The epidemiology of chronic syndromes that are frequently unexplained: do they have common associated factors?

Abstract

Background Syndromes for which no physical or pathological changes can be found tend to be researched and managed in isolation although hypotheses suggest that they may be one entity. The objectives of our study were to investigate the co-occurrence, in the general population, of syndromes that are frequently unexplained and to evaluate whether they have common associated factors.

Methods We conducted a population-based cross-sectional survey that included 2299 subjects who were registered with a General Medical Practice in North-west England and who completed full postal questionnaires (response rate 72%). The study investigated four chronic syndromes that are frequently unexplained: chronic widespread pain, chronic oro-facial pain, irritable bowel syndrome, and chronic fatigue. Validated instruments were used to measure the occurrence of syndromes and to collect information on a variety of associated factors: demographic (age, gender), psychosocial (anxiety, depression, illness behaviour), life stressors, and reporting of somatic symptoms.

Results We found that 587 subjects (27%) reported one or more syndromes: 404 (18%) reported one, 134 (6%) reported two, 34 (2%) reported three, and 15 (1%) reported all four syndromes. The occurrence of multiple syndromes was greater than would be expected by chance (P < 0.001). There were factors that were common across syndromes: female gender [odds ratio (OR) = 1.8; 95% confidence interval (95% CI) 1.5–2.2], high levels of aspects of health anxiety like health worry preoccupation (OR = 3.5; 95% CI 2.8–4.4) and reassurance seeking behaviour (OR = 1.4; 95% CI 1.1–1.7), reporting of other somatic symptoms (OR = 3.6; 95% CI 2.9–4.4), and reporting of recent adverse life events (OR = 2.3; 95% CI 1.9–2.8).

Conclusion This study has shown that chronic syndromes that are frequently unexplained co-occur in the general population and share common associated factors. Primary care practitioners need to be aware of these characteristics so that management is appropriate at the outset.

Introduction

Unexplained symptoms are defined as symptoms for which no objective physical or pathological changes can be found. Over the years several terms have been coined to describe these symptoms, e.g. psychosomatic syndromes and symptoms, conversion disorders, somatoform disorders, functional somatic syndromes and symptoms, and medically unexplained symptoms.¹ Symptoms often present in different anatomical regions of the body leading to variations in presentation and reporting between individuals. Medical specialties have, therefore, further split these symptoms into a number of diagnostic categories or syndromes e.g. irritable bowel syndrome (IBS) (gastroenterology), chronic widespread pain (CWP) (rheumatology), temporomandibular joint dysfunction, atypical facial pain (dentistry), chronic fatigue (infectious diseases).² These are only a few examples and in fact most medical specialties seem to have at least one unexplained syndrome.²

There is overwhelming evidence to show that these syndromes have psychological risk factors. Prospective epidemiological study has shown that somatic symptoms and Illness behaviour strongly predict onset of CWP³ whereas psychosocial factors including high levels of fatigue, psychological distress, aspects of illness behaviour, and high levels of anxiety strongly predict onset of non-specific abdominal pain.⁴ Similarly, another study showed that subjects developing chronic fatigue had a 10-fold odds of having current psychiatric disorder [odds ratio (OR) = 10.1; 95% confidence interval (95% CI) 4.0–26.0].⁵ However, these studies have investigated specific syndromes in isolation. Studies in clinic settings^6–9 have shown that these syndromes are common in persons attending hospital outpatient clinics and that some patients report multiple syndromes. A study of outpatient clinics found that 52% of new attendees to these clinics had at least one medically unexplained symptom.⁶ In addition, patients reporting these syndromes shared common characteristics notably female predominance, younger age group, and altered health-seeking behaviour. Psychiatric morbidity was more common in those reporting multiple syndromes.⁶

The findings from such studies indicate that there may be common factors associated with unexplained syndromes. It is unclear whether these syndromes are discrete separate entities or whether they encompass a continuum of syndromes with common risk factors. Consider, for example, the scenarios for two seemingly different syndromes: (i) the syndromes may be discrete entities with independent risk factors, (ii) they may have common risk factors, or (iii) there may be common and syndrome-specific risk factors. Although the true relationship remains to be elucidated, evidence from current studies¹⁰ of unexplained syndromes appears to favour the scenario for the existence of common and syndrome-specific factors. However, these studies that have been conducted in clinic and hospital settings have limited external validity because they include small sample sizes of a selected group of patients referred from a primary care setting. This usually represents the more severe or intractable cases that may be more likely to share certain features, e.g. psychological factors, and importantly excludes persons who do not consult for their syndromes and have successfully developed strategies to cope with their pain and fatigue.

