Managing Systems, Neighborhood and Population Health

1
Managing Systems, Neighborhood and Population
Health

• Eliseo J. Pérez-Stable, MD
• EPI 222 Health Disparities Research Methods
• April 28, 2011

2
Conceptual Framework Multi-level Determinants of
Health Disparities
Contextual
Individual-level
Demographics - age, gender, race, ethnicity,
education, income
Physical environment
Social environment
Psychosocial - beliefs, attitudes, adherence,
coping, personality
Health health care disparities
Health care system
Technical aspects of health care
Behavior - exercise, diet, alcohol, smoking,
sexual behavior, substance use
Communication with clinicians
Biological - genetics,stress, allostatic load,
opiate receptors, metabolism, telomeres
Economic resources
3
Health Care System
4
Structure-Process-Outcome Paradigm
Process of care -technical care -interpersonal
care
Structure of care
Patient outcomes

• Structure - system of care
• Technical process - knowledge
• and judgment skills
• Interpersonal process - the way care is provided

Donabedian A. Quality Review Bulletin, 1992, p.
356
5
Hospice Use Differences by Race/Ethnicity

• Medicare Beneficiaries, Diagnosis of heart
  failure in 2001
• 98,258 patients, age 80, 39 new
• Any Hospice Use OR
• All 3.9
• White 4.1
• Black 2.8 0.59
• Latino 2.4 0.49
• Gives, Arch Intern Med 2010 170 427

6
Readmission Rates by Race

• Medicare Beneficiaries, 30-day readmission for
  MI, CHF and pneumonia, 3.16 million patients
• Minority-serving hospitals worse
• Readmit W B OR
• MI 22.6 24.8 1.13
• CHF 27.1 27.9 1.04
• Pneumonia 21.3 23.7 1.15
• Joynt KE, Orav EJ, Jha AK, JAMA 2011 305 675

7
Admissions to High-Quality Hospitals for CHD

• Medicare data, 2002-2005, from markets with
  top-ranked cardiac hospitals
• Evaluate role of race, SES of area, distance to
  hospital
• Black with acute MI more likely to go to top
  ranked hospitals (OR 1.12 1.08 - 1.16)
• No difference in care for CABG
• Blacks from disadvantaged zip codes were less
  likely to go to top ranked hospitals (OR 0.75
  0.64 0.86)
• Popescu, Arch Intern Med 2010 170 1209

8
Limited English Proficiency is a Risk factor for
Readmission

• Retrospective review of registry of 7023
  hospitalized patients 2001-2003
• 8 Chinese, 4 Spanish, 4 Russian
• Similar LOS, cost, mortality
• LEP patients had higher adjusted odds of
  readmission OR 1.3 (1.0 - 1.7)
• Chinese and Spanish speaking LEP patients had
  increased odds (1.7 and 1.5) of readmission
• Karliner L, et al. J Hosp Med 2010 5 276-282

9
Use of Interpreters in Language Discordant
Encounters

• Patients using interpreters ask less, say less,
  answer less (professional)
• Encounters take twice as long or do half as much
  cost and time
• Who translates matters professional interpreter
  should be required for all important
  interactions better accuracy
• Technology can help dual head set telephones,
  Video conferencing, new technology?

10
VMI Studies of Interpretation Experiences in
Clinical Settings

• Survey of clinicians experience with non-Spanish
  interpreter-mediated visits
• Hospitalized patients (Spanish, Chinese)
  interviewed admission post- discharge
• Survey of professional interpreters working at 3
  medical centers

11
Characteristics of 283 LEP Visits Primary Care
Community Clinics

• Mode of Interpretation
• In-person 114 (40)
• VMI 107 (38)
• Ad hoc 62 (22)
• 20 different languages
• Half were with women, Mean age 56 y
• 25 first visit with clinician
• No differences in demographics of patients by
  interpretation mode

12
Clinician Ratings by Interpreter Mode

• In-Person VMI Ad hoc
• High quality interpretation 89 93 81
• (Good/V.Good/Excellent)
• High quality communication 77 89 66
• (Good/V.Good/Excellent)
• High patient engagement 95 97 94
• (Fairly well, Well, Very well)
• plt0.01

