710730

research-article2017
ANP0010.1177/0004867417710730ANZJP ArticlesLee et al.

Research

Australian & New Zealand Journal of Psychiatry

Cost of high prevalence mental 2017, Vol. 51(12) 1198­–1211

https://doi.org/10.1177/0004867417710730
DOI: 10.1177/0004867417710730

disorders: Findings from the 2007 © The Royal Australian and

New Zealand College of Psychiatrists 2017

Australian National Survey of Mental Reprints and permissions:

sagepub.co.uk/journalsPermissions.nav
journals.sagepub.com/home/anp
Health and Wellbeing

Yu-Chen Lee, Mary Lou Chatterton, Anne Magnus, Editor’s Choice

Mohammadreza Mohebbi, Long Khanh-Dao Le and
Cathrine Mihalopoulos

Abstract
Objective: The aim of this project was to detail the costs associated with the high prevalence mental disorders (depres-
sion, anxiety-related and substance use) in Australia, using community-based, nationally representative survey data.
Methods: Respondents diagnosed, within the preceding 12 months, with high prevalence mental disorders using the
Confidentialised Unit Record Files of the 2007 National Survey of Mental Health and Wellbeing were analysed. The use
of healthcare resources (hospitalisations, consultations and medications), productivity loss, income tax loss and welfare
benefits were estimated. Unit costs of healthcare services were obtained from the Independent Hospital Pricing Author-
ity, Medicare and Pharmaceutical Benefits Scheme. Labour participation rates and unemployment rates were determined
from the National Survey of Mental Health and Wellbeing. Daily wage rates adjusted by age and sex were obtained from
Australian Bureau of Statistics and used to estimate productivity losses. Income tax loss was estimated based on the
Australian Taxation Office rates. The average cost of commonly received Government welfare benefits adjusted by age
was used to estimate welfare payments. All estimates were expressed in 2013–2014 AUD and presented from multiple
perspectives including public sector, individuals, private insurers, health sector and societal.
Results: The average annual treatment cost for people seeking treatment was AUD660 (public), AUD195 (individual),
AUD1058 (private) and AUD845 from the health sector’s perspective. The total annual healthcare cost was estimated at
AUD974m, consisting of AUD700m to the public sector, AUD168m to individuals, and AUD107m to the private sector.
The total annual productivity loss attributed to the population with high prevalence mental disorders was estimated at
AUD11.8b, coupled with the yearly income tax loss at AUD1.23b and welfare payments at AUD12.9b.
Conclusion: The population with high prevalence mental disorders not only incurs substantial cost to the Australian
healthcare system but also large economic losses to society.

Keywords
Cost-of-illness, cost, depression, anxiety disorders, substance use disorders, Australia, NSMHWB 2007, National Survey
of Mental Health and Wellbeing 2007

Introduction
Mental and substance use disorders have drawn global
attention in the last decade due to their health, social and Deakin University, Geelong, VIC, Australia
economic consequences (Whiteford et al., 2013; World
Health Organization, 2014). According to the 2010 Global Corresponding author:
Cathrine Mihalopoulos, Deakin Health Economics, School of Health
Burden of Disease study, mental and substance use disor- and Social Development, Deakin University, 221 Burwood Highway,
ders were the leading cause of non-fatal burden of disease Burwood, VIC 3125, Australia.
(Whiteford et al., 2013). Globally, depression, anxiety Email: cathy.mihalopoulos@deakin.edu.au

Australian & New Zealand Journal of Psychiatry, 51(12)

Lee et al. 1199

Table 1.  Subgroup diagnoses within each principal diagnosis used in the 2007 NSMHWB.

Substance use (including substance abuse,

Depression Anxiety-related disorder dependence, and harmful use)

Minor depressive episode Obsessive-compulsive disorder Alcohol

Moderate depressive episode Post-traumatic stress disorder Opioids

Major depressive episode Panic disorder Sedatives

Dysthymia (mild long-term depression) Social phobia Marijuana

Agoraphobia Stimulant
Generalised anxiety disorder  

NSMHWB: National Survey of Mental Health and Wellbeing.

disorders and substance use disorders comprised 40.5%,
14.6% and 20.5% (including both illicit drug use and alco-
hol use disorders) of the total disability adjusted life years
(DALYs) attributed to mental and substance use disorders,
respectively (Whiteford et al., 2013). In Australia, mental
and behavioural disorders accounted for 12.9% of total dis-
ease burden, ranking third after cancer and cardiovascular
diseases. When limited to non-fatal diseases, they made up
of 22.3% of total DALYs in Australia (Australian Institute
of Health and Welfare, 2016).
While burden of disease studies measure health impacts,
they comprise only one element of ‘burden’. Cost of illness
(COI) studies measure the economic burden associated
with a health condition including both direct and indirect
costs. Direct costs usually represent the monetary value of
healthcare resources that could be used for other purposes
if the disease was not present (e.g. hospitalisations, com-
munity medical costs, pharmaceutical costs, allied health-
care costs). Indirect costs usually refer to the value of lost
productivity due to morbidity and mortality. Other costs,
such as direct non-health sector costs can also be included
in COI studies (e.g. criminal justice system). COI studies
can be very informative to policy-makers as they not only
provide estimates regarding the cost associated with the
treatment but also economic impact on different social sec-
tors (Tarricone, 2006).
Although numerous international COI studies for high
prevalence mental disorders (HPD), including depression,
anxiety and substance use disorders, are available, there are
major impediments to their usefulness in the Australian
decision-making context. For example, there are methodo-
logical inconsistencies between the studies as well as impor-
tant structural differences in healthcare systems (Luppa
et al., 2007). The mental health expenditure series published
by Australian Institute of Health and Welfare (2014) and
Australian Government Department of Health and Ageing
(2013), while useful, are based on ‘top down’ expenditure
allocations, which do not report total expenditure by diag-
nostic category and do not include costs outside of govern-
mental expenditures (such as productivity impacts).
Furthermore, these costs are dependent on institutional
diagnostic coding systems which may or may not fully
account for costs associated with mental health problems.
The aim of the current study is to estimate the annual
costs of HPD from multiple perspectives within the
Australian health sector and society using individual data
from the most recent Australian community-based mental
health survey, the 2007 National Survey of Mental Health
and Wellbeing (NSMHWB). This is the first study, to our
knowledge, which has costed service use and productivity
losses using data from this survey.
Methods
This analysis utilised the 2007 NSMHWB conducted by
the Australian Bureau of Statistics (ABS). The 2007
NSMHWB surveyed Australians aged 16–85 with a focus
on three major groups of mental disorders: anxiety disor-
ders, affective disorders and substance use disorders (ABS,
2008c). The information collected included individual
diagnosis, healthcare resource use, physical conditions,
severity of the mental issues, impact of mental health on
daily life, and demographic and socio-economic status. The
response rate of the 2007 survey was 60% with a final sam-
ple of 8841 respondents. Since the response rate was less
than expected (in comparison with the first NSMHWB con-
ducted in 1997 where there was a 78% response rate), con-
cerns regarding the representativeness of the survey were
addressed by extensive non-response analyses and adjust-
ment in the weighting strategy (ABS, 2008b). The age and
gender of the respondents was weighted to match the age
and gender distribution in Australia.
Population
We estimated the costs associated with HPD including
depression, anxiety-related disorders and substance use dis-
orders (see Table 1). People with comorbid psychosis were
excluded from the analysis due to exceptionally high man-
agement costs that might introduce bias. The International

Australian & New Zealand Journal of Psychiatry, 51(12)

1200 ANZJP Articles

Table 2.  Resources included in the analysis from each perspective.

