How Data-Informed Risk Stratification Can Support Suicide Prevention

February 20, 2024 - As the mental health crisis in the United States persists, supporting patients and preventing adverse outcomes are top priorities for healthcare providers. However, the national behavioral health workforce shortage and access barriers to mental healthcare mean that many are unable to receive the care they need, a phenomenon with potentially life-threatening consequences.

Listen to the full podcast to hear more details. And don’t forget to subscribe on iTunes, Spotify, or Google Podcasts.

Statistics from the Centers for Disease Control and Prevention (CDC) indicate that suicide is one of the leading causes of death in the US, as the suicide death rate increased 37 percent between 2000 and 2018. This rate decreased roughly five percent between 2018 and 2020, but returned to pre-2018 levels again in 2021.

In response, mental health advocates have called on government and healthcare stakeholders to invest in improved behavioral healthcare and robust suicide prevention initiatives.

Health systems are uniquely positioned to help guide these efforts, as access to patient data has significant potential to help identify and support those at risk for suicide at the patient population level. But building and deploying this risk stratification infrastructure presents a daunting task for healthcare organizations.

Parkland Health & Hospital System (PHHS) and Parkland Center for Clinical Innovation (PCCI) in Dallas, Texas have successfully achieved this with their Universal Suicide Screening Program. Two experts from the initiative sat down to discuss the program’s hurdles, successes, and future plans in a recent episode of Healthcare Strategies.

TAKING A PROACTIVE APPROACH TO SUICIDE SCREENING

The Universal Suicide Screening program was launched in 2015 to better flag patients at risk of suicide who may have flown under the health system’s radar.

“[The program’s launch] came from our recognition that many individuals who eventually die by suicide have contact with our healthcare systems in the days, weeks, and months leading up to their deaths,” explained Kimberly Roaten, PhD, ABPP, professor in the Department of Psychiatry at University of Texas Southwestern Medical Center and associate chief quality and safety officer for Behavioral Health - Parkland Health. “We came to the recognition that it was a missed opportunity to identify risk and do something to prevent suicide attempts and suicide deaths.”

At the crux of this prevention strategy is proactive, universal screening. Since the program’s launch, all Parkland patients aged 10 or older are screened for suicide risk regardless of their presenting problems or reasons for the clinical encounter.

During the screening, a nurse asks patients a set of standardized, validated questions to assess their suicide risk. That information is then documented in the electronic health record (EHR), and appropriate clinical resources are provided to at-risk patients.

“If a patient indicates imminent risk factors for suicide, somebody we need to be concerned about right now, the nurse would be prompted to keep the patient in eye contact, make sure that they're safe, and immediately let the rest of the healthcare team know that they need to respond to safety issues and do a full suicide risk assessment,” Roaten noted.

However, in the event of an immediate risk situation, being able to efficiently deploy the necessary resources to help the patient can be a challenge.

“The last thing we want to do is violate anybody's rights,” she stated. “We want people to be able to make choices about their own healthcare, but we also want to provide them with access to the resources they need to address things like depression and suicide risk.”

The screening process helps achieve this balance. Once the patient’s responses to the screening questions are put into the EHR, the process is automated, providing recommendations for next actions to the care team to ensure that patients receive the most appropriate resources and interventions.

This can help alleviate some of the additional burdens clinicians face when health systems deploy new risk stratification and workflow tools.

PHHS also uses a tiered approach to its suicide prevention screening to ensure that it is effective without adding unnecessary workloads to care teams.

“As part of our clinical decision support, we created a risk stratification tiered approach,” Roaten indicated. “Patients who say ‘no’ to all of our screening questions don't get any further intervention. That patient is probably getting maybe two or three additional minutes of questions from nursing staff, and they don't require a higher level of care from anybody else.”

The approach allows clinicians to better serve patients in moderate- or high-risk categories who may require higher-cost resources like one-to-one observation or, in extremely rare cases, inpatient psychiatric hospitalization.

“These patients are people that we need to take care of and keep safe in this moment,” she said, noting that such patients typically undergo a full suicide risk assessment with a trained social worker, a psychologist, or a psychiatrist after being flagged by the screening.

Roaten emphasized that such an approach does not necessarily require significant investments in additional resources for health systems looking to improve their suicide risk assessment approach.

“[Health] systems who are thinking about implementing these sorts of tools can think creatively about who can provide these interventions, assessments, and resources because it doesn't have to be a physician. It doesn't even have to be a psychologist. It can be somebody else with good mental health training,” she explained. “There are a lot of different people who have the kind of training to provide that assessment and then provide the connection to the resources that the patient needs.”

THE ROLE OF DATA

By providing these screenings and interventions, PHHS and PCCI have collected a wealth of data to inform and improve their work in these areas over time.

