Relationship between employee engagement scores and service quality ratings: analysis of the National Health Service staff survey across 97 acute NHS Trusts in England and concurrent Care Quality Commission outcomes (2012-2016).

OBJECTIVE: This research explores measures of employee engagement in the National Health Service (NHS) acute Trusts in England and examines the association between organisation-level engagement scores and quality ratings by the Care Quality Commission (CQC).
DESIGN: Cross-sectional.
SETTING: 97 acute NHS Trusts in England. PARTICIPANTS: 97 NHS acute Trusts in England (2012-2016). Data include provider details, staff survey results and CQC reports. Hybrid Trusts or organisations affected by recent mergers are excluded. OUTCOME MEASURES: Analysis uses organisation-level employee engagement and CQC quality ratings.
RESULTS: Employee engagement is affected by organisational factors, including patient bed numbers (β=-0.46, p<0.05) and financial revenue (β=0.38, p<0.05). CQC ratings are predicted by overall employee engagement score (β=0.57, p<0.001) and financial deficit (β=-0.19, p<0.05). The most influential employee engagement dimension on provider ratings is 'advocacy' (λ=0.54, p<0.001). Analysis supports the notion that employee engagement can be predicted from advocacy scores alone (eigenvalue=4.03). Better still, combining advocacy scores from the previous year's survey or adding in motivation scores is a highly reliable indication of overall employee engagement (95.4% of total variance).
CONCLUSIONS: NHS acute Trusts with high employee engagement scores tend to have better CQC ratings. Trusts with a high financial deficit tend to have lower ratings. Employee engagement subdimensions have different associations with CQC ratings, the most influential dimension being advocacy score. A two subdimension model of engagement efficiently predicts overall employee engagement in NHS acute Trusts in England. Healthcare leaders should pay close attention to the proportion of employees who would recommend their organisation as a place to work or receive treatment, because this is a proxy for the level of engagement, and it predicts CQC ratings. © Author(s) (or their employer(s)) 2019. Re-use permitted under CC BY-NC. No commercial re-use. See rights and permissions. Published by BMJ.

Engagement data are taken from a large national survey of National Health Service (NHS) employees.

The survey results coincide with the first national inspection programme of all acute NHS Trusts in England, by the Care Quality Commission (CQC).

A conceptual model is used to analyse the association between the subdimensions of employee engagement and perceived quality (as measured by the CQC).

The predictor variables are taken from a single self-reported source, which risks common method variance.

The sample and cross-sectional design limit conclusions about causation or generalisability.

Introduction

This study considers organisation-level measures of work engagement taken from the annual National Health Service (NHS) staff surveys (NSS) 2012–2016.1 It examines the effect that organisational size, status and financial revenue have on overall engagement and compares engagement scores with provider ratings for NHS acute Trusts in England. Employee engagement research typically uses a multidimensional construct of engagement, so the study applies this approach to the NHS. It investigates the associations between NSS engagement subdimensions and the perceived quality of provider organisations (as reported by the Care Quality Commission (CQC)).

Organisational factors

Saks2 suggests that employees repay their organisation for the resources they receive through their levels of engagement and to some extent engagement reflects the relationship employees have with their organisations. Maslach et al 3 emphasised the important role that organisations play in providing these valued resources by allowing employees some autonomy or by providing feedback and learning opportunities. The engagement-promoting capacity of job resources was also identified by Bakker et al,4 who described job resources as the physical, social or organisational factors, which reduce job demands and play a motivational role at work. Job resources reduced the perceived demands of a job and appeared to protect employees from burnout, particularly during sustained periods of high work intensity. From a healthcare perspective, Hakanen et al 5 in a longitudinal study of over 2500 Finnish dentists reported a positive, stepwise relationship between job resources, engagement and personal initiative or innovativeness. In another Finnish study of 409 healthcare workers, Mauno et al 6 concluded that the best predictors of engagement were self-esteem and the ability of employees to control some aspects of their work (job resources were more influential than job demands). Relatedly, job resources have been linked to engagement among nurses and doctors working in hospital environments.7 8

In the UK, recruitment and retention of staff has historically been more successful in large, prestigious teaching hospitals with considerable resources at their disposal. Several studies have linked the structure of healthcare organisations to measures of performance including efficiency, patient outcomes, staff and patient satisfaction.9 10 There has been a reporting bias towards ‘bigger is better’ but the evidence for such a general assertion is weak.11 West et al 12 found that the type of NHS organisation influenced employee engagement but the key organisational characteristics which predict employee engagement in an acute healthcare environment are uncertain. Parsimoniously, this research examines the results of recent NHS staff surveys for evidence that employee engagement is linked to organisational characteristics which may be a proxy for available resources, using the following hypothesis H1(a) and H1(b):

H1(a) NHS employee engagement will be related to Trust size (bed numbers or revenue).

