Cost-effectiveness of interventions for medically unexplained symptoms: A systematic review

Background In primary and secondary care medically unexplained symptoms (MUS) or functional somatic syndromes (FSS) constitute a major burden for patients and society with high healthcare costs and societal costs. Objectives were to provide an overview of the evidence regarding the cost-effectiveness of interventions for MUS or FSS, and to assess the quality of these studies. Methods We searched the databases PubMed, PsycINFO, the National Health Service Economic Evaluation Database (NHS-EED) and the CEA registry to conduct a systematic review. Articles with full economic evaluations on interventions focusing on adult patients with undifferentiated MUS or fibromyalgia (FM), irritable bowel syndrome (IBS) and chronic fatigue syndrome (CFS), with no restrictions on comparators, published until 15 June 2018, were included. We excluded preventive interventions. Two reviewers independently extracted study characteristics and cost-effectiveness data and used the Consensus on Health Economic Criteria Checklist to appraise the methodological quality. Results A total of 39 studies out of 1,613 articles met the inclusion criteria. Twenty-two studies reported costs per quality-adjusted life year (QALY) gained and cost-utility analyses (CUAs). In 13 CUAs the intervention conditions dominated the control conditions or had an incremental cost-effectiveness ratio below the willingness-to-pay threshold of € 50,000 per QALY, meaning that the interventions were (on average) cost-effective in comparison with the control condition. Group interventions focusing on MUS (n = 3) or FM (n = 4) might be more cost-effective than individual interventions. The included studies were heterogeneous with regard to the included patients, interventions, study design, and outcomes. Conclusion This review provides an overview of 39 included studies of interventions for patients with MUS and FSS and the methodological quality of these studies. Considering the limited comparability due to the heterogeneity of the studies, group interventions might be more cost-effective than individual interventions. Registration Study methods were documented in an international prospective register of systematic reviews (PROSPERO) protocol, registration number: CRD42017060424.


Introduction
Patients with medically unexplained symptoms (MUS, i.e. physical symptoms for which no pathological cause can be found after adequate physical examination) are highly prevalent in primary and secondary care in all medical settings [1,2]. The classification of these physical symptoms is problematic as numerous overlapping diagnoses and syndrome labels show [3]. Almost each medical specialty has defined its own syndrome(s) based on symptoms that relate to their organ of interest [4]. Psychiatry uses the designation somatic symptom disorder, while most medical specialties have patients with clusters of MUS within so called 'functional somatic syndromes' (FSS) [1] e.g. fibromyalgia (FM) [5], irritable bowel syndrome (IBS) [6], chronic fatigue syndrome (CFS) [7], chronic benign pain syndrome and multiple chemical sensitivity (MCS) [4,8]. The most well-known FSS are FM, IBS, and CFS [9] and most primary care physicians and researchers are familiar with the umbrella term MUS [10].
MUS are often accompanied by psychological distress, social isolation and reduced quality of life [11,12]. Severe MUS are associated with multiple functional impairments and psychiatric morbidity [13][14][15]. Patients with MUS and FSS suffer from their symptoms, are functionally impaired [9] and are at risk for false-positive diagnostic tests, potentially harmful additional testing and treatment procedures [16]. Therefore, these symptoms constitute a major burden on patients and society with considerable societal costs, health care costs and costs of lost productivity [9]. In a Dutch study (2005)(2006)(2007)(2008)) the mean total cost, both the use of healthcare services (direct costs) and productivity-related costs (indirect costs), was estimated to be € 6,815 per patient per year [12]. In a German study (2007)(2008)(2009), outpatient physician visits were the most expensive single cost category of the direct costs and indirect costs were predominantly caused by productivity reduction at work [17].
Little is known about the cost-effectiveness and methodological quality of economic evaluations of interventions for patients with MUS and FSS. Although helpful for policy makers, systematic reviews of cost-effectiveness data in this area are scarce. Earlier, Konnopka et al. [18] published a systematic review of health economics studies for MUS. The aim of that systematic review was to give an overview of both cost-of-illness studies and economic evaluations for patients with MUS. Since in the review by Konnopka et al. [18] the quality of the included studies was not addressed and the included studies were only up to 2008, we consider an update is due. Therefore, the objectives of this review are to provide an overview of the evidence regarding the cost-effectiveness of interventions for patients with MUS and FSS, and to assess the methodological quality of the identified economic evaluations.

