Medical and nursing students' intentions to work abroad or in rural areas: a cross-sectional survey in Asia and Africa

Abstract Objective To assess medical and nursing students’ intentions to migrate abroad or practice in rural areas. Methods We surveyed 3199 first- and final-year medical and nursing students at 16 premier government institutions in Bangladesh, Ethiopia, India, Kenya, Malawi, Nepal, the United Republic of Tanzania and Zambia. The survey contained questions to identify factors that could predict students’ intentions to migrate. Primary outcomes were the likelihoods of migrating to work abroad or working in rural areas in the country of training within five years post-training. We assessed predictors of migration intentions using multivariable proportional odds models. Findings Among respondents, 28% (870/3156) expected to migrate abroad, while only 18% (575/3158) anticipated a rural career. More nursing than medical students desired professions abroad (odds ratio, OR: 1.76; 95% confidence interval, CI: 1.25–2.48). Career desires before matriculation correlated with current intentions for international (OR: 4.49; 95% CI: 3.21–6.29) and rural (OR: 4.84; 95% CI: 3.52–6.66) careers. Time spent in rural areas before matriculation predicted the preference for a rural career (20 versus 0 years: OR: 1.53, 95% CI: 1.19–1.98) and against work abroad (20 versus 0 years: OR: 0.69, 95% CI: 0.50–0.96). Conclusion A significant proportion of students surveyed still intend to work abroad or in cities after training. These intentions could be identified even before matriculation. Admissions standards that account for years spent in rural areas could promote greater graduate retention in the country of training and in rural areas.


Introduction
Shortages of physicians and nurses jeopardize health system advances in many low-and middle-income countries (LMIC). 1,2 Sub-Saharan Africa has only 2 doctors and 11 nurses or midwives per 10 000 people, compared with approximately 30 physicians and 84 nurses or midwives per 10 000 people in high-income countries. 3,4 Emigration of health professionals from LMIC to countries with less need of such professionals exacerbates the global workforce imbalance. 1,2,5 Meanwhile, rural-to-urban migration of those professionals continues to increase provider shortages in rural areas where the need is the greatest. 6,7 To address the human resource problems in LMIC's health sector, the Global Health workforce alliance was formed in 2006 to identify, implement and advocate solutions to the crisis. 8 Then in 2010, the World Health Organization (WHO) issued a global Code of Practice that intended to curb international migration of health professionals. 9 WHO has also published recommendations for attracting, recruiting and retaining health workers in rural areas. 6 Since then, partnerships between governments, institutions and funding organizations have emerged to strengthen LMIC health workforces. [10][11][12] New medical and nursing schools have been established and existing ones expanded, 13 while curricular reforms have been instituted to enhance graduate retention. 13,14 Yet, these investments have been made without systematic analysis of the students' migration intentions. We know very little regarding the characteristics of students inclined to work in rural areas or remain in the country in which they train. 6 Without such information, resource-intensive interventions might promote training of graduates with no plans to practice in areas of need. Scholarship-bonding schemes and compulsory service obligations to work in areas with professional shortages struggle to achieve long-term retention, 15 with most of the health-care professionals leaving shortly after required service terms. 16 Greater understanding of factors associated with students' intentions to work in high-demand regions is needed, 6,17 and this could help direct admissions' policies towards selecting individuals most likely to serve in these settings long-term.
Previous studies on health worker retention are mainly from high-income countries [18][19][20] and are not applicable to LMIC. Compared to graduates in high-income countries, LMIC graduates face greater resource disparities between rural and urban settings, as well as the additional lure of providing support to family through remitted salaries earned abroad. Assessments from LMIC are limited in size and fail to compare class years, degree programmes, institutions and countries. 6 To Objective To assess medical and nursing students' intentions to migrate abroad or practice in rural areas. Methods We surveyed 3199 first-and final-year medical and nursing students at 16 premier government institutions in Bangladesh, Ethiopia, India, Kenya, Malawi, Nepal, the United Republic of Tanzania and Zambia. The survey contained questions to identify factors that could predict students' intentions to migrate. Primary outcomes were the likelihoods of migrating to work abroad or working in rural areas in the country of training within five years post-training. We assessed predictors of migration intentions using multivariable proportional odds models. Findings Among respondents, 28% (870/3156) expected to migrate abroad, while only 18% (575/3158) anticipated a rural career. More nursing than medical students desired professions abroad (odds ratio, OR: 1.76; 95% confidence interval, CI: 1. 25-2.48). Career desires before matriculation correlated with current intentions for international (OR: 4.49; 95% CI: 3.21-6.29) and rural (OR: 4.84; 95% CI: 3.52-6.66) careers. Time spent in rural areas before matriculation predicted the preference for a rural career (20 versus 0 years: OR: 1.53, 95% CI: 1.19-1.98) and against work abroad (20 versus 0 years: OR: 0.69, 95% CI: 0.50-0.96). Conclusion A significant proportion of students surveyed still intend to work abroad or in cities after training. These intentions could be identified even before matriculation. Admissions standards that account for years spent in rural areas could promote greater graduate retention in the country of training and in rural areas.

