Association of Empathy of Nursing Staff With Reduction of Seclusion and Restraint in Psychiatric Inpatient Care

This study examined variables that might be related to use of seclusion or restraint on an acute inpatient psychiatric unit at San Francisco General Hospital. The study was reviewed and approved by the University of California, San Francisco, institutional review board. The hospital reflects the great ethnic diversity of San Francisco, with many immigrants without English proficiency, and the psychiatry service at that time included four inpatient units: a Hispanic unit, an African-American unit, an Asian unit, and a mixed HIV/gay/lesbian/transgender unit. Subsequent budget cuts eventually led to one large inpatient unit, although language-capable teams remain to some extent. This research was done in the Asian unit, which limits somewhat the findings’ generalizability but reflected the first author's interest in Asian meditation and empathy. Non-Asian patients were admitted when the Asian ward had the only open bed, for an average of 60% Asian Americans (range 20%–91%) during the study.

Seclusion and restraint usage as well as other variables were examined in all three shifts each day (night, day, and evening shifts every day of the week) during two six-month periods (July 1 to December 31 in 2003 and 2004; 549 shifts in 183 days in each period). After the first data period, the first author offered mindfulness-based empathy training (15) to the day and evening shift members of the nursing staff. (The night shift rarely initiated seclusion or restraint.) In 2004, an early group of staff volunteers was trained during March and April, and a second group received training during July through October.

Recorded data included all patients and permanent nursing staff who spent any time on the unit during each shift. Data on nursing staff who were present on the unit focused on core staff permanently assigned to the ward, except that float staff assigned during periods of high patient acuity were included in the count of total nursing staff hours worked on each shift. Presence of each attending and resident psychiatrist was recorded for each day.

The independent variable of primary interest was the empathy of nursing staff members who were working either day or evening shifts. The nurse supervisor, three other senior nurses, and one attending psychiatrist (C-PPY) independently rated 32 core nursing staff members on their ability and motivation to engage patients in a respectful, caring, therapeutic relationship. Nurses were not aware they were being rated. Raters gave a score of 1 (below average) to 5 (above average) to each of the nursing staff whom they felt they knew well enough to rate. No nursing staff members had fewer than two ratings. For each nursing staff member, ratings received were averaged to provide an estimate of their empathy with patients, which ranged from 2.5 to 5.0. Agreement among raters was moderate, with an intraclass correlation of only .22. Therefore, we simply dichotomized the nursing staff at the mean rating of 4.0 or higher (rounded from 3.72) as above-average ability and motivation related to empathy when working with patients and counted how many empathic nursing staff members were present on each shift.

The study outcome was a binary variable that represented any versus no entry of a patient into seclusion or restraint in each day or evening shift. Our hypotheses were that a greater mean empathy rating among the nursing staff on a shift would be associated with a lower probability of any entry into seclusion or restraint on that shift and that empathy training would reduce restraint in the second data collection period, after the training.

We first adjusted for factors other than empathy skill and motivation that could strongly predict the probability of use of seclusion or restraint. These factors included previous aggressive behavior, the number of patients very recently admitted to the ward, seclusion or restraint in the previous shift, and shift type (day versus evening, because entry into seclusion was higher in the evening shift in preparation for the low staffing of the night shift). Therefore, we adjusted for these predictors so they would not obscure the possibly subtle influence of staff empathy skills and motivation.

We used hierarchical logistic regression to test these hypotheses and used SAS Proc Logistic, version 9.2. We analyzed four sets of independent variables sequentially: shift variables, patient variables, staff variables, and then time period and empathy training and the interaction between them. Independent variables within each set were analyzed simultaneously.

The first set, shift effects, was categorized into three types of nursing shift: weekday day, any night, and other shifts (evening, weekend day, and holiday day). Weekday day shifts had the most staff members present and the lowest number of seclusion or restraint incidents. The second set of independent variables captured patient mix. Daily severity measures were not available for each patient, but these were approximated by using historical data about each patient before admission, while in the emergency department, and since admission. Initially, there were 13 patient variables, but we dropped three of them that were highly correlated with another patient variable, leaving ten for analysis. Third, we introduced staff measures, including the number of nursing staff with above-average empathy scores among the nursing staff on each day or evening shift, with zero used for the night shift. Fourth, we factored in the pre-post training period and the number of staff who were trained and the interaction of these two factors with the time period—whether before or after empathy training—to examine whether empathy training reduced the number of shifts with any incidents of seclusion or restraint.

