Quality of care delivered by general internists in US hospitals who graduated from foreign versus US medical schools: observational study
BMJ 2017; 356 doi: https://doi.org/10.1136/bmj.j273 (Published 03 February 2017) Cite this as: BMJ 2017;356:j273
All rapid responses
Rapid responses are electronic comments to the editor. They enable our users to debate issues raised in articles published on bmj.com. A rapid response is first posted online. If you need the URL (web address) of an individual response, simply click on the response headline and copy the URL from the browser window. A proportion of responses will, after editing, be published online and in the print journal as letters, which are indexed in PubMed. Rapid responses are not indexed in PubMed and they are not journal articles. The BMJ reserves the right to remove responses which are being wilfully misrepresented as published articles or when it is brought to our attention that a response spreads misinformation.
From March 2022, the word limit for rapid responses will be 600 words not including references and author details. We will no longer post responses that exceed this limit.
The word limit for letters selected from posted responses remains 300 words.
We are grateful to Chadha et al. for comments to our recent article on the differences in patient outcomes between U.S. and international medical graduates.1 They point out that USMLE pass rates are lower for international medical graduates (IMGs) than for U.S. medical graduates (USMGs). However, what we compared in our study was USMGs with IMGs who were successful in passing USMLE exams and the matching process and became board-certified physicians in the U.S., which is different from comparing USMGs with IMGs who are taking USMLE exams. Chadha et al. also raised a concern about the differences in hospitals where IMGs and USMGs practice and physician characteristics. However, these concerns have been addressed in our study. We used a method called “hospital fixed effects”, which adjust for all characteristics of hospitals as long as they do not vary over time, allowing us to effectively compare physicians within the same hospital.2-4 Therefore, there was no need to adjust for individual hospital characteristics (e.g., teaching status of hospitals) in our study. We also adjusted for many patient characteristics including Diagnosis-related groups (DRGs).
Peer commented about how we attributed patient outcomes to physicians. We used three different methods: (1) assigning to physicians who accounted for the largest Part B spending, (2) assigning to physicians who billed the largest number of evaluation-and-management [E&M] claims, and (3) assigning to physicians who billed the first E&M claim for a given hospitalization. And we found that our findings were not affected by the approach we used. In our main analyses, we assigned to physicians who accounted for the largest Part B spending in our main analyses because a similar approach has been used in previous studies to attribute patients to Accountable Care Organizations (ACO).5 6 There is no single way to attribute patient outcomes to physicians, and that is why we believe that the assessment of different attribution rules is critically important to test the robustness of our findings.
The discharging physicians could be different from assigned physicians in our study. We examined the impact of physicians who played a major role in the treatment of the patient for a given hospitalization, because our focus was the quality of care provided during the hospitalization, and not the quality of discharge planning or quality of care provided after discharges. Although we agree that there are many different ways to measure the quality of hospital care, both 30-day mortality and readmission rates are widely-accepted standard measure of quality of hospital care, which the Centers for Medicare and Medicaid Services use to evaluate quality and reimburse hospitals.7 8
Nathanson et al. suggested that different ways to adjust for potential confounders may be better. We agree that there are multiple methods for adjusting for the illness severity of patients, but there is no single method that is substantially better than others. We adjusted for age, sex, race, MS-DRG, Elixhauser comorbidity index, median household income, and Medicaid coverage, which we believe accounted for major potential confounders. We also used a natural experiment by focusing on patients who were treated by hospitalists, who are “quasi-randomized” to hospitalists based on their hospitalists’ schedule, to address the issues of unmeasured confounders.1 We used this approach in other studies as well,9 10 and we believe that our approach may account for confounders better than using the 3M propriety APR DRG model which cannot address unmeasured confounders. Moreover, our data showed that IMGs were, in general, treating sicker patients, and therefore, it is likely that, if any unmeasured confounders do exist, they were biasing our estimates toward the null (i.e., the mortality difference may be larger if we could adjust for all confounders).
We agree with Chowdhury and Payal that the fact IMGs practice in resource-poor settings may be one of the causal mechanisms why they could deliver high-quality care. As we wrote in our article, we also agree that what we are observing is not due to the differences in medical education between the U.S. and elsewhere, but instead, observed differences in patient outcomes can be explained by the rigorous selection processes for the IMGs, which is what we need to keep in our mind when interpreting these findings.
Mummadi argues that we assumed that most of the intensivists are likely to be IMGs. However, what we explained in our article was “international graduates might be more or less likely to work as intensivists in intensive care units”, and we were not making an assumption that IMGs are more likely to be intensivists. It is reasonable to assume that intensivists treat sicker patients than hospitalists, and if the probability of becoming intensivists differ between IMGs and USMGs, that would introduce a bias in our estimates, which was what we addressed in our sensitivity analysis.
