Hawthorne effect

Rich and defining stories describing the professionals’ roles showed remarkable consistency with many supporting parents and patients through distress including bereavement, or making other professionals act to prevent a disaster. More experienced nurses tended towards the former structure while less experienced preferred the latter. Many described patients’ and parents’ needs at the centre of best healthcare. One interview was irretrievably interrupted before feedback, 5 respondents found the process sad or painful but none sought counselling.

All except 1 and the interrupted interview had at least one positive response. All complex feedback (n=27 respondents) involved reflection on what respondents had done or felt, or the events or feelings for others. (table 1 and Fig 1)

Vacancy rates (%) fell significantly on B and M wards in the period after the project but not on P. M; before (median and standard error) 22. 7, 1. 9; during 22. 6,1. 7; after 16. 2,1. 8: B; before 14. 1,4. 5; during 16. 0, 3. 9; after -1. 4,4.2: P; before 10. 3, 3. 7; during -1. 6,3. 2; after 6. 7,3. 4: There was a trend to an increase in leavers from B and P but not M during the project; M, 5 (0. 76 leavers/month/100 WTE) during, with 11 (0. 78 leavers/month/100 WTE) outside the project versus 11 (1. 47 leavers/month/100 WTE) and 14 (0. 87 leavers/month/100 WTE) from B and P (p=0. 11) respectively. SAR showed a significant reduction on M during the project but rose again to previous levels on completion. (Fig 2 here approx).

Fourteen individuals of 23 who were interviewed and could be located still working in the Trust replied to the two questions to provide follow up over 2 years later. Responding to the question ‘What do you remember about the Appreciative Inquiry interview we undertook 2 years ago? ‘ all 14 remembered the interview with varied degrees of detail, 2 recalled it as emotional, 2 felt good about it, 2 felt it useful/helpful, 1 recalled it as interesting.

Detailed responses to the question “What difference do you think the interview has made to you since then?” are shown in Figure 3. (Fig 3 here). At that time SAR for M was 4. 9%, for B was 3. 5% and for P was 0. 0%. Discussion The qualitative data describe the collective values of healthcare at its best held by the ward staff of a UK National Children’s Liver service. Widely endorsed themes – empathy including trusting, caring, efficiency and use of time, expertise, enjoyment and teamwork reflect aspects of the 6 facets of quality of healthcare: patient-centredness, effectiveness, efficiency, safety, equitablility, and timeliness (7).

Respondents considered equitability and timeliness in immediate personal needs, and efficiency and safety in avoiding errors and inappropriate care and treatments, while almost all stories in response to questions1, 2 & 3 had features of patient-centredness. It is reassuring that when exploring values to motivate best care, professionals have a good idea what quality is in practice. Replies were inevitably constrained by the interviewer who may have been seen as an insider, connected to the management, personally powerful and with his own agenda.

It was unlikely that a story of success could have included conflict with managers or senior doctors, for example. While 23 respondents described positive feedback or appreciation, only 9 described feeling respected or valued by the institution, despite its progressive HR policies and being an “Investor in People” (IIP). Perhaps they were not fed back information on their value or were detached from the opinions of their managers who changed frequently, so that such feedback was not heard, or do not see their own value as part of good healthcare.

Ai may be a better means for such feedback than providing it as a solution to a problem. Despite being emotional, the interviews were often enjoyed with no adverse consequences. Refusal rate was very low (6%). Staff quickly learned the interviews were occasionally associated with tearfulness but worthwhile, and some were impatient for their turn. Although in feedback only 3 expressed pleasure that someone was interested it seemed that this was an important motivating factor. It is reassuring that even when uncomfortable levels of emotion developed no harm derived from the Ai process.

When recalled over 2 years later the process had affected many profoundly. It had precipitated or contributed to life-changing decisions. Improved communication with enhanced feelings of belonging to a team and improved relationships with colleagues and patients mentioned by 6 were all positive contributions to clinical microsystems. Greater confidence and a feeling of being valued were mentioned by 5. It served as an introduction to, or reminder of, reflection as a tool particularly for dealing with stress and strong emotions through developing insight.

One-to-one Ai interviews can therefore have powerful long-term positive effects permeating the entire environment where they were undertaken. There were no significant effects on staffing numbers as the fall in vacancy rate was also seen on B. Effects from the project depend on willingness of nurses to join M based on factors such as its reputation and morale. Although 4 other studies could not show SAR responding to interviews (8), we have shown that very high SAR can be managed by Ai interviews.

The average SAR in 299 NHS trusts was 4.5%; ancillary staff, and nursing including midwifery staff had the highest rates (6. 0% & 5. 4%), while medical and administrative staff had the lowest (1. 3 % & 2. 3 %) (9). Thus SAR on M was very high and greater than B or P, possibly due to the intensity of work, M being designated high dependency, having the highest vacancy rate and frequent changes of nursing leadership. In favour of improved morale on M being associated with lower SAR during the project, we received unsolicited feedback of a more positive atmosphere from 3 independent sources, and a ward newsletter was started.

