Monday, August 26, 2019

In child welfare, it’s the scientists who want their peers to "think less"

Too much thinking stands in the way of an ever-larger child welfare surveillance state

In the climactic scene of Inherit the Wind a film loosely based on the Scopes Monkey Trial, the character based on Clarence Darrow – making the case for the right to teach evolution - is questioning the character based on William Jennings Bryan, who makes the case for creationism.

But the Darrow character argues that more than creationism vs. evolution is at stake.  The case really is about the right to think:

I love that scene.  I suspect a lot of my fellow liberals love it, too.  Science, after all, is all about thinking, and only fundamentalists could oppose that, right?  In fact, it’s preposterous to even imagine a field in which women and men of science – doctors, no less – would actually urge their fellow professionals to do less thinking – uh, right?  Right?

Well, not exactly.  Because in the upside-down inside-out world of child welfare, there are medical professionals who are doing exactly that – literally.

Andrew Brown, director of the Center for Families and Children at the Texas Public Policy Foundation, first brought this to light in an excellent op-ed column for The Hill called “The doctor will accuse you now.” He writes about how two doctors, Richard Klasco, a professor of emergency medicine, and Daniel Lindberg, a professor of pediatrics, are urging their colleagues to adopt an approach they themselves call “think less, screen more.”

The idea is this: If a child has certain injuries, sometimes even something as minor as a bruise, these injuries are more likely – or, a better way to put it – less unlikely to be caused by abuse than other injuries. 

Up to now, doctors have been told to consider these injuries in the context of things like family medical history and the plausibility of a parent’s explanation.  If there is still genuine concern that the injury might be a sign of another problem – be it abuse or a medical condition – they’re supposed to refer the child for additional medical tests, in particular x-rays – which then are reviewed by professionals who, again, are supposed to consider all possible explanations.

Getting rid of that pesky context

But under the think less-screen more approach, we get rid of all those other pesky considerations. 

● First, the proponents say, we have to vastly expand the definition of what kinds of injuries are deemed suspicious – or maybe even eliminate definitions altogether and deem any visit by any child to an emergency room suspicious.

● Then any such injury, or ER visit, regardless of family medical history or explanation, should automatically prompt, at a minimum, a demand for a “skeletal survey” – equivalent to 20 standard x-rays of the child.

As a family profiled by Brown in his op-ed put it in their own post, it’s like leaping from: “smoking causes lung cancer” to “anyone who has lung cancer is a smoker.” 

Actually, it’s worse. The treatment for lung cancer is likely to be the same regardless of the cause.  But if one leaps to the conclusion that any bruise of a certain nature is probably child abuse, the consequences for the child can be far worse.

Lindberg and Klasco summarize their case, and use the term “think less, screen more” in an op-ed for Time magazine.  But Lindberg goes into far more detail in an essay for the Journal of Pediatrics.  That essay is co-authored by Dr. Rachel Berger. She runs the “Child Advocacy Center” in the Department of Pediatrics at the University of Pittsburgh Medical Center Children’s Hospital.  She also is prone to minimize the enormous harm of foster care, as can be seen in a commentary she co-authored that is discussed toward the end of this previous NCCPR Blog post.

It’s not that Lindberg, Klasco and Berger are ill-motivated. On the contrary.  The nature of their work means they see the very worst that parents can do to their children (though the nature of their work also means they see it so often that it may distort their perception of how often it happens). Like all of us, they are genuinely horrified and want to do something about it.  What they don’t seem to see, however, are the horrifying consequences for children of false allegations and needless foster care.

So in their current article, Berger and Lindberg say referring a child for a skeletal survey, and sometimes even more tests, based on suspicion of child abuse should be routine whenever there is a so-called “sentinel injury,” a term typically applied to a narrow range of bruises and other injuries in infants.

