Validation of the gastrointestinal symptom score for the assessment of symptoms in patients with functional dyspepsia

Transcription

1 Aliment Pharmacol Ther 2005; 22: doi: /j x Validation of the gastrointestinal symptom score for the assessment of symptoms in patients with functional dyspepsia B. ADAM*, T. LIEBREGTS*, K. SAADAT-GILANI, B.VINSONà & G. HOLTMANN* *Department of Gastroenterology, Hepatology and General Internal Medicine, Royal Adelaide Hospital, University of Adelaide, Adelaide, SA, Australia; Division of Gastroenterology and Hepatology, University Hospital, Essen, Germany; àresearch Department, Steigerwald Arzneimittelwerk GmbH, Darmstadt, Germany Accepted for publication 13 May 2005 SUMMARY Aim: To validate the gastrointestinal symptom score as an outcome measure for functional dyspepsia. Methods: In focus groups, 10 dyspepsia-specific items including nausea, sickness, vomiting, bloating, abdominal cramps, early satiety, acidic eructation/heartburn, loss of appetite, retrosternal discomfort, epigastric pain/ upper abdominal pain were identified. Ninety-five patients with functional dyspepsia and 56 healthy controls were recruited and responsiveness evaluated by analysing gastrointestinal symptom score data from 357 patients from previous placebo-controlled trials. Gastrointestinal symptom score response data were correlated with the patient s global assessments of efficacy. Convergent validity was assessed by correlating the gastrointestinal symptom score with the results obtained by the Nepean Dyspepsia Index. Results: Sensitivity: In patients and healthy controls gastrointestinal symptom score yielded consistently different scores (all P < ). Test retest reliability: Gastrointestinal symptom score determined at the two time points were significantly correlated (r-values ranging from to 0.901). Convergence validity: Gastrointestinal symptom score of both rating groups were significantly correlated with the symptom-specific component of the Nepean Dyspepsia Index (r-vales ranging from to 0.764, P < 0.01). Responsiveness: Responses of gastrointestinal symptom score during treatment were different for patients with a global self assessment as responders compared with non-responders (all P < ). Conclusion: The gastrointestinal symptom score is a valid and reliable instrument to assess symptom intensities in patients with functional dyspepsia. INTRODUCTION Traditional disease concepts are based on the presence of structural lesions or biochemical abnormalities that are hallmarks of a specific disorder or closely linked to the clinical manifestation of the diseases. Interestingly, Correspondence to: Prof. G. Holtmann, Department of Gastroenterology, Hepatology and General Internal Medicine, Royal Adelaide Hospital, University of Adelaide, North Terrace, Adelaide, South Australia 5000, Australia. gerald.holtmann@adelaide.edu.au utilizing clinically available diagnostic measures, structural lesions or biochemical abnormalities are only found in a proportion of patients presenting for evaluation and treatment. This in particular holds true for patients with chronic or relapsing abdominal symptoms. In more than 50% of these patients, diagnostic measures do not identify a cause of symptoms. 1 In the absence of structural lesions, these disorders are addressed as functional disorders. In this context a patient with chronic or relapsing upper abdominal symptoms (symptoms centred between the navel and the lower end of the sternum) is referred to as suffering from functional dyspepsia. 2 Ó 2005 Blackwell Publishing Ltd 357

2 358 B. ADAM et al. In the absence of structural lesions that can be used to guide therapy and determine the response to therapy, the systematic assessment of symptoms is an important issue and there is the need to validate these instruments. 3, 4 Thus this study aimed at validating a previously developed instrument for the assessment of dyspeptic symptoms. MATERIALS AND METHODS Development of the gastrointestinal symptom score (GIS) In panel discussions and interviews with patients and gastroenterologists, items characteristic of functional dyspepsia were identified. Six gastroenterologists thus conducted focus group discussion with three to five patients with functional dyspepsia and symptoms identified that were considered by these patients as relevant. These symptoms were listed and symptoms identified that were reported by more than two investigators. The 10 items finally included into the instrument (Table 1) were indeed independently reported by the majority of investigators. The 10 score items identified during the development process included nausea, sickness, vomiting, bloating, abdominal cramps, early satiety, acidic eructation/ heartburn, loss of appetite, retrosternal discomfort and epigastric pain/upper abdominal pain. In a face to face interview patients were asked by the doctor to rate the intensity of each individual symptom on a validated 5-point Likert scale (0 ¼ no problem, 1 ¼ mild problem, 2 ¼ moderate problem, 3 ¼ severe problem and 4 ¼ very severe problem). This method has been found to meet the key criteria reproducibility, responsiveness to change and validity compared with corroborating measures. The sum score, which ranges from 0 to 40 points with higher scores representing severe disease manifestations, were utilized in several clinical trials as a measure for the severity of functional dyspepsia at the start and end of therapy for calculating treatment effects. The instrument was developed for administration by a doctor (or a trained health professional). In addition, data available from clinical trials are summarized in this article. Patients Prior to study entry all patients and volunteers