Sanabria‑Mazo et al. Systematic Reviews (2023) 12:66 Systematic Reviews https://doi.org/10.1186/s13643-023-02230-4 RESEARCH Open Access Over 40 years (1981–2023) assessing stigma with the Community Attitudes to Mental Illness (CAMI) scale: a systematic review of its psychometric properties Juan P. Sanabria‑Mazo1,2,3, Eduardo Doval4, Albert Bernadàs1,3, Natalia Angarita‑Osorio5, Ariadna Colomer‑Carbonell1,2,3, Sara Evans‑Lacko6, Graham Thornicroft7, Juan V. Luciano1,2,8*    and María Rubio‑Valera1,2 Abstract Background The Community Attitudes to Mental Illness (CAMI) scale measures social stigma towards people with mental illness. Although it has been used worldwide, the psychometric properties of the CAMI have not been system‑ atically reviewed. The main aim of this study was to systematically review the psychometric properties of the different versions of the CAMI more than 40 years after of its publication. Methods A systematic search was conducted in MEDLINE, PsycINFO, Web of Science, and EMBASE from 1981 (year of publication) to 2023 (present). A double review was performed for eligibility, data extraction, and quality assessment. Results A total of 15 studies enrolling 10,841 participants were included. The most frequently reported factor struc‑ ture comprises 3 or 4 factors. Overall, the internal consistency seems adequate for the global scale (α ≥ 0.80), except for CAMI-10 (α = 0.69). Internal consistency of the subscales are not supported, with authoritarianism being the weak‑ est factor (α = 0.27 to 0.68). The stability over time of the total scale has been assessed in the CAMI-40, CAMI-BR, and CAMI-10 (r ≥ 0.39). Few studies have assessed the temporal stability of the CAMI subscales. Most of the correlations with potentially related measures are significant and in the expected direction. Conclusions The 3 and 4 factor structure are the most widely reported in the different versions of the CAMI. Even though reliability and construct validity are acceptable, further item refinement by international consensus seems warranted more than 40 years after the original publication. Systematic review registration PROSPERO identification number: CRD42018098956. Keywords Stigma, Mental health, Community attitudes to mental illness, Psychometric properties, Systematic review *Correspondence: Juan V. Luciano juanvicente.luciano@uab.cat Full list of author information is available at the end of the article © The Author(s) 2023. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://​creat​iveco​mmons.​org/​licen​ses/​by/4.​0/. The Creative Commons Public Domain Dedication waiver (http://​creat​iveco​ mmons.​org/​publi​cdoma​in/​zero/1.​0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Sanabria‑Mazo et al. Systematic Reviews (2023) 12:66 Page 2 of 19 Introduction disagree”. According to its developers, the CAMI con- Stigma towards people with mental disorders is a socio- tains four subscales: authoritarianism, benevolence, cultural phenomenon [1, 2] that negatively affects qual- social restrictiveness, and community mental health ide- ity of life, self-esteem, interpersonal relationships, health ology [23]. Each subscale contains 10 items (5 positively care seeking and provision, and workplace integration formulated plus 5 negatively formulated) on the opinions [3, 4]. In Europe, it is estimated that there are about 165 of treating and caring for people with a serious mental million people with mental disorders, and it is calculated disorder. Therefore, subscale scores can range from 10 to that around 38% of people will experience a mental dis- 50, with higher scores indicating less stigma towards peo- order in their lifetime [5]. A recent systematic review ple with mental disorders. confirmed that mental disorders cause a substantial eco- Given that the CAMI has been available for more than nomic burden for societies, with developmental disor- four decades, with hundreds of citations, the time is right ders, schizophrenia, and intellectual disabilities obtaining for a systematic review of the psychometric properties the top median societal cost per patient [6]. of its different versions. As far as it is known, there are Although there is not a universally accepted defini- no previous reviews summarizing available psychomet- tion of stigma, it can be considered a multidimensional ric information on the CAMI. This systematic review construct composed of negative elements of knowledge, bridges this gap by synthesizing and critically appraising attitudes, feelings, beliefs, and behaviours towards a the psychometric properties of this stigma scale. group of people [7]. Stigma is a powerful barrier to social participation and professional help-seeking for people Method with mental disorders and there is a widespread social Protocol and registration belief that such people are aggressive and uncontrollable This systematic review was conducted according to the [8–12]. In the last decade, various countries have imple- Preferred Reporting Items for Systematic Reviews and mented antidiscrimination campaigns to reduce stigma Meta-analyses guidelines (PRISMA) [26]. The review and improve the integration of people with mental illness protocol was registered in Prospective Register of Sys- into communities [13]. Examples of such programmes tematic Reviews (PROSPERO) on July 25th, 2018, under are Time to Change in England [14]; Obertament in Spain identification number: CRD42018098956. [15]; Schizophrenia has many faces in Austria [16]; Like minds like mine in New Zealand [17]; One of us in Den- Search strategy mark [18]; and Opening minds in Canada [19]. Searches were conducted through four electronic data- There is a long history of scales developed to assess atti- bases: MEDLINE (PubMed), PsycINFO (ProQuest), Web tudes towards mental illness. The Opinion about Mental of Science (Core Collection), and EMBASE (Elsevier). Illness (OMI) [20] and the Custodial Mental Illness Ideol- The search strategy included terms related to psychomet- ogy Scale (CMI) [21] were developed in the 1950s–1960s rics (psychometrics OR factor analysis OR reliability OR as the first scales to measure stigma. More recently, the intra-class OR test–retest OR internal consistency OR Community Mental Health Ideology (CMHI) [22], the validity OR dimensionality OR sensitivity to change OR Community Attitudes to Mental Illness Scale (CAMI) responsiveness OR sensibility OR specificity) and to the [23], the Mental Health Knowledge Schedule (MAKS) original scale name ((attitude* AND toward* AND men- [24], and the California Assessment of Stigma Change tal* AND Ill*) OR ("CAMI")), found using keywords in all (CASC) [25] were designed. Since its publication in 1981, fields and in the Medical Subject Headings (MeSH). The the CAMI [23] has been the gold standard measure for search string used in MEDLINE (PubMed) is shown in assessing stigma towards people with mental disorders. the Supplementary Table S1. Limits and filters were not It has been translated into several languages (Spanish, activated in any of the database searches to avoid loss Italian, Swedish, Portuguese, Greek, and Persian, among of potential eligible studies. The references of included others) and used to measure stigma in a wide variety of studies were screened by reverse citation search to iden- samples (e.g., nurses, psychiatrists, and relatives of psy- tify studies not detected in the electronic searches. chiatric patients). The original version of the CAMI was partially derived Eligibility criteria from a brief, revised, and updated version of the OMI The search in the databases incorporated studies pub- [20], and it was initially developed to predict the reac- lished in peer-reviewed journals from 1981 (when the tions of the general population to local services for peo- original version was published [23]) to February 28th, ple with severe mental disorders. This original version is 