Intervention

The intervention took place between March and May 2016 for Group A and between June and August 2016 for Group B. One experienced facilitator of singing groups was recruited. One weekly, one-hour programme of singing was designed to run over three months, with 12 sessions in total. The singing programme included warm-up exercises and an initial range of musical styles with varying degrees of complexity to determine the group’s preferences. The facilitation style of the sessions encouraged a relaxed and ‘fun’ time away from work. Handouts were distributed to remind the participants of the lessons they had learned and to encourage home practice. In addition, each week a ‘mission’ was given for participants to work on, with feedback given during the next session. The songs were learned and sung either ‘a cappella’ or accompanied by guitar, with the lyrics written on a flipchart. Venues were booked within the trust for weekly late-afternoon sessions. Attendance records were kept to enable later analysis of changes in participants’ well-being in relation to the number of sessions attended.

Data collection and analysis

Quantitative data were collected on recruitment, retention, records of participants’ attendance at the singing groups, response rates to questionnaires and the following measures:

These measures were combined into one questionnaire and given to all participants electronically at baseline, three months and six months. Qualitative data were also collected, through free comments on the questionnaire and on anonymous ‘post-it’ notes after the final sessions.

As a feasibility study, the data analysis was largely descriptive in terms of identifying adequacy of recruitment, retention and acceptability. Thabane et al (2010) suggested setting targets for decision-making in relation to whether to proceed to a full study, and these were agreed and set as:

Since previous studies found that benefits from singing were related to frequency of engagement in the group, only those attending for at least half of the singing sessions were included in the analysis. All data were transferred to an Excel spreadsheet, and the mean group scores were calculated for the two groups for each domain of the WHOQOL-BREF (Skevington et al 2004), the Measurement of Engagement and Burnout (Schaufeli et al 2006) and the Organizational Commitment Questionnaire (Mowday et al 1979).

Ethics

Confirmation of compliance with the requirements for proportionate ethical review was received from the researchers’ university and the trust’s research and development departments. Participants were provided with an information sheet, which they signed to provide informed consent before participation. Anonymity of questionnaire data was assured. All data were stored either electronically on a password-protected computer or on paper in a locked filing cabinet.

Recruitment and retention

Initial interest in the project was high and exceeded the target number of participants at pre-intervention. However, after excluding participants who attended for less than half of the sessions, the numbers declined considerably, with a further decline in the number of questionnaires returned. Table 1 shows the participant recruitment, retention and questionnaire completion.

Table 1.

Participant recruitment, retention and questionnaire completion

At baseline, there were few differences in the demographics between the participants who completed questionnaires in Groups A and B. Across the two groups, 20 participants completed the questionnaires at baseline (Table 1), which had decreased to 10 participants at six months. Most participants were over 40 years old (n=19/20), which broadly reflects the age profile of the NHS workforce as a whole (NHS Employers 2013), and most had been employed in the trust for more than five years (n=16/20). The largest employment group was nursing and midwifery and support (n=10/20), and more than half of participants worked full-time (n=12/20). Only three out of 20 participants were already singing in a choir.

Feedback from participants on the intervention and research procedures

At baseline, the idea of a staff singing group was well-accepted, and its time and venue suited most participants well. However, one participant who was unable to attend regularly suggested having a shorter singing group session during the day as an alternative:

‘I love the idea of the singing group – it made me feel pleased with the organisation for supporting this type of project. I just wish we could have a 30-minute one during the day’ (Group A, baseline questionnaire).

There was general support for the recruitment methods used, although it was noted that some staff did not access their emails regularly. One participant commented that items in the trust’s newsletter were often too embedded and easily missed. Additional suggestions included use of the information ‘hubs’ (dedicated information exchange rooms) to promote the singing groups and recruit participants.

