Overcoming Healthcare Disparities:The Role of Patient-Centered Care

Lisa A. Cooper, MD, MPH

Professor of Medicine, Epidemiology, and Health Policy and Management

Johns Hopkins Medical Institutions

Racial and ethnic disparities in health are documented

• Life expectancy at birth – Blacks vs. Whites,10 year gap for men, 5 year gap for women

• Infant mortality rate – Blacks and Native Americans vs. Whites: twice as high

• Death rate – Blacks vs. whites: greater for cancer, diabetes, heart disease, HIV/AIDS, homicide; Hispanics vs. Whites: greater for diabetes

• Morbidity – most ethnic minorities vs. Whites: higher for cancer, diabetes, hypertension, obesity, HIV/AIDS, tuberculosis, hepatitis

Potential Reasons for Disparities in Health

•Biologic factors•Socioeconomic status•Environmental factors•Discrimination/Stress•Cultural factors•Health risk behavior•Access to healthcare•Quality of healthcare

Race Health

Access to Health Care for Racial and Ethnic Groups

Barriers Health Care Processes

Use of Services

Mediators

Outcomes

Visits

primary care specialty emergency Procedures preventive diagnostic therapeutic

Health Status mortality morbidity well-being functioning

Equity of Services Patient Views of Care

experiences satisfaction effective

partnership

Quality of providers cultural competence communication skills medical knowledge technical skills bias/stereotyping Appropriateness of care Efficacy of treatment Patient adherence

Personal/Family acceptability cultural language/literacy attitudes, beliefs preferences involvement in care health behavior education/income

Structural availability appointments how organized transportation

Financial insurance coverage reimbursement

levels public support

Modified From Access to Health Care in America (1993, Millman M, ed).Cooper LA, Hill MN, and Powe NR. JGIM 2002; 477-486

Unequal Treatment: A Report of the Institute of Medicine*

Whites Ethnic minorities

Clinical Appropriateness and Need, Patient Preferences

Systems, Legal, Regulatory

Discrimination, Bias, Clinical uncertainty

Disparity

Difference

Qua

lity

of

Car

e

*National Academy Press, Washington DC, 2003

Racial and ethnic healthcare disparities are pervasive

• Conditions: cancer, diabetes, heart disease, kidney disease, HIV/AIDS, mental health, respiratory diseases (e.g., asthma)

• Populations: young, old, urban, rural, men, women, immigrants, non-immigrants

• Settings: primary care, emergency care, hospital care, specialty care, nursing homes

• Levels and types of care: preventive, acute care, chronic disease management

• Dimensions of healthcare quality: timeliness, effectiveness, safety, patient-centeredness

Dimensions of Health Care Quality

• Structure: “characteristics of the settings in which care is delivered…”

• Process: “ …the care itself, or activities undertaken by the health care system…”

• Outcome: “the effect of care on the health and welfare of individuals or populations…”

Donabedian A. JAMA 1988;260:1743-1748

Process

Structure Outcome

interpersonal, technical care, or

appropriateness of care

race concordance, staff expertise, availability, organization, coordination,

patient ratings of care, equity of services

death, complications

Examples of Structure, Process, and Outcome Variables

Disparities in Process of Care• Technical care – many studies

– Ethnic minorities receive fewer preventive services, diagnostic and therapeutic tests and procedures, and fewer appropriate medications

• Patient-centered or interpersonal care – fewer studies– Ethnic minority patients rate interpersonal care from

physicians more negatively than whites– It is unclear whether this is due to ethnic/racial

discordance, poor communication, bias, or mistrust

• Few disparities studies make links between structure, processes, and outcomes

Process

Structure Outcome

Interpersonal or Patient-centered Care

Race Concordance

Patient ratings of PDM

* physicians’ participatory decision-making style

Concordance• What is it?

