The Impact of Telepsychiatry on Spatial Equity of Medicare‐Reimbursed Private Psychiatric Service Utilisation Across States and Territories in Australia (2005–2023): A Time Series Analysis

ABSTRACT

Objective

To examine the trend of locational equity in utilising Medicare‐reimbursed psychiatric consultations (face‐to‐face and telepsychiatry) across states/territories.

Design

Time series analysis of administrative data.

Setting

Medicare‐reimbursed private psychiatric consultations.

Main Outcome Measures

Greater coefficients of variation (CVs) represented higher state/territory variations in per capita consultations (2005 Q3–2023 Q4). We ran an autoregressive integrated moving average model for telepsychiatry‐related CV change, with restricted rural‐remote telepsychiatry items (2005–2022) and unrestricted telepsychiatry items (2020 Q1–2023 Q4) as the predictors. We examined the association between distance from mean per capita consultations and the fraction of per capita telepsychiatry consultations using state/territory‐level data in another regression analysis, with telehealth legislative policy changes as another independent variable.

Results

Including telepsychiatry items since their introduction resulted in lower all‐item CVs (i.e., reduced state/territory variation) than the corresponding CVs for face‐to‐face items only. The rural‐remote‐restricted telepsychiatry items had a larger effect than the unrestricted ones. Every 1000 restricted consultations were associated with 0.287 units of CV improvement (i.e., reduction in regional variation), in contrast to 0.007 units per 1000 consultations for unrestricted items. Telehealth‐enabling policy changes significantly modified the association between distance from mean per capita face‐to‐face consultations and the fraction of telepsychiatry consultations. Lower per capita face‐to‐face consultations were associated with higher telepsychiatry fractions.

Conclusion

Telepsychiatry may help improve mental health equity. Bulk‐billed, incentivised telepsychiatry for rural and remote areas may enhance regional health access.

What is already known on this subject

• There was a large increase in Medicare‐reimbursed telepsychiatry consultations following the expansion and consolidation of new Medicare Benefits Schedule (MBS) telehealth items after the onset of the COVID‐19 pandemic.

• Telepsychiatry can potentially improve access for rural and remote patients, but the effects of telepsychiatry on improving regional access equity still require further research.

What this paper adds

• We found that MBS telepsychiatry items, especially those restricted to rural remote areas, were associated with reduced regional variability in per capita Medicare‐reimbursed psychiatric consultations.

• We demonstrated that with telehealth‐enabling policies, telepsychiatry appeared to supplement face‐to‐face consultations as lower per capita face‐to‐face consultations were associated with higher fractions of telepsychiatry consultations.

1. Introduction

Among the main objectives of Medicare is to achieve equitable access to Australian health services through the national universal health insurance scheme that reimburses patients consulting private healthcare practitioners, including psychiatrists [1]. It is critical to clarify whether Medicare‐reimbursed (Medicare Benefits Schedule—MBS) private psychiatric services have led to equitable access to psychiatric care across geographic areas and socio‐economic categories. Although Medicare can improve financial access to healthcare, the distribution of healthcare providers, in this case, psychiatrists, can restrict access for some Australians [2, 3].

An early health economic study found high variations in per capita MBS psychiatric service utilisation across states and territories from 1984 to 2001 [4]. The prevalence of mental illness is fairly uniform across Australia. As shown in the National Study of Mental Health and Wellbeing, the 12‐month prevalence of mental disorders ranged from 19.5% to 28.8%, but with overlapping margins of error across states and territories, resulting in no statistically significant differences [5]. Thus, regional variations in service utilisation rates might be due to other factors such as remoteness and economic variables (e.g., a low supply of psychiatrists and high out‐of‐pocket costs). Telepsychiatry may improve access for rural and remote patients by potentially addressing the supply‐side aspect [6]. However, the effects of telepsychiatry on improving regional equity in service utilisation have not yet been researched in detail.

