| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From The Research Clinic for Functional Disorders and Psychosomatics (P.F., T.T., M.S.T., E.Ø.), Aarhus University Hospital, Denmark.; The Research Unit for General Practice (F.O.), University of Aarhus, Denmark.
Address correspondence and reprint requests to Per Fink, The Research Clinic for Functional Disorders, Aarhus University Hospital, Noerrebrogade 44, 8000 Aarhus C, Denmark. E-mail: flip{at}as.aaa.dk
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONNOTESREFERENCES |
|---|
Methods: The 978 consecutively admitted patients from a neurological department (n = 120), a medical department (n = 157), and from primary care (n = 701) were interviewed using the Schedules for Clinical Assessment in Neuropsychiatry (SCAN) diagnostic instrument.
Results: Patients complained of a median of five functional somatic symptoms; women of six, men of four (p < .0001). No single symptoms stood out as distinctive for patients with multiple symptoms. Principal component factor analysis identified a cardiopulmonary including autonomic (CP), a musculoskeletal (MS), and a gastrointestinal (GI) symptom group explaining 36.9% of the variance. Latent class analysis showed that the symptom groups are likely to materialize in the same patients, suggesting that they are different manifestations of a common latent phenomenon. Inclusion of a group of five additional general, unspecific symptoms in latent class analysis allowed construction of clinical diagnostic criteria for ‘bodily distress disorder’ dividing patients into three classes: nonbodily distress (n = 589), modest bodily distress (n = 329, prevalence 25.3%, men 20.4%, women 25.6%), and severe bodily distress (n = 60, prevalence 3.3%, men 1.2%, women 4.8%).
Conclusion: The study suggests that bodily distress disorder as defined here may unite many of the functional somatic syndromes and some somatoform disorder diagnoses. Bodily distress may be triggered by stress rather than being distinct diseases of noncerebral pathology.
Key Words: somatization • somatoform disorders • functional somatic symptoms • functional somatic syndromes • bodily distress • classification
Abbreviations: SCAN = Schedules for Clinical Assessment in Neuropsychiatry; CP = cardiopulmonary; MS = musculoskeletal; GI = gastrointestinal; CFS = chronic fatigue syndrome; DSM = Diagnostic and Statistical Manual of Mental Disorders; ICD = International Classification of Diseases; SCL = Symptoms Check List; CAGE = cutting down, annoyance by criticism, guilty feeling, and eye-openers; WHO = World Health Organization; STATA = statistical software; IQR = interquartile range; GS = general symptoms; IBS = irritable bowel syndrome.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONNOTESREFERENCES |
|---|
The classification of patients presenting with functional somatic symptoms remains contested. Many different functional syndromes have been introduced, e.g., chronic fatigue syndrome (CFS), fibromyalgia, irritable bowel syndrome (IBS), and chronic benign pain syndrome, and new syndromes are intermittently introduced (7). In the International Classification of Diseases (ICD)-10 and the Diagnostic and Statistical Manual of Mental Disorders (DSM)-IV psychiatric classification, the symptoms are mainly classified under somatoform disorders (8–11). There is a considerable overlap in symptoms and published diagnostic criteria between functional syndromes and between somatoform disorders (8–11). This may be so because most of these diagnostic labels and syndromes were developed in small, unrepresentative samples from sub-specialty clinics or by consensus. Their justification has been tested by use of inappropriate designs or statistics like case control design or a confirmatory analytical approach (8,12–14). As patients referred to a specialized service are selected according to the symptoms they present, the various functional somatic syndromes may thus simply be an artifact of medical specialization reflecting the referral process and specialists’ tendency to focus only on symptoms pertinent to their specialty. Patients with identical symptom patterns may therefore receive different diagnostic labels depending on the specialty of the doctor and the department they are admitted to.
The diagnostic criteria for the somatization disorder diagnosis including patients with multiple functional symptoms have varied with the permutations of the DSM and ICD classifications. Both the number of symptoms and the listed symptoms to be counted have varied, and cutpoints for symptom counts have been disputed in the medical literature (15, 16). The somatization disorder diagnosis springs from an exploratory study by Perley and Guze in 1962 based on 39 patients admitted to a psychiatric ward and diagnosed with "hysteria" (17). To our knowledge, only one later study by Fink (18) has explored a broad spectrum of bodily symptom complaints. Other studies have relied on short, predefined symptom lists deriving from the Perley and Guze study (17), or have used paper and pencil questionnaires, which allow no distinction between medically explained and functional symptoms.
