Improving access to mental health care: a system dynamics model of direct access to specialist care and accelerated specialist service capacity growth

Abstract Objective To simulate the impact on population mental health indicators of allowing people to book some Medicare‐subsidised sessions with psychologists and other mental health care professionals without a referral (direct access), and of increasing the annual growth rate in specialist mental health care capacity (consultations). Design System dynamics model, calibrated using historical time series data from the Australian Bureau of Statistics, HealthStats NSW, the Australian Institute of Health and Welfare, and the Australian Early Development Census. Parameter values that could not be derived from these sources were estimated by constrained optimisation. Setting New South Wales, 1 September 2021 – 1 September 2028. Main outcome measures Projected mental health‐related emergency department presentations, hospitalisations following self‐harm, and deaths by suicide, both overall and for people aged 15–24 years. Results Direct access (for 10–50% of people requiring specialist mental health care) would lead to increases in the numbers of mental health‐related emergency department presentations (0.33–1.68% of baseline), hospitalisations with self‐harm (0.16–0.77%), and deaths by suicide (0.19–0.90%), as waiting times for consultations would increase, leading to disengagement and consequently to increases in adverse outcomes. Increasing the annual rate of growth of mental health service capacity (two‐ to fivefold) would reduce the frequency of all three outcomes; combining direct access to a proportion of services with increased growth in capacity achieved substantially greater gains than an increase in service capacity alone. A fivefold increase in the annual service growth rate would increase capacity by 71.6% by the end of 2028, compared with current projections; combined with direct access to 50% of mental health consultations, 26 616 emergency department presentations (3.6%), 1199 hospitalisations following self‐harm (1.9%), and 158 deaths by suicide (2.1%) could be averted. Conclusion The optimal combination of increased service capacity growth (fivefold) and direct access (50% of consultations) would have double the impact over seven years of accelerated capacity growth alone. Our model highlights the risks of implementing individual reforms without knowledge of their overall system effect.


Figure 1. Overview of the causal structure of the system dynamics model
The bolded arrows represent the main causal pathways involved in the tested scenarios.The plus and minus signs indicate the effect.COVID-19 temporarily decreases migration, employment, and access to services, while increasing psychological distress.Developmental vulnerability and unemployment increase psychological distress, which increases demand for mental health services.Mental health services decrease suicidal behaviour and psychological distress, except if people disengage from services, in which case the psychological distress increases.Psychological distress increases suicidal behaviour, which in turn increase demand for services.
The model was calibrated using historical, time-series data from the Australian Bureau of Statistics (population sizes, birth and mortality rates, overseas and internal migration rates (1), employment rates and transitions (2, 3), education and non-school qualification data (4), family characteristics (5), and the impact of the pandemic on psychological distress prevalence (6)), HealthStats NSW (psychological distress prevalence (7), intentional self-harm hospitalizations (8) and suicide deaths (9)), the Australian Institute of Health and Welfare (AIHW) (mental-health-related ED presentations (10), and mental health service usage statistics (11)), and the Australian Early Development Census (childhood developmental vulnerability prevalence data (12)).Parameters that could not be derived from available data were estimated by constrained optimisation.

