## Abstract

Background: Fatigue in cancer is very common and can be experienced at all stages of disease and in survivors. There is no accepted definition of cancer-related fatigue (CRF) and no agreement on how it should be measured. A number of scales have been developed to quantify the phenomenon of CRF. These vary in the quality of psychometric properties, ease of administration, dimensions of CRF covered and extent of use in studies of cancer patients. This review seeks to identify the available tools for measuring CRF and to make recommendations for ongoing research into CRF.

Methods: A systematic review methodology was used to identify scales that have been validated to measure CRF. The inclusion criteria required the scale to have been validated for use in cancer patients and/or widely used in this population. Scales also had to meet a minimum quality score for inclusion.

Results: The reviewers identified 14 scales that met the inclusion criteria. The most commonly used scales and best validated were the Functional Assessment of Cancer Therapy Fatigue (FACT F), the European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire (EORTC QLQ C30) (fatigue subscale) and the Fatigue Questionnaire (FQ).

Conclusions: Unidimensional scales are the easiest to administer and have been most widely used. The authors recommend the use of the EORTC QLQ C30 fatigue subscale or the FACT F. The FQ gives a multidimensional assessment and has also been widely used. A substantial minority of the scales identified have not been used extensively or sufficiently validated in cancer patients and cannot be recommended for routine use without further validation.

## introduction

Cancer-related fatigue (CRF) is a subjective symptom experienced by patients at all stages of disease. It can occur during treatment [1], in advanced disease [2] and in disease-free survivors [3]. However, the prevalence of fatigue can vary widely depending on which measurement tool is used [4]. This is in part a reflection of the lack of an agreed definition of CRF. Diagnostic criteria for a syndrome of CRF have been proposed for inclusion in the International Classification for Diseases (ICD 10) (Cella ref). These criteria can be applied using a short semistructured interview [5]. The widespread adoption of a syndrome approach using agreed diagnostic criteria would help with the interpretation of prevalence figures in different populations and across different studies. However, the syndrome approach has not yet been widely used, probably because a strict application of the criteria requires the use of a semistructured psychiatric interview to ensure that symptoms of CRF are not related to underlying psychiatric co-morbidity.

The European Association of Palliative Care has developed a working definition [6] of CRF on the basis of the European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire (EORTC QLQ C30). This defines fatigue as a subjective feeling of tiredness, weakness or lack of energy. This is a pragmatic approach which provides a working understanding of CRF for clinicians. However, other organisations and authorities have defined CRF in their own ways, e.g. the National Comprehensive Cancer Network [7], and there is no universally accepted definition.

The lack of consensus in this area has led to the development of a number of scales to measure CRF. These scales have usually been validated originally in cancer patients. Other investigators have used scales that were originally validated in non-cancer populations and have then independently validated their use in patients with cancer. These scales can differ widely in the number of items that they contain, the dimensions of CRF that they cover (e.g. physical, affective and cognitive) and their psychometric properties.

In the light of this multiplicity of different assessment methods, we undertook a review to identify which scales have been best validated and to make recommendations about which instruments should be used in research and/or in routine clinical practice.

## methods

A systematic methodology was used. The following terms were used to search Medline, CINAHL and Psychinfo (1950—week 1 February 2008):

1. exp “Outcome Assessment (Health Care)”/

2. exp Psychometrics/

3. exp “Outcome and Process Assessment (Health Care)”/

4. (fatigue adj (scale or inventory or instrument or measurement or assessment)).mp. [mp = title, original title, abstract, name of substance word, subject heading word]

5. (fatigue adj2 (scale or inventory or instrument or measurement or assessment)).mp.

6. (fatigue adj3 (scale or inventory or instrument or measurement or assessment)).mp.

7. (fatigue adj4 (scale or inventory or instrument or measurement or assessment)).mp.

8. exp Questionnaires/

9. or/1-8

10. exp Fatigue/

11. fatigue (ti,ab,kw)

12. ((lack$or loss or lost) adj2 (energy or vigour or vigor)).mp. 13. (tired$ or weary or weariness or exhaustion or exhausted or lacklustre or astheni$).mp. 14. (apathy or apathetic or lassitude or letharg$ or (feeling adj3 (drained or sleepy or sluggish or weak4))).mp.

15. or/10-14

16. 9 and 15

The titles and abstracts of the papers identified using this search strategy were screened by one of the authors (O.M.) and where necessary the full text articles were retrieved. The reference lists of included articles were also examined. O.M. was responsible for the search strategy and retrieving studies. An ad hoc quality score was created before searching to ensure that included scales reached a minimum level of psychometric properties on the basis of the ideal characteristics of measurement scales described in detail by Norman and Streiner [6, 8]. These characteristics are detailed in Table 1. The final list of included studies was agreed by both authors.

Table 1.

