Significant outcomes
-
• In this randomised survey experiment, patients’ approval ratings of staff were highly similar across attire conditions (personal clothes vs. uniforms).
-
• Given the apparent parity in patient approval of staff wearing personal clothes versus uniform, and the advantages in terms of hygiene and correct identification of staff, we find that more consistent use of uniforms in psychiatric services should be considered.
Limitations
-
• The response rate was relatively low, which introduces a risk of selection bias.
-
• We did not employ a validated questionnaire to assess patient approval of staff attire because, to our knowledge, no such instrument is available.
Introduction
The therapeutic alliance is fundamental to effective treatment across medical specialties. It facilitates treatment adherence and may augment the placebo response (Lee & Lin, Reference Lee and Lin2008; Thompson & McCabe, Reference Thompson and McCabe2012). It is influenced by several factors, including institutional trust, prior experiences and both verbal and non-verbal communication, such as professional attire (uniforms) (Petrilli et al., Reference Petrilli, Mack, Petrilli, Hickner, Saint and Chopra2015; Lerch et al., Reference Lerch, Hänggi, Bussmann and Lörwald2024).
Uniforms were originally introduced in medical specialties to ensure hygiene, but have subsequently come to be associated with authority, professionalism and responsibility. In general medicine, such symbolism has been associated with perceptions of competence and trustworthiness (Rehman et al., Reference Rehman, Nietert, Cope and Kilpatrick2005; Gherardi et al., Reference Gherardi, Cameron, West and Crossley2009; Petrilli et al., Reference Petrilli, Mack, Petrilli, Hickner, Saint and Chopra2015). In psychiatry, where treatment is more prominently rooted in interpersonal engagement, these symbols of authority may, inadvertently, create professional distance and diminish perceived approachability (McCabe & Priebe, Reference McCabe and Priebe2004; Eikhom et al., Reference Eikhom, Torsaeter and Wik2006). Thus, in psychiatric settings, uniforms may provide clarity and structure, whereas personal attire has been associated with increased approachability and reduced social barriers (Jones et al., Reference Jones, Kahn and Wolcott1964; Hawkins et al., Reference Hawkins, Claghorn and Zentay1966; Lavender, Reference Lavender1987; Gledhill et al., Reference Gledhill, Warner and King1997; Eikhom et al., Reference Eikhom, Torsaeter and Wik2006; Chu et al., Reference Chu, Chang, Dai, Hui, Ip, Kwok, Ying, Chan and Bressington2020). Despite these considerations, decisions regarding uniforms, especially in psychiatry, are more often guided by traditions and practical considerations (including costs) than by evidence.
In Denmark, healthcare staff at general hospitals usually wear scrubs (standard clinical attire consisting of short sleeved tunic and trousers) or white coats. In contrast, staff at most psychiatric hospitals wear their own clothes (personal clothes henceforth) and no white coat except in specific clinical situations (e.g. during electroconvulsive therapy, when doctors are on call and during liaison visits to general hospital departments). However, it has not been examined whether this dress code in Danish psychiatric services optimally supports patients’ perceptions of staff professionalism, competence, empathy and approachability. Therefore, this study aimed to examine the effect of attire on patient approval of staff empirically via a randomised online survey experiment.
Materials and methods
Study design and preregistration
The study protocol was preregistered on the Open Science Framework platform (https://osf.io/bvgt5/files/7dvhq). A flowchart of patient recruitment and the study design is shown in Figure 1.
Flowchart of patient recruitment and study design. The Roman numerals refer to the five intervention groups: (I) a doctor wearing personal clothes, (II) a doctor wearing a white coat over personal clothes, (III) a Doctor wearing a white coat over scrubs, (IV) a nurse wearing personal clothes and (V) a nurse wearing scrubs. e-Boks is the secure digital mailing system used by Danish authorities to communicate with citizens.
Figure 1. Long description
Flowchart of patient recruitment and study design. The process begins with all adult patients receiving treatment in the Psychiatric Services of the Central Denmark Region, excluding forensic psychiatric departments, totaling 10688 patients. Patients with specific diagnoses or treatments are excluded: 544 with organic mental disorder, 332 with mental retardation, 61 who received coercive treatment in the 3 days preceding the study, 433 with forensic sanction receiving treatment at a general psychiatric department, and 8 who previously declined participation in surveys. This leaves 9310 patients eligible for invitation. These patients receive a digital mail invitation via E-Boks, followed by a baseline questionnaire. The study includes five intervention groups: (I) a doctor wearing personal clothes, (II) a doctor wearing a white coat over personal clothes, (III) a doctor wearing a white coat over scrubs, (IV) a nurse wearing personal clothes, and (V) a nurse wearing scrubs. After the intervention, patients complete a post-intervention questionnaire to measure outcomes.