had to give informed consent concerning participation in the trial and the use of the data. The study protocol was approved by the Ethics Committee of the University of Essen, Germany. For patients with a diagnosis of functional dyspepsia, appropriate investigations [including upper endoscopy, colonoscopy in all patients above 50 years of age and all patients with concomitant irritable bowel syndrome (IBS) symptoms] were required within the last 9 months to exclude structural lesions as the cause of symptoms. Any patient with chronic peptic ulcer disease or reflux oesophagitis, erosive duodenitis, gastrointestinal bleeding or cancer was excluded. Diagnosis was made consistent with the Rome II criteria, however, symptoms of gastro-oesophageal reflux disease (GERD) or IBS were Table 1. Items included in the GIS Definition Nausea Vomiting Bloating Abdominal cramps Early satiety Acid eructation/heartburn Sickness Loss of appetite Retrosternal discomfort Epigastric or upper abdominal pain Urgent feeling of need to vomit but vomit does not actually occur Vomiting of mucus and gastric contents or strong unproductive retching Feeling of congestion of food without relation to prior food intake which could explain this feeling Spasmodic or colic-like stomach pain without specified localization Feeling that the stomach is overfilled soon after starting to rest, unproportional to the quantity of food taken, so that the meal cannot be finished Belching with acid taste, burning sensation in the oesophagus Discomfort combined with the impression for the need to vomit Listless for food intake Unpleasant feeling behind the sternum, painful or drawing Pain localized between costal arches, below the sternum/pain localized in the upper abdomen GIS, gastrointestinal symptom score.

3 VALIDATION OF THE GASTROINTESTINAL SYMPTOM SCORE 359 allowed if these symptoms did not dominate the clinical picture and did not necessitate specific treatment. At the time of recruitment, all patients were required to have symptoms and were required to have at least one additional disease episode during the last 12 weeks. All patients with relevant psychiatric comorbidities that interfered with the aims of the study or the inability to understand this trial or being unable to give informed consent were excluded. For the assessment of responsiveness, GIS data obtained in patients with functional dyspepsia during recent trials 5 8 were used. Validity For this purpose, item sensitivity, test retest reliability and responsiveness of the score as well as convergent validity were examined. In this context, sensitivity is the ability of the questionnaire to differentiate between groups with different health status, i.e. patients with functional dyspepsia and healthy controls. It was predicted that patients with functional dyspepsia would show higher scores compared with healthy controls. To determine sensitivity the rater had to assess after the interview whether the person he interviewed was a patient with functional dyspepsia or a healthy control. Results were correlated with source data of the interviewed person. Test retest reliability This refers to the stability of the measures over time and concerns with the extent to which measurements are repeatable and consistently produce the same results. Because the instrument was administered by an external rater two aspects of reliability, i.e. intrarater and interrater reliability, had to be assessed. For testing intrarater reliability the interview was repeated 1 week later by the same rater who was blinded for the dyspeptic score of the first interview. For testing interrater reliability the persons were interviewed by one rater and the interview was repeated on a separate occasion by a second rater who was blinded for the score of the first interview. The different specialist groups were general practitioners (rating group A) and gastroenterologists/internal specialists (rating group B). The study design is summarized in Figure 1. Day 0 Day 1 Day 7 Intrarater-reliability Figure 1. Study design. Responsiveness Responsiveness is the ability of the questionnaire to detect clinically relevant changes of disease-specific parameters over time as an aspect of outcome. In this study responsiveness was examined using pooled data of 357 patients from former randomized, placebo-controlled, double-blind therapeutic trials with the herbal preparation STW 5. Responsiveness was determined by comparing GIS data at baseline and after 4 weeks of therapy (DGIS). These data were correlated with the global assessments of efficacy by the patients at the end of therapy. Depending on their global assessment on a 6-point Likert scale, patients were assigned to a responder group (assessed therapy as good or very good ) and a non-responder group (assessed therapy effect as moderate, acceptable, poor or very poor ). Differences in the baseline-adjusted GIS data for responders and non-responders at the end of therapy were compared. Convergent validity Screening examination Patients or healthy controls Interview by rater A and B Inclusion / exclusion criteria Randomly assignement of identification number Interrater-reliability NDI score Patients or healthy controls Interview by rater A and rater B Interrater-reliability Randomly assignement to interview termins Convergent validity was determined by comparing the results of the GIS with the similar subscale score (domain: symptom) of the disease-specific self-report instrument for quality of life, the Nepean Dyspepsia Index (NDI) 9 which was administered in a validated German version. The criteria for convergent validity were fulfilled when scale scores for related concepts show Pearson correlation coefficients >0.4.

