Discussion
The interpretation of clinical studies undertaken to investigate the efficacy of therapy is often constrained by the fact that the typical study setting differs from conditions in ordinary clinical practice. GORD studies undertaken on patients with oesophagitis which have used 'relief of heartburn' as the principal symptom outcome fail to take into account two important facts now widely acknowledged: first, that many GORD patients do not have oesophagitis and second, that the symptom burden experienced by GORD patients is much more complex than heartburn alone[
31]. In addition, despite differences in healthcare systems it is now generally accepted that the use of PPIs to treat GORD is not conditional on first obtaining an upper GI endoscopy. The diagnosis is usually made and treatment started on the basis of clinical history, very often in a primary care setting.
To obtain information relevant to this clinical situation, our study was therefore undertaken with a pragmatic rather than explanatory trial design and patients were enrolled in the trial on the basis of a clinical history of GORD. In explanatory trials, participating patients are typically enrolled on the basis of precisely defined symptom criteria, perhaps with positive results from investigations, rather than on the basis of the way diagnoses are made in everyday clinical practice. The relevance of findings made in explanatory trials to everyday clinical practice may therefore be impaired because the patient population being studied is not the population being treated in ordinary practice[
10,
11]. Classical explanatory trials of PPI therapy in GORD have mostly overestimated the success of treatment in comparison with the therapeutic outcomes achieved in ordinary clinical practice[
32].
A diagnosis of GORD made on the basis of the clinical history has specificity of around 65%[
33,
34], implying that misdiagnosis is not infrequent and this will certainly be responsible for some instances of poor treatment responses, sometimes called 'PPI failure'[
35,
36]. Our focus in this study was not to identify causes of PPI failure, but rather to address questions that are often at the forefront of a physician's mind when initiating PPI therapy for GORD. We selected features identifiable at the time of starting treatment that would possibly have influence on therapeutic outcome and then prospectively tested whether they did so. These features were: age, body mass index (BMI), gender, geographical location,
H. pylori status (determined serologically), symptoms suggesting IBS as well as GORD, presence of oesophagitis before treatment, the grade of oesophagitis, if present, and anxiety and depression. Response to treatment was defined in terms of symptom relief, because in clinical practice it is not usual to perform endoscopy to assess healing of any pre-treatment oesophagitis unless troublesome symptoms persist on treatment or alarm features appear.
The patient self-administered ReQuest™ questionnaire used in this study is a validated, systematic and comprehensive assessment of frequency and severity of symptoms shown to be GORD-related[
18‐
20]. Response to treatment is defined as a decrease in symptom score to one equal to or below the score that has been identified in healthy individuals[
21].
Patients with NERD have been found in previous studies to respond to PPI treatment less well than those with ERD[
37‐
39]. This is usually attributed to inclusion of some patients in the NERD group who do not have reflux disease, notably patients with 'functional heartburn', who are expected to respond poorly to acid suppression. In line with these findings, our data show a better response rate in patients with ERD than in NERD. In addition, our patients with ERD showed no significant differences in response rates according to grade of oesophagitis. If NERD patients with 'true' reflux disease were to respond similarly to those with ERD, these observations suggest a functional heartburn prevalence of around 4% among the patients recruited to this study.
Previously published studies are inconsistent regarding the existence of a gender difference in response to PPI treatment[
26,
39]. Our findings in the univariate analysis indicate that males show a better response than females. Of course, NERD is more common in females and so the lower response rates illustrated in Figure
1 may be partly explained by confounding. However, as described above, the p-values shown in Table
3 were determined from logistic regression and relate to each factor considered independently.
IBS is known to be more frequent in patients with GORD than in control populations[
40‐
42] and is also more prevalent in females[
43‐
45]. In our patients, concurrent IBS symptoms were associated with a lower response rate to the PPI when compared with patients without concurrent IBS symptoms. There are uncertainties about the interpretation of this observation, however. It may be supposed that the inclusion of lower abdominal complaints in the spectrum of symptoms assessed by the ReQuest™-GI questionnaire will inevitably imply a higher rate of symptom 'non-response' if these lower abdominal symptoms are unaffected by acid suppression. Surprisingly, though, previous studies using ReQuest™ have found that the lower abdominal symptoms in GORD patients do improve during PPI treatment[
20,
46‐
48] suggesting that these symptoms are not independent. Perhaps the mechanisms underlying classical reflux symptoms and the lower abdominal symptoms are somehow linked or the symptoms are linked within the patients' perceptions of overall symptom burden.
