Kales, Helen C. MD; DiNardo, Andrew R.; Blow, Frederic C. PhD; McCarthy, John F. PhD; Ignacio, Rosalinda V. MS; Riba, Michelle B. MD, MS
Dr. Kales is core investigator, Serious Mental Illness Treatment Research and Evaluation Center (SMITREC), Health Services Research and Development, VA Ann Arbor Healthcare System, Ann Arbor, Michigan; assistant professor, Department of Psychiatry, University of Michigan, Ann Arbor, Michigan; and affiliated investigator, Geriatric Research Education and Clinical Center, VA Ann Arbor Healthcare System, Ann Arbor, Michigan.
Mr. DiNardo is a medical student at Wayne State University School of Medicine, Detroit, Michigan. At the time of the study, he was research assistant at the Serious Mental Illness Treatment Research and Evaluation Center (SMITREC).
Dr. Blow is director, Serious Mental Illness Treatment Research and Evaluation Center (SMITREC), Health Services Research and Development, VA Ann Arbor Healthcare System, Ann Arbor, Michigan; and associate professor, Department of Psychiatry, University of Michigan, Ann Arbor, Michigan.
Dr. McCarthy is core investigator, Serious Mental Illness Treatment Research and Evaluation Center (SMITREC), Health Services Research and Development, VA Ann Arbor Healthcare System, Ann Arbor, Michigan; and research investigator, Department of Psychiatry, University of Michigan, Ann Arbor, Michigan.
Ms. Ignacio is research analyst, Serious Mental Illness Treatment Research and Evaluation Center (SMITREC), Health Services Research and Development, VA Ann Arbor Healthcare System, Ann Arbor, Michigan; and research analyst, Department of Psychiatry, University of Michigan, Ann Arbor, Michigan.
Dr. Riba is clinical professor and associate chair for Integrated Medical and Psychiatric Services, Department of Psychiatry, University of Michigan, Ann Arbor, Michigan; and past president, American Psychiatric Association.
Please see the end of the article for information about the authors.
Correspondence should be addressed to Dr. Kales, Psychiatry Service (116A), VAMC, 2215 Fuller Road, Ann Arbor, MI 48105; telephone: (734) 761-7926; fax: (734) 769-7410; e-mail: 〈firstname.lastname@example.org〉.
Editor’s Note: A Commentary on this report is on page 176.
International medical graduates (IMGs) account for an increasing proportion of physicians practicing in the United States, and currently make up nearly 25% of these physicians.1 In underserved specialties such as psychiatry, internal medicine, and family practice, IMGs constituted more than 30% of resident physicians in 2002–2003.2 IMGs work more often than do USMGs in shortage areas3,4 and in the public sector, where they treat higher proportions of minorities and supply up to two-thirds of the care for the poor.5 A survey of psychiatrists who treated large numbers of geriatric patients found that one-third of these physicians were IMGs.6 Thus, IMGs provide an important safety net for the current physician shortages in the United States.
To our knowledge, surprisingly little research has specifically addressed IMG status as a factor in physicians’ clinical decision making. One recent study examining how IMGs were perceived by other physicians found that a primary care physician sample (which included 12% IMGs) presented with clinical vignettes describing patients who needed specialist referral were significantly less likely to refer to IMG than to USMG specialists.7 However, no studies to our knowledge have studied mental health diagnosis by IMG physicians.
Late-life depression is an all too common disorder that affects its sufferers’ quality of life and causes considerable morbidity and mortality. The prevalence of clinically significant depressive symptoms is estimated to be as high as 20% of elderly in the community,8 and even higher numbers in medical settings (outpatient, inpatient, nursing home, etc.). The majority of elders with depression receive their treatment in primary care settings,9 where depression often has been shown to be inadequately diagnosed and treated.10,11 Physicians’ failure to diagnose and treat late-life depression is a complex issue influenced by a number of illness-, patient-, system-, and physician-level factors. While the impact of physician-level characteristics on the management of late-life depression has received limited study, no studies to our knowledge have examined the effect of U.S. versus international medical school training. Because of cultural differences in the acceptance and manifestations of mental disorders, depression may be less recognized by physicians trained in other countries than in the United States.
