Open access follow up for inflammatory bowel disease: pragmatic randomised trial and cost effectiveness study

Protocol

Comprehensive guidelines for the shared management of inflammatory bowel disease were distributed to all local general practitioners before the study started. These covered diagnosis, medical treatment of mild to severe disease, laboratory monitoring, the place of surgery, stoma care, follow up and surveillance, communication, documentation, and audit. For patients due for follow up by open access we transferred responsibility for care back to the general practitioner and stopped routine appointments at outpatient clinics. In return, we guaranteed rapid access to specialist care when necessary. The normal recall system continued for patients needing regular surveillance by colonoscopy because of the increased risk of colorectal cancer.

Patients were recruited by three consultant gastroenterologists, one staff doctor, two senior registrars, and four registrars from outpatient clinics at the two hospitals during October 1995 to November 1996. Patients aged over 18 with inactive or mildly active but stable inflammatory bowel disease were invited to take part. Those with active disease requiring treatment, a stoma, other disease which required regular follow up, or who were thought unable to comply with data collection were excluded. Three general practitioners declined to collaborate, and their patients were also excluded.

Patients randomised to routine follow up made their next appointment at the end of each hospital visit as usual. Our practice in both hospitals has been to see patients at short notice between appointments if requested by the patient or general practitioner. It was made clear to patients in the routine arm that this policy still applied. Those randomised to follow up through open access were asked to contact their general practitioner about problems or to contact the hospital directly if they were unable or unwilling to see the general practitioner first. Appointments were made by telephoning outpatient clerks or gastroenterology secretaries, who were made aware of the need to offer an early appointment.

Patients were reviewed in the outpatient clinic in the normal way. A relapse did not require withdrawal, but the patient was seen and treated as appropriate and remained in the same study group. All patients were called for review 24 months after entry into the study.

Primary outcome was measured by the generic and disease specific quality of life questionnaires SF-36 6 7 and UK inflammatory bowel disease questionnaire (UKIBDQ).8 The questionnaires were completed in clinic at recruitment and at the end of the study and by post at six monthly intervals between. Two reminders were sent to non-respondents. Those who failed to attend the end of study appointment after two reminders were sent the final questionnaire by post.

Resource use was estimated from patient questionnaires and case notes. Medical staff abstracted data from hospital notes, and practice staff abstracted them from general practitioner records. Questions on patient borne costs within the quality of life questionnaires covered travel, parking, time off work for the patient and any accompanying person, and other costs such as baby sitting. Patient and general practitioner satisfaction, preferences, and views were assessed by postal questionnaires at the end of the study, supplemented by semistructured interviews with a sample of general practitioners. To minimise bias these were undertaken by general practitioners during audit visits.

When the trial was designed no disease specific health related quality of life scale had been validated for use in the United Kingdom. As we found no previous study of inflammatory bowel disease which used SF-36 as an outcome, we estimated sample size using SF-36 scores from patients with suspected peptic ulcer,7 which showed a standard deviation of 20 for most subscales. We considered a difference of 10 points to be clinically important, equivalent to a standardised difference of 0.5. Hence, 170 patients (85 per group) would yield 90% power to detect a significant difference with a significance level of 0.05.9 A target of 180 was set to allow for loss to follow up. Limited information on the distribution of key resource variables meant that the sample size could not reflect likely differences in costs.10

Analysis was by intention to follow up. To counteract the effect of possible differences in baseline health related quality of life scores, we analysed changes in individual scores from baseline using t tests. Preference data were analysed by χ2 tests.

Because data on use of resources tend to be highly skewed, routine parametric statistics are not appropriate. We therefore assessed significance by the Mann-Whitney U test. As economic analysis is mainly concerned with a comparison of means, however, means and standard deviation are reported for each variable.

Valuation of hospital resources was based on estimates provided by the trusts. Costs of outpatient and general practitioner home visits were derived from Netten et al,11 drug costs from the British National Formulary,12 and costs of general practitioner surgery visits from Graham and McGregor.13 Patients' lost work time was valued by using average wages,14 and their motoring costs were estimated from Automobile Association figures.15 Total costs to society were derived by summing primary care, secondary care, and patient borne costs.

Progress of participants through trial

• Download figure
• Open in new tab
• Download powerpoint

Random allocation

To ensure balance in type of follow up, patients were first stratified by centre and between four diagnostic groups: ulcerative or idiopathic proctitis; ulcerative or indeterminate colitis affecting more than the rectum; Crohn's disease of the small or large bowel; and Crohn's disease of the small and large bowel. The computer generated allocation lists were securely held by one independent researcher in each centre. When the clinician had established the eligibility of the patient and received informed consent, the local researcher was contacted for the random assignment and the patient immediately informed of the follow up arrangements.