To examine the impact of telepsychiatry services, we must consider the stages of telehealth provision in the MBS because governmental Medicare legislative changes have determined their accessibility (Figure 1). The initial MBS telepsychiatry (videoconferencing) items were introduced in 2002 to improve patient access to care in rural and remote areas. Co‐claim item 288, which came with a 50% fee‐loading incentive, was introduced in 2011 to encourage uptake through subsidy for specific areas classified as rural and remote using the Monash Modified Model [7, 8]. Prompted by the unprecedented disruptions caused by the COVID‐19 pandemic, telehealth items (including video and telephone items for telepsychiatry) were expanded considerably during the pandemic in March 2020 [9]. These newer telehealth items could be accessed without geographical restrictions. Item 288 ceased in January 2022 but was replaced by another fee‐loading co‐claim item, Item 294, in November 2022 [10, 11].

FIGURE 1.

Phases of Australian governmental telehealth legislation for the Medicare Benefits Schedule (MBS).

We hypothesised that variations in overall per capita consultations would be lessened with sequential telehealth items included than with face‐to‐face items alone. Access would be more equitable with additional care provision via telepsychiatry. We, therefore, compared variations in per capita psychiatric consultation items with and without telepsychiatry items from 2005 onwards.

We further examined the strength of associations of two categories of telepsychiatry items, restricted and unrestricted, with the change in CV in per capita consultations attributed to telepsychiatry. Restricted items (initial video items, previous and current co‐claim items) had limited application. These items could only be used by patients residing in approved rural and remote areas (as classified by the Monash Modified Model) and meeting the required distance from the psychiatrist to be eligible for Medicare reimbursement [8]. Unrestricted, nationwide items (new video and telephone items) did not have geographical restrictions.

Finally, we hypothesised that telepsychiatry decreased (improved) per capita psychiatric consultation variations by supplementing face‐to‐face consultations when these were low.

2. Materials and Methods

We reported this study in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines [12] (see Table S1: STROBE Checklist).

2.1. Data Sources

Publicly available reports of quarterly per capita (per 100 000 population) MBS items for psychiatric consultations by states/territories (Quarter 3, 2005–Quarter 4, 2023) were downloaded from the Services Australia website (http://medicarestatistics.humanservices.gov.au/statistics/mbs_item.jsp). The MBS item list is shown in Table S2. The data were aggregated into time series for all (face‐to‐face and telepsychiatry) consultations, face‐to‐face consultations and telepsychiatry consultations. Separate time‐series for restricted and unrestricted telepsychiatry items were also compiled.

A suitable measure was needed to quantify the variations in per capita consultations across states/territories over the study period. For the per capita consultation variable, there were 118 quarterly data points for the eight states/territories. The mean per capita consultations across states/territories changed greatly in this period. We used the coefficient of variation (CV) as our outcome measure to account for this difference in measuring the variability of per capita consultations. CV is a relative measure of variability that indicates the size of the standard deviation (σ _t) of a data series to its mean value (μ _t) [13, 14]. It is standardised and unitless, thus allowing direct comparisons of data dispersion between different data series. Using CV, we could compare the variability of state/territory per capita consultations temporally and between categories of consultations.

Telehealth policy was a categorical variable representing four phases of Australian telehealth legislative policies governing MBS telepsychiatry reimbursement: initial items (before 2011 Q3), co‐claim item introduction (2011 Q3 onward), expansion (2020 Q1 onward) and consolidation (2022 Q1 onward).

A summary of the definition and operationalisation of the variables used in the analyses is shown in Table 1.

TABLE 1.

Definition and operationalisation of study variables.