There is an urgent need for establishing the validity of and defining borders between the various functional somatic syndromes and diagnoses (9,19,20).
This study aims to explore which symptoms patients presenting with functional symptoms complain of, and to investigate if the symptoms cluster into patterns that may support the existence of various distinct syndromes and diagnostic entities.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONNOTESREFERENCES |
|---|
All the participating patients received written and oral information and gave written informed consent.
Patients of nonScandinavian descent and patients who could not be interviewed were excluded. Exclusion was mainly due to patients declining the invitation to participate, the severity of physical disease, deafness, disorientation, aphasia, or death. For details, see (2,21,22).
The Danish Health Care System is almost entirely tax financed, and almost all medical care is free of charge. General hospital departments provide all hospital-based neurological and medical service for the general population within its catchment area. Denmark has a family doctor system, and 98% of the population is registered with a family physician, whom they must consult for medical services and for referral to specialized care.
Procedures
A two-phase design was used in all three samples. At admission or first contact, patients received a screening questionnaire (first phase) which included, among others, the eight-item version of the Symptom Check List (SCL-8d) (23) assessing anxiety and depression, and the seven-item Whiteley-index (24) measuring worrying and conviction of illness. In the primary care sample, the somatization subscale of the SCL-90, which checks for 12 common physical symptoms (25), was included as well as the CAGE, which consists of four questions screening for alcohol abuse (26).
Selection of Patients for Diagnostic Psychiatric Interview
Patients with high scores on the screening instruments were selected for Schedules for Clinical Assessment in Neuropsychiatry (SCAN) interview (second phase). In order to produce a stratified sample, we also selected a random sample of one ninth of the remaining primary care patients and one third of the neurological and internal medical patients respectively for interview. For more details, see the statistical analysis section and (2,21,27).
Of the 130 neurological patients who were selected for interview, 10 declined and 120 were successfully interviewed. Of the 170 internal medical patients who were selected for interview, two died, 11 declined, and thus 157 were interviewed. Of the 894 primary care patients who were selected for interview, 193 declined and the remaining 701 were interviewed. For details on attrition analysis, see (2,21,22,27,28).
The Psychiatric Research Interview
The psychiatric interviews were performed as soon as possible after the index contact, i.e., for the vast majority of patients during the first week after initial contact. Only for very few patients it was more than one month after. In those cases, the interviewer did not include the period between index contact and day of interview. We used the World Health Organization (WHO) endorsed psychiatric SCAN interview, version 2.1 (29) covering all types of psychiatric disorders and including an extensive section on physical health.
As for the hospitalized patients, the SCAN interviews were conducted by two psychiatric physicians, the primary patients by six. All the interviewers had been certified at the WHO SCAN training center in Aarhus. Besides psychiatric residency, they had at least two years of medical and surgical residency.
The physical health chapter of the SCAN interview explores 76 physical symptoms distributed among seven symptom groups. The interviewer rates each symptom to be either absent, attributable to a medical condition/dysfunction, or functional. Symptoms, for which this distinction cannot be made reliably, are rated with separate codes. Symptoms with these codes were not included in the analyses in this study. Also, symptoms were only counted if they had been present at some point during the past 2 years and had caused significant impairment, or had attracted medical attention.
At the beginning of the chapter, the patients are asked some screening questions about physical health, and if there is no indication of significant functional symptoms, the interviewer skips the rest of the chapter.
The interviewers were free to explore aspects that were not fully clarified in the interview, e.g., by reviewing medical records or discharge letters. The interviewers met regularly during the course of the study to confer about specific cases and rating ambiguous responses and symptoms. They were free to consult various specialists of the Aarhus University Hospital.
The interrater agreement on diagnosis-level was high (
= 0.86 in the neurological/medical sample, and 0.88 in the primary care sample) (21,22).