Perspective Type of items

Health sector’s perspective Any costs associated with mental health-related admission
Any cost associated with mental health-related consultation
Any costs associated with pharmaceuticals

Public (i.e. government) Welfare benefit

Income tax forgone
Public mental health-related admission
Mental health-related consultation reimbursed by MBS
Pharmaceuticals reimbursed by PBS

Private (i.e. private insurance companies) Private mental health-related admission

Mental health-related consultation reimbursed by private insurance

Individuals (i.e. out-of-pocket) Out-of-pocket costs for mental health-related consultation

Out-of-pocket costs for pharmaceuticals

Societal perspective Any costs associated with mental health-related admission

Any cost associated with mental health-related consultation
Any costs associated with pharmaceuticals
Productivity loss

MBS: Medicare Benefits Schedule; PBS: Pharmaceutical Benefits Scheme.

Classification of Diseases–10 (ICD-10) was selected to
identify the target population instead of the Diagnostic and
Statistical Manual of Mental Disorder–Fourth Edition
(DSM-IV). This was consistent with other existing studies
which used the 2007 NSMHWB (ABS, 2008d; Burgess
et al., 2009). In addition to the specified diagnosis, the
recency of symptoms further categorised diagnosis as (1)
30-day diagnosis, (2) 12-month diagnosis and (3) lifetime
diagnosis. The 12-month diagnosis category was adopted
in the analysis to match the collected 12-month resource
use data.
Costs
A bottom-up costing approach based on individual-level
data was adopted in the estimation of costs associated with
HPD. The type of costs included in the analysis was mental
health-related treatment costs (i.e. hospitalisation, consul-
tation and medication use), productivity loss, distributed
welfare benefits and income tax forgone. The perspectives
taken in the study were public (i.e. government), private
(i.e. insurance), individual (i.e. out-of-pocket payment by
consumers), health sector (i.e. public + private + individ-
ual) and societal (i.e. productivity impact). Table 2 sum-
marises the resource use included in each of perspectives.
All the costs presented were calculated as annual costs and
expressed in 2013–2014 AUD
Mental health treatment costs
Twelve-month mental healthcare-related resource use
(including hospitalisations, consultations and medications)
was collected in the 2007 NSMHWB using a self-report
format and costed for the current study. Total costs were
determined for the whole population meeting the criteria
for a HPD. Average costs per person were based on people
who met the criteria for a HPD and sought treatment.
Treatment seeking refers to people who meet the criteria for
a HPD and stated that they used at least one healthcare
resource specifically for their mental health.
Data from the Independent Hospital Pricing Authority
(IHPA) were used to assign costs to each hospitalisation
reported by survey respondents. The IHPA was established
by the Commonwealth of Australia as part of the National
Health Reform Act 2011 to determine an annual National
Efficient Price (NEP), which is a major determinant of the
cost to the Australian Government for public hospital ser-
vices. In addition, the IHPA also developed the National
Weighted Activity Unit (NWAU) as a measure of activity
expressed as a common unit, against which the NEP is paid
(Australian Institute of Health and Welfare, 2015). For each
hospital admission, a NWAU was calculated using
Australian refined Diagnosis Related Groups (AR-DRGs)
in the IHPA algorithms. The NWAU derived from different
AR-DRG codes was based on the reason for hospital admis-
sion reported in the survey. The resulting NWAU per case
was then multiplied by the 2013–2014 published NEP price
of AUD4993 (IHPA, 2015). Public hospital cost was used
for both public and private hospital admissions since many
cost items occurring in the public sector such as medical
salaries, pathology, pharmacy, imaging or allied health ser-
vices were not accounted for in private hospitals’ unit costs.
The use of up to five medications with duration of use
for each survey respondent was collected in the 2007

Australian & New Zealand Journal of Psychiatry, 51(12)

