Eur J Health Econ DOI 10.1007/s10198-014-0614-4

ORIGINAL PAPER

Chronic pain patients’ treatment preferences: a discrete-choice experiment Axel C. Mu¨hlbacher • Uwe Junker • Christin Juhnke Edgar Stemmler • Thomas Kohlmann • Friedhelm Leverkus • Matthias Nu¨bling

•

Received: 18 November 2013 / Accepted: 30 May 2014  Springer-Verlag Berlin Heidelberg 2014

Abstract Objective The objective of this study was to identify, document, and weight attributes of a pain medication that are relevant from the perspective of patients with chronic pain. Within the sub-population of patients suffering from ‘‘chronic neuropathic pain’’, three groups were analyzed in depth: patients with neuropathic back pain, patients with painful diabetic polyneuropathy, and patients suffering from pain due to post-herpetic neuralgia. The central question was: ‘‘On which features do patients base their assessment of pain medications and which features are most useful in the process of evaluating and selecting possible therapies?’’ Methods A detailed literature review, focus groups with patients, and face-to-face interviews with widely A. C. Mu¨hlbacher (&)
C. Juhnke IGM Institute Health Economics and Healthcare Management, Hochschule Neubrandenburg, Brodaer Straße 2, 17033 Neubrandenburg, Germany e-mail: [email protected] A. C. Mu¨hlbacher
M. Nu¨bling GEB Empirical Consulting mbH, Freiburg, Germany U. Junker Department of Anesthesiology, Intensive Care Medicine, Pain Therapy and Palliative Care, Sana Klinikum Remscheid, (Academic) Teaching Hospital, University Cologne, Remscheid, Germany E. Stemmler Pfizer Pharma GmbH, Berlin, Germany T. Kohlmann Ernst-Moritz-Arndt-University Greifswald, Greifswald, Germany F. Leverkus Pfizer Deutschland GmbH, Berlin, Germany

recognized experts for pain treatment were conducted to identify relevant treatment attributes of a pain medication. A pre-test was conducted to verify the structure of relevant and dominant attributes using factor analyses by evaluating the most frequently mentioned representatives of each factor. The Discrete-Choice Experiment (DCE) used a survey based on self-reported patient data including sociodemographics and specific parameters concerning pain treatment. Furthermore, the neuropathic pain component was determined in all patients based on their scoring in the painDETECT questionnaire. For statistical data analysis of the DCE, a random effect logit model was used and coefficients were presented. Results A total of 1,324 German patients participated in the survey, of whom 44 % suffered from neuropathic back pain (including mixed pain syndrome), 10 % complained about diabetic polyneuropathy, and 4 % reported pain due to post-herpetic neuralgia. A total of 36 single quality aspects of pain treatment, detected in the qualitative survey, were grouped in 7 dimensions by factor analysis. These 7 dimensions were used as attributes for the DCE. The DCE model resulted in the following ranking of relevant attributes for treatment decision: ‘‘no character change’’, ‘‘less nausea and vomiting’’, ‘‘pain reduction’’ (coefficient: [0.9 for all attributes, ‘‘high impact’’), ‘‘rapid effect’’, ‘‘low risk of addiction’’ (coefficient *0.5, ‘‘middle impact’’), ‘‘applicability with comorbidity’’ (coefficient *0.3), and ‘‘improvement of quality of sleep’’ (coefficient *0.25). All attributes were highly significant (p \ 0.001). Conclusions The results were intended to enable early selection of an individualized pain medication. The results of the study showed that DCE is an appropriate means for the identification of patient preferences when being treated with specific pain medications. Due to the fact that pain

123

AC. Mu¨hlbacher et al.

perception is subjective in nature, the identification of patients´ preferences will enable therapists to better develop and implement patient-oriented treatment of chronic pain. It is therefore essential to improve the therapists´ understanding of patient preferences in order to make decisions concerning pain treatment. DCE and direct assessment should become valid instruments to elicit treatment preferences in chronic pain. Keywords Patient preferences
Discrete-Choice Experiment
Neuropathic pain
Chronic pain JEL codes I100 Health: General
I180 Health: Government Policy; Regulation; Public Health
D700 Analysis of Collective Decision-Making: General
D710 Social Choice

Background Chronic pain is a widespread medical and economic burden that appears at all ages and in all populations. It has a substantial impact on the quality of a patient’s daily life as well as their physical and mental function. Besides considering individual susceptibility to specific side effects due to individual health status, comorbidities, or age, the individual patient’s circumstances and, if possible, the patient’s preferences should be taken into account when choosing a specific drug for pain treatment [5, 87]. There is an increasing focus on involving patients in treatment decisions as a form of shared decision making. Hence, it is becoming increasingly important to elicit, understand and take into account the patients’ preferences as helpful treatment characteristics and options when making decisions [25, 26]. To ensure effective pain management, patient adherence is highly essential and thus patients’ preferences should be specifically integrated into the regimen of the pain management [20, 23]. This means that their pain will be systematically recorded and assessed using standardized, reliable, and valid instruments. As patients are highly experienced in concern of their individual health status, integrating their preferences regarding treatments and outcomes is an important part of both quality management and evidence-based practice [7, 8]. A healthcare system based on preferences could significantly improve patient satisfaction, therapy compliance, and therefore treatment outcomes [8, 20]. A better understanding of patients’ needs and expectations for pain relief and continued commitment to patient education could improve the treatment of pain [20].

123

Decision-making context: epidemiology and treatment alternatives of chronic pain Chronic pain In the etiology of chronic pain one can mechanistically distinguish between nociceptive and neuropathic pain. However, these two types of pain do not only differ with respect to their mechanism of origin but also in the medical options for their therapeutic approach [3]. However, many patients experience individual pain syndromes (e.g., in back pain or in tumor-related pain syndromes) that are composed of both nociceptive and neuropathic components. Therefore, the concept of the so-called ‘‘mixed-painsyndrome’’ has been developed for the description of such a mixture, or even transition of nociceptive and neuropathic pain components [46]. In contrast, neuropathic pain develops as a consequence of modified neuronal activity based on a direct consequence of a lesion or disease affecting the somatosensory system. Thus, in neuropathic pain states, the nervous system itself is involved in causing the pain. Often, neuropathic pain arises due to damage or lesion of the central or peripheral nervous system (e.g., by cut, bruised or injured nerve fibers). Nerve injury may cause far-reaching structural and anatomical changes of the nociceptive system, which are partly irreversible and may therefore persist. Typical symptoms are burning persistent pain, paroxysmal attacks of pain, and evoked pain. These symptoms are called positive symptoms. Negative symptoms are defined as sensory deficit [3, 46]. Furthermore, the nervous system can also be damaged as a result of infectious diseases such as shingles (herpes zoster). Varicella-zoster viruses infect dorsal root ganglia of the spinal cord. The consequences are painful rashes on the affected skin. Because of the frequency and the often very pronounced pain symptoms, painful diabetic polyneuropathy (DPN) and postherpetic neuralgia (PHN) are the focus of scientific interest [31]. Epidemiology of neuropathic pain Neuropathic pain arises due to lesion, damage or disease of the somatosensory system. The symptoms are typically characterized by burning persistent pain or sudden lancinating pain attacks. Epidemiological reviews have shown that there are no reliable estimates on the prevalence of neuropathic pain. Accordingly, in study results from the US and the UK, the prevalence of neuropathic pain is estimated to be 1–2 % of the general population [69]. Torrance et al. [92] found a prevalence of 8 % in a general population survey in the UK. Other estimates suggest that

