Alimentary Pharmacology & Therapeutics
Clinical trial: a randomized trial of early endoscopy, Helicobacter pylori testing and empirical therapy for the management of dyspepsia in primary care A. E. DUGGAN*,, C. A. ELLIOTT*, P. MILLER*, C. J. HAWKEY & R. F. A. LOGAN*,
*Division of Epidemiology and Public Health, University of Nottingham, Nottingham; Wolfson Digestive Diseases Centre, University of Nottingham, Nottingham, UK Correspondence to: Prof. R. Logan, Division of Epidemiology and Public Health, Queens Medical Centre, University of Nottingham, Nottingham NG7 2UH, UK. E-mail:
[email protected]
Publication data Submitted 20 June 2008 First decision 7 July 2008 Resubmitted 10 September 2008 Accepted 10 September 2008 Epub Accepted Article 17 September 2008
SUMMARY Background Early endoscopy, Helicobacter pylori eradication and empirical acid suppression are commonly used dyspepsia management strategies in primary care but have not been directly compared in a single trial. Aim To compare endoscopy, H. pylori test and refer, H. pylori test and treat and empirical acid suppression for dyspepsia in primary care. Methods Patients presenting to their general practitioner with dyspepsia were randomized to endoscopy, H. pylori ‘test and treat’, H. pylori test and endoscope positives, or empirical therapy with symptoms, patient satisfaction, healthcare costs and cost effectiveness at 12 months being the outcomes. Results At 2 months, the proportion of patients reporting no or minimal dyspeptic symptoms ranged from 74% for those having early endoscopy to 55% for those on empirical therapy (P = 0.009), but at 1 year, there was little difference among the four strategies. Early endoscopy was associated with fewer subsequent consultations for dyspepsia (P = 0.003). ‘Test and treat’ resulted in fewer endoscopies overall and was most cost-effective over a range of cost assumptions. Empirical therapy resulted in the lowest initial costs, but the highest rate of subsequent endoscopy. Gastro-oesophageal cancers were found in four patients randomized to the H. pylori testing strategies. Conclusions While early endoscopy offered some advantages ‘Test and treat’ was the most cost-effective strategy. In older patients, early endoscopy may be an appropriate strategy in view of the greater risk of malignant disease. Aliment Pharmacol Ther 29, 55–68
ª 2008 The Authors Journal compilation ª 2008 Blackwell Publishing Ltd doi:10.1111/j.1365-2036.2008.03852.x
55
56 A . E . D U G G A N et al.
INTRODUCTION Dyspepsia is common with about 40% of adults in the UK reporting some symptoms.1 About a quarter of those with dyspepsia report consulting their general practitioner (GP) and their management has been estimated to cost over £600 million a year to the NHS which in 2001 consisted of £463 million for prescribed drugs and over £130 million for endoscopies.2 As dyspepsia is so common and its management so costly, it is important that it should be managed in an efficient as well as effective manner. Four main strategies exist for the initial management of those presenting in primary care: Endoscopy to establish a diagnosis and appropriate treatment (early endoscopy), testing for H. pylori and referring for endoscopy only those who are positive (test and refer), a policy of testing for H. pylori and giving H. pylori eradication to those positive (test and treat) and pragmatic treatment with a proton pump inhibitor (PPI) without investigation (empirical therapy).3 Potentially, early endoscopy improves management by ensuring diagnostic precision, reassuring doctors and patients about the underlying lesion and by detecting unsuspected malignancies. Whilst some studies have found some evidence of such benefit, others have questioned whether endoscopy is reassuring to patients and it is costly.4–7 One way of obtaining some of these benefits whilst reducing the demand for endoscopy lies in the test-and-refer strategy, in which patients testing H. pylori positive are referred for early endoscopy.8, 9 At the other end of the spectrum is the strategy of empirical therapy. However, while this is associated with symptom improvement in many patients, it can leave doctors and patients anxious as to important pathologies going undetected and untreated. Consequently, the test-and-treat strategy has become popular and is generally recommended as the most cost effective approach in younger patients.10, 11 Its effect as a treatment is limited and much of its attraction lies in the fact that it appears cheap.12–16 In addition, eradication of H. pylori might ultimately have the potential to reduce the risk of gastric cancer. However, because it does not detect existing malignancies, many guidelines recommend endoscopy in older patients such as those aged 55 or more.10, 11 While there have now been a number of randomized trials that reported comparing these four management strategies, all have involved pairwise comparisons,
mainly of early endoscopy vs. empirical therapy or H. pylori testing.3–9, 13–17 These trials also have often been performed in a secondary care setting, which would be expected to influence patient responses.4, 9, 13–15 The trial reported here was performed in primary care and is the first trial to compare directly all four management strategies simultaneously.
