1
Physician Wages across Specialties:
Informing the Physician Reimbursement
Debate
J. Paul Leigh, PhD,1, 2 Daniel Tancredi, PhD,1, 3 Anthony Jerant, MD,1, 4
Richard L. Kravitz, MD, MSPH,1,5
1
Center for Healthcare Policy and Research, University of California Davis School of Medicine
2 Department
of Public Health Sciences, University of California Davis School of Medicine
3 Department
of Pediatrics, University of California Davis School of Medicine
4 Department
of Family and Community Medicine, University of California Davis School of Medicine
5
Division of General Medicine, Department of Internal Medicine, University of California Davis School of Medicine
Funding/Support: This study was funded in part by grants to Dr. Leigh from the National Institute for Occupational Safety and Health (1
R01 H008248-01) and Dr. Tancredi and Dr. Kravitz from the UC Davis Office of the Vice Chancellor for Research.
Center for Healthcare Policy and Research, UC Davis
2
Center for Healthcare Policy and Research
Speaker’s Verbal Disclosure Statement:
Have you (or your spouse/partner) had a personal financial
relationship in the last 12 months with the manufacturer of the
products or services that will be discussed in this CME activity?
___ Yes
_x_ No
(If yes, please state disclosures and resolutions)
Center for Healthcare Policy and Research, UC Davis
3
Center for Healthcare Policy and Research
Educational objectives for this seminar:
 To understand method to rank specialties
 To identify specialties at the top and bottom of the wage lists.
 To discuss causes and cures for disparities .
Center for Healthcare Policy and Research, UC Davis
4
Introduction
 Previous reports document income disparities between
primary care and other physician specialties.
 The need to address generalist-specialist income gap based on
several lines of research.
 Generalist physicians may provide more cost-effective care
 Health of public better and more equitably distributed among
groups in geographic regions with higher ratios of generalists
to specialists.
Center for Healthcare Policy and Research, UC Davis
5
Intro Con’t
 Shortfall in number of U.S. medical students entering
generalist careers is looming.
 Causes for shortfall multi-factorial, reimbursement has
contributed to rapid declines in medical student interest.
 Many policymakers believe reforming the payment system to
include greater incentives to generalists might increase the
overall value of aggregate medical care to patients.
Center for Healthcare Policy and Research, UC Davis
6
Intro Con’t
 Considerable data available that rank physician specialties by
annual income.
 However, some data in these studies were collected either from
physicians employed in medical schools or from nonrepresentative community practitioner samples.
 Much data collected a decade or more ago, before managed care.
 Most studies separately ranked only a handful of specialties.
 Most prior studies adjusted for either few or none of the covariates
associated with physician income, such as age and gender.
Center for Healthcare Policy and Research, UC Davis
7
Intro Con’t
 Annual income may not be best indicator.
 Does not account for variations in work hours among specialties.
 Physician wages (earnings-per-hour) offer more useful and
intuitive metric.
 Lower income and more hours been linked with physician job
dissatisfaction, undesirable patient outcomes, strikes, medical
problems for physicians themselves and leaving the practice of
medicine.
Center for Healthcare Policy and Research, UC Davis
8
Intro Con’t
 We examined differences in wages across physician specialty categories.
 Employing recent (2004-5) data from nationally representative
Community Tracking Study (CTS) physician survey.
 Controlling for known socio-demographic, community, and
practice correlates of physician remuneration.
 We compared wages between four aggregated generalist and subspecialist physician categories (primary care, surgery, internal
medicine and pediatric sub-specialties, and all others), as well as
across 41 disaggregated specialties.
Center for Healthcare Policy and Research, UC Davis
9
Data
 Physicians responding were representative sample of physicians who
participated in Round 4 (2004-2005, the latest years available) of Community
Tracking Study (CTS) physician survey.
 Physicians responding resided in continental U.S., practiced patient care, and
not employed by federal government.
 Survey followed complex sampling design of 60 communities selected with
weights proportional to population size from strata defined by geographical
region, community size, and whether the community was metropolitan or nonmetropolitan.
 Some hospital-based specialists such as radiologists, anesthesiologists, and
pathologists as well as all residents and fellows were excluded.
 Primary care physicians and responders to previous survey rounds were oversampled.
 The overall response rate was 53% in 2004-2005.
Center for Healthcare Policy and Research, UC Davis
10
Data Con’t
 CTS dataset included information on 6,628 physicians in 2004-2005.
 We required valid (non-missing) answers for income, hours, and
annual weeks worked and for all control variables.
 We restricted analysis to physicians who reported at least 20 hours
but not more than 100 hours per-week and who reported at least 26
weeks worked per-year.
 We reasoned these restrictions enhance robustness of results by
limiting outliers, data misclassifications, decimal misplaced.
 Sample = 6,381.
Center for Healthcare Policy and Research, UC Davis
11
Dependent Variables
 Annual income from medical practice measured as pre-tax, net of
practice expenses, including malpractice insurance.
 Because income measured before taxes, was not “take-home pay.”
