Factors influencing the migration of West African health professionals

Introduction The West African health sector is characterized by a human resource base lacking in numbers and specialized skills. Among the contributory factors to this lack of human resource for health workforce include but not limited to the migration of health professionals. Methods This cross-sectional survey targeted 118 young professionals who have participated in the Young Professional Internship Program (YPIP) of the West African Health Organization (WAHO), from (2005-2013). It inquired about their socio-demographic characteristics associated with migration and reasons for going to their preferred or most likely destinations through online survey. Results Of the 118 young professionals, 100 responded to the online survey, of which (28%) have migrated and (72%) did not migrate. Migration was more common among males and those (age≤31 years old), single with high dependency level and no previous work experience. Having a medical profession and being posted to urban or semi-urban area was also associated with their emigration. Their most important reasons for going to preferred or most likely destinations were to have fair level of workload, job promotion and limited occupational risks. Conclusion This finding suggests that the migration of health professionals is situation dependent, mediated by basic socio-demographic variables and work related conditions. These issues have implications for curbing the brain drain potential of health professionals in the West African health sector.


Introduction
The West African health sector is characterized by a human resource base lacking in numbers and specialized skills [1]. A major contributory factor to this lack of human resource for health workforce includes but not limited to the migration of health professionals to rich countries. Anecdotal evidence suggests that West African-trained physicians in particular have been migrating primarily to the US and the UK since the time medical education began in Nigeria and Ghana in the 1960s [2]. Since then, a flurry of scholarly work has been devoted to understanding the migration patterns of health professionals from sub-Saharan African countries.
Many studies have revealed a number of factors responsible for the migration of health professionals. These include career development-related factors [3], financial considerations such as better salary [4], poor and difficult working conditions, including long working hours [5], high patient workload and inadequate resources [6]. Political instability [7] and other personal related factors such as perceived social status and lifestyle [8], as well as the prevalence of diseases such as HIV/AIDS [3] have also been documented. Although many of these studies have accurately Evaluation research suggested that YPIP has substantial potential to contribute to strengthened health systems [1]. However, no empirical study has been conducted to examine the emigration potential of young professionals who have graduated from the program. This study aims to examine whether socio-demographic characteristics such as sex, age, marital status, dependency level, professional occupation, work experience and region will make important difference in the migration of health professionals who have participated to the program. The study also investigates some of the most important reasons for going to their preferred or most likely destinations. Variable studied: The dependent variables for the study included migration or no migration. Migration is defined as when a respondent's country of first job position is different from his or her second job position following graduation from YPIP. We considered second job country to include not only countries outside of West Africa (international), but also within (intra) Africa since YPIP was designed specifically to address the lack of human resource for health workforce in the West Africa region. No migration is defined as when a respondent's first job position is the same as his or her second job position following graduation from YPIP. This means migration did not take place or only internally within the same country. Respondents were asked to provide two job positions they have held and for each job to specify which country. First job position means a job title or position a respondent has held prior to participating in the YPIP. Second job position was considered as job title or position a respondent has occupied after he or she has graduated from YPIP. The different countries provided by respondents were coded (in numbers) and if the code for the country of first job position differs from the second job position, it means the respondent has changed job at country of first job position to second job position in another country. Our independent variables included are sex, age, dependency level, professional occupation, work experience and posting region. The variable for sex was coded as 1 for male and 0 for female. The age category was divided into (≤31; 32-34; and ≥35 respectively). Dependency level was classified into two different categories: low and high dependence. High dependency means a respondent has many people (such as a wife, husband, father or mother) to support and the opposite is true for low dependency. Professional occupation was divided into two categories: medical and public health officer. A medical officer was considered to be a medical doctor whereas the public health officer category comprises of public and environmental health officers and other allied health professionals such as nurses, laboratory technicians, pharmacists, nutritionist and biologists. This combination was done because there were few observations for allied health professionals. The data distinguish if a respondent was employed with a "yes" response, unemployed with "no" response, studying or in other type of economic inactivity as "others". The variable of work experience was obtained from this construction, in which, respondents were asked if they have had previous work experience in the public or private health sector prior to participating in the YPIP or after graduation from the program.

Methods
Posting region was divided into three categories: rural (coded as 1), urban (2) and semi-urban areas as (3). However, in the analysis we combined urban and semi-urban together because there were few observations for semi-urban.

Statistical analysis:
We first conducted a descriptive analysis of respondents' socio-demographic characteristics and of the outcome variables of interest (migration or no migration) using mean and standard deviation for continuous variables and frequencies and percentages for categorical variables [13]. All the variables that correlate with migration were entered. Second, we conducted bivariable analyses using the Chi-Square test to determine which socio-demographic variables were associated with each of the outcome variables of interest (migration or no migration). The adjusted OR was estimated by entering all the variables simultaneously, whereas the crude OR was estimated by entering all variables one-by-one. The 95% confidence intervals were calculated using fisher's exact test P for variables less than or equals to five.
We considered two-sided p-values and no statistical significant

