A Preliminary Validation of the Korean Version of the Mental Illness: Clinicians’ Attitudes Scale (MICA-4): Psychometric Properties and Healthcare Provider Stigma

Abstract

Objective

Mental illness-related stigma among healthcare providers constitutes a critical barrier to care. The Mental Illness: Clinicians’ Attitudes Scale (MICA-4) is a validated measure of healthcare provider stigma, but no Korean-language version exists. Our objective was to develop and validate a Korean version of the MICA-4, examining psychometric properties and associations with healthcare provider characteristics.

Methods

Ninety-two healthcare workers completed the Korean MICA-4. Following rigorous forward and back translation, exploratory factor analysis examined the instrument’s structure. Group comparisons tested associations between MICA-4 scores and provider characteristics (specialty, experience, education).

Results

Internal consistency was acceptable (α=0.673). An exploratory five-factor structure, broadly consistent with the original instrument, was identified and provisionally retained for interpretation, given the modest sample size and preliminary nature of the study. Psychiatry specialists demonstrated significantly less stigmatizing attitudes than non-psychiatry professionals (M=42.90 vs. 50.08, p<0.001). Mental health education completion within the past 12 months showed the strongest association with reduced stigma (p<0.001). Longer clinical experience was associated with higher stigma. Age, gender, and contact frequency were not significantly associated with stigma scores.

Conclusion

The Korean MICA-4 shows preliminary evidence of reliability and validity for assessing healthcare provider stigma in Korea, although further validation with larger and more diverse samples is needed before firm conclusions can be drawn. The substantial association between recent mental health education and lower stigma suggests that targeted training programs could be a promising strategy to reduce stigma among healthcare professionals.

INTRODUCTION

Stigma toward mental illness remains a significant obstacle to care-seeking and treatment engagement worldwide [1,2]. However, research attention has disproportionately focused on public stigma and self-stigma among individuals with mental illness [3], leaving the role of healthcare provider stigma relatively understudied. Provider-level stigma is particularly consequential: it directly influences clinical decision-making, treatment recommendations, and the quality of care delivered [2,4].

Healthcare worker stigma differs fundamentally from stigma in the general population [5]. Unlike the public, healthcare providers have direct power to translate stigmatizing beliefs into clinical action—whether through withholding treatments, abbreviating clinical encounters, or providing inadequate information to patients [6]. In this sense, provider stigma constitutes not merely personal prejudice but a systemic obstacle embedded within healthcare delivery itself [2].

Stigma comprises multiple interconnected components [7]. The cognitive dimension includes factual knowledge, beliefs, and misconceptions about mental illness—for example, whether recovery is believed to be possible [1]. The affective dimension encompasses the emotional reactions (fear, disgust, anger, pity) that arise from stigmatizing beliefs. The behavioral dimension refers to actual discriminatory practices and social distancing. Understanding stigma requires attending to all three, as they interact in complex ways. A provider’s misconception that mental illness is inevitably chronic may generate hopelessness, which then manifests clinically as therapeutic pessimism and reduced treatment initiation [2].

Evidence from multiple healthcare systems demonstrates that non-psychiatric healthcare workers hold more stigmatizing attitudes than psychiatrists and mental health specialists [8,9]. This differential has been attributed to less specialized training, limited clinical exposure to individuals with mental illness, and insufficient understanding of mental health conditions [6,8]. The clinical consequences are substantial: patients report experiences of dismissal, paternalistic treatment, exclusion from decision-making, and inadequate information about their conditions [4,6].

To measure stigma among healthcare professionals systematically, the Mental Illness: Clinicians’ Attitudes Scale (MICA) was developed [10]. The MICA-4, an evolution of the earlier MICA-2, extends beyond medical students to encompass diverse healthcare professions [10,11]. This instrument has been translated and validated in multiple countries—Turkey [8], Brazil [11], South Africa [12], Japan [9], and others—yielding consistent findings that professional specialization and educational exposure to mental health content significantly predict less stigmatizing attitudes.

