Patient value: its nature, measurement, and role in real world evidence studies and outcomes-based reimbursement

Abstract

The assessment of “patient value” is fundamental to clinical trials, real world evidence studies, and outcomes-based reimbursement schemes. Measures of health-related quality-of-life (HRQoL) are widely used in health research. Such measures are effective in determining the presence or absence of symptoms and functional ability. However, HRQoL measures were not intended, nor designed, to determine the value to patients of alternative health states. Functions have no intrinsic value—they are a means to fulfil human needs. However, needs can be met in a variety of ways, for example by adopting different functions or by the provision of social services. It is possible to analyze all functions in terms of the needs they satisfy. A needs model has been applied in health research since the 1990s. It is concerned with the extent to which human needs are fulfilled in the presence of disease and its treatment. It is argued that this is the major concern of the patient. Needs-based measures are patient-centric and produce a valid unidimensional index of outcome. Consequently, they provide a direct means of measuring patient value. This approach provides the possibility of evaluating health services in terms of the value they provide to consumers and payers. It also has a role to play in real-world evidence studies and outcomes-based reimbursement. It is recommended that greater attention is given in future to the development of patient-reported outcome measures that provide direct assessments of patient value.

Keywords:

• Health-related quality-of-life
• health service evaluation
• needs-based quality-of-life
• outcomes-based reimbursement
• patient reported outcome measures
• patient value
• real world evidence

“Value” is neither an abstract ideal nor a code word for cost reduction¹.

Background

Since the beginning of “modern medicine” in the late 18th century, clinical outcomes have focused on medical criteria². In the late 19th century, even greater reliance was placed on biological assessments of health status. Advances in medicine and the availability of expensive new treatments in the second half of the 20th century did not seem to translate into effects noticeable to the patient³. This led to the emergence of the concept of “health-related quality-of-life” (HRQoL), although it was referred to as “quality-of-life” until the 1980s. Its origins were the functionalist approach of Durkheim⁴ in the late-19th century. Function-based measures appeared particularly attractive to the medical world of the 1950s and 1960s, as there was still a reluctance to accept subjective aspects of outcome assessment. The concept was developed further by Parsons⁵, who defined health as “the state of optimum capacity for the effective performance of valued tasks”.

HRQoL provided a framework for contrasting improvements in health status with potential adverse effects of interventions. It also signaled a move in outcome assessment from the technological to the social arena. HRQoL can be defined as the functional effect of illness and its treatment, as perceived by the patient. It is a multidimensional concept embracing physical, emotional, and social components⁶.

As HRQoL focuses on the measurement of symptoms and functions, it addresses aspects of life considered to be the province of health professionals. Indeed, most of the widely used HRQoL measures were developed by physicians. Consequently, its main value is in the assessment of health status^7,8. Figure 1 shows how both disease and its treatment influence symptoms and the ability to function. It should be noted that adverse effects associated with the treatment can also influence symptoms and functioning.

Figure 1. Relationships between disease, treatment, non-clinical influences, and outcomes.

Information about health status is clearly valuable to the clinical assessment of outcome. HRQoL categorizes, rather than evaluates, health status. In this respect, it resembles measurement in the physical sciences. A kilometer is longer than a meter, which is longer than a centimeter. However, none of these lengths is “better” than the others. The 6-minute walk test⁹ tells us nothing about patient value, only providing an inexact measure of walking capacity. It does not necessarily matter whether we can walk 1 km or 5—the distance has no intrinsic value. Of course, we might gain value in our lives by walking the extra distance, such as enjoying a walk in the countryside, shopping for food, or visiting friends and family.

Length of life has no intrinsic value to the individual. The benefit of living until 80 rather than 60 years of age must be inferred. The potential value of the additional years is clear, but the presence of chronic illness may limit their actual value. HRQoL measurement does not provide assessments of patient value. Instead, value must be inferred from scores obtained on HRQoL measures. Such inferences are not standardized, and actual value is likely to vary from person to person. It is not surprising that HRQoL measures do not value functions, as they were developed to determine health status, not patient value.

