A randomised controlled trial, cost-effectiveness and process evaluation of the implementation of self-management for chronic gastrointestinal disorders in primary care, and linked projects on identification and risk assessment

Health-related quality of life

Health-related quality of life was measured in the trial using quality-adjusted life-years (QALYs). QALYs were generated as the product of the health state of each individual and the time spent in that state. The health state of each individual in the study was assessed at entry to the trial (baseline), and at 6- and 12-month follow-up using the EQ-5D descriptive system.

Quality-adjusted life-years were estimated using the area under the curve approach,63 with linear interpolation between EQ-5D scores at each follow-up point. The QALY estimates are presented adjusted for baseline EQ-5D score,64 but also without the adjustment.

Resource use and unit costs

At each follow-up (6 and 12 months post randomisation), patients were asked to recall their use in the last 6 months of hospital services (including inpatient stays and outpatient attendances), visits to GP surgery (GP or practice nurse), home visits (from GP, physiotherapist, or occupational therapist) and other health/social sector resource use.

The unit costs of health services, for example the cost of a visit to a GP, were estimated using the published literature and are presented in Table 5. The unit costs were then applied to the appropriate resource use item.

Findings

Summary of organisational-level interviews

Overall, the implementation seems to have progressed well, although more slowly than had been anticipated. The initial problems relating to communication, extra funding for GP time and recruitment of practices appear to have been resolved. However, there was a consistent view that the WISE model had not developed a high enough profile within the PCT. This seems to have arisen because of the tensions created by placing the WISE model outside the Commissioning Directorate. Thus, commissioning managers did not develop much knowledge of the aims of the WISE model. In addition, there is a sense that the fact that it was perceived as a pilot may have contributed to the lack of buy-in at managerial level.

Recruitment of practices

The project started with a list of 56 practices. Practices were not eligible if they shared clinicians or administrative staff with other practices or if they served a very specific population (e.g. drug users, the homeless or young patients). Forty-four practices in the Salford PCT were identified as potentially eligible for the study, and all were contacted. Of these, 32 initially agreed to participate in the trial and were randomised, and three subsequently withdrew from the study before data were collected from 29 practices. To ensure that sufficient power for the study was maintained, we recruited 12 practices from Bury (a neighbouring PCT serving a similar population to Salford).

Salford has a large number of single-handed practices, and 17 were identified as eligible for the trial: 14 of the early sign-ups were from single-handed practices and two of the ineligible practices were single-handers. High sign-up from single-handed practices may reflect the fact that it proved easier to negotiate times of meeting and set dates for training when only one GP was involved.

From the start, practices had different reactions to the introduction of the WISE model training. Reactions ranged from interest and enthusiasm to complete lack of interest and hostility. Few perceived that training in self-management support was needed, so the approach to engaging practices involved the use of incentives (costs of attending training to be met by the PCT), as well as targeting practices known to be well disposed to involvement in research early on and using this to encourage more reluctant GPs to join the programme. The involvement in a research project was deliberately downplayed because of the known difficulties of getting GPs to engage with research. Barriers to engagement included:

• Practical problems such as staff sickness, arranging time out of the practice to undertake the training and the implications for practice finances (although the PCT provided funding for locum and out-of-hours cover to allow participation in training).

• Practices faced with incorporating other changes, at the same time, to practice and management.

• Self-management not being seen as a priority for the practice and a lack of belief that supporting self-management would be effective.

• A perception of the WISE model as providing nothing new. Staff claimed that they were already providing good care for patients with systems and strategies for long-term conditions in place. Furthermore, they could see no need to change practice and no potential benefit for patients in their attending training.

• A perception among some practices that patients were unlikely to take-up, or benefit from, a self-management approach.

• Training being considered by some to be inappropriate for support staff.

• Signing up to the training and the link to research meaning that clinicians would be expected to invest additional time with patients providing self-management support (i.e. there was concern that using the WISE model approach would generate more work).

• Negativity towards the inclusion of IBS as one of the chronic conditions being included.

Training impact

Evaluation forms were completed by staff at the end of each training session. The scores indicated high satisfaction with the training (Tables 12 and 13).

Questionnaire survey of the use of the WISE model tools

The questionnaire was sent to 302 staff at the 31 practices where training took place (25 in Salford and six in Bury), including 163 administrative staff and 139 clinicians. There was an overall response rate of 48% (88 administration staff and 67 clinicians) (Table 14).

Face-to-face in-depth interviews with practice staff

The analysis presented here is based on interviews with 11 practice nurses and one assistant practitioner; all were female with between 2 and 21 years of experience of working in general practice. Interviews with other staff, documents and notes collected during the study and the training sessions and the observations carried out in the development stage of the WISE model helped to provide context for the analysis.

Findings

Table 15 provides a summary of how the systematic identification of work undertaken by nurses mapped on to the NPT framework. The analysis focused on the ways nurses spoke about the work of managing patients with long-term conditions and how using the WISE model tools changed their everyday self-management practices. Some elements of the WISE model were reported to work well (such as distribution of the guidebooks), but other elements (such as using the PRISMS tool to help address patient needs and priorities) were rarely taken up. A number of themes emerged. Illustrative data are presented and labelled with the respondent identifier.

Meaning and use of the PRISMS tool

The majority of participants in both studies had not used the PRISMS form. When shown the form there was a range of responses about its utility. The participants who did consider it something they could use primarily saw it as a memory prompt and mechanism of introducing relevant issues into the consultation which otherwise would be forgotten or not raised (e.g. if it was a sensitive issue or something that was considered embarrassing, this provided an opportunity that might not occur naturally during the consultation). Of most concern was the receptivity of the GPs to patient-initiated prompts and thus the feasibility of making this work in practice. In this respect patients were rehearsing the needs of professionals and conscious of the problems of disrupting consultation practices established over time.

I haven’t used it. But, it’s um, it is a good idea actually because if you’re like me, you keep forgetting, you can go through it, tick it all off and then just hand it in to your GP when you go. Yeah that’s a good idea that actually . . . Yeah, yeah, that’d be better that, fill it in at your leisure and then when you go down there you can just hand it in. And then they’d know what, rather than trying to remember what you were going to say when you get there . . . it’s a good idea. But, whether the GPs would like it you don’t know, they should do, they can just, alright then, write you a prescription and give it you and you’re away aren’t you?

ID 43309, female, index condition COPD

The perceived usefulness of the PRISMS tool by patients was influenced by the participants’ relationship with their GP and the expected response. One participant who had used the PRISMS form during a primary care consultation reported that this had enabled her to raise ongoing health problems that were not condition specific (i.e. increasing tiredness), and to be referred to appropriate services (Box 2). Use of the PRISMS tool was considered by respondents with reference to previous consultations and encounters with their GP or practice nurse. Participants’ perceived relationship with their GP ranged from good to poor. When participants reported a good relationship with their GP, they described being able to raise concerns easily. However, for the majority of participants, their relationship ranged from ambivalent to negative and this seemingly influenced the perceived acceptability of introducing the PRISMS form. In addition, some participants who considered their conditions to be stable did not feel that the PRISMS tool was applicable to them at that time but might be in the future. This points to the need for tools to be sensitive to patient trajectories and changing needs and circumstances.

The PRISMS form. Do you think that’s useful?

Not to me.

How come? . . .

Most of this is not applicable . . . Maybe because of my age, maybe because it’s diet controlled diabetes . . . It might be more applicable later on.

ID 12188, index condition: diabetes, male

The Salford online self-care support guide

None of the participants reported using the self-care guide or being referred to it by their GP or practice nurse.

Condition guidebooks

Participants in the longitudinal study had not received the guidebooks; however, five participants provided feedback on the content in the final interview (Box 3). The majority of participants in the process evaluation of the main trial did not recollect being sent the guidebooks through the post or being given them during a consultation. However, in dialogue with the interviewer, the following participant had received the guidebook and found it useful.

Yeah I sat down and read them. I did sit down, that one [guidebook], that yeah. Yeah you read them and if there’s anything that catches the eye, what I’ll do, yeah I do take notice of what people say and what advice . . . Yeah, they’re good. They are, they are useful.