Assuming Ai caused the mean reduction in SAR of 1. 3%, 15. 6 hours per week were saved at the expense of less than one hour each of staff and interviewer time. Thus even without the additional cognitive benefits the process was cost effective. It is possible that the transient improvement in SAR was due to a Hawthorne effect of observation, although no particular scrutiny or discussion of sickness or absence was undertaken at any point and the outcome measures were not revealed until the final analysis.

“Instead of referring to the ambiguous and disputable Hawthorne effect ….. researchers should introduce specific psychological and social variables that may have affected the outcome under study but were not monitored, along with the possible effect on the observed results” (10). The wide swings in SAR seen during the entire period were possibly echoed on B where vacancies also fell, but not on P suggesting common perhaps cyclical factors fortuitously amplified by Ai to lower SAR.

There were 3 different nurse ward managers with periods of vacancy between resignations and appointments between months 1 and 50 with one leaving at month 28 during the project. In their absence, the interviews may have acted as a form of managerial supervision, motivating staff to change sickness behaviour through a sense of individual scrutiny, even though the intervention was designed to avoid scrutiny or judgment. AB left M at month 42, possibly reducing any effect, which might have been maintained by a rolling programme of interviews including with a different interviewer or stories.

There is a current debate among Ai practitioners whether Ai is a tool or a philosophy. We showed change on M while Ai was used as a tool but did not show any sustained change in the underlying behaviour once Ai conversations dissipated, although individuals still retained effects within themselves. This is in keeping with a social constructivist explanation that social structure and behaviour are created by the language being used, suggesting that the closer Ai can become to a philosophy, the more pervasively it will affect language and behaviour.

In a meta-analysis of 45 studies of management interventions related to stress at work (8) with outcome variables based on ’cause for concern’ including SAR and retention rates, 17 showed positive effect, 11 of them major. Employees improve quality of life at work, psychological resources and complaints most from short interventions of cognitive design when they have high degree of decision latitude to use new skills. At least 12 weeks are necessary to detect benefits (11). This project fulfilled the above criteria well except that no direct assessment of change in work satisfaction was made.

From the form of words in the feedback there was evidence of second loop learning (e. g. “makes you think.. ” “didn’t realize… “), shifting the meaning of work towards congruence with respondents’ values and skills giving potential for self-actualisation in work (12,13). These effects were still evident two years later. Despite discovering relative conformity in ward values, the Ai process may promote autonomy and personal values giving the confidence for individuals to stand out against the group when appropriate.

For example religious values are not part of the culture of M but motivate a significant minority of staff. Ai may be a means to ensure personal values are not over-run by group culture, avoiding situations such as the Bristol scandal, when senior figures in an hierarchical system performing paediatric cardiac surgery with very poor results refused to hear the appropriate concerns of team members and punished them for dissenting. Appraisal can contribute to lower hospital mortality (14), but is mostly used in the NHS without 360degree feedback, as a combination of appraisal, assessment, and performance management (15,16).

Even if the relationship between the parties is good the process may be threatening. A problem-solving approach may identify the appraisee as the source of problems risking negative feedback, with deterioration of the relationship and worsening performance. Using appropriately designed Ai, a two way synthesis may be achieved with appraisees identifying personal goals aligned with strategic intentions, linking strategy with the micro-systemic level.

Meanwhile the simultaneous emphasis of their own achievements may empower them to lead the relationship in favour of their own values and capabilities, engaging themselves with their work. Ai should therefore form the primary structure of appraisal in staff development leaving deficit based solutions for situations with certain cause and effect (2). Ai may also have the potential to facilitate the development of positive social interactions as bridges between groups with differences in goals and mental models.

Poor interactions between such groups are currently spoken of as the ‘silo culture’ in the NHS. Many respondents identified with parents’ experiences, followed by learning about their professional identity, actions and meaning in a ‘community of practice’ at work to which parents and children also belonged (17). They recognised that families became members of a group with the same aims, practices and ideas as themselves. This is encouraging in terms of the primary intention to find an effective link between strategy and operation.

It implies the link may be found in relationships between people mutually seen as ‘different to us’. For example, Ai is an ideal strategy for managers outside clinical microsystems to exert a positive influence on professionals within them, and staff from different silos could be invited to recount stories about, and reflect on when their interactions together were successful, co-creating new, shared possibilities made into actions by continued Ai conversations.

A simple but very robust format that we have used for initiating Ai stakeholder group work including patients is illustrated in appendix B (18). Perhaps the successful future of the NHS lies in recognising what we, including patients, do well together, talking about it and doing more of it.