Interventions for health workforce retention in rural and remote areas: a systematic review

Inclusion and exclusion criteria are summarised in Table 1. Studies were confined to 2010 or later because of available substantive reviews that synthesised the evidence relating to retention up to that time [1, 11].

A three-step search strategy was used to locate both published and unpublished studies [21]. The searches were undertaken 11–12 April 2019 and repeated on 1 July 2020 to capture any additional published studies. MEDLINE (OVID), CINAHL (EBSCO), Embase, Web of Science, Scopus, and Informit databases were searched as were ProQuest Dissertations and Theses, Trove and MedNar and the websites of government and peak non-government organizations. The MEDLINE search strategy is available as an Additional file 1.

All identified citations were collated and uploaded into EndNote Version X9 (Clarivate Analytics, PA, USA) and duplicates removed. Titles and abstracts were screened by two independent reviewers against the inclusion criteria. Potentially relevant studies were retrieved in full and their citation details were imported into the Joanna Briggs Institute’s System for the Unified Management, Assessment and Review of Information (JBI SUMARI; JBI Adelaide, Australia). Using the inclusion criteria, the full text of each citation was assessed independently by two independent reviewers. In the few instances where more than one paper was from the same research study only one paper was included. Reasons for exclusion of full text studies were recorded and are reported in the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) flow diagram (Fig. 1) [24].

Eligible studies were critically appraised and scored by two independent reviewers for methodological quality using the appropriate JBI critical appraisal instrument [25]. Disagreements were resolved through discussion, or with a third reviewer.

Data were extracted by two independent reviewers using the standardized data extraction tool from JBI-SUMARI. Study citation details, study objective, participant information, details of the setting/context, details of the retention intervention, and study results for the relevant outcomes were extracted.

The structure of the narrative synthesis of extracted data was based on categories for rural health workforce interventions used in the WHO Global Policy Recommendations [26]: education; regulatory interventions; financial incentives; and personal and professional support. A further category—health systems—was added, as proposed by Putri et al. [27].

The search strategy identified 2592 papers, with a further 57 papers identified from other sources (Fig. 1). After duplicates were removed 2043 papers remained. 1901 papers were excluded by title and abstract screening and 142 articles underwent full text assessment. 108 were excluded, leaving 34 articles. The main reasons for exclusion on full text review were a lack of quantifiable retention outcome measures, no intervention or ineligible study type (Additional file 2).

Most were cohort studies (29/34, 85%). Methodological quality of included studies was generally low (Additional file 3). The median score for included cohort studies was 16 out of a maximum of 22 (interquartile range 13–20). Only one-third of included studies applied appropriate statistical analysis, with less than half adjusting for key potential confounders. Many studies had no comparator group.

Study characteristics, participants and sample size, interventions, outcome measures and main findings are shown in Table 2. Of the 34 included studies, 13 (38%) were from Australia, 11 (32%) from USA, five from Canada and five originated from Nordic and nearby northern European countries. Most (n = 28, 82%) studies exclusively examined retention of doctors. Three studies were exclusively of nurses [28‐30], one exclusively of dentists [31] and two studies included mixed health professions [32, 33]. There were a total of 58,188 participants or participant observations in included studies. Four studies were outliers in terms of their comparatively large sample size [34‐37]. Most (n = 29, 85%) studies measured actual retention (or turnover), with the remainder measuring health professional preferences, intentions or simulating interventions. Actual turnover and retention were measured/defined over very variable periods of time: for example, one study measured only 6 months rural practice as being ‘long-term’ [38] whereas another documented up to 38 years of rural retention [39]. The outcome measures in 12 studies were retention rates [28‐30, 33, 40‐47], while five studies used survival probabilities or hazard ratios [37, 48‐52], and a further five used odds or relative risk ratios for staying or leaving [31, 35, 36, 53, 54]. The turnover or retention profile used most frequently in included studies was at the level of rural or remote practice anywhere within a country [31, 35, 36, 38, 46, 47, 54‐59]. Next most frequent was at the level of a largely rural jurisdiction [29, 41, 48, 50, 52] and rural practice within a jurisdiction [34, 40, 42, 49, 51], followed by turnover or retention within a rural community [33, 37, 45, 53].

A meta-analysis could not be conducted because the studies were highly heterogeneous, reporting different interventions and retention outcomes. Hence, a narrative approach was taken.