We, therefore, wished to determine in an unselected sample, the co-occurrence of four frequently unexplained syndromes [CWP, IBS, chronic oro-facial pain (OFP), and chronic fatigue] and whether there are common factors associated with the reporting of these syndromes.

Methods

Study design and participants

The study was a population-based cross-sectional survey of randomly selected subjects aged 18–75 years and registered with a general medical practice in Handforth, a commuting suburb on the outskirts of Manchester, UK. Subjects were requested to complete a postal questionnaire between September 2003 and June 2004. Ethical approval for the survey was granted by the Macclesfield Research Ethics Committee, East Cheshire National Health Service Trust.

Mailing strategy

All sampled subjects received, by mail, a copy of the study questionnaire that was headed by a covering letter from their General Practitioner inviting them to participate. Also included was a reply paid stamped-addressed envelope. Non-responders were mailed a postcard reminder 2 weeks after the first mailing. Subjects who had still not responded within 2 weeks of receiving the postcard were mailed another full study questionnaire. Finally, a short (two page) questionnaire was mailed to the remaining non-responders after a further 2 weeks. This inquired about facial pain symptoms only. A telephone questionnaire was also used to interview subjects who had still not participated. Again this inquired about facial pain symptoms only.

Study questionnaire

The full study questionnaire inquired about demographic details (age, gender), chronic syndromes (CWP, chronic facial pain, IBS, and chronic fatigue), and associated factors (anxiety, depression, health anxiety, sleep disturbance, somatic symptoms, and adverse life events).

Definition and measurement of syndromes

Chronic widespread pain

This was defined according to the American College of Rheumatology (ACR) criteria for classification of widespread pain for fibromyalgia.¹¹ Briefly, the criteria require the presence of pain in the left and right side of the body, pain above and below the waist, and pain in the axial skeleton (cervical spine, anterior chest, thoracic spine, or low back). To be defined as chronic, pain was required to have been present for at least 3 months. To satisfy the definition for fibromyalgia,¹¹ in addition to having widespread pain subjects must have pain in 11 of 18 tender point sites on digital palpation. Tender point examination was not carried out in the current study, which measured the presence of CWP and not fibromyalgia.

Pain was recorded using four blank body manikins outlining the front, back, and sides (right and left) of the body. Subjects were asked to shade on the manikin where they had felt the pain. The construct validity of these methods has been demonstrated in previous epidemiological studies.^12–14 For the purpose of the current study, head pain was excluded from the definition of CWP, because chronic facial pain was being considered as a separate outcome.

Chronic OFP

OFP was defined as pain in the face, mouth, or jaws, which had been present for a day or longer in the past month. Chronic OFP was defined as OFP that had been present for 3 months or longer. OFP was recorded using four blank manikins comprising the face and mouth. Three manikins of the head and neck recorded extra-oral pain while a further manikin including the teeth and associated soft-tissues identified location of intra-oral pain. Subjects were asked to shade on the manikin(s) where they had felt the pain. The construct validity of these methods has been demonstrated in a previous population-based study¹⁵ and in further studies investigating the classification of unexplained OFP (unpublished data).

Irritable bowel syndrome

This was defined using the Rome II criteria¹⁶ that require the presence of abdominal pain or discomfort for 12 weeks of the previous 12 months associated with at least two of the following: (i) relief with defecation, (ii) looser or more frequent stools, and (iii) harder or less frequent stools. The Rome II criteria are the most recent for identifying subjects with IBS and have resulted in standardization across studies, aiding comparability.