 CTS physician survey contained upper limit of $400,000 (n = 378;
weighted percent = 7.5%) and a lower limit of $0 (n = 28; weighted
percent = 0.5%).
 Work hours measured per-week, via self-report week before the survey.
 Work hours included “all medically related activities” which included
“direct patient care, administrative tasks, and professional duties.”
 Weeks worked measured the number of weeks worked per-year.
Center for Healthcare Policy and Research, UC Davis
12
Specialty Variables
 Specialty codes classified physicians according to specialty they
reported spending most time weekly.
 To enhance integrity of findings, we combined specialty
classifications with fewer than 20 respondents into related
specialty classifications to achieve minimum of 20 respondents
in each of 41 specialties.
Center for Healthcare Policy and Research, UC Davis
13
Specialty Variables Con’t
 Created four broad specialty categories: primary care, surgery, internal medicine
and pediatric sub-specialties, and all other specialties.
 Primary care included pediatrics, geriatrics, family practice, internal medicine,
general practice, and internal medicine and pediatrics (combined).
 Surgery contained neurological, plastic, orthopedic, otolaryngologic, thoracic,
urologic, general, and vascular, obstetrics/gynecology, and “other surgical.”
 Internal medicine and pediatric subspecialties included allergy and immunology,
hematology, gastrointestinal, cardiovascular, nephrology, rheumatology,
pulmonary critical care, pulmonary, endocrinology, infectious diseases, critical
care internal medicine, hospitalists, medical oncology, neonatal, and other
pediatric sub-specialties.
 Final category, “other medical,” included radiation oncology, physical medicine
and rehabilitation, occupational medicine, emergency medicine, adult and child
psychiatry, pediatric emergency, neurology, ophthalmology, and dermatology.
Center for Healthcare Policy and Research, UC Davis
14
Control Variables
 Selected control variables based on review of medical and economic literature.
 Classified variables as physician characteristics, community factors, and
practice factors.
 Physician characteristics included age, gender, race, whether board
certified, and whether graduated from foreign medical school.
 Community factors included residence in areas with less than 200,000
population (roughly 9% of sample) and residence in nine regions of the country.
Center for Healthcare Policy and Research, UC Davis
15
Control Variables Con’t
 Practice factors included practice ownership, whether employed by
medical school, and experience with managed care.
 Three-level CTS practice ownership variable parameterized with two
dummy variables - full owner (sole proprietor) and part owner (partner) –
using non-owners as reference.
 Physician’s experience with managed care captured by variable “percent
of revenue from managed care,” which we expressed in 20 percentage
point units.
Center for Healthcare Policy and Research, UC Davis
16
Statistical Methods
 Stata 11 used.
 Survey design effects arising from unequal probability sampling,
stratification and clustering were accounted for by using weighting
and survey data analysis procedures.
 In multivariable analysis, control variables were accounted for in
Tobit and least squares regressions.
Center for Healthcare Policy and Research, UC Davis
17
Statistical Methods Con’t
 CTS questionnaire did not allow physicians income values lower than $0
or greater than $400,000.
 Some might have had negative income (perhaps due to business
expense) others may have had incomes above $400,000.
 Tobit regression designed for “censored” variables.
 Tobit regression used to generate predicted values below and above the
limits based on covariates in the model.
Center for Healthcare Policy and Research, UC Davis
18
Statistical Methods Con’t
1) Ran Tobit regressions for income with all specialty variables and control
variables as covariates.
2) Obtained individual conditional expectations of income.
3) Used expected values in place of censored response values at $0 and
$400,000.
 To construct wages, we divided annual income by the product of weekly
work hours multiplied by the number of reported weeks worked per-year.
Center for Healthcare Policy and Research, UC Davis
19
Statistical Methods Con’t
 General surgery chosen as reference category because it contained many
incumbents (n = 202) and was near middle of specialties when ranked
by mean wages.
 Linear regressions used when wage was dependent variable.
 Tobit income regression model used to impute incomes for values left
censored at $0 (n=23) or right censored at $400,000 (n=369) and
included all covariates in the wage regression as well as work hours and
annual weeks worked.
Center for Healthcare Policy and Research, UC Davis
20
Results
 Table 1 presents frequencies and percents for binary variables.
 Tobit-adjusted mean annual income was $187,857 (2004-2005 dollars).
 Mean weekly work hours was 53.1 (median=50; mode=60), and mean
weeks worked was 47.3 (median=48; mode=48).
