“Patient-relevant endpoints” - assessment from the patient’s perspective?

Most decision makers, including IQWiG, consider various patient-relevant endpoints to demonstrate the superiority of a therapy over a placebo or a comparator. The question arises: How can contradictory results in different endpoints be interpreted? The problem in benefit assessments with more than one endpoint is the weighting and the subsequent calculation of a key figure for the overall benefit. Without the aggregation of several endpoints into a score/index, the derivation of recommendations or regulatory decisions could be questionable.

>> The term “preference” generally refers to the preference, precedence or favoring of an alternative or the preference that someone has for something. As a one-dimensional indicator for explaining (choice) decisions, preferences represent the extent to which an object of assessment is preferable to a specific person during a specific period of time (Voeth 2000). Preferences are therefore the basis for choosing an option; the utility of an option is the basis of preference. Known preferences therefore allow conclusions to be drawn about the individual benefits of the available options (Telser 2002).
An evaluation of care processes that goes beyond measuring patient satisfaction would have to make the patients' priorities and the preferability of alternatives the basis for the evaluation. The priorities can only be measured on the basis of the concrete voting decisions of those affected. Election decisions in this regard may relate to the following issues:
• Patient-relevant endpoints (cost-benefit trade-off),
• Risks/safety during use (cost-risk/safety trade-off) as well
• Properties of the therapy (properties trade-off).

A methodological approach to developing an adequate procedure is the discrete choice analysis used in preference research, which enables the evaluation and prediction of preferences based on individual selection decisions. Multiattribute evaluation methods, such as the discrete choice experiment (DCE), are based on the evaluation of stated preferences. There are two types of multi-attribute evaluation methods:
(1) preference-based methods, in which individuals are asked to evaluate and/or rank each product alternative.
(2) choice-based procedures, in which the respondent is given a number of product alternatives from which he then chooses an alternative.

In general, preference-based methods are referred to with the generic term conjoint analyses, while the sub-form of choice-based methods is called discrete choice analyses. The discrete choice experiment is a frequently used method for quantifying the preferences of citizens, insured persons or patients (Ryan et al. 2001; Marshall et al. 2010; de Bekker Grob et al. 2010). Discrete choice analyzes are decompositional methods (Ryan et al. 2008) made possible by the theoretical work of Lancaster (1966) and McFadden (1973). Only this theoretical work makes it possible to replace the ranking or rating of many different product or service variants with a pairwise comparison, ie with a (selection) choice decision. This method of measuring preference uses a stochastic utility function (random utility function), in which the random component includes all undefined factors that influence the choice. Random utility theory assumes that the utility of a good or service consists of a systematic or observable component and a random, unobservable component (Thurstone 1927; McFadden 1973; Manski 1977). DCEs are based on the demand theory formulated by Lancaster, according to which the benefit that a product or service provides to the consumer (or demander or patient) depends on the properties of the product and the characteristics of the consumer. The benefit of a therapy therefore depends on the properties of the therapy, e.g. effect, reduction of symptom burden or side effects, etc.
Discrete choice experiments are specifically designed to provide information about an individual's willingness to accept trade-offs between individual properties (attributes) of a product. It is based on the hedonistic principle that products are made up of a number of different attributes and that the attractiveness of a product to an individual is a function of these attributes. DCEs are based on the assumption that individuals choose the alternative that guarantees them the highest utility in choice situations. However, the observer does not know this benefit with certainty and must therefore treat it as a random variable (Ben-Akiva/Lerman 1985; Anderson et al. 1992). Statistical analysis of individual choices can identify the relative importance of attributes and trade-offs (i.e., the amount of one attribute that respondents are willing to give up to get more of another attribute). Individuals' relative preferences between product features and levels vary, and thus they are willing to accept compromises (Thurstone 1974; Johnson et al. 2000; Telser/Zweifel 2002; Hensher et al. 2005).
IQWiG pilot study: The German method concept for cost-benefit assessment provides for the use of an efficiency limit within an indication area. The benefit is assessed by IQWiG based on patient-relevant endpoints. However, when making decisions based on multiple efficiency limits, it remains unclear how the results will be interpreted by the decision maker. The aim of the IQWiG pilot project was to present the possible uses of conjoint/discrete choice analysis to identify patient-relevant endpoints. The first question was how can we identify these decision criteria (endpoints)? Furthermore, it should be examined whether this method can be used for weighting and prioritization across multiple endpoints. Building on this, it was examined to what extent this method can be used to prioritize health technologies based on multiple endpoints. An indication-specific comparison of treatment alternatives is conceivable. The aim of the feasibility study was the identification, weighting and aggregation of patient-relevant endpoints using the example of antiviral hepatitis C therapy (Mühlbacher et al. 2012).
The results showed that the discrete choice method offers a theoretically sound method for identifying, weighting and prioritizing patient-relevant outcomes. Furthermore, with the help of DCEs, preference differences between experts and patients can be shown and compared. Furthermore, the value function can be approximated and a higher-level utility value can be formed. The graphical derivation of a higher-level efficiency limit can therefore be represented (Mühlbacher et al. 2011).
Discussion: DCEs for identifying and weighting patient-relevant outcomes
The discrete choice experiment method offers numerous practical advantages (Ryan et al. 2008), making it an internationally established method for eliciting preferences. It is considered to be well established based on years of research and is increasingly being used in healthcare (Louviere et al. 2000; Ryan/Farrar 2000; Phillips et al. 2002). The use of conjoint analysis is increasing rapidly in medicine (Bridges et al. 2008), and professional societies have developed guidelines for the use of conjoint analysis in health services research (Bridges et al. 2011).
Discrete choice experiments provide the most comprehensive information about the trade-offs individuals are willing to make between treatment effectiveness, safety, and other factors that determine patient well-being (Louviere et al. 2000; Flynn et al. 2008; Louviere/ Lancsar 2009).
Despite the growing popularity of stated preference methods among health economists, they are not yet widely known among industry decision-makers and healthcare providers. They have also only been used to a limited extent in health technology assessment or health economic evaluation (Bridges 2003).
However, the possible applications of discrete choice models to determine patient preferences and willingness to pay are increasingly being investigated in the context of cost-benefit analysis and cost-effectiveness analysis of medical interventions (Johnson/Hauber 2008; Lancsar/Louviere 2008; Mühlbacher et al . 2009; Mühlbacher et al. 2011; Mühlbacher et al. 2011). Despite the specific characteristics of the healthcare markets, a large number of studies have been carried out on preferences with regard to service provision, therapy characteristics and compliance.
Today there are, among other things, studies based on this approach on:
• Psoriasis (Ashcroft et al. 2006; Seston et al. 2007),
• Overweight (Roux et al. 2004),
• Chronic pain (Mühlbacher et al. 2011),
• Multiple myeloma (Mühlbacher et al. 2008; Mühlbacher/Nübling 2011),
• HIV/AIDS (Mühlbacher et al. 2011; Mühlbacher et al. 2012),
• ADHD (Mühlbacher et al. 2009; Mühlbacher/Nübling 2010),
• Multiple sclerosis (Johnson et al. 2009),
• the quality of life in cancer patients (Johnson et al. 2006),
• Crohn's disease (Johnson et al. 2010)
• Hepatitis B (Mohamed et al. 2012)
• Non-small cell lung cancer (Bridges et al. 2012) or
• Type II diabetes (Hauber et al. 2009),

