Participation in a clinical trial influences the future management of patients with gastro-oesophageal reflux disease in general practice
Dr V. Meineche-Schmidt, Christiansholmsvej 5, DK-2930 Klampenborg, Denmark.
Background The long-term effects of participation in trials has not been reported. A randomized-controlled trial (the ONE study) reported on the management of gastro-oesophageal reflux disease with esomeprazole in primary care, testing on-demand treatment vs. treatment courses.
Aim To evaluate the impact of participation in a trial on General Practitioners management and patient behaviour.
Methods Management of gastro-oesophageal reflux disease was compared between General Practitioners who participated in ONE (ONE-GPs) and a random sample of General Practitioners who did not participate in ONE (Other-GPs). Symptom presentation and satisfaction with treatment was compared between patients who had participated in ONE (ONE-patients) and patients who had not (Other-patients).
Results ONE-GPs prescribed on-demand treatment with proton-pump inhibitors to 47% of the patients, Other-GPs to 27%. ONE-patients consulted for significantly less symptoms compared with Other-patients. ONE-patients reported significantly higher patient satisfaction compared with Other-patients. ONE-patients used 98 doses during 6 months whereas Other-patients used 76 doses.
Conclusions Participation in a clinical trial influenced both doctors and patients. Treatment modalities introduced by the trial were used in daily practice by the General Practitioners. Patients who had participated in the trial consulted for less symptoms and used more medication, compared with patients who did not participate.
Management recommendations for a defined disease or symptom complex can be based on randomized trials or consensus among specialists. Regarding gastro-oesophageal reflux disease (GERD) several trials1–3 have been published as well as consensus reports from specialists,4 systematic reviews5 and Cochrane reviews.6 Results from randomized-controlled trials are considered best evidence. However, randomized-controlled trials are limited according to their protocol, often testing one defined treatment or strategy against another. The relationship between the trial and daily routine treatments are often not addressed. A trial will often introduce doctors and patients to a new treatment option, which may be incorporated in the daily clinic. But what follows a randomized-controlled trial? Does the trial result influence doctors to change management? Does the trial influence the patient's health-seeking behaviour? Or is everything ‘back to normal’ after the trial?
In a prospective, open, randomized multicentre study, the ONE trial3 a total of 1357 patients with symptoms suggestive of GERD were randomized to 6-month treatment with either on-demand treatment (esomeprazole 20 mg) or treatment courses (with esomeprazole 40 mg daily for 2 or 4 weeks), i.e. one study arm reflected patient control and the other two arms General Practitioner (GP) control with the management. The study demonstrated that on-demand treatment had significantly lower total costs, compared with treatment courses of both 2- and 4-week duration. These results, however, were based on the parameters of the trial, which did not reflect the daily clinical practice in primary care in at least three aspects: (i) the only treatment option was esomeprazole in contrast to other proton-pump inhibitors (PPIs), H2-blockers, antacids or advice, (ii) patients allocated for treatment courses had to consult the GP by appointment, in contrast to telephonic contacts and (iii) during the trial medication was free, in contrast to co-payment in real life.
The idea of the present study (ONETWO) was to compare the results from ONE3 to daily clinical practice (in order to assess to what extend the trial's economic results were representative for ‘real life’), to compare symptom presentation, management and patient satisfaction in ONETWO, according to the GPs’ or the patients’ participation in ONE (in order to reveal any effect of former participating in a trial on the daily clinic for the following period) and finally to compare patients who had participated in ONE to patients who had not (to address the question of patient-related consequence of participating in trials).
- 1To compare management strategies in daily clinical practice between GPs who had and GPs who had not participated in the ONE trial.
- 2In GPs who had participated in ONE, to compare management strategies in patients who had participated and patients who had not participated in ONE.
- 3To compare symptom presentation by patients who had participated and patients who had not participated in the ONE trial.
- 4To compare patients satisfaction, treatment response and use of medication according to participation in ONE.
