Sunday, December 20, 2020

RETREAT FROM REFORM: When Marc Cherna first took over child welfare in Pittsburgh it was a national disgrace. Nearly 25 years later, he leaves it – a national disgrace all over again

● “I’m glad I’m leaving this place in very good shape,” Cherna says. In fact, he’s leaving it almost as bad as he found it.  Pittsburgh tears children from families at a rate nearly as high as it was when Cherna arrived in 1996; a rate far higher than many other cities, including Philadelphia.

 ● There were some good years in between, and things aren’t AS bad as they were in 1996. But make no mistake; Cherna has failed – and not just because of predictive analytics.  That’s why Pittsburgh needs to find a new leader who will radically change direction – as Cherna did, before he retreated from reform. 

By Richard Wexler, NCCPR Executive Director 

A story in The Imprint about the decision by Marc Cherna, the longtime director of the Allegheny County (metropolitan Pittsburgh) Pa., Department of Human Services to retire begins with one of Cherna’s favorite lines: 

When Marc Cherna first came to work in Pittsburgh, Allegheny County’s child welfare system was floundering. Plagued by child deaths, burdensome caseloads, staff burnout and attrition and a lot of negative media coverage, it was, Cherna readily acknowledged, “a national disgrace.” 

The Pittsburgh Post-Gazette also used the quote, calling it “a phrase he’s used frequently.” Indeed he has,  But there’s more to it. 

Actually, the full quote is: “Allegheny County once was a pathetic national disgrace. Today, it is a shining national model.”  I know this because I’m the one who said it, in a CNN interview in 2002.  At the time it was true.  

But it hasn’t been true for awhile.  So let me update my quote: “Allegheny County once was a pathetic national disgrace. Then it became is a shining national model.  Now it’s a disgrace again.”  And almost all of it, good and bad, is because of Cherna. 

Marc Cherna

I’ve watched him in action from the beginning; in fact, slightly before the beginning.  In the
mid-1990s, I served on a search committee that unanimously, and with great enthusiasm, recommended Cherna and one other candidate for the job. 

For awhile that was something to brag about. Early on, Cherna significantly reduced the number of children in foster care and made more modest, but real, reductions in entries into care. He put housing counselors in child welfare offices and provided transportation to visits in a county where getting around without a car is notoriously difficult. He supported open court hearings and was a pioneer in emphasizing kinship care. 

But by 2010 the whole effort had stalled – even though it hadn’t gone nearly far enough.  Then things started going backwards. 

Yes, the number of children in foster care was cut in half.  But that’s because the original number was insanely high.  The reduction actually stopped dead in its tracks in 2011.  The most recent number published by the county, for January 1, 2018 – the number Cherna brags about – is, proportionately nearly, 50 percent higher than the state and national averages.  It’s even a little higher than Philadelphia, long excoriated, rightly, for being an outlier in holding children in foster care.  All of these comparisons factor in rates of child poverty.  Oh, and by September 30, 2018, the number of children trapped in foster care in Allegheny County had risen again

That isn’t even the worst of it. 

The point-in-time number is not the most important when determining if a community is ripping apart families too quickly. For that, you need to look at entries into care over the course of a year.  The picture there is even more dismal.  The reduction in entries during Cherna’s early years was more modest – about one-third.  Again, the county’s own data on first entries into care show the reduction in first entries ended in 2011 – in almost every year since they’ve gone up.  By 2017 almost all the gains had been erased – children were being torn from their parents at nearly the same rate as when Cherna started. (Look closely: Although line superimposed on the county’s own graph, below, suggests steady decline, the bars themselves show a pattern more like a repeating U. The line distorts the visual impression made by the actual bars in the bar graph.) 


These data measure first entries into care.  The county report doesn’t include early data for entries and re-entries combined. By that measure the county might look a little better, since, in recent years the county has reduced the proportion of re-entries.  Nevertheless, when comparing total entries into care – entries and re-entries combined -- to other cities, the rate at which children are taken from their parents in Pittsburgh is obscene. 