13
Higher Quality of Interpretation or Communication

• In-Person In-Person VMI
• vs. VMI vs. Ad hoc vs. Ad hoc
• Interpretation 1.93 6.92 3.59
• (95 CI) (0.90, 4.14) (1.88, 25.5) (1.30, 9.91)
• Communication 1.96 3.18 1.62
• (95 CI) (0.90, 4.14) (1.88, 25.5) (1.30, 9.91)
• Adjusted for site, pts language, pts sex, pts
  health status, pts emotional distress,
  clinicians age and sex
• Napoles AM, et al, J Health Care Poor and
  Underserved 2010 21 301-317

14
Use of Interpreters in 234 Hospitalized Patients
With MD at Admission With MD during stay With RN
Total 57 60 37
65 Years 78 81 51
Predictors of Use Predictors of Use Predictors of Use Predictors of Use
10 yrs age 1.4 (1.1-1.8) 1.2 (0.9-1.4) 1.2 (1.0-1.5)
lt HS Grad _______ 1.4 (0.7-2.8) 2.2 (1.0-4.9)
Chinese 1.5 (0.5-4.6) 1.8 (0.6-5.1) 3.3 (1.2-9.3)
15
Interpreters satisfaction with communication by
modality
16
Does VMI represent a significant improvement?
Scenario responding telephonic interpretation is at least adequate Odds ratio for responding VMI is at least adequate (vs telephonic)
Family meeting 26 3.4 (1.5-7.7)
Physical Therapy 47 4.5 (1.6-12.1)
Inpatient RN Teaching 60 3.0 (1.2-7.9)
MD Evaluation in ED 64 1.4 (0.8-2.5)
Consent for Procedure 70 5.0 (1.6-15.8)
Hospital Discharge 70 3.2 (1.1-9.2)
Case Man/ Social work 70 2.5 (1.2-5.4)
17
Neighborhood and Place

• Physical and Social Environment

18
What is neighborhood?

• People composition, poverty level, average
  incomes, unemployed, LEP
• Social environment People relationships,
  Collective efficacy, Social support
• Physical or Built environment
• Green space / parks,
• Roads and walkways
• Housing

19
What is neighborhood part II

• Public services
• Public transportation
• Police and safety

• Commercial services
• grocery
• Retail stores

20
Neighborhood-Health Connections

• Main outcomes of interest
• Self-reported health
• Chronic disease
• Injury, alcohol, violence
• Mortality
• Health behaviors diet, physical activity,
  tobacco

21
Neighborhood Environment Health Disparities

• Policies that create diversity
• in neighborhoods
• public investment
• zoning

• Characteristics of neighborhoods
• Composition poverty, segregation
• built environment stores, parks, trees
• Social environment threats, social cohesion

21
22
Neighborhood - Definitions

• Census boundaries
• Zip code
• Half-mile around home
• Perceived no concrete definition

22
23
Neighborhood - Measures
Direct observation
Geospatial data
24
Administrative Data Census Variables

• Family income - Median income for all
  households
• Poverty - Proportion of persons whose annual
  income falls at or below 175 of the poverty
  line
• Education - Proportion of persons 25 years and
  older with less than a high school education
• Housing value - Median value of owner- occupied
  housing units
• Limited English Proficiency households where
  language other than English spoken

24
25
Alameda County Study
Is there an association of neighborhood
environment with mortality? 1983 data - 50
sample (1,799) Alameda County residents in 1983
(1,129) Neighborhood defined as census tract
Yen IH, Kaplan GA. American Journal of
Epidemiology 1999
26
Is neighborhood environment associated with
mortality?

• Commercial Services
• Supermarkets
• Laundromats / Dry cleaners
• Beauty Parlor / Barber Ships
• Pharmacies

• Environment/Housing
• population of census tract
• Area of tract
• renters
• single family dwellings

• Community SES
• Per capita income
• white collar employees
• crowding

Yen IH, Kaplan GA. American Journal of
Epidemiology 1999
27
Low social environmental quality and 11-year
mortality risk 2-level logistic regression
Alameda County Study 1983 (n996)
Yen IH, Kaplan GA. American Journal of
Epidemiology 1999
28
Commercial services and 11-year mortality risk
ACS
Yen IH, Kaplan GA. American Journal of
Epidemiology 1999
29
Community SES 11-yr Mortality Risk
By individual income level
Odds Ratio
Yen IH, Kaplan GA. American Journal of
Epidemiology 1999
30
Chicago School of Sociology
City Center
Transition
Workers homes
Residential
Commuter
Yen IH, Syme SL. Annual Review of Public Health
1999
31
Survey Assessment of Perceived Neighborhood What
happens when we measure perceived neighborhood
environment? Self-administered questions
over-ride limitation of census boundaries
32
(No Transcript)
33
UCSF Asthma and Rhinitis Panel

