|Home | About | Journals | Submit | Contact Us | Français|
In modern medicine the traditional way of assessing change in patients has been to focus on laboratory or clinical tests. At its most simple this involves measuring pulse, blood pressure, and temperature, and carrying out physical examinations. At more complex levels it may include haematological analysis, computed tomography, radiography, organ function tests, genetic analysis, and other investigations. While these give important information about the disease, especially about chronic and progressive diseases, it is impossible to separate disease from an individual's personal and social context. No illness exists in a vacuum.
One way of capturing the personal and social context of patients is to use quality of life measures.1 These are accepted as outcome measures in clinical research but are rarely used in routine clinical practice, despite the fact that Florence Nightingale was one of the first clinicians to insist on measuring the outcome of routine care to evaluate treatment.2 This article reviews the challenges of using quality of life measures in clinical practice including selecting appropriate measures, analysing data, providing feedback, interpreting results, and incorporating these measures into clinical decision making. Practical ways of resolving the tension between the need for approaches suitable in the clinical encounter and the highly individualised nature of quality of life are also examined.
Quality of life measures have eight potential uses in aiding routine clinical practice. They can be used to prioritise problems, facilitate communication, screen for potential problems, identify preferences, monitor changes or response to treatment, and train new staff (box). They can also be used in clinical audit and in clinical governance. The first five of these are of immediate value in the clinical encounter, while the last three contribute to training, reviewing care, and improving care in the future. (A summary of how quality of life measures can be used to improve the quality of care appears on the BMJ's website.)
The underlying reason for using quality of life measures in clinical practice is to ensure that treatment plans and evaluations focus on the patient rather than the disease. Quality of life is not the only way to measure patient centred outcomes; measures of disability, social interaction and support, and psychological wellbeing may be appropriate. Quality of life measures are not a substitute for measuring outcomes associated with disease but are an adjunct to them: for example, rheumatologists do not treat rheumatoid arthritis with antirheumatic drugs simply on the basis of quality of life scores. Similarly, broad, multidimensional quality of life measures may be less effective, accurate, and responsive than measures of specific patient outcomes (for example, anxiety and depression) in situations where treatment is aimed at achieving a particular outcome. Quality of life is a highly individual concept. Mount and Scott likened the assessment of it to assessing the beauty of a rose6: no matter how many measurements are made (for example of colour, smell, and height), the full beauty of the rose is never captured. Quality of life measures will never capture all aspects of life that are important to an individual, although systems in which patients specify at least some of the qualities are likely to come closest. The individual nature and the shortcomings of many existing measures are discussed further in the next paper in this series.7
The routine use of quality of life measures is no substitute for training staff. There is the danger that staff may see the use of the assessment as an alternative to communicating with patients rather than as an aid to care. Training in the use of quality of life measures is something that is generally lacking in undergraduate and postgraduate education. In clinical governance and audit caution is needed in interpreting the results of these assessments and other outcome measures for a different case mix of patients.
The breadth of quality of life as a concept means that problems might be identified that are outside the usual remit of medical care.8 This raises a number of ethical concerns. Firstly, the act of measuring quality of life in a clinical setting may generate the expectation that the clinician will be able to influence it: otherwise, what would be the purpose of measuring it? In situations where this is not possible, patients may be seen to be harmed by the process of measurement. Secondly, some pressure groups, such as the movement for independent living in the United States, have opposed the clinical measurement of quality of life on the grounds that it represents the “overmedicalisation” of life and clinical interference in aspects of patients' lives that should not be the concern of the clinician. However, data from quality of life measures could be used to lobby for sufficient resources or to inform health and social policy. Thirdly, chronic disease affects and is affected by broader aspects of people's lives, such as their relationships and social support, and information on these aspects can influence treatment decisions and assessments of healthcare need.
Identifying and prioritising problems
Because the measure records information on a range of problems the patient and the doctor or nurse can identify which problems are most important. They can thus agree priorities. This is particularly useful when patients have multiple problems.3 Additionally, these measures can be used to capture information that superficially seems to have no clinical relevance but might explain disease severity or coping problems4
Because the measure presents clear information on a range of problems it can help patients to communicate their problems. If correctly applied it may speed the clinical encounter and help staff to focus on the patient's main concerns.