Explanation of psychometric properties of an ideal scale (on the basis of Norman & Streiner [6, 7])

 Psychometric evaluation Internal consistency Do items within the scale correlate with each other and with the total score? This is usually measured by Cronbach's alpha. A low value indicates a poor consistency of items; a very high score indicates item redundancy. Test–retest reliability Does the scale provide a similar score when administered to the same population in the same setting at different times? Discriminant validity Does the scale distinguish between groups expected to have differing levels of fatigue, e.g. cancer patients and a control population. Responsiveness to change Does the scale measure change during the course of an intervention that alters the level of fatigue, e.g. during chemotherapy. Convergent validity Does the scale provide a similar measure to previously validated fatigue tools? This is usually measured by the degree of correlation of the scores. The value usually given is Pearson's r correlation.
 Psychometric evaluation Internal consistency Do items within the scale correlate with each other and with the total score? This is usually measured by Cronbach's alpha. A low value indicates a poor consistency of items; a very high score indicates item redundancy. Test–retest reliability Does the scale provide a similar score when administered to the same population in the same setting at different times? Discriminant validity Does the scale distinguish between groups expected to have differing levels of fatigue, e.g. cancer patients and a control population. Responsiveness to change Does the scale measure change during the course of an intervention that alters the level of fatigue, e.g. during chemotherapy. Convergent validity Does the scale provide a similar measure to previously validated fatigue tools? This is usually measured by the degree of correlation of the scores. The value usually given is Pearson's r correlation.

Scales were included only if they met the following a priori criteria:

• 1) Self-assessment scales originally validated in cancer patients and/or subsequently widely used in cancer populations (n > 50 patients).

• 2) Scales must have been used in a second test population for independent validation of their use in cancer patients.

• 3) 90% of participants must be >18 years.

• 4) Scales obtained a minimum quality score. The paper describing the scale must have contained information on at least three of the following: internal consistency, test–retest reliability, known group validity (discriminant validity), responsiveness to change or convergent validity (against other scales).

• 5) Scales must be in English or translated and validated for English language use.

There were also explicit exclusion criteria:

• 1) Objective rating scales testing power/strength, etc. as opposed to subjective fatigue.

• 2) Single-item scales, including visual analogue scale (VAS)

• 3) Fatigue subscales as part of a broader quality-of-life measure. Unless specific data were available relating to the psychometric properties of the subscale.

In addition, the original reference was cross-referenced for citating articles via Medline and Web of Knowledge. This was carried out to quantify the number of times the scales had been used and the type of populations studied. This procedure allowed us to approximately quantify the number of participants who had been tested with each scale. For ease of reference, populations of cancer patients were divided into three groups:

• 1) Active—on treatment with curative intent.

• 2) Palliative—on treatment with palliative intent or supportive care only.

• 3) Disease free—medium to long-term survivors.

This methodology allowed us to make firmer recommendations for ongoing research into CRF by being able to compare different levels of scale usage as well as the scales’ psychometric properties.

## results

The search strategy identified 7889 abstracts in total. These were screened and 116 studies were identified where the main focus of the paper was on fatigue measurement. This shortlist generated data on 22 different scales that have been used in the assessment of CRF (see Figure 1). Eight of these scales did not meet our minimum quality requirements. The specific reasons for excluding these instruments are itemised below.

Figure 1.

Flow chart of study identification and selection.

Figure 1.

Flow chart of study identification and selection.

The Fatigue Symptom Control Checklist [9]: Although used in some studies of CRF (e.g. by Morrow et al. [10]), this scale has neither been validated in English nor been specifically validated for use in cancer patients.

The Swedish Occupational Fatigue Inventory [11]: This scale has been used in one study of patients undergoing radiotherapy [12] but was found to have limitations and has not been further evaluated.

The original Piper Fatigue Scale (PFS) [13]: This scale was not included as it was found to have a number of flaws in its use. It also required the use of an initial screening tool to identify patients with fatigue. The revised version of the scale was included (see below).

The Cancer Fatigue Scale [14] and the Fatigue Assessment Questionnaire [15] were both excluded as validated English language versions have not been published.

The Fatigue Management Barriers Questionnaire [16], the Clinical Survey for CRF (QFAS) [17] and the Cancer-Related Fatigue Distress scale [18] were all excluded because although they have undergone pilot evaluation in cancer patients no further published reports regarding their validity could be identified.

This left 14 scales which met our criteria and have been included in the review. The scales have been divided into unidimensional and multidimensional for ease of presentation. The unidimensional scales invariably cover the physical aspects of fatigue only. The multidimensional scales cover anywhere between two and five different aspects of fatigue. This division is possibly artificial as CRF has been shown in a recent study to be a unidimensional construct, even after measuring all relevant dimensions of fatigue [19]. However, this method of categorising measurement scales is still widely used and has been adopted in this review in order to structure the presentation and discussion of the different instruments.

### unidimensional scales

We found five scales that reached our minimum quality standard. Full details of their psychometric properties and areas of use are detailed in Table 2.

Table 2.

