Title: Racial disparities in pain: A social-cognitive perspective Diana Burgess, PhD Core Investigator Center for Chronic Disease Outcomes Research (CCDOR) Assistant Professor University of Minnesota, Department of Medicine
1Racial disparities in pain A social-cognitive
perspective Diana Burgess, PhDCore
InvestigatorCenter for Chronic Disease Outcomes
Research (CCDOR)Assistant ProfessorUniversity
of Minnesota, Department of Medicine
2Overview of talk
- Background
- Social-cognitive model How site of care may
contribute to racial disparities in pain
management - Current Research Understanding presence and
correlates of racial disparities in pain
treatment using administrative data - Opportunities/Future Directions
- Discussion
3Evidence of racial disparities in pain pain
treatment
- Greater prevalence of pain and greater
impairment/severity of symptoms among nonwhites - Contributors include
- Greater exposure to discrimination other
stressors (Burgess, in press Edwards, 2008) - Poorer pain treatment
- Racial/ethnic disparities in pain treatment
(acute, chronic, bodily injury, postoperative,
and cancer) - E.g., Analysis of National Ambulatory Medical
Care Survey from 1992 to 2001 - lower odds of
receiving an opioid from a primary care physician
for non-whites (Olsen, 2006) -
- Systematic review by Green et al (2003)
4Evidence of racial disparities in VA
- Black veterans experience more pain, seek more
treatment for pain report greater severity
disability (2001 National Survey of Veterans
Golightly, 2005, Dobscha, in press) - Compared to whites, black veterans w/ chronic
pain - less likely to rate effectiveness of treatment as
very good or excellent (Dobscha, in press) - less likely than to be prescribed Schedule 2
opioids (more potent) and were more likely to be
prescribed Schedule 3 opioids (Burgess, 2009) - Black veterans were less likely to have pain
assessed than whites (Burgess, 2009) - exploratory studies
5Racial disparities in pain management consistent
with disparities in other domains
- Over 500 peer-reviewed studies have found racial
disparities in medical care (e.g., IOM report,
Unequal Treatment) - Systematic review (Saha, 2008) - evidence of
disparities in the VA
61. Patient preference2. Site of care 3.
Provider contribution
Potential sources of healthcare disparities
71. Patient preferences (e.g., Non-whites more
likely to refuse treatment)
- Does not have strong support
- IOM report concluded that patient preferences are
unlikely to be major sources of healthcare
disparities - Studies that have examined the role of patient
preferences find that racial differences in
refusal rates are small and that disparities
persist controlling for patient preferences (e.g.
Ayanian, 1999, Conigliaro, 2002 Hannan, 1999,
Kressin, 2002, Petersen van Ryn, 2000 2006
Whittle, 1997)
82. Site of care (minorities treated in settings
w/ lower quality care)
- Growing evidence for this(e.g., Bach, 2004,
2005 Skinner, 2005 Epstein, 2004 Clarke, 2007
Lucas, 2006 Konety, 2005 Barnato 2005) - Although disparities have been documented
independent of treatment site - Some evidence that racial disparities are more
likely in healthcare settings with higher
concentration of minority patients (Silber, 2007
Groeneveld, 2007)
8
93. Provider contribution to disparities
- Some evidence that providers diagnostic
treatment decisions are influenced by patients
race ethnicity - likely to be unintentional, unconscious due to
normal cognitive processes - More research is needed to understand the
underlying mechanisms - Evidence of poorer quality of communication for
non-white vs non-white patients
10How site of care may contribute to healthcare
disparities A social-cognitive model
- Model posits that characteristics of healthcare
settings that lead providers to experience
excessive levels of cognitive load will increase
the likelihood that providers will
unintentionally contribute to racial/ethnic
disparities - Cognitive load the amount of mental activity
imposed on working memory - can come from the task itself
- also from fatigue, stress, multi-tasking, time
pressure, etc. - Burgess, in press, Medical Decision Making
11This model is grounded in dual process models of
social cognition
- Automatic vs controlled processes
- Controlled processes relatively intentional,
conscious, effortful, controllable - Automatic processes relatively unintentional,
unconscious, effortless, uncontrollable - Cognitive load can interrupt, impair, or prevent
execution of controlled processes - This leads to a greater reliance on automatic
processes, which are not disrupted under high
levels of cognitive load - Use of racial stereotypes is one of these
automatic processes
12Primary Hypotheses of Model
- Hypothesis 1 Providers who experience excessive
levels of cognitive load will make poorer
clinical decisions and provide poorer care - Hypothesis 2 Providers who experience excessive
levels of cognitive load will be more likely to
be influenced by racial stereotypes, which will
lead to poorer processes outcomes of care - Hypothesis 3 Racial minorities are more likely
to be treated in settings in which providers
experience excessively high levels of cognitive
load (i.e., levels that harm performance) - Hence, racial minorities will be more likely to
receive poorer care
13Hypothesis 1 Providers who experience excessive
levels of cognitive load will make poorer medical
decisions and provide poorer care (for all
patients)
- Experienced clinicians rely on automatic
processes (e.g., generating diagnosing, use of