Lee et al. 1201
NSMHWB. Since the reason for taking each medication
was not collected in the survey, medications commonly
prescribed for people diagnosed with HPD were costed in
the analysis including sleeping pills (including antihista-
mine), tranquillisers, antipsychotics, antidepressants, hor-
mones and mental health-related symptom relievers (i.e.
beta-blockers, proton pump inhibitors, etc.). Additionally,
as the strength and the form of medication was not col-
lected in the survey, the weighted average cost of all
strengths and forms for each medication obtained from
2013–2014 Pharmaceutical Benefits Scheme (PBS) item
reports was assigned as the monthly unit cost (Department
of Human Services, 2015). Note a co-payment (i.e. out-of-
pocket cost) is generally required for prescribed medication
in Australia and the amount varies by the type of health
card held, the total cost of medication and whether the
annual safety-net limit has been reached by individuals
within a calendar year. According to the PBS in 2013–2014,
for people with a general health card (i.e. a Medicare card),
the maximum co-payment for each drug in each prescrip-
tion was AUD36.90 and the annual safety net was
AUD1421.20. For concession card holders, the maximum
co-payment for each drug in each prescription was AUD6
and the annual safety net was AUD360 (Australian
Government Department of Health, 2015). Since the health
card status and the number of prescriptions were not col-
lected in the survey, two assumptions were made to esti-
mate the medication costs: (1) each prescription covered 1
month and (2) people who reported receiving welfare ben-
efits were holding concession cards.
The visits to different health professionals for mental
health problems during the past 12 months were collected
and costed in the analysis by multiplying the number of
visits by a unit cost. Consultations were reported as being
funded by three possible sources. For consultations reported
as being reimbursed by Medicare (Government-funded
universal public insurance scheme), the unit cost was
obtained from the Medicare Benefits Schedule (Australian
Government Department of Health, 2014). The rebate of
each consultation was determined by type of healthcare
professional and length of contact. For health professionals
without an exact item in the Medicare schedule, the
weighted average of all the allied health visits reimbursed
by Medicare was adopted as a proxy. When private health
insurance was reported as the funding source, unit costs
were obtained from The Private Health Insurance
Administration Council (2015). Any cost reported as self-
funded (i.e. out-of-pocket) was included within the indi-
viduals’ perspective.
Productivity costs
The human capital method was adopted in the current
analysis to estimate the productivity loss, based on the
self-reported total lost working days in the preceding
12 months attributed to HPD (van den Hout, 2010). For
respondents who reported their status as being employed,
a daily wage adjusted by age and gender, expressed in
2013–2014 AUD, was applied to estimate the loss associ-
ated with HPD (ABS, 2014). For people younger than
65 years old who reported their status as being unem-
ployed or not in the labour force, productivity loss was
estimated following methods used by Neil et al. (2013)
and Tarricone (2006). This method assumed that an indi-
vidual with HPD should expect the same level of employ-
ment at similar wage rates as the general population. The
cost of productivity loss was based on differences in
labour participation and employment rates between the
general population and the population with HPD in
Australia. Since the recommended annual leave in
Australia is 4 weeks, 48 working weeks were used as the
annual working weeks (Fair Work Ombudsman, 2016).
The following formula was used to estimate the produc-
tivity loss in people not employed:
Productivity loss of unemployed and not in labour force HPD
population  = 
daily adjusted wage rate*5*48*number
unemployed and not in labour force *(proportion employed in
general population − proportion employed in HPD population)
/ (proportion of general population unemployed + proportion
of general population not in labour force)
Using conventional productivity costs methodology, the
average daily wage was used (which includes a tax
component).
Transfer payment/welfare benefits
Transfer payments, such as welfare benefits, refer to the
flow of money from government to individuals that does
not arise from the production of resources (Drummond
et al., 2005). As the government perspective was also
adopted in the analysis, this financial flow was included.
In the 2007 survey, respondents were asked whether a
government pension or allowance was a source of income
without further specification. Hence, the average of the
welfare benefits that were most likely to be claimed by the
people with mental illness was adopted (with age used as
the indicator for the amount of benefit received). Survey
participants with a HPD, who were below 22 years of age
and reporting as claimants of welfare benefits, were
assigned a fortnightly Youth Allowance of AUD444. Those
of working age between 22 and 65 were assigned the aver-
age of a NewStart Allowance, Disability Support Pension
and Sickness Allowance of AUD588.84 fortnightly. People
with HPD, who were over retirement age (65+), were
assigned no welfare payments as the planned nature of
retirement was unknown and this is unlikely to change as a
consequence of HPD (e.g. people are moved to an old age
pension).
Australian & New Zealand Journal of Psychiatry, 51(12)
1202 ANZJP Articles
Income tax forgone
From the government’s perspective, forgone income tax asso-
ciated with HPD was estimated based on the lost productivity
of only the respondents who were unemployed or not in the
labour force, as the productivity loss of the employed popula-
tion with HPD was borne by the employer. The individual
income tax rates were sourced from the Australian Taxation
Office (2015) and were applied to the average daily wage
rates used in the productivity costs estimation.
Data analysis
All analyses were conducted in Stata SE version 13.1
(College Station, Texas, USA). Since the survey was con-
ducted with complex sample designs and weighting proce-
dures, both person weight and replicate weights were
applied to determine unbiased results and more precise
variance for the total in-scope population (ABS, 2008d).
Given this, the jackknife resampling technique for unequal
probability of selection was used to estimate standard errors
and perform hypothesis tests (ABS, 2009b). This technique
was applied for both population and subgroup level. The
jackknife procedures were applied to the data since they
were recommended by ABS (2009b).
The descriptive analyses undertaken included the mean
and total costs of each healthcare resource utilised as well
as the mean and total cost incurred from multiple perspec-
tives as listed in Table 2. We conducted independent analy-
sis to assess the impacts of comorbidities by grouping the
HPD population into seven mutually exclusive diagnostic
groups, namely, (1) Depression only (D), (2) Anxiety-
related disorders only (ANX), (3) substance use disorders
only (SUB), (4) Depression + anxiety-related disorder (D +
ANX), (5) Depression + substance use disorder (D + SUB),
(6) Anxiety-related disorder + substance use disorder (ANX
+ SUB), and (7) Depression + Anxiety-related disorder +
substance use disorder (D + ANX + SUB). Proportion data
(i.e. the proportion of the total HPD population who
reported the use of healthcare, being employed, etc.) were
reported with 95% confidence interval and costs data as
mean and standard deviation. The parametric test (adjusted
Wald test) was applied to detect the statistically significant
differences across diagnostic subgroups (described above)
for the weighted proportion and cost data (Tu and Zhou,
1999). If there were statistically significant differences
across these subgroups, post-hoc pairwise comparisons
were conducted to examine whether there was statistically
significant differences between specific diagnostic groups.
Results
Diagnostic profile
Approximately 18.5% of Australians were diagnosed with
HPD in 2007. Within the HPD population, the majority
Australian & New Zealand Journal of Psychiatry, 51(12)
were diagnosed with a single diagnosis (80%). ANX
accounted for more than 50% of the diagnoses in the HPD
population, followed by SUB (17%), D + ANX (11%) and
D (9%). Overall, men accounted for less than half (43%) of
the HPD population. This gender disparity was reversed for
the other mental health diagnoses without SUB, as a comor-
bidity. Additionally, the mean age of people diagnosed with
SUB alone or in combination with SUB was significantly
lower in comparison with other diagnoses.
Health service utilisation
Table 3 summarises the utilisation rate of mental health-
related resource use for the HPD population by type of
resource and diagnostic group. Overall, 39% of the HPD
population in Australia reported the use of mental health-
related healthcare resources. There were statistically sig-
nificantly higher proportions of people diagnosed with D +
ANX with or without SUB who sought any treatment (81%
and 70%, respectively) than for those diagnosed with any
single mental health diagnosis (D or ANX or SUB)
(p < 0.0000). There were also significant differences in the
proportions seeking treatment across the groups with a sin-
gle mental health diagnosis, with SUB lowest (13%) fol-
lowed by ANX (36%) and D (53%). However, no significant
difference of these proportions was found between people
with SUB in combination with D or ANX and those who
were diagnosed with a single mental health diagnosis. The
health funding source for 76% of people reporting use of
healthcare resources was the public sector (i.e. Medicare or
admitted as public patients). Hospitalisation was the least
utilised resource, with less than 2% of respondents with
HPD reporting being admitted to hospital in the preceding
12 months due to their mental health.
A larger proportion of people in the D + ANX + SUB
subgroup reported being hospitalised (28%) compared to
the remaining diagnostic subgroups; however, this was not
statistically significant. Only people with D + ANX had
statistically significantly higher proportion of hospitalisa-
tion compared to those diagnosed with the single diagno-
ses of ANX or SUB. Consultation services were the most
frequently utilised healthcare services among all, with
more than 33% of the HPD population reporting ever
receiving a consultation from a healthcare professional for
mental health problems, and 82% reporting that these were
funded exclusively by the public sector. People with D +
ANX with or without SUB reported significantly higher
use of consultation services (67% and 77% respectively)
compared to the remaining diagnostic subgroups. There
were statistically significant differences in the proportion
utilising consultations across single diagnostic subgroups
with SUB having the lowest rate (12%), followed by ANX
(29%) and D with the highest rate at 49%. Nearly 23% of
people with HPD reported the use of medication relevant
to their mental health. People with two or more mental
Lee et al. 1203

Table 3.  Proportion of HPD population reporting the use of different type of healthcare resources for their mental health in the
2007 NSMHWB.

