Introduction

Continuing to strengthen the rural health workforce is crucial as part of building universal health coverage and achieving Sustainable Development Goals [1,2,3,4]. Even when rural people, who account for nearly half of the world’s population, are covered by universal health insurance, service coverage may be poor without a sufficient number of skilled local health workers [3, 5].

Higher doctor-to-population ratios correlate with lower maternal, child, and neonatal mortality [6, 7] and lower all-cause morbidity and mortality rates [8, 9], suggesting that access to tertiary qualified doctors is essential. Countries at all levels of socioeconomic development are investing in strategies to improve the supply and retention of qualified doctors in rural areas. High-income countries such as the United States, Canada, and Australia have implemented numerous policies and published extensively about various interventions, including financial incentives, rural education pathways, regulatory, and personal and professional support strategies to address rural doctor shortages [10,11,12,13,14,15]. In low- and middle-income countries (LMICs), stand-alone policies of compulsory rural healthcare-professional placements have also been implemented [16, 17]. However, the range of research informing how to improve access to qualified rural doctors in LMICs remains to be summarized. An additional quality issue is the lack of evidence relating to doctors at various career stages, since medical workforce dynamics may change by stage professional development [18].

The Asia-Pacific region is home to more than half of the global population, with approximately 98% of the Asia-Pacific population living in 29 LMICs, and just over half of these LMIC populations living rurally [19]. Doctor-to-population ratios in Asia-Pacific LMICs are well below the World Health Organization (WHO)’s benchmark of 1.15-to-1000 population [19], which is essential to achieve its Sustainable Development Goals [1]. Thus, it is critical to understand the effectiveness of strategies implemented to increase rural medical workforce supply in Asia-Pacific LMICs.

With this background in mind, this review summarizes and synthesizes existing evidence about factors associated with preferences and actual work locations of medical students and doctors in Asia-Pacific LMICs. This is done with a view to identifying effective strategies for recruiting and retaining doctors in rural areas. Additional aims are to describe how studies define rural or remote and to determine the spread of evidence by career stage, so as to inform how to target strategies better.

Methods

Nature of review

The scoping review method was used as it was most relevant to answering the primary research question about the range and extent of existing evidence. Scoping reviews, unlike traditional systematic reviews, place less emphasis on the critical appraisal of the included evidence, thus allowing the inclusion of a broader range of literature potentially relevant to capturing emerging evidence in Asia-Pacific LMICs [20]. The protocol for this review was developed iteratively by the authorship team according to Preferred Reporting Items for Systematic Reviews and Meta-analysis Extension for Scoping Review (PRISMA-Scr) [21].

Search strategy

The authors, with assistance of an experienced librarian, developed a Boolean string from key search terms (Table 1) and tested hits against ten key articles known to the first author with 100% sensitivity. Included were terms that covered LMICs, sub-regions and country groups in the Asia-Pacific as at 2019, using the World Bank 2019 definition of Asia-Pacific LMICs [22]. Other search terms addressed the population of interest (medical doctors), exposures of interest, and location of practice.

Table 1 Search terms applied in the scoping review
Full size table

Both peer-reviewed and non-peer-reviewed literature published in the last 20 years (July 1999–June 2019) were retrieved. Pubmed, Medline, CINAHL, EMBASE, PsycINFO, Web of Science, and SCOPUS were searched. Human Resources for Health and Rural and Remote Health journals were also searched. Grey literature searches included Proquest dissertations; first 10 pages of Google Scholar for each country; hrhresourcecenter.org (category: rural/urban imbalance, deployment); WHO website; and Global Health Workforce Alliance website. We searched the eligible articles’ references to identify any additional materials.

Study selection

Included were studies investigating the following outcomes: (1) actual work, referring to current work, and; (2) preference, referring to attitude towards, intention to work and remain, in the rural and remote areas. As the 2010 WHO global policy recommendations to improve rural health worker recruitment and retention emphasize the importance of educational interventions, and to specifically explore career stage [23], the review included studies of doctors, medical students and interns. This review was restricted to university-qualified doctors or students undertaking tertiary (university-level) degree training (Table 2).

Table 2 Inclusion and exclusion criteria applied in the scoping review
Full size table

After retrieving articles and removing the duplicates, we screened titles and abstracts against inclusion/exclusion criteria. We worked independently, then in pairs, to compare assessments and reach agreement. When eligibility for inclusion differed, the team conferred to resolve difference.