b) Population sector
The population sector (figure 2) models the changes in population sizes across different age groups as a result of births, migration, aging, and mortality.The total national population is represented as 5 stocks (state variables), corresponding to numbers of people aged 0−14 years, 15−24 years, 25−44 years, 45−64 years, and 65 years and above.The size of the population increases via births (which flow into the stock of 0−14-yearolds) and immigration and decreases through emigration and mortality.Aging is modelled as a first-order delay, in which people flow out of each stock (except the stock of people aged ≥ 65 years) at a rate n⁄d, where n is the number of people in the stock at any particular time point and the delay time d is the mean number of years a person spends in the stock.Births occur at the rate bP, where b is the per capita birth rate and P the total population.Deaths occur at the rate M i P i , where M i and P i are respectively the mortality hazard ratio and the population for age group i.The per capita birth rate and per capita mortality rate for the total population m decline at constant fractional rates per year.Net migration for age group i is equal to I i _-e i P i , where I i is agespecific immigration per year and e i is the age-specific per capita emigration rate per year.Figure 3 shows the structure of the psychological distress sector, which models flows between states of low psychological distress (Kessler 10 [K10] scores 10−15) and moderate to very high psychological distress (K10 scores ≥ 16) in each age group.Numbers of people currently experiencing moderate to very high levels of psychological distress are modelled as stocks with inflows corresponding to psychological distress incidence and outflows corresponding to recovery.Psychological distress incidence is equal to bhL, where b is the base per capita rate of distress onset per year, h is the product of the effects of developmental vulnerability during childhood, unemployment or (for people aged 15−24 years) non-participation in education or employment, and underemployment on psychological distress onset, and L is the number of people experiencing low levels of psychological distress.Moderately to highly distressed people in each age group recover at a yearly rate rH+T, where r is the per capita spontaneous recovery rate per year, H is the number of people currently experiencing moderate to very high psychological distress, and T is the number of people moving from a state of moderate to very high psychological distress to a state of low psychological distress per year due to effective mental health services.Aging of people experiencing moderate to very high levels of distress is modelled using a first-order delay like the population sector.The developmental vulnerability sector (Figure 4) models the effect of exposure to adversity during childhood on the risk of developing mental disorders in adolescence and adulthood.The number of developmentally vulnerable children aged 0−14 years is modelled as a stock that grows as children at low risk of psychopathology transition to a state of higher risk.The model assumes that the onset of psychopathological vulnerability depends on cumulative exposure to adverse experiences, such as (e.g., parental psychological distress, physical and sexual abuse, domestic violence, poverty) and is irreversible.
The incidence of significant psychopathological vulnerability is equal to bpL, where b is the base per capita rate at which children at low risk of psychopathology transition to a state of higher risk per year, p is the effect of parental psychological distress on the risk of developing mental disorders in later life, and L is the number of low-risk 0−14-year-olds in the population.The suicidal behaviour sector, represented in Figure 5, models captures self-harm hospitalisations and suicide deaths.Note that we equate suicide attempts with intentional self-harm hospital admissions due to data availability constraints.Age-specific suicide attempt rates are calculated as s i L i + r S i H i , where L i and H i are the numbers of people in age group i experiencing low psychological distress and moderate to very high psychological distress, respectively, S i is the per capita suicide attempt rate for mildly distressed people in age group i, and r is the suicide attempt rate ratio.
The number of suicide deaths per year is calculated as aF, where a is the suicide attempt rate and F is the suicide attempt lethality, that is the proportion of suicide attempts that are fatal.Numbers of people currently studying for a post-secondary qualification are modelled as stocks with inflows corresponding to enrolment and outflows corresponding to completion and discontinuation (drop out prior to completion).Age-specific enrolment rates are calculated as bdN, where b is the base per capita enrolment rate per year, d is the effect of psychological distress on entry into post-secondary study, and N is the number of people not currently studying.Completion and discontinuation rates are equal to cS and hdS, respectively, where c is the per capita completion rate per year, d is the base per capita discontinuation rate per year, h is the effect of psychological distress on the discontinuation rate, and S is the number of people currently studying for a post-secondary qualification When turning 15, adolescents are assumed to enter the population of those not in the labour force (NILF), which corresponds to people who are neither employed nor seeking employment.People not in the labour force who decide to seek employment enter the stocks of unemployed people, while those seeking employment (the unemployed) may leave the labour force.For each age range, net flows from the unemployed population to the population of people not in the labour force are calculated as fbU-hrN, where U and N are, respectively, the numbers of unemployed people and people not in the labour force, b is the base per capita rate that unemployed people leave the labour force per year, f is the effect of the unemployment rate on labour force participation (assumed to be greater than 1, so that increases in the unemployment rate reduce participation, r is the base per capita rate that people enter the labour force per year, and h is the product of the effects of psychological distress and completion of post-secondary education or vocational training on the labour force entry rate.

Figure 7. Structure of the employment sector h) Mental health services sector
A high-level view of the mental health services sector is presented in Figure 8.This sector models the movement of people through the mental health care system.People with low or moderate to very high psychological distress engage with mental health services in two ways: they may either perceive a need for mental health care and seek help (for example from a general practitioner or online services), or they may present to an emergency department (ED) (for example for self-harm) without having previously perceived a need for treatment.
After engaging with mental health services, people may: -recover following treatment, returning to the general population of people with low psychological distress and no perceived need for care, -be treated but not recover, or -disengage due to excessive waiting times, as a result of insufficient service capacity, or because they are dissatisfied with the care they receive.
People who are treated but do not recover return to perceiving a need for services and will eventually seek help again if they do not recover spontaneously; thus, people entering the mental health care system continue receiving treatment until they recover, disengage, or die (captured in the model, but not shown in figure 8).

