Article Text

Download PDFPDF

What does patient feedback reveal about the NHS? A mixed methods study of comments posted to the NHS Choices online service
  1. Gavin Brookes1,
  2. Paul Baker2
  1. 1 School of English, University of Nottingham, Nottingham, UK
  2. 2 Department of Linguistics and English Language, Lancaster University, Lancaster, UK
  1. Correspondence to Dr Gavin Brookes; gavin.brookes{at}


Objective To examine the key themes of positive and negative feedback in patients’ online feedback on NHS (National Health Service) services in England and to understand the specific issues within these themes and how they drive positive and negative evaluation.

Design Computer-assisted quantitative and qualitative studies of 228 113 comments (28 971 142 words) of online feedback posted to the NHS Choices website. Comments containing the most frequent positive and negative evaluative words are qualitatively examined to determine the key drivers of positive and negative feedback.

Participants Contributors posting comments about the NHS between March 2013 and September 2015.

Results Overall, NHS services were evaluated positively approximately three times more often than negatively. The four key areas of focus were: treatment, communication, interpersonal skills and system/organisation. Treatment exhibited the highest proportion of positive evaluative comments (87%), followed by communication (77%), interpersonal skills (44%) and, finally, system/organisation (41%). Qualitative analysis revealed that reference to staff interpersonal skills featured prominently, even in comments relating to treatment and system/organisational issues. Positive feedback was elicited in cases of staff being caring, compassionate and knowing patients’’ names, while rudeness, apathy and not listening were frequent drivers of negative feedback.

Conclusions Although technical competence constitutes an undoubtedly fundamental aspect of healthcare provision, staff members were much more likely to be evaluated both positively and negatively according to their interpersonal skills. Therefore, the findings reported in this study highlight the salience of such ‘soft’ skills to patients and emphasise the need for these to be focused upon and developed in staff training programmes, as well as ensuring that decisions around NHS funding do not result in demotivated and rushed staff. The findings also reveal a significant overlap between the four key themes in the ways that care is evaluated by patients.

  • Patient feedback
  • NHS Choices
  • Quantitative linguistics

This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See:

Statistics from

Request Permissions

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.

Strengths and limitations of this study

  • This study examines the largest (228 113 comments and approximately 29 million words) and most recent (March 2013 to September 2015) collection of online patient comments on NHS services analysed to date.

  • Building on previous research, the feedback data examined relate to a wider range of areas of healthcare service provision, including: acute trusts, care organisations, care providers, clinical commissioning groups, clinics, dentists, general practitioner practices, hospitals, mental health trusts, opticians and pharmacies. Although the comments relating to these various areas of provision are not compared in the analysis, this nonetheless makes for a more widely representative dataset.

  • The use of quantitative computer-assisted linguistic techniques produces large-scale, generalisable insights into this vast dataset, while more fine-grained, qualitative analysis helps elucidate nuances and areas of difference and overlap that have been overlooked by research employing solely quantitative approaches.

  • Further data and research are required to assess possible demographic trends in the feedback given.


Since the 1980s, patient feedback exercises have been undertaken by an increasing number of healthcare providers worldwide in order to monitor the quality of the services they provide and stimulate improvements where needed.1 Although the reliability of patient feedback as an indicator of the technical quality of care remains a topic of debate,2 patient feedback exercises have nonetheless become a staple way of measuring and regulating healthcare standards,3 4 as well as ensuring public involvement in the design and improvement of healthcare provision.5 Patient empowerment is, as Gann puts it, ‘here to stay’ (p. 150),6 and policy makers over the world have come to recognise the potential of active patient involvement to drive service improvements, improve self-care and ultimately improve the affordability and sustainability of the services they provide.7 In England, since 2002, patient feedback has played an increasingly significant role in the way that care quality is assessed, with all National Health Service (NHS) trusts required to collect and report the results of feedback on their services to the regulatory body, the Healthcare Commission. The importance of the insights gained from patient feedback exercises is all the more pronounced in this context, where reductions in government expenditure in areas of social provision have required healthcare providers to constantly demonstrate both the quality and financial viability of the services they provide.

