YouTube Transcript:
What Practitioners Need to Know to Help Maltreated Children and their Families

so I think we're going to get started I
want to welcome everybody to um our
webinar series reducing risk for child
Mal treatment and treating its
consequences and National Training
series and the application of causal
science um this is our second session uh
for those of you who came to our first
session um we're really excited that you
came back and for those of you who are
new to this um our first session is
available um online and we actually
after this webinar and we send out um
the video from this webinar we also
include the first one so you can get
caught up the first session really did
focus on kind of the basics of causal
data science um and how that applies to
um children um and this session is going
to really focus on what practitioners
need to know um to help maltreated
children and their families so we're
going to be a little bit uh closer to
like on the ground uh still pretty
science heavy but a little closer to on
the ground um so we're really excited to
have all of you here Glenn if you could
move to the next slide please that'd be
fantastic all right just a couple
housekeeping things before we dive into
the content today um the slides will be
available after the webinar and one
thing I will start with is we are doing
this as a as a zoom meeting so people
can ask questions live with their voices
uh but I will also highlight that
because it's a zoom meeting that means
you have the right the ability to take
yourself on and off mute so I would ask
everyone um to keep themselves on mute
uh when they're not talking um you can
have yourself on camera or off camera as
you would like uh but
um please keep yourself on mute uh so
after the webinar the slides will be
available we'll send out to everybody
who register for the webinar we'll send
out the slides and it is also being
recorded and we will send that out to
you as well um this slide is inaccurate
actually because it says it's a webinar
but it's a meeting so please keep
yourself on mute uh you can throw things
into the chat over the course of the
webinar if there's a question or comment
that you have during the webinar at any
point feel free to throw it into the
chat we're keeping track all of of all
of that in the background and so we'll
log all of those and bring those
questions up uh Dr saxs is going to have
a couple spots um where he's going to
ask for feedback in people's thoughts um
so at that point um when he opens it up
if you want to say something via chat
throw it in the chat if you want to say
something um by talking just raise your
hand uh which is down on the bottom you
can uh raise your hand just raise your
hand and uh via we you know via the the
webinar not via you know your own actual
hand um and we will call on you and you
can ask your question or make your
comment um I think with that that is the
housekeeping um if we want to go on the
next slide we're not going to have
everybody uh go around and introduce
themselves like we did last time but
this is the The Faculty that's
associated with this webinar series and
also with the champ Center and the champ
Network as well which Kate Sullivan
leads so I believe with that I'm going
to turn it over to Dr Glenn saxs who is
going to present
today well uh thanks Andy so um welcome
back everyone uh really excited you're
here and really excited to share with
you what we have to share with you today
and so as you know as we talked about
last time we're we're talking about
causality um causal science all of that
from the framework of what the ground
what what clinicians what practitioners
need to know to help people so this
session is going to be focused on
practitioners children families what's
needed because that builds the case for
causal day the science and for a sort
sort of science that is needed to add
value to the work so with that the goal
today um we'll review key Concepts from
last time so that we're all in the same
page if people have joined us and missed
last time it's good to be grounded in
the concepts and then for those who were
here last time this will be a good
review then I'm going to talk about how
we generally
practice and where
causality might fit into that
practice I'm going to then go into
diagnosis and how important it is and
problems with it in the curring
practice and then talk about
personalizing practice which is
absolutely
indispensable in cases of
multicausality as as I'll be talking
about so here are some
disclosures about
uh my support and here's a
review so remember last time we defined
child Mal treatment outcomes these are
deleterious events that undermine safety
development Health mental health
adaptive functioning that have to do
with maltreatment exposure or their
consequences so this is what we're
aiming to uh prevent and so the two c
categories
exposures consequences and you could see
examples of
that now I
defined what's in the middle of
everything in practice and it's really
two things when you're seeing someone a
kid family is she at risk for a bad
maltreatment outcome if so what can I do
to reduce the
risk one and two
and I defined one requires predictive
knowledge two is very different
requiring causal
knowledge and then I
defined and this is you know let's I'll
be grounded in basic definitions because
everything Builds on top of this if you
see two variables two factors A and B A
would be any risk or protective Factor B
could be any Mal treatment outcome let's
think of the difference between
predictive and causal knowledge for a
factor a predicts B if knowledge about
it informs knowledge about B and we see
statistical associations correlations
Etc and clinically it's the use of
assessment information to classify an
individual for their risk for treatment
outcome and that's very
different than causal knowledge where
Factory is causal of B if Factor the
value of B is dependent on the value of
a and this all has to do with
intervention so if B is dependent on a
then if I
intervene uh on a b can change however
if B is not dependent on a I could
intervene all I want and B will never
change and so we there's many many many
many ways that a can predict B without
being causal of it and that's an
essential thing to understand for this
uh training
series and you know I defined why this
is through the common cause principle
any Factor that's an effect of a cause
of another variable because are to
correlate with that variable confounded
by their common cause I talked about
lung cancer and yellow
fingers which correlate and the
correlation would be replicated in any
study and of course it there is a common
cause
smoking and this is built the this
triangle of these three variables is
critically critically important and when
we talk about mult
causality boy is of important and I
showed you an example of how this works
how confounds work in observational
data like is telling you about this stol
metaanalysis of uh child um physical
abuse and 29 factors in this