That’s not because children with sentinel injuries probably were abused.  In fact, in any given year, of all Americans under age 18, fewer than two-tenths-of-one-percent were believed by child protective services workers to be victims of any form of physical abuse, from the most minor (some CPS workers consider a spanking to be abuse) to the most severe.  (Even if we assume that the “real number” is twice as high, which is unlikely,* that still means more than 99.5 percent of American children are not physically abused in any way in any given year.)  But, Berger and Lindberg argue, the percentage who show up with “sentinel injuries” and were abused isn’t as low.

Nevertheless, Berger and Lindberg call for vastly broadening the types of injuries that should be deemed “sentinel injuries” and raising the age for labeling them “sentinel injuries” to age 4. And maybe not just actual injuries. Berger and Lindberg say even a baby’s “fussiness” could be a sign of serious abuse.  

If the family doctor – or any other medical practitioner - sees a "sentinel injury," the authors argue, the doctor should, literally “think less, screen more” -- put the family under suspicion and send them to the hospital for a battery of tests on the child.  Those tests may be evaluated by doctors who are so-called “child abuse pediatricians.” Even when they don’t have that designation, the doctors looking at all those x-rays may be predisposed to find abuse by the very fact that child abuse is the suspicion that prompted the referral in the first place.

But even that isn’t enough for Berger and Lindberg.  They go on to cite, with approval, a mechanism Berger developed in which every child who is brought in to an emergency room is automatically “screened” for abuse and, presumably, further tests, via a checklist of “risk factors” – regardless of whether there are “sentinel” injuries. The results go into the child’s electronic health record. And here’s the great news, they write: Berger’s model is based on one already “validated” in the Netherlands.  But you have to follow the endnote to the actual Dutch study to learn that the checklist got it wrong more than 89 percent of the time.

But so what? say Berger and Lindberg. Lots of medical tests are performed routinely for screening and come up negative in most cases.  But none of those other tests can result in a child being confiscated on-the-spot and consigned to the chaos of foster care.  Berger and Lindberg's approach would place many more families under suspicion. That, in turn, would place their children at exactly that risk.

It gets even more absurd

But the absurdity is just beginning. 

Berger and Lindberg actually call for less thinking on the grounds that it will eliminate bias!  In other words, no need to worry about, say, being too suspicious of Black parents and needlessly demanding they get their kids x-rayed if you simply suspect everyone and demand that all parents whose children have certain symptoms subject those children to testing.

On the one hand, I suppose it’s progress that these doctors acknowledge that bias among their colleagues is so pervasive and so deep that the solution is to bar them from even thinking.  But even if one assumes vastly widening the net of a cruel, stressful intervention into children’s lives is the best way to eliminate bias, there’s still a problem – it doesn’t eliminate bias; it just kicks that particular can down the road.

X-rays don’t read themselves.  Conclusions are drawn by doctors.  In this scenario, the doctors already know that the referral is based on a suspicion of child abuse.  If you add to that the fact that the family walking through the door for those x-rays is a family of color you have just as much, if not more, potential for bias than existed before.

Bias may be even more likely if the person looking at those x-rays is a so-called “child abuse pediatrician” and/or part of a “child abuse team.” This is a new subspecialty in which people who already have the certification supervise the training of other doctors who want the same certification. So whatever biases might have been there when the subspecialty was created, in 2009, are likely to be perpetuated.

I am aware of no studies concerning who becomes a child abuse pediatrician and why.  I do know that I have never read a story in which a child abuse pediatrician says: “I kept finding families who were destroyed, and children who suffered enormously when those parents were wrongly accused of child abuse. So I vowed to learn everything I could and become a child abuse pediatrician so I could stop children from being hurt that way, and make sure we were focusing on children in real danger.”

One need only look at how readily the field overdiagnosed “shaken baby syndrome” to see the potential for bias. 

In her book, They Took the Kids Last Night, family defense attorney Diane Redleaf describes case after case of misdiagnosis of child abuse.  But, she notes, “Not one child abuse pediatrician I knew of ever admitted their opinions about abuse were ever mistaken.”   She calls the field “a specialty stacked in favor of finding child abuse.”

