Introduction

Mental disorders and service use across nations

Large, well-designed epidemiological studies, focussing on access and equity, can provide national prevalence estimates of psychiatric disorders, and associations of interest such as service use, hence benefiting the planning of services as well as advocacy for the needs of people suffering from mental health disorders. Recent information from the Lancet Global Mental Health Group [42] shows that annually 30% of the world’s population has some form of mental disorder, and of those people, about two-thirds receive no treatment, even in high-income countries. Similar findings have been echoed in numerous national and cross-national studies, including the European study of the epidemiology of mental disorders (ESEMeD) project in 2004 [3, 6, 16, 61]. In a study of 17 countries, Wang et al. [67] found that among the ten high income countries in the study, between 40 and 61% of individuals with severe mental disorders had received services during the previous 12-months, compared to 24–40% of individuals with moderately severe mental disorders and 13–27% of people with mildly severe mental disorders. The World Bank considers 60 countries, including Canada and Australia, to have high-income economies, defined as having gross national income (GNI) per capita in 2006 of (U.S.) $11,116 or more [71]. Thus, it can be assumed that the rates of mental health services utilisation for Canada and Australia also fall within the same ranges as found by Wang et al. [67]. The study findings suggest further research in determining if some mental health disorders would have utilised more services than others and if differences exist between countries.

Numerous studies have examined the varied rates of mental disorders found throughout the world. In their comprehensive literature review, De Girolamo and Bassi [22], noted 65 large-scale mental health surveys conducted over 25 years across 33 countries. However, they found that cross-national studies were relatively rare and most research papers documented findings from individual surveys within individual countries. Indeed, the literature shows that some studies have investigated disorder-specific prevalence among defined cultural/economic groups [52], while others have concentrated on disorder-specific prevalence within one or several regions of a country [40, 41, 44, 55]. Yet other researchers have investigated the prevalence in general or of specific mental disorders within an entire country [19, 23, 36, 51, 54, 57, 65]. With the development in 1998 of the World Health Organization (WHO) World Mental Health version of the Composite International Diagnostic Interview (WMH-CIDI), enabling cross-national comparisons, more studies have been documented recently that investigate prevalence and severity of diagnosed mental disorders in two or more countries [24, 60]. This is in accord with the recommendation from the WHO and the Organization for Economic Cooperation and Development (OECD) for more cross-national studies to be conducted comparing health-related issues [32].

Comparing Australia and Canada

Canada and Australia have many geographic, demographic, and economic similarities and differences [12, 62, 71]. With regard to mental health care, both countries have similar ratios of mental health professionals [12, 62] and each country formulated mental health policy; Canada in 1988 and Australia in 1992. Both policies address advocacy, promotion, prevention, treatment and rehabilitation.

Throughout the mid- and late-1990s, the Australian federal government actively promoted changes in mental health through strategic plans, targeted funding and work with all the States and Territories, guided by the Australian National Mental Health Policy [9, 10, 14, 15, 53]. Mental health care policy implementation in Canada was directly contrasted with that in Australia through the work of a national committee inquiry on mental health in 2006. The committee’s work included hearings on the systems in Australia, the United Kingdom, and the United States of America. It was concluded that the impetus behind change in Canada at the federal level had not been as vigorous as in Australia and hence, the Canadian system should adopt similar initiatives in facilitating the delivery of mental health care [39].

Large mental health surveys have been conducted in Canada (the Canadian Community Health Survey on Mental Health and Well Being, cycle 1.2 (CCHS 1.2) [29]) and in Australia (the Australian National Survey of Mental and Well Being (NSMHWB) [5, 68]) using versions of the Composite International Diagnostic Interview (CIDI) [73]. In Australia, a then-current version of the WHO-CIDI was employed as the survey instrument in 1997, whereas in Canada, the CCHS1.2 employed an adapted version of the revised and more comprehensive WMH-CIDI in 2002. The WMH-CIDI included assessment of the same target conditions as the earlier versions but also broader areas of assessment and allowed for expansion of the diagnostic sections to include dimensional information rather than just categorical. It had a modified question flow intended to promote episode recall and reduce opportunities to purposively influence the interview duration with strategies that would also lead to false negatives [38].

Levels of service utilisation for mental health care from the two surveys have already been reported in the literature at 10.1% in Canada [64] and 11.1% in Australia [48] but service utilisation has not been systematically compared between the two surveys. Extending results from the Australian national survey into comparisons with Canada and across a range of diagnostic groups would bring Australia further into the body of international epidemiological comparative work.

