Heterogeneity of Symptom Pattern, Psychosocial Factors, and Pathophysiological Mechanisms in Severe Functional Dyspepsia

Transcription

1 GASTROENTEROLOGY 2003;124: Heterogeneity of Symptom Pattern, Psychosocial Factors, and Pathophysiological Mechanisms in Severe Functional Dyspepsia BENJAMIN FISCHLER,*, JAN TACK, VÉRONIQUE DE GUCHT, Z IV SHKEDY, PHILIPPE PERSOONS,*, DORINE BROEKAERT,* GEERT MOLENBERGHS, and JOZEF JANSSENS Departments of *Psychiatry and Internal Medicine, University Hospital Gashuisberg, Leuven, Belgium; Department of Neurosciences and Psychiatry, University of Leuven, Leuven, Belgium; Center for Statistics, Biostatistics, Limburgs Universitair Centrum, Diepenbeek, Belgium Background & Aims: Categorization of functional dyspepsia into subgroups is based on expert opinion according to (dominant) symptoms or on underlying pathophysiological mechanisms. We used an evidence-based approach to the determination of subtypes of functional dyspepsia. Methods: Consecutive functional dyspepsia patients were recruited from a tertiary referral center. The following were performed: (1) exploratory (EFA) and confirmatory factor analysis (CFA) of symptom patterns in a large group of patients with functional dyspepsia; (2) external validation of these factors by the determination of their association pattern with physio- and psychopathological mechanisms, and with health-related quality of life and sickness behavior; and (3) cluster analysis of their distribution in this population. Results: Both EFA and CFA do not support the existence of functional dyspepsia as a homogeneous (unidimensional) condition. A 4-factor model is found to be valid, with differential distribution within the patient population according to cluster analysis. Factor 1 is characterized by nausea, vomiting, early satiety, and weight loss and factor 2 by postprandial fullness and bloating. Both factor 1 and 2 are associated with delayed emptying, but only factor 1 is associated with younger age, female sex, and sickness behavior. Factor 3 is characterized by pain symptoms and associated with gastric hypersensitivity and several psychosocial dimensions including medically unexplained symptoms and health-related quality of life dimensions. Factor 4, characterized by belching, is also associated with hypersensitivity, but is unrelated to psychosocial dimensions. Conclusions: In a tertiary care population, functional dyspepsia is a heterogeneous condition characterized by 4 major dimensions differentially associated with psychopathological and physiopathological mechanisms. Functional dyspepsia, a clinical syndrome defined as chronic or recurrent upper abdominal symptoms without identifiable cause by conventional diagnostic means, is one of the most prevalent functional gastrointestinal disorders. 1,2 It is generally presumed that functional dyspepsia according to the current definition is a heterogeneous disorder, consisting of different subgroups. In subdividing functional dyspepsia into subgroups, 2 different approaches have been applied. Expert opinion subdivided patients, based on the predominant symptoms, into ulcer-like dyspepsia, dysmotility-like dyspepsia, and unspecified dyspepsia. 2 Others have used underlying pathophysiological mechanisms to subdivide functional dyspepsia patients. 3 5 An alternative approach would be to conduct factor analysis of symptom patterns in a large group of functional dyspepsia patients. To the best of our knowledge, factor analytical studies related to functional dyspepsia have only been performed in the general population Moreover, these factor analytical studies were exploratory, and no confirmatory factor analysis has ever been undertaken in functional dyspepsia. Several studies have shown an association between functional dyspepsia and psychopathological features such as neuroticism, somatization, and abuse Although these studies suggest a contribution of psychopathology in functional dyspepsia, they did not assess the relationship between psychopathological dimensions and specific dyspepsia-symptom clusters. Several recent studies have established the relationship of dyspepsia-symptom patterns to underlying physiopathological mechanisms such as delayed gastric emptying, impaired accommodation of the proximal stomach to a meal, and hypersensitivity to gastric distention. 3 5 To substantiate the possible existence of separate dyspepsia factors, the association between physiopathological mechanisms and dyspepsia factors was also analyzed. Abbreviations used in this paper: CFA, confirmatory factor analysis; EFA, exploratory factor analysis; IBS, irritable bowel syndrome by the American Gastroenterological Association /03/$30.00 doi: /gast