To value functional ability, it is necessary to determine the benefits patients gain from functioning. For example, social interaction allows us to have fun, share experiences, discuss problems, and find friendship, love, and partnerships. By measuring these added values of healthcare, inferences would no longer be needed. In fact, it is possible to analyze any function in terms of the value that it provides to the individual.

Patient value is fundamental to the evaluation of health services, whose only role is to increase patient value. In a world where there are limited resources, medical care is rationed. Consequently, it has become essential to maximize the patient value gained per dollar spent. For this to be achieved it is necessary to determine patient value accurately—not to infer what it might be.

How can patient value be measured? The obvious approach is to bypass “physical” measures of health status and determine value directly. This requires a clear definition of what patient “value” is.

The role of human needs in valuing life

Throughout history, theorists and philosophers have been interested in the relationship between quality-of-life (QoL) and human needs. In 1516, Sir Thomas More¹⁰ argued that life quality was dependent on the satisfaction of certain human needs. These included the need for good health, mobility, adequate nutrition, and shelter. In the second half of the 20th Century, psychologists realized that individuals were driven or motivated by their needs^11,12. The role of human needs in motivating behavior and providing satisfaction has been recognized in many fields of research. Some examples follow.

The Human Needs Theory shows how conflicts can be managed and resolved through the satisfaction of basic human needs¹³. In the Palestinian–Israeli conflict, the Palestinians believe that their identity (their land) is denied them, while the Israelis feel a lack of security due to attacks from their neighbours. Needs Theory is concerned with satisfying the needs of all parties involved in a conflict. Burton¹³ identified security/safety, identity, recognition, and personal development as needs that must be addressed in conflict resolution. At a local level, needs are generally fulfilled by family and community, and by policies and services provided by the state. However, if the state fails to address these needs appropriately, or if a group feels that their needs are unmet, violence can result¹⁴.

Jahoda’s¹⁵ latent deprivation model argued that employment is important as it provides high levels of time structure, social contact, collective purpose, physical activity, and status. Unemployment is detrimental to health, because these needs are not met¹⁶. Jahoda’s¹⁵ work has been confirmed more recently by Karsten and Batinic¹⁷, who showed that employees reported higher levels of satisfaction of these needs (except for status) in comparison to unemployed persons and others outside the workforce.

The contribution of need fulfilment to the quality of human life has been of equal concern in the context of human rights and justice. Hassoun¹⁸ highlighted the connections between need fulfilment and health when considering what a good society enables its members to secure. Malnutrition inhibits the immune system. Inadequate shelter exposes people to environmental hazards. People whose emotional and social needs are unmet are at high risk of mental and physical illness.

The role of needs in health outcomes became recognized in the late 20th Century. Minshull et al.¹⁹ proposed the Human Needs Model of Nursing. This placed emphasis on patient problems arising from both unmet basic and psychological needs. They argued that the application of such a model of nursing would acknowledge the holistic and dynamic nature of people. Need fulfilment was also found to mediate the relationship between work–life balance and wellbeing²⁰. The key factor was whether the individual could fulfil his or her needs in non-work time.

The importance of needs to health has been identified in obesity research. The Basic Psychological Needs in Exercise Scale²¹ was developed to investigate the role of need fulfilment in exercise participants. Need for competence was the main predictor of exercise attendance, adherence, and dropout. Need fulfilment also underpinned the relationship between health-enhancing physical activity and wellbeing²². Greaves et al.²³ argued that behavior changes required for weight loss cause psychological tension, as the new behaviors are incompatible with fulfilling psychological needs. Management of this tension requires constant effort. They concluded that non-obesogenic methods of fulfilling needs and (potentially) changing self-concept are required.

Needs-based measures of patient value

The needs-based approach to QoL assessment was proposed in the 1990s²⁴. The measurement model grew out of qualitative research into the impact of depression on the lives of patients. Interviews with depressed people indicated that a majority were unemployed and/or living alone. Interviewees were probed to describe how they were affected by these functional limitations. They responded in terms of needs that were not fulfilled. Absence of employment meant that interviewees did not have shared goals or the opportunity to work as part of a team. Their opportunities for socialization were also restricted. Those without partners were deprived of the opportunity to share pleasant experiences, to have fun, and to be accepted and loved. Research in additional diseases indicated that all functions can be analyzed in terms of the needs that they satisfy^25–27.