Chronic fatigue

Subjects were defined as having chronic fatigue if they experienced fatigue for 6 months or longer with fatigue scores of 8 or greater on the Fatigue Scale.¹⁷ The Fatigue Scale has been used extensively in large community samples to determine the prevalence of chronic fatigue and its construct validity has been demonstrated in identifying associated factors.^18–20 The 11-item scale is easy to administer, has good face validity and reasonable discriminant validity.¹⁷ It inquires about symptoms of physical and mental fatigue. Each item is allocated four responses: (i) better than usual, (ii) no more than usual, (iii) worse than usual, and (iv) much worse than usual. For the purpose of analysis, the four responses were dichotomized with responses (i) and (ii) having scores of 0 and (iii) and (iv) a score of 1. This gave a total fatigue score of 11 from which a cut-off score of 8 or more was used to define chronic fatigue cases. It is important to note that we were not measuring the occurrence of chronic fatigue syndrome as this requires exclusion of medically explained causes of fatigue by clinical evaluation, which was not feasible. Chronic fatigue unexplained by current medical knowledge that fails to meet the criteria for chronic fatigue syndrome has been classified as idiopathic chronic fatigue^19–21 and is the definition used in the current study.

Measurement of associated factors

Psychological distress (anxiety and depression)

The hospital anxiety and depression (HAD) scales²² were used to measure psychological distress. The 14 items measure degrees of anxiety and depression and concentrate on psychological symptoms in the past week. Importantly, the HAD scale avoids items that may be attributed as physical symptoms arising from, e.g. pain symptoms and concentrates on the psychological symptoms of moods. The scale includes seven items for anxiety and seven for depression. Each item records a response on a 0–3 scale giving a total score of 21 for each subscale. Scores, on each subscale, of 0–7 represent non-cases, 8–10 indicate a borderline case, whereas 11–21 represent probable cases of a mood disorder (anxiety, depression). It is important to note that the HAD scale like any other self-report measure does not allow a clinical diagnosis of a mood disorder although symptom scores on the HAD scale correlate well with clinical diagnosis of an anxiety and depressive disorder with correlations of 0.74 and 0.70, respectively.

Health anxiety

The health anxiety questionnaire²³ was used to assess aspects of illness behaviour and identify individuals with persistent health anxiety who may be relatively unresponsive to routine medical reassurance. The scale comprises 21 items, 14 of which are derived from the illness attitude scale.²⁴ The items are divided into four subscales: health worry and preoccupation (8 items), fear of illness and death (7 items), reassurance seeking behaviour (3 items), and interference with life (3 items). Each item is scored on a four point Likert scale in a range of 0–3 giving a total score between 0 and 63.

For the purpose of analysis, the overall health anxiety score and the two larger subscales were categorized into tertiles while the remaining subscales were dichotomized.

Sleep problems

Sleep disturbance was measured using a validated 4-item sleep problem scale.²⁵ This was used to assess the impact of various syndromes on sleep patterns and quality of sleep. Each question in the scale asked about problems with sleep in the past month and responses were scored in a range of 0–5 giving a total score of between 0 and 20. For the purpose of analysis, the total sleep score was divided into tertiles.

Somatic symptoms

The somatic symptom checklist comprising six items common to both sexes²⁶ was used as a screening tool for somatization disorder. It has been used in population-based studies³ to examine the propensity to report the lifetime experience of a number of somatic symptoms and its construct validity demonstrated. For the purpose of analysis, subjects with no somatic symptoms were used as the reference category while subjects with one or more symptoms were dichotomized. Further, the question in the scale that related to pain symptoms was excluded because pain was being considered as a separate outcome.

Life stressors

The list of threatening experiences, a 12 event inventory initially modified by Brugha et al.²⁷ from a 67 life events inventory,²⁸ was used to gather information on recent adverse life events. The 12 categories are associated with a significant marked or moderate long-term contextual threat rating. The categories ask about personal relationships, employment, illness, and financial and legal problems in the last 6 months. Each item response scores 0–1 giving a maximum life events score of 12. For the purpose of analysis, the life events score was dichotomized and the question on personal illness and injury was omitted as this was being analysed as a separate outcome.

