Transcript patient expressives

CASES
Comprehensive Assessment of the
Structure of Clinical Encounters System
Understanding Clinical Communication by Time and Task
M. Barton Laws
Department of Health Services Policy and Practice
The evolving view of clinical
communication
‘50s – ’70s: “Sick role” means follow “doctor’s orders”
’80s – ’90s: “Patient-centeredness”
The fabulous 21st Century: “Concordance,” “Shared
decision-making”*
Regardless:
The bulk of physicians’ time is spent talking with patients.
It is indispensable for diagnosis, choice of treatment,
adherence to treatment, consolation and reassurance,
patient satisfaction and retention in care.
It is little understood and haphazardly taught.
* “SDM” 1st defined in President’s Commission for the Study of Ethical Problems in Medicine in 1982, but
doesn’t commonly appear in PubMed indexed lit. until late 1990s. (Makoul & Clayman, 2006)
Talk: The Final Frontier of Clinical Science
Physics (atoms) → Chemistry → Biology → Psychology → Society
Emergent properties at each level
Greater complexity
Less predictability
Less regularity – entities are more variable, boundaries inexact
Interpersonal
communication
Words → Syntax → Speech Acts → Conversational Turns → Interactions
→ Relationships
Qualitative Approaches to
Interpersonal Communication
Conversation Analysis
– How speakers manage the task of conversation
e.g the organization of sequences, turn-taking,
repair, syntax, the structure of speech events,
integration of speech and gesture
Discourse Analysis
– A social constructionist approach; extracts themes
and organizes them according to a theoretical
perspective (e.g., gender role norms).
– Explicitly holds that there are no stable or essential
facts, that discourse creates reality
Elliot Mishler’s classic work
applies Habermas’s theory of
Communicative Action to clinical
encounters.
Finds a “voice of medicine” and
“voice of the lifeworld.”
Physicians habitually thwart the
voice of the lifeworld and channel
discourse into the voice of medicine.
Quantitative Approaches
• All existing approaches (that I know of) code
for events, typically at the speech act level or
presence of specific content categories,
and then simply count them.
• DYAD, Roter Interactional Analysis System,
ad hoc systems for specific research questions
• Criticized as lacking a clear theoretical basis,
not simultaneously capturing speech act and
content, and not capturing the dynamic process
of interaction
The Generalized Medical Interaction
Analysis System (GMIAS)
• Explicit theoretical basis in speech act theory
• Codes for both speech act and content (topic)
• Structure is hierarchical; categories can be
expanded or collapsed depending on research
question, while retaining comparability across
studies
Speech Acts and Topics
Talk is action
physicians, patients,
and people in all roles
do things with words.
Talk has a focus
– a topic
Speech acts
Speech sometimes represents reality
but it always creates reality
Some “perlocutions”
•
•
•
•
•
•
obligation
belief
doubt
intention
resentment
gratitude
Speech Acts -- Illocutions
• Questions
– representative – Is this is a mango?
– expressive – Do you like mangoes?
– knowledge – Do you know what a mango is?
• Representatives
– facts in general – This is a pipe.
– own behavior – I smoke a pipe.
– conclusion/deduction – Smoking caused your cough.
Illocutions – cont.
Expressives
– knowledge
I understand calculus.
– value/belief
Math is useful.
– preference
I like math.
– goal, desire, intention
I want to learn set theory.
– affect
I’m happy.
– relational statements
criticism, praise, agree, disagree,
empathy – validation, reassurance
More illocutions
• Directives
“soft form” – recommend, suggest, approve
“hard form” – order, convince, “aspiration”
• Commissives / promises
• Conversation
management
• Ritual
• Jokes
What’s it all about?
• The domains
– Biomedical
– Psychosocial
– Logistics
– Socializing
• Traditional (RIAS) and empirically valid
• Medicine now incorporates the psychosocial,
but does not integrate it
Biomedical breakdown
• Treatment
– pharmaceutical
• ARV
– non-pharmaceutical
– non-allopathic
•
•
•
•
•
•
Diagnosis
Symptoms, history
Tests
Screening/vax
Behavioral risk
Reproduction
Psychosocial breakdown
•
•
•
•
•
•
•
•
•
Substance abuse
Recovery
Mood or emotions (non-DSM)*
Relationships/natural supports
Health of others
Social Services
Religion
Health insurance/costs
Other
*Is life hard, or is it bad chemicals?
Logistics and Socializing
• Logistics (general)
– physical exam
– studies/trials
– pharmacy/refills
– certifications
• Socializing (general)
– Dr/Pt relationship
– Boundary crossing
So what can we do with this?
Spearman R= 0.36, p-value=0.0067
Relationship between patient expressives
and reported “doctor’s knowledge of you as a person”
Role asymmetry
Speech act distribution
6000
5000
4000
3000
2000
1000
0
Doctor
Patient
r
l
s
e
s
s
nt thy ve s v es
a
o
g
e
n
e
e
u
v
d
o
v
m
t
i
u l Ri
sti ntat ow le es si ge m mpa r ecti is si
H
e
E
cia
Di Co m
Qu res e Kn Expr a na
o
S
m
p
e
n
R
tio
a
s
r
ve
n
co
N=116 routine outpatient visits of PLWHA
But . . .