Variable	Definition	Operationalisation
Per capita consultations	Total MBS private psychiatric consultations per 100 000 population in state/territory i at quarter t.	$\frac{\mathrm{MBS}{\text{consultations}}_{\mathrm{i},\mathrm{t}}}{\text{Population}}\times 100,000$
Coefficient of variation (CVt)	Measure of relative dispersion across states/territories at quarter t. Lower values indicate lower dispersion.	${\mathrm{CV}}_{\mathrm{t}}=\left(\frac{{\sigma }_{\mathrm{t}}}{{\mu }_{\mathrm{t}}}\right)\times 100$
CV improvement (CVdiff,t)	Reduction in spatial variation achieved by adding telepsychiatry to face‐to‐face items.	${\mathrm{CV}}_{\text{diff},\mathrm{t}}={\mathrm{CV}}_{\mathrm{t},\mathrm{F}2\mathrm{F}}-{\mathrm{CV}}_{\mathrm{t},\mathrm{all}\text{items}}$
Distance from mean (DMi,t)	Relative deviation of state/territory face‐to‐face per capita consultations from the national quarterly mean.	$\mathrm{D}{\mathrm{M}}_{\mathrm{i},\mathrm{t}}=\frac{Y_{\mathrm{i},\mathrm{t},\mathrm{F}2\mathrm{F}}-{\overline{Y}}_{\mathrm{t},\mathrm{F}2\mathrm{F}}}{{\overline{Y}}_{t,F2F}}$
Telepsychiatry fraction (TFi,t)	Ratio of telepsychiatry consultations to total psychiatric consultations in state/territory i at quarter t.	${\mathrm{TF}}_{\mathrm{i},\mathrm{t}}=\frac{{\text{Telepsychiatry consultations}}_{\mathrm{i},\mathrm{t}}}{{\text{Total consultations}}_{\mathrm{i},\mathrm{t}}}$
Restricted items	MBS telepsychiatry items restricted to rural/remote areas.	Sum of quarterly claims for items: 353, 355, 356, 357, 358, 359, 361, 90 262, 90 268, 288, 294.
Unrestricted items	Universal MBS telepsychiatry items without geographical restriction.	Sum of quarterly claims for items: 91827, 91 828, 91 829, 91 830, 91 831, 92 434, 92 435, 92 436, 92 437, 92 458, 92 459, 92 460, 92 455, 92 456, 92 457, 92 162, 92 172, 91 837, 91 838, 91 839, 91 840, 91 841, 92 474, 92 475, 92 476, 92 477, 92 498, 92 499, 92 500, 92 495, 92 496, 92 497, 92 166, 92 178.

2.2. Data Analysis

Statistical analyses were performed using Stata 18 (Statacorp LLC, College Station, TX). We mapped variations in per capita psychiatric consultations in Australian states/territories from Quarter 3, 2005 (when per capita statistics for telepsychiatry consultations were first reported) until Quarter 4. In the first part of the study, descriptive statistics and time series plots of CV were generated. Differences in CV (CV improvement, CV_diff) with or without telepsychiatry were calculated and plotted. CV differences did not show seasonality. To model the association of restricted and unrestricted telepsychiatry items with CV_diff, we fitted an Autoregressive Integrated Moving Average with Exogenous Variables (ARIMAX) model. Augmented Dickey‐Fuller unit‐root tests indicated non‐stationarity in the raw CV improvement series, which achieved stationarity following first‐differencing (d = 1, p < 0.001). Using the arimasoc command, we evaluated combinations of autoregressive (AR) and moving‐average (MA) terms through automated information criteria selection. The ARIMAX(2,1,2) model minimised both the Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC).

In the second part, panel data consisting of quarters (level‐1 units) and states/territories (level‐2 units) for distance from mean (DM) for face‐to‐face consultations and fraction of per capita telepsychiatry consultations (TF) were constructed. At each time point, the DM values for every state/territory were calculated by finding the difference between each value and mean and dividing it by the mean. Per capita counts higher than mean produced positive values, and vice versa. TF values were calculated by dividing telepsychiatry counts by total (face‐to‐face and all telepsychiatry) consultation counts. The states/territories with larger deviations from the mean (“outliers”) contributed more to the variability in per capita consultations than the states/territories with smaller deviations. If telepsychiatry had a supplementary effect, we should see outliers deviating negatively from mean values having larger TF values, i.e., an inverse relationship between per capita consultations and the fraction of telepsychiatry consultations. Linear cross‐sectional time series regression with panel‐corrected standard errors [15] was performed, with TF as the dependent variable, and DM and telehealth policy phases as the independent variables. An interaction term between the independent variables was also included. We used Prais–Winsten regression with panel‐specific AR(1) autocorrelation adjustment to account for temporal autocorrelation and heterogeneous time‐series dynamics across states and territories. Models assuming no autocorrelation and common AR(1) autocorrelation were conducted as sensitivity analyses.

2.3. Ethical Considerations

The Australian government approved the MBS data used in the analysis for public availability in an aggregate, deidentified format through Services Australia. Formal ethics exemption was granted for this study.

