Daniel Flaming on Cost Studies & Triage Tools

Total Time 19:13
0:36 – Give me some background on the cost-of- homelessness studies and triage tool.
2:50 – What is the value of linking data between public agencies?
3:35 – What specific data do you need?
5:13 – What are the challenges in getting this  data?
6:02 – What are the challenges in processing the data?
7:04 – What was the main conclusion of the cost study?
8:24 – What else did the study show?
9:02 – What are the benefits of a cost study?
10:38 – If we did a cost study in San Diego, wouldn’t we just get  the same results as Santa Clara County?
11:10 – Tell me about the Silicone Valley Triage Tool?
12:29 – Why is a record linkage so important?
14:08 – Give me an example of what can come from a better understanding of costs.
15:55 – Why do you think there is such momentum in Santa Clara County around solving homelessness?
17:47 – Why does LA not have as much momentum on solving homelessness?
18:40 – Aren’t both counties doing bond measures?

Video Transcript

President, Economic Roundtable
Interviewed on September 23, 2016

Tell me about your background and how you ended up in your current role?

I’m president of the Economic Roundtable.  I worked for Los Angles County for about 21 years.  I directed HUD housing programs and job training programs and created this research unit in the county.  We spun off from Los Angeles County in 1991.

We’re a public benefit research organization.  One of the areas of research that we focus on is homelessness and we’ve done cost studies in Los Angles County and Santa Clara County.

Our cost study in Los Angles began as a very modest study to look at a program that had given intensive help to about 50 people.  Sometimes it’s about the same amount of work to do a big study as a little study because the problem is getting ahold of the data.  So we expanded it very substantially and we looked at costs for a representative group of about nine thousand homeless people in Los Angeles, and another thousand who had been in permanent supportive housing.  So we had costs for folks when they were homeless and costs for them when they were housed.

We looked at the cost distribution amongst people who had experienced homelessness, we found this hockey stick pattern with most people not having very much cost, and then a small number, about 10%, accounting for most public costs, over half of public costs for homelessness.  There was a lot of interest in who are these people and how do you find them.  Then we developed a screening tool, a triage tool, for identifying these people.  We launched a pilot program to use the tools in hospitals and to house these very expensive people directly out of hospitals.

Santa Clara County was interested in bringing it to scale.  They are doing record linkages which gives you the data for finding these folks.  These are the most complex cost studies that have been done, the most complete.  Other regions like New York, Philadelphia, and Denver have done much simpler studies that don’t break out costs in as much detail.  The cost study in Los Angeles and Santa Clara County show costs by public agency, and cost savings after they are housed by public agency.

If a cost study has been done in Los Angeles also so that government officials know that it costs less to house these homeless people, why is a focus on housing the homeless not as much of a focus in LA as it is in Santa Clara County?  

One is I think there is a really strong set of champions of the cause in Santa Clara County, so there’s maybe more public unity around the issue.  Santa Clara County had also deferred acting on this problem for a very long time, so their inventory of permanent supportive housing is very small and it’s an extremely expensive area in which to find housing.  I think they felt is was a problem they couldn’t put off addressing any longer.

Los Angeles has been developing modest amounts of permanent supportive housing over a long period.  We’re still far, far short of what we need, but we had made a bit more progress than Santa Clara in building up a small inventory.  With there bond issue, Santa Clara is really getting their arms around this problem, so if that’s approved, that will be a very powerful force.

There is also a bond in the City of Los Angeles for building housing, which is just a little larger than Santa Clara’s, so for a much bigger city, a comparatively smaller bond, but there is one on the city ballot.

What does doing a cost study entail?  

To start with, linked records are extraordinarily powerful.  Homeless people tend to be invisible.  In census data, data is collected based on where we live, so if you don’t have a home, they really don’t collect data.  Linked records are the key tool that we use for screening, for identifying these high need, high cost people. When we link public records, we see a lot of dimensions of people’s lives over time.  We see all of their health system encounters, all of their medical diagnoses, their stints in jail, their use of public assistance, their episodes of homelessness.  We have information that lets us know both about people’s problems, their attributes, their age, their gender, as well as their system use, and that then we can use to target people for intervention.