Motivation is the reason for people’s actions and desires²⁸. We are motivated by our needs, and satisfied when they are met. The needs model hypothesizes that the value of individuals’ lives is dependent on the extent to which their human needs are fulfilled²⁴. Value is low when few needs are met.

The major influences on need fulfilment are the presence of disease and its treatment (see Figure 1). These effects can be summarized by HRQoL measures. Disease prevents need fulfilment, and effective interventions benefit the patient by allowing the satisfaction of more needs. Clearly, there is a need for good health, but this is taken as read in the context of clinical research. HRQoL measures are intended to assess only those factors that directly result from healthcare. Such a distinction is not possible, as ability to function is influenced by many additional factors such as social support, financial resources, aids and adaptations, assistance from others, education, technology, and environment (marked as non-clinical influences in Figure 1). Where an individual has a chronic disease, it is likely to be the major influence on his or her need fulfilment. Consequently, health services are crucial. By reducing symptoms of ill health and improving the ability to function, individuals should become better able to fulfil their needs.

Consider an individual with severe rheumatoid arthritis treated with a biologic product. If the treatment is effective, mobility may improve and pain reduce. HRQoL measures should detect these changes. However, the improvements may be insufficient to allow the person to walk very much further or to wash themselves. Consequently, HRQoL scores may improve without patient value increasing. Providing a care worker to assist the person to get out of bed, wash, dress, and prepare breakfast will not improve their HRQoL. However, several aspects of needs may be fulfilled, including; personal hygiene, social interaction, and nutrition. In this situation, patient value would increase with the provision of social care.

Items for needs-based measures are generated from qualitative interviews conducted with relevant patients. Interviewers ask how the patient’s life has been affected by the disease in question, and probe how limitations of functioning adversely affect the interviewees. This provides statements concerning need fulfilment. Interview transcripts are analyzed to identify common issues identified by the patient group. After testing for face and content validity, large samples are used to identify the final item set, and to establish reliability and validity.

The importance of unidimensionality

Item response theory (IRT) and particularly Rasch analysis represent modern psychometric theory, the successor to classical test theory²⁹. Rasch analysis³⁰ is widely used in the measurement of ability and achievement in education, and it is now becoming more widely used in clinical research. A basic assumption of rating scales is that their items measure a common underlying construct. Where this is the case, they are unidimensional³¹. If they are not unidimensional, they are technically invalid³². Health authorities accept that unidimensionality is essential for scales used as clinical trial endpoints³³. This quality is also fundamental to measuring patient value.

Where a scale is unidimensional, Rasch analysis places both patients and questionnaire items on the same measurement continuum. Consequently, it is possible to see whether the items are well targeted to the study population. The individual’s score is the point on this continuum where there is a 50% chance that they will affirm the item (assuming a dichotomous response format, as employed with needs-based measures). A good scale will have items equally spaced along the continuum, meaning that it is not necessary to measure all possible needs—but a representative sample of those needs that are most likely to be affected in the specific study population.

Rasch analysis is applied to data collected in the validation survey to ensure that the needs-based scales are unidimensional. Measures that fit the Rasch model provide fundamental results and operate at the interval, rather than ordinal, level³⁴.

Patient value measures should be truly patient-centric. As the content of needs-based measures is entirely derived from qualitative unstructured patient interviews, interviewees decide what is important to their lives and the issues that should be addressed by the outcome measure.

A measure of patient value must provide an index rather than a profile of different outcomes to allow changes in score to be interpretable. Needs based measures provide such an index, as they determine the extent to which human needs are met. This is confirmed by fit to the Rasch model. Furthermore, as all needs measures have the same theoretical measurement basis, scores on the different scales can be co-calibrated to allow valid direct comparison across diseases^35,36.