Patterns of care and co-morbidity

The importance of primary care as a location for mental health care delivery has long been recognized [26], and epidemiological work has only served to reaffirm its salience. General Practitioners (GPs) are the most commonly consulted health care providers, not only for physical problems but also for emotional complaints [43, 56]. As such, GPs constitute a key figure, with unique potential to identify and manage mental health problems existing in about a third of their patients [58, 66]. Considerable effort has been made in the United Kingdom [63], Australia [31], and Canada [21] to increase the efficacy and efficiency of mental health care in primary care settings. Given this large contribution of primary health care to the mental health care system, exploration of the epidemiology of mental health care use should include mapping patterns of care defined by the role of primary care.

Two or more psychiatric disorders can commonly occur in the same individual [25]. Anxiety and affective disorders occurring together are common [75]. Among individuals being treated for a substance dependence disorder, co-morbidity with another mental disorder occurs in 30–90% of cases [27, 28, 45]. In Edmonton, 52.9% of survey participants had more than one psychiatric disorder [18]. In Montreal, respondents with co-morbid disorders were 11 times as likely to have used services in the past year [43]. In the Australian NSMHWB, it was found that 8.2% of adults suffered from more than one mental disorder, and those with co-morbidities were significantly more likely to report mental health consultations than people with a single mental disorder [5, 50].

Summary

The literature shows that there is a need for more cross-national studies that examine the prevalence of mental disorders and utilisation of mental health services, and specifically identify which levels of the mental health care system are being utilised the most and for which co-morbidity patterns.

Aims

To investigate prevalence of mental disorders diagnosed among residents of Australia and Canada, and to examine the utilisation of mental health services in these two countries.

Methods

Design

This study involved secondary data analyses of two national surveys: the Canadian CCHS1.2 [29] and the Australian NSMHWB [5, 68]. The key information for the two national surveys, as described below, is presented in Table 1.

Table 1 Key information describing the CCHS 1.2 and the NSMHWB
Full size table

The CCHS 1.2 covered representative samples of the Canadian general population at a provincial level. The Australian NSMHWB was designed to yield national estimators of prevalence with acceptable precision down to levels of approximately 1% of the population. Both population-based surveys were cross-sectional, and collected information from countrywide general populations about common mental disorders, factors affecting mental health, and the extent and patterns of mental health care service use. They were also both conducted by national statistics offices. Both surveys used forms of the CIDI [70, 72], a diagnostic tool frequently used for large epidemiological studies of mental health, as exemplified in the ESEMeD project [2]. The two versions of the CIDI have differences but also commonalities for diagnostic criteria assessed and information collected on service utilisation so work with data from the two survey instruments presents opportunities for comparisons at common levels for service utilisation and prevalence of mental disorders. While the Australian NSMHWB survey collected data for participants 18 years and older, the Canadian CCHS 1.2 survey participants were over 15 years of age. Thus for the purpose of comparability, this study restricted data from the CCHS 1.2 survey to participants 18 years of age and older.

The CIDI versions employed in both surveys produced diagnoses following a Computer-Assisted Personal Interview (CAPI) [12, 29]. The instrument used in the NSMHWB was an adaptation of the 12-month version of the CIDI 2.1, which on one CAPI platform is also referred to as the CIDI–Auto, as developed by the WHO collaborating centre in Sydney [8]. This centre also programmed the entire NSMHWB field interview. The instrument used in the CCHS 1.2 was programmed by Statistics Canada (StatsCan) around selected diagnostic modules of the pencil and paper version of the lifetime instrument, WMH-CIDI [34, 38]. Despite the WMH-CIDI instrument’s ability to create lifetime and 12-month prevalence profiles of mental disorders in individuals, the mental health disorders studied in the CCHS 1.2 survey were selected according to the criteria that their anticipated prevalence rate for 12 months would be at least 1% [29], which was a similar design constraint to the Australian survey.

The WMH-CIDI has a number of modifications from the CIDI 2.1 intended to improve problems with underreporting in the CIDI 2.1 [73] consequent on effects such as reliance on semantic rather than episodic recall and on the participants learning to reduce interview length by denying symptoms asked about in the screening sections of the diagnostic modules. All screening sections are grouped in the early part of the WMH-CIDI rather than being at the start of modules in the CIDI 2.1. The WMH-CIDI is a substantially longer interview than the CIDI 2.1-12-Month, and following field trials, interview length constraints led to removal of several diagnostic modules in Canada.