Depressive disorder was defined as patients fulfilling ICD-10 diagnostic criteria for F31.3 to 5, F32, F33.0 to 3, F33.8 to 9, and anxiety disorder F40.0 to 42.9.
Physician Ratings
At discharge, the primary caring physicians were asked to assess whether the neurological and internal medical patients’ illness was functional or if symptoms had a medical explanation, and whether the patients were overly preoccupied with or worried about illness. Family physicians were also asked whether they judged the patients to have a low threshold for consultation, and whether the patients consulted frequently for functional symptoms.
Statistical Analysis
The SCAN interviews were used for computerized ICD-10 and DSM-IV psychiatric diagnoses based on the algorithms developed by the WHO. We processed the data in STATA version 8.2 (30). Groups were compared using the Kruskal-Wallis test or Mann-Whitney test for nonnormally distributed continuous data. Clustering of symptoms was explored using principal component factor analyses with varimax rotation and Scree plot to identify the number of factors. Patient grouping was analyzed by latent class analysis with dichotomous and polytomous variables. Model selection was based on information criteria and goodness-of-fit values.
Overall prevalence figures were calculated using weighted logistic regression with the observed sampling fractions of the study phase two patients as weights, thus correcting for sample skewness introduced by stratified sampling procedure (31,32). Individual symptom frequency is reported unweighted. In the analyses of symptom clustering and analyses of associations, we did not use weighting because the stratified sampling procedure is not expected to introduce any bias in those results. This conclusion was supported by the fact that the screening negative patients and the screening positive patients had a similar symptom pattern in the principal component factor analyses, and the symptom frequencies among the screening negative were similar to other patients with few symptoms.
Ethical Approval
In Denmark, all biomedical studies have to be approved by an Ethics Scientific Committee, which is equivalent to an institutional review board. This study was approved by the Ethics Scientific Committee of the County of Aarhus.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONNOTESREFERENCES |
|---|
2 2.118; df = 2; p = .3468). Women complained of a median of six symptoms (IQR: 2–13), men of a median of four (IQR: 2–9) (Mann-Whitney U-test, p < .0001). Analysis of the neurological and medical samples alone showed that patients older than 65 had significantly fewer symptoms (median 3.5, IQR 1–7) than younger patients (median 6, IQR 2–12) (Mann-Whitney U-test, p = .0122). The overall number of functional somatic symptoms rose significantly with illness duration, but only during the first six months (data not shown, but available from authors).
Symptom Specificity
Neurological symptoms were more frequent in the neurological sample, symptoms suggesting a severe medical condition were more frequent in the medical sample, and symptoms commonly found in the general population were more frequent in the primary care sample. These differences seem trivial and presumably reflect the selection of patients during the referral process, and data are therefore not shown (data available from authors).
Figure 1 shows the number of functional symptoms of which each patient complained. The gradual decline of the curve shows absence of a natural cutpoint for grouping patients according to number of functional somatic symptoms. We therefore arbitrarily defined one fifth of the patients with ten or more functional symptoms as being multi-symptomatic patients, and the fourth with one to three symptoms as oligo-symptomatic. In the group with 10 or more symptoms, no particular symptoms stand out as being particularly characteristic, and in the group with one to three symptoms only tension headache stands out (Figure 2). The graphs for symptom frequency is on a constant higher level among patients with 10 or more symptoms compared with patients with one to three symptoms, but the symptoms appear in approximately the same order in both groups (Figure 2). The frequency of individual symptoms in each patient group is hence a simple function of the overall number of symptoms (formula expression of the association available from the authors).
|
|
Functional Somatic Syndromes
Underlying empirical symptom patterns were identified by principal component factor analysis performed separately for men and women. The analysis only included the 62 symptoms eliciting a positive response from 10 or more patients. Symptoms of menstrual disturbances, unusual vaginal discharge, pain during sexual intercourse, and sexual indifference make up a separate factor among women explaining 3.4% of the 25.1% variance in a four-factor model. Excluding these four symptoms and another two gynecological symptoms, the Scree plots supported a three-factor model both for men and women. Furthermore, the same factor pattern was supported for both genders, except for the two symptoms ‘excessive and distressing fatigue (following physical exercise)’ and ‘excessive and distressing fatigue (following mental exercise),’ which loaded high (factor loads >0.47) on one factor among the women, but not among the men (factor loads <0.11). Excluding those two symptoms, combining the male and female sub-sample and including the 25 symptoms with a factor load of 0.45 or higher only, yields the final three-factor model shown in Table 1. This model explains 36.9% of the variance.
