o
R
G
N
A
L
COM
M
U
N
C
A
T
o
N
Physician-Patient Race Concordance from the Physician Perspective Alan E. Simon, MD; Jill A. Marsteller, PhD, MPP; Susan X. Lin , DrPH
Disclaimer: The findings and conclusions in this paper are those of the authors and do not necessarily reftect the views of the Centers for Disease Control and Prevention, Financial Disclosure: The authors have no financial relationshipsrelevant to thisarticle to disclose
Background: The benefits of racial/ethnic physician'patient concordance have been cited to support increasing the number of minority physicians, Few studies have examined the rates at which physicians of different race/ethnicity groups or specialties see concordant visits, We aim to determine whether differences exist in rates at which physicians of different race/ethnicity groups and physician specialties see visits by patients of concordant race/ethnicity, Methods: We used data from the NationalAmbulatory Medical Care Survey, 2001-2006, a nationally representative survey of visits to priva te physician's offices, For physicians of each race/ethnicity group, the percentage of visits by pa tients in each race/ethnicity group was calculated, A concordant visit was defined as one in which a physician in a particular race/ethnicity group saw a patient of the same race/ethnicity group, Concordance rates were calculated overall. and for visits to primary care, medical specialties, and surgical specialties individually, Results: White physicians see a higher percentage of concordant visits than any other race/ethnicity of physician (84.3%, p0,05 for comparison), with non-Hispanic Asian physicians having the lowest rate of concordant visits (14,5%, p
J Noll Med Assoc, 2013; 105: 150-156,
Author Affiliations: National Center for HealthStatistics, Centersfor Disease Control and Prevention, Hyattsvill e, MD, USA (Dr Simon): Department of Health Policy and Management, Johns Hopkins Bloomberg School of Publi c Health, Baltimore, MD, USA (Dr Marsteller): Center for Family and Community Medicine, Columbia University Medical Center, New York, NY, USA (Dr Lin)
150 VOL, 105, NO, 2, SUMMER 2013
Correspondence: Alan E, Simon (Corresponding Author), Medical Officer/Senior Service Fellow, National Center for Health Statistics, 33 11Toledo Road, Rm 6122, Hyattsville, MD, 20782, Phone: (301) 458-4338, Fax: (30 1) 458-4038 (fpa8@cdc,gov): Jill Marsteller, Associate Professor, Department of Health Policy and Management, Johns Hopkins, Bloomberg School of Public Health, 624 N, Broadway, Rm, 433, Baltimore, MD 21205-1900, Phone: (410)6 14-2602 limarstel@jhsph,edu): Susan X, Lin, Assistant Professor, Center for Family and Community Medicine, Columbia University Medical Center, 100 Hoven Ave" ApI. 27C, New York, NY 10032, Phone: (212)304-7032 (xI18@columbia,edu)
IN TRODUCTIO N
R
acial or ethnic concordance, meaning simply that the race or ethnicity of a physician and a patient are the same, has been examined as a possible factor leading to differences in health care quality and patient satisfaction, Findings from studies on the benefits to patients of concordant relationships are mixed. ' -8 Despite debate concerning the benefits of concordance, it has been shown to be preferred by some patients9 and the possible benefits of concordant relationships have been cited by the report of the Sullivan Commission (Missing Persons: Minorities in the Health Profession) as a rationale for improved minority recmitment to increase the numbers of minority physicians. 'o The Sullivan Commission on Diversity in the Healthcare Workforce, comprised of leaders in healthcare, was formed to examine the reasons for a lack of minorities in the health care workforce. 10 Recommendations such as these have supported policies that further this aim, such as increased educational loans for individuals from disadvantaged backgrounds, as stipulated in the Patient Protection and Affordable Care Act. " However, significant gaps exist in our understanding of physician-patient race concordance. Most notably, concordance has generally been assessed from the patient perspective. 12.1 3 That is, what percentage of patients of each race/ethnicity has a doctor of the same race/ethnicity? Little research, however, has examined whether physicians in different racel ethnicity groups see visits by patients of the same
JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
PHYSICIAN PERSPECTIVE OF RACE-CONCORDANCE