2023 (present). This systematic review included all studies composed of 40 items that are responded on a 5-point that provided evidence on the psychometric properties Likert scale, ranging from “strongly agree” to “strongly (content validity, factor structure, internal consistency, Sanabria‑Mazo et al. Systematic Reviews (2023) 12:66 Page 3 of 19 test–retest reliability, construct validity, floor/ceiling tive of the questionnaire and the target population effects, and interpretability) of the different versions of were explicitly defined. For this criterion to be met, the CAMI. No restrictions were placed on the character- it was necessary to develop the questionnaire items istics of the participants and the type of sampling used in in consensus with the general population and stigma the search. To ensure the rigour of the included studies, experts [27]. A score of 1 was given if some of the non-original studies (reviews, books, doctoral disserta- aspects mentioned above were missing and a score of tions, commentaries, conference abstracts, study proto- 0 if none of the above information was described. cols, case reports, and qualitative studies, among others) 2. Factor structure refers to the dimensionality of the and grey literature (i.e., non-peer-reviewed manuscripts) scale [27]. A score of 2 was given if an exploratory were excluded. Non-English, non-Spanish, or non-Italian factor analysis (EFA) and a confirmatory factor analy- papers were also excluded. sis (CFA) had been performed on different samples or if the CFA had been calculated considering a theo- Data management and study selection retical model. This score was only given if the factor In the first phase, duplicate articles in the databases analyses supported the structure promoted by the were removed using Mendeley. In the second phase, two authors. A score of 1 was awarded if only the EFA reviewers (AB and NA–O) independently assessed the had been carried out, and if the EFA supported the articles based on their title and abstract according to the factor structure. A score of 0 was awarded if factor eligibility criteria. In the third phase, the full text of those analysis has not been conducted or if EFA or CFA articles that met the second phase was reviewed to ver- does not support the proposed dimensionality. ify compliance with the eligibility criteria. In the fourth 3. Internal consistency is used to indicate the degree of phase, discrepancies in study selection were resolved reliability of a scale. For health scales, Cronbach’s α with the help of two additional external reviewers should be between 0.70 and 0.95 [27]. For greater rig- (JVL and MR-V). In the fifth phase, relevant data were our in the findings, the Nunnally and Bernstein [28] extracted from the selected documents with a stand- criteria were used in this study, which present scores ardised data extraction form and the respective quality above 0.80 as acceptable. A score of 2 was given if assessment was carried out for each study. Cronbach’s α was calculated for each dimension and if it was between 0.80 and 0.95. A score of 1 was Data extraction assigned if internal consistence was calculated only Data extraction from the selected articles was performed for some dimensions and if it was below 0.80. A score independently by two reviewers (AB and NA–O), using of 0 was reported if no internal consistency informa- a template containing the following sections: authors, tion was found or if the evidence was questionable. year of publication, country, CAMI version, study design, 4. Test–retest reliability is a measure used to vali- target population, sample type, sample size, age, gender, date the stability of the scale over time. For accept- results depending on sociodemographic variables, and able temporal stability, the test–retest needs to be psychometric results about CAMI. The authors of the at least r = 0.70 [27, 29]. The intraclass correlation study were contacted to obtain additional information coefficient (ICC) is the most recommended statisti- on the psychometric properties of the scale when it was cal index for continuous measures in the assessment necessary. of temporal stability. To reduce possible recall bias, a score of 2 was indicated if the time interval between Quality assessment test administration was 1–2 weeks. A score of 1 was The quality of the included studies was assessed using the assigned if the time interval between test administra- criteria proposed by Terwee et al. [27] for health meas- tion was less than 1 week or greater than 2 weeks and ures. Each of the 7 criteria is scored 2 if the criteria are a score of 0 if no information on test–retest reliability fulfilled, 1 if they are partially fulfilled, and 0 if no crite- was reported. ria are fulfilled. The total score can range from 0 to 14. 5. Construct validity addresses whether scores on a The quality assessment was carried out by two review- questionnaire are significantly associated with poten- ers (JPS-M and AB), with the supervision of two exter- tially related measures. A theoretical underpinning nal reviewers (JVL and MR-V). Specifically, the following is needed to verify the hypotheses of expected cor- psychometric properties were assessed: relations between different scales. At least two of the correlations between two theoretically related 1. Content validity indicates whether the construct constructs had to have a minimum of r = 0.50 [27]. of interest is sampled by the questionnaire items. A A score of 2 was given if information about conver- score of 2 was assigned if the measurement objec- gent validity and divergent validity was provided. A Sanabria‑Mazo et al. Systematic Reviews (2023) 12:66 Page 4 of 19 score of 1 was assigned if information was provided if mean and SDs of less than four relevant groups for only one of the concepts mentioned above and were informed and score of 0 if no information about score of 0 if no construct validity information was interpretability was found. provided. 6. Floor and ceiling effects is a measure to detect the number of participants achieving the highest or low- est possible scores [27]. A score of 2 was assigned if Results less than 15% of respondents achieved the highest or Selection and inclusion of studies lowest possible scores. A score of 1 was reported if As displayed in Fig. 1, the initial database search yielded more than 15% of respondents achieved the highest a total of 537published articles. In addition, 5 articles or lowest possible scores and a score of 0 if this infor- were included by reverse citation and 5 by experts. mation was not provided. After removal of duplicates, 498 titles and abstracts 7. Interpretability indicates how differences in scores on were reviewed, of which 25 was selected for full-text the CAMI can be interpreted or the degree to which review. After this process, 10 articles were excluded, 4 qualitative meaning can be obtained from quantita- articles because they were not related to the CAMI, 2 tive scores. A known-groups validity approach is because they did not focus on the psychometric prop- suggested with means and standard deviations (SDs) erties of the scale, 2 because they did not provide rele- of scores of relevant subgroups of participants who vant information, and 2 because it was written in other are expected to differ in the CAMI [27]. A score of languages (German or Chinese). Finally, a total of 15 2 was assigned if mean and SDs of four or more rel- studies were included in this systematic review. evant groups were reported. A score of 1 was given Fig. 1 PRISMA flowchart from record identification to study inclusion Sanabria‑Mazo et al. Systematic Reviews (2023) 12:66 Page 5 of 19 Characteristics of included studies that items loaded on different subscales depending on The 15 included studies were conducted in 12 different the version of the CAMI. Considering suggested “Rules countries: Spain (n = 2), United Kingdom (n = 2), Italy of thumb” [41, 43], in the original CAMI (40 items dis- (n = 1), France (n = 1), Sweden (n = 1), Ireland (n = 1), tributed in 4 factors) only 26 of the 40 items had a fac- Canada (n = 1), China (n = 1), Chile (n = 1), Argentina tor loading greater than 0.40, with a difference ≥ 0.15 in (n = 1), Kenya (n = 1), and Iran (n = 1). Participants in the factor loadings of each item in the different factors. In most studies were healthy individuals (n = 13) and the addition, some items load more strongly on other factors most frequent type of sampling was non-probability than on the original factor assignment reported by Taylor (n = 12). The sample size of the studies ranged from and Dear [23]. 