Healthcare providers can obtain feedback from their patients using a range of methods which can be implemented in different settings and at differing times following an episode of treatment. Ziebland and Coulter8 provide a list of such methods, which include (but are not limited to): face-to-face interviews, postal questionnaires, telephone interviewers (using automated and live interviewers), web-based online questionnaires, diaries, questions on handheld portable devices, touch screen kiosks and bedside consoles. Moreover, feedback can be collected on-site, at the point of service contact or at patients’ homes, some days, weeks or months later.8 The analysis reported in this study focuses on feedback given in the form of online patient comments. In recent years, increasing attention has been paid by researchers and healthcare providers to the internet as a site for patients to recount their express of healthcare services and to draw attention to what was good and bad about those experiences.9–11 One such recent study was undertaken by Greaves et al,12 who compared patients’ ratings of care posted to the NHS Choices online service with the results of non-experiential measures of service performance, such as morality rates. The researchers reported that, overall, patients’ ratings tended to correlate with the non-experiential measures. For example, hospitals that were poorly evaluated by patients were found to have higher mortality rates. This research therefore supports the value of online forms of patient feedback for assessing care quality and targeting areas for improvement.

Given the increasing significance of patient feedback to the ways that healthcare services are designed, delivered and regulated, there is a pressing need for research that accounts for the concerns expressed by patients in their feedback. However, rather than explore the content of patient feedback itself, the majority of existing research in this area is concerned chiefly with: reviewing the suitability of instruments and methods of collecting and analysing feedback13; considering the reliability of feedback data for assessing healthcare quality14; reflecting on the extent to which insights gained from such exercises have actually improved service provision15 and recommending how such insights might be translated into positive clinical outcomes in the future.16 The comparatively few studies that have examined the content of patient feedback (even fewer of which relate to healthcare in England) have reported recurring drivers of feedback to include the technical quality of care, accessibility to care, and the interpersonal and communication skills of practitioners (with the latter two often conflated).17–19

The present study identifies and examines the key drivers of positive and negative feedback on healthcare services given in patients’ online comments posted to the NHS Choices website between March 2013 and September 2015. Our findings build on existing patient feedback research in several important ways. At 228 113 comments and approximately 29 million words, the feedback data we analyse are considerably larger than those examined in previous research on this topic, which have mainly accounted for hundreds of comments20–22 and at the most in the tens of thousands.23 Moreover, the data we analyse represent feedback relating to a wider range of healthcare services than considered in previous research, which has often focused on specific areas of healthcare provision.24 The lion’s share of research on patient feedback was conducted using data collected in the 1990s and early 2000s, while our dataset contains comments made as recently as September 2015, making this dataset the most up-to-date of its kind. More broadly, given the ever changing landscape of healthcare provision in England and the UK, the present study responds to the need for regular and up-to-date research that assesses patient attitudes towards the NHS and specifically identifies the key drivers of feedback about the particular services they access.



We studied the written feedback posted to the NHS Choices online service ( between March 2013 and September 2015 (data made available to the researchers). The comments were collected from the NHS Choices service’s comprehensive RSS feed for posted comments, using a developer key provided to us for this purpose, and then converted from RSS/XML to suitable structured corpus/database format for analysis. The data comprise a total of 228 113 comments, amounting to 28 971 142 words. The comments relate to a variety of healthcare organisations, including acute trusts, care organisations, care providers, clinical commissioning groups, clinics, dentists, general practitioner (GP) practices, hospitals, mental health trusts, opticians and pharmacies. However, the majority of the comments (27 005 715; 93.21%) relate to three primary care services: GP practices, hospitals and dentists. A numerical breakdown of the data is provided in table 1.