Twenty-one studies investigated the impact of educational interventions, including: selecting university students with rural backgrounds [31, 32, 38, 58, 60]; location of university training and its duration (rural, within a jurisdiction, within the same country) [31, 32, 36, 38, 43, 48, 54, 56, 58, 60]; multi-faceted interventions providing support for advancing education to a degree level (paid tuition with ROS obligation, scheduling flexibility, locally accessible, academic support locally; off-campus, decentralised education opportunities) [28‐30]; location of postgraduate training [44‐47, 52, 53, 58, 59]; and multi-faceted interventions supporting skills training for fully qualified rural health professionals (backfill, payment for supervisors) [42]. One study measured retention intentions; the remainder measured actual retention or turnover outcomes [32].

Selecting university students from rural backgrounds was consistently associated with increased rural retention [31, 32, 38, 58, 60]. Undertaking 1 or more years of university health training in non-metropolitan regions was associated with longer retention [43, 48, 54, 56, 60]. The retention of students who chose longer duration of rural training exposure during their basic training (2 years) were approximately double (odds ratio 5.38, 95% CI 3.15–9.20) those of students who chose 1 year (odds ratio 2.85, 95% CI 1.77–4.58) [38]. Opting in to a much shorter 1-month rural clinical placement during final year as a dental student was associated with increased prevalence of working rurally as a dentist in both 2015 and 2017 (prevalence ratio 1.93, 95% CI 1.19–3.15) compared to those who had no rural clinical placement [31]. Undertaking basic health professional training at an international versus a domestic university had no association with rural retention of graduates in two studies [36, 48]. One study showed internationally trained doctors were more likely to leave the province than graduates trained within that province [48].

Three studies with 392 nurses [28‐30] showed that supporting existing employees of rural health services and rural residents to undertake further university study using distributed models of education in conjunction with paid tuition, flexible schedules which allowed concurrent part-time or full-time paid employment, local teaching and tailored academic support was associated with high retention or low turnover and vacancy rates. Annual turnover of Licensed Practice Nurses at a US rural medical centre decreased from 16.8 to 6.8% following program implementation [28]. Nilsen reported 4-year nurse retention of 92.5% in a northern Norwegian county in an off-campus rural training group, compared to 70% for those who trained on-campus [29].

Nine studies investigated associations between retention and rural or remote location of postgraduate training. Two studies, specifically examining internship location and retention [41, 54], showed there was considerable variation in the proportion of postgraduates retained in rural or remote locations (ranging from approximately 0.35 up to 0.8). Undertaking postgraduate training in smaller rural sites (population < 10,000) was associated with marked increase in retention (odds ratio 36, 95% CI 12–109) [53]. A positive association between rural or remote postgraduate training and retention was reported to be stronger amongst rural origin registrars [58]. One Canadian study found that postgraduate (residency) training in a largely rural province was not a significant predictor of retention in the province, after adjusting for undergraduate training in that province [48]. An Australian study found an association between postgraduate rural generalist training (training which includes developing advanced skills, for example in Aboriginal health) and remote retention [54].

A small study of a skills enrichment program for fully qualified rural physicians, with provision for backfill and funding for preceptors for up to 1 year, reported all 29 were retained 5 years later, whereas significantly fewer matched physicians (22/29) not participating in the program were retained (risk ratio 1.3 95% CI 1.1–1.6) [42].

Nine studies examined the effectiveness of regulatory interventions on retention of doctors. One study simulated the impact of different types of health workers on doctor turnover. Increasing access to mid-level practitioners such as Physician Assistants and Advanced Nurse Practitioners was associated with a significantly increased probability of physicians moving away from the area after 1 or 2 years of service as these providers are a substitute for physicians [34]. In contrast, increasing rural supply of registered nurses, who provide services that complement those of physicians, was associated with a significant decrease in the likelihood of rural physicians leaving [34].

Eight studies investigated interventions which required service in rural areas (for a varying length of time) in return for a benefit. Several studies demonstrated that interventions comprising ROS in a rural area in exchange for highly valued visa waivers or access to professional licenses or provider numbers were associated with comparatively low rural retention/high turnover, especially once the ROS period was complete. Visa waiver recipients in Nebraska, USA, were almost four times more likely to leave rural areas of the state than state loan repayment recipients [51]. In contrast, the loan repayment recipients remained in rural areas for many years, with more than half still there 17 years later. Half of the visa waiver recipients left within 2 years of completing the 3-year minimum obligatory period. A study of loan repayment recipients who had completed ROS obligations in Colorado, USA, found that approximately two-thirds of rural recipients were still practising in a rural community compared to almost 100% of urban recipients who were still practising in an urban community [33]. Almost half (n = 10/21) of the loan repayment recipients who had finished their rural service obligations stayed less than 1 year beyond their ROS obligation [33]. One study found that international medical graduates (IMGs) who had yet to complete their ROS obligations had a substantially higher hazard of turnover than IMGs without locational restrictions [37]. Another study, however, found no statistically significant difference in the risk of leaving rural for IMGs, whether they had work location restrictions or not, compared to non-restricted graduates [36].