race/ethnicity at different rates. Examination of concordance from the physician perspective may be important to understanding the effects of minority recruitment on concordance. Concordance rates are likely affected by numerous factors, including local and national supply of minority physicians, area demographics, physician choice in location of practice, physician acceptance of insurance, and patient choice of physician. These factors, and others, create the rate at which physicians in different minority race/ethnicity groups see patients oftheir own race/ethnicity group. If differences in this rate exist between physicians of varying race/ethnici ty groups, concordance will be affected differentially by recruitment into the medical profession of different minority groups. Similarly, to our knowledge, all national studies of concordance to date have focused on concordance rates for primary care physicians. Studies have not examined concordance for surgeons or medical specialists. If concordance rates in medical and surgical specialties differ from those in primary care, recruitment into different areas of medicine may also have varying effects on concordance rates achieved. This study aims to determine if variation exists between physicians of different racial/ethnic groups and different specialties in how often they see patients of the same race/ethnicity. Examining concordance rates from the perspective of the physician may provide insight into the changes in concordance rates that one might expect as the numbers of physicians froln different racial/ethnic groups increase in different areas of medicine.
METHODS We conducted a secondary data analysis of the National Ambulatory Medical Care Survey (NAMCS), 2001-2006. The NAMCS is a nationally representative annual probability sample survey of visits to office-based physicians in the United States conducted by the National Center for Health Statistics (NCHS). The NAMCS uses a three-stage probability sampling procedure, with geographic areas as the first stage, physicians as the second stage, and visits to a sampled physician during a one week reporting period as the third stage. Visits, rather than patients (or number of individual people), are the unit of observation. Physicians are selected from the master files of the American Medical Association (AMA) and the American Osteopathic Association (AOA)
JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
after stratification by physician specialty. NAMCS does not sample anesthesio logists, radiologists, or pathologists. Sample weights were used to make national estimates, and weights were adjusted for non-response. We conducted all analyses in STATA using SVY commands to adjust for the weighting and complex sample design of the survey. We combined data from 200 I to 2006 to provide a sufficient sample size among minority groups and to improve the reliability of the data. There were no significant differences between 200 I and 2006 data in the concordance rates of visits, of any racelethnicity group, overall or in any of the three medical specialty areas. Physician race and ethnicity in the NAMCS are determined by physician self-report as provided on the AMA and AOA master files while patient race is reported by the physician or abstracted from the medical chart by office staff or a federal abstractor. Both physicians and patients were classified into the following racelethnicity categories: non-Hispanic white, non-Hispanic black, non-Hispanic Asian, Hispanic (any race), and Other. We classified physician visits into three physician-type categories: primary care, medical specialties, and surgical specialties based on the classification categories specified by the AMA in 1999,1 4 and updated by the National Center for Health Statistics since that time for use with the NAMC S survey (Table I). Patient racelethnicity was missing for 26.4% of visits and physician race/ethnicity was missing for 26.5% of visits. Complete data were available for 55.8% ofvisits, ora total of84 , 174 visits. We examined the distribution of missing cases and found no statistically significant difference in the missing rates of patient race/ethnicity among visits to physicians of different race/ethnicity categories. Similarly, we found no statistically significant difference in the missing rates of physician racelethnicity among visits by patients of different race/ethnicity categories. Thi s was true for visits to all physicians collectively, and for visits to physicians in each of the three specialty types, individually. Missing categories of both physician racelethnicity and patient racelethnicity were eliminated from the analyses to best represent the true percentages of visits in each category. We calculated concordance rates from the perspective of physicians of different racelethnicity groups. Among visits to all physicians in each racel ethnicity group, we calculated the percent of visits by