2 904 FISCHLER ET AL GASTROENTEROLOGY Vol. 124, No. 4 The aims of this study were to examine (1) the existence of several factors or subtypes of functional dyspepsia in a large dyspepsia population by exploratory and confirmatory factor analysis; (2) the external validity of these factors by examining their differential association with psychopathological and psychosocial dimensions such as neuroticism, somatization, abuse, medical visits, sick leave, and health-related quality of life and physiopathological mechanisms reported to be associated with functional dyspepsia; and (3) the distribution of these factors within the patient population using cluster analysis Materials and Methods Four hundred thirty-eight consecutive patients of the department of gastroenterology diagnosed with functional dyspepsia according to the Rome II criteria 2 were recruited from the general gastroenterology clinic and from the motility clinic during 2 consecutive years. The patients presented because of severe epigastric symptoms, and all underwent careful history taking and clinical examination, upper gastrointestinal endoscopy, routine biochemistry, and upper abdominal ultrasound. Inclusion criteria were fulfillment of the Rome II criteria for functional dyspepsia with the presence of symptoms for at least 3 months during the last year, in the absence of organic, systemic, or metabolic disease. Dyspeptic symptoms had to be present at least 3 days per week, with 2 or more symptoms scored as relevant or severe on the symptom questionnaire (see later). Exclusion criteria were the presence of esophagitis, a history suggestive of gastroesophageal reflux disease, gastric atrophy or erosive gastroduodenal lesions on endoscopy, heartburn as a predominant symptom, a history of peptic ulcer, major abdominal surgery, underlying psychiatric illness, and the use of nonsteroidal anti-inflammatory drugs, steroids, or drugs affecting gastric acid secretion. The intensity of 8 dyspeptic symptoms (epigastric pain, postprandial fullness, bloating, early satiety, nausea, vomiting, belching, epigastric burning) was assessed by a single gastroenterologist ( J.T.) and scored on a Likert scale with scores between 0 and 3 (absent, mild, moderate, and disabling) as previously reported. 4,5 This scale was shown to have an excellent interrater reliability. 15 In addition, body weight and the percentage of weight loss since the onset of the dyspepsia symptoms were determined. During a consecutive 18-month period, all newly diagnosed functional dyspepsia patients were provided a questionnaire assessing psychosocial dimensions, which was handed out at the clinic or mailed home during the next 2 weeks. One hundred seventy-two out of 249 patients (69.1%) completed the questionnaire. No differences were observed between responders and nonresponders in gender ratio ( , P 0.61), age (z 0.81, P 0.42), weight loss, and the intensity of all dyspepsia symptoms (data not shown) except for nausea that was more intense in the group that filled out the questionnaire (z 2.26, P 0.024). Neuroticism has been assessed with the neuroticism subscale of the NEO five factor inventory (NEO-FFI). 16 Abuse has been assessed with a self-rating scale for physical and sexual abuse 17 and a selfrating scale for psychological abuse. 18 Somatization has been assessed by summing up the symptoms reported by the patient on a self-report scale consisting of 45 nongender-specific symptoms. This scale consists of the symptoms included under the heading of DSM-IIIR and DSM-IV somatization disorder (Diagnostic and Statistical Manual of Mental Disorder). 19,20 In addition, the key symptoms of 4 functional somatic syndromes, namely functional dyspepsia, irritable bowel syndrome, chronic fatigue syndrome, and fibromyalgia, 21 were included. Current somatization (past month) was measured. The sum of the symptoms excluding those related to the gastrointestinal tract was computed (remaining number of symptoms 37). The organicity of extraintestinal symptoms was evaluated in 2 ways: by the self-report of associated organicity and by careful examination of the medical file of the patients. The decision regarding borderline organic cases was taken by consensus between 3 clinicians. Discrete somatoform conditions were also included as the following somatization categories: (1) the diagnostic criteria for chronic fatigue (self-rated) was chronic disabling fatigue for 6 months or more, and (2) irritable bowel syndrome (IBS) was assessed with a questionnaire assessing the Rome 2 criteria for IBS. 22 Health-related quality of life has been assessed with the classical Medical Outcomes Study Short Form 36 (MOS SF- 36) scale. 23 The number of visits to the general practitioner and to specialists during the last 12 months were assessed by self-rating. The number of days of sickness was also assessed by self-report and this for the last 12 months. Pathophysiological Mechanisms Results of gastric-emptying studies scintigraphy or breath test were available for all patients; only those emptying tests performed at our institution according to a standardized protocol were taken into account. All patients recruited from the motility clinic underwent a gastric barostat study to assess gastric sensitivity to distention and accommodation to a meal. Gastric-emptying studies. Gastric emptying for solids was measured using the previously validated 14 C octanoic acid breath test. 24 Briefly, all studies were performed in the morning after an overnight fast. The test meal consisted of 60 g of white bread; 1 egg, the yolk of which was doped with 74 kbq 14 C octanoic acid sodium salt; and 300 ml of water. Breath samples were taken before the meal and at 15-minute intervals for a period of 240 minutes postprandially. Gastric half emptying time (t1/2) was calculated as previously described. 24 Barostat studies. After an overnight fast of at least 12 hours, a double-lumen polyvinyl tube (Salem sump tube 14 Ch.; Sherwood Medical, Petit Rechain, Belgium) with an adherent plastic bag (1200-mL capacity, 17-cm maximal diameter) finely folded was intro-