Material and methods
This study (‘ONETWO’) was carried out exactly as the ONE trial3 with regard to inclusion and exclusion criteria, observation period and analyses, but management of the patients were left to the GPs according to their daily routines, with no restrictions. Thus, the study was designed to observe in a prospective manner, how the participating GPs managed patients with symptoms suggestive of GERD during a period of 6 months.
Patient-driven consultation because of symptoms suggestive of GERD (heartburn as the predominant symptom with or without regurgitation and/or epigastric pain) for 3 days or more during the last week. Main exclusion criteria were: known complications to GERD, alarm symptoms, history of upper gastrointestinal (GI) surgery, pregnancy and lactation.
General Practitioners who had participated in ONE were invited to participate in ONETWO and the first 75 GPs, who volunteered were enrolled (‘ONE-GPs’). Another 75 GPs were enrolled following invitation to a randomly selected number of GPs throughout Denmark (‘Other-GPs’). A total of 28 GPs did not enrol sufficient number of patients, hence 64 ONE-GPs and 58 Other-GPs, included patients to ONETWO. The two groups of GPs were compared according to age, gender, experience (years as MD) and geography.
A total of 1206 patients were included, some had participated in the ONE trial (‘ONE-patients’) and others had not (‘Other-patients’). The study population was grouped into three different subgroups (Table 1).
Table 1. Subgroups of patients in ONETWO according to participation in ONE
|Yes||Group A (N = 247)|| |
|No||Group B (N = 451)||Group C (N = 469)|
- 1Group A: ONE-patients who were included into ONETWO by ONE-GPs (N = 247).
- 2Group B: Other-patients who were included into ONETWO by ONE-GPs (N = 451).
- 3Group C: Other-patients who were included into ONETWO by Other-GPs (N = 469).
Study objective 1 was addresses by comparing groups A + B to group C, study objective 2 by comparing group A to group B and study objective 3 by comparing group A to groups B + C.
At visit 1, the following GERD symptoms were assessed. Heartburn: severity (none, mild, moderate, severe) and frequency (number of days) within the last week. Epigastric pain and regurgitation: severity (none, mild, moderate, severe) within the last week. At visit 2, 6 months ±7 days later, GERD symptoms were assessed as in visit 1.
Patient satisfaction was addressed at visit 2 by two questions: (i) ‘How satisfied have you been with the symptom relieving effect of the treatment during the last six months'? (ii) ‘How satisfied have you been with the way your GP has managed your GORD symptoms'? Both questions could be answered by using: completely satisfied, very satisfied, quite satisfied, satisfied, dissatisfied, very dissatisfied and completely dissatisfied.
The response to management was assessed at visit 2 and defined as: good response = no more than 1 day with mild heartburn during the last week; fair response = no more than 1 day with moderate heartburn or 2 days with mild heartburn during the last week; no response = all other outcome.
The proportion of patients with different management strategies was compared between groups of patients according to participation of GP and patient in ONE. All tests were two-tailed at 5% significance level and 95% confidence intervals were applied. Patients with missing assessment were considered treatment failures in respect to that parameter in the analysis.
A total of 1206 patients were enrolled in the study. Altogether 39 patients (five, 11 and 23 in groups A, B and C, respectively) were excluded from analysis: six patients had <3 days of symptoms before inclusion, 13 patients had a history of disease that should exclude the patient from the study, 12 patients did not use effective contraception and eight patients for other reasons, leaving 1167 patients for analysis. Patient allocation according to subgroup is shown in Table 1.
Baseline data (Table 2) demonstrated no significant differences in age and sex among the subgroups in the study. Comparison between ONE-GPs and Other-GPs showed no difference in sex, geography, number of years as MDs and distribution according to region or type of practice (Table 3).