• Began in 1992
• Sample of patients with asthma from northern
  California physicians
• RDD sample added in 1999
• Ages 18 to 50 at baseline
• Wave 5 (Feb 2000 May 2001) n 439

Yen IH, et al. Am J Public Health 2006.
34
Assessing neighborhood environment

• Thinking about your neighborhood as a whole, how
  much of a problem do you feel each of the
  following is in your neighborhood? on a scale
  of 0 to 5, 0 is not a problem, 5 is a serious
  problem
• Too much traffic
• Excessive noise
• Trash and litter
• Smells or odors from factories or farms
• Smoke from fires or burning

35
Neighborhood problems and physical functioning
REF0
SF-12 score difference


adjusted for age, sex, income, education,
ethnicity, asthma severity
Yen IH, et al. Am J Public Health 2006.
36
Direct Observation in Collecting Data on
Neighborhoods
Kaiser CYGNET Study 444 7-year old girls
Follow for 5 years pubertal transition
Collect data on diet, physical activity, height,
and weight
Live in Alameda, San Francisco, Marin, and Contra
Costa Counties
37
Neighborhood Observations in CYGNET
Select half of the 7-year old girls
Send trained observer to girls neighborhood.
Walk around to collect information about
presence/absence of food stores, fast food and
other restaurants, recreation opportunities, and
walkability and bikeability.
37
38
Example of a map given to a street observer
39
(No Transcript)
40
Results Kaiser CYGNET Study
Street observation data for 213 girls
Observed 3 to 32 street blocks per address total
of 2,301 street blocks
Created combination variables for items in audit
form, e.g. food stores, public services (e.g.
library, post office), walkability/bikeability
(e.g. sidewalks, cross walks, speed bumps)
Used factor analysis to see how variables cluster
together
41
Social context of Pregnancy

• 39-item survey representing four categories of
  neighborhood attributes
• neighborhood physical conditions
• social interactions
• nonresidential land use (commercial property)
• public, residential and nonresidential space.

Laraia BA, et al. Int J Health Geographics 2006.
42
Examples of items recorded
Adult activity (Code all that apply) If no adults
present Walking
Socializing (Talking with neighbors)
Socializing in mixed racial groups Home
repair, landscaping, or car care
Sitting/standing on porch or stoop
Supervising children Patronizing
business establishments Standing on the
sidewalk Sitting/standing at the bus stop
Getting into or out of vehicles
Walking a dog . Conducting
home-based vending Recreational activity
(e.g., jogging)
Laraia BA, et al. Int J Health Geographics 2006.
42
43
Practical Definition of neighborhood concepts
physical incivilities

• Combination of physical disorder and poor housing
  conditions theorized to communicate decreased
  local social control and may contribute to crime
  and further neighborhood deterioration.
• Items
• fences
• hedges
• Physical and symbolic signs that demarcate
  residential property

43
44
Definition of Neighborhoods Territoriality

• Ownership and social control leading to
  protective effects against crime and other
  adverse events
• Items
• condition of housing, yards, commercial and
  public spaces,
• vacant or burned property
• litter and graffiti

44
45
Our Space GIS database UCSF, UC Berkeley, and
Kaiser DOR

• census data
• retail store data (InfoUSA or Dunn Bradstreet)
• pollution
• traffic
• green space / parks

46
OUR SPACEFood store and eating establishments

• Supermarkets
• Produce vendors
• Small groceries
• Convenience stores
• Specialty food service (e.g. bagel, deli)
• Restaurants
• Fast food

47
Examples of buffer differences
Alameda County
Marin County
47
48
(No Transcript)
49
(No Transcript)
50
Food stores/eating places by county,
race/ethnicity, household income