3 April 2003 FACTOR ANALYSIS OF SEVERE DYSPEPSIA 905 duced through the mouth and secured to the subject s chin with adhesive tape. The polyvinyl tube was then connected to a programmable barostat device (Synectics Visceral Stimulator, Stockholm, Sweden). After unfolding the bag, with a 2-minute fixed-volume inflation of 300 ml of air, the balloon was again deflated and subjects were positioned sitting with the knees bent (80 ) and the trunk upright in a specifically designed bed. After a 30-minute adaptation period, minimal distending pressure (MDP) was first determined by increasing intrabag pressure by 1 mm Hg every 3 minutes until a volume of 30 ml or more was reached. This pressure level equilibrates the intra-abdominal pressure. Subsequently, isobaric distentions were performed in stepwise increments of 2 mm Hg starting from MDP, each lasting for 2 minutes, while the corresponding intragastric volume was recorded. Subjects were instructed to score their perception of upper abdominal sensations at the end of every distending step by using a graphic rating scale that combined verbal descriptors on a scale graded 0 6. The end point of each sequence of distentions was established at an intrabag volume of 1000 ml or when the subjects reported discomfort or pain (score 5 or 6). Discomfort threshold was defined as the first level of pressure, expressed relative to MDP, that provoked a score of 5 or more. 5 We previously established that more complex distending protocols yield similar results. 25 After a 30- minute adaptation period with the bag completely deflated, the pressure level was set at MDP 2mmHg during at least 90 minutes. After 30 minutes, a liquid meal (200 ml, 300 kcal, 13% proteins, 48% carbohydrates, 39%, Nutridrink; Nutricia, Bornem, Belgium) was administered. Gastric tone measurement was continued for at least 60 minutes after the meal. The meal-induced gastric relaxation was quantified as the difference between the average volumes during 30 minutes before and 60 minutes after the administration of the meal. 4 Statistical Methods Statistical methods: exploratory and confirmatory factor analysis. We used both exploratory factor analysis (EFA) and confirmatory factor analysis (CFA) to assess the existence of subtypes of functional dyspepsia and to identify these subtypes. The purpose of exploratory factor analysis 26 is to describe the relationship among the observed 9 items of the dyspepsia symptom complex and few underlying latent variables (the factors; i.e., the subtypes of functional dyspepsia). Based on the result obtained from the exploratory stage of the analysis, the latent structure of the data can be constructed. CFA allows evaluating to what extent the observed data agree with the theoretical model that was assumed. The CFA approach requires us to assume an underlying structure (established here with the EFA) and to test whether the observed data support the theoretical model CFA models are commonly used and discussed in the literature whenever a hypothesised behavioral model is of primary interest in the study. 30,31 Subsequently, nonhierarchical cluster analysis based on the factor scores derived earlier was performed to examine their distribution within the total patient population. Cluster analysis for 3, 4, and 5 cluster solutions were investigated, yielding similar results. Only the results of the 4 clusters solution are given in detail. One-way analysis of variance was used to investigate possible association between the 4 clusters and the 4 factors that were identified. All statistics have been performed with SPSS 9.0 for Windows with the exception of the confirmatory factor analysis for which SAS PROC CALIS has been used. Parametric analyses were used if not otherwise specified. All P values were 2-tailed and considered as statistically significant at a level of 5%. The ethical committee of the University Hospital approved the present research protocol. Results Patient characteristics. Four hundred and thirtyeight consecutive functional dyspepsia patients (138 men; mean age, 42 1 year) filled out the dyspepsia questionnaire. One hundred and seventy-nine patients (mean age, 43 1 year; 56 men) filled out the psychosocial questionnaire. The symptom pattern did not differ between both subgroups (data not shown). Figure 1 summarizes the patient selection and additional procedures used in the study. Table 1 summarizes the grading of dyspeptic symptoms in the patient group. Postprandial fullness and bloating were the most prevalent symptoms, present in 87% and 83%, respectively, of the patients. Epigastric pain (62%), nausea (64%), early satiety (60%), and belching (58%) were also frequently reported. Vomiting and epigastric burning sensation were present in 29% and 55%, respectively, of the patients. Weight loss in excess of 5% was present in 146 patients (34%). ECA and FCA on the total population. The results obtained with the exploratory factor analysis with principal component analysis and varimax rotation showed that the 9 items (8 symptoms and percentage weight loss) could be reduced into 4 factors with an eigenvalue 1.0 (Table 2).