Participants reported that the programme was well-organised and described the facilitator as ‘excellent’, a ‘good motivator’ and ‘professional and inspiring’ (Group A and B post-intervention questionnaires). The song selection was judged to be good and some individuals felt inspired to either join a choir or purchase CDs of songs that the facilitator had introduced them to. A few participants found some of the material to be challenging. Several participants suggested that the song lyrics could be written on the flipchart in larger print or sent via email, while a minority preferred being provided with the musical notation.

The procedures and instructions were regarded as clear, and the questions easy to complete. One participant commented on the prompt response to emails sent to the researchers, although another participant reported that they found responding to some of the questionnaire questions challenging, since she had only been employed in the trust for a short time. No comments were made about the allocation of participants to the initial intervention and control groups.

Effects of the staff singing for well-being group

At the final sessions, all of the participants made positive comments in relation to having taken part in the staff singing group. Remarks related to emotional and psychological well-being, social benefits, work attitude and learning. In terms of emotional well-being, participants reported ‘fun’, ‘enjoyment’, ‘lifting spirits’ and ‘feeling more positive and cheerful’ (Group A and B post-it notes). One participant commented:

‘I have really enjoyed participating in this singing programme and found it has lifted my spirits. I talk about it to my family on a regular basis… Thank you for offering this opportunity to be part of the singing programme. I have had great fun!’ (Group A, post-intervention questionnaire).

Another participant commented that ‘without a doubt these sessions have been the best part of my week’, while other comments related to well-being included ‘feeling alive’, ‘feeling de-stressed after a singing session’ and ‘gaining confidence in singing’ (Groups A and B post-it notes).

There were several comments relating to the social benefits of the group:

‘I have really enjoyed the singing project and the benefits it has brought me in relation to friendships’ (Group A, follow-up questionnaire).

‘I enjoyed taking part in the project and hope that the choir carries on. It was nice to meet staff from other areas’ (Group A, follow-up questionnaire).

These comments indicate that, three months on from the end of the project, there was an enduring feeling of connectivity with staff in other areas of the trust. Similar sentiments were expressed by the participants in terms of ‘a community of people who love to sing’ and a ‘sense of special community’ (Group A post-it notes).

It was also indicated that work well-being increased after the singing programme, both immediately after a session and cumulatively over time. Participants mentioned feeling increasingly enthusiastic, positive and inspired to meet the challenges they experienced at work, and one participant felt that this could lead to improved relationships with colleagues. Feeling ‘de-stressed’ and relaxed could also contribute to this positive work ethic. The mechanism by which participating in a singing group could affect the workplace was rationalised by one participant:

‘I’m sure singing does affect well-being. I always felt cheerful after a session… maybe attending the sessions is confidence-boosting and therefore inspires/motivates us to meet challenges that we have to cope with while at work’ (Group A, post-intervention questionnaire).

One participant’s comment on the post-intervention questionnaire illustrates the general response to taking part in the singing group:

‘It definitely impacts on well-being. On several occasions I had to really push myself to attend because life was being difficult at the time and curling away seemed preferable. I always left feeling much better and more able to put the rest into perspective. I think it has an immediate effect on how you feel but also a longer-term impact on confidence and equilibrium. The experience of having the group begin to form into a community is something else I will also remember and apply elsewhere’ (Group B, post-intervention questionnaire).

Outcomes for well-being measures

World Health Organization Quality-of-Life Scale

Table 2 shows the mean scores for each group for the WHOQOL-BREF (Skevington et al 2004) domains. It should be noted that the scores at baseline are immediately pre-intervention for Group A and three months before the intervention for Group B. At three months, the scores are immediately post-intervention for Group A and immediately pre-intervention for Group B. At six months, the scores are three months after the intervention for Group A, and immediately post-intervention for Group B.

Table 2.

Mean group scores for the World Health Organization Quality-of-Life Scale (Skevington et al 2004) domains

At baseline, the mean scores for the physical health domains were similar for the two groups. At three months, the intervention group (A) improved on this measure, while the control group (B) demonstrated a slight deterioration. At six months, the intervention group (B) improved while the control group (A) plateaued (Table 2).