Many physicians would predict that patients with concurrent anxiety or depression would respond poorly to GORD treatment and previous studies have borne this out[
26,
49]. Our data likewise show that a high HADS score before treatment was associated with a poorer response, although anxiety seemed to be more relevant than depression. Nevertheless, the anxiety sub-score fell to normal in the patients who responded to treatment, consistent with a contention that GORD symptoms were causing some of the anxiety while the higher anxiety sub-scores seen in non-responders than responders, especially at baseline, are consistent with the possibility that anxiety may itself contribute to the perceived GI symptom burden. The possibility thus exists of a two way relationship between anxiety and GORD symptoms.
Our data showing an association between BMI and response rate was statistically highly significant and this observation was not expected. Although published data clearly link obesity and GORD[
50‐
52], and indeed a correlation between BMI and GORD is also evident within the normal BMI range[
53], there has been little investigation of the effect of BMI on responses to antisecretory treatment[
54,
55]. However, one recent study found a positive correlation between BMI and symptom relief when lansoprazole therapy was given to patients with a wide range of upper GI symptoms and negative upper GI endoscopy[
56]. The authors suggested their findings could be explained by an association of higher BMI with GORD and a good response of acid reflux symptoms to treatment; this explanation could be relevant to our findings also.
An aspect of our results worth comment was the failure to find any link between age or
H. pylori status on response rates. The literature is inconclusive with respect to
H. pylori: Hatlebakk et al.[
57] found
H. pylori infection to have no effect on response to treatment whereas Labenz et al.[
58] found greater age and a positive
H. pylori status were both positively associated with better resolution of heartburn over a 4-week treatment period. The failure to find a link between
H. pylori status and response may have resulted from the use of the serology for
H. pylori diagnosis, which does not distinguish between cured and active infection[
16].
Impaired quality of life is well-established as a consequence of GORD and improves with treatment[
59‐
61]. Our results show that the baseline GERDyzer™ score was higher (indicating greater quality of life impairment) in the patients who subsequently responded poorly to treatment and that these non-responders also showed a smaller absolute improvement (fall in score) during treatment.
It is interesting to compare the symptom control determined by the ReQuest™-GI questionnaire with the investigators' assessments of symptom control and with the patients' own assessments of satisfaction with symptom control. The biggest disparity, as shown in Figure
5, was in the proportion of patients judged by the physician to have been 'well-controlled' who nevertheless were non-responders according to ReQuest™-GI. This suggests physicians tend to overestimate patients' responses to treatment, as has been previously reported[
62,
63]. The 'satisfaction' findings are also of interest, however, in that some 14% of patients who rated themselves to be 'very satisfied' with symptom control were 'non-responders' according to ReQuest™-GI. Patients may thus be satisfied with symptom control that is less than total, which raises an issue about whether attainment of a predefined level of symptom suppression or a 'very-satisfied' patient should be the objective of therapy. It may certainly be argued that GORD has been adequately treated if the residual symptoms a patient still has while on treatment are judged by that patient not to be troublesome. Moreover, in ordinary clinical practice it is to be expected that a patient's satisfaction with symptom control will influence the physician's assessment: a very satisfied patient may reasonably lead the doctor to consider that the GORD symptoms are 'well-controlled'. Nevertheless, the discrepancies between these different measures of treatment success illustrate the difficulties of measuring the occurrence and impact of symptoms and of establishing unequivocally what the most clinically relevant treatment goal should be. Of course, the dilemma is well recognised in other areas of clinical medicine also[
64].