Statistical analysis

Multi-level modelling was used to take into account that there were four syndromes measured on each individual, and these syndromes may not be independent from each other. This technique takes into account that the correlation of syndromes within individuals will be greater than that between individuals. Each syndrome was thus treated as a within subject factor called ‘type’ with four levels representing the four syndromes. Other variables measured at the subject level formed the between-subject covariates. These included associated factors like anxiety, depression, somatization etc. Initially, syndrome-specific associations were calculated using a population average model. A term for the interaction between ‘type’ and the associated factor was then added to the model, and a Wald test carried out to investigate whether the effect of the associated factor was similar across all four syndromes, while taking into account within-subject correlation of having multiple syndromes. The Wald test provides P-values to assess the interaction of ‘type’, which represented the four syndromes, with the associated factor, e.g. anxiety, depression, etc. The null hypothesis is that there is no difference across syndromes, i.e. the effect of the associated factor is common. Therefore, small P-values (P < 0.05) would indicate that differential effects are likely, i.e. reject the null hypothesis while larger P-values (P ≥ 0.05) indicate that a common effect is plausible, i.e. accept the null hypothesis. In cases where there was no evidence of a different magnitude of association (P ≥ 0.05), the common effect estimate was obtained from the model. All associations are presented as ORs with 95% CIs and relationships were adjusted for age and gender. Associated factors like anxiety, depression, sleep disturbance, and interference with life were not included in the multivariate analysis for all four syndromes because items on the fatigue scale were similar to some items on the corresponding scales used to measure these factors: HAD, sleep scale, and the interference with life subscale of health anxiety. Indeed the correlation of the fatigue scale with these scales (sleep scale, HAD, interference with life subscale of health anxiety) was >0.5 while for the other scales the correlations were 0.3 or lower. The common effect of these factors was, therefore, measured across three syndromes (CWP, chronic OFP and IBS) with the exclusion of chronic fatigue. All analysis were carried out using STATA (version 8).²⁹

Results

Study response rates and participation

Prior to the mailing, we found 732 subjects ineligible as they had migrated, died, were not at the mailing address, or were not on the electoral roll and, therefore, unlikely to be resident at the address and to receive the questionnaire. This left 3468 eligible subjects of whom 2505 returned completed questionnaires resulting in a study participation rate of 72% (Figure 1). Full questionnaire data was available for 2299 (92%) subjects who were included in the analysis.

Prevalence and co-occurrence of unexplained syndromes

CWP was reported by 340 subjects (15%), chronic OFP by 163 (7%), IBS by 211 (9%), and chronic fatigue by 173 (8%). Overall, 587 subjects (27%) reported one or more syndromes: 404 (18%) reported one syndrome, 134 (6%) reported two, 34 (2%) reported three, while 15 (1%) reported all four syndromes (Table 1). The occurrence of multiple syndromes (n = 183) was significantly greater (χ²; P-value <0.001) than would be expected by chance (n = 102) based on the prevalence of syndromes (Table 1).

Factors associated with unexplained syndromes

Age and gender

The age-related distribution for the syndromes is displayed in Figure 2. Whereas the prevalence of subjects with CWP clearly increased with age that of IBS and chronic OFP generally decreased with age. Chronic fatigue had a fairly constant prevalence across the 5th–7th decades with lower prevalence at younger and older ages (Figure 2). The proportion of subjects with no syndromes was similar across age groups (19–22%). Two-thirds of the subjects reporting each syndrome were female compared with only half of those reporting none of the syndromes (Table 2).

Associated features

The proportion of subjects in the highest scoring category for all the measured factors was much greater for those who reported syndromes when compared with subjects with no syndromes (Table 2). For example, the proportion of subjects reporting the highest level of sleep disturbance ranged from 54 to 71% for the four syndromes while the corresponding proportion for those with no syndromes was only 23% (Table 2). Similarly, the proportion of subjects in the lowest scoring category for all associated factors was much lower for those with syndromes when compared with subjects with no syndromes (Table 2). Again using sleep score as an example, the proportion of subjects reporting the lowest level of sleep disturbance ranged between 5 and 18% for those with syndromes while the corresponding proportion for those without a syndrome was much larger at 49%.