Center for Healthcare Policy and Research, UC Davis
21
Table 1
Characteristics of physician respondents to the 2004-2005 Community Tracking Study
Variable/Description
Frequency,
un-weighted
Percent,
un-weighted
6381
100.00
<$50k income
434
6.80
$50-$99k income
745
11.68
$100-$149k income
1617
25.34
$150-$199k income
1362
21.34
$200-$249k income
874
13.70
$250-$299k income
512
8.02
>$300k income
837
13.12
Work hours <=50
3496
54.79
51≤work hours ≤60
1666
26.11
Work hours ≥61
1219
19.10
Entire sample
Dependent Variables
Center for Healthcare Policy and Research, UC Davis
22
Table 1 Con’t
Characteristics of physician respondents to the 2004-2005 Community Tracking Study
Variable/Description
Frequency,
un-weighted
Percent,
un-weighted
age <35
342
5.36
age 35 to 44
1963
30.76
age 45 to 54
2214
34.70
age 55 to 64
1323
20.73
age 65 to 74
432
6.77
age 75+
107
1.68
White non-Hispanic
4728
74.09
African-American non-Hispanic
259
4.06
Hispanic all races
328
5.14
Asian, Pacific Islander, other, non-Hispanic
817
12.80
Missing race
249
3.90
Gender male
4617
72.36
Gender female
1764
27.64
Control Variables
Center for Healthcare Policy and Research, UC Davis
23
Table 1 con’t
Characteristics of physician respondents to the 2004-2005 Community Tracking Study
Variable/Description
Frequency,
un-weighted
Percent,
un-weighted
New England
515
8.07
Middle Atlantic
983
15.41
East North Central
1061
16.63
West North Central
211
3.31
South Atlantic
1362
21.34
East South Central
234
3.67
West South Central
679
10.64
Mountain
397
6.22
Pacific
939
14.72
Rural/town
793
12.43
Board certified
5834
91.43
Foreign medical school graduate
1288
20.18
Control Variables
Center for Healthcare Policy and Research, UC Davis
24
Table 1 con’t
Characteristics of physician respondents to the 2004-2005 Community Tracking Study
Variable/Description
Frequency,
un-weighted
Percent,
un-weighted
Full owner (sole proprietor)
1924
30.15
Part owner (partner)
1419
22.24
Not an owner
3038
47.61
Employed in medical school
597
9.36
Percent revenue managed care 0-20%
1841
28.85
Percent revenue managed care 21-40%
1622
25.42
Percent revenue managed care 41-60%
1240
19.43
Percent revenue managed care >60%
1678
26.30
Control Variables
Center for Healthcare Policy and Research, UC Davis
25
Results Con’t
 Table 2 provides frequencies on specialties and ranks them based
upon the population weighted wages without adjustment for control
variables.
 Top panel provides information on four broad specialty categories.
 Next panel provides information on 41 narrow specialty categories.
 From highest to lowest wages, ordering of broad categories: surgery,
“other medical”, internal medicine and pediatrics sub-specialties, and
primary care.
 Wages for surgery, internal medicine and pediatric sub-specialties, and
other specialties were 52%, 40%, and 46% higher, respectively, than for
primary care specialties.
Center for Healthcare Policy and Research, UC Davis
26
Results Con’t
 Bottom panel of Table 2 (ranking results for 41 specific
specialties), the top two specialties (significantly different from
general surgery) were neurological surgery (wage=$132, p<0.05)
and radiation oncology ($126, p<0.01)).
 Bottom two specialties were internal medicine and pediatrics
(combined practice) ($50, p>0.01) and other pediatric
subspecialties ($52, p<0.01).
Center for Healthcare Policy and Research, UC Davis
27
Table 2
Broad Category and Specialties Ranked by Wages, Unadjusted for Control Variables
Broad Category and Specialty
Broad Category
Primary Care (referent)
Surgery
Internal Medicine and Pediatric Sub-specialties
Other Medical
Frequency Mean Wage, dollars Mean Hours
3179
1064
836
1302
60.48
92.10***
84.85***
88.08***
51.2
59.0***
57.9***
47.4***
P-value legend: ***<=0.01; **=(0.01-0.05); *=(0.05-0.10), compared to Primary Care in top panel and compared
to General Surgery in second panel
Center for Healthcare Policy and Research, UC Davis
28
Table 2 Con’t
Broad Category and Specialties Ranked by Wages, Unadjusted for Control Variables
Broad Category and Specialty
Frequency Mean Wage, dollars Mean Hours
Rank and Specialty
1. Neurological Surgery
22
132.33**
58.0
2. Radiation Oncology
42
126.32***
52.9***
3. Medical Oncology
47
114.21
60.2
4. Plastic Surgery
51
54.3**
5. Allergy & Immunology
54
113.78
111.71*
48.7***
6. Thoracic Surgery
20
110.45**
64.9
7. Orthopedic Surgery
167
107.93***
60.6
8. Ophthalmology
180
103.63**
51.7***
9. Dermatology
97
102.68*
44.9***
10. Hematology & Oncology
21
100.17
55.0**
11. Other Surgical Subspecialties
32
99.77
62.2
12. Cardiovascular Diseases
145
93.74
59.7
P-value legend: ***<=0.01; **=(0.01-0.05); *=(0.05-0.10), compared to Primary Care in top panel and compared
to General Surgery in second panel
Center for Healthcare Policy and Research, UC Davis
