Patient-reported outcome instruments
A patient-reported outcome is a report of a patient's health that comes from the patient, and an instrument is the tool that captures that report and turns it into a score. Instruments differ in what they measure and what each kind offers, so the choice starts from the construct to be measured and the decision the data has to support. Instruments are used for medicines and devices alike, in trials, observational studies and registries, and aggregating their results at population level is a difficulty of its own.
What a patient-reported outcome is
Patient-reported outcomes reflect what patients think and how they feel about their disease and the treatments they receive. The US Food and Drug Administration defines a patient-reported outcome as any report of the status of a patient's health condition that comes directly from the patient, without interpretation of the patient's response by a clinician or anyone else. The European Medicines Agency adds that patient-reported outcomes are based on the patient's perception of a disease and its treatments, and that the term is an umbrella covering both single-dimension and multidimension measures of symptoms, health-related quality of life, health status, adherence to treatment and satisfaction with treatment.
A utility, a measure of a patient's or another stakeholder's preference for an outcome, often overlaps with a patient-reported outcome, although the terms are not synonyms. Patient-reported outcome is an umbrella term for input provided directly by patients about their treatment preferences, and its instruments typically rely heavily on qualitative input for development and are then translated into a validated numeric scoring system. Sources of patient-reported outcome information can include assessment tools that capture patients' global impressions, functional status, well-being, symptoms, treatment satisfaction, health-related quality of life and treatment adherence.
What patient-reported outcomes reveal
Patient-reported outcomes provide information on the aspects of health, disease and treatment that only the patient suffering from the condition knows, such as the frequency, severity and emotional repercussion of symptoms, the impact of the illness on everyday life, and the factors determining beliefs and behaviours towards treatments. They let investigators and clinicians know their patients' thoughts and perceptions of the healthcare process and of the expected results, which tests, technologies or other observers cannot unveil. Accounting for them helps to empower patients and to enhance communication among patients, healthcare providers and decision makers, and helps to anticipate the probable effectiveness of therapies.
Patient-reported outcomes are very useful for providing information about what patients think, how they feel, what their preferences are and why they behave as they do towards their disease and treatments, especially in chronic, disabling, progressive and other difficult-to-treat conditions. They may contain information little known to clinicians, policy makers and regulatory authorities, and their data will help to determine the effectiveness and success of treatments in usual clinical practice and in real life.
In usual clinical practice, clinicians and patients understand the effect of disease and treatment differently, for example the prevalence and severity of patients' symptoms, their functional impairments and the influence of the disease on everyday life, and these differences have been extensively researched and reported. Patient-reported outcomes bridge them. Patients' direct self-reporting on health problems also facilitates the discussion of important symptoms and quality-of-life aspects with healthcare professionals, which supports documentation and can help to improve disease management and positively influence clinicians' decisions.
Instruments and their kinds
A patient-reported outcome instrument is the tool patients use to assess their health condition, health status and other physical, social and mental functions as they perceive them. It is administered by the patient, by a healthcare provider through an interview, or by a combination of both. Responses take the form of a score, such as 1 to 5, an ordering, such as highest to lowest, or a choice among a set of answer options, such as the most and the least important, which can consist of pictures, numbers or categories. Instruments can be classified by the concept they measure, and a classification based on the Oxford University Patient Reported Outcomes Measurement Group gives six kinds, each with its advantages, disadvantages and an example:
- Generic instruments measure health status and the outcomes of illness. They are applicable to the general population and to a wide range of patient groups, assess an extensive variety of aspects of health and disease, are suitable across a broad range of health problems and for comparing treatments for different disease groups, and are useful for assessing the impact of healthcare technologies whose therapeutic effects are still uncertain. Some levels of detail that may be relevant to specific disease groups are sacrificed, and they are not sensitive to changes in health that may be clinically important. Example: the 36-Item Short Form Survey (SF-36).
- Utility measures capture preferences or values attached to individual health states. They produce information on the overall value of health states to society and are useful for economic evaluation studies. They are labour-intensive and time-consuming, and respondents may have difficulty understanding the tasks they are required to perform. Example: the five-dimension five-level EuroQol questionnaire (EQ-5D-5L).