H1(b) NHS employee engagement will be related to Trust type or status (Foundation/non-Foundation, teaching/non-teaching hospital).

Higher levels of employee engagement will be associated with large teaching hospitals, Foundation Trusts or organisations in strong financial positions.

Employee engagement and performance (CQC ratings)

The proposition that employee engagement has a positive effect on organisational performance is not new.13 Employee engagement has been associated with improved performance in many industries, although there is limited healthcare evidence. Engaged employees tend to be intrinsically motivated, are more likely to achieve their goals and learn from mistakes and engagement has been associated with organisation-level quality outcomes.14–16 Engagement among healthcare professionals is considered high compared with other industries and hospitals with more engaged nurses tended to deliver better patient care and have superior safety records compared with those with less engaged employees.14 Although there are few studies from the NHS, the historical link between NHS engagement and quality was reported in the study by West et al.12 Staff engagement had a significant effect on patient satisfaction, hospital mortality rates, infection rates, absenteeism, staff turnover and annual health check ratings (a forerunner of QCC ratings).

One controversial measure of NHS provider quality is the use of CQC healthcare ratings by the Department of Health and Social Care. In 2016, the CQC completed the first national inspection programme of NHS acute Trusts in England and rated each organisation as: outstanding, good, requires improvement or inadequate. The CQC highlighted the variation in quality and attributed this to factors such as culture, leadership and staff engagement.17 It follows that employee engagement is worth investigating as a predictor of CQC ratings. Based on the literature, the direct relationship between engagement and perceived quality of NHS acute Trusts is expressed as our second hypothesis (H2):

H2: there is a positive relationship between the levels of staff engagement and Trust performance so overall employee engagement in acute NHS Trusts will predict their CQC ratings.

Engagement subdimensions

Although Kahn18 was the first to define work engagement as a multidimensional construct related to meaningfulness, safety and availability, subsequent developmental theories conceptualised engagement as the positive antithesis to burnout. Contemporary research has been strongly influenced by the definition of engagement by Schaufeli and Bakker15as a ‘positive, fulfilling, work-related state of mind characterised by vigour, dedication and absorption’.Vigour was associated with energy, resilience, persistence and greater effort. Dedication was characterised by involvement and associated with a personal sense of significance, pride, inspiration and challenge. The third subdimension (absorption) was linked to being happily engrossed at work so that time passed quickly.

The NSS questionnaire was based on the Utrecht Work Engagement Scale (UWES), which operationalises the definition of engagement by Schaufeli and Bakker.15 The final survey questions were influenced by the NSS Improvement Board to reflect employee engagement in the context of the organisation and its environment. These modifications were tested by cognitive interviewing for validity.19 Overall-engagement scores were synthesised from three subdimension scales: motivation, advocacy and involvement. NSS motivation is similar to psychological engagement and includes elements of intrinsic motivation, dedication and absorption at work. Advocacy is strongly linked to care standards and reflects the perceptions that staff have of the organisation’s patient-centredness and the level of pride they feel at work. It also reflects the willingness to recommend the organisation as an employer or healthcare provider. Involvement is a ‘practitioner’ measure, which covers employee involvement in decision-making, change management and relationships with supervisors.20

Recent studies have suggested that employee engagement is better represented by a two-subdimension model. For example, Salanova et al 21 reported that only two of the15 dimensions by Schaufeli and Bakker predicted employee engagement, namely vigour and dedication. The ‘absorption’ subdimension was considered a consequence of employee engagement, not an antecedent. The debate about the antecedents of employee engagement prompted the research question:

In NHS survey instruments, overall engagement is calculated from three subdimension scales. Which subdimensions are the ‘core dimensions’ for NHS employee engagement? (figure 1)

Figure 1

Conceptual model. In annual employee surveys, the National Health Service engagement scores are synthesised from three subdimension scores. The dimensions of engagement may have differential associations with Care Quality Commission ratings (which are a controversial indicator of the perceived quality of NHS providers in England).