ID 43309, female, index condition COPD

Stepped-up care

The stepped-up care option was available to IBS participants in the process evaluation only (n = 10) and none had been referred to stepped-up care.

Patient expectations of self-management support

Patients’ expectations of help with self-management were shaped by previous experiences with their GP and expectations of support available. There were generally low expectations of support based on poor previous experiences, and difficult and unhelpful relationships.

. . . but when your doctor says something that really doesn’t help, you know, and I have had her say ‘well you just need to control what you put in your bloody mouth’.

ID 62153, female, diabetes patient

Such events resulted in people avoiding going to their GPs because of their low opinions of the help they would get.

. . . she’s no good as a doctor basically . . . I’d rather suffer than have to go and see her. I only go if I really need to go.

ID 63325, female, IBS patient

The restrictions imposed by the way in which primary care is organised militates against meaningful and supportive discussions and adds to the lack of confidence in obtaining support.

Well, first time I went to her was Thursday morning and apparently you’ve got 10 minutes, but it’s like 2 weeks to get in to see her. So I had a little list, because my memory’s not very good with this fibromyalgia. Sometimes I’m alright, other times I can’t remember things. And I wanted advice on my IBS, seeing if she could come up with something different while I was in the process of suffering as well as this fibromyalgia, as well as menopause, things like that, but she said right, start with the most important, and then when my 10 minutes was up she said would you like to make another appointment please. I thought right, I’ve got another 2 weeks to wait. So I’m not pleased. Not when the person in front of me was in more than 10 minutes.

ID 33304, female, IBS patient

In order to get support that they felt was appropriate, patients often had to work hard to persuade the doctor and the build-up of negative experiences contributed to the low expectations of help.

I had to go to [Town] to a pharmacy that does the programme there . . . you can only purchase it through a pharmacy. And they weigh you every week, they advise you, but they will only let you do the programme with the agreement of the doctor. So they approached my doctor, she was reluctant to do it at first, and it took a month to get her agreement for me to do it.

ID 62153, female, diabetes patient

The examples of positive support occurred when existing relationships were good and when the doctor or nurse was perceived to have genuinely listened; this appeared most likely in nurse consultations.

I usually ask if I can make an appointment with the nurse, and obviously they ask me why. She’s very good. The nurse at my GP centre is very good. They will sit and listen. Listening makes a big difference, while a doctor’s looking at you and looking right through you. You’re just sat there and you’re making me money.

ID 61015, female, COPD patient

Managing additional conditions

Additional conditions were more readily accommodated if people established cognitive links between existing management practices. Thus, multiple conditions were not inevitably experienced as an increasing burden but subject to considerable flux and change. Prioritising one condition over another at a particular time together with a transfer and amalgamation of practices appears to facilitate accommodation of multiple conditions. Clinicians might usefully engage with patients’ understanding to reduce complexity, and enhance engagement of condition management.

However, patients who developed additional long-term conditions could experience that as disruptive.

I don’t know whether it’s I am carrying too much weight or I’m having too much sugar . . . but I’m trying to find out . . . and all I get from [GP], it is your diabetics, they give it a different name all together, I forget, I can’t get my bloody tongue round it and yeah, that’s one main thing and the other is, I have had my eyes done cataract done and that, that were all right but um, it’s just niggly.

ID 0900 (diabetes, knee problems and kidney problems, male, age unknown)

I have, yeah just get on with it. It’s a chronic condition, there’s not a lot I can do about it other than, except I’d like to take more exercise, walking, but at the moment as I say because of my feet problems I can’t, so it’s a double whammy as they say.

ID 110s (COPD, oesophageal problems, feet problems, male, age 69 years)

Conversely, accommodation of further conditions could occur as there was a flow of existing management practices reducing the impact of further conditions on the daily management.

I have to sort of eat regularly for the epilepsy, eat regularly, not get over tired, not drink too much alcohol . . . don’t skip meals, so that is what I do for me epilepsy anyway, so that’s what helps with the diabetes as well.

ID170a (diabetes and epilepsy, female, age 55 years)

Practices that learned and adapted from the management of one condition allowed some respondents to accommodate new conditions if new routines were congruent with existing routines. This contrasts with a view that multimorbidity by definition requires more complex clinical management accompanied by a new set of practices for each ‘new’ condition. Of greatest salience to our respondents was a process of continuation and flow of actions, reactions and self-preservation of strategies that are deemed to work for one condition applied to another.

Well she talked about diet and um, yes, really it was diet really you know, just be careful what I eat.

And has that been a big change?

No not really ‘cos I was already on quite a healthy diet . . . You know, before that I’d um, I’d decided to lose a bit of weight if I could because with having MS it’s better not to carry a lot of weight with it.

ID 100a (diabetes, multiple sclerosis, underactive thyroid and high cholesterol, female, age 66 years)

The dynamic prioritisation of conditions

Shifting prioritisation between conditions was key to the way management was framed and the process of prioritisation over time required renegotiation of available resources. Although for a few participants there was little description of change where one condition remained routinely prioritised, the majority of descriptions were of changing prioritisation of conditions.

But it’s hard (to lose weight) I don’t know. So I put it down me personally to the tablets, whether that’s putting weight on I don’t know . . . just watching what I eat now you know with diabetes really more than anything like . . . But this breathing, this COPD that’s what will affect me in another few weeks when it starts getting cold; it’s the cold that affects me.

ID 600s (diabetes, COPD, high blood pressure, high cholesterol, male, age 65 years)

Interactions with health professionals

Most respondents described a dynamic and variable process of self-management with little separation between conditions. Respondents’ accounts of management practices were viewed across conditions with links in practices and tensions that required active choices between options. This suggests a difference between health professional accounts and patient approaches to multimorbidity, given that participants talked about seeking information and management strategies across conditions and were given little or no information.

The degree of congruence or conflict between views of health professionals and patients regarding management priorities impacted on subsequent management practices and therapeutic relationships. For participants who described similar management goals to their health professionals, contact reinforced and maintained priorities.

What’s kind of, your main priority at the moment with [GP]? . . .

Probably blood pressure, probably and cholesterol . . . so I’m more worried about those because they are more serious things. IBS didn’t kill anybody, you know, but blood pressure is serious and cholesterol is serious so IBS has gone into the background, you know.

ID 110a (IBS and high blood pressure, female, age 57 years)

However, for participants whose illness management priorities conflicted with health professionals, established relationships were challenged. One participant with diabetes and chronic depression described being ‘upset’ and ‘angry’ when her GP told her that she ‘just had to have willpower’ to control her diet for diabetes. However, the respondent’s priority was depression, which she recognised affected her diabetes control, and it was something she had actively tried to change over a number of years but was a complex problem. When she tried to obtain information about managing both conditions ‘they couldn’t offer any help’. The GP’s response conflicted with her priorities and had a negative impact on what she would engage with in managing her health.

I mean, I’ve got to say, when the doctor, when the GP said it to me, I actually came home and felt, er, why should I stick to this diet, because I was depressed I felt, I’ve nothing to live for anyway, the diabetes will kill me anyway.

ID 300s (diabetes, chronic depression, female, age 50 years)

Tipping points

To manage the complexity of multiple conditions, most participants reacted to changes in a process of prioritisation and reprioritisation over time. This involved an interaction of factors such as timing between diagnoses, prior experiences, recommended self-management activities, bereavement, contact with health services and flare-up of conditions.

. . . there’s been a lot of things going on, because I started with epilepsy in 2000 and it’s like I was, I had like a cluster and then they settled down . . . and then out of the blue I started with another cluster, but I think it’s hormonal because that started up, I started with diabetes and blood pressure you know, it seemed just, and I had gone through a period of depression because of [losing] my dad.

ID 170a (diabetes, epilepsy, female, age 55 years)

Prioritisation and reprioritisation occurred around key transition points when the relative importance of different management practices shifted, having either negative or positive effects. These ‘tipping points’ particularly arose around issues of medication management, lifestyle changes and understanding of conditions.

Medication management

Participants’ accounts of medications were framed in one of two ways: as an external event that was ‘being done to them’ and over which they had little control or, conversely, as an internal process whereby they gained control of their illness management by establishing routines for taking tablets and personalising their medication management. For participants whose narratives emphasised taking control, the presence of multiple long-term conditions played a minor role in everyday life.

