Participants

Participants were from a panel of 200 people drawn from the UK general population recruited using direct mailshots and advertisements in local newspapers as described in a previous study. ^{Reference McGrouther28}

Instruments

The five-item Attitude to Mental Illness Questionnaire (AMIQ) is a brief, self-completion questionnaire. ^{Reference Atkinson, Atkinson, Smith, Bem and Nolen-Hoeksema27,Reference Till and Busler29} Respondents read a short vignette describing an imaginary person and answered five questions (Appendix). Individual questions were scored on a five-point Likert scale (maximum +2, minimum −2) with blank questions, ‘neutral’ and ‘don't know’ scored zero. The total score for each vignette ranged between −10 and +10. The AMIQ has been shown to have good psychometric properties in a sample of over 800 members of the UK general public (one component accounted for 80.2% of the variance; test-retest reliability was 0.702 (Pearson's correlation coefficient); alternate test-reliability v. Corrigan's attribution questionnaire was 0.704 (Spearman's rank correlation rho); Cronbach's alpha was 0.93). ^{Reference McGrouther28} Further details of the AMIQ are included in the Discussion.

Procedure

Participants were block randomised using the randomisation function of the Stats Direct Statistical Package (Version 2.4 for Windows Excel, StatsDirect, www.statsdirect.com). The control group were asked to complete the AMIQ with the following description: ‘Oliver has Down syndrome. He is 32 years old and lives with his parents. He cannot read or write but he is happy and cheerful and keen to help people’. The experimental group were asked to look at a picture of a young man with the physical appearance of Down syndrome wearing a shirt and tie in an office (the picture was a model from a commercial photo-image gallery, see cover, this issue). Participants were then asked to complete the AMIQ. Six months later the experiment was repeated but all participants were crossed between experimental and control group.

Data analysis

Randomisation, correlation coefficients and non-parametric (Mann-Whitney) tests were used to generate and compare differences in subgroups using the StatsDirect statistical computer software package (Version 2.4). Arithmetic mean scores are used to summarise the ordinal data (Likert scale scores) for convenience according to common statistical practice, although analysis involves use of the median scores in the Mann-Whitney tests.

Demographic data

Overall, 26% of respondents were male; the mean age was 47.9 years (s.e. = 1.5); 55% of participants were in paid employment, 17% were retired; 92% described their ethnic group as White British.

Completed questionnaires were received for 360 people (response rate 93% at baseline; 87% at 6-month crossover). The mean AMIQ score in the control groups was 1.56 (s.d. = 2.85, s.e. = 0.21, n = 186). The mean score in the experiment group (after looking at the pictures) was 2.43 (s.d. = 2.59, s.e. = 0.12, n = 174, median difference 1; P = 0.0016 Mann-Whitney U-test, effect size 0.23). The order of exposure (control or experimental) had no significant effect on the outcome (there was no significant difference between outcomes when the control groups were compared before and after the crossover, P = 0.87).

Strengths and limitations

The AMIQ was used in this project as it is convenient and has been well-validated. ^{Reference McGrouther28-Reference Wilmshurst and Mackay30} Other instruments are available, although these tend to be much longer, involve interviews or tend to address the experience of stigma by people with mental illness. ^{Reference McGrouther28,Reference Kahle and Kim31} The AMIQ was designed as a brief instrument to measure stigmatised attitude in members of the general public. It draws from Corrigan's attribution model of public discrimination that proposes that people have negative attitudes towards the mentally ill because people with a mental illness have responsibility for their illness and are more blameworthy than people with a physical illness. ^{Reference Corrigan, Markowitz, Watson, Rowan and Kubiak4} Other features identified in Corrigan's model that may distinguish between the perception of mental and physical illness include their tendency to illicit emotional responses such as pity, anger, fear, or helping/rejecting responses. The AMIQ was validated in a postal survey of a sample of 1079 UK adults. The AMIQ is a self-report questionnaire with five-point Likert scale responses originally validated in response to short fictional vignettes. The AMIQ showed highly stigmatised attitudes to people with addictive disorders but more positive attitudes to those with depression or self-harm. ^{Reference McGrouther28} Results from a smaller follow-up sample showed that attitudes towards people with alcohol dependence and schizophrenia were intermediate. It has good test-retest reliability as well as face and construct validity. The AMIQ was also compared with the 21-item Corrigan's attributions questionnaire that has been validated as a measure of stigmatised attitudes towards people with mental illness. ^{Reference Corrigan, Markowitz, Watson, Rowan and Kubiak4,Reference McGrouther28,Reference Patzer32} Fifty participants received the AMIQ and attributions questionnaire and were asked to complete both in relation to the schizophrenia vignette. The attribution questionnaire was scored from 0 to a maximum of 64, with a mean score of 38.7 (s.e. = 1.8). Kendall's tau b = 0.563 (P<0.001) and Spearman's rank correlation rho = 0.704 (P<0.001) indicated good alternative test reliability.

