Transcript Socioeconomic and Racial/Ethnic Differences in the Discussion of Cancer Screening:

Socioeconomic and Racial/Ethnic
Differences in the Discussion of Cancer
Screening:
“Between-” vs. “Within-” Physician Differences
Yuhua Bao, Ph.D.†,
Sarah Fox, Ed.D.†,
Jose Escarce, M.D., Ph.D. ‡
† Center for Community Partnerships in Health Promotion,
UCLA General Internal Medicine/Health Services Research
‡ UCLA GIM/HSR
Funded by the NIH EXPORT Center at UCLA/DREW (YB), NCI (SAF) and
AHRQ (JE)
Socioeconomic and Racial/Ethnic Disparities
in Cancer and Cancer Screening
 Recent years saw steady decline in cancer death
rates and improvement in cancer survival
 However, disparities by patient socioeconomic
status (SES) are substantial in
 Adherence with cancer screening guidelines
 Stage of diagnoses
 Mortality and survival
 Differences by patient race/ethnicity are less
consistent, but
 Mortality from all cancers is highest among Blacks
2
The Role of Physician-Patient
Communication
 Differential adherence to cancer screening is partly
due to differences in access to care. However,
 Disparities in cancer screening utilization exist even
among people with a usual source of care
 People of low-SES more likely to cite
 “I didn’t know I need it” and
 “Dr did not recommend it”
as barriers to cancer screening (Finney et al. 2003)
 Disparities in cancer screening communication may
have played a role
3
The “Within-” vs. “Between-” Physician
Differences
 Once patients get access to health care,
treatment disparities arise because
 Patients of different SES or race/ethnicity are
treated differently by the same physicians
(“within-physician” differences),
AND / OR
 They are treated by a different group of physicians
(“between-physician” differences)
4
“Within-physician” Differences
5
Possible Mechanisms for
“Within- Physician” Differences
 Patient-level factors
 Low-SES and/or racial/ethnic minority patients are less
aware of the need for cancer screening (Finney et al. 2003)
 They are less assertive / proactive in clinical encounters
 Physician-level factors
 Physicians may perceive them to be less interested (van
Ryn and Burke 2000)
 Physicians may have greater difficulties in assessing their
needs and preferences (Balsa and McGuire 2001; 2003)
 Patient-physician interaction
 Patient preferences and physician attitudes and perceptions
are reinforced (IOM 2002)
6
“Between-physician” Differences
Dr. A
Dr. B
7
Possible Mechanisms for “BetweenPhysician” Differences
 Physicians serving disproportionately more minority or
low-SES patients
 May be less well trained
 Are less likely to be board-certified (Bach et al. 2004)
 Are more likely to be foreign medical school graduates
(Bellochs and Carter 1990)
 May be less knowledgeable about national preventive care
guidelines (Ashford et al. 2000)
 They may also have less resources in the community such
as
 Specialty groups with cancer screening capabilities
 Institutional support for preventive care
 Some of the “within-physician” differences may be
reinforced to become practice patterns
8
Research Question
How much of the differences in cancer screening
discussion were due to “within-” vs. “between-“
physician differences?
