References Australian Government Department of Health and Aged Care. (n.d.). Telehealth. https://www.health.gov.au/topics/health-technologies-and-digital-health/about/telehealth Australian Institute of Health and Welfare. (2018). Mental health services provided by general practitioners: 2018 update. https://www.aihw.gov.au/getmedia/0e102c2f-694b-4949-84fb-e5db1c941a58/aihw-hse-211.pdf?v=20230605175041&inline=true Australian Institute of Health and Welfare. (2022, December 13). Australian Burden of Disease Study 2022: Summary. https://www.aihw.gov.au/reports/burden-of-disease/australian-burden-of-disease-study-2022/contents/summary Australian Institute of Health and Welfare. (2022). Mental health. Prevalence and impact of mental illness [internet]. https://www.aihw.gov.au/mental-health/overview/prevalence-and-impact-of-mental-illness Australian Institute of Health and Welfare. (2025, December 2). Mental health. Australia’s mental health system. https://www.aihw.gov.au/mental-health/overview/australias-mental-health-system Bartholomaeus, J. D., Collier, L. R., Lang, C., Cations, M., Kellie, A. R., Inacio, M. C., & Caughey, G. E. (2023). Trends in mental health service utilisation by Australia’s older population. Australasian Journal on Ageing, 42(1), 159–164. https://doi.org/10.1111/ajag.13118 Bruzzo-Gallardo, S., Genie, M. G., Gallagher, R., & Paolucci, F. (2025). Telehealth evolution and policy response in Australia: Insights from the COVID-19 pandemic. In P. Ordóñez De Pablos, M. N. Almunawar, & M. Anshari (Eds.), Information technologies in healthcare industry (Vol. 5, pp. 561–592). Academic Press. https://doi.org/10.1016/B978-0-443-30168-1.00002-5 Carswell, O., Morgan, L., Wait, S., Ruszanov, A., & Valiotis, G. (2023). Health system readiness for innovation: Putting research into practice to drive effective implementation. The Health Policy Partnership. https://www.healthpolicypartnership.com/app/uploads/Health-system-readiness-for-innovation-putting-research-into-practice-to-drive-effective-implementation.pdf Davis, F. D. (1989). Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quarterly, 13(3), 319–340. https://doi.org/10.2307/249008 De Guzman, K. R., Snoswell, C. L., Caffery, L. J., Wallis, K. A., & Smith, A. C. (2022). Costs to the Medicare Benefits Schedule for general practitioner consultations: A time-series analysis. Journal of Telemedicine and Telecare, 28(10), 726–732. https://doi.org/10.1177/1357633X221122135 De Guzman, K. R., Snoswell, C. L., & Smith, A. C. (2022). The impact of telehealth policy changes on general practitioner consultation activity in Australia: A time-series analysis. Australian Health Review, 46(5), 605–612. https://doi.org/10.1071/AH22058 Garavand, A., Mohseni, M., Asadi, H., Etemadi, M., Moradi-Joo, M., & Moosavi, A. (2016). Factors influencing the adoption of health information technologies: A systematic review. Electronic Physician, 8(8), 2713–2718. https://doi.org/10.19082/2713 Gayathri, S., & Buvaneswari, P. S. (2019). The Technology Acceptance Model: A review of theories and models. https://ijrar.org/papers/IJRAR19K3028.pdf Górczak, K., Burzykowski, T., & Claesen, J. (2025). A hierarchical negative-binomial model for analysis of correlated sequencing data: Practical implementations. Bioinformatics Advances, 5(1), vbaf126. https://doi.org/10.1093/bioadv/vbaf126 Hall Dykgraaf, S., Desborough, J., Sturgiss, E., et al. (2022). Older people, the digital divide and use of telehealth during the COVID 19 pandemic. Australian Journal of General Practice, 51(8).* https://www1.racgp.org.au/getattachment/9863b056-b31f-4f0d-bc06-053cec8f6b5b/Older-people-and-use-of-telehealth.aspx Hashmi, R., Alam, K., Gow, J., Alam, K., & March, S. (2023). Inequity in psychiatric healthcare use in Australia. Social Psychiatry and Psychiatric Epidemiology, 58(4), 605–616. https://doi.org/10.1007/s00127-022-02310-1 Jayawardana, D., & Gannon, B. (2021). Use of telehealth mental health services during the COVID 19 pandemic. Australian Health Review, 45(4), 442–446. https://doi.org/10.1071/AH20325 Klaic, M., Kapp, S., Hudson, P., Chapman, W., Denehy, L., Story, D., & Francis, J. J. (2022). Implementability of healthcare interventions: an overview of reviews and development of a conceptual framework. Implementation science: IS, 17(1), 10. https://doi.org/10.1186/s13012-021-01171-7 Lee, A. T., Ramasamy, R. K., & Subbarao, A. (2025). Understanding psychosocial barriers to healthcare technology adoption: A review of TAM technology acceptance model and unified theory of acceptance and use of technology and UTAUT frameworks. Healthcare, 13(3), 250. https://doi.org/10.3390/healthcare13030250 Lee, J. S., Bhatt, A., Pollack, L. M., Jackson, S. L., Chang, J. E., Tong, X., & Luo, F. (2024). Telehealth use during the early COVID 19 public health emergency and subsequent health care costs and utilization. Health Affairs Scholar, 2(1), qxae001. https://doi.org/10.1093/haschl/qxae001 Le, J. T., Mahoney, A. E. J., Court, J. L., & Shiner, C. T. (2025). Barriers and facilitators of digital mental health use in regional, rural, and remote Australia. Australian Journal of Rural Health. https://doi.org/10.1111/ajr.70011 Mbatha, B. (2024). Diffusion of innovations: How adoption of new technology spreads in society. In Information, knowledge, and technology for teaching and research in Africa (pp. 1–18). Springer. https://doi.org/10.1007/978-3-031-60267-2_1 MBS Review Advisory Committee. (2024). Telehealth post-implementation review: Final report. Australian Government Department of Health. https://www.health.gov.au/sites/default/files/2024-06/mbs-review-advisory-committee-telehealth-post-implementation-review-final-report.pdf McMaster, K. (2025). Australia’s shift to digital health—How telehealth is redefining mental health care. https://healthhub.hif.com.au/mental-health/australia-s-shift-to-digital-health-how-telehealth-is-redefining-mental-health-care RACGP. (2024). The RACGP position on the use of telehealth in general practice. https://www.racgp.org.au/advocacy/position-statements/view-all-position-statements/clinical-and-practice-management/racgp-position-telehealth-general-practice Rogers, E. M. (2003). Diffusion of innovations (5th ed.). Free Press. https://books.google.com/books?id=9U1K5LjUOwEC Services Australia. (n.d.). Statistics—Item reports. http://medicarestatistics.humanservices.gov.au/statistics/mbs_item.jsp Wang, C. P., Mkuu, R., Andreadis, K., Muellers, K. A., Ancker, J. S., Horowitz, C., Kaushal, R., & Lin, J. J. (2024). Examining and addressing telemedicine disparities through the lens of the social determinants of health: A qualitative study of patient and provider during the COVID 19 pandemic. AMIA Annual Symposium Proceedings, 2023, 1287–1296. https://pmc.ncbi.nlm.nih.gov/articles/PMC10785927/ World Health Organization. (2022). Health systems resilience toolkit. https://www.who.int/publications/i/item/9789240048751

Overall, substantial telehealth use was associated with COVID 19 telehealth expansion. However, this was followed by a reversion to a stable low-level range. Adjustment for state, age, and gender, with the ITS augmented Negative Binomial GEE model, ensured a better capture of the national pattern. Significant differences persisted across states/territories, age groups, and gender throughout the study period. Higher telehealth share was noted in VIC, SA, younger and mid-adult age groups, as well as among females.

Discussion Principal findings The study was conducted to understand the evolution of telehealth for selected mental health services in Australia using national, population‐level Medicare data disaggregated by state/territory, age, and gender. The study indicates that there was a sudden rise in telehealth for GP mental health plan reviews due to COVID 19 expansion in early 2020, followed by a decline and a stabilisation at a lower level through the study period. At the national level, telehealth accounted for 0.5% of services at baseline (2020 Q1), peaked at almost 30% in 2020 Q2, before settling at about 13% by 2025 Q3. The ITS augmented negative binomial GEE models captured the abrupt step change and subsequent attenuation better than time-only or composition-adjusted models, with incremental gains from seasonal adjustment. Persistent differences were evident by state (higher in VIC and SA; lower in WA and QLD), age (higher in 0–4 and mid adult groups; lower in the oldest age groups), and gender (slightly higher in females). A large positive ITS step effect observed in 2020 Q2 was due to the immediate and unrestricted availability of universal telehealth funding and the shift to remote health services on account of pandemic-related restrictions, more of a substitution of in-person services. The negative post-interruption slope seen is suggestive of a general return of in-person services. The stability achieved through the study period in telehealth share, as indicated by 2025 Q3, suggests that the integration of telehealth has become a permanent feature of mental health services in Australia. Differences in telehealth share in geography and demography may reflect the disparities in digital access and literacy, provider supply, and patient mix (Wang et al., 2024; Hall Dykgraaf et al., 2022). The stark disparities between VIC and SA versus WA and QLD in telehealth share may map onto differences in urbanicity, practice models, and historical telehealth infrastructure. Higher shares in telehealth use among young children and mid adults, and lower use in the oldest groups, may be related to caregiver-mediated access (for young children), and related to employment among mid adults. Comparison With Previous Literature The findings of this study extend the previous studies on telehealth use in the initial pandemic-era (Jayawardana & Gannon, 2021; RACGP, 2024), revealing long-term stability after the early surge (Lee et al., 2024). This study also adds to the literature the quantification of population mean rates with nested NB GEE and ITS specifications. While previous studies reported rapid uptake of telehealth in the immediate pandemic period (Jayawardana & Gannon, 2021; RACGP, 2024), we have shown by a 23-quarter perspective analysis the sustained, moderate telehealth use over time, with clear geographic and demographic disparities and questions on equity. Strengths and contributions Leveraging on comprehensive national administrative datasets, this study was able to prepare a growth curve GLM, characterising the telehealth long-run adoption arc, and estimated population mean telehealth rates using NB GEE models. Furthermore, the study reinforced the robustness of the nested model through comparison and in-sample error metrics (MAE/RMSE). Limitations The limitations associated with this study include the issues involved in the use of administrative datasets. Hence, the study could not capture clinical appropriateness, symptom severity, socioeconomic circumstance, or patient-level outcomes. Furthermore, the ITS approach assumes a common interruption point (2020 Q2) and a linear post-interruption trend on the scaled time index, introducing unobserved heterogeneity. Residual confounding from unmeasured factors (e.g., the digital capability of a general practice) may have limited the findings of this study. Finally, the study focused only on GP mental health plan review items. Hence, the study findings may not be generalisable to other mental health services. Policy and practice implications The stable midrange telehealth share suggests that telehealth has become a normal part of routine GP mental health care, rather than a transient substitute. Policymakers need to continue to invest in stable reimbursement, reliability of platforms, and improved practice workflow, with a view to mitigating geographic and age-related disparities. It is also imperative to embed telehealth monitoring into mental health services monitoring and evaluation, towards improving hybrid mental health care models and reducing inequities. Future research Future studies should consider the linkage of item data to patient-level sociodemographic indicators in the analysis of equity-focused data, highlighting who benefits more from sustained telehealth use. Researchers should also explore whether the choice of modality affects care continuity, referrals, and outcomes. Conclusions Telehealth use for the review of mental health plans by GPs in Australia experienced a sharp pandemic-era expansion. This was followed by a stable, durable telehealth share. An ITS augmented NB GEE framework best captured the national trajectory and shed more light on enduring differences by state/territory, age, and gender. These findings support telehealth’s ongoing role within hybrid primary mental health care and highlight the need for policies that sustain access while addressing uneven adoption across populations and geographic locations.

share had settled at 12.7%.

Figure 2: National mental health GP consultations, by modality

Geographic variation Across states/territories, telehealth share ranged from 8.2% in WA to 17% in VIC. There was a general rise in telehealth in early 2020, followed by stabilisation, which differed in magnitude and persistence. VIC and SA had persistently higher telehealth share. On the other hand, WA and QLD (10.8%) had comparably lower telehealth shares throughout the study period. The other states/territories had telehealth share with the range: NSW (12.5%), TAS (11.9%), ACT (11.8%), and NT (11.4%).

Table 3: Key characteristics of in-person and telehealth equivalents (national and subgroups) Characteristic Value Study period (quarters) 2020Q1–2025Q3 (N = 23 quarters) National total services per quarter Range: 133,415–208,958 services per 100,000 population National telehealth services per quarter Range: 955–55,901 services per 100,000 population Telehealth share (national) Mean 14.5%; median 13.1% (share = (phone+video)/(in-person+phone+video)) Baseline telehealth share (2020Q1) 0.526% (955 / 181,507) Peak telehealth share (quarter) 29.52% in 2020Q2 (55,901 / 189,357) Latest quarter (national) 2025Q3: share 12.67% (18,229 / 143,907) Latest by gender – Female 13.35% (observed share, national) in 2025Q3 Latest by gender – Male 11.41% (observed share, national) in 2025Q3 Latest by age – Highest share 0–4 years: 18.53% (observed share) in 2025Q3 Latest by age – Lowest share 75–84 years: 7.83% (observed share) in 2025Q3 Latest by state – Highest share VIC: 17% (observed share) in 2025Q3 Latest by state – Lowest share WA: 8.2% (observed share) in 2025Q3

Age Group Patterns In the most recent period (2025 Q3), the highest telehealth share occurred in people aged 0-4 years and 35-44 years. Older groups (75-84 years) as well as ≥85 years had a telehealth share of 7.8 and 7.9% respectively. There were sustained differentials in telehealth share across the study period. Gender Differences There were slight differences in the telehealth share by gender. In the latest quarter, the observed telehealth share was 13.3% in females and 11.4% in males, compared to 13.4% (females) and 11.4% (males) predicted by the NB GEE model. Model Effects (GEE / ITS) Model M1–M4 comparison results (Appendix 1) showed that the ITS augmented model (M3) indicated a markedly improved fit (MAE 0.0203; RMSE 0.0275) compared with the time-only and demographic-adjusted models (M0–M2). Model fit appears to improve with adjustment for seasonality in M4 (MAE 0.0200; RMSE 0.0272). The ITS step term was positive (β = 4.09, 4.17), suggesting an immediate change in both M3 and M4 models. The slope term following the interruption was negative, indicating a gradual decline of the early surge.<br /> The models developed here reveal a limited fit using time only spline (M0), an improved fit with demographic adjustment (M1), a minimal additional gain from time interactions (M2), and a significant improvement with ITS (M3, M4). The models incorporating ITS aligned more closely with observed values (Figure 3). Table 4: Model coefficients Model Term Coef_logRR M3 ITS_step 4.09 M3 ITS_slope_scaled -2.38 M4 ITS_step 4.17 M4 ITS_slope_scaled -1.02

Figure 3: Telehealth share trends

Results Study Population and Descriptive Characteristics Study population and overall service volumes Across the study period (2020Q1–2025Q3), the panel dataset comprised state × gender × age × quarter service counts (per 100,000 population) with 3,680 valid strata, excluding cells with zero total mental health services. National quarterly total mental health services ranged from 134,866 to 208,958 services per 100,000 population, and quarterly telehealth counts ranged from 955 to 55,901 per 100,000 (Table 1). Overall trends in telehealth use At baseline (2020 Q1), the proportion of telehealth use was 0.5 % of all mental health service use. There was a marked change in 2020 Q2, where telehealth use rose to 29.5% (Figure 2). From mid 2020 onward, a steady decline in the telehealth shares was observed until it became stable. By the end of the study period (2025Q3), the national telehealth share had settled at 12.7%.

Diagnostics and sensitivity We conducted diagnostics and sensitivity tests for the model using observed versus fitted national shares for the GLM and GEE models, respectively, assessing trend capture and smoothing behaviour. We used Python (pandas, patsy, statsmodels, matplotlib) for data manipulation and modelling.<br /> Ethical Considerations This study was part of a larger study for which ethical clearance has been obtained (ETH2023-0357) from the UniSQ Human Research Ethics Committees (HREC), Toowoomba. We used publicly available data and de-identified datasets. Hence, informed consent was not applicable in the study.

Excess telehealth (Excessq) indicates additional telehealth services in quarter 𝑞 compared with what was expected to have occurred if the baseline 2020Q1 telehealth share had persisted. This was defined as, Excessq = Tq − E (Tq), where Tq is the observed number of telehealth services (phone + video) in quarter 𝑞, E (Tq), the expected telehealth activity in quarter 𝑞 based on the baseline share. We excluded strata with zero total services, preventing undefined proportions. Covariates Age group: This was captured in intervals (0-4, 5-14, 15-24, 25-25-34, 35-44, 45-54, 55-64, 65-74, 75-84, >=85). Gender: This was categorised as male and female. State/territory: This includes the Australian Capital Territory (ACT), New South Wales (NSW), Northern Territory (NT), Queensland (QLD), South Australia (SA), Tasmania (TAS), Victoria (VIC), and Western Australia (WA). Time: The quarterly time used in this study was encoded using an integer index representing sequential quarters since 2000 (e.g., 2020Q1 → 80). This was then normalised to the unit interval to improve numerical stability in spline estimation. This scaled time index (t_scaled) was used in all model specifications. Statistical Modelling Growth Curve Model Using aggregated per-quarter national totals, the study derived the national telehealth total as well as the national total mental health service as described above. Estimating the temporal trajectory of telehealth adoption at the national level, we fitted a binomial GLM with a logit link to stratum-level (gender × age × quarter) telehealth proportions. In this, we used outcome (telehealth proportion for each stratum), weights (stratum total service counts, i.e, frequency weights), predictor (flexible smooth function of time using a B spline basis with 5 degrees of freedom), and covariance (HC1 robust sandwich estimator for inference on spline coefficients), accounting for heteroskedasticity from differing stratum sizes. Generating the spline design matrix and model fit, we used patsy and statsmodels. The study computed the predicted telehealth shares and 95% confidence intervals for each quarter using the fitted GLM.

Negative Binomial Generalised Estimating Equations (GEE) Core specification Data was reshaped into one record per state × gender × age group × quarter, summing modalities as above to construct telehealth total, total mental health services, and telehealth share, underpinning the GEE and all nested model variants (M0–M4). Setting all negative values in the dataset to zero, we removed rows where the total mental health services (telehealth + in-person) were equal to zero. The study complemented the share model with a rate-based analysis on counts, estimating population mean telehealth rates using GEE with Negative Binomial mean–variance structure and log link, at the state × gender × age panel level: i) Outcome (telehealth counts 𝑌𝑖, 𝑞); ii) Offset (log (𝑁𝑖,𝑞) - modelling the rate per total mental health services); iii) Time (B spline, df=5 in scaled quarter); iv) Covariates (state, gender, age group - categorical main effects); v) Working correlation [AR(1) within each panel 𝑖 across quarters]; and vi) Inference [empirical (sandwich) covariance for robust standard errors]. Nested multivariable models & ITS To explain trends and quantify the interruption (assumed at 2020 Quarter 2), the study estimated a set of nested models: M0 (time only): Yi,q ∼ spline(time); M1 (+ composition): M0 + State + Gender + Age (main effects); M2 (+ differential slopes): M1 + Gender×time + Age×time (linear time interactions); M3 (ITS): M2 + ITS step (post 2020Q2) + ITS slope (quarters since 2020Q2; scaled); and M4 (ITS + seasonality): M3 + quarter of year fixed effects (C(qnum)). For each model, we computed national predictions. A consolidated model comparison table (MAE and RMSE on the national share scale) was also computed. In-sample predictive error using MAE and RMSE on the national share scale was summarised.

Variables Outcome variables, including measures derived The study defined telehealth total as the sum of the MBS telephone item (92127) and videoconference (91115), while total mental health services represent the sum of in person (2713) and telehealth total. For each statexgender×age×quarter stratum, we computed telehealth total (telephone + videoconference) and total mental health services (telehealth total + in person) Telehealth share: This is the proportion of total mental health services that were offered via telehealth, expressed in percentages (telehealth total/total mental health services X 100). This was computed for each quarter. Synthetic Baseline Telehealth Activity: This was defined as the estimated “excess” telehealth adopted relative to the patterns observed before the pandemic. Applying a fixed baseline telehealth share observed at Quarter 1, 2020, we computed the expected telehealth activity under the baseline for each quarter as follows: E (Tq) = p2020Q1 * total mental health servicesq, where E (Tq) is the expected number of telehealth services in quarter 𝑞 under the fixed baseline telehealth share, p2020Q1 is the baseline telehealth share estimated from 2020Q1 estimated in (i) above.