Chronic pain patients’ treatment preferences

2–3 % of the population in industrialized countries are affected [62]. The point prevalence of neuropathic pain syndromes is estimated at 5 % of the general population [3, 58]. A study of the German general population found 6.5 % having neuropathic pain features (NeP) [70]. According to estimates from professional pain societies about 20 million people in Germany suffer from chronic or recurrent pain. Approximately 6–8 million of these people are considerably impaired in their quality of life. Chronic pain is described as pain duration for more than 3–6 months and will lead to a tremendous deterioration and impairment of the patients’ quality of life. Chronic pain is significantly associated with typical comorbidities like sleep disturbances, anxiety, and depression. Neuropathic pain represents an extraordinary type of chronic pain: Many of the debilitating comorbidities are the more characterized the more the neuropathic pain component is present [30]. Moreover, the prevalence of chronic pain increases with increasing patient age. Generally, the elderly (over 65 years) and women are especially affected. In addition, the socio-economic status or income appears to have a negative impact on the incidence of chronic pain [41]. As specified by Baron [3], diabetes or post-herpetic neuralgia accounts for a share of 13 % of neuropathic pain patients. As has been shown in an epidemiological analysis, neuropathic pain is present in 37 % of patients with chronic low back pain [30]. Neuropathic pain occurs in approximately 67 % of all spinal cord injuries, in 28 % of patients with multiple sclerosis, and in 8 % of all stroke patients [35]. Chronic back pain One of the most common reasons for pain complaints in industrialized nations is back pain. Consequently, back pain is also one of the leading reasons for treatment in outpatient and rehabilitative care and causes high numbers of sick leaves and invalidity. In Germany, the point prevalence of back pain is between 27 and 40 %, with an annual prevalence of 70 % and a lifetime prevalence of 80 %. In particular, psychosocial and physical burdens significantly contribute to the development of chronic back pain [81]. It has been demonstrated that socio-demographic factors such as age, female gender, low education, and work-related stress do highly correlate with the prevalence of back pain [68]. Health economic studies have shown that the total costs to society caused by the medical treatment of back pain as well as the costs of disability are comparable to the costs of diabetes, depression, and heart disease. However, a large part of these back pain-associated costs can be attributed to only a very small proportion of chronic back pain symptoms [68].

Painful diabetic (poly) neuropathy In several epidemiological studies the prevalence of peripheral neuropathy in patients with diabetes was estimated to be 26–47 % [4]. An Australian study concludes that about 13 % of diabetes patients suffer from neuropathy. A multicenter study from the UK showed that the proportion of diabetic patients with neuropathy is between 22 and 32 %. An Italian study yielded similar results with 32.3 % of diabetes patients showing a painful diabetic neuropathy [100]. About 16–26 % of diabetes patients suffer from painful diabetic neuropathy [42, 101]. Veves et al. [98] reported a pain prevalence of 40–50 % among diabetics who have developed diabetic neuropathy. Thus, painful diabetic neuropathy is a considerable health problem worldwide. Country-specific studies show the dimension of the problem [39, 85]. Postherpetic neuralgia Herpes zoster may cause significant suffering due to acute and chronic pain, or postherpetic neuralgia [82]. As many as 15–30 % of all people are estimated to suffer from herpes zoster once in their life time [97]. The studies on herpes zoster consistently show that the incidence increases with increasing age. Although young people are not infrequently affected the mean age of patients with herpes zoster is approximately 64 years. Especially in the elderly, aged above 85 years, the incidence rate is nearly 50 % [102]. It is likely, however, that the incidence will continue to rise in the coming decades [24]. In addition, genderspecific shifts can be seen: women in all age groups are at higher risk than men [27, 40]. Therapies for the treatment of chronic neuropathic pain Chronic pain often seriously impacts the working ability of patients. According to a European study on chronic pain, 20 % of patients suffering from chronic pain lost their jobs due to the impact exerted by the pain [33]. Depression is a well-known and strongly associated comorbidity of chronic pain. Nearly 20 % of Europeans suffering from chronic pain have been diagnosed with depression. Neuropathic pain due to postherpetic neuralgia may cause fatigue, insomnia, depression, anxiety, interference with social roles and leisure activity, and impaired basic and instrumental activities of daily living [3, 30]. Hence, it can profoundly diminish the quality of life of affected persons [83]. Breivik et al. [10] showed that approximately 17 % of patients rate their pain so severe that they would rather die than continue to endure such pain.

123

AC. Mu¨hlbacher et al.

Next to the pain-related impairment on quality of life, analgesic medications administered for pain treatment have various side effects, including dizziness, weight gain, attention deficit disorder, memory problems, nausea, and blurred vision [3, 22, 62]. Recommended pharmacotherapies for the treatment of neuropathic pain include antidepressants, anticonvulsants/ antiepileptics, opioid analgesics, topical therapies, and miscellaneous like cannabinoids [3]. These medications may lead to an alleviation of the pain, but complete resolution of symptoms is usually not achieved. The following outcomes are generally accepted as realistic treatment goals in neuropathic pain [22, 41, 62]: • • • • •

Pain reduction by [30–50 % Improvement of quality of sleep Improvement of quality of life Maintenance of social activity and social relationships Maintenance of working capacity

In addition to pharmacological treatment, non-pharmacological therapies are also utilized in the treatment of neuropathic pain. Amongst these alternative therapies are acupuncture, electro-stimulation, herbal remedies, magnetic therapy [73], and transcutaneous electrical nerve stimulation [1, 19].

Methods and study design: how did we elicit preferences? Introduction to preference elicitation using DiscreteChoice Experiments Revealed and stated measurements are the two approaches for the measurement of patient preferences. The revealed method is based on observed decisions being evaluated (e.g., prescription data) [93]. It shows how patients decide, but cannot explain why they do so [13]. In stated analyses, patient preferences are collected using direct questioning: patients have to assess objects with respect to several characteristics, which are presented in different combinations. Stated methods, such as conjoint analyses, aim to measure the influence of therapy characteristics on patient preferences [13, 56, 66]. The Conjoint Analysis is a statistical method to determine how people weigh and assess different characteristics of a product or service. The preference measurement assumes that each product or service can be specified by one or more characteristics (attributes), which in sum represent this product or service [50]. One stated preference technique to quantify patients’ priorities is the Discrete-Choice Experiment (DCE). The DCE is a choice-based method, and a variant of the Conjoint Analysis, which was made possible through the

123

theoretical work of Lancaster [50] and McFadden [59]. The value of goods and services depends on the nature and level of the underlying attributes [48, 49]. A key feature is the specification of virtual good or service, the benefit of which can be analyzed through the statistical computation of the choice data in the experiment [6]. DCEs are an attributebased measure of benefit, because an individual’s valuation depends on the levels of these characteristics. They focus on investigating the trade-offs between crucial attributes [77, 78, 80]. Discrete-Choice models are specifically designed to provide useful and appropriate information about individuals’ willingness to accept trade-offs among attributes of multi-attribute products [75]. It is based on the hedonic principle that products are composed of a set of various attributes and that the attractiveness of a product to an individual is a function of these attributes. People’s relative preferences among product attributes and levels vary, and, thus, they are willing to accept trade-offs among them [37, 43, 90, 91]. Based on subjects’ choices, an endpoint-based utility score of certain alternative could be derived. Preference measurement using DCE: Discrete-Choice methods provide the richest information regarding the trade-offs individuals are willing to make among treatment efficacy, treatment safety, and other factors that affect patient well-being [29, 56, 57]. The application of Conjoint Analysis is rapidly increasing in medicine and health economics [14, 21] and professional societies have developed guidelines for the application of Conjoint Analysis in outcomes research [12, 66]. Discrete-Choice analysis is a method for surveying preferences that is highly recognized in international research and is regarded as well established on the basis of years of research and is already finding application in healthcare [56, 72, 77]. DCEs have recently gained importance in the study of innovative health technologies and non-market goods [51, 52, 79] or where market choices are severely constrained by regulatory and institutional factors [76]. This method offers strategies for eliciting preferences for the evaluation of health and healthcare. It is already used to elicit preferences in primary care [61, 84, 99]. Healthcare interventions, services, or policies can be described by their attributes [36]. The application of DCEs has been extended to measure provider preferences [96], eliciting preferences for hospital on various aspects or insured preferences for health system attributes [89]. Moreover, the technique has been used to evaluate patientcentered outcomes in the provision of care [65, 67]. For policy analysis, it might be interesting to calculate how choice probabilities vary with changes in attributes or attribute levels, or to calculate secondary estimates of money equivalence (willingness to pay or willingness to accept) [47], risk equivalence (maximum acceptable risk)