Given the prevalence of late-life depression and the numbers of IMGs treating elderly patients in both primary care and psychiatric settings, differences between U.S. medical graduates (USMGs) and IMGs in the diagnosis and treatment of depression would have clear implications for outcomes of late-life depression. As part of a larger project examining the effects of patients’ race and gender on physicians’ diagnosis and treatment of late-life depression,12,13 our aim in this study was to assess the effect of the location of medical school training on the diagnosis and management of late-life depression. Given the number of elderly patients seen exclusively in a primary care setting, our sample included primary care physicians as well as psychiatrists. Because of cultural differences in acceptance and manifestations of mental disorders, we hypothesized that IMGs would be less likely to diagnose and treat late-life depression than USMGs.
We developed a computerized survey instrument with several components, incorporating video-recorded interviews and text to present descriptions of late-life depression and modeled on a study by Schulman and colleagues that examined race and gender effects on treatment recommendations in cardiology.14 First, participants answered a questionnaire about their demographic characteristics. Next, they viewed a video-recorded interview with an actor portraying an older patient with depressive symptoms. Finally, a second questionnaire segment assessed their diagnosis and management recommendations for the patient and examined their judgment of patient characteristics.
The interview script for major depression was written and reviewed by four board-certified geriatric psychiatrists. Established DSM-IV (Diagnostic and Statistical Manual of Mental Disorders, fourth edition)15 criteria for major depression were met; the script was also designed to capture the frequent medical and neuropsychiatric complexity of late life depression,16 as well as to introduce ambiguity such that the diagnosis would not be entirely obvious. Thus, in addition to the depressive symptoms presented, other symptoms suggesting possible psychosis, cognitive problems, and alcohol use were included. We then piloted the written vignette individually with two board-certified psychiatrists (blind to the study purpose), both of whom diagnosed the patient with major depression.
We recruited four actors to portray the patients in the vignettes to represent four possible combinations of race (African American and white) and gender. As in the Schulman et al. study,14 we employed actors, rather than patients, because of their ability to read the scripts verbatim and to express consistent affect across the vignettes. The interviews were recorded at a single studio (Russell Video Services, Ann Arbor, Michigan) with all four actors following an identical script and direction, dressing in similar clothing, and displaying the same affect within video segments. Video recordings were produced by ImageWeaver Studios,17 a company with expertise in educational medical video production. The four video vignettes (white female, white male, African-American female, and African-American male) were then each piloted separately with at least two additional board-certified psychiatrists (12 psychiatrists in total) who were also blind to the purpose of the study; the rate of agreement on diagnosis across all observers was 93%.
The video patient interview segment was introduced by a computer screen that listed the patient’s age (73 years old), medical conditions (diabetes, arthritis, and hypertension), insurance (Medicare and Blue Cross-Blue Shield supplemental insurance), and the information that the patient was “brought in” by his or her daughter. We asked physicians to choose the single best DSM-IV diagnosis for the patient based on the information presented in the video. They were then asked their degree of certainty in this primary diagnosis. We then presented them with possible treatment choices and asked them to select their initial treatment based upon the information presented. The final portion of the survey instrument assessed physicians’ judgment about patient characteristics (modified from the Schulman et al. survey instrument), including attributes and predicted behaviors such as likelihood of compliance with treatment plans, benefit from treatment, understanding recommendations, side effects, socioeconomic status, and demeanor. The software program required that all components of the 15-minute survey (including the entire vignette) were presented to each physician prior to their answering the survey questions. The study protocol was approved by the institutional review boards at the University of Michigan (UM IRBMED) and Ann Arbor VA Health care System.
Participants and data collection
We invited attendees at the 2002 American Psychiatric Association Annual Meeting in Philadelphia, Pennsylvania, and the 2002 American Academy of Family Physicians Annual Meeting in San Diego, California, to participate in this study. We recruited participants on-site who were postresidency training and currently practicing in the United States. We told these physicians that they were taking part in a “clinical decision making” study.
Physicians were randomly assigned to view one of the four vignettes. At each meeting, the survey instrument was administered in a booth in the main meeting exhibit hall that had six individual computer stations. The layout of the stations was designed to provide privacy to the participants and to prevent them from viewing each other’s video interviews and questionnaires.
We performed the following analyses on physicians’ diagnosis, treatment recommendations, and assessments of patient characteristics based on the randomly-assigned video vignettes. First, univariate (i.e., mean, standard deviation) and bivariate (i.e., Spearman and Pearson correlations) descriptive statistics were calculated overall and by group (i.e., vignette, specialty, and medical school location). We used chi-square tests to examine differences in categorical characteristics between groups. Analysis of variance and t-tests were used to examine group differences in continuous outcome measures. We used the Tukey-Kramer method for correction of multiple comparisons to test post hoc differences among groups. For all analyses, alpha was set at 0.05 and all tests were two-sided. All analyses were performed using SAS statistical software version 8.2 (SAS Inc., Cary, North Carolina).
