This team is constantly reviewing labs, doctor visits, and other data to help patients manage often complex comorbid conditions, ensuring they understand their disease and potential imminent risks to their health.

Predictive analytics help identify most at-risk of hospitalization

Interwell uses proprietary machine learning tools designed specifically to identify people with CKD who are most at risk for hospitalization. In 2023, Interwell doubled the size of its data science team, adding the most advanced technology, as well as experts in artificial intelligence who have further improved the company's existing predictive models.

Predictive analytics is the use of algorithms that can evolve with trends to make accurate assumptions about future events. Advanced machine learning models like those created by Interwell can continuously learn over time as new data is added to improve its accuracy and better understand the nuance of each individual event, enabling the delivery of more personalized care.

The company's hospitalization model for late-stage CKD can accurately predict this risk many months in advance using claims data. The machine learning model is further enriched by multiple sources across the patient journey. This work is helping deliver even greater results for our patients in value-based contracts with private payers, reducing hospitalizations in 2023 to almost 26 percent below the national average for people on hemodialysis, and 8 percent lower for people with CKD (3-5).¹

“It’s really about developing personalized care and treatment,” said Alex Ruterbories, director of data science. “Our hospitalization model is being used across our payer contracts to determine which patients need to be engaged directly with our care team as soon as possible. It is one of many places where we can combine our models with our own rich and robust data to customize our patient outreach and interventions at every step of the patient journey.”

For patients with ESKD who are dialyzing at Interwell-aligned dialysis centers, the company also leverages an Imminent Hospitalization Predictive Model (IHPM), which can analyze more than a thousand variables. This predictive tool, developed several years ago and evolved over time, delivers a risk score and the top reasons why an ESKD patient is at risk for hospitalization, providing a guide to how care teams can intervene and reduce that risk.

The user-friendly dashboard offers a quick snapshot review for nurses and care coordinators and is updated daily, using near real-time data from these aligned dialysis centers. This allows Interwell’s integrated care teams to respond immediately and perform the necessary clinical interventions needed for a patient to avoid a costly hospital admission.

Reducing readmissions with targeted engagement

Patients with CKD who have recently left the hospital are at the highest risk of being admitted again over the next three months. More than 30 percent of patients on dialysis will be readmitted within 30 days of discharge, according to data from the Centers for Medicare and Medicaid Services. One recent study suggests that up to 50 percent of all readmissions are potentially avoidable.²

This is why Interwell ensures prompt and proper follow-up care is given to anyone in a value-based care program who has been recently discharged from the hospital.

“If we see a hospital admission, we are able to flag that patient for timely intervention and help them schedule a follow-up visit with their nephrologist or other physicians,” said Arone. “It’s critical we ensure those patients are given a higher level of attention.”

These efforts have already led to a drop in readmissions in 2023 across Interwell’s 125,000 eligible CKD and ESKD patients, with the expectation for further reductions in 2024. For patients on hemodialysis in 2023, Interwell’s all-cause readmission rate is 25.7 percent, compared to the national average of 34.9 percent for Medicare patients (a 27 percent reduction).¹

Identifying risk of progression to kidney failure

The population health team has also implemented an improved predictive model to better identify patients most at risk of progressing from Stage 4 or 5 kidney disease to ESKD. This helps target patients early who should be preparing for a transition to transplant, home dialysis, conservative care, or in-center treatment with permanent access in place.

If CKD progresses too quickly without detection or intervention, it can result in a “crash” into dialysis in the hospital, which leads to poor health outcomes and is also incredibly costly to the healthcare system. One recent study suggests a planned transition, or optimal start, resulted in a $49,168 savings in the 12 months pre-dialysis compared to a crash start, and an additional $16,565 savings in the following year.³ Interwell’s predictive tool empowers the care team to target patients at the right time and ensures the proper use of resources. Interwell continues to vastly outperform the national average in planned transitions.

"If we were to try to intervene with every single patient in the exact same way, we would have limited impact,” said Dr. Carmen A. Peralta, Chief Clinical Officer. “We can prioritize patients depending on their specific risk factors and needs. This also includes identifying areas of social determinants of health, such as food and housing insecurity, that we know increases the risk for disease progression and hospitalization.”

All the company’s predictive models were built with health equity at the forefront during the design phase, helping eliminate biases when suggesting patients most at-risk. It also helps with more meaningful and effective interventions.

The team is looking for ways to add even more data points to the model, with the goal to include lab data, pharmacy data, and the updated data available through Interwell’s aligned physicians, 73 percent of whom use the company’s Acumen Epic Connect electronic health record. Acumen is the most-adopted nephrology-specific health record on the market.

Conclusion: The power of data

Population health initiatives to reduce hospitalizations work best by harnessing enormous amounts of data across different touch points. The resulting insights could not be determined simply by looking at patient charts or individual claims.

Interwell’s strength is in its unique ability to easily leverage data from participating Acumen providers, along with that of more than 2,600 dialysis centers and our payer partnerships. These relationships offer far greater access to accurate, timely data that is critical to informing predictive modeling than other value-based kidney care management companies can provide.

"Everything hinges on the ability to produce robust analytics that have the power to drive change," said Arone. "We have the capability to bring in more data sources than anyone else, from claims data to lab results to Acumen data. We are working to bring all that data together to communicate insights effectively, which can help guide physicians and our care team around the changes and interventions that will have the most valuable impact.”

Notes

1. 2023 Report (2019 Data). USRDS.
   https://usrds-adr.niddk.nih.gov/2023/end-stage-renal-disease/5-hospitalization. Note: we compare results to 2019 data as the latest 2020 and 2021 data was significantly impacted by the COVID-19 pandemic and would not provide an accurate baseline for comparison. This is specifically addressed in the report which suggests lower hospitalization rates were most likely due to hospital avoidance.
2. Strategies to Reduce Rehospitalization in Patients with CKD and Kidney Failure. Clinical Journey of the American Society of Nephrology. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7863646/
3. Dialysis Costs for a Health System Participating in Value-Based Care. American Journal of Managed Care.
   https://www.ajmc.com/view/dialysis-costs-for-a-health-system-participating-in-value-based-care