 Over the course of a year, Pittsburgh tears apart families at a rate 50 percent higher than Pennsylvania’s longtime exemplar of bad child welfare practice: Philadelphia. (See data tables for all Pennsylvania counties here.)  The rate of removal in Pittsburgh (total entries divided by impoverished child population) is more than triple the rate of New York City and Chicago.  In fact, were Pittsburgh big enough to be among the nation’s ten largest cities, its rate of removal – more than 32 total entries of children into foster care every thousand impoverished children - would be worse than at least nine of them.  Again, these figures factor in rates of child poverty.  

There is no evidence that Pittsburgh is a cesspool of depravity with vastly more child abuse than all these other cities.  There is no evidence from anywhere in America that taking away huge numbers of children makes them safer – and considerable evidence to the contrary

The kinship care excuse 

The response from Allegheny County to these figures is likely to rely heavily on the fact that Pittsburgh places a larger proportion of children in kinship foster care – relatives instead of strangers – than the other cities.  

That’s true – thanks largely to the pioneering work of Dr. Sharon McDaniel and her agency, A Second Chance, Inc.  That agency really is a shining national model. It may be the only genuinely positive legacy of Cherna’s tenure.  But here, too, there are limits. 

For starters, kinship foster care is still foster care.  Yes, it’s almost always the least harmful form of foster care, but a kinship placement still can be enormously disruptive and still do harm. 

But also, Pittsburgh is such an extreme outlier that even if you compared only the number of entries into care in which children are placed with strangers in each city, Pittsburgh still would come out badly.  Indeed, Pittsburgh’s seemingly outstanding record in using kinship care may be in part because Cherna’s agency places children with relatives who, in some other cities, would not have been taken from their homes at all. 

How it all plays out 

Case in point: A case portrayed in this story in the Post-Gazette as a success.  The story is all about how much a caseworker for a private agency, Three Rivers Youth, did to help this family – including yes placement with a nearby relative, after the children were taken, and how grateful the mother is for that assistance. 

But look at why the Allegheny County Department of Human Services took the children in the first place: 

There were several holes in Ms. Roberson’s cramped home in the Hill District, giving rodents and other pests unfettered access to rooms where she and her five children ate and slept. She said she was having trouble getting her landlord to repair them. 

Once Allegheny County’s Office of Children, Youth and Families caught wind of the disrepair, along with other issues such as some of her children’s truancy, [her children] were placed into foster care. 

This is a classic example of one the biggest single problems in child welfare – the confusion of poverty with neglect. (Where were the housing counselors?) All of the help from Three Rivers Youth could have been provided without resorting to taking away the children.  And it appears that housing still is the reason this family has not been reunited.  According to the story: 

With [the Three Rivers Youth caseworker’s] help, Ms. Roberson said she is focusing on goals to help get her life back on track — moving to a new, safer home in West Mifflin so she can get her children back, getting her GED and eventually finding a job and learning how to drive. [Emphasis added.] 

This is exactly the kind of misuse and overuse of foster care that some of us on that long-ago search committee hoped hiring Marc Cherna would prevent.  For a few years, it did; but no longer.  And the problems don’t stop there. 

● Cherna’s deputy and, presumably, a possible successor, Erin Dalton wrote an appalling apologia for foster care, dismissing the enormous, and enormously well-documented trauma inflicted on children when they are torn from their homes. 

● Cherna’s agency works hand-in-glove with the University of Pittsburgh Medical Center (UPMC) – where the extremism can be seen in their support for urging doctors to do less thinking before jerking their knees and reporting their slightest suspicion of child abuse.  UPMC and Cherna’s agency also are being sued for an alleged “plan or agreement” that amounts to harassment of pregnant women who so much as smoke marijuana. 