We conducted multivariate logistic regression analyses to assess the effects of physicians’ characteristics on their diagnosis and treatment recommendations, controlling for patient race and gender as well as physician specialty. The physician characteristics used as predictors were age, gender, race, medical school location, practice location, board certification, and years since graduation. Separate logistic regression models were estimated for each characteristic due to the high multicollinearity between the predictors.
Three hundred twenty-one psychiatrists and 178 primary care physicians volunteered to participate in the study, for a total of 499 respondents. They were randomly assigned to each patient vignette group and did not differ significantly on any of the physician characteristics we measured.
Characteristics of IMGs and USMGs
Among the entire combined physician sample, there were 353 (71%) USMGs and 146 (29%) IMGs. We compared the two groups for the following characteristics: age, gender, race, board certification, and practice location. The mean age of the two groups was comparable: 42.1 for USMGs and 43.3 for IMGs. As shown in Table 1, USMGs and IMGs did not differ on gender or practice location. However, USMGs were 2.6 times more likely to be white than were IMGs (χ2 = 148.4, p < .0001). USMGs were also 1.29 times more likely to be board certified than were IMGs (χ2 = 29.9, p < .0001). Table 2 shows further the racial composition of our study population.
By country of medical school location, there were 46 IMGs from South Asia (India, Pakistan, Sri Lanka); 27 from Southeast Asia (the Philippines, Myanmar, Indonesia, Vietnam, Thailand); 15 from Latin America (Mexico, Dominican Republic, Brazil, Argentina, Colombia); 12 from Western Europe (Belgium, Ireland, France, England, Germany, Switzerland, Italy, Greece); 11 from the Middle East (Egypt, Syria, Turkey, Israel, Iran); eight from the Caribbean (Montserrat, St. Lucia, Grenada, Dominica, Haiti); seven from Eastern Europe (Russia; Romania, Hungary, Belarus, Bosnia-Herzegovina); six from Africa (Nigeria, South Africa, Ghana, Zambia, Senegal); five from East Asia (China, Hong Kong, Japan, Korea); four from Canada; and one from Australia. Four IMGs did not provide a medical school location.
Depression diagnosis, treatment, and management
IMGs were significantly less likely to make the correct diagnosis of depression than were USMGs. Of 146 IMGs, 109 (75%) made the correct diagnosis, whereas 302 of 353 USMGs (86%) did (χ2 = 8.44, p = .0037). Among physicians who gave a correct diagnosis of depression, the mean level of certainty of diagnosis did not differ significantly between IMGs (69%) and USMGs (68.6%) (t = 0.20, df = 408, p = .8401).
We examined treatment recommendations according to physicians’ two possible recommendations for medications: either for a newer antidepressant (SSRI, SNRI, bupropion, etc., considered first-line agents for late-life depression) or other medications (tricyclics, MAO inhibitors, neuroleptics, cholinesterase inhibitors, etc.). Again, we found significant differences between IMGs and USMGs, where 100 IMGs (68%) compared to 291 USMGs (82%) prescribed treatment with a newer antidepressant (χ2 = 11.8, p = .0006). Although IMGs (54%) were less likely than USMGs (63%) to recommend follow-up for six months or longer for treating the patient’s symptoms, these differences were not significant (χ2 = 7.20, p = .07). Further analysis of diagnosis and treatment recommendations within the IMG group by physician race did not yield significant results. There were also no significant differences between vignettes (e.g., by the race or gender of the patient) in rates of depression diagnoses and treatment by IMGs compared to USMGs.
Physicians’ assessment of patient characteristics did not differ significantly between IMGs and USMGs, except in their prediction of their ability to form a treatment alliance with the patient. USMGs were significantly more likely to indicate that they would be able to form such an alliance with the patient than IMGs (t = –2.01, df = 496, p = .0449). All other patient characteristics such as likelihood of compliance with treatment plans, benefit from treatment, understanding recommendations, side effects, socioeconomic status, and demeanor were assessed by the IMGs and USMGs similarly.
Post hoc analyses
Controlling for physician specialty (psychiatry or primary care), vignette (patient race and gender), and other physician characteristics (age, gender, board certification, years since graduation, and practice location) in multivariate logistic regression models, we found that IMGs were still significantly less likely to give the correct diagnosis of depression than were USMGs (odds ratio: 0.49; 95% CI: 0.29, 0.82; c2 = 7.41, df = 1, p = .006) and significantly less likely to recommend treatment with newer antidepressants than USMGs (odds ratio: 0.52; 95% CI: 0.32, 0.83; c2 = 7.47, df = 1, p = .006).
