A third of Utah overdose deaths are actually suicides, researchers find
https://www.deseret.com/utah/2019/12/20/21027135/utah-overdose-suicide-opioid-deaths-research-johns-hopkins
SALT LAKE CITY — A recent study using artificial intelligence found that a third of overdose deaths in Utah are actually suicides that haven’t been tracked as such.
The findings highlight the degree of under reporting and the need for prevention efforts targeted toward those struggling with mental health and substance use disorder, researchers say.
“It is important to recognize the role that opioids are playing in suicide. Because they are increasingly prevalent and can be much more lethal on overdose than other drugs or medicines, we might compare them to firearms,” said Dr. Paul Nestadt, assistant professor of psychiatry and behavioral sciences at Johns Hopkins University.
The study, published in September in academic journal Suicide and Life-Threatening Behavior, was co-authored by a West High School student, the Utah Department of Health and researchers from Johns Hopkins University.
Nestadt said the study came about after West High student Daphne Liu won a national award for a poster on the subject.
Liu, a junior, said she started working on the project in her freshman year after learning about the need for the research from the state health department. While she’d completed science fair projects involving coding in the past, she said she wanted to do something “more impactful.”
Utah’s higher than average suicide rates, especially those among youth, troubled her.
“Especially because I know that impacts a lot of people my age,” Liu said.
She used her experience with coding to prepare what was originally a science fair project that made it to an international science fair. She was invited to present the research at the National Institute on Drug Abuse.
Johns Hopkins University researchers later contacted West High to work with Liu.
Liu, with help from the other researchers, built upon her original project to prepare it for publication and used data from the Centers for Disease Control and Prevention’s National Violent Death Reporting System, an anonymous database of information on violent deaths gathered from state and local sources.
Overdose suicides are often mis-classified as accidents or undetermined
according to the study. Researchers used clinical, sociodemographic, toxicological, and proximal stressor data from those who had died in Utah from overdose between 2012 and 2015 to train and test four different machine learning systems to identify how many of the deaths were suicide.
Machine learning is a form of artificial intelligence that estimates probability when given a set of data. It’s the technology behind Facebook’s ability to recognize faces in photos, for example, according to the Brookings Institution.
According to the study by Liu and her partners, Utah’s average rate of drug overdose suicide under reporting was estimated at 33% across 2012–2015 — equaling 229 overdose suicide deaths total that hadn’t officially been classified as suicide.
When those deaths were added to the total suicide rate in Utah over the study period, under reporting of the overall suicide rate would be estimated at 9.2%, researchers said.
All four machine learning models achieved overall accuracy of 92.3% or higher. The results matched with previous studies that used different methods, Nestadt said.
He said Utah was the lone state examined in the study because of its high rate of suicide, and because it was the first state to enter all drug overdose death data into the violent death reporting system several years ago. Other states are gradually following suit, he said.
Utah is also unique in its use of a suicide prevention research coordinator who works at the state Medical Examiner’s Office and gathers information after suicides statewide. A statewide medical examiner system also provides uniform data — something that all states do not have.
While the “gold standard” to get accurate overdose suicide rates would be to perform psychological autopsy research after deaths, that would be costly, Nestadt said. Machine learning, however, is inexpensive.
“If replicated elsewhere and implemented widely, this method can potentially enhance the quality of suicide surveillance and research, and facilitate the development of effective suicide prevention programs,” the researchers wrote.
Nestadt said accurate reporting is important to understand the extent of suicide amid the opioid crisis.
“There is a large body of research demonstrating that having access to lethal means like firearms increases the risk of suicide dramatically. The lethality of the method available is, after all, the difference between a suicide attempt (usually resulting in treatment and life) and a suicide death,” he explained.
“Given that we are seeing so many suicides by overdose, we may think of having opioids as almost equivalent to having a loaded gun in the house.”
Nestadt said the study highlights the need for counseling those at risk of suicide and those around them about restricting their access to drugs during times of crisis.
“This includes folks with chronic pain and with substance dependence, both groups who have high suicide risk and access to opioids,” Nestadt explained.
Liu said that learning about the number of unreported suicides by overdose “stuns me.” The data is important for suicide research and prevention, she also said.
Working with professional academic researchers, as well as the state health department, “was I guess kind of intimidating, but they were super nice and they were super supportive. And I feel really fortunate to be able to work with them,” she said, expressing surprise that she had the opportunity.
“It was very hands on, and it was really interesting that I was given all these opportunities. I feel so fortunate, and it’s really amazing. And I’m proud of my work,” Liu said.
Like this:
Like Loading...
Filed under: General Problems | 4 Comments »
folks who do not understand the industry. I know when I submit a bid to a PBM there is not an exorbitant profit built into the equation…in fact, after all of the rebates and discounts, the profit margin is unbelievably slim. I also hear how the gov. Does not negotiate prices, well here is what I do know, when I submit anything to the government I have to use a formula to get a GNUP and use that number. It is far from retail. Most of the drugs that are at issue are specialty drugs and can be more of a supply and demand play. I have actually had to accept a 64% rebate demand from a PBM, not to mention administration fees of 4.5% and an early pay discount of 2%. There was another fee as well ~3%. 
I had always heard that the PBM’s demanded rebates/kickbacks/discounts from the pharmas of up to 50% to have one of their drugs on the PBM’s formulary and being approved without going thru a PA.
