Surrogate outcome measurement

Until relatively recently, clinician-based or laboratory measures (e.g. foot alignment or plantar pressure), referred to as surrogate outcomes (Guyatt & Rennie 2002), were often used to evaluate outcomes of treatment. Unfortunately, patients do not generally consider these surrogate measures important for understanding their health. Surrogate outcomes place little or no emphasis on an appraisal from the person being treated (Jenkinson & McGee 1998), whose only concern might be that they want their pain to reduce. Consequently, decisions regarding the effect of an intervention based on surrogate outcomes rely on the health professional’s judgement, rather than on the patient’s appraisal of their own health (Friedman et al 1998, Staquet et al 1999). Reasons for using surrogate outcomes include:

To illustrate this issue further, a common example of a surrogate outcome from medicine is that of measuring tumour size as the primary outcome of success of a cancer treatment (Gebski et al 2002). Of course there would be a high likelihood that tumour size and progression of the disease are highly correlated. However, it is not necessarily the case that a decrease in tumour size correlates to a good therapeutic outcome from the patient’s perspective. Indeed, adverse effects of treatment can have a negative impact on the individual’s quality of life, both in the short and long term. This clinician-based focus is an historical problem in medicine and medical research where patient input had customarily been considered too subjective.

Bias towards clinician-based measures also exists in professions that care for foot and foot-related disorders, for example measuring arch height, frontal plane calcaneal position or plantar pressure distribution. Clearly, these surrogate outcomes have little meaning to the average patient who wants more relevant feedback about whether a prescribed treatment is working. Another key issue also needs to be considered when evaluating surrogate outcomes, particularly in relation to foot-related disorders. Because little is known about the actual relationship between these surrogate measures and a person’s health status, it is unclear whether a change in, for example, foot alignment, correlates to an improvement in the patient’s health status. Often, good-quality evidence that provides certainty about these links is not available. With this in mind, clinicians and researchers need to be careful when interpreting changes in these surrogate outcomes as being meaningful to the patient.

Due to the limitations of surrogate outcome measures, which tend to focus on disease, health outcome assessment has developed rapidly over the past two decades. Not surprisingly, health outcome assessment focuses more on the concept of what is important to the patient. The World Health Organization’s definition of ‘health’ is that which is ‘not merely the absence of disease or infirmity, but a state of complete physical, mental, and social well-being’ (World Health Organization 1958). Clearly, this definition of health is more holistic and consumer-centred than clinician-based measures (Bowling 2005). However, in their early development, health outcome measures were created by clinicians, incorporating issues that they thought were important to patients. Whether these clinician-generated health outcome measures were, indeed, important to patients is an area of great conjecture. In the light of this, health status measurement has developed to include a broader account of ‘health’, which encompasses the patient’s (or consumer’s) perspective of their health (Jenkinson & McGee 1998). Put more simply, health status (and health-related quality of life) measurement attempts to evaluate what is important to patients, ideally from the patient’s perspective.

Health status and quality-of-life measurement

With the aforementioned concerns about surrogate outcomes in mind, health status measurement has taken on increasing importance. Health status incorporates the notion of health-related quality of life, which is an important, and some would argue, fundamental measure of the impact of a disease or abnormality on an individual (Testa & Simonson 1996). Health-related quality of life characterises and measures what consumers experience as a result of healthcare (Staquet et al 1999). Emphasis is taken away from the disease process, while highlighting the broader appraisal of an individual’s health by that individual (Muldoon et al 1998). Measurement of health-related quality of life is usually by self-report (i.e. self-reported questionnaire), although in paediatrics it can be acceptable to use a proxy (e.g. a parent/guardian). Health-related quality of life includes such areas as somatic sensation (e.g. pain), physical function, cognitive function and psychological well-being, social function and life satisfaction (Jenkinson & McGee 1998, Bowling 2005). A summary of health outcome terms used thus far in this chapter can be found in Table 4.1.

