Responding to the COVID-19 pandemic at Kaiser Permanente and in Washington state

The challenge

When the first case of COVID-19 was confirmed in Washington State, life as we knew it changed. Our health system rapidly transitioned from in-person to virtual care but needed ways to understand the clinical implications of stay-at-home orders or who might be at high risk of negative health outcomes because they were not getting in-person care or were more vulnerable to COVID-19. In addition, our health system leaders needed a systematic way to remain informed about the rapidly emerging evidence on testing, treatment, and vaccine development. Even though the pandemic has ended, the need for evidence-based guidance on COVID-19 treatment remains essential to support informed decision-making and ensure that effective medications are prescribed to those who will benefit most.

The work

In the early days of the pandemic, the ACT Center moved quickly to support Kaiser Permanente Washington’s rapid response to COVID-19:

• We used predictive analytics to identify Medicare-age patients at highest risk for complications of COVID-19, so care teams could proactively reach out to these patients offering care and support.
   
• We helped Kaiser Permanente Washington’s Business Command Center to understand, improve, and evaluate virtual care models that quickly evolved from the system’s rapid COVID-19 response.
   
• We conducted rapid reviews of scientific evidence related to COVID-19 care during the first year of the pandemic. Our team systematically searched through peer-reviewed research articles and preprints to find the most notable studies related to COVID-19 treatments and vaccines. Senior researchers reviewed the evidence and shared summaries with health system leaders. Monthly briefings helped Kaiser Permanente leaders stay up to date on the latest evidence in the midst of constantly changing knowledge.
   
• We shared insights with Washington state’s COVID-19 Health System Response Management team in developing a database to track hospital and intensive care unit use statewide.
   
• We partnered with Kaiser Permanente Washington Community Health and various community partners to identify and share resources with care providers in the region’s federally qualified health centers, including in-clinic workflows for COVID-19 and collaboratives to plan for emerging social needs resulting from the pandemic.

After the height of the pandemic, we continued to use our advanced analytic capabilities to help guide informed treatment decisions for patients diagnosed with COVID-19. Partnering with Kaiser Permanente Washington’s Communicable Disease Ambulatory Therapy (CDAT) team, the ACT Center developed, validated, and implemented a risk model to predict the risk of hospitalization or death from COVID-19 among patients age 50 to 64.

What we learned

Leveraging our Learning Health System Program capabilities, we were able to support Kaiser Permanente Washington's early response to the pandemic — and to help guide ongoing treatment decisions with a validated COVID-19 risk model in use since August 2023 among newly diagnosed members age 50 to 64.

Impact

Our work enabled care teams to efficiently reach out to thousands of patients who were at high risk of negative health impacts during the pandemic. It also supported health system leaders in making evidence-informed decisions during a quickly evolving public health crisis and helped our safety net partners to put tested strategies and workflows in place.

More recently, our advanced analytics team has helped Kaiser Permanente Washington meet the unique challenges of caring for patients newly diagnosed with COVID-19. Many clinicians see patients with whom they don’t have an established relationship, so they are learning about the patient and assessing their risk, often in less than 15 minutes. The COVID-19 risk model we developed uses data from the patient’s history and the current visit to identify those who are at high risk of hospitalization or death. The model is responsive enough to produce a risk score on demand and to recalculate the score as needed if the clinician learns something about the patient that would affect their risk.

---