Volume 36, Issue 4 , Pages 367-373, October 2008
Psychometric Properties of the Brief Fatigue Inventory in Greek Patients with Advanced Cancer
Article Outline
Abstract
To validate the Greek version of the Brief Fatigue Inventory (BFI-Gr) in a sample of cancer patients, the scale was translated with the “forward–backward” procedure to Greek. It was administered twice, at a three-day interval, to 102 eligible patients with cancer. Together with the BFI-Gr scale, the patients also completed the European Organization for Research and Treatment of Cancer QLQ-C30 (version 3.0) subscales of fatigue and emotional functioning, and the M. D. Anderson Symptom Inventory. The BFI-Gr had an overall Cronbach alpha for the nine items of 0.954. The assessment of the relationships between the BFI-Gr and the other measurements showed statistically significant correlations between all the assessed measurements (r values between 0.47 and 0.76, P
<0.0005), except with the emotional subscale of the European Organization for Research and Treatment of Cancer scale. Factor analysis yielded a one-factor solution, explaining 73.6% of the variance. Interitem correlations were high and ranged from 0.567 to 0.882 (P<0.0005). The test–retest reliability of scale showed that the coefficient agreement was 0.901 (P<0.0005). Univariate analysis revealed significant correlations between hemoglobin and fatigue (r=−0.21, P=0.037), and between performance status (P<0.0005) and opioids (P=0.009). These results support that the BFI-Gr is an instrument with satisfactory psychometric properties, and is a valid research tool for cancer-related fatigue in Greek cancer patients
Key Words: Fatigue, cancer quality of life, symptom distress
Introduction
Fatigue is one of the most common symptoms in cancer patients. The prevalence of fatigue has been reported to be over 50% in those with advanced disease1, 2, 3 and over 30% in newly diagnosed cancer patients4 and cancer survivors.5 It is also known that patients who undergo anticancer treatment modalities such as chemotherapy, radiotherapy, and bone marrow transplantation often experience fatigue as a significant treatment side effect.6, 7 Patients who experience severe fatigue may withdraw prematurely from potentially curative treatment, or be unwilling to take adequate doses of various forms of treatment.8 It frequently and significantly interferes with quality of life.9
Both physical and psychological factors are thought to be associated with fatigue.10 Although the physiological mechanisms are unknown, it has been recognized that a low hemoglobin level is associated with severe fatigue. An association between fatigue and albumin level also has been reported,11 but is still controversial. Numerous other factors may contribute.
Because fatigue is not often assessed and is undertreated, concise assessment is the key to better management of this symptom. Self-rating scales have enabled physicians to assess patient symptoms without introducing observer bias. The Multidimensional Fatigue Inventory12 is a 24-item questionnaire in which the patient has to indicate on seven-point scales to what extent the particular statement applies to him or her; the statements refer to aspects of fatigue experienced during the previous days, and higher scores indicate a higher degree of fatigue. The Functional Assessment of Cancer Therapy-Fatigue13 measures multiple fatigue characteristics and their impact on function. The Schwartz Cancer Fatigue Scale14 is a 28-item scale comprising four subscales (physical, emotional, cognitive, and temporal). The Fatigue Symptom Inventory15 is a 13-item self-report measure designed to measure the intensity and duration of fatigue and its impact on quality of life. The revised Piper Fatigue Scale16 is a 22-item scale comprising four subscales (behavioral/severity, affective meaning, sensory, and cognitive/mood). The main characteristic of these recently developed scales is their multidimensionality. However, multidimensional scales are often too long for very sick patients to complete.
There currently exists no gold standard for assessing and managing cancer-related fatigue, although guidelines have been developed.17 These guidelines emphasize the importance of assessment.