|
The three factors are cardiopulmonary (CP), including autonomic symptoms, gastrointestinal (GI), and musculoskeletal (MS) tension or pain. These factors also emerged from a nonorthogonal, promax factor analysis, and furthermore they showed some degree of correlation with correlation coefficients in the range of 0.29 to 0.44. Having thus defined three different symptom groups or hypothesized syndromes, we pose the question whether these three syndromes are independent, distinct phenomena, or three different expressions of a common, but more basic phenomenon in the causal hierarchy. Significantly intercorrelated correlation coefficients (Spearman’s rho) of 0.40 to 0.45 were observed between the number of symptoms from the CP, GI, and MS symptom groups. This means that patients with symptoms from one group were also inclined to complain of symptoms from one of the other symptom groups and other organ systems as displayed in Figure 2. Only two to three (1.2% to 2.3%) patients with three or more symptoms from each symptom group were ‘pure’ types, i.e., presented symptoms from one symptom group only. This indicates that the three symptom groups do not define three independent, distinct diagnoses.
To further investigate this hypothesis, we used latent class (LC) analysis to ascertain whether symptoms from the three groups occurred in the same patients, or if different syndromes materialized in different patients. We first performed a latent class analysis including all 62 SCAN symptoms. Due to convergence problems, only a two-class model was actually estimated. Their overall number of symptoms largely characterized the patients in the two classes that emerged, i.e., patients in one of the classes displayed few symptoms, and patients in the second class presented many symptoms. Furthermore, the same symptoms as found in the principal component factor analysis stand out in the LC analysis in the way that patients belonging to the few symptoms’ class had a low probability of displaying any of these symptoms, whereas patients belonging to the other class had a high probability of presenting any of the symptoms from the three groups.
To reduce the dimensionality of the data, we therefore decided to make use of the three factors from the principal component factor analysis in the following way. We included the number of symptoms presented by the patients from the CP, GI, and MS symptom groups as three separate variables. First we included the total number of symptoms presented by the patients from each of the three symptom groups (i.e., as polytomous variables). Then we divided the number of symptoms into three categories within each symptom group, and finally dichotomized the numbers at different cut-points. All three analyses showed that the patients fell into two or three different classes defined by the total number of symptoms they presented from any of the three symptom groups, i.e., total number of symptoms regardless of symptom group. This finding suggests that the CP, GI, and MS syndromes do not express distinct disorders, but rather are different representations of a common latent phenomenon.
Bodily Distress Disorder, Clinical Criteria
In the current absence of any pathophysiological markers for identifying this common latent phenomenon, clinical criteria to identify patients are desirable, both from a clinical and from a research point of view. We therefore looked for more simple and useful clinical criteria for identifying and classifying patients.
The above latent class analysis suggested that the best model fit was achieved when dichotomizing the number of symptoms from each symptom group into patients with zero to one and two or more symptoms, or into patients displaying zero to two and three or more symptoms from each of the symptom groups displayed in Table 1. From a clinical point of view, prioritizing specificity instead of sensitivity, we chose a cutpoint of three or more symptoms. The three factors that emerged in the principal component factor analysis only accounted for 36.9% of the variance, and as we wished to further improve the sensitivity for identifying multi-symptomatic patients, and to discriminate between oligo-symptomatic and multi-symptomatic patients, we constructed a ‘general symptom’ (GS) group with the five most frequent general, unspecific symptoms not already included in the CP, GI, and MS symptom groups. These symptoms were headache, dizziness, memory impairment, concentration difficulties, and fatigue. Correlation coefficients (Spearman’s rho) between number of symptoms in GS and the CP, GI, and MS symptom groups were 0.46, 0.39, and 0.48 respectively. A variable identifying patients with four or more symptoms from any of the CP, GI, MS, or GS symptom groups was also formed. Inclusion of this variable together with the CP, GI, MS, and GS groups, each dichotomized into patients with zero to two or three or more symptoms from the CP, GI, MS, and GS groups, showed a good statistical fit (Pearson’s 2 (14) = 16.35, p = .2923) with a three-class model (Table 2).