Sakurai commented that most 2-group comparisons would result in significance by traditional statistical analyses with “big data.” However, this is true only if there is a true difference between two groups. If the true patient mortality rates were the same between the two groups, no matter how large the sample size, we would not observe statistical significance (except for the 5% false positive that applies to all statistical tests regardless of size). The use of big data does not introduce a bias in our estimate in any way – all it does is to reduce the noise and make it easier to detect the signal (even the small signal) from the data. In other words, when we increase the number of observations, the estimate we get will be closer to the true parameter, the concept known as “consistency” in statistics. The caveat of interpreting the results we get from analyzing the big data is that because of the high-precision of the estimates, we can detect a difference between two groups even if the difference was too small that it does not have any clinical significance. We also explained in our article why we think that the difference we observed is as of clinical significance. This difference in mortality between IMGs and USMGs was comparable to the reduction in mortality we have seen between 2000 and 2010 across US hospitals.11
References
1. Tsugawa Y, Jena AB, Orav EJ, et al. Quality of care delivered by general internists in US hospitals who graduated from foreign versus US medical schools: observational study. BMJ 2017;356:j273. doi: 10.1136/bmj.j273
2. Allison PD. Fixed effects regression methods for longitudinal data using SAS: Sas Institute 2005.
3. Gunasekara FI, Richardson K, Carter K, et al. Fixed effects analysis of repeated measures data. Int J Epidemiol 2014;43(1):264-9. doi: 10.1093/ije/dyt221
4. Kaufman JS. Commentary: Why are we biased against bias? Int J Epidemiol 2008;37(3):624-6. doi: 10.1093/ije/dyn035
5. McWilliams JM, Hatfield LA, Chernew ME, et al. Early Performance of Accountable Care Organizations in Medicare. N Engl J Med 2016;374(24):2357-66. doi: 10.1056/NEJMsa1600142
6. McWilliams JM, Landon BE, Chernew ME, et al. Changes in Patients' Experiences in Medicare Accountable Care Organizations. N Engl J Med 2014;371(18):1715-24.
7. Centers for Medicare & Medicaid Services. Outcome Measures [updated September 29, 2015. Available from: https://www.cms.gov/medicare/quality-initiatives-patient-assessment-inst... accessed February 28 2017.
8. Centers for Medicare & Medicaid Services. Hospital Value-Based Purchasing Washington DC2015 [Available from: https://www.cms.gov/Outreach-and-Education/Medicare-Learning-Network-MLN... accessed March 1 2017.
9. Tsugawa Y, Jena AB, Figueroa JF, et al. Comparison of Hospital Mortality and Readmission Rates for Medicare Patients Treated by Male vs Female Physicians. JAMA Intern Med 2017;177(2):206-13. doi: 10.1001/jamainternmed.2016.7875
10. Tsugawa Y, Jha AK, Newhouse JP, et al. Variation in Physician Spending and Association With Patient Outcomes. JAMA Intern Med 2017 doi: 10.1001/jamainternmed.2017.0059
11. Hall MJ, Levant S, DeFrances CJ. Trends in Inpatient Hospital Deaths: National Hospital Discharge Survey, 2000–2010. NCHS data brief, no 118. Hyattsville, MD: National Center for Health Statistics, 2013.
Competing interests: No competing interests
Tsugawa et al. argue that the current standards for selecting international medical graduates to practice in the United States of America are sufficiently rigorous. One of the key study findings supporting this conclusion is that older Medicare patients treated by international graduates had a lower mortality rate compared to those treated by United States graduates (11.2% vs 11.6%, adjusted odds ratio 0.95, 95% confidence interval [0.93, 0.96], p < 0.001, n = 1,215,490 patients). The authors touch on the “at most modest clinical significance” of this finding in the discussion. However, in the abstract and “what this study adds” section, the authors instead reaffirm their claim that patients cared for by international graduates had lower mortality rates than those by US graduates. While such a statement is technically correct, I argue this conclusion inadvertently confuses statistical significance for clinical significance and is potentially misleading.
With a sample size over 1 million patients, almost any observed differences in the study will be statistically significant. For example, a mortality rate difference of just 0.1% will likely be statistically significant and the corresponding p-value will likely be <0.001, again simply because of the sample size of the study. In fact, if the authors computed mortality rates for the other variables in their logistic regression model (male vs female physicians holding all other variables constant, older vs younger physicians holding all other variables constant, etc…), I suspect all these differences will be statistically significant too. Would the authors then report that patients from one gender of physicians have a higher mortality rate than those from the other?
To better appreciate the significance of the observed difference in mortality rates presented in the paper, a supplementary table should be provided showing calculated mortality rates between the other variables in the logistic regression model (physician sex, physician age, etc…). Variables with the largest difference in mortality rates should be prioritized. If the difference in mortality rates is highest between international and US graduates, that would provide evidence justifying the authors’ claim. However, if a handful of other variables have a similar or higher difference, that would cast additional doubt on the authors’ claim.