Fatigue assessment has rarely been part of routine cancer care, as both patients and health care professionals historically have regarded cancer-related fatigue as an expected part of the disease. The science of measuring fatigue is only recently becoming well developed. A challenge in measuring fatigue in cancer patients is distinguishing patients having severe fatigue from those having moderate or mild fatigue. In studies of another validated questionnaire, the Brief Fatigue Inventory (BFI), Levels 1–3 fatigue indicated mild fatigue, Levels 4–6 fatigue was moderate, and Levels 7–10 fatigue was severe.18 In a study of American cancer patients, BFI responses showed that 50% of patients with hematological malignancies experienced severe fatigue (defined as seven or greater on a 0–10 scale), as did 34% of patients with solid tumors. In contrast, severe fatigue was reported by only 17% of the community-dwelling sample.18, 19, 20
The BFI18 was specifically developed to measure fatigue in cancer populations. The BFI has been translated into a variety of languages (Chinese, Japanese, German, and Taiwanese) and the psychometric properties have been established.20, 21, 22, 23 The validation and translation of the BFI in Greek will provide further evidence for the measurement's adaptability in different cultures, and also will allow study results to be compared across different countries.
The aim of the present study was to assess the psychometric properties of the BFI translated into Greek, including validity and reliability in a sample of Greek cancer patients receiving palliative care.
Patients and Methods
A total of 190 consecutive patients suffering from incurable cancer were treated in a palliative care unit in Athens, Greece, for pain relief and cancer-related symptoms during the study period. Criteria for inclusion were histologically confirmed malignancy, age >18 years, ability to communicate effectively with health care professionals, and patient informed consent. Patients were excluded if there was a history of drug abuse, a diagnosis of a psychotic illness, or significant cognitive impairment. A total of 155 patients fulfilled the study criteria and were considered eligible for entry into the study. Participants were invited on a face-to-face basis. Forty-eight (31%) patients did not complete the assessment forms, either due to refusal or due to long distance, and thus were excluded from the study. The final sample consisted of 102 cancer patients.
Patients were seen individually either at outpatient clinics or on the wards. Participants were asked to complete three self-report questionnaires: the BFI, the European Organization for Research and Treatment of Cancer (EORTC) QLQ-C30 (version 3.0) subscales of fatigue and emotional functioning, and the M. D. Anderson Symptom Inventory (MDASI). Research workers recorded data on disease status, treatment regimen, and demographic characteristics (age, sex, family status, and educational level). The sociodemographic and clinical data of the participants are presented in Table 1. Disease status information included cancer diagnosis, anticancer treatment (chemotherapy, radiotherapy), and performance status as defined by the Eastern Cooperative Oncology Group.24 Treatment regimen data consisted of the opioid analgesics the patients were already receiving. The study was performed in accordance with the Helsinki Declaration and according to European guidelines for good clinical practice, and was approved by the institutional review board.
Table 1. Demographic and Disease-Related Patient Characteristics
| n | % | |
|---|---|---|
| Age | ||
| Mean±SD: 64.32±11.58 years | Range (39–84) | |
| Years of education | ||
| Mean±SD: 10.20±3.80 years | Range (6–16) | |
| Gender | ||
| Male | 55 | 53.9 |
| Female | 47 | 46.1 |
| Diagnosis | ||
| Urogenital | 32 | 31.4 |
| Lung | 27 | 26.5 |
| Breast | 19 | 18.6 |
| Other | 13 | 12.7 |
| Gastrointestinal | 11 | 10.8 |
| Family status | ||
| Married | 90 | 88.2 |
| Unmarried | 12 | 11.8 |
| ECOG score | ||
| 0–1 good | 59 | 57.8 |
| 2–3 poor | 43 | 42.2 |
| Metastasis | ||
| No | 23 | 22.5 |
| Yes | 79 | 77.5 |
| Chemotherapy | ||
| No | 34 | 33.3 |
| Yes | 68 | 66.7 |
| Radiotherapy | ||
| No | 31 | 30.4 |
| Yes | 71 | 69.6 |
| Opioids | ||
| Mild | 20 | 19.6 |
| Strong | 82 | 80.4 |
Translation
To develop the BFI-Greek (BFI-Gr), the forward–backward translation method was used. The items were first translated into Greek by two independent translators whose native language was Greek and then back-translated into English by another two independent translators whose native language was English and who had not seen the original English version. Next, the English back-translated items were compared with the originals.