We’re still just counting events
•
No dimension of time:
• No use of information about
the sequence of events
• No representation of process
– how one event influences another
• No information about specific tasks and
outcomes, beyond what is captured in
topic codes
So . . .
Providers and patients have goals – they meet in a
clinical encounter in order to get things done – to
accomplish purposes and solve problems.
For any given goal in a clinical setting, interpersonal
communication can carry out any or all of four
categories of functions, which we call Processes.
Processes
Presentation (P)
Developing relevant information
about the individual patient – such
as symptoms, circumstances, or
needs. These contribute to problem
definition and to diagnosis.
Information (I)
Presenting general information –
i.e., not specific to the patient –
that is essential to managing a
condition, solving a problem, or
just satisfying the desire to be
informed.
Resolution (R)
Making decisions – selecting
among alternatives (including do
nothing) and (possibly) committing
to future action.
Engagement (E)
Interpersonal and emotional
exchange such as empathy,
reassurance, or feelings about the
other party.
Threads
“Threads” are all of the dialogue pertaining to a specific problem or issue
such as a symptom or complaint, lab results, medication adherence, or a
psychosocial problem, which may be raised by either participant.
Threads may be very brief – e.g. the physician may inquire if the patient
practices safe sex, the patient may answer in the affirmative, and the thread
ends – or very lengthy, taking up any proportion of a visit.
Threads frequently reappear multiple times in a visit – the subject changes,
then comes back.
They may traverse multiple GMIAS topic codes, in other words the concept
of the “subject” of a thread is different from the concept of a priori topic
categories.
Outside of Thread Space . . .
And not coded to processes are:
The GMIAS “socializing” topic – social ritual such as greeting and
parting, small talk, and the occasional boundary-crossing exchange
such as physician confiding in patient. (Yes, odd things do happen.)
Utterances related to the business of the physical examination, e.g.
“deep breath.” (Clinical observations made during the exam are coded
to an appropriate thread.)
General agenda setting or wrap up. (There is often a final review of
many or all of the decisions made in the visit.)
Extraneous material such as 3rd party interruptions.
Sample presentation process
Patient: Well that medication's making me still kind of nauseated.
I was up 3:30 this morning throwing up. So, I think I'm
getting used to it but it’s just taking awhile.
Doctor: Okay.
Patient: Back when I first started, it really made me queasy and
wanting to throw up, so.
Doctor: Now was that the pain medication?
Patient: No, it's the Abacavir.
Doctor: Okay. Do you think it's getting a little bit better?
Patient: Just a little bit, just a little yeah, but not much. I still feel
like I'm gonna throw up during the day.
Sample information process
Patient: Just a question and I’ve been trying to avoid it not to
upset myself up. It’s, you know the fat redistribution?
Doctor: Yes.
Patient: Is, is that a viral orDoctor: Yes, that’s a good question. I think that the current
consensus is that fat redistribution issue is probably
driven by HIV medication, Mostly by medications that
we are prescribing.
Patient: Okay.
Doctor: And this is 10 years later in 2007. And some of the
developments and advances in the field have been a
development of drugs, which do not appear to cause
injury to fat cells, long periods of time.
Sample resolution process
(The blue part)
Doctor: Now are you trying the nausea medicine?
Are you taking that before you take the
medicine?
Patient: I thought you said take it after?
Doctor: Oh no. Thirty minutes before, because I'd
kind of like to prevent you, rather than
waiting 'til you get nauseated and taking it.
I'd like you to take it 30 minutes before you
take your other medicine.
Patient: Yeah. Yeah. Okay I can do that.
(This is quite typical in that it’s a one-way street.
All of the “decision making” happens in Doctor’s cerebral
cortex. Regardless, we call it a process.)
Sample engagement process
Doctor:
Patient:
Doctor:
Patient:
I'm really glad you're getting that hip fixed.
Yeah, I can't wait.
Yeah, yeah it's been a long time you've been
dealing with that.
Yeah. Yeah, it's been awhile.
Thread Ownership
The person who introduces the thread
is the “owner” (P or D).
Who first started talking about this?
Induction codes
1 Request
Pt. makes a request such as signing a
form, or a drug
2 Problem presentation
New or old symptoms or other
problem
3 Inquiries
– focused or open question.
An “invitation” question, e.g. “What
would you like to talk about?” gives
ownership to the other party.
A focused question that names the
subject gives ownership to the speaker.
4 Cued
Something mentioned in one
thread starts another; also
test results, checklist, exam
finding
5
Narrative without
identified problem
Sometimes people just tell
stories.
There are sub-codes with decimal
places for some of these.