In principle, patient preferences can be included in the decision-making processes at all levels. In the international context, there is increasing discussion about how the perspective of the affected consumers can be taken into account. The spectrum of participation efforts extends from the qualitative assessment of patient needs (NICE Citizen Councils in the UK) to the scientifically based documentation of quantitative patient preferences (PCORI Patient-Centered Outcomes Research in the USA). Canada and Australia are also working on reforms to encourage patient involvement in decision-making. Internationally, the methods of multi-criteria or multi-attribute decision making (multi-criteria decision making) are increasingly being discussed as alternatives to the concept of cost-effectiveness studies. What all approaches have in common is that patient benefit (overall benefit) is viewed as a multidimensional construct that can only be evaluated with the involvement of the affected patients. As a one-dimensional indicator for explaining (choice) decisions, patient preferences represent the extent to which a characteristic or an object of evaluation is preferable. The perceived or expected benefit from the patient's perspective is therefore the basis for preferences.
Against the background of the inclusion of patient preferences in the context of the approval, assessment of reimbursement and price setting of medicines, the question for the public is “How unsafe is safe enough?”, while for the individual patient the question is “What risks do I take for what benefit?” on me?” is of great importance (Klinke/Renn 2002). As part of the clinical and economic decision-making process, the question is how the interests of the patient can be integrated into the benefit-risk assessment. It should be noted that the perspectives and risk attitudes of political decision-makers, doctors and patients can be different (Hart 2005).
In general, three areas of application for the results from preference measurements seem conceivable:
(1) Utility Assessment/Health Technology Assessment: Description of the benefits of different alternatives by adding the existing preference weights of a actually existing alternative.
(2)Product design & evaluation: Information about how certain products should be designed so that they produce the greatest benefit for certain patient groups.
(3) Prediction: Providing predictive information about how much a hypothetical product would influence patient acceptance.
The method is also suitable for assessing health conditions (health valuation) as well as for operationalizing a multidimensional overall benefit or determining willingness to pay as an additional decision criterion in the context of Health Technology Assessment (HTA), for example: B. in the different phases of the development of drug innovations or health technologies (Mühlbacher et al. 2012). The use of preference data in product design has already been tested as part of the description of an optimal care program. For example, in two similar studies in the USA (Mühlbacher et al. 2011) and in Germany (Mühlbacher 2012), 21 attributes of integrated care networks were assessed on a patient-specific basis. In a study dedicated to the third area of application, the focus would be on compliance or adherence: How can patient behavior be controlled through preference-based care offerings? In addition to recording patient preferences, the discrete choice experiment has already been used to determine social preferences (Ratcliffe 2000; Green/Gerard 2009; Whitty et al. 2011). Social preferences are measured under the implicit assumption that individuals act altruistically in their decision-making to maximize their individual utility by maximizing societal utility as a whole. Furthermore, with the help of discrete choice experiments, differences in preferences between experts and patients can be shown and compared (Mühlbacher/Nübling 2011; Mühlbacher/Juhnke 2012).
Summary
With the inclusion or consideration of clinical endpoints in the economic benefit assessment, two currently independent study worlds are united. In the future, study results could be analyzed on the basis of evidence-based medicine and interpreted in the context of medicine and economics. The state, authorities or public institutions as administrators of citizens, insured persons or patients decide on the design of the care system and the catalog of services. This requires the assessment and evaluation of service offerings and the forecast of demand and acceptance of the available alternatives. The reimbursement or ability to reimburse services and products is a central area that must be decided. When considering costs, it is important to determine whether it makes sense to accept a higher price for a new product. A higher price is only justified if a service is associated with greater patient benefit or less harm to the patient. When a provider wants to introduce a procedure, a product or a technology into the German healthcare system, the primary question is “What will the patient get out of it?” This makes it clear that preferences, as an expression of patient benefit, play a decisive role in decision-making in the healthcare system have to play. One method for recording and documenting latent preference structures and the deliberation processes with regard to treatment methods is the economically sound discrete choice analysis (McFadden 1974; Thurstone 1974), which can be used to demonstrate the additional patient benefits of products, services or innovative technologies. <<