We found significant differences for the diagnosis and treatment of late-life depression between USMGs and IMGs who viewed standardized vignettes of older patients portraying late-life depression. IMGs were significantly less likely than were USMGs to diagnose depression and recommend treatment with newer antidepressants.
Differences in diagnosis and patient management may reflect whether or not physicians are board-certified or their overall knowledge base regarding depression. In our analysis, although IMGs were significantly less likely to be board-certified than were USMGs, differences between the two groups in the diagnosis and treatment of late-life depression remained even after controlling for board-certification. In terms of knowledge base, it has been noted that one reason for the lower percentage of IMGs with board-certification may be poorer initial training in medical school.5 Undergraduate training may differ greatly around the world; emphasis may be placed on aspects of training different from the North American norm.18 Thus, it is possible that inadequate psychiatry training at the medical school level is more common at medical schools outside of the United States.
Notably, a majority of residents accepted after the National Resident Matching Program (the Match) in underserved specialties are IMGs.19 One family practice study found that residents accepted after the Match tended to perform worse in residency (scoring lower on tests and needing remedial training) than those accepted through the Match. In some cases, residents’ specialty choices may have been dictated less by a guiding interest in a particular specialty such as family practice or psychiatry, and more by the availability of an open residency position.20
However, given that many IMGs are outstanding physicians who may have had to sacrifice a great deal to be able to practice in the United States,21 another possible explanation for different rates of depression diagnosis and treatment may not lie with physicians’ abilities, but with their acculturation. Difficulties with acculturation may put IMGs at a disadvantage in how they approach psychiatry, and providing cultural competency education to IMGs in their first year of training may produce significant improvements in performance.21 Myers22 has noted that a majority of IMG psychiatry residents come from Eastern cultures, which may be culturally the most disparate from the United States with regard to regulation of interpersonal interaction. The documented rate of prevalence of mental disorders and depression in the United States is higher than international rates.23 In countries such as Japan, it has been noted that the rate of mental disorders diagnoses is significantly lower than and inconsistent with the rate of psychotropic drug prescription, suggesting that assignment of diagnosis may be lower due to stigma.24 There also may be different concepts associated with depression in other countries that might affect how IMGs are trained and eventually affect their rates of diagnosis and treatment of late-life depression. The decreased diagnosis and treatment of late-life depression by IMGs in our study could be due to decreased familiarity with depressive symptoms due to different training in or cultural paradigms of depression.
There are two main limitations to our study. First, we used video vignettes of actors portraying patients. Although we meticulously standardized and piloted the vignettes, subtle differences between the actors’ appearances or nonverbal communication might have affected our results. The use of case vignettes to assess clinical decision making25,26 and quality of physician practice27 is supported by several studies; video (rather than written) vignettes may enhance the accuracy of physician decisions.26 Such techniques also permit a degree of researcher control unattainable in observational studies, allowing determination of the influence of nonmedical factors on physician decision making.28 However, it is not known whether our survey instrument correlates with actual physician decision making. Second, we recruited physicians at national meetings, which may have resulted in a nonrepresentative sample.
These limitations notwithstanding, the results of this study provide new data on how the location of residents’ medical training may contribute to differences in the diagnosis and treatment of late-life depression. The lower rates of depression diagnosis and recommendation of gold-standard depression treatments for older patients by IMGs than by USMGs may point to inadequacies in foreign medical school psychiatry training and/or problems with acculturation, leading to difficulties in some psychiatric decision making. Residency training may provide an optimal environment for initiatives that are targeted toward competency for IMGs in cultural and aging mental health issues.
This research was supported by a Health Services Research & Development Research Career Development Award from the Department of Veterans Affairs (HCK); and the Summer Training on Aging Research Topics-Mental Health (START-MH) program (ARD).
Portions of this research were presented at the American Association for Geriatric Psychiatry Annual Meeting, March 2005, in San Diego, California.
1 Viaiano M, Makarechchi F. Globalization and the Physician Workforce in United States. Proceedings of the Sixth International Medical Workforce Conference, 25 April 2002. Ottawa, Canada: U.S. Department of Health and Human Services, Health Resources and Services Administration, Bureau of Health Professionals.
