General practice registrars’ practice in outer metropolitan Australia: a cross-sectional comparison with rural and inner metropolitan areas

ReCEnT

ReCEnT is an ongoing inception cohort study of general practice registrars’ in-consultation clinical and educational experience, and behaviours. From 2016, the study has been conducted in three training organisations in three states and one territory. Data included in this analysis were confined to registrars training in New South Wales (NSW) and the Australian Capital Territory (ACT). The study participants are registrars in general practice training terms.

The detailed methodology is presented elsewhere (Davey et al. 2022). Briefly, registrars record details of 60 consecutive clinical consultations, with data collection representing approximately 1 week of a registrar’s clinical work. This is repeated in each of three terms of training for each registrar. Data, collected through an online portal, includes patient demographics, diagnoses and problems managed, investigations, management, and educational training aspects including use of resources and supervisor assistance.

Outcome factor

The outcome factor was a measure of practice geographic location: rural, outer metropolitan (OM) and inner metropolitan (IM). We used the Australian Government Department of Health Inner and Outer Metropolitan (IOM) classification to identify regions of training by geographical location (Australian Government Department of Health and Aged Care 2022b). In the IOM, only the Sydney and Canberra regions are classified as OM or IM in NSW and ACT, respectively. This does not reflect the registrar experience of GP vocational training undertaken in Newcastle and Wollongong (large cities to the north and south of Sydney) training practices, which is not consistent with rural training experience. Thus, for our main analysis, we omitted data from Newcastle and Wollongong practices. In a sensitivity analysis, we included these practices as additional OM practices, as the judgement of Medical Education members of our team is that the training experience is more consistent with outer- than with inner metropolitan-defined locations. These are also areas of relative workforce need, similar to OM areas. Both the Newcastle and Wollongong regions are identified as locations with a shortage of medical practitioners by the Distribution Priority Area (DPA) classification. All inner metropolitan areas are automatically classified as non-DPA areas, regardless of relative workforce need (Australian Government Department of Health and Aged Care 2022c).

Independent variables

Independent variables were related to registrar, practice, patient, consultation content and consultation action (see Table 1).

Statistical methods

This was a cross-sectional analysis of NSW and ACT registrar data from 2016 to 2020. Analysis was at the level of individual consultation.

Descriptive statistics included frequencies with percent for categorical variables and mean with standard deviation (s.d.) for continuous variables.

Multinomial logistic regression was used within the generalised estimating equations framework to account for repeated measures within registrars, assuming an exchangeable working correlation structure. Multinomial logistic regression for a nominal outcome with three (non-ordered) levels produces two sets of explanatory variable effect estimates, one for each outcome level compared to the reference level. In this analysis, ‘OM’ was used as the reference level. The model thus estimated odds ratios with 95% confidence intervals for the two comparisons, ‘IM vs OM’ and ‘Rural vs OM’. For ease of interpretation of estimates in the models, beta-coefficients have been multiplied by −1 and then exponentiated (with the result that odds ratios have been inverted).

Univariable analyses were conducted to assess the association of each covariate with the outcome. Covariates with a univariable P-value < 0.20 were considered for inclusion in the multivariable regression model.

After an initial multivariable model with all significant covariates fitted, the potential for model reduction was assessed. Covariates that were no longer significant (at P < 0.2) in the multivariable model were tested for removal from the model. If the covariate’s removal did not substantively change the resulting model, the covariate was removed from the final model. A substantive change to the model was defined as any covariate in the model having a change in the effect size (odds ratio) of greater than 10%.

To examine different facets of our research question, four sequential models were built, with each OM/IM/Rural location as the dependent variable.

The four models, each sequentially including an extra factor group included: (a) registrar and practice factors, before (b) patient factors, then (c) consultation content factors, and then (d) consultation ‘action’ factors were added

The rationale for building four models was that associations of a registrar’s consultation being in an OM, IM or rural area will include registrar and practice factors, but estimates of these associations may be attenuated by inclusion of patient factors, if the latter lay on the causal pathway from registrar and practice factors to the outcome. Similar reasoning applied to the sequential addition of consultation content, before consultation ‘action’ factors were added

Analyses were programmed using STATA (ver. 16.0, StataCorp, College Station, TX, USA) and SAS (ver. 9.4, SAS Institute Inc., Cary, NC, USA).

Ethics approval

Ethics approval was obtained from the University of Newcastle Human Research Ethics Committee (H-2009-0323).

Main analysis

In multivariable analysis, regarding ‘registrar and practice factors’, registrars in OM practices were more likely to have worked at the practice previously than those in IM practices, but less likely than those in a rural setting (adjusted odds ratio, aOR = 1.62 [95% CI:1.19, 2.21], P = 0.002; and aOR = 0.32 [95% CI:0.23, 0.44], P < 0.001, respectively). Registrars in OM practices were less likely to work in a bulk-billing practice than those in IM practices, but more likely than those in rural areas (aOR = 0.71 [95% CI:0.55, 0.91], P = 0.008; and aOR = 14.95 [95% CI:10.26, 21.77], P < 0.001). Registrars in OM, compared to those in rural areas, were more likely to be in Term 2 or Term 3 of training, respectively, than in Term 1 (aORs = 2.00 [95% CI:1.53, 2.61], P < 0.001; and aOR = 1.52 [95% CI:1.10, 2.11], P = 0.011).