Research
Medical and nursing students' intentions to migrate in Africa and Asia David M Silvestri et al.
address this research gap, we conducted a multinational assessment of medical and nursing students' migration intentions in LMIC by surveying first-and final-year students at leading government institutions (Appendix, available from: http://biostat.mc.vanderbilt.edu/ StudentMigration).

Study Design
We considered countries in sub-Saharan Africa and south-east Asia that were classified by WHO as having a critical shortage of health service providers (less than 2.28 physicians, nurses, or midwives per 1000 population). 2 To avoid confounding effects of language on migration intentions and because most health professional emigration is to English-speaking nations, 21,22 we included only countries where English is the language of instruction. To limit the influence of conflict and political turmoil on the results, 22 we excluded countries with an active United Nations peacekeeping mission or bottom-decile rank in either the Global Peace Index or World Bank Worldwide Governance Indicators. [23][24][25] To enhance similarity between study institution governance structure and founding principles, we excluded countries without both a government medical and nursing school, or in which either school was established after 1993. This date corresponds to a period of increased attention to health sector reform in developing countries that may have affected guiding missions of health training institutions established thereafter. 26,27 Three nations meeting selection criteria were excluded due to study resource constraints, leaving eight countries: Bangladesh, Ethiopia, India, Kenya, Malawi, Nepal, the United Republic of Tanzania and Zambia (Appendix). In each country, we selected one government medical school and affiliated government nursing school in the nation's capital or major commercial city. In countries with multiple qualifying institutions, we used commonly accepted rankings to choose a highly reputed school where institutional approval could also be obtained. Affiliated nursing schools were selected to minimize confounding differences in institutional values, resources and faculty. We focused exclusively on premier government institutions, since they profess a longstanding mission to train future national leaders to address domestic health challenges and employ public funds towards this mission.
The 16 study sites selected are listed in Table 1. Research ethics committee approval was obtained from all sites and from the Vanderbilt University Institutional Review Board.

Procedures
We conducted the study from September 2011 to April 2012. Students eligible for our study were first-or final-year students enrolled in medical (Bachelor of Medicine and Surgery; Medical Doctor) or nursing (Bachelor of Science) degree programmes. At each institution, a selfadministered questionnaire was given to all eligible students attending a mandatory class lecture. Written informed consent was obtained before survey administration. Survey items assessed student background characteristics such as socioeconomic status and place of origin, attitudes towards rural and international careers and student career intentions (Appendix). Questions, derived from literature review (Appendix), consisted largely of multiple-choice items and five-point Likert scales. Surveys were in English. To minimize potential bias from variable English proficiency, official language translations were provided as an aide alongside the English questions in five countries (Bangladesh, Ethiopia, Malawi, Nepal and the United Republic of Tanzania) where English is not an official language. Translations were validated using independent backtranslation. Field testing was performed in each country and country-specific Medical and nursing students' intentions to migrate in Africa and Asia David M Silvestri et al.
modifications were made to ensure relevant terminology. The two primary outcomes were self-reported likelihood of choosing to work outside the country or in a rural setting inside the country within five years post-training. Questionnaire results were entered electronically and data entry audited by randomly selecting 22 electronic questionnaires from each country for comparison with hard copies. Audit sample size was calculated to ensure an estimated data entry error rate less than 5% with 95% confidence interval (CI). Questionnaires with more than 25% of responses missing were excluded from analysis.