Unidimensional scales

 Scale name BFI [18] EORTC QLQ (FS) [19] FSS [20] FACT F [21] POMS F [22] Number of items 9 3 9 13 7 Scale type Numerical Likert Numerical Numerical Likert Population studied in original validation study Mixed cancer Three populations: lung cancer, bone marrow transplant, metastatic cancer Chronic illness population Mixed cancer patients on treatment Mixed work population and psychiatric patients Internal consistency 0.96 0.80–0.85 0.96 0.95 0.90–0.94 Dimensions fatigue covered Physical functioning Physical functioning Physical functioning Physical functioning Physical functioning Test & retest reliability – – Subset 87 patients, 3/52 later measurement, error 4.7 units r = 0.90 (3–7 days later) r = 0.66 (12–16 weeks later) Known group comparison (discriminant validity) Distinguished CRF from community adults Distinguished between local and regional disease Distinguished CRF versus healthy controls Distinguished between three known haemoglobin levels (>13, 11–13 and <11) Distinguished between chronic fatigue syndrome and controls (N = 6275) Responsiveness to change Not done On the basis of performance status deterioration Not done Minimum clinical important difference (MCID) derived (3 points) MCID derived (5.6 points) Convergent validity (against other scales) POMS F, r = 0.84 Fatigue Questionnaire, r = 0.49–0.75 EORTC, r = 0.83 Piper Fatigue Scale, r = −0.75; POMS F, r = 0.74 Not done Number of participants tested in original validation study 305 366 227 50 695 Populations studied (active/palliative/survivors) A/P/S A/P/S A/P A/P/S A/P/S Approximate number of further cancer studies and participantsa 4 (∼2300) >10 (>5000) 1 (100) >50 (>5000) >20 (>2000)
 Scale name BFI [18] EORTC QLQ (FS) [19] FSS [20] FACT F [21] POMS F [22] Number of items 9 3 9 13 7 Scale type Numerical Likert Numerical Numerical Likert Population studied in original validation study Mixed cancer Three populations: lung cancer, bone marrow transplant, metastatic cancer Chronic illness population Mixed cancer patients on treatment Mixed work population and psychiatric patients Internal consistency 0.96 0.80–0.85 0.96 0.95 0.90–0.94 Dimensions fatigue covered Physical functioning Physical functioning Physical functioning Physical functioning Physical functioning Test & retest reliability – – Subset 87 patients, 3/52 later measurement, error 4.7 units r = 0.90 (3–7 days later) r = 0.66 (12–16 weeks later) Known group comparison (discriminant validity) Distinguished CRF from community adults Distinguished between local and regional disease Distinguished CRF versus healthy controls Distinguished between three known haemoglobin levels (>13, 11–13 and <11) Distinguished between chronic fatigue syndrome and controls (N = 6275) Responsiveness to change Not done On the basis of performance status deterioration Not done Minimum clinical important difference (MCID) derived (3 points) MCID derived (5.6 points) Convergent validity (against other scales) POMS F, r = 0.84 Fatigue Questionnaire, r = 0.49–0.75 EORTC, r = 0.83 Piper Fatigue Scale, r = −0.75; POMS F, r = 0.74 Not done Number of participants tested in original validation study 305 366 227 50 695 Populations studied (active/palliative/survivors) A/P/S A/P/S A/P A/P/S A/P/S Approximate number of further cancer studies and participantsa 4 (∼2300) >10 (>5000) 1 (100) >50 (>5000) >20 (>2000)

Where internal consistency—Cronbach's alpha unless otherwise stated; convergent validity—Pearson's correlation value unless otherwise stated. A, active treatment population; P, palliative care population; S, disease-free survivors; CRF, cancer-related fatigue. Scale abbreviations: BFI, Brief Fatigue Inventory; EORTC QLQ C30 FS, European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Fatigue subscale; FSS, Fatigue Severity Scale; FACT F, Functional Assessment of Cancer Therapy Fatigue subscale; POMS F, Profile of Mood States Fatigue subscale.

a

This involved cross-referencing studies in Medline and Web of Science and identifying studies in cancer where the scale has been used subsequently.

The Brief Fatigue Inventory (BFI) [23] is a nine-item VAS that was validated for use in a mixed cancer population. It has reasonable psychometric properties but has had limited ongoing use. The scale has cut-off scores to differentiate between mild, medium and severe fatigue but these have not been validated and are likely to be of use for screening purposes only.

The EORTC QLQ C30 [24] is a 30-item quality-of-life questionnaire. The full tool has been used extensively as a quality-of-life instrument. The three-item fatigue subscale has been independently validated as a separate fatigue measure. While the psychometric properties are weaker than more extensive scales, their brevity and ease of administration may outweigh this disadvantage. There have also been two large-scale studies (>2000 patients in each) independently assessing its use [20, 21], so there are extensive data in a variety of settings. However, it has been noted to have a ceiling effect in advanced cancer patients [20] and is not recommended as a single measure in this group.

Another advantage of this measure is that it is possible to extract fatigue data from studies that have used the full quality-of-life scale. This includes the large number of studies that have used this scale in cancer chemotherapy trials (note: these studies have not been included in the current analysis).

The Fatigue Severity Scale [25] is a nine-item scale that was originally validated in a chronic illness (non-cancer) population and while it has been extensively used in neurological disease and chronic fatigue, it has had very limited use in cancer patients. It has only been used and validated by one author in a few related studies [1, 26]. However, no other studies were identified and it is not recommended for use in the measurement of CRF.

The Functional Assessment of Cancer Therapy Fatigue (FACT F) subscale [27] is a 13-item stand-alone scale but is one of a range of the FACIT series of quality-of-life and tumour-specific symptom questionnaires [22]. It has been used in a number of intervention studies to treat CRF, and the original authors have been able to derive change in scale scores that correspond to minimum clinically significant differences [28]. This application makes it an especially useful measure for intervention studies.

The Profile of Mood States [29] was originally used as a measure of workforce health. It contains a number of scales including a fatigue subscale of seven items which has been independently examined in both a non-cancer [30] and a cancer population [31]. It has also been used to provide convergent validity in the validation of a number of other scales used in CRF. It has a defined minimum clinically significant difference [32]. Its extensive use in studies has meant there are ample data available to recommend its continued use in this setting.

### multidimensional scales

There were nine scales included in this category. Full details of these scales can be found in Table 3.

Table 3.