Table 2. Baseline characteristics of participating patients in ONETWO
|<39||20 (8)||89 (20)||121 (26)|
|40–54||72 (29)||117 (26)||144 (31)|
|55–64||74 (30)||122 (27)||108 (23)|
|>65||81 (33)||123 (27)||96 (20)|
|Females||135 (55)||275 (61)||220 (47)|
|Days with heartburn/week|
| 3||71 (29)||82 (18)||115 (25)|
| 4||53 (21)||100 (22)||79 (17)|
| 5||42 (17)||49 (11)||47 (10)|
| 6||26 (11)||41 (9)||37 (8)|
| 7||55 (22)||179 (40)||191 (41)|
|Mean days with heartburn||4.8||5.3||5.2|
|Heartburn last week|
| Mild = 1||40 (16)||44 (10)||95 (20)|
| Moderate = 2||170 (69)||312 (69)||289 (62)|
| Severe = 3||37 (15)||95 (21)||85 (18)|
|Mean severity of heartburn||2.0||2.1||2.0|
|Epigastric pain last week|
| None = 0||40 (16)||66 (15)||64 (14)|
| Mild = 1||103 (42)||127 (28)||124 (26)|
| Moderate = 2||85 (34)||212 (47)||208 (44)|
| Severe = 3||19 (8)||46 (10)||73 (16)|
|Mean severity of epigastric pain||1.3||1.5||1.5|
|Regurgitation last week|
| None = 0||32 (13)||63 (14)||108 (23)|
| Mild = 1||101 (42)||159 (35)||131 (28)|
| Moderate = 2||91 (37)||187 (41)||169 (36)|
| Severe = 3||23 (9)||42 (9)||61 (13)|
|Mean severity of regurgitation||1.4||1.5||1.4|
|NSAID in anti-inflammatory dose||46 (19)||63 (14)||129 (28)|
|Satisfaction: symptom control|
| Completely satisfied||120 (49)||205 (45)||173 (37)|
| Very satisfied||69 (28)||111 (25)||110 (23)|
| Quite satisfied||28 (11)||56 (12)||79 (17)|
| Satisfied||17 (7)||31 (7)||57 (12)|
| Dissatisfied||8 (3)||19 (4)||17 (4)|
| Very dissatisfied||1 (<1)||2 (<1)||4 (<1)|
| Completely dissatisfied||0||1 (<1)||1 (<1)|
| Missing||4 (2)||26 (6)||28 (6)|
| Completely satisfied||145 (59)||243 (54)||229 (49)|
| Very satisfied||60 (24)||115 (25)||117 (25)|
| Quite satisfied||27 (11)||43 (10)||55 (12)|
| Satisfied||11 (4)||23 (5)||35 (7)|
| Dissatisfied||0||1 (<1)||4 (<1)|
| Very dissatisfied||0||0||1 (<1)|
| Completely dissatisfied||0||0||0|
| Missing||4 (2)||26 (6)||28 (6)|
|Response to treatment|
| Good response*||162 (67)||288 (68)||275 (62)|
| Fair response†||28 (12)||29 (7)||24 (5)|
| No response‡||53 (22)||109 (26)||143 (32)|
| Missing||4 (2)||26 (6)||28 (6)|
Table 3. Comparison between ONE-GPs and Other-GPs in the ONETWO study
| Female||11 (19)||7 (11)||18 (15)|
| Male||47 (81)||57 (89)||104 (85)|
| Urban||26 (45)||30 (47)||56 (46)|
| Suburban||22 (38)||24 (38)||46 (38)|
| Rural||10 (16)||10 (16)||20 (16)|
|Years as MD|
| 1–9||3 (5)||0||3 (3)|
| 10–19||16 (28)||14 (22)||30 (25)|
| 20–29||30 (52)||41 (64)||71 (58)|
| 30–||9 (16)||14 (9)||18 (15)|
|Region in DK*|
| 1||25 (43)||28 (44)||53 (43)|
| 2||8 (14)||13 (20)||21 (17)|
| 3||15 (26)||15 (23)||30 (25)|
| 4||10 (17)||8 (13)||18 (15)|
The management strategy at first consultation (according to subgroup) is shown in Table 4. The management strategy by ONE-GPs differed between ONE-patients and Other-patients: 13% of ONE-patients were treated with advice only, compared with 8% of Other-patients (P = 0.04). If prescription was given, no significant difference was found according to use of H2-blockers or PPIs. The prescription mode was different: ONE-patients were significantly more often prescribed medication on demand: 58% vs. 40% (P < 0.0001) and treatment courses were significantly more often prescribed in Other-patients: 29% vs. 11% (P < 0.0001). Prescription as daily maintenance treatment did not differ significantly (P = 0.14).