– a structural dimension of health care quality– shared identities between patients and health

professionals

• Why do we care? – Because most ethnic minorities see physicians who

differ from them in key social characteristics

• Patients and physicians may be concordant in:– Visible demographic factors such as race/ethnicity,

gender, age, education, social class, language– Less visible factors such as beliefs, values,

expectations, preferred roles

Patient-centered Care

“Providing care that is respectful of and responsive to individual patient preferences, needs, and values, and ensuring that patient

values guide all clinical decisions…”

*Institute of Medicine, “Crossing the Quality Chasm, 2001

Race, Gender, and Partnership in the Patient-Physician

Relationship• Design: Cross-sectional telephone survey • Subjects: 1816 adults (784 W, 814 AA, 218 Other)

who had seen their MD (n=65) within the past 2 weeks• Setting: 32 primary care practices, large network style

managed care organization in Washington D.C. area• Predictor variables: race and gender concordant or

discordant status in patient-physician relationship• Main Outcome: patients’ ratings of their MD’s

participatory decision-making (PDM) styleCooper-Patrick L et al, JAMA 1999;282:583-589

Measurement of Physicians’ Participatory

Decision-Making Style*Patient is asked:• If there were a choice between treatments, how

often would this doctor ask you to help make the decision?

• How often does this doctor make an effort to give you some control over your treatment?

• How often does this doctor ask you to take some of the responsibility for your treatment?

*Kaplan SH et al, Medical Care 1995;33:1176-1187Each item contributes 33.3 points. Maximum score is 100 points.

Ethnic minorities rate their visits with physicians as less participatory

77.1

73.9 73.8

72

73

74

75

76

77

78

PD

M s

core

Whites

Blacks

Others

PDM scores range from 0-100. A higher score means visit is more participatory.

Cooper-Patrick L , JAMA 1999;282:583-589

P=0.007

P=0.05

Patients in race-concordant relationships rate their physicians as

more participatory

61.1

63.3

58.5

61.7

56

57

58

59

60

61

62

63

64

Race Gender

concordant

discordant

Adjusted for patients’ age, gender, education, marital status, health status, length of the patient-physician relationship, physician gender (race concordant analysis) and physician race (gender concordance analysis). Cooper-Patrick L, JAMA 1999;282:583-589

P=0.02

Mea

n P

DM

Sty

le S

core

P-value NS

Process

Structure Outcome

Interpersonal or Patient-centered Care:

Communication

Race Concordance Patient ratings of PDM* and Satisfaction

* physicians’ participatory decision-making style

Patient-physician communication is related to important outcomes

• Patient adherence

• Patient satisfaction

• Clinical outcomesGlycemic controlBP controlPain reductionDepression resolution

Roter 1988, Greenfield 1988, Kaplan 1989, Stewart 1995, Kaplan 1995

Patient-Centered Communication, Ratings of Care and Concordance

of Patient and Physician Race• Design: cross-sectional study using pre-visit and

post‑visit surveys and audiotape analysis • Participants: 458 African American and white

adult patients receiving care from 61 PCPs• Setting: urban primary care practices serving

managed care and fee-for-service patients• Patient recruitment: ~10 patients per MD

recruited consecutively from waiting roomsCooper LA, Roter DL, Johnson RL, Ford DE, Steinwachs DM, Powe NR.

Ann Intern Med 2003;139:907-915

Functions of Clinical Communication

• Data-gathering

• Educating and counseling patients

• Relationship-building

• Partnering with patients to negotiate diagnostic and treatment decisions

Lipkin, Putnam, & Lazare, 1995

Measuring Clinical Communication*

• Content (questions and information-giving)– Biomedical talk– Psychosocial talk

• Affect – Emotional Talk - Negative talk– Positive talk - Social talk

• Process– Orientation (directions or instructions)– Facilitation (includes partnership-building)

*Roter Interaction Analysis System (RIAS)

Roter D, Larson S. Patient Educ Couns 2002;46:243-51

Examples from RIASCommunication Categories

• Biomedical talk “Your blood pressure is 100 over 70.”“I was in the hospital last year for ulcers.”

• Psychosocial talk“You really need to get out and meet more people.”“I guess every marriage has its ups and downs.”

• Emotional talk“This must be very hard for you.”“I hope you’ll be feeling better soon.”

• Partnership-building“Do you follow me?” “How does that sound to you?”