3. Results

Table 2 provides the descriptive statistics of the study variables. Face‐to‐face consultations vastly outnumbered telepsychiatry consultations (measured as MBS items accessed). Although unrestricted items were more recently introduced, these consultations were more numerous than restricted remote and regional items. Telepsychiatry consultations had a higher median CV than face‐to‐face consultations. The time series plots for the CVs of per capita psychiatric consultations are shown in Figure 2. The CV of all‐item per capita consultations remained comparatively stable. In contrast, the CV levels of telepsychiatry items were much higher in the early period and fluctuated greatly before approaching similar levels as face‐to‐face items from 2020 onward (Figure 2A). The CV of all items became lower than that of face‐to‐face items since 2005 when telepsychiatry consultations appeared in per capita data. This was in addition to a gradually decreasing trend shown by face‐to‐face items from around 2015 onward.

TABLE 2.

Sums, medians and interquartile ranges (IQRs) of study variables.

Variable		Sum	Median	IQR a
All telepsychiatry consultations		3 704 857	8360	427, 20 286
	Restricted items	684 470	6201	407, 14 778
	Unrestricted items a	3 020 387	191 522	171 623, 225 433
Face‐to‐face consultations		28 838 029	407 582	377 604, 427 660
Coefficient of variation, CV (all consultations)		—	38.74	37.24, 41.91
	CV (telepsychiatry consultations)	—	83.87	65.53, 112.77
	CV (face‐to‐face consultations)	—	40.88	38.24, 44.18

^a 2019 Q4 to 2023 Q4.

FIGURE 2.

Time series plots of the coefficient of variations of per capita consultations for (A) All items, face‐to‐face items, and video items and (B) CV improvements when telepsychiatry consultations were included (left‐side y‐axis), with total psychiatric consultations, face‐to‐face psychiatric consultations, all telepsychiatry consultations and restricted and unrestricted telepsychiatry items plotted in total numbers (right‐side y‐axis).

Since the inclusion of new telehealth items from 2020 onward, there has been a very marked increase in telepsychiatry consultations, with corresponding large but fluctuating improvements in the CV values (Figure 2B—double y‐axis graph, CV on left y‐axis). The median CV improvement was 1.42 (IQR: 0.06–3.26), with a maximum improvement of 11.32 in 2021 Q3. The newly introduced unrestricted items registered a great increase in consultations since early 2020 (Figure 2B—numbers of telepsychiatry consultations on the right y‐axis). The old, restricted items were also used at slightly higher levels during the first year of the pandemic. By early 2022, there was a marked drop in the number of restricted items as they were abolished (with a corresponding decline in CV improvement) before the numbers rose again when the new co‐claim item was introduced in late 2022.

Pearson's correlation coefficients indicated strong positive bivariate correlations between CV improvement and restricted telepsychiatry items (rho = 0.88, p < 0.001) and unrestricted telepsychiatry items (rho = 0.76, p < 0.001). Associations of CV improvement (i.e., reduction) with restricted telepsychiatry items and unrestricted telepsychiatry items are visualised in scatterplots (Figure S1).

The ARIMAX(2,1,2) model fitted the data well. Residual diagnostics demonstrated no significant remaining autocorrelation (Portmanteau test Q = 19.94, p = 0.974) (See Tables S3 and S4 for details of the ARIMA model). Both types of telepsychiatry consultations (restricted and unrestricted) had statistically significant associations with CV improvement (Table 3A). Restricted telepsychiatry items displayed a considerably larger effect size than the unrestricted items. Every 1000 restricted telepsychiatry consultations accounted for 0.287 units (95% CI: 0.241–0.332) of the reduction in CV when telepsychiatry was included in quarterly per capita consultations. Meanwhile, every 1000 unrestricted telepsychiatry consultations accounted for 0.007 units (95% CI: 0.003–0.011) in the reduction.

TABLE 3.

(A) Summary of the ARIMAX(2,1,2) model estimates for predictors of Coefficient Variation Improvement (CV_diff); (B) Summary of the results of linear cross‐sectional time series model for telepsychiatry fraction. Results from the model with panel‐specific autocorrelation are shown.