130 [30] to 2000 [31], enrolling a total of 10,841 partici- pants, and the mean age ranged from 15 [32] to 48 [15] Internal consistency years old. The proportion of women in all studies was The internal consistency was assessed in 9 of the included higher than 50%. The design of all included studies was studies. It was assessed for both the global CAMI (n = 7) observational. [23, 30, 32, 34, 35, 39, 40] and its subscales (n = 7) [15, The included studies were published between 1981 [23] 23, 30, 32, 39, 40, 42]. All studies that assessed the Cron- and 2023 [30]. The original CAMI-40 (n = 5) [23, 30, 32– bach’s α or McDonald`s Omega (ω) of the global scale, 34] and the CAMI-20 (n = 3) [33, 35, 36] were the most both original CAMI-40 and its versions (i.e., CAMI-BR, psychometrically analysed versions, followed by CAMI- CAMI-20, and CAMI-10), exceeded the minimum estab- W (n = 2) [33, 37], CAMI-26 (n = 2) [15, 38], CAMI-BR lished by Nunnally and Bernstein [28] of 0.80, except for (n = 1) [39], CAMI-31 (n = 1) [31], CAMI-10 (n = 1) [40], three studies: (ω = 0.78 for the CAMI-40) [34], (α = 0.59 CAMI-24 (n = 1) [41], CAMI-22 (n = 1) [42], CAMI-1 for the CAMI-40) [30], and (α = 0.69 for the CAMI- [33], CAMI-2 [33], and CAMI-W (n = 1) [33]. In total, 14 10) [40]. Studies that assessed the Cronbach’s α of the studies examined the psychometric properties of CAMI subscales in the original CAMI-40 [23, 30, 32] and its using classical test theory [15, 23, 30–33, 35–42] and one versions (i.e., CAMI-26, CAMI-BR, CAMI-10, and using item response theory (IRT) [34]. Table 1 provides a CAMI-22) [15, 37, 39, 40] showed some heterogeneity in detailed description of the included studies. their coefficients. Of these, some (n = 4) [15, 30, 40, 42] reported internal consistency values below the recom- Quality assessment mended minimum cut-off point in the CAMI subscales. As shown in Table 2, the overall methodological qual- On one hand, the study assessing the subscales of the ity of the included studies was low. On a scale of 0 to 14 CAMI-22 [42] reported internal consistency values points, 8 studies scored 5 or less [30, 31, 33, 35, 37, 38, below the minimum recommended score on benevolence 41, 42] and 6 scored 6–7 [15, 23, 32, 34, 39, 40]. Only one (α = 0.63), non-authoritarianism (α = 0.52), non-social study [36] scored higher than 10 on the quality assess- restrictiveness (α = 0.53), normalisation (α = 0.43), and ment, which indicates that in general the psychomet- community rehabilitation (α = 0.61). The study examin- ric properties of the CAMI have not been adequately ing the subscales of the CAMI-10 [40] obtained alpha assessed. values below the minimum recommended cut-off point on acceptance (all α < 0.70). The study assessing the Content validity subscales of the CAMI-26 [15] reported alpha values The construct of interest in all included studies (n = 15) below the recommended cut-off point on benevolence was the assessment of attitudes towards people suffering (α = 0.63), authoritarianism (α = 0.54), and support for from mental disorders. The developers of the CAMI gen- the mental health community (α = 0.72). erated part of the scale by extracting items from previ- On the other hand, there were studies that reported ously published measures [23], whereas the subsequent a Cronbach’s α value below the recommended cut-off versions were adaptations with different length of the only in some specific subscales of the original CAMI- original scale. 40 [23, 30, 32] and the CAMI-BR [39]. The three studies [23, 30, 32] that evaluated the original CAMI Factor structure indicated values below the cut-off on the dimen- The dimensionality of the CAMI was assessed in most of sions of benevolence (α = 0.64 [23]; α = 0.76 [32]; the included articles (n = 11). Three studies computed a α = 0.49 [30]), authoritarism (α = 0.27 [23]; α = 0.68 CFA [34, 35, 38], six opted for an EFA or principal com- [32]; α = 0.61 [30]), social restrictiveness (α = 0.67 ponent analysis [15, 23, 31, 37, 40, 42], and 2 computed [32]; α = 0.64 [30]), and CMHI (α = 0.76 [30]). The EFA and CFA [38, 39]. The number of dimensions ranged study [39] focused on the subscales of the CAMI- between 2 and 5. Regarding item allocation, it was found BR reported very low alpha values on benevolence Table 1 Characteristics of the studies included in the systematic review (n = 13) Author (year). Country CAMI version (items) Target population and study Sample size. Mean age (SD). Results depending on Psychometric results about design (sample type) % female sociodemographic variables CAMI Taylor and Dear (1981) [23]. CAMI-40 • General population from • 1090 participants • Less considerate attitudes Factor structure Canada (40 items) Toronto with separate samples • Mean age: NA (NA). Female: toward the mentally ill: elder • Benevolence (10 items) from areas with and without NA people, divorced people, users • CMHI (10 items) mental health facilities with children (not including • Social restrictiveness (10 items) • Observational (non-probabilis‑ the ones with children older • Authoritarianism (10 items) tic sample) than 18 years old), regular Internal consistency church attenders (depending • Global (NA) Sanabria‑Mazo et al. Systematic Reviews on the religious determination). • Subscales: benevolence The Pentecostal and Greek (α = 0.76), CMHI (α = 0.88), social Orthodox groups showed more restrictiveness (α = 0.80), and authoritarian attitudes and the authoritarianism (α = 0.68) Pentecostal and Greek Ortho‑ Test–retest reliability dox had the least benevolent • Global (NA) (2023) 12:66 attitudes • Subscales (NA) • More considerate attitudes Validity toward the mentally ill: female • A strong degree of correspond‑ respondents, higher sociode‑ ence between the factor scales mographic status, and higher and the theoretically ones. It is education. The users that had important to emphasize that used mental health facilities the authoritarianism and social or are related with friends that restrictiveness scales equally had used those facilities. The correlated with the first factor, ad Baptists and Salvation Army to a lesser extent, with the fourth showed fewer authoritarian factor attitudes and the Baptists with the United Church, showed more benevolent attitudes Brockington et al. (1993) [31]. CAMI-31 • General population from the • 2000 participants • CMHI: (factor 1). Higher educa‑ Factor structure United Kingdom (31 items) Malvern and Bromsgrove area • Bromsgrove mean age: NA tion and professional experi‑ • Benevolence (NA) • Observational (non-probabilis‑ (NA). Female: 51% ence with people with mental • CMHI (NA) tic sample) • Malvern mean age: NA (NA). health problems • Authoritarianism (NA) Female: 58% • Authoritarianism (factor 2). Internal consistency Lower scores are related with • Global (NA) higher education • Subscales (NA) • Benevolence (factor 3). High‑ Test–retest reliability est scores were found within • Global (NA) the groups with higher educa‑ • Subscales (NA) tion, knowledge of mental Validity illness, and an age between 55 • NA and 64 Page 6 of 19 Table 1 (continued) Author (year). Country CAMI version (items) Target population