Data charting and analysis methods

A spreadsheet of key factors was developed to extract relevant data. Two authors tested and refined the data extraction tool using 10 eligible studies. Information extracted covered key areas such as country, sample, rural work outcomes, and factors related to the outcomes, and organized using the categories and sub-categories of the WHO global policy recommendations [23].

To understand how eligible articles defined ‘rural’ or ‘remote’ areas, we searched for explicit and implied definitions in the text according to categories described in previous studies such as: non-metropolitan area, population density and characteristics, distance from the nearest town and environmental characteristics [24, 25].

The authors discussed, agreed on, summarized and synthesized findings and implications for future research, policy and practice. Although not the main purpose of this scoping review, an overall exploration of study quality was undertaken to identify issues in research quality and support ongoing research.

Results

Source of evidence

The search retrieved 3425 articles. After removing duplicates and screening titles and abstracts, 71 articles were included in this review (Fig. 1). Ninety-two percent of the 71 eligible articles (see Additional file 1) were published after 2009 (Fig. 2).

Fig. 1
figure 1

PRISMA diagram of article selection process

Full size image
Fig. 2
figure 2

Peer- and non-peer-reviewed articles on effective rural medical workforce strategies by country and year. *Countries included in multi-country studies: Bangladesh (3 articles), Cambodia (1 article), China (2 articles), India (3 articles), Nepal (2 articles), Thailand (2 articles), and Vietnam (2 articles)

Full size image

Although the search was for low- and middle-income country studies, Nepal was the only low-income country (LIC; as classified in 2019) addressed in articles that met the search criteria. However, Nepal has been classified as a middle-income country (MIC) since 2020. No study from Pacific Island nations met the inclusion criteria.

Study characteristics

The majority (69%) of studies were quantitative, about one-quarter (24%) were qualitative, with the remainder (7%) policy analyses and review. More than half of studies (62%) included only tertiary-qualified doctors (at any career stage) as respondents, one-quarter (25%) of studies included only medical students and nine studies included both. Of 53 studies involving doctors (44 studies with doctors only and 9 studies with both doctors and students), half (53%) explored preference for rural work while the remaining (47%) investigated actual rural work (Table 3).

Table 3 Summary of the characteristics of the eligible articles
Full size table

Factors associated with doctors working in rural locations

We present results of factors associated with preference or actual work in rural locations according to the WHO areas (educational, regulatory, financial incentives, and professional and personal support), and an additional category of factors related to health system contexts. Some rural work predictors identified in the MIC-based studies—such as rurally located medical school, rural clerkship and rural internship—were not identified in the studies of Nepal (the only LIC in studies that met the search criteria). Differences were due to the scope of covariates explored. Given this, and because only one LIC was included in the studies considered, we present the results of both low- and middle-income countries as a unified analysis.

There were mostly concurrences in factors associated with preference and actual rural work when studies of similar factors were explored as to their findings. Therefore, we discuss the findings in an integrated way while noting them separately in Table 4. Where relevant, any differences in preference and actual rural work outcomes are discussed.

Table 4 Factors associated with actual/preferred work in rural locations of Asia-Pacific LMICs medical graduates and students
Full size table

Educational

Eighty-two percent of the articles—from 12 Asia-Pacific LMICs—were about educational factors, categorized into 2 areas: (1) student selection, and; (2) delivering medical education. For student selection, most studies demonstrated rural background was associated with both rural preference [33,34,35,36,37,38,39,40,41,42,43,44,45,46, 54] and actual work [26], while several found no association with rural preference [47,48,49, 56]. Being enrolled through the 'special track', which consists of rural student recruitment, scholarships and receive a rural-oriented curriculum, were associated with actual work in rural areas [27,28,29,30,31,32]. Other student selection factors associated with rural preference were: having parents with lower educational level or wealth [38, 40, 43, 46] or with an income source from the agricultural sector [37, 44], entering medical school through a graduate track [47], and having graduated from a government-owned high school [43]; however, there was no evidence for such association with the actual work. Studies exploring the delivery of medical education found associations for both rural preference and actual work with rurally located medical schools [40, 46, 61], rural clerkship [37], and rural-oriented curricula combined with other educational strategies [27,28,29,30,31, 62,63,64,65]. Students in public medical schools were more likely to prefer rural work compared to those from private ones [43, 55]. Rural internship was cited as a negative experience leading to poorer intention to work rurally in Indonesia [80], while, when delivered as part of a rural-oriented curricula, it was associated with better rural doctor supply [64].