I am quite vigilant about taking my medication um, and if I’ve missed it I do know that I do start feeling tired um, so I am quite compliant with me medication, but I have to take my epilepsy medication anyway so yeah, I take me medication morning and take my medication at night, like I say I am quite compliant about that.

ID 170a (diabetes and epilepsy, female, age 55 years)

However, other accounts of medicines portray a sense of an ever-increasing burden of tablets and resistance to further tablets. These accounts depicted medicines as objects and a process which they are subject to but have little direction over.

. . . for different ailments and that like and I’ve got bags of tablets there and now I’ve got that many tablets and I don’t know you know, swapping and changing tablets, getting tablets from the hospital and I don’t know where I’m up to with them like, you know.

ID 600s (diabetes, high blood pressure, high cholesterol, male, age 65 years)

Medications were also a point through which change could be negotiated. Some respondents described seeking additional sources of information and verification around medicines to those provided by GPs or specialists, as they did not feel confident that the latter were aware of all medicines being taken or their interactions. The main sources of verification were from the internet, pharmacist and others with similar conditions. These acted as legitimation points through which participants could change their routines as they were able to engage in focused discussions with health professionals to take control and direct illness management. Narratives depicted a change to becoming active self-managers and taking control by seeking out enough information to enable them to make decisions about medications.

I asked the pharmacy yesterday, I rang them up, ‘cos I’m taking so many different drugs, you’ve gotta be careful. And they were saying, I was saying could you take them at the same time, ‘cos you’re supposed to take these with food but you can’t take ‘em all with food . . . there’s one or two more, but . . . you’ve gotta keep a track.

ID 0700 (diabetes, rheumatoid arthritis, high blood pressure, male, age 82 years)

I take a lot of tablets now, every time I go and see him, ‘Don’t give me another tablet.’ . . . He tries me on different things and what you know doesn’t agree with me, he doesn’t give me . . ., he knows that, you know what I mean. Same as painkillers, I won’t take painkillers, I only take paracetamol.

ID 120a (COPD and IBS, female, age 57 years)

Information use in patient self-management

Existing knowledge and direct experience of conditions from other sources played a role in illness prioritisation and interpretation of information. Information was often viewed as contradictory or too general with insufficient information about interactions across conditions. Participants described integrating information from a number of sources. This led respondents either to feel confused or to actively monitor the suitability of information provided by practitioners. Information and advice had to be reinterpreted and verified for it to be useable and allow development of pragmatic routines.

Again, if you start to research it you come across dietary and nutrition, so that one kills all the sugars off and that kills all the fat off and the two of them just clash all the way down . . . You end up with you would be eating absolutely nothing except fruit and the odd slice of bread.

ID 150a (diabetes, high cholesterol, male, age 52 years)

Exercise as part of self-management

Exercise was considered important but also unrealistic because of functional limitations. When conditions placed a physical limitation on an individual, this was often talked about as having the greatest impact on their lives. It caused a tension between which self-management activities they should prioritise and restricted daily activities. Restrictions could accumulate and produce further problems, which influenced what could be attempted and the prioritisation of management practices.

Breathing. Um, only that I can’t do the things I used to, but whether that’s breathing or whether that’s old age I’m not sure. In truth, I suppose I used to walk up to [town centre], now I get a taxi . . . Or somebody takes me, but 70% of that is because of my feet, maybe 30% [because of COPD]. If my feet were fine, I’d still walk up to [town].

ID110s (COPD, oesophageal problems and feet problems, male, age 69 years)

Food and diet as part of self-management support

Dietary information was described as conflicting by many respondents and was not necessarily translatable across conditions, resulting in the accommodation of new practices within pre-existing regimes. Advice was often considered to be too generic and insufficiently individualised. Health-care professionals were not seen as always addressing the impact of other conditions and respondents were left to navigate through conflicting information to find the most appropriate course of action or seek further information.

Well she [practice nurse] gave me, um, information on diet which you could read in any woman’s magazine, there was nothing specific. I did say to her that my problem was binge eating and she had nothing to offer at all. She could not comment you know she knew nothing about it, so she wasn’t able to help. So as I say it was a complete and utter waste of time.

ID 300s (diabetes and chronic depression, female, age 50 years)

Social networks and self-management support

Interviews conducted as part of the longitudinal study with patients also revealed something of the wider relationships influencing self-management support. Analysis reinforced evidence that who is in the network, and the types of relationships that are present, influence how management practices are framed and the extent of engagement. 79 Resources available to an individual through the network can support or undermine engagement, and these change over time. Networks included family, friends, GP, nurses and companion animals. The amalgamation of different relationships that constituted the social networks were characterised by three network typologies. The three types of networks are characterised by differing combinations of features that influence priorities for health, the various roles that network members occupy and the degree to which the network might support and facilitate the normalisation of illness management into everyday life. Networks can be categorised as a family-focused network, a friend-focused network or a health-care professional-focused network.

In the family-focused network, the main source of support and information is provided by multiple family members with various roles in supporting the individual. There are usually multiple ties between network members. For networks that can be defined as family-focused networks, health-care professionals are viewed as important but are not so regularly seen. In a friend-focused network, there is greater variance in the types of relationships present; friends appear to have a more significant role for the individual than members of either the family-focused network or the health-care professional-focused network. For younger participants with a friend-focused network, parents and partners are also important, yet their role is subject to change (e.g. becoming more restricted because of specific events which have altered their relationship). For the participants whose social networks are characterised as friend focused and who are older, family members are less relevant. For these first two types of networks, management is normalised into existing practices and routines through negotiation and discussion with other network members, but, despite multiple conditions, participants in this study identified only a single GP or nurse as significant. The networks described in these first two categories frame health management within existing dimensions of everyday life that are not separate but integrated as routines and practices in daily life. The friend-focused network seems to emerge when there is an absence or loss of adequate family support and alternative lay support is sought instead.

The third type of network, the health-care professional-focused network, represents a network in which health is considered to be something managed more separately from the wider social network. Health and its management are framed as private, which can be a burden to others, and so seeking support is avoided, or limited as much as possible.

Setting

The setting for this study was the city of Salford, in north-west England, with an estimated population of 228,992 in 2010.

Sampling frame

The sampling frame comprised all patients registered with the 52 general practices in Salford PCT.

Data collection and coding

The Salford Integrated Record (SIR) is a local patient-sharing record system that integrates primary care, community care and secondary care information into one continuous electronic health record per patient. An anonymised data set is made available for research through NorthWest EHealth. Patients may opt out of the SIR; however, the proportion who do so is relatively small.

We obtained electronic clinical records from the SIR for January 2003 to December 2009. The data set comprises anonymised patient identifiers, anonymised general practice identifiers and the set of Read Codes used to code clinical consultations. In total, the data set contained > 136 million Read Codes derived from 34,200 distinct codes. Two gastroenterologists independently identified all Read Codes potentially relating to gastrointestinal disorders or IBS.

Ethics permission for this study was granted by the NorthWest EHealth Board.

Analysis

The analysis was conducted in two parts. First, we conducted analyses of patients with gastrointestinal symptoms. Second, we refined this analysis to consider only those patients with IBS.

Gastrointestinal symptom space

First, we selected patients with evidence of at least one Read Code for gastrointestinal disorders in 2009, which generated 19,869 patient records. The records for these patients were then searched for those with at least one diagnosis of IBS (Read Code J521) at any time since 2003. For each of these patients we collected together all the Read Codes from their records in general practice in 2009. Each patient was therefore represented by their set of Read Codes from 2009. To investigate the importance of gastrointestinal symptoms we then generated a second data set, which included only the Read Codes for gastrointestinal symptoms for these patients.

These data sets were then mapped onto a two-dimensional plane using a methodology that calculates the Resnik similarity between the concepts. It then maps these data from similarity space to the appropriate vector space and uses principal components analysis (PCA) to represent the data on a plane.

Each patient was therefore represented as a point on the plane and patients with similar symptoms occupied similar positions on the plot. These plots were then presented according to particular sets of Read Codes contained within the records.