29
Table 2 Con’t
Broad Category and Specialties Ranked by Wages, Unadjusted for Control Variables
Broad Category and Specialty
Frequency Mean Wage, dollars Mean Hours
Rank and Specialty
13. Gastroenterology
113
93.27
56.8*
14. Otolaryngology
81
93.00
54.6**
15. Neurology
91
92.52
51.5***
16. Physical Medicine & Rehabilitation
68
91.69
46.4***
17. Emergency Medicine
391
87.47
47.4***
18. Nephrology
49
86.61
58.2
19. General Surgery (referent)
202
85.98
60.6
20. Urology
91
85.62
58.2
21. Occupational Medicine
52
84.71
44.0***
22. Obstetrics and Gynecology
365
83.40
57.1*
P-value legend: ***<=0.01; **=(0.01-0.05); *=(0.05-0.10), compared to Primary Care in top panel and compared
to General Surgery in second panel
Center for Healthcare Policy and Research, UC Davis
30
Table 2 Con’t
Broad Category and Specialties Ranked by Wages, Unadjusted for Control Variables
Broad Category and Specialty
Frequency Mean Wage, dollars Mean Hours
Rank and Specialty
23. Vascular Surgery
33
80.47
68.0**
24. Rheumatology
34
78.28
52.6***
25. Pulmonary Critical Care Medicine
29
76.32
61.5
26. Neonatal & Perinatal Medicine
62
75.86
64.8
27. Psychiatry
294
72.24**
44.7***
28. Hospitalists
36
71.84*
60.3
29. Pulmonary Diseases
55
71.67*
63.0
30. Endocrinology
37
71.65
47.7***
P-value legend: ***<=0.01; **=(0.01-0.05); *=(0.05-0.10), compared to Primary Care in top panel and compared
to General Surgery in second panel
Center for Healthcare Policy and Research, UC Davis
31
Table 2 Con’t
Broad Category and Specialties Ranked by Wages, Unadjusted for Control Variables
Broad Category and Specialty
Frequency Mean Wage, dollars Mean Hours
Rank and Specialty
31. Pediatrics
729
69.24**
45.1***
32. Critical Care Internal Medicine
23
68.84**
67.5**
33. Child & Adolescent Psychiatry
58
67.36**
46.3***
34. Infectious Diseases
36
66.26**
52.1
35. Pediatric Emergency Medicine
29
65.76
41.1***
36. Family Practice
1309
58.25***
52.2***
37. Internal Medicine
974
58.18***
54.3***
38. General Practice
88
57.55***
47.8***
39. Geriatric Medicine
31
56.84***
52.8***
40. Other Pediatric Subspecialties
95
51.62***
55.7**
41. Internal Medicine & Pediatrics
48
49.90***
57.0
P-value legend: ***<=0.01; **=(0.01-0.05); *=(0.05-0.10), compared to Primary Care in top panel and compared
to General Surgery in second panel
Center for Healthcare Policy and Research, UC Davis
32
Results Con’t
 Table 3 presents results of linear regression of wages on the
broad categories together with control variables.
 Coefficients on broad categories indicate how much more the mean
wage is in that category compared to primary care, adjusted for
control variables.
 Results on broad categories were similar to those that did not
account for control variables: surgery (+$29.00 more than primary
care, p<0.01), other medical (+$27.13, p<0.01), and internal
medicine and pediatrics subspecialties (+$21.59, p<0.01).
Center for Healthcare Policy and Research, UC Davis
33
Results Con’t
 Next panel of Table 3 reports results on control variables.
 No race category was statistically significant from non-Hispanic
whites except for the combined group for “other” and “missing.”
 Women’s wages were roughly $9 below those of men’s (p<0.01).
 Being a partner in private practice was positively (+$14; p<0.01)
whereas current employment in a medical school (-$18; p<0.001)
was negatively associated with wages.
Center for Healthcare Policy and Research, UC Davis
34
Table 3
Linear Regression Results for Broad Categories and for Control Variables with Wage as Dependent Variable
Covariate
Broad Categories
Primary Care
Surgery
Internal Medicine and Pediatric Sub-specialties
Other Medical
Regression
Coefficient
Lower 95%
Confidence Limit
Upper 95%
Confidence Limit
Referent
29.00***
21.59***
27.13***
Referent
23.78
15.05
22.64
Referent
34.21
28.14
31.62
P-value legend: ***<=0.01; **=(0.01-0.05); *=(0.05-0.10)
a. F-test for entire regression was statistically significant, p-value <0.01.
Center for Healthcare Policy and Research, UC Davis
35
Table 3 Con’t
Linear Regression Results for Broad Categories and for Control Variables with Wage as Dependent Variable
Covariate
Control Variables
age <35
age 35 to 44
age 45 to 54
age 55 to 64
age 65 to 74
age 75+
White, non-Hispanic
African-American, non-Hispanic
Hispanic
Asian, Pacific Island, non-Hispanic
Missing race
Regression
Coefficient
Lower 95%
Confidence Limit
Upper 95%
Confidence Limit
-10.55***
-7.02***
Referent
0.91
0.58
2.69
Referent
-2.70
4.04
2.96
-10.80**
-16.03
-11.99
Referent
-4.41
-8.15
-15.65
Referent
-13.01
-4.89
-3.60
-20.98
-5.07
-2.04
Referent
6.23
9.31
21.02
Referent
7.61
12.96
9.53
-0.62
P-value legend: ***<=0.01; **=(0.01-0.05); *=(0.05-0.10)
a. F-test for entire regression was statistically significant, p-value <0.01.
