Events Preceding Changes in Conflict and Containment Rates on Acute Psychiatric Wards

The Patient-Staff Conflict Checklist—Shift Report (PCC-SR) is a validated (7), end-of-shift report that nurses complete to log the frequency of attempted patient conflict behaviors and the staff containment measures used in response. It was used during the Tompkins Acute Ward Study, a longitudinal, mixed-methods research project investigating care on 13 acute psychiatric wards and three psychiatric intensive care units in one large inner-London National Health Service (NHS) Trust (8,9). The trust served a population of 650,000 from three London boroughs, each of which had high proportions of residents from racial or ethnic minority groups (approximately 60%) and high levels of social deprivation. The median number of beds per ward was 18 (range between nine and 20), and a majority of wards had mixed-gender occupancy (one psychiatric acute ward was female only, and one psychiatric intensive care unit was male only). Overall, a total of 15,006 PCC-SRs were completed over a two-year period (August 2004–July 2006), with response rates by ward varying from 18% to 84% (mean response rate of 45%, although response rates fluctuated across all wards throughout the two years). By aggregating each of the conflict-specific and containment-specific items from the PCC-SR, a total conflict score and a total containment score for each shift were calculated.

Using these data, we generated a total conflict and total containment smoothed time-series line graph for each ward, resulting in 32 graphs. The prominent total conflict and containment transitions were then visually identified and marked on each graph. This involved marking the starting point of an upturn or downturn transition. The ending point of the transition was selected to be the point at which there was a prominent shift toward the opposite score direction (for example, an upturn that shifted to a downturn) or when the score horizontally flattened out. We applied no overall (cross-graph) rule concerning the minimum change in scores that would constitute a transition. As the total conflict and containment score means and variance differed from ward to ward, prominent score changes that emerged were assessed in relation to each graph's norms. When there was uncertainty about whether a particular transition should be identified as prominent, the transition was marked in order to decrease the likelihood of missing an important score transition.

An interrater reliability test was subsequently carried out to establish the level of reliability in the identification of key transitions. This was particularly important given the subjective nature of the transition identification process. The second rater (new to the research team) was provided with a random sample of 16 graphs. This provided a large enough sample of transitions to sufficiently detect a kappa of at least .70 (with N set at 90% power) (10). Transitions that were marked by both raters within a one-month time frame were considered rater agreement, whereas those identified outside of this time frame were considered rater disagreement. Both the Spearman's rho and Cohen's kappa statistics showed good rater agreement (ρ=.747; p<.001; κ=.738, p<.001).

A series of repeated qualitative semistructured interviews with ward managers and consultant psychiatrists on the 16 wards was also carried out during the Tompkins Acute Ward Study. The role of a typical ward manager of an acute psychiatric inpatient unit based in the United Kingdom is to manage and coordinate the quality of the clinical environment, including the management of nursing care and general patient care as well as management of all staff on the ward. The role of a typical consultant psychiatrist, of whom there are usually at least two associated with a particular ward, is as a specialist in the diagnosis, assessment, and management of psychiatric symptomatology. Contact and dialogue between ward managers and consultant psychiatrists are often minimal, and consequently most negotiations over bed management, admissions, discharges, and the management of the ward take place between the other psychiatrists and the ward manager (9).

The interviews occurred during the same two-year period that the PCC-SR data were collected from each ward and were carried out on a rolling six-month basis. In total, 77 ward manager interviews and 43 consultant psychiatric interviews were conducted. The response rate from ward managers was excellent, with only a few missed interviews (96% response rate). Interviews with consultant psychiatrists were more difficult to obtain (46% response rate). The mean age of the ward managers at the time of the interview was 40 years, compared with a mean age of 44 years for consultant psychiatrists. A majority of the interviewees were male (68%). The mean duration of time in the position was 6.6 years for ward managers and 6.9 years for consultant psychiatrists.

The interviewees were asked about events that had occurred during the previous six months. They were free to volunteer any information about events but were also prompted about changes in staff personnel, community services, catchment areas (boundary and population changes), ward profile (such as new location, bed numbers, and team structure), and ward policies. They were also asked about staff training, any particularly influential people on the ward (patients or staff), the social dynamics of the team, and whether any major incidents had occurred and the impact they may have had. All of the reported events were subsequently verified by cross-checking case notes and other available sources. Disagreements between the consultant psychiatrists' and ward managers' reports of when events occurred were also resolved through this process. Dates accurate to calendar month were obtained for all responses. Using these data, we constructed a retrospective timeline of events for each ward. The institutional review boards of all participating institutions approved all study procedures, and informed consent was obtained from all study participants.