Table 1

Breakdown of the NHS comments database


We examined the comments using computer-assisted methods of linguistic analysis afforded by CQPweb,25 an online tool that offers a range of techniques for quantitatively and qualitatively analysing large collections of digitised language data. We began by identifying the 10 most frequently occurring linguistic markers of positive and negative evaluation across the comments. These words were manually identified from a list of all the words occurring in the data provided by the ‘frequency’ function of CQPweb. Evaluation is a complex linguistic phenomenon, and can be made according to a variety of parameters, including the extent to which things are important, expected, comprehensible, possible and reliable. To ensure that our analysis captured the broadest range of themes concerning the positive and negative evaluation in the comments, we focused on the most generic evaluative items, that is, words that were broadly used to describe something as either being good or bad.

Using CQPweb, we then generated a list of those words that tend to occur frequently alongside the positive and negative evaluative words in the comments, that is, their most frequent ‘collocates’. Collocation refers to ‘the characteristic co-occurrence patterns of words’.26 By analysing the collocates of the evaluative words, we were able to get a sense of what tended to be the target of the evaluation in the feedback—that is, of what was evaluated as ‘good’ and ‘bad’ in the comments. These words therefore reflect the key themes of positive and negative feedback in the data.

Building on this, the next more qualitative step in our procedure involved closely reading a randomly selected sample of comments in which each theme was evaluated positively and negatively to determine the more specific reasons or ‘drivers’ of the evaluation. Each sample consisted of 100 comments and contained comments relating to all organisations represented in the data. To ensure that 100 comments provided a sufficiently representative sample for this stage in our analysis, we adopted a saturation point procedure, well established in such quantitative linguistic research,27 of randomly selecting 30 comments, analysing the emergent patterns, proceeding to examine another 30 randomly selected comments and continuing the process until saturation point was reached and new patterns had ceased to emerge. New patterns, or drivers, were no longer emergent by the time we analysed the 100th comment (positive and negative) for each theme, and so this sample size was deemed sufficiently large to account for the common drivers of positive and negative feedback, yet small enough to facilitate fine-grain qualitative examination.


Quantitative findings

Table 2 displays the 10 most frequent positive and negative evaluative words used in the comments. The comparatively higher frequencies of the positive words (total: 223 439) compared with the negative words (total: 73 363) provide a quantitative indication that the patients are more likely to evaluate the services they access positively than negatively. The positive evaluation words occur, on average, across almost three times as many comments as the negative words.

CQPweb was then used to generate lists of those words occurring frequently within the three words preceding and following the positive and negative evaluation words in table 2 (ie, their collocates). The collocational span of 3 is the default value in CQPweb and is fairly standard in collocation analyses, which tend to operate with spans ranging from three to five words.28 As we mentioned earlier, evaluation is a complex linguistic phenomenon, and the positive words below could, in some circumstances, be used to evaluate something negatively (‘not good’) and vice versa (eg, ‘isn’t bad’). Such cases comprised a tiny proportion (under 1%) of cases and were removed from the remainder of the analysis below.

Table 2

Ten most frequent positive and negative evaluative words in the comments 

The 100 most frequent words occurring alongside the positive and negative words were thematically coded to reflect the most frequently evaluated areas of concern for patients giving feedback. Four areas emerged as frequent across the comments (corresponding words in brackets): (1) treatment (care, treatment, dental); (2) communication (communication, attention, listener(s), advice); (3) interpersonal skills (atmosphere, attitude(s), manner(s)) and (4) system/organisation (system, appointment, management, waiting time(s)). As the forthcoming qualitative analysis shall demonstrate, feedback concerning communication and interpersonal skills related to a mixture of medical and non-medical staff groups, with the latter including staff members such as receptionists and managers. Note that we combined waiting and time(s) together into one linguistic item, as references to time(s) by itself often appeared in statements like ‘I had a really bad time’, which were too vague to be categorised. Based on the corresponding words (in brackets), we then examined how often each concern featured alongside the positive versus negative evaluative words (figure 1).