Methods

For consistency, this research focuses on NHS acute Trusts, while organisations in unusual circumstances (recent mergers, acquisitions or significant reconfigurations) and hybrid organisations (mixed community and acute services) are excluded. The resulting sample is 97 NHS acute Trusts in England. This study does not include NHS services in the rest of the UK, as they fall under different regulatory arrangements. Although this study uses Trust-level data, representativeness and comparability are assumed because a weighting procedure is applied to NSS returns based on a hypothetical national staff profile for each type of organisation. To allow for historical comparisons, data weighting is regularly reviewed.22

Organisation characteristics

All data are publicly available in the UK on NHS acute Trust websites, including NHS Trust Board papers and quality accounts. Organisational characteristics are selected as follows (to identify or reduce confounding effects). Acute Trusts in England have a wide range of operating incomes that directly affects available resources. Trusts with significant financial deficits can have constraints on resources so the size of Trust deficit as a percentage of financial turnover is used. Although bed numbers are an indicator of organisational size, it may also reflect an element of work intensity, and so could affect performance. Teaching hospitals affiliated to reputable academic organisations are associated with higher performance, so teaching status is included.9 10 ‘Foundation Trust’ status is awarded to higher-performing NHS organisations and is intended to give them more autonomy and greater financial flexibility; and therefore is likely to impact on culture, climate and resources.

Engagement and performance (CQC ratings)

NHS acute Trust data are extracted from NSS reports. For each Trust, survey data corresponding to the year of their CQC inspection and the previous year is used. The mean average annual response rate for acute Trusts for 2012–2016 is: 49% (2012–2013), 42% (2013–2014), 41% (2014–2015) and 43% (2015–2016), respectively, comprising between 269 000 and 456 000 respondents per survey year.

Trust-level engagement scores are summarised for each organisation using the weighting procedure described above. Organisation scores are then compared with the national average for organisations of a similar type. Benchmark data are obtained from the summary reports provided to individual NHS Trusts.22

The CQC inspected all 136 acute Trusts and 17 specialist Trusts in England between September 2012 and June 2016 and published the results on their website. This included a total of 265 non-specialist hospital sites or locations and 27 specialist hospitals operated by these Trusts. Assessment of core health services included: children and young people, intensive/critical care, maternity and gynaecology, end-of-life care, outpatients and diagnostic imaging, surgery, urgent and emergency services and medical care including older people. In making their assessments, the CQC uses a set of 150 indicators obtained from various sources (including inspection visits). They rate organisations under five domains (safe, effective, caring, responsive and well-led). Each organisation receives an overall rating as: outstanding, good, needs improvement or inadequate17 (see online supplementary table 1).

NSS dimensions of employee engagement

As discussed, a three subdimension model of employee engagement is captured in the NSS. Each dimension is scored across a number of items, using a 5-point scale or yes/no answers. ‘Overall engagement’ scores for each respondent is created by taking the mean average from the three subdimension scores (motivation, advocacy and involvement). The subdimension scores have strong intercorrelations (Pearson’s, p<0.001) and convergent validity. Factor loadings are all >0.7, Bartlett’s test p<0.001, Kaiser-Myer-Olkin=0.71. Overall engagement score for each organisation is calculated using a weighted mean average (to account for occupational differences between Trusts). Overall engagement Cronbach’s α=0.70 and the standardised regression weight delta <0.2 for subdimension scales (using a single common factor approach) (table 1).

Table 1

The National Health Service staff survey calculates overall engagement from three scales: motivation, advocacy and involvement

Dimension	Description
Motivation	Staff motivation at work (Cronbach’s α=0.81)
Advocacy	Recommend the organisation as a place to work or receive treatment (Cronbach’s α=0.74).
Involvement	Ability to contribute towards improvement at work (Cronbach’s α=0.86).

Results

Employee engagement is treated as an organisation-level variable. Scatter plots of standardised residuals show a roughly rectangular distribution with central clustering, so the assumption of linearity is met. There is no evidence of a systematic pattern of residuals and there are no residuals outside the accepted range for Trust-level data. Only two organisations have Mahalanobis distances greater than the critical value (eg, χ2=18.47 for 4 df), which is around the 2% recommended tolerance. Similarly, Cook’s distances are <1, so the outliers do not have an undue influence on the predictability of the model. As a result, no acute Trusts with complete data sets are excluded.