Methods
The methods and reporting of this systematic review are in concordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (S1 Table) [19]. Prior to the start of article inclusion, we documented study methods in an international prospective register of systematic reviews (PROSPERO) protocol (S1 Text), registration number CRD42017060424.

Literature search and study selection
We performed a literature search until 15 June 2018 in the following databases: PubMed, Psy-cINFO, the National Health Service Economic Evaluation Database (NHS-EED), and the CEA registry. The NHS-EED is a health economic database including economic evaluations. The CEA registry includes studies in which a cost-effectiveness analysis was performed. In addition to free-text terms, we used Medical Subject Headings (MeSH) terms and Psychological Index Terms for searches within the PubMed and PsycINFO databases, respectively. In order to identify economic evaluations on MUS, we selected key terms that were used in a Cochrane review on non-pharmacological interventions for somatoform disorders and medically unexplained physical symptoms (MUPS) in adults [20] and combined these with health-economic key terms. An information specialist was involved in the development of the search strategy. A detailed description of the search strategy for every database can be found in the supplementary files (S1 Appendix). Additionally, we checked existing systematic reviews and the references of studies included in our review manually for relevant studies.
We included studies on adult patients with MUS, reporting on psychological, physical/exercise, internet-delivered, pharmacological and combined interventions compared with usual care, waiting list, other physical or psychological treatment and describing health care use or societal costs.
Only studies reporting on full economic evaluations were included, meaning that the studies compared both costs and effects of two or more conditions [21]. We excluded studies when interventions focused on prevention or screening. We limited the scope of the studies to adult patients with undifferentiated MUS and the three most common specific functional syndromes FM, IBS and CFS. We excluded studies with medically (partly) explained symptoms or medically unexplained symptoms as secondary diagnosis. The literature search was limited to publications written in English, Dutch and German. We included both trial-based economic evaluations (TBEEs) and model-based economic evaluations (MBEEs). In TBEEs costs and effects are measured alongside an effectiveness trial, whereas in MBEEs available evidence is synthesized and used to simulate (often long term) effectiveness and costs. We excluded study protocols and included only original research.
Titles and abstracts of the search results were independently screened by two reviewers (MSHW and JL). Studies that were in agreement with the inclusion criteria based on title and abstract were retrieved as full text. Disagreements about the eligibility of studies were resolved in a consensus meeting. A third reviewer (BV) was available in case of disagreement.
The full text articles were evaluated independently by the two reviewers (MSHW and JL) to assess eligibility. In a consensus meeting the full text articles were discussed and discrepancies between the two researchers were resolved by consensus, and when needed a third researcher (BV) was consulted.

Data extraction
Two reviewers (MSHW and JL) independently extracted data and assessed the methodological quality of each study. The articles excluded on full-text level were documented and are provided in the supplementary files (S2 Table). The development of the data extraction form was based on a previous review by one of the authors (JL) [22]. To pilot this data extraction form, the reviewers screened the first eight articles together. After adaptation of the draft extraction form, we extracted from each of the included articles the following information: name first author, country, study design economic evaluation, target population, perspective, time horizon, treatment alternatives (intervention, comparators and sample size), effect measurement and valuation, discount rates, valuation year, costs categories, incremental costs, incremental effects and health economic results.

Quality assessment
The methodological quality of the studies was assessed with the extended Consensus on Health Economic Criteria (CHEC) list [23], which is recommended by the Cochrane Handbook for Systematic Reviews of Interventions [24] for critical appraisal of the methodological quality of health-economic evidence. The checklist contains 20 items covering the quality of the design and reporting of the economic evaluation studies. Although the CHEC is not optimal for assessing the methodological quality of MBEEs, we chose the CHEC for the quality assessment of both TBEEs and MBEEs in order to optimize comparability of the results. Each question on the CHEC checklist was scored with either 'Yes' (score 1), 'Suboptimal' (score 0.5), 'No' (score 0), 'NA' (not applicable) or 'Uncertain' (no score). The 'Uncertain' option was used only when information on an item was not entirely clear. We did not contact authors when the published information was insufficient to assign a score.
Prior to the quality assessment, to improve uniform scoring, two reviewers (MSHW and JL) independently assessed and discussed eight included studies (two of each target population: MUS, FM, IBS and CFS). A detailed description of the scoring instructions is provided in the supplementary files (S2 Appendix). Two reviewers (MSHW and JL) assessed the quality of each study independently. Disagreements between the two reviewers were resolved in a consensus meeting.