Multidimensional scales

 Scale name FQ [30] FSI [33] VASF [34] MAF [35] Number of items 11 13 18 16 Scale type Likert Numerical Visual analogue scale Numerical Original validation sample General practice Breast cancer on treatment Patients with sleep disorders Arthritis patients Internal consistency 0.88–0.90 0.94 0.91–0.96 0.93 Dimensions fatigue covered Physical and mental Physical and mental Physical and mental Distress, interference, severity, cognitive Test & retest reliability Not done Range of values: patients, r = 0.35–0.75; controls, r = 0.10–0.74 Not done Not done Known group comparison (discriminant validity) Distinguished between Hodgkin's disease and general population Distinguished between healthy controls and breast cancer patients Distinguished between patients with sleep disorders and healthy controls Distinguished between patients with arthritis and healthy controls Responsiveness to change Not done Distinguished between pre- and post-treatment fatigue levels Limited ability to detect changes over treatment period Not done Convergent validity (against other scales) Fatigue item on clinical interview schedule POMS F, r = 0.75 POMS F, r = 0.58–0.78 POMS F, r = 0.78 Number of participants tested in original sample 374 270 122 50 Populations studied (active/palliative/survivors) A/P/S A/S A only A only Approximate number of further cancer studies and participants >10 (>2000) 7 (∼700) 3 (∼250) 3 (∼760) Scale name MFI-20 [36] MFSI-30 [37] PFS (revised) [38] Schwartz CFS [39] Wu CFS [40] Number of items 20 30 22 28 9 Scale type Likert Likert Likert Likert Likert Original validation sample Mixed cancer population Breast cancer population Breast cancer population Mixed cancer population Breast cancer population Internal consistency 0.84 0.87–0.96 0.97 0.96 0.91 Dimensions fatigue covered Cognitive, physical and emotional Cognitive, physical and mental Temporal, intensity, cognitive, affective and sensory Physical, emotional and cognitive Physical and emotional Test & retest reliability Not done Three occasions, r > 0.50 r = 0.98 Not done Not done Known group comparison (discriminant validity) Distinguished between cancer patients, healthy controls and doctors on shift work Distinguished between cancer patients and healthy controls Significant differences between lung and breast cancer patients Significant differences in fatigue pre- and post-first chemotherapy cycle None Responsiveness to change Monitored change over course of radiotherapy Not done Not done MCID calculated on treatment population (5 points) Not done Convergent validity (against other scales) VASF, r = 0.77 Short form 36, r = 0.55; FSI, r = 0.74 MFI subscale r = 0.84 POMS F—not quantified Schwartz CFS, r = 0.84 Number of participants tested in original validation 111 cancer patients (1200 controls) 275 382 166 82 Populations studied (active/palliative/survivors) A/P/S A/S (breast only) A/S (breast only) A only A (breast only) Approximate number of further cancer studies and participantsa >20 (∼2000) 3 (∼380) 9 (∼800) 2 (∼300) 1 (∼170)
 Scale name FQ [30] FSI [33] VASF [34] MAF [35] Number of items 11 13 18 16 Scale type Likert Numerical Visual analogue scale Numerical Original validation sample General practice Breast cancer on treatment Patients with sleep disorders Arthritis patients Internal consistency 0.88–0.90 0.94 0.91–0.96 0.93 Dimensions fatigue covered Physical and mental Physical and mental Physical and mental Distress, interference, severity, cognitive Test & retest reliability Not done Range of values: patients, r = 0.35–0.75; controls, r = 0.10–0.74 Not done Not done Known group comparison (discriminant validity) Distinguished between Hodgkin's disease and general population Distinguished between healthy controls and breast cancer patients Distinguished between patients with sleep disorders and healthy controls Distinguished between patients with arthritis and healthy controls Responsiveness to change Not done Distinguished between pre- and post-treatment fatigue levels Limited ability to detect changes over treatment period Not done Convergent validity (against other scales) Fatigue item on clinical interview schedule POMS F, r = 0.75 POMS F, r = 0.58–0.78 POMS F, r = 0.78 Number of participants tested in original sample 374 270 122 50 Populations studied (active/palliative/survivors) A/P/S A/S A only A only Approximate number of further cancer studies and participants >10 (>2000) 7 (∼700) 3 (∼250) 3 (∼760) Scale name MFI-20 [36] MFSI-30 [37] PFS (revised) [38] Schwartz CFS [39] Wu CFS [40] Number of items 20 30 22 28 9 Scale type Likert Likert Likert Likert Likert Original validation sample Mixed cancer population Breast cancer population Breast cancer population Mixed cancer population Breast cancer population Internal consistency 0.84 0.87–0.96 0.97 0.96 0.91 Dimensions fatigue covered Cognitive, physical and emotional Cognitive, physical and mental Temporal, intensity, cognitive, affective and sensory Physical, emotional and cognitive Physical and emotional Test & retest reliability Not done Three occasions, r > 0.50 r = 0.98 Not done Not done Known group comparison (discriminant validity) Distinguished between cancer patients, healthy controls and doctors on shift work Distinguished between cancer patients and healthy controls Significant differences between lung and breast cancer patients Significant differences in fatigue pre- and post-first chemotherapy cycle None Responsiveness to change Monitored change over course of radiotherapy Not done Not done MCID calculated on treatment population (5 points) Not done Convergent validity (against other scales) VASF, r = 0.77 Short form 36, r = 0.55; FSI, r = 0.74 MFI subscale r = 0.84 POMS F—not quantified Schwartz CFS, r = 0.84 Number of participants tested in original validation 111 cancer patients (1200 controls) 275 382 166 82 Populations studied (active/palliative/survivors) A/P/S A/S (breast only) A/S (breast only) A only A (breast only) Approximate number of further cancer studies and participantsa >20 (∼2000) 3 (∼380) 9 (∼800) 2 (∼300) 1 (∼170)