4 360 B. ADAM et al. Statistics All demographic data were analysed in a descriptive manner. Sensitivity of the GIS was assessed utilizing analysis of variance for repeated measures and Kendall s K-statistic. Test retest reliability as well as correlation of GIS with the NDI scale (convergent validity) were analysed with correlation coefficient of Pearson and Spearman s rank correlation. Responsiveness was analysed using a covariance analysis model with the factor response and the covariate baseline. In addition, factor analysis was performed to determine the factorial structure of the questionnaire. All statistical analyses were performed utilizing Statistical Analysis System (Version 8.1). Based upon previous data 10, 11 a target population of 90 patients and 45 controls was considered appropriate for the necessary data analysis. RESULTS Patients A total of 151 people were enrolled into this study. Data of 95 patients and 56 healthy controls were eligible for the final analysis. Mean age of the patients and controls was 53.6 ± 14.0 and 39.6 ± 14.8 years respectively. Exactly 52.6% of the patients and 67.9% of the controls were female. Mean duration of the disease was 22 months. With variance analytic methods, we confirmed that the group differences in age and sex had no significant influence on the results. Sensitivity. Gastrointestinal symptom score was significantly higher in patients with dyspepsia than in healthy controls (Table 2). The rater was asked after the interview to globally assess whether the person is a patient or a control. Correlation coefficients of scores and external assessment (Kendall s K-statistic) ranged from to and was highly significant for all groups (P < ) (Table 2). Test retest reliability. Sum scores of GIS determined by the two rating groups and at both rating times (Table 2) were highly correlated. Correlation coefficients for both rating groups were at rating time 1 and week later at rating time 2. Correlation coefficients for both rating times were Table 2. Sensitivity: mean GIS of patients and controls for rating group A and for rating group B. The Spearman rank correlation coefficients were significant for all (P < ). Responsiveness. Data of 357 patients from four previous trials 5 8 were eligible for analysis. The data from four treatment trials are summarized in Table 3. The baselineadjusted, post-treatment GIS values were significantly lower among treatment responders compared with nonresponders, with the difference between groups ranging from 4.5 to 6.7 points (P < to P < ). Convergent validity Patients (n ¼ 95), mean GIS (points ± s.d.) Controls (n ¼ 56), mean GIS (points ± s.d.) Sensitivity: comparison of patients and controls Rating group A Rating time ± 6.9* 2.1 ± 3.4 Rating time ± 6.4* 1.8 ± 3.1 Rating group B Rating time ± 6.4* 2.3 ± 3.3 Rating time ± 5.6* 1.6 ± 2.8 Sensitivity: correlation of GIS with the source data Rating group A Rating time * Rating time * Rating group B Rating time * Rating time * Correlation coefficient Gastrointestinal symptom score (GIS) values were significantly higher in patients for both measurements and for both groups (all * P < for the comparison of patients and controls or the correlation coefficient between GIS and source data). The symptom-specific scale of the NDI showed a high correlation with the results of the GIS at both rating times and for both rating groups. The mean score of the NDI symptom domain was 60.2 (±39.5) for the patients and 12.9 (±16.9) for the controls. Correlation coefficients of both scores were and for rating group A and and for rating group B, and were significant (P < ) for all (Figure 2). These results confirm that the GIS as an external rating score correlates highly with the symptom domain of the NDI.