However, cross-cultural research reveals important variability in the instrument’s factor structure [8,9]. While the original MICA-4 was conceptualized as having five factors, translations into Turkish and Japanese have yielded three-factor solutions. Such variation may reflect genuine cultural differences in how stigma is organized and experienced, or differences in healthcare system structures and professional training [8]. This underscores the importance of culture-specific validation studies [13].

South Korea presents a particularly important context for this work. Despite having one of the highest prevalences of mental disorders among developed nations (27.8%), mental healthcare service utilization remains strikingly low—only 12.1% of individuals with lifetime diagnoses have ever accessed mental health services [14]. This treatment gap reflects multiple contributing factors, including structural barriers and high stigma [2]. In Korean healthcare, non-psychiatric primary care physicians and emergency department staff serve as gatekeepers to mental health referrals; their attitudes and knowledge therefore directly shape treatment access. Additionally, the Korean healthcare system’s specific structural features—such as official psychiatric diagnosis coding practices—create stigma-related barriers distinct from Western contexts [15]. To date, no validated Korean-language measure of healthcare provider stigma exists, limiting the capacity to assess intervention outcomes and compare findings internationally.

This study had several aims: 1) to develop a Korean-language version of the MICA-4 through rigorous translation and adaptation procedures; 2) to conduct a preliminary examination of the instrument’s psychometric properties in a Korean healthcare sample, acknowledging the modest sample size and exploratory nature of the factor analysis; and 3) to identify which provider characteristics are most strongly associated with stigma. Establishing a valid Korean MICA-4 would enable evidence-based anti-stigma policy development and participation in comparative research initiatives, while providing an initial, exploratory foundation for future large-scale validation studies rather than a definitive account of the instrument’s dimensional structure in Korea.

METHODS

Study design and setting

This was a cross-sectional validation study conducted from April to December 2025. The study was approved by Samsung Medical Center’s Institutional Review Board (SMC 2025-07-120-001, 2025-07-120-003, approved August-October 2025).

Participants

We recruited 92 healthcare workers from psychiatry and non-psychiatry departments using convenience sampling. Inclusion criteria were: healthcare workers aged 19–65 years with direct patient care responsibilities and adequate Korean proficiency [13]. Responses were excluded from analysis if participants withdrew consent during the survey or if response reliability was deemed excessively low. Low reliability was defined as meeting one or more of the following criteria: 1) more than 30% of items on the total scale were left unanswered or marked as “I do not know”; 2) identical response patterns were observed across both positively and negatively worded items, suggesting insufficient engagement with item content; or 3) two or more internal inconsistencies were identified in basic sociodemographic variables, such as age, gender, or marital status.

The final sample was 33.70% psychiatry staff (n=31) and 66.30% non-psychiatry professionals (n=61). The majority were in their 30s (52.17%), with balanced gender representation (58.70% female). Most held 6–15 years of experience (52.17%), and 58.70% reported mental health training within the past year. Over 70% reported professional contact with individuals with mental illness.

Measures and translation

The MICA-4 is a 16-item self-report scale measuring healthcare provider attitudes toward mental illness, with items rated on a 6-point Likert scale (1=strongly disagree to 6=strongly agree). Higher scores indicate more stigmatizing attitudes (range: 16–96). The English version has demonstrated adequate reliability and validity across diverse samples [10].

We followed international best practices for translation [13]. Two professional translators produced independent forward translations, which were synthesized into a consensus version. Two bilingual translators then produced back-translations to verify semantic equivalence. A five-member expert panel (2 psychiatrists, 2 clinical psychologists, 1 mental health stigma researcher) evaluated semantic equivalence using 7-point ratings; items below 4.0 underwent revision. The finalized version was reviewed in consultation with members of the Indigo Partnership Project team. Following the expert panel review, we obtained further feedback on the cultural appropriateness and clarity of the Korean translations through an interim review at the National Center for Mental Health, Ministry of Health & Welfare, Republic of Korea, and incorporated this feedback into the finalized version.