Chronic pain patients’ treatment preferences

[45], or time equivalence for various changes in attributes or attribute levels [44]. Findings on the reliability and validity of DCEs in healthcare settings are encouraging [16, 17]. Preference elicitation: In a first step, all characteristics relevant for each target group have to be identified. These may be physician expertise, medical-technical equipment, or waiting time for an appointment [64]. The relevant characteristics are then combined to define hypothetical products or treatment options. In the DCE, different therapies are presented pair-wise and the subjects have to choose one of the options [6]. As the number of possible combinations increases exponentially, not all possible combinations can be presented. It is important to cover all the relevant characteristics when selecting the items for the DCE. Therefore, in most cases, a reduced sample of alternatives is used, which, however, allows for a reliable evaluation of preferences [86, 88]. The treatment alternatives are presented to the patient and the patient has to choose one of the presented options. Based on the decision behavior, the relevance of the different characteristics for the decision can be calculated and described by coefficients [11]. Study design Research question: patient preferences in chronic pain? The purpose of the study was to determine treatment preferences of patients with neuropathic pain. More precisely, the study aimed at the identification and weighting of multiple endpoints of a pain medication that are relevant from the perspective of patients with chronic pain. Within the indication ‘‘chronic neuropathic pain’’, three groups were analyzed in depth: neuropathic back pain, diabetic polyneuropathy, and zoster. Moreover, parameters most relevant for patients as well as varying preference structures possibly inherent in different subgroups were to be analyzed. The central question was: ‘‘On which features do patients base their assessment of pain medications and which features are most useful in the process of evaluating and selecting possible therapies?’’. In the present study of chronic neuropathic pain, preferences of affected persons regarding different medical treatments were to be elicited. While paying special attention to the three main groups within this indication, great interest was attached to the aspect of differential preference structures. This investigation was expected to provide information regarding patient-relevant endpoints in effectiveness/efficacy studies on analgesics. A further study objective was to obtain a detailed description of potential barriers to and prerequisites for useful medical treatments. Adopting the perspective of affected patients

required an extended focus that extended pharmacological aids. Patients were concerned not only to find out: ‘‘How effective is the drug?’’ but also ‘‘What will help me?’’ and ‘‘What are the side effects?’’. Decision model: attributes and levels Attribute identification: literature search and expert interviews: A literature review was conducted to ascertain quality aspects of pharmacologic treatments in neuropathic pain/chronic pain from the patients’ perspective. As a result, 151 full texts were included in the further search. To identify relevant treatment attributes of a pain medication, the aspects identified were then evaluated in face-to-face interviews with two highly regarded experts in the field of chronic pain. The hypotheses and findings from these interviews were used as the basis for the following parts of the study. Attribute identification: focus groups: To enable the assessment of unexpected patient-relevant aspects, a qualitative phase was conducted in the form of focus group interviews with affected patients prior to the quantitative research. Based on a previously developed interview guide, focus groups were conducted in 6 German cities between November 2009 and February 2010. Patients were recruited from chronic neuropathic pain self-help groups. A total of 20 patients participated in six focus groups. The objective of this phase was to obtain a complete collection of possibly relevant characteristics of treatments from the patients’ viewpoint. During the discussion with the participants, the criteria used by neuropathic pain patients to evaluate drugs were assessed. The discussion centered on the complaints and problems of those affected as well as on their treatment needs, expectations, and personal goals. The open, thematically grouped collection of criteria was followed by a targeted inquiry into the relevance of previously undisclosed aspects. As far as possible, the treatment characteristics were concretized in the form of a workable formulation and description for the subsequent DCE. In the following discussion phase, the treatment characteristics were weighted again by the group to create a first and approximate ranking. Quantitative pilot study: factor analysis: Based on the results of the qualitative study phase, a questionnaire was developed for the quantitative pilot study. The primary objective of this pilot study was to extract central dimensions in the assessment of medications from the entire set of therapy characteristics relevant from the patients’ perspective (a total of 36 individual aspects) by means of factor analysis (principal component analysis, varimax rotation). The pilot study was conducted in summer of 2011, both as a classic pencil and paper version and as an online survey.

123

Pain reduction by 1 point on a 0–10 scale

Longer interval of nausea and vomiting (approx. 4 h per day)

5 out of 7 nights per week relaxing sleep

Risk of character change of 30 % (30/100 people)

Pain reduction by 3 points on a 0–10 scale

Risk of addiction of 10 % (10/100 people)

Incompatibility with comorbidity in 10 % of all cases (10/100 people)

Short interval of nausea and vomiting (approx. 2 h per day)

A couple of hours relaxing sleep every night

Risk of character change of 10 % (10/100 people)

Pain medications may vary in terms of pain reduction. This describes the amount of pain reduction that the medication causes on a 10-point pain scale

‘‘Risk of addiction’’ describes the risk of occurrence of addiction due to the pain medication

Pain medications may be tolerable and/or compatible with other drugs or comorbidities to varying degrees. This attribute describes the degree to which the medication is feasible with other existing concurrent diseases (comorbidities)

Pain medications can cause gastrointestinal problems. These include in particular nausea and vomiting. Nausea and vomiting may occur over varying time periods

Pain medications can affect the length of deep sleep and sleep quality

‘‘Character change’’ describes the risk of occurrence of character and personality changes (concentration disorders, disorders in social interaction etc.) due to the pain medication

Pain reduction

Less risk of addiction

Applicability with comorbidity

Less nausea and vomiting

Improvement of quality of sleep

No character change

123

Incompatibility with comorbidity in 30 % of all cases (30/100 people)

Slow onset of action: 3-4 days Rapid onset of action: 1–2 days Pain medications have different onsets. ‘‘Rapid effect’’ describes the time until the full onset of action of the drug is achieved Rapid effect

Level 2 Level 1 Description Attribute

Table 1 Attributes and levels in the Discrete-Choice Experiment (7 dichotomous attributes, fold-over design)

Risk of addiction of 30 % (30/100 people)

AC. Mu¨hlbacher et al.

The questionnaire for the pilot study was divided into three sections: socio-demographic parameters (including age, sex, education, health status, diagnostic group, and year of first diagnosis), information about pain, and importance ratings of characteristics of therapies for neuropathic pain. The latter included a direct assessment of relevance of a total of 36 items using a 5-point Likert scale. For the analysis, all 36 items of the relevance rating were transformed to a scale of 0 (not important) to 100 (very important). In addition, two free-text questions on content validity and comprehensibility (practicality) were included in the questionnaire. Attributes and levels used for preference elicitation: As a result of the detailed review of literature, interviews, and factor analyses, seven characteristics were identified and assessed as most significant by the patients. These seven final criteria were then used for the quantitative DCE survey. The attributes and characteristics used in the final study are shown in Table 1. Data collection plan: stratification, and recruitment Sample-size calculations represent a challenge in DCEs [11]. Minimum sample size depended on a number of criteria, including the question format, the complexity of the choice task, the desired precision of the results, and the need to conduct subgroup analyses [55, 71]. As the purpose of this study was to determine treatment preferences of patients with neuropathic pain, patients from all three groups of interest were included in the survey. The main survey was performed in November and December 2011. The questionnaire was available in a paper and pencil version as well as in an online survey at www. schmerz-befragung.de, in both cases completely anonymous. The recruitment was carried out by cooperating physicians and treatment centers as well as a large mailing to the members of the German Pain League (Deutsche Schmerzliga e.V.). Data collection: instrument, elicitation technique, tasks and experimental design For the identification of neuropathic pain, the widely tested and validated painDETECT questionnaire was used for both the pilot study in terms of socio-demographics and the indication-specific parameters, as well as for the quantitative main study to identify neuropathic pain. The painDETECT questionnaire has been developed and validated by the German Research Network for Neuropathic Pain [30]. The questionnaire represents a reliable screening tool for the identification of neuropathic pain and could easily be completed by patients within a few minutes