Proportion used any Proportion Proportion consulting

type of healthcare hospitalized % any type of healthcare Proportion prescribed
Diagnosis resources % (95% CI)** (95% CI)* professional % (95% CI)** medications % (95% CI)**

D 53 [43–64] 2 [0–4]a 49 [38–59] 35 [26–43]

ANX 36 [31–41] 1 [0–1]a 29 [24–34] 21 [17–26]

SUB 13 [8–19] 1 [0–1]a 12 [7–17] 5 [2–7]

D + ANX 70 [60–80] 4 [1–7] 67 [57–77] 44 [34–54]

D + SUB 47 [6–88] 0 47 [6–88] 0

ANX + SUB 44 [29–59] 4 [0–7] 32 [20–45] 22 [12–33]

D + ANX + SUB 81 [62–100] 28 [0–70]a 77 [56–99] 64 [36–92]

HPD 39 [36–42] 2 [1–3] 33 [30–36] 23 [20–26]

HPD: high prevalence mental disorders; NSMHWB: National Survey of Mental Health and Wellbeing; CI: confidence interval; D: Depression; ANX:
Anxiety-related disorders; SUB: substance use disorders.
aDespite using the jackknife technique to estimate the standard error (SE) which is a robust non-parametric resampling approach, we still observed

negative values for the lower boundary of the CI and truncated them to zero.
*Statistically significant difference across subgroup as described in the text: p < 0.0429.
**Statistically significant difference across subgroup as described in the text: p < 0.00001.

health disorders had significantly higher use of mental
health medications compared to those with single diagno-
ses (except for those with D + SUB). Citalopram was the
most frequently prescribed medication (17%), followed by
venlafaxine (15.65%), sertraline (14.43%) and temazepam
(10.51%).
Table 4 presents the average cost by healthcare service,
perspective and diagnostic subgroup for people with HPD
who reported the use of any healthcare resources for their
mental health in the past 12 months. The average annual
treatment costs of people with HPD was estimated at
AUD845 from the health sector perspective. People diag-
nosed with D + ANX + SUB incurred the highest average
annual mental health-related treatment costs from the
Health Sector perspective which was significantly greater
than for people with ANX (F = 4.79; p < 0.0327) or SUB
(F = 4.51; p < 0.038) but not D (F = 3.61; p < 0.0622). No
statistically significant differences in average costs were
found across single diagnostic subgroups. The key findings
from this table show that from both the public and the pri-
vate perspective, people with three diagnoses have the
highest cost. In contrast, from the individual’s perspective,
the average total out-of-pocket spending was similar among
all diagnostic subgroups.
Table 5 presents the total mental health–related treat-
ment cost for the population diagnosed with HPD in
Australia by perspective (health sector, public, private
and individuals’ out of pocket) and diagnosis group.
Figure 1 shows the proportion of the total mental health
treatment costs incurred by each diagnostic group. From
the health sector’s perspective, total annual healthcare
costs of people with HPD were estimated at AUD974m,
with nearly 72% of these costs coming from the public
sector, followed by individuals and private sector at 17%
and 11%, respectively. People with ANX and D + ANX
incurred the highest total costs among all diagnostic sub-
groups, comprising 39% and 29% of total costs, respec-
tively. The lowest total cost of all the diagnostic groups
was D + SUB comprising 0.4% of the total cost. Similarly,
from the public and private perspective, people with
ANX and D + ANX accumulated the majority of the total
costs, and D + SUB remained as the group with the low-
est costs among all diagnostic subgroups. From the indi-
vidual’s perspective, medications accounted for one-third
of the overall out-of-pocket costs, with the balance attrib-
uted to consultation services. Statistically significant dif-
ferences in total costs across subgroups were found
regardless of perspective or cost categories. An exception
was for hospitalisation costs which demonstrated non-
statistically significant differences across subgroups irre-
spective of perspectives.
Productivity loss
The total productivity loss attributed to HPD in Australia
was estimated at AUD11.8b, with 60% made up of forgone
earnings due to higher rates of unemployment in the HPD
population compared to the general population and the
remaining 40% attributed to lost productivity in the
employed HPD population (Table 6).

Australian & New Zealand Journal of Psychiatry, 51(12)

 1204

Table 4.  Average cost per person of HPD population reporting the use of healthcare resources for their mental health in the 2007 NSMHWB (2013–2014 AUD (Mean [SD]).

Diagnosis/ D + ANX +
Type of resource perspective D ANX SUB D + ANX D + SUBa ANX + SUB SUBa HPD

Unweighted sample size 154 906 249 190 12 108 26 1645

Hospitalisation Public* 9266 (1659) 7998 (1779) 5374 (1265) 7491 (2675) 0 4399 (2717) 3718 (1797) 6,023 (3035)
(AUD (Mean [SD]) Private 4645 (-)b 4521 (-)b 0 7237 (2355) 0 0 7405 (-)b 6318 (2294)
Health Sector** 8073 (2661) 9113 (142) 5374 (1265) 7810 (1830) 0 4399 (2717) 3894 (2052) 6432 (2946)

Consultation (AUD) Public*** 289 (335) 320 (331) 348 (424) 499 (454) 160 (146) 390 (394) 607 (306) 373 (382)
(Mean (SD)) Private 184 (136) 551 (547) 219 (44) 196 (174) 373 (-)b 276 (256) 193 (272) 407 (464)

Australian & New Zealand Journal of Psychiatry, 51(12)

OOP 261 (558) 246 (338) 215 (272) 321 (338) 246 (188) 315 (393) 108 (-)b 268 (370)
Health Sector 385 (576) 453 (489) 396 (403) 606 (597) 303 (169) 491 (553) 594 (319) 482 (523)

Medication (AUD) Public 103 (188) 148 (237) 169 (235) 163 (194) 0 128 (200) 240 (99) 149 (217)
(Mean (SD)) OOP 75 (103) 85 (87) 127 (174) 95 (109) 0 92 (76) 111 (74) 89 (97)
Health Sector 178 (243) 233 (280) 295 (359) 258 (247) 0 220 (234) 350 (126) 238 (265)

Total (AUD) Public**** 631 (1999) 482 (1166) 664 (1829) 847 (1792) 160 (146) 737 (1783) 2,013 (2317) 660 (1600)
(Mean, (SD)) Private 1388 (2255) 736 (989) 219 (44) 2,284 (4139) 373 (-)b 276 (256) 3691 (6060) 1058 (2068)
OOP 173 (395) 181 (277) 209 (333) 250 (336) 246 (188) 201 (321) 110 (72) 195 (306)
Health Sector** 808 (2113) 643 (1306) 678 (1511) 1217 (2254) 303 (169) 816 (1744) 2195 (2440) 845 (1761)

HPD: high prevalence mental disorders; NSMHWB: National Survey of Mental Health and Wellbeing; SD: Standard Deviation; D: Depression; ANX: Anxiety-related disorders; SUB: substance use
disorders; OOP: out of pocket.
aThis diagnostic categorisation is based on a small sample size therefore results may be less robust than the other diagnostic categories.
bStandard error (SE) was not calculated due to limited sampling units in a stratum.

*Statistically significant difference across subgroups: p < 0.0239.

**Statistically significant difference across subgroups: p < 0.0060.
***Statistically significant difference across subgroups: p < 0.0143.
****Statistically significant difference across subgroups: p < 0.0314.
ANZJP Articles
 Lee et al.

Table 5. Total costs people with HPD reporting the use of healthcare resources for their mental health in the 2007 NSMHWB (2013–2014 AUDmillion) (95% CI).