Outcomes
For each study, we extracted the incremental costs, incremental effects and incremental cost-effectiveness ratio (ICER), indicating the costs per additional quality-adjusted life year (QALY) or any other (clinical) outcome. To enhance comparability of the health economic results between studies conducted in different countries and at different years, ICERs were converted to 2016 Euro using Purchasing Power Parity (PPP) rates [25] and the Consumer Price Index [26]. To assess cost-effectiveness for studies reporting the cost per additional QALY, one year in perfect health, we applied an overall willingness-to-pay (WTP) threshold of € 50,000 per QALY, a commonly used threshold in the Netherlands [27]. WTP thresholds are not available for other outcome measures. The WTP threshold refers to the maximum amount a country or society is willing to pay for a particular health gain [28]. When an ICER is below the WTP threshold, the intervention can be regarded as on average cost-effective in comparison with the comparator. In accordance with the Consolidated Health Economic Evaluation Reporting Standards (CHEERS), we distinguished healthcare and societal economic perspectives [29]. Due to heterogeneity, a meta-analysis could not be conducted.

Literature search and study selection
In total, the search strategy yielded 1,713 articles. One study was found by additional reference searching. After excluding 101 duplicates, the titles and abstracts of 1,613 articles were screened for relevance. Title and abstract screening resulted in the exclusion of 1,535 articles, mainly because they were not (full) economic evaluations or not primarily focused on MUS. Of the 78 articles that were assessed full-text, 39 were excluded for being not full-economic evaluations (n = 25), not primarily focused on MUS or FSS (n = 3), or not being original research (n = 11). Finally, 39 articles were included for analysis. A flow diagram of the study identification process is presented in Fig 1.

Data extraction
Overview of the included studies. The main characteristics of the included studies are presented in Table 1. The most recent study was published in July 2017 [30], and the oldest study was published in 1992 [31]. Most studies were conducted in Europe (n = 30): UK (n = 11), the Netherlands (n = 6), Germany (n = 3), Spain (n = 4), Sweden (n = 2), Denmark (n = 2), Scotland (n = 1) and Norway, Sweden, Finland, Denmark (n = 1). The remaining studies (n = 9) were conducted in the USA (n = 7), Mexico (n = 1) and one study had a global scope.
The majority (n = 30) of the studies were trial-based economic evaluations (TBEEs) of which 24 originated from randomized trials. Nine economic evaluations were based on economic modelling (MBEE).
In the included studies a societal perspective (n = 14) or a healthcare perspective (n = 15) or both perspectives (n = 9) were used. One study did not explicitly report the study perspective [32].
Studies focusing on patients with FM (n = 10) compared a pharmacological intervention with another pharmacological intervention (n = 4) [42,47,52,53], a group-based therapy (n = 2), acceptance and commitment therapy (ACT) [30] and cognitive behavioral therapy (CBT) [36] compared with a pharmacological intervention and treatment as usual or waiting list. An educational intervention (n = 2) was compared with usual care or an educational discussion group [58,60] and an aquatic exercise program (n = 1) [37] or spa treatment (n = 1) [59] was compared with usual care.
Effects, costs and uncertainty. Information on effect measurement and valuation is described in Table 2. Seventeen studies included only a cost-effectiveness analysis (CEA), fifteen studies only a cost-utility analysis (CUA), seven studies included both a CEA and a CUA. In studies with CEAs, outcomes were expressed as costs per unit improvement on a (clinical) outcome measure. In studies with CUAs, outcomes were expressed as costs per QALY gained, where the majority of the CUAs (n = 14) elicited utilities using the EuroQol 5D (EQ-5D).

(Continued)
Cost-effectiveness of interventions for medically unexplained symptoms Cost-effectiveness of interventions for medically unexplained symptoms

Outcomes
The incremental costs, incremental effects and health economic results in terms of ICERs of the reference-cases are presented in Table 2.  [42,47] or IBS (n = 2) [34,49] and two individual interventions focusing on IBS (n = 1) [51] or CFS (n = 1) [68] appeared to be cost-effective in comparison with the control condition.