Chronic pain patients’ treatment preferences

and does not require clinical examination. Three questions related to current pain intensity and the intensity in the last 4 weeks. Besides questions on course and profiles of pain, the painDETECT questionnaire also contained seven basic questions on specific neuropathic pain characteristics (sensory descriptor) which were to be rated by the patients according to intensity on a 0–5 scale (never to very severe). Based on ratings, four patterns allowed identification of the pain course. Finally, two items relating to the spatial (radiating) and temporal characteristics of the individual pain pattern were included, allowing for topological consideration of the neuropathic pain genesis. With a positive predictive accuracy of 83 % as well as a sensitivity of 85 % and a specificity of 80 %, this questionnaire is widely accepted as a reliable screening tool for the identification of neuropathic pain [7, 30, 32]. Discrete-Choice Experiment There are several design properties that needed be considered in the development of the experimental design [38, 55]. DCE scenarios were developed using a fractional factorial design [11]. To appreciate the importance of possible statistical correlation between main effects and interactions, the number of combinations was reduced to a more manageable size without losing essential information through an orthogonal design (if certain assumptions about interaction effects are made). The choice sets were generated automatically using the software SPEED [9]. The design demonstrated uncorrelated main effects and an efficiency of 100 % [18]. The maximum dissimilarity between therapy alternatives was achieved by creating the associated sets using the foldover method. Dichotomous levels were used as a basis for each attribute. Each respondent had to make eight choices between binary choice sets (Fig. 1). A pre-test (n = 24) was conducted to verify the structure of relevant and dominant attributes as well as the comprehensibility and practicality. Data analysis The data analysis used both descriptive methods (frequency counts, statistical parameters of distributions) and bivariate methods (crosstabs, comparison of means: ANOVA/analysis of variance) and complex multivariate techniques (principal component analysis with varimax rotation). SPSS (IBM, Armonk, NY, USA) and STATA (StataCorp, College Station, TX, USA) were used for data analysis. For all analyses p \ 0.05 (two-sided) was judged to be statistically significant.

The statistical data analysis of the DCE used random effect logit models based on dummy coding [8] of the levels as well as probit models.

Results: how do patients prioritize? Results of quantitative pilot study: factor analysis The pilot study was conducted in June 2011 with 104 patients. Almost all of the 36 items reached an average relevance of C75 points. Figure 2 shows the mean values of the 18 items that were rated highest, sorted by relevance. Items with a mean value of \75 points (n = 7) were considered less relevant and therefore not included in the factor analysis to determine the assessment dimensions for the DCE. The remaining 29 items were subjected to an exploratory factor analysis. This resulted in a 4-factor solution according to the eigenvalue criterion, but also in a well-defined 6-factor solution in terms of content interpretation. With the goal of integrating all quality criteria from the patients’ perspective into the DCE, one aspect (item) from each resulting factor was taken into account. This was to ensure the representation of the entire content spectrum of the treatment preferences. Those items were selected as attributes that had a high loading on the factor, that affected most patients, and that could be operationalized in a positive and negative pole (Table 1). The aspect of ‘‘applicability with comorbidities’’ was not clearly classifiable to one factor but since it was designated as important in the focus groups and the direct assessment, it was added as 7th attribute. For each of the seven factors, two poles were given [28], based on the characteristics of the pain medication. Results of Discrete-Choice Experiment Sociodemographics A total of 1,324 German patients participated in the selfadministered DCE survey (Table 2). The vast majority of patients with chronic pain were older than 45 years. Only about 15 % of the study sample was younger than 45 years, while 62 % of the respondents were female. The respondents gave a critical self-assessment of their health status: more than 60 % rated their own health as fair or poor, only 10 % as very good or excellent. The mean year of diagnosis was 1998 with a standard deviation (SD) of 10.3. Division into diagnostic groups showed that the group of patients with neuropathic back pain (including mixed pain syndrome) was the largest with 44 %. Ten percent of respondents were suffering from diabetic polyneuropathy and 4 % from herpes zoster/

123

AC. Mu¨hlbacher et al. Fig. 1 Example choice set: Discrete-Choice Experiment

Fig. 2 Factor analysis, mean, 18 highest ranked items

shingles. A fourth group comprised neuropathic pain patients (not back pain) (Table 3). Due to the high number of cases in the study, the preference models were differentiable for these subgroups despite the uneven distribution.

123

Information on pain (painDETECT) The widely tested and extensively validated questionnaire painDETECT [7, 30, 32] was used to determine the neuropathic pain component. However, n = 156 respondents

Chronic pain patients’ treatment preferences Table 2 Sample characteristics Sample characteristic

n = 1,324 %

neuropathic pain component ([90 %). The remaining patients (n = 320) had a score of 13–18. Discrete-Choice Experiment

Gender Female

62.0

Male

38.0

Age 18–24 years

1.4

25–34 years

2.7

35–44 years 45–54 years

10.7 25.7

55–64 years

28.4

65 and older

31.2

School leaving qualification None

1.1

Basic leaving certificate

34.7

Intermediate secondary school leaving certificate

41.4

Higher education entrance qualification

22.9

Self-rated health Excellent

1.6

Very good

8.4

Good

28.9

Fair

41.0

Poor

20.1

Table 3 Patient characteristics–diagnostic groups Characteristic

Total n = 1,324

Proportion (in %)

Neuropathic back pain (including mixed pain syndrome)

584

44.1

Diabetic polyneuropathy

127

9.6

58

4.4

Diagnostic group

Herpes zoster/shingles Neuropathic pain (not back pain)

74

5.6

Other

316

23.9

Not sure

123

9.3

42

3.2

Missing/not specified

did not answer this section of the questionnaire. Hence, n = 1,168 patients were included in the painDETECT analysis. As Fig. 3 shows, this questionnaire revealed a substantially normal distribution with a center at about 18–19 points. The mean score was 17.67 with an SD of 7.58 (min: 0, max: 38). The grouping of the painDETECT numbers according to the cutoffs of the tests resulted in 25.3 % of patients (n = 295) having a score \13. Most patients (n = 553) achieved a score [19. This means that 47.3 % of all included patients were very likely affected by a

Overall, a total of n = 9,732 valid observations were obtained from n = 1,324 patients. For the analysis the ‘‘better’’ level (pole) was encoded with the positive value ‘‘1’’, the negative pole was coded with the value ‘‘0’’ (‘‘dummy coding’’). After examining the Hausman test, a final model of the main effects was created using a ‘random effects logit model’. This model was seen to be more efficient than the fixed effect models. In addition, probit models were also tested; all calculated models showed very similar results in terms of content interpretation. All seven attributes of the model resulted in highly significant (different from zero) estimates for each of the positive poles. Hence, all coefficients showed the expected direction. The positive pole was always chosen significantly more often than the negative one. Thus, all seven attributes contributed significantly to the treatment decision of the patients. Nonetheless, as the results in Table 4 show, the choice of a treatment by the patient was influenced by the seven characteristics to varying degrees. The following ranking of the seven attributes for treatment decision resulted from the DCE: ‘‘No character change’’ (coeff.: 0.990) was the most important attribute to the patients. ‘‘Less nausea and vomiting’’ (coeff.: 0.973) and ‘‘pain reduction’’ (coef.: 0.934) are of high importance followed by ‘‘rapid effect’’ (coeff.: 0.519) and ‘‘less risk of addiction’’ (coeff.: 0.518). ‘‘Applicability with comorbidity’’ had a coefficient of 0.380. The least important attribute was ‘‘improvement of quality of sleep’’ (coeff.: 0.239). All attributes were highly significant (p \ 0.001) (Fig. 4). Significance tests were performed to determine which attributes differed significantly from each other. The importance of the three features ‘‘No changes in character’’, ‘‘Less nausea/vomiting’’ and ‘‘Pain reduction’’ do not differ from each other. However, all three coefficients are significantly higher than those of the other four features (thus: rank 1–3). The two features ‘‘Rapid effect’’ and ‘‘Low risk for addiction’’ are equally relevant. Nevertheless, they are more important in the classification than attributes 4 and 6, and less important than attributes 7, 5 and 2 (resulting in ranks 4–5). The attribute ‘‘Applicability with comorbidities’’ was ranked 6th. It is significantly more important than the attribute ‘‘Quality of sleep’’ (ranked 7th), and both are significantly less important than the other five attributes. When interpreting these results, it must be remembered that the presented comparison (coefficients) reflects the

123

AC. Mu¨hlbacher et al.