Organisational factors and employee engagement

Employee engagement is compared with Trust size (financial turnover, bed numbers), type (Foundation/non-Foundation) and teaching status (teaching/non-teaching hospital). Univariate analysis of variance (ANOVA) shows that the model is statistically significant compared with chance (p<0.05). Regression analysis is used to quantify the % variance in employee engagement explained by the predictor variables (R2=0.104, adjusted R2=0.064, SE=0.09). This suggests that the combined predictors explain 6%–10% of the variance. The significant contributions are bed numbers (β=−0.46, p<0.05) and financial turnover (β=0.38, p<0.05).

The data are analysed using hierarchical multiple regression: block 1 comprises control variables (financial deficit as % turnover, bed numbers and Trust status) and block 2 Trust engagement scores (see online supplementary table 2).

In this way, the model assesses the contributions of predictor variables to the variance in the dependent variable. The model is a statistically significant predictor of CQC ratings (ANOVA; F=11.42, p<0.001). The combined effect of the model’s variables is 39% of CQC ratings variance. The control variables account for approximately 10% of CQC ratings. The change in R2 (ΔR2) in block 2 shows that engagement scores account for an additional 29% of variance (p<0.001, SE=0.53). The statistically significant predictors are financial deficit (β=−0.19, p<0.05) and engagement score (β=0.57, p<0.001). The regression coefficients are shown in table 2.

Table 2

Hierarchical multiple regression—the conceptual model predicts CQC ratings

	Unstandardised coefficients		Standardised coefficients	Change statistics
	R	R2	Adjusted R2	SE	ΔR2	ΔF	P value
Controls†‡	0.312	0.097	0.057	0.638	0.097	2.423	0.054
Full model†§	0.625	0.391	0.357	0.527	0.294	42.891	<0.001*

Engagement scores and Trust financial deficits are the significant predictors of CQC ratings, (*p<0.001)

†Dependent variable: CQC rating.

‡Predictors: (constant), deficit %, FT or non-FT, teaching status, bed numbers.

§Predictors: (constant), deficit %, FT or non-FT, teaching status, bed numbers, engagement score.

CQC, Care Quality Commission; FT, Foundation Trust

Employee engagement dimensions and perceived performance

Discriminate analysis is used to assess the ability of the employee engagement dimension scores to predict CQC ratings (see online supplementary table 3). The assumption of multivariate normality is met with Box’s M, p>0.05. Univariate ANOVA suggests a statistically significant difference between the three engagement dimensions. Canonical discriminate functions show a statistically significant relationship between the discriminating function (1) and the engagement subdimension scores. The (eigenvalues) canonical correlation=0.67, demonstrating good group separation by a discriminate function. That function explains 95.7% of the variance between the engagement dimensions (λ=0.54, p<0.001). Analysis shows that the factor driving discriminate function 1 is advocacy score, with the largest absolute correlation=0.96 (table 3).

Table 3

Discriminate analysis—the intercorrelations and correlations between engagement subdimensions and Care Quality Commission ratings can be represented by a non-correlated discriminate function (function 1)

Function	Eigenvalue	% of variance	Cumulative %	Canonical correlation
1	0.801	95.7	95.7	0.667
2	0.036	4.3	100.0	0.185
3	0.000	0.0	100.0	0.020

Core dimensions of employee engagement in the NHS

Principal component analysis is used to test for a latent effect using data from the year of and the year before (Yb4) CQC inspections. Standard assumptions are met (sample >10 subjects per variable, strong intercorrelations r>0.3, Bartlett’s test p<0.001, Kaiser-Myer-Olkin=0.76). The component matrix supports retaining a 1-factor solution since only one component had eigenvalue >1. The retained factor is advocacy score (eigenvalue=4.03), which explains approximately 67% of the total variance in the engagement data. By adding advocacy scores from the year before CQC inspections this increases to 80%. Combined advocacy and motivation scores from both years explain 95% of the total variance (table 4).

Table 4

Principal component analysis—advocacy scores from the year of and year before CQC inspections effectively predict employee engagement

Component	Initial eigenvalues			Extraction sums of squared loadings
	Total	% of variance	Cumulative %	Total	% of variance	Cumulative %
Advocacy	4.033	67.223	67.223	4.033	67.223	67.223
Advocacy Yb4*	0.748	12.462	79.684	–	–	–
Motivation	0.545	9.088	88.772	–	–	–
Motivation Yb4*	0.398	6.629	95.401	–	–	–
Involvement	0.182	3.042	98.442	–	–	–
Involvement Yb4*	0.093	1.558	100.000	–	–	–

Combined advocacy and motivation scores are a reliable indicator of overall engagement which can be efficiently represented by a two-dimension model.