Study Period Nationwide expansion of telehealth services occurred in March 2020 in response to the COVID-19 pandemic, its effects, and related containment measures (Bruzzo-Gallardo et al., 2025), marking the introduction of unrestricted telehealth services in Australia. Furthermore, GP mental health consultation items 2713, 92115, 92127 were replaced by time tiered general attendance items from November 2025. Hence, the period of the study was limited to Quarter 1, 2020, to Quarter 3, 2025. The primary aims of the analyses in this study include: (i) Describing the national evolution of telehealth share over time; and (ii) Estimating population mean telehealth rates, adjusting for state/territory, age group, and gender. The study included quantification of the impact of a policy interruption (COVID-19 pandemic-related) using an Interrupted Time Series (ITS) specification, embedding it in a Negative Binomial GEE.

Methods Study setting Source of data The study analysed publicly available data on GP mental health consultations using national, administrative datasets from the Medicare Benefits Schedule (MBS), capturing all Medicare claims of outpatient mental health services in Australia. The dataset includes information on mental health consultations provided by GPs. The aggregated datasets include reported month of service use, age group, and gender by State/Territory. The national totals of the services were also included. The MBS item codes indicate the type of service, its duration, and the modality of service offered (in-person, videoconference, and telephone). Datasets used for the study were obtained from the website of Services Australia (Services Australia - Statistics - Item Reports, n.d.). These included monthly counts of services per 100,000 population. The three modalities of GP mental health consultation of at least 20-minute duration were examined: In-person, telephone, and videoconferencing, represented respectively by MBS items 2713, 92115, and 92127. These are defined as follows: i) 2713 - This refers to In-person consultation by a GP for a mental condition; ii) 92115- This indicates videoconference consultation by a GP for mental health condition; and iii) 92127 - This represents telephone consultation by a GP for mental disorders. These MBS items were used for mental health services, including relevant history taking, treatment, and advice, referrals for additional services or treatments where indicated, and recording the consultation's outcomes.

Quantifying the use of telehealth over a period by modality and demographic stratum requires the application of rigorous scientific methods capable of detecting patterns and changes (Bartholomaeus et al., 2023)Therefore, this study aimed to examine whether the use of telehealth services was sustained over time by applying two complementary analytical approaches: a growth curve binomial model and a hierarchical Negative Binomial Generalised Estimating Equation (NB-GEE) (Górczak et al., 2025). By combining national Medicare administrative datasets, disaggregated by age, gender, and States/Territories, this study aimed to provide novel information on telehealth use in mental health services in Australia. The findings of this study will provide policymakers with actionable evidence on the evolution, course, magnitude, and stability of mental healthcare services offered through various modalities in Australia.

Conceptual framework for the adoption and sustenance of telehealth services for mental health The conceptual framework for this study was based on the integrated evidence in the literature on the Diffusion of Innovations (Mbatha, 2024; Rogers, 2003), the Technology Acceptance Model (TAM)(Davis, 1989; Lee et al., 2025 ), the Normalization Process Theory (NPT)(Gayathri & Buvaneswari, 2019), and Health Systems Readiness Models (Carswell et al., 2023; World Health Organization, 2022). Studies have shown that new health service innovations, such as telehealth, digital tools, and other new products, services, and clinical pathways, often encounter challenges within the health system due to existing constraints and resources (MBS Review Advisory Committee, 2024). Hence, the key elements to consider in the context include the following: i) The policy environment and structures for reimbursement: The capacity of the workforce, digital literacy (in the context of telehealth), and the readiness for change in the organisational hierarchy determine the success of the new service or policy. Other factors relevant in this environment include the expectations of care and the sociocultural dimension, and the availability of infrastructure to implement the innovation. These factors determine whether the new service can be considered feasible and whether it would be implemented with speed and achieve sustainability in the long term (Klaic et al., 2022; MBS Review Advisory Committee, 2024; McMaster, 2025).<br /> ii) Adoption: The factors driving adoption include the perceived relative advantage of the new service (e.g. telehealth) in terms of convenience of use, clinical value compared with routine care (e.g. in-person consultation), opportunities to test the service before full implementation, alignment with current workflows, amongst others (Klaic et al., 2022; MBS Review Advisory Committee, 2024; McMaster, 2025; Garavand et al., 2016). iii) Implementation and integration: Successful integration depends on the extent to which the service fits into routine workflows, shared responsibilities of stakeholders, and the operational capacity of the system (Le et al., 2025). iv) Sustainment: The factors necessary for sustainment of the new service include the quality of the service, provider uptake, the satisfaction of patients who use the service, the extent to which the system is being monitored or evaluated, continuous quality improvement, organisational learning, and feedback loops (McMaster, 2025; Garavand et al., 2016) (Figure 1). The pathway a new service often follows includes adoption, implementation, sustainment, evaluation, and redesign. Continuous learning and improvement help in ensuring the refinement and sustainability of new services and products, including telehealth use in mental health services. This study is concerned with the evaluation of the use of telehealth in mental health services in Australia, where telehealth is part of the services available for mental health in the country (McMaster, 2025; Garavand et al., 2016).

Introduction Mental health is a significant public health challenge in Australia. Mental health disorders rank among the top causes of morbidity in the country. Almost half of Australians have experienced a mental health disorder at some point in their lives (Australian Institute of Health and Welfare, 2022). A previous study has also shown that the rate of use of mental health services increased significantly over time (Bartholomaeus et al., 2023). Inequities persist in access to and the use of mental health care services in Australia, especially in remote and very remote parts of the country (Hashmi et al., 2023; Australian Institute of Health and Welfare, 2025). Telehealth, designed to address barriers to mental health service use, continues to undergo significant changes with time (Australian Government Department of Health and Aged Care, n.d.). The March 2020 rapid expansion of telehealth services across Australia facilitated the uptake of mental health services offered through videoconferencing or telephone channels at an unprecedented scale. The adoption of telehealth as an alternative means of service delivery indicates an opportunity to offer services to target populations irrespective of their geographical locations. This represents a landmark shift in the organisation of mental health services nationwide. However, early adoption of health services does not guarantee the sustained use of such services. General practitioners play a key role in providing first-line mental health care services in Australia. An estimated 18 million mental health GP encounters occur annually in Australia, representing more than a tenth of all consultations by general practitioners (GPs) (Australian Institute of Health and Welfare, 2018). It was also cited that GP mental health consultation data in Medicare does not fully reflect GP mental health service provision, as GPs often provide mental health services in routine consultations (Australian Institute of Health and Welfare, 2018). Previous studies have shown the increased use of telehealth for mental health services post COVID-19 pandemic (De Guzman, Snoswell, Caffery, et al., 2022; De Guzman, Snoswell, & Smith, 2022). These studies examined the effects of telehealth on GP mental health consultations in the immediate pandemic impact period. However, there is a paucity of studies to ascertain whether telehealth was sustained in the long term.

Abstract Introduction: Telehealth expansion in 2020 in Australia led to an increase in its use in mental health services. This study was conducted to examine the evolution, magnitude, and persistence of telehealth services for mental health from 2020 Q1 to 2025 Q3. Methods. Analysing the national Medicare Benefits Schedule (MBS) administrative data, disaggregated by state/territory, gender, and age group, the study compared telehealth services with in-person consultations across 23 quarters. We used statistical methods, including a growth curve binomial model and a hierarchical Negative Binomial Generalized Estimating Equation (NB-GEE), to capture the effect of COVID-19 telehealth expansion and its sustainability, as well as disparities among different groups. Results From a baseline of 0.5% of services in 2020 Q1, telehealth use peaked at 29.5% in 2020 Q2 and stabilised at about 13% by 2025 Q3. The ITS augmented models captured the spike in telehealth and its subsequent decline. Disparities in use persisted, with higher use in Victoria and South Australia. Young children (0-4 years), middle-aged adults, and females also had comparably higher use. Conclusions Telehealth has become a stable feature of primary care mental health services. To improve equity in service delivery, continued policy support for telehealth is required.

Author Summary Why was this study done? There was a rapid expansion of telehealth services in Australia in March 2020, aimed at maintaining access to healthcare services during COVID-19. Mental health services delivered by general practitioners (GPs) were strongly affected. While telehealth use rose sharply during the pandemic, it remains unclear whether this shift was temporary or became a lasting feature of care. Understanding whether telehealth use was sustained and whether it differed by region, age, or gender is important for equitable digital health policy. What did the researchers do and find? We analysed national Medicare Benefits Schedule (MBS) administrative datasets covering all GP mental health consultations in Australia from 2020 to 2025. Telehealth use was reported to be 0.5% of total services offered in early 2020, rising to nearly 30% during the initial pandemic period, then declining and stabilising at about 13% by 2025. There was sustained telehealth use with variations across populations, States/Territories with higher uptake in Victoria and South Australia, among young children and middle-aged adults, and among females. What do these findings mean? Telehealth has become a stable component of primary mental health care in Australia. However, differences across regions and demographic groups suggest unequal access and system readiness. Policies that strengthen digital infrastructure and support equitable access are needed to ensure telehealth improves care for all populations.

Abstract Introduction: Telehealth expansion in 2020 in Australia led to an increase in its use in mental health services. This study was conducted to examine the evolution, magnitude, and persistence of telehealth services for mental health from 2020 Q1 to 2025 Q3. Methods. Analysing the national Medicare Benefits Schedule (MBS) administrative data, disaggregated by state/territory, gender, and age group, the study compared telehealth services with in-person consultations across 23 quarters. We used statistical methods, including a growth curve binomial model and a hierarchical Negative Binomial Generalized Estimating Equation (NB-GEE), to capture the effect of COVID-19 telehealth expansion and its sustainability, as well as disparities among different groups. Results From a baseline of 0.5% of services in 2020 Q1, telehealth use peaked at 29.5% in 2020 Q2 and stabilised at about 13% by 2025 Q3. The ITS augmented models captured the spike in telehealth and its subsequent decline. Disparities in use persisted, with higher use in Victoria and South Australia. Young children (0-4 years), middle-aged adults, and females also had comparably higher use. Conclusions Telehealth has become a stable feature of primary care mental health services. To improve equity in service delivery, continued policy support for telehealth is required.

Author Summary Why was this study done? There was a rapid expansion of telehealth services in Australia in March 2020, aimed at maintaining access to healthcare services during COVID-19. Mental health services delivered by general practitioners (GPs) were strongly affected. While telehealth use rose sharply during the pandemic, it remains unclear whether this shift was temporary or became a lasting feature of care. Understanding whether telehealth use was sustained and whether it differed by region, age, or gender is important for equitable digital health policy. What did the researchers do and find? We analysed national Medicare Benefits Schedule (MBS) administrative datasets covering all GP mental health consultations in Australia from 2020 to 2025. Telehealth use was reported to be 0.5% of total services offered in early 2020, rising to nearly 30% during the initial pandemic period, then declining and stabilising at about 13% by 2025. There was sustained telehealth use with variations across populations, States/Territories with higher uptake in Victoria and South Australia, among young children and middle-aged adults, and among females. What do these findings mean? Telehealth has become a stable component of primary mental health care in Australia. However, differences across regions and demographic groups suggest unequal access and system readiness. Policies that strengthen digital infrastructure and support equitable access are needed to ensure telehealth improves care for all populations.

This preprint represents an early version of this research. Following further analysis, data updates, methodological refinement, and reference verification, the study was substantially revised and expanded. A revised manuscript with updated analyses, corrected references, and revised findings was subsequently prepared and submitted elsewhere. Readers should interpret this preprint as an earlier draft rather than the final version of the research.

1. Schulz T, Long K, Kanhutu K, Bayrak I, Johnson D, Fazio T. Telehealth during the coronavirus disease 2019 pandemic: rapid expansion of telehealth outpatient use during a pandemic is possible if the programme is previously established. J Telemed Telecare. 2022;28(6):445 51. doi: https://doi.org/10.1177/1357633X20942045
2. Nalkar S, Chandak A. Telemedicine for specific populations: evaluating effectiveness and barriers in enhancing healthcare access and outcomes. J Public Health (Berl). 2025. doi: https://doi.org/10.1007/s10389-025-02625-8
3. Nicholas J, Bell IH, Thompson A, Valentine L, Simsir P, Sheppard H, et al. Implementation lessons from the transition to telehealth during COVID 19: a survey of clinicians and young peole from youth mental health services. Psychiatry Res. 2021;299:113848. doi: https://doi.org/10.1016/j.psychres.2021.113848

Appendices Appendix 1: MBS Items for Psychological Therapy Services and Focused Psychological Strategies (Mental Health and Allied Mental Health) MBS Item Description MBS Item Description In-person services Equivalent telehealth services Clinical Psychologist 80010 Professional attendance for psychological assessment and therapy lasting at least 50 minutes, provided by a clinical psychologist. The patient is referred by a medical practitioner, and the service is provided in consulting rooms. 80011 Like 80010 but provided via video conference. The patient must be in a telehealth eligible area, at least 15 kilometres from the psychologist. Psychologist 80110 Professional attendance for focussed psychological strategies services for an assessed mental disorder by a psychologist, lasting more than 50 minutes. 80111 Like 80110 but provided via video conference for patients located in telehealth eligible areas and lasts more than 50 minutes.

Appendix 2A: Trends of Costs for Clinical Psychology Services (Video), by State/Territory

Appendix 2B: Trends of Costs for Clinical Psychology Services (Video), by Age Group.

Appendix 2C: Trends of Costs for Clinical Psychology Services (Video), by Sex

Appendix 2D: Trends of Costs for Psychology Services (Video), by Age Group

       Appendix 2E: Trends of Costs for Psychology Services (Video), by State/Territory

Appendix 3: Model Summary for Vector Autoregressive Moving Average with exogenous inputs (VARMAX) model Metric Value Dependent Variables ['80011', '80111'] Observations 4608 Model VARMAX (5) + intercept Log Likelihood -78047.798 AIC 156205.596 BIC 156559.551 HQIC 156330.166

Appendix 4: Telehealth Policy Changes Relating to Mental Health and Allied Mental Health in Australia, 2017.

Appendix 5: Telehealth Policy Changes Relating to Mental Health and Allied Mental Health in Australia, 2020. Appendix 6: MBS Items for Psychological Therapy Services and Focused Psychological Strategies (Allied Mental Health), March 2020

In-person Videoconference item Telephone item (when videoconferencing facilities are not available) Duration Clinical Psychologists 80010 91167 91182 Attendance lasting at least 50 minutes Psychologists 80110 91170 91184 Attendance lasting at least 50 minutes Occupational Therapists 80135 91173 91186 Attendance lasting at least 50 minutes Social workers 80160 91176 91188 Attendance lasting at least 50 minutes

Results Cost comparison Pre-Intervention Period (November 2017 – February 2020) Before the intervention, in-person clinical psychology services showed a mean benefit paid of AU$98,199.5 (SD = 12,275) per 100,000 and a median (Q1, Q3) of AU$100,899 (93,024.5, 106,368.8) per 100,000. Psychology in-person services had a mean (SD) of AU$79,385.1 (9,670.6) and a median (Q1, Q3) of AU$81,710 (74,573.8, 86,065.8). Video consultations for MMM 4-7 had clinical psychology services mean (SD) costs of AU$169.5 (112) and psychology a mean (SD) of AU$117.9 (63.6), with respective medians of AU$128.5 and AU$112.5 (Table 1). Post-Intervention Period (March 2020 – June 2022) Post-intervention, in-person clinical psychology services showed a reduced mean benefit paid of AU$71,325.2 (SD = 14,804.7) and a median of AU$73,512.5 (Q1 = 61,343.8, Q3 = 78,740). Psychology in-person services had a mean of AU$62,901 (SD = 11,849.6) and a median of AU$65,226.5 (Q1 = 57,365.8, Q3 = 68,939). Video services for MMM 4-7 exhibited increased benefits for clinical psychology (mean = AU$1,557.6, SD = 371.8) and psychology (mean = AU$1,313.5, SD = 347.5), with higher respective medians (Table 1).

Table 1: Results of descriptive analysis of benefits paid per 100,000 for mental and allied healthcare services in Australia, Nov 2017 to Jun 2022.

Statistic Pre-Intervention (AU$) Post-Intervention (AU$) T-Statistic P value Clinical Psychology <br /> In-Person Services <br /> Mean (SD) 98199.5 (12275) 71325.2 (14804.7) 7.39 <0.001 Median (Q1, Q3) 100899 (93024.5, 106368.8) 73512.5 (61343.8, 78740) <br /> Video Services for MMM 4-7 <br /> Mean (SD) 169.5 (112) 1557.6 (371.8) -18.92 <0.001 Median (Q1, Q3) 128.5 (78.2, 272.8) 1470.5 (1334, 1821) <br /> Video Services for General Population (no geographical restrictions) <br /> Mean (SD) 25,066.8 (8,590.6) <br /> Median (Q1, Q3) 26,178 (18,060.8, 30,616.5) <br /> Phone Services for General Population (no geographical restrictions) <br /> Mean (SD) 8,212.4 (3,449.7) <br /> Median (Q1, Q3) 7,314.5 (5,923.2, 9,264.5) <br /> Psychology <br /> In-Person Services <br /> Mean (SD) 79385.1 (9670.6) 62901(11849.6) 5.7 <0.001 Median (Q1, Q3) 81710 (74573.8, 86065.8) 65226.5 (57365.8, 68939) <br /> Video Services for MMM 4-7 only <br /> Mean (SD) 117.9 (63.6) 1313.5 (347.5) -17.91 <0.001 Median (Q1, Q3) 112.5 (65.8, 158.8) 1256.5 (1105.5, 1557.5) <br /> Video Services for General Population (no geographical restrictions) <br /> Mean (SD) 16,887.4 (5,937.5) <br /> Median (Q1, Q3) 17,802 (12,945.8, 20,287.2) <br /> Phone Services for General Population (no geographical restrictions) <br /> Mean (SD) 7,173.6 (2,915.2) <br /> Median (Q1, Q3) 6,381.5 (5,326.5, 8,582.2)

Telehealth utilisation impact The utilisation of services across the three periods (March 2020 - February 2021, March 2021 - February 2022, and March 2022 - February 2023) varied significantly (Table 2). For Clinical Psychology, in-person services showed stable Utilisation across all periods, with no significant differences (F = 0.08, p = 0.927). Video services for MMM 4-7 displayed a significant increase in usage over time (F = 5.27, p = 0.010), while video services for the general population exhibited consistent Utilisation without significant changes (F = 0.35, p = 0.705). Phone services for the general population decreased significantly across the study period (F = 5.20, p = 0.011). In Psychology, in-person services demonstrated steady usage across all periods with no significant differences (F = 0.13, p = 0.879). Video services for MMM 4-7 experienced a significant rise in Utilisation (F = 8.49, p = 0.001), whereas video services for the general population remained stable (F = 0.40, p = 0.673) (Table 2, Appendices 2A-E). Table 2: Service Utilisation Trends and related costs for Clinical Psychology and Psychology Services, by Consultation Mode (March 2020– February 2023)

Service Period Mean (SD) Median (Q1, Q3) F-Statistic P-Value Mean (SD) Median (Q1, Q3) F-Statistic P-Value No. of Services Offered Per 100,000 Benefits Paid (AUD) Per 100,000 <br /> Clinical Psychology <br /> In-Person Services March 2020 - February 2021 541 (110) 566 (453, 589) 0.08 0.927 <br /> March 2021 - February 2022 525 (141) 504 (455, 640) <br /> March 2022 - February 2023 527 (69) 529 (493, 567) <br /> Video Services for MMM 4-7 March 2020 - February 2021 10 (2) 10 (10, 11) 5.27 0.01 1304.0 (276.9) 1333.5 (1224.8, 1440.8) 9.27 0.001 March 2021 - February 2022 12 (2) 12 (11, 13) 1647.8 (295.5) 1603.0 (1395.5, 1824.3) <br /> March 2022 - February 2023 13 (2) 13 (11, 14) 1787.1 (275.7) 1777.0 (1619.8, 1954.3) <br /> Psychology <br /> In-Person Services March 2020 - February 2021 665 (119) 699 (588, 721) 0.13 0.879 <br /> March 2021 - February 2022 689 (164) 677 (601, 817) <br /> March 2022 - February 2023 687 (84) 701 (650, 751) <br /> Video Services for MMM 4-7 March 2020 - February 2021 12 (3) 12 (11, 13) 8.49 0.001 1064.8 (258.8) 1084.5 (998.5, 1230.5) 11.97 <0.001 March 2021 - February 2022 15 (2) 15 (13, 16) 1411.9 (271.4) 1355.5 (1175.3, 1559.3) <br /> March 2022 - February 2023 16 (3) 16 (15, 18) 1546.5 (212.4) 1525.5 (1462.3, 1666.0)

Cost trends Interrupted time series analysis (ITSA) Interrupted time series analysis (ITSA) revealed distinct trends in healthcare costs pre- and post-intervention. An immediate reduction in total costs was observed following the intervention in March 2020 (Table 3). The Interrupted Time Series analysis for Clinical Psychology revealed an intercept of -2.83 (95% CI [-156.59, 150.92], p = 0.971), a pre-intervention trend of 12.76 per month (95% CI [2.99, 22.54], p = 0.011), and a post-intervention trend change of 16.08 per month (95% CI [2.26, 29.9], p = 0.023). For Psychology, the intercept was 16.62 (95% CI [-122.22, 155.45], p = 0.811), and trend change 20.48 (95% CI [8, 32.96], p = 0.002). The intervention increased costs for video-based services: Clinical Psychology by +29,874 (actual: 45,337; counterfactual: 15,463) and Psychology by +22,834 (actual: 38,297; counterfactual: 15,463), reflecting expanded Utilisation post-intervention (Figure 1). Residual diagnostics confirmed the model's validity, with no significant patterns suggesting autocorrelation. Seasonal adjustments accounted for cyclical variations, ensuring observed changes were attributed to the intervention.