In the psychological domain, the intervention group (A) at baseline demonstrated higher well-being scores than the control group (B). Both demonstrated improvement at three months, with greater improvement for the intervention group (A). At six months, the control group (A) scores reduced, while the intervention group (B) scores remained stable (Table 2).

In the social relationships domain, the pattern was similar to the psychological domain initially, with both groups improving their scores at three months, although the intervention group (A) demonstrated greater improvement in their scores. However, at six months, the scores for the control group (A) continued to improve, while those for the intervention group (B) decreased (Table 2).

In the environment domain, the intervention group (A) demonstrated slightly greater improvement than the control group (B) at three months. At six months, the pattern was similar to the physical health domain, but with intervention group (B) scores higher than those of the control group (A) (Table 2).

Work and well-being

Data from the Measurement of Engagement and Burnout (Schaufeli et al 2006) demonstrated similar slight improvements for both the intervention A and control B groups at three months, with the difference between the two groups decreasing over this time (Table 3). At six months, both groups had continued to show an improvement in work and well-being scores, with the intervention group (B) showing a greater improvement than the control group (A).

Table 3.

Mean group scores for the Measurement of Engagement and Burnout (Schaufeli et al 2006)

Organisational commitment

Data from the Organizational Commitment Questionnaire (Mowday et al 1979) suggested a slight difference to the other patterns (Table 4). The intervention group (A) and control group (B) both registered higher organisational commitment at three months than at baseline, with the control group (B) showing a higher score than the intervention group (A) at both of these points. At six months, the data demonstrated an improvement in the intervention group (B) scores, but some decline in the control group (A) scores.

Table 4.

Mean group scores for the Organizational Commitment Questionnaire (Mowday et al 1979)

Limitations

As a piece of conventional research, this study was limited in its ability to measure the effects of group singing on staff well-being. However, this was not its purpose as a feasibility study. Even so, the sample size was small and participant retention was challenging, which limits any firm conclusions regarding the recommendation to proceed to a full research trial, if the same research design and procedures were followed. In addition, further recruitment to Group B took place ahead of that group’s intervention, which meant that this group’s participants changed between baseline and the pre-intervention measure, potentially affecting these findings. Furthermore, a crossover design is arguably not suitable for this type of intervention, where the effects of singing may be long-lasting. Another limitation was that the research design did not enable follow-up of specific individuals, since the data were fully anonymised.

Recommendations for future practice and research

While there is insufficient evidence to strongly recommend singing groups for staff in healthcare organisations, this study suggests that they are likely to be beneficial for those who wish to attend. Where singing groups are to be set up, the experience from this project suggests that attention should be given to recruitment by investing in face-to-face publicity with staff across departments. Furthermore, this study shows that the implementation of staff singing for well-being groups should be preceded by a survey to identify the best time and venue for staff, followed by an evaluation of the benefits of the singing group for participants.

More broadly, greater resourcing of occupational health departments could ensure that staff well-being gains greater prominence in NHS trusts (National Nursing Research Unit 2013, Sprinks 2013). Singing for health could be signposted as one option within a staff well-being strategy. Where lack of time to attend a singing group is an issue, consideration should be given to integrating singing and music into care practices, which can be mutually beneficial to staff and patients (Vella-Burrows 2009, Sonke et al 2017).

Regarding further research, a larger trial would identify with greater certainty whether singing could contribute to staff well-being. Guaranteeing participants’ confidentiality rather than full anonymity would enable any differential outcomes for those already engaged in singing groups to be identified, as well as the tracking of individuals through the data collection phases, follow-up of less frequent attendees and greater ability to compare variables across the two groups. In addition, looking to a future where there may be wider acceptance of social prescribing – which enables GPs, nurses and other primary care professionals to refer people to a range of local, non-clinical services (The King’s Fund 2017) – it may be possible to avoid individuals self-selecting into singing groups, which tends to result in those predisposed to singing taking part.