*Yb4, year before CQC inspection.

CQC, Care Quality Commission.

Discussion

Engagement is a popular but imprecise term with various definitions, models and measurement tools used in academic research. Conceptual models tend to consider engagement in terms of job demands, job resources and personal resources.4 5 23 In this context, the demands of a job or available resources extend beyond management styles, work intensity, materials or equipment to include employee autonomy, social support, optimism, coaching, feedback, personal development, self-efficacy and self-esteem. Adequate resources are an important motivational force at work because they reduce the perceived demands of a job, particularly when work intensity is consistently high.5 Mauno et al 6 reported that the loss of these resources can produce a downward spiral, particularly when employees sense a loss of autonomy or the inability to control aspects of their work.

Simpson’s review emphasised how organisational factors affected the engagement of nurses (in addition to job attributes and leader behaviours).7 Our findings emphasise that nuancing is required over the assumption that organisational structure has a strong effect on engagement. Our analysis shows that Trust size, type and status explain 6%–10% of the variance in engagement scores which partially supports our first hypothesis, H1(a). Although the size of NHS acute Trusts is related to engagement scores, the two indicators of organisational size have opposite associations. Trusts with higher incomes (turnover) tended to have more engaged employees but organisations with more beds are associated with lower engagement. Although prestigious UK teaching hospitals or Foundation Trusts (with considerable resources at their disposal) have historically found recruitment and retention of employees less challenging than small providers, in our study organisation status is not a significant predictor of employee engagement. This does not support H1(b). It is possible that personal resources (positive self-evaluations) have a strong influence on perceived job resources/organisation status when determining overall engagement. We speculate that the most influential organisational factors on engagement are related less to structure and more to employees’ perceptions of the culture, leadership style and their working environment.24

There is growing interest in workplace factors that influence employee engagement due to the apparent effect that engagement has on organisation-level performance and personal well-being. Reported benefits to businesses included: improved productivity, profitability and customer satisfaction.4 16 25 26 Engagement research in healthcare settings tends to focus on health outcomes or quality metrics rather than business performance.

While comparison between countries and industrial sectors is problematic, worldwide, the NHS is ranked fifth for number of employees and the NSS is considered to be the largest annual employee survey of its kind. West and Dawson20 reported that levels of employee engagement predicted hospital quality ratings (in addition to mortality, infection rates, patient satisfaction and absenteeism). These studies were based on survey data from 2008 to 2009 but despite a change in the regulatory regime and some outcome measures, our study suggests that the link between employee engagement and quality ratings in UK secondary healthcare has been maintained.20 This finding supports our second hypothesis. The CQC highlights that financial pressures in provider organisations are associated with lower quality ratings, which is supported by the finding in our study: NHS acute Trusts with higher financial deficits as a proportion of their turnover tend to achieve lower CQC ratings. Although the CQC reported that good internal financial management is linked with better hospital ratings, this should be taken in the context of the prevailing external environment (particularly the pressure to control costs and prioritise effectiveness). CQC ratings do not decipher between organisations that are ‘better at balancing their budgets’ and the root causes of the financial ‘deficits’ in challenged Trusts.17 There are widespread calls for investment in the NHS but it remains to be seen what level of investment is required before a ‘quality dividend’ is apparent in secondary care providers in the UK.

Engagement is generally considered to be a multidimensional construct and, adopting this approach, the NSS uses a three subdimension model to assess overall engagement. Schaufeli and Bakker15 described the subdimensions of engagement as vigour, dedication and absorption. This model has been widely adopted, indeed Simpson7 recommended it be applied to all nurse-related research in order to provide a consistent approach. Although not directly compatible, the vigour dimension by Schaufeli and Bakker links to NSS motivation and the dedication dimension is related to NSS involvement or advocacy. Our study suggests that advocacy is the most influential dimension on CQC ratings. Better ratings tended to occur in organisations where employees thought the care of patients or service users was the organisation’s top priority and where they recommend their organisation as a place to work or receive treatment. This is consistent with CQC,17 which reported that ‘staff in Trusts that have received higher ratings tend to recommend their organisation as a place to work and/or receive treatment’. Furthermore, our analysis suggests that advocacy scores explain most of the variance in overall engagement scores. This implies that NHS employee engagement could be efficiently predicted by simply determining advocacy scores in future surveys. Alternatively, overall engagement could be reliability assessed by using a two subdimension model of engagement (advocacy plus motivation). This is consistent with Salanova et al 21 and Mauno et al,6 who reverted to a two-dimension model of engagement, concluding that vigour and dedication were the core dimensions of engagement in healthcare workers. Engaged employees choose to employ their energy while at work, they tend to be aware of their business context, identify with their role, are attentive and absorbed when performing their job.