Participant flow and follow up

The figure shows the progress of the trial. No patients refused to participate, although five subsequently withdrew. Quality of life questionnaires were completed by 170 patients at six months (94%), 160 at 12 months (89%), 159 at 18 months (88%), and 164 at two years (91%). The number of patients who failed to complete the study differed significantly between the two hospitals (12 in Morriston v 4 in Neath; P<0.05). There was no significant difference between groups at baseline in age, sex, diagnostic group, or quality of life.

Quality of life

There was no significant change in mean health related quality of life scores in either group over the two years of the study, although there was some deterioration in both groups in most subscales. We found no significant differences between groups in changes in health related quality of life scores at 6, 12, 18, or 24 months compared with baseline (table 1).

Patients' preferences

Patients had a strong preference for open access follow up (103/164, P<0.01); 69/81(85%) in the open access follow up group preferred open access follow up, and 34/83 (41%) in the routine group would have preferred open access follow up. The main reason given for this preference was the appropriateness of attending only when ill. The reason most commonly given for keeping routine appointments was for reassurance. Some patients had difficulty arranging open access appointments, and a few would probably have been lost to follow up if they had not been called for the end of study visit.

General practitioners' preferences

Study patients were registered with 53 practices. Forty practices returned postal questionnaires relating to 155 patients, including 12 patients who did not complete the final patient questionnaire (86% response rate). Sixty nine general practitioners indicated their preferred method of follow up for 143 patients (including eight who did not complete the final patient questionnaire). The general practitioners preferred open access follow up for 108 patients (55 in open access follow up, 53 routine) and routine follow up for 35 patients (15 open access, 20 routine). This difference was highly significant even after potential correlation between multiple responses from individual practitioners was allowed for (P<0.001). Preference for open access follow up was associated with sensible patients, stable disease, and the effective booking of urgent review. Forty four general practitioners (64%) favoured a gastrointestinal nurse practitioner as point of contact; 10 were opposed to this, eight wanted further discussion of the role, and seven did not express a view.

Use of resources

Comprehensive data on resource use in both primary and secondary care was available for 155 patients. Table 2 shows use of hospital facilities. Open access patients had fewer day visits (P=0.019), fewer outpatient visits (P=0.002), and cost less in total investigations (P=0.032). There were no significant differences in numbers of inpatient days or specific investigations. Patient borne costs were lower for open access patients (P=0.002). Mean total cost for hospital care was significantly lower for open access patients than routine outpatients (£582 v £611, P=0.012).

Analysis of resource use in primary care showed no significant differences in general practitioner visits, patient borne costs, or drugs prescribed (table 3). Although more maintenance drugs (5-aminosalicylates) were prescribed in the open access follow up group, this did not reach significance. Primary care costs were higher for open access patients, but not significantly so. When primary and secondary care costs were considered together there was no difference in total costs to society between study groups (table 4).

Problems with open access

Despite the strong preference for open access follow up, some patients experienced difficulties in making urgent appointments, largely because of pressure on clinics and the inexperience of clerical staff in managing open access follow up. Letters from general practitioners were effective but took time. We believe that the best way to overcome this problem is to have a single telephone point of contact for patients that is staffed by a specialist gastrointestinal nurse practitioner. General practitioners were generally supportive of this proposal, and we intend to introduce and evaluate this approach. Chronic inflammatory disease is a well recognised risk factor for the development of gastrointestinal malignancy. 17 18 As well as managing open access, a nurse practitioner could ensure that patients are called back at appropriate intervals for assessment and colonoscopy if necessary.

There were no significant differences in patient characteristics between the two hospital sites. However, significantly more patients at Neath completed the study. The dedicated clinics and computerised clinical information system at Neath may have contributed to this.

As far as we are aware, this is the first randomised trial comparing open access and routine follow up for patients with inflammatory bowel disease, although Probert et al recommended such reorganisation in 1993.3 Their survey reported that most gastroenterologists in Britain cared for at least 100 patients with inflammatory bowel disease, and nearly a quarter of them for 200 or more. Thus the reduction in outpatient attendances which we have documented could save each consultant 25–50 visits a year.

Wider applicability

A study of shared care of patients with moderately severe asthma also found that it was equally effective as hospital care and produced cost savings in secondary care without a significant increase in primary care workload.19 Similarly, a randomised trial of patients with breast cancer showed that follow up of patients in remission by general practitioners was not associated with increased time to diagnosis of relapse, increased anxiety, or deterioration in health related quality of life. However, resource use and preferences were not evaluated.20

Although much has been written about shared care,19^–24 further studies are needed to evaluate whether our findings can be extrapolated to conditions such as arthritis, epilepsy, heart failure, and multiple sclerosis. Much would depend on the ability and willingness of general practitioners to shoulder the increased responsibility.