Instruments
The BFI was developed to measure fatigue in cancer populations. The three important characteristics of this scale are (1) it is brief and easy for patients to complete, (2) it is easily translated into other languages, and (3) it includes interference items. The validity and reliability of the original scale has been established.18 The questionnaire uses an 11-point scale (0=“no fatigue” to 10=“fatigue as bad as you can imagine”) to measure the specific symptom of cancer-related fatigue in a single dimension. Three items ask patients to rate the severity of their fatigue at its “worst,” “usual,” and “now” during the past 24hours. Cutpoints for fatigue severity were defined in two categories: a “fatigue worst” rating of ≥7 indicates “severe” and 0–6 indicates “nonsevere.” Six additional items describe how much fatigue has interfered with different aspects of the patient's life during the past 24hours. These items include general activity, mood, walking ability, normal work, relationships with other people, and enjoyment of life. Interference is measured with 0=“does not interfere” and 10=“completely interferes.” The global score for the BFI is calculated as the mean value of these nine items.
The EORTC QLQ-C30 is one of the most frequently used self-rating questionnaires in assessing patients' quality of life.25 It has 30 items and consists of five multi-item function subscales and a global health status/quality of life (physical, role, emotional, cognitive, and social function), four multi-item symptom scales (fatigue, pain, nausea, and vomiting) and six items that assess symptoms (dyspnea, sleep disturbance, appetite loss, diarrhea, and constipation) and financial impact. In the present study, we used the fatigue subscale and the emotional subscale in the analysis, using a four-point Likert scale for each. All scales range in a transformed score from 0 to 100. A high score for a functional scale represents a high/healthy level of functioning. A high score for the global health status/quality of life represents a high quality of life. A high score for a symptom scale/item represents a high level of symptomatology/problems. Originally, the EORTC QLQ-C30 (version 3.0) was designed for prospective randomized trials in cancer patients, but today, it also is used as a screening instrument in cross-sectional studies. The questionnaire has been validated, by the authors of the present study, in a Greek sample of cancer patients attending a palliative care unit.26
The MDASI is a brief measure of the severity and impact of cancer-related symptoms.19 The Greek version of the MDASI was used for the assessment of symptom prevalence and distress.27 The MDASI consists of 13 core symptom items that are rated based on their presence and severity. Each symptom is rated on an 11-point scale (0–10) to indicate the presence and severity of the symptom, with 0 meaning “not present” and 10 meaning “as bad as you can imagine” in the last 24hours. It also includes six symptom interference items that are rated based on the level of symptom interference with the function of a patient's life in the last 24hours. The interference items were also measured on scales from 0 to 10, with 0 meaning “did not interfere” and 10 meaning “interfered completely.” The MDASI was administered to all patients.
Statistical Analysis
Descriptive statistics, including means, and counts and percentages for the variables were calculated. The psychometric properties of the BFI-Gr were assessed. Exploratory factor analysis (principal axis factoring with varimax rotation) was used to examine the structure of the questionnaire. Criterion validity was evaluated by calculating the Pearson product moment correlation coefficient between the BFI-Gr scores and the EORTC QLQ-C30 fatigue and emotional subscales and the MDASI total score, the MDASI fatigue item, and the MDASI total score omitting the fatigue item. Convergent validity was evaluated by calculating the interinstrument correlations. The internal consistency was evaluated by calculating the Cronbach alpha coefficient, which ranges from 0 to 1, higher values indicating less measurement error. The test–retest reliability was evaluated by calculating the Pearson product moment correlation coefficient with a three-day interval in the whole sample. A three-day interval was chosen because fatigue severity is not expected to change within this time period. We also investigated the association between fatigue scale scores and the demographic and clinical data.
Results
Of the 102 patients, 53.9% were males and 88.2% were married. Urogenital and lung cancer were diagnosed in 31.4% and 26.5%, respectively. A significant proportion (42.2%) had a poor performance status, 77.5% had metastasis, 66.7% had undergone chemotherapy, and 69.6% had received radiotherapy. A total of 80.4% were already receiving strong opioids (Table 1).
