Dyspepsia is common in the general population. Studies have reported prevalence estimates from 20 to 50 percent (2;42–44), the broad spectrum prevalence covered being mainly explained by the variation in definition of symptoms (36). Dyspepsia-related healthcare costs are high due to many sick-leave days, diagnostic investigations, and primary care consultations (71;72;75), and especially considerable costs for pharmacological treatment (1;92). Because dyspepsia is a frequent reason for consulting a general practitioner (GP) (37;47), the choice of the initial management strategy has a major influence on healthcare costs.
The available management strategies for uninvestigated dyspepsia include prompt endoscopy, the Helicobacter pylori (H. pylori) test-and-eradicate strategy, and empirical antisecretory therapy. Several guidelines have recommended the H. pylori test-and-eradicate strategy for initial management of young patients with dyspepsia, reserving the endoscopy strategy for older patients (>45 years) and those with alarm symptoms (62;85). However, such recommendations have been made in the absence of definitive cost-effectiveness data and despite conflicting clinical trial results (61).
H. pylori test-and-eradicate has been shown to be as effective as early endoscopy and to reduce costs (5;59;68), but uncertainty remains as to its cost-effectiveness in primary care compared with empirical acid suppression (21;23). A recent decision analytic model found that a combination strategy consisting of initial test-and-eradicate for H. pylori, followed by empirical antisecretory therapy for nonresponders may be more cost-effective than H. pylori test-and-eradicate or empirical antisecretory therapy alone (82). Discrepant results, attributed to differences in data used in the modeling studies (11;26;28;55;61;73;74;81;82), however, indicate that the analyses may not always be useful and highlight the need for direct evidence obtained from randomized trials (77) in a primary care setting.
We aimed to provide an economic evaluation comparing empirical antisecretory therapy with H. pylori test-and-eradicate for management of dyspepsia in general practice. A randomized controlled trial was conducted providing data on resources used and patient outcome.
METHODS
Clinical Study
A prospective cluster-randomized trial was performed in 106 general practices in the County of Funen, Denmark, between June 2001 and October 2003. Patients were eligible if they presented with dyspeptic symptoms (epigastric pain or discomfort with or without heartburn, regurgitation, nausea, vomiting, or bloating) for more than 2 weeks, and if the GP found indication for investigation or treatment. Exclusion criteria were alarm symptoms, ongoing treatment with antisecretory drugs, ongoing use of NSAID, serious or terminal disease, alcohol or drug abuse, previous surgery of the upper gastrointestinal tract, non-Danish speaking, age below 18 years, and pregnancy or lactation.
Each practice was randomized to one of three management strategies: Strategy A, Empirical antisecretory therapy, esomeprazole 20 mg twice a day for 1 week; Strategy B, H. pylori test-and-eradicate, H. pylori testing, eradication treatment if positive; Strategy C, Combination strategy, esomeprazole 20 mg twice a day for 1 week followed by H. pylori testing if symptoms improved, eradication treatment if H. pylori-positive.
We used the [13C]urea breath test (UBT) for H. pylori test (79). Due to the risk of a false-negative result, patients treated according to Strategies B and C were denied proton pump inhibitor (PPI) for at least 1 week before UBT. The eradication treatment was esomeprazole 20 mg, amoxicillin 1,000 mg, and clarithromycin 500 mg twice daily for 1 week. The effect of the 1 week PPI was evaluated by the GP; treatment response was defined as improvement in symptoms.
Effectiveness Outcome Measures
The main clinical outcome measure was the proportion of days without dyspeptic symptoms as registered in patient diaries during a 1-year follow-up period. Secondary outcome measures were gastrointestinal symptom scores (Gastrointestinal Symptom Rating Scale) (25), Quality-of-Life scores (SF-36) (91)), and patient satisfaction; all stated in status questionnaires upon entry, after 6 weeks, and after 1 year. The global rating of the severity of symptoms was recorded in the status questionnaires; after 6 weeks and 1 year, patients were asked to compare their symptoms with those at entry (no symptoms, improved, unchanged, or worse).
Resource Utilization
During the 1-year follow-up phase of the study, patients were managed by their own GP according to standard clinical practice. The health resource utilization data were collected prospectively and were performed from a societal perspective, which suggests that also the indirect costs were included.