Service utilisation data was collected with survey-specific instruments. Instrumentation was sufficiently similar to allow derivation of comparable classes of broad service use patterns although neither survey allowed for attribution of service delivery to any specific disorder.

Further details on methods of the two surveys can be found elsewhere [5, 29, 68].

Analysis

Statistical analyses of the two surveys were conducted somewhat differently. Following the Australian NSMHWB, a series of Confidentialised Unit Record Files (CURFs) was released, the final one in 2000 [11]. This CURF included refined replicate weights that enabled researchers to work on the data and calculate accurate estimates of design error without information on sampling units which were not included on confidentiality grounds. Public release of the Canadian dataset was accompanied with tables of coefficients of variation. For more accurate estimation, copies of the data including sampling unit information were made available at a number of Statistics Canada Regional Data Centres where investigators could work and use bootstrapping methods on the survey data. For the purposes of the comparisons in this paper, however, each of these methods could generate population and subgroup prevalence estimates, with associated standard errors and 95% confidence intervals. For Australia these were calculated using supplied replicate weights using STATA version 9 and for Canada they were calculated by StatsCan using the SAS software package [59].

Past-year-presence of a mental disorder and substance dependence was studied not only according to single diagnostic categories but also co-morbidity, using five co-morbidity categories: (1) major depressive disorder (MDD) only; (2) Anxiety disorders only; (3) Alcohol and/or Drug Dependence only, summarized as substance dependence only; (4) MDD and any anxiety disorder; (5) Substance dependence with any other mental disorder. These co-morbidity classifications were presented because (i) they respected the criteria set by StatsCan on the maximum value of the coefficient of variation as well as the minimum subgroup size for a given cell that can be reported; and (ii) typically in line with the design, the Australian NSMHWB had poor relative standard errors for prevalence rates below 1% and exploratory analyses showed that none of these co-morbidity classifications presented a population estimate substantially below 1%.

Mental health service utilisation in this study was defined as using any outpatient and community clinic service for mental health reasons in the 12 months prior to the interview. Past-year service use for mental health care was studied according to three health provider categories: (1) General Practitioner (GP) only, (2) GP and other health professional (prompts given here included: psychiatrist, psychologist, other physician specialist, nurse, and social worker, so some of the delivery may have been from the social and welfare sector as well as the health sector), and (3) any health professional (as above) but not including a GP.

Population and subgroup prevalence rates and standard errors were estimated and the Australian findings were standardised directly to the Canadian national population 2002. Standardisation in all analyses used the smallest age and sex subpopulation disaggregation that yielded acceptable cell sizes from the NSMHWB data. This varied with the detail of the tables, and the strata used are given as notes to each table. Cross-national comparisons were carried out by estimating event rate differences for Canadian and standardised Australian data with their 95% CI and P-values. The calculation of event rate differences relies on the difference having the property of pooled variance of the two independent samples [20].

One measure of overall severity of mental health problems, the Kessler Psychological Distress Scale (K10) [13, 30, 35] was common to both surveys and so here allows some cross-national comparisons of severity of mental disorders. The K10 has been cross-validated against diagnosis in several countries, including Canada [13].

For diagnostic categories, comparative analyses could only be performed for those categories assessed in both surveys, these being MDD (single episode or recurrent- mild, moderate or severe), panic disorder and agoraphobia (each singly or in combination), alcohol dependence, and drug dependence which included dependence on opioids, cannabis, sedatives, and amphetamines. All analyses used the DSM-IV as the chosen diagnostic classification and reported 12-month retrospective prevalence only.

Results

Prevalence rates

Australian and Canadian 12-month prevalence rates of mental disorders were compared in general and common categories of mental disorder with relevant results presented in Table 2. Anxiety disorders were significantly more common by a margin of 1.9% in Canada (4.6%) than in Australia (2.7%). Estimated rates for MDD, alcohol dependence and drug dependence were greater in Australia than in Canada but these findings did not reach conventional statistical significance, with P-values in the range of 0.1–0.2. Much of the excess of anxiety disorders found in Canada is attributable to disorders without co-morbidity from other major groups, with a significant positive difference of 1.7% for the category of people with anxiety disorders only. Apart from this group, estimates for all other co-morbidity classifications were higher in Australia than Canada, though the differences were not statistically significant.