The NSS advocacy scale contains a strong element of employee endorsement and has two out of three questions that are specific to healthcare. In this context, it is consistent with self-assessment by employees of their psychological state at work but diverges from other measurement tools (in that the emphasis is organisational engagement). Although this approach has utility, it could reflect a general attitude of employees towards the employer rather than the traditional view of work engagement used in most empirical research.27 28 This study does not seek to replace existing theory but rather explores the functional relationships acting on and from engagement in a healthcare setting. In broad terms, engagement as a construct may reduce to its dimensions but these dimensions cannot reduce to overall engagement. By the same token, the dimensions of engagement do not entirely explain engagement but may be used to measure it. This type of non-representativeness is permissible because it is apparent, pragmatic and accommodative to the context.

Conclusion

This study provides evidence that the NSS measure of ‘overall’ employee engagement predicts regulator’s ratings of NHS acute Trusts in England. Organisations with higher engagement scores tend to have better CQC ratings. This research also provides new evidence that the NSS engagement dimensions have different associations with these ratings. Specifically, it shows that the most influential predictor of CQC ratings is advocacy score (employees think the care of patients and service users is the organisation’s top priority, they would recommend their organisation to others as a place to work and would be happy with the standard of care provided by the organisation if a friend or relative needed treatment). Overall engagement in future NHS surveys could be reliability assessed by using a two subdimension model (advocacy and motivation) rather than current three subdimension model.

Implications

Theoretically, senior managers are best placed to modify the working environment, provide resources, moderate job demands and create the conditions that foster employee engagement. However, the pressure to control costs and prioritise effectiveness may limit the impact of lessons learnt from engagement research. Many people currently working in or being cared for in NHS acute Trusts in England are aware of the changing environment and recognise the pressures caused by externally driven reforms, high work-intensity and rising job demands experienced by many healthcare professionals. Alarmingly, these are the very conditions which have been associated with higher levels of employee burnout.14

Healthcare organisations interested in improving engagement and quality ratings should pay close attention to the proportion of employees who would recommend their organisation as a place to work or receive treatment, because this is a proxy for the level of employee engagement, and it predicts CQC ratings.

Study limitations

The study uses a cross-sectional design which limits any conclusions about causation. Although the study period is 2012–2016, the data extraction was determined by the timing of CQC inspections, so a longitudinal design would better identify the factors which consistently influence employee engagement and organisational performance. It is difficult to directly compare this study with research based on the UWES because the NHS measure of engagement is a synthesis of psychological engagement (used in most academic research) and organisational engagement (used in most practitioner research). Although the risk of reverse causality is acknowledged (CQC ratings predict engagement), during the study period the annual organisation-level engagement scores were stable while most Trusts were only rated once by the CQC.

The sample is limited to NHS acute Trusts in England, which limits the generalisability of the conclusions. The predictor variables are taken from a single self-reported source, therefore using different sources, more instruments or adding objective measurements would reduce common method variance.7 However, the criterion variable was from a different source and time which mitigates this bias.29 Self-reported observations can exaggerate relationships among variables and cannot exclude effects due to latent variables. This research uses aggregated engagement scores, whereas most previous studies have used non-aggregated scores. Shuck and Wollard30 claimed that looking at engagement at the level of an organisation rather than the individual may be necessary but it ‘distorts the nature of the concept’.

Directions of future research

Improving healthcare quality is a high priority in high-income economies and so research designed to identify factors that predict quality performance should be encouraged. Engagement as a predictor of employee or organisation performance is supported by several empirical studies.15 25 Research that overcomes the methodological issues identified in our study may provide stronger empirical evidence of the economic and healthcare benefits of an engaged workforce. Research has tended to focus on individual engagement but there is a clear need for more group-level studies particularly in service industries where many people work in teams.21 To facilitate this research, the instruments currently used to measure individual engagement need to be tested at different levels of an organisation.

While acknowledging the recommendation that a common definition of engagement be used in future research, it would nevertheless be interesting to test if the strong influence of employee endorsement seen in our study is more generalisable.7

Finally, future studies designed to identify the interventions that increase and maintain staff engagement will be of value to academia, business schools and human resource management professionals alike.