Setting
This online survey experiment was performed within the Psychiatric Services of the Central Denmark Region, which has a catchment area of approximately 1.4 million individuals. It comprises five public psychiatric hospitals, which provide free (tax-financed) inpatient, outpatient and emergency psychiatric treatment.
Participants
Patients were eligible to participation if they were aged 18 years or older and received treatment in the Psychiatric Services of the Central Denmark Region. Patients were ineligible if they had a forensic sanction, received coercive treatment, or had an organic mental disorder (ICD-10 subcategory: F0) or mental retardation (ICD-10 subcategory: F7).
Survey distribution
Invitations were sent using the secure digital mailing system ‘eBoks’ used by Danish authorities to communicate with citizens (Ebert et al., Reference Ebert, Huibers, Christensen and Christensen2018). The survey was conducted using the SurveyXact online survey platform (SurveyXact, 2026). The survey was distributed on December 11, 2025, and was open until January 10, 2026. A reminder was sent to those who had not provided a complete response by December 18, 2025.
Intervention
Respondents were randomised to one of five intervention groups, which were shown a photograph of a doctor or nurse wearing different attires with associated questions (Figure 2): I. A doctor wearing personal clothes, II. A doctor wearing a white coat over personal clothes, III. A doctor wearing a white coat over scrubs, IV. A nurse wearing personal clothes and V. A nurse wearing scrubs. The two models, a female doctor and a female nurse, were not employed in the Psychiatric Services of the Central Denmark Region. Female models were chosen deliberately as most doctors and nurses in the Psychiatric Services of the Central Denmark Region are female. The photographs shown to groups III and V were the originals; those shown to the remaining groups were modified from the originals using Google Gemini (version 2.5 Flash), Midjourney (version 7.0), and Adobe Photoshop (version 2026). Notably, the scrubs (light grey) shown in the photographs were purchased for the purpose of the survey and differ from those worn by staff at the general hospitals in the Central Denmark Region (sky blue colour).
The photograph shown to each of the five randomisation groups. The Roman numerals refer to the five intervention groups: (I) a doctor wearing personal clothes, (II) a doctor wearing a white coat over personal clothes, (III) a Doctor wearing a white coat over scrubs, (IV) a nurse wearing personal clothes and (V) a nurse wearing scrubs.
Randomisation to intervention groups
As SurveyXact does not support standard 1/5, 1/5, 1/5, 1/5, 1/5 random allocation, the platform was configured to assign participants to one of the five intervention groups based on the time they accessed the survey link in the invitation letter. Specifically, participants who accessed the link in the millisecond intervals (within each second) from 0 to 199, >199 to 399, >399 to 599, >599 to 799 and >799 to <1000 were assigned to groups I, II, III, IV and V, respectively. We have previously demonstrated that this pseudo-randomisation method works well on the SurveyXact platform (Perfalk et al., Reference Perfalk, Bernstorff, Danielsen and Østergaard2024, Reference Perfalk, Bernstorff, Danielsen and Østergaard2025). The particiants were unaware (blinding) that a randomisation and intervention took place and that other participants were shown other photographs.
Baseline questionnaire
Upon entering the survey, regardless of the intervention group, participants initially completed a baseline questionnaire about their education level, current employment status, household composition (adults and children), birthplace (Denmark or abroad), birthplace of both parents (Denmark or abroad), general trust, trust in various Danish institutions and satisfaction with treatment in the Psychiatric Services of the Central Denmark Region. The items on trust and treatment satisfaction were rated on Likert scales from 0 to 10 (minimum to maximum).
