A calculator that predicts the risk of someone hurting or killing themselves could be the next tool to combat a surge in suicides in Minnesota.

Researchers with Bloomington-based HealthPartners Institute and two other U.S. groups developed the tool by crunching the medical records of 3 million patients and finding common traits among the 24,133 people in the group who attempted suicide and the 1,240 who died by suicide.

Feeding that information into a predictive calculator, the researchers found that patients with the highest scores had a 13 percent chance of attempting suicide within 90 days. Only 0.1 percent of patients with the lowest risk scores attempted suicide.

“We’re taking something that has historically been thought of as unpredictable, and we actually have found pretty good predictability,” said Dr. Rebecca Rossom, a psychiatrist with HealthPartners, which joined in the study with five West Coast medical centers aligned with Kaiser Permanente and the Henry Ford Health System in Detroit.

Results were published last month in the American Journal of Psychiatry.

HealthPartners leaders are discussing how to use the calculator in its health plan and clinics. An upgrade to its electronic medical record system will make it possible to use during routine patient visits, Rossom said. “We can apply it for a particular patient at a particular visit and have an estimate of their suicide risk at that moment in time.”

Suicide is a growing problem, punctuated by the high-profile deaths this month of celebrities Kate Spade and Anthony Bourdain. Death certificate data analyzed by the Star Tribune showed 788 people died by suicide in Minnesota in 2017, an increase from 728 suicides in 2016 and from 573 a decade earlier. (The Star Tribune numbers for earlier years differ slightly from official totals provided by state and federal agencies. Official numbers for 2017 have not yet been reported.)

Much of that increase is due to self-inflicted deaths involving people, mostly white men, who are 55 and older, the data show. The state’s per capita rate of suicides has increased steadily since 1999, and it even closed the gap on traditionally higher national rates.

‘We need to do something’

News of rising rates and celebrity deaths compound the problem and contribute to the “contagion” of despondent people attempting suicide, but the reality is that most people with known risks don’t kill themselves, said Melissa Heinen, a suicide epidemiologist for the Minnesota Department of Health.

“People find help and hope, and that’s a big piece of this,” Heinen said. “You can have a risk factor, and that doesn’t mean you’re going to end up with an attempt or death.”

Existing screening tools for depression include the nine-question PHQ-9 survey, which is commonly used by Minnesota doctors. While the questionnaire asks about thoughts of self-harm, a more precise means of measuring suicide risk is needed given the rising number of deaths, said Sue Abderholden, executive director of the Minnesota chapter of the National Alliance for Mental Illness.

“We need to do something,” she said.

Rossom’s research verified that a history of major depression was one predictor of suicide. The study didn’t find any surprising or new risk factors, but it found that combinations of well-known factors — such as depression, prior attempts at self-harm and alcohol abuse — were meaningful, she said. “It’s not like ‘Ooh, you like chocolate doughnuts, that says something about you.’ ”

The final PHQ-9 question asks whether people feel they would be better off dead or hurting themselves. Prior research found that people who felt that way every day had a 2.5 percent risk of attempting suicide within 90 days.

The new calculator proved better than the questionnaire at identifying high-risk patients, and it did so based on medical records rather than on patient surveys, Rossom said. That could prove valuable when considering the number of depressed patients who don’t show up for appointments to answer questions, she added.

More accurate and aggressive screening sounded like a good idea to Kara Nelsen, a Minneapolis veterinarian whose brother died of a suicide that nobody in her family had anticipated. Perhaps the right screening tool might have identified his risk in time, based on drug abuse and some personal losses that occurred before his death, but she noted that it makes a difference only if doctors are prepared to act on the results.

“It’s the acting upon it too that is important,” she said.

Heinen agreed. The state launched a Zero Suicide campaign with 16 health care organizations, specifically so people with identified risks receive both social and medical support. She hopes the new calculator could help identify more people needing support.

The state also is in its second year of collecting nonclinical data on suicides, checking to see if events such as foreclosures or drunken-driving arrests precede deaths.

“They’re not the cause,” Heinen said. “People lose their jobs every day. People go to court and they don’t end up dying by suicide. So it’s more than just that, but it does seem to be a pressure point.”

Rossom cautioned that screening tools cannot replace physician judgment. The calculator was based on data from people who sought mental health care from primary care doctors or specialists, for example, but one third of suicides involve people who don’t seek mental health care.

The calculator also identified the highest-risk patients, but they make up a minority of the population. The greatest number of suicides in the study group involved patients with more moderate risk scores.

Rossom said it will be up to clinics to decide which risk scores are meaningful and indicate that patients need interventions such as counseling or written safety plans.