Does the Redesign of a Psychiatric Inpatient Unit Change the Treatment Process and Outcomes?

Participants were 290 adults receiving care for mood and anxiety disorders at a large mental health facility in Toronto, Canada. The recruitment period (March 2007–May 2009) bracketed the redesign, such that a full year of data collection was conducted for both the preexisting and the redesigned units. Data collection took place prior to the redesign in the original 16-bed unit and after the redesign in the new 24-bed unit. All individuals admitted (N=377) were eligible for participation. The final sample represented 77% of those approached for participation. Reasons for nonparticipation included not feeling well enough, being too busy, or being unavailable to complete the surveys when approached.

Recruitment and data collection were conducted by the research team. Self-administered questionnaires were completed at admission and discharge. Of the 272 participants enrolled at admission, 185 (68%) were followed up at discharge. An additional 22 individuals were recruited into the study at discharge but did not complete the admission surveys. Data from the surveys were supplemented by chart review. The study was approved by the hospital’s research ethics board, and all participants provided written informed consent.

Those who completed discharge surveys were older than those who did not (39.8 versus 36.4 years old, t=2.14, df=283, p=.033). No significant differences (α=.05) were found in the rates of follow-up by study phase (before or after the redesign), gender, employment status, marital status, diagnosis, prior hospitalizations, or baseline quality of life or functioning. Further, the data collected via chart review were available for all patients, regardless of whether they completed the surveys.

Participants’ perceptions of ward atmosphere were assessed at discharge by using the Ward Atmosphere Scale (WAS) (4). Of the ten WAS dimensions, we focused on three that correspond closely to the objectives of the redesign: peer support, patient autonomy, and practical skill development. The average scores of items in each dimension were calculated to yield subscale scores; possible scores range from 0 to 1, with higher scores indicating a more positive ward atmosphere. Treatment satisfaction was assessed at discharge with the eight-item Client Satisfaction Questionnaire (CSQ-8) (9). The average score of the items was calculated to yield an overall score between 1 and 4, with higher scores indicating greater satisfaction. Mental health–related quality of life was assessed at admission and discharge with the mental health component score from the Short-Form 36 (SF−36) (10). Raw scores were standardized and normed to general population levels (50±10). Functioning was rated by physicians at admission and discharge with the Global Assessment of Functioning (GAF) (11). Possible scores range from 1 to 100, with higher scores indicating better functioning. The GAF was abstracted from participants’ charts, along with patient characteristics and diagnosis.

Path analysis was used to test the hypothesis that the redesign would have an indirect effect on patient outcomes through changes in ward atmosphere. We calculated a single index for ward atmosphere by summing the WAS subscales, which were significantly correlated (r=.42–.48). We report the total (unmediated) effect of redesign on outcomes by comparing outcomes before and after the redesign. We also report the indirect effect of the redesign on outcome (that is, the product of the two pathways linking redesign with ward atmosphere with outcomes) and the direct effect of the redesign after controlling for ward atmosphere. Standard errors for the indirect effects were bootstrapped (500 replications) (12). For the models predicting quality of life and global functioning, the corresponding baseline score was included as a covariate. As such, these models assess change during treatment in each domain. Analyses were conducted in Mplus 6.2 using full information maximum likelihood to minimize potential selection biases associated with missing data (13).

Participants ranged in age from 18 to 71 (mean±SD=38.8±12.3), and a slight majority were female (N=154, 54%). Over half (N=153, 54%) were single; 84 (30%) were married or partnered; and 46 (16%) were separated, divorced, or widowed. At admission, 91 (32%) were employed, 42 (15%) were unemployed, and 114 (40%) were not currently in the workforce. Most participants were diagnosed as having major depressive disorder (N=137, 49%) or bipolar disorder (N=110, 39%). Seventy percent (N=195) had been previously hospitalized for a mental disorder.

Participants recruited after the redesign had better functioning at admission than participants recruited before the redesign (GAF score of 54.4±11.5 versus 46.9±10.5, t=−5.74, df=283, p<.001). There were no differences by gender, age, employment status, diagnosis, or quality of life. [A table describing participants’ demographic and clinical characteristics is available online as a data supplement to this report.] After adjustment for participants’ baseline functioning, the redesigned unit was perceived to offer greater peer support and autonomy, with more emphasis on practical skill development. [A figure summarizing WAS scores obtained before and after the redesign is available in the online data supplement.]

Table 1 summarizes findings from the path models. The total effect of the redesign was significant only in relation to global functioning. Mean GAF scores for participants recruited before the redesign were 22.5±13.7 points higher at discharge (69.4±9.5) than at admission. Among participants recruited after the redesign, GAF scores at discharge (65.3±10.6) were only 11.0±11.9 points higher than at admission. The bivariate association between unit redesign and global functioning at discharge was significant (b=−4.15, SE=1.26, p=.001), suggesting that the smaller change in functioning was not due entirely to ceiling effects. The direct path linking unit redesign with global functioning remained significant after the influence of ward atmosphere was partitioned out, while the indirect effect including ward atmosphere was nonsignificant.