metaanalysis multiple studies and
showing how easy it is for these factors
to be confounded
and misguide our approach to
intervention and then I reviewed
problems with the state of our knowledge
and it boils down to a misalignment a
poor alignment between how we generate
knowledge and the needs for knowledge
and that
causal knowledge can really help in
distinguishing predictive from causal
and then I noted that
all of us here in the this training
series all the participants we have such
a diversity between practitioners
administrators scientists and everyone
is necessary so I'm I'm really looking
forward as we move ahead in this series
to get people's input to have people
roll up their sleeves and wrestle with
the ideas because everyone is important
to move us forward
and it's all about scaling the
translational wall we have our science
as we read about in our
literature then we have our practice and
what's needed and we have this wall and
that wall needs to be scaled and we we
really believe that predictive and
causal knowledge is indispensible for
this okay oh and the last thing because
this I'm going to be referring to again
and again in this series um I wanted to
Define effective care requires the
capacity to make accurate predictions of
Mal treatment outcomes
under intervention conditions and to
learn from the predictive results so we
gather our data we then decide what it
means we want to reduce risk or say this
person doesn't need intervention but
it's all about predicting outcomes under
intervention conditions so we we
intervene because we expect things will
get better it's a prediction then we
intervene if necessary then we look at
what happens we compare results about
what happens to what we
predicted and then we refine our
understanding and our approach to
intervention based on that especially if
we there were prediction errors so this
is going to be a central thing that we
return to again and again and
again and you know we have to get
everything
aligned so let's define how we
practice and I'm I understand of course
everyone is different and practices
differently so I'm making some general
comments to build to build the case for
how things could be adjusted differently
so a general principle We Gather
clinical assessment data to lessen the
risk for maltreatment outcomes for if we
decide there is a risk
intervention so we start with clinical
assessment we get to Mal treatment
outcomes this is a long way to get there
how do we get there
well we have to make clinical decisions
so how does this fit
together what to assess that's the
question what would be most informative
for our clinical decisions and I'm just
going to give some examples of what we
might assess but this becomes critically
important
because whatever we choose is our
material for making our decisions
do does she have symptoms that will help
us uh diagnose a mental
disorder do either of her parents have
symptoms for this
diagnosis has she been exposed to a
particular social factor or
trauma is she a member of a particular
demographic
group does her family have a current or
previous child welfare service plan does
does she carry an alil for a particular
Gene does she have an unconscious
conflict or a particular defensive style
does she have a particular brain Imaging
finding you could see this is all about
information and anything might be
relevant or irrelevant so we have to be
very careful in the limited time we have
for assessment what would be most
informative because this leads to our
clinical decision and then it's the time
we're seeing someone to make these
decisions and then what's going to
happen it's all about time and can we
intervene in
time and
again these definitions are about the
use of the information for certain
purposes and now let's think
what does our literature tell us about
what would be most important and how to
use
that so
let's talk about a kid and let's see how
this might
work here is Eddie I'm going to tell you
a little bit about him but first here is
you or someone like you um a
practitioner again has to make two
basic decisions is he at risk for a bad
maltreatment outcome if so what can I do
to reduce the risk everything you do if
you're working with kids and families is
reducible to that essentially if if if
we want to be in the position of
helping so Eddie is an 8-year-old boy
who has repetive episodes of of hid and
kicking biting other kids and adults
most recently he ran as his teacher
knocking her off uh her
stool he is a history of physical abuse
from his father and was brought to
assessment by his mother for concerns
about episodes of pushing teacher off
her stool kicking his soccer coach
biting another kid recess at
school so of course his maltreatment
outcome is aggressive behavior that's
our Target for intervent that's our
Target and we need to figure out how to
change
that
so be thinking now about well what
information would you gather to
help here's some further pieces of
information at times it's hard for Eddie
to sit still his teacher reports he has
to go to the bathroom too
often at times he reports he can't focus
his teacher reports he Daydreams a
lot his mother reports he has nightmares
of his father hitting
him his mother reports that she believes
his aggressive episodes are due the
rejection sensitivity because of how
much she observed his father rejecting
him he gets really anxious when he sees
someone that reminds him of his father
and sometimes he even feels uh the belt
landing on him that his father used to
strike him
with Eddie's father is in jail now for
assaulting other
people and his parents are separated but
before his father went to jail Eddie was
very anxious about visits with his
father and tried really hard to avoid
them so we have to think about this what
it means how we're going to help him and
you know I wish I could say this was the
most serious case I've ever seen but I'm
I'm sure people you know in in the the
room here um you know this if we're
working with maltreated kids and
families these are the kids and families
we're called to help um this is sadly
tragically not an atypical case and what
that means
is what would you do how would you how
do you think about it and maybe I'll ask
now for people either in the
chat or um or even to say
what what what does this make you think
of how would you begin to approach the
case what would you do with this
information any any
thoughts so folks can throw it either
directly into the chat the chat just
comes back to the moderators but we will
see it and read it out um or you can
raise your
hand and we will call on you
must be someone out there with a
thought anything's
relevant I'm
waiting all right we'll give people one
more a little bit more time if they want
to either raise their hand or uh throw
something into the chat about what they
would do if it's okay I am muted
couldn't find perfect go for iten um I
would probably look for and talk to Mom
and Eddie too teachers about pro-social
behaviors and attributes that we can