The intervention groups in two studies were subject to different types of regulatory interventions [50, 60]. Playford et al. were unable to differentiate the association of retention with two different interventions: Bonded Medical Places (access to a government subsidised university place) and Medical Rural Bonded Scholarships (scholarships paid to students during university training), [60]. Mathews et al. similarly did not differentiate between fellowships (funding provided to postgraduates training in a particular speciality), a bursaries program (scholarship to university students), and another program which provided special access to a postgraduate training place [50]. Hence it is not possible to determine which particular intervention(s) may have been associated with rural retention in these studies.

Five studies investigated associations between retention or turnover and various financial incentives, including having guaranteed access to paid locums [57], subsidized school fees for children [57], receiving retention incentive payments [35, 57], rural skills loading [57], increasing the salary of rural health professionals [34], reduced costs of malpractice insurance [61], and receiving Rural Doctors’ Association Settlement Package payments as a New South Wales (NSW) Visiting Medical Officer [37]. All studies involved doctors and took some account of potential confounders. One study was a simulation, another a discrete choice experiments and a further study recorded retention intentions [34, 57, 61]. Only two Australian studies entailed observed retention or turnover behaviour [35, 37].

Both a 50% increase in retention payments (β coefficient 1.423, p < 0.001) and 20% rural skills loadings (β coefficient 0.363, p < 0.001) were associated with increased probability of retaining General Practitioners (GPs), but were not as effective as providing guaranteed paid locum relief for 6 weeks every 12 months (β coefficient 1.624, p < 0.001) [57]. While locum relief incentives were important for retention of all rural GPs, regardless of location or on-call frequency, rural skills loadings were most important for GPs also doing hospital work. GPs with dependent children were also more responsive than GPs without dependent children to subsidised school fees [57]. Australian Government Rural Incentive Program payments were more effective in recruiting new GPs to incentivised rural areas rather than increasing the retention of existing GPs [35]. GP workforce retention was also significantly negatively associated with geographical remoteness in NSW, Australia, where GP retention incentives are scaled according to remoteness [37]. NSW GPs who were Visiting Medical Officers (and thereby received payments according to the Rural Doctors’ Association Settlement Package), had a 50% lower risk of leaving rural communities compared to GPs who were not [37]. In the USA, a 5% increase in rural county physician salary, simulated as an increase in reimbursement rate, was found to significantly decrease the probability of moving away from the same rural area; male physicians were more receptive to a policy change in reimbursement than their female counterparts [34]. Pepper et al. found that financial disincentives, in the form of high malpractice insurance rates, were associated with physicians planning to move their practice out of the (largely rural) state of Wyoming, rather than remaining in the state [61].

Four studies examined the effectiveness of personal or professional support interventions on actual retention or turnover of rural health professionals [39‐41, 55]. None adjusted for potential confounders. Offering medical students early sign-up to internships in a specific rural region, rather than going into a lottery for the opportunity to choose their preferred internship location, was associated with double the proportion of interns still working as physicians in the study area (29% versus 15%) [41]. However, retention of early sign-up interns was entirely in the most densely populated municipalities and none were recruited to any of the 15 remote municipalities. In a study of a cognitive behavioural coaching program (advertised as a work-life balance retreat), 94% of rural GPs who voluntarily attended were subsequently retained in rural general practice compared with 80% of the general rural doctor population (p = 0.027), despite their higher intention to leave rural general practice before coaching [40]. One study of an enhanced professional support network for rural surgeons added little to the extant literature as it lacked a comparator [39] and another low quality study reported that approximately two-thirds of survey respondents to a rural medical practice survey indicated that more reasonable hours of work, availability of locum tenens, availability of professional backup and educational opportunities for children would influence their retention intentions [55].

One USA study, where there is no universal health care insurance, used simulation to examine how expansion of publicly funded access to health care for some segments of the population (age 65 or older or younger but with disabilities—Medicare; low income—Medicaid) was associated with physician turnover [34]. Expansion of Medicaid and Medicare in rural areas was found to increase likelihood that physicians moved away from the rural county they worked in and became clinically inactive [34].