Data collection

Participants completed an online questionnaire via Qualtrics. Demographic variables collected included: age, gender, discipline, specialty (psychiatry vs. non-psychiatry), years of experience, frequency of contact with individuals with mental illness, mental health education completion in the past 12 months, personal mental health history, and family/acquaintance mental health history. All responses were informed that the study formed part of a larger project to validate a Korean mental health stigma scale for multi-level intervention development and international comparison. The study aimed to assess healthcare professionals’ attitudes toward mental illness and to generate foundational data for improving mental health awareness among healthcare providers. Voluntary participation, as well as details regarding compensation and its provision, were clearly explained. There was no missing data on the 16 MICA-4 items among the 92 participants retained for analysis, since responses with substantial missingness had already been excluded under the criteria in Section 2.2. A few demographic items used a skip-logic design—for example, contact frequency was asked only of participants who reported contact experience, and personal treatment history only of those who reported a personal mental health history—so analyses of these items were limited to the responding subgroup, with sample sizes reported alongside each result.

Statistical analysis

We computed Cronbach’s alpha to assess internal consistency [16]. Although the MICA-4 was originally conceptualized as a five-factor scale representing attitudes, knowledge, and awareness dimensions of mental illness stigma [10], cross-cultural validation studies have demonstrated structural instability with considerable item cross-loadings [8,9]. To examine the factor structure in the Korean sample, we conducted an exploratory factor analysis using principal components extraction and Varimax rotation. Data suitability for factor analysis was confirmed via Kaiser-Meyer-Olkin (KMO) sampling adequacy [17] and Bartlett’s test of sphericity [18]. Factors with eigenvalues >1.0 were initially retained; the final solution incorporated eigenvalue magnitude, scree plot inspection, and variance explained as decision criteria. Model-fit comparisons between candidate factor solutions (three- vs. five-factor) were evaluated using maximum-likelihood extraction, whereas principal components extraction with Varimax rotation, as described above, was used for the primary five-factor solution.

To evaluate criterion validity, we examined associations between provider characteristics and MICA-4 scores across three conceptual domains: demographic factors (age, gender), professional variables (psychiatric specialty status, clinical experience, patient exposure), and mental health-related experiences (personal or familial psychiatric history, mental health training) [8,19]. Binary and categorical comparisons were analyzed using t-tests and a one-way analysis of variance, respectively, with nonparametric alternatives employed when normality assumptions were violated [20]. Pairwise comparisons were corrected using Tukey–Kramer adjustment, and effect sizes (Cohen’s d, η²) were calculated to assess practical significance alongside 95% confidence intervals.

Due to our sample size (n=92, approximately 5.75:1 item-to-participant ratio), we limited statistical analyses to those sufficient to answer our primary research questions. Confirmatory factor analysis on an independent sample, measurement invariance testing, and differential item functioning analysis was beyond the scope of this study and represent important directions for future work [20].

All statistical analyses were performed using SAS (SAS Institute Inc.) and R (R Foundation for Statistical Computing). Statistical significance was set at α=0.05.

RESULTS

Participant characteristics

The sample comprised 41.30% male and 58.70% female participants, with 52.17% in their 30s. Professional discipline ranged from physicians (30.43%) to nurses (26.09%), social workers (21.74%), and others (21.74%). Regarding clinical experience, 52.17% had 6–15 years, 31.52% had ≤5 years, and 16.30% had ≥16 years. Mental health training in the past 12 months was reported by 58.70%, and 70.65% had professional contact with individuals with mental illness. Detailed demographic characteristics appear in Table 1.

Table 1.

Demographic and professional characteristics of participants (N=92)

Characteristic	Category	N (%)
Gender	Male	38 (41.30)
	Female	54 (58.70)
Age (yr)	20s	12 (13.04)
	30s	48 (52.17)
	40s	20 (21.74)
	50s and older	12 (13.04)
Professional discipline	Physician	28 (30.43)
	Nurse	24 (26.09)
	Social worker	20 (21.74)
	Clinical psychologist	2 (2.17)
	Other	18 (19.57)
Professional specialty	Psychiatry	31 (33.70)
	Non-psychiatry	61 (66.30)
Clinical experience (yr)	≤5	29 (31.52)
	6–15	48 (52.17)
	≥16	15 (16.30)
Mental health education	Yes (past 12 months)	54 (58.70)
	No	38 (41.30)
Contact with mental illness	Yes	65 (70.65)
	No	27 (29.35)
Contact frequency*	Less than monthly	14 (21.88)
	Occasional	22 (34.38)
	Frequent	8 (12.50)
	Very frequent	11 (17.19)
	Almost daily	9 (14.06)
Personal mental health experience	Yes	25 (27.17)
	No	67 (72.83)
Family mental health experience	Yes	31 (33.70)
	No	61 (66.30)
Acquaintance mental health experience	Yes	40 (43.48)
	No	52 (56.52)
Personal mental health treatment†	Yes	11 (44.00)
	No	14 (56.00)