Resolutions
1. Information (often no resolution process; only need is for info)
1.1 Diagnosis (no treatment)
1.2 Instructions
1.3 All other
2. Treatment recommendation, prescribing, behavior change
2.1 Patient agrees
2.2 Patient refuses (so far rare)
3. Referral (with sub-categories)
4. Other provider actions (with subcategories)
5. No need to be resolved/no problem
6. Not resolved
Representation of a coded visit
JC09P03
Threads
20
10
0
-10
0
50
100
150
200
250
300
350
Utterances
400
450
500
550
600
650
And another
JC02P03
Threads
14
12
10
8
6
4
2
0
-2
-4
-6
0
50
100
150
Utterances
200
250
300
# of threads per transcript
N=50(JHU24, WSU4, OHSU8, SLR14)
20.0
17.5
15.0
Count
12.5
10.0
7.5
5.0
2.5
0
3
9
15
21
# of threads per transcript
27
33
% of Doctor owned threads per visit
N=50(JHU24, WSU4, OHSU8, SLR14)
15.0
12.5
Count
10.0
7.5
5.0
2.5
0
18.00
30.00
42.00
54.00
66.00
% of DR owned threads per visit
78.00
90.00
Socializing and physical exam are included in “all” but not shown in table.
Eleven visits contain no engagement; 4 have no information process. Data
refers only to those visits in which these processes exist.
All
N=50
Dr
45.8
16%
Pt
11.4
5%
Engagement
N=391
Dr
Pt
0.6
0.1
8.6%
1.2%
Open question
8.7
3.5%
1.9
0.9%
0.4
2.8%
0
0%
0*
0.2%
0.2
2.3%
6.8
4.5%
1.2
0.7%
0.5
0.9%
0.3
1.4%
Leading question
18.8
6.2%
5
2.1%
0.5
5.7%
0.1
1.2%
0.3
0.7%
0.8
5.2%
14.9
8.2%
3.4
1.9%
1.8
2.7%
0.7
2.4%
Total questions
56.4
20.2%
13.6
6%
1
11.5%
0.1
1.2%
1.3
3.9%
1.8
12.6%
41
24.2%
8.2
4.9%
2.8
4.7%
2.3
9%
Representatives
99.4
32.1%
108.5
49.8%
1.2
9.7%
1.1
20.9%
24.1
71.3%
2.9
13.1%
53
28.3%
91.4
56.9%
17.3
33.2%
4.6
17%
Expressives
18.2
5.9%
23.8
9.7%
2
23.8%
1.4
22.4%
2
5.8%
0.8
2.6%
9.4
5%
17.5
9.8%
3.2
4.8%
2.8
8.4%
Empathy
2.2
0.7%
0.1
0%
1.3
15%
0
0.4%
0.1
0.9%
0
0%
0.7
0.4%
0
0%
0.3
0.4%
0
0%
Directives
28.5
10%
1.9
0.8%
1.1
11.5%
0.2
2.9%
2.9
8.6%
0
0.1%
7.4
4.1%
1.1
0.6%
12.1
30%
0.4
1.7%
Commissives
7.2
2.8%
2.8
1.4%
0.2
0.8%
0
1.7%
0.1
0.2%
0.1
0.9%
2.5
1.6%
0.9
0.5%
2.8
9.7%
1.6
12.4%
Other2
83.3
28.2%
73.6
32.8%
2.8
29%
2.8
44%
4.4
12.6%
5.6
34.3%
30.4
17.1%
45.4
27.6%
8.9
19%
11.2
49%
Total
295.2
224.3
9.6
6.3
34.9
16.3
177.6
164.5
47.4
22.9
Closed question
Information
N=461
Dr
Pt
0.9
1.6
2.1%
10.3%
Presentation
N=50
Dr
Pt
37.6
7.3
21.6%
4.3%
Resolution
N=50
Dr
Pt
3.5
2.1
6%
8.1%
Other = Conversation management, social ritual, jokes, and missing values.
Other Points
23% of all threads have at least one 2.x resolution code,
indicating that some treatment decision was made
More than half of all D-owned threads consist only of presentation
Correlations between indicators of “patient centeredness”
in the entire visit, and in the resolution process only:
Indicator
Pearson’s r
P value
Pt/Dr utterance ratio (verbal
dominance)
.42
.003
Pt. questions/Dr. questions
.32
.044
% of Pt. speech acts which are
expressives
.32
.024
% Pt. control parameter utterances
-.02
.89
Conclusions
CASES creates a far more detailed, and in many respects more accurate
representation of an encounter than simply counting events.
In this data, decision making (resolution) processes are physician
dominated, highly directive, and constitute a small portion of most visits
This method immediately suggests many research questions, some of
which seem novel, others of which have been susceptible only to
qualitative approaches
The method can readily be extended or modified to address particular
questions
However, it is laborious!
Thanks to
The National Institute of Mental Health
(ARRA Supplement to R01MH083595)
Ira B. Wilson, Tatiana Taubin, Tanya Bezreh, Yoojin Lee,
William H. Rogers
Also, for development of the GMIAS, Amanda Barrett,
Emily Howe, and Annabelle Lee.
And, for help and advice throughout, Ylisabyth Bradshaw