Results
Of the 100 young professionals who responded to the online survey,  ( Table 1). This suggests that these health professionals are distributed in a way which disadvantaged rural population in their countries. In the bi-variable analyses, we evaluated the basic set of demographic variables (sex, age, dependency level, professional occupation, work experience and region) against the two main dependent variables (migration or no migration) to see if there were important differences. This evaluation revealed no statistical significant differences at the P=0.05 level, which could be due to the effect of low statistical power considering the sample size of the study. However, this was not true in the logistic regression results (Table 2). Although not very strong, we found slight differences. It was that migration was more common among males than females. With regard to age, compare with the reference category (i.e. age ≤31), migration was less common among young professionals age (≥35) years old. This effect could be because relatively older young professionals are close to their retirement age, but it could also be due to the grouping of the respondents. We also found that compared with being single, migration was less common among married respondents (OR=0.910; 95% CI 0.172-4.825). Also, compare to those with low level of dependency, those with high level of dependency were more migratory (OR=1.99; 95% CI 0.561-7.096).
We also found that compare with public health officers, migration was more common among medical doctors (OR=1.48; 95% CI 0.42-5.170). Being posted to urban or semi-urban area was statistically associated with migration (OR=3.676; 95% CI 1.118-12.08). In terms of attitude to migration (Table 3), we asked respondents to respond on a five-point Likert scale on some of the most important reasons for going to their preferred or mostly likely destinations. We found some variability in responses, which were sorted out starting first with the highest or most attractive responses from the respondents. The most attractive response with the highest score was to have fair level of workload with a mean score of (2.48), followed by to have fair level of job promotion and limited occupational risks respectively. The least ranked response or bottom score was to have career prospects with a mean score of (1.41).
This result could be because these young professionals have already obtained the needed professional skills for their work from their previous training and the YPIP, which could facilitate their opportunities for career prospects or entry into the labor market compared with other health professionals who have not participated to such programs as the YPIP.

Discussion
Factors associated with the migration of health professionals: Although there is growing concern in the number of female lead migrants from African countries [14], we found that migration in this study was more common with males than with females. This difference in migration pattern may be due to the restricted geographic mobility of women [15] and their informal care Page number not for citation purposes 5 giver responsibility [16]. Research has also shown that many female African migrant health workers are unable to work according to their qualifications due to gender discriminatory policies and the lack of recognition for their qualifications in the country of destination [17].
Our finding postulates that compared with being married, migration was also more common with young professionals who were single. This finding is in consistent with another study conducted in Lebanon [13] but contradicts other studies, such as [18], who argues that the likelihood of migration of unmarried men is higher than it is for single people. We also found that migration was more common among younger respondents age (≤31) years old compared with relatively older respondents, age (≥35) years old.
Although we do not have concrete evidence, we believe that relatively older respondents tend to migrate less, not only because the costs of migrating are higher at older ages, but also because the gain in terms of the expected earnings is smaller [19]. Research has also shown that household membership size, relative wealth, and the age and sex of its inhabitants may play a part in the decision to migrate [20]. This confirms our finding that among the respondents; those with high dependency level from their family members were more migratory compared with those that have less family members to support. Therefore, it seems that the possibility of instituting mean-tested family welfare benefits and other allowances may be at the root of reducing the migration of health professionals who have high dependency levels from their family members. The literature has shown that a mix of both financial and non-financial incentives is effective in addressing motivation [21]. This could also have implications for health worker migration. For migration between particular places, [22] argues that the relationship between occupation and migration may not be linear because the observed difference may be a consequence of migration, rather than a constraint to it. We, however, found that professional occupation makes a slightly important difference in the migration pattern of health professionals. Our findings showed that migration was more common among medical doctors compared to public health officers.
One plausible explanation for this observed occupational difference in migration pattern could be due to the high demand of medical doctors in developed countries. With regard to work experience, we found that migration was more common among respondents without previous work experience prior to joining the YPIP compared to those with previous work experience. This, in a way, contradicts the notion that more experienced workers are geographically more mobile. We postulates that respondents without previous work experience may be relatively younger and are likely to be single, which unlike married and older respondents may increase their physics cost of moving and decreases the length of the accrual period for benefits, thus inhibiting their geographic mobility [23]. The geographical push factors inhibiting posting preference is also associated with migration. We found that compared with rural posting, being posted to urban or semi-urban area was statistically associated with migration. One plausible explanation for this statistical association could be because migration information is more accessible in urban areas than in rural areas due in part to the consequence of rural-urban inequality in development [24]. All these issues put together might explain at included. The study, therefore, recommends a more sophisticated study methodology with a larger sample size.

Conclusion
Among the most salient points of this study is its multi-national nature and reasonable inclusion of different health professional Page number not for citation purposes 6 categories across the fifteen member countries of the Economic Community of West African States (ECOWAS). The study has also provided both post and pre-migration information of health professionals. Pre-migration data, in particular provides valuable information about motivations, perceptions and expectation concerning migration, which could be used by home and recipient countries in the formulation of migration policies [26].
What is known about this topic  The West African health sector is characterized by a lack of human resource for health;  Among the contributory factors to this shortage of human resource for health workforce includes the migration of health professionals to rich countries.

What this study adds
 That the migration of health professionals is situation dependent, mediated by basic socio-demographic variables and work related conditions;  These issues have implications for curbing the brain drain potential of health professionals in the West African health sector.

Authors' contributions
Mat Lowe participated in the conceptualization and design of the study, collects and analyzes the data and draft of the manuscript.
Duan-Rung Chenalso participated in the conceptualization and design of the study and data analysis. Both authors proofread and approved the final manuscript.

Acknowledgments
We acknowledge the contribution of Professor Abdoulaye Diallo, acting Human Resource Director of the Division of Human Resource of the West African Health Organization (WAHO) and Madam Adeola Awoyale Florence, the Program Coordinator of YPIP. We also owe gratitude to the study respondents.