The proportion of subjects in the highest scoring category for each of the associated factors was similar across syndromes although chronic fatigue had notably higher proportions of subjects with the highest level of sleep disturbance and anxiety cases when compared with the other syndromes (Table 2).

Associated features: common or syndrome specific

Table 3 shows the strength of relationships between the associated features and the type of syndrome reported. The strength of associations estimated using ORs showed a similar distribution. Subjects reporting the highest scores for each of the associated factors had significantly increased odds of reporting each of the syndromes. The ORs were similar across syndromes, although there were stronger relationships between chronic fatigue and sleep disturbance, anxiety, depression, and the ‘interference with life’ subscale of health anxiety (Table 3). The data relating to reporting of somatic syndromes, adverse life events, and high levels of health anxiety (health worry preoccupation and reassurance seeking behaviour) were consistent with common effects across the syndromes, i.e. Wald test P-values >0.05. The common strength of effect estimated was highly significant for the highest scoring category of the common associated factors. For example, for subjects reporting high levels of aspects of health anxiety such as health worry preoccupation it was 3.5 (95% CI 2.8–4.4) and reassurance seeking behaviour it was 1.4 (95% CI 1.1–1.7), amongst subjects reporting more than one somatic symptom it was 3.6 (95% CI 2.9–4.4) and amongst subjects reporting two or more life events it was 2.5 (95% CI 2.0–3.0) (Table 3). Where common effects were measured across three syndromes (CWP, chronic OFP, and IBS), there was also no evidence of different associations (Wald test P-value >0.05) and the common effect estimated was 3.8 (95% CI 2.9–5.0) for syndrome reporting associated with the highest levels of depression, 3.2 (95% CI 2.6–4.0) for the highest levels of anxiety, 4.4 (95% CI 3.5–5.6) for the highest level of sleep disturbance, and 5.6 (95% CI 4.6–7.0) for the highest scores on the interference with life subscale of health anxiety.

Comparison of responders and non-responders

Non-responders included those subjects who did not reply but were on the electoral roll (n = 604), those who returned blank questionnaires (n = 353), and those who had ‘wasted questionnaires’ (n = 6), i.e. subjects for whom questionnaires were returned but had been completed by somebody else (Figure 1). When compared with non-responders, responders (n = 2505; Figure 1) were more likely to be female (55% of responders were female compared with 46% of non-responders; P < 0.001) and older [median age 48 (IQR 38–60) vs median age 40 (IQR 30–53); P < 0.001]. In addition 186 (7%) subjects who responded returned a short questionnaire only and were not included in the analysis. Consistent with the profile of non-responders, these subjects were significantly younger [median age 42 (IQR 38–60); P < 0.001] and more likely to be male (57% males compared with 44% of those who returned full questionnaires; P = 0.001).

Discussion

This study suggests the co-occurrence of chronic syndromes that are frequently unexplained in an unselected general population sample. In addition, these chronic syndromes were found to share common associated factors such as aspects of health anxiety, for example health worry preoccupation and reassurance seeking behaviour, reporting of other somatic syndromes, reporting of recent adverse life events, and gender with a female predominance.

There are some methodological issues that warrant consideration. First, a cross-sectional study design only allows evaluation of the relationship between associated factors and syndrome reporting at a specific point in time. Therefore, some factors studied may not be risk factors for syndrome reporting but rather a consequence of the syndromes. Nevertheless, prospective population-based studies on specific syndromes^3,4 have shown that health anxiety and reporting of other somatic symptoms were found to predict the onset of CWP and non-specific abdominal pain. This reinforces the findings from our study and may suggest that health anxiety and reporting of other somatic symptoms, which were common factors associated with the four unexplained syndromes in our study, may be risk factors rather than a consequence of syndromes. This will need confirmation in prospective studies.