really highlight and celebrate um focus
a lot on the good to get some Rapport
building happened for him to feel
comfortable and safe and then probably
utilizing some child- centered play
therapy with some collaterals with the
parent so they can see opportunities for
them to inflect that like into the home
not every day no parent can do child
centered play therapy all the time time
um but just to kind of incorporate
pieces of that into the home so Eddie
has those safe spots to land you know
ever so often throughout the
week yeah
so part of it is engagement part of it
it's a really good point is collecting
more information that you might use
through play and and through through
more engagement to get other information
now how would it help you around are two
essential questions about getting
information to know his risk and how you
might help
him and you don't have to answer maybe
if you're thinking about an answer it
would be but everyone should be thinking
about the information I might want to
get how would it help us for those two
things and I I absolutely agree um
ultimately helping them feel safer with
you and also knowing more about the
positives is going to be eventually
essential uh for sure and we'll talk a
lot more about
that any other any other thoughts before
I move
on hello this is Quinta I couldn't find
where I could raise my hand I apologize
um thanks sir yeah uh I would want to
you know uh rule out
any maybe organic reasons um that could
explain the behavior any
neurological issues if possible um I'm
in Community Mental Health which is it's
not always possible to get a a medical
doctor or a neurologist to rule out
maybe he's had a concussion in the past
or a TBI from being beaten um so I would
I would do that um as part of the
data collection um and then I would get
a comprehensive history of his behavior
um just to sort of pinpoint if maybe
some of this uh behavior that we're
seeing existed in his formative
developmental years or if there is a a
point in time where it started becoming
more
prominent yeah th those are great
thoughts about get more information that
might be critical and I'm assuming if
you thought it was organic that that
might help you know a little more about
how that reduces risk if the organic
cause was remediable maybe it's a a
seizure too and or or or something like
that that you could intervene on um and
and also getting more detail about his
history so those are really good points
um I what about the information that we
have right now in front of us is any is
any of it relevant we think for how to
know how to help him reduce his risk for
aggression anything that jumps out at
anyone about any of the the the pieces
of data that I've I've offered so
far well I I this is anws I also don't
know how to raise my hand here I can't
see the button for that but it seems
like you mentioned that he seems to be
having like U intrusive thoughts about
his father hitting him past and so
um it um it seems like you know he's got
some I mean this seems like so obvious I
feel kind of ridiculous saying it but he
clearly has symptoms of postmatic stress
disorder and so I mean I would focus on
and he seems to be overwhelmed by those
things and so um you know I think about
self-regulation skills about how to
regulate some of those symptoms
um and like you know in the classroom I
think you know you you know like he he
um ped over his teacher and he kicked
the soccer coach I suspect that there
were you know cues maybe that reminded
him of his of his past experiences and
so I would want to do like a like a
analysis of what specifically happened
which I I
mean it seems like that's consistent
with your approach to to like really
like look yeah well what exactly
happened when that happened like could
you tell me more detail about that
particular incident and and um find out
well like did was there like a loud loud
noises or something that precipitated
those things and then and then CH try to
try to address the environment and then
so that that external environment and
then also build his regulation skills
like sometimes you know like when you
hear those things maybe you could you
know you know like go or his his the
adults in the environment can say you
know I forgot his name Eddie that's
Eddie like maybe you can go to a
fireplace that was created to you know
like that to to I mean you can't do that
overnight but like you know build those
things to deal with his you know his
traumatic symptoms I mean you may be
dissociating or something as well
something like that and and you we'll
take time so my thoughts those are those
are really good points you're you're
using the clinical data here and
thinking you know this seems consistent
with the diagnosis of post-traumatic
stress disorder you know that there's if
that's the case then there's probably
triggers for it that you'd look for
you'd gather more data and and think
that it's the PTSD that's caus you know
is related to the the aggression so
again this is the exercise is about
using clinical data and seeing what it
might tell you about what you should do
so let's um keep going with that and and
again our cycle here we're Gathering
clinical data and then using it to say
okay maybe I'll do this with it and then
the aggressive episodes will presumably
get
better
so I I'll just
um start here by organizing the clinical
data and people have made comments
they're consistent and one of the things
is I totally agree there's more data to
get and actually we're going to carry
Eddie into the next session too where
I'll be talking a lot about other data
but let's talk about the data we have
right
now
okay
so there's nightmares in attention
rejection sensitivity
impulsivity anxiety hyperactivity
avoidance intrusive memory okay that's
what we have so far more data to gather
but you know might this inform us about
what to do might any of it be
causal because then we could use it to
reduce the risk of um
aggression and what about our um well
how does our science help us organize
this so we might see that some of these
fit
together under a diagnosis of PTSD as as
was just stated
before we also might see that other
symptoms fit together the Triad of
inattention impulsivity and
hyperactivity it's part of a diagnosis
of
ADHD and then you know this clinician
you know where this is a story about
her she might go to the literature and
remember she read something you know the
Ripple effects of trauma valuing
vulnerability PTSD symptoms and
aggression with the Child and Adolescent
population or she could look at ADHD and
aggression impulsive aggression it's
comorbid for adhc and children
adolescence
so
great I think often in the world we try
and arrive at the diagnosis and and then
that just points us to the intervention
so maybe she has it or you know all of
that so how does this help us CU this is
often how practice
works so here's what she
did based on what she understood as I've
just talked
about she has could organize the data
into two diagnoses both are related to