Table 4.1 Terms used in health outcome assessment

Measurement of health status and quality of life via self-reported questionnaire is a complex area that is developing rapidly (Scientific Advisory Committee of the Medical Outcomes Trust 2002). These questionnaires or instruments pose relevant questions relating to a person’s quality of life. Each question or item is then scored individually or added together to produce a summary score. Scores can also be weighted to place more importance on certain sections of the instrument (Streiner & Norman 1995). In addition, many questionnaires group items together into like areas, known as domains or subscales (e.g. ‘pain’, ‘function’, ‘vitality’, etc.). Domains can also be scored individually or combined to give an overall numerical score of health and well-being, although this feature is not available for all instruments (McDowell 2006). Questionnaire development is a science of its own and rigorous methods should be employed to develop an instrument that will measure appropriately and accurately. Questions and their grouping into domains need to be developed using sound measurement principles, including item inclusion, psychometrics, validity and reliability testing (Streiner & Norman 1995). Appropriate development will ensure that instruments measure what they are supposed to measure and that they do so in a reproducible manner (Muldoon et al 1998).

Health status measures can be classified under two broad categories: (i) generic measures, which assess universal aspects of general health and well-being; and (ii) specific measures, which assess a specific medical condition or region of the body (Patrick & Deyo 1989). It is generally accepted that when measuring health status, both a generic and a specific measuring instrument should be used, as they measure distinct but complementary aspects of patients’ health (Kantz et al 1992, Bombardier et al 1995, Hawker et al 1995). Generic instruments allow some commonality of measurement, and as such, comparison between different conditions (Patrick & Deyo 1989). However, due to their non-specific nature, they are generally less responsive to change compared with specific instruments (Patrick & Deyo 1989). Table 4.2 presents a summary of the categories discussed above.

Table 4.2 Categories of health status measures

Before discussing specific examples of health outcome measures in more detail, a brief look at the important issues of validity and reliability is needed.

Visual analogue scale

Often clinicians and researchers want to assess a single (uni-dimensional) construct, such as pain, in as simple a manner as possible. One such method that is well accepted is the visual analogue scale (VAS). The VAS is a 100 mm straight line, which has at each end labels that define the extreme limits (i.e. range) of the sensation or response being measured (McDowell 2006). The line is most frequently presented horizontally on a page, but it can be presented vertically (Scott & Huskisson 1979b).

Although primarily used to measure pain, the VAS can be used to measure other constructs as well (e.g. anxiety, satisfaction, comfort). When used for the measurement of pain, the extreme left end of the scale (i.e. 0 mm) is labelled ‘no pain’ and the right end of the scale (i.e. 100 mm) is labelled ‘pain as bad as it could be’ (Huskisson 1974, Scott & Huskisson 1976). Alternatively, the labels ‘worse pain imaginable’ or ‘agonising pain’ can be used at the extreme right end of the scale (McDowell 2006). Figure 4.1 shows a typical example of a VAS formatted for the measurement of pain.

Figure 4.1 An example of a VAS used for measuring pain.

The patient rates the degree of pain they are experiencing for the condition in question by marking a point on the 100 mm line that corresponds to their level of pain relative to the two extremes. The distance from the 0 mm point to the mark that the patient draws is then measured to determine the level of pain. VAS are relatively simple for clinicians to explain to patients and are easy for patients to use. In addition, they can be easily used by patients at home (i.e. away from the clinical setting) if ongoing assessment is required.

The VAS has been validated as a measure for the following types of pain: experimentally induced pain; pain intensity; unpleasantness of pain; and chronic clinical pain (Duncan et al 1989, Price et al 1983, Price & Harkins 1987). When used to measure pain, the VAS has also been shown to be reliable, with test–retest correlations ranging from 0.95 to 0.99 (Revill et al 1976). With respect to serial assessment of pain over longer periods of time (e.g. every month for a year), one study found that it is acceptable to show patients their initial scores (Scott & Huskisson 1979a). Caution is needed with non-literate patients where reliability, while still acceptable, is not as good as with literate patients (Ferraz et al 1990).