4 906 FISCHLER ET AL GASTROENTEROLOGY Vol. 124, No. 4 Figure 1. Patient selection and additional questionnaires and procedures used in the data analysis. *Symptom pattern and demographics in this subgroup were not different from the whole population. The first factor captures 27.18% of the total variance; factors 2, 3, and 4 capture 14.66%, 12.99%, and 11.56% of the total variance, respectively. All other factors capture less then 10% each. The proportion from the total variance explained by the first 4 factors is 66.3%. This indicates that a single-factor structure of functional dyspepsia factor is not appropriate for the data. Moreover, the first 4 factors have clear clinical interpretations. Factor 1 included nausea, vomiting, early satiety, and weight loss; factor 2 included bloating and fullness; factor 3 included epigastric burning and pain; and factor 4 included belching. The factor loading of items on other factors was well below the level of 0.25 with the exception of a positive loading of early satiety on factor 2 (0.36) and a negative loading of weight loss on factor 4 ( 0.25) (Table 2). The results of the exploratory analysis suggest that there are 4 underlying components of functional dyspepsia. The path diagram in Figure 2 shows the theoretical model that was assumed and was under investigation in the confirmatory stage of the analysis. A CFA model with 4 factors was fitted to the data ( , df 15; P 0.114). This indicates that the model fits the data well. Goodness to fit has also been assessed by revealing high values of the adjusted goodness-of-fit index (0.96) and low value for the mean square residuals (0.034). Although the EFA suggested that a single-factor model for dyspepsia was not appropriate, the hypothesis of 1-dimensional latent structure was tested by fitting a second-order CFA model to the data. This model assumes the existence of the 4 subtypes but also the association of these factors with a global dyspepsia factor. Although the 2 statistic ( , df 13) indicates a good fit to the data (P 0.108), the difference in goodness of fit between the 2 models (first- and secondorder CFA) is not statistically significant. The first-order factors are all positively (and significantly) associated with the global factor (data not shown). Hence, the second-order CFA model supports the structure of 4 subtypes of functional dyspepsia and provides evidence that these 4 subtypes are associated (in a second order fashion) with an underlying global factor of dyspepsia. Association between dyspepsia factors and external variables. An EFA was also computed for the group of subjects who had filled out the psychosocial questionnaire. Symptom pattern and demographics did not differ significantly in this subgroup. The same 4 factors were found, explaining 67% of the variance with very similar factor loading for the individual items (data not shown). A statistically significant negative association was found between age and factor 1 only (Pearson s r 0.21; P 0.008). Women were found to have a significantly higher score than men on factor 1 (t 2.56, df 161; P 0.011). The association between age and factor 1 was observed in women only (women, Pearson s r 0.26, P 0.005; men, r 0.09, P 0.53). Because not all external variables could be normalized, nonparametric statistics were used for the study of the Table 1. Frequency of Severity Grading for Each of 8 Dyspepsia Symptoms in 438 Patients with Functional Dyspepsia 0 (absent) 1 (mild) 2 (relevant) 3 (severe) Postprandial fullness 54 (12.4) 47 (10.8) 175 (40.1) 161 (36.8) Bloating 73 (16.7) 42 (9.6) 187 (42.8) 135 (30.9) Epigastric pain 166 (38.0) 43 (9.8) 107 (24.5) 121 (27.7) Early satiety 175 (40.1) 46 (10.5) 112 (25.6) 104 (23.8) Nausea 157 (36.0) 63 (14.5) 133 (30.5) 83 (19.0) Vomiting 311 (71.2) 20 (4.6) 42 (9.6) 64 (14.7) Belching 182 (41.8) 57 (13.1) 148 (34.0) 48 (11.0) Epigastric burning 196 (44.9) 84 (19.2) 99 (22.7) 58 (13.3) NOTE. Numbers between parentheses represent row percentages.

