Mixed Methods Research in Practice: Communication about Prognosis in Intensive Care Units

1
Mixed Methods Research in PracticeCommunication
about Prognosis in Intensive Care Units

• Douglas B. White, MD, MAS
• Assistant Professor
• Division of Pulmonary and Critical Care Medicine
• Investigator, UCSF Program in Medical Ethics

2
Overview

• Background (brief)
• Aims Study Methods
• Practical Issues
• Research on family members of dying patients
• Training research coordinator
• Data management
• Methodological issues
• Why mixed methods?
• Why grounded theory?

3
Co-Investigators

• Anita Stewart, PhD
• John M. Luce, MD
• Randy Curtis MD, MPH
• Seth Landefeld, MD

• Bernard Lo, MD
• Director, UCSF Program in Medical Ethics
• Research Physician-patient communication
  decision-making.
• Ken Covinsky, MD, MPH
• Director, Geriatrics Research Training Program
• Research determinants of prognosis in community
  dwelling elders

4
An Example

• Previously healthy 71-year man admitted to the
  ICU with a large stroke. He develops severe
  pneumonia w/ resp failure, sepsis and renal
  failure.
• Aphasic, R hemiparesis
• APACHE II 35 In-hospital mortality 70
• Significant functional impairment
• Patient decisionally incapacitated

5
Should life support be continued?

• Surrogate decision-making
• No clear right medical answer
• Preference-sensitive decision

6
Why study communication of prognosis?

• Patients/Families have
• A right to know
• autonomy informed DM
• A need to know
• Prognostic info affects treatment choices
• Prognostic misunderstandings are common

7
I Shouldn't Have Had To Beg for a Prognosis With
all the conflicting reports on his health, I
didn't know if he was holding steady or dying.

Aug. 22, 2005 issue - I was once a stalker. My
victimsyes, there were several were high on the
social scale, but they were not celebrities. They
were doctors.
8
What causes misunderstandings about prognosis?

• Little empirical research about mechanisms
• Poor MD communication skills?
• No information from physicians?
• Optimism bias in MD communication?
• Optimism bias by families?
• Lack of trust in physicians?
• Low health literacy/numeracy?
• Different attitudes about predicting future?

9
The Structure-Process-Outcome Paradigm
Prognosis Communication in the ICU

• Physiciancharacteristics
• Demographics
• Skills
• - Attitudes

Process of care - prognosis discussions -
Content of discussion
Outcome MD-family agreement re prognosis
Family characteristics - literacy/numeracy -
optimism - depression - prior experiences -trust
in physician -Beliefs about future telling
10
What causes misunderstandings about prognosis?

• How do surrogates arrive at an understanding of a
  patients prognosis?
• -what sources of information?
• -cultural/religious influence?
• -attitudes about prognostication?

11
Specific AimsProject 1

• Aim 1 To determine the prevalence and predictors
  of misunderstandings about prognosis between
  physicians and family of ICU patients at high
  risk for death.
• Aim 2 To determine what factors contribute to
  families assessment of a patients prognosis.

12
K12 Project 1- Study Design

• Design Cross sectional study
• Setting 4 ICUs at UCSF (60 ICU beds)
• Subjects
• 175 ICU patients at high risk of death
• Attending MDs
• Family decision-maker(s)
• Measurements
• Questionnaires from MDs family members
• Chart review
• Audiotaped interview with family members

13
K12 Project 1- Subjects

• Eligible Patients
• Lack decision-making capacity
• Mechanically ventilated 3 days and 5 days
• 40 mortality predicted mortality (APACHE II)
• Why study these patients?

14
K12 Project 1- Subjects

• Eligible family decision-maker(s)
• Traditional hierarchy of surrogates is inadequate
• Question to family Who would be involved in DM
  if patient couldnt participate?
• Potential for multiple respondents per patient
• Physician
• Primary Attending Physician

15
Recruitment Data Collection Strategy

• Daily screening
• RA identifies pts intubated for 72 hours
• calculates APACHE scores
• 1st Contact- Attending MD
• Oral consent/permission to approach family
• Answer prognosis questions by phone
• Complete written questionnaire

16
Recruitment Data Collection Strategy

• Contact with Family
• 30 minute questionnaire
• 20 minute semi-structured interview (audiotaped)
• Conducted in private room adjacent to ICUs

17
Outcome Measure- Prognostic Discordance

• What do you think are the chances that the
  patient will survive
• this hospitalization if the current treatment
  plan is continued? Place a mark on
• the line