And it is not exactly reassuring that Berger, Lindberg and Klasco all repeatedly minimize the harm of a false accusation.  Lindberg and Klasco claim in their op-ed that the worst that will happen is that “some non-abused children will be screened, and some non-abusive parents will be offended.”


As Andrew Brown explains, the worst that can happen is what happened to the children of Rena and Chad Tyson – after the parents were told to get the child x-rayed and the x-rays found fractures:

Rather than attempting to find a medical explanation, doctors at the hospital contacted Child Protective Services. All three of the Tysons’ children were removed by the state and placed in a kinship foster placement. The family would be separated for five months while Rana, Chad, and their children’s pediatrician worked to figure out the cause of the fractures.

The real causes were a series of medical conditions.  But the children were trapped in foster care and the legal and medical bills bankrupted the family.

Oh, wait, that’s not even the worst.  The Tysons’ children were placed with relatives and the parents could see them every day.  Anyone care to bet how this “bias-free” system would have responded had they been poor and nonwhite?  Oh, wait again. We know exactly how it would respond – the same way it always responds - as described here.

There are many other cases, and they all, as the doctors might say, present with similar symptoms: doctors whose specialty is child abuse jump to the conclusion that the cause of the “suspicious” injury is child abuse. 

Adding even more bias

Berger and Lindberg support approaches that would make the process even more biased.  For page after page they go on and on about how decisions can be based on looking objectively at physical injuries.  But then they speak admiringly of a frightening surveillance-state process already in place in Britain thanks to the use of electronic health records.

Berger and Lindberg write:

[Child protective services] shares information with the National Health Service about children who are on a “Child Protection Plan.” If that child then receives medical care in any unscheduled care setting, such as an [emergency department] or urgent care center, the healthcare team is alerted and given access to the contact details for the CPS caseworkers and service providers. In addition, CPS is automatically notified that the child has been to the ED, and both parties can see details of the child’s previous 25 visits to EDs or urgent cares.

Here’s why that’s so scary.

1. The overwhelming majority of cases in which families are under some kind of child welfare agency supervision (our equivalent of a “Child Protection Plan”) don’t involve abuse at all – they involve neglect, where the determinations are most subjective of all.  Indeed, what CPS workers call neglect often is simply poverty.

2. The people most likely to have “receive[d] medical care in any unscheduled care setting” are, of course, poor people. 

So now, under this plan – which Berger and Lindberg seem to love – any poor parent “under supervision” because of “neglect” automatically is doubly suspect if s/he has to take a child to the ER.  And visiting the ER, for any reason, becomes part of the child welfare agency’s case file, ratcheting up suspicion of the family.

Berger and Lindberg also write approvingly of a system in New South Wales, Australia, in which a “Child-At-Risk” alert is in the electronic health record for any child whose parents were subject of any report alleging child abuse or neglect.  But in the United States, more than 80 percent of such reports are false reports.

Now throw in "predictive analytics"

As I read all these glowing accounts of building a bigger and bigger child welfare surveillance state, I thought to myself: Oh God, imagine what would happen in Pittsburgh, where Dr. Berger works, and where the child welfare agency already harvests vast troves of data about poor families Cambridge Analytica-style – that is, without their consent, and uses it against them in its “scarlet number” predictive analytics algorithm.

Sure enough, just a few paragraphs later, Berger and Lindberg start singing the praises of the algorithm.  They repeat the misleading claim that the Pittsburgh algorithm may have reduced racial bias; in fact any reduction was solely a result of screening in more white people.

The solution to bias is not to refer more and more families to “child abuse pediatricians” for less and less reason.  The whole biased process starts the moment the family doctor pulls the trigger and sends the family for additional “screening” when it isn’t necessary.  Berger and Lindberg's approach winds up making all doctors trigger-happy.

But wait, say Berger and Lindberg – at least under the “think less” approach, if the x-rays don’t show abuse, the family can be cleared – so, isn’t it worth getting those x-rays?

