The head of Rhode Island's largest homeless shelter argued in a Nov. 12, 2014 Providence Journal commentary that the costs of not finding someone a home can be very high.
Anne Nolan, president of Crossroads Rhode Island, reported that, "In Rhode Island, a recent study showed that among a group of people who had been homeless for a year or more, Medicaid costs averaged about $60,000 per person, which is far higher than the typical $18,096 per person per year for disabled adults on Medicaid and $9,240 per person per year for the average Medicaid recipient."
That's a huge difference. We were curious about whether those numbers were accurate and why the gap was so large.
Medicaid is a joint state-federal, government health-care program for the very poor. Nolan’s commentary referenced the 2013 "Rhode Island Annual Medicaid Expenditure Report," developed by the state's office of Health and Human Services. We located the report, but we couldn't find any numbers that matched Nolan’s. So we emailed her to make sure that was really her source.
She replied that the information actually came from a June 1, 2014, report by Providence College sociologist Eric Hirsch, in which he identified 5,986 Rhode Islanders who had spent at least one night in an emergency shelter over a 28-month period and discovered that 2,308 were covered by Medicaid. Their Medicaid expenses totaled $58 million.
Hirsch doesn't calculate the cost per person, so we did. It's $10,795 per person per year. That's pretty close to the average for all Medicaid recipients and a far cry from $60,000.
So where did Nolan get her $60,000 figure?
In his report, Hirsch wanted to make the point that those costs were not evenly distributed among the homeless. "In fact," he wrote, "the 67 individuals who were homeless for one year or more and had the highest Medicaid bills had total charges of $9,325,375." That's $139,185 per person for these 67 people" over 28 months.
"The corresponding annual charge was just under $60,000 per person," Hirsch wrote.
But we immediately noticed a problem -- $60,000 isn't the average for all long-term homeless people. It's only for the 67 people who racked up the highest Medicaid bills, a distinction Nolan didn't make in her commentary.
When we contacted Hirsch, he explained that, "We do want to be able to argue that it would be most cost effective to house the most vulnerable." In other words, finding housing for 'high cost' homeless are going to save the most money.
He said he used an arbitrary cutoff -- expenses of $47,000 or more -- to identify the high-cost users.
Because Nolan was talking about all long-term homeless on Medicaid -- not just the 67 with the highest bills -- we asked Hirsh for the information on the larger group.
He went back to his data and reported to us that there were 442 long-term homeless people who had, on average, $23,650 in Medicaid charges per person per year.
Although that's double the amount for the typical person who was homeless for at least one night, it's dramatically less than $60,000.
When we wrote back to Nolan asking if she was aware that the $60,000 figure only applied to homeless Medicaid patients with the biggest bills, she said she had misinterpreted the report.
Finally, for the record, Hirsch and Nolan erred when they said the average Medicaid cost is $18,096 per person per year for disabled adults. As Hirsch reported when we questioned the figure based on the state Medicaid report, the correct amount is $21,696.
Anne Nolan said Rhode Islanders who have been homeless for at least a year cost Medicaid an average of $60,000 per person per year.
But the $60,000 cost she cites applied only to the 67 homeless people with the highest medical bills, a major caveat she didn't mention in her commentary.
The annual dollar amount for all Medicaid recipients in the study who were homeless for at least a year was $23,650 -- a substantial cost to taxpayers, but far less than what Nolan indicated.
Because Nolan's statement makes it sound like it applies to all long-term homeless people on Medicaid when, in fact, it only focuses on the most-expensive subset, we rate it False.