Objective: To examine whether staff feedback on quality, safety and workforce issues is reflected in patient-reported experience.
Setting: 166 NHS acute trusts in England.
Participants: Respondents to the 2006 surveys of adult inpatients and staff in 166 NHS acute trusts in England.
Methods: Multiple linear regression was used to model the relationship between responses by “frontline” staff and inpatients at trust level. Staff survey items were the explanatory variables and inpatient responses the dependent variables. Adjustments were made for location (London and non-London) and trust type.
Results: 69 500 staff and 81 000 patients responded to the surveys. There were several significant associations between staff and patients’ responses, including that staff availability of hand-washing materials was positively associated with patient feedback on cleanliness and hand washing by doctors/nurses (p<0.00). It was a significant predictor of patient experience also in several other models. Other significant predictors of patient experience were managerial support, witnessing and reporting of errors (positively associated with patient experience), working extra hours and stress (negatively associated). London trusts performed worse on patient experience than trusts outside London and specialist trusts performed better than other acute trusts.
Conclusions: Staff feedback was associated with patient-reported experience. Positive staff feedback on availability of hand-washing materials was broadly reflective of positive patient experience. Negative staff experience was reflected in poorer patient experience and vice versa. Although we cannot demonstrate causality, the consistent direction of the findings is indicative of it. Management boards of trusts and clinicians and other staff should monitor and act on the results of their staff surveys. This has the potential for improving quality, safety and patient experience.
Statistics from Altmetric.com
“Empowered patients and empowered staff are the key to world-class standards” (Lord Darzi, on publication of Leading Local Change, May 2008)
Safe, high-quality NHS care and improvements in patient experience are key elements of government policy in healthcare. The NHS Next Stage Review (the Darzi report) explicitly acknowledges the role that NHS staff play in achieving these aims.12 The NHS Operating Framework 2008/2009 for England builds on these themes, setting out five priorities, including “improving patient experience, staff satisfaction and engagement”.3 Staff feedback on organisational practices and service quality is critical for identifying factors that impact on the safety and quality of services and patients’ experiences of them. However, research on the relationship between the experiences of healthcare staff and patients is limited. West noted that appraisal, training and continuing professional development for doctors has a positive impact on patient care and mortality and that these processes need to be informed by patient feedback.4 Cleopas et al examined patient assessments of a hypothetical medication error scenario and concluded that poor patient ratings of healthcare were associated with slow staff response to the error, non-disclosure of the error and suffering serious health consequences.5 Fottler et al compared hospital staff and patient perceptions of services in the USA and found strong correlation between staff and patient perceptions of customer services.6
The present study is uniquely different from previous research in that it uses evidence from two large, contemporaneous surveys to explore the relationship between staff behaviours, attitudes and experiences and a range of patient experiences and observations. We examined associations between the responses to the 2006 national surveys of staff and adult inpatients in NHS acute and specialist trusts in England. The results show some significant patterns, indicating that the safety and quality of services and patient experience could improve if trusts act on feedback from their staff. The analysis also examines the impact on patient experience of location (London, non-London) and trust type (general acute, teaching, specialist).
In 2006, almost 81 000 patients responded to a survey asking about their recent experiences as an inpatient in one of the 167 acute or specialist NHS trusts in England, with a response rate of 59%.7 The numbers of respondents ranged from 240 to 660 between trusts. Responses to key questions were scored on a scale of 0–100, where higher scores represented better patient experience. The scores were adjusted for age, sex and method of admission (planned vs emergency) to take account of differences in the distribution of these patient characteristics between trusts.
For this analysis, we used scored responses to 17 questions relating to the following aspects of patient experience: respect and dignity (four questions), staff-patient relationships (four questions), information (four questions), cleanliness (four questions) and an overall rating of care (one question). The questions, their mean scores and SD are given in online Appendix A.
In 2006, almost 69 500 staff from 171 acute and specialist trusts in England took part in a survey asking about their views and experiences of working for the NHS, with a response rate of 52%.8 Data for the 166 trusts that matched the patient survey were used for this analysis.