|
The classes are easily identified clinically. Class 1 patients’ (except eight) symptom presentation was below cutpoint for any symptom group (Table 3). In class 2, the patients presented symptoms above cutpoint from one to three symptom groups (except 11 patients), and patients belonging to class 3 had above cutpoint number of symptoms from four or five symptom groups. These classes seem to represent a severity dimension, and we will accordingly label the 3 classes: 1, ‘nonbodily distress’; 2, ‘modest bodily distress’; and 3, ‘severe bodily distress.’
|
The number of symptoms from each symptom group (CP, GI, MS, and GS) were included in a principal component factor analysis, and a one-factor model explaining 56% of the variance emerged. All variables had loadings larger than 0.7. This one-dimensional finding supports that the three classes represent a severity dimension.
Characteristics of Patients With Bodily Distress
Class 1 patients had a median of one (min. zero, max. eight) of the 76 functional somatic symptoms, class 2 a median of nine (min. two, max. 30), and class 3 a median of 21.5 (min.10, max. 50) (Table 4). Patients belonging to class 2 and 3 had long illness durations (Table 4). Class membership was significantly associated with interference with everyday activities; thus 80.0% of class 3 patients and 60.8% of class 2 patients were modestly or severely impaired (Table 4).
|
Association With Emotional Distress
The bodily distress classes were highly associated with mood and anxiety disorders and unspecific emotional distress (Table 4). Thus, 50% to 60.0% of class 3 patients with severe bodily distress also had unspecific emotional distress or a mood or anxiety disorder.
Physician Rating of Patients
The primary care physicians, the neurologists, and internists found patients with bodily distress to be significantly more preoccupied with illness and bodily sensations than the patients with nonbodily distress and their admissions or consultation to be caused by medically unexplained symptoms (Table 5). Compared with the nonfunctional patient group, the primary care physicians found that significantly more of their patients with bodily distress had a low threshold for consultation, and that these patients frequently consulted because of functional symptoms.
|
Prevalence
Bodily distress was prevalent in the overall sample: 25.3% had modest and 3.3% had severe bodily distress (Table 6. The syndrome was more prevalent among women than among men, and less prevalent among patients older than 65 than among younger patients. Modest bodily distress was more prevalent in the neurological and the primary care sample than in the medical sample.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONNOTESREFERENCES |
|---|
In biomedicine, clustering of symptoms is called a syndrome. Presumably, a syndrome is caused by a common latent factor, i.e., the symptoms are all rooted in a single organic or physiological abnormality (9). The MS factor may simulate fibromyalgia and the GI factor irritable bowel syndrome (IBS), but we have no commonly used analogue syndrome alias for the third CP factor. However, we find little support for viewing these symptom groups as three completely independent, distinct diagnostic entities attributable to abnormalities in specific organ systems. First, they were significantly intercorrelated. Other studies simultaneously exploring more than one functional syndrome confirm this observation (8,10–12,14,33). Second, though latent class analysis including the three symptom groups showed that patients fell into three classes, these classes were defined by the overall number of symptoms that the patients presented, regardless of which symptom groups (CP, GI, or MS) they presented from. This suggests that the three functional somatic syndromes identified in this study are different representations of a common latent phenomenon. We have named this phenomenon bodily distress, because the symptom profiles indicate that they must be ascribed to centrally controlled mechanisms rather than abnormalities in specific peripheral organ systems. The CP and GI factors may be accounted for by hyperactivity of the autonomic nervous system. The MS factor may be attributed to malfunction of the reticular system located within the brain stem and the medulla. The hypothalamic-pituitary-adrenocortical axis may be involved as well. We therefore hypothesize that the three syndromes as well as the general tendency to experience multiple functional somatic symptoms may represent a stress response mediated through physiological and cerebral pathways. This hypothesis does not exclude the possibility of the symptoms being caused by an acquired or inherent abnormality in the mediation pathways, i.e., that they may arise with or without the presence of any evident external or internal stressor. However, further studies are needed to identify why patients manifest different GI, MS, or CP variations of bodily distress or bodily distress disorder and other types of studies are needed to uncover the precise nature of the biological and physiological intrinsic mechanism involved (12,34,35).