Type of Diagnosis/
resource perspective D ANX SUB D + ANX D + SUBa ANX + SUB D + ANX + SUBa HPD

Unweighted sample size 154 906 249 190 12 108 26 1645

Hospitalisation Public 40 [0, 63.3] 71.8 [0, 176] 14.9 [0, 36.8] 63 [0, 131] 0 28.7 [1.67, 55.8] 51.4 [0 to 115] 270 [121, 420]
(AUD million) Private 6.99 [–]b 10.1 [–]b 0 45.5 [0, 108] 0 0 5.14 [–]b 67.8 [0.46, 135]
(95%CI)
Health 47 [0, 101] 82.6 [0, 204] 14.9 [0, 36.8] 109 [19, 198] 0 28.7 [1.67, 55.8] 56.5 [0, 126] 338 [168, 509]
Sector

Consultation Public* 34.9 [22.8, 47] 134 [93.5, 175] 14.7 [7.05, 22.2] 100 [67.9, 132] 1.08 [0, 2.32] 19.5 [8.7, 30.4] 23.3 [1.69, 45] 328 [266, 390]
(AUD million) Private 0.75 [0, 2.02] 30.1 [0, 76.1] 2.28 [0, 7.06] 3.65 [0.95, 6.35] 0.9 [–]b 1.47 [0, 3.8] 0.28 [0, 0.83] 39.4 [0, 85.8]
(95%CI)
OOP* 12.5 [3.13, 21.8] 48.6 [29.1, 68.1] 5.76 [1.22, 10.3] 30.6 [17.7, 43.5] 1.46 [0, 3.99] 8.26 [1.32, 15.2] 0.11 [–]b 107[76.2, 138]
Health 48.1 [30.2, 66.1] 213 [127, 299] 22.7 [11.5, 33.8] 134 [95.3, 173] 3.43 [0.03, 6.83] 29.3 [13.8, 44.7] 23.7 [1.95, 45.5] 474 [395, 590]
sector*

Medication Public* 9.17 [4.03, 14.3] 51.2 [32, 70.4] 3.88 [0.58, 7.18] 23.8 [14.7, 32.9] 0 5.3 [2.47, 8.13] 7.9 [0, 16.4] 101 [77.3, 125]
(AUD million) OOP* 6.71 [3.22, 10.2] 29.3 [19.6, 38.9] 2.92 [0.41, 5.43] 13.9 [8.6, 19.2] 0 3.78 [1.79, 5.77] 3.64 [0, 7.5] 60.2 [45.2, 75.3]
(95%CI)
Health 15.9 [8.3, 23.5] 80.4 [53.5, 107] 6.8 [1.39, 12.2] 37.7 [24.7, 50.7] 0 9.08 [4.64, 13.5] 11.5 [0, 23.2] 161 [125, 198]
sector*

Total Public* 84.1 [22.4, 146] 258 [141, 374] 33.5 [4.08, 62.8] 187 [100, 274] 1.08 [0, 2.32] 53.6 [19.2, 87.9] 82.6 [0, 168] 700 [517, 882]
healthcare Private 7.74 [0, 22.1] 40.2 [0, 90.5] 2.28 [0, 7.06] 49.2 [0, 112] 0.9 [–]b 1.47 [0, 3.8] 5.42 [0, 16.6] 107 [26.7, 188]
resource
(AUD million) OOP* 19.2 [9.56, 28.8] 77.9 [53.3, 102] 8.68 [2.41, 15] 44.5 [27.8, 61.2] 1.46 [0, 3.99] 12 [4.4, 19.7] 3.75 [0, 7.63] 168 [128, 207]
(95%CI) Health 111 [47.5, 175] 376 [213, 538] 44.4 [12.7, 76.2] 281 [169, 393] 3.43 [0.03, 6.83] 67.1 [27.8, 106] 91.8 [0, 186] 974 [743, 1210]
sector*

NSMHWB: National Survey of Mental Health and Wellbeing; CI: confidence interval; D: depression; ANX: anxiety-related disorders; SUB: substance use disorders; OOP: out of pocket; HPD: high
prevalence mental disorders.
aThis diagnostic categorisation is based on a small sample size therefore results may be less robust than the other diagnostic categories.
bStandard error (SE) was not calculated due to limited sampling unit in a stratum.

*Statistically significant difference across subgroups: p < 0.00001.

Australian & New Zealand Journal of Psychiatry, 51(12)

1205
1206 ANZJP Articles
Figure 1. The proportion of total treatment costs incurred by the different diagnostic groupings.
D: Depressive Disorder; ANX: Anxiety Disorder; SUB: Substance Use Disorder only.
Productivity loss in the employed HPD
population
In all, 70% of the HPD population below the age of 65 indi-
cated they were employed. The number of mental health
diagnoses was negatively correlated with the proportion
reporting being employed. Within the employed HPD pop-
ulation, nearly 26% reported loss of working days associ-
ated with HPD in the preceding 12 months. The results also
suggested the number of comorbidities was positively cor-
related with the proportion of people who reported lost
working days. For people with any single diagnosis, on
average only 18% of people reported loss of working days
associated with HPD. The proportion increased substan-
tially to 58% and 71% for people with one and two comor-
bidities, respectively. Within the 26% of the employed
HPD population who reported lost working days, the aver-
age annual loss was 38 days, costing AUD4.64b. The esti-
mated productivity loss of people diagnosed with any two
HPD was AUD2.3b, which accounted for the highest pro-
portion of lost productivity costs (around 50%) in the
employed population, followed by people with any single
diagnosis and three diagnoses at 36% and 14%, respec-
tively (Table 6).
Productivity loss in the unemployed HPD
population
The estimated cost of lost productivity for those with a
HPD who reported they were not in employment (estimated
at 148,000 individuals) amounted to AUD7.18b to
Australian & New Zealand Journal of Psychiatry, 51(12)
Australian society (Table 6). Productivity costs include
both income accruing to the individual and taxes accruing
to the government. Below we separately report income tax
losses to governments as these may be of particular area of
interest to governments. Of the total productivity loss
attributed to the unemployed group, the population with a
diagnosis of ANX accounted for the highest proportion of
all diagnostic groups with nearly 55%, followed by D +
ANX (14.9%), SUB (11%) and D (10%).
Transfer payment/welfare benefits
Approximately 31% of the HPD population under 65 years
of age reported receiving welfare benefits from the govern-
ment. Within the group of people who reported their status
as not being employed (i.e. unemployed or not in the labour
force), nearly 70% were claiming welfare benefits.
Conversely, only 17% of the employed HPD population
were claimants of welfare benefits. Across single diagnosis
subgroups, a statistically significantly lower proportion of
people with SUB received welfare benefits (19%) com-
pared to those with D (42%; p < 0.0003) or ANX (31%;
p < 0.0047) but not statistically significant between depres-
sion and anxiety subgroups (p < 0.0924). The proportion of
people with D or ANX was not statistically significantly
different to those with D + ANX or those with three comor-
bidities but was statistically significantly higher compared
to those with D + SUB or ANX + SUB. It is worth noting
that only 3% of people with D + SUB reported receiving
welfare benefits, which was statistically significantly lower
 Lee et al.

Table 6.  Productivity loss associated with HPD in the 2007 NSMHWB in Australia 2013–2014 AUD.