VOL. 105, NO. 2, SUMMER 2013
151
PHYSICIAN PERSPECTIVE OF RACE-CONCORDANCE
patients in each race/ethnicity category, both overall, and within each physician specialty area. Estimates were considered unreliable if they were based on a sample size of less than 30 cases or the relative standard error of the estimate was greater than 30%. All estimates presented had relative standard errors less than 50%. Ninety-five percent confidence intervals were calculated for each estimate. Statistical comparisons of estimates were conducted using Wald tests, with p<0.05 considered significant. No adjustments were made for multiple comparisons. As a sensitivity analysis , imputed values for patient race and ethnicity provided on the NAMCS file were included in an additional analysis. Physician race has never been imputed by NCHS. This yielded 106,128 observations and a missing rate of3l. 7%. For 20012002, race, but not ethnicity, was imputed by NCHS. For 2003-2006, both race and ethnicity were imputed by NCHS. Prior to 2006, imputations of race and ethnicity were conducted in the NAMCS file using a "cold deck" single imputation approach which draws donor values from the previous year 's data based on matching physician specialty, geography, and ICD-9 codes of primary diagnosis. 15. 16 In 2006, NAMCS used a multi-round imputation based matching ICD-9 of primary diagnosis at first the ZIP-code, then county, state, and regional levels until a donor observation was obtained from either the current year or the previous year 's data. 17
RESULTS The percent of concordant visits seen by white physicians was higher than any other race/ethnicity of physician (84.3%, p0.05 for comparison) . Non-Hispanic Asian physicians had the lowest rate of concordant visits (14.5%, p0.05 for both comparisons) (Table III). Within medical specialties, concordance rates were higher for nonHispanic black physicians (49.3%) than for Hispanic
Table I: AMA Physician Type Classification Categories
152
Primary care
General practice, family practice, internal medicine, adolescent medicine, pediatrics , sports medicine, obstetrics and gynecolog y, combined internal medicine and pediatrics , maternal fetal medicine, geriatric medicine
Medical sub-specialties
Both adult and pediatric w here appropriate: dermatology, psychiatry, critical care medicine, neonatology, perinatology, neurology, allergy/ immunology, cardiology, endocrinology, infectious disease, pulmonology, emergency medicine, gastroenterology, hematology/oncology, nephrology, rheumatology, toxicolog y, reproductive endocrinology, preventive medicine, hepatology, genetics, nutrition, occupational medic ine, palliative medicine, physical medicine and rehab, sleep medicine
Surgical specialties
General surgery (abdominal , colon, rectal), urology, ophthalmology, hand surgery, otolaryngology (head/neck), orthopedics (adult reconstructive, foot and ankle, pediatric, sports, spine, and trauma) , gynecology (critical care, oncology) , cardiovascular/cardiothoracic surgery, plastic surgery, neurosurgery, surgical oncology, thoracic surgery, transplant surgery, trauma surgery, vascular surgery, critical care surgery, pediatric surgery, pediatric otolaryngology, and pediatric ophthalmology
VOL. 105, NO. 2, SUMMER 2013
JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
PHYSICIAN PERSPECTIVE OF RACE-CONCORDANCE
Table II: Percent visitsof each patient race/ethnicity group for each physician roce/ethnicity group [95% Confidence Intervals] and (number of sampled visits)
PATIENTS
Non-Hispanic white
Non-Hispanic black
Asian / Pacific Island er
Hispanic (all ra ces)
Non-Hispanic white
84.3 [83.0,85.7] (56,804)
7.4 [6 .5, 8.2] (4,664)
1.9 [1.5,2.2] (1 ,166)
5.9 [4.8, 6.9] (3.417)
Non -Hispanic b lack
40.7 [33 .3, 48.0] (947)
46.8 [38.6, 55.0] (962)
1.6 [0 .8,2.5] (41)
10.7 [5.2, 16.3] (203)
Asian/ Pacific Islander
63.8 [58.1, 69.5] (6,582)
11.3 [8.7, 13.9] (1,059)
14.5 [10.0,19.0] (1 ,1 08)
9.3 [6 .6, 11.9] (808)
Hispanic (all races)
40.7 [31.0, 50.4] (1 .500)
7.6 [4.0, 11.2] (248)
1.2 [0.7, 1.8] (44)
50.0 [38.4, 61. 7] (1.476)
All Race/Ethnicities
77.2 [75 .2. 79.2] (67.620)
9.2 [8.3. 10.0] (7 .188)
3.4 [2.7. 4.1] (2.494)
9.4 [7 .6. 11 .3] (6.277)
PHYSICIANS
' Number of sampled visits for All Race/Ethnicity category does not equal sum of the sampled visitsamong the listed Race/Ethnicity categories because "Other" of physician race/ethnicity isnot shown. "Sum of rows does not equal 100%. as "Other" category for patient race/ethnicity is not shown.