• San Francisco has more food stores and eating
  establishments than the other counties
• Racial/Ethnic differences in proximity to food
  stores and eating establishments
• Where there are county and racial/ethnic
  differences, not always household income
  differences

51
Access to Markets with Healthy Foods for
Diabetics in New York City

• Food targets Fruit, vegetables, 1 fat milk,
  diet drinks, high fiber bread
• 173 stores in East Harlem and 152 stores in Upper
  East Side
• Had all 5 categories 9 vs. 48
• More likely to live on a block with no store
  selling foods in E Harlem50 vs. 24
• Example of disparities in environmental justice
  issues complicating behavior
• AJPH 2004 94 1549-54

52
Stephen A. Matthews (Pennsylvania State
University) - Multiple Activity Spaces (and
Temporal Rhythms)
52
53
Death Rate by Race/Ethnicity, US 2000

• W B L A/PI
• Heart Disease 130 191 89 72
• Stroke 25 44 20 24
• Diabetes 12 29 19 9
• Age-adjusted per 100,000 NCHS

54
Causes of Death, US 2001

• Latinos
• Heart Disease 23.9
• Cancer 19.7
• Injury 8.4
• Stroke 5.7
• Diabetes 5.0
• Homicide 2.9
• Liver Disease 2.9

• Whites
• Heart Disease 29.7
• Cancer 23.3
• Stroke 6.8
• COPD 5.6
• Injury 3.9
• Flu/pneumonia 2.6
• Diabetes 2.6

55
Causes of Death, Latinos and Whites, 65 y and
over, US 2003

• Latinos rate
• Heart Disease 1036
• Cancer 682
• Stroke 240
• Diabetes 203
• Alzheimers 87
• ESRD 74
• COPD plus 128

• Whites rate
• Heart Disease 1587
• Cancer 1089
• Stroke 385
• Diabetes 143
• Alzheimers 185
• ESRD 93
• COPD plus 323

56
Cancer Mortality and Poverty in Latino Women,
1990-94/1995-00
Site All (rates) lt 10 poverty gt 20 poverty
Lung 14.2 14.9 12.7 13.6 15.2 16.3
Breast 18.6 18.0 16.3 15.0 20.0 20.8
Colon 11.6 11.4 9.4 9.9 11.3 11.6
Chu K, et al. JNMA 2007 1092-1104
57
Health Related Quality of Life by Ethnicity - Los
Angeles 1999
Poor and Unhealthy Activity N
Fair Health Days Limitation D White
3376 13.1 7.1 2.7 Latino
3267 35.6 6.3 2.4 AA 835 21.2
8.3 3.5 API 716 15.3 4.7
1.7 MMWR 2001 50556-9
58
Screening for Colon Canceradults age 50-74,
BRFSS, 2008

• Percent Tested
• White 64
• Black 62
• Asian / PI 56
• AI / AN 54
• Latino 50
• lt High School 46
• High School / GED 58
• Some College / Tech 64
• College Graduate 71
• MMWR 2010 59 810

59
Screening Mammography US 2008Women 50 74, BRFSS

• Percent
• White 81
• Black 82
• Asian / PI 80
• AI / AN 70
• Latina 81
• lt High School 73
• High School / GED 79
• Some College / Tech 81
• College Graduate 86
• MMWR 2010 59 814

60
Prevalence of Elevated LDL-C, Treatment, and
Control, NHANES, 2005-8

• Percent
• High Rx Cont
• White 35 50 35
• Black 30 45 26
• Mexican Am 28 34 20
• lt High School 41 46 28
• High School 42 52 36
• Some College 36 47 32 College Grad 29 49 39
• MMWR 2011 60 109-112

61
Breast Cancer Death Rates Among Women Aged 45-64
Years, by RaceUnited States, 1990-2007
41 decrease in White and 24 in Black women.
MMWR 2010 59 29
62
Conceptual Framework Multi-level Determinants of
Health Disparities
Contextual
Individual-level
Demographics - age, gender, race, ethnicity,
education, income
Physical environment
Social environment
Psychosocial - beliefs, attitudes, adherence,
coping, personality
Health health care disparities
Health care system
Technical aspects of health care
Behavior - exercise, diet, alcohol, smoking,
sexual behavior, substance use
Communication with clinicians
Biological - genetics,stress, allostatic load,
opiate receptors, metabolism, telomeres
Economic resources