Where internal consistency—Cronbach's alpha unless otherwise stated; convergent validity—Pearson's correlation value unless otherwise stated. MCID, minimum clinically important difference; POMS F, Profile of Mood States Fatigue subscale; A, active treatment population; P, palliative care population; S, disease-free survivors. Scale abbreviations: FQ, Fatigue Questionnaire; FSI, Fatigue Severity Inventory; VASF, Visual Analogue Scale for Fatigue; MAF, Multidimensional Assessment of Fatigue; MFI-20, Multidimensional Fatigue Inventory; MFSI-30, Multidimensional Fatigue Symptom Inventory short form; PFS, Piper Fatigue Scale (revised); Schwartz CFS, Schwartz Cancer Fatigue Scale; Wu CFS, Wu Cancer Fatigue Scale.

a

This involved cross-referencing studies in Medline and Web of Science and identifying studies in cancer where the scale has been used subsequently.

The Chalder Fatigue Scale—also known as the Fatigue Questionnaire (FQ) [33]—is an 11-item scale that was originally validated in a general practice setting. However, its main use has been in the investigation of chronic fatigue syndrome. It is brief and easy to administer while still covering two aspects of fatigue (mental and physical). It has been used in population studies and so has normative data available for comparison with cancer patients [34, 41].

The Fatigue Symptom Inventory (FSI) [35] is a 13-item scale that was originally validated in a breast cancer population. A subsequent validation study involved 342 mixed cancer patients [42]. It has reasonable psychometric properties but there is a question over its test–retest reliability. It has been used in a number of studies but with overall small numbers of patients. Its areas of measurement have been limited to patients undergoing active treatment and survivors.

The Lee fatigue scale (or Visual Analogue Scale for Fatigue—VASF) [36] is an 18-item tool that was originally validated in a group of patients with sleep disorders. It has had very limited use in cancer patients. Its psychometric properties were assessed in a sample of 212 mixed cancer patients undergoing treatment [31]. However, because of a potential overlap with measures of sleep disturbance it is not recommended for use in CRF measurement.

The Multidimensional Assessment of Fatigue (MAF) [43] is a 16-item scale that was originally validated in rheumatoid arthritis patients. Its psychometric properties were assessed in the same sample of cancer patients previously discussed with respect to the VASF [31]. It was also used (as one of a number of fatigue measures) by Morrow and colleagues in two subsequent intervention studies examining the role of antidepressants in treating fatigue [10, 37]. Compared with other fatigue instruments included in this review, it has been relatively poorly validated and its use in further studies cannot be recommended unless further validation work is undertaken.

The Multidimensional Fatigue Inventory (MFI-20) [44] is a 20-item scale that was designed for use in cancer patients. Its original validation had a number of group comparisons including both healthy controls and groups of subjects who at various time points were expected to be fatigued. This included army trainees and doctors undertaking shift work. This means there are normative data available for reference. It was further validated for use in cancer in a subsequent study of 275 patients undergoing radiotherapy [38]. It has since been used in a number of studies but with small numbers of patients in each study.

The Multidimensional Fatigue Symptom Inventory short form (MFSI-30) [39] is a 30-item scale that was originally investigated in a group of breast cancer patients undergoing treatment. It was further validated in a mixed cancer population of 304 patients [40]. It has favourable psychometric properties; however, its use in clinical studies has been somewhat limited. There have only been two additional studies reporting its use in breast cancer survivors.

Its use beyond this group will require further validation.

The revised PFS [45] is a shorter version of the original scale [13]. Piper reported some limitations to her original fatigue scale [13] and she went on develop a revised version. This revised scale consists of 22 items and has been validated in a group of breast cancer survivors [45]. There are limited data on the psychometric properties of the revised scale for use in cancer patients although data are available in other populations. While the scale has since been used in a small number of studies of cancer patients, the majority of these studies have been undertaken in breast cancer patients undergoing treatment and in breast cancer survivors. There are little or no data in other cancer populations.

The Schwartz Cancer Fatigue Scale [46] is a 28-item scale that was validated in a mixed cancer population undergoing treatment. Its psychometric properties were further examined (along with the VASF and the MAF) in the study of 212 mixed cancer patients discussed previously [31]. Schwartz has also determined the minimum clinically significant difference in a further study [32] on 103 mixed cancer patients undergoing treatment. We did not find any other studies that have used this scale. Its usefulness despite extensive psychometric data must therefore be questioned.

The Wu Cancer Fatigue Scale [47]: The original scale had 16 items but a secondary validation study found redundant items and the scale was revised to include only nine items [48]. The instrument has undergone limited psychometric evaluation and has not been used in any subsequent studies. The relative lack of data means that it cannot be recommended for further use.

## discussion

This review has demonstrated the range and number of scales available to specifically measure CRF. A number of other tools have been validated for the measurement of fatigue in nonmalignant conditions and in the general population but evaluation of these scales was beyond the scope of this review.