Questions following the intervention (outcome measurement)
Participants were shown a photograph corresponding to their allocated group eight times. The first time, they were only instructed to take a good look at the photograph of the doctor/nurse and told that they would be asked questions about it in the subsequent part of the survey. On the following seven pages of the survey, the photograph (always the same for each participant) was accompanied by the following questions (translated from Danish): 1. How competent (skilled) do you find this doctor to be? 2. How credible do you find this doctor to be? 3. How empathic do you find this doctor to be? 4. How trustworthy do you find this doctor to be? 5. How likely will you be to follow this doctor’s recommendations about your treatment? 6. Imagine that you are an inpatient at a psychiatric ward and that you would like to discuss your treatment with the staff. How likely would you be to reach out to this doctor? 7. Taken together, how professional do you find this doctor to appear? For those allocated to one of the two nurse groups, ‘doctor’ was replaced by ‘nurse’ in all seven questions. All seven questions were rated on Likert scales from 0 to 10 (minimum–maximum).
Handling of survey responses
At the start of the survey, the participants provided their unique social security number and name to enable the extraction of supplementary clinical data (see the next section) and to handle multiple responses. In cases where a participant responded to the survey multiple times, only their first response was considered. If a participant initially provided a partial response, including the randomisation element, and then completed the survey under a different randomisation group, their full response was excluded from analysis as they had become unblinded to the randomisation. Since the questions serving as outcome measures were placed at the end of the survey, only those who completed the full survey were included in the analyses. Participants were required to answer all questions (no missing values allowed).
Supplementary clinical data from electronic health records
After reviewing the study information, participants provided consent for the extraction of sociodemographic data (age, sex and civil status) and clinical data (number and type of contacts [inpatient/outpatient] with the Psychiatric Services of the Central Denmark Region, ICD-10 diagnoses associated with these contacts and time since first contact with psychiatric services) from their electronic health records for use in this study. The electronic health record system has been in place since 2011. Consistent with Perfalk et al., (Perfalk et al., Reference Perfalk, Bernstorff, Danielsen and Østergaard2024), the following ICD-10 diagnostic subcategory hierarchy was used to define the most severe diagnosis over time: F2 (psychotic disorders) > F3 (mood disorders) > F4 (anxiety- and stress-related disorders) > F5 (eating, sleeping and other behavioural syndromes associated with physiological disturbances) > F6 (personality disorders) > F8 (developmental disorders including autism) > F9 (child and adolescent mental disorders) > F1 (substance use disorders).
Statistical analysis
All data management and statistical analyses were performed using RStudio (version 2025.05.0, Build 496), and a p < 0.05 was considered statistically significant. Correction for multiple testing was not performed as the analyses were highly interdependent (Rothman, Reference Rothman1990).
Sociodemographic and clinical characteristics were summarised using descriptive statistics. The time spent responding to the survey was compared across intervention groups using the Kruskal–Wallis H-test.
To compare responses across intervention groups, a composite approval score was created based on the seven survey items. First, a confirmatory factor analysis was performed, and items with strong loadings (>0.7) on the latent factor were included in the composite approval score (Brown, Reference Brown2006). Then, the ratings of the included items were summed for each participant to generate the composite approval score. Finally, a multi-group confirmatory factor analysis was performed to assess measurement invariance across intervention groups.
In the primary analyses, composite approval scores were compared pairwise across the three groups allocated to the three different photographs of the doctor and between the two groups allocated to the two different photos of the nurse using two-sample t-tests. In the secondary analyses, item scores were compared pairwise among the three doctor intervention groups and between the two nurse groups using two-sample t-tests. Finally, the primary analyses were repeated for the following subgroups: males, females, younger (age below the median split), older (age above the median split), completed higher education, not completed higher education, born in Denmark and both parents born in Denmark, born outside Denmark or at least one parent born outside Denmark, general trust above the median, general trust below the median), trust in the Psychiatric Services of the Central Denmark Region above the median, trust in the Psychiatric Services of the Central Denmark Region below the median, at least one inpatient stay, no inpatient stays, most severe diagnosis being affective/anxiety disorder, and most severe diagnosis being psychotic/other disorder. Only preregistered secondary analyses were performed.
Ethics
Survey-based research studies are exempt from ethical review board approval in Denmark. This study received a waiver of ethical approval from the Health Research Ethics Committee of the Central Denmark Region (no. 1-10-72-13-24). It was further approved by the Legal Office of the Central Denmark Region (reg. no. 1-45-70-99-24) and registered on the internal list of research projects having the Central Denmark Region as the data steward (reg. no. 1-16-02-109-25).