The total data set available for analysis in diagnosis space was too large to make an analysis computationally tractable. We therefore examined 16,584 patient records from the three GP surgeries with the highest number of patient records with a diagnosis of IBS. All diagnosis codes for all these patients were collected and projected into diagnosis space using the same computational strategies discussed above.

Representation of patient records in gastrointestinal symptom space

Patient records were represented in the space of gastrointestinal symptoms. Figure 7 shows the results obtained by taking just the gastrointestinal symptoms from the 2009 Salford data set, and performing the semantic similarity analysis and PCA for dimensionality reduction.

FIGURE 7.

Principal component analysis plot of gastrointestinal symptom data. Each dot represents a particular set of patient data from 2009. These were then coloured according to the types of code found within each set of records, as indicated in the figure legend. GIT, gastrointestinal tract.

The gastrointestinal symptoms recorded by GPs were separated on the basis of two predominant symptoms, diarrhoea and gastrointestinal tract (GIT) pain. Patient records marked in light blue do not have any record of diarrhoea or GIT pain. The region highlighted in light green represents the patient records that are positive in recording of diarrhoea symptoms and negative GIT pain. Therefore, the largest source of variation within the data set concerns diarrhoea and GIT pain. Furthermore, patients with particular sets of symptoms occupy particular regions of gastrointestinal diagnosis space.

We considered next how other symptoms are distributed in symptom space. In Figure 8 we present several common gastrointestinal problems in this space: heartburn, flatulence, abdominal distension and indigestion. These symptoms are scattered across most of the diagnosis space without being clustered in any particular area.

FIGURE 8.

Common gastrointestinal problems mapped into symptom space. Shown in green are patients who have been given a diagnosis of IBS. (a) Heartburn; (b) flatulence/wind; (c) abdominal distension symptom; and (d) indigestion.

Representation of irritable bowel syndrome patient records in gastrointestinal symptom space

Figures 7 and 8 provided some insights into symptom space and the ways in which the patients are distributed. Next, we examined whether or not there was any clustering of patients diagnosed with IBS in this space by looking for the IBS diagnosis code in all patient records from 2003 to 2009 and highlighted whether or not the IBS code was present. These records were then mapped onto gastrointestinal symptom space from 2009, as previously calculated (Figure 9).

FIGURE 9.

A representation of patients in gastrointestinal symptom space.

The IBS patients were scattered across all of symptom space. There was no evidence that IBS patients cluster in any part of this space.

Irritable bowel syndrome patients in diagnosis space

Figure 10 shows the results of the PCA applied to all the diagnosis data from all the patients from the three general practices in Salford that had the most patients with an IBS diagnosis. Patient records were classified in different regions of vector space in terms of three different groups of diseases: ‘nervous system and sense organ diseases’; ‘genitourinary system diseases’; and ‘infectious and parasitic diseases’. The light-green area belongs to the patients who were recorded to have ‘infectious and parasitic diseases’ but not ‘nervous system and sense organ diseases’, or ‘genitourinary system diseases’. The light-blue area encompasses those patients who have none of the diseases. The area in black represents the patients who were recorded to have ‘nervous system and sense organ diseases’ but not ‘genitourinary system diseases’. The dark-green region contains the patients who were recorded as having ‘genitourinary system diseases’ but not ‘nervous system and sense organ diseases’. Finally, the dark-blue area belongs to patients who were recorded as having both ‘nervous system and sense organ diseases’ and ‘genitourinary system diseases’. The various diagnoses do cluster very effectively in the plot.

FIGURE 10.

Classification of patient records in terms of three different criteria (F%, K%, A%). F% refers to the patients who have any Read Code describing ‘nervous system and sense organ diseases’ in their records; K% presents the patients who have any Read Code describing ‘genitourinary system diseases’ in their records; and A% indicates patients who has any Read Code encoding the ‘infectious and parasitic diseases’ in their records.

Figure 11 Shows this same data, but this time with patients labelled either by the presence of a gastrointestinal symptom code in their data (see Figure 11a), or by a diagnosis of IBS (see Figure 11b).

FIGURE 11.

A representation of all diagnosis information collected in 2009 for patients from three GP surgeries in Salford. (a) any patients showing gastrointestinal symptoms are shown in green and (b) patients with a diagnosis of IBS are shown in green.

Comparing Figures 10 and 11 provides an insight into the areas of diagnosis space in which patients with gastrointestinal symptoms and a diagnosis of IBS fall. It is clear that gastrointestinal symptoms are found in patients across all of this space. The picture for patients with IBS is more informative. There is a cluster of IBS patients in the area associated with infectious disease, another in the area associated with nervous system and sense organ disease and a third area clustering in genitourinary system disease.

Design

The risk assessment study was a 12-month prospective case–cohort study in which patients consulting their GP with newly identified (incident) or existing (prevalent) IBS were recruited prospectively in order to validate the risk assessment questionnaire.

Setting

Initially we approached those practices that had a service-level agreement with the Greater Manchester Comprehensive Local Research Network. We then took a staged approach to roll the study out to general practices within Greater Manchester.

Study population

Patients consulting their GP for existing (prevalent) or new episodes (incident) of IBS.

Sample size calculation

The sample size calculation was based on separate samples of newly diagnosed and prevalent (i.e. existing) patients. Assuming an intracluster correlation coefficient of 0.05, 320 patients would be required to provide a 95% CI of ± 10 points or less for true sensitivity of ≥ 70%. The CI for an equivalent level of specificity is considerably smaller (± 6%). Assuming a dropout rate of approximately 25%, 420 newly diagnosed and 420 prevalent (i.e. existing) participants would be required.

Recruitment method: via consultation

General practitioners were asked to invite potential participants to take part in the study during their consultation for IBS and forwarded a permission to contact slip for those participants who expressed an interest in taking part to the study team.

During the consultation, GPs were asked to recruit patients with new or prevalent cases using the following definitions.

1. New diagnoses of functional lower gastrointestinal symptoms were defined clinically as a patient reporting symptoms for at least 3 months but less than 1 year by using Read Codes for:

   1. IBS

   2. bloating

   3. AP

   4. diarrhoea

   5. constipation

   6. change in bowel habit.

2. Prevalent episodes were defined clinically as patients with an onset of functional lower gastrointestinal symptoms/IBS more than 1 year ago for:

   1. IBS

   2. bloating

   3. AP

   4. diarrhoea

   5. constipation

   6. change in bowel habit.

In addition, patients had to meet the following inclusion criteria:

• adults aged > 18 years

• must have consulted the health-care team in the past 3 months for functional lower gastrointestinal symptoms/IBS as defined at (1) and (2) above.

Recruitment method: Read Code searches

Owing to the difficulties recruiting patients via consultation, GPs were asked to conduct a Read Code search to identify participants. Initially we focused on those practices that had not referred any patients and assistance from the Primary Care Research Network (PCRN) was offered. GPs were asked to screen participants to ensure their suitability for the study.

For the Read Code searches in practices, the inclusion criteria differed slightly from those for GPs who recruited patients via the consultation. Practices were asked to recruit patients on the basis of the IBS Read Code, by identifying those who had received this code within the last 10 years, but patients must have consulted their GP within the past 3 months for functional lower gastrointestinal symptoms (e.g. bloating, AP, diarrhoea, constipation or change in bowel habit). By restricting the search to those who had received the IBS label within the last 10 years and who had consulted recently for lower functional gastrointestinal symptoms, it was felt that this would produce a more manageable number of patients and, therefore, reduce the burden on practices.

Potential participants were sent a study pack containing an invitation letter (either from the GP or the research team), study information sheet, questionnaire and consent form. Those participants who wished to take part in the study were asked to complete the consent form and return it with the questionnaire to the research team. Those who did not wish to participate were asked to return a blank copy of the questionnaire. One reminder was sent to non-responders after approximately 2 weeks. Participants were followed up at 3, 6 and 12 months. One reminder was sent to non-responders after approximately 2 weeks.