In contrast, the total effects of the redesign were not significant in relation to quality of life or treatment satisfaction. Quality-of-life scores for participants recruited before the redesign were 5.6±12.1 points higher at discharge (24.3±13.9) than at admission (18.7±13.4). Among participants recruited after the redesign, quality-of-life scores at discharge (27.7±12.3) were 7.2±11.4 points higher than at admission (20.4±11.6). Treatment satisfaction was high for both groups (3.2±.6 and 3.3±.5).

The redesign, however, had a significant indirect influence on both outcomes through its influence on ward atmosphere. That is, an improved ward atmosphere accounted for the greater improvement in quality of life after the redesign. The redesign had no significant direct effect on quality of life, indicating there was no residual impact of redesign after accounting for the indirect effect transmitted through ward atmosphere. Likewise, after adjustment for baseline functioning, treatment satisfaction was higher after the redesign, and this was partially explained by the improvement in ward atmosphere. The direct effect of redesign on treatment satisfaction indicated that after partitioning out the positive influence transmitted through ward atmosphere, the redesign had a residual negative influence on treatment satisfaction.

This work offers partial support for a conceptual framework linking treatment setting with patient outcomes (4). As expected, participants on the newly redesigned unit perceived a more positive atmosphere in terms of greater peer support and autonomy and more emphasis on practical skill development. This, in turn, was associated with greater improvement in mental health–related quality of life during treatment and greater treatment satisfaction. In contrast, global functioning improved to a slightly lesser degree after the redesign, and this association operated independently of ward atmosphere.

Our study sheds light on the complexity of the relationships linking treatment setting with outcomes. The results suggest that ward atmosphere acted as a suppressor of a negative association between the redesign and treatment satisfaction. The reasons underlying the decline in treatment satisfaction require further investigation, given that they may offset the potentially positive effects of improvements to the treatment setting. In a companion study exploring staff perceptions of the impact of the unit redesign on service delivery and the work environment, clinicians reported a number of challenges (14). Many challenges concerned the tension between supporting patient autonomy and ensuring patient well-being and safety. It is possible that the challenges experienced by staff affected patients and were reflected here in ratings of treatment satisfaction. With data collection taking place on the redesigned unit during the first year of operations, it is also possible that our findings reflect to some degree the challenges associated with the transition and process of reestablishing clinical and operational routines. Nonetheless, it appears that the intended positive influence of the redesign on patients’ experiences of treatment was to some extent realized in terms of patient satisfaction and quality of life.

In contrast, the negative association between redesign and change in global functioning was independent of ward atmosphere. The mechanism linking these factors is also worthy of further investigation, although an examination of mean GAF scores at admission and discharge provides some clarification. Whereas the change in GAF scores during hospitalization was smaller after the redesign, participants exhibited significantly higher functioning at admission, thus limiting to some degree the amount of improvement that could be achieved. Further, although GAF scores at discharge were significantly lower after the redesign, the difference was relatively small. In light of these figures, it seems unlikely that the smaller change exhibited after redesign represented a real negative impact of the new environment on patient functioning. To rule out the latter explanation, it would be helpful to investigate other potential mediators, such as clinician stressors related to environmental uncertainty and challenges relating to the new physical layout (14), and to consider their persistence following the initial period of transition.

Strengths of this study included the use of both patient and clinician perspectives in assessing outcomes, the robust analytical approach toward testing for mediation, and the use of recommended procedures to handle missing values. Nonetheless, study attrition (32%) remained a limitation. Randomization of participants to the traditional and redesigned units would have provided a stronger test of the redesign effort; however, this was not possible within the context of the evaluation.

An understanding of the ways in which physical features and clinical routines affect the psychosocial environment of inpatient units and patient outcomes is critical to the effective design of therapeutic spaces. Given the potentially high cost of facility renovations, more work is needed to highlight the specific areas that redesign efforts should target. Efforts to improve the inpatient environment by supporting patient autonomy, peer support, and practical skill development may be expected to meet with improvements in patient outcomes, at least in terms of quality of life and satisfaction with treatment. Such findings are consistent with a design focus that is patient centered (1) as well as with broader perspectives on recovery-oriented services (15).

This work was financially supported by the research program at the Centre for Addiction and Mental Health (CAMH). Support to the centre for salary of scientists and for infrastructure has been provided by the Ontario Ministry of Health and Long-Term Care. The views expressed do not necessarily reflect those of the ministry.

Dr. Mulsant is a member of the board of directors of the CAMH Foundation. He receives research support from Bristol-Myers Squibb and Pfizer (medications for a clinical trial funded by the National Institutes of Health). He owns stock in General Electric and has received some travel support from Roche. The other authors report no competing interests.