Regarding ‘patient factors’, there was greatest exposure for trainees in the care of older (>65 years) patients in rural areas, followed by OM area and then those in IM areas (aOR = 0.47 [95% CI:0.41, 0.53], P < 0.001 and aOR = 1.18 [95% CI:1.05, 1.31], P = 0.004). Management of Aboriginal and/or Torres Strait Islander patients also displayed a similar difference, with exposure greatest for registrars working in rural areas, and lowest for those in IM areas (aOR = 0.71 [95% CI:0.57, 0.88] and aOR = 1.93; [95% CI:1.47, 2.54]; P < 0.001 and P = 0.002, respectively). Conversely, highest exposure to non-English speaking background (NESB) patients was to registrars in IM practices, declining to OM and least in rural practices (aOR = 0.43 [95% CI:0.36, 0.52], P < 0.001; and aOR = 7.60 [95% CI:5.85, 9.88], P < 0.001). Patients were less likely to be new to OM practices than to IM and rural practices (aOR = 0.68 [95% CI: 0.61, 0.76], P < 0.001; and aOR = 0.76 [95% CI:0.66, 0.87], P < 0.001).

Regarding ‘consultation’ and ‘consultation action’ factors, chronic problems were managed less often by trainees in OM than those in rural practices (aOR = 0.84 [95% CI:0.77, 0.92], P = 0.001). Imaging was less likely to be ordered by trainees in OM practice than in rural practice (aOR = 0.82 [95% CI:0.75, 0.89], P < 0.001). Referrals ordered by trainees working in OM areas were less than those working in IM and rural areas (aOR = 0.80 [95% CI:0.74, 0.86], P < 0.001; and aOR = 0.84 [95% CI:0.77, 0.91], P < 0.001). Pathology was less likely to be ordered by trainees in OM areas compared with those in IM and rural areas (aOR = 0.89 [95% CI:0.83, 0.95], P = 0.001; and aOR = 0.75 [95% CI:0.69, 0.82], P < 0.001).

Sensitivity analysis

The sensitivity analysis (see Supplementary Tables S1 and S2) produced similar results to the main analysis.

Summary and interpretation of main findings

The salient findings of this study centre around the distribution of the training workforce, their experiences while working in OM practices, and the evidence this provides regarding patient access to health services in OM areas. There are also implications for trainee education and support in OM practices, inferred from observed referrals for pathology, imaging and specialist support.

Implications for practice and policy

OM areas are distinct from IM and rural regions. There is a difference in training experiences and service provision by GP registrars across these areas; some aspects of which highlight health inequality affecting OM areas. This study has demonstrated that with respect to training location of GP registrars, the patient populations seen, the types of healthcare issues managed and the types of services provided, that OM areas are distinct, with distinct demographics and healthcare needs. Discussions concerning medical workforce, health service provision and GP training location thus need to extend beyond the urban versus rural divide and include OM areas.

Encouraging and increasing GP registrar presence in OM areas would not be a ‘zero-sum’ initiative, where registrars are simply fulfilling a workforce need at the expense of educational utility. Registrars are not being disadvantaged from an educational perspective; in fact, our findings suggest the opposite. Effective and targeted policy is required to increase the OM registrar presence, fulfilling both a workforce and social imperative. Such policy could be improved by framing the need for an OM registrar workforce in the context of learning and educational advantages, where providing rich and supportive educational environments will help increase GP trainee numbers. If retained in OM areas (as previous research regarding rural and lower socioeconomic status location suggests is likely) (Moad et al. 2022b), these fellowed GPs will go some way to addressing the growing health inequality between OM and other regions. Currently, equity as well as workforce distribution considerations, are determinants of registrar distribution across training practices, with trainees required to work at least 12 months in an OM area or MM2–7 region. This socially responsible policy should be retained, if not strengthened, and, as our findings suggest, is educationally sound.

Part of the reason for the drift of GP trainees away from rural areas in which they may have trained and worked is that unless they had pre-existing rural connections, family and other commitments were more likely to be in metropolitan areas (Ogden et al. 2020). GPs and trainees could live in metropolitan areas and commute to outer metropolitan areas, if this was appropriately incentivised and their educational experiences in OM areas had given them the skills and confidence to work there effectively.

Strengths and limitations

This study is the first to identify the distinct general practice training and working environment presented by OM regions. The evidence presented is supported by a large sample size, high response rate for studies of GPs (85%) and a representation of practices across the Australian Statistical Geographical Classification – Remoteness Area (ASGC-RA) spectrum. This study is limited by using indirect markers of the ‘richness’ of the training environment, a concept which remains difficult to define.

Implications for future research

Further research would examine whether the distinct OM training environment presents challenges of its own that could prove to be opportunities for professional growth or potentially predispose to greater workload, responsibility and burnout. Now that a distinct training environment in OM areas has been identified, the exact nature of its impact on registrars and their subsequent retention in the workforce in this area needs to be explored to inform future planning and policy development.