^*contact frequency was asked only of the 65 participants who reported contact experience with individuals with mental illness; 64 of these provided a valid frequency response (1 missing) and are reflected here;

^†among 25 participants who reported personal mental health problem experience (n=25 answered).

MICA-4 reliability

Cronbach’s alpha for the 16-item scale was 0.673. While slightly below the conventional 0.70 threshold, this value is considered acceptable for exploratory research and multidimensional scales [21,22]. Item means ranged from 1.58 to 4.59, with total scores (M=47.66, SD=8.16) showing a range of 30–75.

Factor structure

Prior to conducting an exploratory factor analysis on the Korean MICA-4 (n=92), we assessed the factorability of the data using the KMO index and Bartlett’s test of sphericity. The KMO measure was 0.623, suggesting marginally adequate sampling adequacy. Bartlett’s test of sphericity was statistically significant (χ²(120)=368.43, p<0.001), indicating that item intercorrelations were sufficiently non-trivial to justify factor analysis.

Examination of eigenvalues, the scree plot (Figure 1), and competing model fit tests were conducted to determine the optimal number of factors. Eigenvalue inspection revealed that four factors exceeded the Kaiser criterion (eigenvalue >1.0): 6.03, 2.10, 1.61, and 1.30. Competing model fit tests via chi-square comparisons indicated that a three-factor solution inadequately fit the data (χ²(75)=110.09, p=0.005), whereas a five-factor solution achieved acceptable model fit (χ²(50)=53.67, p=0.34). Although the fifth factor’s eigenvalue (0.66) fell below the Kaiser threshold, the five-factor structure was retained for further analysis based on this model-fit comparison, which favored five factors over three. It also keeps the solution conceptually aligned with the original MICA-4 framework [10].

Figure 1.

Scree plot of MICA-4 exploratory factor analysis (N=92). The scree plot demonstrates eigenvalues extracted from principal component analysis of the 16-item MICA-4 scale (N=92). Four factors exceeded the Kaiser criterion (eigenvalue=1.0), with eigenvalues of 6.03, 2.10, 1.61, and 1.30. The fifth factor had an eigenvalue of 0.66, below the Kaiser criterion. Subsequent factors showed declining eigenvalues, with factors 11–16 demonstrating negative eigenvalues. MICA-4, Mental Illness: Clinicians’ Attitudes Scale.

A principal components analysis with Varimax rotation produced the following exploratory five-factor structure (Table 2): Factor 1 (eigenvalue=6.03) encompassed items 1, 2, 4, 5, 10, 15, and 16, representing general knowledge and prognosis beliefs regarding mental illness. Factor 2 (eigenvalue= 2.10) included items 4 and 7, both assessing reluctance to disclose mental illness to peers. Factor 3 (eigenvalue=1.61) comprised items 3 and 9, reflecting respect for mental health professionals and ethical standards. Factor 4 (eigenvalue=1.30) contained items 1, 2, 10, 11, 14, and 16, capturing clinical comfort and competence. Factor 5 (eigenvalue=0.66) was dominated by item 12 (λ=0.99) and supplemented by item 13, reflecting public safety concerns.

Table 2.