Screening for hidden problems
Some patients' problems can be overlooked unless specifically inquired about, especially psychological and social problems.5 For example, a measure that asks, “Would you describe your mood as depressed most of the time,” is a sensitive and specific screening tool for depression
Facilitating shared clinical decision making
Used in this way assessments help identify the patient's preferred outcome or treatment goals. If these are not known, then the treatment may not meet the patient's expectations, and this may affect adherence to treatment and the patient's satisfaction with care
Monitoring changes or responses to treatment
Change is usually monitored through laboratory or clinical tests rather than the patient's perception of change. Inability to bring about improvements that are seen as relevant to the patient may affect adherence to treatment
In addition to the properties needed when using a measure in research, such as validity and reliability, in clinical practice a wider range of properties are required to ensure that a measure can be used routinely. These include the appropriateness and acceptability of the measure, its responsiveness to clinical change, and its interpretability (box). (A more detailed version of this box appears on the BMJ's website.)
Validity: does the instrument measure what it is intended to measure, such as quality of life?
Appropriateness and acceptability: is the measure suitable for its intended use? This is crucial in clinical practice
Reliability: does the measure produce the same results when repeated in the same population?
Responsiveness to change: does the measure detect clinically meaningful changes? This is sometimes called sensitivity
Interpretability: can results from the measure be interpreted clinically and are they relevant?
Barriers to the routine clinical use of outcome measures, such as quality of life, include concerns about cost, feasibility, and clinical relevance.9 For a measure to have clinical usefulness it must not only be valid, appropriate, reliable, responsive, and able to be interpreted, but it must also be simple, quick to complete, easy to score, and provide useful clinical data.10 Most existing measures were developed for use in clinical research11 where time and budgetary constraints are different from those in clinical practice. Some quality of life measures require trained staff to administer them and are time consuming, taking 20-30 minutes to complete. Similarly, since the purpose of measuring the quality of life in clinical trials is to compare groups of patients (usually over relatively short periods) assessments of existing measures have focused on their performance in these situations. Such situations are different from clinical practice where the purpose of measurement is to assess change in individual patients, in some instances over many years.12 Furthermore, measures that quantify the broader context of a patient's life are likely to be influenced by events occurring throughout the patient's life, and it is not yet clear how changes in these measures should be interpreted over long periods. This is the problem of “response shift,” discussed in the first paper in this series and which will be revisited in a later paper about who should measure quality of life.13 14
A small but growing number of instruments used to measure quality of life specifically in clinical practice are available. The disease repercussion profile15 assesses a patient's perception of handicap in rheumatoid arthritis, osteoarthritis, and back pain. Other examples include the support team assessment schedule,16 the Edmonton symptom assessment scale,17 and the palliative outcome scale,18 all of which were developed specifically for palliative care, and the measure yourself medical outcome profile (MYMOP),19 designed for use in primary care. Many of these instruments are known to be reliable and valid but trials are required to evaluate their routine use in clinical practice.
Scores from quality of life measures in studies are often presented as means. While this is useful in testing one treatment against another in groups of patients, it is of less value in clinical practice. At what point is a problem considered severe? Is it when the score is above the mean? Or when scores are in the top quarter? The decision is clinical. Some screening scales have cut-off points for clinical intervention (for example, depression scales), but for others what is more important is whether the problem is rated as severe.18 Reducing the number of patients with severe pain was considered to be the clinically important aspect of the support team assessment scale.20
Introducing quality of life measures into clinical practice often means that staff need to change their practice. Change can be threatening, especially if staff believe that they may be judged adversely. The organisation's culture becomes important.21 Staff will need training in using and interpreting the measure, as they would for any new assessment tool. To be of most value quality of life measures should be incorporated into the clinical record and the results discussed at clinical review meetings.22 Suggested steps in choosing a quality of life measure and introducing it into clinical practice are shown in the boxes.
The potential benefit to patients of using these measures in clinical practice is that their problems are identified and dealt with and that treatment decisions are based on their priorities and preferences. Evidence for these benefits is lacking because these measures are rarely used in clinical practice. In rheumatology, where quality of life has been an important outcome in clinical trials for 15 to 20 years, surveys in the United Kingdom suggest that little use is made of these measures in clinical practice.23 Moreover, there is some suggestion that even when quality of life measures are used they do not influence clinical decision making. Analyses of clinical judgment have highlighted discrepancies between the ways some clinicians think they make decisions and the way they actually do.24 The effect of information from these measures on clinical decision making seems small,25 but these data were collected before the introduction of high profile quality of life measures (such as the medical outcomes survey short form 36 (SF-36)).