Table 4. Management strategy of reflux symptoms at first consultation according to subgroups in ONETWO
|Advice and/or OTC medication|| 32 (13)|| 35 (8)|| 38 (8)|
|Prescription(s) at first visit||215 (87)||416 (92)||431 (92)|
|H2-blocker|| 12 (5)|| 2 (1)|| 5 (2)|| 13 (3)|| 10 (2)|| 12 (3)|| 19 (4)|| 23 (5)|| 8 (2)|
|PPI||131 (53)||25 (10)||40 (16)||166 (37)||120 (27)|| 93 (21)||110 (23)||166 (35)||102 (22)|
|Total||143 (58)||27 (11)||45 (18)||179 (40)||130 (29)||105 (24)||129 (27)||189 (40)||110 (24)|
Depending on participation in ONE, the GPs management was (Table 5): ONE-GPs prescribed medication in 90% of the patients, compared with 92% of Other-GPs (P = 0.44). The preferred prescription was PPIs in 82% and 81% (P = 0.48), respectively, and H2-blockers were prescribed in 8% and 11%, respectively (P = 0.10). The prescription mode was different. The preferred prescription mode among ‘on demand’ ONE-GPs was 47% compared with 27% among Other-GPs (P < 0.0001). The comparable figures for intermittent prescription and daily prescription were: 23% vs. 40% (P < 0.0001) and 21% vs. 24% (P = 0.47). A total of 81% were ‘completely’ or ‘very’ satisfied with the management strategy by ONE-GPs, compared with 74% by Other-GPs (P < 0.001). The comparable figures for satisfaction with symptom control were 72% vs. 60% (P < 0.01; data not shown).
Table 5. Management strategies of reflux symptoms at first consultation according to subgroups in ONETWO
|H2-blocker||25 (4)||12 (2)||17 (2)||54 (8)||19 (4)||23 (5)||8 (2)||50 (11)|
|PPI||297 (43)||145 (21)||133 (19)||575 (82)||110 (23)||166 (35)||102 (22)||378 (81)|
|No prescription|| || || ||67 (10)|| || || ||38 (8)|
|Total||322 (47)||157 (23)||150 (21)||696 (100)||129 (27)||189 (40)||110 (24)||466 (100)|
Symptoms at baseline were dependent on participation in ONE (Table 6). ONE-patients reported significantly fewer days with heartburn compared with Other-patients: 67% vs. 51% had reported these symptoms for <6 days during the latest week before inclusion (P < 0.0001). The severity of heartburn was reported equal in ONE-patients and Other-patients: 16% reported mild and 84% moderate or severe heartburn. The severity of regurgitation did not differ, either: 54% of ONE-patients compared with 50% of Other-patients (P = 0.33). Epigastric pain was significantly less severe among ONE-patients compared with Other-patients: 58% of ONE-patients reported none or mild epigastric pain compared with 42% of Other-patients (P < 0.0001).