Center for Healthcare Policy and Research, UC Davis
36
Table 3 Con’t
Linear Regression Results for Broad Categories and for Control Variables with Wage as Dependent Variable
Covariate
Control Variables
Female
New England
Middle Atlantic
East North Central
West North Central
South Atlantic
East South Central
West South Central
Mountain
Pacific
Rural/town
Regression
Coefficient
Lower 95%
Confidence Limit
Upper 95%
Confidence Limit
-9.45***
0.30
-1.64
2.27
5.71
Referent
3.54
4.38
3.12
1.22
-7.23***
-13.26
-7.04
-5.97
-2.56
-4.78
Referent
-5.75
-0.86
-5.04
-4.79
-12.66
-5.64
7.64
2.68
7.09
16.19
Referent
12.83
9.62
11.28
7.23
-1.79
P-value legend: ***<=0.01; **=(0.01-0.05); *=(0.05-0.10)
a. F-test for entire regression was statistically significant, p-value <0.01.
Center for Healthcare Policy and Research, UC Davis
37
Table 3 Con’t
Linear Regression Results for Broad Categories and for Control Variables with Wage as Dependent Variable
Covariate
Control Variables
Board certified
Foreign school graduate
Full owner
Partner
Not owner or partner
Employed by medical school
Percent revenue managed care 20 percent units
Intercept
Regression
Coefficient
Lower 95%
Confidence Limit
Upper 95%
Confidence Limit
12.52***
0.62
-1.86
14.42***
Referent
-17.97***
-0.12
53.19***
5.19
-4.21
-6.93
8.81
Referent
-23.39
-1.55
44.41
19.85
5.44
3.21
20.02
Referent
-12.56
1.31
61.97
P-value legend: ***<=0.01; **=(0.01-0.05); *=(0.05-0.10)
a. F-test for entire regression was statistically significant, p-value <0.01.
Center for Healthcare Policy and Research, UC Davis
38
Figure 1 summarizes results on wages for the four broad categories
before and after adjustment for control variables and indicates
adjustment did not appreciably alter estimated wages.
Center for Healthcare Policy and Research, UC Davis
39
Figure 1: Wages for Broad Categories Before and After Adjustment
for Control Variables
100.00
92.10
89.65
90.00
84.85
88.08
82.25
87.79
Dollars per Hour
80.00
70.00
60.49
60.65
60.00
50.00
Unadjusted
40.00
30.00
Adjusted
20.00
10.00
0.00
Primary Care
Surgery
Internal Medicine
Pediatric Subspecialties
Other Medical
(95% Confidence Intervals indicated by the error bar)
Center for Healthcare Policy and Research, UC Davis
40
Results Con’t
 Table 4 presents 41 specialty rankings.
 Highest statistically significant coefficients indicated that the wages for
physicians in neurological surgery, radiation and medical oncology,
dermatology, orthopedic surgery and ophthalmology reported $17 to $50
more per-hour than physicians in general surgery.
 Specialties with the lowest wages included internal medicine and
pediatrics (combined), internal medicine, family medicine, general
practice, other pediatric subspecialties, general practice, geriatric
medicine, and clinical care internal medicine ( -$18 to -$24).
Center for Healthcare Policy and Research, UC Davis
41
Table 4
Linear Regression Results for Specialties Adjusted for Covariates, with Wage as Dependent Variable
Covariate
Regression
Coefficient
Lower 95%
Confidence Limit
Upper 95%
Confidence Limit
1.
Neurological Surgery
50.39***
14.69
86.08
2.
Radiation Oncology
42.65***
17.92
67.38
3.
Medical Oncology
31.66*
-4.69
68.01
4.
Plastic Surgery
29.26*
-2.72
61.25
5.
Allergy & Immunology
23.15*
-4.06
50.37
6.
Thoracic Surgery
22.70**
4.09
41.32
7.
Dermatology
21.54***
5.95
37.12
8.
Orthopedic Surgery
20.42***
5.31
35.53
9.
Ophthalmology
17.02**
1.69
32.35
10. Hematology & Oncology
15.28
-16.52
47.07
11. Other Surgical Subspecialties
13.41
-10.95
37.76
P-value legend: ***<=0.01; **=(0.01-0.05); *=(0.05-0.10)
a. F-test for entire regression was statistically significant, p-value <0.01. All control variables in Table 3---age through
revenue from managed care--- also included in this regression, but omitted in the interest of brevity.