One way of ensuring that quality of life assessments influence clinical decision making is to use them as a basis for making choices about treatment. This can be effected by using measures to identify individual problems and priorities for treatment and then negotiating treatment goals based on them. An evaluation of the role of these measures in setting clinical goals in patients with rheumatoid arthritis has just been completed (RA Hughes et al, personal communication, 2000).
There is a lack of evidence showing that findings from audit or similar initiatives have resulted in a change in practice. Realising that a problem exists is not enough to indicate what exactly needs to be changed in a clinical service.26 Evaluation of audits in one health region in the United Kingdom identified changes in clinical services in the development and implementation of new standards of care; improvements in documentation; and specific changes in clinical practice, such as prescribing, managing accidents, and seeking information by health professionals.27 However, there was no analysis of whether patients' outcomes changed, and further work is needed to assess this. Using quality of life measures, such as quality adjusted life years (QALYs), to determine the relative value of different services or interventions is difficult because of the “disability paradox.” People with severe or even life threatening disease may not rate their quality of life as significantly poorer than people with mild disease or people who are healthy. This makes it difficult to directly compare groups of patients with different diseases in order to allocate resources. The implications of the paradox in measuring quality of life are discussed in more detail in several papers in this series.
Many of the practical problems associated with measuring quality of life in clinical practice may be overcome by the use of new technologies. Computerised approaches to data storage and retrieval will simplify the collection, storage, and monitoring of data. The ability to administer measures over the internet, using touch screen or palm top computers, will overcome some of the problems of administering and scoring them. Data will be able to be automatically downloaded to the records of individual patients and then reviewed in the context of their treatment and clinical outcomes.
The increasing interest in developing individualised measures reflects the perception that quality of life is unique to individuals and cannot be adequately assessed using standardised measures that ask every patient the same questions and require responses to be selected from a predetermined set. The extent to which existing measures capture the quality of life of individual patients is discussed in the next paper in this series.7 Individualised measures such as the schedule for the evaluation of individualised quality of life (SEIQoL)28 and the patient generated index29 ask the same questions of all patients but allow them to specify their own responses.
The use of individualised measures in research has been limited by difficulties in administering and scoring them, but in clinical practice they have immediate relevance. They are designed to detect individuals' problems and as such are more readily interpreted in ways that are clinically meaningful. They also provide a basis for sharing clinical decision making between patients and clinicians, identifying patients' priorities for treatment, and facilitating the setting of realistic treatment goals. There may be opportunities to combine these approaches with short, standardised measures that include screening questions. Further evaluation of the performance of individualised measures in clinical practice is required. This should be developed in parallel with statistical methods that allow data from individualised measures to be analysed.
A number of questions about the clinical utility of quality of life measures remain unanswered. Firstly, are existing measures appropriate and adequate for clinical practice or are new measures required? Answering this will involve evaluating existing measures in clinical settings with appropriate psychometric assessment of their performance in individual patients over time. This raises the second question, which relates to the definition and assessment of changes in the quality of life of individual patients: how do existing measures take account of changes in expectations, adaptation, and normalisation when assessing changes? These issues were highlighted in the first paper in this series.13 The third question concerns the clinical interpretation of these measures: what constitutes an important change in quality of life (and to whom is the change important)? Answering these questions will enable existing measures to be calibrated with respect to thresholds for intervention. They will thus communicate better information to patients and their families about the likely benefits of treatment.
We thank our colleagues in the Interdisciplinary Research Group in Palliative and Person Centred Care at King's College London, in particular Peter Robinson, Barry Gibson, Stanley Gelbier, Robert Dunlop, Julia Addington-Hall, Lalit Kalra, and Alan Turner-Smith, who participated in discussions and commented on an earlier draft of this work.
This is the second in a series of five articles
Series editors: A J Carr, I J Higginson, P G Robinson
Funding: AJC's post is funded by the Arthritis Research Campaign.
Competing interests: None declared.
Additional information about using measures for clinical improvement and properties needed for use in clinical practice can be found on the BMJ's website