Regulatory

One-fifth (20%) of articles, from China, Nepal, Thailand, and Timor-Leste, examined regulatory strategies. Compulsory rural service periods, whether implemented as a stand-alone strategy [31, 81, 82], combined with scholarships only [43, 47, 66, 81], or combined with combined with scholarship and recruiting students from rural areas [27, 29,30,31, 47,48,49, 72], were associated with higher rural preference or actual work.

Financial incentives

Forty-five percent of the studies—from 12 Asia-Pacific LMICs—explored associations between rural preference or actual work and financial incentives. Despite many demonstrating that appropriate financial incentives were essential for rural doctor recruitment, it was not clear what increment of incentive was needed for optimal results. One study revealed the actual income is higher among urban than rural doctors [46]. Across studies applying discrete choice experiment (DCE) methods, the proportion of increased salary or allowances tested varied, ranging from 0 to 300%. One study demonstrated doctors and medical students were 1.1–1.3 times more likely to consider rural jobs if offered 16% higher salaries [56], while others suggested that incentives worth 45% or 50% of doctors’ salary had the highest coefficient for rural work preference [54, 74, 76]. Nonetheless, some studies found that salary increases had lower utility compared to other recruitment/retention strategies such as good working environment [37], study assistance and supportive management [75], and support for professional development [58]. Opportunities to do private work were associated with better doctor supply or preference to work in rural areas [32, 45, 50, 90].

In term of retention, good salary was the second highest reason for Filipinos doctors’ willingness to stay in rural areas after completing the rural deployment program [96]. Likewise, a 50% salary increase had the highest utility to influence rural retention among Lao doctors [76]. However, in Thailand, financial incentives were deemed less valuable for retention compared to other factors such as working environment, community and personal factors [72]. Nor did an increase in salary associate with Timorese doctors’ preferences to remain working in rural locations [58].

Personal and professional support

Over half (57%) of the articles—from Bangladesh, China, India, Indonesia, Nepal, Pakistan, Timor-Leste and Vietnam—investigated personal and professional supports. The three most important personal and professional support strategies were working environment, living conditions and career development opportunities. Working environment included adequate facility infrastructure, equipment, drugs, and technology [34, 40, 58, 60, 75, 85, 91], sufficient number of health workers, availability of supportive mentoring or supervision, as well as availability of, and good relationships with, other health professionals [34, 46, 53, 91, 93]. Better housings [76, 92], electricity, water, and communications [52, 57, 59, 88], transportation [57, 72, 74, 76], schooling facilities [50, 53], employment opportunities for spouses [53] were the key living amenities important for doctors to work in rural locations. Clear career promotion schemes such as guarantee of permanent employment, transfer to more developed areas, or promotion opportunities were preferred to overcome rural doctor shortages [31, 54, 56, 76]. While relationship between gender and rural work preference showed mixed results (i.e., the majority found no difference, some found male prefers rural work, and others found the opposite), studies on actual rural work demonstrated that being male was associated with working in rural locations [27, 28]. Another frequently raised issues deterring doctors from rural practice was the lack of security [40, 50, 72, 73, 79, 88, 92], especially among women and in conflict-afflicted areas [57, 83].

Health systems

Some (13%) studies covered factors related to health systems issues that did not fit well into the existing WHO strategy categories [97]. These were from Bangladesh, China, Cambodia, India, Nepal, Thailand, Timor-Leste, and Vietnam, and included governance, service delivery, and health financing issues impacting on rural workforce supply [32, 50, 59, 70, 73, 83].

Definitions of rural (or remote)

Definitions of rural used to describe the outcomes were grouped according to four themes identified inductively: (1) inferred with no clear description; (2) facility-related, if differentiated by health facility factors such as facility resources; (3) non-facility-related, if characterized by demographic structure, environmental characteristics, population characteristics, topography or accessibility, and; (4) combination, if combined according to points (2) and (3) above (Table 5).

Table 5 Definitions of ‘rural’ as the actual/preferred work locations of Asia-Pacific LMICs doctors and medical students
Full size table

Many included studies (39%) did not provide a definition of rural. Many (39%) also defined ‘rural’ based on non-facility aspects such as level of socioeconomic deprivation [32, 51, 53, 57, 62, 70, 86, 90, 91], being located outside of metropolitan areas [26,27,28,29, 48, 82], population size [42, 65], administrative unit definitions [36, 77, 84], and geographical access [50, 52, 56, 58, 83]. Some (14%) studies defined rural as working in primary care or community-level facilities or smaller hospitals [31, 47, 49, 61, 63, 64, 67, 95].