The index provided by needs based measures differs from that provided by measures of utility in several ways. Rather than determining need fulfilment, utility assesses the value of an individual’s HRQoL. Furthermore, measures such as the EQ-5D³⁷ and willingness-to-pay³⁸ generally focus on valuations made by lay people rather than relevant patients. Finally, utility values are claimed to provide measurement at the ratio level. This implies that there is a meaningful zero point to the scale. Needs-based measures assess at the interval level and do not identify an absolute zero. However, such a value is conceivable.

Since the 1990s, over 30 disease-specific needs-based measures have been developed. Example items from these measures are shown in Table 1. Many of these measures have been widely adapted for use in up to 60 languages. A needs-based measure is easier to adapt, as human needs are universal, while functions performed to meet needs can vary considerably across cultures. Wagner et al.³⁹ report that the most difficult items to translate in the SF-36 were “physical functioning items that refer to activities not common outside the United States”. For example, the Taiwan adaptation of the SF-36 replaces “playing golf” with “playing Tai-Chi”⁴⁰.

Table 1. Sample of needs-based questionnaire items.

I’ve lost interest in food

I feel vulnerable when I’m on my own

I feel that I’m losing my role in life

I rarely feel clean

I feel dependent on others

I can’t do things on the spur of the moment

Needs-based measures of patient value have been widely used in international clinical studies. The QoL-AGHDA²⁵ is used in the UK to decide which patients gain sufficient value from growth hormone replacement to benefit from treatment. If their QoL-AGHDA scores do not improve with treatment (i.e. they do not benefit) their treatment is discontinued. The ASQoL²⁶ was used to obtain a QoL label for Humira for the treatment of ankylosing spondylitis. Needs-based measures have also shown the patient value gained from non-clinical interventions, where HRQoL measures failed to detect change^41,42.

The potential value of needs-based measures

The requirement for direct measures of patient value is increasing rapidly as healthcare funding and research methods are changing.

Clinical studies and routine clinical care

Use of patient-centric outcome measures allow clinical trials to generate information on patient value resulting from different interventions. Such data are complementary to information on clinical outcomes that are already measured in the trial and/or with HRQoL measures. Similarly, when used in routine clinical practice, the clinician can see whether the patient is gaining value from his/her care.

Real-world evidence and pragmatic clinical trials

Manufacturer’s remuneration is not connected sufficiently with the real-life performance of their products⁴³. Clinical trials carefully select participants and avoid as many potential confounders as possible. In real life, the use of medicines is quite different. People may have several different conditions, particularly as they grow older. They may use numerous additional medications and miss doses from time to time, or regularly. Consequently, drugs may not work as well in the real world as in clinical trials. Furthermore, it is possible that, in real life, new drugs may not be superior to older, cheaper drugs. Schwartz and Lellouch⁴⁴ distinguished explanatory trials that confirm physiological or clinical hypotheses from pragmatic trials that provide the evidence necessary to inform decisions on real-world clinical practice. In the latter case, it is necessary to take account of all factors that may influence the impact of an intervention on the patient. These may include the availability of carers, social services, financial resources, adaptations to housing, additional medications and diseases, and employment status. A robust measure of patient value is required for such studies. Needs measures determine the extent to which needs are met in the individual’s real life, irrespective of the nature of interventions and resources.

Non-clinical interventions

Relatively little attention has been paid to the effects of non-clinical interventions on patient value. However, such interventions may improve need-fulfilment, even where symptoms do not change. For example, increases in social interaction may well lead to needs for sharing experiences or working as part of a team to be fulfilled.

Combined clinical and non-clinical interventions

Healthcare research has traditionally investigated the impact of a single intervention. The move to real-world evidence studies requires outcome measures that can determine the impact of a potentially wide range of health and social services and personal behaviors. Needs-based measures are unrelated to the nature of the intervention, simply determining whether relevant needs are fulfilled. Consequently, they can establish the value gained by patients from any combination of different types of intervention. Indeed, patient value measures could provide a method for identifying the best combination of interventions. This is particularly important when considering beyond the pill services.

All individuals are influenced by both clinical and non-clinical interventions and other changes that may occur in their lives. The restrictions imposed by clinical trial protocols will not overcome all such confounding variables. It is neither possible nor appropriate to attempt to restrict changes in outcome to those that directly result from treatment. The true value of a new treatment to an individual is determined by their personal circumstances and existing level of need fulfilment. It is for this reason that real world trials are needed⁴⁵.