Figure 1: ITS Analysis with Counterfactuals for Clinical Psychology and Psychology Video Services Table 3: Interrupted Time Series Model Results for Clinical Psychology and Psychology Video Services (Nov 2017-Jun 2022) Variable Coefficient (β) Standard Error t-value p-value 95% CI Clinical Psychology <br /> Intercept (Baseline Level) -2.83 76.62 -0.04 0.971 [-156.59, 150.92] Time (Trend Pre-Intervention) 12.76 4.87 2.62 0.011 [2.99, 22.54] Intervention (Immediate Effect) 363.33 219.71 1.65 0.104 [-77.56, 804.22] Time x Post-intervention 16.08 6.89 2.33 0.023 [2.26, 29.90] Psychology <br /> Intercept (Baseline Level) 16.62 69.19 0.24 0.811 [-122.22, 155.45] Time (Trend Pre-Intervention) 7.5 4.4 1.71 0.094 [-1.33, 16.32] Intervention (Immediate Effect) 135.93 198.39 0.69 0.496 [-262.17, 534.02] Time x Post-intervention 20.48 6.22 3.29 0.002 [8.00, 32.96]

Drivers of costs Data analysis showed that sex, age categories, state, and COVID-19 waves had varying costs per 100,000 for clinical psychology services. Significant differences were observed across groups: Sex (U=3688844, p<0.001), Age Category (H=2418.09, p<0.001), and other variables (Table 4). Similarly, for psychology services, there also significant differences between variables. Females had higher medians (1754.5) than males (673), with a significant U=3688844 (p<0.001). Furthermore, age categories and states exhibited significant variances (H=2241.71, H=328.92; both p<0.001) (Table 5).

Table 4: Potential sociodemographic and temporal predictors of costs of video consultations for clinical psychology services in MMM 4-7 (March 2020 – February 2023) Variable Median (Q1, Q3) Test Statistic P-Value Sex <br /> Female 1754.5 (382.8, 3256.5) 3688844 <0.001

Male 673 (0, 1232.3) <br /> Age Category <br /> 15-24 1673 (850.3, 3174.3) 2418.09

<0.001

25-34 2226.5 (1246.5, 4064.3) <br /> 35-44 1992 (1162.5, 3499.8) <br /> 45-54 1615 (890.8, 2537.5) <br /> 55-64 1072.5 (652.5, 1920.8) <br /> 65-74 530 (174, 895.3) <br /> 75-84 118 (0, 371)

=85 0 (0, 0) <br /> State <br /> ACT 737 (0, 2481.3) 81.65 <0.001 NSW 973 (394.5, 2235.5) <br /> NT 0 (0, 2677.5) <br /> QLD 969 (367, 2003.3) <br /> SA 1028 (218.8, 2033) <br /> TAS 743.5 (0.0, 1543.8) <br /> VIC 1202.5 (393.8, 2143.3) <br /> WA 998.5 (414.5, 2111.3) <br /> COVID-19 Wave <br /> First Wave 771.5 (162.5, 1661) 96.26 <0.001 Low Transmission 822.5 (104.5, 1730.3) <br /> Omicron Wave 952.5 (231.3, 2403.5) <br /> Second Wave 721.5 (0, 1632) <br /> Subsequent Waves 1182 (259.8, 2737) <br /> Third Wave 1021.5 (202.8, 2310.3)

Table 5: Potential sociodemographic and temporal predictors of costs of video consultations for psychology services in MMM 4-7 (March 2020 – February 2023) Variable Median (Q1, Q3) Test statistic P value Sex <br /> Female 1283.5 (225, 2594.3) 3688844 <0.001 Male 463 (0, 1021.3) <br /> Age Category (years) 1331.5 (690, 2780.5) <br /> 15-24 1795.5 (946.3, 3161.8) 2241.71 <0.001 25-34 1432 (890.8, 2553.3) <br /> 35-44 1182.5 (587.3, 1874.5) <br /> 45-54 818.5 (415.3, 1479.5) <br /> 55-64 395.5 (100.8, 904) <br /> 65-74 0 (0, 236.3) <br /> 75-84 0 (0, 0)

=85 <br /> State <br /> ACT 116 (0, 1023.8) 328.92 <0.001 NSW 964.5 (307.5, 1884.5) <br /> NT 507.5 (0, 1737.5) <br /> QLD 889 (288.8, 1658.3) <br /> SA 588 (158.8, 1272.8) <br /> TAS 1096 (0, 2653.3) <br /> VIC 1154.5 (450.5, 2317.8) <br /> WA 431.5 (115.3, 1015.5) <br /> COVID-19 Wave <br /> First Wave 672.5 (179.5, 1412.5) 76.65 <0.001 Low Transmission 538 (0, 1325.5) <br /> Omicron Wave 706.5 (44.3, 1555.8) <br /> Second Wave 580.5 (0, 1308) <br /> Subsequent Waves 972.5 (149.8, 2131) <br /> Third Wave 732.5 (0, 1787.3)

Figure 3: Median costs of video services per 100,000 individuals, by COVID-19 wave for each state (Psychology and Clinical psychology Services, March 2020 – February 2023) Concerning COVID-19 Waves and costs per 100,000 for services, states and territories like Australia Capital Territory (ACT), Tasmania (TAS), and Victoria (VIC) exhibited the highest costs across both services, while the Northern Territory (NT) and Queensland (QLD) show moderate costs. The "Subsequent Omicron Waves" account for the largest share of costs across states, highlighting sustained telehealth demand in later pandemic waves. In contrast, the Delta Variant Wave and First Wave reflected lower initial telehealth adoption (Figure 3). Clinical psychology services consistently incurred higher per capita costs than psychology. TAS shows notably high costs for psychology. Western Australia (WA) and South Australia (SA) reported relatively lower costs (Figure 3). Vector Autoregressive Moving Average with exogenous inputs (VARMAX) model A Vector Autoregressive Moving Average model with exogenous inputs (VARMAX) employed to analyse multivariate time-series data (4608 observations) and explore the temporal and demographic effects on the costs of video consultations for clinical psychology (80011) and psychology (80111) services showed notable findings. The VARMAX (5) model with an intercept term captured dynamic interactions over time. For clinical psychology services, the baseline cost was $1,398.14 (p < 0.001). Significant lagged effects indicated cost persistence. Costs declined notably for individuals aged 65–74 (-$1,068.83, p < 0.001) and were lower for males (-$968.12). However, an interaction effect (65–74 × Male) showed increased costs (+$777.64, p < 0.001) (Table 6). For psychology services, the baseline cost was $1,597.82 (p < 0.001). Lagged predictors revealed persistent but diminishing effects over time. Costs decreased significantly for the ≥85 age group (-$1,564.14, p < 0.001) (Table 6). Model fit metrics (AIC: 156205.596, BIC: 156559.551, HQIC: 156330.166) and a log-likelihood of -78047.798 demonstrated strong predictive capability with balanced complexity (Appendix 3). Diagnostics showed no residual autocorrelation (Ljung-Box, p > 0.05), but heteroskedasticity was present (p < 0.001). Overall, the VARMAX (5) model effectively captured cost dynamics and demographic influences. Table 6: VARMAX Model Results for Costs of Video Consultations for Clinical Psychology Services and Psychology Services (March 2020-February 2023)

Variable Clinical Psychology Services Psychology Services Coefficient (95% CI) Coefficient (95% CI) Intercept 1398.14 (1254.65, 1541.64) 1597.82 (1494.66, 1700.98) Age Category (Ref: 15-24 years) <br /> 25-34 397.43 (272.45, 522.41) 243.13 (140.57, 345.7) 35-44 473.58 (348.2, 598.95) -377.89 (-483.03, -272.75) 45-54 -323.71 (-494.52, -152.9) -577.26 (-695.71, -458.8) 55-64 -623.48 (-813.69, -433.27) -713.7 (-831.5, -595.91) 65-74 -1068.83 (-1317.21, -820.45) -1077.52 (-1221.68, -933.35) 75-84 -1305.78 (-1858.45, -753.12) -1453.83 (-1740.37, -1167.29)

=85 -1375.8 (-2540.37, -211.24) -1564.14 (-2086.93, -1041.36) Sex (Ref: Female) <br /> Male -968.12 (-1192.69, -743.55) -1082.34 (-1244.72, -919.96) Interactions <br /> 25-34 x Male -176.03 (-465.83, 113.77) -195.97 (-416.42, 24.47) 35-44 x Male -448.88 (-774.34, -123.43) 549.47 (331.87, 767.08) 45-54 x Male 328.96 (-29.7, 687.62) 608.35 (388.97, 827.72) 55-64 x Male 472.1 (89.46, 854.75) 583 (340.27, 825.74) 65-74 x Male 777.64 (373.63, 1181.66) 814.97 (537.13, 1092.82) 75-84 x Male 1019.31 (364.45, 1674.18) 954.4 (425.77, 1483.04) =85 x Male 962.77 (-887.58, 2813.13) 1060.77 (-39.19, 2160.72)

Generalised Linear Mixed Model (GLMM) For clinical psychology, the final Generalised Linear Mixed Model (GLMM) identified significant state and COVID-19 wave effects on square root-transformed healthcare costs. The baseline cost for the reference state and baseline wave was estimated at β₀ = 34.017 (p < 0.001). State effects varied substantially, with higher costs in NSW (β = 4.472), VIC (β = 4.465), and WA (β = 4.768) compared to the reference state (p < 0.001 for all). In contrast, NT (β = -1.804, p = 0.006) and TAS (β = -2.178, p = 0.004) exhibited lower costs, while QLD and SA showed no significant differences. COVID-19 wave effects highlighted increased costs during the Omicron Wave (β = 4.222, p = 0.006), Subsequent Waves (β = 7.385, p < 0.001), and the Third Wave (β = 4.083, p = 0.003), with no significant effects during Low Transmission or the Second Wave (Table 7). For psychology services, significant COVID-19 wave effects included decreased costs during Low Transmission (β = -2.838, p = 0.019) and the Second Wave (β = -2.535, p = 0.041) but increased costs in Subsequent Waves (β = 3.967, p < 0.001). Effects during the Omicron and Third Waves were not statistically significant (Table 8). Random Effects The models demonstrated significant variability across states and sexes for clinical psychology, with state variance (σ² = 450.025) and sex variance (σ² = 250.233). Covariance between state and sex was negligible. For psychology, state variability was moderate, with random intercept variance at σ² = 34.873. Residual Analysis Residual diagnostics confirmed well-fitting models for both clinical psychology and psychology services. Residuals were symmetrically distributed around zero, with no discernible patterns against fitted values, indicating homoscedasticity. QQ-plots showed alignment with normality, with only minor tail deviations, confirming robust model performance. The results of multiple analyses in this study show that the COVID-19 pandemic's telehealth expansion significantly increased healthcare costs, particularly for clinical psychology and psychology services. Video consultations surged in usage and cost, rising from AU$169.5 to AU$1,557.6 for clinical psychology and from AU$117.9 to AU$1,313.5 for psychology post-intervention. Despite lower operational costs, telehealth's broader accessibility drove total expenditures higher. Females and younger adults (25–34) incurred higher median costs, while costs declined with age. Regional disparities showed NSW, VIC, and WA with higher costs, while NT and TAS had lower costs. Temporally, costs peaked during the Omicron and Subsequent Waves, highlighting telehealth’s role in reshaping access and healthcare spending patterns.

Table 7: Results of Generalised Linear Multilevel Effects Model- Video Services for Clinical Psychology Fixed Effect Coefficient (β) Standard Error p-value Intercept 34.017 1.105 <0.001 State <br /> ACT (Ref) <br /> NSW 4.472 0.451 <0.001 VIC 4.465 0.456 <0.001 WA 4.768 0.481 <0.001 NT -1.804 0.659 0.006 TAS -2.178 0.645 0.004 COVID 19 Wave <br /> First wave (Ref) <br /> Omicron Wave 4.222 1.312 0.006 Subsequent Waves 7.385 1.094 <0.001 Third Wave 4.083 1.153 0.003 Low Transmission -0.389 0.958 0.582 Second Wave -1.879 1.331 0.156

Random Effect Variance (σ²) State Variance 450.025 Sex Variance 250.233

Table 8: Results of Generalised Linear Multilevel Effects Model- Video Services for Psychology Effect Coefficient (β) Standard Error P-value Intercept 26.196 3.102 <0.001 Low Transmission -2.838 0.921 0.019 Second Wave -2.535 1.052 0.041 Subsequent Waves 3.967 0.854 <0.001 Omicron Wave -0.374 1.402 0.794 Third Wave 0.472 1.239 0.713 Random Effect Variance (σ²) State Variance 34.873

Discussion The telehealth policy changes introduced in Australia, particularly during the COVID-19 pandemic, were effective in maintaining healthcare access while transforming cost and utilisation patterns. Video consultations experienced high cost increases post-intervention, rising from AU$169.5 to AU$1,557.6 per 100,000 for clinical psychology and from AU$117.9 to AU$1,313.5 per 100,000 for psychology. This reflects the success of telehealth initiatives in addressing the need for alternative healthcare delivery during a period of reduced in-person accessibility. Concurrently, in-person services showed reduced costs, with clinical psychology and psychology services seeing declines of AU$26,874.3 and AU$16,484.1 per 100,000, respectively, suggesting a patient and provider shift toward telehealth models. These findings align with policy goals aimed at improving healthcare access during the pandemic. However, the increased reliance on telehealth has introduced new cost dynamics, emphasizing the need for ongoing policy evaluation and adjustment (34). One of the primary objectives of telehealth expansion was to improve access for underserved populations, particularly in rural and remote areas. The data showed a significant rise in video consultation utilisation in MMM 4-7 regions, with mean benefits for clinical psychology increasing from AU$169.5 to AU$1,557.6 per 100,000, highlighting telehealth's potential in reducing geographic barriers. However, the persistent disparities between states (e.g., higher costs in NSW, VIC, and WA compared to NT and TAS) suggest that equitable access remains incomplete, potentially due to varying levels of digital infrastructure and healthcare resource availability across regions. Demographic factors also played a role in equity. Younger adults (25–34 years) and females incurred higher costs, likely reflecting greater adoption and acceptance of telehealth among these groups. In contrast, older adults, particularly those aged 75+, demonstrated lower utilisation, which aligns with existing barriers such as digital literacy and access to technology (29). Addressing these gaps will be crucial to achieving broader equity. While telehealth reduces operational costs, such as travel and physical infrastructure, the study revealed unintended cost increases driven by higher demand. Post-intervention, total expenditures on telehealth services exceeded pre-intervention levels due to the broader accessibility and convenience of telehealth. This aligns with international findings, where expanded telehealth offerings led to increased utilisation, even for patients who might not have sought in-person care previously (30, 31). The increased demand raises concerns about the sustainability of telehealth expansion. Without appropriate measures, such as optimized service models or stricter triaging protocols, cost growth could outpace the savings derived from operational efficiencies. These findings underscore the importance of balancing access with cost control in telehealth policy design. The findings align with existing literature on telehealth’s potential to improve equity and access, particularly in rural areas. Previous studies have demonstrated that telehealth reduces barriers to care in underserved populations, including rural communities in Australia and globally. However, consistent with this study, the literature highlights challenges such as uneven adoption across demographics and regions, often tied to digital infrastructure disparities (31). The cost dynamics observed in this study are also reflected in the broader telehealth literature. While some studies note cost reductions due to efficiency gains (32), others, similar to these findings, report increases linked to higher service utilisation post-expansion (2, 12). This duality highlights telehealth’s capacity for both cost savings and demand-driven expenditure growth. Lastly, the sociodemographic patterns observed here, with greater adoption among younger and female populations, align with global trends, as these groups are generally more tech-savvy and proactive in seeking healthcare services (32). Conclusion This study showed that the introduction of telehealth improved access to mental health services in MMM 4–7 regions. However, it led to increased costs due to higher service utilisation. While video consultation costs for psychology services rose, in-person expenses declined, reflecting a shift in care delivery. In addition, cost variations were influenced by demographics, geography, and COVID-19 waves, with younger adults, females, and certain states (NSW, VIC, WA) seeing higher expenditures. It is important to consider policy adjustments to balance financial sustainability with accessibility. Future research should explore how telehealth models can be refined to enhance cost efficiency while maintaining benefits.