I agree with the fundamental conclusion of the article in that the current standards for selecting international medical graduates to practice in the United States of America are sufficiently rigorous. International medical graduates play a vital role in providing much needed healthcare services to millions of Americans, and I believe they will be increasingly important in the years to come. However, it is not necessary to claim that international graduate patients have a lower mortality rate to support this argument. Given the potentially controversial nature of such a claim, additional results should be made available so the results provided in the paper can be better interpreted.
Competing interests: No competing interests
"Data on older Medicare patients admitted to hospital in the US showed that patients treated by international graduates had lower mortality than patients cared for by US graduates."
Even though the results are statistically significant, we as IMGs (International Medical Graduates) still find it conflicting. Yes, the thought process is good, since United States attracts top performers from other countries. However, one would wonder if the gaps are due to differences in approach between the countries.
"Patients treated by international graduates had lower mortality (adjusted mortality 11.2% v 11.6%; adjusted odds ratio 0.95, 95% confidence interval 0.93 to 0.96; P<0.001) and slightly higher costs of care per admission (adjusted costs $1145 (£950; €1080) v $1098; adjusted difference $47, 95% confidence interval $39 to $55, P<0.001)." Evaluating healthcare expenditure in GDP (Gross Domestic Product), United States seems to have the highest amount devoted to healthcare, almost 3-7 times of most countries. Amazingly, most countries other than United States and Mexico have less than 1 physician per 1000 people.
Country, Health expenditure of GDP in % for 2014{1}, Physicians per 1,000 people for 2011{2}
USA- 17.14%, 2.45
Mexico- 6.30%, 2.09
Egypt- 5.64%, not reported
China- 5.55%, 1.49
Philippines- 4.71%, not reported
India- 4.68%, 0.73
Nigeria- 3.67%, not reported
Syria- 3.25%, not reported
Pakistan- 2.61%, not reported
Now, the higher cost is hard to explain given the fact that during medical school IMGs see more patients with less resources than AMGs(American Medical Graduates).
There could be many arguments for and against lower mortality seen in case of IMG Hospitalists. The differences are hard to believe, given all physicians got their training at a United States based residency program. On the other hand, medical school training, accreditation and board certification process for IMGs is largely different for their medical degree.
If we look at USMLE pass rates from 2001 to 2015 {3}; the IMGs have done worse than AMGs as first test takers or repeaters. Although, there is a trend of IMGs getting better slowly. One can probably conclude, that test taking skill for IMGs is getting better. This does indicate that the methods of teaching, learning and content of both might be different. Authors quote from references 16-18 about the inconsistent data regarding scores seem arguable. Two of the references are from non-United States health systems. The third reference is regarding in-service exams, the comparison would be mostly relevant for the first year residents only.
The difference in mortality and cost seem debatable based on below mentioned observations.
Table 1: The mean ages of IMGs and AMGs were 46.1 and 47.9 respectively; bringing us to an assumption that on average the evaluated physicians had about 10 years of medical practice experience after residency training. It is hard to believe that education and training 13 years ago would have any significant impact on current practice.
Table 3 and Table 4: The model 2 and 3 do account for patient volumes and compare hospitalists within same hospital, but Case Mix Index (CMI) or Diagnosis Related Groups (DRG) do not seem to be accounted for.
Table 1: Larger hospitals had primarily AMGs and medium size hospitals had higher percentage of IMGs. The acuity of patients is usually higher in large hospitals. These characteristics do not seem to be adjusted for.
Table 1: Majority of teaching was done by AMGs and majority of non-teaching practice was with IMGs. It is not clarified if resident teams were carrying sicker patients. These characteristics if so did not seem to be adjusted for.
As per 2014 data of Young et al {4}, majority of IMGs (23%) were from India. There have been some concerning publications {5-7} about medical education system in India. Which would bring us to the question of a largely flawed medical education system producing superior practicing physicians. This would make us wonder that the current practice of IMGs and AMGs could be a reflection for their residency training. Another plausible explanation could be post residency practice trends like, community versus academic medicine.
In the end, we do not believe in assigning people in buckets based on gender, religion, race, nationality, ethnicity, caste or creed. One of the recent publication by same author correlated the effect of gender on the patient mortality and readmissions {8}. If we pick race or religion, we will again find some small statistically significant but clinically insignificant differences. We need further studies adjusting for a number of variable to understand these trends.
References:
{1} http://data.worldbank.org/indicator/SH.XPD.TOTL.ZS?end=2011&start=2011&v...
{2} http://data.worldbank.org/indicator/SH.MED.PHYS.ZS?end=2011&start=2011&v...