Outcomes based reimbursement

“We must move away from a supply driven healthcare system organized around what physicians do and toward a patient-centered system organized around what patients need”⁴⁶.

The lack of connection between the amount paid for drugs and the benefits to patients has led to significant pressure to constrain spending on specialist drugs in the UK⁴³. Rather than paying for services provided, the intention is that payment should be related to the benefit that accrues. Presumably, it is the benefit to the patient (increase in patient value) that matters. Patients are the payers, either directly or indirectly, through insurance premiums or tax. The provider must demonstrate the value gained by the intervention. In contrast, the payer needs to know the patient value gained by the intervention, to justify the payment made. Such a process requires accurate assessment of patient value.

Carlson et al.⁴⁷ argued that, due to the current emphasis on health technology assessment, payers and decision-makers are requiring more and better evidence to inform coverage and reimbursement decisions. They report that reimbursement schemes between medical product manufacturers and payers have been set up. The performance of the product is tracked in a defined patient population for a specified time. The level of reimbursement is then related to the observed outcome. In the UK, the National Health Service has set up such a scheme for the treatment of multiple sclerosis⁴⁸. Johnson & Johnson agreed to refund the NHS for multiple myeloma patients who failed to respond to bortezomib⁴⁹. Similar schemes have been established in the US⁵⁰. Such agreements rely on meeting certain clinical targets, and this can lead to disagreements between providers and payers about what the targets should be. Again, the targets may have little significance to patients, and there is a tendency to track what is easily measured rather than what matters to the patient (termed “outcome first/patient second”)⁵¹. It is arguable that the availability of true measures of patient value would lead to better reimbursement decisions being taken.

Cost–utility analysis (CUA)

Cost–utility analysis commonly uses preference weights that are converted into Quality-Adjusted Life Years (QALYs). The intention is to determine changes in both quantity and quality of life⁵². The technique has been seriously questioned by many for theoretical, ethical, and methodological reasons⁵³. Perhaps the most serious concern is whether the tools commonly used to determine utility (such as the EQ-5D³⁶ and SF-6D⁵⁴) measure quality-of-life. These generic HRQoL measures have major shortcomings, as they are short and, consequently, unlikely to measure the factors that patients consider most important in specific diseases. Furthermore, they assess HRQoL rather than how individuals’ lives are affected (see Figure 1). There is potential for using needs-based measures to determine the “quality-of-life” variable in QALY calculations. This would ensure that a disease-specific, patient-centric outcome was used to calculate a QALY. A utility measure has been developed from the needs-based measure specific to pulmonary hypertension (CAMPHOR)⁵⁵. This Utility Index was able to distinguish between clinical severity classes, and was a more sensitive measure of outcome than the 6-minute walk test⁵⁶.

Conclusions

Porter¹ argued that the value of health services remains largely unmeasured and misunderstood. Value must be defined “around the customer”, and the creation of value for patients should determine the rewards for all other actors in the system. As value depends on results, not inputs, value in healthcare is measured by the outcomes achieved, not the volume of services delivered.

This article has argued that traditional HRQoL measurement, while valuable, is not capable of determining patient value. It identifies the presence or absence of symptoms and disabilities, rather than how these affect the patient. Needs-based outcomes provide an effective method of measuring patient value that could be particularly important in generating real-world evidence and enabling true outcomes-based reimbursement.

While HRQoL still has a role to play in describing the impact of health status from the patient’s perspective, modern healthcare systems also need a new type of outcome measure. This must provide unidimensional, patient-centric, valid, and reliable assessments of patient value. It is hoped that greater attention will be given to developing disease-specific measures of patient value for a wide range of conditions. Better models of patient value may well be developed in time. However, needs-based measures currently provide the possibility of evaluating interventions in terms of their true impact on patient value.

Transparency

Declaration of funding

This manuscript was not funded.

Declaration of financial/other relationships

The authors develop needs-based outcome measures. JME peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Acknowledgments

None reported.