5 April 2003 FACTOR ANALYSIS OF SEVERE DYSPEPSIA 907 Table 2. Factor Loading of the 9 Functional Dyspepsia Symptoms Factor I Factor II Factor III Factor IV Nausea Vomiting Early satiety Weight loss Bloating Fullness Epigastric burning Epigastric pain Belching association between the dyspepsia factors and external variables. Association between dyspepsia factors and psychosocial characteristics. Results of these analyses are summarized in Figure 3. No association was found among sexual and physical abuse and the dyspepsia factors. Being psychologically abused during childhood or adulthood was found to be associated with a higher level of factor 3 (respectively z 2.14, P 0.035; z 2.32, P 0.020). Statistically significant correlations were found between neuroticism as well as the somatization variable Figure 2. Schematic representation of first-order confirmatory factor analysis model for dyspepsia symptoms ( , df 15; P ). Four different factors are identified, with differential symptom loadings. Full arrows indicate positive loading of a symptom on a factor; dotted arrow indicates negative loading of a given symptom on a factor. Figure 3. Schematic overview of differential associations of the 4 dyspeptic symptom factors with pathophysiological (left side) and psychosocial or psychopathological (right side) dimensions. and factor 3 only (respectively, Spearman s 0.22, P 0.01; 0.28, P 0.001). A statistically significant higher score of factor 3 only was found to be associated with chronic fatigue and IBS (z 3.55, P 0.001; z 2.37, P 0.018) (Table 3). The MOS SF-36 health-related quality of life dimensions physical functioning and role physical were found to be associated with factor 1 only (respectively, n 155, 0.17, P 0.036; 0.27, P 0.001); body pain, general health, social functioning, and role emotional were found to be associated with factor 1 (respectively, 0.30, P 0.001; 0.22, P 0.008; 0.20, P 0.014; 0.16, P 0.046) and factor 3 (respectively, 0.21, P 0.011; 0.23, P 0.003; 0.18, P 0.028; 0.20, P 0.015). Vitality was found to be associated with factor 3 only ( 0.25, P 0.002). No statistically significant association was observed for the mental health dimensions of SF-36 and the 4 factors. Because factor 1 and several dimensions of the SF-36 were found to be higher in women (data not shown), correlations were re-analyzed in women only. Only the association between factor 1 and body pain remained statistically significant (data not shown). A statistically significant association was found between the frequency and the total number of days of sickness leave and factor 1 only ( 0.32, P 0.009; 0.33, P 0.009, respectively). The number of visits to the general practitioner and to the specialist were also found to be associated with factor 1 ( 0.19, P 0.021; 0.24, P 0.007, respectively). When adjusting for gender, all these findings remained statisti-

6 908 FISCHLER ET AL GASTROENTEROLOGY Vol. 124, No. 4 Table 3. Clinical Correlates of Dyspeptic Symptom Factors Factor I Factor II Factor III Factor IV Gender Female male Age Younger age Neuroticism Somatization Physical functioning a ( ) Role physical a ( ) Body pain a Vitality a General health a ( ) Social functioning a ( ) Role emotional a ( ) Mental health a Inability to work Episodes Days Medical visits GP Specialists ( ) Psychological abuse Child Adult, 0.01, P 0.05; ( ), not significant any more, after adjustment for gender., 0.001, P 0.01., P a Dimensions of MOS SF-36 scale. cally significant, with the exception of the number of medical visits to the specialists (data not shown). Association between dyspepsia factors and pathophysiological mechanisms. Symptom pattern and demographics did not differ significantly in the subgroups that underwent standardized testing of pathophysiological mechanisms, and factor analysis yielded similar results (data not shown). Standardized gastricemptying breath test studies were performed in 204 patients. The half-time of gastric emptying was significantly correlated to factor 1 (n 204, 0.24; P 0.001) and factor 2 (n 204, 0.18; P 0.009). Gastric accommodation to a meal was not found to be significantly associated with the dyspepsia factors. Gastric hypersensitivity (discomfort threshold) was found to be significantly associated with factor 3 (n 100, 0.24; P 0.017) and with factor 4 (n 100, 0.20; P 0.047) (Figure 3). Cluster analysis of distribution of dyspepsia factors in the patient population. Four separate clusters were identified which resulted in total R square equal to One-way analysis of variance was used to investigate possible association between the 4 clusters and the 4 factors