- Disease-specific instruments capture a patient's perceptions of a specific disease or health problem. They are relevant to patients suffering from the disease, and to clinicians because they are responsive to clinically important changes resulting from interventions directed at controlling the health problem. Their health status scores cannot be compared with those obtained for the general population, comparisons across treatments for different diseases are not possible, and they may not be sensitive for detecting side effects or unforeseen effects of treatment. Example: the Audit of Diabetes Dependent Quality of Life (ADDQoL).
- Population-specific instruments capture particular demographic groups' perceptions of disease, such as children's or elderly people's. Their content is more relevant to the group in question, their format is specifically tailored, with cartoon illustrations for example, and they are more sensitive to systematic differences between population groups. Their health status scores cannot be compared with those obtained for the general population, and comparisons across population groups may not be possible. Example: the Child Health and Illness Profile, Child Edition (CHIP-CE).
- Dimension-specific instruments capture the severity of symptoms, or of another dimension of disease, and their impact on functioning, role activities and psychological and social well-being. They provide a more detailed assessment of a particular dimension of health, and are not appropriate as the sole outcome measure for evaluating the effectiveness of treatment. Example: the Brief Symptom Inventory (BSI).
- Individualised instruments capture the issues, concerns or domains of personal concern to the respondent. Their content validity is high, and they have to be administered by interview, which makes them labour-intensive and time-consuming. Example: the Schedule for the Evaluation of Individual Quality of Life (SEIQoL).
Ideally, an instrument should be specific to the concept being measured, such as health status, within a framework of robust evidence gathered earlier, contain an optimum number of items to minimise response overload, have scales that are easy to use and if possible the simplest for the intended population to understand, be reproducible, and maintain patient confidentiality.
The construct to be measured
The chair of PROMIS Germany holds that the most problematic thing in developing a patient-reported outcome is that people jump straight into it without thinking about what construct they really want to measure. Parties frequently approach his group wanting to measure "quality of life" without really understanding what is meant by that, and in his view people bypass the first steps in developing the conceptual measurement model much too often, picking an established instrument out of the literature without questioning its appropriateness for their particular research question. He holds that the longer one is in the field, the more clearly one sees the need for the basics to be correct, and that initiatives such as the International Society for Pharmacoeconomics and Outcomes Research (ISPOR) guidelines for identifying and evaluating an existing measure are very useful in ensuring that the basic elements needed in a patient-reported outcome are present.
When writing, he advises keeping proximal outcomes, such as symptoms and function, apart from distal outcomes such as quality of life. In heart failure, shortness of breath and physical function are proximal outcomes likely to change under medical interventions, while a distal outcome like quality of life might not be affected by the intervention, because aspects also relevant to that construct, such as level of job satisfaction or environmental factors, are not what the intervention targets. He also advises care with terms, so as not to confuse outcomes with predictions or determinants: an outcome is something expected to change or vary based on other factors, yet people who write "patient-reported outcomes" are often thinking not of outcomes but of predictions.
Choosing an instrument
Appropriately selecting patient-reported outcomes and their instruments, measuring them, and accurately interpreting and reporting their results are fundamental, and essential to the reliability of the evidence generated. Verifying that a proper instrument has been selected is vital for interpreting the results adequately, and so enhances the chances of publication. If the results are to be used in labelling claims, they should also satisfy regulatory requirements. Three properties of an instrument are fundamental: validity, reliability and responsiveness. The calculation of the minimally important difference is also relevant.
Beyond the fundamental properties, a series of further aspects should be taken into account to assess whether the chosen instrument is appropriate, each with its question:
- Documentation: Is there formal written documentation, and are there publications on the use of the instrument, covering the type of research, the objectives and aims, the limitations and the findings? Is there a user manual on how to administer it, score it and interpret the results?
- Development: How was it developed, in terms of the methods and findings for content and concept development, validation in the original and other languages, and the robustness of the validity, responsiveness and reliability findings?
- Feasibility: Are the questions, tasks and scoring easily understandable? Is the mode of administration or data collection too long?