Practice questionnaire

Practices that agreed to take part in study 2 were asked to complete a short questionnaire about the characteristics of their practice. This included details on the number of male and female partners, assistants, retainers, salaried GPs and practice nurses working full or part time; numbers of patients on the practice list; and the number of sites that the practice operated from. Practices were also asked to describe the type of location of their practice (rural, semi-rural, suburban, town/city or inner city), as well as the type of practice population (deprived, mixed – poor, average, mixed – well off or affluent). Finally, they were asked about their contract: General Medical Services, Personal Medical Services, Alternative Provider Medical Services or other; and whether or not the practice was a teaching/training practice.

Patient questionnaire

This consisted of the risk assessment questionnaire, the General Health Questionnaire (GHQ), IBS Symptom Severity Scale, Irritable Bowel Syndrome Quality of Life Questionnaire (IBS-QOL) and the EuroQol-5 Dimensions, three-level version (EQ-5D-3L), as well as patient demographics (see Appendix 4).

The risk assessment questionnaire consisted of the following questions:

• When your symptoms occurred how long did they last? (≤ 2 hours or less, > 2 hour)

• When your symptoms occurred had you changed your diet? (yes, no)

• Did you (i) have gastroenteritis or food poisoning (yes, no) or (ii) have contact with anyone with diarrhoea and vomiting? (yes, no)

• Have your activities (e.g. work or social activities) been interrupted in the last year because of problems with your bowels? (yes, no)

Participants were asked to answer the questions thinking about the symptoms they consulted their GP with most recently.

Participants were also asked to complete the 12-item GHQ. 91 The GHQ is a self-administered questionnaire aimed at detecting psychological distress. It consists of 12 questions in relation to symptoms or behaviours, which participants are asked to score from four categories ranging from ‘not at all’ to ‘much more than usual’. Scores are then summed for each of the 12 items to produce a total score out of 12, with higher scores representing increasing levels of distress.

The scoring for the risk assessment tool was based on participants’ responses to the above questions (see Appendix 5). 82 In addition, those who scored ≥ 3 on the GHQ were scored an additional point on the risk assessment tool. Thus, participants were given a total score (0–5).

The IBS Symptom Severity Scale92 is a four-item questionnaire. Participants are asked to score on individual visual analogue scales the severity of their AP and distension, satisfaction with bowel habit and interference with life in general. They are also asked how many days out of the last 10 days they had AP. They are given a total score up to a maximum of 500. Those with a score of < 75 are considered healthy patients, 75–174 indicates mild IBS, 175–299 moderate IBS and ≥ 300 severe IBS.

The IBS-QOL93 is a 34-item questionnaire that asks participants to provide responses to statements about their bowel problems over the last 30 days. Responses range from ‘not at all’ to ‘extremely’. The IBS-QOL produces an overall score and eight subscales, including dysphoria, interference with activity, body image, health worry, food avoidance, social reaction, sexual and relationships. Scores are transformed to a 0–100 scale ranging from poor quality of life (0) to maximum quality of life (100).

The EQ-5D-3L58,59 is a self-administered standardised measure of health status. It consists of five dimensions: mobility, self-care, usual activities, pain/discomfort, anxiety/depression. Participants are asked to rate each of these dimensions as having no problems, some problems, extreme problems. Participants are also asked to indicate their own health state today by completing a visual analogue scale from worst (0) to best imaginable health (100). 59

The questionnaire also contained questions in relation to age, gender, postcode (as a measure for socioeconomic status), ethnicity, occupation and employment status. Follow-up questionnaires at 3, 6 and 12 months consisted of the IBS Symptom Severity Scale, GHQ, IBS-QOL and EQ-5D-3L.

Data cleaning and validation

Questionnaires were entered into the study database; data were double entered and checked. This process did not reveal any systematic data entry discrepancies. Frequencies and cross-tabulations were run to check for out-of-range values and data inconsistencies.

Responders versus non-responders

Where possible, anonymised data were collected on non-responders, to provide comparison between participants and non-respondents. These data included age, gender and postcode sector. IMD scores were derived using GeoConvert (UK Data Service Census Support, Colchester, UK),94 which provides an IMD score for individual postcodes. For those non-responders identified via Read Code searches, we had information only on postcode sector rather than full postcode. To ensure comparability between responders and non-responders, we used postcode sector for both groups. For each postcode sector GeoConvert was used to obtain a list of postcodes within that postcode sector and assigned an IMD score based on the weighted mean of the postcodes within that postcode sector.

Referrals and response rates

Referrals and response rates were reported for those recruited via the consultation and Read Code searches.

Sensitivity and specificity

Receiver operating characteristic curve analysis was used to examine the ability of the risk assessment questionnaire to predict patients classified as having severe IBS according to the irritable bowel severity scoring system 3, 6 and 12 months later. Sensitivity and specificity were calculated for a range of cut-off points on the questionnaire, from which it was determined whether or not a cut-off point existed that yielded adequately high levels of both sensitivity and specificity. Positive predictive values were used to assess the extent to which those patients rated as high risk by the risk assessment tool went on to have high symptom distress.

Comparison of participants at baseline, 3 and 6 months

Symptom distress, IBS quality of life and EQ-5D scores were presented for all participants at baseline (unmatched), as well as for those who completed the questionnaire at baseline and at 3 and 6 months (matched). We do not report changes at 12 months because responses were not collected from all participants.

Referrals

A total of 58 individual practices were recruited. Table 16 gives further details of practices recruited to the study.

The majority of practices recruited described their location as suburban (48%) or town/city (35%). The population type was described as average by 43% of practices, deprived or mixed (poor) by 33% and mixed (well off) or affluent by 24% of the practices. The majority of practices worked under a General Medical Services contract (62%) and were training practices (69%). The mean list size was 7166 patients, ranging from 1344 to 16,583 patients. The median number of full- and part-time GP partners was two and the median number of full- and part-time nurses was zero and one, respectively.

Of the 58 practices recruited to the study, 34 referred patients to the study either via permission to contact slips or through the use of Read Code searches. In total, there were 606 referrals to the study, of which just over half (n = 325, 54%) were referred via practices using permission to contact slips (Table 17). The remaining 281 (46%) referrals were made through the use of Read Code searches at the practice. The mean number of referrals from all practices was 10 and ranged from zero referrals to 207 referrals from one practice.

Response rates

Of those referred to the study, 257 patients from 30 different practices agreed to take part (the overall response rate was 42%). The response rate was higher among those patients who were identified via the consultation (47%) than among those who were identified via Read Code searches (37%). For those practices that recruited patients to the study, the mean number of patients recruited was nine, the mode was one and the number of patients recruited ranged from 1 to 94 from one practice.

Completed responses at 3, 6 and 12 months were received from 188 (73%), 161 (63%) and 73 (28%) participants, respectively. Owing to time constraints, complete 12-month data were not available for all participants; of those who posted a 12-month questionnaire, the response rate was 60% (73/121). Of those who returned the initial questionnaire, 54 (21%) completed the initial questionnaire only, 38 (15%) completed the initial and 3-month questionnaires, 82 (32%) completed the initial, 3- and 6-month questionnaires and 65 (25%) completed the initial, 3-, 6- and 12-month questionnaires. A further 18 (7%) participants completed the initial questionnaire but missed one or more follow-up questionnaires.

Table 18 shows the patient characteristics for those recruited to the study. The majority of participants were female (80%) and white (88%). The mean age was 41 years and the mean IMD score was 23.65.

Responders versus non-responders

Demographic data were available for 81% of non-responders (281/345). Males were less likely to respond (56%) than females (52%), and non-responders tended to be slightly younger (39.8 years) than responders (41.3 years). These differences were not statistically significant. There was also no significant difference in the mean IMD scores for responders and non-responders based on the postcode sector analysis (24.9 vs. 26.5, respectively).

Sensitivity, specificity and positive predictive values

Table 19 shows the number and percentage of participants by symptom severity and risk assessment score at baseline, and at 3, 6 and 12 months. At baseline, one-quarter of participants were classified as having severe symptom distress; this proportion fell to about 15% at each of the follow-up periods. The majority of participants scored 4 or 5 on the risk assessment tool at each time point.