Medically unexplained symptoms.
In studies focusing on MUS, four studies assessed the cost-effectiveness of group interventions [46,48,54,56] and one study an individual intervention [55]. The group interventions, group CBT program (STreSS) [46], collaborative group intervention [54], and cognitive-behavioural group training [48] appeared to be cost-effective, but the mindfulness-based cognitive group therapy [56] was not cost-effective. The individually administered psychodynamic interpersonal therapy [55] was not cost-effective either. Each of these interventions was compared to enhanced usual care or enhanced medical care, except for the use of a wait-list condition in Visser et al. [48].
Fibromyalgia. In the studies focusing on FM, the aquatic exercise program [37] appeared to be cost-effective whereas spa treatment [59] was not cost-effective; both group interventions were compared to usual care. Three other studies on group-based cognitive behavioral therapy (CBT) [36], additional psychoeducation [58] or group based acceptance and commitment therapy (ACT) [30] were cost-effective compared to recommended pharmacologic treatment or treatment as usual or waiting list or usual care. In an older study [60] educational cognitive intervention was not cost-effective in comparison with an educational discussion group. In two studies, both funded by the pharmaceutical industry, a pharmacological intervention was compared with another pharmacological intervention. Duloxetine as second-line treatment was cost-effective [42] and pregabalin 450 mg appeared to be cost-effective in comparison with duloxetine 120 mg, but pregabalin 450 mg was not cost-effective compared to tramadol, amitriptyline or placebo [47].

Irritable bowel syndrome.
Three studies focused on IBS [34,35,49], all funded by the pharmaceutical industry, compared a pharmacological intervention with another pharmacological intervention or placebo. Linaclotide appeared to be more cost-effective in comparison with antidepressants [49] and with lubiprostone [34]. Tegaserod [35] did not appear to be more cost-effective than placebo. Sacral nerve stimulation, a non-pharmacological intervention [51], appeared to be cost-effective in comparison with no treatment. Another non-pharmacological intervention, acupuncture as adjunct to usual care [61], was not cost-effective in comparison with usual care.
Chronic fatigue syndrome. In studies including patients with CFS, the interventions pragmatic rehabilitation and supportive listening appeared to be not cost-effective compared Table 3. Quality assessment CHEC-extended. Cost-effectiveness of interventions for medically unexplained symptoms to treatment as usual [44]. On average, individual cognitive behavior therapy (CBT) was more cost-effective in comparison with specialist medical care [68]. In another study [45] CBT was compared with guided support groups (SG) and the natural course. CBT was less costly and more effective than SG and even cost-effective in comparison with the natural course of the disease. Compared to CBT, multidisciplinary rehabilitation treatment [66] appeared not to be cost-effective.

Main findings
To our knowledge, this is the first systematic review of cost-effectiveness of interventions for undifferentiated MUS and the three most well-known functional syndromes FM, IBS and CFS with a methodological quality assessment of the included studies. We identified 39 full economic evaluations of interventions for treating patients with MUS and FSS. Heterogeneity of the included studies concerning interventions, time horizon, and outcome was high. Twentytwo out of 39 studies included a CUA with QALYs as outcome. In 13 CUAs the intervention conditions dominated the control conditions or had an ICER below the WTP threshold of €  Cost-effectiveness of interventions for medically unexplained symptoms 50,000 per QALY, meaning that the interventions were (on average) cost-effective in comparison with the control conditions. In nine CUAs the intervention condition was not cost-effective compared with the control conditions. Group interventions focusing on MUS (n = 3) or FM (n = 4) might be more cost-effective than individual interventions.