5 VALIDATION OF THE GASTROINTESTINAL SYMPTOM SCORE 361 Table 3. Responsiveness: baseline-adjusted GIS differences for responders and nonresponders in clinical trials with STW 5 and upper and lower confidence intervals for baseline adjusted GIS differences of responders and non-responders after therapy Study code Patients (n) Baseline-adjusted GIS difference R vs. NR (points) 95% CI P-value Madisch et al < Melzer et al < Rosch et al < STW 5/311 D 02 II-S (dof) < dof, data on file; s.d., standard deviation; R, responder; NR, non-responder. Factor analysis. Factor analysis identified four independent factors (factor load): factor 1: nausea (0.61), vomiting (0.78), sickness (0.611); factor 2: early satiety (0.64), bloating (0.48), loss of appetite (0.57); factor 3: retrosternal discomfort (0.58), acid eradication/heartburn (0.66); and factor 4: epigastric/upper abdominal pain (0.48), abdominal cramps (0.48). Subgroup analysis. To ensure that the results of this study were not biased by enrolment of patients with endoscopy-negative GERD, a subgroup analysis excluding all patients with acid-related symptoms or heartburn was conducted (n ¼ 35). This did not change any of the above findings. DISCUSSION The results of this study can be summarized as follows. The 10-item GIS is a valid and reliable instrument for the assessment of a broad spectrum of symptoms in NDI score (points) Patients Healthy controls GIS score (points) Figure 2. Correlation of the gastrointestinal symptom score (GIS) and the Nepean Dyspepsia Index (NDI) symptom score (r ¼ 0.887; P < ). patients with functional dyspepsia. Our data confirm that the GIS is a valid tool for detecting response to therapy in patients with functional dyspepsia. Besides the test characteristics, this instrument is compared with other tools extremely efficient and time saving when used in clinical trials to precisely assess response to therapy. It usually needs no more than 2 3 min to precisely assess the symptom severity utilizing the GIS. In contrast, for the NDI min were needed. The NDI, however, also covers aspects of health-related quality of life. Considering the ease of use, the GIS is not only an instrument to assess the response to treatment in clinical trials it can also be used as a tool to monitor treatment effects in clinical practice. Compared with more global response ratings (e.g. sufficient improvement of symptoms), this instrument allows to precisely assess various domains of improvements. While the factor analysis revealed four independent factors (that could be labelled nausea, early satiety, upper abdominal pain and GERD), this opens the opportunity to assess separately various domains. However to simplify the assessment of a global response the sum score can be used without any further adjustments. Thus a relatively broad spectrum of symptoms is covered by GIS. However, it might be argued that a symptom such as regurgitation is being included. While regurgitation is believed to be more indicative of GERD, there is a considerable overlap of GERD and functional dyspepsia. Consequently, it can be considered advantageous to include at least one item that addresses this issue. In the present study, a separate analysis was conducted in patients with functional dyspepsia without any GERD symptoms. The performance of the GIS was identical in this subgroup. Therefore, it cannot be argued that the instrument simply measures or our results might be due to a response of GERD symptoms. The tool was specifically designed to assess symptoms in patients with functional dyspepsia. Thus all validation

6 362 B. ADAM et al. was carried out in these patients. It might be argued that the instrument might also properly assess symptoms in patients with uninvestigated dyspepsia as a majority of these patients actually suffer from functional dyspepsia. However, data on the performance of this instrument in uninvestigated dyspepsia are scarce. It is beyond doubt that functional dyspepsia is characterized by pain or discomfort centred between the navel and the lower end of the sternum. However, these symptoms considerably overlap with GERD and IBS symptoms. 12, 13 In particular, the overlap with GERD is sometimes striking. 14 While from the perspective of definition GERD with an increased exposure of oesophageal mucosa with acidic gastric content represents a completely distinct entity, the underlying disease concepts might share some mechanisms. In functional dyspepsia there is the widespread belief that delayed gastric emptying plays a central role in the manifestation of symptoms in at least a subgroup of patients. 