{3} http://www.usmle.org/performance-data/
{4} Young et al. A Census of Actively Licensed Physicians in the United States, 2014. Journal of Medical Regulation Vol 101 No 2 Page 8-23
{5} Jain et al. Corruption: medicine’s dirty open secret. BMJ 2014; 348
{6} Jocalyn Clark. Indian medical education system is broken, Reuters investigation finds. BMJ 2015; 350
{7} Nagral et al. A radical prescription for the Medical Council of India. BMJ 2016; 352
{8} Tsugawa et al. Comparison of Hospital Mortality and Readmission Rates for Medicare Patients Treated by Male vs Female Physicians. JAMA Intern Med. 2017;177(2):206-213.
Competing interests: No competing interests
This is the second recent study by Tsugawa et al, using similar methodology to evaluate individual inpatient physician care (1). In both studies the physician who billed the highest dollar amount, the “assigned physician”, is assumed to be responsible for 30 day readmissions and mortalities. Yet neither study offers a reference or data to support this assumption.
Although readmissions have been, controversially, used as a surrogate for hospital system quality indicators I am unaware of any study showing that readmissions reflect an individual inpatient physician’s care. Previous studies have concluded that most readmissions are due to unmodifiable causes (2). And of the modifiable factors that could have been changed during hospitalization, resolution of the main problem, adequacy of post discharge destination, stability of doses of therapy and appropriate timing of the first follow up visit were the highest predictors of readmission (2). These are only partly under inpatient physician control, and the discharging physician is most responsible for these elements. Were the discharging physicians the same as the “assigned physician” in this study?
Multiple studies have showed the value of outpatient care and care coordination in prevention of hospitalizations and readmissions for Medicare patients (3) (4). Yet this study seems to ascribe all of the difference in readmissions to the “assigned” inpatient physician.
I also question the use of patient death as an indicator of “assigned physician” care unless the patient died during the index admission while being treated by the “assigned physician”. The same unmodifiable and modifiable factors involved in readmissions would be expected to apply to mortalities. And we should expect that outpatient care would have an influence on mortality rates (5).
The authors have made adjustments for many possible patient and hospital variables. But was the “assigned physician” the physician most responsible for patient care? Were the number and expertise of specialists involved a factor in the results? It would be difficult to determine the effects of these variables from this study.
Gary W. Peer MD
Department of Medicine, Stanford Health Care Valley Care, Pleasanton CA
Conflict of Interest Discloser: None
1) Tsugawa Y, Jena AB, Figueroa JF, Orav EJ, Blumenthal DM, Jha AK. Comparison of Hospital Mortality and Readmission Rates for Medicare Patients Treated by Male vs Female Physicians. JAMA Intern Med. Published online December 19, 2016.
2) Benbassat J, Taragin M. Hospital Readmissions as a Measure of Quality of Health Care Advantages and Limitations. Arch Intern Med. 2000;160(8):1074-1081.
3) Jack BW, Chetty VK, Anthony D, Greenwald JL, Sanchez GM, Johnson AE, et al. A Reengineered Hospital Discharge Program to Decrease Rehospitalization: A Randomized Trial. Ann Intern Med. 2009;150:178-187
4) Kasper E , Gerstenblith G, Hefter G, Van Anden E, Brinker A, Thiemann D, Terrin M, Forman S, Gottlieb S. A randomized trial of the efficacy of multidisciplinary care in heart failure outpatients at high risk of hospital readmission. Journal of the American College of Cardiology. Volume 39, Issue 3, 6 February 2002, Pages 471–480
5) Fonarow G, Albert N, Curtis A, Gheorghiade M, Heywood JT, Liu Y, Mehra M, O'Connor C, Reynolds D , Walsh M and Yancy C. Associations Between Outpatient Heart Failure Process-of-Care Measures and Mortality.
Circulation. 2011;123:1601-1610.
Competing interests: No competing interests
It is with great interest that we read Tsugawa et al’s recent article on quality of care among international versus US trained physicians.1 The authors should be commended for an ambitious and novel study. While we agree with their conclusions that differences between international and US trained physicians were modest and suggest the two groups have equivalent outcomes, their study does warrant additional comments. First, the authors had limited ability to adjust for patient acuity. For example, lab values and patient level treatments (eg, inotropes, pain medications, dialysis etc) seem absent. Furthermore, severity of illness was not defined and while the Elixhauser comorbidity index was used, the use of comorbid conditions may not truly define illness severity unless they used 3M propriety APR DRG model for this analysis. Consequently, unmeasured confounding may explain the small, but statistically significant results favoring internationally trained physicians. Similarly, when the authors conducted two sensitivity analyses that tried to make the patient populations more homogeneous: examining subgroups by disease states and excluding hospitals with ICUs, both showed either predominantly non-significant results or results where the confidence intervals are very close to 1. This further implies the two groups perform equally well rather than internationally trained physicians’ confer a survival benefit to their patients.