Fig. 3 Results of painDETECT questionnaire

Table 4 Coefficients of main effects logit calculation Attribute: level

Coeff.

p value

95 % CI

Rapid effect

0.519 \0.001

0.43; 0.61

Pain reduction

0.934 \0.001

0.84; 1.02

Less risk of addiction

0.518 \0.001

0.43; 0.61

Applicability with comorbidity

0.380 \0.001

0.29; 0.47

Less nausea and vomiting

0.973 \0.001

0.88; 1.06

Improvement of quality of sleep

0.239 \0.001

0.15; 0.33

No character change

0.990 \0.001

0.90; 1.08

model-constant

-2.094 \0.001

-2.27; -1.91

difference between the levels of an attribute. In respect of dichotomous attributes, this describes the difference between the positive and the negative pole. The correct Fig. 4 Chronic pain main effects logit model (positive poles)

123

framing in the interpretation would be that ‘‘a reduction of pain by two points on the pain scale’’ (difference between positive and negative pole) is much more important to the patient than the ‘‘improvement of quality of sleep for 2 days’’. In additional analyzes (interaction models), the modifying effects of socio-demographic factors or parameters like the individual information on pain on the general preference structure were tested. In total, 20 parameters (e.g., age, sex, diagnostic group, current medication) were analyzed resulting in 140 different coefficients (20 parameters 9 7 attributes) with a potential influence on the preferences. Since the artifact risk increases with multiple testing, the significance level was hence reduced to p \ 0.001 for the analyses.

Chronic pain patients’ treatment preferences

Fig. 5 Coefficients in DCE: three different diagnostic groups

A total of 28 of the tested interactions were significant in the bivariate tests (main effects model plus seven interactions of one parameter from the table). A reduced multivariate model was then created, which included only the multivariate significant subgroup effects in addition to the main effects (most parsimonious model). Within the subgroup analyses a special focus had been placed on three parameters: – – –

PainDETECT score (3 groups) Current adjustment of medication (2 groups) Diagnostic group (3 groups)

However, none of these parameters had a significant impact on one of the seven attributes in the multivariate model. Within the painDETECT score there were hardly any noteworthy differences in the preference structure between the three subgroups (painDETECT negative, positive, marginal). This justifies the inclusion of all respondents in the overall analysis regardless of the painDETECT score. The same applied to the analysis of medication influences. The two groups indicated that they were either currently well-medicated or that their medication was not well attuned. Again, there are only marginal differences that do not significantly influence the preference structure. The most visible differences were apparent in the analysis of diagnostic groups (Fig. 5). The largest differences were seen in terms of attributes ‘‘Less nausea and

vomiting’’ and ‘‘No character change’’. Both attributes were weighted significantly lower by patients suffering from ‘‘diabetic polyneuropathy’’ (Group 2) than by those with neuropathic back pain (Group 1). All other differences between the three diagnostic groups were not significant. However, it became obvious that herpes zoster patients place the attributes in a different order than the other patients. For them, the attribute ‘‘Rapid effect of medication’’ is of great importance, while they place less importance on the attribute ‘‘Less nausea and vomiting’’ (Table 5). However, the group was too small to produce a significant result. Pain equivalents: pain reduction as ‘‘currency’’ Using ‘‘pain reduction’’ as ‘‘currency’’ (price is the parameter conventionally used in marketing), it is possible to evaluate the marginal rate of substitution in relation to the other attributes. For example, the patients would accept an 18 % higher risk of addiction if the product were to reduce their pain by 1 point on the pain scale (Table 6). In principle, this would be possible for all attributes in relation to each other. On this basis it would be possible to calculate the ‘‘value’’ of every hypothetical product in comparison to another. This means that the probability with which a patient chooses one or the other product can be predicted. The only prerequisite is that the levels of all seven attributes of the two alternatives are known.

123

AC. Mu¨hlbacher et al. Table 5 Results of subgroup analyses (separated according to diagnostic groups)

Attribute

Neuropathic back pain

Diabetic polyneuropathy

Herpes zoster/ shingles

Rank

Odds ratio

Rank

Odds ratio

Rank

Odds ratio

1. Rapid effect

1,665

5

1,699

4

2,203

2. Pain reduction

2,460

3

2,096

1

1,859

3

3. Less risk of addiction 4. Applicability with comorbidity

1,804 1,462

4 6

1,350 1,405

6 5

1,433 1,584

7 4

5. Less nausea and vomiting

2,691

1

1,840

2

1,553

5

2

6. Improvement of quality of sleep

1,259

7

1,323

7

1,522

6

7. No character change

2,535

2

1,822

3

2,363

1

Table 6 Marginal rate of substitution against pain reduction Attribute

Meaning of 1 level difference

1 point pain scale (Att. 2, ‘ level)

Equivalent for 1 point on pain scale

Means: in exchange to 1 point on pain scale

Rapid effect

2 days to onset of action

1 day

1.8

1.8 days slower onset of action

Pain reduction

2 points of pain scale

1 point

1

Not applicable

Less risk of addiction Applicability with comorbidity

20 % risk of addiction 20 % risk of problems

10 % 10 %

1.8 2.45

18 % higher risk of addiction 24.5 % higher risk of comorbidity problems

Less nausea and vomiting

2 h nausea and vomiting per day

1h

0.96

0.96 h more nausea and vomiting per day

Improvement of quality of sleep No character change

2 nights of relaxing sleep per week 20 % risk of character change

1 night

3.91

10 %

0.94

3.91 nights with less relaxing sleep per week 9.4 % higher risk of character change

Discussion In the assessment of medical services (e.g., medications, treatment modalities), the perspective of patients, who are affected by all measures and procedures, is often not adequately taken into consideration [16]. However, considering the present preoccupation with concepts such as health competence, patient empowerment, patient participation [54], and ‘shared decision-making [25, 26], it is important to analyze and recognize patients’ preferences with regard to therapies and medications. This study assessed the preferences of patients in the pharmacological treatment of chronic pain. Moreover, 44.1 % of the patients in this study had a neuropathic pain component, including mixed pain syndrome. Based on a qualitative study to ensure the content validity and a pilot study to identify the key quality dimensions, a quantitative measurement of preferences, using DCE, was conducted. Within the quantitative pilot study to determine the dimensions, the patients were initially presented with a total of 36 quality aspects for the evaluation, which had been found as particularly relevant in the literature and the qualitative study phase. As

123

expected, this resulted in classifications in the upper scale range for almost all items. From these 36 single items seven dimensions were determined using factor analysis. These were then used to create a DCE with seven dichotomous attributes. Thus, the part-worth utilities of the various aspects of the decision model could be determined. Since all the attributes and levels were assigned with quantitative characteristics, exchange values of the individual features can be calculated (e.g., what is the equivalent of ‘‘1 point reduction on pain scale’’ in ‘‘hours of nausea and vomiting’’). With this knowledge, exchange rates of each fictitious product can be evaluated/estimated in terms of patient preferences—the only prerequisite is that the characteristics of the seven parameters of the DCE are known for this product. Within the experiment, three aspects had the significantly highest relevance for the choice of the patients: ‘‘No change of character by medication’’, ‘‘significant reduction of pain scale’’ and ‘‘less nausea/vomiting as a side effect’’. This was followed by the aspects of ‘‘rapid effect’’ and ‘‘low risk of addiction’’. The aspects ‘‘applicability with comorbidities’’ and ‘‘improvement of quality of sleep’’ were weighted least.