Statistical analysis
For each country, we used summary statistics to describe respondent characteristics and multivariable proportional odds models to estimate the independent relationship between 14 selected characteristics (chosen before the analysis) and the likelihood to work internationally or in rural areas. The number of characteristics selected was computed using the smallest country sample size. 28 The characteristics were: degree programme, class year, sex, number of languages spoken fluently, native language, years spent in a rural setting, maternal tertiary education, first-degree extended family location, self-reported family economic status, expected number of dependents within five years of graduation, post-graduate study plans, pre-matriculation desire for rural or international careers, and country of childhood. Missing values were accounted for using multiple imputation, 29 and continuous variables were expanded using restricted cubic splines to avoid assumptions of linearity. 28 Country-specific odds ratios (ORs) were combined using the meta-analysis approach of DerSimonian and Laird, 30 a random-effects method that adjusts for heterogeneity across institutions. We used R-software version 2.15.1 (http:// www.r-project.org) for data analyses.

Results
At the 16 institutions studied, 3822 students were enrolled in first-or finalyear medical or nursing degree classes (Table 1). Of these, 3217 completed the questionnaire, a median class response rate of 87% (range: 66-100%). The primary reason for non-response was student absence during questionnaire administration; six (< 1%) students were present but declined participation. Among collected surveys, 18 (0.6%) were eliminated due to incompleteness and the remaining 3199 were included for analysis. Data entry audits yielded low error rates for all sites (range: 0.02-0.6%).
Multivariable analysis showed that nursing students were more likely than medical students to intend careers abroad (OR: 1.76; 95% CI: 1.25-2.48).
Final-year students were less likely to plan international careers than firstyear counterparts (OR: 0.83; 95% CI: 0.70-0.99; Table 4). The location of a student's extended family did not correlate with international migration preferences. However, the longer that students had resided in rural settings, the less likely they were to want to work abroad (20 versus 0 years: OR: 0.69; 95% CI: 0.50-0.96; Table 4). Sex, economic status, number of languages spoken, primary language, the mother's education and number of expected dependents were not independently associated with international migration intentions (Table 4).
Students who had desired a rural career before matriculating consistently reported having less intention of international migration (OR: 0.85; 95% CI: 0.71-1.01; Table 4), while those who had initially desired careers abroad were far more likely to want to pursue one after graduation (OR: 4.49; 95% CI: 3.21-6.29; Table 4). Final-year students who reported desiring international careers before matriculation remained nearly four times more likely to want to migrate (OR: 3.74; 95% CI: 2.25-6.21; Appendix). The desire for post-graduate specialty education also modestly increased the likelihood of wanting to seek work abroad after training (OR: 1.36; 95% CI: 0.96-1.93; Table 4). Students raised in the country of study desired international careers less than those raised abroad (OR: 0.43; 95% CI: 0.24-0.76; Table 4).

Rural career intentions
Only 18% (575/3158) of all students reported high likelihood of choosing a rural career within five years of training, including 16% (360/2257) of medical and 24% (215/901) of nursing students (Table 3). In sub-Saharan Africa, students who ruled out a rural career (29%, 609/2089) outnumbered those anticipating one (18%, 384/2089). Medical and nursing students' intentions of rural work did not differ (OR: 0.96; 95% CI: 0.76-1.22; Table 4). Final-year students were less likely to choose rural careers after training than first-years (OR: 0.67; 95% CI: 0.55-0.82; Table 4). Students who spent longer durations in a rural setting were more likely to anticipate selecting a rural practice (20 versus 0 years: OR: 1.53; 95% CI: 1.19-1.98; Table 4). Sex, economic status, number of languages spoken, primary language, the mother's education, number of expected dependents and the wish for an international career before starting studies were not independent predictors of rural career intentions.
As with international migration intentions, individuals who upon school entry had desired a rural career were now nearly five times more likely to plan one (OR: 4.84; 95%CI: 3.52-6.66; Table 4). Final-year students who had originally sought rural careers remained over three times more likely to choose one (OR: 3.26, 95% CI: 1.94-5.47; Appendix).