and study Sample size. Mean age (SD). Results depending on Psychometric results about design (sample type) % female sociodemographic variables CAMI Wolff et al. (1996) [37]. United CAMI-W • General population from • 215 participants • Fear and exclusion were Factor structure Kingdom (20 items) London • Mean age: 37 (13). Female: associated with having children • Fear and exclusion (11 items) • Observational (non-probabilis‑ 55% in the domestic establishment • Social control (6 items) tic sample) (< 18), higher age, and a lower • Goodwill (3 items) occupational status Internal consistency • Social control was associated • Global (NA) with a higher age; sociode‑ • Subscales (NA) Sanabria‑Mazo et al. Systematic Reviews mographic lower class; Asian, Test–retest reliability African, or Caribbean descend‑ • Global (NA) ant; lower educational level; • Subscales (NA) with children in the household; Validity lower occupational status; • Predictive validity. Significant being divorced, widowed, or correlation between fear and separated; and be aware of exclusion, social control and (2023) 12:66 somebody with a mental health goodwill, and behavioural inten‑ problem tion items • Goodwill was associated with lower age, higher educational level, and people from ethnical origin Song et al. (2005) [42]. China CAMI-22 • General population from • 1203 participants • NA Factor structure (22 items) Taiwan • Mean age: 42 (14). Female: • Benevolence (6 items) • Observational (stratified pro‑ 51% • Rehabilitation in the community portional sample) (6 items) • Non-authoritarism (5 items) • Non-social restrictiveness (3 items) • Normalization (2 items) Internal consistency • Global (NA) • Subscale: benevolence (α = 0.63), rehabilitation in the community (α = 0.61), non- authoritarism (α = 0.52), non- social restrictiveness (α = 0.53), and normalization (α = 0.43) Test–retest reliability • Global (NA) • Subscale (NA) Validity • NA Page 7 of 19 Table 1 (continued) Author (year). Country CAMI version (items) Target population and study Sample size. Mean age (SD). Results depending on Psychometric results about design (sample type) % female sociodemographic variables CAMI Buizza et al. (2005) [41]. Italy CAMI-24 • General population from • 280 participants • Physical distance and fear. Factor structure (24 items) Brescia • Mean age: 48 (NA). Female: They are associated with • Physical distance and fear (8 • Observational (probabilistic 60% age > 61 years, being divorced/ items) sample) widowed/separated, and never • Social distance and isolation (9 having participated in voluntary items) or social activities • Social responsibility and toler‑ • Social isolation. Associated ance (7 items) Sanabria‑Mazo et al. Systematic Reviews with age > 41 years, higher edu‑ Internal consistency cation, and being unemployed • Global (NA) • Social responsibility and toler‑ • Subscales (NA) ance. There are no significative Test–retest reliability association between the social • Global (NA) demographic variables and this • Subscales (NA) factor Validity (2023) 12:66 • NA Högberg et al. (2008) [35]. CAMI-20 • Student nurses selected based • 421 participants • NA Factor structure Sweden (20 items) on their easy availability • Mean age: 28 (7). Female: NA • Open-minded and pro-integra‑ • Observational (non-probabilis‑ tion (9 items) tic sample) • Fear and avoidance (6 items) • Community mental health ideol‑ ogy (5 items) Internal consistency • Global (α = 0.90) • Subscales (NA) Test–retest reliability • Global (NA) • Subscales (NA) Validity • NA Page 8 of 19 Table 1 (continued) Author (year). Country CAMI version (items) Target population and study Sample size. Mean age (SD). Results depending on Psychometric results about design (sample type) % female sociodemographic variables CAMI Morris et al. (2011) [33]. Ireland CAMI-40 (40 items), • Nurses • 1242 participants • NA Factor structure re-specified CAMI-1 (NA), • Observational (non-probabilis‑ • Mean age: 40 (10). Female: CAMI-40 re-specified CAMI-2 (NA), tic sample) 66% • Authoritarianism (10 items) Högberg CAMI-20 (20 • Benevolence (10 items) items), • Social restrictiveness (10 items) CAMI-W (20 items), • CMHI (10 items) Re-specified Wolff Scale Re-specified CAMI 1 Sanabria‑Mazo et al. Systematic Reviews (NA) • NA Re-specified CAMI 2 • NA Högberg CAMI-20 • Open-minded and pro-integra‑ tion (9 items) • Fear and avoidance (6 items) (2023) 12:66 • Community mental health ideol‑ ogy (5 items) Wolff CAMI-20 • Fear and exclusion (11 items) • Social control (6 items) • Goodwill (3 items) Re-specified Wolff scale • NA Internal consistency • Global (NA) • Subscales (NA) Test–retest reliability • Global (NA) • Subscales (NA) Validity • NA Page 9 of 19 Table 1 (continued) Author (year). Country CAMI version (items) Target population and study Sample size. Mean age (SD). Results depending on Psychometric results about design (sample type) % female sociodemographic variables CAMI Abelha et al. (2015) [39]. Brazil CAMI-BR • General population from the • 230 participants • NA Factor structure (40 items) Rio de Janeiro area • Mean age: 45 (3). Female: 68% • Social restrictiveness (10 items) • Observational (non-probabilis‑ • Benevolence (10 items) tic sample) • CMHI (10 items) • Authoritarism (10 items) Internal consistency • Global (α = 0.84) Sanabria‑Mazo et al. Systematic Reviews • Subscales: social restrictiveness (α = 0.76), benevolence (α = 0.69), CMHI (α = 0.81), and authoritari‑ anism (α = 0.35) Test–retest reliability • Global (r = 0.69) • Subscales: social restrictiveness (2023) 12:66 (r = 0.64), benevolence (r = 0.62), CMHI (r = 0.54), and authoritarian‑ ism (r = 0.37) Validity • NA Ochoa et al. (2016) [32]. Spain CAMI-40 • Elementary school students • 150 participants • NA Factor structure (40 items) • Observational (non-probabilis‑ • Mean age: 15 (1). Female: 51% • Authoritarianism (10 items) tic sample) • Benevolence (10 items) • Social restrictiveness (10 items) • CMHI (10 items) Internal consistency • Global: first assessment (α = 0.86) and second assessment a (α = 0.90) Data provided by Ochoa et al. [34] • Subscales: authoritarianism (α = 0.27), benevolence (α = 0.64), social restrictiveness (α = 0.67), and CMHI (α = 0.81) Test–retest reliability • Global (NA) • Subscales: CMHI (r = 0.88), authoritarianism (r = 0.81), benevolence (r = 0.85), and social restrictiveness (r = 0.81) Validity • NA Page 10 of 19 Table 1 (continued) Author (year). Country CAMI version (items) Target population and study Sample size. Mean age (SD). Results depending on Psychometric results about design (sample type) % female sociodemographic variables CAMI Grandón et al. (2016) [40]. Chile CAMI-10 • Two samples of the general • 749 participants form pooling NA Factor structure (10 items) population from Concepción 2 samples • Acceptance (5 items) • Observational (non-probabilis‑ • First sample: mean age: 39 • CMHI (5 items) tic simple) (13). Female: 56% Internal consistency • Second sample: mean age: 42 • Global (α = 0.69) (13). Female: 56% Subscales: acceptance (α = 0.61) and CMHI (α = 0.66) Sanabria‑Mazo et al. Systematic Reviews Test–retest reliability • Global (r = 0.79) • Subscales: CMHI (r = 0.63), authoritarianism (r = 0.57), benevolence (r = 0.39), and social restrictiveness (r = 0.62) Validity (2023) 12:66 • Factor 1 SDO-factor 1 CAMI (− 0.25) • Factor 1 SDO–factor 2 CAMI (− 0.30) • Factor 2 SDO–factor 1 CAMI (− 0.25) • Factor 2 SDO–factor 2 CAMI (0.16) Rubio-Valera et al. (2016) [15]. CAMI-26 • General population • 1019 participants • NA Factor structure Spain (26 items) • Observational (probabilistic • Mean age: 48 (1). Female: 51% • Authoritarianism (7 items) sample) • Benevolence (6 items) • Support for community mental health ideology (9 items) Internal consistency • Global (NA) • Subscales: authoritarism (α = 0.54), benevolence (α = 0.63), and support for community mental health care (α = 0.72) Test–retest