Allegheny DHS has not earned our trust 

All of this, of course, is before we even reach the issue for which Allegheny County child welfare now is best known: It’s embrace of the Allegheny Family Screening Tool, an Orwellian predictive analytics model.  It began with those subject to reports alleging child neglect.  Now, with some modificiations to the algorithm, Cherna’s agency is attempting to stamp an invisible “scarlet number” risk score on every child in the county at birth – a score that could haunt the child for a lifetime. 

I’ve written about how dangerous this is many times (see the links in the previous paragraph and our overview here). So now, I just want to focus on the fact that the way the Allegheny Department of Human Services has sought to market the scheme, in itself, illustrates why neither the plan nor the agency is worthy of our trust. 

● They said that unlike other algorithms that purport to predict who is likely to abuse a child, AFST is transparent; everyone can see what goes into creating the algorithm. But in fact: You can see the ingredients but the weight for each ingredient and even whether it counts in favor or against the accused remains secret.  That makes the ingredients list nearly useless. 

● They said they did an ethics review and got a glowing report.  Well, yes. That’s likely to happen when one of the reviewers is a faculty colleague of the co-designer of the algorithm. In fact, they co-authored papers together.  The review itself is startlingly superficial, citing only papers written by either the designers of AFST or the reviewer himself.  

Even then, the favorable verdict was premised on the idea that the algorithm would be used only in cases where there had been a report alleging child abuse or neglect, not on every child.  (See below about that.) 

● They said that the algorithm curbed racial bias.  But only because the algorithm led to investigating more white families, not sparing more Black families from the enormous trauma of needless investigations. 

● They brag about their algorithm predicting actual child abuse -- based on a study which found that it fails up to 99.8 percent of the time.  (The study was done by the designers of the algorithm itself – in cooperation with UPMC.) 

● When they went full Orwell and created the version of AFST in which the aim is to stamp the risk score on every child at birth in order to target prevention programs, they had a problem: The ethics review that supported AFST was premised in part on not doing just that.  So, they commissioned another ethics review – but once again, they made sure to choose reviewers who would tell them what they wanted to hear

● To counter the charge of “poverty profiling” they said that unlike the AFST algorithm the one they hope to use on every child at birth, called "Hello Baby" “only relies on data where the County has the potential to have records for every family." But the key weasel word there is “potential.”  Because right before making this claim, the county acknowledges that they probably will use “child protective services, homeless services and justice system data.” 

● They say Hello Baby is voluntary. But it’s voluntary only in the sense that you are automatically assumed to have agreed to surrender your data unless you are constantly on the alert for your one opportunity, in the first days of your new baby’s life to opt out.  

So ask yourself: If Hello Baby is so great, why does Allegheny County have to, in effect, sneak it past the very people its proponents say are most likely to benefit, instead of being open and aboveboard about exactly what it is, and letting people opt in if they really want it?  If it’s as wonderful as Marc Cherna says, people should be lining up to take part. 

Cherna also says Hello Baby is to be used strictly for targeting prevention; the data from one part of Cherna’s agency won’t be shared with that other part of Cherna’s agency that does the child abuse investigations.  But how long will that last?  What happens after the next high-profile child abuse tragedy? Also: The prevention workers who go out – knowing that the family is “high risk,” and therefore possibly predisposed to see neglect whether it’s there or not -- also are mandated reporters of “child maltreatment.” 

The Hello Baby data are kept out of the hands of child protective services only as long as Marc Cherna or his successor or his successor’s successor decide to keep it that way.  If they change their minds Allegheny County parents can’t change their minds and take back their data.  And the way Cherna and his colleagues have sold AFST and Hello Baby doesn’t inspire confidence. 

Next steps 

Cherna plans to retire on March 5.  But there’s no need to wait.  Allegheny County’s political leaders should do right now what they did in 1996: name a committee to conduct a nationwide search.  Perhaps they’ll find another leader like Marc Cherna – the Marc Cherna from 1996, not the one from 2020.