Figure 8. High level view of the mental health services sector
People with low or moderate to very high psychological distress who are not currently considering engaging with mental health services perceive a need for care at rates equal to Pi Di, where Pi is the per capita rate that people with distress level i perceive a need for care per year, and Di is the number of people with distress level i not currently considering treatment.The per capita rates Pi are assumed to increase at a constant rate per year due to increasing public awareness of high-prevalence mental disorders and available treatment options.
After perceiving a need for treatment, people engage with mental health services at per capita rates dependent on their age and levels of psychological distress.Recently treated patients who have not recovered or disengaged from services return to perceiving a need for care and may attend subsequent appointments with a general practitioner (GP) or community mental health care (CMHC) psychiatric services (i.e., psychologists, psychiatrists and allied health services, hospital outpatient services), be admitted to a general or private hospital, commence online treatment, or present to an emergency department (Figure 9).Prospective and current patients may age, recover spontaneously, or transition from a state of low psychological distress to a state of moderate to very high distress.Prior to receiving treatment, people referred to a psychologist, psychiatrist or allied health services by a general practitioner or after completing hospital inpatient care wait for a period of time that depends on service capacity and the total number of people waiting for care.The stock of people waiting for treatment also contains people currently engaged with specialised services who have planned (follow-up) appointments (these patients referred to services after receiving hospital care enter via the flow called 'Additional psychologist, psychiatrist or allied health services' in Figure 10).Service capacity, i.e., the number of psychologist, psychiatrist and allied health services that can be provided per year, increases at a constant rate per year, estimated from MBS claims data.People receiving treatment are referred to psychiatric hospital services, disengage from the mental health services system due to dissatisfaction with the care received, recover, or return to perceiving a need for care (these people flow back into the arrayed stock labelled 'Perceived need for services'.The large impacts of the continuing COVID-19 pandemic were modelled as abrupt change in multiple flows directly affected by infection control measures (lockdowns, social distancing, international and interstate travel restrictions, including: 1.A decrease in the number of people arriving from overseas per year; 2. Increases in the per capita rates at which people transition from employment (including underemployment) to unemployment and from full employment to underemployment; 3. Reductions in per capita rates of non-acute mental health services provision (including general practitioner services, psychologist, psychiatrist and allied health services, public hospital outpatient services, and private mental health services); 4. An increase in the incidence of moderate to very high psychological distress resulting from social dislocation unrelated to job loss (e.g., working from home, not participating in recreational activities, restricted social gatherings) and anxiety about potential unemployment To take into account the resilience of the population as people adapted to the new situation and resumed employment, we allow the per capita spontaneous recovery rate to increase as the prevalence of psychological distress increases above that observed immediately prior to the start of the pandemic.As the direct social and economic effects of the pandemic abate, the incidence of moderate to very high psychological distress declines, and the higher per capita recovery rate results in a relatively rapid decrease in distress prevalence, consistent with the empirical data.

Intervention Description Direct Access program
Program designed to enable a proportion of help-seeking people to have access to subsidized sessions with psychologists or allied workers without requiring a general practitioner referral and mental health plan.Parameters that can be modified are: Starting year -the year in which the Direct Access program commences (the default is 2022, or January 2022.When enabled in combination with specialised mental health service capacity growth, the starting year was changed to 2024, or January 2024).
Implementation time (years) -the time required for the Direct Access program to be fully implemented (the default is 2 years).
Program duration (years) -the duration of the Direct Access program (the default is 1000 years, ensuring the Direct Access program is enabled until the end of the simulation).
Maximum use of Direct Access -maximum proportion of help-seeking people having access to subsidized sessions with psychologists or allied workers without requiring a general practitioner referral and mental health plan.For this intervention, we varied to value from 0.1 to 0.5, corresponding to 10% to 50% of help-seeking people.

Specialised mental health service capacity growth
Multiplies the annual rate of increase in the total number of psychologists, psychiatrists and allied services that can be provided per week.The default value (1) corresponds to the business as usual case, in which service capacity continues to increase at the current rate, estimated using Medicare Benefits Schedule (MBS) data for 2014-2019 assuming services were operating at (near) maximum capacity over this period.For this intervention, we multiplied the annual rate of increase by successively 2, 3 and 5.
This multiplicative increase in service capacity growth rates commences in January 2022 and remains in place until the end of the simulation.