Chronic pain patients’ treatment preferences

Existing findings in the literature support the results of this study and the importance of patient-centered treatment strategies. The study groups of Argoff et al. and Langley et al. showed a great loss of quality of life in patients with chronic pain [2, 53]. Argoff et al. [2] in particular analyzed the association of neuropathic pain with sleep and psychiatric disorders. The investigators found a complex interrelationship between the index neuropathic pain state and certain comorbid conditions such as sleep disturbances, depression, and anxiety: comorbid conditions exacerbate pain, and, in turn, pain exacerbates the comorbid conditions. As comorbidities can negatively impact response to pain treatment, healthcare providers should assess comorbidities as part of the diagnostic work-up. The authors recommend treating the whole patient, not just the pain. Gore et al. [34] demonstrated for the first time a high burden of illness in patients with painful diabetic peripheral neuropathy and also found that 50 % of patients with chronic neuropathic pain were prescribed with nonsteroidal anti-inflammatory drugs or coxibs, a medication class with no proven efficacy in painful DPN. Almost 70 % of these patients were additionally taking over-the-counter medicines. Fewer patients reported taking medications specifically recommended for neuropathic pain, and fewer still received medications for their sleep disorder. The results of Gore et al. suggested suboptimal pain management and low levels of treatment satisfaction. Meyer-Rosenberg as well as Poliakov and Toth also analyzed the quality of life of patients with chronic neuropathic pain [60, 74]. These authors found sleep disturbance and lack of energy to be the most frequently occurring attributes (65 and 57 %, respectively). However, they did not evaluate which of the found attributes represented the greatest burden for the patients. For example, sleep disturbance also occurred in this study, but was not among the attributes which imposed the greatest burden on the patients. Back pain patients in the study of Freynhagen et al. [30] suffered from neuropathic pain (37 %), non-neuropathic pain (35.3 %), or back pain of undetermined etiology (27.7 %). In comparison to the other two groups, the neuropathic pain patients exhibited a considerable incidence of severe depression and anxiety and were receiving psychotherapy. Medications used to treat depression are known to change the character and may over time exert some influence on cognition and/or personality. Since in our study ‘‘No change of character’’ came out to be the most important attribute for a treatment decision, these findings suggest that patients with back pain with a neuropathic component require individualized, patientcentered therapy and more focused therapeutic care.

Limitations The presented methods document and analyze the priorities and assessments of patients with chronic pain, based on their current situation, perception, and experience. The DCE is able to transform information from the subjective viewpoint in a statistical and objectively evaluable form for decision makers. Nevertheless, the present study reveals some limitations that must be considered when interpreting the results.

Stability of preferences The study demonstrated that the stability of preferences depends on the perspective and the experience of the participants [75]. Further research is needed to determine the impact on the temporal, personal, and social distance within the behavioral level in terms of preference elicitations [94, 95].

Reality orientation of the decisions It should be mentioned that the study participants made their decisions alone and not together with their relatives or physicians, although a different situation could exist in healthcare reality, depending on the attitudes of the patient and their healthcare provider. Different stakeholders may have different preferences. As part of an evaluation process, it should be possible to appropriately take account of every perspective, i.e. the perspective of decision-makers, of the society, of the patients, the insured, or the experts. Information on each of the individual viewpoints and priorities is therefore necessary. Various studies have shown that patient preferences and expert judgments can differ [15, 63], and that the results should always be interpreted in the light of actual circumstances. However, this was not the primary concern of the present study.

Treatment experience The patients surveyed in this study had a mean year of first diagnosis of 1998, with earliest diagnosis in 1951 and latest diagnosis in 2010. Given that patients in the sample might be experienced to different degrees, preferences of patients who have received one or more treatments and preferences of those who are newly diagnosed might vary as preferences might change over time. However, the results present a snapshot over a stratified patient population. It needs to be mentioned that an individual patient has dynamic

123

AC. Mu¨hlbacher et al.

preferences which will probably change in the future. This study presents a stratified patient population. Based on this sample treatment, preferences of patients with neuropathic pain were elicited.

Conclusion The patients‘ perspective is becoming more important in healthcare policy decisions worldwide. Because of the lack of information on patient needs in the decision-makers’ assessment of health services, at present individuals’ preferences often play a subordinate role. The patients’ perspectives and expectations in healthcare decisions are often not sufficiently considered. This study has shown that the top three attributes in the patients’ decision are ‘‘No change in character’’, ‘‘Pain reduction’’ and ‘‘No/less nausea and vomiting’’. Due to the fact that pain perception is subjective by nature, the management of chronic pain needs to be patient-oriented. Moreover, the treatment of chronic pain should be geared to the needs and preferences of the patients. An understanding of patient preferences is therefore essential for the inclusion of patients’ needs in treatment decisions. Our study has demonstrated that the combined use of a direct assessment for rating different levels of importance on the one hand and collection of data by a DCE on the other are reliable, productive, and manageable instruments to elicit treatment preferences in chronic pain treatment. These two procedures yield comparable results, with the direct measurement being able to cover a great number of potentially relevant aspects and the DCE allowing for the most important aspects to be weighed against each other. Acknowledgements The authors extend their thanks to the patient advocacy group ‘‘Deutsche Schmerzliga e.V.’’ for their support in conducting this study. The authors benefited from the valuable assistance of Susanne Bethge (IGM Institute Health Economics and Healthcare Management, Hochschule Neubrandenburg). This research was financed with support from Pfizer Deutschland GmbH, Berlin, Germany. Conflict of interest The authors have no conflicts of interest that are directly relevant to the content of this article.

References 1. Alvaro, M., Kumar, D., Julka, I.S.: Transcutaneous electrostimulation: emerging treatment for diabetic neuropathic pain. Diabetes Technol. Ther. 1, 77–80 (1999) 2. Argoff, C.E.: The coexistence of neuropathic pain, sleep, and psychiatric disorders: a novel treatment approach. Clin. J. Pain 23, 15–22 (2007)

123

3. Baron, R.: Diagnostik und Therapie neuropathischer Schmerzen. Dtsch. Arztebl. 103, 2720–2730 (2006) 4. Barrett, A.M., Lucero, M.A., Le, T., et al.: Epidemiology, public health burden, and treatment of diabetic peripheral neuropathic pain: a review. Pain Med. 8, S50–S62 (2007) 5. Bas, H.: Pharmakotherapie bei neuropathischen Schmerzen durch Nichtspezialisten—Neue NICE-Leitlinien. ARS MEDICI 13, 540–542 (2010) 6. Ben-Akiva, M., Lerman, S.R.: Discrete Choice Analysis: Theory and Application to Travel Demand. MIT Press, Cambridge (1985) 7. Bennett, M.I., Attal, N., Backonja, M.M., et al.: Using screening tools to identify neuropathic pain. Pain 127, 199–203 (2007) 8. Bowling, A., Ebrahim, S.: Measuring patients’ preferences for treatment and perceptions of risk. Qual. Health Care 10, i2–i8 (2001) 9. Bradley, M.: User’s manual for the speed version 2.1 stated preference version 2.1 stated preference experiment editor and designer. Hague Consulting Group, The Hague (1991) 10. Breivik, H., Collett, B., Ventafridda, V., et al.: Survey of chronic pain in Europe: prevalence, impact on daily life, and treatment. Eur. J. Pain 10, 287 (2006) 11. Bridges, J., Hauber, A., Marshall, D. et al.: Checklist for conjoint analysis applications in health: report of the ISPOR conjoint analysis good Research practices Taskforce, in press (2009) 12. Bridges, J., Hauber, B., Marshall, D., et al.: Conjoint analysis use in health studies—a checklist: a report of the ISPOR conjoint analysis in health good research practices task force. In: ISPOR TF Report. Baltimore (2011) 13. Bridges, J., Onukwugha, E., Johnson, F., et al.: Patient preference methods—a patient centered evaluation paradigm. ISPOR Connect. 13, 4–7 (2007) 14. Bridges, J.F., Kinter, E.T., Kidane, L., et al.: Things are looking up since we started listening to patients: trends in the application of conjoint analysis in health 1982–2007. Patient 14, 273–282 (2008) 15. Bridges, J.F., Slawik, L., Schmeding, A., et al.: A test of concordance between patient and psychiatrist valuations of multiple treatment goals for schizophrenia. Health Expect. 16, 164–176 (2011) 16. Bryan, S., Gold, L., Sheldon, R., et al.: Preference measurement using conjoint methods: an empirical investigation of reliability. Health Econ. 9, 385–395 (2000) 17. Bryan, S., Parry, D.: Structural reliability of conjoint measurement in health care: an empirical investigation. Appl. Econ. 34, 561–568 (2002) 18. Burgess, L.: Discrete choice experiments (computer software). In: Department of Mathematical Sciences, University of Technology, Sydney (2007) 19. Carroll, D., Moore, R., Mcquay, H., et al.: Transcutaneous electrical nerve stimulation (TENS) for chronic pain. Cochrane Libr. (2002) 20. Crooks, L.K.: Assessing pain and the Joint Commission pain standards. Adv. Emerg. Nurs. J. 24, 1–9 (2002) 21. De Bekker-Grob, E.W., Ryan, M., Gerard, K.: Discrete choice experiments in health economics: a review of the literature. Health Econ. 21, 145–172 (2012) 22. Diener, H.C., Putzki, N., Berlit, P.: Leitlinien fu¨r Diagnostik und Therapie in der Neurologie Thieme Stuttgart (2008) 23. Ducharme, J.: Acute pain and pain control: state of the art. Ann. Emerg. Med. 35, 592–603 (2000) 24. Dworkin, R.H., Johnson, R.W., Breuer, J., et al.: Recommendations for the management of herpes zoster. Clin. Infect. Dis. 44, S1–S26 (2007) 25. Edwards, A., Elwyn, G.: Inside the black box of shared decision making: distinguishing between the process of involvement and who makes the decision. Health Expect. 9, 307–320 (2006)