Unemployed/not in
Labour status Employed labour force All

Average number Proportion of

Employment rate Proportion taking of lost working Total productivity people unemployed Total productivity Total productivity loss
(people < 65 years days off from work days due-HPD loss (AUD million) (people < 65 years loss (AUD million) (AUD million) (95%
Diagnosis old) % (95% CI)* % (95% CI)** (Mean (SD))*** (95% CI)**** old) % (95% CI)* (95% CI)** CI)**

D 61 [49, 73] 40 [27, 54] 35 [75] 463 [140, 786] 39 [27, 51] 822 [445, 1200] 1280 [793, 1780]

ANX 69 [65, 74] 18 [14, 22] 27 [67] 1080 [610, 1560] 31 [26, 35] 3910 [3190, 4630] 4990 [4150, 5840]

SUB 81 [74, 88] 11 [6, 15] 12 [33] 130 [0, 284]a 19 [12, 26] 702 [390, 1010] 833 [475, 1190]

D + ANX 61 [51, 72] 60 [45, 76] 65 [101] 2000 [894, 3100] 39 [28, 49] 1070 [700, 1430] 3070 [1960, 4180]

D + SUB 94 [81, 100]b 62 [13, 100]b 27 [41] 76.3 [0, 156]a 6 [0, 19]b 1.53 [–]c 77.9 [0, 158]

ANX + SUB 74 [62, 86] 55 [40, 71] 16 [43] 224 [0, 468]a 26 [14, 38] 426 [176, 677] 651 [241, 1060]

D + ANX + SUB 62 [–] 71 [27, 100]b 127 [150] 668 [0, 1730]a 42 [4, 80] 252 [0, 569] 920 [0, 2020]

HPD 70 [67, 73] 26 [23, 29] 38 [81] 4640 [3070, 6210] 30 [27, 33] 7180 [6180, 8180] 11,800 [10,000, 13,600]

HPD: high prevalence mental disorders; NSMHWB: National Survey of Mental Health and Wellbeing; CI: confidence interval; SD: Standard Deviation; D: Depression; ANX: Anxiety-related disorders;
SUB: substance use disorders
aDespite using the jackknife technique to estimate the standard error (SE) which is a robust non-parametric resampling approach, we still observed negative values for the lower boundary of the CI and

truncated them to zero.

bDespite using the jackknife technique to estimate the SE which is a robust non-parametric resampling approach, we still observed over 100% for the higher boundary or negative values for the lower

boundary of the CI and truncated this to 100% or 0%, respectively.

cConfidence interval was not calculated due to limited sampling unit in a stratum.

*Statistically significantly difference between subgroups: p < 0.0003.

**Statistically significantly difference between subgroups: p < 0.0000.
***Statistically significantly difference between subgroups: p < 0.0107.
****Statistically significantly difference between subgroups: p < 0.0001.

Australian & New Zealand Journal of Psychiatry, 51(12)