Sick-leave days and visits to GPs in the preceding month were registered by the patients in absolute numbers on the diary cards. Study protocol visits were not included as they occurred equally in the three randomization groups. The number of endoscopies was obtained from the patient administrative system in the County of Funen. Information on use of medication was collected from the Odense University Phamacoepidemiologic Database (OPED). OPED is a research database based on information supplied by community pharmacies containing information on prescriptions of drugs for each individual in the County of Funen since 1990.
Unit Costs
The total costs were estimated from individual patient's use of resources with unit costs applied from national data (activity-based costing methods). All cost estimates are presented in 2004 Danish kroner (DKK; 7.4 DKK ≈1 EUR, 6.2 DKK ≈1 US$; Table 1).
The costs for endoscopies were based on current charges from the National Board of Health and included costs for staff, instruments, and utensils. Provision for depreciation was determined from information about investment cost and durability of the endoscopes. Research and staff training were considered to be fixed costs and, therefore, were not included.
The UBT for H. pylori was carried out in general practice at the study protocol visits. Costs included purchase price for the test kit, laboratory fees, and staff pay, as well as provision for depreciation of apparatus. The cost of a GP consultation was based on current 2004 charges agreed by Danish General Practitioners' Organisation and the National Health Insurance.
Drug consumption was analyzed using the defined daily doses (DDD) and the price per daily dose. For unit costs, we determined the most used drug in the follow-up period and costs for drugs used as project medication.
Transport costs associated with GP visits or visits to outpatient clinics were estimated by multiplying average travel distance with a cost per kilometer. The average distance to a GP was estimated to be 3 km one way. The average distance to outpatient clinics was estimated on the basis of data collected in another study carried out in the County of Funen (7) and was determined to be 8 km one way.
The indirect costs comprised the value of time related to number of sick-leave days, and time for visits to the GPs and outpatient clinics. Cost of time was estimated by applying hourly wages based on sex- and occupation-specific salaries. Time costs were assumed to be zero for patients who were unemployed. We estimated the transport time to be 0.5 hours for visits to the GP and 0.8 hours for visits to outpatient clinics. Time for a GP consultation was estimated to be 0.5 hours, and time involved in an endoscopy was estimated to be 2.5 hours, including waiting time.
Economic Analyses
The main economic objective was to measure prospectively the costs of health resources consumed per patient over 1 year. Mean costs per patient are reported. Discounting is not included in the analyses, as the estimated resource consumption occurs within 1 year.
The cost-effectiveness is calculated as mean cost divided by the outcome measures. Incremental cost-effectiveness ratios (ICER) are calculated as the difference in mean cost (ΔC) divided by difference in effectiveness (ΔE), going from the least expensive to the most expensive strategy. ICER were determined among nondominated alternatives (61).
Handling Uncertainty
We performed one-way sensitivity analyses on unit costs for endoscopies (5,000 DKK and 1,000 DKK), cheapest and most expensive drugs per class, and for the prevalence of H. pylori infection (twice and half of the present infection rate) (10;46). Furthermore, we demonstrated the implication of setting the opportunity cost of time equal to half and one quarter of the wage rate (80).
Utilizing the nonparametric bootstrap method (12), we also present the uncertainty of the sampling estimates of the ICER as cost-effectiveness acceptability curves (9;30). The cost-effectiveness acceptability curves represent the probability that the intervention is cost-effective with society's increasing willingness to pay (WTP) for additional treatment success.
RESULTS
The study comprised 722 patients. The allocation resulted in 222, 250, and 250 patients managed according to Strategy A, B, and C, respectively. Baseline characteristics of the three groups appear from Table 2.
Effectiveness Outcome Assessments
The mean proportion of days without dyspeptic symptoms during the 1-year follow-up period was 0.59 (95 percent confidence interval [CI], 0.54–0.63) in the group treated with empirical antisecretory therapy, 0.57 (95 percent CI 0.53–0.60) in the H. pylori test-and-eradicate group and 0.53 (95 percent CI 0.49–0.58) in the combination group, p=.16. At 12 months 23 percent, 26 percent, and 22 percent, respectively, had no dyspeptic symptoms. The total symptom scores on the Gastrointestinal Symptoms Rating Scale, in the generic Quality-of-Life questionnaire and the satisfaction with management after 1 year are presented elsewhere (41). None of these measures differ significantly across the three strategies.