Chronic pain patients’ treatment preferences 26. Edwards, A., Elwyn, G., Wood, F., et al.: Shared decision making and risk communication in practice: a qualitative study of GPs’ experiences. Br. J. Gen. Pract. 55, 6 (2005) 27. Fleming, D., Cross, K., Cobb, W., et al.: Gender difference in the incidence of shingles. Epidemiol. Infect. 132, 1–5 (2004) 28. Flynn, T.N., Louviere, J.J., Peters, T.J., et al.: Best–worst scaling: what it can do for health care research and how to do it. J. Health Econ. 26, 171–189 (2007) 29. Flynn, T.N., Louviere, J.J., Peters, T.J., et al.: Estimating preferences for a dermatology consultation using Best-Worst Scaling: comparison of various methods of analysis. BMC Med. Res. Methodol. 8, 76 (2008) 30. Freynhagen, R., Baron, R., Gockel, U., et al.: Pain DETECT: a new screening questionnaire to identify neuropathic components in patients with back pain. Current Med. Res. Opin. 22, 1911–1920 (2006) 31. Freynhagen, R., Busche, P., Konrad, C., et al.: Wirksamkeit und Wirkungsbeginn von Pregabalin bei Patienten mit neuropathischen Schmerzen. Der Schmerz 20, 285–292 (2006) 32. Freynhagen, R., To¨lle, T., Gockel, U., et al.: PainDETECT-ein Palmtop-basiertes Verfahren fu¨r Versorgungsforschung, Qualita¨tsmanagement und Screening bei chronischen Schmerzen. Aktuelle Neurologie 32, P641 (2005) 33. Fricker, J.: Pain in Europe. A report. The Pain Society, Cambridge (2003) 34. Gore, M., Brandenburg, N.A., Hoffman, D.L., et al.: Burden of illness in painful diabetic peripheral neuropathy: the patients’ perspectives. J. Pain 7, 892 (2006) 35. Haanpa¨a¨, M., Treede, F.: Diagnosis and classification of neuropathic pain. IASP Clin. Updates 18, 1–6 (2010) 36. Hauber, A.B.: Healthy-years equivalent: wounded but not yet dead. Expert Rev. Pharmacoecon. Outcomes Res. 9, 265–270 (2009) 37. Hensher, D., Rose, J., Greene, W.: Applied choice Analysis: a Primer. Cambridge University Press, Cambridge (2005) 38. Huber, J., Zwerina, K.: The importance of utility balance in efficient choice designs. J. Mark. Res. 33, 307–317 (1996) 39. Icks, A., Rathmann, W., Rosenbauer, J., et al.: Gesundheitsberichterstattung des Bundes, Heft 24: Diabetes mellitus. Robert Koch-Institut (2007) 40. Insinga, R.P., Itzler, R.F., Pellissier, J.M., et al.: The incidence of herpes zoster in a United States administrative database. J. Gen. Intern. Med. 20, 748–753 (2005) 41. Iskedjian, M., Einarson, T., Walker, J. H., et al.: Anticonvulsants, serotonin-norepinephrine reuptake inhibitors, and tricyclic antidepressants in management of neuropathic pain: a metaanalysis and economic evaluation (Technology report). In: Canadian Agency for Drugs and Technologies in Health, Ottawa (2009) 42. Jensen, T.S., Backonja, M.-M., Jime´nez, S.H., et al.: New perspectives on the management of diabetic peripheral neuropathic pain. Diabetes Vas. Dis. Res. 3, 108–119 (2006) 43. Johnson, F.R., Banzhaf, M.R., Desvousges, W.H.: Willingness to pay for improved respiratory and cardiovascular health: a multiple-format, stated-preference approach. Health Econ. 9, 295–317 (2000) ¨ zdemir, S.: Using conjoint ana44. Johnson, F.R., Hauber, A.B., O lysis to estimate healthy-year equivalents for acute conditions: an application to Vasomotor symptoms. Value Health 12, 146–152 (2009) 45. Johnson, F.R., Ozdemir, S., Mansfield, C., et al.: Crohn’s disease patients’ risk-benefit preferences: serious adverse event risks versus treatment efficacy. Gastroenterology 133, 769–779 (2007) 46. Junker, U., Baron, R., Freynhagen, R.: Das ‘‘mixed pain concept‘‘ als neue Rationale. Das Zusammenspiel von nozizeptiven

47.

48. 49. 50. 51.

52.

53.

54.

55.

56.

57.

58.

59.

60.

61.

62.

63.

64. 65.

66.

und neuropathischen Schmerzen erfordert die neuen Wege der Analgesie. Dtsch Arztebl 101, A1393–A1394 (2004) Kleinman, L., Mcintosh, E., Ryan, M., et al.: Willingness to pay for complete symptom relief of gastroesophageal reflux disease. Arch. Intern. Med. 162, 1361–1366 (2002) Lancaster, K.: Consumer Demand: A New Approach. Columbia University Press, New York (1971) Lancaster, K.J.: A New Approach to Consumer Theory. BobbsMerrill, Indianapolis (1966) Lancaster, K.J.: A new approach to consumer theory. J. Polit. Econ. 74, 132–157 (1966) Lancsar, E., Louviere, J.: Conducting discrete choice experiments to inform healthcare decision making: a user’s guide. PharmacoEconomics 26, 661–678 (2008) Lancsar, E., Louviere, J., Flynn, T.: Several methods to investigate relative attribute impact in stated preference experiments. Soc. Sci. Med. 64, 1738–1753 (2007) Langley, P.C.: The societal burden of pain in Germany: healthrelated quality-of-life, health status and direct medical costs. J. Med. Econ. 15, 1201–1215 (2012) Loh, A., Simon, D., Bieber, C., et al.: Patient and citizen participation in German health care-current state and future perspectives. Zeitschrift fu¨r a¨rztliche Fortbildung und Qualita¨t im Gesundheitswesen-German J. Qual. Health Care 101, 229–235 (2007) Louviere, J.J., Hensher, D.A., Swait, J.D.: Stated Choice Methods: Analysis and Application. Cambridge University Press, Cambridge (2000) Louviere, J.J., Hensher, D.A., Swait, J.D.: Stated choice methods: analysis and applications. Cambridge University Press, Cambridge (2000) Louviere, J.J., Lancsar, E.: Choice experiments in health: the good, the bad, the ugly and toward a brighter future. Health Econ. Policy Law 4, 527–546 (2009) Mcdermott, A.M., Toelle, T.R., Rowbotham, D.J., et al.: The burden of neuropathic pain: results from a cross-sectional survey. Eur. J. Pain 10, 127 (2006) Mcfadden, D.: Conditional logit analysis of qualitative choice behavior. in Frontiers of Econometrics, ed. by P. Zarembka, Academic Press, New York, pp. 105–142 (1973) Meyer-Rosberg, K., Burckhardt, C.S., Huizar, K., et al.: A comparison of the SF-36 and Nottingham Health Profile in patients with chronic neuropathic pain. Eur. J. Pain 5, 391–403 (2001) Morgan, A., Shackley, P., Pickin, M., et al.: Quantifying patient preferences for out-of-hours primary care. J. Health Serv. Res. Policy 5, 214–218 (2000) Moulin, D., Clark, A., Gilron, I., et al.: Pharmacological management of chronic neuropathic pain—consensus statement and guidelines from the Canadian Pain Society. Pain Research & Management: The Journal of the Canadian Pain Society 12, 13 (2007) Mu¨hlbacher, A., Juhnke, C.: Patients preferences versus physicians judgments: does it make a difference in health care decision-making? Appl. Health Econ. Health Policy 11, 163–180 (2013) Mu¨hlbacher, A., Juhnke, C., Bethge, S.: Experts’ judgment on patient-centered coordinated care. Value Health 13, A337 (2010) Mu¨hlbacher, A., Lincke, H., Nu¨bling, M.: Evaluating patients’ preferences for multiple myeloma therapy, a discrete choice experiment. GMS Psycho-Social-Med. 5 (2008) Mu¨hlbacher, A.C., Bethge, S., Tockhorn, A.: Pra¨ferenzmessung im Gesundheitswesen: grundlagen von Discrete-ChoiceExperimenten. Gesundheitso¨konomie Qualita¨tsmanagement 1, 17–44 (2013)