Measuring Emotional Tone of Visits using the RIAS

Coders are asked to rate overall emotional tone of the visit for patients and physicians:

• Physician positive affect = (assertiveness + interest + responsiveness + empathy) - hurried

• Patient positive affect = (assertiveness + interest + friendliness + responsiveness + empathy)

The Patient-CenteredClinical Interview

• Visit duration is longer• Speech speed is lower• Physicians are less verbally dominant

• doctor talk to patient talk ratio is close to 1

• Patient-centeredness ratio is high: more psychosocial, emotional, and partnership talk than biomedical talk

• More positive emotional tone

Physicians communicate differently with black and white patients

Communication measure Whites

n=202

Blacks

n=256

p-value*

Physician verbal dominance 1.50 1.73 <0.01

Physician positive affect** 14.1 13.2 0.02

Patient positive affect** 16.7 15.8 <0.01

Patient-centeredness ratio 1.91 1.58 0.08

Adjusted for: patient age, gender, education level, and self-rated health status; and physician gender, race, time since completing training, and report of how well he/she knows each patient.*p-value from linear regression with GEE.** Patient and physician affect scores are derived from audiotape coders’ impressions of the overall emotional tone of the medical visit.

Johnson RL, Roter DL, Powe NR, Cooper LA. Am J Public Health 2004;94:2084-2090.

17.518.2

16.4

13.2

15.4

19.2

15.8

12.7

10

15

20

Visitduration,minutes

Speechspeed per

minute

Patientpositive

affect

Physicianpositive

affect

concordant

discordant

Race-concordant visits are longer with slower speech and more

positive patient emotional tone

Adjusted for patient age, race, gender, and health status, physician gender & yrs in practice Cooper LA et al, Ann Intern Med 2003;139:907-915

P=0.01P=0.03

P=0.05

P=0.19

Patients in Race-Concordant Relationships Rate Their Physicians Better

76.1 73 7368

5157

0

10

20

30

40

50

60

70

80

ParticipatoryDecision-making

Overall Satisfaction Recommend MD to afriend

concordant discordant

Analyses adjusted for patient gender, race, age, and health status, physician gender, years in practice, and patient-centered communication.

Cooper LA et al, Ann Intern Med 2003;139:907-915

Me

an

Sc

ore

/Pro

ba

bili

ty P=.01 P<.01 P=.03

Summary• African American patients experience visits in which

physicians are less patient-centered

• African Americans in race-discordant relationships with their physicians experience:

– Lower levels of satisfaction

– Less participation in medical decisions

– Shorter visits with less positive emotional tone

• Differences in communication do not explain why patients in race-discordant relationships rate their care worse

• Other factors, such as physician and patient attitudes, may play a role

Process

Structure Outcome

Interpersonal Care:Bias

Race Concordance Patient ratings of care

Explicit vs. Implicit Bias

• Explicit (conscious) bias: attitudes and beliefs we recognize and know we have

• Implicit (unconscious) bias: attitudes that are unavailable to introspection and outside of conscious cognition– Can unintentionally affect behavior– Are better predictors of behavior than self

reported measures of prejudice, stereotyping and discrimination

Clinician Racial Bias, Communication Behaviors and

Patient Experiences of Care• Design: Cross-sectional study• Participants: 39 primary care clinicians and

213 of their African American patients• Setting: 24 urban, community-based primary

care practices in Baltimore, Maryland and Wilmington, Delaware

• Main predictor variables: Clinicians’ implicit attitudes about race (race attitude IAT and patient race/medical compliance IAT)

The Race Implicit Association Test(http://www.implicit.harvard.edu)

• An indirect measure of an individual’s implicit (unconscious) attitudes

• Images appear rapidly on computer screen and subjects respond by sorting pairs of images and attributes using right and left keys

• Premise: individuals will respond faster to concepts that are strongly associated compared to those that have weak associations

• If subjects match white+good/black+bad pairings faster than black+good/white+bad pairings, then the race IAT score differs from zero and is positive – labeled implicit white preference

Greenwald, McGhee, Schwartz, 1998

Implicit preference for whites:Response to these pairings is faster…

&African

American unpleasant

pain

death

stink

grief

agony

filth

tragedy

vomit

European American&pleasant

gentle

happy

smile

joy

warmth

pleasure

paradise

rainbow

…than response to these pairings

& unpleasant

pain

death

stink

grief

agony

filth

tragedy

vomit

European American

African American&pleasant

gentle

happy

smile

joy

warmth

pleasure

paradise

rainbow

Implicit association for European American and compliant patient

Response to these pairings is faster…

&

willing

cooperative

compliant

reliable

adherent

helpful

European American

African American&

doubting

reluctant

hesitant

apathetic

resistant

lax

ReluctantPatient

Compliant Patient

…than response to these pairings

&

doubting

reluctant

hesitant

apathetic

resistant

lax

European American

African American&

willing

cooperative

compliant

reliable

adherent

helpful

Compliant Patient

ReluctantPatient

Methods, continued• Main outcomes:

– Audiotaped Measures: Clinician and patient communication behaviors measured by Roter Interaction Analysis System (RIAS)

– Patient ratings of care: overall satisfaction, trust in clinician, participation in decision-making, and quality of interpersonal care measured by post-visit survey

• Analysis: determine whether clinicians’ implicit attitudes predict differences in communication and patient ratings of care*

*Linear and logistic regression with generalized estimating equations to account for clustering of patients by clinician

Measuring Clinical Communication*

• Content (questions and information-giving)– Biomedical talk– Psychosocial talk

• Affect – Emotional Talk - Negative talk– Positive talk - Social talk

• Process– Orientation (directions or instructions)– Facilitation (includes partnership-building)

*Roter Interaction Analysis System (RIAS)

Roter D, Larson S. Patient Educ Couns 2002;46:243-51

Audiotape Ratings of Clinicianand Patient Emotional Tone

• Clinician behaviors– Positive affect – average of 6 items each rated on

a 5-point scale: interest, warmth, engagement, respect, and sympathy

– Negative affect – average of 2 items each rated on a 5-point scale: dominance and hurried/rushed

• Patient behaviors– Positive affect – average of 5 items each rated on

a 5-point scale: interest, warmth, engagement, sympathy, and respect

Patient Ratings of Clinician• Overall satisfaction

– Overall, I was satisfied with this visit– I would recommend this provider to a friend

• Quality of interpersonal care– My provider has a great deal of respect for me– My provider likes me– I like this provider

• Participation in decision-making– If there were a choice, this provider would ask me to help

make the decision

• Trust in provider– I trust this provider to act in my best interests

Responses are on 5-point Likert scale from strongly agree to strongly disagree.

Interpersonal CareQuality Measures

• Patient-centeredness ratio is high: more psychosocial, emotional, and partnership talk than biomedical and procedural talk

• Clinicians and patients exhibit more positive emotional tone and less negative emotional tone

• Patients report higher levels of trust, respect, and satisfaction, and participation in decision-making

Characteristics of CliniciansCharacteristic (N=39)

Mean age, yrs (SD) 44.1(8.2)

Practice experience, yrs (SD) 13.5 (7.4)

Female gender,% 64

Caucasian, % 49

African American,% 21

Asian, % 23

Liberal political idealogy, % 73

Internal medicine training, % 77

Board certified,% 90

Characteristics of Patients

Characteristic N=213

Mean age, yrs (SD) 54.5 (13.3)

High school graduate, % 81

Female gender, % 73

African American,% 100

Have health insurance,% 91

Annual income < $35,000, % 60

Poor/fair self-rated health status 46

Known by clinician (not first visit) 90

Clinician Responses to IAT(N=39)

14%

26%

26%

10%

14%

5%

5%

Strong preference for Whites

Moderate preference for Whites

Slight preference for Whites

Little to no preference

Slight preference for Blacks

Moderate preference for Blacks

Strong preference for Blacks

Percent of respondents with each score

66%

The IAT D (difference score) ranges from -2 to +2, with 0 indicating no relative preference for blacks compared to whites, and positive scores indicating some degree of implicit bias favoring Whites. [mean score for this sample is +0.24 (.49)]

27%

27%

16%

17%

6%

4%

2%

Strong preference for Whites

Moderate preference for Whites

Slight preference for Whites

Little to no preference

Slight preference for Blacks

Moderate preference for Blacks

Strong preference for Blacks

Percent of Harvard website respondents with each score

70%

Implicit Preference for White vs. Black People by732,881 respondents on Project Implicit websites,

July 2000- May 2006

Association of Clinician Implicit Racial Bias with

Communication BehaviorsCommunication behavior β-coefficient P-value

Patient-centeredness -0.67 0.29

Clinician positive affect -0.28 0.21

Clinician negative affect +0.23 0.03

Patient positive affect -0.18 0.03The beta coefficient means for each 1-point increase in the IAT score --indicating more pro-white bias among clinicians – clinician’s negative affect was higher and African American patients’ positive affect was lower . Adjusted for patient age, education, health status, clinician’s gender, race, and the interaction of clinician race with implicit bias.