that were identified. For all factors, significantly different levels were found in each cluster (F scores , 15.18, 50.52, and respectively, all P ). Cluster 1 was characterized by elevated scores on factor 1, whereas cluster 2 had elevated scores for all factors, especially 2, 3, and 4. Cluster 3 and 4 are only slightly elevated on factor 3 and factor 4 respectively, with depressed scores on the remaining factors (Table 4). No significant difference in age between the clusters was observed using F statistics. Pearson 2 was used to investigate the association between gender and clusters. A significantly higher proportion of women was found in cluster 2 (82% women compared with 68% in the total sample). Discussion Functional dyspepsia is generally considered a heterogeneous disorder. By using factor analysis of the symptom pattern in well-characterized functional dyspepsia patients, we showed the existence of 4 separate factors within functional dyspepsia, which are differentially associated with specific psychosocial and pathophysiological characteristics (Figure 3). The existence of 4 separate factors is supported by the large number of subjects studied, the validation of the exploratory factor analysis by a confirmatory factor analysis, and second-order CFA. Moreover, a unidimensional model of functional dyspepsia tested with CFA did not fit the data well. External validity was obtained by means of the specific correlation patterns of these factors with relevant physio- and psychopathological dimensions. The demonstration of 4 factors within functional dyspepsia provides evidence for the heterogeneous nature of this condition in terms of symptom pattern and associated physiopathological and psychopathological mechanisms. Identification of the nature of the 4 factors is beyond the scope of the current study. The factor analysis Table 4. Overview of Factor Summary Scores by Cluster Using a 4-Cluster Solution N Factor 1 Factor 2 Factor 3 Factor 4 Cluster (0.48) 0.16 (1.11) 0.21 (0.98) 0.41 (0.88) Cluster (1.11) 0.66 (0.61) 0.73 (1.04) 1.02 (0.57) Cluster (0.52) 0.14 (1.01) 0.33 (0.85) 0.77 (0.56) Cluster (0.51) 0.17 (0.97) 0.63 (0.66) 0.36 (0.89) NOTE. Differential distribution of the 4 dyspeptic symptom factors across the 4 clusters are found, based on 403 observations without missing values.

7 April 2003 FACTOR ANALYSIS OF SEVERE DYSPEPSIA 909 does not show the existence of 4 separate subgroups among patients with functional dyspepsia with specific clinical and pathophysiological features. Rather, in individual patients, the symptom pattern and the associated physiopathological and psychopathological abnormalities will be determined by the relative contribution of each of these factors. Subsequent cluster analysis confirmed that the dyspeptic population can be subdivided into clusters with different loading scores for the different factors. Factor 1 consists of vomiting, nausea, early satiety, and weight loss. Factor 2 consists of fullness and bloating. Both factors are reminiscent of dysmotility-like dyspepsia. 2 Both factors are associated with delayed gastric emptying, but only factor 1 seems to be more prevalent in young women and is associated with healthseeking behavior and sickness leave. This is in keeping with previous studies showing that functional dyspepsia patients with delayed gastric emptying are more likely to be women and have a higher prevalence of postprandial fullness, nausea, and vomiting. 3,32 Factor 3 includes pain and epigastric burning. This factor is reminiscent of ulcer-like dyspepsia. 2 This factor has been found to be associated with most psychopathological and with several health-related quality-of-life dimensions. Moreover, this factor has also been found to be associated with other somatoform conditions such as chronic fatigue and IBS. Gastric hypersensitivity to balloon distention is associated with this factor, which confirms our previous observations of an association between hypersensitivity to balloon distention and pain. 5 Previous research of our group 33 in functional dyspepsia showed a strong association between abuse and psychopathological dimensions such as somatization on the one hand and gastric hypersensitivity on the other. These findings strongly suggest that so-called ulcer-like dyspepsia may in fact be part of a more global psychopathological-somatization dimension beyond the border of functional dyspepsia. This also means that within functional dyspepsia, variability in somatization does not explain the variability in most dyspepsia symptoms. Factor 4 includes belching. Weight loss was found in both the EFA and CFA to be moderately negatively associated with this symptom. Recently, we also showed an association between visceral hypersensitivity and belching. 