Discussion of the results
Although this study provides valuable information regarding existing evidence on the costeffectiveness of interventions for MUS, the comparability of the studies included in this systematic review was limited due to heterogeneity in terms of interventions, (economic) study design, time horizon and outcome measures. The variety in effect measures used in the economic evaluations limits the comparability of the studies and their results. Only twenty-two studies used both clinical effects and QALYs as outcome. The other studies included CEAs using diagnosis-specific measures. The variety in outcome measures amongst these studies was high. A general limitation of the included studies is that the costs of somatic specialist care were often not taken into account, while it can be expected that proper treatment of MUS can lead to a decrease in these costs. This could potentially result in underestimation of the cost-effectiveness of the interventions.
The studies included in this systematic review contained four studies [31][32][33]41] that were also part of an earlier systematic review on the cost-of-illness and economic evaluations of interventions for MUS disorders by Konnopka et al. [18]. In that systematic review, 13 studies were included with patients with MUS, five cost-of-illness studies and eight economic evaluations, of which only two cost-effective analyses. Similar to our review the comparability of included studies was limited due to the heterogeneity concerning design, methods and year of study conduct.
An intervention may be cost-effective only after a longer period of time. One study [48] did show through a modeling approach that cognitive behavioral group intervention could be cost-effective after 21 months. The time horizon of only 12 months used in many studies may be the reason that interventions were not cost-effective. An important advantage of MBEEs is that they allow cost-effectiveness to be modelled over longer periods, although at the cost of more uncertainty. Six out of nine included MBEEs had a time horizon of at least two years.
Quality assessment. Due to the variability of the study methodological quality, drawing conclusions regarding the cost-effectiveness of different types of interventions focusing on different target populations is difficult. Assessing the methodological quality of the studies using the CHEC-list [23], a validated checklist for the methodological quality assessment of economic evaluations, is partly subjective and we have limited ourselves to discuss all items separately. To enhance the reliability we applied the procedure as mentioned in the methods paragraph. Moreover, the separate items of the CHEC-list were valued equally and the overall quality score therefore does not reflect that certain items, such as the chosen time horizon and the chosen perspective, could be perceived as having a relatively large impact on study methodological quality. In the included studies the chosen time horizon was usually appropriate and the chosen perspective depended on the healthcare system of the country of the study.

Strengths and limitations
This review has several strengths. We used a broad search strategy in which psychological, medical and health economic literature databases were searched thoroughly. We included all health economic studies focusing on interventions for MUS with the three most prevalent FSS. Furthermore, the quality of the included studies was appraised with the CHEC [23].
Additionally, we applied the recommended strategy for conducting and reporting systematic reviews [19]. In this review we only presented base case results and not the results of sensitivity analyses. Therefore decision makers should consider context-specific factors (e.g. cost-reimbursement of interventions) when deciding on implementing interventions.
This study also has some limitations. The literature search had language restrictions (only English, German, and Dutch), but we assume that economic evaluations will be published mainly in international journals. The included patient groups vary in terms of reported severity, i.e. number and duration of symptoms, functional disability or quality of life. This might constitute a limitation in terms of how MUS are defined but also whether the patients have similar characteristics at inclusion in terms of, e.g. length of illness before treatment, severity of symptoms or previous treatments. Cost-effectiveness of different interventions is not only supposed to depend on the intervention itself, but also on underlying medical (and demographic) conditions and prognostic factors such as number of symptoms, number of body systems involved and number of times symptoms are presented. [10].
With regard to the interventions, it should be noted that the availability of the pharmacotherapy and non-pharmacological interventions can differ per country.
Another limitation is the methodological quality assessment with the CHEC. The CHEC list [23] is the best available instrument for the quality assessment of economic evaluations, and is recommended by the Cochrane Handbook for Systematic Reviews of Interventions [24] for critical appraisal of the methodological quality of health-economic evidence. The presented descriptive information and quality assessment gives a broad picture of the study designs and study quality, but the information is incomplete and decision makers should look in detail at the studies of interest, especially so for the model-based economic evaluations, for which the CHEC-list is less appropriate as a quality assessment tool.
As all included studies were conducted in Western countries, the generalizability of the results is presumably limited due to differences in healthcare provision of a country.

Recommendations
This systematic review provides an overview of group interventions and individual interventions for patients with MUS. Further research is needed to investigate the willingness to participate in group interventions. To address the disease burden and societal costs associated with MUS, it is important to know the cost-effectiveness of available interventions. While the current review shows that not all interventions are cost-effective, this could also be the result of choices made in the included studies such as a relatively short time horizon and the chosen perspective. Due to the chronic nature of MUS and the societal costs MUS may cause outside of the healthcare domain (e.g. lost productivity), it is recommended to conduct high quality economic evaluations of interventions for patients with MUS, with a long time horizon and a chosen perspective in line with the national or local guidelines and the decision makers' information requirements. The studies would ideally use both perspectives, healthcare and societal perspective, so that the outcomes become more relevant for decision makers in different settings. High quality economic evaluations are necessary in order to draw robust conclusions about the cost-effectiveness of interventions for MUS.

Conclusion
The current review provides an overview of 39 studies of interventions for patients with MUS, FM, IBS and CFS and the methodological quality assessment of these studies. In 13 out of 22 studies the intervention condition dominated the control conditions, meaning that the interventions were (on average) cost-effective in comparison with the control conditions. Considering the limited comparability due to the heterogeneity of the studies, the group interventions might be more cost-effective than individual interventions.
Supporting information S1