But in order to do that, agencies have to give up their crown jewels which is their data.  They have to agree to share it.  So there’s a policy issue to get people to agree to do that.  There’s anxiety that it’s going to show failures, that they’re giving away something important.  So even within a government agency, like a county, getting people to agree to linkages is hard.

And then executing a good linkage is also hard.  For the 104 thousand people in Santa Clara County, we were working with 25 million linked records.  Different agencies use different kinds of identifiers, people show up with different variants of their name in an agency file.  So de-duplicating records and then to link them across agencies is a technical challenge.  It’s doable, but it’s not trivial.

You need champions at the top to get this to happen, and then you need a set of technical skill to do the linkages to do it well.  Once you do that, though, you have an extraordinarily powerful tool.  You can use it to describe need, to identify outcomes, and you can use it to develop tools for targeting interventions on high-need people.

Who are you working with at the various agencies?  And what agencies are you working with?  

One place to start might be the public social services system.  That is going to take in most of the population in poverty, and then you can potentially expend that list of people with your HMIS records from your CoC.  You want to start by identifying everyone who has been homeless, so you’ll find most of the population in those two agencies.  Also identify additional people out of different elements of the health services system.  It depends on record keeping in the county.

In Santa Clara, we started by creating a spine of everybody who had been homeless and then we attached records from every agency to that spine.  Once we had the spine, we wanted all of the health data.  We wanted the justice system data, so arrest data, jail data, what people were charged with, how long they were in jail, what kind of jail facility was it, their security level in jail.  We wanted their mental health data.  We wanted their substance abuse treatment data.  We wanted all the records for the CoC which were very valuable in showing homeless stints.  We wanted all of the social service data.

There’s getting agreements among the departments to the agencies to link the records.  The records we work with are de-identified, so you have a lot of personal data about people here.  So the handoff to use was data where people’s names, social security numbers, and so forth had been removed because we don’t need them.

Who was your main point of contact in Santa Clara County? 

In Santa Clara County, it was Ky Li (director of the Office of Supportive Housing) who was the staff person who was extremely competent and who spearheaded this.  He was infinitely patients and he was also someone who hung in there to get all of the collaboration that was needed.  He was terrific.  Also, Jennifer Loving from Destination: Home which is the umbrella agency for addressing homelessness.

[back after microphone went out]

We were talking about the kind of linked records we want.  We want really everything that we can get that tells how public agencies and service providers have touched the lives of these people and what they’ve learned about them.  We want health diagnoses.  We want all encounters with any kind of health care, whether it’s a hospital, a clinic, a psychiatric emergency service, emergency medical transportation.  We want justice system information, so time in jail, kind of incarceration facility, security classification, arrest data, charges.  We want the HMIS data about homeless stints.  We want the public social services data which tells us quite a bit.  Anything we can get about their housing is very valuable, particularly if we can get post homeless data for people who have been housed to compare costs.   So we really try to bring together all of the data from every agency that has touched these folks lives.

Please talk about your conclusions from the Santa Clara cost study?  

The Santa Clara data is really by far the best data about the costs of homelessness because it’s a much larger group of people.  It’s about 104 thousand versus about 9 thousand in the Los Angeles study.  What we’re doing is focusing on these most expensive, high need people, which aren’t the only people we should focus on, but in this particular piece of the study, that’s what we did.  What we found was that average costs were about $60 thousand a person when they are homeless, and they go up above a million dollars a person, so there are some people with very, very high costs.  When people are housed, if you take service costs of about $20 or so thousand a year and housing of about $17 thousand, you have savings of $19-$20 thousand per year per person.  So the solution is cheaper than the problem.  For these very high need people, it’s cheaper to house them then to leave them homeless.

About how many people are we talking about in Santa Clara County?  

On a given day, the homeless population there is about 6 thousand people.  Over the course of a year, the number gets bigger.  The number of very high need people out of the annual population would be about 2,800 which is about 10% of the annual population.