This study assumes that the attending physician’s medical school training is a key driver of patient outcome. However, the medical care of hospitalized patients is truly interdisciplinary and it is unknown how similar supporting staff were for each physician in the study, even within the same hospital. Moreover, costs were slightly higher among the internationally trained physicians. Could it be that internationally trained physicians are more likely to do additional tests or more likely to seek a consult from a specialist? Additional study into the practice patterns of physicians by international versus US medical schools is needed to answer these questions.
This study also brings into light the underlying issue of free market medicine in the US with many foreign graduates making lifelong career choices to remain within internal medicine/general practice due to opportunity limitation, immigration issues, inferior working conditions and geographical limitations as these markets are less financially lucrative. To this point, as a group, internationally trained doctors may simply have greater experience caring for medical inpatients which produces good outcomes. Finally, we stress that these results may not be generalizable to other physician specialties like surgeons or intensivists.
1. Tsugawa Y, Jena AB, Orav EJ, Jha AK. Quality of care delivered by general internists in US hospitals who graduated from foreign versus US medical schools: observational study. bmj. 2017 Feb 3;356:j273.
Competing interests: No competing interests
Dear Editors,
Is there concern that international medical graduates (IMGs) deliver lower quality care because they were trained in developing countries? Yes. Is the concern justified? Apparently not, if we go by the evidence presented by Tsugawa et. al. in their recent study, “Quality of care delivered by general internists in US hospitals who graduated from foreign versus US medical schools: observational study.”[1] In fact, their analysis shows that IMGs deliver better outcomes than their domestic counterparts.
What makes this finding remarkable is that IMGs tend to treat patients with higher rates of chronic conditions than the patients treated by US graduates. The authors present compelling reasons that might explain this finding. We would like to add to those explanations.
There are certain advantages to practicing in resource-poor settings, as many IMGs do. The paucity of investigational tools in such settings probably encourages IMGs to rely more on their clinical skills than on diagnostic investigations. This may help them become better skilled as clinicians.
Patients in developing countries face barriers to access medical care and may be more amenable to be treated by physicians-in-training or novice medical graduates. This may result in IMGs gaining more ‘breadth’ of clinical exposure during their primary medical training, compared to US graduates. Perhaps this initial ‘breadth’ prepares them better for the rigors of postgraduate training. In developed countries, compared to developing ones, different patient care standards and relatively higher prevalence of patient-initiated litigation make hands-on training scarce for medical undergraduates.
Many US graduates also train as scientists. Many more undergo formal research training through programs like the master of public health (MPH). This usually implies time spent away from the clinic. In contrast, a medical graduate undergoing doctoral or research training in addition to medical school and residency is highly unusual in countries like India and its neighbors. Thus, while IMGs tend to get longer duration of clinical training, it may be the case that US graduates get a broader set of skills at the expense of more clinical training. The utility and long-term benefits of a physician work-force trained in research as opposed to being trained for longer in clinical medicine is a matter of debate and further research.
As Tsugawa et al[1] point out, the current approach for selecting IMGs to residencies in the US may select for, on average, better physicians. Additionally, these physicians are likely to be only those who can afford the substantial fees, travel and associated costs, time, and resources needed for the prolonged licensure examinations (two out of four parts of the exam are conducted only in the US) and interview process.[2 3] The better physicians among IMGs who cannot afford all this are naturally excluded from the process. Thus the comparison seems to have been between US graduates and a subset of the better physicians among IMGs. This should be taken into account while interpreting the results.
The quality of undergraduate medical training is just one of many variables impacting the outcomes in inpatient and hospital care; much work remains to be done to quantify the others. The authors are to be congratulated for their detailed analysis - a limited, but very substantive step towards bridging the evidence gap.
References
1. Tsugawa Y, Jena AB, Orav EJ, et al. Quality of care delivered by general internists in US hospitals who graduated from foreign versus US medical schools: observational study. BMJ 2017;356 doi: 10.1136/bmj.j273[published Online First: Epub Date]|.
2. Chen PG-C, Curry LA, Bernheim SM, et al. Professional Challenges of Non-U.S.-Born International Medical Graduates and Recommendations for Support During Residency Training. Academic Medicine 2011;86(11):1383-88 doi: 10.1097/ACM.0b013e31823035e1[published Online First: Epub Date]|.
3. Boulet JR, Norcini JJ, Whelan GP, et al. The international medical graduate pipeline: recent trends in certification and residency training. Health affairs (Project Hope) 2006;25(2):469-77 doi: 10.1377/hlthaff.25.2.469[published Online First: Epub Date]|.
Competing interests: No competing interests
The United States is a country of a particular strength because of its ability to attract exceptional expertise and individuals with outstanding skills from around the world. This is true for the healthcare system as well as for the tech industry, science, and engineering, among others.