METHODS General practitioners in 43 practices in Nottingham, UK recruited patients to a randomized trial comparing four strategies for the initial management of dyspepsia. In the first strategy, all patients were randomized to early endoscopy (OGD). In the second strategy, patients had a near-patient test for H. pylori and those testing positive were referred for endoscopy. In the third strategy, patients testing positive received H. pylori eradication therapy without further investigation. Patients testing H. pylori negative in these two strategies received empirical therapy with a PPI. In the fourth strategy, all patients were treated with empirical therapy with a PPI (Figure 1).
Baseline assessment Dyspepsia was defined pragmatically as symptoms which the GP thought were arising from the upper gastrointestinal tract and were of sufficient severity to justify empirical therapy with an H2 antagonist or PPI. Patients randomized were between the ages of 18 and 70 years. Patients were ineligible if they were thought to be unfit for investigation, had ‘alarm’ symptoms suggestive of malignancy, (dysphagia, weight loss > 5 kg, anaemia, haematemesis, melaena or jaundice), had a previous radiological or endoscopic diagnosis of peptic ulcer disease or reflux oesophagitis, had had investigation for dyspepsia in the previous 5 years with either procedure or had symptom onset within 6 months of commencement of NSAID therapy. Previous H. pylori eradication therapy or more than three prescriptions for acid suppression therapy in the previous 6 months were also exclusion criteria.
Recruitment and randomization Patients were identified during their consultation with their GP between 1995 and 1998 with GP record data collected until 2001. After obtaining signed consent, patients completed a baseline questionnaire to confirm ª 2008 The Authors, Aliment Pharmacol Ther 29, 55–68 Journal compilation ª 2008 Blackwell Publishing Ltd
C L I N I C A L T R I A L : I N I T I A L M A N A G E M E N T O F D Y S P E P S I A I N P R I M A R Y C A R E 57
All patients presenting to GP with dyspepsia
762 recruited by GP and randomized to:
187 early OGD
199 H. pylori test & refer
198 H. pylori test & Rx
2 not tested
178 Empiric Rx
1 not tested 1 withdrew
52 H. pylori pos 17 refused OGD
170 to early OGD
142 H. pylori neg
46 H. pylori
149 H. pylori neg
3 refused OGD
49 to OGD
145 Empiric Rx
46 Hpylori
151 Empiric
175 Empiric Rx
2 withdrew
187 completed 2 months
186 completed first
197 completed 2 months
196 completed first year
196 completed 2 months
195 completed first year
176 completed 2 months
176 completed first year
Figure 1. Trial profile.
eligibility, assess symptomatology, past history of dyspepsia, previous dyspepsia investigation and dyspepsia risk factors such as smoking and use of alcohol and medications. Patients were also requested to provide a 7 mL blood sample which was sent to University Hospital, Nottingham for storage. On completion of the trial the stored serum was used to estimate H. pylori sero prevalence using a standard assay (HM-CAP; Enteric Products, Inc., Westbury, NY, USA). Patients were randomized by the GP opening the next sealed and numbered envelope containing the randomized strategy. Randomization was computer generated and performed in blocks of 8, 12 and 16 according to practice size.
Interventions For strategies requiring H. pylori testing, the GP or practice nurse tested serum from the clotted sample using the FlexSure HP near patient test (SmithKline Diagnostics, San Jose, CA, USA) as previously reported.18 Patients testing positive for H. pylori in the test-and-treat strategy received a prepackaged 1-week ª 2008 The Authors, Aliment Pharmacol Ther 29, 55–68 Journal compilation ª 2008 Blackwell Publishing Ltd
course of omeprazole 20 mg b.d., metronidazole 400 mg b.d. and clarithromycin 250 mg b.d., which was then the standard first line eradication regimen in use in Nottingham. Patients testing positive for H. pylori in the test-and-refer strategy were referred to endoscopy in a manner similar to patients randomized to the early endoscopy strategy. Patients testing negative for H. pylori in the two H. pylori testing strategies and those randomized to empirical therapy were prescribed lanzoprazole 30 mg daily for 1 month. Patients randomized to the two strategies requiring endoscopy were endoscoped within 10 days at University Hospital. Endoscopy was performed after lignocaine throat spray and intravenous sedation with midazolam unless declined. During endoscopy, biopsies were taken from both the antrum and gastric body for rapid urease testing (CLO test; Delta West, Beatley, Australia). Patients were treated according to the endoscopic findings such that H. pylori positive patients with duodenal or gastric ulcers were treated with H. pylori eradication therapy and those with H. pylori negative ulcers were treated with lanzoprazole 30 mg daily for 1 month. Erosive duodenitis was
58 A . E . D U G G A N et al.
confirmed on duodenal histology and considered part of the spectrum of duodenal ulcer disease. Biopsies were taken from gastric ulcers and any other lesions that might possibly be malignant. Patients with gastric ulcers had endoscopy repeated at 6 weeks to confirm healing. Patients with ulcers or erosive duodenitis who were H. pylori colonized were given H. pylori eradication. Patients with evidence of reflux disease and those in whom no abnormality was detected were given an information sheet and advised to attend their GP for a prescription for lanzoprazole 30 mg daily for 1 month and to make a follow-up appointment with their GP at 6 weeks.