While this review has used systematic methodology, there are some potential limitations. The titles and abstracts were only screened by one author—this was because of limited time and resources. However, both authors agreed on the final included studies. This includes three scales which were identified but excluded from full analysis because only one published paper was available. These authors were not contacted for further unpublished data as a meta-analysis was not possible with this type of review. As a result, obtaining these additional data would not have materially altered the conclusions of the review (which concerned peer-reviewed published data only).

Our estimate of the number of studies that the various scales have been used in is on the basis of a comprehensive cross-referencing in both Medline and the ISI citation index with full text article retrieval if necessary. The purpose of including these approximate figures is to give the reader an ‘order of magnitude’ assessment about how often these scales have been used in previous studies. However, this method is not foolproof (e.g. authors may have failed to reference the scale correctly) and so this is stated to be approximate numbers only.

The unidimensional scales (which measure the physical impact of fatigue) are the most widely used. They also have some of the most robust psychometric data to support their use. Their limitation is the scope of measurement—subjective fatigue is more than the sensation of physical impairment. Although measuring physical fatigue symptoms may include the social and functional impact of physical fatigue. This extends beyond one aspect of fatigue that could imply that these scales cover more than one dimension of CRF.

However, any theoretical limitation is compensated for by their ease of use and brevity (the scales contain between three and 13 items). The most widely used of these scales are the FACT F and the EORTC QLQ C30 fatigue subscale which have data from over 10 000 patients between them and have been widely used in intervention studies to treat CRF [49]. The FACT F has the advantage of having a validated clinically significant score change—it also covers the social impact of CRF but takes longer to administer than the three-item EORTC QLQ C30 fatigue subscale. The former should be used in a research setting and the latter could be used to monitor clinical effects. The Profile of Mood States Fatigue has also been extensively used but was not originally validated in cancer patients and has no clear advantage over the other two scales. It has been used to help validate six out of the 14 scales included. The widespread use in a healthy population could provide a useful baseline measure of fatigue in the general population [50] for comparison with a cancer patient group.

The multidimensional scales are much more limited in their usage. While they offer the theoretical advantage of covering more aspects of fatigue, such as the cognitive or affective symptoms, this is often at the expense of an increase in the number of items. The more complex administration and completion time may be why their usage has been more limited. Moreover, the advantages of measuring additional ‘dimensions’ of fatigue are not clear [19]. As yet, there is no clinical value in distinguishing between patients with predominantly ‘physical’, ‘cognitive’ or ‘motivational’ fatigue. Until the clinical importance of these proposed dimensions has been clarified there is little incentive to use these scales outside of a research study. Indeed, some authors suggest that this construct is redundant as CRF can be regarded as a unidimensional phenomenom [19]. It is perhaps relevant that a number of multidimensional scales have been developed that have not been subsequently extensively used. Some of these scales have up to 30 items and so could potentially more than double the administration time of all the unidimensional scales. The majority of the scales (seven out of nine) have data on <1000 patients for any individual scale. In addition, three of these seven scales have only been used in breast cancer patients and so their use is even more limited. The two exceptions are the FQ and the MFI-20. These two scales have each been used to measure CRF in >2000 patients. However, this evaluates still considerably fewer patients than the FACT F or EORTC QLQ C30. The FQ, by virtue of consisting of only 11 items, offers a two-dimensional (physical and mental) assessment of fatigue without an increase in the time required for administration. It was not, however, originally designed for use in cancer patients, whereas the MFI-20 was specifically created to measure fatigue in this population. Ultimately, the trade-off between undertaking a comprehensive multidimensional assessment of fatigue and the consequent problems of missing data and questionnaire ‘burden’ will be a decision for individual researchers.

## conclusions

For most purposes, a unidimensional fatigue instrument is the appropriate measure. Of the many unidimensional scales available the best validated and most widely used are the EORTC QLQ C30 fatigue subscale and the FACT F. The EORTC subscale has the virtue of brevity and is usually used as part of the EORTC QLQ C30 quality-of-life assessment instrument. However, its extreme brevity and limited number of response categories may make it less sensitive to changes in fatigue and less able to detect differences in fatigue between groups. For this reason, the authors prefer the FACT F. This instrument is relatively brief, has robust psychometric properties and can also be easily combined with a validated quality-of-life assessment instrument (such as the FACT G). In those circumstances where a multidimensional fatigue instrument is required, the authors recommend the use of the FQ. Although not originally developed for use in cancer patients, this scale has robust psychometric properties and has been extensively used in other populations. It is brief, easy to use and its dimensions have face validity.

## funding

OM was supported by a Wellcome Trust Value in People (VIP) award.

## References

1.
Stone
P
Richards
M
A'Hern
R
Hardy
J
Fatigue in patients with cancers of the breast or prostate undergoing radical radiotherapy
J Pain Symptom Manage
,
2001
, vol.
22

6
(pg.
1007
-
1015
)
2.
Stone
P
Hardy
J
K
, et al.  .
Fatigue in advanced cancer: a prospective controlled cross-sectional study
Br J Cancer
,
1999
, vol.
79

9–10
(pg.
1479
-
1486
)
3.
Bower
JE
Ganz
PA
Desmond
KA
, et al.  .
Fatigue in long-term breast carcinoma survivors
Cancer
,
2006
, vol.
106

4
(pg.
751
-
758
)
4.
Minton
O
Stone
P
How common is fatigue in disease-free breast cancer survivors? A systematic review of the literature
Br Cancer Res Treat
,
2007

doi: 10.1007/s10549-007-9831-1
5.
Cella
D
Davis
K
Breitbart
W
, et al.  .
Cancer-related fatigue: prevalence of proposed diagnostic criteria in a United States sample of cancer survivors
J Clin Oncol
,
2001
, vol.
19