heuristics) but, ideally are able to
strategically shift to controlled processes when
needed - Under cognitive load, clinicians ability to
switch from automatic to controlled processing
may become compromised. - Evidence from aviation, human factors,
educational research shows decreased performance
when cognitive load is too high
14Hypothesis 2 Providers who experience excessive
levels of cognitive load will be more likely to
be influenced by racial stereotypes... This will
lead to poorer processes outcomes of care
14
15Hypothesis 2 Providers who experience excessive
levels of cognitive load will be more likely to
be influenced by racial stereotypes... This will
lead to poorer processes outcomes of care
- Stereotypes concepts that contain our knowledge,
beliefs, expectations, and feelings about a
social group - Salient patient characteristics (e.g., race) may
activate stereotypes , which may influence
providers - interpretation of behaviors and symptoms,
- expectations about patient behaviors,
- behaviors toward patients ...which can influence
patients behaviors - This can occur automatically (or implicitly),
without conscious intent
15
16Everyone engages in stereotypingnot just
providers
16
17Stereotypes are more likely to be activated and
applied under high levels of cognitive load
- Under cognitive load it is
- More likely to rely on automatic processes such
as stereotyping - Less likely that we will
- override or correct for stereotypes that are
activated, via controlled processes - engage in individuation (focus on the unique
features of the person), which may differ from
the stereotype that was automatically activated
18Indirect evidence that cognitive load may
increases healthcare disparities via provider
stereotyping (Muroff, 2007)
- Hypothesis Gender stereotypes will be more
likely to influence mental health diagnoses under
high cognitive load - Methods Retrospective chart review of patients
treated in Psychiatric Emergency Services (N
1236) - Cognitive load was operationalized as high versus
low levels of patient load (experienced by each
provider) - Results Under conditions of high cognitive load,
being female increased the odds of receiving a
diagnosis of depressive disorder (a disorder
that has been shown to be over-diagnosed among
women)
19Hypothesis 3 Racial minorities are more likely
to be treated in settings with excessive levels
of cognitive load
- Study by Varkey et al, 2009 Archives of Internal
Medicine - Physicians in clinics w/ at least 30 minority
patients - (N 27) were more likely than physicians in
other clinics (N 69) to - Lack access to referral specialists
- Have more difficult/complex patients
- Report lower levels of job satisfaction work
control - Report a chaotic workplace (4 X more likely)
- These are all sources of cognitive load that may
contribute to lower performance and increase the
likelihood of bias
20III. Current research Presence Correlates of
Disparities in Pain Management
- Will assess
- 1) the extent to which racial disparities in
pain assessment, treatment, and outcomes exist
across VHA facilities - 2) whether racial disparities are smaller or less
likely in organizations with greater structures
processes that support high quality pain
management. - Such structures/processes free up cognitive
resources for providers, improving the quality
of decision-making/care overall and reducing the
likelihood that racial stereotypes will influence
decisions - VA HSRD Co-investigators Bair, Farmer, Kerns,
Nelson, Partin, van Ryn
21Secondary data analysis
- 2007 Survey of Healthcare Experiences of Patients
(SHEP), ambulatory care model - Sampling frame (AA vs. white, w/ visit in primary
care) - Pain outcomes (perceived effectiveness of pain
treatment functional interference due to pain) - Corporate data warehouse (CDW)
- Pain assessment (presence of a pain score)
- Pharmacy Benefits Management (PBM) database
- Pain treatment (pain medication)
- Clinical Practice Organizational Survey Primary
Care Directors Module (CPOS-PC) - Structures/processes that support pain management
general measures of cognitive load - OQP (Office of Quality Performance)
- Cognitive load (primary care access measures)
22Examine the presence correlates of racial
disparities among the following cohorts
- Pain Assessment Cohort Was pain assessed at
SHEP sampling visit? - Base sample SHEP responders non-responders
whose index visit was in primary care - Pain Treatment Cohort Pain medication issued at
patient encounters one year prior to index visit - Chronic pain sample Patients in base sample
with chronic pain Dx in past year - Pain Outcomes Cohort Perceived effectiveness of
pain among treatment/ functional interference due
to pain - Outcomes sample Patients in chronic pain sample
who responded to the SHEP
23IV. Opportunities/future research - using this
dataset administrative VA data
- Examine variation in pain management among other
vulnerable/stigmatized groups - Obese versus non-obese (Pilot study to be
submitted, P.I. Diana Higgins) - Women (gender stereotypes)
- Age/cohort (elderly, OEF/OIF)
- Mental health comorbidities
- Examine association between treatment outcomes
24Other research questions...
- In OEF/OIF population
- Are there racial differences in presence of pain
or in the relationship among pain, PTSD
post-concussive syndrome? - What is the role of early cumulative exposure
to stress adversity as a mediator/contributor
(e.g., Shonoff, 2009 JAMA)? - How might stereotypes/subtypes based on OEF/OIF
status race/ethnicity affect pain treatment?
25(No Transcript)