Follow-up General practitioners were asked to review all trial patients at 6 weeks. Management then and subsequently was at their discretion. Patients not responding could be referred for routine out-patient appointments or endoscopy. Patients were sent repeat questionnaires at 2 and 12 months. Patients’ satisfaction with investigation and treatment and symptom response was assessed using a five-point Likert Scale. Data on GP consultations for dyspepsia, dyspepsia prescribing and hospital referral, endoscopy or admission were collected from GP records at 1 and 2 years by the trial coordinator (CAE) visiting the practices. Where necessary, hospital records were obtained.
Sample size and statistical analysis At the outset, we estimated that 250 patients randomized to each arm of the trial would be required to detect important differences in clinical outcomes (power = 90%, a 15% difference between strategies where outcomes with empirical therapy were 10–50% and significance level of 5%). All results were analysed using SPSS (SPSS Inc., Chicago, IL, USA). Ninety-five per cent confidence intervals were calculated for proportions. Differences between groups were assessed using a chi-squared test. The Nottingham University Hospital Ethics Committee and the Local Medical Committee for Nottinghamshire approved the trial.
Economic analysis The outcomes for the economic analysis were effectiveness as assessed by symptom response at 12 months and costs of managing dyspepsia from a
health service provider’s perspective. Data on resource use in primary and secondary care were collected from GP records and hospital notes; 12-month cost data were collected on all patients (n = 762). Cost estimates were calculated from the British National Formulary, Queen’s Medical Centre’s Pharmacy and Finance Departments, and NICE Dyspepsia guidelines (2004; Table 1). Economic analysis used dichotomized symptom data at 12 months as the measure of effectiveness (1 or 2 = minimal or no symptoms, 3, 4 or 5 = significant symptoms). Where data were missing at 12 months, last observation (2 months) was carried forward and, where not available, imputed by observed resource use. Cost-effectiveness was calculated in terms of net monetary benefit (NMB) as this allows probabilistic comparison of multiple treatment alternatives.19 NMB may be defined as the monetary value of the clinical benefit from a treatment minus the cost of that treatment: NMB = (k)ÆE ) C [Where E is clinical effect,
Table 1. Costs of dyspepsia related procedures and treatments Unit cost 2004
Source
Hp test (NPT) Helicobacter pylori eradication PPI 20 ⁄ 30 mg b.d. PPI 20 ⁄ 30 mg daily PPI 10 ⁄ 15 mg daily PPI as required Cimetidine 400 mg b.d. Cimetidine 400 mg daily Ranitidine 150 mg b.d. Ranitidine 150 mg daily Nizatidine (high) OPD
*£15.00 £30.49
NICE (2004) BNF (2004)
£25.50 £12.75 £11.40 £25.50 £5.58 £2.79 £8.16 £4.08 £13.70 £66.00
GP visits Barium meal
£18.00 £172.00
BNF (2004) BNF (2004) BNF (2004) BNF (2004) BNF (2004) BNF (2004) BNF (2004) BNF (2004) BNF (2004) QMC R&D Department NICE (2004) QMC R&D Department NICE (2004)
Resource
UBT
£18.80
NPT, near patient test; OPD, outpatient visit; PPI, proton pump inhibitor; UBT, urea breath test. * Cost of near patient breath test as FlexSure test no longer available. Based on cost of prescription of generic omeprazole.