14
(pg.
3385
-
3391
)
6.
L
Strasser
F
Elsner
F
, et al.  .
Fatigue in palliative care patients—an EAPC approach
Palliat Med
,
2008
, vol.
22
(pg.
13
-
32
)
7.
NCCN Clinical Practice Guidelines in Oncology
Cancer Related Fatigue

http://www.nccn.org (8 January 2008, date last accessed)
8.
Streiner
DL
Norman
GR
Health Measurement Scales—A Practical Guide to Their Development and Use
,
1989
1st edition
Oxford
Oxford University Press
pg.
175

9.
Yoshitake
H
Relations between the symptoms and the feeling of fatigue
Ergonomics
,
1971
, vol.
14

1
(pg.
175
-
186
)
10.
Morrow
G
Hickok
J
Roscoe
J
, et al.  .
Differential effects of paroxetine on fatigue and depression: a randomized, double-blind trial from the University of Rochester Cancer Center Community Clinical Oncology Program
J Clin Oncol
,
2003
, vol.
21

24
(pg.
4635
-
4641
)
11.
Ahsberg
E
Dimensions of fatigue in different working populations
Scand J Psychol
,
2000
, vol.
41

3
(pg.
231
-
241
)
12.
Ahsberg
E
Furst
CJ
Ahsberg
E
Furst
CJ
Dimensions of fatigue during radiotherapy—an application of the Swedish Occupational Fatigue Inventory (SOFI) on cancer patients
Acta Oncol
,
2001
, vol.
40

1
(pg.
37
-
43
)
13.
Piper
BF
Piper fatigue scale available for clinical testing
Oncol Nurs Forum
,
1990
, vol.
17

5
(pg.
661
-
662
)
14.
Okuyama
T
Akechi
T
Kugaya
A
, et al.  .
Development and validation of the cancer fatigue scale: a brief, three-dimensional, self-rating scale for assessment of fatigue in cancer patients
J Pain Symptom Manage
,
2000
, vol.
19

1
(pg.
5
-
14
)
15.
Glaus
A
Fatigue in patients with cancer. Analysis and assessment
Recent Results Cancer Res
,
1998
, vol.
145

I–IX
(pg.
1
-
172
)
16.
Passik
SD
Kirsh
KL
Donaghy
K
, et al.  .
Patient-related barriers to fatigue communication: initial validation of the Fatigue Management Barriers Questionnaire
J Pain Symptom Manage
,
2002
, vol.
24

5
(pg.
481
-
493
)
17.
Quick
M
Fonteyn
M
Development and implementation of a clinical survey for cancer-related fatigue assessment
Clin J Oncol Nurs
,
2005
, vol.
9

4
(pg.
435
-
439
)
18.
Holley
SK
Evaluating patient distress from cancer-related fatigue: an instrument development study
Oncol Nurs Forum
,
2000
, vol.
27

9
(pg.
1425
-
1431
)
19.
Lai
J-S
Crane
P
Cella
D
Factor analysis techniques for assessing sufficient unidimensionality of cancer related fatigue
Qual Life Res
,
2006
, vol.
15

7
(pg.
1179
-
1190
)
20.
Knobel
H
Loge
JH
Brenne
E
, et al.  .
The validity of EORTC QLQ-C30 fatigue scale in advanced cancer patients and cancer survivors [see comment]
Palliat Med
,
2003
, vol.
17

8
(pg.
664
-
672
)
21.
Storey
DJ
Waters
RA
Hibberd
CJ
, et al.  .
Clinically relevant fatigue in cancer outpatients: the Edinburgh Cancer Centre symptom study
Ann Oncol
,
2007
, vol.
18

11
(pg.
1861
-
1869
)
22.
Cella
D
Functional Assessment of Cancer Therapy Website

www.facit.org (28 January 2008, date last accessed)
23.
Mendoza
TR
Wang
XS
Cleeland
CS
, et al.  .
The rapid assessment of fatigue severity in cancer patients: use of the Brief Fatigue Inventory
Cancer
,
1999
, vol.
85

5
(pg.
1186
-
1196
)
24.
Aaronson
NK
Ahmedzai
S
Bergman
B
, et al.  .
The European Organization for Research and Treatment of Cancer QLQ-C30: a quality-of-life instrument for use in international clinical trials in oncology
J Natl Cancer Inst
,
1993
, vol.
85

5
(pg.
365
-
376
)
25.
Krupp
LB
LaRocca
NG
Muir Nash
J
Steinberg
The fatigue severity scale. Application to patients with multiple sclerosis and systemic lupus erythematosus
Arch Neurol
,
1989
, vol.
46

10
(pg.
1121
-
1123
)
26.
Stone
P
Richards
M
A'Hern
R
Hardy
J
A study to investigate the prevalence, severity and correlates of fatigue among patients with cancer in comparison with a control group of volunteers without cancer
Ann Oncol
,
2000
, vol.
11

5
(pg.
561
-
567
)
27.
Yellen
SB
Cella
DF
Webster
K
, et al.  .
Measuring fatigue and other anemia-related symptoms with the Functional Assessment of Cancer Therapy (FACT) measurement system
J Pain Symptom Manage
,
1997
, vol.
13