Resource-Utilization
Use of endoscopy was higher in the empirical treatment group, where the mean per patient use of endoscopy was 0.36 (95 percent CI, 0.30–0.43) compared with 0.28 (95 percent CI, 0.23–0.34) and 0.22 (95 percent CI, 0.17–0.27) in the test-and-eradicate group and the combination group, respectively, p=.02. H. pylori test was applied to all patients in the test-and-eradicate group and to 90 percent of patients in the combination group according to the study protocol. The H. pylori infection rates were 24 percent and 23 percent in the test-and-eradicate group and the combination group, respectively, and all H. pylori-positive patients were treated with eradication therapy. Patients randomized to the empirical treatment group had more sick-leave days; however, the difference between the groups was significant for patients in work only. The mean drug consumption and visits to GPs during the 1-year follow-up did not differ significantly across the three strategies.
Table 3 combines the unit costs from Table 1 and the resource consumption, summarizing the average cost per patient. In total costs, the mean yearly average cost per patient for empirical treatment, H. pylori test-and-eradicate, and the combination strategy was 7,157 DKK, 5,458 DKK, and 3,870 DKK, respectively.
Economic Analysis
Table 4 shows the parameter estimates used for the cost-effectiveness calculations. The combination strategy offers the lowest costs, but also a lower effectiveness rate. Empirical treatment provided an additional proportion of days without dyspeptic symptoms at 0.06 during 1 year compared with the combination strategy, but at an extra cost of 3,287 DKK. When the effectiveness outcome was the proportion of patients free of symptoms after 1 year, the H. pylori test-and-eradicate strategy was less expensive and more effective, as well, compared with the empirical treatment strategy.
According to the one-way sensitivity analyses, the opportunity costs of patients' time had the greatest influence on the cost-effectiveness analyses, but none of the sensitivity analyses changed the direction of cost-effectiveness among strategies.
The cost-effectiveness acceptability curve (Figure 1) is a plot of the probability that the incremental step going from the cheapest to a more expensive strategy is cost-effective as a function of increasing WTP. The probability that, for example, the incremental step from Strategy C to Strategy A is cost-effective at WTP equal to 40,000 DKK is 33.7 percent. If society's WTP for an additional treatment success increases to 80,000 DKK, the probability for cost-effectiveness increases to 63.5 percent.
Cost-effectiveness acceptability curves, all costs. Thecost-effectiveness acceptability curves represents the probability that theintervention is cost-effective as a proportion of the willingness to pay(WTP) for additional treatment success. Strategy A: Empirical antisecretorytherapy; Strategy B: testing for Helicobacter pylori; Strategy C: empirical antisecretorytherapy, followed by H. pylori testing if symptoms improved. DKK, Danish kroner.
DISCUSSION
The aim of this economic analysis was to identify, measure, and compare competing alternatives in the management of dyspepsia in primary care. We found the strategies based on eradication of H. pylori infection cost-effective compared with empirical treatment. The result was particularly due to the reduced endoscopic workload and fewer sick-leave days in the H. pylori test-and-eradicate strategies, as the clinical outcomes showed minor differences between the strategies.
Strengths and Weaknesses of the Study
Most economic data in the area of dyspepsia have arisen from modeling studies and have included estimates of treatment success and costs based on expert opinion (19). In contrast, the present study was designed for a prospective collection of all health care utilization data in combination with the effectiveness estimates, which extends the internal validity of results (63).
The perspective of the cost analysis was societal, as recommended for explicit cost calculation methods and sources of economic information (33). The use of a broader perspective that includes indirect costs is relevant, because gastrointestinal diseases account for substantial loss of productivity (80). The currently used approach for calculating productivity costs due to absence from work is to take days absent valued by gross earnings, using the argument that this approach reflects the lost value of production when individuals are absent from work (49). This strategy in fact reflects the potential production costs, whereas the actual costs to society may be lower because the absence is compensated (50;80); for example, losses may be recovered when returning from short-time absence and, for longer periods, the patient can be replaced by an unemployed worker (45). Furthermore, we estimated the time for visits to GPs or outpatient clinics based on hourly wages, irrespective of whether the individual used working hours or leisure time. In the present study, the time costs were assumed to be zero for patients who were unemployed. The time cost, therefore, is considered to be a conservative estimate of the patients' opportunity cost of time. The reason for excluding the time cost of the unemployed is uncertainty about the valuation of their time and the exact meaning of sick-leave days. As recommended (80), we used a sensitivity analysis to explore the implications of setting the opportunity cost equal to half and one quarter of the wage rate and found both analyses diminished but not changing the direction of cost-effectiveness.
The three cost-effectiveness acceptability curves, furthermore, illustrate the uncertainty in the cost-effectiveness analyses. One problem is that the amount society is willing to pay for a unit of effect is unknown. However, the present figure indicates that the empirical treatment strategy is probably not cost-effective at a realistic level of willingness to pay.
