Background
The contribution of the built environment of health-care facilities to treatment outcomes is increasingly acknowledged.189 However, the relationship of psychiatric ward design to patient outcomes, as well as to the overall experience of treatment, is a particularly under-researched area. This is rather surprising as, in UK policy, evidence-based design has become a cornerstone for the construction of new facilities under the Private Finance Initiative.189 Furthermore, current UK policy guidelines for the design of psychiatric wards stipulate that ‘service users . . . should have an early and continuing opportunity to contribute to the new service design and participate in the planning process’.189 However, no study thus far has been based on such a contribution. Although some have elicited service user and front-line staff perceptions of design, no study to date has worked with users to create a robust tool for the evaluation of that environment. The present study grew out of WP1 and applied participatory methods with the aim of eliciting service user and staff priorities concerning psychiatric ward design in order to provide better tools for the future.
Our study had three stages.
Systematic review to evaluate the evidence on the impact of the design of psychiatric environments on treatment outcomes and staff well-being (March 2012).
Measure development using a model of ‘stakeholder involvement’ and participatory methodology (April–August 2012 and July–August 2013).
Qualitative study of service user experience of ward environment using photo elicitation (June–August 2012).
Stage one: literature review
Evidence on the contribution of psychiatric design to clinical outcomes is limited. A total of 23 studies were identified (20 quantitative and three qualitative). Only three studies measured the impact of facility design on symptom reduction. The majority of the remaining studies explored the relationship between the physical environment, social behaviour and treatment or job satisfaction. Several studies suggested that private spaces and homely features might encourage social interaction in service users. However, staff and service users often had conflicting responses to the same features. In addition, many of the findings are inconclusive, as several studies used small samples, not all involved control groups, and many did not account for confounding variables. Moreover, no study developed a user-generated measure for the assessment of ward design.
Stage two: measure development and pilot study190
Setting
The study took place in an inner city NHS mental health trust. Four 18-bed adult acute wards participated. Data collection took place in April–May 2012 (measure development), June–August 2012 and July–August 2013 (pilot study). Inclusion criteria for service user participants were that they had been present on the ward for 3 days prior to data collection and were able to provide informed consent. Inclusion criteria for staff were that they had been working on the ward for at least 4 weeks. Demographic data were collected by self-report from both groups, while clinical data from the service users were collected from NHS records.
Measure generation
Questionnaires were produced through a participatory methodology.17 Two service user researchers reviewed the literature and produced topic guides which formed the basis of interviews with staff and service users on the participating wards. The interviews were audio-taped, transcribed and thematically analysed by the service user researchers using NVivo 9 (QSR International, Warrington, UK). Key themes identified were then used to develop questionnaire items.
Analysis
The internal reliability of the measure was assessed using Cronbach’s alpha83 with data from the questionnaire phase of the study. Test–retest reliability was carried out with participants who completed the measures on two occasions. Test–retest reliability was estimated for each item using weighted Kappa (quadratic weighting to emphasise the importance of minor differences between time points) and alpha statistics. Based on previous research, we expected that service users who were involuntarily admitted and those from BME groups would show more negative perceptions of the ward environment.
We assessed any differences with one-way ANOVA on views on the ward design.
Analyses were carried out using Stata 12.1 and Statistical Product and Service Solutions (SPSS) version 20 (IBM Corporation, Armonk, NY, USA).
Results
Sample characteristics
Ten service users and 10 nurses were recruited into the measure generation phase of the study. Fifty-three service users and 61 staff were recruited into the piloting phase of the study. A total of 50% of service users in the interview phase and 58.5% in the piloting phase were from BME groups, and the most frequent diagnosis was psychosis spectrum disorder (40% and 46%, respectively). The majority of staff participants were also from BME groups in both the interview and the piloting phases (70% and 56.8%, respectively), and all the participants in the interview phase and 83.7% in the piloting phase were qualified nurses.
Measure generation
Two measures were generated: the service user questionnaire included 20 items, while the staff questionnaire included 22 items. The items were rated via a six-point Likert scale. In addition, free-text spaces were included to provide qualitative data.
Service user measure
Feasibility and acceptability
All 34 participants found the questionnaire easy to understand and complete, but 11% (n = 4) felt that it was too short. The Flesch reading ease score was 84.6 (for 8- to 9-year-olds).
Psychometric testing
Following initial reliability checks, it was decided that items Q8 and Q13 should be removed owing to underperformance in terms of kappa. The final measure comprised 18 items. Internal consistency analyses indicated a Cronbach alpha of 0.91 with an overall total scale ICC of 0.67 indicating moderate test–retest reliability.
Exploratory analyses
Service users from a white background were more negative about the ward design than BME service users (F = 5.30; p = 0.026). Those service users diagnosed with a psychosis spectrum disorder had lower scores; that is, they viewed the ward design more positively than those services users with other diagnoses (F = 5.20; p = 0.009).
Staff measure
Feasibility and acceptability
All 53 respondents found the questionnaire easy to understand and complete and 96% felt that the length was appropriate. The Flesch readability score was 70.4 (for 12-year-olds).
Psychometric testing
All items were found to be reliable and, therefore, no items were dropped.
The average item–item correlation was 0.42 and the alpha statistic for the overall standardised scale was 0.94.
Exploratory analyses
Following univariate analyses, we found that staff participants from a white background were similar to service users in their more negative assessment of ward design. No other group differences were found.
Stage three: photo elicitation
Sample
Service users on two out of the four acute wards involved in the measure development participated in the study. Inclusion criteria for service user participants were that they had been present on the ward for 3 days prior to data collection and were able to provide informed consent.
Method
Participants were given a point-and-click digital camera and asked to take two photographs: one depicting the best and one depicting the worst part of the ward. Two service user researchers briefly interviewed participants on their views about the aspects of the ward represented on the photographs. Interviews were audio-taped and transcribed verbatim. The service user researchers then thematically analysed both photographs and transcripts using NVivo 9.
Preliminary results
A total of 36 service users participated in the study. Key themes included concerns with hygiene and maintenance, positive evaluation of items representing leisure activities (TV, ping pong table), appreciation of privacy.
Discussion
Both measures showed good or reasonable test–test reliability and internal validity and this should be further strengthened in a larger study. Service users may change their view as they experience the ward for longer periods of time and, therefore, this in particular needs further exploration to detect relationships with other measures (e.g. length of admission, previous admissions, mental state, etc.). The lack of observable differences according to legal status and the positive ratings among BME service users and users with psychosis spectrum disorders were not anticipated and merit further study. UK studies indicate that BME service users are more negative about their experience of mental health services and about inpatient care in particular.191 However, it is possible that the negative views reported in these studies refer to other aspects of the treatment environment (e.g. relations with staff, continuity of care, violence in the ward). Photo elicitation data suggest that service users prioritise privacy and this may be more important than other design considerations.
Concluding remarks
This study has demonstrated that a robust tool for the evaluation of inpatient psychiatric facilities can be generated through participatory methods, making use of staff and service user perceptions of ward design.