References 1. Namusisi N, University III KI. The impact of telehealth on healthcare delivery. Eurasian Exp J Med Med Sci. 2024;5:37 40. Available from: https://publications.kiu.ac.ug/view/2680/the-impact-of-telehealth-on-healthcare-delivery 2. Anawade PA, Sharma D, Gahane S. A comprehensive review on exploring the impact of telemedicine on healthcare accessibility. Cureus. 2024 Mar 12;16(3):e55996. doi: https://doi.org/10.7759/cureus.55996 3. Australian Government Department of Health. Medicare Benefits Schedule Review Taskforce. Taskforce recommendations.Telehealth. 2020. Available from: https://www.health.gov.au/sites/default/files/documents/2020/12/taskforce-recommendations-telehealth.pdf 4. Thomas EE, Haydon HM, Mehrotra A, Caffery LJ, Snoswell CL, Smith AC. Building on the momentum: sustaining telehealth beyond COVID 19. J Telemed Telecare. 2022;28(4):301 8. doi: https://doi.org/10.1177/1357633X20960638 5. National Rural Health Alliance. Fact Sheet – March 2024. Digital Health and Con-nectivity in Rural Australia. NRHA;2024. Available from: https://www.ruralhealth.org.au/wp-content/uploads/2024/05/nrha-digital-health-connectivity-factsheet-apr-24.pdf 6. Hall Dykgraaf S, Desborough J, de Toca L, Davis S, Roberts L, Munindradasa A, et al. "A decade's worth of work in a matter of days": The journey to telehealth for the whole population in Australia. Int J Med Inform. 2021 Jul;151:104483. doi: 10.1016/j.ijmedinf.2021.104483. 7. Australian Government Department of Health. Modified Monash Model classification. Available from: https://www.health.gov.au/topics/rural-health-workforce/classifications/mmm 8. Fairchild RM, Ferng Kuo SF, Laws S, Rahmouni H, Hardesty D. Telehealth decreases rural emergency department wait times for behavioral health patients in a group of critical access hospitals. Telemed J E Health. 2019;25(12):1154 64. doi: https://doi.org/10.1089/tmj.2018.0227 9. Reed ME, Huang J, Graetz I, et al. Patient characteristics associated with choosing a telemedicine visit vs office visit with the same primary care clinicians. JAMA Netw Open. 2020;3(6):e205873. doi: https://doi.org/10.1001/jamanetworkopen.2020.5873 10. Butzner M, Cuffee Y. Telehealth interventions and outcomes across rural communities in the United States: narrative review. J Med Internet Res. 2021;23(8):e29575. doi: https://doi.org/10.2196/29575 11. Wang S, von Huben A, Sivaprakash PP, Saurman E, Norris S, Wilson A. Addressing health service equity through telehealth: a systematic review of reviews. Digit Health. 2025;11:20552076251326233. doi: https://doi.org/10.1177/20552076251326233 12. Tsou C, Robinson S, Boyd J, Jamieson A, Blakeman R, Yeung J, et al. Effectiveness of telehealth in rural and remote emergency departments: systematic review. J Med Internet Res. 2021;23(11):e30632. doi: https://doi.org/10.2196/30632 13. Australian Government Department of Health. Better Access Fact Sheet for Allied Health Professionals. Better Access to Psychiatrists, Psychologists and General Practitioners through the MBS (Better Access) initiative. Canberra. Available from: https://nqphn.com.au/wp-content/uploads/2023/07/better-access-fact-sheet-professionals-better-access-fact-sheet-professionals.pdf 14. Australian Government Department of Health. Continuing MBS telehealth services – mental health services.2022. Available from: https://www.mbsonline.gov.au/internet/mbsonline/publishing.nsf/Content/81F4D6E6C09A3762CA25887200043384/$File/Factsheet-Continuing-telehealth-Mental-Health.25.01.22.pdf 15. Australian Government Department of Health. Medicare Benefits Schedule Book: Operating from 01 November 2017. Canberra : Department of Health; 2017. Available from: https://www.mbsonline.gov.au/internet/mbsonline/publishing.nsf/Content/40901A5DCA1CD593CA258614007F45C1/$File/201711-MBS-16-Nov.pdf 16. Department of Health and Aged Care. Lessons from the fourth Omicron COVID-19 wave. Canberra. Available from: https://www.health.gov.au/sites/default/files/2023-03/lessons-from-the-fourth-omicron-covid-19-wave-chief-medical-officer-professor-paul-kelly.pdf<br /> 17. Reserve Bank of Australia. The COVID 19 Pandemic: 2020 to 2021. 2021. Available from: https://www.rba.gov.au/education/resources/explainers/pdf/the-covid-19-pandemic-2020-to-2021.pdf 18. Lewis Beck MS, Bryman A, Liao TF, editors. Interrupted time series design. In: The SAGE Encyclopedia of Social Science Research Methods. 2004. p. 516. doi: https://doi.org/10.4135/9781412950589.n446 19. Jiang H, Feng X, Lange S, et al. Estimating effects of health policy interventions using interrupted time series analyses: a simulation study. BMC Med Res Methodol. 2022;22:235. doi: https://doi.org/10.1186/s12874-022-01716-4 20. Corrigan PW, Watson AC. Understanding the impact of stigma on people with mental illness. World Psychiatry. 2002;1(1):16 20. 21. Mojtabai R. Mental illness stigma and willingness to seek mental health care in the European Union. Soc Psychiatry Psychiatr Epidemiol. 2010;45:705 12. doi: https://doi.org/10.1007/s00127-009-0109-2 22. Weisz JR, Sandler IN, Durlak JA, Anton BS. Promoting and protecting youth mental health through evidence based prevention and treatment. Am Psychol. 2005;60(6):628 48. doi: https://doi.org/10.1037/0003-066X.60.6.628 23. Düker M C, Matteson DS, Tsay RS, Wilms I. Vector autoregressive moving average models: a review. WIREs Comput Stat. 2025;17:e70009. doi: https://doi.org/10.1002/wics.70009 24. Korstanje J. The VARMAX model. In: Advanced Forecasting with Python. Berkeley (CA): Apress; 2021. 25. Islam MA, Biswas SC. Generalized linear mixed models. In: Generalized Linear Models and Extensions. Singapore: Springer; 2025. doi: https://doi.org/10.1007/978-981-96-4726-2_7 26. Kwak SK, Kim JH. Statistical data preparation: management of missing values and outliers. Korean J Anesthesiol. 2017;70(4):407 11. doi: https://doi.org/10.4097/kjae.2017.70.4.407 27. Statsmodels. Python statsmodels library [Internet]. 2023 [cited 2025 Jan 1]. Available from: https://www.statsmodels.org 28. McKinney W, Perktold J, Seabold S. Time Series Analysis in Python with stats-models. InSciPy 2011 Jul (pp. 107-113). 29. Lohr KD. Digital literacy and access: equity from a global and local perspective. New Dir Adult Contin Educ. 2025;39 43. doi: https://doi.org/10.1002/ace.20559

Methods Study design and setting This study employed a retrospective observational design to examine the impact of telehealth expansion on the costs of mental health and allied health services in Australia. Data were collected from Medicare Benefits Schedule (MBS) claims between November 2017 and February 2023, focusing on clinical psychology and psychology services. Data sources The study used monthly consultation data obtained from Services Australia, selecting MBS items related to clinical psychology and psychology services. The study collected data for costs (Medicare benefits paid) for services lasting at least 50 minutes in rural, remote, and very remote areas (MMM 4–7 regions) (3,13-15). The list of datasets included the costs per 100,000 for the following: i) In-Person clinical psychology (80010), and psychology (80110) services; and ii) Video consultations for clinical psychology (80011) and psychology (80111) services restricted to eligible areas in rural, remote, and very remote areas (Appendix 1). Datasets for 80010, 80110, 80011, and 80111 were obtained based on the dates of introduction of the MBS items and the objectives of the study. The relevant data on services per 100,000 for the respective datasets were also obtained for the identified MBS items and period. Variables Dependent variable: Costs per 100,000 population were the dependent variable in this study. It refers to the benefits paid by Medicare on account of the delivery of mental and allied healthcare services for eligible individuals. These costs were aggregated for all consultations and described per capita. Independent variables: Independent variables in this study include: i) Rurality: This refers to the proportion of states described as rural, remote, and very remote, measured by Modified Monash Model (MMM) classification. The proportion of people located in MMM 4-7 in each state/territory was computed using population per postcode obtained from national census data and then applying the postcode-based MMM classification obtained from the Department of Health, Disability and Ageing (7). ii) Service Type: This provides a classification of the type of mental health and allied services offered, described as either psychology or clinical psychology. iii) Service Utilisation: This shows the number of service encounters within specific periods, measured in services per 100,000. iv) Demographic factors: This captures the characteristics of the population or individuals using the services and incurring the costs described in item (i) above. These include Age [in terms of groupings such as Youth (15-24 years), Young adults (25-34 years), Adults (35-64 years), and Aged (≥65 years)]. It also includes Gender, described as male/female, as provided on the website of Services Australia (16). v) States/Territory: This refers to the place (state) of domicile of the client at the time of the service provision. vi) COVID-19 Waves: This study defined COVID-19 waves in Australia as distinct periods marked by varying transmission levels, driven by different viral variants and public health measures. The waves included: First Wave (March–May 2020), involving the original strain and strict lockdowns; Second Wave (June–November 2020), which was Victoria-focused and included quarantine breaches; Low Transmission (Dec 2020–May 2021), signifying the Pre-Delta containment period; Third Wave (June–Oct 2021), including Delta outbreaks and prolonged lockdowns; Low Transmission (Nov 2021), with Post-Delta and pre-Omicron periods; Omicron Wave (Nov 2021–Feb 2022), which was marked by high cases despite vaccination; and Subsequent Waves (March 2022–Feb 2023), involving Omicron subvariant surges (16). Categorically coded, these periods enabled analysis of COVID-19's impact on costs of telehealth for mental and allied healthcare services in Australia.

Analytical Framework To investigate the implications of telehealth expansion on mental health and allied health care costs in Australia, an analytical framework comprising cost comparison, telehealth Utilisation impact analysis, cost trend analysis, and cost drivers’ analysis was conducted. Cost comparison: The study analysed datasets spanning November 2017, when video items were introduced for focused psychological interventions in rural, remote, and very remote areas (MMM 4-7), to June 2022 (15). This ensured an equivalent period: 27 months pre- and post-March 2020, when telehealth expansion occurred in Australia (14). Descriptive analyses of costs [mean (SD) and median (Q1, Q3)] were conducted for pre- and post-telehealth expansion periods, while statistical tests, including paired t-tests, independent t-tests, or ANOVA, were conducted to test the significance of differences. Telehealth utilisation impact analysis: Service utilisation trends for Clinical Psychology and Psychology were analysed across three time periods: March 2020–February 2021, marking the immediate COVID-19 pandemic period; March 2021–February 2022, representing the late COVID-19 pandemic period; and March 2022–February 2023, signifying the pandemic recovery period (3,17). Cost trend analysis: Interrupted Time Series (ITS) Analysis evaluates intervention impacts by analysing changes in time series trends before and after an event (18,19). Effective for causal inference in non-randomized settings, it is widely used in public health and policy research. ITS is therefore suitable for conducting this study to evaluate changes in the costs of mental and allied healthcare services over time following telehealth policy changes. The primary outcome variable was the costs per 100,000 individuals, expressed in three categories: i) In-Person consultation costs; ii) Video consultation costs; and iii) Total costs (sum of In-Person and Video consultations). To address skewness in the cost data, a natural logarithmic transformation was applied. For the ITS model, the intervention variable was modelled as a binary indicator, distinguishing pre- and post-March 2020 periods. Time was modelled as a continuous variable, and interaction terms were introduced to capture changes in the slope of trends following the intervention. We performed ITS analysis using ordinary least squares (OLS) regression. The model included: i) Time, to represent the overall trend before and after the intervention; ii) Intervention, to capture the immediate effect of the policy change in March 2020; and iii) Time-Post-Intervention Interaction, to estimate changes in the slope of trends following the intervention. Cost drivers’ analyses The analyses of cost drivers were conducted through multiple approaches, including Vector Autoregressive Moving Average with exogenous inputs (VARMAX) and Generalised Linear Mixed Model (GLMM). For modelling the drivers of costs, records of children aged 0 to 14 years were excluded. Access to mental health services by adults is often complicated by perceptions, stigma surrounding mental health care, workplace constraints, or financial barriers that may not apply to children (20,21). Including children in this study would introduce developmental and contextual factors that could dilute the focus, as their access to psychological services is predominantly mediated by caregivers and schools, which operate under entirely different dynamics (22). Vector Autoregressive Moving Average with exogenous inputs (VARMAX) model. The Vector Autoregressive Moving Average with Exogenous Inputs (VARMAX) model is a multivariate statistical method used to analyse and forecast time-series data. It extends the VARMA model by incorporating external variables (exogenous inputs) to improve predictive accuracy (23). VARMAX captures dynamic interdependencies between multiple time-series variables through autoregressive (VAR) and moving average (MA) components while allowing external factors to influence outcomes (24). The VARMAX model examined interactions between dependent variables and predictors with up to five lags, incorporating autoregressive terms and exogenous effects. Parameters were estimated using maximum likelihood under the assumption of normally distributed residuals. Model fit was evaluated using AIC, BIC, and HQIC, and confidence intervals were derived via the delta method. Sensitivity analyses tested alternative lag structures and interaction terms, while variance inflation factors (VIFs) checked for multicollinearity. Diagnostic tests, including residual analysis and Ljung-Box tests for autocorrelation, ensured robustness. The covariance structure accounted for near-singular matrices using a robust correction (24).

Generalised Linear Mixed Model (GLMM) A Generalised Linear Mixed Model (GLMM) extends Generalised linear models by incorporating both fixed and random effects, allowing for analysis of data with hierarchical or clustered structures. GLMMs handle non-normal response variables and model relationships using various distributions (e.g., Gaussian, Gamma) and link functions (e.g., log, identity) (25). In this study, GLMMs accounted for variability within states and demographic groups by including random intercepts for sex and state, while fixed effects captured state-level and COVID-19 wave influences on healthcare costs. This approach enabled robust analysis of complex datasets with multiple sources of variability. Data were sourced from a publicly accessible healthcare database encompassing eight Australian states and territories across multiple COVID-19 waves, including the Omicron wave, Third Wave, and other specified periods. The dataset included 4,608 observations of healthcare encounters that met the inclusion criteria of complete data on healthcare costs, state identifiers, COVID-19 wave classification, and demographic details. Observations with missing or implausible values were excluded (26). Healthcare costs, the dependent variable, were represented by items 80011 (clinical psychology) and 80111 (psychology). These were subjected to a square root transformation to stabilize variance and mitigate skewness. Independent variables included state (a categorical variable with eight levels) and COVID-19 wave classifications. Random intercepts were incorporated for sex and state to account for within-group variability (27). To ensure robust analysis, preprocessing involved addressing zero and negative values, applying square root transformations, and filtering out outliers. Bias was minimised through standardised procedures: selection bias was mitigated by including only complete records, measurement bias was addressed through uniform preprocessing, and heterogeneity bias was managed using random effects. The sample size of 4,608 observations was used based on the availability of complete and valid records. A GLMM with a Gamma distribution and a log link function was initially specified. Fixed effects for COVID-19 waves and states and random intercepts for sex and state were included in the model. Residual diagnostics evaluated normality, homoscedasticity, and outliers, leading to the exclusion of extreme residuals (greater than three standard deviations) to improve model robustness. Analyses were conducted using Python's statsmodels library for GLMM fitting and diagnostics (26-28). Ethical Considerations The study was granted ethical approval by a relevant Human Research Ethics Committee. It adhered to key ethical principles.

This study aims to investigate the implications of telehealth expansion on mental health and allied health costs in rural populations. Specifically, the study seeks to answer the following research questions: i) What were the implications of telehealth expansion on costs in rural, remote, and very remote areas? ii) What were the of costs post-telehealth expansion in rural, remote, and very remote areas? By addressing these questions, this study contributes to the growing body of evidence on telehealth's role in reshaping healthcare delivery, with a particular emphasis on equity and cost-effectiveness (11,12).

Telehealth has emerged as a transformative approach to healthcare delivery, particularly in mental health and allied health services. Its significance has grown globally, driven by its potential to overcome barriers such as geographic isolation, transportation limitations, and healthcare provider shortages (1-3). The COVID-19 pandemic further accelerated telehealth adoption, serving as a critical tool to ensure continuity of care during periods of restricted mobility and strained healthcare systems (3). Telehealth’s ability to provide virtual consultations has proven especially beneficial in mitigating access challenges during emergencies, including public health crises like COVID-19. In Australia, telehealth was rapidly expanded through significant policy interventions during the COVID-19 pandemic (4). Key among these was the introduction of Medicare Benefits Schedule (MBS) items for telehealth consultations, which enabled government-subsidized access to virtual healthcare services. These policy changes were designed to address accessibility gaps in rural and remote populations, where geographic and logistical barriers have historically limited access to in-person healthcare services (3-6). The Modified Monash Model (MMM) classification system, which identifies areas based on remoteness, underscored the critical need for telehealth in MMM 4-7 regions, where healthcare disparities are most pronounced (7). Despite the increased adoption of telehealth, there is a limited understanding of its economic implications, particularly regarding the costs incurred for mental health and allied health services. Existing studies have primarily focused on access and satisfaction outcomes, leaving gaps in knowledge about telehealth’s financial impact (8,9). Additionally, the cost dynamics among rural populations, who may benefit the most from telehealth, remain underexplored (10). Given the heterogeneity in service utilisation and demographic factors, understanding the drivers of telehealth costs is essential for evidence-based policy development.

Author Update (June 2026)

Following a subsequent audit of this preprint, some references and supporting citations were identified as inaccurate. The manuscript was later comprehensively revised, references were re-verified against original sources, and corrections were incorporated into a revised version submitted subsequently. Readers are advised to consult the annotations within this preprint regarding affected citations.

References 1. Australian Institute of Health and Welfare. Health system overview [internet]. Canberra: AIHW[updated 2024 Jul 2, cited 2026 May 25]. Available from: https://www.aihw.gov.au/reports/australias-health/health-system-overview. 2. Saxby K, Zhang Y. Bulk-billing rates and out-of-pocket costs for general practitioner services in Australia, 2022, by SA3 region: analysis of Medicare claims data. Med J Aust. 2022. Available from: https://onlinelibrary.wiley.com/doi/pdf/10.5694/mja2.52562 3. Australian Institute of Health and Welfare. Medicare-subsidised GP, allied health and specialist health care across local areas: bulk billing and out-of-pocket costs. Canberra: AIHW; 2025 Mar. Available from: https://www.aihw.gov.au/reports/primary-health-care/medicare-subsidised-gp-allied-health-specialist https://www.aihw.gov.au/getmedia/93a8062a-fab2-45f1-917e-c94c20df5933/medicare-subsidised-gp-allied-health-and-specialist-health-care-across-local-areas.pdf?v=20260326060230&inline=true 4. Evans J. Bulk-billing rates flatline, despite billions tipped into Medicare. ABC News. 2025 Nov 21. Available from: https://www.abc.net.au/news/2025-11-21/bulk-billing-rates-flatline-medicare-billions/106032104 5. Australian Government Department of Health. COVID-19 temporary MBS telehealth services: bulk-billed MBS telehealth services for GPs and other medical practitioners. Canberra: Department of Health; 2020 Sep. Available from: https://www.mbsonline.gov.au/internet/mbsonline/publishing.nsf/Content/0C514FB8C9FBBEC7CA25852E00223AFE/%24File/Factsheet-COVID-19-Bulk-Billed-MBS-Telehealth-Services-GPs-OMP-17.09.2020.pdf 6. Lim ECN. Policy meets practice: financial modelling of bulk-billing and practice incentives reform in Australian general practice. Open J Appl Sci. 2025;15:3131-45. doi: https://doi.org/10.4236/ojapps.2025.1510206 7. Australian Government Department of Health, Disability and Ageing. What we’re doing about mental health. Canberra: Australian Government Department of Health, Disability and Ageing; 2025. Available from: https://www.health.gov.au/topics/mental-health-and-suicide-prevention/what-were-doing-about-mental-health 8. Looi JCL, Allison S, Bastiampillai T, Kisely S. Mapping the regional and remote specialised mental health workforce: commentary on the AIHW data for 2022-2023. Australas Psychiatry. 2025;33(2):252-8. doi: https://doi.org/10.1177/10398562251316365 9. Australian Government Department of Health. COVID-19 temporary MBS telehealth services [Internet]. Canberra: Department of Health; 2022. Available from: https://www.mbsonline.gov.au/internet/mbsonline/publishing.nsf/Content/Factsheet-TempBB 10. Australian Government Department of Health. COVID-19 temporary MBS telehealth services: factsheet for GPs and other medical practitioners (post 1 July 2021 version 5). Canberra: Department of Health; 2021 Jul. Available from: https://www.mbsonline.gov.au/internet/mbsonline/publishing.nsf/Content/0C514FB8C9FBBEC7CA25852E00223AFE/$File/Factsheet-COVID-19-GPsOMP-Post-1July2021V5.pdf 11. Chen C, Gu D. Andersen model. In: Gu D, Dupre ME, editors. Encyclopedia of gerontology and population aging. Cham: Springer; 2021. Available from: https://link.springer.com/rwe/10.1007/978-3-319-69892-2_876-1 12. Andersen RM. Revisiting the behavioral model and access to medical care: does it matter? J Health Soc Behav. 1995;36(1):1-10. Available from: https://www.jstor.org/stable/2137284 13. Lederle M, Tempes J, Bitzer EM. Application of Andersen’s behavioural model of health services use: a scoping review with a focus on qualitative health services research. BMJ Open. 2021;11(5):e045018. doi: https://doi.org/10.1136/bmjopen-2020-045018 14. Radhamony R, Cross WM, Townsin L, Banik B. Culturally and linguistically diverse community access and utilisation of the mental health service: an explanation using Andersen’s behavioural model. Issues Ment Health Nurs. 2024. Available from: https://www.tandfonline.com/doi/pdf/10.1080/01612840.2024.2359602 15. Australian Bureau of Statistics. National study of mental health and wellbeing, 2020-2022. Canberra: ABS; 2023 Oct. Available from: https://www.abs.gov.au/statistics/health/mental-health/national-study-mental-health-and-wellbeing/latest-release 16. Services Australia. Medicare Item Reports[internet]. Available from: https://medicarestatistics.humanservices.gov.au/statistics/mbs_item.html 17. Australian Government Department of Health and Aged Care. Bulk billing incentives in general practice. Canberra: Department of Health and Aged Care; 2026 [updated 2026]. Available from: https://www.health.gov.au/our-work/bulk-billing-incentives-in-general-practice 18. Locascio JJ, Atri A. An overview of longitudinal data analysis methods for neurological research. J Neurol Sci. 2011;307(1-2):5-12. Available from: https://www.researchgate.net/profile/Alireza-Atri/publication/51954192_An_Overview_of_Longitudinal_Data_Analysis_Methods_for_Neurological_Research/links/0912f50c0324dc5000000000/An-Overview-of-Longitudinal-Data-Analysis-Methods-for-Neurological-Research.pdf 19. Bolker BM, Brooks ME, Clark CJ, Geange SW, Poulsen JR, Stevens MHH, et al. Generalized linear mixed models: a practical guide for ecology and evolution. Trends Ecol Evol. 2009;24(3):127-35. doi: https://doi.org/10.1016/j.tree.2008.10.008 20. Johnson TR. J.M. Hilbe (2011) Negative binomial regression, 2nd ed.. Psychometrika. 2012;77(3):611-2. Available from: https://link.springer.com/article/10.1007/s11336-012-9263-7 21. Cameron AC, Trivedi PK. Regression analysis of count data. 2nd ed. Cambridge: Cambridge University Press; 2013. Available from: https://www.cambridge.org/core/books/regression-analysis-of-count-data/2AB83B406C5798030F7C91ECC99B1BE4 22. Linden A. Conducting interrupted time-series analysis for single- and multiple-group comparisons. Stata J. 2015;15(2):480-500. Available from: https://journals.sagepub.com/doi/pdf/10.1177/1536867X1501500208 23. Casals M, Girabent Farrés M, Carrasco JL. Methodological quality and reporting of generalized linear mixed models in clinical medicine (2000-2012): a systematic review. PLoS One. 2014;9(11):e112653. doi: https://doi.org/10.1371/journal.pone.0112653 24. National Health and Medical Research Council; Australian Research Council; Universities Australia. National statement on ethical conduct in human research 2023. Canberra: National Health and Medical Research Council; 2023. Available from: https://www.nhmrc.gov.au/about-us/publications/national-statement-ethical-conduct-human-research-2023 25. Australian Institute of Health and Welfare. AIHW data governance framework 2022. Canberra: AIHW; 2022. Available from: https://www.aihw.gov.au/getmedia/3117bc9c-46ee-4891-a423-2ec0d66e12e7/aihw-data-governance-framework-2022.pdf.aspx 26. Rosenberg SP, Hickie IB. The impact of differences in bulk-billing rates: strategies for greater equity in Medicare. Med J Aust. 2025 Feb 17;222(3):133-134. doi: 10.5694/mja2.52580. Available from: https://doi.org/10.5694/mja2.52580. 27. O'Sullivan BG, Kippen R, Hickson H, Wallace G. Mandatory bulk billing policies may have differential rural effects: an exploration of Australian data. Rural Remote Health. 2022 Mar;22(1):7138. doi:https://doi.org/10.22605/rrh7138<br /> 28. Gerhart J, Piff A, Bartelt K, Barkley E. Telehealth visits unlikely to require in-person follow-up within 90 days [Internet]. Verona (WI): Epic Research; 2022. Available from: https://www.epicresearch.org/articles/telehealth-visits-unlikely-to-require-in-person-follow-up-within-90-days