Table 2 Comparisons of psychiatric disorders prevalence rates for Canada and Australia (Adult 18+ years, 12 month prevalence rates)
Full size table

Severity of disorders

Relative severity of disorders is not the primary focus of this paper and details of these analyses will be reported elsewhere but the issue was explored because of its possible bearing on service utilisation. Score rates on the severity bands of the K10 were calculated for Australia and Canada for the diagnostic and co-morbidity categories set out in Table 2. Some key results from this follow concentrating on co-morbidity. Across all the co-morbidity categories the estimation was of a higher percentage of Australians with ‘low distress’ (score 0–5), with all of these comparisons being significant at P < 0.05 except for the final category of co-morbidities in Table 2. For example for people with MDD only, 29.3% of Australian participants had low psychological distress compared to the Canadian 18.0% while for anxiety disorders only the rates were Australia 32.4% and Canada 21.0%. The results of the K10 measure suggest that Australia has a greater proportion of people with disorders who have low psychological distress, and so overall, lower symptomatic severity.

Mental health service utilisation rates

Table 3 presents a comparison between Canada and Australia of service utilisation rates for major categorisations of reported contacts with providers (GP only, GP & Other Health Professional, and Any Other Health Professional but GP) among adults with a mental disorder or addiction diagnosed in the last 12 months. The table suggests that perhaps more people with mental disorders utilise mental health services in Australia (48.7%) than in Canada (35.4%) but this finding falls short of conventional significance at P = 0.07. Across all categories the central estimates are of greater utilisation in Australia but statistical significance criteria are not met.

Table 3 Service utilisation for mental health care among those with mental disorders (12 month prevalence rates for Canada and Australia)
Full size table

Use of care in the context of different psychiatric morbidities

This study also examined the proportionate relationship of Canadians and Australians with specific mental health disorder patterns to the mental health services utilised. Service utilisation rates for a 12-month period are presented in Table 4 for individuals with a diagnosis of anxiety disorders, MDD, substance dependence, and two categories of co-morbid disorders. The table provides 15 comparisons of positive reports of service utilisation as defined by co-morbidity categories and major groupings of health care service providers (five co-morbidity categories x three service provided groupings). In 12 of these 15 comparisons the proportion of those from Canada with a mental disorder receiving particular services was less than the proportion of Australians. Statistical significance at P < 0.05 is met for the findings of overall greater service utilisation among both of the co-morbidity categories and marginal at P = 0.05 for anxiety disorders only.

Table 4 Comparison of service utilization for anxiety, depression, & substance dependence in Canada and Australia
Full size table

Discussion

Twelve month prevalence rates of mental disorders

To facilitate comparison in this study, DSM-IV categories were utilised for all analyses. The prevalence rates generated by the DSM-IV classification, however, tend to be lower than in the ICD-10, which has been more commonly used in reporting outcomes from the Australian survey [4]. For example, the Australian NSMHWB one-year prevalence for all ICD-10 anxiety disorders was typically reported as 9.7% [5, 46]. When the same data was analysed so as to be comparable with the CCHS 1.2, this fell dramatically to 2.8%, as it included both fewer disorders and differences between ICD-10 and the DSM-IV. After attention to these aspects of comparability the combined 12-month prevalence rates for all mental disorders that were examined in this study differ slightly between the two countries with Canada having a lower overall rate of 10.2% compared to the Australian rate of 11.5%. The analysis suggests that Canadians have higher rates of anxiety disorders and, more tentatively, perhaps lower rates of MDD and substance misuse.

Possible causes for apparent differences

Bias

Non response rates were comparable for the two surveys (CCHS1.2 23%, NSMHWB 22%) so although prevalence may be higher among non responders [37], this is not necessarily a source of systematic error in the comparisons. Age and sex standardisation renders demographic differences an unlikely source of error. While the different questionnaires were carefully compared and all efforts made to attain consistency in the ways the data sets would be analysed, it was not possible to control for bias arising from interviewer training techniques or for the differences in design of the two versions of the CIDI. We might comment though that the redesign between the CIDI 2.1 and the WMH-CIDI was intended in part to reduce the capacity of participants to learn to manipulate and skip rules to shorten the interview. If they had been successful, then we should expect that the WMH-CIDI might yield higher prevalence figures for the disorders that were placed later in the CIDI 2.1 interview where the opportunity for learning how to manipulate the interview was greater. Anxiety disorders were assessed before affective disorders in the CIDI 2.1 as used in the NSMHWB so the directionality of comparisons is not what would be expected if this were the sole source of the differences found.