Center for Healthcare Policy and Research, UC Davis
42
Table 4 Con’t
Linear Regression Results for Specialties Adjusted for Covariates, with Wage as Dependent Variable
Regression
Coefficient
Lower 95%
Confidence Limit
Upper 95%
Confidence Limit
12. Physical Medicine & Rehabilitation
11.18
-22.15
44.51
13. Neurology
8.73
-12.60
30.07
14. Gastroenterology
7.68
-7.36
22.72
15. Otolaryngology
7.17
-13.49
27.83
16. Cardiovascular Diseases
6.37
-8.60
21.34
17. Emergency Medicine
5.35
-7.58
18.28
18. Vascular Surgery
3.37
-19.31
26.06
19. Obstetrics & Gynecology
2.96
-9.19
15.11
20. Occupational Medicine
2.23
-11.38
15.84
Referent
Referent
Referent
22. Urology
-0.19
-17.59
17.21
23. Nephrology
-2.00
-18.28
14.28
24. Neonatal & Perinatal Medicine
-6.01
-19.88
7.87
Covariate
21. General Surgery
P-value legend: ***<=0.01; **=(0.01-0.05); *=(0.05-0.10)
a. F-test for entire regression was statistically significant, p-value <0.01. All control variables in Table 3---age through
revenue from managed care--- also included in this regression, but omitted in the interest of brevity.
Center for Healthcare Policy and Research, UC Davis
43
Table 4 Con’t
Linear Regression Results for Specialties Adjusted for Covariates, with Wage as Dependent Variable
Regression
Coefficient
Lower 95%
Confidence Limit
Upper 95%
Confidence Limit
25. Hospitalists
-7.14
-20.50
6.23
26. Rheumatology
-7.64
-26.80
11.52
27. Endocrinology
-7.82
-26.19
10.55
28. Psychiatry
-8.54
-20.69
3.62
29. Pulmonary Critical Care Medicine
-8.89
-35.22
17.44
30. Child & Adolescent Psychiatry
-12.05
-27.03
2.92
31. Pediatric Emergency Med.
-12.81
-43.26
17.64
Covariate
P-value legend: ***<=0.01; **=(0.01-0.05); *=(0.05-0.10)
a. F-test for entire regression was statistically significant, p-value <0.01. All control variables in Table 3---age through
revenue from managed care--- also included in this regression, but omitted in the interest of brevity.
Center for Healthcare Policy and Research, UC Davis
44
Table 4 Con’t
Linear Regression Results for Specialties Adjusted for Covariates, with Wage as Dependent Variable
Regression
Coefficient
Lower 95%
Confidence Limit
Upper 95%
Confidence Limit
32. Pediatrics
-12.96*
-26.96
1.05
33. Pulmonary Diseases
-14.01*
-28.81
0.79
34. Critical Care Internal Medicine
-18.14**
-33.81
-2.47
35. Infectious Diseases
-19.51**
-34.45
-4.57
36. Geriatric Medicine
-19.55**
-35.15
-3.95
37. General Practice
-21.94**
-38.89
-4.99
38. Other Pediatric Subspecialties
-23.44***
-38.24
-8.64
39. Family Practice
-23.70***
-33.87
-13.53
40. Internal Medicine
-24.27***
-35.30
-13.25
41. Internal Medicine & Pediatrics
-24.36***
-38.34
-10.38
Covariate
P-value legend: ***<=0.01; **=(0.01-0.05); *=(0.05-0.10)
a. F-test for entire regression was statistically significant, p-value <0.01. All control variables in Table 3---age through
revenue from managed care--- also included in this regression, but omitted in the interest of brevity.
Center for Healthcare Policy and Research, UC Davis
45
Results Con’t
 A rank correlation coefficient was calculated that compared rankings
before and after control variables were added to the regressions, i.e.
Table 2 versus Table 3 rankings of the 41 specialties. The coefficient
was 0.98 (p<0.001) suggesting few differences in rankings.
Center for Healthcare Policy and Research, UC Davis
46
Discussion
 Debate surrounds disparity in pay among physician specialties,
particularly between primary care physicians and specialists.
 Both annual income and Resource-Based Relative Value Units
(RBRVUs) have been employed as indicators of physician remuneration.
 Neither income nor RBRVUs directly account for work hours and weeks
worked.
 Our wage rankings may provide more complete picture than previously
available.
Center for Healthcare Policy and Research, UC Davis
47
Discussion Con’t
 Our analyses of four broad groups of specialties, and after adjusting for
confounders, found wages of physician specialists roughly 36% to 48%
higher than primary care.
 Our disaggregated adjusted analyses of 41 specialties, we found
specialties with statistically higher-than-average wages either perform
surgery (neurologic, orthopedic, ophthalmologic), deploy sophisticated
technologies (radiation oncology), or administer expensive drugs in
office settings (oncology).
 Lower paid specialties largely non-procedural (i.e., cognitive) as they
rely on talking to and examining patients.
 Major exception is critical care internal medicine.
Center for Healthcare Policy and Research, UC Davis
48
Discussion Con’t
 Our study’s data and design can be compared to others.