physicians (28.1 %, p0.05). Although no statistical difference was found, point estimates for concordance rates among visits to surgical physicians were consistently lower than for primary care physicians for non-Hispanic blacks, Hispanics, and non-Hispanic Asians (p>0.05 for all comparisons) (Table III). Therefore, in order to
JOURN AL OF THE NATIONAL MEDICAL ASSOCIATION
achieve larger sample sizes we grouped non-Hispanic black, Hispanic, and non-Hispanic Asian physicians into a single minority group and non-Hispanic black, Hispanic, and non-Hispanic Asian patients into a single group of minority patients. Minority surgical and medical specialists had significantly lower rates of concordant visits (33.4%, CI: 26.7-40.0% and 33.6%, CI: 26 .6-40.5% respectively) compared to minority primary care physicians (49.5%, CI: 43.355.6%, p0.05).
VOL. 105. NO.2. SUMMER 2013
153
PHYSICIAN PERSPECTIVE OF RACE-CONCORDANCE
Table III: Percent Concordant Visits to Physician Race/Ethnicity Groups [and 95% Confidence Intervals] and (number of sampled visits]
PHYSICIAN RACE/ ETHNICITY
Non-Hispanic white
Non-Hispanic black
Asian/ Pacific Islander
Hispanic (all races)
Total
84.3 [83.0, 85.7] (56 ,804)
46.8 [38.6, 55.0] (962)
14.5 [10.0, 19.0] (1,1 08)
50.0 [38.4, 61 .7] (1,476)
Primary Care
83.0 [8l.0,84.9] (18,000)
48.2 [38.0, 58.3] (594)
16.3 [10.5,22 . 1] (589)
57.4 [44.3, 70.6] (887)
Surgical Specialty
84.7 [83.0, 86.3] (19,000)
29.3 [15.0,43.5] (153)
9.8* [3.8,15.9] (209)
42.0 [24.3,59 .6] (289)
Medical Specialty
87.2 [85 .6, 88.8] (20,000)
49.3 [34.0, 64.6] (215)
12.7* [4.8, 20.6] (310)
28 .1 [14.1, 42.0] (300)
' Estimate is not statisticall y reliable due to Relative Standard Error>30%.