1. Hua X, Erreygers G, Chalmers J, Laba TL, Clarke P. Using administrative data to look at changes in the level and distribution of out-of-pocket medical expenditure: an example using Medicare data from Australia. Health Policy. 2017;121(4):426-33. doi: https://doi.org/10.1016/j.healthpol.2017.02.003
2. Cao Y, Chen D, Smith M. Use telehealth as needed: telehealth substitutes in-person primary care and associates with the changes in unplanned events and follow-up visits. BMC Health Serv Res. 2023;23:426. doi: https://doi.org/10.1186/s12913-023-09445-0
3. Uscher-Pines L, Fischer SH. Key findings from RAND health care research on telehealth policy [Internet]. Santa Monica (CA): RAND Corporation; 2024. Available from: https://www.rand.org/pubs/research_briefs/RBA1402-1-v4.html
4. U.S. Department of Health & Human Services. Telehealth research recap: economic impact [Internet]. Washington (DC): U.S. Department of Health & Human Services; 2024. Available from: https://telehealth.hhs.gov/documents/ResearchRecap-Telehealth_and_Economic_Impact_09-30-24.pdf
5. Koroma MI, Inungu JN, Adu-Serwaah M, Sultana S, Younis MZ, Iheduru-Anderson K. Disparities in adult mental health service utilization in the United States: a cross-sectional study. Electron J Gen Med. 2025;22(5):em683. Available from: https://doi.org/10.29333/ejgm/16747
6. Panchal N, Lo J. Exploring the rise in mental health care use by demographics and insurance status [Internet]. San Francisco (CA): Kaiser Family Foundation; 2024. Available from: https://www.kff.org/mental-health/exploring-the-rise-in-mental-health-care-use-by-demographics-and-insurance-status/
7. National Rural Health Alliance. Digital health and connectivity in rural Australia - fact sheet [Internet]. Canberra: National Rural Health Alliance; 2024. Available from: https://www.ruralhealth.org.au/wp-content/uploads/2024/05/nrha-digital-health-connectivity-factsheet-apr-24.pdf
8. Bradford NK, Caffery LJ, Smith AC. Telehealth services in rural and remote Australia: a systematic review of models of care and factors influencing success and sustainability. Rural Remote Health. 2016;16:3808. doi: https://doi.org/10.22605/RRH3808
9. Osman S, Churruca K, Ellis LA, Luo D, Braithwaite J. The unintended consequences of telehealth in Australia: critical interpretive synthesis. J Med Internet Res. 2024 Aug 27;26:e57848. doi: https://doi.org/10.2196/57848
10. American Hospital Association. Telehealth fact sheet [Internet]. Chicago (IL): American Hospital Association; 2025. Available from: https://www.aha.org/system/files/media/file/2025/02/Fact-Sheet-Telehealth-20250207_0.pdf
11. Mehrotra A, Perkins J. Telehealth policy brief: advancing telehealth - potential policy solutions to ensure the sustainable and equitable growth of telehealth [Internet]. Providence (RI): Brown University School of Public Health; 2024. Available from: https://cahpr.sph.brown.edu/sites/default/files/documents/Telehealth%20Policy%20Brief_09_2024.pdf

Discussion

Key Findings This study assessed the effects of April 2020 and July 2021 bulk billing policies on the use of three selected mental health services in Australia. The analysis revealed that the April 2020 increase in bulk billing incentives was associated with a modest increase in fully subsidised in-person services (MBS 2715) as well as a slight uptick in the trend of the partly subsidised in-person services (MBS 281) without an immediate change. On the other hand, the July 2021 intervention was associated with a notable effect only on the use of fully subsidised telehealth services (MBS 92116). It was associated with a 7.7% decline in the odds of service use as well as a decline in its post-intervention trajectory. There were notable and persistent differences by gender, age group, and States/Territories across all service types, highlighting the role of structural and behavioural factors as important variables in shaping the patterns of service utilisation during widespread policy implementation in the health system.

Connection to Existing Literature Aligning with Andersen’s Behavioural Model of Health Services Use, the results of this study imply that affordability may be a key factor in service use. These findings of this study also support those of Rosenberg and Hickie (26), showing that financial incentives play a major role in service use among socioeconomically vulnerable groups. In addition, this study further buttresses the points made by O'Sullivan et al (27) that reductions in telehealth funding disproportionately harm rural patients with limited alternatives. Conversely, our results diverge from this discrepancy, which may stem from the complexity of policy implementation during COVID-19 or the lag in patient and provider response to changing billing structures. Additionally, the results of a study also indicated that patients are unlikely to revert to in-person care after adopting telehealth (28), whereas our modelled results show a statistically significant decline in telehealth uptake after the reduction in incentives, indicating that financial barriers remain influential. Hua et al(29) showed that there were gradual increases in the use of services following Medicare reforms. In the present study, the observed change in the use of services was thought to be due to several factors, ranging from the complexity of COVID-19 policy implementation to the lag in patient and provider response and the changing bulk billing structures. Additional literature also suggests continued use of telehealth after adoption (30). In contrast, our study showed a significant decline in telehealth use after the reduction in incentives, implying that financial barriers are a major factor in patients’ use of services, whether telehealth or in-person.

Interpretation of Results The link between financial incentives and service utilisation remains strong. The April 2020 bulk billing incentives likely encouraged more providers to offer fully subsidised care, particularly for MBS 2715. In contrast, the July 2021 policy change did not cause an immediate drop in raw service volumes but was associated with a 7.7% reduction in the odds of telehealth use. This suggests a dampening effect on service growth, even amid sustained demand, reflecting the sensitivity of telehealth uptake to financial settings (31,32). The findings of this study suggest that financial incentives contributed to a decline in service use. While April 2020 was associated with an immediate shift in service use counts, the modelled estimates indicate a 7.7% reduction in the odds of telehealth use. The observed contrast between the raw counts of service use and the adjusted odds highlights the significance of accounting for demographic and regional variables. This allowed the model to identify the associated policy effect from broader contextual changes. The results also suggest that the use of telehealth services may have continued in the downward trend had the billing policy not changed. This study also attests to the persistent demographic disparities in where males and older adults record lower mental health service use (33,34). In addition, differences in service use among States/Territories further reveal structural and service delivery factors, with NT residents consistently underserved. Unexpectedly, the pattern of use of the partly subsidised service (MBS 281) was relatively stable throughout the index period, possibly reflecting entrenched practice patterns among service providers or patients.

Limitations There are a few limitations of this study. First, the use of aggregated national administrative data, such as the MBS data) limits the ability to account for individual-level insights, including clinical severity or socioeconomic status, which may influence service use. Second, overlapping policy changes, including COVID-19 pandemic policies and the bulk billing reforms, may have created confounding factors that challenge causal inference. Third, our statistical models accounted for demographic and regional differences, as well as unmeasured variables such as internet access. However, the local health workforce, cultural attitudes toward mental health, and other extraneous variables not accounted for could have influenced service use.

Implications for Practice The results highlight the need for long-term, stable policy planning in mental healthcare financing. Sudden changes to health providers’ motivation in the form of incentives may influence service disruption, exacerbating inequities, in underserved communities. This study suggests that policymakers should consider the impact of financial structures on patients' behaviour and service delivery. Outreach to older adults and men, and digital literacy programs, could mitigate disparities. Continued support for telehealth infrastructure and reimbursement is especially crucial in rural areas where physical access to care remains limited (35–39). Policy implementation may be hampered by funding constraints, provider buy-in, and digital capacity. Overcoming these requires concerted efforts at all levels, including the formulation of relevant national state policies on mental health, investment in workforce development, and community-based digital inclusion efforts.

Conclusion and Future Research In summary, this study demonstrates that bulk billing policy interventions significantly influence mental health service use. In addition, it also shows that financial incentives play a key role in service use patterns. The findings highlight the need for cohesive policy frameworks that avoid abrupt reversals and better account for regional and demographic disparities. Future research should explore patient and provider perspectives on bulk billing policies and practices. Furthermore, researchers should assess the long-term effects associated with the 2023 tripling of bulk billing incentives. Addressing these gaps will support more equitable and effective mental healthcare delivery under Australia’s Medicare system.

Declaration Ethics The study was part of a larger study for which ethical clearance (ETH2023-0357) was obtained. De-identified, publicly available data was used. Hence, informed consent and participant compensation were not relevant for the study..

Hurdle Models

Partly subsidised in-person mental health service (281) The hurdle model used for the evaluation of the associated effects of the April 2020 and July 2021 interventions on service use consisted of two components, including: i) a count model, which assessed service counts where at least one service was reported used, and (ii) a zero model, which estimated the probability of any service use. Age group, gender, and States/Territories, as well as level and slope parameters, were included in the models for the April 2020 and July 2021 interventions. Before 2020, service use showed a slight upward trend (β = 0.03, 95% CI: 0.00–0.06, p = 0.028). There was no significant immediate change in service use associated with the April 2020 intervention (β = 0.65, 95% CI: -0.72 to 2.02, p = 0.355), and the trend following the intervention was positive (β = 0.05, 95% CI: 0.01–0.09, p = 0.031) (Table 2). There were no changes in service use on account of the July 2021 intervention (Immediate impact, β = -0.98, 95% CI: -12.82 to 10.86, p = 0.860); Long-term (β = 0.00, 95% CI: -0.01 to 0.01, p = 0.874) (Table 2). Compared with females, males had consistently had lower service use at both intervention points (April 2020: β = -0.36, 95% CI: -0.63 to -0.07, p = 0.015; July 2021: β = -0.63, 95% CI: -0.69 to -0.57, p < 0.001). Service use was lower among older adults, especially those aged ≥65 (April 2020: β = -2.05, 95% CI: -3.27 to -2.35, p < 0.001; July 2021: β = -2.47, 95% CI: -2.57 to -2.37, p < 0.001) (Table 2). Overall, the hurdle model results revealed that demographic and geographic disparities in service use limited the immediate effects of the interventions (Table 2). Table 2: Hurdle Model Results Showing the Impact on April 2020 and July 2021 Interventions on In-Person Mental Health Service (281) April 2020 Intervention July 2021 Intervention Logistic Regression Negative Binomial GLMM Logistic Regression Negative Binomial GLMM Variable Coefficient Std. Error P-Value Coefficient Std. Error P-Value Coefficient Std. Error P-Value Coefficient Std. Error P-Value Intercept 2.82 0.77 0.000 1.35 0.28 0.000 2.94 0.40 0.000 1.64 0.14 0.000 Time -0.01 0.02 0.540 0.01 0.01 0.513 -0.07 0.02 0.001 -0.03 0.01 0.003 Intervention 0.00 0.00 0.201 0.00 0.00 0.995 0.00 0.00 0.106 0.00 0.00 0.016 Time After Intervention 0.02 0.02 0.434 0.01 0.01 0.522 -0.01 0.03 0.747 0.02 0.01 0.244 Sex (Ref: Female) Male -1.08 0.33 0.001 -0.36 0.14 0.011 -0.79 0.17 0.000 -0.43 0.07 0.000 Age Category, in years (Ref: 15-24) 25-34 -0.33 0.83 0.690 0.13 0.23 0.577 0.50 0.44 0.251 0.10 0.12 0.426 35-44 -1.25 0.81 0.120 -0.04 0.24 0.849 -0.19 0.42 0.657 -0.27 0.12 0.027 45-54 -1.56 0.80 0.051 -0.40 0.24 0.091 -0.45 0.41 0.276 -0.64 0.12 0.000 55-64 -1.99 0.79 0.012 -0.73 0.25 0.003 -0.94 0.41 0.022 -1.15 0.13 0.000

=65 -5.58 0.79 0.000 -2.05 0.24 0.000 -4.55 0.39 0.000 -2.05 0.13 0.000 State (Ref: ACT) NSW 4.15 0.74 0.000 1.79 0.28 0.000 2.39 0.38 0.000 1.96 0.15 0.000 NT -4.83 0.78 0.000 -1.41 0.77 0.066 -3.63 0.36 0.000 -0.97 0.29 0.001 QLD 3.46 0.72 0.000 1.40 0.28 0.000 2.09 0.37 0.000 1.93 0.15 0.000 SA 1.74 0.69 0.012 0.23 0.29 0.428 1.81 0.36 0.000 0.88 0.15 0.000 TAS -1.94 0.59 0.001 -1.10 0.38 0.004 -1.50 0.33 0.000 -0.20 0.18 0.263 VIC 4.49 0.76 0.000 2.25 0.28 0.000 2.67 0.38 0.000 1.93 0.15 0.000 WA 2.33 0.69 0.001 1.33 0.29 0.000 1.49 0.36 0.000 1.18 0.15 0.000 Interactions time:State[NSW] 0.09 0.04 0.039 <br /> time:State[QLD] 0.09 0.04 0.033 <br /> time:State[SA] 0.08 0.04 0.044 <br /> time:State[TAS] 0.08 0.03 0.024 <br /> intervention:('age category')[45-54] -1.85 0.91 0.043 <br /> intervention:State[SA] -0.78 0.39 0.045   Telehealth Mental Health Service (92116) The hurdle model was also used to model the use of the fully subsidised telehealth health service (92116) and the associated effects of the July 2021 intervention. The pre-intervention trend of service use showed a small but statistically significant upward direction (β = 0.15, p < 0.001). While the intervention had no immediate effect (β = 0.00, p = 0.233), the post-intervention trend was in the upward direction (β = 0.09, p < 0.001) (Table 3). The evidence of regional disparities was clear with NSW, VIC, and QLD showing the highest service counts (VIC: logistic β = 4.49, p < 0.001; GLMM β = 3.18, p < 0.001). The lowest service use count was in NT (logistic model: β = -2.03, p = 0.006).<br /> Overall, there was no change in telehealth use immediately after the July 2021 bulk billing intervention. However, demographic and regional disparities remained significant. The AIC and BIC values indicated good model fit, and the residual diagnostics did not reveal any evidence of significant violations of model assumptions.  Table 3: Hurdle model results showing the impact of July 2021 bulk billing change on telehealth, fully subsidised mental health service (92116)

Logistic Regression

Negative Binomial GLMM

Variable Coefficient Std. Error P-Value Coefficient Std. Error P-Value Intercept 0.48 0.43 0.273 0.68 0.28 0.014 Time 0.15 0.02 0.000 0.07 0.01 0.000 Intervention 0.01 0.00 0.000 0.00 0.00 0.233 Time After Intervention -0.08 0.03 0.010 0.09 0.02 0.000 Sex (Ref: Female) <br /> Male -0.88 0.23 0.000 -0.62 0.12 0.000 Age Category, in years (Ref: 15-24) <br /> 25-34 1.20 0.54 0.025 0.33 0.19 0.080 35-44 0.08 0.49 0.876 0.10 0.20 0.612 45-54 -0.38 0.48 0.423 -0.36 0.20 0.072 55-64 -1.00 0.47 0.032 -0.78 0.21 0.000

=65 -3.70 0.45 0.000 -1.65 0.21 0.000 State (Ref: ACT) <br /> NSW 3.66 0.51 0.000 2.36 0.28 0.000 NT -1.48 0.46 0.001 -0.28 0.48 0.561 QLD 3.77 0.51 0.000 1.52 0.29 0.000 SA 2.06 0.45 0.000 0.64 0.30 0.034 TAS 0.80 0.41 0.053 0.40 0.33 0.224 VIC 4.49 0.55 0.000 3.18 0.28 0.000 WA 2.53 0.47 0.000 0.95 0.30 0.001 Interactions <br /> time:State[QLD] -0.11 0.05 0.014 <br /> time:State[SA] -0.10 0.04 0.018 <br /> time:State[TAS] -0.09 0.04 0.023 <br /> intervention:Q('age category')[>=65] -1.74 0.77 0.023 <br /> intervention:State[NT] -2.03 0.74 0.006 <br /> intervention:State[SA] 1.47 0.72 0.043 <br /> time:State[WA] -0.10 0.03 0.000 intervention:State[NT] 1.94 0.63 0.002 intervention:State[QLD] 2.18 0.40 0.000 intervention:State[SA] 2.29 0.43 0.000 intervention:State[TAS] 2.51 0.48 0.000 intervention:State[VIC] 1.44 0.40 0.000 intervention:State[WA] 2.69 0.42 0.000 time:Q('age category')[>=65] -0.04 0.02 0.028 time:State[NT] -0.13 0.04 0.001 time:State[QLD] -0.07 0.03 0.008 time:State[SA] -0.08 0.03 0.003 time:State[TAS] -0.13 0.03 0.000 time:State[VIC] -0.08 0.02 0.002

Results

Participants There were 24,626 observations in the datasets, comprising the MBS service claims from October 2018 to December 2022, stratified by age group, gender, and State/Territory of residence (NSW, VIC, QLD, SA, WA, TAS, ACT, and NT). Each MBS item had the service use count, total services provided across States/Territories. All extracted datasets had completed elements.