Figure 1

Collocation of most frequent feedback themes with positive and negative evaluation words.

Of the four key themes displayed above, treatment exhibited the highest proportion of positive feedback, occurring alongside the positive evaluation words 87% of the time. Communication was also evaluated positively overall (77%), while interpersonal skills were only evaluated positively 44% of the time and system/organisational issues fared worst of all with only a 41% positive evaluation.

Qualitative findings

To understand why the four key themes identified in our quantitative analysis were evaluated positively and negatively across the patients’ comments, we examined a sample of comments (n=100) in which each occurred alongside the positive and then negative evaluation words. Tables 3–10 report the reasons each theme was positively and negatively evaluated in our sample. This section deals with each theme in turn, starting with the theme that fared best in the patients’ comments (treatment), and concluding with the theme that fared worst (system/organisation).


Table 3

Reasons treatment was positively evaluated

Table 4

Reasons treatment was negatively evaluated


Table 5

Reasons treatment was positively evaluated

Table 6

Reasons communication was negatively evaluated

Interpersonal skills

Table 7

Reasons interpersonal skills were positively evaluated

Table 8

Reasons interpersonal skills were negatively evaluated

System and organisation

Table 9

Reasons system and organisational issues were positively evaluated

Table 10

Reasons system and organisational issues were negatively evaluated


Statement of principal findings

The online comments analysed in this study paint a generally positive picture of healthcare services provided by the NHS in England. Our quantitative analysis of the patient comments revealed the most commonly used linguistic markers of positive evaluation occurred approximately three times as often as markers of negative evaluation and, on average, across approximately three times as many comments. Patients’ experiences and impressions of their treatment, communication, staff members’ interpersonal skills and system/organisational issues were identified as key to the ways that healthcare services were both positively and negatively evaluated. Of these key themes, treatment exhibited the highest proportion of positive feedback, occurring alongside the positive evaluation words 87% of the time. Communication was also evaluated positively overall (77%), while interpersonal skills were only evaluated positively 44% of the time and system/organisational issues fared worst of all with only a 41% positive evaluation.

Qualitative examination of the data was able to reveal the more precise nature of the evaluation made in the comments, as well as uncover overlaps and nuances between the key themes of positive and negative feedback. This part of the analysis suggested that as well as constituting a key theme in its own right, staff members’ interpersonal skills also emerged as a frequent driver of both positive and negative feedback in relation to how treatment and staff communication skills were evaluated. Other frequently cited drivers of positive and negative feedback included accessibility to care, patient centredness, and staff-to-staff and staff-to-patient communication. Staff technical competence was a less prominent driver of feedback, cited only in relation to the evaluation of treatment itself, and accounting for a relative minority of these comments. Our findings therefore support the notion that there is a discord between the significance that practitioners and patients place on technical competence when judging the overall quality of care.29

Strengths and weaknesses of the study

This study has examined the largest and most up-to-date collection of patient feedback on NHS services in England in any format. Our use of quantitative computer-assisted linguistic techniques has produced large-scale, generalisable insights into this dataset. Yet at the same time, the more fine-grained, qualitative analysis was able to elucidate areas of difference and overlap that have been overlooked by research employing solely quantitative approaches in the past.