Actual differences between the populations studied

The differences detected may reflect secular trends in prevalence rates. Data for the Australian study was collected in 1997; the Canadian study was conducted in 2002. We might speculate that the political and societal context of life in North America in 2002, including for instance a changed security situation, might have included stressors that could promote anxiety disorders in those with vulnerability to these problems that were not present in Australia in 1997. Alcohol consumption in Australia is higher than in Canada (2004 data: Australia 9.2 l per capita; Canada 8.3 l per capita [74]). The findings in Table 2 would be compatible with the tentative speculation that some of the apparent excess of MDD in Australia arises because people with chronic anxiety disorders are regularly employing alcohol in a maladaptive coping strategy that induces MDD.

Service use amongst those with diagnosed disorders

The majority of people in both Canada and Australia diagnosed as having had a mental disorder in the past 12 months did not receive mental health interventions. According to the CCHS 1.2, 64.6% of Canadians who had been diagnosed as having a mental disorder in the 12 months prior to the survey being conducted had not received a mental health consultation. The equivalent figure for Australia derived from the NSMHWB was 53.8%. These figures support the assertion by the Lancet Global Mental Health Group [42] regarding the utilisation of services for mental health care.

Although in both countries most people with mental health problems did not receive care, the estimates suggest that more Australians with mental disorders received mental health services than did Canadians. This applies particularly to those with co-morbidities and with anxiety disorders, though the findings are compatible with the possibility that this is true across other categories too. One possible explanation of this would be that both countries have comparable proportionality of service response but that disorders are overall more severe in Australia. However this is not supported by the analyses of the K10 data which would point towards lower severity in Australia.

For Australian mental health workers and policy makers, it would be reassuring if higher rates of service utilisation found amongst Australians with mental disorders resulted from the extensive efforts that have been made within Australia to prevent, treat and educate the public about mental health.

Differences in service use by different categories of mental health disorders

It was found that health care service use varies by diagnostic category in both Canada and Australia. This has often been the finding when mental health service use has been examined [7, 17, 46, 47]. We also found that those with co-morbidity were more likely to receive some form of mental health care, replicating the finding of others [6].

In Australia and Canada service use within the diagnostic co-morbidity categories examined in this study was most frequent for people with MDD with any anxiety disorder(s), declining in order through the categories of: substance dependence and any other disorder, MDD only, anxiety disorder(s) only, then substance dependence only. In both countries those with substance dependence disorder were the least likely to have received mental health services. Only 11.5% of Canadians and 21.7% of Australians who had substance dependence disorders had received such services. In both countries, individuals that had both MDD and anxiety (66.3% of Canadians and 86.1% of Australians) were the most likely to have received mental health services.

Variations in service use by disorder categories have been examined in a number of large mental health epidemiological studies [1, 3, 7, 17, 33, 69]. In all of the identified studies where some category of substance dependence disorder (i.e. alcohol, substance, or a combination of these) has been examined (with only one exception [33]), individuals with these disorders are the least likely to have received mental health services. In these studies it was also usually found that individuals with co-morbidity were the most likely to have received services. Furthermore, a higher proportion of those with affective disorders had received services than had those with anxiety disorders. There seems to be a great deal of similarity in the variation in service use by disorder categories across a number of countries.

Limitations of the study

Some possible sources of bias have been discussed above and the comparisons are not fully conclusive. We note that the service activity data collection was not standardized, though our close examination concluded that both surveys collected service utilization data for the last year with reasonably comparable items used. Although reasonably comparable at face estimation service use instrumentation was not standardized and the method does not permit certain identification of the target disorder for services delivered. Service use itself could be a confounder for the K10 analyses given the K10 has a shorter timeframe than the 12-month prevalence as derived from the CIDI. Inferences are possible from the Australian survey regarding adequacy of care received based on assessment of levels of met and unmet perceived need [46, 49, 50], but such instrumentation was not included in the CCHS 1.2 so comparison was not possible across these dimensions.

Conclusion

The analyses presented show significant differences in prevalence of anxiety disorders. Consistently, Australia presents a more favourable picture for access to mental health care than does Canada, though still most people with mental health problems do not receive professional help with them. Based on the fact that Australia has been implementing national policy more vigorously than Canada, the results could be seen to reflect favourably on these efforts.