Validity
Factor Analysis. Factor analysis was performed to determine whether the BFI-Gr measures the same constructs as the original BFI. The Keiser Meyer-Olkin measure of sampling adequacy was equal to 0.921, showing that the data are suitable for factor analysis. The results of the scree test for the BFI-Gr suggested a one-factor solution. The eigenvalue was 6.62 for this factor, followed by 0.75 and 0.45 for the second and third factor, respectively. Moreover, the first factor explained 73.6% of the variability in the data. The single-factor model fits well according to Harman's28 rule that the standard deviation of the residuals be slightly less than or approximately equal to the standard error of a correlation coefficient, which is the reciprocal of the square root of the sample size. The eigenvalues and amount of explained variability indicate that most of the data can be explained by a single construct, consistent with the original BFI. Because the BFI-Gr measures a single construct, the mean of the nine BFI-Gr items can be used as a global BFI-Gr score.
Criterion Validity. Pearson's product moment correlation was used to calculate the relationships among the BFI-Gr, and the EORTC fatigue and emotional subscales, and the MDASI total score, MDASI fatigue item, and MDASI total score without the fatigue item. The results showed statistically significant correlations between BFI-Gr and all the assessed measurements (r values between 0.471 and 0.769, P<0.0005), except with the emotional subscale of the EORTC instrument (Table 2).
Table 2. Construct Validity Between BFI-Gr, EORTC QLQ-C30, and MDASI
| BFI-Gr | ||
|---|---|---|
| Pearson's Correlation Coefficient | P value | |
| MDASI total score | 0.525 | <0.0005 |
| MDASI fatigue item | 0.769 | <0.0005 |
| MDASI total score without fatigue item | 0.471 | <0.0005 |
| EORTC fatigue subscale | 0.695 | <0.0005 |
| EORTC emotional subscale | −0.106 | NS |
Convergent Validity. Convergent validity is one aspect of construct validity of psychological measurements.29 It was demonstrated that the questionnaire was able to show a correspondence or convergence between the items of the scale. Correlation coefficients (Pearson's r) were high and ranged from 0.567 to 0.882 (P<0.0005) (Table 3).
Table 3. Convergent Validity: Interitem Correlation
| Items | 1 | 2 | 3 | 4a | 4b | 4c | 4d | 4e | 4f |
|---|---|---|---|---|---|---|---|---|---|
| 1 | 1.000 | 0.818 | 0.838 | 0.807 | 0.618 | 0.733 | 0.782 | 0.567 | 0.678 |
| 2 | 1.000 | 0.882 | 0.759 | 0.604 | 0.677 | 0.731 | 0.583 | 0.638 | |
| 3 | 1.000 | 0.793 | 0.613 | 0.662 | 0.746 | 0.584 | 0.665 | ||
| 4a | 1.000 | 0.653 | 0.730 | 0.824 | 0.651 | 0.770 | |||
| 4b | 1.000 | 0.589 | 0.625 | 0.678 | 0.762 | ||||
| 4c | 1.000 | 0.813 | 0.599 | 0.595 | |||||
| 4d | 1.000 | 0.669 | 0.736 | ||||||
| 4e | 1.000 | 0.780 | |||||||
| 4f | 1.000 |
Reliability
Internal Consistency. In an attempt to estimate the internal consistency of the BFI-Gr, Cronbach's alpha (estimates of a magnitude of 0.70 or greater were sought) was calculated for the nine items in the scale. The overall alpha was 0.954. The fact that the reliability coefficient of the scale is high emphasized the increased internal consistency and reliability of the scale.
Test–Retest (Stability). Data from 102 patients who responded both at the first and the second assessment (three-day interval) were used to examine the stability of the BFI-Gr over time. The test–retest reliability of scale (Pearson's r), showed that the coefficient agreement was 0.901 (P<0.0005). These results indicate that the BFI-Gr scores were remarkably consistent across the two occasions and were significantly correlated.
Univariate Analysis
Significant correlations were found between the continuous variable hemoglobin and fatigue (r=−0.210, P=0.037), whereas the associations between the categorical variables and fatigue revealed statistically significant correlation between performance status (P<0.0005) and opioids (P=0.009) (Table 4).