DIS C USSION Examination of concordance from the physician perspective shows that physicians in different race/ethnicity groups see different percentages of concordant visits. White physicians see the highest percentage of visits by concordant patients, followed by Hispanic and non-Hispanic black physicians with similar rates of concordant visits, and non-Hispanic Asian physicians with the lowest rates of concordant visits. Within each specialty area (primary care, surgical specialties, and medical specialties), the findings have a similar pattern. Comparing across specialties, for minority physicians as a whole, concordance rates for physicians in primary care appear to be higher than for medical and surgical specialties. The reasons for variations in concordance rates by specialty area are unknown. Primary care physicians may have a greater opportunity to influence the nature of their patient populations by locating in a geographic area with a particular patient demographic, while surgical and medical specialists may have less influence since much of their business comes from referrals. Physician behavior in insurance acceptance, patient choice9 , physicians ' social and professional networks (that lead to some referrals), and changing
154
VOL. 105, NO.2, SUMMER 2013
area demographics may also affect concordance rates . Regardless of the source of the differences, the differential concordance rates among specialties from the physician perspective may have consequences for the presumed effects of minority recruitment on concordance. Recruitment of minority groups into surgical and medical specialties may result in smaller increases in concordance rates for minority patients compared to recruitment into primary care. Said another way, the same increase in the number of minority physicians in the primary care field would be expected to result in a larger increase in concordance rates for minority patients. Similarly, recruitment of non-Hispanic black and Hispanic students into medicine may result in larger increases in concordance rates for patients of those race/ethnicity groups than would be expected from recruitment of non-Hispanic Asian students into the medical field. National estimates of the percent of concordant visits seen by minority physicians are rarely reported . However, Rocheleau used a special supplement of the NAMCS in 1975 to estimate that 87% of vi sits to black physicians were made by black patients,13 higher than our current estimates of 46.8%. Specifically among primary care physicians, Komaromy used data from
JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
PHYSICIAN PERSPECTIVE OF RACE-CONCORDANCE
California in 1993 to show that black primary care physicians had slightly greater than 50% concordant visits,1 2 similar to our estimate of 48.2% for black primary care physicians. These results suggest that black physicians began seeing a more varied racial composition of patients between 1975 and 1993, potentially with little change since 1993. This study also provides data on the percent of concordant visits to non-Hispanic Asian physicians which has not previously been available in other studies. Non-Hispanic Asian physicians see concordant visits at a significantly lower rate than black and Hispanic physicians. Indeed, only 16.3% of visits to non-Hispanic Asian physicians in primary care and 14.5% of visits to non-Hispanic Asian physicians overall are concordant.
LIMITATIONS There are several limitations to this study. Most notably, limited sample size among some minority groups made some estimates unreliable and hence prevented comparisons from being made with those groups. This remained true despite combining 6 years for which NAMCS data on physician race and ethnicity were available. Also, our data use physician report of race and ethnicity for both the physician and the patient. It is unknown how physician report, as opposed to patient report, of race and ethnicity may affect concordance rates. However, it may well be that what is important to realizing the potential benefits of concordance is actually perception of concordance, from the patient perspective, the physician perspective, or both. Certainly, more detailed classification of race and ethnicity, particularly in highly heterogeneous groups such as non-Hispanic Asian, would be of great value. Although this has been a limitation of many studies that explore the effects of race and ethnicity, little progress has been made in this respect because of the limited sample sizes available for more nuanced groupmgs. Finally, the rate of missing data for patient or physician race/ethnicity is high. This is a problem in many national data sets. The percent of visits with missing patient race/ethnicity was similar across visits to physicians of different race/ethnicity groups. Likewise, the percent of physicians with missing race/ethnicity was similar among visits by patients of different race/ethnicity groups. This was true overall, and for visits to physicians in each specialty type.
JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
Further, the imputed data returned similar results. Still, we were not able to evaluate other potential biases resulting from the high level of missing data, and bias cannot be discounted.
CO NCLUS ION The evidence concerning whether, and in what situations, physician-patient racial and ethnic concordance provides benefits is conflicting and incomplete. In any case, to the extent that concordance is desired by some patients, it may be considered a matter of ensuring patients have choice among physicians and the opportunity to select a concordant physician if they wish to do so. This study adds to our understanding of the epidemiology of concordance by assessing concordance rates from the perspective of the physician. This perspective provides insight into the changes in concordance rates that might be expected as the number of physicians from different racial! ethnic groups increases in different areas of medicine. Indeed, the difference between the rates at which nonHispanic Asian physicians have concordant visits and the rates at which black and Hispanic physicians have concordant visits may suggest that recruiting of specific minority groups into the health profession will have varying effects on concordance. Similarly, that medical and surgical specialists have fewer concordant visits suggests that to reach equivalent rates of concordance in visits to surgeons and medical specialists, a greater increase in the number of minorities in these specialties may have to occur than would be needed in primary care. If this were a societal goal, minority recruitment at the medicalschool admissions level would need to be followed with recruitment efforts at the residency level. Further research could examine the reasons why differences in concordance occur among different racial and ethnic provider groups and different specialties.