MICA-4 items and factor loadings after varimax rotation

Item#	Item description	F1	F2	F3	F4	F5	h2
1	I just learn about mental health when I have to, and would not bother reading additional material on it.	0.517*	-0.002	0.091	0.356*	-0.187	0.438
2	People with a severe mental illness can never recover enough to have a good quality of life.	0.766*	0.034	-0.009	0.301*	0.026	0.679
3	Working in the mental health field is just as respectable as other fields of health and social care.	0.065	0.019	0.451*	-0.017	-0.258	0.275
4	If I had a mental illness, I would never admit this to my friends because I would fear being treated differently.	0.557*	0.540*	0.195	-0.059	-0.048	0.644
5	People with a severe mental illness are dangerous more often than not.	0.622*	0.211	0.031	0.046	-0.025	0.435
6	Health/social care staff know more about the lives of people treated for a mental illness than do family members or friends.	0.093	-0.077	-0.010	-0.226	-0.154	0.089
7	If I had a mental illness, I would never admit this to my colleagues for fear of being treated differently.	0.164	0.968*	0.019	0.167	0.089	1.000
8	Being a health/social care professional in the area of mental health is not like being a real health/social care professional.	0.014	-0.029	0.172	-0.148	-0.253	0.117
9	If a senior colleague instructed me to treat people with mental illness in a disrespectful manner, I would not follow their instructions.	0.063	0.005	0.970*	0.215	0.097	1.000
10	I feel as comfortable talking to a person with a mental illness as I do talking to a person with a physical illness.	0.331*	0.239	-0.116	0.390*	0.024	0.333
11	It is important that any health/social care professional supporting a person with a mental illness also ensures that their physical health is assessed.	0.089	-0.240	0.125	0.372*	0.002	0.220
12	The public does not need to be protected from people with a severe mental illness.	0.091	0.047	0.031	-0.002	0.994*	1.000
13	If a person with a mental illness complained of physical symptoms (such as chest pain), I would attribute it to their mental illness.	0.034	0.037	0.274	0.107	-0.308	0.184
14	General practitioners should not be expected to complete a thorough assessment for people with psychiatric symptoms because they can be referred to a psychiatrist.	0.188	0.053	0.062	0.477*	-0.045	0.272
15	I would use the terms “crazy,” “nutter,” “mad,” etc. to describe to colleagues people with a mental illness who I have seen in my work.	0.328*	0.059	0.226	0.262	-0.047	0.233
16	If a colleague told me they had a mental illness, I would still want to work with them.	0.407*	0.094	0.098	0.633*	0.074	0.591

Standardized factor loadings from principal components analysis with Varimax rotation (N=92). Communalities (h2) represent the proportion of each item’s variance explained by the five-factor model. Cross-loadings observed on items 1, 2, 4, 10, and 16; low communalities (h²≤0.20) on items 6, 8, and 13. Items 7, 9, and 12 showed communalities of 1.000, consistent with an improper (Heywood) solution, and Factors 2, 3, and 5 were each defined by only one or two salient items; these factors reflect an over-extracted, statistically unstable solution and should be interpreted with caution pending confirmatory replication.

^*substantial loadings (≥0.30).

MICA-4, Mental Illness: Clinicians’ Attitudes Scale.

The rotated factor structure revealed several notable patterns. Five items [1, 2, 4, 10, 16] demonstrated cross-loadings of ≥0.30 across multiple factors. Three items [6, 8, 13] exhibited low communalities (h² ≤0.20). These patterns are detailed in Table 2. In addition, items 7, 9, and 12 each showed a communality of 1.000, with a single dominant loading (λ=0.97–0.99) on their respective factor; Factors 2, 3, and 5 were each carried by only one or two salient items.

Group differences

Significant differences in total MICA-4 scores emerged across professional specialization and clinical experience. Psychiatry specialists demonstrated substantially lower (less stigmatizing) attitudes compared to non-psychiatry professionals (Table 3): M=42.90 (SD=6.47) vs. M=50.08 (SD=7.90), t(90)=4.68, p<0.001, d=0.97, representing a large effect size. This pattern was consistent: 33.7% of the sample were psychiatrists.

Table 3.

Comparison of MICA-4 total scores by professional specialty

Specialty	N	Mean	SD	95% CI	Test statistic	Effect size
Psychiatry	31	42.90	6.47	40.47–45.33	t(90)=4.68, p<0.001***	d=0.97
Non-psychiatry	61	50.08	7.90	47.90–52.26

^***p<0.001.