aggression we also don't have to think
too much but we'll presume there's some
processes that relate to
this yeah please mute yourselves um and
then just to remind you about this
because it's so important effective care
requires capacity to make accurate
predictions of outcomes under
intervention conditions and learn from
predictive result
so she thinks of the interventions
predicting things will get better when
she delivers them or refers for them so
she'll use tfcbt for
PTSD she'll she's a psychiatrist she'll
deliver methyl venid Rin or conera or
she'll refer to this and she'll expect
things will get better our cycle here
right there deliver the
intervention predict next time you
see Eddie things will be a little
better so what
happened he's seen two weeks after his
assessment and he was perhaps a bit less
hyperactive and
inattentive but he had two significant
aggressive episodes
one in the school yard with those same
kids he was as the mother said doing
typical boy stuff yelling in
schoolyard then Eddie kicked another kit
with the
provocation the school monitor reported
another episode at lunch anded he was
slow the clean up um his tray um he felt
rushed by the lentry Monitor and then he
angrily stood up pushed the tray at her
and Eddie's drink fell over staying in
her clothes yeah how do we think Eddie
is welcomed at school around all these
type of episodes kids like this tend not
to be perceived very well in school and
that that's a real problem the other
thing is anxiety and Nightmares are
unchanged so you know you may think see
this say um you know I want to give it
more time
um maybe it didn't you know I'm not
going to change anything maybe it'll
take another few weeks for it to work or
you may want to look and see whe whether
you missed something that's all part of
it people you know you know some great
comments said should we look for organic
causes should we take more detail about
what happened should we uh look for
other diagnosis should we you know
collect more information should we do
play to get more information but part of
it again is when things don't go as
predicted you think of what you missed
and and that's just a really important
thing here and again that fits here okay
you've intervened measuring the outcomes
is this is what I'm telling you what the
outcomes are then you compare results to
your
expectations and then you decide whether
you should be refining your
understanding or whether you shouldn't
have expected change to occur in this
period of
time
okay so what would you do and I'll keep
moving but be thinking about that again
given this because I want everyone to be
in the space of
after you
assess predict do something an
intervention to be always ready to see
what you may have missed and all of that
and it's also how does our science help
us to predict well because that's really
our need to be grounded clinically in
what's what science how how it might
guide us so with that I want to
introduce the concept of the process
that generated the
data another word for this is the truth
you know there is a process out there
whatever it is that generated the
clinical data or the research data what
exactly is it because we want to align
as
closely as we can to it so that we could
predict well under intervention
conditions so what was the process that
generated the clinical
data here it is just take my word for it
again I this is not a real case but
there is a process and I want to say
well what would this mean if this were
true and it's plausibly true of course
so in this case in the brain there's a
survival circuit let's say it's led by
the amydala the hippocampus involved the
cortex in some ways you know I could
make it up but there's a lot of
literature about it and this is set up
by the experiences of physical
abuse uh maybe certain kids are
vulnerable through genetic
predispositions and then as was
mentioned in one of the comments well
what about triggers so a social fact
fact or a traumatic reminder clinically
relevant that leads to intrusive
memories which spill into impulsivity
and then aggression that's the path
what's also important is that there are
other things here in the clinical data
that if you were to intervene on them
like nightmares hyperactivity avoidance
anxiety and attention they're not going
to change aggression and that's very
very important as we move into what
causality really is it's so important
clinically for this reason what can you
intervene on that will make a difference
and what could you intervene on that
will not change at
all so these are intervention
targets um physical abuse and Gene isn't
changeable so we need to be thinking
about this
but then we need to be thinking
about there are processes that confound
our ability to see what's happening that
obscures our view because nightmares
hyperactivity avoidance anxiety and
attention correlate heavily with
aggression Eddie reported this if we
look at a research article we'll see
these
correlations um but they're not relevant
for changing aggression they're relevant
for predicting aggression but not
changing aggression that was the
difference between predictive knowledge
and causal knowledge why is that again
for the same reason because if you think
of C as the survival circuit and it's
lined to be aggression goes through
intrusive memory and
impulsivity a is let's say think of a as
nightmares just think of this
triangle that's going to correlate but
it will be confounded as I've said
before same thing for all of
these see this process that generated
the data of the truth if you just look
at it you cannot intervene on each of
these things to change
aggression so
it's only on anything in green if it's
changeable any that would possibly
improve aggression and that leads to a
rule I want you to burn into your brain
forever if it's green you can
intervene if it's red get that out of
your head
everyone with me if it's green you can
intervene if it's red get that out of
your head and again with
um this the one in two here R is
predictive you know you you could know
about risk from Red it's not
unimportant but it completely misguides
too about intervention to change
anything and that becomes so
important and
so let's move
into how what happens with
diagnosis because it has to do with
clustering and comorbidity so if you
think about ab and c and the triangle
there inter clusters of
intercorrelations
are to be expected and will happen if
this is necessarily
true
so this will all of this will correlate
intercorrelate with each
other similarly hyperactivity
impulsivity and inattention will
correlate with each other which will of
course
justify our
diagnosis but also explains why our
diagnosis are way less clinically
relevant than we might think
because if you look at what's in red
it's mixed you might say okay it maybe
it's not ADHD it's really PTSD and the
ADHD symptoms come from that so
intrusive memories lead to avoidance
they lead to hyperactivity because
you're he's really active when he's