- Target population: Is the scale suitable for the target population, such as very ill people, children or elderly people? Is a carer needed to help?
- Language and cultures: Are the translations properly validated?
- Scoring: Is the scoring procedure defined, and is it easy to interpret?
- Interpretation: Are guidelines available for interpreting scale scores and for dealing with missing data?
- Licence for use: Is there a fee attached to the use of the instrument, with what copyright protection, holders and extension? What are the conditions for use, considering the number of projects to be conducted with the same instrument, the number of subjects in whom it will be used, the period during which it will be administered, and whether clinical practice, academic, research, private or public entities run or support the project?
Other considerations are the setting, nature and aim of the project and the type of healthcare decision to be made. An instrument for registries of patients' health records should prioritise practicality, easy administration, cost-effectiveness, low participant burden and simple documentation with other clinical data. An instrument for product labelling should reinforce high validity and reliability, sensitivity to changes, stability over time and low rates of missing data. An instrument for economic evaluation should focus on less complexity, speed, and sensitivity to incremental effects on health-related quality of life and to choices in decision making.
Where preferences for health states or for the characteristics of treatment are considered, the method for eliciting patients' preferences should be the most appropriate one. Examples of preference assessment are a ranking or rating scale, best-worst scaling, the standard gamble, the time trade-off, the visual analogue scale, discrete choice experiments and conjoint analyses, and multi-attribute utility instruments.
Some institutions provide accurate information on the characteristics and properties of instruments, together with recommendations for use and reporting and bibliographic references. The Mapi Institute maintains the Patient-Reported Outcome and Quality of Life Instruments Database, a large and comprehensive database in which a user can find the instrument that best addresses a specific research question. The On-line Guide to Quality-of-Life Assessment is another database of existing health-related quality of life and other instruments. The US National Institutes of Health PROMIS Initiative has developed rigorously tested item banks across a broad range of domains and subdomains that allow comparisons, actively evaluates techniques for achieving brevity in instruments, and makes many of its measures publicly available through the PROMIS Assessment Center.
Quality-of-life patient-reported outcome measures have become popular for assessing participants' perception of their own health and wellbeing, and selecting an appropriate measure is important for the reliability and validity of the participants' self-perception and definition of their quality of life. Choosing an appropriate measure of quality of life in clinical trials is complex. A 2023 review of peer-reviewed studies conducted between 2012 and 2023, which compared questionnaires' design and applicability with best practice as described by DeVellis and Thorpe (2021), identified five generic, validated quality-of-life questionnaires: the World Health Organization Quality of Life Assessment Instrument (WHOQOL), the 36-item and 12-item Short-Form Health Surveys (SF-36 and SF-12), the Nottingham Health Profile and the EQ-5D. It found the SF-12 the most suitable, for its ease of use, low participant burden and ability to identify differences between quality-of-life domains. The complexity of a questionnaire and its timing during the trial are factors that may affect the responses. Understanding how participants perceive their quality of life may inform clinical best practices and make them more patient-centric. A trial consists not only of data collection for publication but of processes to improve the participants' health.
Uptake by regulators, payers and developers
A review of evaluations for the approval of new pharmaceutical products by the European Medicines Agency between 1995 and 2003 showed more than a 30% increase in the use of health-related quality of life and other patient-reported outcome instruments, particularly in cancer-related treatments. About 24% of new drug approvals by the US Food and Drug Administration between 2006 and 2010 carried patient-reported outcome labelling, and this figure rose to almost 77% between 2011 and 2015, as most approvals of new products were for diseases that traditionally rely on patient-reported outcomes for evaluating the benefits of treatment.
In 2013 patient-reported outcomes were increasingly common in economic and clinical research, with multiple instruments, both disease-specific and designed to assess general health, being developed and validated. Because patient preferences can substantially influence whether a treatment is properly used or adhered to, payers and regulators were by 2013 recognising the value of patient-reported outcomes in drug development, and drug developers were recognising the potential of such information to improve their products' competitive profiles and likelihood of formulary acceptance. In acknowledgment of their increased relevance in clinical research, the US Food and Drug Administration published guidelines in 2009 for the use of patient-reported outcomes in medical product development.