Increase in demand for mental-health services
Increases (by 5% to 10%) the per capita rates at which people perceive a need for mental health services and seek help from a general practitioner, online services, and from psychologists and allied professionals if Direct Access is enabled.
Default duration: until end of simulation.

Employment programs
As employment programs were implemented by the Australian government in response to the COVID-19 pandemic, this intervention was enabled for the COVID-19 baseline scenario model calibration.
These employment programs are designed to stem rapidly increasing unemployment due to the COVID-19 pandemic (e.g., the JobKeeper Payment).This intervention reduces the increase in the per capita job loss rate resulting directly from the Intervention Description pandemic.The per capita rate of employment initiation can also be increased (or decreased); however, the default settings assume that employment programs have no direct effect on employment initiation.Parameters that can be modified are: Starting year -the year in which employment programs commence (the default is 2020.33 or end April 2020).
Implementation time (years) -the time required for employment programs to be fully implemented (the default is 0.167 years, or 2 months).
Program duration (years) -the duration of employment programs (the default is 1 year).
Effect on job loss -the multiplicative effect of employment programs on the increase in the job loss rate due to the COVID-19 pandemic.The default value (0.56) assumes that employment programs will reduce the increase in the per capita job loss rate by 44% (40).
Effect on employment initiation -the multiplicative effect of employment programs on the per capita employment initiation rate.The default value (1) assumes no effect of employment programs on employment initiation.

Better Access
As this program was implemented by the Australian government in response to the COVID-19 pandemic, this intervention was enabled for the COVID-19 baseline scenario model calibration.
Reform of the existing Better Access to Psychiatrists, Psychologists and General Practitioners through the MBS (Better Access) initiative to provide people with access to a greater number of specialised mental health care consultations per year.This intervention increases the flow of people with a perceived need for mental health care into psychiatrist and allied mental health services.Parameters that can be modified are: Starting year -the year in which the reformed Better Access initiative commences (the default is 2020.75, or October 2020).
Implementation time (years) -the time after commencement required for the reformed Better Access initiative to be fully implemented (the default is 0.167 years, or 2 months).
Program duration (years) -the duration of the reformed Better Access initiative.The default is set to 2.25 years (2 years 3 months).
Services per week -the mean number of specialised mental health care services provided per patient per week.The default value (1) assumes that patients attend 1 consultation per week, so that a patient attending a total of 4 consultations (for example) is assumed to do so over a period of 4 weeks.
Additional services per patient -the mean number of additional specialized mental health care services provided per patient per year under the reformed Better Access scheme.The default value (4) assumes that patients will attend an additional 4 consultations per year when the cap on the number of consultations per patient is increased.The population of NSW in 2028 is projected to be 3.9% lower than what would have been the case if the pandemic had not occurred because of the impact of COVID-19 on overseas and interstate migration (41).Despite this population decrease, Table 1 shows that self-harm hospitalisations are expected to increase by 1.79% for the general population (all ages), and by 4.85% in the 15-24 years age group relative to the no-COVID-19 scenario.These projections highlight that the COVID-19 disruption will have a more severe impact on the mental health and well-being of younger people in coming years.

Figure 2 .
Figure 2. Structure of the population sector

Figure 3 .
Figure 3. Structure of the psychological distress sector

Figure 4 .
Figure 4. Structure of the developmental vulnerability sector

Figure 5 .
Figure 5. Structure of the suicidal behaviour sector

Figure 6 .
Figure 6.Structure of the education sector

Figure 9 .
Figure 9. Flow structure of the help-seeking, general practitioner services, and online services components of the mental health services sector

Figure 10 .
Figure 10.Flow structure of the psychologist, psychiatrist and allied health services component of the mental health services sector

Figure 11 .
Figure 11.Adverse mental health outcome estimates derived from the model and corresponding historical data from HealthStats NSW (8-9)and AIHW (10) Refers to allowing x% of help-seeking people to use the Direct Access program n x Services growth rate Corresponds to multiplying by n the annual growth rate of specialised mental health care consultations

Table 3 . Projected mental health-related emergency department presentations, self-harm hospitalisations and suicide deaths over the period September 2021 to September 2028 with and without COVID-19 No COVID-19 COVID-19 Proportional change
This decrease occurs because of lower population growth due to COVID-19 related border closures.