Trend Analysis The plot of the time series for each mental health service revealed fluctuations. The fully subsidised in-person service (MBS 2715) declined over time, while the telehealth service (MBS 92116) displayed a sharp uptick, and the partly subsidised in-person service (MBS 281) remained stable (Appendices 1 and 2).

Interrupted Time Series (ITS) with Negative Binomial, Generalised Linear Mixed Model (GLMM) Before April 2020, service use counts showed an upward trend (β = 0.03, 95% CI: 0.00 to 0.06, p = 0.028), signifying a gradual increase in the use of services. The April 2020 Intervention was not associated with an immediate change in level of service use (β = 0.65, 95% CI: -0.72 to 2.02, p = 0.355). However, there was an increase in the slope of service use β = 0.05, 95% CI: 0.01 to 0.09, p = 0.031) post intervention, suggesting service use increase continued after the implementation of the intervention. For the July 2021 intervention, there were neither significant effects on service use in the immediate (β = -0.98, 95% CI: -12.82 to 10.86, p = 0.860), nor in the post-intervention (β = 0.00, 95% CI: -0.01 to 0.01, p = 0.874) periods (Table 1, Appendix 2). The Negative Binomial Generalised Linear Mixed Models (NB GLMM) used to assess the association between States/Territories, age group, and gender, and the use of Medicare-subsidised in person service (2715) over time, showed that males had consistently lower values compared to females (April 2020: β = -0.35, p = 0.015; July 2021: β = -0.63, p < 0.001). Relative to ACT, the Northern Territory (April 2020: β = -1.44, p < 0.001; July 2021: β = -1.18, p < 0.001) had lower service counts. Older adults (≥ 65) had a slower post-April 2020 increase (β = -0.18, p < 0.001), highlighting an age-related disparity in the growth of service use (Table 1).   Table 1: Interrupted Time Series (ITS) Analysis with Negative Binomial, Generalised Linear Mixed Models for In-Person Service (2715) April 2020 Intervention

July 2021 intervention

Variable Coefficient Standard Error P-Value Estimate Standard Error P-Value Intercept 4.18 0.37 <0.001 4.94 0.06 <0.001 Time 0.03 0.01 0.028 0.00 0.00 0.041 Intervention 0.65 0.70 0.355 -0.98 5.53 0.860 Time after intervention 0.05 0.02 0.031 0.00 0.01 0.874 Sex (Ref: Female) <br /> Male -0.35 0.14 0.015 -0.63 0.03 <0.001 Age Category, in years (Ref: 15-24) <br /> 25-34 0.11 0.28 0.698 0.08 0.06 0.160 35-44 -0.10 0.28 0.714 -0.20 0.06 0.001 45-54 -0.36 0.28 0.197 -0.53 0.06 <0.001 55-64 -0.81 0.28 0.004 -0.94 0.06 <0.001

=65 -2.81 0.23 <0.001 -2.47 0.05 <0.001 State (Ref: ACT) <br /> NSW 3.09 0.28 <0.001 3.05 0.06 <0.001 NT -1.44 0.29 <0.001 -1.18 0.06 <0.001 QLD 2.71 0.28 <0.001 2.73 0.06 <0.001 SA 1.56 0.28 <0.001 1.54 0.06 <0.001 TAS 0.22 0.28 0.447 0.30 0.06 <0.001 VIC 2.95 0.28 <0.001 2.90 0.06 <0.001 WA 1.77 0.28 <0.001 1.92 0.06 <0.001 Interactions <br /> Time and age category >=65 -0.18 0.04 <0.001 <br /> Intervention and State - VIC -1.54 0.66 0.020

Statistical Analyses Descriptive Analysis Descriptive statistics were conducted, and continuous variables were reported as means (±SD) and medians (Q1, Q3). Before and after interventions, monthly service counts were computed. We also plotted scatter diagrams and lines of best fit. Time series plots, percentage changes over time, and trends across key periods were conducted. For consistent comparison, rates of service use were normalised as a percentage of each service’s peak to allow consistent comparison. Multivariate analysis The multivariate analyses conducted in this study assessed the effects associated with bulk billing reforms across age groups, genders, and States/Territories. We developed models for Interrupted Time Series (ITS) using a Generalised Linear Mixed Model (GLMM) with a Negative Binomial distribution for the fully subsidised in-person mental health service (2715). A preliminary analysis showed that the partly subsidised service (281) and the telehealth service (92116) had many zero counts. Hence, we developed hurdle models to manage excess zeros and overdispersion. We developed separate models for each service and intervention. For in-person services (2715 and 281), we assessed the effects of April 2020 and July 2021 interventions on their use, while the effects of the July 2021 intervention were examined on the use of the telehealth service (92116). We did not examine the effects of the April 2020 intervention on the telehealth mental health service, as the service was introduced the previous month. For all models, we used monthly data spanning 18 months pre- and post-intervention.<br /> Interrupted Time Series (ITS) with Negative Binomial, Generalised Linear Mixed Model (GLMM) Widely acknowledged as a suitable method for analysing quasi-experimental, longitudinal interventions, Interrupted time series (ITS) was used in this study. ( 18). The study was meant to address hierarchical and repeated measures (monthly service counts in States/Territories). Hence, integrating ITS with Generalised Linear Mixed Models (GLMMs) was appropriate, allowing for the inclusion of fixed and random effects (18,19). The preliminary data analysis indicated that overdispersion was evident in the monthly service count of in-person, fully subsidized mental health service (MBS 2715). Hence, we used the Negative Binomial (NB) distribution, which offers a better fit for count data exhibiting high variability (20–23). The study analysed both the immediate and long-term effects of the identified bulk billing interventions over time. A random residual with an autoregressive variance structure was used to account for the correlation of the count of services over time and for overdispersion. Gender, age group, and State/Territory were modelled as fixed effects, controlling demographic and geographic variables. The model was specified as follows: Log (Yit)=β0 + β1(time) + β2(intervention) +β3 (time × intervention) + β4(sex) + β5(age group) + β6(state)+ϵit with Yit representing the number of services claimed, β coefficients the model parameters, and ϵit the error term. Akaike information criteria (AIC) values and Bayesian Information Criterion (BIC) were used for assessing model fit, and residual diagnostics were evaluated.

Hurdle Models Preliminary evaluation also showed that partly subsidized, in-person service (281) and telehealth service (92116) showed zero-inflated count data pattern, indicating overdispersion and excess zeros. Studies have shown that hurdle models are used for handling zero-inflated count data, where some individuals never use a service (structural zeros), while others use it at varying rates. Being an effective method, the hurdle models were used to model these services (281 and 92116). Each model consists of a logistic regression to identify factors linked to any service use, and a Generalised Linear Mixed Model with a negative binomial distribution for non-zero counts. The study modelled age, gender, and State/Territory as independent variables., reported coefficients, standard errors, and p-values of final models and evaluated AIC, BIC, log-likelihood, and goodness-of-fit tests in adjudging model fit.   Bias Aggregated data was used in the study, ensuring uniformity across jurisdictions and services and minimising bias. The study was conducted in keeping with ethical standards (24, 25). All analyses were carried out at 5% level of significance. All analyses were conducted using Python 3.11 with the statsmodels package (version 0.14).

Study design and participants The study was conducted using publicly accessible Medicare Benefits Schedule (MBS) items datasets. The three selected mental health services assessed include: i) MBS 2715, a fully subsidised mental health in-person service for preparing a mental health treatment plan; ii) MBS 92116, a fully subsidised telehealth equivalent of MBS 2715; and iii) MBS 281, a Medicare partly subsidised in-person mental health service. The provider’s fees for each of these services were AUD 103.70. The report of this study was prepared in keeping with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline. Procedures This study assessed the effects of the April 2020 increase in bulk billing incentives and the July 2021 reduction in telehealth incentives on the use of the selected mental health services. Therefore, we downloaded the monthly aggregated datasets for the relevant MBS items from the website of Services Australia. We downloaded datasets from October 2018 to December 2022 (9,10,16), to allow sufficient time for observation pre- and post-intervention, thereby ensuring temporality in the study. It is important to note that MBS item 2715 was already established before 2018, while MBS 281 was introduced in July 2018 (3,10 ).The three MBS items - 2715, 92116, and 281 were selected to explore several dimensions of change in service use. MBS 2715 and MBS 92116 (Appendix 1) relate to initiating Mental Health Treatment Plans, a reflection of help-seeking behaviour. MBS 92116, the telehealth equivalent of MBS 2715, indicates shifts in the modality of service delivery during the COVID 19 pandemic. We included MBS 281 to ensure variations in out-of-pocket expenditure as well as changes in providers' billing practices under changing bulk billing incentives. Socio-demographics The datasets for the selected MBS items were extracted, exported to Microsoft Excel documents, and prepared for analysis, including the service used by age group, gender, and date. Variables The study included explanatory variables, interventions, and an outcome variable. Explanatory variables used in the study include age (15–24, 25–34, 35–44, 45–54, 55–64, 65+), gender (male, female), and State/Territory [New South Wales (NSW), Victoria (VIC), Queensland (QLD), South Australia (SA), Western Australia (WA), Tasmania (TAS), Australian Capital Territory (ACT), Northern Territory (NT)]. The two policy changes (interventions) examined include the April 2020 increase in bulk billing incentives and the July 2021 reduction in telehealth incentives. We created binary indicators, indicating pre-intervention (0) and post-intervention (5, 10, 17). The outcome variable was defined as the monthly total number of services used.

Theoretical Framework The Andersen’s Behavioural Model of Health Service Use is a widely used framework in the literature (11,12)was used to assess the use of mental health services was applied in this study. The model considers the factors that predispose to (e.g., age, gender)(13), enable (e.g., policy changes, affordability), and influence (e.g., severity of mental health conditions) the use of health services (14). This model provides a guide to the analysis of Medicare bulk-billing policy reforms and their effects on the use of mental health services in Australia (Figure 1). Mental health is a key focus of bulk billing reforms (7,8), prompting the need to assess how policies have influenced (6) pressure on the healthcare system(15), out-of-pocket expenses, and service use.

Figure 1: A Conceptual Framework illustrating the Effects of Bulk Billing Policy Changes on the Use of Mental Health Services. (Based on Andersen’s Behavioural Model of Health Services Utilisation)

Australia prides itself on being one of the few countries with a universal health system catering to the needs of all its citizens (1). Central to this system is affordability (1–3). Upticks in the cost of living have given rise to a growing concern regarding the declining bulk-billing practices by healthcare providers, and the consequent rise in out-of-pocket expenditures for healthcare services (3). Out-of-pocket expenditure, reflecting the gap between providers’ fees and Medicare rebates, continues to raise concerns for individuals and families seeking healthcare in Australia (2).<br /> With bulk billing designed to eliminate out-of-pocket costs through the Medicare Benefits Scheme (MBS), Australia aims to improve the health of its citizens (1,3). Bulk billing helps vulnerable groups like pensioners, children, and concession cardholders access healthcare. Frequent changes to bulk billing policies from 2019 to 2023 and the overlapping reforms have made evaluating bulk billing practices a challenge(1,3,4). There were several bulk billing policy reforms in Australia from 2019 to 2023, including the introduction of temporary telehealth bulk billing incentives in 2020 (1,5), and the subsequent addition of further incentives, especially for mental health care (1,3). In 2023, there was an introduction of tripling of bulk billing incentives to reverse the decline in service use among vulnerable groups (3,4). <br /> These multiple reforms were aimed at increasing access to services, improving cost effectiveness, and motivating healthcare providers. However, geographic and socioeconomic disparities persisted as well as inequities in access to healthcare due to high out-of-pocket expenditure in some locations despite the bulk billing policy reforms (1,3,4). While billions of dollars were invested to increase access to services, especially among vulnerable populations, bulk‑billing rates did not significantly improve, highlighting the complex nature of implementation and effects of system-wide policies in the health system (1,2,6). Mental health is a key health challenge in Australia, with almost 50% of Australians aged 18 years and above facing a mental health problem at least one time in their lives (7). Rising to the challenge, the Australian Government enacted policies and framework to tackle this menace (5). These include the development of the Fifth National Mental Health and Suicide Prevention Plan (2017–2022); the National Mental Health and Suicide Prevention Agreement; Vision 2030 for mental health reform; and the National Mental Health Policy National Mental Health Workforce Strategy (2022–2032) (5,7). Highlighting mental health as a national priority, the government also instituted a major Productivity Commission Inquiry into mental health in Australia (7). To improve mental healthcare access, policies developed were targeted at financial and geographic barriers (8). The rapid pace of policy changes and the time lag necessary for results create methodological challenges for isolating effects. Given these complexities, there is a dire need for an empirical study to examine and highlight the effects associated with bulk billing policy reforms. This study aims to highlight bulk billing changes from 2019 to 2023 and assess associated effects on mental health service use and disparities across states and territories. By focusing on fully subsidised Medicare mental health in-person services, partially subsidised Medicare mental health services, and telehealth mental health services, this study seeks to provide a clearer understanding of policy effects towards guiding future reforms.

Brief Review of Literature Highlights of Bulk Billing Policy Changes (2019–2023) The bulk billing policy changes developed under the Medicare Benefits Scheme (MBS) from 2019 to 2023 were designed to improve access to services, promote equity, and protect from catastrophic health expenditure, especially for vulnerable populations (5,9). These include: Introduction of temporary telehealth bulk billing incentives during COVID-19 towards ensuring continued access in March 2020, especially for mental health care.<br /> Addition of extra incentives in April 2020, towards encouraging bulk billing for vulnerable populations, including children younger than 16 years of age, pensioners, and concession cardholders, April 2020. July 2021 scaling back of telehealth incentives, enabling health providers to charge fees; and Tripling of bulk billing incentives in November 2023 for vulnerable populations towards reversing the decline in service use. While these policy reforms were dynamic and reactive, they aimed at addressing immediate challenges. These include the COVID-19 pandemic, the rise in operational costs, and declining use of services (5,9,10).

Theoretical Framework The Andersen’s Behavioural Model of Health Service Use is a widely used framework in the literature (11,12)was used to assess the use of mental health services was applied in this study. The model considers the factors that predispose to (e.g., age, gender)(13), enable (e.g., policy changes, affordability), and influence (e.g., severity of mental health conditions) the use of health services (14). This model provides a guide to the analysis of Medicare bulk-billing policy reforms and their effects on the use of mental health services in Australia (Figure 1). Mental health is a key focus of bulk billing reforms (7,8), prompting the need to assess how policies have influenced (6) pressure on the healthcare system(15), out-of-pocket expenses, and service use.

Figure 1: A Conceptual Framework illustrating the Effects of Bulk Billing Policy Changes on the Use of Mental Health Services. (Based on Andersen’s Behavioural Model of Health Services Utilisation)

Author Update (June 2026)

Following a subsequent audit of this preprint, several references and supporting citations were identified as inaccurate. The manuscript was later comprehensively revised, references were re-verified against original sources, and corrections were incorporated into a revised version submitted subsequently. Readers are advised to consult the annotations within this preprint regarding affected citations.

Appendix 1: Cost-Effectiveness Analysis Components Component Description Reference / Justification Health States Minimal, Mild, Moderate, Severe, Death GAD-7 validated by Spitzer RL et al (12) Transition Probabilities Monthly transitions over 60 cycles Minimal → Mild Mild → Moderate Moderate → Severe Severe → Death Delgadillo J et al (27) i. There is no longitudinal data found in the literature on monthly transition rates from one state to another. ii. NHS Digital (20) shows that less than 60% of the overall recovery rate for those completing treatment over the course of 1 year. With less than 60% over 12 months, a significant shift toward the "Minimal" state may require a 5% monthly rate. iii. Delgadillo et al. (27) provides information on the reduced rates of deterioration (moving left). Clinical evidence shows that while no-treatment cohorts have Reliable Deterioration (RD) rates of about 6%, patients receiving therapy experience reliable deterioration of about 3%, justifying the lower probabilities for moving from Mild to Moderate or Moderate to Severe. iv. Sandin et al. (26) validates the dose-response relationship. The study shows that about 68% of patients recovered after an average of 10 therapy sessions. In comparison, about 18%–24% of treated patients remained "unchanged," supporting a much lower diagonal value for patients in treatment. v. Jankovic et al (36) supports the stepwise improvement logic and related mortality. vi. The intervention improves clinical states. However, it may not completely remove mortality risk. vii. A separate transition probability matrix was assumed for the no-treatment group to ensure that QALYs gained are state-based and not per session.

viii. These reflect clinically plausible transitions over monthly cycles in chronic mental health care, modelled over 60 cycles (5 years). The transition probabilities used in this study were assumptions informed by the literature. ix. The assumption of improvement has been revised to show 3% improvement from severe to moderate category and 1% from moderate to mild monthly. x. An assumption of death rates of 3% among patients with severe anxiety/depression, 1% for people with moderate anxiety/depression per cycle (month), and 0% for other categories per month was made. xi. Vos et al (18) provide information on the maintenance of treatment effect. The study indicated that maintenance therapy may prevent over 50% of the disease burden over five years. This justifies the high retention in the "Minimal" state (0.97) for patients in treatment, as maintenance/booster sessions significantly reduce the natural relapse rates seen in untreated populations. xii. Jankovic et al (36) indicated that patients may improve at an annual rate of 15% in the first year. Hence, a monthly rate of 1% improvement transition was assumed. xiii. Delgadillo et al (27) showed that the rates of deterioration may range from about 3% to 6% for depression and anxiety. xiv. <br /> Utilities (QALY Weights) Minimal (0.90), Mild (0.75), Moderate (0.60), Severe (0.40), Death (0.00) i) Australian Burden of Disease Study (23) cited a utility of 0.00 for death ii) “Minimal” symptoms state utility, representing the health utility of individuals who have recovered or who function at a "normal" community level, was cited as 0.91 by McCaffrey et al (37). iii) Mihalopoulos et al (10) indicated utility (1-disability weight) values for mild, moderate, and severe depression: 0.86, 0.65, and 0.24, respectively. Australian Burden of Disease Study (23) indicated that these values were 0.85, 0.6, and 0.44, respectively. The conditions being represented here are mental health disorders. Hence, we assigned these utilities being guided by evidence.

Costs – Direct Healthcare Telehealth & In-Person: AUD $166.85/session; No Treatment: AUD $0–30/year (minimal On-The-Counter (OTC) or incidental care only) Services Australia (MBS Items 80010, 91167, 91182).

Costs – Societal In-person: AUD $181.20/session (4 hrs × average hourly wage); Telehealth: minimal Australian Bureau of Statistics (ABS) Average Weekly Earnings, 2023. Perspective Healthcare payer and societal PBAC Guidelines (19) Time Horizon 5 years (60 monthly cycles) 10, 18, 36 Discount Rate 5% annually (0.417% monthly) PBAC Guidelines (19). Comparators Telehealth, In-Person, No Treatment 30 Outcomes Total Costs, Total QALYs, ICERs CHEERS Reporting Standards (22) Baseline distribution of the cohort 17.9% - Mild;<br /> 78.6% - Moderate; 3.5% - Severe Janvovik et al (36)

Appendix 2: MBS Items for Psychological Therapy Services and Focused Psychological Strategies (Allied Mental Health) In-person Videoconference item Telephone item (when videoconferencing facilities are not available) Duration Clinical Psychologists 80010 91167 91182 Attendance lasting at least 50 minutes 80011 Like 80010 but provided via video conference. The patient must be in a telehealth eligible area, at least 15 kilometres from the psychologist.

Appendix 2: Key Assumptions in the Conduct of the Budget Impact Analysis Model Assumption Value Justification Population base (aged 15–64) 32,669 individuals per year Derived from MBS telehealth utilisation data for clinical psychology Population growth rate 1.9% annually Based on ABS population statistics (2023) Service uptake Stable baseline uptake across 5 years Assumes steady demand given consistent service access and policy coverage Number of visits per user 1.5 visits per quarter Assumed to be the typical service patterns in mental health care Unit cost per consultation AUD $166.85 (telehealth & in-person) Reflects the MBS rebate schedule for psychology consultations Support cost per user per year AUD $32 Assumed costs to include admin, monitoring, IT and system infrastructure for telehealth delivery Coverage rate by Medicare 85% of service costs Based on standard public healthcare subsidies for psychology consultations Clinical equivalence Telehealth = In-person outcomes Supported by literature showing parity in effectiveness and satisfaction. Model perspective Public healthcare payer Reflects real-world budgetary planning by Medicare and public funders Discount rate 0% Standard BIA approach; short-term (5-year) cash flow model

Appendix 3: Cohort trace for treatment arm  

Appendix 4: Budget impact analysis scenario

Appendix 5: Annual Projection of Telehealth Users Year Projected Users (15-64 years) Year 1 33,290 Year 2 33,922 Year 3 34,567 Year 4 35,224 Year 5 35,893

1. McCaffrey N, Kaambwa B, Currow DC, Ratcliffe J. Health-related quality of life measured using the EQ-5D-5L: South Australian population norms. Health Qual Life Outcomes. 2016 Sep 20;14(1):133. doi: 10.1186/s12955-016-0537-0.