The absence of validated outcome measures and lack of consensus on how to measure outcome or how to define efficacy of an intervention constitutes a methodological problem (90). Because dyspepsia patients may have several symptoms (2) and severity as well as symptoms may vary in the same patient over time (3), a measure of overall change in symptoms should be the primary outcome criterion (90). We used the proportion of symptom-free days during the 1-year follow-up period as the main outcome measure, but also reported the global measure of patients who obtained total symptom relief after 1 year. These measures are easily understood, make clinical sense, and are not biased by methodological difficulties associated with measuring over time. However, the measures may underestimate treatment effects in patients with an incomplete yet beneficial response (16). As dyspepsia is a chronic relapsing disorder, there may be some methodological limitations of using absolute effect measures in cost-effectiveness analysis (51), and the proportion of symptom-free days may be a more exact approach than the total symptom relief at a given moment.
In the case of a fluctuating disorder, management of patients with persisting symptoms is as important as the initial management strategy. This issue is difficult to capture in randomized trials where complex algorithms and planned follow-up may take the trial too far from everyday practice (22). In our study, however, we chose that all additional treatment or triage in the 1-year follow-up should be left at the GP's discretion in accordance with their usual clinical practice and not according to predetermined algorithms, which makes the outcomes very similar to the outcome in daily practice.
Strengths and Weaknesses in Relation to Other Studies
The H. pylori test-and-eradicate strategy has been advocated to patients under the age of 45 years. The guidelines are based in large part on the recognized role of H. pylori in peptic ulcer disease and on considerations regarding the cost-effectiveness of noninvasive versus invasive (e.g., endoscopy) initial management strategies (52). The benefit derived from the test-and-eradicate strategy, attributable to cure of H. pylori infection in patients with peptic ulcer disease, and subsequently the prevention of peptic ulcer recurrence is proved in several studies (6;31;34;39;60;65), whereas the reduction in dyspeptic symptoms and continued use of antisecretory drugs after eradication therapy have been less clear (15;57;67;89).
The challenge in primary care is that the majority of patients presenting with dyspepsia have functional dyspepsia rather than peptic ulcer disease (35;38;87). The efficacy of eradication therapy in resolving dyspeptic symptoms in patients without peptic ulcer disease is still a subject of debate (84), as several studies have evaluated the efficacy of eradication therapy on functional dyspepsia (8;13;32;40;48;56;66;69;70;86;88) and with different results. A recent economic analysis modeled several dyspepsia strategies (61) and found the H. pylori test-and-eradicate approach to be the most effective regarding symptomatic cure, but the most costly initial test in adult dyspeptics. The model also found that the choice of the most cost-effective approach was dependent on the benefits of H. pylori eradication in patients with functional dyspepsia.
We included a combination strategy comprising initial PPI therapy, with test-and-eradicate reserved only for treatment responders. This combination strategy has not been tested in a clinical setting before, but decision analytic models have indicated that combination strategies may be more cost-effective than test-and-eradicate or empirical antisecretory therapy alone (17;82). Performing test-and-eradicate only in those responding to antisecretory therapy would potentially benefit the population of dyspepsia patients with H. pylori–related acid-peptic disease. In the present cost-effectiveness analysis the combination strategy was the most cost-effective approach, and even more cost-effective than the test-and-eradicate strategy alone, despite lower effectiveness (Table 4). Only a minority of the patients initially treated with antisecretory therapy were denied the subsequent H. pylori test, although the reduced costs were not related to saved H. pylori tests and eradication treatments, but to reduced endoscopic workload and fewer sick-leave days. In the sensitivity analysis, we showed that a decline in the costs of antisecretory therapy will further support the combination strategy.
Only a few randomized clinical trials have compared the H. pylori test-and-eradicate strategy with empirical antisecretory therapy, and most studies did not perform regular cost-effectiveness analyses of the management strategies. Manes et al. (64) showed that eradication treatment allowed resolution of symptoms in a large number of patients and reduced the endoscopic workload, but the prevalence of H. pylori was found to be quite high (61 percent) in this study (24) and no economic evaluation was performed. Chiba (18) also showed an improvement in dyspeptic symptoms, but studied H. pylori-positive patients randomized to eradication treatment or antisecretory therapy. In the economic evaluation, eradication therapy for those who were H. pylori-positive led to a cost-effective improvement in dyspepsia symptoms compared with a strategy of no eradication therapy (19). Allison et al. (4) enrolled patients already receiving antisecretory therapy for physician-diagnosed peptic ulcer disease and randomized patients to H. pylori test-and-eradicate or to usual care. The study showed a modest but significant effect in reducing clinical symptoms, but a higher total costs for the test-and-eradicate group. Finally, one U.S. study failed to identify a symptomatic advantage of test-and-eradicate over usual care, which is in accordance with the findings of the present study, but the American study only comprised ninety-three patients (54). No difference was found in total disease-related expenditures per patient in the two groups.