The staff survey results were analysed as scores on 28 separate items. The items (along with their mean percentages/scores and SD) are given in online Appendix B. It was hypothesised that links between staff and patient responses might be clearer if the analysis was restricted to frontline staff with direct contact with patients, in contrast to administrative and managerial staff who may have little or no patient contact. Online Appendix C lists the staff groups included in the analysis. Frontline staff comprised 52%–78% of staff in the 166 trusts, with an average 65%. The vast majority of staff included in the analysis would work wholly or mainly with the inpatient populations covered by the patient survey.
The staff survey data used for the analysis comprised:
percentages of staff answering “yes” to 18 questions (eg, whether they had an appraisal in the previous 12 months) and
mean scores (on a 5-point agreement scale) for 10 multi-item questions.9
The staff survey scores were aggregated by taking a weighted average of the occupational groups comprising frontline staff. The scores used in the regression analysis were rescaled to a scale of 0–100 to enable comparison of the regression coefficients across variables.
Further details about both surveys and the scoring methodology are available at the links provided.
Multiple linear regression was used to model the relationship between responses to the staff and inpatient surveys, using data at trust level. The 28 staff survey items were used as explanatory variables, and inpatient responses to selected questions were the dependent variables. The analysis was conducted at trust level.
Forward stepwise regression was used, with thresholds of p = 0.10 for inclusion and p = 0.15 for removal of the explanatory variables. Dummy variables were included to take account of the variation explained by trust location (London and non-London) and trust type (general acute, teaching, specialist), as these factors are known to influence the responses of patients. The numbers of trusts in these categories are given in table 1. The dummy variables were fixed in the model so that each model was adjusted in the same way.
Respect and dignity
The variation in the four dependent variables explained by the regression models ranged between 38% and 57% (table 2). Support from immediate managers (coefficient 0.41), availability of hand-washing materials (0.24) and witnessing potential errors (0.12) were positively associated with patients’ responses; work-related stress (−0.09) and working extra hours (−0.08) were negatively associated. Both work-related stress (−0.13 and −0.13 respectively) and violence from colleagues (−0.70 and −0.46 respectively) were less prevalent in trusts where fewer patients said that doctors and nurses talked in front of them as if they were not there. Managerial support (0.16) and working extra hours (−0.04) showed positive and negative associations respectively with patient privacy. Work-related stress featured negatively in three of the four models. While most associations were in the expected direction, two variables showed an unexpected association: quality of job design was negatively associated with being treated with respect and dignity (−0.29) and work-related injury was positively associated with fewer patients having doctors talking in front of them as if they were not there (0.19).
London trusts consistently performed significantly worse than trusts outside London (coefficients −1.08 to −4.14), especially in terms of whether staff talked in front of patients. Specialist trusts performed better than other trusts (2.76 to 6.81).
The variation in the four dependent variables explained by the regression models ranged between 31% and 57% (table 3). For doctor-nurse team working, the strongest positive associations were with managerial support (coefficient 0.34), availability of hand-washing materials (0.19) and witnessing potential errors etc (0.12); working extra hours was negatively associated (−0.15). Staff suffering stress was negatively associated with patients saying they received answers from doctors that they could understand (−0.12). Staff reporting errors was positively associated with both doctors and nurses giving comprehensible answers to patients (0.20 and 0.26 respectively). Managerial support (0.60) and availability of hand-washing materials (0.48) showed positive associations with nurses giving comprehensible answers to patients, although the negative association with quality of job design was again unexpected. Availability of hand-washing materials appeared positively in three of the four models. Again, London trusts consistently performed worse than trusts outside London (coefficients −1.40 to –2.89). Specialist trusts and to a lesser extent teaching trusts, performed better than other trusts.
The variation explained by the regression models was highest for the question about patient involvement in decisions about care and treatment (R2 0.51) (table 4). Reporting errors, near-misses etc was positively associated with this question (0.29) and staff intention to leave was negatively associated (−0.20). Staff-related factors explained less of the variation in patient feedback about information on medication; however, both procedures for reporting errors etc (0.39) and staff reporting errors etc (0.48) were positively associated. The latter was also positively associated with patients’ receiving information on discharge (0.45), as was managerial support (0.96). Staff reporting errors etc was positively associated with three of the four dependent variables in this section (coefficients 0.29 to 0.48). London trusts consistently performed worse than trusts outside London (coefficients −1.76 to −5.31). Specialist trusts (coefficients 4.22 to 10.30), and to a lesser extent teaching trusts (coefficients 0.71 to 2.44), performed better than other trusts.