Non-psychiatry professionals reported significantly higher MICA-4 scores than psychiatrists (t(90)=4.68, p<0.001, d [Cohen’s d]=0.97), representing a large effect size. MICA-4, Mental Illness: Clinicians’ Attitudes Scale; CI, confidence interval.

Clinical experience was associated with stigmatizing attitudes (Table 4). Providers with ≤5 years of experience (M=46.14, SD=9.23), 6–15 years (M=47.35, SD=7.82), and ≥16 years (M=51.60, SD=5.93) exhibited increasing stigma scores, F(2,89)=4.12, p=0.020, η²=0.085. Post-hoc testing revealed that providers with ≥16 years of experience had significantly higher stigma than those with ≤5 years of experience (p=0.021), though the difference between 6–15 years and other groups did not reach statistical significance after a multiple comparison adjustment.

Table 4.

Comparison of MICA-4 total scores by clinical experience

Experience	N	Mean	SD	95% CI	Test statistic	Effect size
≤5 years	29	46.14	9.23	42.42–49.86	F(2,89)=4.12, p=0.020*	η²=0.085
6–15 years	48	47.35	7.82	44.96–49.74
≥16 years	15	51.60	5.93	48.85–54.35

^*p<0.05.

Clinical experience was a significant predictor of stigmatizing attitudes (F(2,89)=4.12, p=0.020, η² [Eta squared]=0.085). Specifically, professionals with 16 or more years of clinical experience scored significantly higher than those with ≤5 years of experience (p=0.021). MICA-4, Mental Illness: Clinicians’ Attitudes Scale; CI, confidence interval.

Mental health education completion in the past 12 months showed the strongest association with lower stigma (Table 5). Providers who completed mental health training (M=44.96, SD=7.15) had substantially lower scores than those without recent training (M=51.50, SD=8.04), t(90)=4.95, p<0.001, d=1.04, representing a large effect size.

Table 5.

Comparison of MICA-4 total scores by mental health education completion

Education	N	Mean	SD	95% CI	Test statistic	Effect size
Yes (past 12 months)	54	44.96	7.15	42.69–47.23	t(90)=4.95, p<0.001***	d=1.04
No	38	51.50	8.04	48.78–54.22

^***p<0.001.

Mental health education completion within the past 12 months significantly reduced stigmatizing attitudes compared to those without recent education (44.96 vs. 51.50, p<0.001, d [Cohen’s d]=1.04). MICA-4, Mental Illness: Clinicians’ Attitudes Scale; CI, confidence interval.

Age showed a statistically significant omnibus effect on stigma scores (Table 6), F(3,88)=3.54, p=0.018, η²=0.108, with mean scores generally higher among older age groups. However, none of the pairwise comparisons between individual age groups reached significance after Tukey–Kramer correction.

Table 6.

Comparison of MICA-4 total scores by age group

Age group	N	Mean	SD	95% CI	Test statistic	Effect size
20s	12	45.58	7.03	41.40–49.76	F(3,88)=3.54, p=0.018*	η2=0.108
30s	48	46.06	8.93	43.50–48.62
40s	20	50.45	6.79	47.09–53.81
≥50s	12	51.50	5.90	47.64–55.36

^*p<0.05.

Age showed a significant omnibus effect on stigmatizing attitudes (p=0.018, η² [Eta squared]=0.108), with mean scores generally higher in older age groups; however, pairwise comparisons between individual age groups were not statistically significant after Tukey–Kramer correction. MICA-4, Mental Illness: Clinicians’ Attitudes Scale; CI, confidence interval.

Gender, frequency of professional contact with individuals with mental illness, and personal or familial mental illness history showed no significant associations with MICA-4 scores (all p>0.05) (Table 7). These null findings suggest that contact frequency and personal psychiatric experience were not predictive of clinician attitudes in this Korean sample.

Table 7.