5 These findings are logically reproduced here. In contrast to factor 3, factor 4 is not associated with psychosocial dimensions. In a population study of univestigated dyspepsia, Talley et al. 9 showed the existence of a dyspepsia factor including a largely pain dimension. In another recent study, 1 factor with features of ulcer-like and reflux-like dyspepsia was shown. 10 Almost all symptoms of this factor were also pain-related. Another factor included the symptoms of upper abdominal pain associated with nausea, vomiting, and loss of appetite, resembling dysmotility-like dyspepsia. These 2 factors were shown to be independent from an IBS factor. Our results are partly similar to aforementioned findings in the general population. Unlike previous studies, in the present study, the dysmotility factor is found to be part of 2 independent factors and IBS seems to be associated with the ulcer-like factor. Bennett et al. 34 did not find similar psychosocial differences between the subtypes of dyspepsia. In that study, psychological features were found as globally homogeneously distributed among dyspepsia subtypes. However, Bennett et al. 34 compared several types of functional gastrointestinal disorders established on the basis of existing diagnostic criteria, whereas our analyses were conducted on symptom (and not patient) factors resulting from factor analysis within a large group of functional dyspepsia only. A limitation of the present study is that the questionnaires assessing psychosocial parameters were not always filled out immediately after the assessment of the dyspepsia symptoms, although questionnaires were mailed within 2 weeks after the symptom assessment visit. This problem has partly been overcome by exclusively using, besides the somatization dimension, trait measures of psychopathology or past stressors (such as abuse) and by avoiding the use of state variables such as the intensity of anxiety or depression. Regarding somatization, it is doubtful that the number of somatoform symptoms oscillates a lot over short periods of time especially because the number of symptoms was quite high. However, this constitutes a limitation of this study. Another important limitation is the fact that the study has been performed in tertiary care. It is conceivable that our findings cannot be generalized to nonconsulting dyspepsia patients or to functional dyspepsia in primary or secondary care. Our data are of a cross-sectional nature. The stability over time of these functional dyspepsia dimensions is unknown. Prospective naturalistic and intervention studies are needed to understand whether these 4 factors or dimensions are characterized by different evolution or distinct prognostic patterns. In summary, functional dyspepsia can be considered as a heterogeneous condition on a symptomatic, psychosocial, and physiopathological level. The major contribution of the present study is to show the existence of four major dimensions within functional dyspepsia differen-

8 910 FISCHLER ET AL GASTROENTEROLOGY Vol. 124, No. 4 tially associating symptom pattern, as well as psychopathological and physiopathological mechanisms. References 1. Talley NJ, Silverstein MD, Agreus L, Nyren O, Sonnenberg A, Holtmann G. AGA technical review: evaluation of dyspepsia. American Gastroenterological Association. Gastroenterology 1998;114: Talley NJ, Stanghellini V, Heading RC, Koch KL, Malagelada JR, Tytgat GNJ. Functional gastroduodenal disorders. Gut 1999; 45(Suppl 2):II37 II Stanghellini V, Tosetti C, Paternico A, Barbara G, Morselli-Labate AM, Monetti N, Marengo M, Corinaldesi R. Risk indicators of delayed gastric emptying of solids in patients with functional dyspepsia. Gastroenterology 1996;110: Tack J, Piessevaux H, Coulie B, Caenepeel P, Janssens J. Role of impaired gastric accommodation to a meal in functional dyspepsia. Gastroenterology 1998;115: Tack J, Caenepeel P, Fischler B, Piessevaux H, Janssens J. Hypersensitivity to gastric distention is associated with symptoms in functional dyspepsia. Gastroenterology 2001;121: Agreus L, Svardsudd K, Nyren O, Tibblin G. Irritable bowel syndrome and dyspepsia in the general population: overlap and lack of stability over time. Gastroenterology 1995;109: Whitehead WE, Crowell MD, Bosmajian L, Zonderman A, Costa PT Jr, Benjamin C, Robinson JC, Heller BR, Schuster MM. Existence of irritable bowel syndrome supported by factor analysis of symptoms in two community samples. Gastroenterology 1990;98: Taub E, Cuevas