Observational studies
Because patient-reported outcomes depict the results of treatments in real life, they are most frequently measured in observational studies, and in clinical research they are most frequently the primary endpoint of an observational study. They complement the efficacy and safety data usually generated in clinical trials.
Patient-reported outcome assessment has been shown to be paramount in generating information on situations where either exposing patients to an intervention or preventing them from receiving it is unethical, but where it is conceivable to gather their perspectives on the illness and to value their preferences for other possible disease scenarios. Measuring patient-reported outcomes in observational studies is also insightful in rare diseases, because reachable sample sizes are too small for conducting a clinical trial, while gathering primary data on patients' health-related quality of life and on the perceived determinants of disease burden is very important for healthcare decision making.
Patient-reported outcomes for medical devices
Patient-reported outcomes are important tools for measuring patients' experience with medical devices and the effects of treatment, and they are basically the same, and used in the same way, in the medical device and pharmaceutical sectors. Both sectors use general quality-of-life tools such as the EQ-5D and the Short-Form Survey questionnaires, such as the SF-12 and SF-36, and tools that target specific disease areas, such as the Kansas City Cardiomyopathy Questionnaire (KCCQ) and the Oswestry Disability Index.
The EQ-5D is the most commonly used patient-reported outcome in the medical device sector because it is easy to use, comprising five questions and a visual analogue scale. It is also one of the cheapest questionnaires, and cost plays a role in medical devices, particularly when somebody has to be convinced to use patient-reported outcomes. The EQ-5D can also be used to determine quality-adjusted life-years.
Quality of life after a procedure is important, since it is not enough to survive a treatment only to be left with intolerable symptoms; a successful procedure followed by a disabling stroke is undesirable. Conversely, a therapy might not increase survival but might improve quality of life. In trials in peripheral artery disease or of orthopaedic devices, pain reduction is highly relevant and can be measured with a patient-reported outcome such as the Wong-Baker FACES Pain Rating Scale or a visual analogue scale.
Patient-reported outcomes complement clinical and physiological information and can be used as clinical and physiological endpoints. In trials with rare complications, quantifying a patient's health status may be a critical requirement for assessing treatment benefits. Patient-reported outcomes are also relevant because they represent the patient's voice. By 2018 they had been used increasingly over the years, and their relevance was expected to continue to rise. In 2018, new consensus documents on clinical trial endpoints recommended quality-of-life endpoints: endpoint definitions included patient-reported outcomes in coronary intervention trials, and in transcatheter mitral valve trials an improvement in quality of life, such as an improvement of at least 10 points on the KCCQ, was part of the patient success endpoint.
Population-level data and registries
Patient-reported outcome assessment provides very rich information at the individual level, but aggregating the data to measure the performance of treatments and of the healthcare system delivering care at a target population level is challenging. One weakness is that the results patients give in questionnaires, or that interviewers relay, are outcomes based on the patient's own assessment, which may not be equivalent to the particular concerns of most patients suffering the same disease. Individual patients may also decide when and with whom they share their health and disease-related information, which may impede usability and access to information. These and other social issues, together with economic disparities, must be overcome, and in 2017 initiatives that include advanced analysis systems and predictive analytics were under way to improve data collection and the statistical management of patient-reported outcomes at population level.
Patient-reported outcome assessment has become a vital component in the design of patient registries, which should serve to improve the provision of healthcare, to inform decision makers and to gain knowledge of the true long-term effects of treatments on patients. Patient-reported outcomes are a cornerstone for generating real-world evidence and a vital component of registries designed to improve healthcare quality and the generation of information for decision-making bodies. Their collection is increasingly integrated into clinical registries to produce real-world data on the effectiveness of healthcare interventions, and their routine measurement has become more important for informing future care planning feasibly and efficiently. The challenges include selecting the most suitable outcomes and instruments, overcoming the logistic hurdles of collection, ensuring long-term sustainability and complete data gathering, controlling for selection bias and missing information, and managing data aggregation. To succeed, payers such as insurance systems, policy-makers, clinicians, patients and researchers should cooperate to find valuable and meaningful data in the outcomes collected in registries.