MICA-4 total scores for non-significant variables

Variable	Group	N	Mean	SD	Test statistic	p
Gender	Male	38	48.63	8.50	t(90)=1.00	0.343 (ns)
	Female	54	46.98	7.92
Work contact	Yes	65	47.40	8.29	t(90)=-0.47	0.634 (ns)
	No	27	48.30	7.94
Contact frequency	Monthly	14	49.00	10.00	F(2,61)=0.18	0.852 (ns)
	Occasional	22	47.95	8.13
	Frequent	28	46.79	7.10
Personal mental health	Yes	25	46.96	7.84	t(90)=-0.67	0.616 (ns)
	No	67	47.93	8.32

ns=not significant (p>0.05). Gender, work contact with individuals with mental illness, contact frequency, and personal mental health history did not significantly influence MICA-4 scores (all p>0.05). These findings suggest that stigmatizing attitudes toward mental illness in this sample were not substantially shaped by these demographic and experiential variables. MICA-4, Mental Illness: Clinicians’ Attitudes Scale.

DISCUSSION

This study presents the first validation of the MICA-4 in a healthcare professional sample. The instrument demonstrated acceptable internal consistency (α=0.673) and a five-factor structure, successfully differentiating stigmatizing attitudes across professional groups and supporting its preliminary utility for stigma assessment in Korean healthcare contexts.

Professional specialty emerged as the most striking predictor of stigma. Psychiatry professionals reported substantially lower stigmatizing attitudes than their non-psychiatry counterparts (M=42.90 vs. 50.08; d=0.97). This pattern aligns with international literature [8,9] and likely reflects both selection and socialization mechanisms: individuals with high levels of openness may preferentially pursue psychiatry, while clinical training and direct exposure to individuals with mental illness challenge stereotypes.

Mental health education completed within the past 12 months showed similarly robust associations with reduced stigma (M=44.96 vs. 51.50; d=1.04). While the cross-sectional nature of this data precludes causal conclusions—as clinicians with less stigmatizing attitudes may be more inclined to seek out training—these findings suggest that education is strongly linked to less stigmatizing attitudes. Notably, contact frequency alone showed no significant relationship with stigma scores, indicating that unstructured clinical exposure is insufficient for meaningful attitude change without embedding within broader educational frameworks [19]. The association between clinical experience and stigma presented a counterintuitive pattern: clinicians with 16 or more years of experience demonstrated significantly higher stigma than those with ≤5 years (F(2,89)=4.12, p=0.020, η²=0.085), possibly reflecting cohort effects or erosion of attitudes without ongoing education, though the modest effect size warrants cautious interpretation.

Regarding the scale’s psychometric properties, the observed structural instability-characterized by item cross-loadings and low communalities for certain items-merits careful consideration. These findings are consistent with recent cross-cultural adaptations, such as the Turkish [8] and Japanese [9] versions, which also reported shifting factor structures. Such instability may reflect cultural nuances in how mental health stigma is conceptualized across different healthcare systems. In Korea, the ambiguous conceptual boundaries between factors like “clinical knowledge” and “professional behavior” suggest that healthcare providers may perceive these domains as more integrated than their Western counterparts. Some of this instability may reflect such substantive cultural differences; other patterns, however, point more specifically to statistical artifacts of the sample size and item-to-factor ratio. The factor solution itself also warrants some caution. Only four of the five factors met the Kaiser eigenvalue criterion, and the fifth factor’s retention rests on the converging model-fit comparison and on conceptual correspondence with the original MICA-4 rather than on eigenvalue magnitude—arguably the weakest-supported element of the reported structure. Items 7, 9, and 12 also showed communalities of 1.000, consistent with an improper (Heywood) solution, and Factors 2, 3, and 5 were each carried by only one or two salient items; factors built from so few items can fit the data almost perfectly simply by construction, which is not by itself evidence that they represent real, independent dimensions. We therefore treat the five-factor solution as provisional and hypothesis-generating rather than confirmed, pending confirmatory factor analysis in a larger, independent sample.