Could it be best to just focus on those 2,800 people?  

The problem of homeless isn’t static.  We have a pipelines of social failures that leads people into chronic homelessness, into increasing wreckage in their lives, and into this very high cost category.  So we need to do several things.  One is we need to close this pipeline.  We need to do a better job of intervening earlier before people are homeless.   And when they’re homeless, we need interventions earlier before people’s problems get so severe.

One of the very interesting things we saw in the Santa Clara data is that it’s longitudinal.  We had 6 years of data.  Most people make it out of homelessness pretty quickly under the own steam.  Over half make it out in less than half a year.  If we can increase those rates of early exits, then the pipeline towards chronic homelessness shrinks.  Things that we need earlier and that we need tools for targeting are re-employment, that’s the best way out of homelessness is a job.  Targeting people for temporary subsidized housing until they get on their feet.

So these high cost 10% are in a financial sense the lower hanging fruit because they’re so expensive and we save money when we house them, but we still spend quite a bit on that solution.  We also need to have other interventions earlier that we also can target well.  Figuring out who to target for the right intervention is a difficult problem because you can’t necessarily tell all these people apart when you meet them for the first time.

How is 10% of the population 2,800 people when the population is 6,000?

You’re thinking of the annual point in time count data of 6,000.  2,800 is probably about 10% of the people who experience homelessness over the course of a year.  Your chronically homeless people are there day in and day out.  There a fixture.  There are a lot of people who have short stints such as moms with kids and somebody helps them or the welfare system helps them.  The population that is homeless in a year is a lot bigger than the count on a given day.

Targeting this population is absolutely the best solution because these are the most visible and most distressed people out on the streets.  It does reach the people with the most acute needs and it also makes the community more livable by giving folks who were in distress on the sidewalk a place to live.

How would a city or county obtain the Silicone Valley Triage Tool?

The tool is available free of charge on our website, so the tool is out there for anybody to use.  It takes linked administrative data to drive it.  In particular it takes health and jail data for the past 2 years in order to be able to use the tool.

What are the benefits of a triage tool?  

A cost study has two benefits.  One is that it tells you the exact nature of the problem in your community.  Who are the people who are frequent users of public services?  What kind of services do they use?  What are their costs?  It also provides you with the information you need to develop targeting tools, and also to then apply those tools to be able to flag people so that when they come in to a public assistance office or a hospital emergency room, you can have a flag in their record that says they’re eligible for housing.  The triage tool is predicting who is going to have continuing high costs into the future.  A lot of times people have just onetime cost spikes, so the tool is trying to separate that from those who have continuing high levels of need.   The tool gives people a probability of zero to one, or zero percent to a hundred percent of being in this high cost group.  With the tool assigning these probabilities, you can prioritize people.  If you have 3,000 people who you’ve identified as homeless, but 200 housing units, you can give top priority for these housing units to these first 200.  It’s a tool both for targeting and prioritizing.

What are the benefits of the cost study?

We have a segment of folks who experience homelessness who, through a cost study, we determine have recurrent crisis.  They’re in emergency rooms a lot.  They’re in hospitals a lot.  They’re sometimes in jail and in mental health confinement which is very expensive.  We can see very pragmatically that the problem that we’re leaving unaddressed is costing us a great deal, and in fact the solution saves us money.  The cost study is very compelling for documenting that this problem exists, that it cost a great deal, and that we can actually make peoples lives better and save money.

This study will probably give us the same results as Santa Clara County.  So why should we do it in San Diego?  

I think it’s a strategic decision about how you drive change and the information you need in a community to drive change.  Yes, it’s very likely to produce a similar story to Los Angeles or Santa Clara County, but maybe people need to hear it about San Diego specifically and not about another place.

I think we did a cost study on 25 people in San Diego.  What are the advantages to doing a larger scale cost study?  