Although this dataset has proven to be a valuable resource for learning about individuals’ perspectives on the healthcare services they access, its lack of metadata regarding the demographic information of individual contributors meant that it was not possible to attribute particular types of comment or concern to any demographic group. It is also worth bearing in mind that the majority of the comments we analysed (93.21%) relate to the primary care services of GPs, hospitals and dentists. While this is unlikely to present issues respecting the general trends examined, more specific comments relating to these areas might be said to be over-represented compared with other areas of service provision, such as care providers and mental health trusts. Moreover, the data analysed in this study represent feedback given in one specific form (online comments), posted to one particular website (NHS Choices), about organisations based in one country (England). This raises issues surrounding representativeness; for those who choose to share their experiences online are not necessarily representative of the general population. It is now well documented that, compared with non-internet users, internet users tend to be younger, are more educated and are from higher income brackets.30 Although this digital divide is estimated to have narrowed over time,11 the perspectives of people from these so-called ‘hard-to-reach’ or ‘seldom-heard’ groups are still likely to be under-represented in our data.8 31

Strengths and weaknesses in relation to other studies and important differences in results

Where a great deal of existing research has explored patient feedback in terms of predetermined themes, the data-driven approach adopted in the present study has allowed drivers of feedback to emerge from the comments themselves throughout the course of the analysis. As a consequence, system and organisation issues, which have remained largely unexplored in existing research, have emerged here as significant drivers of positive and negative evaluation with respect to various other aspects of care, including quality of treatment and staff communication skills.

As well as providing fresh insight into the perspectives of patients accessing contemporary healthcare services in England, the findings reported in this study also provide more substantive quantitative evidence to support the findings reported in existing studies of patient feedback that are based on comparatively smaller and older datasets.32 33 However, our findings highlight the centrality of interpersonal skills as a key area of concern in its own right and as significant to the ways that treatment quality and staff communication are evaluated.

Meaning of the study: possible explanations and implications for clinicians and policy makers

While the majority of research into patient feedback has focused principally—in most cases exclusively—on what motivates negative feedback, the present study has elucidated the drivers of positive and negative feedback equally. Accordingly, while the reported drivers of negative feedback might flag up areas that require attention, the specific drivers of positive feedback outlined over the course of the analysis offer insight that can be used to stimulate and guide quality improvement efforts.34–36

The quantitative section of our analysis suggested system and organisational issues to be a prominent theme in negative feedback. This often relates to issues surrounding accessibility of care, such as (emergency) appointment availability, waiting times, technical difficulties experienced with online booking systems, telephone waiting times and practice opening times. Tightening government expenditure in healthcare provision and resultant constraints on practitioner time and availability mean that these issues are unlikely to abate. Such issues arguably lie within the remit of policy makers and governing bodies. However, practitioners and other staff can improve patient feedback in this area by making an effort to ensure that appointments run on time and informing and updating patients and their families in the case of cancellations or delays.

The qualitative section of our analysis suggests that staff interpersonal skills lie central to improving care, as these were shown to motivate negative (and positive) feedback in relation to a variety of areas of concern. On the surface, the findings of this study suggest that developing the interpersonal skills of staff should be a priority in staff training. The interpersonal skills of both medical and non-medical staff were evaluated positively for qualities such as being friendly and approachable, empathetic, for smiling, and not being afraid to laugh and joke with patients. Allied to this, practitioners were frequently positively evaluated for providing care that was patient centred and involved discussing treatment options with patients, as well as explaining treatment plans and listening to their concerns. Conversely, staff were negatively evaluated when they were perceived as being rude, dismissive, lazy, not listening to patients’ concerns, as well as for not smiling and appearing unhappy.

Most professional medical training (eg, in medicine and nursing) includes the development of communication skills as a key element at the undergraduate level and onwards. This kind of interpersonal training is often focused on developing skills such as  information gathering and shared decision making. Our findings suggest that, as far as patients are concerned, the interpersonal aspect of interaction is given a high premium. Such skills might be developed more effectively through greater opportunity for hands-on human engagement, rather than instruction alone, at the early stages of training. In terms of developing the interpersonal skills of other non-medical staff groups, it is likely that many staff working in administrative capacities, such as receptionists, will not have received formal training in interpersonal skills (although healthcare providers are increasingly running courses in ‘customer service’ to address this). Likewise, there is an intermediate group, which includes staff working in healthcare assistance, who may have received limited or no formal training, but who nonetheless engage with patients at very significant levels and may benefit from some form of interpersonal skills training.