15 Delayed gastric emptying can also be found in a considerable proportion of patients 16 and is considered an important factor for the manifestation of symptoms. Thus excluding all symptoms of GERD may not be advisable in order to quantitate the effects of treatment on the relatively broad spectrum of symptoms found in patients with functional dyspepsia. Regardless of these considerations, GIS allows to assess the response of specific items. This might be considered a major advantage compared with more global outcome assessment that simply addresses global improvement. In summary, based upon our data, the GIS is a valid and reliable instrument for the assessment of dyspeptic symptoms that is suitable for the quantification of symptom severity in clinical trials as well for the standardized assessment of symptoms in the clinical setting. ACKNOWLEDGEMENTS The development of the GIS was initiated and funded by Steigerwald Arzneimittelwerk GmbH in close co-operation with a panel of specialists (Gastroenterology and Psychology, Juergen Hotz, Wolfgang Roesch, Juergen Neuser, Richard Raedsch, Harald Goebell) who gave input during various stages of development. The studies used for the testing of responsiveness were funded by Steigerwald Arzneimittelwerk GmbH. The authors appreciate the co-operation of CRO EPA Europharma and in particular of Uwe Hehnke for his expert statistical advice with the data analysis. Steigerwald Arzneimittelwerk holds the copyright of the GIS. The instrument is available free of charge for non-commercial use (i.e. in the clinical setting or for academic research purposes). REFERENCES 1 Talley NJ, Silverstein MD, Agréus L, Nyrén O, Sonnenberg A, Holtmann G. Evaluation of dyspepsia. Gastroenterology 1998; 114: Talley NJ, Stanghellini V, Heading RC, Kock KL, Malagelada JR, Tytgat GN. Functional gastroduodenal disorders. Gut 1999; 45 Suppl 2: II Van Zanten SJOV, Cleary C, Talley NJ, et al. Drug treatment of functional dyspepsia: a systematic analysis of trial methodology with recommendations for design of future trials. Am J Gastroenterol 1996; 91: Veldhuyzen van Zanten SJ, Talley NJ, Bytzer P, Klein KB, Whorwell PJ, Zinsmeister AR. Design of treatment trials for functional gastrointestinal disorders. Gut 1999; 45 (Suppl. 2): Madisch A, Holtmann G, Mayr G, Vinson B, Hotz J. Treatment of functional dyspepsia with a herbal preparation. A doubleblind, randomized, placebo-controlled, multicenter trial. Digestion 2004; 69: Rosch W, Vinson B, Sassin I. A randomised clinical trial comparing the efficacy of a herbal preparation STW 5 with the prokinetic drug cisapride in patients with dysmotility type of functional dyspepsia. Z Gastroenterol 2002; 40: Madisch A, Melderis H, Mayr G, Sassin I, Hotz J. A plant extract and its modified preparation in functional dyspepsia. Results of a double-blind placebo controlled comparative study. Z Gastroenterol 2001; 39: Melzer J, Rosch W, Reichling J, Brignoli R, Saller R. Metaanalysis: phytotherapy of functional dyspepsia with the herbal drug preparation STW 5 (Iberogast). Aliment Pharmacol Ther 2004; 20: Talley NJ, Haque M, Wyeth JW, et al. Development of a new dyspepsia impact scale: the Nepean Dyspepsia Index. Aliment Pharmacol Ther 1999; 13: Junghard O, Lauritsen K, Talley NJ, Wiklund IK. Validation of seven graded diary cards for severity of dyspeptic symptoms in patients with non ulcer dyspepsia. Eur J Surg Suppl 1998; 583: Leidy NK, Farup C, Rentz AM, Ganoczy D, Koch KL. Patientbased assessment in dyspepsia: development and validation of Dyspepsia Symptom Severity Index (DSSI). Dig Dis Sci 2000; 45: Agreus L, Svärdsudd K, Nyren O, Tibblin G. Irritable bowel syndrome and dyspepsia in the general population: overlap and lack of stability over time. Gastroenterology 1995; 109: Holtmann G, Goebell H, Talley NJ. Functional dyspepsia and irritable bowel syndrome: is there a common pathophysiological basis? Am J Gastroenterol 1997; 92:

7 VALIDATION OF THE GASTROINTESTINAL SYMPTOM SCORE Pimentel M, Rossi F, Chow EJ, et al. Increased prevalence of irritable bowel syndrome in patients with gastroesophageal reflux. J Clin Gastroenterol 2002; 34: Stanghellini V, Tosetti C, Paternico A, et al. Risk indicators of delayed gastric emptying of solids in patients with functional dyspepsia. Gastroenterology 1996; 110: Buckles DC, Sarosiek I, McMillin C, McCallum RW. Delayed gastric emptying in gastroesophageal reflux disease: reassessment with new methods and symptomatic correlations. Am J Med Sci 2004; 327: 1 4.