0 chance of survival 100 chance of survival
0 chance of survival 100 chance of survival
18
Outcome Measure- Prognostic Discordance

• What do you think the doctor thinks are the
  chances that the patient will survive
• this hospitalization if the current treatment
  plan is continued? Place a mark on
• the line

0 chance of survival 100 chance of survival
19
Measurements- Physician

• Predictors
• Demographics (age, gender, race)
• Specialty
• Self-rated skill
• Communicating prognosis to family
• End of life communication skills
• Attitudes about
• Prognostication
• Involving family in decision-making

20
Measurements- Family

• Predictors
• Literacy
• Numeracy
• Desire for information
• Preferred Role in DM
• Depression
• Locus of Control
• Dispositional Optimism
• Prior EOL DM experience

21
Statistical Plan- Phase 1

• Overarching goal To identify factors associated
  with overly optimistic prognostic estimates by
  family.
• Approach multivariate analysis
• logistic regression or linear regression
• mixed effects modeling (2 levels of clustering)
• include factors with p0.20 on bivariate

22
Aim 2 To determine what factors contribute to
families assessment of a patients prognosis.

• Semistructured interviews with family
• RA shows family the recorded prognostic estimate
  and asks
• 1) What has made you think this is your loved
  ones chance of surviving?
• -follow up probes
• 2) I notice this is your prognostic estimate,
  but that this is what you think the MD thinks the
  prognosis is. Can you tell me why theyre
  different?

23
Aim 2 To determine what factors contribute to
families assessment of a patients prognosis.

• Analysis
• -transcription by trained qualitative
  transcriptionist
• -multidisciplinary coding team
• -Grounded theory approach to inductively develop
  a conceptual framework
• -multiple investigator meetings
• -Member checking

24
Expectations- Project 1

1. Quantitative determination of predictors of
   discordance
2. Qualitative understanding of how family members
   make an assessment of patients prognosis.
3. Reasons that family hold systematically different
   view of prognosis than physician.

25

• K12 Project 2
• Audiotaped Discussions about Prognosis

26
Specific AimsProject 2

• Aim 3 To determine how physicians communicate
  with surrogates of ICU patients about prognosis.
• Aim 4 To identify communication strategies that
  are associated with physician-family concordance
  about prognosis.

27
Qualitative Data AnalysisCoding Strategy

• Development of framework
• Inductive process ?Grounded Theory approach
• Develop categories of prognosis
• Preliminary framework
• 5 investigators analyzed prognostic statements
  from same 5 conferences ? each developed
  framework
• Multiple investigator meetings ? developed
  consensus regarding framework

28
Sample coding

• Im really concerned about your fathers future.
  
• His chances of surviving this hospitalization are
  poor.
• When I say that, I mean maybe 80 of people in
  your Dads situation dont survive.
• Even if he did survive, his quality of life would
  be poor.

General
Survival
Survival
QOL
29
K12 Study Design- Project 2

• Design (Nested) cross-sectional study
• Subjects
• N60 subset of the 175 physician-family pairs
  from Project 1
• Measurements
• Audiotaped MD-family discussion
• Questionnaires from MDs family members
• Outcome understanding of prognosis after
  discussion

30
Recruitment

• Daily screening
• By RA ? bedside nurse Is a family meeting
  planned for today?
• 1st Contact- Attending MD
• Oral consent/permission to approach family
• Consent from MD and all family
• probable clustering

31
Data Collection Strategy

• Before MD-Family Meeting
• Family prognostic estimate
• Audiotape the meeting
• After MD-Family Meeting
• MD prognostic estimate
• Family prognostic estimate
• Family satisfaction with communication

32
Outcome Measure- Discordance Score
Family Pessimistic Family Pessimistic Family Pessimistic Family Pessimistic Family Pessimistic Family Pessimistic Family Pessimistic Family Pessimistic Family Pessimistic Family Pessimistic Family Optimistic Family Optimistic Family Optimistic Family Optimistic Family Optimistic Family Optimistic Family Optimistic Family Optimistic Family Optimistic Family Optimistic

-100 -90 -80 -70 -60 -50 -40 -30 -20 -10 10 20 30 40 50 60 70 80 90 100
No Discordance
33
Data Analysis

• Possible Predictors
• number/type prognostic statements
• Language used to communicate risk
• MD behaviors (assessing desire for prog info and
  understanding)
• Family behaviors (questions, explicit statement
  of prognosis, disagreement)
• Family satisfaction w/ communication