Results
A flowchart illustrating invitation of the study population and their participation is shown in Figure 1. In brief, a total of 9310 patients were eligible and invited to participate, 1431 (15.4%) completed the full survey and 136 who provided partial responses were excluded. The respondents’ sociodemographic and clinical characteristics, as well as information regarding interpersonal and institutional trust, are presented in Table 1. The clinical characteristics of the 1567 randomised participants (both those with complete and partial responses) are provided in Supplementary Table 1.
Characteristics of participants with complete responses
Table 1. Long description
The table presents sociodemographic and clinical characteristics of participants with complete responses. It includes data on age, sex, marital status, educational level, current work status, sick leave, unemployment, household adults and children, place of birth, mother’s and father’s place of birth, general health, treatment in various services, satisfaction with treatment, and main severe psychiatric diagnosis. The table has 43 rows and 12 columns. Column headers include Age, Sex, Marital status, Educational level, Current work status, Sick leave, Unemployed, Adults in household (except participant), Children in household, Place of birth, Mother’s place of birth, Father’s place of birth, General health, Treatment in psychiatric, Treatment in healthcare services, Satisfaction with treatment in the Psychiatric Services of the Central Region, Years since first contact to Psychiatric services, Main severe psychiatric diagnosis, and Prior admissions. Row labels include categories and subcategories for each characteristic. Notable trends and specific values are detailed within each cell.
1 Most severe main psychiatric diagnosis during the period with available data (2011-time of study).
‘Fx’ refers to the ICD-10 block.
‘-’: small numbers that cannot be reported due to risk of identification.
Among the 1431 participants with complete responses, 305 (21.3%) were allocated to the photograph of the doctor wearing personal clothes, 271 (18.9%) were allocated to the photograph of the doctor wearing a white coat over personal clothes, 260 (18.2%) were allocated to the photograph of the doctor wearing a white coat over scrubs, 293 (20.5%) were allocated the photo of the nurse wearing personal clothes, and 302 (21.2%) were allocated to the photograph of the nurse wearing scrubs. The median time spent responding to the survey was 4 minutes (interquartile range: 3–6) and did not differ statistically significantly across groups (p = 0.292).
The results of the confirmatory factor analysis are presented in Supplementary Table 2 and Supplementary Figure 1. As expected a priori, all survey items loaded strongly on the latent factor (>0.70). Thus, the composite approval score for each respondent was calculated by summing their ratings of all seven survey items yielding a range from 0 to 70.
The results of the primary analyses, which compared composite approval scores across the groups allocated to the three different photographs of the doctor and compared composite approval scores between the groups allocated to the two different photographs of the nurse, are shown in Figure 3. Notably, mean composite approval scores were very similar across the compared groups and did not differ statistically significantly between them. The mean differences in composite approval score (the latter subtracted from the former) were −1.4 (95%CI: −3.6;0.9, p = 0.244) for doctor wearing personal clothes versus doctor wearing a white coat over personal clothes, −0.6 (95%CI: −2.9;1.7, p = 0.608) for doctor wearing personal clothes versus doctor wearing a white coat over scrubs, 0.7 (95%CI: −1.7;3.1, p = 0.543) for doctor wearing a white coat over personal clothes versus doctor wearing a white coat over scrubs, −0.3 (95%CI: −2.6;2.1, p = 0.826) for nurse wearing personal clothes versus nurse wearing scrubs. The individual item scores were also very similar across the groups (see Supplementary Table 3).
Comparison of composite approval scores among the three doctor groups and between the two nurse groups. The p-values are derived from two-sample t-tests among the three doctor groups and the two nurse groups. The dot represents the mean, and the error bars represent 95% confidence intervals around the mean.
The results of the subgroup analyses (sex, age, education level, place of birth, general trust, trust in psychiatric services, prior admission and diagnostic group) are listed in Supplementary Table 4. Composite approval scores did also not differ statistically significantly between randomisation groups in these subgroup comparisons. The only exception was for the subgroup of respondents with general trust above the median, where those allocated to the photograph of the doctor wearing a white coat over personal clothes had higher mean composite approval scores than those allocated to the photograph of the doctor wearing personal clothes (score difference of 2.6 [95%CI: 0.1–5.1], p = 0.039), potentially due to multiple testing).
Discussion
In this randomised survey experiment conducted in a large psychiatric service in Denmark, patients’ mean approval ratings of staff were highly similar across attire conditions (personal clothes vs. uniforms). The results remained stable across item-level and subgroup analyses.
