Determinant factors to the rural and remote medical workforce in Maluku Islands of Indonesia: A cross-sectional study

Many factors contribute to engagement in rural and remote medical practice but little is known about the determinant factors of rural and remote medical practice in the such remote locations as the Maluku Province of Indonesia. This study describes determinants, preferences and intentions of doctors in the rural and remote practice. An online survey of work-related experience and intentions for future rural work was administered to 410 doctors working in the Maluku province of Indonesia. Participant characteristics were described using descriptive statistics, associations between the independent variables with the location of workforce, preference for rural practice and intention to remain in rural and remote Maluku were analysed using Chi-square tests and logistic regression.


Page 3/32
Background A shortage of health personnel and workforce maldistribution means there is unequal access to healthcare for people living in rural and remote (RR) communities, a persistent and signi cant problem globally. [1][2][3] Indonesia continues to struggle with community access to healthcare and insu cient health workforce, especially in its RR areas. [4][5][6] Maluku, a province in the eastern Indonesian archipelago which includes some of the most remote, isolated and poorly served islands in the country 7 , has national data showing maternal mortality rates (above national standards 102/100,000 live births), infant mortality rates (> 22 deaths/1,000 live births) and infectious diseases rates that are many times higher than in countries with acceptable access to care. 7,8 As reported in the provincial health pro le, a signi cant challenge in achieving the national standards for this province is inadequate services due to limited resources, namely the human and capital resources for health and essential physical infrastructure. 9 Maluku has more than a half (6/11) of its regencies classi ed as underdeveloped areas. 10 The national government allocates just 5% of its budget for health. 11 In addition to the remote and isolated nature of the province being a disincentive for medical personnel choosing to live and work there, 4,12,13 the region has less infrastructure, facilities and amenities, di culties with communication, a perceived lower quality of children's education and lower employment income. 4,12,13 In Indonesia, the national health pro le shows that the number of doctors in an area was positively related to population numbers, population density, number of hospitals, and number of community health centers. 13 The ratio of doctors per population is 1:7,269 in Maluku, compare to 1:2,294 nationally. 7 In recent years, the central government has implemented various policies such as compulsory work placements and nancial and career incentives to attract and bond doctors and health professionals to remote and isolated areas of Indonesia. 4,12 Even though Maluku has been one province where doctors may extend their stay after compulsory work placements with goverment incentives 14 , the small numbers of doctors in RR areas of Maluku remains a signi cant issue. 7 Clearly more solutions are required.
Maldistribution of rural health personnel, particularly doctors, occurs globally. 2,15−17 Recruitment and retention of doctors in rural areas is in uenced by many factors including personal factors such as rural background and educational factors including rural exposure during medical training. A scoping review of 61 papers published between 2010 to 2020 18 concluded that rural background and rural training are decisive factors in recruiting and retaining doctors in RR areas.  In Indonesia, before the establishment of Pattimura University Medical School in Ambon, Maluku, doctors who worked in Maluku predominantly graduated from medical schools in Java Island (Jakarta, Surabaya, Yogyakarta, and Bandung) and cities outside Java island such as Medan, Makassar and Manado. In 2008, an undergraduate medical school was established in Pattimura University with the aim of educating medical students to work locally in Maluku. The rst ve cohorts of 50 students were partly funded by the government to encourage graduates to serve the province, with more than 20% of students having a rural background. 44 The vision was to improve the number of doctors (an additional 50) in Maluku by 2015.
Understanding of the determinants of medical workforce taking up RR practice in Indonesia, and speci cally Maluku, is limited and not adequately addressed in either Indonesian and international research. Hence, this study aimed to investigate the determinants of doctors' preferences to work in the RR areas of Maluku Province. Speci cally, this research answers three research questions: 1. What are the characteristic of doctors who work in Maluku Province? 2. Which demographic variables are associated with RR practice location, preference for, and intention to remain in rural practice in rural and remote Maluku Province?
3. What variables predicts the outcomes of RR practice location, preference to RR practice, and intention to remain in practise in rural and remote Maluku Province.
Findings from this study can inform policy and practice for other archipelago regions and developing countries, especially within the Asian region which has similar societal and regional geographical characteristics.