Roaten noted that while many of the risk factors identified in Parkland’s patient population are similar to the established, general risk factors for suicide, she underscored that recognizing unique aspects of patients’ experiences can significantly improve screening and intervention efforts.

Many of Parkland Health’s patients are from historically marginalized communities and have either not had access to appropriate healthcare or have had traumatic experiences with healthcare in the past. This knowledge is particularly useful to help tailor interventions and advance health equity.

Behavioral health inequity is a major contributor to adverse outcomes like suicide, and using data to close these gaps is a significant piece of Parkland’s suicide prevention strategy.

“We’re hyper-focused on social determinants of health (SDOH), nonmedical drivers of health, [and] health-related social needs,” explained Jacqueline Naeem, MD, PCCI’s Senior Medical Director. “As part of the analysis that we're working with [PHHS] on, we have access to all that demographic data that can help provide some insights into how those screenings performed for different groups.”

Those insights are integral to the suicide screening program’s initiatives.

“The interventions we provide are only as good as our patient's ability to use them,” Roaten added. “It is not helpful or effective for us to check a list of things that we're supposed to do for suicide risk without actually checking to see if those things are accessible to a patient.”

“For example, if we recommend outpatient psychotherapy to treat depression, but the patient has no ability to pay or no transportation to get to that appointment, that is not going to be an effective evidence-based intervention for that patient,” she explained.

She emphasized that the relationship between PHHS and PCCI allows program leadership to explore the data, identify nuances in suicide risk and other patient factors, and improve care for at-risk patients.

LESSONS LEARNED

Roaten and Naeem underscored that building a program like this requires significant effort, but that sustaining the program is also a continuous learning experience.

Roaten highlighted that gaining buy-in from leadership is a major part of the program development experience. She noted that as hospital leadership turns over throughout the years, keeping them informed about the suicide prevention program is vital.

“They don't know what this program is, they don't understand the value, and somebody has to be prepared to sort of keep the momentum going and to keep the enthusiasm for the program going,” she noted.

Program deployment also provided some major lessons for PHHS and PCCI early on.

“One of the most supportive groups when we started [the program] was our obstetrics and gynecology colleagues,” Roaten stated. “But they were the most frustrated right off the bat, and we couldn't figure out what was going on.”

Nurses and patients were both initially unhappy with the screening’s implementation, and soon after receiving complaints, program stakeholders found out why: the suicide screening risk questions had inadvertently been placed in an inappropriate section of the intake process.

“For the patients and the nurses, it essentially looked like, ‘Tell me your name, tell me your date of birth, tell me when you had your last menstrual cycle, and do you have thoughts of killing yourself today?’” she explained.

Roaten indicated that learning these “in the moment” lessons has been invaluable in terms of gauging what is feasible, what screening approach is acceptable to patients, and what approach is acceptable for care teams providing the screening. After fixing the screening issues that the obstetrics and gynecology providers flagged, program leadership saw much more success.

However, identifying and addressing the issues necessary to sustain suicide prevention efforts long-term requires access to high-quality data and support from analytics experts. Naeem noted that to date, the suicide risk screening program has analyzed 7 million patient encounters, providing a wealth and depth of data that would be impossible to use without a strong analytics approach.

She further explained that by looking at the data, one can gain insights into patients’ stories, which are critical to improving outcomes.

“We're able to look through the electronic health record, [and] we're able to understand how many people scored positively on the assessment,” Naeem said. “Looking at their utilization, were they people that frequently come to the emergency room? Or where do they go?”

“The hard question to answer is the outcome,” she continued. “So, who are the people that went on to die by suicide? Without knowing that information, it's very hard to understand those people… We have to know the end to understand what happened before.”

By pulling data from multiple sources, including EHRs and state mortality data, program analysts can better understand a patient’s journey and what was happening in their lives before they died. Learning from that and sharing those learnings with others is a key piece of the program’s ongoing efforts.

“What we're really curious to understand is, [for high-risk patients], how were they presenting in the emergency room?” Naeem noted. “When they have had the assessment, what was the outcome of that?”

She indicated that the program is looking to pull in additional data to see whether Parkland patients are also interacting with other health systems as part of their care journeys, which could reveal valuable insights into suicide rates and other relevant trends that could be used to improve suicide prevention screening in the future.

While these efforts are large-scale and have taken years to build, Roaten emphasized that other healthcare organizations looking to create a suicide risk screening program can do so by creatively utilizing their existing resources, like social workers.

“Many of us work in systems that don't have abundant mental health resources, but that doesn't mean that we can't get patients connected with what they need,” she stated. “We just may need to think more creatively not only about our connections in the community that can be helpful – like crisis lines, lifelines, [and] peer support – but also, who's already operating in our system who could actually take part in this program?”

“Don't reinvent the wheel,” Roaten continued. “Find a system that's similar to yours that's doing things that you hope to be able to do, and ask lots of questions. You do not have to do this on your own, and you do not have to start from scratch.”