 Until 2003, American Medical Association (AMA) published annual
Physician Socioeconomic Statistics bulletin which contained information
on income across specialties.
 Using AMA data from 1993-1998 for income , Dorsey found income
ranking for broadly defined specialties: orthopedic surgery ($323,000),
urology ($245,000), otolaryngology ($242,000), general surgery
($238,000), ophthalmology ($225,000),obstetrics and gynecology
($224,000), dermatology ($221,000), emergency medicine ($183,000),
neurology ($172,000), internal medicine ($158,000), pediatrics
($138,000), psychiatry ($ 134,000), and family practice ($132,000).
 For these same specialties we found urology to be much lower and
dermatology to be somewhat higher.
Center for Healthcare Policy and Research, UC Davis
49
Discussion Con’t
 Bodenheimer used data from the Medical Group Management
Association (MGMA) to rank numerous specialties as well as a broadly
defined primary care group (family practice, internal medicine and
pediatrics) and a specialties group (14 specialties, including
obstetrics/gynecology).
 In 2004, the specialties group income was 83.5% higher than the primary
care income group.
Center for Healthcare Policy and Research, UC Davis
50
Discussion Con’t
 Bureau of Labor Statistics(BLS) used large data sets to estimated income
for a handful of specialties.
 BLS ranking for 2005 was: general surgery ($282,504), obstetricsgynecology ($247,348), psychiatry ($180,000), internal medicine
($166,420), pediatrics ($161,331), and family practice ($156,010).
 Our ranking similar, but places obstetrics-gynecology ahead of the others
on this short BLS list.
Center for Healthcare Policy and Research, UC Davis
51
Discussion Con’t
 Our findings broadly consistent with these prior studies and surveys.
 But each of these studies and surveys has limitations.
 Most recent AMA data are from 2001.
 Dorsey data from 1990s.
 Medical school salaries (MGMA) unlikely to represent great majority
physicians not employed in medical schools.
 The BLS only produces estimates for a handful of specialties.
 Physician salary surveys that do not contain a larger scientific purpose
may not representative.
 Few studies adjust for more than a few of the covariates.
 We are not aware of any that directly account for work hours and
annual weeks of work.
Center for Healthcare Policy and Research, UC Davis
52
Discussion Con’t
 The low relative remuneration of generalist physicians been observed
in several prior studies.
 Relative positions among specialties were similar in our ranking to
those in rankings based on annual income.
 Magnitude of differences between generalist and other specialties
somewhat smaller in our study.
 Our wage (earnings-per-hour) analyses resulted in substantially lower
remuneration ranking for general internal medicine than have prior
annual income-focused analyses.
Center for Healthcare Policy and Research, UC Davis
53
Discussion Con’t
 Some might interpret our findings as reflecting labor markets
responding to supply and demand.
 However, market forces constrained in regulated physician environment.
 Additionally, experts projected substantial shortfalls in numbers of
primary care physicians in the near future.
 Earlier research repeatedly demonstrates that perceived remuneration
is significant predictor of medical student specialty choice.
 Our findings suggest that legislators, health insurance administrators,
medical group directors, health care plan managers and executives,
residency directors, and health policymakers should consider taking
action to increase incomes and/or reduce work hours for specialties near
the bottom of our wage ranking list, particularly generalist specialties.
Center for Healthcare Policy and Research, UC Davis
54
Discussion Con’t
 Limitations
 CTS excluded radiologists, anesthesiologists, and pathologists and our
results thereby compress the gap between generalists and subspecialties.
 Data self-reported. However, virtually all large, nationally representative
data sets rely on self-reported income and hours.
 Data cross-sectional, precluding confident assertions regarding causal relations.
 We considered longitudinal CTS, but ultimately rejected that strategy
realizing that too few physicians would change specialties.
 Moreover, only 29% of sample available for all three earlier rounds.
 Survey response rates may differ across specialties.
 Overall response rate is 53%.
 However, CTS administrators believe these data are representative of
physicians in the nation.
Center for Healthcare Policy and Research, UC Davis
55
Discussion Con’t
 Work hours variable limited.
 Captured only total hours per week, not variability across day, swing,
or night shifts or for weekends or weekdays.
 Nor did we allow hours-on-call to be included in work hours.
 We could only adjust for hours worked, not intensity of work (which
has numerous components including patients seen per hour).
 Hours variable included administrative tasks and professional duties.
 Future researchers may want to limit hours to only direct patient care,
but we felt all medically related activities were relevant.
Center for Healthcare Policy and Research, UC Davis
56
Discussion Con’t
 A few specialties in sample had few incumbents, including pediatric
emergency medicine (n = 29), thoracic surgery (n = 20), critical care
internal medicine (n = 23), hematology and oncology (n = 21), and
neurological surgery (n = 22).
 Caution should be exercised when interpreting results for these specialties.
 Despite possible limitations, worth noting prior researchers have found
these data to be reliable.