Study population
The study population comprised quali ed doctors who were currently working in Maluku Province.
Inclusion and exclusion: Doctors employed in Maluku Province (hospitals and health clinics, administrators and academics), of any age, gender and discipline/specialty area were included. Those temporarily unemployed or working in another province were excluded.  45 with the location of practice converted to the category (rural = 1 or urban = 0). 4. Outcome variables i.e., location of current practice, preferred future practice location, and intention to remain in the rural and remote Maluku areas.

Data collection
Online-based delivery was the most feasible option given the doctors' geographical distribution with the survey open for a 2-month period to mitigate limited internet access in the Maluku area.

Data analysis
All analyses were undertaken based on valid cases using IBM SPSS statistics version 26. Descriptive statistics were used to determine participants' charracteristics. A Shapiro Wilks test was used to con rm normality of the data distribution. Chi-square tests were performed after a test of independence between outcome variables and independent variables (Tables 2-4) to determine the relationships between both. The outcomes of interest were de ned as Binary logistic regression was based on variables with a signi cance level < 0.20 to estimate odds ratios associated with factors predictive of the target outcomes. A con dence interval of 95% and alpha signi cance level of 0.05 were used.

Characteristic of respondents
There were 324 recorded responses (79% response rate) with 241 completed surveys. Most respondents being younger, female, married, and born in Maluku (Table 1). Of the females, 61% were working in a RR practice location. Of all respondents born in Maluku Province, less than a third were born in RR areas.
More than half the respondents had never lived in rural areas before commencing medical school. Most Page 6/32 of the doctors graduated from medical schools in cities outside Java ( Fig. 1) and almost all respondents experienced some rural exposure during their medical program (Table 1). Yes 62 (25) *Ordinal categories were then adjusted to dichotomous for further analysis Not all participants answered every question so the numbers do not add to 324.

Figure 1. Location of participants' medical schools
Most respondents had worked less than ve years since graduation and under ve years in their current post, had only temporary contracts, and did not undertake additional practice besides their main job (Table 1). Of the 11 regencies in Maluku province, respondents predominantly worked in Ambon (39%), the capital city of Maluku Province, and the nearest regency from the capital, Central Maluku regency (21%) (Fig. 2).  Figure 3 showed how salary had not changed over duration of work and take home pay increased after work for more than 10 years. The analysis of duration of work with monthly take home pay and location of current work and additional practice showed that of doctors who had worked for more than 10 years, 50% had an additional practice and 78% were located in more developed urban areas, Ambon city (59%) and Central Maluku Regency (15%). Additionally, those doctors with more years of work and higher take home pay (greater than median IDR 8,5 million) were mostly specialists who need specialist facilities (n = 28; 57%) so are permitted to have up to an additional 2 practices. 46   Logistic regression revealed that shorter length of work since graduation, salary < IDR 6 million, take home pay < IDR 12 million, and having no additional practice were predictive of the current RR practice location. Doctors who started work within the last ve years were 7 times more likely than their seniors (OR 7.07, CI 95% 2.46-20.3, p = 0.000) to have current practice in rural and remote Maluku. Doctors with a smaller salary were almost 6 times more likely (OR 5.61, CI 95% 2.26-13.9, p = 0.000), and doctors with smaller take home pay were almost 4 times more likely (OR 3.85, CI 95% 1.38-10.7, p = 0.010) than those with higher salaries and take home pay to have current practice in rural and remote Maluku. Further, doctors with no additional practice were nearly 4 times more likely (OR 3.89, CI 95% 1.68-9.04, p = 0.002) than those with additional practice to be working in rural and remote Maluku. Preferred location of practice in rural and remote areas in Maluku (Table 3) There was an association between younger age (nearly three times as likely to prefer future RR practice in Maluku than their seniors, OR 2.82, CI 95% 1.