reliability • Global (NA) • Subscales (NA) Validity • NA Page 11 of 19 Table 1 (continued) Author (year). Country CAMI version (items) Target population and study Sample size. Mean age (SD). Results depending on Psychometric results about design (sample type) % female sociodemographic variables CAMI García et al. (2017) [38]. France CAMI-26 • Undergraduate nursing • 268 participants • NA Factor structure (26 items) students • Mean age: NA (NA). Female: • Benevolence (7 items) • Observational (non-proba‑ NA • CMHI (5 items) bilistic) • Authoritarism (7 items) • Restrictiveness (7 items) Internal consistency • Global (NA) Sanabria‑Mazo et al. Systematic Reviews • Subscales (NA) Test–retest reliability • Global (r = 0.79) • Subscales: CMHI (r = 0.63), authoritarianism (r = 0.57), benevolence (r = 0.39), and social (2023) 12:66 restrictiveness (r = 0.62) Validity • CAMI is related to the housing control scale, RIBS, and MARKS Tong et al. (2020) [36]. China CAMI-20 (20 items) • Medical students (MS) and • 1228 MS and 1092 PGW NA Factor structure primary healthcare workers • MS sample: mean age: 21 (2). • Benevolence (5 items) (PHW) Female: 63% • Fear and exclusion (8 items) • Observational (non-proba‑ • PHW sample: mean age: 36.3. • Support and tolerance (7 items) bilistic) Female: 70% Internal consistency • Global: MS (α = 0.82) and PHW (α = 0.85) • Subscales: MS and PHW, respec‑ tively: benevolence (α = 0.79 and 0.83), fear and exclusion (α = 0.79 and 0.85), support and tolerance (α = 0.74 and 0.85) Test–retest reliability • Global: MS (r = 0.79) and PHW (r = 0.75) • Items: MS (r = 0.29 to 0.61) and PHW (r = 0.45 to 0.74) Validity • Convergent validity. Correlation with potentially related measures was not assessed • Discriminant validity. Good discriminant validity between subgroups of interest (MS and PHW) Page 12 of 19 Table 1 (continued) Author (year). Country CAMI version (items) Target population and study Sample size. Mean age (SD). Results depending on Psychometric results about design (sample type) % female sociodemographic variables CAMI Bitta et al. (2022) [34]. Kenya CAMI-40 • Population from the Kilifi • 616 participants • NA Factor structure (40 items) County Hospital • Mean age: 37 (14). Female: • Authoritarianism (13 items) • Observational (non-proba‑ 51% • Benevolence (7 items) bilistic) • CMHI (3 items) Internal consistency • Global (ω = 0.78) • Subscales (NA) Sanabria‑Mazo et al. Systematic Reviews Test–retest reliability • Global (r = 0.39) • Subscales (NA) Validity • RIBS-factor 1 CAMI (− 0.26) • RIBS-factor 2 CAMI (0.13) • RIBS-factor 3 CAMI (0.31) (2023) 12:66 • RIBS-CAMI total (− 0.08) • MAKS total-factor 1 CAMI (− 0.10) • MAKS total-factor 2 CAMI (0.16) • MAKS total-factor 3 CAMI (0.34) • MAKS total-CAMI total (0.09) Item responses • People with mental illness expe‑ riences reported significantly bet‑ ter attitudes than those without • Women had significantly higher scores in factor one scores com‑ pared to men • Level of education had a sig‑ nificant association with overall scores for factor one, but not factor two or three Page 13 of 19 Sanabria‑Mazo et al. Systematic Reviews Table 1 (continued) Author (year). Country CAMI version (items) Target population and study Sample size. Mean age (SD). Results depending on Psychometric results about design (sample type) % female sociodemographic variables CAMI (2023) 12:66 Kafami et al. (2023) [30]. Iran CAMI-40 • General population • 130 participants • NA Factor structure (40 items) • Observational (non-proba‑ • Mean age: 35 (11). Female: NA bilistic) 79% Internal consistency • Global (α = 0.59) • Subscales: authoritarianism (α = 0.61), benevolence (α = 0.49), social restrictiveness (α = 0.64), and CMHI (α = 0.76) Test–retest reliability • Global (r = 0.93) • Subscales: authoritarianism (r = 0.97), benevolence (r = 0.92), social restrictiveness (r = 0.95), and CMHI (r = 0.95) Validity • NA a In those psychometric studies where a factor analysis has been carried out, resulting, for the most part, in several different factors and items, it has been decided, in the CAMI section used, to indicate the version resulting from the factor analysis previously carried out. Brockington et al. [38] – Not all the items are indicated due to their non-specification in the original article Page 14 of 19 Sanabria‑Mazo et al. Systematic Reviews (2023) 12:66 Page 15 of 19 Table 2 Quality assessment of the included studies using the criteria proposed by Terwee et al. [27] Articles Content Factor Internal Test–retest Construct Floor/ Interpretability Final score validity structure consistency reliability validity ceiling (0–14) effects Taylor and Dear (1981) [23] 2 1 1 0 1 0 1 6 Brockington et al. (1993) [31] 1 1 0 0 0 0 1 3 Wolff et al. (1996) [37] 2 1 0 0 0 0 2 5 Song et al. (2005) [42] 2 1 0 0 0 0 1 4 Buizza et al. (2005) [41] 2 1 0 0 0 0 2 5 Högberg et al. (2008) [35] 2 1 1 0 0 0 0 4 Morris et al. (2011) [33] 2 2 1 0 0 0 0 5 Abelha et al. (2015) [39] 2 1 1 0 1 0 1 6 Ochoa et al. (2016) [32] 2 0 1 2 0 0 1 6 Grandón et al. (2016) [40] 2 2 1 0 1 0 1 7 Rubio et al. (2016) [15] 2 1 1 0 0 0 2 6 García et al. (2017) [38] 2 1 0 1 1 0 0 5 Tong et al. (2020) [36] 2 2 2 2 1 0 2 11 Bitta et al. (2022) [34] 1 1 1 1 1 0 1 6 Kafami et al. (2023) [30] 2 0 1 1 0 0 0 4 The quality of the included studies was assessed using the criteria proposed by Terwee et al. [27] for health measures. Each of the criteria is scored 2 if the criteria are fulfilled, 1 if they are partially fulfilled, and 0 if no criteria are fulfilled. The total score can range from 0 to 14 (α = 0.69), authoritarism (α = 0.35). In general, (r = 0.62), authoritarianism (r = 0.37), social restric- authoritarism is the least reliable subscale [23, 32, 39, tiveness (r = 0.64), and CMHI (r = 0.54) did not exhibit 40, 42]. good temporal stability. The temporal stability of the Regarding the original CAMI, two [23, 32] out of three CAMI-10 [40] was acceptable only for the overall scale studies that assessed its subscales reported an acceptable (r = 0.79), but not in the case of the subscales: benevo- internal consistency on the CMHI subscale (α = 0.88; lence (r = 0.39), authoritarianism (r = 0.57), social restric- α = 0.81, respectively), while only one [23] of the two tiveness (r = 0.62), and CMHI (r = 0.63). studies reported an adequate value in social restrictive- ness (α = 0.80). The study [39] that assessed the reliabil- Construct validity ity of the CAMI-BR subscales also reported an adequate Three studies found positive statistically significant corre- value on the CMHI subscale (α = 0.81). lations of low magnitude (< 0.50) between the CAMI and potentially related instruments such as the MAKS and Test–retest reliability the Reported and Intended Behaviours Scale (RIBS). One The temporal stability was assessed in five studies [30, 32, study [40] identified correlations between the factors of 34, 38, 39] by computing the ICC coefficient. One study the CAMI-10 (i.e., CMHI, authoritarianism, benevolence, that used the original CAMI [32] calculated temporal sta- and social restrictiveness) and the Social Dominance Ori- bility on all subscales, without reporting the global coeffi- entation (SDO), with values ranging from − 0.31 to 0.16. cient: benevolence (r = 0.85), authoritarianism (r = 0.81), Other study [38] found negative correlations between social restrictiveness (r = 0.81), and CMHI (r = 0.88). the CAMI-26 and instruments such as the RIBS (− 0.44) Other study [34] explored the temporal stability of the and the MAKS (− 0.30). Finally, one study [34] identified original CAMI in the global coefficient (r = 0.39), with a negative correlation between original CAMI and RIBS questionable results, but not in its subscales. The study (− 0.08) and a positive correlation with MAKS (0.09). that evaluated the temporal stability of the original CAMI [30] reported excellent results in both the global Floor and