It has been argued that, as the prevalence of H. pylori infection falls (78) and the likelihood of H. pylori-negative ulcer increases (20;76;93), empirical antisecretory therapy will become more cost-effective (52) and quite compelling in the setting of a low background prevalence of ulcer disease and H. pylori (14;82). In a cost-minimization study of the two strategies, antisecretory therapy was consistently less costly than H. pylori test-and-eradicate, when the H. pylori prevalence became lower than 20 percent (53). In the present study, 24 percent of H. pylori–tested patients were infected, which is in accordance with the rates reported in most northern European countries and North America (24;29;59;83).
H. pylori testing may have benefits that are not only attributable to the cure of H. pylori infection in patients with peptic ulcer disease. Farkkila et al. (27) as well as McColl et al. (68) suggested that patients with a negative H. pylori test could be reassured that they probably did not have underlying ulcer disease and could be treated symptomatically, as would occur after an endoscopic examination showing no abnormality. H. pylori test-and-eradicate may even have longer term benefits that are not accounted for in the present 1-year follow-up study. A long-term (6.7 years) follow-up study of strategies based on H. pylori test-and-eradicate and prompt endoscopy showed a reduction in both endoscopic workload and use of antisecretory medication in the test-and-eradicate group (58).
CONCLUSION
In summary, this primary care study demonstrated that the empirical treatment strategy with PPI entailed greater 1 year total costs, but only small and statistically insignificant improvements in outcome. The ICER of the combination strategy versus empirical treatment was 54,783 DKK per proportion of days without dyspeptic symptoms; an amount that probably exceeds the realistic level of willingness to pay for the value of the benefit of achieving one extra treatment success and strategies based on H. pylori test-and-eradicate might be cost-effective compared with empirical treatment. Helicobacter-based strategies, however, are challenged by a decreasing prevalence of peptic ulcer disease and of the infection (14) and the cost-effectiveness of the strategies requires periodic re-evaluations in the relevant clinical settings.
POLICY IMPLICATIONS
The present result can be used in healthcare decision making at a clinical practice level as well as in a public health perspective. For the general practitioner, the economic analysis can contribute to the choice of management strategy for the dyspepsia patients, where assessments of feasibility, patient acceptability, and daily practice also form part of this decision making. In a public health perspective, availability of resources will influence the choice of strategy, for example, the access to endoscopy facilities and the practicability for implementing the H. pylori test.
CONTACT INFORMATION
Dorte Ejg Jarbol, MD, PhD (djarbol@health.sdu.dk), Senior Researcher, The Research Unit for General Practice, Mickael Bech, PhD (mbe@sam.sdu.dk), Associate Professor, Institute of Public Health, Health Economics, Jakob Kragstrup, PhD (jkragstrup@health.sdu.dk), Professor, The Research Unit for General Practice, University of Southern Denmark, J.B Wisløws Vej 9A, DK-5000 Odense, Denmark
Troels Havelund, MD, PhD (troels.havelund@ouh.fyns-amt.dk), Associate Professor, Faculty of Health Sciences; Ove B. Schaffalitzky de Muckadell, MD, DMSci (sdm@ouh.fyns-amt.dk), Professor of Medicine, University of Southern Denmark, 19 Winslowparken; Consultant, Department of Medical Gastroenterology S, Odense University Hospital, 29 Sdr Boulevard, DK-5000 Odense, Denmark
We thank all the general practitioners and patients who participated in this study. The study was funded by the Danish Centre for Evaluation and Health Technology Assessment, The National Board of Health, The Danish Medical Association Research Fund/The Lundbeck Foundation Grant, The Institute of Clinical Research, University of Southern Denmark, and The Foundation for Medical Science Research at the County of Funen Hospital Authority. AstraZeneca and Abbott Laboratories supplied most of the medication for the project. The entire funding source had no involvement in the study or in the approval of the manuscript.