Next, the timeline events were overlaid with the identified line graph transitions. This involved searching through the qualitative timelines for events occurring at the starting point of each identified graph transition (Figure 1). Further, any events that occurred after the score transition had begun were not matched to the graphs. These procedures were carried out in order to ensure causality between events and transitions. The successfully overlaid qualitative accounts were then thematically analyzed with the qualitative data analysis software package NVivo 8 by grouping emerging themes into a hierarchical coding system. A second researcher reviewed a random sample of 10% of the interview transcripts, and any differences in interpretation were agreed upon by consensus. The transcripts were continually reread throughout the analytical process to ensure that emerging themes were firmly rooted in the original data, to check for deviant cases, and to identify quotations that encapsulated the meaning of the identified themes. Using Stata, version 11.0, we ran a series of Pearson chi square tests to contrast each theme's level of association with upturns versus downturns in total conflict and containment scores. The level of significance was set at .05.

This study's findings support our research program's working model of what determines the likelihood of conflict and containment in acute inpatient psychiatric wards. The model centers on the influence of staff factors, the most important of which are the staff's positive regard for patients, their own emotional self-regulation, and their provision of an effective structure of ward rules and routines. These three factors are influenced by their psychiatric philosophy, moral commitments, technical mastery, teamwork skill, and organizational support. These findings support the model in that most of the significant themes found to precede conflict and containment score transitions were staff centered.

It is unclear how these themes link with the model's components, although it is likely that there are multiple relationships. Positive appreciation is likely to be linked with positive staff practice, because being compassionate and treating patients with value constitutes positive practice. A relationship between emotional regulation and staff morale, staff support, staff-staff conflict, positive teamwork, and staff pressure, stress, and workload level probably exists; compared with other teams, those that support one another and work together effectively without conflict or pressure should be better able to maintain their emotional equilibrium with patients who are considered “difficult.” A ward with an effective structure that includes a well-directed routine and rules that are fairly and consistently applied is likely to be run by a ward manager who is proactive and reinforced by an active team that works positively together. Positive or negative staffing changes are likely to affect all elements of the working model. The finding that the City Nurses intervention was significantly associated with transitions also lends support to the model given that the intervention was constructed on the model's principles (27).

This study's findings also indicate that a ward's physical environmental influenced conflict and containment. This is a theme that is not currently included within our working model and warrants further empirical investigation.

This study's methodological approach has afforded quantitative patterns in a large and statistically reliable longitudinal data set to be explained by qualitative data. Furthermore, unlike cross-sectional studies, this study enables one to draw inferences about causality because events were matched against time points that preceded score transitions. However, this study's findings can be attributed to accurately represent only approximately a quarter of the possible data because only 27% of the score transitions were successfully matched with qualitative data. Furthermore, we were unable to ascertain how accurate the interviewees' six-month recall of events was, in terms of both what and when events occurred. This means that there remains some uncertainty about the precision between the temporal ordering of the events and the score transitions.

It is also important to consider that the factors identified in this study were inherently linked with the perceptions of the consultant psychiatrists and ward managers. For example, it is not surprising that a theme of poor leadership did not emerge from the interviews. Interviews with patients would have yielded a richer and more balanced qualitative data set of events; however, this was not possible because inpatients in an acute phase of psychiatric illness do not usually remain within the same ward beyond one month.

Another potential limitation of this study was that the marking of the score transitions was a subjective process; however, an interrater reliability test had a good result. In addition, although the overall mean response rate of the PCC-SR was reasonable, there was considerable variation of response rates across the 16 wards. However, conflict and containment rates have been shown to vary up and down across the 16 wards, regardless of response rates (9). Furthermore, previous research has established that the PCC-SR does not suffer from the underreporting of incidents associated with using official reports (28).

Another potential criticism is that the findings may be explained by reverse causality (for example, higher conflict leading to lower staff morale). However, this study has shown that across the 16 wards, low staff morale was significantly more likely to precede increases in conflict compared with decreases in conflict. This is a statistical finding across the data set and implies that reverse causality is not a possible explanation for the study findings. However, it is possible that certain factors (such as injuries, absenteeism, staff turnover, high contact census, changes in leadership, and patient volume) may initially influence staff morale. Future research clearly is needed to investigate the influence that such factors have on staff, given that staff have been shown to play a crucial role in influencing the likelihood of conflict and containment rates in this study.

It could also be argued that aggregating conflict behaviors and containment responses into two total scores is too broad a method given the range of severity in conflict behaviors and containment responses. However, previous research has provided evidence that using aggregated scores on the PCC-SR does not undermine the tool's reliability or validity (7). Finally, although the 40 themes were counted 467 times across the score transitions, for many of the significant themes the counts were fairly small (range 0–19), albeit large enough in each instance to be significant.