Ignorance of or insufficient knowledge about such bodily distress reactions may cause the symptoms to be misinterpreted as physical disease located in peripheral organs and medical attention to be misdirected with the ensuing risk of exposing the patients to iatrogenic harm and further distress. Categorical diagnoses based on positive criteria supplementing the exclusion of all other possibilities are therefore useful and, indeed, needed.
This study showed that such a distinct bodily distress syndrome could be defined for clinical use. Besides the three empirically defined symptom groups, we incorporated a general symptom group (GS) including five frequent, unspecific general symptoms, i.e., symptoms associated with no specific organ or symptom group because they are concomitant to most other symptoms. It may be hypothesized that this symptom group is a proxy for a generally increased sensitivity to symptoms or sensations. A latent class model analysis produced a statistically valid as well as easily interpretable and clinically identifiable classification. This classification divided the patients into three classes: a) nonbodily distress, b) modest bodily distress, and c) severe bodily distress. For clinical use, the physician merely has to decide whether the patient has any of the CP, GI, MS, or GS syndromes (i.e., three or more symptoms from any symptom group), and to decide the severity based on the number of different symptom groups from which the patient presents symptoms. The CP, GI, and MS symptom groups or syndromes may be viewed as different manifestations or types of bodily distress disorder. Only few, if any, conventionally defined diseases seem to manifest themselves with this symptom profile, and we may have established positive criteria for a bodily distress syndrome. We did not explore symptoms accounted for by medical or surgical diseases, and studies testing the differential diagnostic potential of our approach with symptom profiles are therefore needed. We applied principal component factor analysis for identifying symptom patterns as this statistical analysis, in contrast to other methods, includes numerous items simultaneously, and it is an often-used method for grouping symptoms. For classification of patients, we used latent class analysis, which is a commonly used probabilistic model approach (36). We tried to analyze the symptom pattern by introducing all of the 62 symptoms into the latent class analysis, but we were faced with problems. The interpretation for all models with more than two classes resulted in two large classes of patients identified by having few contra many symptoms. Adding further classes to the LC analysis, apart from the two before-mentioned, yielded very few patients with many symptoms arranged in odd patterns. These small classes were vitiated by large standard errors, which led to convergence problems. We also performed hierarchical cluster analysis and k-means cluster analysis of all the data, which produced similar interpretations as mentioned above. Furthermore, we used k-means cluster analysis on the symptom groups deriving from the principal component factor analysis, which yielded very similar results to the latent class analysis. Because the latent class analysis, contrary to k-means cluster analysis, offers the possibility of testing for model fit statistically, we preferred this method. The results of the statistical analysis seem very robust and easily interpretable, but the data may benefit from applying even more advanced and specialized statistical methods, e.g., hierarchical class analysis (37), which can group symptoms and patients simultaneously or state grade of membership (38).
The bodily distress disorder identified in this study is highly associated with emotional distress, i.e., mood disorder, anxiety, and unspecific emotional distress, which supports our stress hypothesis. This association was also found in studies on the association between various functional somatic syndromes and depression (12,39–41).
We identified no other significant symptom patterns that might support the existence of other functional syndromes described in the medical literature (8,12,42). Excessive and distressing fatigue, from which the patient is unable to recover normally, was a very common symptom, but our findings indicate that it cannot be associated with a distinct disorder or a syndrome. Nor did we find any support for the DSM-IV or ICD-10 diagnostic criteria for somatization disorder as the syndromes detected in this study were different from the symptom groups stated in the DSM-IV and ICD-10 diagnostic criteria. In a worldwide primary care multi-center study, Simon et al. (39) disclosed four symptom groups of which three matched those found in this study. Although our sample included neurological patients contrary to the before-mentioned study, we found no indication of symptoms clustering into a distinct neurological symptom group, nor in the primary care sample alone, a finding which is in accordance with the study of Gara et al. (43). The result of this study indicates that the somatization disorder and related diagnoses and the various functional syndromes described in the medical literature unbacked by observer-independent ‘hard evidence’ may be an artifact of inappropriate research methodology (9,19,20). Multiple symptom profiles or ‘syndromes’ or ‘diagnoses’ may be defined among patients with multiple symptoms if only a certain predefined symptom profile is explored. Likewise, the number of syndromes that may be defined based on principal symptoms like fatigue, pain, dizziness, or headache is abundant among multi-symptomatic patients. Our patient sample is too small to uncover rare syndromes, and we therefore cannot rule out the existence of some rare functional syndromes that may hide behind the broader bodily distress disorder described in this study. Another limitation of the study is that we did not include a general population sample and included only Scandinavians, which means that potential cultural differences in symptom presentations cannot be detected.