REFE RE NCES 1. Laveist TA, Nuru-Jeter A. Is doctor-patient race concordance associa ted with greater sa tisfaction with care? J Health Soc 8ehav. 2002;43:296-306. 2, Saha S, Komaromy M, Koepsell TD, et 01. Patient-physician racial concordance and the perceived quality and use of health care. Arch Intern Med. 1999; 159:997-
1004. 3. Cooper-Pa trick L Gallo JJ, Gonzales JJ, et 01. Race, gender, and partnership in the patient-physician relationship. JAMA. 1999;282:583-589. 4. Cooper LA. Roter DL. Johnson RL, et 01. Patient-centered communication, ratings of care, and concordance of patient and physician race. Ann Intern Med. 2003;139;907-91 5.
VOL. 105, NO.2, SUMMER 2013
155
PHYSICIAN PERSPECTIVE OF RACE-CONCORDANCE
5. LaVeist TA. Nuru-Jeter A, Jones KE. The association of doctor-patient roce concordance with health services utilization. J Public Health Policy. 2003;24:312323.
12. Komaromy M, Grumbach K, Drake M, et 01. The role of black and Hispanic physicians in providing health care for underserved populo lions. N Engl J Med. 1996:334: t 305- 131 O.
6. Saha S, Arbelaez JJ, Cooper LA. Patient-physician rela tionships and racial disparities in the quality of health care. Am J Public Health. 2003:93: 171 3- 1719.
13. Rocheleau B. Black Physicians and Ambulatory Care. Public Health Reports 1978:93:278-282
7. Schnittker J, Liang K. The promise and limitsof racial/ethnic concordance in physician-patient interaction. J Health Polit Policy law. 2006:31 :81 1-838.
14. Pasko 1 Seidman B. PhYSician Characteristics and Distribution in the U.S" 1999 Edition. Chicago, III: American Medical Association, 1999.
8. Stevens GD, Mistry R, Zuckerman B, et 01. The parent-provider relationship: does race/ethnicity concordance or discordance infiuence parent reports of the receipt of high quality basic pediatric preventive services? J Urban Heolth. 2005;82:560-574.
15. Cherry DK, Burt CW, Woodwell DA. National Ambulatory Medical Care Survey: 2001 summary. Adv Data. 2003;337:1-44.
9. Saha S, Taggart SH. Komaromy M, et 01. Do patients choose physicians of their own race? Health Aff (Millwood) . 2000:19:76-83.
16. Cherry DK, Woodwell DA, Rechtsteiner EA. National Ambulatory Medical Care Survey: 2005 summary. Adv Data. 2007:387:1 -39.
17. Cherry DK, Hing E, Woodwell DA, Rechtsteiner EA. Nalianal Ambulatory Medical Care Survey: 2006 summary. Nat! Health Stat Report. 2008:31 -39.
10. Missing Persons: Minorities in the Health Profession, A Report of the The Sullivan Commission on Diversity in the Healthcare Workforce. Atlanta, GA: Sullivan Commission on Diversity in the Healthcare Workforce: 2004. 11 . United States Congress House Committee on Ways and Means, United States Congress. House Committee on Energy and Commerce, United States Congress House Committee on Education and Labor. Compila tion of Patient Protection and Affordable Care Act: as amended through November I, 2010 including Patient Protection and Affordable Care Act health-rela ted portions of the Health Care and Education Reconciliation Act of 2010. Washington, DC: U.S. Government Printing Office.
156
VOL. 105, NO.2, SUMMER 2013
JOURNAL OF THE NATIONAL MEDICAL ASSO CI ATION