Association of Clinician Race/Medical Compliance Bias with Communication Behaviors

Communication behavior β-coefficient P-value

Patient-centeredness -3.12 0.004

Clinician positive affect -0.14 0.38

Clinician negative affect +0.02 0.95

Patient positive affect -0.04 0.81

The beta coefficient means for each 1-point increase in the IAT score --indicating more pro-white bias among clinicians – the communication in the visit was less patient-centered. Adjusted for patient age, education, health status, clinician’s gender, race, and the interaction of clinician race with implicit bias.

Clinician Racial Bias andPatient Reports of Care

0 0.5 1.0 1.5 2.0 4.0 6.0 8.0 10.0

I trust this doctor

I would recommend this doctor to a friend

I like this doctor

This doctor asks me to help decide my treatments

This doctor respects me

I was satisfied with this visit

Odds Ratio

0.63

0.24

0.32

0.22

0.47

0.48

As the implicit bias score increases the patient has lower odds of strongly agreeing

Clinician Race/ Medical Compliance Bias and Patient Reports of Care

0 0.5 1.0 1.5 2.0 4.0 6.0 8.0 10.0

I trust this doctor

I would recommend this doctor to a friend

I like this doctor

This doctor asks me to help decide about my treatments

This doctor respects me

I was satisfied with this visit

Odds Ratio

0.49

0.48

0.57

0.89

0.55

0.20

As the implicit bias score increases the patient has lower odds of strongly agreeing

Summary• This is the first study to explore links among

implicit bias, clinician behaviors, and patient ratings in actual patient encounters

• Primary care clinicians in this sample display implicit attitudes about race that are similar to those measured in large samples of society

• Implicit bias favoring whites and the association of white race with medical compliance predicts:– less patient-centered communication – more negative clinician emotional tone– less positive patient emotional tone– poorer ratings of care by African-American patients

Implications• Research – Examine links among clinician

attitudes, behaviors, and health outcomes• Health Professional Education - employ patient-

centered communication skills programs that emphasize rapport building and affective dimensions and enhance awareness of bias and intercultural skills

• Clinical Practice - implement patient activation programs; improve scheduling, increase time to build rapport and develop continuity of care

• Policy - increase numbers of underrepresented ethnic minorities among health professionals

Minority Health Policy Timeline

1970 20081990

1985 DHHS Heckler Report

Minority Health and Health Disparities Research and Education Act of 2000

2003 IOM Report “Unequal Treatment” and first National Healthcare Disparities Report published

1972Tuskegee Syphilis Study becomes public

1980

Health Revitalization Act of 1993 establishes the Office of Research on Minority Health

2000

Evolution of Research on Health Disparities

1980 20001990Describing the problem

Understanding mechanisms Designing interventions Evaluating outcomes

Patient-Physician Partnership to Improve

HBP Adherence• Design: Randomized controlled trial• Population: 50 primary care MDs and 500

patients (60% AA) with high blood pressure• Setting: 15 urban, community-based clinics in

East and West Baltimore• Interventions: Communication skills training on

interactive CD-ROM for MDs; Patient activation by community health worker

• Main Outcomes: patient adherence, BP controlSupported by the National Heart, Lung, and Blood Institute

R01HL69403, 09/01/01-08/31/07

• Design: Randomized controlled trial• Population: 30 primary care clinicians and 250

African American patients with depression • Setting: Urban, community-based clinics in

Delaware and Maryland• Interventions: standard quality improvement vs.

patient-centered, culturally tailored program• Main Outcomes: Depression remission,

depression level, guideline-concordant care

Blacks Receiving Interventions for Depression and Gaining Empowerment

Supported by the Agency for Healthcare Research and Quality R01 HS13645-01, 09/30/03-09/29/09

Funding Sources• National Heart, Lung, and Blood Institute

– R01HL69403 and K24HL083113

• Agency for Healthcare Research and Quality– R01HS013645

• National Center for Minority Health and Health Disparities (P60MD000214)

• Robert Wood Johnson Foundation Amos Medical Faculty Development Program

• The Commonwealth Fund• Fetzer Foundation

Acknowledgments

• Debra Roter• Neil Powe• Daniel Ford• Rachel Johnson• Don Steinwachs

• Mary Catherine Beach• Thomas Inui• Anthony Greenwald• Janice Sabin• Kathryn Carson