The January 27, 2017 presidential executive order resulting in the ban on immigration from seven Muslim-majority countries challenges this ability and hinders the inclusion of some of this international expertise in the healthcare system as well as the advancement of medical science and research.
International medical graduates (IMGs) have been vital against doctor shortage for a very long time. According to a 2014 U.S. census by the Federation of State Medical Records (FSMB), out of the 916,264 physicians with active license, 207,840 (22.7%) were IMGs.1 They also comprised 25% of all active residents and fellows in programs accredited by the Accreditation Council for Graduate Medical Education (ACGME) during the Academic Year 2015-2016 (31,095 out of 124,409).2 In the most recent (2016) residency Match program, 21% of all applicants (7,460 out of 35,476) were IMGs.3 During the first year the NRMP and ECFMG started to jointly report this data in 2013, this percentage was even higher at more than a third (36.8%)4. We used the Match Program Data reports5 to plot the size of the IMG applicant pipeline into both residency and fellowship over all the years the NRMP has published this data (2007-2016, see Figures). This shows that the proportion of IMGs who successfully matched has been a significant one, and also fairly constant over the past decade.
These applicants come from all around the world, but a considerable number of them come from the seven countries specified in the ban. There are currently over 10,000 licensed physicians in the US who graduated from these seven countries, including 1,800 physicians-in-training enrolled in ACGME accredited residency and fellowships programs6. In 2013, a total of 668 IMGs from these countries applied to the Match, and 258 of them successfully matched (38.6%). University of Damascus in Syria, in particular, was among the top 10 international medical schools that contributed IMGs to the U.S. residency programs.1
Pursuing medical training in the U.S. is a daunting journey, considerably more difficult for IMGs. They would have to satisfy higher standards than their U.S. graduate counterparts. This is observed across almost all specialties, and includes higher USMLE exam scores, requiring more research experience and publications, other degrees (eg, PhD), and sometimes even having completed a residency program elsewhere.3 7 This already puts these IMGs at a disadvantage of having to take longer to achieve those qualifications after graduation. There are currently 260 Match applicants from these seven specified countries.8 However, despite acquiring this higher level of competency, those applicants will now face a complete shutdown of their ambitions of a medical training in the US and being able to contribute positively to the care of patients.
The notion that foreigners are taking over work opportunities that Americans are more deserving of has been used many times in defense of this immigration and visa ban. We argue that, for the healthcare work force, this is certainly not true. IMGs have long been filling a large number of positions in U.S. residency programs that cannot be filled by U.S. graduates. For example, in the academic year 2014-2015, there were 21.7% more open residency positions than there were U.S. graduates. Additionally, although this figure is projected to decrease due to the opening of new medical schools, it will remain a high figure of 13.5%.9 Top physician-supply experts who conducted the analysis and made these projections argued that for the past 50 years, US medical graduates have enjoyed a "selection subsidy" which made matching to their favorable program an easy process. For IMGs, on the other hand, it was a much more difficult process that included additional exams, many legal and immigration hurdles, and eventually to settle for places and programs less favorable by US graduates. This includes working in rural areas, where doctor shortages are most severe.6 10-12 Therefore, despite increases in the number and size of US medical schools and Medicare caps on GME funding, projections through 2023-2024 suggest that there would still be a substantial number of available GME positions than US graduates. If anything, these projected changes will make the process of obtaining a residency position for IMGs even more competitive. All these hurdles in the face of IMGs ensure that those of them who successfully make it into joining a residency program are "crème de la crème". This is reflected by how IMGs outperform US graduates in some aspects, such as how patients with congestive heart failure and acute myocardial infarction who were cared for by IMGs had lower mortality than those cared for by US graduates.13 Caring for elderly patients by IMG hospitalists also showed lower mortality than those cared for by U.S. graduated hospitalists, despite the former group of patients having more chronic conditions than the latter.14
The executive order shuts down the entry of these bright minds into the US. It puts those who have applied this year in danger of rejection without considering their qualifications. It also threatens those already serving in medical training programs of collapse of their careers by losing their positions should they need to leave the country. Needless to say, this will have devastating consequences on their programs which would need to scramble to cover the shortages. Patients cared for by these skillful and capable hands would suffer their share as well, giving that about 900,000 patients are currently receiving medical care by the physicians-in-training from the seven specified countries alone.6
References
1. Aaron Young PHJC, DO, MS; Xiaomei Pei, PhD; Katie Halbesleben, PhD; Donald H. Polk, DO; and Michael Dugan, MBA. A Census of Actively Licensed Physicians in the United States, 2014. Journal of Medical Regulation 2014;VOL 101(2):8-23.