Table 20 shows the results from the receiver operating characteristic curve analysis. A score of ≥ 4 on the risk assessment tool yielded the best cut-off point to predict severe symptom severity at follow-up; however, the number of participants included in the analysis was much lower than that proposed in the original sample size calculation and included both those with prevalent and incident IBS. At 3 and 6 months, sensitivity (the probability that the score on the risk assessment tool was high when symptom severity was severe) was > 80%, but specificity (the probability that the score on the risk assessment tool was low when symptom severity was not severe) was low (36% and 42% at 3 and 6 months, respectively). Positive predictive values (the probability that symptom severity was severe when the score on the risk assessment tool was high) was poor (approximately 20%) at all three time points.

Comparison of participants at baseline and at 3 and 6 months

Of those who completed the baseline questionnaire, 147 (57%) participants also went on to complete questionnaires at 3 and 6 months.

Figure 12 shows the IBS-QOL subscales and overall score for those participants who completed the baseline questionnaire and did not have matched data for 3 and 6 months’ follow-up (baseline unmatched) and those with matched data for baseline, 3 and 6 months. Those with unmatched data tended to have lower scores on all subscales and on the overall IBS-QOL scale. At all time points the lowest scores were reported on the food avoidance subscale and were highest on the relationship subscale. The data also suggest that there were small improvements in most HRQoL subscales over time.

FIGURE 12.

Irritable Bowel Syndrome Quality of Life Questionnaire subscales. Higher values indicate better HRQoL.

Likewise, Figure 13 shows the data for participants according to their symptom severity classification. The proportion of patients within each category was similar for both matched and unmatched participants at baseline, with the majority being classified as having moderate disease. The proportion of those with moderate and severe disease fell at 3 and 6 months, as more participants reported disease remission or mild disease.

FIGURE 13.

Symptom severity classification.

Figure 14 shows the proportion of patients reporting problems (some or extreme problems) on the individual EQ-5D scales. The majority of patients reported no problems on the mobility and self-care scales, about one-third reported problems with their usual activities and a much larger proportion reported problems with pain and/or discomfort or with anxiety and/or depression. The chart suggests that there was a small improvement in the usual activities, pain/discomfort and anxiety/depression scales at 3 months’ follow-up, but this was less obvious at 6 months’ follow-up.

FIGURE 14.

Per cent reporting problems on EQ-5D scales.

Summary of findings

The number of patients recruited to the study was much lower than anticipated. However, we were able to calculate sensitivity and specificity based on the sample recruited. The risk assessment tool appeared to be sensitive in predicting those participants with severe disease; however, it was not specific in predicting those without severe disease. Patients reported reduced quality of life; however, the data suggested that there were modest improvements over time.

Strengths and limitations of the study

The main strength of this study was the large number of practices that agreed to take part. Practices were recruited from across Greater Manchester and, therefore, covered a wide geographical area. However, despite the large number of participating practices, the number of referrals to the study was disappointing. Initially, GPs were asked to recruit patients during the consultation and we therefore relied on GPs to remember about the study during the consultation. For a small number of practices this method worked well, but the majority of practices did not refer any patients via this method. This recruitment strategy was later supplemented with Read Code searches at a number of practices. As a result, recruitment to the study was boosted to some extent, but not all practices agreed to carry out a Read Code search despite the offer of assistance from the local PCRN. An inadvertent and unexplained limitation of the current study was selection bias, as the majority of referrals came from a single practice, which is likely to affect the validity of the results. This practice was a particularly large one, with a large student population.

Why is the prevalence of irritable bowel syndrome so low?

Even with the use of Read Code searches, the prevalence of IBS was much lower than expected when compared with the literature. Prevalence estimates of IBS in Western countries range from 2% to 22%. 95 More specifically in the UK, a questionnaire survey estimated IBS prevalence to be 10.5%. 90 However, estimates of the prevalence of IBS are likely to be influenced by a number of factors, including the criteria used, population studied, mode of study delivery and type of prevalence estimate.

Prevalence estimates reported in the literature vary widely depending on the diagnostic criteria (e.g. Kruis,96 Manning97 or Rome I,98 II99 or III4). A study across eight European countries with > 40,000 participants reported that the overall prevalence of IBS was 11.5%; however, only 2.9%, 4.2% and 6.5% of participants met the Rome II, Rome I or Manning criteria, respectively. 100 The usefulness of diagnostic criteria is frequently debated, and they tend to be of little relevance within primary care, as few GPs use them to make a diagnosis. We adopted a symptom-based approach to identify patients consulting with IBS within primary care and, therefore, we would have expected the prevalence of IBS to be higher than in those studies using formal diagnostic criteria such as Rome II. 99 However, there are a number of further possible explanations.

First, many studies used to estimate the prevalence of IBS have been self-reported questionnaire surveys and are therefore likely to overestimate the proportion of patients who have been formally diagnosed with IBS. Nevertheless, Wilson et al. 90 reported that over half of patients with IBS consulted their GP within the past 6 months. In addition, Rey and Talley95 suggest that consultation rates are higher in more recent studies owing to increasing public awareness. They suggest that 30–70% of patients have consulted their GP owing to their symptoms in the previous year. Furthermore, a survey of UK GPs also suggests that IBS is not an insignificant problem. 3 Based on the Read Code searches, we were able to calculate an estimate of IBS prevalence in a 3-month period. This was estimated at 0.2%, and it ranged from 0.03% to 0.4% across practices. However, patients recruited via the Read Code searches had to have had the IBS Read Code recorded in the last 10 years and to have consulted within the past 3 months for symptoms of abdominal pain, bloating, constipation, diarrhoea or change in bowel habit. Despite this, the rate was much lower than expected.

Second, IBS or gastrointestinal symptoms are not a part of the QOF, a scheme that incentivises practices and rewards them according to how well they care for patients. Consequently, they are not likely to be a priority for GPs.

Third, GPs may be reluctant to code patients for IBS or lower gastrointestinal symptoms. One reason for this may be the uncertainty of the diagnosis. Although IBS is no longer regarded as a diagnosis of exclusion, and NICE has produced guidelines,81 GPs may still take a ‘diagnosis of exclusion’ approach. As a result, GPs may not use Read Codes when they first consult and use other Read Codes or free-text fields. Consequently, if patients disengage with primary care they would not receive an IBS label.

The predictive value of the risk assessment tool

We were able to calculate sensitivity, specificity and positive predictive values for the risk assessment tool based on the data collected. Based on a cut-off point of 4 on the risk assessment tool, it was possible to obtain sensitivity values of > 80%. However, specificity was around 40% and the positive predictive value was only about 20%. Therefore, the tool appeared to be quite good at predicting those who did actually have severe IBS, but was not good at predicting those that did not actually have severe IBS. However, the small number of patients recruited to the study meant that the study was underpowered. The sample size calculation estimated that a sample size of 420 prevalent and 420 incident cases would be required in order to detect an effect. Early on it became apparent that we would not be able to recruit incident cases to the study because of the difficulties in establishing a diagnosis and the fact that many patients present with longstanding symptoms. Indeed, of the 257 patients recruited to the study, 95% reported having symptoms of > 3 months’ duration. As a result, the study was expanded to include prevalent cases; however, difficulties with recruitment persisted. Despite these difficulties, it is questionable whether or not such a tool would be useful in predicting which patients have severe disease at a later stage.

Natural history of irritable bowel syndrome

The data suggest that patients with IBS have poor HRQoL, as measured by the IBS-QOL and the EQ-5D. These findings are comparable with those reported in the literature. 101 Those patients who did not complete the questionnaire at all three time points tended to score lower on these measures at baseline. Data from participants who completed the baseline, and 3- and 6-month questionnaires tended to suggest that there were small improvements in the IBS quality of life, symptom severity scores and EQ-5D over time. Although we did not formally test these differences, this appears to contrast with previous findings of no significant difference in the IBS-QOL subscales and overall score from baseline to follow-up 3 months later. 101 No specific intervention was administered in that study. Similarly, in our study, no specific intervention was administered between baseline and 3 and 6 months. However, GPs were asked to recruit patients during the consultation and it may be that treatment prescribed during the course of that consultation resulted in improved outcomes for patients. Alternatively, these improvements may simply reflect the fluctuating nature of the condition.