A.
Predictor variable	Coefficient (β)	Std. error	p‐value	95% CI
Restricted items	0.2867	0.0233	< 0.001	0.2411 to 0.3324
Unrestricted items	0.0071	0.0019	< 0.001	0.0034 to 0.0107
Constant	0.0091	0.1280	0.943	−0.2417 to 0.2599

B.
	Without interaction				With interaction
	Coefficient	SE	p‐value	95% CI	Coefficient	SE	p‐value	95% CI
Difference from the mean (DM)	−0.113	0.011	< 0.001	−0.135, −0.091	−0.018	0.016	0.273	−0.05, 0.014
Telehealth policy
Co‐claim item	0.050	0.016	0.002	0.019, 0.082	0.06	0.018	0.001	0.025, 0.096
Expansion	0.353	0.024	< 0.001	0.306, 0.400	0.375	0.027	< 0.001	0.323, 0.427
Consolidation	0.386	0.025	< 0.001	0.337, 0.434	0.402	0.027	< 0.001	0.348, 0.455
Interaction of DM with:
Co‐claim item introduction	—	—	—	—	−0.1	0.021	< 0.001	−0.141, −0.059
Expansion	—	—	—	—	−0.187	0.025	< 0.001	−0.236, −0.138
Consolidation	—	—	—	—	−0.164	0.032	< 0.001	−0.227, −0.101
Constant	0.016	0.013	0.189	−0.008, 0.041	0.006	0.014	0.682	−0.022, 0.033
rho	0.600, 0.128, 0.858, −0.176, 0.215, … 0.302				0.567, 0.151, 0.704, 0.036, −0.083, … 0.0.272

Note: N = 73. All variables are first‐differenced (d = 1). Coefficients show CV_diff reduction per 1000 consultations. Models estimated using Prais–Winsten regression with correlated panels and panel‐specific AR(1) autocorrelation adjustment. Dependent variable is telepsychiatry fraction.

The median TF (i.e., the ratio of telepsychiatry to total consultations) was 0.014 (IQR = 0.001–0.166), while the median DM for per capita consultations was 0.043 (IQR = −0.244–0.330). The relationship between TF and DM for per capita consultations became more apparent in the later policy phases (Figure 3).

FIGURE 3.

Scatterplots of coefficient between telepsychiatry fraction and difference from mean per capita consultations in four policy phases: (A) Initial items (before 2011 Q3), (B) Co‐claim item introduction (2011 Q3 onward), (C) Expansion (2020 Q1 onward) and (D) Consolidation (2022 Q1 onward).

In the regression analysis of TF with interaction terms, DM on its own was not a statistically significant factor during the baseline initial policy phase (β = −0.018, p = 0.273), demonstrating that regional deficits in face‐to‐face consultations did not initially drive higher telepsychiatry uptake. (Table 3B). Each telehealth policy phase further widening access to telepsychiatry services was associated with increased telepsychiatry fraction as expected, especially the later expansion and consolidation phases. Statistically significant inverse interactions between distance from mean and telehealth policy phases were observed across all subsequent periods: co‐claim item introduction (β = −0.100, 95% CI: −0.141 to −0.059, p < 0.001), expansion (β = −0.187, 95% CI: −0.236 to −0.138, p < 0.001) and consolidation (β = −0.164, 95% CI: −0.227 to −0.101, p < 0.001). The inverse associations meant that a bigger negative difference from mean per capita consultations was associated with a higher telepsychiatry fraction, and this relationship was modified by telehealth policy changes. Sensitivity analyses confirmed that the directionality and statistical significance of all interaction terms remained robust across alternative autocorrelation specifications (Table S5).

Marginal effect analysis demonstrates that newer telehealth policies increased telepsychiatry's share of total consultations, with greater deficits in face‐to‐face consultations (negative DM values) associated with large fractions of telepsychiatry consultations. This compensatory pattern was strongest during the expansion and consolidation phases (Figure S2).

4. Discussion

In this time series analysis of per capita MBS psychiatric consultations, we focused on the degrees of variation among states and territories, represented by the CV, examining how its pattern changed over time, especially with the advent of new telepsychiatry items to the MBS. We sought to measure the strength of the association of restricted telepsychiatry items for rural and remote areas and unrestricted nationwide items with CV improvement. We further tested the hypothesis that telepsychiatry improved regional per capita consultation variations through complementary greater telepsychiatry usage when total per capita consultations were low.