Although there was an excess of female respondents, age and employment status of participants were reasonably matched to that from UK census surveys. Hence the sample appears to be a reasonable cross-section of the British public. However, it is self-selecting and may not generalise across the whole population. Ideally, interviews could be conducted using a quota survey of households with repeat visits for non-responders. ^{Reference Crisp, Gelder, Rix, Meltzer and Rowlands6} Unfortunately, this is prohibitively expensive.

The crossover trial took place after 6 months - a significant period of time which meant it was unlikely that the effects of the first phase would still be present. Furthermore, there was no significant difference between outcomes when the control groups were compared before and after the crossover. The study presented a hypothetical person with intellectual disability - the picture was a model from a commercial photo-image gallery. The picture was of a young man with Down syndrome wearing a suit. It is possible that participants were responding positively towards features other than his dysmorphic facial appearance such as the suit, background or his cheerful expression. Nevertheless, this report does indicate that including a photograph of a dysmorphic person with an intellectual disability can elicit a positive response and, as such, people with intellectual disabilities and their advocates should be aware that promoting a photograph of a dysmorphic person may be advantageous.

The results rely on a hypothetic individual and situation. This is less accurate than real experience - it was not possible to measure stigmatised behaviour towards real people with intellectual disabilities. Moreover, the written views and expressed attitudes may not translate into any enduring behavioural change. Although there was no direct contact between participants and researchers, participants are likely to make some assumptions about the potentially liberal beliefs of researchers. Hence social desirability bias may affect the results. However, the results from other similar studies show a negative view of people with active substance use disorder and suggest that participants had little reservation about indicating their disapproval of these disorders. ^{Reference McGrouther28,Reference Wilmshurst and Mackay30} This is confirmed in other reports. ^{Reference Lin and Yeh33} This would indicate that social desirability bias had only a modest effect. Furthermore, social desirability bias would affect both the experimental and control groups equally.

Main findings

The impact of visual imagery has been widely accepted in marketing since the 1920s: indeed the expression, ‘a picture paints a thousand words’ was popularised by Frederick R. Barnard in a marketing journal. ^{Reference Luty, Fakuda, Umoh and Gallagher34} Stigmatised attitudes are widely reported towards people as a result of disfigurement or dysmorphic facial appearance. ^{Reference Houston and Bull18-Reference Strauss, Ramsey, Edwards, Topolski, Kapp-Simon and Thomas20,Reference Patzer32} Hence the researchers, wrongly, assumed that the picture of a man with Down syndrome would produce a more negative stigmatised attitude than simply asking participants to complete the scoring instrument without viewing any pictures. In fact, looking at a picture of the young man with Down syndrome actually reduced stigmatised attitudes by a small but statistically significant extent. These results are in contrast to most marketing strategies that invariably recommend associating a product with a successful, physically attractive individual. ^{Reference Kaney, Mason, Carlisle, Watkins and Whitehead21-Reference Atkinson, Atkinson, Smith, Bem and Nolen-Hoeksema27} Hence looking at pictures of an individual may actually personalise the description and evoke a more sympathetic outcome than simply asking participants to rate a hypothetical person based purely on a written description.

Although the difference was small (effect size 0.23), it was highly statistically significant. Wolff and colleagues ^{Reference Cunningham, Sobell and Chow35} reported a labour-intensive controlled study of the effect of a public education campaign on community attitudes towards people with mental illness. It produced effects sizes of 1.23 for fear and exclusion, 1.22 for social control and 0.66 for goodwill. The scores for both groups were less than 3 on the AMIQ scale (range −10 to +10). This is significantly less than the scores reported for someone with diabetes or a practising Christian both of whom consistently score above 5. ^{Reference McGrouther28} However, the score of 2.43 was almost identical to that previously reported towards a fictitious individual who had taken an overdose while depressed (2.35). Presenting a picture and a description of a man with an intellectual disability was the least stigmatising presentation of mental illness so far reported using the AMIQ. Hence, although there is likely to be a significant degree of stigma associated with intellectual disability, presenting members of the public with pictures of people with intellectual disabilities significantly reduced stigmatised attitudes.