 Additionally, our specialty rankings are broadly consistent with those in
studies of physician income.
 Our results on many covariates (age, gender, rural/town) are also
consistent with those of other studies both inside and outside medicine.
Center for Healthcare Policy and Research, UC Davis
57
Discussion Con’t
 Question of whether obstetrics/gynecology should be regarded as
primary care is controversial.
 In main analysis we excluded obstetrics/gynecology from the broad
category of primary care.
 However, in sensitivity analyses, we found that moving
obstetrics/gynecology to primary care modestly increased the disparity
between primary care and general surgery.
 Marginally decreased the disparity between primary care and the other
two broad categories.
 More importantly, however, we agree with Rosenblatt and Fink that
obstetrics/gynecology does not represent the full spectrum of primary care.
Center for Healthcare Policy and Research, UC Davis
58
In Conclusion
 Previous studies compared annual income across specialties.
 Wage (earnings-per-hour) comparisons provide more complete picture
 Cross-sectional analysis of data from 6,381 physicians in the 2004-2005
Community Tracking Study .
 Tobit and linear regressions were run.
 Four broad specialty categories (primary care, surgery/surgical
subspecialties, internal medicine and pediatric sub-specialties, and
other) as well as 41 specific specialties were analyzed together with
demographic, geographic, and market variables.
Center for Healthcare Policy and Research, UC Davis
59
Conclusion Con’t
 Analyses on broad categories, wages for surgery, internal medicine and
pediatric sub-specialties, and other specialties were 48%, 36%, and 45%
higher, than for primary care specialties.
 Analyses for 41 specific specialties, wages for the following were
significantly lower than for the reference group, general surgery (with
wage near median, $85.98/hour): internal medicine and pediatrics
(combined) (-$24.36), internal medicine (-$24.27), family medicine
(-$23.70) and other pediatric subspecialties (-$23.44).
 Primary care-specialty wage gap was likely conservative due to
exclusion of radiologists, anesthesiologists, and pathologists.
 In light of low and declining medical student interest in primary care,
these findings suggest the need for payment reform aimed at increasing
incomes and/or reducing work hours for primary care physicians.
Center for Healthcare Policy and Research, UC Davis
60
Kaiser News
 Little-Known AMA Group Has Big Influence On Medicare
Payments
 By Joe Eaton, Center For Public Integrity, Oct 27, 2010
Center for Healthcare Policy and Research, UC Davis
61
Kaiser News
 Early this month, a group of 29 doctors gathered in Chicago.
 Over four days, the little-known group of mostly specialists made
a series of decisions crucial to Medicare.
 Issuing recommendations for precisely how Medicare should
value more than 200 different medical procedures.
Center for Healthcare Policy and Research, UC Davis
62
Kaiser News
 Members of organization waded through technical discussions ranking
procedures by how much time, skill, and mental effort they required.
 More than 100 invitation-only consultants, lawyers, and medical society
representatives hunched over their laptops taking notes.
 Chicago group’s decisions weigh heavily on how much Medicare pays doctors.
 But members of this powerful panel are not employed by the federal
government.
 Instead, group comprised almost exclusively of physicians themselves,
the very people who have the most to win or lose based on how Medicare
values the work they perform.
Center for Healthcare Policy and Research, UC Davis
63
Kaiser News
 Known as American Medical Association/Specialty Society Relative
Value Scale Update Committee, or RUC.
 The group is unknown to much of the medical profession.
 Yet for almost two decades, the committee has had a powerful influence
on Medicare payment rates.
 Since 1991, the RUC has submitted more than 7,000 recommendations
to the Centers for Medicare and Medicaid Services (CMS) on the value of
physician work.
 CMS has overwhelmingly rubber-stamped RUC recommendations.
 Accepting more than 94 percent, according to AMA numbers.
Center for Healthcare Policy and Research, UC Davis
64
Kaiser News
 That record, critics say, means CMS is handing over some of its payment
policy decisions to a physician organization with a massive and obvious
conflict of interest.
 The arrangement has been criticized by bodies like the Government
Accountability Office and MedPAC, the independent agency that advises
Congress on Medicare issues.
 But no group has cried out louder than family doctors, who say CMS
reliance on the secretive group has undervalued face-to-face
consultation between doctors and patients while overvaluing expensive
high-tech medical procedures and imaging.
Center for Healthcare Policy and Research, UC Davis
65
Kaiser News
 In response to proposed federal rules for verifying medical procedure values,
The American Academy of Family Physicians suggested in July that CMS find
experts outside of the RUC to collect and analyze evidence for validating values.
 It suggested looking for experts who are “less invested financially in the outcome.
 The AMA argues that the committee is nothing more than an independent
group practicing its First Amendment right to petition the federal government.
 But the AMA does not try to downplay the RUC’s goal.
 An AMA physicians'’ guide puts it this way: “From the AMA’s perspective, the
RUC provides a vital opportunity for the medical profession to continue to
shape its own payment environment.”
Center for Healthcare Policy and Research, UC Davis