ceiling effects coefficient (r = 0.93) and its subscales: authoritarianism None of the included studies reported information on (r = 0.97), benevolence (r = 0.92), social restrictiveness ceiling and/or floor effects. (r = 0.95), and CMHI (r = 0.95). The study assessing temporal stability on the CAMI-BR Interpretability [39] found acceptable temporal stability for the overall The interpretability was analysed from a known-groups scale (r = 0.69). However, the subscales of benevolence validity approach in several studies (n = 11) [15, 23, 31, Sanabria‑Mazo et al. Systematic Reviews (2023) 12:66 Page 16 of 19 32, 36–42]. For instance, older participants with low [40]. In contrast, the studies that computed Cronbach’s employment status and low social class have higher α of each subscale [15, 23, 30, 32, 39, 40, 42], usually CAMI scores compared to younger participants and obtained values below the recommended cut-off of 0.80. those with high employment and social status [31, 33]. More specifically, the authoritarism subscale was the Men also reported higher scores than women [15]. As least reliable subscale [23, 30, 32, 39, 40, 42]. expected, those participants who had undertaken volun- The CAMI-40 was the only version in which temporal teering or social activities scored lower than those who stability of its subscales was analysed, showing adequate had not [41]. stability over time. CAMI-BR [39] and CAMI-10 [40] did not present good stability in the subscales but had Discussion an acceptable test–retest on the global scale. In general, Principal findings and interpretation there is a lack of longitudinal studies, therefore this psy- The results of this systematic review can be summarized chometric aspect has not been exhaustively addressed. as follows. The CAMI has been used in a wide variety The pattern of correlations between the CAMI and of settings and in diverse samples from many different other potentially related constructs was statistically sig- countries (i.e., Spain, United Kingdom, Italy, France, Swe- nificant and in the expected directions, partially sup- den, Ireland, Canada, China, Chile, Argentina, Kenya, porting the construct validity. Following Terwee et al. and Iran). The target population of included studies var- criteria [27], some correlations were not of the expected ied from students [32] and primary healthcare workers magnitude. Finally, the interpretability of the CAMI scale [33, 36] to the general population [15, 30, 31, 34, 37, 38, assessed by 11 studies highlight that young female partic- 40–42]. These aspects, besides the different CAMI ver- ipants who have undertaken volunteering activities and sions used, might account for the heterogeneous findings have high employment and/or social status, present less in the psychometric data of this stigma measure. stigma towards people with mental disorders. In the current systematic review, a total of 15 papers met the inclusion criteria and provided data on several Strengths and limitations psychometric indices. Although the 3-factor model was As far as it is known, this systematic review is the first the most reported structure [15, 31, 34, 35, 37, 41] fol- to summarize the psychometric properties of the differ- lowed by a 4-factor model [23, 38, 39], the items consid- ent versions of the CAMI since it was published 40 years erably varied in their expected allocation among studies. ago. The CAMI has been administered to a wide variety Only 3 out of 15 studies presented the same number of of populations. Adapted criteria based on the consensus- items [23, 30, 32]. based standards for the selection of health measurement There were three studies using the 40-item version that instruments (COSMIN) were applied to evaluate the differed from the original dimensions proposed for the quality of the CAMI measurement properties and pro- CAMI. Two studies [15, 34] supported a 3-factor struc- vided a comprehensive and qualitative synthesis of its ture, whereas the other supported a 4-factor structure current evidence. For transparency purposes, the review [38]. The studies with more dramatic changes in item protocol was registered in PROSPERO, and an exhaus- allocation were those using the CAMI-24 [41], CAMI- tive search strategy was carried out, as well as a clear data 10 [40], and CAMI-26 [15]. The estimation method of extraction procedure. In contrast, the scope of this results maximum likelihood can only be used in CFA when mul- might be considered as limited due to the exclusion of tivariate normality is met, but in the CAMI-24 study [41] papers not written in English, Spanish, or Italian. The this assumption was violated. Moreover, the different relatively few included studies in this systematic review approaches to analyse dimensionality (principal compo- underly the need of addressing with more emphasis some nent analysis, EFA, and CFA) and the diverse methods psychometric properties of this stigma measure, such as used to estimate the factor models might explain the het- the factorial invariance across age, gender, or cultures. erogeneous findings in dimensionality. Regarding the internal consistency of the CAMI, 9 out Conclusions of 15 studies addressed this psychometric aspect. Unex- The CAMI psychometric properties have been exam- pectedly, 7 studies reported the α or ω for the total scale, ined mainly using classic test theory as a framework. whereas 7 reported α values only for subscales. Accord- This methodological approach does not allow an assess- ing to Nunnally and Bernstein criteria [28], those stud- ment of the quality of individual CAMI items and fac- ies that assessed the internal consistency of the total tors. Even though some evidence on its psychometric scale (CAMI-BR [39], CAMI-40 [32], and CAMI-20 [35]) soundness is beginning to emerge from IRT, the evi- obtained adequate alpha values, except for two studies dence is limited at present. The study that explored the with CAMI-40 [22, 34] and one study with the CAMI-10 CAMI properties by means of IRT methods indicated Sanabria‑Mazo et al. Systematic Reviews (2023) 12:66 Page 17 of 19 significant item bias according to health status, gen- forty years, it is suggested that future systematic reviews der, and education level. In general, people with men- update the available evidence every 10 years. tal health experiences, women, and people with lower levels of education scored higher on some of the three Abbreviations CAMI factors. IRT-based methods can provide valu- CAMI Community Attitudes to Mental Illness able information for gauging the quality of individual CASC California Assessment of Stigma Change CAMI items and response options across different CMI Custodial Mental Illness Ideology Scale CMHI Community Mental Health Ideology levels of stigma. This methodology is also very useful COSMIN Consensus-based standards for the selection of health measure‑ for assessing differential item functioning according to ment instruments sociodemographic variables of interest. In this sense, EFA Exploratory factor analysis CFA Confirmatory factor analysis more evidence based on a IRT approach is needed for MAKS Mental Health Knowledge Schedule the different versions of the CAMI. MESH Medical Subject Headings The 3- and 4-factor structure are the most widely OMI Opinion about Mental Illness PRISMA Preferred Reporting Items for Systematic Reviews and reported in the different versions of the CAMI. The gen- Meta-analyses eral lack of fit of the existing data to the four-factor scale PROSPERO Prospective Register of Systematic Reviews originally proposed by Taylor and Dear (1981) [23] could RIBS Reported and Intended Behaviours Scale SDO Social Dominance Orientation be related to the diversity of CAMI versions explored and to the changes that have been introduced in the Supplementary Information adaptations of this inventory during the last years. Addi- The online version contains supplementary material available at https://​doi.