2. FORGING AHEAD - ACGME ANNUAL REPORT. http://www.acgme.org/Portals/0/PDFs/2015-2016-ACGMEAnnualReport.pdf: Accreditation Council for Graduate Medical Education, 2015-2016.
3. Charting Outcomes in the Match for International Medical Graduates - Characteristics of International Medical Graduates Who Matched to Their Preferred Specialty in the 2016 Main Residency Match. 2nd ed. http://www.nrmp.org/wp-content/uploads/2016/09/Charting-Outcomes-IMGs-20... National Resident Matching Program (NRMP), 2016.
4. Charting Outcomes in the Match International Medical Graduates - Characteristics of Applicants Who Matched to Their Preferred Specialty in the 2013 Main Residency Match. 1st ed. http://www.nrmp.org/wp-content/uploads/2014/01/NRMP-and-ECFMG-Publish-Ch... National Resident Matching Program (NRMP) and Educational Commission for Foreign Medical Graduates (ECFMG), 2014.
5. NRMP HISTORICAL REPORTS http://www.nrmp.org/match-data/nrmp-historical-reports/2007-2016 [Available from: http://www.nrmp.org/match-data/nrmp-historical-reports/ accessed February 2 2017.
6. Thomas J Nasca M, MACP. Second Letter to GME Community on Immigration Policy - ACGME. http://www.acgme.org/Portals/0/PDFs/Nasca-Community/Nasca-Letter-Immigra..., 2017.
7. Charting Outcomes in the Match for U.S. Allopathic Seniors Characteristics of U.S. Allopathic Seniors Who Matched to Their Preferred Specialty in the 2016 Main Residency Match. 1st ed. http://www.nrmp.org/wp-content/uploads/2016/09/Charting-Outcomes-US-Allo... National Resident Matching Program (NRMP), 2016.
8. American College of Physicians Issues Comprehensive Statement on U.S. Immigration Policy. https://www.acponline.org/acp-newsroom/acp-comprehensive-statement-us-im... America College of Physicians (ACP) 2017.
9. Mullan F, Salsberg E, Weider K. Why a GME Squeeze Is Unlikely. New Engl J Med 2015;373(25):2397-99.
10. Recruitment of Non-U.S. Citizen Physicians to Rural and Underserved Areas through Conrad State 30 J-1 Visa Waiver Programs - Rural Health Research Center. http://depts.washington.edu/uwrhrc/uploads/RHRC_FR148_Patterson.pdf, 2015.
11. Fink KS, Phillips RL, Jr., Fryer GE, et al. International medical graduates and the primary care workforce for rural underserved areas. Health Aff (Millwood) 2003;22(2):255-62.
12. (ACP) ACoP. The Role of International Medical Graduates in the U.S. Physician Workforce. Philadelphia: American College of Physicians, 2008.
13. Norcini JJ, Boulet JR, Dauphinee WD, et al. Evaluating the quality of care provided by graduates of international medical schools. Health Aff (Millwood) 2010;29(8):1461-8. doi: 10.1377/hlthaff.2009.0222
14. Yusuke Tsugawa ABJ, Ruth L Newhouse, E John Orav, Ashish K Jha, K T Ki. Quality of care delivered by general internists in US hospitals who graduated from foreign versus US medical schools: observational study. British Medical Journal (BMJ) 2017;356:j273. doi: https://doi.org/10.1136/bmj.j273
Competing interests: No competing interests
The authors state the most of the intensivists are likely to be International Medical Graduates and therefore, ICU patients were excluded from the analyses. While this could be true, support for this statement is not provided with a retrievable reference.
Competing interests: No competing interests
Dear Editors
This research was written to add to the base of knowledge and "evidence" involving the performance of international medical graduates (IMGs) compared to US graduates (USG) in the US health system. Many will use this in support of the continuing presence of IMGs in the US amidst the controversy of the latest political administration involving immigration from several countries of origin.
It is surprising how such research can be done without prejudice. The authors attempted to justify their methods as follows:
"First, to address the possibility that the international graduates might treat patients with lesser or greater unmeasured severity of illness, we restricted the study population to patients treated by physicians who specialize in the care of patients admitted to hospital (“hospitalists”) (we focused on general internists …… for main analyses, and hospitalists for the sensitivity analysis). These physicians typically work in shifts, and therefore, within the same hospital, patients treated by them are plausibly quasi-randomized to a given physician based on that hospitalist’s work schedule."
There is a selection bias while trying to find a study population of elderly patients in hospitals employing general internists with rotating shifts and a substantial IMG presence in the hospital. It skews the study towards certain facility types; as the authors themselves admit, there is evidence of clustering of IMG employment within the US health system ("international graduates tend to practice in hospitals with higher readmission rates").
"Second, since there are multiple ways to assign physicians to patients, we tested two alternative methods to assign physicians to patients: assigning physicians who had the largest number of evaluation and management claims, and assigning physicians who billed the first evaluation and management claim for a given hospital admission."