In addition to these findings, several methodological strengths of the present study warrant consideration. First, the translation and adaptation procedures adhered to rigorous international best practices, including forward–backward translation, expert panel review, and pilot testing with healthcare workers [13], thereby ensuring cultural and linguistic equivalence. Second, examining several provider characteristics through a series of independent comparisons allowed us to identify specific predictors, showing that education effects exceed those of professional specialty. This analytical approach provides critical insights for intervention design. Third, the inclusion of multiple disciplines beyond physicians (nurses, social workers, clinical psychologists) enhances ecological validity and representativeness, marking a significant advance over prior single-discipline validation studies.

However, important limitations require acknowledgment, particularly regarding the study’s preliminary nature. First, the sample size (n=92) and the modest item-to-participant ratio (5.75:1) limit the definitiveness of the structural solution. Consequently, the factor structure identified should be interpreted as tentative, serving as a pilot validation that requires confirmation in larger, representative samples. Second, the cross-sectional design fundamentally precludes causal inference regarding the relationship between education and stigma. Third, self-report measurement invites social desirability bias [23], with respondents potentially providing socially acceptable rather than truthful answers. Additionally, we did not measure behavioral outcomes such as diagnostic accuracy, treatment recommendations, or therapeutic alliance quality that would establish whether MICA-4 score changes correspond to observable improvements in clinical care. The KMO index (0.623) and internal consistency coefficient further suggest that while the current model is reasonable for a pilot study, independent replication is essential. Fourth, participants were recruited via convenience sampling, which may have introduced selection bias: healthcare workers with less stigmatizing attitudes may have been more willing to volunteer, which could underestimate stigma levels in the broader provider population. Fifth, although Tukey–Kramer correction controlled the familywise error rate within each set of post-hoc pairwise comparisons, no correction was applied across the approximately eight independent comparisons conducted across provider characteristics (Tables 3–7); given the exploratory framing of this study we report nominal p-values, but this should be considered when interpreting the number of significant associations identified. Sixth, the contact-frequency analysis was based on a further-reduced subsample (n=64) of participants who reported contact experience and provided a valid frequency response, which should be considered when interpreting that null finding. Finally, as discussed above, three of the five exploratory factors (Factors 2, 3, and 5) should be treated with particular caution pending confirmatory replication in a larger, independent sample.

Should these findings prove robust in larger longitudinal samples, the strong association between education and lower stigma suggests that systematic integration of mental health literacy into healthcare curricula could promote meaningful attitude change. However, a critical gap remains: whether improved attitudes translate into measurable clinical improvements—such as increased treatment initiation, patient satisfaction, or better outcomes—requires empirical investigation. Given observed specialty differences, targeted mental health education for non-psychiatry professionals warrants particular emphasis, as these groups substantially influence access to mental health services across primary care, emergency medicine, and hospital settings. The association between longer clinical experience and higher stigma further suggests that continuing education policy should mandate periodic mental health training throughout professional careers rather than single-exposure initiatives.

Future research should prioritize confirmatory factor analysis in an independent, larger sample, alongside test-retest reliability and convergent and discriminant validity assessments, to further establish the scale’s measurement properties. Such research should also employ longitudinal designs to clarify causal mechanisms and examine whether MICA-4 improvements predict clinical outcomes. Qualitative research exploring mechanisms through which education reduces stigma could inform more effective curriculum design, while inclusion of implicit attitude measures would address social desirability concerns. Despite these limitations, this preliminary validation provides an essential foundation for systematic stigma assessment in Korean healthcare settings, enabling ongoing monitoring, cross-professional comparison, and international collaboration. The evidence for education’s substantial effects on stigma offers a promising intervention pathway and underscores the necessity of maintaining mental health literacy training across professional careers.

In conclusion, the Korean MICA-4 demonstrates preliminary evidence of reliability and validity as a measure of healthcare provider attitudes toward mental illness. Mental health education and professional specialization emerge as the strongest identified predictors of less stigmatizing attitudes. While current data indicates a strong association between mental health literacy and positive attitudes, confirmation in larger, longitudinal studies with behavioral outcome measures would be necessary before implementing substantial healthcare system changes. The Korean MICA-4 provides a foundation for such future investigations and for participation in international research comparing healthcare provider attitudes across diverse healthcare systems.