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 Race • Biologic factors • Socioeconomic status • Environmental factors • Discrimination/Stress • Cultural factors • Health risk behavior • Access to healthcare • Quality of healthcare Health Access to Health Care for Racial and Ethnic Groups Barriers 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 Health Care Processes Use of Services Visits primary care specialty emergency Procedures preventive diagnostic therapeutic Mediators Quality of providers cultural competence communication skills medical knowledge technical skills bias/stereotyping Outcomes Health Status mortality morbidity well-being functioning Equity of Services Appropriateness of care Efficacy of treatment Patient adherence Patient Views of Care experiences satisfaction effective partnership 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 Quality of Care Difference Clinical Appropriateness and Need, Patient Preferences Systems, Legal, Regulatory Disparity Discrimination, Bias, Clinical uncertainty *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 interpersonal, technical care, or appropriateness of care Structure race concordance, staff expertise, availability, organization, coordination, Outcome 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 Interpersonal or Patient-centered Care Structure Race Concordance Outcome 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) style Cooper-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;33:1176-1187 Each item contributes 33.3 points. Maximum score is 100 points. Ethnic minorities rate their visits with physicians as less participatory 78 77.1 PDM score 77 P=0.007 76 P=0.05 75 74 73.9 73.8 Whites Blacks Others 73 72 PDM scores range from 0-100. A higher score means visit is more participatory. Cooper-Patrick L , JAMA 1999;282:583-589 Mean PDM Style Score Patients in race-concordant relationships rate their physicians as more participatory 64 63.3 63 62 P=0.02 P-value NS 61.7 61.1 61 60 59 58.5 concordant discordant 58 57 56 Race Gender 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 Process Interpersonal or Patient-centered Care: Communication Structure Outcome 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 outcomes Glycemic control BP control Pain reduction Depression 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 rooms Cooper 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 RIAS Communication 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-Centered Clinical 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. Race-concordant visits are longer with slower speech and more positive patient emotional tone 20 P=0.01 17.5 P=0.05 19.2 18.2 P=0.03 16.4 15.4 15 15.8 P=0.19 13.2 12.7 concordant discordant 10 Physician Patient Speech Visit positive positive speed per duration, affect affect minute minutes 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 Patients in Race-Concordant Relationships Rate Their Physicians Better Mean Score/Probability concordant 80 70 60 76.1 P=.01 68 73 discordant P<.01 51 73 P=.03 57 50 40 30 20 10 0 Participatory Decision-making Overall Satisfaction Recommend MD to a friend 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 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 Interpersonal Care: Bias Structure Outcome 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 European pleasant & American pain gentle death happy stink smile grief joy agony warmth filth pleasure tragedy paradise vomit rainbow …than response to these pairings European American & unpleasant African pleasant & American pain gentle death happy stink smile grief joy agony warmth filth pleasure tragedy paradise vomit rainbow Implicit association for European American and compliant patient Response to these pairings is faster… European & American Compliant Patient Reluctant Patient willing doubting cooperative reluctant compliant hesitant reliable apathetic adherent resistant helpful lax African & American …than response to these pairings European Reluctant & American Patient Compliant Patient doubting willing reluctant cooperative hesitant compliant apathetic reliable resistant adherent lax helpful African & American 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 decisionmaking, 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 Clinician and 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 Care Quality 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 Clinicians Characteristic Mean age, yrs (SD) Practice experience, yrs (SD) Female gender,% Caucasian, % African American,% Asian, % Liberal political idealogy, % Internal medicine training, % Board certified,% (N=39) 44.1(8.2) 13.5 (7.4) 64 49 21 23 73 77 90 Characteristics of Patients Characteristic N=213 Mean age, yrs (SD) High school graduate, % Female gender, % African American,% Have health insurance,% Annual income < $35,000, % Poor/fair self-rated health status Known by clinician (not first visit) 54.5 (13.3) 81 73 100 91 60 46 90 Clinician Responses to IAT(N=39) Percent of respondents with each score Strong preference for Whites 14% Moderate preference for Whites 26% Slight preference for Whites 26% Little to no preference 10% Slight preference for Blacks 14% Moderate preference for Blacks 5% Strong preference for Blacks 5% 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)] Implicit Preference for White vs. Black People by 732,881 respondents on Project Implicit websites, July 2000- May 2006 Percent of Harvard website respondents with each score Strong preference for Whites 27% Moderate preference for Whites 27% Slight preference for Whites 16% Little to no preference 17% Slight preference for Blacks 6% Moderate preference for Blacks 4% Strong preference for Blacks 2% 70% Association of Clinician Implicit Racial Bias with Communication Behaviors Communication behavior Patient-centeredness β-coefficient -0.67 P-value 0.29 Clinician positive affect -0.28 0.21 Clinician negative affect +0.23 0.03 Patient positive affect -0.18 0.03 The 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 Patient-centeredness β-coefficient -3.12 P-value 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 and Patient Reports of Care 0.63 I was satisfied with this visit 0.32 I would recommend this doctor to a friend 0.24 This doctor respects me 0.48 This doctor asks me to help decide my treatments 0.22 I like this doctor 0.47 I trust this doctor 0 0.5 1.0 1.5 2.0 4.0 6.0 8.0 10.0 Odds Ratio As the implicit bias score increases the patient has lower odds of strongly agreeing Clinician Race/ Medical Compliance Bias and Patient Reports of Care 0.49 I was satisfied with this visit 0.57 I would recommend this doctor to a friend 0.48 This doctor respects me 0.20 This doctor asks me to help decide about my treatments 0.89 I like this doctor 0.55 0 I trust this doctor 0.5 1.0 1.5 2.0 4.0 Odds Ratio 6.0 8.0 10.0 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 patientcentered 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 1972 Tuskegee Syphilis Study becomes public 1970 1980 Health Revitalization Act of 1993 establishes the Office of Research on Minority Health Minority Health and Health Disparities Research and Education Act of 2000 1985 DHHS Heckler Report 1990 2000 2003 IOM Report “Unequal Treatment” and first National Healthcare Disparities Report published 2008 Evolution of Research on Health Disparities 1980 1990 2000 Describing 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 control Supported by the National Heart, Lung, and Blood Institute R01HL69403, 09/01/01-08/31/07 Blacks Receiving Interventions for Depression and Gaining Empowerment • 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 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