1207
1208 ANZJP Articles
Table 7. Welfare benefits and tax forgone due-HPD in the 2007 NSMHWB.
Proportion of people on
welfare benefit (population
under 65) %(95% CI)*
D 42 [30, 53]
ANX 31 [27, 36]
SUB 19 [12, 26]
D + ANX 44 [31, 57]
D + SUB 3 [0, 9]
ANX + SUB 24 [13, 35]
D + ANX + SUB 56 [23, 90]
Everyone 31 [28, 34]
HPD: high prevalence mental disorders; NSMHWB: National Survey of Mental Health and Wellbeing; CI: confidence interval; D: Depression; ANX:
Anxiety-related disorders; SUB: substance use disorders.
aConfidence interval was not calculated due to limited sampling unit in a stratum.
*Statistically significant difference across subgroups: p < 0.00001.
compared to the rest of the diagnostic groups. Further
details are presented in Table 7.
Income tax forgone
Overall, the estimated income tax loss associated with HPD
was AUD1.23b annually. People with ANX contributed
more than 50% of the total income tax loss compared to the
remaining subgroups and this was statistically significant.
People diagnosed with D + SUB incurred no income tax
loss because the annual income of the unemployed indi-
viduals in the D + SUB subgroup was less than the taxable
income (see Table 7 for further detail).
Discussion
This COI provides an in-depth analysis of the health
resource use and economic impact of HPD to Australia
through multiple payer perspectives using the most recent
comprehensive community-based survey, the NSMHWB,
conducted in 2007. Of the approximate 18% of Australians
diagnosed with HPD in the preceding 12 months, nearly
39% reported the use of healthcare resources for their men-
tal health, resulting in AUD974m of treatment costs to the
health sector. The HPD population also accounted for
AUD11.8b of annual productivity loss, AUD1.23b in for-
gone of income taxes and AUD12.9b of welfare benefits.
The annual total cost to Australian society was estimated at
AUD12.8b.
While the estimated total societal cost of HPD in this
analysis was nearly three times greater than the AUD4.91b
estimated cost of psychosis in Australia, the estimated aver-
age costs per person was much lower compared to people
Australian & New Zealand Journal of Psychiatry, 51(12)
Total welfare benefits Total tax forgone in Australia
received in Australia (2013– (2013–2014 AUD million)
2014 AUD million) (95% CI)* (95% CI)*
1410 [850, 1970] 148 [73, 223]
7070 [5910, 8230] 648 [515, 781]
1280 [770, 1790] 120 [64, 176]
2080 [1230, 2930] 180 [114, 246]
10.8 [–]a 0
652 [338, 966] 78.6 [27, 131]
436 [34, 839] 52.5 [0, 123]
12,900 [11,500, 14,400] 1230 [1030, 1420]
with psychosis (Neil et al., 2013). Considering the preva-
lence of psychosis is significantly lower than HPD (0.45%
vs 18%) (Department of Health, 2011), psychosis has a
greater economic impact on the individual than HPD.
The estimated 12-month HPD prevalence from our anal-
ysis was slightly lower than the value reported by the ABS
and other published literature due to the differences in the
diagnoses included (ABS, 2008e; Burgess et al., 2009;
Slade et al., 2009; Sunderland et al., 2012). Only people
with depression, anxiety-related disorder and substance use
disorders were included in our analysis, whereas in the
other analyses, additional low prevalence disorders such as
bipolar disorder, mania and hypomania were also included.
This resulted in a higher overall prevalence compared to
our estimates. Furthermore, our use of the psychosis
screener to exclude people with 12-month psychosis from
our target sample further reduced the estimated HPD preva-
lence (ABS, 2008e; Burgess et al., 2009; Slade et al., 2009;
Sunderland et al., 2012).
The estimates of healthcare resource utilisation in the
current analysis were slightly higher than those published
by ABS (39% vs 35%) (ABS, 2008a). The difference was
mainly driven by the inclusion of medication use in this
analysis as well as the exclusion of low prevalence disor-
ders. In the previous analyses based on the 2007 NSMHWB,
the resource utilisation rate only considered hospitalisations
and consultations, but not medication (ABS, 2008e; Burgess
et al., 2009; Slade et al., 2009). The rate of medication use
in this study was 23% and the percentage of total health sec-
tor costs spent on medications was 12%, which is similar to
other COI studies in depression (Luppa et al., 2007).
The presence of a positive correlation between the num-
ber of comorbidities and healthcare utilisation rates in the
Lee et al. 1209
current analysis is consistent with the existing literature
(Burgess et al., 2009; Mihalopoulos et al., 2005). Similarly
to healthcare utilisation patterns, the average total cost of
people with mental health comorbidities was significantly
higher than people with a single diagnosis. The analysis
also demonstrated the impact of HPD on time taken off
work and employment status (ABS, 2012). The proportion
of people who reported being unemployed was positively
correlated with the number of diagnoses, and the overall
proportion of the HPD population reporting being employed
was statistically significantly different to the general popu-
lation (70% vs 75%). This finding is echoed by Inder et al.
(2012), which suggested that people with mental illness on
average were 5%–7% less likely to be employed than indi-
viduals without mental illness based on the same data.
However, data published by the Department of Health in
2013 reported an 18% difference in the proportions being
employed between people with and without mental ill-
nesses (The Department of Health, 2013). The difference
between our estimates and the data reported by the
Department of Health may be due to their use of self-report
for mental health diagnoses as well as the inclusion of dif-
ferent mental health diagnoses. We only included people
with HPD and excluded disorders with greater economic
disease burden such as psychosis and bipolar disorder. It is
unclear what types of mental illness were included in the
analysis conducted by the Department of Health.
The current study estimated the total cost of lost pro-
ductivity associated with the HPD population was
AUD11.8b, which was significantly lower than AUD20b
and AUD49b estimated respectively by the ABS (2009a)
and Inder et al. (2012). This difference was likely due to
the methodological differences in estimating lost produc-
tivity. In the current analysis, only reported absence from
work and the potential loss of employment associated with
HPD were defined as productivity loss and costed. While
the method used to derive the cost of productivity loss in
ABS (2009a) was not presented, the analysis by Inder et al.
(2012) included not only absence from work and reduced
employment as loss of productivity, but also the income
difference between the general population and the popula-
tion with mental disorders. This suggests that our results
are a rather conservative estimate compared with Inder
et al. (2012). Other factors including inflation, prevalence
of the disorders, population growth, carers’ time, aware-
ness of mental health disorders and the diagnoses included
in the analysis were also likely to contribute to the differ-
ences between the estimates.
Finally, our results highlight the great economic burden
associated with anxiety disorders. While depressive disor-
ders have received considerable attention both in the scien-
tific and popular media due to the high disease burden
associated with such disorders, anxiety disorders tend to
feature far less prominently. Our study has demonstrated
that while disease burden for anxiety disorders may not be
as high as depressive disorders, the economic burden of
such disorders is considerable.
Strengths and limitations
This is the first study to our knowledge which has estimated
the costs associated with HPD in Australia based on the
2007 NSMHWB. A strength of the survey method was the
community-based sampling and diagnostic assessment
ensuring accuracy of diagnoses classification. Additionally,
this is the first study to report medication use from the 2007
survey. The economic burden from health sector and soci-
etal perspectives was able to be assessed due to the compre-
hensiveness of the data which covered not only the quantity
of healthcare resources used but also the source of payment
allowing costs borne by different sectors to be calculated.
Most importantly, as a community-based national mental
health survey with individual data available, it allowed us
to estimate the costs using a ‘bottom up’ (i.e. micro-cost-
ing) approach, which does not have the limitations associ-
ated with macro-costing (i.e. top-down costing).
However, there were also several limitations in the
analysis due to the data collection methods used in the
survey. First, it must be highlighted that the resource use
collected in the survey was based on self-report which is
associated with recall bias. Future community-based stud-
ies should also consider the addition of administrative
resource use databases such as Medicare and the PBS,
which do not have the problem of recall bias. Second is
the potential underestimation of hospitalisation costs for
private admissions. As stated in the method section, due to
the limited data on the reimbursement of hospitalisation
costs from the private sector, public admission cost was
used as a conservative proxy for the private admissions,
which might lead to the underestimation of the private
admission costs. Third, in our analysis, the HPD-
associated unemployment was estimated by the difference
of the employment rate of the general population (i.e.
including people with and without mental disorders) and
HPD population multiplied by the whole population. As
the HPD population had a higher probability of comorbid
physical restrictions, it is unclear whether the potential