having intrusive memories Etc you know
that that's how it works but again in
red it's both ADD symptoms and PTSD
symptoms and
similarly in
green one of them is impulsivity the
other is intrusive memory so if we use
our
diagnosis
to totally guide our treatment we're
going to miss big things we're not going
to be able to if we say is
PTSD uh targeting avoidance isn't going
to help even
anxiety it's only intrusive memory that
goes into
impulsivity again if this model is the
process that generated the
data so again we think of our clinical
utility
here so now let's um dive into diagnosis
a little um more um
clearly and let me just ask if if anyone
has any questions right now about
anything I said because I I I I want to
make sure that it's um it's clear for
people and Jenna put this in the chat
but you can raise your hand by um
clicking on react and then one of the
options is raise hand or if you just
want to unmute or throw any questions
into the chat by all means feel free to
do so and if we get questions we will
answer them and if not we will uh we'll
keep
going I just want to make sure that it's
really clear because we were building on
this every
session I have more of a comment than a
a question um and I put it in the chat
in the sense that when we look at
um uh what is considered a mental
disorder if we're going by diagnosis
we're essentially looking at being
symptom free so we're looking at all
those criteria and we're going do they
have that do they have that do they is
there evidence of that um however what I
worry about is being symptom free
doesn't necessarily mean mentally
healthy um so what I'm wondering about
is that I think we're missing a
causal framework of what mental health
really
is um because only if we if we
understand what causes mental health
Then I then we have a better way of
saying all right this variable and that
variable may need to be uh targeted
because it
explains the aggression and so that's
what I'm struggling with yeah so it's a
really good point so I'd say um two
things about that in in a way you could
substitute aggression for anything else
you you might say however you might
Define mentally healthy for Eddie you
could put that in here and then you
would you would have the same problem
still because then you would need to
collect whatever information you think
would be most relevant for how you would
improve his mental health
however you might Define it and I agree
with you for sure and in some of our
studies we're we're doing exactly what
you're saying because I think the focus
on pathology is is a limiting factor no
doubt about it um one reason I put you
know aggression here is because you know
this is you know Eddie's undoing it's
really affecting his health his mental
health is in School how socially ease
regarded obviously we we should be
thinking as as we think of interventions
what we may want to put what positive
things might we collect and fit in here
again um in next in in the next session
I'll be talking a lot about what might
be missing and how that might really
help you so I'm building a case for some
of what you're saying for sure um but
again I want everyone to be thinking
that whatever the outcome
is it's it's the same problem everyone
what's the information that allows you
to um know what to do to make it
better and and and it's all about
causality in that
case uh
does does that help for what you were
saying um it it it does um I think it's
just a foundational
issue with mental health in general um
the framework that we're operating on
doesn't really have a causal definition
of mental health so um that leaves us
with the problem that you're identifying
cuz my other my other thought was if
aggression was considered socially
appropriate this wouldn't be an issue
then because we can say oh aggression is
not a maladaptive maladaptive Behavior
because you know in society we we we
think aggression is
fine right so um that's where I I'm um
but I hear what you're saying though
yeah a little no I get the social
construction and I also get the the
Paradigm focused on pathology misses big
things and and in a way you
know I'm looking for a different
Paradigm as well and then hopefully we
can all engage in thinking about it but
I I'll also say and and I hope this will
become clear as we I lay this out
whatever Paradigm we land in causality
is going to need to be in the middle of
it and we could talk more about it but
but I I think that's that's very true
agree yeah thank you
sure okay so and in fact that's a nice
uh leadup to um some of the problems
with our Paradigm having to do with
diagnosis I've I've just shown you how
diagnosis can be misleading because it
misses the essential things and it
misses like it's based on the clustering
of symptoms which may or may not be able
to be targets for
intervention so here's a definition to
use for a
diagnosis the classification of patients
clients
people into groups defined by
pathological processes again this is
pathology but it's it's there in in at
least in medicine there's there's a
problem causing a
problem and this contributes to
particular health risks to indicate
interventions to lessen those risks so
many of us are in the mental health
world and in
maltreatment consequences we're thinking
of exposure to Mal treatment and mental
health mental disorders consequential of
that so that's really important because
how does that fit with this
definition big problems and so I want to
lead into that discussion by talking
about diagnosis in medicine and um and
and how this works and why it works and
why it's so important to help people um
you know improve their health so I'll
just tell you 58- year old man presents
to an emergency room with exertional
chess pain nausea and shortness of
breath these are symptoms an EKG and
blood worker order revealing ST segment
elevation for people not in the medical
world that's uh an important sign on the
EKG about a heart attack and similarly
elevated blood tro levels that's what
doctors look for and so a diagnosis of a
heart attack or what's called acute
myocardial infarction is made so we have
outcomes of heart attacks like
congestive heart
failure uh arhythmia dangerous
heartbeats uh pre premature death all of
that and we have these symptoms and
signs that fit together together but
those symptoms and signs are from death
to a heart muscle from lack of oxygen
from blockage of
arteries that's why these symptoms are
there none of them cause any of these
outcomes if you intervened on shortness
of breath you know it it's uh not going
to help in less get sometimes getting
more oxygen or nausea or or you know EKG
you know all of this isn't going to help
you if you g for chest pain like a
painkiller it's not going to
help because the these are the causal
Pathways so
again
red green if it's green you can
intervene if it's intervenable if it's
red it's not going to change things and
again R and
green and this is all based on
confounded
associations now that doesn't mean that
what's in red isn't