The variation in the four dependent variables explained by the regression models ranged between 39% and 54% (table 5). Availability of hand-washing materials as reported by staff was positively associated with patient feedback about the cleanliness of inpatient accommodation (coefficients 0.56 and 0.50) and doctors/nurses washing hands between patients (0.23 and 0.37). Managerial support was positively associated with three of these dependent variables. Procedures for reporting errors and staff reporting errors showed some positive associations with availability of hand-washing facilities, as did health and safety training. The negative association of staff working in a well-structured team environment with two of the dependent variables was unexpected. Similar effects as for the other analyses were apparent here for trust location (range -0.23 to –2.87). The strong association between specialist trust type and patient experience was again largely apparent (coefficients 0.66 to 7.73).
Overall rating of care
For patients’ overall rating of care (table 6), five staff survey items emerged as significant predictors: support from immediate managers (coefficient 0.41), availability of hand-washing materials (0.23), staff witnessing potential errors (0.14) and health and safety training (0.06) were positively associated with overall patient experience, while staff working extra hours (−0.13) was negatively associated. London trusts performed significantly worse on overall patient experience than trusts outside London. Both teaching trusts and, in particular, specialist trusts performed significantly better than general acute trusts. The variation in the dependent variable explained by the model was 59%.
The national surveys of patient experience are designed to provide direct feedback from patients about their experiences of NHS services. The NHS staff surveys are designed to provide direct feedback on employment, working experience, attitudes and behaviours. The aim of these surveys is to help NHS trusts to identify areas for improvement. The surveys use a robust, standardised design and sampling methodology and have large, nationally representative samples. The data therefore offer unique opportunities for secondary analysis such as this. This article is based entirely on self-reported feedback from staff and patients.
Limitations of the study
The same predictor variables from the staff survey were used throughout the analysis and stepwise regression was used to identify those that were significant. We recognise that stepwise regression aims to maximise the “goodness of fit” of any model. However, as there was no established theory or specific hypothesis to test, stepwise regression was used as an exploratory tool to generate models that include more than one predictor and thereby provide a basis for discussing which factors are important when considered together.
There were correlations within the staff survey data; hence, some predictive variables on the staff side may be displaced by others in the regression analyses, while others may be indicative of a relationship between inpatient scores and some underlying level of quality on the staff side. Further, as our analyses explain only up to 59% of the variation in patient experience, clearly there are other determinants operating also.
Finally, it was not possible to adjust the patient survey data for diagnosis; it is possible that, for example, surgical patients respond differently to medical patients.
We recognise that this analysis does not demonstrate causality. Further, while we are generally assuming that staff factors impact on patient experience, it is possible for causality to also operate in the reverse direction (ie, patient dissatisfaction or behaviour could impact on staff responses).
Associations between staff and patient responses
Most of the statistically significant associations observed between staff and patient responses went in the “expected” direction (ie, positive staff experience was associated with positive patient experience). The associations were also generally plausible (eg, managerial support was positively associated with patients being treated with respect and dignity and good team working between clinical staff). Most of the staff factors emerging as predictors of patient experience were related to working conditions (eg, working extra hours, stress, managerial support). Workforce issues such as appraisal and training did not appear as significant; however, some factors could be proxies for others (given the high degree of correlation between the staff survey items).
Most striking of all was the consistent positive association between staff feedback on availability of hand-washing materials and patient feedback on cleanliness of the wards and toilets/bathrooms and hand washing by doctors/nurses between touching patients. Availability of hand-washing materials was associated also with a range of other patient experience dimensions, which could indicate that this staff variable acts also as a proxy and reflects wider organisational factors conducive to positive patient experience. In the current climate of concerns about safety and hospital-acquired infections, we consider this finding to be highly pertinent and it reinforces cleanliness and hygiene as priority areas for action by trusts.
Another pertinent finding was the positive association between staff reporting/witnessing potential errors, near misses, etc and aspects of patient experience, which may reflect the “safety culture” in an organisation. Research indicates that higher reporting rates to the National Reporting and Learning System of the National Patient Safety Agency correlate positively with some independent measures of a safety culture, such as feedback on incident reporting from the NHS staff survey and risk management ratings from the NHS Litigation Authority.10 The authors conclude that higher reporting rates may be indicative of a positive organisational culture of safety and reporting. Trust boards should therefore encourage and support the reporting of errors and the learning from them.