Definitions of rural varied across different studies from the same country. For example, in India, a study conducted in Odisha state considered the entire state as rural [91], while another study conducted in Andhra Pradesh [41], only classified positions in community health centers or lower-level facilities as rural. Likewise, in Indonesia, while one study defined rural districts as < 25,000 population size [42], other studies considered any areas outside of Java and Bali—the most developed regions—as rural, regardless of population size [51, 90]. Studies involving respondents from more than one country relied on respondents’ self-reporting ‘rural’ via questionnaire [33,34,35].

Rural preference and career stages

Few studies considered and the impact of career stage. Of 53 studies involving medical graduates, only 2 analyzed outcomes by length of medical career. There was no association between being in early, mid, or later career and intention to stay working rurally among doctors in rural India [91]. Among early career doctors in Thailand, rural preference was higher among a cohort of doctors finishing 2-year compulsory rural service compared to those finishing 1 year [54].

Study quality

While most of the included studies had a clear research question and coherent methods, some were of poorer quality. Only half (n = 26) of 49 quantitative studies applied multivariate analysis (Table 3) with the remainder analyzing data at a univariate level or applied descriptive statistics without adjusting for potential confounding variables. Furthermore, among studies which adjusted for confounders, several relied on subjective definitions of rural location [33,34,35, 39, 75], thereby reducing study quality. Almost all qualitative studies explained data collection methods and respondent recruitment in detail. However, less than half (n = 7) clearly described the theoretical framework used. Some qualitative studies also did not report the relationship between interviewers and respondents, qualifications of the interviewers nor whether training was conducted to ensure consistency in the quality of the interviewing, thereby weakening the credibility of reported findings [99].

Discussion

This review is the first published study to undertake a detailed synthesis of factors associated with rural medical workforce supply in Asia-Pacific LMICs. Seventy-one articles from 12 countries published between 1999 and 2019 were included. Most evidence was from India, published within the last 10 years and mainly focused on doctors in practice. Around one-third of evidence related to medical students. The spread of evidence was reasonably even across the globally recognized WHO categories of strategies for rural retention, although this review identified a new category: health systems, including government policies and political climate, found to affect the rural medical workforce.

A broad range of educational factors were associated with rural work, especially related to rural background. Both preference and actual work in rural locations were associated with having resided in rural areas during the school-age period, having graduated from a rurally located high school, or being a native of a particular area, consistent with evidence on the importance of recruiting rural-origin students to increase rural doctor supply from other regions [10, 14, 100]. Despite this widely acknowledged evidence, there remains great opportunity for selecting rural background students into medical schools in the Asia-Pacific LMICs. In some countries, such as Indonesia, where more than 50% of medical students are enrolled in private institutions [101] and most medical schools are city-based, executing such rural-focused student selection could require government to provide more financial support for rural students as rural students are less able to afford a medical education. In Thailand, only 1 out of 19 medical schools is privately owned, an easier context in which to implement rural background selection targets [102].

There was limited research isolating the effectiveness of delivering a rural-oriented curriculum, and few examples of rurally located medical schools, rural clerkships, and rural internships. The scant available evidence about rural clerkships showed a positive association with rural work preference; yet, this only applied to those with an urban background. This evidence is not currently strong enough to recommend rural clerkships, nor how to go about rural-focused education. Nonetheless, the evidence available in Asia-Pacific LMICs does suggest that combining rurally based medical education strategies with other strategies, or with other compulsory and incentivizing strategies, can improve rural supply. This is consistent with evidence from other regions that combinations of rural workforce strategies are more effective than single strategies in increasing rural doctor availability [23, 103, 104].

Financial incentives and opportunities to earn income from additional jobs, a conducive work environment, and ongoing supports for professional development were also associated with rural intention, preference, and practice. These benefits were generally coveted by doctors and compensated for the perceived disadvantages of practicing in rural areas. The availability of local amenities such as housing, road infrastructure, and schooling facilities, as well as working-environment considerations such as facility readiness, and adequacy of drugs and equipment, were also associated with doctors’ decisions about work location, supporting the widely documented evidence from around the globe [105]. These strategies—financial incentives, supportive working environments, decent local amenities, and clear career ladder as well as effective human resource management practices—are in the governments’ scope of authority and make practical sense for local governments, rural health services and communities to implement. Thus, this could be especially important because many Asia-Pacific LMICs are decentralized nations in which the management of human resources, health resources, infrastructure, and finance is devolved to local government, thereby providing local governments with a specific role in rural medical workforce management.