Discussion
Our data suggest that in nations with critical shortages of health professionals,

Research
Medical and nursing students' intentions to migrate in Africa and Asia David M Silvestri et al.
nearly a quarter of medical students and over a third of nursing students surveyed felt very likely to leave their country within five years post-training. Meanwhile, less than one fifth of students anticipated a rural career. Much of this intent appears suggested by characteristics evident even before enrolment.
Our findings extend data from LMIC that students who have spent significant time in rural settings are more likely to practice in their country and in rural areas. 6,31-33 We found new evidence that this likelihood correlated with the duration of residing in rural areas. Conversely, students raised abroad are most likely to emigrate. 34 As reported elsewhere, 33,35 sex was not independently associated with migration intentions, suggesting that women are making geographic career decisions independent of perceived family obligations. 36 Confirming other studies, 33,37 self-reported economic status did not alone predict desired career location nor did high maternal education. 37 The association between desire for post-graduate education and intended emigration may be due to the limited supply of specialty training opportunities in many countries studied and indicates the need to strengthen these post-graduate training programmes, while carefully considering which students receive scholarships for additional training.
Although we measured student intentions, our results closely parallel actual rates of migration observed among Asian and African health professionals. Among medical students we surveyed, 26% in sub-Saharan Africa and 21% in South Asia were planning to seek international careers. This resembles school administrative data indicating that 28% of recent sub-Saharan African medical graduates had migrated within five years post-training, 13 and population estimates suggesting 19-23% of Af-rica-trained physicians work abroad. 2,38 Among South Asian medical graduates -for whom data are limited -11% work in just four developed countries: Australia, Canada the United Kingdom of Great Britain and Northern Ireland and the United States of America. 5 Likewise, the 16% of sub-Saharan African medical students planning rural careers in our survey mirrors observed rural practice rates reported by school administrators (16%) and regional population data (13%). 13,39 Estimates of nurse migration are more elusive. 36,40 Some report 5-11% of all sub-Saharan African nurses (from diploma or degree training programmes) presently practice abroad. 22 The higher rate of intended international migration that we observed in the same region (32%) likely reflects a greater opportunity among bachelor-degree students to move abroad compared to diploma-only graduates. Similarly, the nursing students we surveyed had  (14) 32 (5) 10 ( (36) 200 (46) 159 (37) 78 (18) 245 (39) 162 (42) 51 (25) (12) 34 (8) 50 (12) 68 (16) 61 (10) 43 (11) 29 (14) 43 (12)  Medical and nursing students' intentions to migrate in Africa and Asia David M Silvestri et al.
lower rural practice intentions (24%) than the observed rural practice rates in sub-Saharan Africa (49%) or South Asia (31%). 39 This indicates either a stronger urban preference among bachelordegree nurses or an increasing trend among students to value urban careers more highly than their predecessors. This study has several strengths. It is large and comprehensive, with 3199 students surveyed from 32 classes in eight countries (Appendix). Our aggregate class response rate is high (84%). Our study is systematic, employing rigid yet relevant selection criteria to identify study sites. All participating nations face significant health worker shortages and ongoing emigration, but possess stable environments where retention policies are not superseded by larger systemic sociopolitical motivators of migration. We independently analysed 14 student characteristics to identify predictors for a career in the country or in rural areas. Our results have implications for education and health-care policy-makers in LMIC and donor nations.
Our sample cannot be generalized to areas where internal conflict or political turmoil may drive migration regardless of student characteristics, nor can it be extrapolated to non-Anglophone countries. Sub-Saharan African countries have comparable physician emigration rates regardless of national language, 22 indicating that different languages may not impede migrants' mobility. However, languages do influence the destination that the migrants select, 21 and possibly rural re-tention rates of graduates. 13 The effect of languages on students' migration plans warrants further research.
Our data suggest that students' career desires before matriculating may persist into the last year of training. Although our study design cannot exclude recall bias, the direction of recall error is unclear and should be further investigated through longitudinal assessment. Similarly, under-reporting of true migration intentions due to perceived values of the school professionals might create a social bias. However, if such bias is present it would strengthen our results, since the true number of students intending to migrate may be higher and those planning to work in rural areas even lower than we report. Social bias was mitigated by survey anonymity and