Discussion
This survey of doctors living and practicing in Maluku province Indonesia offers valuable insights into factors signi cant to both recruiting and retaining a sustainable rural workforce.
Doctors currently working in rural and remote Maluku were more likely to be of younger age, single, graduated from Pattimura University, working for less than ve years in current practice and since having graduated, report no additional practice, and have a smaller monthly salary and take home pay. Rural and remote practice in Maluku were predicted by shorter length of work since graduation, the absence of additional practice, and smaller salary and take home pay.
Length of work since graduation up to 5 years was the strongest predictor of RR practice location, and currently practising rural was the strongest predictor for future RR practice preference and intention to remain in RR practice.
These data inform an argument for dedicated government support of recruiting people from areas of medical workforce shortage into medical school, and support for students and graduate doctors through training and attractive opportunities to sustain their practices, which can subsequently retain the doctors in areas of workforce shortage.
This study identi ed more doctors practicing rurally were female (70%). This nding is notably different from other international studies describing characteristics of rural doctors, where male doctors tend to be the majority. 21,30,47 While not predictive in this sample, another study in Australia has reported that more women are applying to Rural Clinical Schools 34 and proportionately more are going on to rural work 26 , thus beginning to redress rural workforce shortages in female practioners. 26 The majority of doctors practicing in Maluku graduated from medical schools in regional or provincial capitals, including almost a third from Pattimura University, Maluku. Given that most of our respondents were Maluku-born, this nding con rms that regional students come back to their regions 48-50 , and implies that signi cant efforts should be put into development of regional medical schools in order to improve distribution of doctors to the regions. 48-50 While not signi cantly associated with rural practice location, we found that most doctors experienced rural exposure during their medical training. This is similar to the Australian requirement that all students experience rural work during their medical training. 51 Clearly more directed positive strategies are needed.
Interestingly, rural exposure during medical training which is one of the most widely reported factors in other studies, was not associated with or predictive for RR practice location, rural preference, and intention to stay in a rural post. This was also found a study from Canada 52 , however, that study was from 1999 when medical schools were in the early stages of advancing rural experiences for their students. A 10 year longitudinal cohort study from Australia showed that rural exposure during medical training related to rural work. 51 Another important factor reported elsewhere as a determinant of RR practice uptake is rural background [19][20][21][22][23][24][25][26][27][28][29][30][31][32][33][34][35] , but in this study it was not associated with or predictive for rural practice location. However, rural born and rural living experiences were associated with the preference of future rural practice location also intention to remain in RR practice. Rural living experience was also a predictor for prefered future RR practice, and born rurally was a predictor for intention to remain in rural practice among doctors working in Maluku Province. All of these factors should inform development of pro-rural work policies for this archipelago.
These positive policies are not only the domain of Western countries, as we con rm that a geographical maldistribution exists even in this developing province. 9 But since our ndings suggest that rural living experience and being rural born were associated with future rural work preference and intention to remain in rural practice, more focus and attention should be given to the recruitment of students and doctors with a rural background. Although this nding is not novel, this study importantly con rms the pattern among the limited number of studies from developing countries.
Additionally, evidence suggests that widening access to medical course enhanced care to underserved communities. [53][54][55][56][57] A more comprehensive approach is needed to widen the participation and aspirations for medicine of under-represented socio-economic and educationally disadvantaged groups. This could include regent government early education programs and support including scholarships aimed at these under-represented groups.
Regarding future preference of practice type, doctors in Maluku preferred to work in specialist practices, namely Internal Medicine, Paediatrics, Obstetrics and Gynaecology, and Surgery. Specialist practice, especially within the four major specialties, is an opportunity to earn more income and the four specialty areas align with international preferences among doctors and medical students. 58-60 Although there are currently few specialists in Maluku and there is a great need for more specialists, priority should be given to primary care, rural practice, rural generalist, and family medicine for these areas of practice are associated with improved recruitment and retention of RR medical human resources. 21,23,24,27,28,32,61,62 We found that monthly salary and take home pay were relatively low regardless of the length of work (IDR 5 million, equal to USD 350/GBP 250 and IDR 34.5 million equal to USD 2400 or GBP 1750 respectively), and low salary was associated with and predicted current rural practice. Considerable evidence shows that a medical school intentionally established in a region of workforce shortage pays much greater attention to the region's health status and concern. This is the case in both the developed. 22,66−68 and developing world. 49,64,69 The medical school in The Philippines, Zamboanga 64 , has a similar rural, archipelago, and developing country context as Indonesia so its ndings are likely to be immediately relevant. This medical school showed that effective and sustainable medical education is possible in poor rural areas. 64,70 Compared to James Cook University which strongly favours applicants with rural backgrounds and requires a commitment to work rurally after graduation 22,66−68 , Pattimura University only stresses the philosophical value to practise rurally and offers rural exposure during medical training without any requirement for a rural background or commitment to work rurally.
From the successful experience of other universities 24,26,28,49,64,69,71 , and based on this study's ndings, Pattimura University could expand even further into the rural areas of the region. Stressing the university's values by increasing the proportion of students with a rural background and ensuring rural exposure is offered at different year levels for a range of disciplines during the medical course will likely increase the number of graduates serving the RR areas of Maluku Province.
It is evident from this study that younger doctors were more likely to work rurally and to stay rurally, perhaps re ecting this generations' ethical stances around the world. 72,73 Those who start working in the last ve years were more likely to take up and prefer rural practice, however, they received a lower salary and smaller take-home pay which reduces their intention to stay in RR practice. The rural work experience requirement and recommendation from the rural government for scholarship in specialist training from the Ministry of Health 74 means these rural posts are likely to temporarily attract younger doctors to rural service in Maluku. However, retaining doctors in rural and remote Maluku Province requires more than nancial incentives. Evidence from elsewhere shows that educational 61,75−80 , multi dimensional 81-85 , and professional development strategies 86,87 improves retention of doctors in RR areas. Consequently, collaboration between medical schools and local government is required to ensure relevant strategies are implemented to improve the recruitment and retention of doctors in RR areas.