​ tionally, it is frequent to find in the literature that mod- org/​10.​1186/​s13643-​023-​02230-4. ern psychometric approaches (i.e., based on CFA) fail to replicate the dimensionality of old instruments devel- Additional file 1: Supplementary Table S1. Detailed search strategy in oped in the seventies or eighties, which were originally PubMed (it was adapted to each database). analysed with exploratory techniques [44]. Although this inventory has been employed in many samples from Acknowledgements The authors are grateful to the CIBER of Epidemiology and Public Health different cultures and with different languages, some (CIBERESP CB22/02/00052 & CB16/02/00322; ISCIII) for its support. aspects have been scarcely addressed. For example, time needed for completion, difficulties in understanding the Authors’ contributions ED, JVL, SE-L, GT, and MR-V designed the study. JPS-M, AB, and AC-C drafted items, or the scale’s acceptability have not been explored. the manuscript. AB and NA–O contributed to the eligibility, systematic data As stated before, measurement invariance has not been extraction, and quality assessment. JPS-M, ED, AB, NA–O, AC-C, SE-L, GT, JVL, assessed in more than 40 years of history. Additionally, and MR-V critically revised the manuscript. All authors revised and approved the final manuscript. as only a few of the included studies had a longitudi- nal design, it was not possible to draw firm conclusions Funding about the temporal stability of the CAMI. In the lim- JPS-M has a PFIS predoctoral contract from the Institute of Health Car‑ los III (ISCIII; FI20/00034). AC-C has a FI predoctoral contract from AGAUR ited evidence available, a lack of good test–retest reli- (FI_B/00216). GT is supported by the National Institute for Health Research ability was detected. This low temporal stability could (NIHR) Applied Research Collaboration South London at King’s College be related to factors such as social desirability or that London NHS Foundation Trust, and by the NIHR Asset Global Health Unit award. The views expressed are those of the author(s) and not necessarily stigma is not an enduring trait. those of the NHS, the NIHR or the Department of Health and Social Care. GT In the opinion of the team of researchers of this article, is also supported by the Guy’s and St Thomas’ Charity for the On Trac project the next step should focus on item refinement to create (EFT151101), and by the UK Medical Research Council (UKRI) in relation to the Emilia (MR/S001255/1) and Indigo Partnership (MR/R023697/1) awards. SEL is a uniform set of items, especially considering the influ- supported by funding from the Medical Research Council UK, the Economic ence of culture and context on the expression of stigma. and Social Research Council and the NIHR. This study did not receive any This task would imply collaboration among an interna- specific grant from funding agencies in the public, commercial, or not-for- profit sectors. tional panel of stigma experts. The resulting candidate scale should be evaluated by using cognitive interviews Availability of data and materials and surveying different samples to reach a final version All data included in this systematic review were extracted from published papers. The data used for the study are made available within the manuscript of the inventory with adequate dimensionality, reliabil- and supplementary materials. ity, and validity that would allow cross-cultural compari- sons. In this sense, the development of tools with room Declarations for contextual adaptation could be a valuable scientific contribution for the future. Finally, considering the sci- Ethics approval and consent to participate N/A. entific interest that CAMI has generated for more than Sanabria‑Mazo et al. Systematic Reviews (2023) 12:66 Page 18 of 19 Consent for publication 13. Walsh DAB, Foster JLH. A call to action. A critical review of mental health N/A. related anti-stigma campaigns. Front Public Heal. 2020;8:569539. https://​ doi.​org/​10.​3389/​fpubh.​2020.​569539. Competing interests 14. Evans-Lacko S, Malcolm E, West K, Rose D, London J, Rüsch N, et al. The authors declare that they have no competing interests. Influence of Time to Change’s social marketing interventions on stigma in England 2009–2011. Br J Psychiatry Suppl. 2013;55:s77-88. https://​doi.​ Author details org/​10.​1192/​bjp.​bp.​113.​126672. 1 Teaching, Research & Innovation Unit, Parc Sanitari Sant Joan de Déu, Catalo‑ 15. Rubio-Valera M, Fernández A, Evans-Lacko S, Luciano JV, Thornicroft G, nia, Spain. 2 Centre for Biomedical Research in Epidemiology and Public Health Aznar-Lou I, et al. Impact of the mass media OBERTAMENT campaign (CIBERESP), Madrid, Spain. 3 Department of Basics, Developmental and Edu‑ on the levels of stigma among the population of Catalonia. Spain Eur cational Psychology, Autonomous University of Barcelona, Barcelona, Spain. Psychiatry. 2016;31:44–51. https://​doi.​org/​10.​1016/j.​eurpsy.​2015.​10.​005. 4 Department of Psychobiology and Methodology of Health Sciences, Autono‑ 16. Grausgruber A, Schöny W, Grausgruber-Berner R, Koren G, Apor BF, mous University of Barcelona, Barcelona, Spain. 5 Mental Health Research Wancata J, et al. “Schizophrenie hat viele Gesichter” – Evaluierung der Group, Hospital del Mar Medical Research Institute (IMIM), Barcelona, Spain. österreichischen Anti-Stigma-Kampagne 2000–2002. Psychiatr Prax. 6 Care Policy and Evaluation Centre, London School of Economics and Political 2009;36:327–33. https://​doi.​org/​10.​1055/s-​0029-​12203​86. Science, London, UK. 7 Centre for Global Mental Health and Centre for Imple‑ 17. Vaughan G, Hansen C. “Like Minds, Like Mine”: A New Zealand project to mentation Science, Institute of Psychiatry, Psychology & Neuroscience, Kin’s counter the stigma and discrimination associated with mental illness. College London, London, UK. 8 Department of Clinical and Health Psychology, Australas psychiatry Bull R Aust New Zeal Coll Psychiatr. 2004;12:113–7. Autonomous University of Barcelona, Barcelona, Spain. https://​doi.​org/​10.​1080/j.​1039-​8562.​2004.​02083.x. 18. Bratbo J, Vedelsby AK. ONE OF US: The National Campaign for Anti- Received: 1 June 2022 Accepted: 4 April 2023 Stigma in Denmark. In: Gaebel W, Rössler W, Sartorius N, editors. The Stigma of Mental Illness - End of the Story? Cham: Springer International Publishing; 2017. 317–38. https://​doi.​org/​10.​1007/​978-3-​319-​27839-1_​17. 19. Stuart H, Chen S-P, Christie R, Dobson K, Kirsh B, Knaak S, et al. Opening minds in Canada: background and rationale. Can J Psychiatry. 2014;59(10 References Suppl 1):S8-12. https://​doi.​org/​10.​1177/​07067​43714​05901​s04. 1. Hinshaw SP, Stier A. Stigma as related to mental disorders. Annu Rev Clin 20. Cohen J, Struening EL. Opinions about mental illness in the personnel of Psychol. 2008;4:367–93. https://​doi.​org/​10.​1146/​annur​ev.​clinp​sy.4.​022007.​ two large mental hospitals. J Abnorm Soc Psychol. 1962;64:349–60. 141245. 21. Rogers AH, Cohen M, Naranick CS. A validation study of the custodial 2. Vistorte AOR, Ribeiro WS, Jaen D, Jorge MR, Evans-Lacko S, Mari JDJ. mental illness ideology scale. J Clin Psychol. 1958;14:269–70. https://​doi.​ Stigmatizing attitudes of primary care professionals towards people with org/​10.​1002/​1097-​4679(195807)​14:3%​3c269::​aid-​jclp2​27014​0313%​3e3.0.​ mental disorders: a systematic review. Int J Psychiatry Med. 2018;53:317– co;2-1. 38. https://​doi.​org/​10.​1177/​00912​17418​778620. 22. Baker F, Schulberg HC. The development of a community mental health 3. Becker AE, Kleinman A. Mental health and the global agenda. N Engl J ideology scale. Community Ment Health J. 1967;3:216–25. https://​doi.​ Med. 2013;369:66–73. https://​doi.​org/​10.​1056/​NEJMr​a1110​827. org/​10.​1007/​BF014​34874. 