Owing to the problems with recruitment, we felt that it was important to gather the views of GPs about the management of IBS in primary care. We therefore proceeded to the qualitative study to explore how GPs diagnose and manage patients with IBS (study 3). We also decided to investigate further the patient profiles of patients with gastrointestinal symptoms (IBS, AP and IBD) using data from the SIR (study 4).

Design

Face-to-face, in-depth, semistructured interviews were conducted with GPs in north-west England who were participating in the study of the risk assessment tool. Data were collected between March and December 2011.

Sampling

General practitioners from practices invited to take part in the validation of the risk assessment tool were invited to participate in a semistructured interview. Purposive sampling was used to maximise the variation of the sample in terms of age, length of experience, ethnicity, practice size and location. Recruitment was continued until theme saturation was reached.

Data collection

Interviews were conducted, with written consent, at GP premises by two of the researchers. Interviews lasted between 20 and 70 minutes (mean duration 42 minutes). The topic guide was developed from review of the existing literature and discussion within the research team. The topic guide, designed to be used flexibly, was modified in the light of emerging themes. The main areas explored were GPs’ views on the aetiology of IBS, how the diagnosis of IBS is made, how GPs explain IBS to patients, treatments offered and the potential usefulness of a tool to predict chronicity.

Interviews were fully transcribed. Data coding was by constant comparison across interviews by individual researchers and emerging themes were agreed through discussion among all authors from different professional backgrounds (academic primary care, health services research, epidemiology). An iterative approach to data collection and analysis was taken: coding and conceptual categories were constantly reviewed and refined in the light of new interview data and ongoing discussion in the research team. The topic guide was modified to allow for further exploration of emerging themes. NVivo 9 (QSR International, Warrington, UK) was used to store and manage the data.

Understanding irritable bowel syndrome

All GPs recognised the existence of IBS as a syndrome, a collection of symptoms, affecting different people in different ways.

I do think it exists, I really do. I think people’s lives can be made – it’s a label isn’t? People who have got constipation predominant IBS might – you could just call them constipated. It’s a real condition, people get a lot of pain from it. There are people that seem to get more anxious about it. But we all get a dodgy tummy if you’ve got an exam or something like that. So I definitely think it exists.

GP 11

General practitioners described IBS as a complex condition with a biopsychosocial aetiology, alluding to the notion of susceptibility, triggers and precipitants, with patients often presenting symptoms related to ‘stress’ and lifestyle.

I think there’s a lot of psychology with irritable bowel, not that necessarily the psychology causes the irritable bowel but I think for a fairly benign condition it can cause a lot more upset than you’d expect.

GP 9

General practitioners defined IBS by the lack of demonstrable organic pathology and focused on abnormal behaviour of the bowel.

I think it is an exclusion of other pathology, but it’s often a change in bowel habit, people become constipated, they get increased frequency of bowel motions, bloatedness, colicky abdominal pains, sometimes I think there’s a bit of reflux associated with that, with IBS.

GP 11

So they’d be likely to have, sort of, crampy abdominal pain with bloating, sensitive to bowel disturbance whether it be diarrhoea, constipation, erm, nausea, erm, those types of symptoms, but none of the serious red flag-type symptoms, not usually any sign of any bleeding from the back passage, no sign of weight loss, erm, and that type of pain, sort of, colicky pain.

GP 1

General practitioners were aware of the chronicity of IBS, with the potential for ‘flare-ups’.

. . . well irritable bowel syndrome often it just flares up, they come and they get their month of Mebeverine, they take it and then they’re not troubled for another year.

GP 14

Some GPs alluded to their own experiences, which made symptoms in patients more real and more understandable.

I think I’ve dealt with it so much, I’ve experienced it as well and so I think when you’ve experienced it you know how severe some of the symptoms can be.

GP 11

Summary of main findings

This study illustrates how the perspectives of GPs about the diagnosis and management of IBS influence their views on the value of a risk assessment tool to predict chronicity. GPs reported that IBS is not a difficult condition to diagnose or manage, yet most described reluctance to add the Read Code for IBS to the record. Respondents acknowledged the link between IBS and psychological distress, but were reluctant to refer for therapies such as CBT or hypnotherapy and did not see the value of a risk assessment tool to predict chronicity. Most GPs suggested that patients did not return because their symptoms had settled, rather than because they may be dissatisfied with care.

Strength and limitations of the study

Data are presented from interviews with GPs over a wide geographical area; this purposive sampling enabled us to access a range of views. However, because we employed a theoretical, rather than a statistical, sampling approach, the proportion of GPs holding different views cannot be inferred. Seventeen out of 19 GPs were already taking part in an evaluation of the risk assessment tool for IBS and were possibly more likely to feel comfortable managing IBS. The two GPs who agreed to be interviewed but not to participate in the main study might have been expected to have more negative views about IBS; however, this was not apparent. Data were analysed by researchers from different professional backgrounds, increasing trustworthiness. 107

Comparison with existing literature

The diagnosis of medically unexplained symptoms is often contentious;108 however, GPs in the current study did acknowledge the existence of IBS and reported that they had no difficulty in diagnosis or management. These findings are similar to published data reporting that compared with pelvic and back pain, IBS was not considered difficult in terms of distinguishing functional from organic disease. 109 When GPs experienced IBS themselves, or knew someone with IBS, it helped them to understand the condition, resonating with studies describing a similar finding in GPs who have first-hand experience of chronic fatigue syndrome. 110

Respondents recognised the link between IBS and psychological distress;109 however, they were uncertain of the benefit of psychological therapies in patients with IBS. It has been reported that GPs in the UK do not refer patients with IBS for psychological treatment,109 whereas doctors in the Netherlands are more likely to do so. 102

Although the diagnosis of IBS is often made by excluding red flag indicators, in one study 72% of GPs considered that they were usually, or often, able to diagnose IBS at the initial visit. 109 This contrasts with a previous study reporting that 19% of patients formally diagnosed with IBS had been given the diagnosis on their first visit and 56% after a further 1–5 visits. 100 These studies were conducted prior to the introduction of the NICE guideline81 for the diagnosis and management of IBS. However, although most GPs in our study were aware of this guideline, many were reluctant to apply a label of IBS to patients. Read Codes were reported as being applied only if the patient reconsults and probably reflecting the fluctuating nature of the condition. 111

General practitioners emphasised the importance of self-management in patients with IBS and felt that offering advice and reassurance was important in the first instance. Dietary advice was emphasised by respondents, but it is not always appreciated by patients, who are likely to have already tried such measures before consulting. 105

A strong physician–patient relationship and empowering explanations given by GPs are reported to be important to the successful management of IBS. 112,113 General practitioners often give the problem back to the patient102 and health-care professionals tend to distinguish between ‘good’ and ‘bad’ patients, with so-called ‘bad’ patients being unaccepting of the diagnosis of IBS, recurrent attenders, demanding further investigations, failing to cope or respond to treatment and resentful of the IBS label and psychological explanations given. 103 ‘Good patients’, on the other hand, were those who GPs suggested had a sense of relief with the label given and were accepting of the diagnosis. 103 In the current study, GPs suggested that most patients did not return, attributing this to patients either getting better or learning to live with their symptoms. However, there is evidence to suggest that patients with IBS disengage with services because of dissatisfaction with GP interactions,114 a belief that there is little the NHS can offer, or attributing the onset or worsening of symptoms to previous poor medical care. 104

Of those patients who consult their GP with a FGID, approximately 30% go on to develop chronic symptoms. 82 Providing a self-help guidebook, designed with the aid of patients, reduces primary care consultations by 60%. 15 In addition, for patients with intractable symptoms resistant to conventional medical therapy, there may be therapeutic benefit from CBT or hypnotherapy. 16,17 A risk assessment tool may aid GP decisions about the management of patients at risk of poorer outcomes and fast-track them to therapies that are known to be effective, such as CBT or hypnotherapy, leading to lower levels of distress and health-care utilisation in the long term. However, the majority of GPs interviewed did not feel that a risk assessment tool would be useful to them, whereas others felt that it would be impracticable to administer such a tool in a 10-minute consultation.