Second, we could not examine the association of factors like sleep disturbance, anxiety, depression, and interference with life across all four syndromes because the scales used to measure these factors were highly correlated with the fatigue scale. However, the strength of the relationship for these factors was similar across the other three syndromes: CWP, chronic OFP, and IBS. The occurrence of anxiety and depression as common factors is supported by previous studies in clinic settings.³⁰

Third, some cases of reported syndromes may have an organic cause. However, the inclusion of such cases with a different aetiology will have diluted the magnitude of effects observed in our study. Further, population-based studies that have investigated some of these syndromes in isolation have shown that the majority of cases identified by completion of questionnaire-based criteria are indeed unexplained.^4,20 In addition, the use of questionnaire-based criteria will result in random misclassification of cases, i.e. subjects will not be misclassified according to their exposure status. This may not hold true for clinic/hospital-based studies where cases of unexplained syndromes are verified by clinical examination and, therefore, the clinician/interviewer may introduce bias as he/she is often not blinded to the subject's exposure status.

The findings of our study are in agreement with other epidemiological studies that have investigated these syndromes in isolation. Age and gender variations of syndrome reporting were like previous studies whereby the prevalence of reported CWP was found to increase with age³¹ while that of IBS³² and OFP³³ was found to decrease with age. Studies on chronic fatigue also found its prevalence to be lowest in the younger and older age groups.²⁰ These studies also found that females were more likely to report syndromes. Furthermore, associated factors were similar to other studies that found overwhelming evidence of psychological factors (reporting of somatic symptoms, illness behaviour, and high levels of psychological distress) associated with each of these syndromes^3,4,20,34 and in some studies these factors were also found to predict the onset of syndromes.^3,4 The age variation of syndrome reporting raises an important issue—whether syndromes like IBS and chronic OFP present in earlier years and may be risk factors for the development of other syndromes like CWP, which manifest more frequently in later years; a question that needs to be addressed by future research.

The findings of our study have important clinical implications. Patients initially present to primary care practitioners with unexplained syndromes and are often treated for their presenting syndromes while co-existing syndromes are ignored. This results in referral to hospital clinics where management is focused on finding an organic cause for these syndromes, e.g. patients with chronic OFP presenting to dental clinics. The co-occurrence of these syndromes suggests that a thorough history may reveal co-existing syndromes for which the patient is not seeking help and, therefore, that management of regional syndromes is unlikely to be useful. Further, inquiry into illness behaviour, other somatic symptoms, and a social history regarding recent adverse life events may shed light on associated factors. This can lead to early recognition of these syndromes and management can then focus on the use of techniques like cognitive behavioural therapy, which attempt to change aspects of patients' thoughts and behaviour to help them cope with their syndromes. Such techniques have been tested in intervention trials on patients who consult repeatedly for medically unexplained complaints^35,36 and have shown significant reductions in consultations, symptom reporting, and psychological distress. Reduction in consultations among primary care patients with unexplained physical symptoms has also been achieved by other techniques like group aerobic exercise programmes.³⁷

Future research, therefore, needs to be conducted in the form of prospective cohort studies to determine whether associated factors identified in this study, such as health anxiety and reporting of other somatic symptoms and recent adverse life events, predict the onset of these syndromes. This may also shed light on whether having one unexplained syndrome is a risk marker for the development of multiple syndromes. In addition, randomized controlled trials will be required to determine whether early intervention in the natural history of unexplained syndromes can improve outcome.

Overall, this population-based study has unravelled important characteristics, in the general population, of syndromes that are frequently unexplained. Primary care practitioners (both medical and dental), to whom persons with these syndromes initially present, need to be aware of these characteristics so that management is appropriate at the outset.

The authors are most grateful to Miss Lisa Fulluck, Mr Phil Steer, and staff/patients of The Handforth Health Centre and Handforth Clinic for their help with the study. V.R.A. is supported by a Wellcome Trust Fellowship in clinical epidemiology. The study was funded by the Wellcome Trust and The Arthritis Research Campaign, Chesterfield, UK.

References