4. Evans-Lacko S, Brohan E, Mojtabai R, Thornicroft G. Association between 23. Taylor SM, Dear MJ. Scaling community attitudes toward the mentally ill. public views of mental illness and self-stigma among individuals with Schizophr Bull. 1981;7:225–40. https://​doi.​org/​10.​1093/​schbul/​7.2.​225. mental illness in 14 European countries. Psychol Med. 2012;42:1741–52. 24. Evans-Lacko S, Little K, Meltzer H, Rose D, Rhydderch D, Henderson C, https://​doi.​org/​10.​1017/​S0033​29171​10025​58. et al. Development and psychometric properties of the Mental Health 5. Wittchen HU, Jacobi F, Rehm J, Gustavsson A, Svensson M, Jönsson B, Knowledge Schedule. Can J Psychiatry. 2010;55:440–8. https://​doi.​org/​10.​ et al. The size and burden of mental disorders and other disorders of the 1177/​07067​43710​05500​707. brain in Europe 2010. Eur Neuropsychopharmacol J Eur Coll Neuropsy‑ 25. Corrigan PW, Gause M, Michaels PJ, Buchholz BA, Larson JE. The California chopharmacol. 2011;21:655–79. https://​doi.​org/​10.​1016/j.​euron​euro.​ Assessment of Stigma Change: A Short Battery to Measure Improve‑ 2011.​07.​018. ments in the Public Stigma of Mental Illness. Community Ment Health J. 6. Christensen MK, Lim CCW, Saha S, Plana-Ripoll O, Cannon D, Presley F, 2015;51:635–40. https://​doi.​org/​10.​1007/​s10597-​014-​9797-5. et al. The cost of mental disorders: a systematic review. Epidemiol Psychi‑ 26. Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, atr Sci. 2020;29:e161. https://​doi.​org/​10.​1017/​S2045​79602​00007​5X. et al. The PRISMA 2020 statement: an updated guideline for reporting 7. Subramaniam M, Abdin E, Picco L, Pang S, Shafie S, Vaingankar J, et al. systematic reviews. Int J Surg. 2021;88:105906. https://​doi.​org/​10.​1016/j.​ Stigma: dimensions and correlates in a multi-ethnic population. Eur ijsu.​2021.​105906. Psychiatry. 2016;33:S522–S522. https://​doi.​org/​10.​1016/j.​eurpsy.​2016.​01.​ 27. Terwee CB, Bot SDM, de Boer MR, van der Windt DAWM, Knol DL, Dekker 1933. J, et al. Quality criteria were proposed for measurement properties of 8. Perry BL, Pescosolido BA, Martin JK, McLeod JD, Jensen PS. Comparison of health status questionnaires. J Clin Epidemiol. 2007;60:34–42. https://​doi.​ public attributions, attitudes, and stigma in regard to depression among org/​10.​1016/j.​jclin​epi.​2006.​03.​012. children and adults. Psychiatr Serv. 2007;58:632–5. https://​doi.​org/​10.​ 28. Nunnally JC, Bernstein IH. The Assessment of Reliability. Psychom Theory. 1176/​ps.​2007.​58.5.​632. 1994;3:248–92. 9. Walker JS, Coleman D, Lee J, Squire PN, Friesen BJ. Children’s stigmatiza‑ 29. Barker C, Pistrang N, Elliott R. Research methods in clinical psychology. tion of childhood depression and ADHD: magnitude and demographic London: Wiley; 2002. variation in a national sample. J Am Acad Child Adolesc Psychiatry. 30. Kafami Z, Farhoudi F, Alimoradi M, Sangsefidi Z, Delshad N, Khadem- 2008;47:912–20. https://​doi.​org/​10.​1097/​CHI.​0b013​e3181​79961a. Rezaiyan M. Persian translation and validation of community attitudes 10. Stuber JP, Rocha A, Christian A, Link BG. Conceptions of mental illness: toward the mentally ill scale: a tool for assessing social stigma about attitudes of mental health professionals and the general public. Psychiatr mental illness. Int Clin Psychopharmacol. 2023;38(1):40–4. https://​doi.​org/​ Serv. 2014;65:490–7. https://​doi.​org/​10.​1176/​appi.​ps.​20130​0136. 10.​1097/​YIC.​00000​00000​000419. 11. Martin JK, Pescosolido BA, Tuch SA. Of fear and loathing: the role of 31. Brockington IF, Hall P, Levings J, Murphy C. The community’s tolerance of “disturbing behavior”, labels, and causal attributions in shaping public the mentally ill. Br J Psychiatry. 1993;162:93–9. https://​doi.​org/​10.​1192/​ attitudes toward people with mental illness. J Health Soc Behav. bjp.​162.1.​93. 2000;41:208–23. https://​doi.​org/​10.​2307/​26763​06. 32. Ochoa S, Martínez-Zambrano F, Vila-Badia R, Arenas O, Casas-Anguera 12. Perkins R, Repper J. Prejudice, discrimination and social exclusion: Reduc‑ E, García-Morales E, et al. Spanish validation of the social stigma scale: ing the barriers to recovery for people diagnosed with mental health community attitudes towards mental illness. Rev Psiquiatr y Salud Ment. problems in the UK. Neuropsychiatry. 2013;3:377–84. 2016;9:150–7. https://​doi.​org/​10.​1016/j.​rpsmen.​2015.​02.​002. 33. Morris R, Scott PA, Cocoman A, Chambers M, Guise V, Välimäki M, et al. Is the Community Attitudes towards the Mentally Ill scale valid for use Sanabria‑Mazo et al. Systematic Reviews (2023) 12:66 Page 19 of 19 in the investigation of European nurses’ attitudes towards the mentally ill? A confirmatory factor analytic approach. J Adv Nurs. 2012;68:460–70. https://​doi.​org/​10.​1111/j.​1365-​2648.​2011.​05739.x. 34. Bitta MA, Baariu J, Fondo E, Kariuki SM, Lennox B, Newton CR. Validating measures of stigma against those with mental illness among a commu‑ nity sample in Kilifi Kenya. Global Ment Health. 2022;9:241–8. https://​doi.​ org/​10.​1017/​gmh.​2022.​26. 35. Högberg T, Magnusson A, Ewertzon M, Lützén K. Attitudes towards men‑ tal illness in Sweden: adaptation and development of the Community Attitudes towards Mental Illness questionnaire. Int J Ment Health Nurs. 2008;17:302–10. https://​doi.​org/​10.​1111/j.​1447-​0349.​2008.​00552.x. 36. Tong Y, Wang Z, Sun Y, Li S. Psychometric properties of the Chinese version of short-form community attitudes toward mentally illness scale in medical students and primary healthcare workers. Front Psychiatry. 2020;11:337. https://​doi.​org/​10.​3389/​fpsyt.​2020.​00337. 37. Wolff G, Pathare S, Craig T, Leff J. Community attitudes to mental illness. Br J Psychiatry. 1996;168:183–90. https://​doi.​org/​10.​1192/​bjp.​168.2.​183. 38. Garcia C, Golay P, Favrod J, Bonsack C. French translation and validation of three scales evaluating stigma in mental health. Front psychiatry. 2017;8:290. https://​doi.​org/​10.​3389/​fpsyt.​2017.​00290. 39. Abelha L, Gonçalves Siqueira SR, Legay L, Yang LH, Valencia E, Rodrigues Sarução K, et al. Analysis of psychometric properties of the brazilian portuguese version of the community attitudes towards the mentally ill (CAMI-BR). Rev Fac Cien Med Univ Nac Cordoba. 2015;72:211–9. 40. GrandónFernández P, SaldiviaBórquez S, Cova Solar F, Bustos C, Turra Chávez V. Análisis psicométrico y adaptación de la escala de actitudes comunitarias hacia la enfermedad mental (CAMI) en una muestra chilena. Univ Psychol. 2016;15:153–62. https://​doi.​org/​10.​11144/​Javer​iana.​upsy15-​ 2.​apae. 41. Buizza C, Pioli R, Ponteri M, Vittorielli M, Corradi A, Minicuci N, et al. Community attitudes towards mental illness and socio-demographic characteristics: an Italian study. Epidemiol Psichiatr Soc. 2005;14:154–62. https://​doi.​org/​10.​1017/​s1121​189x0​00064​00. 42. Song L-Y, Chang L-Y, Shih C-Y, Lin C-Y, Yang M-J. Community attitudes towards the mentally ill: the results of a national survey of the Taiwanese population. Int J Soc Psychiatry. 2005;51:162–76. https://​doi.​org/​10.​1177/​ 00207​64005​056765. 43. Worthington RL, Whittaker TA. Scale development research: a content analysis and recommendations for best practices. Couns Psychol. 2006;34:806–38. https://​doi.​org/​10.​1177/​00110​00006​288127. 44. Coyne JC, van Sonderen E. The Hospital Anxiety and Depression Scale (HADS) is dead, but like Elvis, there will still be citings. J Psychosom Res. 2012;73(1):77–8. https://​doi.​org/​10.​1016/j.​jpsyc​hores.​2012.​04.​002. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in pub‑ lished maps and institutional affiliations. Ready to submit your research ? Choose BMC and benefit from: • fast, convenient online submission • thorough peer review by experienced researchers in your field • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations • maximum visibility for your research: over 100M website views per year At BMC, research is always in progress. Learn more biomedcentral.com/submissions