How Wildflower Finds and Supports High-Risk Pregnancies as a Critical Component of Value-Based Maternity Models

An important part of improving maternal health outcomes within a given population is being proactive and staying ahead of complications that can lead to worse, and more costly, outcomes for mom and baby. Traditional methods for stratifying risk in healthcare aren’t dynamic or comprehensive enough to ensure success. Wildflower's Chief Actuary Sara Teppema writes about how we approach risk stratification in maternity care, and what we’ve learned from our model.

Comprehensive

Many organizations talk about serving the “whole patient” but to do that, you must be able to identify clinical, behavioral and SDoH risk factors, all of which have a significant impact on a person's health. Legacy risk models often focus on specific clinical metrics but miss incredibly important and impactful pieces of the puzzle.?

For a robust risk score, you also must pull in a combination of self-reported data, as well as key data from electronic health records, to develop a deep understanding of every patient we’re supporting. This includes:

• Demographic Information – such as age, race and ethnicity. With this data, we can determine whether there are inherent risks with the pregnancy (advanced maternal age, teenage pregnancy or health equity concerns).
• Comorbid Conditions – to gauge whether there are existing diagnoses such as hypertension, diabetes or other health concerns that would contribute to a higher risk pregnancy
• Past Pregnancies – gathering history on the patient’s previous experiences, particularly complications, miscarriages, premature births, multiples or emergency C-sections.
• SDoH Needs – understanding whether there are food, transportation, housing or other social needs that would affect access to care and ability to follow care plans.
• Mental Health – screening for perinatal depression to monitor the patient’s mental wellbeing
• Patient Activation – measuring how engaged the patient is in their own healthcare to determine whether she will need more assistance navigating the system and managing their care.?
• At-Home Monitoring – for patients who are prescribed devices for remote patient monitoring, keeping a close eye on them between appointments.?

Continuous?

It’s not enough to screen patients early in pregnancy or upon some discrete event, such as downloading an app or completing a survey. You need to continuously monitor for risk throughout the pregnancy and dynamically respond to changes in their risk score based on new data.

Dynamic

The risk model has to be capable of integrating and weighting multiple inputs to smartly assign patients to the right level of risk. It also has to learn and refine the scoring algorithm over time as you glean new insights from the overall data.?

Connective?

Stratifying patients based on their risk profile is not very helpful unless you have the right resources at the ready and can match those resources effectively to help those in need. In addition to in-app prompts for individuals, Wildflower uses Health Advocates to proactively reach out to patients at high risk and specifically address their most pressing concerns. In some cases, this might be a referral to a provider or health plan program, a connection with a community resource, coaching sessions to address key issues or escalations to the patient’s OB.?

What We’ve Learned

As Wildflower has refined its risk score over time in support of value-based maternity care models, we’ve learned many things. Some of our key takeaways include:?

> You don’t need full EHR integration to effectively risk stratify a population of pregnant people. With limited key data from the EHR, coupled with self-reported data from the patient, we can very accurately predict which patients are most at risk for adverse outcomes during pregnancy.?

>> Single data inputs can be misleading when it comes to a person’s overall risk. For instance, it would be faulty to assume every patient who is remotely monitoring her blood pressure is in the high risk category. While this can be a good indicator for heightened risk, there are scenarios where this is more precautionary, and the patient is actually doing well overall. Without a comprehensive view of the patient, and a risk model that integrates multiple factors, you have no way to make this determination.?

>>> You can never stop learning. As mentioned above, you have to continue refining and adapting your scoring algorithm in order to make it more accurate, and more personalized for each patient. Every new piece of data offers an opportunity for the model to get smarter and better. Our Becker's Value-Based Maternity Care 2.0 webinar touches on the key learnings we've had with our New Jersey provider and health system partners.

>>>> Pregnant patients who are at high risk are very open to support from health advocates. ?We are able to successfully engage with upwards of 80% of patients who are categorized as high risk. This is a telling stat. Patients who are struggling during pregnancy want more support. They welcome it. Compare this level of engagement with other forms of advocacy or coaching, such as the corporate wellness industry. It’s much harder to convince someone to get help with lifestyle management goals, such as diet and exercise. Pregnancy is a different matter altogether. The patient, in most cases, is ready for support. You just have to find them and make the right connections to resources. This is especially true for rural populations, where both maternity care deserts and lower reimbursement rates can impact their level of care. We explore that in our Maternity Care Access for Rural Medicaid Beneficiaries report.

-- Sara Teppema , Chief Actuary at Wildflower Health

#maternalhealth #maternity #valuebasedmaternitycare #valuebasedcare #riskstratification #healthcare #digitalhealth #healthcareinnovation #pregnancy #healthequity #sdoh