The population experiencing homelessness is diverse, so a cost study can show you the complete range of experiences of homelessness.  It can help you understand the dynamics of homelessness, the movement in and out of homelessness.  That’s very important to understand who’s stuck and also to understanding who gets out and how to get that to happen more often.  It’s based on an assessment of the environment. What does it take to drive change and to document the need for change in San Diego.

What other data does the cost study show and what are some of the practical applications of that data?

One of the things is patterns of service use.  Often multiple agencies are touching the lives of the same individual and they don’t know it, so mental health services, medical clinics, hospitals, homeless service providers, the public assistance system, the jail and sheriff system, the probation department, substance abuse treatment may all be touching a person’s life and they’re not coordinating the services.  It can be a tool for service integration and more effective service.  I can also show services people need and they’re not getting. For example substance abuse is prevalent.  Providing behavioral health services to help people reduce the level of harm they do to themselves through substance abuse can also be targeted through this kind of data.

It’s a lot of work for a County to get you this data.  How did Santa Clara County do it?

It took a lot of effort on the part of Santa Clara County.  Staff there spent quite a bit of time on it.  It does take a champion at the top to get that to happen.  It takes the county board of supervisors and the county executive office to make people drop something else they are doing and do this.  It does take staff time which cost money.  It also takes experts to analyze the data and to develop tool out of it.

Anything else?

The problems of poverty and the cost to the public as peoples levels of distress get more extreme, in a lot of communities, there really is a drive to change this.  To do that, you need to get better results.  You need more effective outcomes for the services provided.  That really means that you need to integrate services.  A record linkage is what underlies a cost study.  A record linkage is a tool for a county understand service delivery and to integrate it and coordinate it.

It’s an ongoing and useful tool for the county to understand who’s touching the lives of residents in the county, where services should be integrated, and where there are gaps, where services are needed and they’re not happening.

One of the things that’s happened in Los Angeles County is that the health care system has also become a housing provider.  They are using some of their dollars to lease housing and put high need homeless patients into it.  One of the things a study like this can support is thinking more broadly. What kind of solutions do we need to bring down total costs?  Housing becomes part of that solution.

A challenge with a study like this is making the dollars fungible, able to be used for another purpose.  So you identify the high need people, and you house them so they’re using the jail less, they’re using the hospital less, but how then do you use those savings to pay for housing.  One of the things that is starting to happen is that health care providers are starting to see is that housing is a solution for health.  Also, people coming out of jail with histories of substance abuse and self medication, with mental health problems, part of the reentry solution is for them to have housing.  Some of the dollars from the state for the movement of people out of the prisons to jails and out of jails into the community need to be spent on housing.  It’s a tool for thinking about solutions for broadly.

I’ve had conversation with San Diego based non-profit housing providers and may have had some conversations with the County about it.

I think one of the things that Santa Clara County has done very well is to create a sort of center of gravity for carrying about this issue.  Destination: Home pulls together foundations, the county, the housing authority, the non-profits providers.  They’ve become a powerful point of consensus and they represent a lot of leadership from the county.  Finding a way to create a powerful voice for people who care about homelessness and that change is possible.  That it’s not an insolvable problem.  It’s a problem we can make headway on.

Destination: Home is a small group of about 3 or 4 people.  

Yes, but her board is broad.  It’s about 20 people.  So the county’s chief operating officer is on it.  Foundation people are on it.  She has a very good board.  They’re board is the Continuum of Care board.

Here in Los Angeles our Continuum of Care is a weak organization.  It’s a joint powers of the city and the county, but on its own, it doesn’t have much clout.  It’s board is appointed by the mayors office at the city and the county board of supervisors.  They have a comparatively small budget.  The HUD money for homelessness is very, very small compared to public social services or health care.  It’s, here, politically not very strong.  For whatever reason, the group in Santa Clara has brought together the real powerful leaders in the community and it has a lot of clout.

Part of people not having clout is that they are fragmented.  Find the people who care about this who care about this who are in significant positions, city council, board of supervisors… It Santa Clara, the dynamic was that the top leadership had decided that they wanted a campaign and this was a tool that gave them ammunition for the campaign.  They wanted the ability to have a very strong argument.