1. Jankovic D, Saramago Goncalves P, Gega L, Marshall D, Wright K, Hafidh M, et al. Cost Effectiveness of Digital Interventions for Generalised Anxiety Disorder: A Model-Based Analysis. Pharmacoecon Open. 2022 May;6(3):377-388. doi: 10.1007/s41669-021-00318-y.

1. Powell RE, Henstenburg JM, Cooper G, Hollander JE, Rising KL. Patient perceptions of telehealth primary care video visits. J Med Internet Res. 2017;19(2):e76. PMID: 28213341. DOI: https://doi.org/10.1370/afm.2095

2. Snoswell CL, Taylor ML, Comans TA, Smith AC, Gray LC, Caffery LJ. Determining if Telehealth Can Reduce Health System Costs: Scoping Review. J Med Internet Res. 2020;22(10):e17298. doi: 10.2196/17298.

1. Reay RE, Looi JCL, Keightley P. Telehealth Mental Health Services during COVID-19: Summary of Evidence and Clinical Practice. Australas Psychiatry. 2020;28:514–16. Available from: https://doi.org/10.1177/1039856220943032
2. Yellowlees P, Nakagawa K, Pakyurek M, Hanson A, Elder J, Kales HC. Rapid conversion of an outpatient psychiatric clinic to telepsychiatry in response to COVID-19. Psychiatr Serv. 2020;71(7):749–752. Available from https://doi.org/10.1176/appi.ps.202000230
3. Kruse CS, Krowski N, Rodriguez B, Tran L, Vela J, Brooks M. Telehealth and patient satisfaction: a systematic review and narrative analysis. BMJ Open. 2017;7(8):e016242. doi:10.1136/bmjopen 2017 016242. PMID: 28775188. Available from: https://doi.org/10.1136/bmjopen-2017-016242

1. Australian Bureau of Statistics (ABS). Average Weekly Earnings, Australia. Canberra: ABS; 2023. Available from: https://www.abs.gov.au/statistics/labour/earnings-and-working-conditions/average-weekly-earnings-australia/may-2023
2. Gamst-Klaussen T, Lamu AN, Chen G, Olsen JA. Assessment of outcome measures for cost-utility analysis in depression: mapping depression scales onto the EQ-5D-5L. BJPsych Open. 2018;4(4):160-166. doi: 10.1192/bjo.2018.21.
3. Australian Institute of Health and Welfare (AIHW). Australian Burden of Disease Study: methods and supplementary material 2018 — Years lived with disability (YLD). Canberra: AIHW; 2021. Available from: https://www.aihw.gov.au/reports/burden-of-disease/abds-methods-supplementary-material-2018/contents/estimating-burden-of-disease-measures/years-lived-with-disability-yld
4. Whiteford HA, Ferrari AJ, Degenhardt L, Feigin V, Vos T. The global burden of mental, neurological and substance use disorders: an analysis from the Global Burden of Disease Study 2010. PLoS One. 2015;10(2):e0116820. doi:10.1371/journal.pone.0116820. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC4320057/

5. Clark DM, Canvin L, Green J, Layard R, Pilling S, Janecka M. Transparency about the outcomes of mental health services (IAPT approach): an analysis of public data. Lancet. 2018;391(10121):679‑686. Available from: https://doi.org/10.1016/S0140-6736(17)32133-5

6. Sandin K, Shields G, Gjengedal RGH, Osnes K, Bjørndal MT, Reme SE, et al. Responsiveness to change in health status of the EQ-5D in patients treated for depression and anxiety. Health Qual Life Outcomes. 2023;21(1):35. doi: 10.1186/s12955-023-02116-y.

7. Delgadillo J, Overend K, Lucock M, Groom M, Kirby N, McMillan D, et al. Improving the efficiency of psychological treatment using outcome feedback technology. Behav Res Ther. 2017;99:89-97. doi: 10.1016/j.brat.2017.09.011.
8. Australian Bureau of Statistics (ABS). National Study of Mental Health and Wellbeing, 2020–22. Canberra: ABS; 2022. Available from: https://www.abs.gov.au/statistics/health/mental-health/national-study-mental-health-and-wellbeing/latest-release
9. Services Australia. Medicare Benefits Schedule (MBS) Item Reports[internet]. [updated 2026 May 27; cited 2026 May 30]. . Available from: https://medicarestatistics.humanservices.gov.au/statistics/mbs_item.html
10. Department of Health and Aged Care. Medicare Benefits Schedule Book – Category 8. Operating from 1 July 2023. Canberra: Australian Government; 2023. Available from: http://www6.health.gov.au/internet/mbsonline/publishing.nsf/Content/4C767FC45843CB02CA25898500825ACA/$File/PDF%20version%20of%20the%20July%202023%20Category%208%20-%20Miscellaneous%20Services.pdf

1. Australian Institute of Health and Welfare. Mental Health[internet]. Canberra: AIHW[cited 2026 May 29]. Available from: https://www.aihw.gov.au/mental-health

2. Productivity Commission 2020. Mental Health, Report no. 95, Canberra: Australian Government; 2020. Available from: https://www.pc.gov.au/inquiries-and-research/mental-health/report/

3. Australian Government Department of Health, Disability, and Ageing. Better Access Initiative. Canberra; 2023. Available from: https://www.health.gov.au/our-work/better-access-initiative

4. Duckett S, Breadon P. Access all areas: new solutions for GP shortages in rural Australia. Grattan Institute; 2013. Available from: https://grattan.edu.au/report/access-all-areas-new-solutions-for-gp-shortages-in-rural-australia/

5. Australian Government. MBS Telehealth Services. Department of Health; 2022. Available from: https://www.mbsonline.gov.au/internet/mbsonline/publishing.nsf/Content/Factsheet-Telehealth-Arrangements-Jan22
6. Snoswell CL, Arnautovska U, Haydon HM, Siskind D, Smith AC. Increase in telemental health services on the Medicare Benefits Schedule after the start of the coronavirus pandemic: data from 2019 to 2021. Aust Health Rev. 2022;46(5):544‑549. Available from: https://doi.org/10.1071/AH22078

7. Australian Bureau of Statistics. Patient experiences [internet]. Canberra: 2025[updated 2025 Nov 18]. Available from: https://www.abs.gov.au/statistics/health/health-services/patient-experiences/latest-release

8. Backhaus A, Agha Z, Maglione ML, Repp A, Ross B, Zuest D, et al. Videoconferencing psychotherapy: a systematic review. Psychol Serv. 2012;9(2):111–131. Available from: https://europepmc.org/article/MED/22662727

9. Olthuis JV, Watt MC, Bailey K, Hayden JA, Stewart SH. Therapist-supported internet cognitive behavioural therapy for anxiety disorders in adults. Cochrane Database Syst Rev. 2016;3(3):CD011565. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC7077612/
10. Mihalopoulos C, Vos T, Pirkis J, Smit F, Carter R. Do Indicated Preventive Interventions for Depression Represent Good Value for Money? Aust N Z J Psychiatry. 2011;45(1):36-44. doi:10.3109/00048674.2010.501024
11. Husereau D, Drummond M, Augustovski F, de Bekker‑Grob E, Briggs AH, Carswell C, et al. Consolidated Health Economic Evaluation Reporting Standards 2022 (CHEERS 2022) statement: updated reporting guidance for health economic evaluations. BMJ. 2022;376:e067975. Available from: https://doi.org/10.1136/bmj-2021-067975
12. Spitzer RL, Kroenke K, Williams JBW, Löwe B. A brief measure for assessing generalized anxiety disorder: the GAD‑7. Arch Intern Med. 2006;166(10):1092‑1097. Available from: https://doi.org/10.1001/archinte.166.10.1092

13. Le LK-D, Esturas AC, Mihalopoulos C, Chiotelis O, Bucholc J, Chatterton ML, et al. Cost-effectiveness evidence of mental health prevention and promotion interventions: A systematic review of economic evaluations. PLOS Medicine. 2021;18:e1003606. Available from: https://doi.org/10.1371/journal.pmed.1003606

14. Richards DA, Bower P, Pagel C, Weaver A, Utley M, Cape J, et al. Delivering stepped care: an analysis of implementation in routine practice. Implement Sci. 2012:3. doi: 10.1186/1748-5908-7-3.

15. Krzyzaniak N, Greenwood H, Scott AM, Peiris R, Cardona M, Clark J, et al. The effectiveness of telehealth versus face-to face interventions for anxiety disorders: A systematic review and meta-analysis. J Telemed Telecare. 2024;30(2):250-261. doi: 10.1177/1357633X211053738
16. Stratton E, Lampit A, Choi I, Calvo RA, Harvey SB, Glozier N. Effectiveness of eHealth interventions for reducing mental health conditions in employees: A systematic review and meta-analysis. PLoS ONE, 12(12):e0189904. Available from: https://doi.org/10.1371/journal.pone.0189904
17. Ngo PJ, Wade S, Banks E, Karikios DJ, Canfell K, Weber MF. Large-Scale Population-Based Surveys Linked to Administrative Health Databases as a Source of Data on Health Utilities in Australia. Value Health. 2022 Sep;25(9):1634-1643. doi: 10.1016/j.jval.2022.03.026.
18. Vos T, Haby MM, Barendregt JJ, Kruijshaar M, Corry J, Andrews G. The burden of major depression avoidable by longer-term treatment strategies. J Affect Disord. 2004;79(1–3):263–272. Available from: https://doi.org/10.1001/archpsyc.61.11.1097
19. Pharmaceutical Benefits Advisory Committee (PBAC). Guidelines for Preparing Submissions to the PBAC. Version 5.0. Canberra: Department of Health; 2016. Available from: https://pbac.pbs.gov.au/content/information/files/pbac-guidelines-version-5.pdf
20. Clark DM, Canvin L, Green J, Layard R, Pilling S, Janecka M. Transparency about the outcomes of mental health services (IAPT approach): an analysis of public data. Lancet. 2018;391(10121):679‑686. Available from: https://doi.org/10.1016/S0140-6736(17)32133-5

Strengths and Limitations A key strength of this study is its policy relevance, employing real-world Australian Medicare data and simulating outcomes using a state-transition Markov model validated through internal and external processes. By including both healthcare and societal perspectives, the study provides a more comprehensive understanding of the value of telehealth. We also conducted extensive sensitivity analyses, confirming the robustness of results under wide-ranging assumptions, including variations in productivity costs, utility weights, and uptake rates. However, several limitations must be noted. First, the model assumed clinical equivalence between telehealth and in-person care, supported by evidence but not tested within this model [8, 9, 33]. Second, the transition probabilities were informed by literature and by assumptions due to the limited availability of Australian longitudinal data [26, 27]. Third, indirect costs (e.g., productivity loss) were estimated, not directly measured, although we applied conservative assumptions supported by ABS data [21]. Fourth, the model used average session costs, without stratifying by condition severity or provider practice variation, which may limit generalisability to specific populations. Uncertainty and Model Validation The probabilistic sensitivity analysis (PSA) revealed over 99% probability that telehealth is cost-effective at a willingness-to-pay (WTP) threshold of AU $50,000 per QALY, consistent with Australian health economic thresholds. The tornado diagram demonstrated that the model was most sensitive to changes in utility values and productivity costs. Nonetheless, in all tested scenarios, telehealth remained below WTP thresholds, affirming its robust cost-effectiveness. The model was internally validated through formula auditing and independently reviewed for logical consistency. Policy and Practice Implications This study provides strong economic justification for retaining and expanding telehealth for mental health care within Australia’s Medicare framework. The relatively low budget impact (AU $2.6 per member per month) and superior cost-effectiveness under a societal lens suggest that telehealth is not only financially sustainable but also equity-promoting, especially for rural or mobility-impaired populations. Policymakers should consider these findings when designing long-term digital health strategies, reimbursement models, and service delivery frameworks. Future Research Further research is needed to refine estimates of real-world telehealth uptake, patient preferences, and long-term outcomes across subpopulations, including culturally and linguistically diverse communities. Studies should also explore integration with digital mental health tools, such as blended care models and asynchronous platforms, and examine outcomes in underserved regions [8,16, 31-33]. Furthermore, future studies should focus on the exploration of integrated digital mental health tools, such as blended care models and asynchronous platforms, examining outcomes in underserved regions. Future work should also assess differential treatment response and subgroup outcomes to ensure digital transformation benefits all populations. Conclusion Telehealth for psychological services is a cost-minimising, clinically effective, and fiscally responsible strategy. From a societal perspective, telehealth provides substantial public health value. These findings support its continued funding and expansion within the Australian mental healthcare system.  

Discussion Principal Findings This economic evaluation examined the cost-effectiveness and budget impact of telehealth-delivered psychological services compared to in-person services and no treatment for adults with mental health conditions in Australia. Using a five-year Markov model, we found that both telehealth and in-person psychological services resulted in substantial QALY gains over no treatment. However, from a societal perspective, telehealth was the dominant strategy, delivering equivalent health outcomes at significantly lower total cost due to reduced productivity losses. These findings suggest that continued investment in telehealth services for mental health is an efficient use of healthcare resources and supports its long-term integration into national policy.

Interpretation of Findings in Context Our results are consistent with previous literature showing that evidence-based psychological therapies are cost-effective interventions for depression and anxiety [10, 13,18]. Importantly, our study extends this evidence by evaluating the cost and outcome implications of different modes of delivery rather than only comparing treatment to non-treatment. Prior studies confirm that telehealth is clinically equivalent to in-person care in terms of symptom improvement, therapeutic alliance, and patient satisfaction [8,9,16,27,31-34], a key assumption underpinning our cost-minimisation analysis. From a healthcare payer perspective, the cost of providing telehealth was comparable to in-person services, given identical MBS rebates per session. From a societal perspective, however, telehealth yielded substantial cost savings due to reduced indirect costs, principally productivity losses from travel and time off work. This aligns with findings from Snoswell et al [35], who observed that telehealth modalities reduced economic barriers to access and improved service efficiency.

Key Findings Telehealth was cost-effective relative to no treatment, with an ICER of AU $4, 640/QALY, well below the Australian WTP threshold. This was significantly below the commonly applied Australian willingness-to-pay threshold of AU$50,000 per QALY. Since telehealth is equivalent to in-person services, comparisons were made between both interventions, using cost-minimisation analysis Comparison with Current Standard of Care Assuming cost parity per consultation (AU $166.85), the only incremental cost introduced by telehealth is support-related. Thus, the model isolates the marginal financial burden attributable to telehealth as administrative overhead. Effectiveness Comparison Evidence from literature showed that mental health outcomes delivered via telehealth were comparable to in-person care, particularly when digital services are well-integrated [8, 9, 16, 29, 30].

Figure 1: Tornado diagram highlighting the influence of key parameters on cost-effectiveness outcomes

Figure 2: Cost-effectiveness acceptability curves (CEAC)

Budget Impact Analysis (BIA) A national budget impact analysis (BIA) showed that the 5-year projected cost to the Australian government for supporting telehealth services (including MBS rebates and digital infrastructure) was about AU $37.1 million. On a per capita basis, the incremental government cost of national telehealth implementation was AU $2.6 per member per month (PMPM), AU $31 per member per year (PMPY).<br /> Sensitivity analyses testing alternative rates of uptake (1.0–5.0% growth), telehealth support costs variation, and population growth, showed the total costs possibly ranging from AU$35.5 million to AU$43.1 million by the fifth year, indicating moderate variation.

Societal Perspective With productivity losses estimated, telehealth was less costly than in-person services. While both interventions had equivalent QALYs, telehealth had substantially lower productivity costs (AU$494 per patient versus AU$3,954 per patient over 5 years). Compared with no treatment, telehealth remained highly cost-effective, with an ICER of about AU$4,640 per QALY gained.

Sensitivity Analyses Deterministic Sensitivity Analysis (DSA) The DSA results were most sensitive to: i) The QALY gain per cycle (driven by time spent in high-utility states); and ii) The indirect cost per session, particularly for in-person services. A tornado diagram (Figure 1) illustrates the range of ICERs for telehealth vs no treatment across one-way sensitivity inputs. Even under conservative assumptions such as higher productivity losses, telehealth remained cost-effective below AU $50,000/QALY.

Probabilistic Sensitivity Analysis (PSA) With 10,000 Monte Carlo simulations (probabilistic sensitivity analysis), reflecting parameter uncertainty, a mean ICER of about AU $4,640 per QALY was obtained for telehealth compared to no treatment. The cost-effectiveness acceptability curve (CEAC) showed that telehealth was cost-effective in 99.85% of simulations from the healthcare payer perspective. It was also cost-effective in 99.92% from the productivity losses (societal perspective), at a willingness-to-pay threshold of AU $50,000/QALY (Figure 2). Framing the comparison between telehealth and in-person services as a cost-minimisation analysis, telehealth had lower productivity costs (30 minutes per session loss vs 4 hours for in-person services) – Appendix 1.  

Table 4: Base-Case Cost-Effectiveness Results, Per Patient Strategy QALYs per patient Healthcare Cost per patient (AU$) Societal Cost per patient (AU$) Incremental QALYs Incremental Cost (AU$) ICER (AU$/QALY) No treatment 1.94 91.97 0 – – – Telehealth 2.7 3,757.41 494.27 0.79 3,665.43 4,640 In-person 2.7 3,641.02 3,954.17 0.79 3,549.05 4,492  

Table 2: Probabilistic Sensitivity Analysis Summary Perspective Mean Incremental QALY Mean Incremental Cost (AU$) Probability Cost-Effective (AU$50,000/QALY) Telehealth – healthcare 0.79 3,665 0.916 Telehealth – productivity-only (societal) 0.79 494 0.916 In-person – healthcare 0.79 3,549 0.916 In-person – productivity-only (societal) 0.79 3,954 0.916  

Table 3: Descriptive statistics for selected clinical psychology services and related costs per 100,000

Statistic Telehealth (video and phone) services unrestricted to any geographical area Telehealth (video) services for rural, remote, and regional areas only Combined telehealth (unrestricted and restricted) services In-person services for all areas Services per 100,000 population Mean (SD) 733 (251) 37 (6) 769 (251) 1586 (246) Min 514 28 545 1299 Median (Q1, Q3) 690 (551, 783) 37 (32, 41) 756 (676, 827) 1596 (1346, 1741) Max 1251 44 1289 2011 Costs per 100,000 population Mean (SD) 26,650 (24,965) 408 (282) 30,456 (28,254)<br /> Min 107 41 178<br /> Median (Q1, Q3) 20,239 (5,476, 43,360) 349 (184, 603) 23,547 (7,192, 47,882) <br /> Max 111,392 975 122,430<br />

Cost-Effectiveness Analysis Model Cohort and Base-Case Estimates The model simulated a cohort of 100,000 adults aged 15–64 years presenting with psychological symptoms over 5 years, comparing three strategies: i) No treatment; ii) Telehealth psychological services (video and telephone), and iii) In-person services delivered by clinical psychologists. receiving mental health services. Health outcomes were expressed in quality-adjusted life years (QALYs), and costs were presented from both the healthcare payer and societal perspectives. All results were discounted at 5% annually, and costs were expressed in 2023 AUD.   Total Costs and Health Outcomes Healthcare Payer Perspective Under the healthcare payer perspective (including MBS-funded session costs only), both telehealth and in-person psychological services produced equivalent QALYs per patient (2.70 over a period of 5 years) in the 100,000 adults cohort, using the state-based utility monthly accrual over the period (Table 4). The incremental cost-effectiveness ratio (ICER) of telehealth vs no treatment was AU$ 4,640 per QALY. Telehealth and in-person services produced equal QALYs. Hence, cost-minimisation analysis (CMA) was used for their comparison. In-person care incurred no additional healthcare costs; thus, both strategies were cost-equivalent from the payer perspective.