Table 4. Means of Categorical Demographics, Clinical Variables, and BFI-Gr
| BFI-Gr | P value | |||
|---|---|---|---|---|
| n | Mean | SD | ||
| Gender | ||||
| Male | 55 | 50.91 | 12.67 | NS |
| Female | 47 | 49.51 | 12.05 | |
| Family status | ||||
| Married | 90 | 49.92 | 11.44 | NS |
| Unmarried | 12 | 53.33 | 18.38 | |
| ECOG | ||||
| 0–1 | 59 | 45.89 | 12.02 | <0.0005 |
| 2–3 | 43 | 56.25 | 10.17 | |
| Chemotherapy | ||||
| No | 34 | 50.03 | 11.97 | NS |
| Yes | 68 | 50.38 | 12.62 | |
| Radiotherapy | ||||
| No | 31 | 50.74 | 13.06 | NS |
| Yes | 71 | 50.06 | 12.11 | |
| Metastasis | ||||
| No | 23 | 49.43 | 13.37 | NS |
| Yes | 79 | 50.51 | 12.12 | |
| Opioids | ||||
| Mild | 20 | 51.69 | 11.92 | 0.009 |
| Strong | 82 | 43.70 | 12.07 | |
| NSAID | ||||
| No | 27 | 49.22 | 13.36 | NS |
| Yes | 75 | 50.64 | 12.04 | |
| Primary site | ||||
| Gastrointestinal | 11 | 55.45 | 8.39 | NS |
| Lung | 27 | 52.33 | 10.73 | |
| Urogenital | 32 | 46.78 | 13.48 | |
| Breast | 19 | 47.26 | 13.37 | |
| Other | 13 | 54.54 | 11.70 | |
Discussion
Fatigue, weakness, and lack of energy are three very frequent symptoms in advanced cancer patients.30 Fatigue is one of the top complaints in Western cancer patients19 and is also a major concern to Greek cancer patients. It has been recognized by many oncology professionals that fatigue almost always clusters with other significant symptoms, either caused by disease or therapy.31 Fatigue is commonly conceptualized as a multidimensional symptom that incorporates sensory, cognitive, affective, behavioral, and physiologic components.32 Although the BFI-Gr does not capture the multiple dimensions of fatigue, it is sufficient to screen for patients with high levels of fatigue.
The one-factor solution provided a good model fit for the nine BFI-Gr items, in accordance with the original English, Japanese, German, and Taiwanese versions of the BFI. The construct validity analysis confirmed that the BFI-Gr composite score was highly correlated with items of the MDASI, as well as with the fatigue scale from the EORTC QLQ-C30 (version 3.0). This result encourages the use of BFI-Gr in clinical studies on fatigue as a simple measure of fatigue-related physical functional impairment. Moreover, this finding shows that using a multisymptom or quality-of-life questionnaire may be beneficial in screening for fatigue. Furthermore, patients in the present study related fatigue with symptom distress rather than affective areas, as seen in the findings of construct validity, where BFI-Gr had no significant correlation with the emotional subscale of the EORTC QLQ-C30 (version 3.0).
An interesting finding of the present study is the statistically significant correlation between BFI-Gr and opioids. Patients receiving mild opioids scored higher in BFI-Gr. As participants had been referred to the Pain Relief and Palliative Care Unit for pain relief, among other symptoms, it can be assumed that these patients were undertreated, because pain has been found33 to be a significant predictor for fatigue. Consistent with previous research, the current study showed that cancer patients, and those with anemia had a strong relationship with fatigue.20 In addition, patients having poor Eastern Cooperative Oncology Group performance status scores had higher BFI-Gr scores, as in the study of the Japanese version of the BFI.21
The study had a few limitations. The relatively small sample of the study means a further investigation is needed in a larger sample. Moreover, the present study did not examine the ability of the BFI to distinguish patients according to disease severity.
Conclusion
The BFI-Gr compares favorably to other fatigue assessment tools. Moreover, it is short, easily understood, with comprehensible and standardized rules for administration and scoring, and well-documented reliability and validity. These qualities make the BFI-Gr a useful screening and monitoring instrument even for patients with reduced performance status. This study proved that the BFI-Gr is a reliable and valid self-rating assessment tool for fatigue.