important it has a role in in the way we
work with people obviously because this
tells you this predicts the outcome it
it helps you know that myocardial cell
death is
going on but you can't intervene that's
again the same difference between
prediction and
causation and here's another case just
to make the point an 8-year-old
girl goes to her
pediatrician she complains of
fatigue she complains of um thirst and
frequent urin
ation uh including bed wedding that just
started blood test elevated um blood and
urinary test uh Lev blood urinary
glucose and a blood test for uh high
glucose and the diagnosis of insulin
dependent diabetes melodus is
made so again diabetes
causes vast peripheral vascular disease
diabetic retinopathy that can cause you
know blindness all of that this is
diabetes but we also see these symptoms
and signs are not causally related and
diabetes is related to the autoimmune
death killing cells in the pancreas that
make insulin that metabolize glucose so
insulin you you give insulin because of
this and it prevents the if the
downstream outcomes of of
diabetes it's the same thing you can't
intervene on any of these
things here's where you might
intervene you certainly with insulin so
again making the same
point and so we could look at the
problem with our current diagnostic
system fermental disorders what are the
problems it's
non-causal it's by definite in 1980 with
dsm3 um Psychiatry decided it shouldn't
be causal because we didn't have enough
causal information well it's uh what 45
years later it's still
non-causal which is a massive problem
it's why you know as we talk later in um
sessions we'll talk about how to build
the causal mental health through
research that is able to discover causes
with the type of data that we have but
what happens in our practice is we will
typically conflate three distinct
Concepts that should be considered
separately diagnosis is a cause so you
might ask the question or Eddie's mother
may ask the question to you if you were
the
clinician why is Edie so aggressive and
your answer could be well if you're were
using diagnosis in the way that DSM is
organized well he's aggressive because
he has PTSD and ADHD
I've talked about the problem diagnosis
is outcome what's your goal for Ed's
treatment well I think this is part of
one of the comments which is absolutely
true well if our treatment works he'll
be symptom free he'll no longer have
PTSD and add which we assume will help
reduce his aggression because we make
lots of assumptions about correlates
between diagnosis and lots and lots of
outcomes
and this is how
diagnosis developed in medicine in the
history of medicine a means of patient
classification which is
different the question how can I use the
clinical data to classify him into a
category of risk for aggressive behavior
indicating how I could reduce the risk
and so to do that Eddie has to be
classified
by
causes now it gets more
difficult with multicausality which
which really defines also why we need to
personalize
things
so I want you to think about any of the
maltreatment outcomes I've talked about
and um any that you you know you you
know a lot
about so just think about the factors
that you believe might cause it think of
any any sort of factor developmental
social different sorts of TR whatever it
is for any of them and think
about how many factors I think we we
could do a PLL now
even um let's let's do the poll J that
can we do it yeah why don't
people just you know give your best
guess of any of
the uh Mal treatment outcomes you know a
lot about
um
okay yeah just whatever it's just don't
sweat it
some more people just put your best
guess this will be
intered okay you have 15 more seconds
and then we'll close the
poll this is good so we're seeing you
know
multicausality most of you and and a lot
of people it's like 11 to 20 or over 20
even
yeah some four the 10 so a lot of
factors here okay I'm going
to end the poll
now
okay okay so we could see this okay
let's put our poll away
so yeah we are in a multicausal world
and many of you think many of you pick D
or C you know vastly
multicausal so let's let's stick with
that and I'm going to talk about what
this
means and I haven't even talked about
the interaction between causes so the
greater the number of causes the greater
the risk of um false conclusions from
research you might read
about and I've already defined this so
let's talk about the Peril and
promise of malte causality I'm going to
start with the
Peril remember the common cause
principle any Factor that's an effect of
the cause cause of another variable will
be observed to correlate with that
variable confounded by the common cause
we remember this so we could
substitute you know heart diseas you
know it's not just um lung cancer but I
talked about heart attack yellow fingers
through smoking causal will
be have a confounded correlation with it
we can also think about you know our
story here urine gluc blood glucose is
causal but blood glucose causes urinary
glucose which is
correlated we also talked about the
survival circuit
nightmares it's not going to change
aggression that's
confounded
and this because we are in a multicausal
world almost certainly you all believe
that or almost all of you it's this this
is our Bermuda Triangle where we could
get lost and never found again so and
I'll tell you why that's
true so here's our Bermuda
Triangle let's here's our Mal treatment
outcome so any effect of this cause
maybe it's a survival
circuit we've already talked about that
any of these CA any other causes
EF any of their
effects so if we're reading you know
correlational studies in our
observational
literature any of them and then any of
the uh Upstream causes and you know all
of
that and
so all of
these are
confounded and then even something not
measured in any study it's still part of
reality part of the process that
generated the data so any of their
effects will also be confounded it's a
massive problem I know I haven't talked
much to the researchers in the room and
I know there's several of you but so
let's let's then talk about the
implications of this for a research
study you might conduct let's say
just part of the process that generated
the data so let's just clean it up a
little bit again we know what red and
green mean
right
so I'm not going to repeat it but burn
it into your brain so just think about
it if you discover that factor H has a
highly significant correlation with
Factor B how would you determine that
factor Factor e confounded it because
you would need to know that or else you
would simply you'd write a paper talking
about H's uh intervention
implications which happens so often or
if you discovered that uh group
stratified by Factor V yielded
significantly different mean scores on
Factor
B how would you determine or rule out
that factor P can founded it making V
predictive not causal and meaning
nothing about
intervention this is the hardest one if
you discover that factor R had a
significant correlation with Factor