123

AC. Mu¨hlbacher et al. 67. Mu¨hlbacher, A.C., Stoll, M., Mahlich, J., et al.: Patient preferences for HIV/AIDS therapy-a discrete choice experiment. Health Econ. Rev. 3, 1–8 (2013) 68. Neuhauser, H., Ellert, U., Ziese, T.: Chronische Ru¨ckenschmerzen in der Allgemeinbevo¨lkerung in Deutschland 2002/2003: pra¨valenz und besonders betroffene Bevo¨lkerungsgruppen. Das Gesundheitswesen 67, 685–693 (2005) 69. Nice Neuropathic pain: the pharmacological management of neuropathic pain in adults in non-specialist settings. Clinical guideline 96. In: National Institute for Health and Clinical Excellence London (2010) 70. Ohayon, M.M., Stingl, J.C.: Prevalence and comorbidity of chronic pain in the German general population. J. Psychiatr. Res. 46, 444–450 (2012) 71. Orme, B. Sample size issues for conjoint analysis studies. Sawthooth Software Research paper Series. Squim, WA, USA: Sawthooth Software Inc (1998) 72. Phillips, K.A., Johnson, F.R., Maddala, T.: Measuring what people value: a comparison of ‘‘attitude’’ and ‘‘preference’’ surveys. Health Serv. Res. 37, 1659–1679 (2002) 73. Pittler, M.H., Ernst, E.: Complementary therapies for neuropathic and neuralgic pain: systematic review. Clin. J. Pain 24, 731–733 (2008) 74. Poliakov, I., Toth, C.: The impact of pain in patients with polyneuropathy. Eur. J. Pain 15, 1015–1022 (2011) 75. Ryan, M., Bate, A., Eastmond, C.J., et al.: Use of discrete choice experiments to elicit preferences. Qual. Health Care 10(Suppl 1), i55–i60 (2001) 76. Ryan, M., Farrar, S.: Eliciting preference for healthcare using conjoint analysis. BMJ. 320, 1530–1533 (2000) 77. Ryan, M., Farrar, S.: Using conjoint analysis to elicit preferences for health care. BMJ (Clinical research ed) 320, 1530–1533 (2000) 78. Ryan, M., Gerard, K.: Using discrete choice experiments to value health care programmes: current practice and future research reflections. Appl. Health Econ. Health Policy 2, 55–64 (2003) 79. Ryan, M., Gerard, K., Amaya-Amaya, M.: Using discrete choice experiments to value health and health care. Springer, Dordrecht (2008) 80. Ryan, M., Hughes, J.: Using conjoint analysis to assess women’s preferences for miscarriage management. Health Econ. 6, 261–273 (1997) 81. Sachversta¨ndigenrat Fu¨r Die Konzertierte Aktion Im Gesundheitswesen (Svr) Gutachten 2000/2001: Bedarfsgerechtigkeit ¨ ber-, Unter- und Fehlverund Wirtschaftlichkeit. Band III. U sorgung. In:BT-Drucksache 14/6871 vom 31. 08. 2001 (2001) 82. Schmader, K., Gnann, J.W., Watson, C.P.: The epidemiological, clinical, and pathological rationale for the herpes zoster vaccine. J. Infect. Dis. 197, S207–S215 (2008) 83. Schmader, K.E.: Epidemiology and impact on quality of life of postherpetic neuralgia and painful diabetic neuropathy. Clin. J. Pain 18, 350–354 (2002)

123

84. Scott, A., Vick, S.: Patients, doctors and contracts: an application of principal-agent theory to the doctor-patient relationship. Scottish J. Polit. Econ. 46, 111–134 (1999) 85. Siegel, E.: Versorgungsstrukturen, Berufsbilder und professionelle Diabetesorganisationen in Deutschland. Deutscher Gesundheitsbericht Diabetes 2011, 23–33 (2010) 86. Street, D.J., Burgess, L.: The construction of optimal stated choice experiments: theory and methods. Wiley, London (2007) 87. Tan, T., Barry, P., Reken, S., et al.: Pharmacological management of neuropathic pain in non-specialist settings: summary of NICE guidance. BMJ 340, c1079 (2010) 88. Telser, H.: Nutzenmessung im Gesundheitswesen: die Methode der Discrete-Choice-Experimente. Dr. Kovacˇ, Hamburg (2002) 89. Telser, H.B.K., Zweifel, P.: Validity and reliability of willingness-to-pay estimates: evidence from two overlapping discretechoice-experiments. Patient, Patient-Cent. Outcome Res. 1, 283–293 (2008) 90. Telser, H., Zweifel, P.: Measuring willingness-to-pay for risk reduction: an application of conjoint analysis. Health Econ. 11, 129–139 (2002) 91. Thurstone LL: A law of comparative judgment. In Maranell GM (ed): Scaling: A Sourcebook for Behavioral Scientists. Chicago, IL, Aldine, pp 81–92 (1974) 92. Torrance, N., Smith, B.H., Bennett, M.I., et al.: The epidemiology of chronic pain of predominantly neuropathic origin. Results from a general population survey. J. Pain 7, 281–289 (2006) 93. Train, K.: Discrete Choice Methods with Simulation. Cambridge University Press, Cambridge (2009) 94. Trope, Y., Liberman, N.: Construal-level theory of psychological distance. Psychol. Rev. 117, 440 (2010) 95. Trope, Y., Liberman, N., Wakslak, C.: Construal levels and psychological distance: effects on representation, prediction, evaluation, and behavior. J. Consum. Psychol 17, 83 (2007) 96. Ubach, C., Scott, A., French, F., et al.: What do hospital consultants value about their jobs?A discrete choice experiment. BMJ 326, 1432 (2003) 97. Va´zquez, M., Shapiro, E.D.: Varicella vaccine and infection with varicella–zoster virus. N. Engl. J. Med. 352, 439–440 (2005) 98. Veves, A., Backonja, M., Malik, R.A.: Painful diabetic neuropathy: epidemiology, natural history, early diagnosis, and treatment options. Pain Med. 9, 660–674 (2008) 99. Vick, S., Scott, A.: Agency in health care. Examining patients’ preferences for attributes of the doctor-patient relationship. J. Health Econ. 17, 587–605 (1998) 100. Wong, M.-C., Chung, J.W., Wong, T.K.: Effects of treatments for symptoms of painful diabetic neuropathy: systematic review. BMJ 335, 87 (2007) 101. Ziegler, D.: Painful diabetic neuropathy: treatment and future aspects. Diabetes/metabolism Res. Rev 24, S52–S57 (2008) 102. Zussman, J., Young, L.: Zoster vaccine live for the prevention of shingles in the elderly patient. Clin. Interv. Aging 3, 241 (2008)