Descriptive analysis Our study analysed data of service use per 100,000 collected over nine quarters (April 2020-June 2022), 27 months. Telehealth services unrestricted to any geographical area had a mean uptake of 733 (SD 251) services per 100,000 per quarter. On the other hand, telehealth services limited to regional, rural, and remote areas showed a much lower mean usage of 37 (6) services per quarter. The combined telehealth (restricted and unrestricted) services had a median (IQR) value of 756 (676, 827 (Table 3). The median (Q1, Q3) costs per 100,000 of unrestricted telehealth services were AUD 20,239 (5,476, 43,360), vastly exceeding that of restricted rural services, AUD 349 (184, 603).

Cost Reporting We reported the results using the following measures: a) Per Member Per Month (PMPM); ii) Per Member Per Year (PMPY); and iii) Per Treated Member Per Month (PTMPM). An initial modelled population of 32,669 people per year was assumed, with annual population growth of 1.9%. The analysis was conducted from a public healthcare payer perspective, which assumes 85% coverage for clinical psychology services under Medicare [19]. All analyses were conducted in Python 3.11 within the Google Colab environment

Ethics Approval This study was approved by the Human Research Ethics Committee (HREC) of the University of Southern Queensland (UniSQ), Toowoomba (Approval Number: ETH2023-0357). The analysis was informed by publicly available, de-identified data from the Medicare Benefits Schedule (MBS). As such, individual informed consent and participant compensation were not required.  

Budget Impact Analysis A budget impact analysis (BIA) was conducted to estimate the financial implications of introducing telehealth services for mental health in Australia. The analysis was performed from the perspective of the Australian healthcare system. In Australia, 43% of persons aged 16–85 years have reported at least a lifetime history of a mental health disorder, with 1 in 5 persons having experienced at least one mental health condition in the past 12 months [28]. The most common mental health conditions are anxiety disorders (17%), affective disorders including depression (7.5%), and substance use disorders (3.3%) [28]. The target population for this study includes individuals with mental health conditions who have accessed clinical psychology services and could benefit from telehealth services. The demographics of the population in this study include people in the active-service age (15–64 years), and males and females who accessed clinical psychology services during the index period (April 2020 to June 2022). The BIA was conducted over a 5-year time horizon to capture both short-term and long-term budget impacts.

Intervention In addition to the video consultation services for clinical psychology restricted to regional, remote, and rural areas in Australia (MBS 80011), video conferencing (MBS 91167) and telephone (MBS 91182) services were introduced in March 2020 at the onset of the COVID-19 pandemic in Australia. These services were not geographically restricted and were billed similarly to existing in-person clinical psychology services [5,29,30].

Comparator The comparators in this study were the existing in-person clinical psychology services without telehealth. Data Sources We used usage data for the relevant MBS services to estimate the effectiveness and utilisation of telehealth services. Cost data for clinical psychology services were retrieved from the Services Australia MBS website [30]. A population base of 32,669 individuals per year was assumed, based on unrestricted telehealth usage data (MBS items 91167 and 91182) (Appendix 3).

Outcomes The primary outcome is the budget impact of introducing telehealth services, measured in terms of total costs to the healthcare system. Sensitivity Analyses Sensitivity analyses were conducted to assess the robustness of the results to changes in key parameters, including: i) ±10% variation in uptake rate; ii) ±20% variation in support costs; and iii) ±1 percentage point change in population growth. Validation The model was validated both internally (through consistency checks) and externally (via comparison with published studies and government-reported service volumes and cost estimates).

Sensitivity Analysis Deterministic Sensitivity Analysis (DSA) A one-way sensitivity analysis varied key parameters individually within plausible bounds: i) Utilities - ±10%; ii) Productivity loss (in-person) – 3 to 5 hours; iii) Productivity loss (telehealth) – 0 to 1 hour; iv) Service cost - ±20%; v) Discount rate – 3 to 5%; and vi) Baseline utility for no-treatment - 0.30 to 0.50. A tornado diagram was constructed to visualise the impact of each parameter on the ICER for telehealth versus no treatment (Figure 1).

Probabilistic Sensitivity Analysis (PSA) A probabilistic sensitivity analysis using 10,000 Monte Carlo simulations was performed to account for parameter uncertainty (Table 2). The following distributions were applied: i) Beta distributions for probabilities; ii) Gamma distributions for costs; and iii) Normal distributions for utility values. Results were summarised using a cost-effectiveness acceptability curve (CEAC), which displays the probability that each strategy is cost-effective across a range of willingness-to-pay thresholds. A five-year budget impact scenario analysis was also conducted (Appendix 4).

Cost-Effectiveness Analysis Setting This study employed a model-based economic evaluation to assess the cost-effectiveness and budget impact of telehealth-delivered psychological services for adults with mental health conditions in Australia. The intervention was evaluated from both the healthcare payer and societal perspectives, using a Markov cohort model over a 5-year time horizon. Reporting of the study adheres to the Consolidated Health Economic Evaluation Reporting Standards 2022 (CHEERS 2022) [11].

Target Population and Setting The target population consisted of adults aged 15–64 years accessing psychological services in Australia between April 2020 and June 2022. This age range was selected to align with the working-age population for whom productivity costs are most relevant. Individuals aged under 15 and over 65 were excluded to minimise bias when estimating indirect societal costs, such as lost wages. Model Structure A state-transition (Markov) model was constructed to simulate the progression of psychological symptom severity across five mutually exclusive health states: i) Minimal symptoms; ii) Mild symptoms; iii) Moderate symptoms; iv) Severe symptoms; and v) Death (Table 1). These states were informed by validated severity criteria from the GAD-7 scale and aligned with classifications used in primary mental healthcare settings [12]. Transitions between states were modelled on a monthly cycle for 60 months (5 years), capturing medium-term clinical dynamics and health system implications of psychological treatment [10,13]. The model was designed to reflect the natural course of mental health symptoms in the presence or absence of psychological therapy. Transitions were influenced by literature [14]. The model cohort size was set at 100,000 individuals, which allowed for stable estimates across transition states and scalable budget impact analysis (Appendix 1). There is a paucity of studies on longitudinal data on mental health showing monthly transitions between mental health severity states. Hence, we used transition probabilities assumption informed by evidence on response to treatment, deterioration, and maintenance effects. Identical probabilities of transition were used for telehealth and in-person mental health services, in keeping with the assumption of clinical equivalence (Table 1).   Table 1: Markov Model (Assumed) Transition Probability Matrix for Psychological Health States

From \ To Transition to Severe Transition to Moderate Transition to Mild Transition to Minimal Transition to Death Treatment arm Severe 0.92 0.03 0.01 0.01 0.03 Moderate 0.04 0.91 0.02 0.02 0.01 Mild 0.02 0.03 0.91 0.04 0 Minimal 0 0.01 0.02 0.97 0 Death 0 0 0 0 1 No-treatment arm Severe 0.95 0.01 0 0 0.04 Moderate 0.03 0.93 0.02 0.01 0.01 Mild 0.01 0.04 0.92 0.03 0 Minimal 0 0.02 0.04 0.94 0 Death 0 0 0 0 1  

Comparators and Intervention Strategies The model compared three service delivery strategies: i) No treatment (comparator) – Reflects individuals with psychological symptoms who are unable to access formal therapy due to geographic, financial, or systemic barriers. ii) Telehealth services – Includes unrestricted access to video (MBS 91167) and telephone (MBS 91182) consultations delivered by clinical psychologists, introduced under temporary Medicare reforms in March 2020. These also include clinical psychology video services restricted to regional, remote, and rural areas in Australia (MBS 80011). iii) In-person services – Traditional face-to-face sessions (MBS 80010) delivered at clinics by clinical psychologists.

Clinical Effectiveness Assumptions For both telehealth and in-person services, the model assumed equal clinical effectiveness in terms of symptom improvement and health-state transitions [8,9]. This assumption was based on multiple lines of evidence: i) Meta-analyses and systematic reviews indicate that telehealth-delivered psychological therapies are non-inferior to in-person care in terms of symptom reduction, patient satisfaction, and therapeutic alliance [9,15,16]; and ii) Studies have confirmed equivalence in outcomes with patient and clinician acceptance [9]. The interventions evaluated (e.g., cognitive behavioural therapy) are transdiagnostic and applicable to a range of mental health conditions, further supporting the assumption of modality equivalence. As a result, the comparison between telehealth and in-person care was evaluated using a cost-minimisation analysis (CMA). In contrast, incremental cost-effectiveness ratios (ICERs) were calculated only when comparing each treatment strategy against no treatment, consistent with health economic best practice [10, 14].

Health Outcomes and Utilities Health outcomes were expressed as quality-adjusted life years (QALYs), derived from utility values assigned to each health state and accumulated monthly. Informed by literature , utility values were assigned to the health states: i) Minimal: 0.9; ii) Mild: 0.7; iii) Moderate: 0.6; iv) Severe: 0.6; v) Death: 0.0. Utility accrual was adjusted using a 5% annual discount rate, applied monthly (0.417%), consistent with PBAC guidelines [17-19]. QALYs are often assigned per year. Hence, a division by 12 was applied to account for the monthly cycle used in this model.

Transition Probabilities A previous study conducted provided information on the transition of health states [20]. However, annual cycles were used in the study. In the present study, monthly transition probabilities between severity states were assumptions informed by literature [20] The transition probability matrix was developed to represent plausible movement among minimal, mild, moderate, and severe symptom states. There were two arms of the transition matrix, the treatment arm (telehealth and in-person psychological services) and the no-treatment arm (Table 1). While the original data on the number of consultations in the index period were for mental health disorders generally, anxiety and depression represent most presentations for Medicare-subsidised psychological services in Australia. Moreover, the treatments under evaluation are generally transdiagnostic in scope. Hence, the transition probabilities were judged applicable to the broader population accessing clinical psychology services under the MBS.   Costing Approach Direct Costs Direct healthcare costs included the standard MBS rebate for a clinical psychologist consultation lasting ≥50 minutes, fixed at AU $166.85 per session across all modalities (telehealth and in-person). These costs were applied per cycle based on average service use and scaled to the cohort size. All costs were expressed in 2023 Australian dollars and discounted at 5% per annum. Indirect Costs (Societal Perspective) Indirect productivity costs, regarded as productivity losses, were incorporated to reflect time lost from work.Initially, the model assumed that: i) In-person sessions incurred 4 hours of productivity loss per session, based on average hourly wages (AU $45.30/hour), yielding AU $181.20 per session; and ii) Telehealth sessions were initially assumed to incur no indirect cost. To improve balance and realism, the base-case model was revised to apply a 30-minute productivity loss for telehealth (AU $22.65), assuming minor workplace disruption or setup time [21].

Model Validation We subjected the model to both internal and external validation processes in accordance with best practice for health economic modelling. Internal Validation Internal consistency checks were performed to ensure that: i) Transition probabilities between health states were correctly applied and summed to one in each cycle; ii) QALY calculations were derived from the correct application of utility weights per health state and per cycle; iii) The time horizon (60 monthly cycles) and discounting at 5% annually (0.417% monthly) were implemented accurately; iii) Session-based utility accrual was avoided; all utilities were applied state-based and aggregated per cohort; iv) Cost and QALY estimates scaled appropriately for the modelled population (n = 100,000) and were internally consistent across scenarios.

External Validation The model structure, health states, and transition assumptions were informed by and aligned with literature [9,17,19,20,22]. Health-state definitions and utility values were cross-validated against Australian burden of disease data and prior economic evaluations of psychological therapies [17,19,23,24]. Transition probabilities were informed by literature [20,25-27]. Cost inputs (e.g. MBS item fees, wage rates) were estimated using publicly available sources from the Australian Department of Health and the Australian Bureau of Statistics [21,28-30]. Furthermore, the assumption of clinical equivalence between telehealth and in-person psychological care was supported by meta-analyses and systematic reviews [15,16]. Model outputs (total QALYs, costs per person) were benchmarked against plausible ranges from prior mental health economic evaluations in Australia. These comparisons showed that the outputs were within expected boundaries and consistent with published estimates.

Study Design and Setting This study employed a retrospective design to examine the cost-effectiveness of the expansion of telehealth services provided by clinical psychologists in Australia. Data were collected from Medicare Benefits Schedule (MBS) claims from Quarter 2 (April to June) 2020 to Quarter 2 (April to June) 2022. This period was selected as the most significant time when the COVID-19 pandemic affected the population's health and health care services. Data Sources Monthly consultation data obtained from Services Australia (selecting MBS items related to clinical psychology services) were used. The study used datasets related to services per 100,000 population, lasting at least 50 minutes, and the associated costs (Medicare benefits paid). The list of datasets included the services and related costs per 100,000 for the following: i) In-person services (80010); ii) Video services (80011) restricted to eligible areas in rural, remote, and very remote areas (MMM 4–7); and iii) Video (91167) and phone (91182) services not restricted to any geographical area. Datasets for 80010, 80011, 91167, and 91182 were obtained from the Services Australia website. Datasets for MBS items 91167 and 91182 were aggregated and analysed as a single variable, as both were telehealth items used in the same period in Australia. Analytical Framework The analytical framework for this study comprises: i) Descriptive analysis of the selected services; ii) Cost-effectiveness analysis using Markov simulation modelling; and iii) Budget impact analysis. Descriptive Analysis The study described the datasets in terms of minimum and maximum values, means with corresponding standard deviations, and medians with Q1 and Q3 values.

Study Design and Setting This study employed a retrospective design to examine the cost-effectiveness of the expansion of telehealth services provided by clinical psychologists in Australia. Data were collected from Medicare Benefits Schedule (MBS) claims from Quarter 2 (April to June) 2020 to Quarter 2 (April to June) 2022. This period was selected as the most significant time when the COVID-19 pandemic affected the population's health and health care services. Data Sources Monthly consultation data obtained from Services Australia (selecting MBS items related to clinical psychology services) were used. The study used datasets related to services per 100,000 population, lasting at least 50 minutes, and the associated costs (Medicare benefits paid). The list of datasets included the services and related costs per 100,000 for the following: i) In-person services (80010); ii) Video services (80011) restricted to eligible areas in rural, remote, and very remote areas (MMM 4–7); and iii) Video (91167) and phone (91182) services not restricted to any geographical area. Datasets for 80010, 80011, 91167, and 91182 were obtained from the Services Australia website. Datasets for MBS items 91167 and 91182 were aggregated and analysed as a single variable, as both were telehealth items used in the same period in Australia.

Moreover, there is limited modelling of service delivery reforms using decision-analytic frameworks that incorporate real-world data, productivity losses, and long-term outcomes.

This study addresses this evidence gap by evaluating the cost-effectiveness and budget impact of telehealth-delivered psychological services for common mental health disorders in Australia. We developed a Markov model to simulate outcomes under three strategies: (i) no treatment, (ii) telehealth, and (iii) in-person care. We assessed results from both a healthcare payer and societal perspective over 5 years, using Australian epidemiological, economic, and Medicare service data. By providing robust, policy-relevant estimates of cost and value, this study aims to inform Medicare funding decisions, promote efficient service delivery, and support equitable access to mental health care in the post-pandemic era.

Mental disorders, particularly depression and anxiety, are leading contributors to the global burden of disease, affecting one in five Australians annually and costing the economy over AUD 70 billion each year in health services, productivity losses, and social impacts [1,2]. Evidence-based psychological therapies such as cognitive behavioural therapy (CBT) are effective in treating common mental health conditions and are publicly subsidised through Australia’s Medicare Benefits Schedule (MBS) [3]. However, longstanding barriers, especially geographic and socioeconomic, have limited timely access to care, particularly in rural and underserved populations [4]. The COVID-19 pandemic catalysed a rapid expansion of telehealth services, including MBS-subsidised psychological therapy via video and phone consultations, with policies implemented in March 2020 and made permanent from 2022 [2,5,6]. This shift has substantially changed the delivery landscape, with over 40% of mental health consultations now occurring remotely [7]. While studies have shown that telehealth achieves clinical outcomes comparable to in-person care [8,9], the economic implications of this policy change, including costs, quality-adjusted life years (QALYs), and value for money, remain under-evaluated [2]. A previous cost-effectiveness analysis (CEA) of psychological therapies in Australia has demonstrated favourable results across a range of population groups [10]. However, few studies have specifically examined the comparative economic value of telehealth versus in-person psychological services, nor considered both healthcare and societal perspectives.

Mental disorders, particularly depression and anxiety, are leading contributors to the global burden of disease, affecting one in five Australians annually and costing the economy over AUD 70 billion each year in health services, productivity losses, and social impacts [1,2]. Evidence-based psychological therapies such as cognitive behavioural therapy (CBT) are effective in treating common mental health conditions and are publicly subsidised through Australia’s Medicare Benefits Schedule (MBS) [3]. However, longstanding barriers, especially geographic and socioeconomic, have limited timely access to care, particularly in rural and underserved populations [4]. The COVID-19 pandemic catalysed a rapid expansion of telehealth services, including MBS-subsidised psychological therapy via video and phone consultations, with policies implemented in March 2020 and made permanent from 2022 [2,5,6]. This shift has substantially changed the delivery landscape, with over 40% of mental health consultations now occurring remotely [7]. While studies have shown that telehealth achieves clinical outcomes comparable to in-person care [8,9], the economic implications of this policy change, including costs, quality-adjusted life years (QALYs), and value for money, remain under-evaluated [2]. A previous cost-effectiveness analysis (CEA) of psychological therapies in Australia has demonstrated favourable results across a range of population groups [10]. However, few studies have specifically examined the comparative economic value of telehealth versus in-person psychological services, nor considered both healthcare and societal perspectives. Moreover, there is limited modelling of service delivery reforms using decision-analytic frameworks that incorporate real-world data, productivity losses, and long-term outcomes.

Background: Mental health disorders, particularly anxiety, constitute a significant burden in Australia, affecting 1 in 5 individuals annually. While telehealth has emerged as a strategic solution to expand access, evidence on its economic impact within the Australian context remains limited. Objective: This study evaluates the cost-effectiveness and budget impact of telehealth-delivered psychological services by clinical psychologists compared to in-person care and no treatment among adults with mental health disorders in Australia. Methods: A retrospective analysis was conducted using Medicare Benefits Schedule data from April 2020 to June 2022. A Markov cohort model simulated health transitions over a five-year horizon, incorporating healthcare payer and societal perspectives. Health outcomes were measured in quality-adjusted life years (QALYs). Results: Telehealth services for mental health were found to be cost-effective compared with no treatment, producing an estimated ICER of AUD $4,640 per QALY gained, reflecting an acceptable value below the commonly applied Australian willingness-to-pay threshold of AUD $50,000 per QALY. While telehealth and in-person services may produce equivalent health outcomes, telehealth was associated with substantially lower productivity losses. The budget impact analysis showed that the implementation of telehealth services may be financially sustainable, with annual expenditure projected to increase from about AUD $34.4 million in the first year to about AUD $37.1 million in the fifth year 5 under the base-case scenario. Conclusions: Telehealth for mental health is highly cost-effective in Australia and is associated with lower productivity losses than in-person care while achieving equivalent health outcomes. These findings support the continued funding and integration of telehealth into national mental health policy to improve access and equity.

Telehealth for mental health is highly cost-effective in Australia and is associated with lower productivity losses than in-person care while achieving equivalent health outcomes. These findings support the continued funding and integration of telehealth into national mental health policy to improve access and equity.

Telehealth services for mental health were found to be cost-effective compared with no treatment, producing an estimated ICER of AUD $4,640 per QALY gained, reflecting an acceptable value below the commonly applied Australian willingness-to-pay threshold of AUD $50,000 per QALY. While telehealth and in-person services may produce equivalent health outcomes, telehealth was associated with substantially lower productivity losses. The budget impact analysis showed that the implementation of telehealth services may be financially sustainable, with annual expenditure projected to increase from about AUD $34.4 million in the first year to about AUD $37.1 million in the fifth year 5 under the base-case scenario.

Author Update (June 2026)

Following a post-publication audit of this preprint, a number of references and supporting citations were identified as inaccurate. These issues arose during manuscript preparation and reference verification processes applied to this early draft. The affected citations were subsequently reviewed against original sources and corrected in later manuscript revisions. Furthermore, several sections of the manuscript were also revised significantly. Readers are advised to consult the annotations throughout this preprint for clarification regarding affected references.