To our knowledge, this study is the first to examine the effect of staff attire in a psychiatric setting using a randomised design. In a real-world randomised trial in Kyunggido, South Korea, Chung et al. (Reference Chung, Lee, Chang, Kim, Lee, Park and Chae2012) evaluated the effect of staff attire among patients seeking treatment for pain at a traditional Korean medical clinic. They randomly assigned patients to be treated by the same doctor who wore four different attires: a traditional uniform (a more than 2000-year-old custom), a white coat, a suit, or personal clothes. Participants assigned higher competence and empathy scores when the doctor wore the traditional uniform or a white coat than when they wore a suit or personal clothes (Chung et al., Reference Chung, Lee, Chang, Kim, Lee, Park and Chae2012).
Several other studies have used photographs to investigate the effect of staff attire in psychiatric settings in other countries, reporting somewhat mixed findings. Among patients attending an urban psychiatry resident’s clinic in Memphis (TN, USA), Long et al., found a preference for personal professional attire over personal professional attire plus a white coat (Long et al., Reference Long, Morton and Taylor2017). Among patients referred to outpatient psychiatric treatment in Zonguldak, Turkey, Atasoy et al., reported the following order of preference for psychiatrist attire: white coats > personal casual clothes (shirt, sweater and jeans) > suits > personal casual/sports clothes (T-shirt and capri pants) (Atasoy et al., Reference Atasoy, Senormanci, Saracli, Dogan, Kaygisiz, Atik and Orsel2015). Among inpatients at a psychiatric hospital in London, UK, Gledhill et al. found a preference for consultant/trainee psychiatrists wearing personal professional attire or white coats over personal casual clothing (e.g. T-shirt and jeans) (Gledhill et al., Reference Gledhill, Warner and King1997). Using the same photographs as Gledhill et al., Eikhom et al., found that inpatients and outpatients at a psychiatric hospital in a rural part of Norway preferred personal professional attire (shirt/blouse and trousers) over personal casual attire and white coats (Eikhom et al., Reference Eikhom, Torsaeter and Wik2006). Furthermore, for inpatients and outpatients at a psychiatric hospital in Croatia, Škaric et al., found that patients preferred that psychiatrists wear personal professional attire or white coats rather than personal casual clothes (e.g. T-shirt and jeans) (Škarić, Reference Škarić2024). Notably, all patients participating in the studies discussed above (Gledhill et al., Reference Gledhill, Warner and King1997; Eikhom et al., Reference Eikhom, Torsaeter and Wik2006; Atasoy et al., Reference Atasoy, Senormanci, Saracli, Dogan, Kaygisiz, Atik and Orsel2015; Long et al., Reference Long, Morton and Taylor2017; Škarić, Reference Škarić2024) were shown more than one photograph in a non-randomised, non-blinded design, which likely introduces a social desirability bias. Such bias is substantially less likely to have affected our findings, as the participants were shown only one photograph and were blinded to the fact that other participants were shown other photographs.
We speculate that the apparent lack of superiority in patients’ approval of uniformed staff in our study is likely partly driven by the fact that patients in Denmark are accustomed to staff wearing personal clothes in psychiatric services and, therefore, respond negatively to situations that counter this expectation (uniformed staff), that is, a confirmation bias (Pilgrim et al., Reference Pilgrim, Sanborn, Malthouse and Hills2024). This bias has been briefly discussed in other studies (Gledhill et al., Reference Gledhill, Warner and King1997; Eikhom et al., Reference Eikhom, Torsaeter and Wik2006; Gherardi et al., Reference Gherardi, Cameron, West and Crossley2009; Chung et al., Reference Chung, Lee, Chang, Kim, Lee, Park and Chae2012; Atasoy et al., Reference Atasoy, Senormanci, Saracli, Dogan, Kaygisiz, Atik and Orsel2015; Petrilli et al., Reference Petrilli, Mack, Petrilli, Hickner, Saint and Chopra2015; Škarić, Reference Škarić2024) that examined the effects of the most widely used attire in different countries, such as suits in England (Gledhill et al., Reference Gledhill, Warner and King1997) and more casual attire in Norway (Eikhom et al., Reference Eikhom, Torsaeter and Wik2006). However, none of these studies have adequately tested or accounted for this bias. Furthermore, the personal clothes shown in our survey is not representative of the large variety of clothing worn by the hundreds of clinical staff members in the psychiatric services that served as the survey setting. Indeed, the personal clothing shown in our survey is likely more neutral and appropriate than that generally worn by staff, potentially upwardly biasing respondents’ ratings of the doctor and nurse wearing personal clothes. Notably, in studies where casual clothes received poor approval from participants, the clothes shown were very casual (e.g. T-shirt and capri pants) (Atasoy et al., Reference Atasoy, Senormanci, Saracli, Dogan, Kaygisiz, Atik and Orsel2015), jeans and T-shirts with print (Gherardi et al., Reference Gherardi, Cameron, West and Crossley2009; Škarić, Reference Škarić2024) and short dress/skirt (Rehman et al., Reference Rehman, Nietert, Cope and Kilpatrick2005)).