Our findings support the view that trusts should monitor their staff survey results closely. Action plans can be produced by human resource advisors, in conjunction with representatives from individual care groups and directorates, which identify the key areas where working conditions and practices can be improved and the actions that are required to achieve these improvements. Progress on the action plans can then be reviewed by boards during their routine meetings with each care group and directorate.
Findings of other studies
While there is evidence of commercial sector approaches linking employee and customer satisfaction,11 there is a paucity of research on the correspondence between staff and patient perceptions of the quality of healthcare services. Based on US data, Fottler et al found that staff and patient perceptions of customer satisfaction with healthcare services were highly correlated—although they also noted that staff were more negative in their perceptions of patient satisfaction than patients themselves.6
However, most studies focus on the association between staffing levels and patient outcomes, namely mortality. In a systematic review of 27 studies investigating the impact of the working environment on patient mortality, Kazanijan et al found that while some studies report a positive relationship between nursing workload and patient mortality, this is not a consistent finding.12 Studies also identify processes such as “burnout” and job satisfaction as mediating between the nursing environment and patient outcomes. West et al investigated the relationship between staff management policies and practices and patient outcomes in 52 NHS hospitals. They conclude that, after controlling for confounding factors including prior mortality rates and doctor-patient ratios, their findings suggest that human resource management practices (especially appraisal, job security and Investors in People status) are related to the quality of healthcare and specifically to hospital patient mortality.13 The present study is uniquely different in that it uses evidence from two contemporaneous, large-scale surveys to explore a range of staff inputs (behaviours, attitudes and experiences) and a range of patient experiences and observations.
Effects of other factors
The location and type of trust were also significant predictors of patient experience. London trusts performed significantly worse on the patient experience measures than trusts outside London. This pattern is also apparent in our analyses of surveys of other patient groups.14 The reasons are unclear and could be related to factors specific to London, such as staffing issues, or factors external to healthcare, such as deprivation, the diverse ethnic and cultural mix of London’s population, language issues, transient populations, etc. It is perhaps unsurprising that specialist trusts perform better than general acute trusts, given their specialised case mix. Although the patient survey scores are adjusted for admission method, this may not compensate adequately for the effects of a diverse case mix, provision of emergency services, etc.
Overall, our findings show that positive staff feedback is associated with good patient experience. These findings are especially pertinent in the context of Lord Darzi’s vision of listening to and empowering both patients and staff. Although this analysis cannot demonstrate causality, the consistent direction of the findings is indicative of it, and it seems reasonable to conclude that staff feedback is a useful marker of patient experience. Some findings are of particular interest, for example, that patient and staff perceptions about cleanliness correspond. Other feedback from staff—such as managerial support, working extra hours and work-related stress—also suggest that the working conditions and morale of staff impacts on patient experience. We therefore recommend that the management boards of trusts, clinicians and other staff in trusts, monitor closely and act on the results of their staff surveys. This has the potential for improving patient experience and safety. The importance of good leadership, effective management and systematic use of information was highlighted in the Healthcare Commission’s report Learning from Investigations, the recommendations of which included that senior managers must elicit views about safety from frontline staff.15
The national patient and staff surveys were managed by the Healthcare Commission, on behalf of the Department of Health. These surveys were developed and coordinated on behalf of the Healthcare Commission by Aston University Business School and by the Picker Institute Europe. The analysis was done under the aegis of the Healthcare Commission. We are grateful to all these organisations and to the patients and staff who participated in the surveys. The views are those of the authors. The survey programme is now managed by the Care Quality Commission.
▸ Additional appendices are published online only at http://qshc.bmj.com/content/vol18/issue5
Funding The NHS Staff Survey Advice Centre at Aston University Business School and the Acute Patient Surveys Co-ordination Centre at the Picker Institute were both funded by the Healthcare Commission. The NHS staff and patient surveys were funded locally by the NHS trusts involved.
Competing interests None.
Ethics approval The NHS Staff survey and the acute hospital inpatient surveys both have approval from the North-West Multi-Centre Research Ethics Committee.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.