B how would you
determine that it was caused by an
unmeasured common common cause here
because unless we can do this our our
research isn't trustworthy for the
reasons I'm talking about here and and
in not next session the session after
I'll be then talking more about methods
to allow us to do this but this is a
teaser for this and you know and
something to you know talk to our our
the researchers in our our our room
about so
why does this model matter
well to predict the outcome to know
whether someone's at risk for an outcome
and it could be a healthy outcome a
positive outcome whatever outcome you
want predicting it would be about all of
this but to prevent it through
intervention these drop out here's
another thing just because it's a Cause
doesn't mean it has an effect that's big
enough that it will matter if you change
it so we have to measure and estimate
the the effects and the size of them so
let's drop out those that in this toy
model it's uh don't have a large enough
effect so these are your intervention
targets It also says that you have 20
statistically significant
associations which would replicate in
every study because it's part of the
process that generated the data um but
you only had seven causes with
clinically significant effects so we
have to we have to figure this out
within our data and that's what causal
data science will help us with and you
know I'm just um I wrote an editorial in
the
my voice is getting scratchy um in the
orange Journal talking about the
research is you navigating a causal
Labyrinth and that we need the sort of
methods that could really rule out
confounding uh in a rigorous way in
order to navigate this Labyrinth but
this
also is the justification for why we
need to be
personalized why we can't have an
intervention like it's the same
intervention for everyone with a
diagnosis because people are complex and
their complexity is
causal that's why they're complex so we
have our model here and this tells us
something let's say we have three kids
for a given
outcome cu could
be aggression could be substance abuse
suicidality or an outcome could be
adaptive functioning it could be doing
well in school it could be anything like
that relate it to the complex people
anytime you think that what contributes
to this is is
multicausal you have to be
personalized and this is one and two as
we've talked talked about the same thing
so one each of these will be predictive
telling you the risk but to reduce the
risk we can't assume that everyone would
have exactly the same ones and so part
of it is we assess every kid and and
then we personalize the interventions
based on that in our last session we'll
be talking about work we're doing to
create clinical decision support Tools
in order to personalize um intervention
based on knowledge of of causes in a
complex way and um you know I'm just
wrapping up here uh I hope we'll have a
good discussion but again I want to put
this here about um I've gone over it but
this this is something to burn in your
brain because even though
our our as I've talked about our science
is lacking to say the least about
causality it's still you could still use
these ideas and next session I'll be
talking a lot about how you use these
ideas now to help um the people you're
working with we will you know I'm going
to continue with Eddie so that and and
to anticipate we need better scientific
knowledge but even with knowledge we
have now we could move forward with it
so next
session how to treat Edie through causal
understanding including assessing some
of the things that people talked about
today and introduce a framework that can
help guide assessment in targeting
causes for Mal treatment outcomes in
your practice I'm going to present an
example of an intervention model that
uses this framework actually the model
that I created with my team called
trauma systems therapy we'll be talking
about that next
time and I'll be presenting tools that
you could use um uh to implement the
causal framework in your practice
whether you're using TST trauma systems
therapy or not or any any to give you
some tools that will help so why um
let's open this up for um some
questions or
comments thank you Glenn so you can uh
again you can either uh unmute yourself
raise your hand um by clicking on the
react button and then selecting raise
your hand or you can throw any questions
or comments into the
chat Quinta raised her hand so by all me
quiny you can unmute yourself and ask
your question or comment um Dr Glenn um
Dr sax I apologize I I appreciate your
when is fine don't
apologize uh appreciate all the
information I look forward to the next
session um for me um as a practitioner I
almost um have in mind when I assess uh
my clients or or patients depending on
where people work um I almost want to
put the diagnosis aside and then say
okay what's the
problem yeah because yeah yeah so that
that's sort of where I've um I'm I'm
leaning because uh I get a client
referred to me because they were charged
with so many know they have a legal
problem but then okay I have to do a
diagnosis for the um the uh uh insurance
companies but the fact that they picked
up a charge it's not necessarily the
problem you know so that's where I'm
leaning no and I I'm with you completely
and and actually showed you why what
you're doing makes sense I mean I I
showed you that with PTSD and add as a
diagnosis you know they some of the some
components of them were not useful for
intervention and some were but you
couldn't tell from diagnosis alone but
you need to intervene to change the
outcome and so part of it is figuring
out how to do this and so the the
diagnosis we have in mental health
because it's non-causal it doesn't give
you this and part of it is wherever you
are you're classifying People based on
the information you get about them and
what I mean as I've said before by
classification you take the information
you get and you look at the kid and you
say are they at risk for this thing I'm
worried about or not so are they within
a group of people given my experience
that I'm worried about and then
classification will they respond to this
intervention I might think about or not
so the question is you know the
diagnostic system we have is really
limited for that so we have to figure
out which A system that is going to be
better for helping us classify kids and
and families like
this thank you um so while we're waiting
for other questions or comments um in
the chat or by raising your hand just
want to highlight that the feedback
survey is in the chat um if we really um
appreciate your feedback this is only
our second one of these so we're really
really interested we took your feedback
very seriously from the first one for
those of you who attended the first one
so please fill out the feedback survey
uh while we're doing questions um and I
will get to you in one second um and
then also uh the registration link for
upcoming webinars is also in the chat
and if you're interested in CES and you