In terms of the practical implications of this study, the results can be read in two ways. First, it can be argued that the lack of difference between the intervention groups support the current practice of clinical staff wearing their private clothes. On the other hand, it can be argued that the results are in favour of the use of uniforms. Indeed, despite confirmation bias, which is likely to be quite strong, and the models wearing personal clothes that are likely more neutral/appropriate than those generally worn by staff, composite approval scores were very similar for the groups allocated to photographs of staff wearing uniforms and those allocated to photographs of staff wearing personal clothes. Given this parity, and the obvious advantages in terms of hygiene and safe identification of staff, we would argue that more consistent use of uniforms should be considered. Currently at the hospitals in the Psychiatric Services of the Central Denmark Region, where the survey took place, staff working in cleaning and in the staff cafeteria can be more readily identified than the clinical staff, as the former wear uniforms and the latter wear personal clothes (with ID cards). At hospitals treating patients with severe mental illness and associated social cognitive dysfunction (Weightman et al., Reference Weightman, Air and Baune2014; Green et al., Reference Green, Horan and Lee2015) who may struggle to distinguish people, this practice is highly questionable.
On January 1, 2026, in accordance with the ‘Agreement on equality and integration of psychiatry and somatic medicine’ (The Ministry of the Interior and Health, 2025) between the Danish Government and the Danish Regions, psychiatric services and general (somatic) hospital services were merged at the managerial and organisational level. The agreement states (translated from Danish) that ‘The parties agree that psychiatry should be placed on an equal footing with other medical specialties, and that psychiatry and somatic medicine should be integrated in terms of management and organisation. This should primarily ensure that patients experience greater coherence in their patient journey, also across specialties.’ (The Ministry of the Interior and Health, 2025) and that ‘It will also support the continued destigmatization of mental illness – also among healthcare personnel – and contribute to people with mental illness experiencing a greater sense of being part of the overall healthcare system’ (The Ministry of the Interior and Health 2025). We argue that ensuring equality across settings in staff attire, through the use of uniforms in psychiatric services, is required to realise these ambitions.
The study had some limitations that should be acknowledged and considered when interpreting the findings. First, the response rate was relatively low, which introduces a risk of selection bias. Legal restrictions prevented us from obtaining clinical and sociodemographic information for all invitees, and we consequently were unable to adjust the analyses for attrition. However, in a prior study involving the same population that examined psychological distress among patients during the COVID-19 pandemic (Kølbæk et al., Reference Kølbæk, Jefsen, Speed and Østergaard2021, Reference Kølbæk, Gil, Schmidt, Speed and Østergaard2023), we were allowed to extract clinical and sociodemographic data on non-responders (a force majeure decision due to the pandemic) for attrition adjustment via inverse probability weighting, which did not affect the results. Although this does not eliminate the risk of selection bias in the present study, it suggests that such bias is unlikely to be a major concern. However, it could rightfully be argued that patients with pronounced social cognitive problems may have been more likely to opt out of participation. The direction and magnitude of any resulting selection bias cannot be determined from the available data. Second, another source of selection bias is by design. Specifically, patients were ineligible for participation if they had a forensic sanction, received coercive treatment, or had an organic mental disorder or mental retardation. This decision, which is in accordance with prior work (Perfalk et al., Reference Perfalk, Bernstorff, Danielsen and Østergaard2024), was made to ensure that the consent for participation was in fact voluntary and informed. Third, a power calculation based on a small effect size (Cohen’s d = 0.2) indicates that 400 participants per intervention group are needed to reduce the risk for type II error. Thus, as each group comprised 260–300 participants, small effects may have