Implications for future research and clinical practice

Patients might be appreciative of such a tool within the consultation – it may demonstrate that the doctor is taking their symptoms seriously and demonstrate patient involvement in decision-making about treatment. There is a parallel with the use of the Patient Health Questionnaire-9, a tool to which GPs are less receptive than patients,115 and the value of a tool based on prognostic information to determine the appropriate intervention for patients with back pain has been shown to be effective116 but not well used by GPs. 117

This study highlights tensions between research evidence identifying the potential role of a risk assessment tool to support primary care management of IBS and GPs’ perspectives that it is not needed. The study reinforces the need to take account of current clinical practice and practitioners’ perspectives and researchers and clinicians should work together when developing research programmes to tackle conditions that create challenges for clinicians.

Setting

The setting for this study was Salford, north-west England, with an estimated population of 228,992 in 2010.

Sampling frame

The sampling frame comprised all patients registered with the 52 practices in Salford. Patients may opt out of the SIR; however, the proportion that do so is relatively small.

Data collection and coding

An application was made to NorthWest EHealth requesting details on a cohort of patients based on various codes for gastrointestinal disorders.

The SIR data were obtained for the period January 2002 to December 2011. Information was supplied in two separate files. The first file contained anonymised patient identifiers, gender and year of birth, whereas the second file contained journal entry identifier, anonymised patient identifier, date of journal entry, Read Code description and Read Code. These two files were matched on patient identifier. The file was checked for duplicates and frequencies were run on all variables to check for any anomalies. Duplicate records and patients under the age of 17 years were removed. Read Codes were then recoded into a new variable to distinguish between symptom and/or diagnostic codes, medication codes and referral codes. Ethics permission for this study was granted by the NorthWest EHealth Board (reference number 177) and the individual health-care organisations.

Symptoms pre and post irritable bowel syndrome, abdominal pain or inflammatory bowel disease

Table 27 shows the number of symptoms/diagnoses 1 year prior to and 1 year after the index episode of IBS, AP and IBD. The most commonly reported symptoms in the year prior to IBS were AP (19.5%), diarrhoea (6.1%) and bloating (3.8%). However, a large majority of patients (69.4%) did not have any gastrointestinal symptoms recorded in the year prior to their IBS. Likewise, in the year after the index episode of IBS, the majority of patients did not have any gastrointestinal symptoms recorded (79.1%). AP was the most commonly recorded symptom (13.4%), and 14% of patients had a further episode of IBS recorded in the year after their index episode. IBD showed a similar pattern to IBS with the majority of patients (69.6%) not having any symptoms recorded in the year prior to their IBD diagnosis. The most commonly reported symptoms were AP (15.9%) and diarrhoea (11.8%). In the year after the index episode of IBD, AP and diarrhoea were still the most commonly reported symptoms, but the proportions tended to be lower (10.4% and 5.4%, respectively). Almost one-third of patients had a further episode of IBD recorded in the year after their index episode.

Most patients with AP (92.5%) had no symptoms recorded in the year prior to their index episode of AP. In the year after their index episode, 23% of patients had a further AP episode and 2.8% of patients had an IBS diagnosis.

The proportion of patients who reported two or more symptoms in the previous year or the year after IBS, AP or IBD was < 5% in most cases.

Medications pre and post irritable bowel syndrome, abdominal pain or inflammatory bowel disease

Table 28 shows the number of NICE medications prescribed in the year prior to, and after, the index episode of IBS, AP or IBD. NICE medications for IBS were prescribed for 31% of patients in the year prior to their index episode of IBS and for 54% in the year after the index episode. The change was most marked for antispasmodics, prescriptions for which increased from 17% in the year prior to the index episode of IBS to 44% in the year after the index episode. IBS NICE medications were also commonly prescribed in patients with AP and IBD, with about one-third of patients being prescribed NICE medications in the year after the index episode of AP or IBD. Prescribing of antispasmodics increased markedly in those with AP, from 5% before the index episode to 19% after. Approximately 10% of patients with any of the three conditions were prescribed selective serotonin reuptake inhibitors (SSRIs) and about 5% tricyclic antidepressants.

Tables 29 and 30 show the number of NICE medications prescribed in the year prior to, and the year after, the index episodes of IBS, AP and IBD for those with and without symptoms, respectively. Where gastrointestinal symptoms were recorded in the year prior to, or the year after, the index episode, the proportion of patients prescribed NICE medications was much higher than in those without symptoms. However, there were still a number of patients with no symptoms who received NICE medications in the year prior to their index episode of IBS (22%), AP (18%) or IBD (20%). Among those with IBS, approximately 50% of those with no symptoms recorded in the year after their index episode were prescribed NICE medications, compared with approximately 30% of those with AP or IBD.

Referrals pre and post irritable bowel syndrome, abdominal pain or inflammatory bowel disease

About 4% of patients had a gastrointestinal referral in either the year prior to or the year after their index episode of IBS (Table 31). This was in contrast to patients with IBD, of whom 9% had a gastrointestinal referral in the year prior to their index episode of IBD and 22% had a gastrointestinal referral in the year after their index episode of IBD. Few patients with AP had a gastrointestinal referral in the year prior to their AP. The number of mental health and lifestyle referrals were low for all three conditions.

Prevalence estimates

Figures 15–17 show the age-specific rates per 100,000 population for IBS, AP and IBD, respectively, for those coded in the SIR data. IBS was much more prevalent in females than in males and peaked in those aged 25–29 years. In females, after this age there is a gradual decline in prevalence by age. In males, the prevalence remains stable until age 50–54 years, after which there is a staged decline in the prevalence rates.

FIGURE 15.

Irritable bowel syndrome age-specific rates: 2002–11.

FIGURE 16.

Abdominal pain age-specific rates: 2002–11.

FIGURE 17.

Inflammatory bowel disease age-specific rates: 2002–11.

The prevalence of AP is much higher than that of IBS or IBD, and higher in females than in males. In females, rates are highest in the youngest age groups up to the age of 30–34 years, after which rates decline until the age of 55–59 years, where the rates gradually increase again to the age of 65–69 years before declining in the oldest age groups. In males, there is a gradual increase in the age-specific rates up to about 60 years, when the rates level out, until they decline in the eldest age group. The rates of IBD are much more similar in males and females and tend to fluctuate across the age groups, with a decline in the older age groups.

Figure 18 shows the age-standardised rates per 100,000 population for IBS, AP and IBD in Salford for the years 2002–11. Rates have been standardised to the Greater Manchester population. The rates for IBD remained stable across the 10-year period for both males and females. IBS shows a similar pattern in males and females, with a slight increase in prevalence rates up to 2006 before levelling out over the second half of the 10-year period. The rates of AP, on the other hand, increased considerably in both males and females from 2002 to about 2009, after which they levelled out for both males and females.

FIGURE 18.

Age-standardised rates: 2002–11. F, female; M, male.

Strengths and limitations

The strengths and limitations of this study are similar to those reported for study 1. This study confirms that the recorded prevalence of IBS in patients who consult, and are recorded by, their GP is low. Several possible explanations for this are given in the Discussion section.

This study also found that the majority of patients who were coded as having IBS did not have gastrointestinal symptoms recorded in the year prior to, or the year after, their index episode. NICE recommends assessment for IBS in patients having any of the following for at least 6 months: AP or discomfort, bloating or change in bowel habit. Therefore, one might expect evidence of this in the recording of symptom codes in the year prior to, or after, the index episode of IBS. GPs in the qualitative study also alluded to the fact that patients may not receive a Read Code for IBS initially, but that they would use Read Codes for AP, diarrhoea or constipation. However, this was not reflected in the data from the SIR. Interestingly, however, IBS medications as recommended by NICE were often prescribed in those without symptoms in the year prior to their IBS. This suggests that the use of medication codes may be an alternative approach to identifying patients via symptom or diagnostic Read Codes. For patients to receive a prescription, GPs must enter a description of the medication, which is then issued electronically. However, some IBS medications (e.g. laxatives) may also be prescribed for other conditions.

The proportion of patients with IBS who were referred to a gastrointestinal specialist was relatively small compared with the proportion of patients with a diagnosis of IBD. Few patients in all three groups had been referred to mental health specialists or for lifestyle advice. This concurs with the findings from the qualitative study where GPs reported being fairly comfortable in managing patients with IBS.