The findings of this study support but do not confirm the hypothesis that the repatriation of Lebanese physicians educated abroad during the post-war 1990s decade has contributed to the international emigration of recent Lebanese medical graduates. In fact, the 1990s have witnessed the following concurrent trends: (1) a dramatic increase in the number of new registrants in the early 1990s; (2) about half of the physicians registered with the Lebanese Order of Physicians joined it during the 1990s decade; (3) only a quarter of those who joined the 1990s decade graduated from Lebanese medical schools during that decade; (4) a "dip" in the percentage of Lebanese graduates of the early 1990s registered with the two Orders as of 2007; (5) a "bump" in the number of Lebanese medical graduates registered as residents in the US in the early 1990s . These conclusions are obviously weakened by the potential confounding by other factors such as the civil disturbances in the late 1980s.
Figure 1 suggests three sources for the growth of the number of registered physicians: graduates of Western Europe who may have returned after the end of civil war (post-war 1990s decade); graduates of Eastern Europe whose numbers appear to have increased after the 1990s dissolution of the Soviet Union and the liberalization of medical recruitment; and the increasing numbers of Lebanese medical graduates (steady increase from 99 in 1977 to 288 in 2006 with a small drop in the late 1990s; unpublished data).
There is an apparent discrepancy between the increasing number of Lebanese graduates joining the workforce in the 1990s and the "dip" in the percentage of Lebanese graduates of the early 1990s registered as of 2007. This could be explained by that most Lebanese graduates joining the workforce in 1990s likely graduated in previous decades and had been in migrant status before their return. Their increasing influx from other countries consequently pushed Lebanese graduates of the early 1990s to get out of Lebanon.
This study has a number of strengths. First, it covers a time span of three decades allowing the depiction of long term trends for the number of new and total registered physicians and for physician density. Second, it analyzes the characteristics of all physicians registered in Lebanon. Third, the two Orders collect the data directly from registrants and require official documentation (e.g., diplomas, certificates) to verify the educational credentials, which strengthens the validity of the data. Finally, we were not able to identify published literature on a similar phenomenon of repatriation of physicians driving the emigration of recent graduates.
This study has a number of limitations. First, the evidence we identified is indirect. Second, registration with the two orders of physicians may not be a highly accurate measure of actual membership of the physician workforce. Physicians who permanently or temporarily reside abroad might have kept their Lebanese registration active. Additionally, some physicians might be illegally practicing in Lebanon, especially in remote areas, without being registered. We expect these numbers to be very low, however, given the requirement to be registered to bill major payers such as the National Social Security Fund. Third, physicians may opt to delay their registration until they complete their residency training; it is thus not clear whether a delayed registration versus increasing emigration explains the relatively low percentage of registration by Lebanese medical graduates of recent years (Figure 5). Fourth, the variable "year of registration" (used in the 3rd set of analysis) was not available for those registered as of 2007 with one of the two Orders. However these constituted only 12% of the total number of registered physicians. Finally, because of the nature of the administrative data, we did not plan to study the "oversaturation of the physician job market" suggested by the qualitative study and focused instead on the growth of the number of physicians. However, the leveling off of the physician density since the early 2000s (Figure 4) supports the suggestion that the market is oversaturated .
While the findings are consistent with earlier reports of a dramatic growth of the workforce in the early 1990s , they show a slowing of this growth since the early 2000s. However registration data do not clarify whether the reduction in registration rates is related to a slowing on the supply end (i.e. decreased repatriation of immigrant physicians or fewer Eastern European graduates entering the workforce) or an acceleration of emigration out of the workforce, or both. Nevertheless, there is indirect evidence that an accelerated emigration is contributing to this phenomenon. In a 2005 survey of graduating Lebanese medical students, 96% intended to travel abroad either for a specialty (78%) or a subspecialty (18%) training . It is well known that residency training abroad is a critical step in the migration of physicians . In addition, only 35% of survey respondents intended to return to Lebanon directly after finishing training abroad.
The Lebanese physician workforce has failed to assimilate about 40% of the country medical graduates of the last three decades. This however is not a new phenomenon. Kronfol et al. found that in 1974, 49% of the 1935–1974 medical graduates of the American University of Beirut, a Lebanese university with cultural and educational ties to the US, were in the US . In 1984, 70% of graduates of the same university were in the US . The author attributed this sharp increase in immigration to the US on the civil war of that time.
The rapid growth of the number of physicians can lead to a number of serious workforce problems besides emigration. These problems include unemployment, employment in other fields, overuse of services and illegal practice. Unfortunately, we were not able to locate any studies assessing these issues.
There are two major concerns with the shift in the workforce demography. First, the "oversaturation of the market" is motivating Lebanese medical graduates to emigrate , and is likely discouraging migrant Lebanese physicians in other parts of the world to consider resettling in Lebanon. This would likely exacerbate Lebanon's brain drain problem leading to losses in educational costs and returns on investment as described for other brain drained countries . Second, it is believed that the certification process of medical graduates entering the Lebanese workforce is not rigorous enough to filter out those who received low quality training abroad, which could affect the quality of care in Lebanon.
Most registered physicians who completed specialty training did so abroad, mostly in Eastern and Western Europe. Specialty training abroad was associated with, among other factors, the region of medical school being abroad, particularly Eastern Europe. This observation raises the question of why those graduating from Eastern European medical schools are not completing residency training in Lebanon, A possible answer could be related to a problem of availability and quality of residency training in Lebanon, as students of Lebanese medical schools have suggested .
Northern Africa, Eastern Europe, and Western Europe as the regions of medical school and Western Europe as the region of specialty training were associated with age and male sex. Because subjects being analyzed are those currently registered in Lebanon, it is hard to judge whether those who are older (and thus graduated earlier from medical schools) or those who are male were more likely to complete specialty training abroad, or were more likely to return to Lebanon after completing specialty training in Western Europe, or both. It is however likely that females are in a weaker socio-economic position to travel abroad for medical education.
It is important to note that while non-Lebanese citizens can enter and graduate from a Lebanese medical school, they can't register or practice medicine in Lebanon. This implies that repatriating physicians would all be Lebanese citizens unlike emigrating Lebanese medical graduates. However, the implications of this phenomenon should be minimal; in a recent survey non-Lebanese citizens constituted less than 6% of students of Lebanese medical schools .