JL, Cook EW 3rd, Crowell M, Whitehead WE. Irritable bowel syndrome defined by factor analysis. Gender and race comparisons. Dig Dis Sci 1995;40: Talley NJ, Boyce P, Jones M. Identification of distinct upper and lower gastrointestinal symptom groupings in an urban population. Gut 1998;42: Whitehead WE, Gibbs NA, Li Z, Drossman DA. Is functional dyspepsia just a subset of the irritable bowel syndrome? Baillieres Clin Gastroenterol 1998;12: Wilhelmsen I, Haug TT, Ursin H, Berstad A. Discriminant analysis of factors distinguishing patients with functional dyspepsia from patients with duodenal ulcer. Significance of somatization. Dig Dis Sci 1995;40: Nyrèn O, Adami HO, Gustavsson S, Loof L. Excess sick-listing in nonulcer dyspepsia. J Clin Gastroenterol 1986;8: Langeluddecke P, Goulston K, Tennant C. Psychological factors in dyspepsia of unknown cause: a comparison with peptic ulcer disease. J Psychosom Res 1990;34: Talley NJ, Fett SL, Zinsmeister AR, Melton LJ 3rd. Gastrointestinal tract symptoms and self-reported abuse: a population-based study. Gastroenterology 1994;107: Cuomo R, Sarnelli G, Grasso R, Alfieri M, Niccolai E, Pumpo R, Tack J, Budillon G. Functional dyspepsia symptoms, gastric emptying and satiety provocative test: analysis of relationships. Scand J Gastroenterol 2001;36: Costa PT Jr, McCrae RR. Stability and change in personality assessment: the revised NEO Personality Inventory in the year J Pers Assess 1997;68: Drossman DA, Leserman J, Nachman G, Li ZM, Gluck H, Toomey TC, Mitchell CM. Sexual and physical abuse in women with functional or organic gastrointestinal disorders. Ann Intern Med 1990;113: Pitzner JK, Drummond PD. The reliability and validity of empirically scaled measures of psychological/verbal control and physical/sexual abuse: relationship between current negative mood and a history of abuse independent of other negative life events. J Psychosom Res 1997;43: American Psychiatric Association. Diagnostic and statistical manual of mental disorders. 3rd ed. Washington DC: American Psychiatric Association, American Psychiatric Association. Diagnostic and statistical manual of mental disorders. 4th ed. Washington DC: American Psychiatric Association, De Gucht V, Neuroticism, alexithymia, negative affect and positive affect as determinants of medically unexplained symptoms in primary care: a cross-sectional study. Acta Neuropsychol 2002;14: Thompson WG, Longstreth GL, Drossman DA, Heaton KW, Irvine EJ, Müller-Lissner SA. Functional bowel disorders and functional abdominal pain. Gut 1999;45(Suppl 2): Ware JE Jr, Sherbourne CD. The MOS 36-item short-form health survey (SF-36). I. Conceptual framework and item selection. Med Care 1992;30: Ghoos YF, Maes BD, Geypens BJ, Hiele MI, Rutgeerts PJ, Vantrappen G. Measurement of gastric emptying rate of solids by means of a carbon labeled octanoic acid breath test. Gastroenterology 1993;104: Sarnelli G, Vos R, Cuomo R, Janssens J, Tack J. Reproducibility of gastric barostat studies in healthy controls and in dyspeptic patients. Am J Gastroenterol 2001;96: Johnson RA, Wichern DW, eds. Applied multivariate statistical analysis. London: Prentice-Hall, Mueller RO. Basic principles of structural equation modelling. New York: Springer, Bollen KA. Structural equations with latent variable. New York: John Wiley & Sons, Liu G, Clark MR, Eaton WW. Structural factor analysis for medically unexplained somatic symptoms of somatization disorder in the epidemiologic cathmebt area study. Psychol Med 1997;27: Robbins JM, Kirmayer LJ, Hemami S. Latent variables models of functional somatic distress. J Nerv Ment Dis 1997;185: Deary IJ, Scott S, Wilson JA. Neurotism, alexithymia and medically unexplained symptoms. Person Individ Diff 1997;22: Sarnelli G, Caenepeel P, Geypens B, Janssens J, Tack J. Symptoms associated with impaired gastric emptying of solids and liquids in functional dyspepsia. Am J Gastroenterol 2003 (in press). 33. Fischler B, De Gucht V, Tack J, Janssens J. Psychosocial and psychopathological correlates of hypersensitivity to gastric distention in functional dyspepsia. Gastroenterology 1999; 116:A163 (abstr). 34. Bennett EJ, Piesse C, Palmer K, Badcock CA, Kellow JE. Functional gastrointestinal disorders: psychological, social, and somatic features. Gut 1998;42: Received August 27, Accepted January 9, Address requests for reprints to: Jan Tack, M.D., Ph.D., Department of Internal Medicine, Division of Gastroenterology, University Hospital Gasthuisberg, Herestraat 49, B-3000 Leuven, Belgium. Jan.Tack@med.kuleuven.ac.be; fax: (32) The authors thank R. Vos, A. De Greef, and K. Luyckx for excellent technical support.