It is uncertain if there is a difference between USGs and IMGs in claims submission and evaluation, which would then result in different outcomes in certain types of conditions.
Given that rotating shifts are involved, we are unsure if there is a difference in distribution of day-time vs night-time cover shifts between the two; this can be manipulated by the hospital administration who may allocate shifts differentially as part of risk management.
"Third, to account for the influence of international medical graduates who were US citizens, we excluded international graduates who graduated from medical schools in Central America and the Caribbean, because three quarters of US citizen international medical students graduate from medical schools in these countries."
By such quasi-rationalisation, the spotlight is shifted to IMGs who graduate from medical schools in non-American continents. If we are to differentiate performance of graduates from medical school based in the US compared to those outside the US, there can be no reasonable justification for these exclusions.
"Fourth, since differences in length of stay, utilization of care (total part B spending per hospital admission), or discharge location might explain differences in patient outcomes between the graduates, we further adjusted our regression models for these variables."
IMGs are linked with higher rates of readmission and higher costs of care; can this be due to difference in management style--for example, cost shifting or transfer of care to others (in other hospitals) when patients are deteriorating.
"Fifth, to deal with the impact of unobserved care preferences of patients, we excluded patients with cancer and patients who were discharged to hospice care."
For patients discharged to hospice care, it is uncertain if the decision to palliate is related to cultural preferences both the patient's (and the family) and the internist's.
"Sixth, as international graduates might be more or less likely to work as intensivists in intensive care units, we excluded hospitals with a medical intensive care unit."
Excluding hospitals with a medical ICU also excludes most major hospitals with tertiary referring facilities, where most critical patients are managed.
Can we extrapolate this to major referral centres or more critically ill patients? Not likely.
"Seventh, it is possible that residents bill Medicare claims on behalf of their attending physicians at teaching hospitals, and differences in patient outcomes might be related to the quality of care delivered by residents. To address this, we stratified our sample by teaching status of hospitals (major teaching, minor teaching, and non-teaching hospitals), and within each group we compared international medical graduates with US medical graduates (adjusted for patient and physician characteristics and hospital fixed effects)."
I wonder if IMGs are more likely to be supervised than US graduates, even as attendings? Being in a teaching hospital also involves more counterchecks amongst peers and more clinical and administrative protocols in place.
"Finally, we examined whether patient outcomes varied by countries where international medical graduates were trained, after restricting to eight countries with the largest number of international medical graduates going to the US (India, Pakistan, Philippines, Syria, Nigeria, Mexico, Egypt, and China) to avoid unstable estimates"
There are significant performance difference by origin within doctors of IMG categories; the reasons may be multifactorial: cultural, training, age, experience, etc. (see supplement)
The perceived divide in performance between USGs and IMGs wouldn’t be resolved easily, the authors’ attempt to discount the impact of their selection bias reinforced my opinion their interrogation inflicted on the data cannot yield meaningful results.
While IMGs represent 26% of the US physician workforce (national average), in this study they account for 44% of all doctors in this study group (and yet accounting for 54% of all patients in the study). This reflects a skewed observational study of localities which the hospital working arrangement and patient load exists neither by practical choice or design.
I believe supporters of IMGs should be cautious in interpreting the significance of conclusions and their applicability elsewhere. While I support any evidence on clinical integrity and standards of IMGs in the US health system (as I am a potential IMG myself), I am wary of assumptions made that have led to these assertions.
These are the first steps into a better review of the performance of USGs and IMGs, but the implications of observational studies like this is neither fait accompli or ejusdem generis to the entire US workforce.
Competing interests: No competing interests
Re: Quality of care delivered by general internists in US hospitals who graduated from foreign versus US medical schools: observational study
I have been following this article for a while now and the other responses are interesting as well.
This article comes at a time when the world is holding dangerous views against people who don't share similar viewpoint and xenophobia is rampant!
Ever since I came to the UK, I have yet to be a part of a gathering that I have been to that highlights how much IMGs contribute to this country's healthcare! Some NHS trusts have got supportive groups for IMGs but one always hear the study figures and numbers, which show the higher number of complaints and disciplinary actions taken against IMG doctors in comparison to UK medical graduates. Even though I found the UK a friendly nation for a foreigner, positive reviews for non-UK doctors are not as common as they might be expected to be, given the large population of IMG doctors.
It is also interesting that often this fact is underestimated that IMGs have passed the licensing examination of the country where they are working as well having had experiences of working in varied healthcare systems and being more flexible and openminded in their approach.
As an IMG doctor in the UK, I appreciate the support NHS trusts provide for the IMG trainee doctors but such studies definitely boost morale. I hope this study will help to change the viewpoint.
Competing interests: No competing interests