But that claim is debunked - by accident - in an editorial in the same issue of the Journal of Pediatrics as the Berger/Lindberg essay.  The editorial, written by Dr. Mary Clyde Pierce, a child abuse pediatrician, is not a critique – it’s highly-supportive of the “think less” approach. But, according to the editorial:

Importantly, the sentinel injury in and of itself may be enough to not only prompt a further workup for other abusive injuries (eg skeletal survey) but also to report the event to social services regardless of the results of the additional workup (eg, negative skeletal survey).

(Emphasis added. And by the way, did you notice how Clyde Pierce says other abusive injuries? In a classic example of the bias that pervades the field, the immediate assumption is that all sentinel injuries are abuse.)

But wait, there’s more:

“When the screening studies are negative, the assumption is sometimes erroneously made that “abuse is ruled out” or it is not abuse because the screening studies were negative. These screening studies … cannot “rule out abuse.”

In fact, Clyde Pierce declares, if an infant with a sentinel injury has no history of trauma and a negative skeletal scan, that actually might put the infant at even greater risk – by lulling professionals into a false sense of security that he wasn’t abused!

So you see, in the world of child abuse pediatrics not only are you guilty until proven innocent, there’s no way to prove your innocence!

The all-purpose fallback argument: Horror stories

In the end, Berger and Lindberg are forced to fall back on the argument that’s always used to get us to “think less” – horror stories.

So they tell us all these elements of the child welfare surveillance state are needed because of the “regular occurrence of fatalities in children previously reported and screened out by CPS…”  “Regular occurrence" is not, of course, a scientific term.  It can mean whatever you want it to mean.  But here’s what we do know: More than 7.5 million children become “known to the system” every year.  An enormously generous estimate of the proportion of those children who die is two-tenths-of-one-percent. 

Each of these deaths is the worst form of tragedy, and the only acceptable goal for such “occurrences” is zero.  But the idea that, in the course of sifting through reports on about 7.5 million children every year, agencies “regularly” screen out cases in which children later die is not borne out by those numbers.  On the contrary, it is graphically obvious that it is an extremely rare occurrence. Here’s the graphic:

For a much more detailed discussion of how figures about child abuse fatalities are regularly misused to get us to “think less” see this previous post.

But the best evidence that the surveillance state approach won’t work comes from the Berger-Lindberg article itself.

The modern process of constantly ratcheting up surveillance of families began more than half a century ago with the first wave of mandatory reporting laws requiring certain professionals, especially doctors, to report their slightest suspicion of abuse or neglect. The scope of the laws and the professions they cover have steadily increased since, despite the fact that there is no evidence these laws actually make children safer.

In 2009, the medical profession created the subspecialty of child abuse pediatrics.  Since then, Berger and Lindberg say, the number of such doctors has more than doubled and the scholarly literature has “increased dramatically.” 

This vast expansion of the child welfare surveillance state gotten to extreme that one study estimates that at some point during their childhoods one-third of American children – and more than half of African-American children will have to endure a child abuse investigation.

Yet Berger and Lindberg themselves admit it hasn’t worked: As examples, they cite studies showing no change in the proportion of cases of “abusive head trauma” (AHT) that are being overlooked. (AHT is the new term the field came up with after all that misdiagnosis of “shaken baby syndrome”) Then they add: “Perhaps most disturbing is that the number of deaths related to physical abuse has remained stable at approximately 600 annually.”

By their own admission, all that additional suffering inflicted on children by the child welfare surveillance state hasn’t done a damn thing to make children safer.  Yet their solution is to make the surveillance state vastly bigger and more intrusive.

Perhaps Berger and Lindberg would have seen the problem with this – if only they’d given it a little more thought.

*-While it is, of course, likely that some allegations of physical abuse are mistakenly labeled unfounded, the only study I know of to second-guess these decisions found that caseworkers are two to six times more likely to wrongly substantiate an allegation of child abuse or neglect than to wrongly label it unfounded