2
(pg.
63
-
74
)
28.
Cella
D
Eton
DT
Lai
JS
, et al.  .
Combining anchor and distribution-based methods to derive minimal clinically important differences on the Functional Assessment of Cancer Therapy (FACT) anemia and fatigue scales
J Pain Symptom Manage
,
2002
, vol.
24

6
(pg.
547
-
561
)
29.
McNair DM and Lorr M EITS Manual for the Profile of Mood States
,
1971
San Diego, CA
Educational and Testing Service
30.
Norcross
JC
E
JO
, et al.  .
Factor structure of the Profile of Mood States (POMS): two partial replications
J Clin Psychol
,
1984
, vol.
40

5
(pg.
1270
-
1277
)
31.
Meek
PM
Nail
LM
Barsevick
A
, et al.  .
Psychometric testing of fatigue instruments for use with cancer patients
Nurs Res
,
2000
, vol.
49

4
(pg.
181
-
190
)
32.
Schwartz
AL
Meek
PM
Nail
LM
, et al.  .
Measurement of fatigue. Determining minimally important clinical differences
J Clin Epidemiol
,
2002
, vol.
55

3
(pg.
239
-
244
)
33.
Chalder
T
Berelowitz
G
Pawlikowska
T
, et al.  .
Development of a fatigue scale
J Psychosom Res
,
1993
, vol.
37

2
(pg.
147
-
153
)
34.
Loge
JH
Abrahamsen
AF
Ekeberg
O
, et al.  .
Hodgkin's disease survivors more fatigued than the general population
J Clin Oncol
,
1999
, vol.
17

1
(pg.
253
-
261
)
35.
Hann
DM
Jacobsen
PB
Azzarello
LM
, et al.  .
Measurement of fatigue in cancer patients: development and validation of the Fatigue Symptom Inventory
Qual Life Res
,
1998
, vol.
7

4
(pg.
301
-
310
)
36.
Lee
K
Hicks
G
Nino-Murcia
G
Validity and reliability of a scale to measure fatigue
Psychiatry Res
,
1991
, vol.
36
(pg.
291
-
298
)
37.
Roscoe
JA
Morrow
GR
Hickok
JT
, et al.  .
Effect of paroxetine hydrochloride (Paxil®) on fatigue and depression in breast cancer patients receiving chemotherapy
Breast Cancer Res Treat
,
2005
, vol.
89

3
(pg.
243
-
249
)
38.
Smets
EM
B
Cull
A
de Haes
JC
Application of the multidimensional fatigue inventory (MFI-20) in cancer patients receiving radiotherapy
Br J Cancer
,
1996
, vol.
73

2
(pg.
241
-
245
)
39.
Stein
KD
Martin
SC
Hann
DM
, et al.  .
A multidimensional measure of fatigue for use with cancer patients
Cancer Pract
,
1998
, vol.
6

3
(pg.
143
-
152
)
40.
Stein
KD
Jacobsen
PB
Blanchard
CM
, et al.  .
Further validation of the multidimensional fatigue symptom inventory-short form
J Pain Symptom Manage
,
2004
, vol.
27

1
(pg.
14
-
23
)
41.
Loge
JH
Ekeberg
O
Kaasa
S
, et al.  .
Fatigue in the general Norwegian population: normative data and associations
J Psychosom Res
,
1998
, vol.
65
(pg.
53
-
65
)
42.
Hann
DM
Denniston
MM
Baker
F
, et al.  .
Measurement of fatigue in cancer patients: further validation of the Fatigue Symptom Inventory
Qual Life Res
,
2000
, vol.
9

7
(pg.
847
-
854
)
43.
Belza
B
Comparison of self-reported fatigue in rheumatoid arthritis and controls
J Rheumatol
,
1995
, vol.
22

4
(pg.
639
-
643
)
44.
Smets
EM
B
Bonke
B
, et al.  .
The Multidimensional Fatigue Inventory (MFI) psychometric qualities of an instrument to assess fatigue
J Psychosom Res
,
1995
, vol.
39

3
(pg.
315
-
325
)
45.
Piper
BF
Dibble
SL
Dodd
MJ
, et al.  .
The revised Piper Fatigue Scale: psychometric evaluation in women with breast cancer
Oncol Nurs Forum
,
1998
, vol.
25

4
(pg.
677
-
684
)
46.
Schwartz
AL
The Schwartz Cancer Fatigue Scale: testing reliability and validity
Oncol Nurs Forum
,
1998
, vol.
25

4
(pg.
711
-
717
)
47.
Wu
HS
McSweeney
M
Wu
H-S
McSweeney
M
Assessing fatigue in persons with cancer: an instrument development and testing study
Cancer
,
2004
, vol.
101

7
(pg.
1685
-
1695
)
48.
Wu
HS
Wyrwich
KW
McSweeney
M
, et al.  .
Assessing fatigue in persons with cancer: further validation of the Wu Cancer Fatigue Scale
J Pain Symptom Manage
,
2006
, vol.
32

3
(pg.
255
-
265
)
49.
Minton
O
Stone
P
Richardson
A
, et al.  .
Drug therapy for the management of cancer related fatigue
Cochrane Database of Systematic Reviews. Reviews 2008 Issue 1
,
2008
(1)
Chichester, UK
John Wiley & Sons, Ltd
50.
David
A
Pelosi
A
McDonald
A
, et al.  .
Tired, weak, or in need of rest: fatigue among general practice attenders
Br Med J
,
1990
, vol.
301

6762
(pg.
1199
-
1202
)