loss of employment was associated with the physical
comorbidities or HPD. Hence, a conservative approach of
using the employment rate of the general population rather
than the mental disorder-free population to estimate the
incremental loss was adopted.
Finally, while the 2007 NSMHWB is the most recent
population-based survey of high prevalence disorders in
Australia, it is important to note that this survey predated
important mental healthcare reforms such as the better
access initiative which provided publically subsidised com-
munity-based psychological care. Hence, the current results
may be an underestimate of total costs or even an overesti-
mate if there has been a substitution away from hospital
Australian & New Zealand Journal of Psychiatry, 51(12)
1210 ANZJP Articles
care. An update of this survey is urgently required to answer
such questions.
Conclusion
HPD places great economic burden on Australian society
with an estimated total cost of AUD12.7b each year. In the
current analysis, we only estimate the treatment cost and
economic burden of the population with confirmed clinical
diagnoses. However, if we further consider people who do
not meet the full diagnostic criteria of such disorders, the
economic burden would be even greater.
Acknowledgements
The authors would like to sincerely thank Jessica Bucholc for
invaluable assistance in preparation of the manuscript for submis-
sion. C.M. was the senior economic evaluator of this study. She
conceptualised and developed the study design in terms of the
model framework, data sources and analytic techniques. She
supervised Y.-C.L. in all aspects of the analyses. Y.-C.L. under-
took all the empirical analyses and wrote the first draft of the cur-
rent manuscript. M.L.C. and A.M. provided expert opinion in the
method adopted in the analysis, assistance of write-up of the
report as well to the analyses. M.M. and L.K.-D.L provided assis-
tance with the statistical methods used in the analysis. All authors
read and provided input into the earlier versions of the submitted
manuscript. The authors would like to thank the Australian Bureau
of Statistics for access to the 2007 National Survey of Mental
Health and Wellbeing.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with
respect to the research, authorship and/or publication of this
article.
Funding
The current study was funded by the Health Research Fund of
Medibank Private. Medibank Private had no influence on the
methods nor the presentation of results. C.M. was funded on a
National Health and Medical Research Council (NHMRC) Early
Career Research Grant (APP1035887) during the conduct of this
study.
References
Australian Bureau of Statistics (2008a) 4326.0 – National Survey of
Mental Health and Wellbeing: Summary of results, 2007. Available
at: http://www.abs.gov.au/AUSSTATS/abs@.nsf/DetailsPage/4326.0
2007?OpenDocument (accessed 17 May 2017).
Australian Bureau of Statistics (2008b) Main features. Available at: http://
www.abs.gov.au/ausstats/abs@.nsf/mf/4326.0 (accessed 18 May 2017).
Australian Bureau of Statistics (2008c) Main features – Introduction. Available at:
http://www.abs.gov.au/AUSSTATS/abs@.nsf/Latestproducts/4326.0
Main%20Features22007?opendocument&tabname=Summary&
prodno=4326.0&issue=2007&num=&view= (accessed 17 May 2017).
Australian Bureau of Statistics (2008d) National Survey of Mental Health
and Wellbeing: Summary of Results. Canberra, ACT, Australia:
Australian Bureau of Statistics.
Australian & New Zealand Journal of Psychiatry, 51(12)
Australian Bureau of Statistics (2008e) Summary of findings of 2007 National
Survey of Mental Health and Wellbeing. Available at: http://www.abs.
gov.au/AUSSTATS/abs@.nsf/Latestproducts/4326.0Main%20Feature
s32007?opendocument&tabname=Summary&prodno=4326.0&issue
=2007&num=&view= (accessed 17 May 2017).
Australian Bureau of Statistics (2009a) 4102.0 – Australia social trends,
March 2009. Available at: http://www.abs.gov.au/AUSSTATS/abs@.nsf/
Lookup/4102.0Main+Features30March%202009 (accessed 1 September
2010).
Australian Bureau of Statistics (2009b) National Survey of Mental Health
and Wellbeing: Users’ Guide. Canberra, ACT, Australia: Australian
Bureau of Statistics.
Australian Bureau of Statistics (2012) 4433.0.55.006 – Disability and
labour force participation. Available at: http://www.abs.gov.au/aus-
stats/abs@.nsf/mf/4433.0.55.006 (accessed 25 August 2015).
Australian Bureau of Statistics (2014) 6302.0 – Average weekly earnings,
Australia, May 2014. Available at: http://www.abs.gov.au/ausstats/
abs@.nsf/detailspage/6302.0May%202014 (accessed 25 August 2015).
Australian Government Department of Health (2014) MBS Online.
Available at: http://www.mbsonline.gov.au/internet/mbsonline/pub-
lishing.nsf/Content/Home (accessed 21 October 2015).
Australian Government Department of Health (2015) Pharmaceutical
Benefits Scheme (PBS) | 2014 PBS co-payment and safety net
amounts. Available at: http://www.pbs.gov.au/info/news/2014/01/co-
payment-safety-net-amounts-update (accessed 26 June 2015).
Australian Government Department of Health and Ageing (2013) National
Mental Health Report 2013. Canberra, ACT, Australia: Australian
Government Department of Health and Ageing.
Australian Institute of Health and Welfare (2014) Expenditure on men-
tal health services. Available at: https://mhsa.aihw.gov.au/resources/
expenditure/ (accessed 25 August 2015).
Australian Institute of Health and Welfare (2015) Australian Refined
Diagnosis-Related Groups (AR-DRG) Data Cubes. Canberra, ACT,
Australia: Australian Institute of Health and Welfare.
Australian Institute of Health and Welfare (2016) Prevalence, Impact and
Burden. Canberra, ACT, Australia: Australian Institute of Health and
Welfare.
Australian Taxation Office (2015) Individual income tax rates. Available at:
https://www.ato.gov.au/Rates/Individual-income-tax-rates/ (accessed
25 August 2015).
Burgess PM, Pirkis JE, Slade TN, et al. (2009) Service use for mental
health problems: findings from the 2007 National Survey of Mental
Health and Wellbeing. Australian and New Zealand Journal of
Psychiatry 43: 615–623.
Department of Health (2011) Estimates of the prevalence of psychotic dis-
orders. Available at: http://www.health.gov.au/internet/publications/
publishing.nsf/Content/mental-pubs-p-psych10-toc~mental-pubs-p-
psych10–2#t22
Department of Health (2013) Department of Health | Indicator 1a:
Participation rates by people with mental illness of working age in
employment: general population. Available at: http://www.health.
gov.au/internet/publications/publishing.nsf/Content/mental-pubs-n-
report13-toc~mental-pubs-n-report13-3~mental-pubs-n-report13-3-
2~mental-pubs-n-report13-3-2-ind1a (accessed 2 September 2015).
Department of Human Services (2015) Medicare Australia – Statistics:
Pharmaceutical benefits schedule item statistics. Available at: http://
medicarestatistics.humanservices.gov.au/statistics/pbs_item.jsp
Drummond MF, Sculpher MJ, Torrance GW, et al. (2005) Cost analysis.
In: Drummond MF, Sculpher MJ, Torrance GW, et al. (eds) Methods
for the Economic Evaluation of Health Care Programmes. Oxford:
Oxford University Press, pp. 55–101.
Fair Work Ombudsman (2016) Annual leave. Available at: https://www.
fairwork.gov.au/leave/annual-leave
Independent Hospital Pricing Authority (2013) IHPA Explanatory Notes
for the 2013-14 Excel NWAU Calculators. Available at: https://www.
Lee et al. 1211
ihpa.gov.au/sites/g/files/net636/f/publications/nwau13-calculator-user-
guide.pdf (accessed 17 May 2017).
Inder B, Meadows G and Cornwell K (2012) Labour market costs of men-
tal illness in Australia. Available at: https://www.beyondblue.org.au/
about-us/research-projects/research-projects/labour-market-costs-of-
mental-illness-in-australia (accessed 25 August 2015).
Luppa M, Heinrich S, Angermeyer MC, et al. (2007) Cost-of-illness stud-
ies of depression: A systematic review. Journal of Affective Disorders
98: 29–43.
Mihalopoulos C, Meadows G, Stiller A, et al. (2005) Attaching unit costs
to Australia’s National Survey of Mental Health and Wellbeing. The
Journal of Mental Health Policy and Economics 8: 61–69.
Neil AL, Carr VJ, Mihalopoulos C, et al. (2013) Costs of psychosis in
2010: Findings from the second Australian National Survey of
Psychosis. Australian and New Zealand Journal of Psychiatry 48:
169–182.
Private Health Insurance Administration Council (2015) Private Health
Insurance Administration Council: Annual report 2015. Available
at: http://phiac.gov.au/wp-content/uploads/2015/02/PHIACAaust-
Dec14.pdf (accessed 21 May 2015).
Slade T, Johnston A, Oakley Browne MA, et al. (2009) 2007 National
Survey of Mental Health and Wellbeing: Methods and key findings.
Australian and New Zealand Journal of Psychiatry 43: 594–605.
Sunderland M, Newby JM and Andrews G (2012) Health anxiety in
Australia: Prevalence, comorbidity, disability and service use. British
Journal of Psychiatry 202: 56–61.
Tarricone R (2006) Cost-of-illness analysis: What room in health econom-
ics? Health Policy 77: 51–63.
Tu W and Zhou XH (1999) A Wald test comparing medical costs based on
log-normal distributions with zero valued costs. Statistics in Medicine
18: 2749–2761.
van den Hout WB (2010) The value of productivity: Human-capital versus
friction-cost method. Annals of the Rheumatic Diseases 69: i89–i91.
Whiteford HA, Degenhardt L, Rehm J, et al. (2013) Global burden of
disease attributable to mental and substance use disorders: Findings
from the Global Burden of Disease Study 2010. The Lancet 382:
1575–1586.
World Health Organization (2014) WHO | Mental disorders. Available at:
http://www.who.int/mediacentre/factsheets/fs396/en/ (accessed 10 March
2016).
Australian & New Zealand Journal of Psychiatry, 51(12)