9
Data: the Communication in Medical
Care (CMC) Studies
 A research series that promotes physician-patient
communication on important preventive care topics
 Aimed at developing and testing a physician-patient
communication model to change patient health
behaviors
 The second and third studies in the series (CMC2&3)
are both randomized controlled community trials that
 Teach the model in a Continuing Medical Education (CME)
program
 Focused on cancer screening behaviors
10
Data: Patient and Physician Samples
 Physicians: office-based, primary care, practicing at least
50% of the time
 CMC2: Los Angeles County
 CMC3: all southern California except LA County
 Patients: having seen and expect to see study physician
regularly, speaking either English or Spanish
 CMC2: 50-80
 CMC3: 65-79
 Data pooled from CMC2&3 baseline




Physicians: N=191
Patients: N=5978
On average, patients had seen their physicians for 5 years
Number of patients per physician: mean=31, median=30,
range: [2, 83]
11
Outcomes of Interest:
Cancer Screening Discussion
“Did Dr. ever talk to you
about …”
Rate of
Discussion (%)
Fecal Occult Blood Test (FOBT)
36.8
Sigmoidoscopy
30.9
Mammogram (female only)
67.1
Prostate Antigen Test (PSA)
(male only; CMC2)
46.0
12
Statistical Strategies
 Probit model of cancer screening discussion
 Two specifications for each cancer screening discussion
outcome
 Model 1: Patient characteristics only
 to assess the “overall” differences
 Model 2: Model 1 + Physician Fixed Effects
 Differences that remain reflect “within-physician” differences
 “Between-physician” differences=“Overall” – “Within”
 We report
 probabilities of discussion for each racial/ethnic or SES
group compared to a reference group
 Bootstrapped standard errors (and p-values) to provide
statistical inferences
13
By Education:
Discussion of FOBT
Compared to college graduates
Difference in FOBT Diss. Rate
0
-0.02
-0.02
-0.03 -0.03
-0.04
*
-0.04
*
**
*
-0.06
-0.08
-0.09
-0.1
**
-0.12
-0.13 **
-0.14
"Within-" Differences
Lt high school
"Between-" Differences
High school
Some college
* p <0.05; ** p<0.01
14
By Education:
Discussion of Mammogram
Difference in Mammogram Diss. Rate
Compared to college graduates
0
-0.004
-0.01
-0.005
-0.013
-0.02
-0.03
-0.028
-0.04
-0.037
-0.05
-0.06
-0.07
-0.071
*
"Within-" Differences
-0.08
Lt high school
"Between-" Differences
High school
Some college
* p <0.05; ** p<0.01
15
By Education:
Discussion of PSA
Difference in PSA Diss. Rate
Compared to college graduates
0.05
0.026
0
-0.01
-0.05
-0.012
-0.063
-0.1
-0.15
-0.144
**
-0.2
-0.192
**
-0.25
"Within-" Differences
Lt high school
"Between-" Differences
High school
College graduates
* p <0.05; ** p<0.01
16
By Income:
Discussion of FOBT
Compared to annual income of $75+
0.02
Difference in FOBT Diss. Rate
0.03
0.02
0.01
0.00
-0.01
-0.02
-0.03
-0.02
-0.02
-0.03
-0.04
-0.05
-0.04
*
*
-0.06
-0.07
-0.06
**
"Within-" Differences
Less than $15k
"Between-" Differences
$15k - $35k
$35k - $75k
* p <0.05; ** p<0.01
17
By Income:
Discussion of Mammogram
Difference in Mammogram Diss. Rate
Compared to annual income of $75+
0.06
0.05
0.05
0.04
0.03
0.02
0.013
0.01
0
-0.01
-0.001
-0.004-0.003
-0.007
"Within-" Differences
Less than $15k
* p <0.05; ** p<0.01
"Between-" Differences
$15k - $35k
$35k - $75k
18
By Income:
Discussion of PSA
Difference in PSA Diss. Rate
Compared to annual income of $75+
0.04
0.023
0.02
0.019
0
-0.02
-0.021
-0.04
-0.028 -0.03
-0.06
-0.08
-0.1
-0.12
-0.108
"Within-" Differences
Less than $15k
"Between-" Differences
$15k - $35k
$35k - $75k
* p <0.05; ** p<0.01
19
Summary of findings
 Disparities by education
 Strong education gradient in the discussion of all three types of
cancer screening
 Most of the education differences arose within physicians
 Disparities by income
 Less consistent across different screening methods, but
 Seemed to have arisen because of “between- physician”
differences
 Differences by race/ethnicity
 Asian/white differences in the discussion of FOBT and PSA
were mostly “within-physician” differences
 Same physicians were much more likely to have discussed
mammogram with black than white patients
20
Study Limitations
 Patient self-report of clinical encounter experience may
not be consistent with what really happened
 If low-SES patients tend to under-report physicians’ discussion
 Both within- and between- differences by SES are biased up
 But hard to say how that might change the relative magnitude of
the two types of differences
 It depends on the distribution of low (vs. high) SES patients
across physicians
 On the other hand, it is arguable that what patients recall is
what matters
 Small sample sizes for some racial/ethnic groups
 Findings regarding racial/ethnic differences should be
interpreted with caution
21
Implications
 Patient education plays an important role in determining
what happens in a clinical encounter
 Tailor patient informational materials to the needs of loweducation patients
 Raise the awareness of physicians about the challenges faced
by low-education patients
 Physicians are not evenly distributed across
communities of different levels of income
 Targeting physicians practicing in low-income communities may
be especially promising
 Geographic accessibility of providers is important to lowincome patients
22