Clearly, the VAS has been extensively validated and been shown to be a reliable tool to measure pain, but how are the results obtained interpreted? For example, what scores on a VAS relate to mild, moderate or severe pain? One study pooled the findings of 11 randomised trials (1080 patients) that evaluated the effects of analgesia using VAS (Collins et al 1997). The results showed that patients who scored their pain as ‘moderate’ had a mean of 49 mm, with 85% of them scoring over 30 mm. Those reporting ‘severe’ pain had a mean of 75 mm, and 85% scored over 54 mm.

In addition, what amount of change in pain can be viewed as positive for the patient? Although this is a highly complex and contentious area, there has been some research on this issue. For example, one study found that a 33% decrease in postoperative pain represented a reasonable standard for determining a meaningful change from the patient’s perspective (Jensen et al 2003). Another study established that patients experiencing pain postoperatively of less than 40 mm on a VAS were ‘adequately’ satisfied, and the complete elimination of pain was not required for good patient satisfaction (Jensen et al 2005).

The issue of meaningful change has received some attention of late, with researchers attempting to determine what change in pain level is important to patients. More broadly, the amount of change required on any outcome measure – not just pain – to be considered important to a patient has been recently referred to as the ‘minimal important difference’ (Schunemann & Guyatt 2005), although it has also been referred to by many other terms including: minimal clinically important difference, clinically significant difference and meaningful change. Unfortunately, there are few examples of this type of research (i.e. calculation of the minimal important difference) in the foot and ankle literature. The literature on emergency medicine indicates the minimal important difference for the VAS ranges between 9 mm and 13 mm (Kelly 1998, Kelly 2001). That is, a change in pain of 9–13 mm on a VAS is, on average, considered important by patients. The minimal important difference is a valuable piece of information, not only when interpreting findings from instruments such as the VAS, but also for clinical trials, in which it can be used to assist in prospective sample size calculations.

Although the VAS is one of the most frequently used instruments to measure constructs such as pain, there are other methods such as the numerical rating scale, the verbal rating scale and the Oucher scale. The numerical rating scale is a simple mechanical device where the patient moves a slider and the numerical rating is read from the back of the device (McDowell 2006). Alternatively, the verbal rating scale can be used where the patient indicates their level of pain, for example, using the numbers 0–10. This simple and clinically useful method correlates well with the VAS (Murphy et al 1988). It has also been shown to have adequate validity and reliability (Jensen et al 1999). The Oucher scale is a pain scale that is specifically used for children between the ages of 3 and 12 years (Knott et al 1994). To indicate their level of pain, children can choose a face from a series of faces with varying expressions of pain.

Finally, more complex, multi-dimensional measures of pain have also been developed, and the McGill Pain Questionnaire (Melzack 1975) is one of the most well known. In contrast with the VAS, the McGill Pain Questionnaire measures beyond just pain intensity; however, in its long form it is significantly more time consuming to use (Kahl & Cleland 2005). Moreover, the newer health status measures also generally have a pain component embedded in them, making more complex pain measurements, such as the McGill Pain Questionnaire, somewhat redundant. If a specific component of pain is required to be measured in addition to the more generalised pain assessed in the newer health status measures, a simple VAS would probably be sufficient. For example, in a randomised trial evaluating low-Dye taping for plantar heel pain (Radford et al 2006), ‘first step’ pain – the pain experienced in the heel on first stepping out of bed in the morning – was measured using a VAS, as well as more generalised foot pain using a foot specific health status measure (the Foot Health Status Questionnaire, covered below).

In summary, the VAS is a highly useful and easy to use instrument to measure constructs such as pain. It has demonstrated good validity and reliability, and is a mainstay in clinical research where an assessment of pain is required. An easier method, the verbal rating scale is a satisfactory alternative in clinical practice. However, the VAS has some limitations, including its unidimensionality and the fact that it does not evaluate broader quality-of-life issues.

Clinician-based outcome measures

Related posts:

The presenting problem Assessment of the surgical patient Laboratory tests The painful foot Neurological assessment Diagnostic imaging

Stay updated, free articles. Join our Telegram channel

Join