Methods to reduce the stigma of mental illness

The Royal College of Psychiatrists' Changing Minds campaign aimed to promote positive images of mental illness, challenge misrepresentations and discrimination and educate the public to the real nature and treatability of mental disorder. ^{Reference Crisp, Gelder, Rix, Meltzer and Rowlands6} Crisp et al's large survey showed that people with substance use disorders were the most stigmatised of all those with mental disorder. ^{Reference Lin and Yeh33} More recently, national anti-stigma campaigns have been launched in Scotland and England (e.g. www.seemescotland.org.uk/ and www.time-to-change.org.uk). Unfortunately, there have been recent reports that national anti-stigma campaigns are not particularly effective. ^{Reference Luty, Umoh, Sessay and Sarkhel13,Reference Luty, Rao, Arokiadass, Easow and Sarkhel36,Reference Luty, Umoh and Nuamah37} These reports discuss the disappointing results to date from the Defeat Depression campaign, the Changing Minds campaign and the Scottish See Me campaign. The English Time to Change campaign is sponsored by the National Lottery and mental health charities including MIND and Rethink. Although a significant amount of work has been carried out on the stigmatising effect of a diagnosis of schizophrenia, very little research has been done on stigmatised attitudes directed against people with intellectual disability. ^{Reference Sperlinger, O'Hara and Sperlinger8,Reference Luty, Umoh, Sessay and Sarkhel13,14,Reference Corrigan and Penn38} For example, the stigma of intellectual disability is not addressed by either the Changing Minds or Time to Change campaigns or the UK government publication Action on Mental Health. ¹ Methods of dealing with stigma are not addressed in the UK Department of Health report, Valuing People, which specifically concerns people with intellectual disability. ¹⁷

Action on Mental Health - A Guide to Promoting Social Inclusion provides 12 individual factsheets to reduce stigma. ¹ This supplements the efforts of the Changing Minds campaign. Both give practical advice to health agencies, employers and stakeholders to tackle stigma. Providing factual information in brief factsheets ^{Reference Crisp, Gelder, Goddard and Meltzer39,Reference Barnard40} or through extensive interventions, such as educational courses, has been reported to reduce stigma. ^{Reference Patzer32,Reference Wolff, Pathare, Craig and Leff41,Reference Paykel, Hart and Priest42} Unfortunately, responses tend to be small, especially if the negative consequences of mental illness are also disseminated. Knox et al showed that addressing stigmatised attitudes to mental illness among 4 million members of the US armed forces with mandatory training to recognise and treat mental illness significantly reduced suicide rates but not stigmatised attitudes. ^{Reference Mehta, Kassam, Leese, Butler and Thornicroft43} Pinfold et al reported a project in which 472 English secondary school children attended mental health awareness workshops. ^{Reference Penn, Kommana, Mansfield and Link44} Overall, there was a small but positive shift in their understanding of mental illness. However, it was possible in both these settings to insist that participants engaged in the anti-stigma training, whereas involvement of the general public is entirely voluntary.

Penn et al reported a study of 163 US undergraduates who were assigned randomly to four groups: three watched a documentary - about schizophrenia (represented realistically), polar bears or being overweight - and the fourth was a ‘no video’ control group. ^{Reference Penn, Guynan, Daily, Spaulding, Garbin and Sullivan45} The schizophrenia documentary did not change attitudes. Depicting the negative consequences of schizophrenia may be realistic but may not be the best way to reduce stigma. Depicting a success story may be more effective, although viewers may then classify this as an exception to the rule. ^{Reference Patzer32} By contrast, a study showed that presenting recovered patients in a positive manner can significantly reduce the stigma of alcoholism, although the effects on schizophrenia were negligible. ^{Reference Wilmshurst and Mackay30} Luty et al also used a brief face-to-face intervention (motivational interviewing) to reduce the stigma of alcoholism, although the effect was modest. ^{Reference Kahle and Kim31}

Promoting direct interpersonal contact with people who are mentally ill may be an effective strategy, but the amount of contact required remains unknown. ^{Reference Patzer32,Reference Cunningham, Sobell and Chow35,Reference Crisp, Gelder, Goddard and Meltzer39,Reference Penn, Kommana, Mansfield and Link44} It would be difficult, in practice, to ensure that a significant proportion of the public had contact with people with a severe mental illness. Our report indicates that looking at a picture of a dysmorphic man does reduce stigmatised attitudes significantly and this may act as an effective substitute for direct contact.

Attitudes to Mental Illness Questionnaire (AMIQ) ^{Reference McGrouther28,Reference Till and Busler29}

This is a fictitious report: ‘Oliver has Down syndrome. He is 32 years old and lives with his parents. He cannot read or write but he is happy and cheerful and keen to help people.’

1. 1 Do you think that this would damage Oliver's career?

   Strongly agree^-2/Agree^-1/Neutral⁰/Disagree⁺¹/Strongly disagree⁺²/Don't know⁰

2. 2 I would be comfortable if Oliver was my colleague at work?

   Strongly agree⁺²/Agree⁺¹/Neutral⁰/Disagree^-1/Strongly disagree^-2/Don't know⁰

3. 3 I would be comfortable about inviting Oliver to a dinner party?

   Strongly agree⁺² /Agree⁺¹ /Neutral⁰ /Disagree^-1/Strongly disagree^-2 /Don't know⁰

4. 4 How likely do you think it would be for Oliver's wife to leave him?

   Very likely^-2/Quite likely^-1/Neutral⁰/Unlikely⁺¹/Very unlikely⁺²/Don't know⁰

5. 5 How likely do you think it would be for Oliver to get in trouble with the law?

   Very likely^-2/Quite likely^-1/Neutral⁰/Unlikely⁺¹/Very unlikely⁺²/Don't know⁰