References
- . Prevalence of symptoms among patients with advanced cancer: an international collaborative study. J Pain Symptom Manage. 1996;12(1):3–10
- Fatigue in advanced cancer: a prospective controlled cross-sectional study. Br J Cancer. 1999;79:1479–1486
- Correlates of fatigue in people with breast or lung cancer. Oncol Nurs Forum. 1991;18(1):81–87
- . Symptom distress in newly diagnosed ambulatory cancer patients and as a predictor of survival in lung cancer. J Pain Symptom Manage. 1995;10:423–431
- Fatigue in breast cancer survivors: occurrence, correlates, and impact on quality of life. J Clin Oncol. 2000;18:743–753
- . Fatigue side effects and treatment and quality of life. Qual Life Res. 1995;4:8–13
- Fatigue in women treated with bone marrow transplantation for breast cancer: a comparison with women with no history of cancer. Support Care Cancer. 1997;5:44–52
- . Measuring and understanding fatigue. Oncology. 1999;13(11A):91–97
- . A Canadian survey of cancer patients' experiences. Are their needs being met?. J Pain Symptom Manage. 1998;16:298–306
- Factors correlated with fatigue in disease-free breast cancer patients: application for the Cancer Fatigue Scale. Support Care Cancer. 2000;8:215–222
- Clinical factors associated with cancer-related fatigue in patients being treated for leukemia and non-Hodgkin's lymphoma. J Clin Oncol. 2002;20(5):1319–1328
- . The multidimensional fatigue inventory (MFI) psychometric qualities of an instrument to assess fatigue. J Psychosom Res. 1995;39:315–325
- . Measuring fatigue and other anemia-related symptoms with Functional Assessment of Cancer Therapy (FACT) measurement system. J Pain Symptom Manage. 1997;13:63–74
- . The Schwartz cancer Fatigue Scale: testing reliability and validity. Oncol Nurs Forum. 1998;25:711–717
- Measurement of fatigue in cancer patients: development and validation of the Fatigue Symptom Inventory. Qual Life Res. 1998;7:301–310
- The revised Piper Fatigue Scale: psychometric evaluation in women with breast cancer. Oncol Nurs Forum. 1998;25:677–684
- Cancer-related fatigue clinical practice guidelines in oncology. J Natl Comprehensive Care Network. 2003;1(3):308–331
- The rapid assessment of fatigue severity in cancer patients: use of the Brief Fatigue Inventory. Cancer. 1999;85(5):1186–1196
- Assessing symptom distress in cancer patients: the M.D. Anderson Symptom Inventory. Cancer. 2000;89(7):1634–1646
- Validation study of the Chinese version of the Brief Fatigue Inventory (BFI-C). J Pain Symptom Manage. 2004;27(4):322–332
- Validation study of the Japanese version of the Brief Fatigue Inventory. J Pain Symptom Manage. 2003;25(2):106–117
- Validation of the German version of the Brief Fatigue Inventory. J Pain Symptom Manage. 2003;25(5):449–458
- Validation of the Taiwanese version of the Brief Fatigue Inventory. J Pain Symptom Manage. 2006;32(1):52–59
- Toxicity and response criteria of the Eastern Cooperative Oncology Group. Am J Clin Oncol. 1982;5:649–655
- 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;85(5):365–375
- The EORTC Core Quality of Life Questionnaire (QLQ-C30) in terminally ill cancer patients under palliative care: validity and reliability in a Hellenic sample. Int J Cancer. 2001;94:135–139
- M.D. Anderson Symptom Inventory: validation and utility in cancer patients. Oncology. 2004;67:203–210
- . Modern factor analysis. 2nd ed.. Chicago, IL: University of Chicago Press; 1967;
- . Psychological testing. New York, NY: Macmillan; 1988;
- . The symptoms of advanced cancer: relationship to age, gender, and performance status in 1,000 patients. Support Care Cancer. 2000;8:175–179
- . Symptom clusters and their effect on the functional status of patients with cancer. Oncol Nurs Forum. 2001;28(3):465–470
- . Measuring fatigue. In: Frank-Stromborg M, Olsen SJ editor. Instruments for clinical research in health care. Boston, MA: Jones & Barlett; 1997;p. 282–296
- Fatigue during preoperative chemoradiation for resectable rectal cancer. Cancer. 2001;92:1725–1732
PII: S0885-3924(08)00148-6
doi:10.1016/j.jpainsymman.2007.10.021
© 2008 U.S. Cancer Pain Relief Committee. Published by Elsevier Inc. All rights reserved.
Volume 36, Issue 4 , Pages 367-373, October 2008