Geographical and ethnic differences in the nature, prevalence and presentation of GORD are recognised[
65] but less is known about possible differences in response to treatment. Unfortunately, it is not clear whether the geographical variation in responses is a reflection of fundamental differences in GORD biology or an artefact of the study. Minor differences in local systems of patient recruitment, for example, may have had an effect on patient outcomes. Consequently, we do not wish to suggest any interpretation of the observed geographical variation.
Prompted by reports that treatment outcome may be predicted by symptom severity before or shortly after commencing treatment [
25‐
29], the potential for the ReQuest™ questionnaire to predict the response to treatment was comprehensively evaluated and the results show that some predictive capability is possible both in respect of response and non-response (Table
4). The results that relate to the baseline are especially interesting in that a prediction of treatment outcome before the treatment begins could potentially help the physician in managing patient expectation. However, the probabilities of correct prediction shown in Figure
4 are perhaps lower than physicians would wish and the proportion of patients for whom no prediction could be made was higher. However, using an abbreviated form of ReQuest™, (ReQuest in Practice™)[
66], scores obtained from the questionnaire were more accurate than the physicians' conventional clinical enquiry in identifying patients whose symptoms would still be controlled after stepping down from full dose to half dose PPI[
67]. In this context at least, therefore, prediction based on the systematic assessment of symptom burden is potentially valuable, being more reliable than ordinary clinical judgment.
Acknowledgements
This study was funded in full by Nycomed GmbH. Data analyses were undertaken by Pierrel Research Europe GmbH who received funding from Nycomed GmbH. The writing and preparation of this paper was funded in full by Nycomed GmbH. Writing support was provided by Stefan Loth and Dr. Ilona Vogt-Humberg (Pierrel Research Europe GmbH). The authors would like to thank all investigators who contributed to patient enrolment and study conductance at each centre. The participating centres were:
C.H.U. de Liege - Hopital Sart Tilman, U.L.B. Clin. Univ. Erasme, V.U.B. - Universitair Ziekenhuis, Fachambulatorium der GKK Graz, Hanusch-Krankenhaus, Clinique St. Luc, Cantonal Hospital Liestal, Practice Dr. Meier-Gräub, Practice Dr. Roten, Hopital Brugmann, Hospital Saint-Pierre, Practice Dr. Italo Fumagalli, Centro Riferimento Oncologico, Universita di Genova, KFJ-Spital, U.I.A. Univers. Ziekenhuis, K.U.L. U.Z. Gasthuisberg, Hospital Italiano, Hospital Gral de Agudos Dr.C.G. Durand, Universidade de Sao Paulo-Ribeirao Preto, Praxis Baumann, Praxis Meinecke, Clínica Puerta de Hierro, Clinica Universitaria de Navarra, Landeskrankenhaus Deutschlandsberg, Southey Green Medical Centre, Vergelegen Clinic, Hospital ABC, Medicina Privada, Hospital Universitário Clementino Fraga, Hospital das Clønicas, Hospital de Clønicas, Algemeen Ziekenhuis Maria Middelares, Hospital "La Paz", Hospital General de Castellón, Dr. Bayol, Dr. Bord, Dr. Cazals, Dr. Ghedighian, Dr. Colonna, Royal Victoria Hospital, Clinique de Gastro Enterologie, Hospital San Lorenzo de el Escorial, Villa Medica, Praxis Dr. Margulies, Klinika Gastroenterologii, St. Michael's Hospital, Praxis Lerche, Ospedale Civile, Praxis Dr. Hans-Joachim Ulmer, Dr. Arramon-Tucoo, Dr. Maignan, Dr. Salas, Klinika Gastroenterologii, KH Krems, North London Clinical Studies Centre, Practice Meyer, Practice Dr. Alain Aisene, Castemilk Health Centre, Praxis Dr. Stadler, Centro Medico Teknon, Saint Augustine