Study limitation
Pertaining to the sampling frame, we noted difference in the number of doctors working in Maluku compare to the list provided from provincial and regents health o ces, and the medical school. There was no integrated database listing all doctors working in this province. The use of the Pattimura University alumni database to identify additional medical graduates augments the denominator for sampling but may potentially bias the participant sample.
This study is cross sectional, so it is not possible to draw inferences of causality and outcomes of individual preferences and inclination to remain in rural practice which can easily change. A longitudinal study is needed to track whether the participants are still in rural practice 5 to 10 years from now. Secondly, despite the high response rate, the study may not have had power to detect less strong associations between rural exposure in medical training and subsequent practice location.

Page 22/32
The de nition of rurality used in this study was the Indonesian national classi cation which may differ to that used by other countries and make comparison challenging.

Conclusion
This study provides evidence that rural background predicts doctors' RR preference and intention, and that a regional medical school helps supply doctors to the RR areas in its region. Sustained collaboration between medical schools and local government implementing relevant strategies are needed to widen participation and improve the recruitment and retention of rural and remote doctors. Consent to participate to this study was embedded in the online survey administered to the participants, in which participation in the survey means consent to participate.

Consent for publication
Not Applicable Availability of data and materials The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Figure 1
Location of participants' medical schools Map of Maluku with population and respondents number per regency/city.

Figure 3
Average monthly salary and take home pay over duration of work Page 32/32

Figure 4
Participants future practice type preference

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. SurveyQualtricsBahasaIndonesia.pdf