Research
Medical and nursing students' intentions to migrate in Africa and Asia David M Silvestri et al.
is unlikely to have affected class years differently. Our cross-sectional questionnaire of students' intentions is not validated for predicting actual migration behaviour. Nonetheless, the similarity of our results to existing migration statistics suggests that students' intentions may resemble such behaviour. 2,13,22,38,39 Therefore intention could be a feasible measure for migration and does not require long-term follow-up. Formal validation through longitudinal design is warranted.
Further research is needed to understand the migration intentions of students. Our study focused on students at premier public schools, where institutional missions focus on producing practitioners for domestic service and health-care leadership, and where public funds offset educational expenses. Migration ambitions may differ at private or newly-established public institutions with varying school resources, values and admissions standards. Also, longitudinal analysis is needed to discern whether differences in career plans between first-and final-year classes represent an evolution of students' preferences during the schooling period or a change in class composition resulting from recent enrolment expansions at many institutions studied. Finally, additional migration routes of health professionals in LMIC remain to be studied: from public to private sector and from clinical to administrative sector. This is important research since health professionals' movements from clinical public-sector work threaten already fragile public-health systems. 1,2 Increased demand for health professionals in developed countries is projected to attract even more LMIC graduates, [41][42][43] and is enabled by slow legislative uptake of WHO recommendations on health personnel recruitment. 44,45 Given the human resource needs in LMIC, migration between countries and from rural to urban settings deserves attention from policymakers both in countries with health worker shortages and partnering nations that provide technical and financial resources. 46 Multiple approaches have been recommended by WHO, such as education strategies, regulatory interventions, financial incentives and personal and professional support (Table 5). They must be combined for an effective outcome. 6 Simply increasing student volume without considering student selection ignores the labour-market dynamics after training, 47 and may also threaten education quality. Holding graduates in the country or in rural areas with compulsory service schemes does not seem to work in the long term, 15 since students who migrate leave promptly after such obligations end. 16 Additional curricular Lived majority of childhood in country 0.43 (0.24-0.76) NA c CI: confidence interval; NA: not applicable; OR: Odds ratio. a Rural population < 50 000 people, semi-urban population 50 000-200 000 people, urban population > 200 000 people. b Self-reported economic status of one's family compared to the rest of the country's population. c For the rural career outcome, the characteristic majority of childhood in country was not selected a priori before analysis.
Medical and nursing students' intentions to migrate in Africa and Asia David M Silvestri et al.
reforms incorporating rural coursework may be inadequate to attract students to rural areas, 6,19,20 as our data suggest short rural exposures have minimal impact compared with selection of students raised in these areas.

La intención de trabajar en el extranjero o en zonas rurales de los estudiantes de medicina y enfermería: una encuesta transversal en Asia y África
Objetivo Evaluar las intenciones de los estudiantes de medicina y enfermería de emigrar al extranjero o trabajar en zonas rurales. Métodos Se encuestó a 3199 estudiantes de medicina y enfermería de primer y último curso de 16 instituciones gubernamentales de primer nivel en Bangladesh, Etiopía, India, Kenya, Malawi, Nepal, Tanzania y Zambia. La encuesta incluyó preguntas para identificar los factores que podrían predecir las intenciones de emigrar de los estudiantes. Los resultados primarios fueron las probabilidades de emigrar para trabajar en el extranjero o en zonas rurales en el país de formación en los cinco años posteriores a esta. Los indicadores de las intenciones de migración se evaluaron por medio de modelos de probabilidades proporcionales multivariables.