The yield from national surveys of mental health

Abstract

At least 21 countries have now carried out national surveys of mental health under the aegis of the World Health Organization’s World Mental Health Surveys. This has meant interviewing some 157 000 people in their homes. The countries are as varied as Australia, China, Iran, six continental European nations, Nigeria, the UK and the USA (Andrews et al, 2001; Demyttenaere et al, 2004; Mohammadi et al, 2005). It is therefore timely to consider what this very large body of information has yielded and to what use it can be put, especially in relation to the costs and human resources expended in a field where unmet need is so conspicuous.

Methods

The surveys have many attributes in common. They were all undertaken to inform health policy. Most used the same standardised interview with the same (DSM–IV) diagnostic criteria (American Psychiatric Association, 1994) for the common mental disorders. The latter are anxiety, depressive disorders and alcohol or substance misuse. All surveys were conducted by lay interviewers and the data are derived solely from self-report. The morbidity described is categorical (present or absent) rather than dimensional, and refers to symptoms both in the previous 12 months and across the respondent’s lifetime. Independent variables have usually been confined to age, gender, marital status and indicators of socio-economic status. Most surveys covered the age range 18–65 years, and only a minority included children or the elderly. Most included measures of disability and recent health service use.

Findings

There have been some consistent findings. The total prevalence rates for adults have been much higher than might be expected, with the median 12-month prevalence for all disorders being 12.2%. The range, however, is considerable, from 4.3% in Shanghai to 26.4% in the USA. Anxiety and depressive disorders are more common in women and substance use disorders are more common in men. The median age at first onset of anxiety disorders is 11 years, of substance use disorders 20 years, and of mood disorders 30 years. Mental disorders are therefore disorders that begin in the young. The burden of disability, in terms of the number of days lost from work or family life, is very much greater than might have been expected and yet the majority of sufferers received no treatment. This is so in higher-income as well as lower-income countries, even though inexpensive and effective treatments exist.

Response rates ranged greatly, from 51% in Belgium to 88% in Colombia (Demyttenaere et al, 2004). Some of the differences in total prevalence may also depend on the number of disorders included in the interview. But there are some striking differences in the prevalence estimates of individual disorders. For example, the 12-month prevalence of depressive disorder is 9.6% in the USA, 6.6% in Lebanon (conducted in 2002–03), 3.1% in Japan, 1.7% in Shanghai and 0.8% in Nigeria. The survey in Iran (Mohammadi et al, 2005) reported lifetime rather than 12-month prevalence, which was only 4.3% for depressive disorder. When the US 1991–92 national survey was repeated in 2001–02, the prevalence of depressive disorder had increased from 3.3% to 7.1% and treatment rates from 12% to 20% (Compton et al, 2006).

The value of the surveys

What does this massive body of data mean? It shows that psychiatric morbidity is common, although it would be wrong to conclude that a 1-year prevalence rate of 12% means that one in eight citizens of a country wish for or need treatment. Where the investigators have included a measure of disability, this has allowed a better estimate of unmet need. It has shown that the economic and social burden arising from mental disorders is highly significant for health policy. These two findings, the high prevalence and the proportion of all disability that comes from mental illness, have been the most influential products of the surveys. They have been noted by senior administrators and politicians, often accompanied by an increased allocation of funds for services and research. The increased prevalence of major depression observed in the USA shows the value of repeating surveys to monitor the mental health of a nation.

By contrast, the scientific advance in understanding the causes of these mental disorders has been slight. The gender differences in anxiety and depressive symptoms were already known. The data on age of first onset in community samples are new, but, that apart, no new hypotheses on aetiology of any fundamental significance have been generated. The psychoses, which are so disabling, are too low in prevalence for useful data on their aetiology to be obtained from a cross-sectional survey in the community. For aetiological research, data on environmental risk factors and temperament could have been included, but it has proved difficult to persuade funding bodies to include these measures to inform aetiology when what they want are answers to three questions:

• How many people have which disorder?

• How disabled are they?

• What services do they need and want?

As a result, new information on causation has been sparse, despite the massive sample size.

It is always tempting to compare prevalence rates between countries. Where differences are observed, as in the above examples, it is tempting to put them down to intrinsic attributes of the population, such as lifestyle or social cohesion. However, the estimates must inevitably be affected by differences in method, such as the response rates in each survey, the use of different instruments or versions thereof, and the readiness of respondents in different countries to acknowledge the presence of psychological or physical symptoms. Also, it may be a mistake to assume the diagnostic criteria are equally applicable across countries. Four decades ago, in his influential paper ‘Are international comparisons timely?’, Kessel (1965) concluded that they were not and that epidemiological work on the aetiology of mental disorders would be better directed to within-country studies. We believe the situation remains the same today. It has to be concluded that the massive effort to obtain data on the mental health of nations, involving over 150 000 respondents, has contributed little to the understanding of the causes of mental illness.

Future surveys

It is likely that some countries will repeat their national surveys and others will conduct their first. The justification must be largely to keep mental health on the political agenda and to monitor changes in service use. The use of standardised interviews, matched to ICD–10 (World Health Organization, 1992) and DSM–IV classifications (American Psychiatric Association, 1994) or their successors, will continue, because rates for such categorical diagnoses seem to have an administrative impact.

The scientific value of this categorical approach is less certain. Much greater scientific value will come from the analysis of dimensional scales such as the Kessler Psychological Distress Scale (K–10; Kessler et al, 2002) and 12-item General Health Questionnaire (GHQ–12; Goldberg & Williams, 1988). Unlike the diagnostic interview, which can take over an hour, these take only a few minutes to administer, even for persons with limited literacy. The scales cover the full range of symptoms of epidemiological interest and they detect the sub-clinical level of morbidity that is of such relevance to disease burden. Their continuous scores avoid the loss of statistical information that comes with categorical diagnoses. They are also less open to the biases that make international comparisons of prevalence rates so difficult to interpret. As Rose (1993) emphasised, the distribution of scores can be used to characterise whole populations. The K-10 has been included in many of the surveys, but has not yet been used to test aetiological hypotheses.

Further large-scale epidemiology needs to contribute to knowledge of aetiology. For example, in addition to obtaining psychiatric measures and exposures to adversity, obtaining genetic data is now feasible. This would allow the study of interactions between genotype and environment in very large samples. National surveys of mental health are major epidemiological undertakings. Their yield can now be expanded.