gone undetected. Fourth, while the method employed (an electronic survey with participants randomly allocated to different photographs) allowed for a large sample (n = 1431) at low cost, it is suboptimal in terms of face validity. Indeed, a randomised controlled study in the actual clinical setting, exposing patients to staff wearing different attires, would be preferable. As discussed above, such a ‘real-world’ study was conducted by Chung et al. (Reference Chung, Lee, Chang, Kim, Lee, Park and Chae2012). Notably, the study by Chung and colleagues was preceded by that of Chang et al. (Reference Chang, Lee, Lee, Park and Chae2011), which addressed the same research question in the same geographical and clinical setting, but rather than actually being confronted with staff wearing different attires, the participating patients rated photographs of a doctor wearing similar attires as those used by Chung et al., (traditional uniform, white coat, suit and casual clothes) displayed on a computer screen. The results were similar to those found by Chung et al. (Reference Chung, Lee, Chang, Kim, Lee, Park and Chae2012), suggesting that the use of electronic surveys displaying photographs is a reasonable approach to assessing the impact of attires on patient perception of staff. Fifth, we did not employ a validated questionnaire to assess patient approval of staff attire because, to our knowledge, no such instrument is available. Instead, questions were phrased to reflect the construct of interest, and a latent dimension was extracted through confirmatory factor analysis. Although this approach has an inherent limitation, we consider the resulting composite approval score to be a reasonable outcome measure. Sixth, the survey lacked attention-check questions, meaning that some respondents may have responded inconsistently/arbitrarily to the questions, potentially introducing noise into the data and complicating signal detection. However, respondents received no incentive to complete the survey and were free to exit at any point, likely decreasing the risk of non-attentive responding.
In conclusion, in this randomised survey experiment conducted in a large psychiatric service system in Denmark, patients’ mean approval ratings of staff were highly similar across attire conditions (personal clothes vs. uniforms). This result remained stable across item-level and subgroup analyses. Given this parity, and the obvious advantages in terms of hygiene and correct identification of staff, we believe that more consistent use of uniforms in psychiatric services should be considered.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/neu.2026.10103.
Data availability statement
The data cannot be shared as the participants have not consented to data sharing.
Acknowledgements
The authors are grateful to Bettina Nørremark for data management. The authors are grateful to Henriette Andreasen Svendsen and Marie Hvas Ellerbæk for survey distribution and handling of questions from invitees. The authors are grateful to medical student Linnea Sehested Skafte for assistance with the literature search. The authors are grateful to the doctor and the nurse modelling for the photographs used in the survey.
Author contributions
The contributions are reported according to CRediT.
Conceptualisation: SDØ, RH, PK, EP
Data curation and formal analysis: EP, AØ
Methodology: SDØ, RH, AØ, PK, EP
Project administration: SDØ, PK, EP
Supervision: SDØ, PK
Visualisation: EP, AØ.
Writing – original draft: SDØ, EP
Writing – review & editing: RH, AØ, PK,
Funding statement
There was no specific funding for this study.
Competing interests
Outside this study, SDØ reports funding from the Lundbeck Foundation (grants R358-2020-2341 and R344-2020-1073), the Novo Nordisk Foundation (grant NNF20SA0062874), the Danish Cancer Society (grant R283-A16461), the Central Denmark Region Fund for Strengthening of Health Science (grant 1-36-72-4-20), the Danish Agency for Digitisation Investment Fund for New Technologies (grant 2020-6720) and Independent Research Fund Denmark (grant 7016-00048B and 2096-00055A). SDØ received the 2020 Lundbeck Foundation Young Investigator Prize, and SDØ owns/has owned units of mutual funds with stock tickers DKIGI, IAIMWC, SPIC25KL, DKIEUIXBNP, DK0016023229 and WEKAFKI, and owns/has owned units of exchange traded funds with stock tickers BATE, TRET, QDV5, QDVH, QDVE, SADM, IQQH, IQQJ, USPY, EXH2, 2B76, IS4S, OM3X, MCHI and EUNL. Outside this study, PK reports funding from the Independent Research Fund Denmark (grant 5336-00022B). The remaining authors have no financial interests to disclose.
Open access