Robins and Guze (44) and later Kendell (19) have listed a range of strategies for establishing the validity of clinical syndromes. The first strategy is to identify and describe the syndrome by ‘clinical intuition’ or cluster analysis, and the second is to demonstrate boundaries or ‘point of rarity’ between related syndromes by statistical methods. We have demonstrated that the criteria for bodily distress disorder enjoyed high statistical validity. Clinical and ‘face’ validity also seem good, even compared with physicians’ assessments (Table 5), but confirmation from daily clinical practice is, of course, needed. A third point in the validation process is to perform follow-up studies to establish a distinct course or outcome. This has not been undertaken in the present study, but the included patients are being followed. However, as the symptoms had lasted for a very long period in a high fraction of the bodily distress classes, the present data indicate that the symptoms pursue a distinct course and that the diagnosis remains stable. Kendell (19) proposes three more strategies: a) therapeutic trials to establish a distinct treatment response, b) family studies establishing that the syndrome ‘breeds true,’ and c) demonstration of the association with some more fundamental abnormalities, i.e., anatomical, biochemical, or molecular. Such validation studies as well as cross-validation studies in other samples have still to be planned.
The strengths of this study are that it is based on large, stratified samples of consecutive patients from three different medical settings including primary care, and that we explored all types of functional somatic symptoms and emotional symptoms simultaneously. We used the SCAN interview, which is presently the most comprehensive standardized psychiatric diagnostic tool with a comprehensive section on physical health and symptoms (45). It includes rules on how to exclude medical explanations for symptoms. The interviewers all had several years of training in medicine and surgery in addition to their psychiatric training. In the analysis, we only included the symptoms that, beyond any reasonable doubt, could not be explained by a conventional medical/surgical condition, based on the interview or if necessary medical record review.
|
CONCLUSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONNOTESREFERENCES |
|---|
The study indicates that it is crucial to take all types of subjective symptoms into consideration when defining new syndromes and that it is of paramount importance to test the validity of a new, diagnostic construct in representative patient populations using an appropriate statistical analysis strategy.
We wish to thank the participating patients, the physicians and their secretaries, and the interviewers.
|
NOTES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONNOTESREFERENCES |
|---|
The study was funded by a grant from the Danish Medical Research Council (grant number 9801278 and 9601898), the Health Service of Aarhus County (project number 0871), the Hede Nielsen Foundation, the fund "Puljen til Styrkelse af Psykiatrisk Forskning" and by Biomed1 grant BMHI-CT93-1180. The funding sources had no role in the study design, collection, analysis and interpretation of the data, or writing of the manuscript, or the decision to submit for publication.
DOI:10.1097/PSY.0b013e31802e46eb
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONNOTESREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  S. Skovbjerg, S. Brorson, A. Rasmussen, J. D. Johansen, and J. Elberling Impact of self-reported multiple chemical sensitivity on everyday life: A qualitative study Scand J Public Health, August 1, 2009; 37(6): 621 - 626. [Abstract] [PDF]
|
|
|
|
 |
|
 B. Van Houdenhove and P. Luyten Customizing Treatment of Chronic Fatigue Syndrome and Fibromyalgia: The Role of Perpetuating Factors Psychosomatics, November 1, 2008; 49(6): 470 - 477. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Duncan and M. Oto Predictors of antecedent factors in psychogenic nonepileptic attacks: Multivariate analysis Neurology, September 23, 2008; 71(13): 1000 - 1005. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 W. Rief and G. Rojas Stability of Somatoform Symptoms Implications for Classification Psychosom Med, November 1, 2007; 69(9): 864 - 869. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