Overall, the variations in per capita psychiatric consultations across states and territories were mostly consistent, indicating persistent regional inequity in MBS‐reimbursed psychiatric care utilisation, concurring with previous findings [4]. Importantly, since the inclusion of MBS telepsychiatry items, their usage was associated with a noticeable decrease in the CV of per capita psychiatric consultations. This indicated an improved regional equity of service utilisation across states and territories. In the second analysis, we demonstrated an inverse association between the fraction of telepsychiatry consultations and a negative deviation from the average per capita consultations when the effects of telepsychiatry‐enabling policy changes were considered. This finding supports the hypothesis that telepsychiatry improves regional equity in service utilisation.

Both restricted telepsychiatry items tailored specifically to serve rural and remote areas (initial video items and co‐claim items) and unrestricted new telepsychiatry items were associated with a decreased per capita consultation CV. However, telepsychiatry services targeting rural and remote populations appeared to be more important in achieving equity of access. This inference is further strengthened by the observation that when the previous co‐claim item 288 was removed, CV improvement reversed rapidly despite the ongoing availability of the newer unrestricted items. CV improvement only returned when the new co‐claim item 294 was introduced. The 50% fee loading to bulk‐billed telepsychiatry services for patients in rural and remote areas with the co‐claim items could be an important incentive. Given the chronic shortage of psychiatrists in rural and remote areas [3], understandably, limited access for patients in these areas is a major aspect of inequality in psychiatric care [16]. Our findings show that telepsychiatry items specifically targeting rural and remote areas through incentive schemes could help address this problem. Targeted telepsychiatry items catering to patient populations with poor access to care should be further refined and enhanced.

Nevertheless, our study findings suggest that the new unrestricted items, although not targeting rural and remote areas, were also utilised by patients in these areas, contributing to reduced geographical variations in per capita consultations. This is despite the fear that opening MBS telepsychiatry access to metropolitan patients could exacerbate health inequity by diverting finite clinical psychiatry resources [17]. In another analysis, we have shown that the usage of all telepsychiatry items since MBS telehealth expansion was inversely associated with the availability of rural psychiatrists [18]. While higher utilisation levels indicate greater availability, a distinction must be drawn between service availability/utilisation and true equity of access. Healthcare access is multidimensional, encompassing availability, affordability, accommodation, acceptability and appropriateness [19]. While the reduction in state/territory CV indicates greater geographic parity in service volume, it does not guarantee equitable access. As rural patients with fewer options for accessing face‐to‐face consultations turn to telepsychiatry, they often bear greater out‐of‐pocket costs than their metropolitan counterparts when they use the unrestricted items [20]. Consequently, while unrestricted items improve geographic availability, they risk widening metro‐rural financial health inequities.

Assuming a similar prevalence of mental illness and per capita patient population size across states and territories [21], negative deviations from mean per capita consultations represented unmet needs for psychiatric care in states/territories, presumably due to barriers to face‐to‐face consultations. This implied a potential demand for telepsychiatry services as alternatives to face‐to‐face consultations. Funding support is important for telepsychiatry services, especially in rural and remote areas, with specifically increased incentives for these areas [22]. When reimbursements for telepsychiatry were not easily available, this potential demand did not translate into an actual increase in telepsychiatry consultations. With each legislative policy change extending Medicare funding to more telepsychiatry services, we could see increased uptake of telepsychiatry in states/territories with lower per capita consultations.

The financial landscape of MBS telepsychiatry has evolved considerably across the four telehealth policy phases. During Phase 1 (Initial) and Phase 2 (Co‐Claim 2011–2020), remote telepsychiatry was tethered to geographical criteria and supported by fee‐loading incentives, which constrained corporate entry and encouraged bulk‐billing. In Phase 3 (Expansion, 2020–2022) and Phase 4 (Consolidation, 2022–present), universal unrestricted items allowed private telehealth services to expand rapidly nationwide, even though another fee‐loading item (294) is still available. This shift possibly facilitated the emergence of commercialised telepsychiatry consultations focusing on brief or single‐issue diagnostic assessments (e.g., adult ADHD evaluations) charging high gap fees [20, 23, 24]. While these emerging services might have expanded utilisation in historically underserved states, they could be disproportionately catering to socioeconomically advantaged individuals, potentially crowding out long‐term, complex psychiatric care for low‐income rural residents. Appropriate government policies regulating financial support for access to care may be needed.