ª 2008 The Authors, Aliment Pharmacol Ther 29, 55–68 Journal compilation ª 2008 Blackwell Publishing Ltd
C L I N I C A L T R I A L : I N I T I A L M A N A G E M E N T O F D Y S P E P S I A I N P R I M A R Y C A R E 59
C is cost and k is the value placed on the clinical effect (in monetary terms)]. In this type of analysis, the monetary value of the clinical benefit is based on the decision maker’s ‘willingness-to-pay’ for each unit of benefit (i.e. a symptom-free patient). As the ‘willingness-to-pay’ value (k) is generally unknown, results can be presented for a range of values for k, from zero to infinity. Where the monetary value of the clinical benefit exceeds the cost of the treatment (i.e. NMB > 0), the treatment can be considered costeffective.20 To determine which treatment option was most costeffective, we calculated the proportion of times that each treatment strategy had the highest NMB of the four comparators after 2000 reiterations of the sample data in a bootstrap simulation. The bootstrap technique randomly selects a new sample of the same size (with repeat observations allowed) from the originally observed sample; this can be repeated thousands of times, and ‘bootstrapped’ statistics can then be estimated. These results were presented graphically to define a cost-effectiveness acceptability curve, showing the level of certainty that each treatment will be cost-effective for different values of k. The interaction of the cost-effectiveness acceptability curves for each strategy forms the so-called cost-effectiveness acceptability frontier. Sensitivity analysis was also conducted in the NMB framework for the cost of endoscopy.
RESULTS Seven hundred and sixty-two patients were recruited to the trial (Figure 1). All groups had similar demographic characteristics and dyspepsia symptoms as shown in Table 2. The mean age of those recruited was 42 years (range 18–73). At recruitment, 65% of patients reported symptoms of dyspepsia at least 50% of the time and for 56% of patients, symptoms were of sufficient severity to interfere with normal activity. The serological prevalence of H. pylori was 37%. At 12 months from recruitment to the trial, GP data were available on 753 (99%) of recruited patients (Figure 1). Hospital data were available on all patients referred to hospital for further investigation. The 2-month symptom and satisfaction questionnaire was returned by 80% (610 ⁄ 762) of patients and the 12-month questionnaire by 74% (565 ⁄ 762); 143 patients (76%) from the early endoscopy strategy, 143 patients (72%) from the treat and refer strategy two, 142 patients (72%) from the test-and-treat ª 2008 The Authors, Aliment Pharmacol Ther 29, 55–68 Journal compilation ª 2008 Blackwell Publishing Ltd
strategy and 137 patients (77%) from the empirical therapy strategy.
Endoscopic findings (Table 3) Early endoscopy group. One hundred and seventy of 187 patients in the early endoscopy strategy were randomized as per protocol. In all, 17 patients initially refused endoscopy of whom two subsequently attended endoscopy. Of the 199 patients in the testand-refer strategy, 52 were detected as positive by near patient testing and referred for endoscopy. Of these, 49 were endoscoped and three refused. In the early endoscopy group, 41% had normal findings, 37% had oesophagitis and 23% had evidence of ulcers or erosions in the stomach or duodenum. In the H. pylori positive patients who were endoscoped in the test and refer group, 36% had evidence of ulcers or erosions, 31% oesophagitis and 31% were normal.
Other groups. Around 31% (163 ⁄ 521) of those whose initial management did not involve endoscopy subsequently underwent endoscopy during the first year as shown in Table 3. While 70 of 178 (39%) of patients recruited to the empirical therapy strategy were later referred for endoscopy, only five (7%) were found to have peptic ulcer disease. In this group, the main indications GPs gave for referral for endoscopy were failure to respond to PPI therapy or a recurrence of symptoms on ceasing PPI treatment.
H. pylori eradication Management of those in the early endoscopy and the test and refer group resulted in H. pylori eradication treatment within 1 month of trial entry in 30 and 22 patients respectively. Amongst the test and treat patients, 46 tested positive and underwent eradication treatment within 1 month of recruitment. None of those allocated empirical PPI treatment underwent H. pylori eradication within 1 month of trial entry.
Clinical course over the first year (Table 4) Consultations. Of those randomized to endoscopy, fewer subsequently consulted for dyspepsia compared with the other strategies (v2 = 13.7, three df, P = 0.003). Eighty (43%) made a total of 263 further consultations for dyspepsia in the first year (1.4 per
60 A . E . D U G G A N et al.
Table 2. Patient characteristics Strategy Endoscopy
Test & refer
Total 187 199 Demographic M:F 102:85 102:97 No (%) ‡ 50 years 49 (26) 61 Helicobacter pylori positive by premier ELISA (%)* 79 (44) 73 % current smokers 63 (34) 81 Occupation I ⁄ II (professional ⁄ managerial) 45 (24) 63 III (skilled manual ⁄ nonmanual) 56 (30) 61 IV ⁄ V (semi or unskilled manual) 42 (22) 32 Unclassifiable 44 (24) 43 Symptoms reported at recruitment Upper abdominal discomfort 133 (71) 138 Heartburn 121 (65) 139 Frequency of symptoms 50% of the time or more 109 (58) 126 Severity of symptoms Interferes with normal activity 101 (54) 110 Days lost from work because of dyspepsia§ Patients (%) 33 (27) 34 History of dyspepsia prior to recruitment 1st episode 46 (25) 50 For