Table 6. Relation between duration of heartburn and severity of heartburn, regurgitation and epigastric pain in ONE-patients and Other-patients
|Severity of heartburn|
| 3–5||33 (13)||133 (54)||166 (67)||91 (10)||381 (41)||472 (51)|
| 6–7||7 (3)||74 (30)||81 (33)||48 (6)||400 (43)||448 (49)|
| Total||40 (16)||207 (84)||247 (100)||139 (16)||781 (84)||920 (100)|
|Severity of regurgitation|
| 3–5||99 (40)||67 (27)||166 (67)||272 (30)||200 (22)||472 (51)|
| 6–7||34 (14)||47 (19)||81 (33)||189 (21)||259 (28)||448 (49)|
| Total||133 (54)||114 (46)||247 (100)||461 (50)||459 (50)||920 (100)|
|Severity of epigastric pain|
| 3–5||103 (42)||63 (26)||166 (68)||229 (25)||243 (26)||472 (51)|
| 6–7||40 (16)||41 (16)||81 (32)||152 (17)||296 (32)||448 (49)|
| Total||143 (58)||104 (42)||247 (100)||381 (42)||539 (58)||920 (100)|
Patient's satisfaction with symptom control and management was highest in subgroup A and lowest in subgroup C (Table 2). A total of 205 of 247 ONE-patients (83%) were completely or very satisfied with the management, compared with 704 of 920 Other-patients (76%, P = 0.04). The comparable figures for satisfaction with symptom control were 77% (189 of 247) vs. 65% (599 of 920, P < 0.001, data not shown).
‘Good’ and ‘fair’ response to management was obtained by 190 of 247 (77%) ONE-patients compared with 616 of 920 (67%) Other-patients (P < 0.01, data not shown).
A total of 305 patients did not accomplish good or fair response. Still high numbers stated that they were satisfied with both the treatment and the management (Table 7). Among Other-patients 176 of 252 patients (70%) were completely or very satisfied with the management and 132 of 252 (52%) were completely or very satisfied with the symptom control, although they reported more than 2 days of symptoms in the last week before visit 2. The comparable figures among ONE-patients were: 70% and 51%.
Table 7. ONETWO patients, who did not accomplish response and their reported satisfaction with symptom control and management
|Completely satisfied||11 (21)||23 (43)||72 (29)||103 (41)|
|Very satisfied||16 (30)||14 (26)||60 (24)||73 (29)|
|Quite satisfied||11 (21)||13 (25)||54 (21)||42 (17)|
|Satisfied||7 (13)||3 (6)||38 (15)||29 (12)|
|Dissatisfied||7 (13)||0||21 (8)||3 (1)|
|Very dissatisfied||1 (2)||0||5 (2)||1 (<1)|
|Completely dissatisfied||0||0||2 (1)||0|
The management over 6 months resulted in significant decreased symptoms. In ONE-patients the average number of days with heartburn decreased from 4.8 to 1.2, compared with Other-patients from 5.3 to 1.3. The severity of heartburn, regurgitation and epigastric pain decreased comparable in both groups of patients.
This was accomplished by prescription of 98 Defined Daily Dose (DDD) of PPIs during 180 days, compared with 76 DDD of PPIs in Other-patients (data not shown).
Prescriptions by doctors can be influenced in various ways
In a nested case–control study from tertiary care7 it was demonstrated that request by doctors for a drug to be added to a hospital formulary was strongly associated with the doctors interaction with the companies manufacturing the drugs. In primary care ‘early experience of using new drugs’8,9 and ‘availability of drug samples’10 has been shown to influence future prescription habits, whereas influence of commercial sources of information has been shown to be associated with short-term use of drugs rather than long-term use.11 From the patient's perspective it has been demonstrated that patients who receive prescriptions for self-limiting conditions will be more likely to expect and receive a new prescription if the same symptoms occur.12
The influence of participation in a trial has recently been reported in a Danish study:13 GPs who participated in a asthma trial did not change their treatment, but they prescribed the sponsor's drug more often than GPs who did not participate in the trial.
There was two reasons for designing ONETWO study: One was to look at the impact of trial participation on both doctors and patients, and the other was to address the external validity of ONE, by comparing ONE with routine management in primary care, i.e. if H2-blockers were the preferred treatment in daily practice, the result from ONE would not be relevant for primary care.