2 Brotherton SE, Rockey PH, Etzel SI. U.S. graduate medical education, 2002–2003. JAMA. 2003;290:1197–202.
3 Mick SS, Lee SY. International and U.S. medical graduates in U.S. cities. J Urban Health. 1999;76:481–96.
4 Mick SS, Lee SY. Are there need-based geographical differences between international medical graduates and U.S. medical graduates in rural U.S. counties? J Rural Health. 1999;15:26–43.
5 Blanco C, Carvalho C, Olfson M, et al. Practice patterns of international and U.S. medical graduate psychiatrists. Am J Psychiatry. 1999;156:445–50.
6 Colenda CC, Pincus H, Tanielian TL, et al. Update of geriatric psychiatry practices among American psychiatrists. Analysis of the 1996 National Survey of Psychiatric Practice. Am J Geriatr Psychiatry. 1999;7:279–88.
7 Kinchen KS, Cooper LA, Wang NY, et al. The impact of international medical graduate status on primary care physicians’ choice of specialist. Med Care. 2004;42:747–55.
8 Gallo JJ, Lebowitz BD. The epidemiology of common late-life mental disorders in the community: themes for the new century. Psychiatr Serv. 1999;50:1158–66.
9 Unutzer J, Katon W, Russo J, et al. Patterns of care for depressed older adults in a large staff model HMO. Am J Geriatr Psychiatry. 1999;7:232–43.
10 Higgins ES. A review of unrecognized mental illness in primary care. Prevalence, natural history, and efforts to change the course. Arch Fam Med. 1994;3:908–17.
11 Gallo JJ, Ryan SD, Ford DE. Attitudes, knowledge, and behavior of family physicians regarding depression in late life. Arch Fam Med. 1999;8:249–56.
12 Kales HC, Neighbors HW, Valenstein M, et al. Effect of race and sex on primary care physicians’ diagnosis and treatment of late-life depression. J Am Geriatr Soc. 2005;53:777–84.
13 Kales HC, Neighbors HW, Blow FC, et al. The effect of race and sex on psychiatrists’ diagnosis and treatment of late-life depression. Psychiatr Serv. [In press]
14 Schulman KA, Berlin JA, Harless W, et al. The effect of race and sex on physicians’ recommendations for cardiac catheterization. N Engl J Med. 1999;340:618–26.
15 American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition. Washington, DC: American Psychiatric Association, 1994.
16 Katz IR, Streim J, Parmalee P. Psychiatric-medical comorbidity: implications for health services delivery and for research on depression. Biol Psychiatry. 1994;36:141–45.
18 Bates J, Andrew R. Untangling the roots of some IMGs’ poor academic performance. Acad Med. 2001;76:43–46.
19 Blonski J, Rahm S. The relationship of residency performance to match status and U.S. versus international graduate status. Fam Med. 2003;35:100–4.
20 Pugno PA, McPherson DS. The role of international medical graduates in family practice residencies. Fam Med. 2002;34:468–69.
21 Norton J. Buy American: choosing psychiatry residents. Acad Psychiatry. 2001;25:181–83.
22 Myers GE. Addressing the effects of culture on the boundary-keeping practices of psychiatry residents educated outside of the United States. Acad Psychiatry. 2004;28:47–55.
23 Demyttenaere K, Bruffaerts R, Posada-Villa J, et al. World Health Organization World Mental Health Survey Consortium. Prevalence, severity, and unmet need for treatment of mental disorders in the World Health Organization World Mental Health Surveys. JAMA. 2004;291:2581–90.
24 Tajima O. Mental health care in Japan: recognition and treatment of depression and anxiety disorders. J Clin Psychiatry. 2001;62:39–44.
25 Wigton RJ, Poses RM, Collins M, et al. Teaching old dogs new tricks: using cognitive feedback to improve physician’s diagnostic judgments on simulated cases. Acad Med 1990;65:Suppl:S5–6.
26 McNutt RA, O’Meara JJ, de Bliek R. The effect of visual information and order of patient presentation on the accuracy of physicians’ estimates of acute ischemic heart disease: a pilot study. Med Decis Making. 1992;12:342.
27 Peabody JW, Luck J, Glassman P, et al. Measuring the quality of physician practice by using clinical vignettes: a prospective validation study. Ann Intern Med. 2004;141:771–80.
28 Feldman HA, McKinlay JB, Potter DA, et al. Nonmedical influences on medical decision making: an experimental technique using videotapes, factorial design, and survey sampling. Health Serv Res. 1997;32:343–66.