attended the whole session um you'll get
a notification in about a week to
register and get your CES and if you
don't mind unmuting and asking your
question or
comment yeah so I was just wondering in
terms of did determining what was what
is red and what is green did you use
statistical analysis to do that or is it
something else so well right now this is
a hypothetical example that you know
there is going to be something that's
true and the way to know it's true is by
using certain uh statistical and data
science techniques to be able to know or
model it's closely as possible the
causal process that generated the data
so we'll be talking about in next
sessions what these techniques are
giving you different examples of
different ways of getting models that
look like this because you you you need
to use very specialized models uh
techniques in order to get that okay
thank
you other folks with questions or
comments feel free again to throw them
in the chat or raise your
hand I really appreciate everybody's
thoughts and comments so
far and if you haven't please fill out
the feedback survey and register for
future events and could I just say
debate is welcome like if anyone wants
to argue that like this it's hard to get
this right we're in a complex world we
have to figure figure out how to get it
right and so you know I I love that
we're a mixture of
providers
administrators and scientists here and
and other other other uh uh people are
represented as well because the solution
is going to involve us mixing it up
because everyone has a a piece of the
elephant so to speak to get it right so
while we're uh waiting for any other
questions or comments that come in Glen
do we want to go just go through the
next session real quick and and ask
people you know just so they know what
that is and question we'll get yes so
this is you know looking ahead where
we're going here
um so we finished one and
two again
we're we introduced the concepts again
we're framing everything by what will be
valuable for what PR practitioners need
to help kids all of that now the next
session as I've said it's going to be
okay I've talked about problems with
clinical application what how people
might be thinking of this and tools they
have so the next session is about
well what our ways to use causal and
predictive knowledge in practice and
there are there's pieces of information
there's ways of using predictive tools
uh and the most to me the most important
thing because in the complex world we
should expect that we're going to be
wrong a lot be wrong a lot we need to we
try an intervention then we don't keep
it going in perpetuity or until someone
drops out of treatment and all of that
we proactively look at it did the did
change happen as we expect it would when
we launched the intervention and if not
what have we missed so that that's next
session and again I'll give you very
practical examples and tools for how you
do
this then we're going to be moving more
into the science and with an examination
of our Paradigm including
diagnosis including the methods we use
in our research uh all of that and and
looking at
limitations that will build to session
five where I'll be talking about very
powerful tools that might that will make
a difference here and how they use these
tools to build knowledge that may put us
in even a better framework to establish
a a diagnostic system based on
causality and then nothing nothing
matters nothing nothing unless it can
really help people in practice and in
the champ Center which is sponsoring uh
this champ Center and champ Network as
we've talked about before it's all about
using these models to build decision
support tools tools that could be used
in practice to guide decision making as
as we've talked about and and our you
know our six session is going to be
about about
that thank you so much and a reminder
that everybody can register for any of
the any or all of those events through
the link in the chat will also include
it when we send out the slides and the
video um from today's which we encourage
you to uh share broadly um if anybody
has any additional questions or comments
again please feel free to throw them
into the chat or raise your hand um we
appreciate all the folks who dived in
today um and raise questions and
comments uh and also please fill out the
survey the uh provider survey I believe
last time we sent it out after to but we
always love to get them today because I
know that once once you have left today
um you know opening up a new email is is
always an extra challenge so um please
feel free to uh fill out that provider
survey right or the feedback survey
right now um and register for any future
event so I'm just looking through the
chat and also for any raised hand I am
not seeing anything Jen I want to make
sure I'm not missing anything before
yeah there is one last question um have
you considered how substance use might
impact behavior in a
diagnosis yeah I mean that's that's
that's a question and again it it fits
here as well if you think substance
abuse is important for any outcome it's
it's really an empirical question we we
want to see it it needs to be causal it
can't be a correlate or just a
predictive Factor um certainly I've seen
it important for lots of outcomes but
again we we in this world we have to
presume nothing and really follow the
data and the way we may practice where
we're we're set to always be learning
and then about substance abuse it's also
if you you could then put you could also
put substance abuse as your Mal
treatment outcome because it could be
related to Mal treatment but it's the
same thing you what are the causes of it
whether only the predictors of it if I
want to intervene to reduce substance
abuse I have to intervene on the
causes that makes sense all right that
was a really good question I'm glad
somebody raised that because oftentimes
we keep our mental health World on one
side of the aisle and our substitut
world on the other side of the aisle and
often they uh intersect and just one
thing for everyone to think about is you
could substitute anything for your
outcome anything you think is relevant
positive things pathological things uh
negative whatever words you want to use
and um substitute anything you think is
important but it's still the same thing
it's still causality is in the middle of
it if you want to change it wonderful
all right I'm not seeing any more
comments or raised hands so I think we
will give um Everybody let them go back
to their day I really want to thank uh
Glenn for presenting and thank everybody
for taking the time today and joining um
we look forward to our next webinar next
month um again all of the various ways
to sign up for that we we like that
we're carrying people from one to the
other um I think that's really great uh
fill out the feedback survey um and if
you're waiting on cus those will come
through as well so thank you everybody
for taking the time to join today and we
look forward to seeing you at the next
webinar