General Practitioners with and GPs without experience from ONE were recruited by invitation. This procedure was not a true ‘randomization’, as it favoured GPs with an interest in trials. Formal matching of the two groups of GPs was not undertaken, but comparison of the total number of investigators did not reveal significant differences according to age and sex, type of practice and settling (Table 3). We consequently believe that ONE-GPs were otherwise comparable with Other-GPs.
ONETWO was designed exactly as ONE except no specified treatment was prompted. Non-pharmaceutical treatment, antacids or H2-blockers could be used parallel to PPIs.
Management of GERD
The study showed that independent on whether or not the GP had participated in ONE, nine of 10 patients were given a prescription at the first consultation. The prescribed drugs were H2-blockers or PPIs (antacids were not used) and PPIs were used in nine of 10 prescriptions with no significant difference between the GPs. This indicates that PPIs are the routine treatment of GERD symptoms in primary care in Denmark. ONE-GPs prescribed PPIs significantly more often as on-demand therapy, compared with Other-GPs, who tended to use treatment courses. There may be several reasons for that, i.e. the GPs had been introduced to on-demand therapy by participation in ONE and they felt comfortable with the treatment. However, as the use of on-demand treatment by ONE-GPs was higher in ONE-patients compared with Other-patients, on-demand treatment may not be used for the first time (to secure the diagnose, a full treatment course were given) or on-demand treatment was not seen as an understandable treatment for a ‘new’ patient. On the other hand, the more widespread use of on-demand treatment could be a result of the experienced patient's demand, they know that they do not need to take a full course to get rid of the symptoms.
The study has demonstrated a significant difference in baseline symptoms between ONE-patients and Other-patients. ONE-patients reported the same severity of heartburn, but for significantly lesser days and they reported less severe regurgitation and epigastric pain compared with Other-patients. It seems, though, that patients with knowledge of efficient treatment do not accept to be symptomatic, but seek to re-medicate more promptly. This finding is probably not limited to experience from a trial, as any successful treatment will be interpreted by the patient as a possibility to reduce symptoms promptly.
Between one-quarter and one-third of the patients were neither completely nor very satisfied. When the subgroups were compared, ONE-patients were satisfied to a higher degree, both regarding symptom control and management, which probably reflect a selection of patients: they were treated with success in the past and they seek the same treatment again. Of Other-patients, one-third was less than very satisfied. It has not been possible to analyse this finding according to treatment, due to the overall high numbers of patients treated with PPIs.
Overall patients reported higher satisfaction with management (the way the GP handled the patient) than with symptom control (the decrease in symptoms). The difference was 7% in ONE-patients and 11% in Other-patients. This indicates problems with ‘satisfaction’ as an end point in trials.
Response to treatment
Response to treatment, as defined in this study, differed from patient's satisfaction in two aspects: it is not a subjective statement, but an objective report of number of days with symptoms, and it does not account for a 6-month period, but for the last week. The overall tendency is that good and fair response is reported with lower figures than the 6 months satisfaction, indicating that even if symptoms are present at a certain point in time, the patient may be satisfied with the treatment. However, it is remarkable that half of the patients, who do not accomplish an ‘objective’ response reports to be completely satisfied or very satisfied. No difference was found between ONE-patients and Other-patients, indicating that experience does not change the gap between a patient's subjective feeling and an objective measure.
Use of medication
The overall use of PPIs was 1 DDD each second to third day. This has been found in other studies3 as well. In this study, ONE-patients had a higher use of PPIs compared with Other-patients. This may be due to the fact that ONE-patients who participated in ONETWO have demonstrated that their symptoms are recurrent, in contrast to Other-patients in ONETWO, who may or may not experience a relapse in the future.
Participation in a clinical trial influences both doctors and patients. Experienced GPs import treatment modalities from the trial into the daily clinic and experienced patients seek health care for fewer symptoms. In the present study, the consequence was increasing use of on-demand treatment with PPIs, resulting in increased consumption. Patient satisfaction as an end point was not consistent with objective measures.
AstraZeneca, Denmark has supported the study.