While good salary was of high importance for rural doctor retention in the Philippines and Lao [76, 96], studies in Thailand and Timor-Leste found no such association [58, 72]. The majority of respondents in the Thai and Timorese studies had received scholarships (88% and 93%), which may have influenced their preference regarding financial incentives. This calls for more research to explore the role of financial factors, whether given upfront as a scholarship or at the time of employment, in increasing rural doctor retention in Asia-Pacific LMICs.

Past studies indicate that initial employment experiences could play a critical role in influencing doctors’ work performance and retention [18, 106]. In South Africa, doctors who had worked in rural locations at early career stages, even as part of a compulsory assignment, were more likely to have rural work intentions [107]. However, this review identified the paucity of evidence on the association between the different length of employment and actual rural work. A better understanding on the difference of rural work across doctor’s career stages would better inform health workforce planning and decision-making, hence calls for more inquiries on this topic.

We identified that male students or doctors were more likely to prefer or actually work in rural or remote locations [27, 29, 38, 43, 47, 77] and female doctors were more affected by perceptions of lack of security than were male doctors [57, 83]. Since women dominate the doctor population in many Asia-Pacific LMICs [38, 48, 58, 76, 108], understanding the environmental and personal attributes that influence female doctors’ willingness to work rurally is crucial to inform effective policy.

The disparity between the factors related to the two outcomes—preference or actual work in rural areas—was mostly due to difference on variables studied across both outcomes. While there is some evidence that graduation from a public medical school or having lower-income parents was associated with rural work preference, the absence of evidence of these associations for actual work should encourage further investigations. This could become an invaluable policy input, especially for countries where a significant proportion of doctors attended high-fee private schools.

The review identifies some weaknesses of study methodologies. First, almost one-third of the studies did not define rural location, and, for studies that did, the definitions were non-standardized and varied significantly. Some of the definitions used may not reflect geographical and demographic aspects of rurality. This variation may not affect the utility of the study for adding to the emerging evidence about effective rural workforce strategies in Asia-Pacific LMICs; however, it does reduce the capacity to validate and generalize the findings to other contexts. The diverse rural definitions for health policy and research purposes have been previously identified and widely discussed, including in developed countries [24, 109]. The usefulness of future studies in this field could be increased by standardizing definitions of rural.

Second, the majority of included studies only asked whether respondents ever lived in rural areas, without considering any particular rural area. This did not allow any conclusions about whether exposure to any rural area, or familiarity with a specific area, has a stronger influence on doctors’ decisions about where to work. Exploring the nature of doctors’ backgrounds, and whether it is rurality in general, or attachment to a hometown, is an important topic that could provide substantial evidence for policymakers in selecting characteristics for medical students recruitment or doctor deployment.

Finally, the included studies lacked multivariate analyses that is needed to isolate the strength of association of factors related to doctors’ rural work and preference. Adjusting for potential confounders in studies on rural workforce strategies will allow policymakers to understand which of the strategies or sociodemographic characteristics need more emphasis to improve rural doctor availability.

We acknowledge that Asia-Pacific LMICs included in this review range from the world’s most populous countries with strong economic growth (e.g., China) to comparatively smaller, poorer nations with less than 2 million people (e.g., Timor-Leste). Also, no study from Pacific island nations was identified; there remains a need for more studies from these countries. The findings should be generalized with caution given the range of included material from different settings and contexts. By focusing this work on one region, and the context of LMICs being more similar than including all country types, the work substantially adds to the existing evidence for guiding rural medical workforce development in this region, including according to the WHO guidelines.

Almost all of the studies we included involved local-origin doctors, except for Timor-Leste where some doctors are from Cuba. As the poaching of medical staff between high-, middle-, and low-income countries could impact a country’s doctor supply, further research should consider supply chains, including cross-country migration and its impact on different LMICs.

Conclusions

This study provides critical new evidence, drawn from 20 years’ research, about a range of factors which can be used to target strategies to increase rural medical workforce supply in Asia-Pacific LMICs. The evidence has grown substantially, especially over the last 10 years, but remains confined to 12 Asia-Pacific LMICs. Achieving rural medical workforce growth in Asia-Pacific LMICs required multi-level approaches including selecting more medical students with a rural background, combining this with rural-focused or -located education and return-of-service scholarships, workplace and rural living support and ensuring an appropriately financed rural health system. The review identifies the need for more studies in a broader range of Asia-Pacific countries which define rural clearly, expanding on all strategy areas, use multivariate analyses, and test how various strategies relate to doctor’s career stages.