Medical Centre 2, Panorama Medi-Clinic, Dr. Christa Wutschitz, Dr. Schöngut, Dr. Julio Otero, Policlinico di Bari, Hosp. d. Clinicas da Univ. de Sao Paulo, Prince of Wales Hospital, University of Malaysiam, Irmandade Santa Casa de Misericórdia-RS, Centro Medico Coyoacan, Hospital Angeles del Pedregal, Dra. Ligia Grau Cobos, Dra. Alma-Rosa Zarate Negrete, Hopital Avicenne, A. Z. Damiaan Campus St. Joseph, AZ Jan Palfijn, Praxis Dr. Ursula Frank, Hospit. General de Mexico, Dr. Guillermo Arceo Perez, Dr. Luis-Humberto Lopez Salazar, Ziekenhuis Groeninge/St. Niklaaskliniek, Universita di Roma Tor Vergata, Praxis Dr. Karl Adenauer, Hosp. de Clin. Univ. Fed. Parana, St. George's Medical Suite, University of the Free State, Fundacion Hospital da Barbanza Parroquia de Oleiros, Hospital Ntra. Sra. deValme, Dr. Jacques Diaz, Klinika Chorob Przewodu Pokarmowego, Dr. Pluta, Kloof Hospital, St. James Surgery, Bradford Road Medical Centre, Asan Medical Center, Gastroenterology Center Louis Leipoldt, Clinica IPENSA, CEMIC, Vadamalayan Hospitals, LTMG Hospital, Vikram Hospital & Heart Care, Gastroenterology & Endoscopy Center, Panorama Medi-Clinic, University Hospital Basel, Praxis Dr. Gesine Avenarius, Layton Medical Centre, National Cheng Kung University Hospital, Taichung Veterans General Hospital, Faculdade de Medicina de Botucatu, Praxis PD Dr. Schmidt, Praxis Dr. Bank, Praxis Dr. Botzler, Praxis Dr. Deuster, Instituto Universitario Dexeus, Dr. Protte, Queen Elizabeth Hospital, Klinika Gastroenterologii Wewnetrznych, Klinika Gastroent. Przemiany Materii, Azienda Ospedaliera di Parma, Istituto Clinico Humanitas, Azienda Ospedaliera Garibaldi, Praxis Dr. Wunderlich, Praxis Dr. Müller, Asian Institute of Gastroenterology, SMS Medical College & Hospital, Deenanath Mangeshkar Hospital, PSG Institute of Med. Sciences& Research, Kangnam St. Mary's Hospital, Hospital das Clinicas de Porto Alegre, Tri-Service General Hospital, McGill University Health Centre, London Health Sciences Centre, Department of Medicine of Community Health Sciences, St. Joseph's Healthcare Hamilton, Ctr. Hosp Affilie-St.-Sacrament, Nepean Hospital, Royal Melbourne Hospital, Alfred Hospital, St Vincents Hospital, Western Hospital, Flinders Medical Centre, Penisula Clinical Research Centre, Fremantle Hospital, GEDyT, Centro Integral de Gastroenterología, Sanatorio Otamendi, Singapore General Hospital, Dr. Rahman Bacchus, Toronto Digestive Diseases Associates, Gastroenterology Center Quebec, Unigastro, Springer, Toronto Digestive Association, Health Sciences Centre, Campus St. Vincentius, St. Nikolaus Hospital, Grant Medical College & Sir JJ Group Hos, Poona Hospital & Research Centre, Sanatorio Julio Mendez, Grupo M.dico Las Lomas, Asoka Plaza Clinic, Seoul National University Hospital, Yongdong Severance Hospital, Samsung Medical Center, South Eastern Gastroenterology
Competing interests
RCH has served as a speaker, a consultant and an advisory board member for Nycomed GmbH (former ALTANA Pharma) and for Reckitt Benckiser. He owns ordinary shares in Novartis, Reckitt Benckiser and Proctor & Gamble.
HM served as a speaker, a consultant and an advisory board member for Nycomed GmbH (formerly ALTANA Pharma), Reckitt Benckiser, Steigerwald, Novartis and Movetis.
HS and AT are employees of Nycomed GmbH and Nycomed International Management GmbH.
Authors' contributions
RCH, HM, AT and HS were all involved in the study concept and design, analysis and interpretation of data, and the critical revision of the manuscript for important intellectual content; all provided approval of the final manuscript. Additionally, RCH took a lead role in drafting of the manuscript. HS provided administrative, technical, and material support, and were responsible for study supervision.