However, many of these specific interpersonal or ‘soft’ skills can be linked to the concept of emotional labour,37 which involves the regulation of emotion to create a publicly visible facial and bodily display within the workplace. These kinds of attributes might seem more like individual character traits, and so incorporating these into training poses challenges, especially as members of the public are often unimpressed by ‘scripted’ interactions which are rightly seen as inauthentic.38 Soft skills training, while clearly helpful in some areas, may sometimes be a ‘sticking plaster’ solution to cover for wider structural problems involving overstretched systems. Other positively evaluated interpersonal aspects of care, such as involving patients in communication and decision making and ensuring that patients have sufficient time to interact with medical staff and are not made to feel as though they are ‘rushed,’ might constitute more tangible and attainable targets for skills development programmes.

In an effort to stimulate such improvements, the findings from the project from which this research derives, including the results reported in this article, have been presented to the Insight and Feedback Team at NHS England as well as the Care Quality Commission in the UK. However, in reality, translating such findings into practice is seldom straightforward. After all, although most major public sector healthcare providers collect feedback on their services at least annually, this information is not always used to improve service quality.3 Suggested reasons for this include a lack of attention to patients’ experience at senior levels,39 as well as feedback data not being specific to particular wards or teams.40

As well as gesturing towards areas for improvement in healthcare provision, the findings reported here also provide insights into patient feedback more generally; insights that likely bear implications for how such feedback should be interpreted in the future. Our quantitative examination of the data revealed significant overlap between the drivers of positive and negative feedback. As an example, although treatment fared best of these four themes of feedback in terms of positive evaluation, it was only by examining the comments relating to treatment that we were able to show that 47% of the positive comments relating to this theme actually praised interpersonal aspects of care, rather than the technical competence of staff, which accounted for only 10% of these comments. It is therefore beneficial, where possible, to gauge feedback at a granular level. This is where combining quantitative and qualitative approaches can bear significant advantages for researchers, allowing us to deal with large datasets in a way that is sensitive to subtle nuances and overlaps and to  point to specific areas for praise or improvement that only become apparent at the more granular level.

Unanswered questions and future research

Although the data contain feedback relating to a variety of healthcare organisations, such distinctions have not figured in the analysis undertaken in this study. Future studies on this particular dataset should therefore take a modular, even comparative approach, to ascertain similarities or differences in the ways that care is evaluated in each of these areas. Furthermore, we did not have access to the demographic information of the comment posters. Future research should endeavour to collect and examine data that comprise this kind of demographic metadata in order to determine whether particular concerns are attributable to people living in certain locations or belonging to particular age, ethnic or sex-related groups. Future research should also assess feedback given in other, particularly non-digital, mediums in order to help account for the perspectives of patients from ‘hard-to-reach’, ‘seldom-heard’ groups who might be less likely to give feedback online. Ziebland and Coulter8 recommend that non-traditional methods of data collection, such as pictures, stories and drama, might be used to incorporate the views of such groups in future feedback collection exercises.


We wish to thank Dick Churchill for his advice concerning the pedagogical implications of our findings.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
View Abstract


  • Contributors GB co-planned and co-conducted the research, and took charge of writing the paper. He is responsible for the content as guarantor. PB co-planned and co-conducted the research and contributed to the writing of the paper. Both authors had access to the data.

  • Funding Economic and Social Research Council; grant number: ES/K002155/1.

  • Disclaimer All authors have completed the ICMJE uniform disclosure form and declare: all authors had financial support from the Economic and Social Research Council for the submitted work; no financial relationships with any organisations that might have an interest in the submitted work in the previous 3 years; no other relationships or activities that could appear to have influenced the submitted work.

  • Competing interests None declared.

  • Ethics approval Research ethics approval for the study was obtained from Lancaster University, Lancaster, UK.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement No additional data available.