Of 1,098 shifts, 79 (7.2%) included one or more new incidents of seclusion or restraint. Among these shifts, 21 were day shifts, 53 were evening shifts, and five were night shifts. This pattern reflects the need to protect the night shift, when there may be only one nurse present and safe implementation of seclusion or restraint requires off-ward help. Mean ratings on the empathy scale were averaged on each shift. When these means were averaged by shift type (day, evening, or night), the mean ratings were similar: day, 3.77; evening, 3.65; and night, 3.95. Table 1 shows the logistic regression results, indicating which shift characteristics increased the probability that one or more patients would be placed in seclusion or restrained.

Weekday day and night shifts had similarly low rates (odds ratio [OR]=.99, p=.98), but evening, weekend day, and holiday day shifts were more likely than weekday day shifts to have new incidents of seclusion or restraint (OR=2.18, p<.01), perhaps driven by the high evening incident rate. More patients restrained or secluded during the previous shift also predicted more incidents in the current shift (OR=1.82, p<.01), as did higher mean hours of seclusion or restraint in the emergency department among current patients. However, a larger proportion of Asian patients reduced the likelihood of use of seclusion or restraint (OR=.09, p<.01), as did higher numbers of patients secluded in the emergency department (OR=1.74, p<.03). The proportion of patients with a suicide attempt in the past two weeks also reduced use of seclusion or restraint (OR=.01, p=.04).

Staffing characteristics had an overall significant effect on use of seclusion or restraint (OR=.67, p<.01) after analyses controlled for shift and patient effects. This was driven primarily by a single staffing variable, number of nursing staff present with high empathy skill (ratings ≥4; OR=.67, p<.01). Although the result was clearly significant, the OR was modest, indicating a consistent but not large reduction of use of seclusion or restraint. A similar trend was found by other researchers in a study offering voluntary restraint-reduction classes (8). There was only a small, nonsignificant effect of the additional empathy training provided. We also found that shifts with more staff who had participated in empathy training had fewer incidents of seclusion or restraint, but that effect was not enhanced after the training. This is probably an artifact resulting from empathic staff being more likely to participate in the training (mean empathy rating of 3.4 among trainees but mean rating of 3.0 among those who did not participate in the training). This is similar to other findings that have shown that in voluntary restraint reduction training, those already preferring to reduce the use of restraint were more likely to participate and that further training did not change those attitudes (7–9).

The most important finding was the impact on reduced use of seclusion and restraint when more strongly empathic nursing staff members were present. Their influence resulted in a 33% reduction in the number of shifts with any new incidents of seclusion or restraint. This was at best a modest effect, and it is disappointing that the training did not enhance that effect. But the take-home message is that recruiting and retaining empathic nursing staff members is a strategy that can reduce the use of seclusion and restraint. The primary limitation of the study was generalizability, because we worked with only one inpatient ward.

In summary, we found fewer incidents of seclusion and restraint when more staff members with above-average empathy skills worked a shift. To our knowledge this specific association has not been previously reported. However, our finding is similar to Owen and colleagues’ (10) finding that the presence of younger staff and staff with higher levels of psychiatric experience has a negative association with violence. It seems likely that staff members’ ability to engage patients in a respectful, caring, therapeutic relationship may reduce disruptive patient behavior and thus reduce the use of seclusion and restraint. In conclusion, our results indicate that nursing staff with strong empathy skills and motivation reduced the use of seclusion and restraint in acute psychiatric inpatient care. Therefore, it is important to attract and retain nursing staff with empathic motivation and skill in acute inpatient psychiatric care.

This project was supported by a University of California, San Francisco, REAC grant. The authors thank Deanne Beuger-Moudgil, Jennifer Bozenski, and Amy Beddoe for assistance in the study.

The authors report no competing interests.