An important limitation of this study is that the MBS data only covered private psychiatric consultations reimbursed by the Australian Government Medicare scheme. Therefore, the data did not provide information on psychiatric consultations in the public sector. It is possible that public psychiatric services helped to improve disparities in per capita consultations to some extent. However, as reported by the Australian Institute of Health and Welfare (AIHW), there were 9.6 million contacts with about 468 800 patients in 2021–2022 by public sector community mental health services, but above 2.7 million patients were accessing nearly 13.2 million Medicare‐subsidised mental health services in 2022–2023 [25]. Given the relatively smaller coverage of public services, equitable access to Medicare‐reimbursed services is highly important, and targeted (incentivised) telepsychiatry items can contribute to this end.

Inequities in access to MBS psychiatric services based on gender, rurality and socioeconomic disadvantage were found in previous studies [16, 26]. Unfortunately, these variables were unavailable in the state/territory‐level aggregated data analysed for this study. Research on the impact of telepsychiatry on sociodemographic, economic and geographical disparities, as well as clinical categories in the access to psychiatric care based on more granular data, is needed. Within the constraints of aggregated MBS claims data, our analyses captured the trend of variations in per capita psychiatry consultations across states/territories. Our findings suggest that the role of telepsychiatry and the possible mechanism in improving this variability is a novel and valuable contribution to this research area.

5. Conclusion

MBS‐reimbursed telepsychiatry may contribute to equity of access to psychiatric treatment. Contrary to the fear that expanded telepsychiatry items introduced during the COVID‐19 pandemic open to metropolitan patients could jeopardise regional access equity, these services might also help reduce the health access gap for rural and remote Australians, but with possible negative financial implications. Our study findings suggest that maintaining incentivised telepsychiatry items for rural and remote areas may promote regional access equity across Australian states and territories.

Author Contributions

Luke Sy‐Cherng Woon: conceptualization, methodology, data curation, formal analysis, writing – original draft, writing – review and editing, project administration, investigation, visualization, software. Tarun Bastiampillai: writing – review and editing, supervision. David Smith: writing – review and editing, formal analysis, validation. Paul A. Maguire: writing – review and editing. Rebecca E. Reay: writing – review and editing. Jeffrey C. L. Looi: conceptualization, supervision, resources, writing – review and editing.

Funding

The authors have nothing to report.

Ethics Statement

Formal exemption from ethics approval was granted by the Australian National University Human Ethics Office, as this study exclusively utilises publicly accessible, aggregate and de‐identified Medicare Benefits Schedule (MBS) data from Services Australia.

Conflicts of Interest

J.L. provides private sector face‐to‐face psychiatric consultations and telepsychiatry. T.B. briefly provided telepsychiatry services during the COVID‐19 pandemic lockdown period but has engaged in no telepsychiatry clinical practice within the past five years. R.E. provides allied health face‐to‐face and telehealth consultations in the private sector. L.W., D.S., and P.M. do not provide Medicare‐reimbursed telepsychiatry consultations. The authors declare no other financial relationships, commercial interests, or affiliations with telehealth software providers that influenced the design, analysis, or interpretation of this study.

Supporting information

Figure S1: Scatterplots of coefficient of variation improvements against (A) Restricted items (2005 Q3–2023 Q4) and (B) Unrestricted items (2019 Q4–2023 Q4).

Figure S2: Predictive margins of telepsychiatry fraction with 95% CIs after co‐claim item introduction, telehealth expansion and telehealth consolidation across a range of differences from mean per capita face‐to‐face consultations. Based on the primary Prais–Winsten model with panel‐specific AR(1) autocorrelation adjustment.

Table S1: STROBE Statement—Checklist of items that should be included in reports of cross‐sectional studies

Table S2: MBS items for psychiatric consultations included in the study.

Table S3: Lag‐order model selection criteria.

Table S4: ARIMAX(2,1,2) parameter estimates and model diagnostics.

Table S5: Summary of the results of additional linear cross‐sectional time series models for telepsychiatry fraction: without autocorrelation and with common autocorrelation.

Data Availability Statement

The data that support the findings of this study are available in Medicare Statistics at https://www.servicesaustralia.gov.au/medicare‐statistics?context=22. These data were derived from the following resources available in the public domain: Medicare item reports, http://medicarestatistics.humanservices.gov.au/statistics/mbs_item.jsp.