An examination of cancer-related fatigue through proposed diagnostic criteria in a sample of cancer patients in Taiwan

Data about cancer patients suspected to be fatigued were recruited from outpatient oncology and palliative clinics and oncology and palliative inpatient units at a medical center and a district teaching hospital in northern Taiwan. Selection criteria required that patients have a pathological diagnosis of cancer, be at least 18 years old, and be able to communicate in Mandarin or Taiwanese. Patients were excluded if they were cognitively impaired, if they refused to participate, or if they could not understand the intent of this study. A final sample of 265 patients was recruited for this study. Fatigue was assessed using the proposed ICD-10 criteria for cancer-related fatigue. Other data collected included demographics (sex, age), educational level, history of disease (type of tumor), staging of disease, and current treatment (e.g., curative treatment, palliative treatment, chemotherapy, radiotherapy, etc.). Written consent was given before the assessment of CRF.

This study was reviewed and approved by the Institutional Review Board of Mackay Memorial Hospital and Cardinal Tien Hospital.

The International Classification of Diseases recently included diagnostic nomenclature for CRF. These criteria have been proposed as a draft of a structured interview guide to help ensure the diagnosis of CRF, and were made in a standardized way across research and clinical applications by Dr. Cella [11]. In 2005, several researchers validated the use of the proposed ICD-10 criteria for fatigue through comparison with other scales, namely the FACT-F and VAS, and discovered that the internal consistency of P-ICD 10 criteria was very good, with an alpha coefficient of 0.82, and concluded that the ICD-10 criteria can be recommended as a diagnostic tool for cancer-related fatigue [12]. We were authorized by Dr. Cella to use the diagnostic interview guide for CRF, and its Chinese-language version, for application in a clinical and research setting to determine the prevalence of CRF in Taiwan.

The English version of the diagnostic interview guide questionnaire was initially translated into Chinese by one of the professional translator, who is bilingual in Chinese and English and experienced in the study topic. A second qualified and blinded bilingual expert back-translated the Chinese version of the diagnostic interview guide questionnaire, creating a new English version. A monolingual reviewer on the research team then compared the original and back translated versions of the diagnostic interview guide questionnaire. The initial question inquired whether the patient had experienced significant fatigue, a lack of energy, or an increased need to rest for a least a 2-week period during the last month. Given the many causes of fatigue and its multidimensional manifestations, the next 10 questions included 10 common clinical manifestations of fatigue reported by cancer patients. The last 3 questions were arranged to exclude non-CRF (Table 2).

Prior to the data collection for the current study, the coordinator of this study (Y.L.) met the interviewers of this study (E.Y, Y.T., and W.S.) for training purposes. The training involved a review of the diagnostic criteria, the completion of simulated interviews, and the completion of actual interviews conducted with patients who were not participants of this study. The interviews of this study were conducted by 2 physicians and 1 nurse who was also currently a graduate student. They were all experienced in working with cancer patients. Sixteen interviews were tape-recorded for the purpose of determining inter-rater reliability. Ratings obtained from the interviewers were compared with those from the independent rater (S.L.) who was trained for the administration of the interviews but not involved in the data collection of this study. The level of the overall agreement between the study interviewers and the independent rater regarding the presence/absence of cancer-related fatigue was 95%. After calculation, we obtained that the coefficient of kappa, a measurement adjusted for base rates, was 0.88. The level of the overall agreement between the study interviewers and the independent rater regarding the presence/absence of the 14 individual criteria that comprise the diagnostic interview guide for Cancer-Related Fatigue (kappa = 0.96) was 90%. According to the rule of thumb, values of Kappa from 0.40 to 0.59 are considered moderate, 0.60 to 0.79 substantial, and 0.80 outstanding (Landis & Kochi, 1977). Most statisticians prefer Kappa values to be at least 0.6 and most often higher than 0.7 before claiming a good level of agreement.

A pretest was performed in which 21 Chinese-language versions of the diagnostic interview guide were completed before the study. Descriptive statistics were computed per patient group as P-ICD-positive or P-ICD-negative. Internal consistency of the P-ICD-10 was assessed by computing Cronbach's alpha coefficient. The correlations of each question were examined, and the values of the alpha coefficient showed medium strength correlations (ranging from 0.11 to 0.3). These indicated the instrument is a reliable tool for measuring fatigue.

Demographic and disease-related characteristics of the patients are presented in Table 3. Data on the 265 patients, of whom 64% were women, were collected between 21 October 2008 and 28 October 2009 at two Taipei clinical centers (Mackay Memorial Hospital, n = 200, Cardinal Tien Hospital, n = 65). Of the 265 patients, 228 (86%) answered "yes" to the first question of the questionnaire, which corresponds to criterion A1 of the P-ICD criteria of CRF, and requires a person to report a minimum of 2 weeks of fatigue in the past month. Criterion A specifies that six or more symptoms must have been resent every day or nearly every day during the same 2-week period of time over the past month and at least one symptom must be significant fatigue. Table 4 lists individually by symptom, the number and the percentages from criteria A1 to A11, which corresponds to questions number 1 to 11, respectively. Criterion B specifies that the fatigue symptoms must cause clinically significant distress or impairment in social, occupational, or other important areas of functioning. Thirty patients (11%) had six or more of the symptoms in criterion A but responded that the symptoms did not cause significant distress or impairment in social, occupational, or other important areas of functioning (Table 5). The part in Table 4 in bold fonts (A1 + any 5 + B) shows that criteria A and B of the P-ICD-10 diagnostic criteria were met by 28 (11%) patients. This result is compared with a western study of 375 patients conducted by Cella D, Davis K, Breitbart W, et al. showing that 17% of patient respondents met both Criteria A and B [11]. In addition, of the 265 patients, 132 (49.8%) were classified as fatigue-positive according to the P-ICD-10. 66 subjects were more than 70 years old (24.9%), 61 subjects were between 61 to 70 years old (23%), 81 subjects were between 51 to 60 years old (30.5%), 37 subjects were between 41 to 50 years old (13.9%), 14 subjects were between 31 to 40 years old (5.2%), and 3 subjects were below 30 years old (1/1%). 15% of the patients did not receive education, and 22% had some college or higher education. The different age groups and education levels were not statistically significant if they were classified as P-ICD 10-positive and P-ICD 10-negative. The participants were diagnosed with various types of cancer, including head and neck (7.9%), lung and mediastinum (10.5%), breast (36.6%), liver (3.7%), GI tract (16.6%), GU tract (3.3%), GYN (6.4%), prostate (3.3%), hematological malignancy (0.3%), and others (8.6%). Notably, the breast (p = 0.033) and GI tract (0.035) were statistically significant if they were classified as P-ICD 10-positie and P-ICD 10-negative (p < 0.05). Among the 132 patients who were classified as fatigue-positive according to the P-ICD-10, 77 patients are under palliative treatment, 18 patients received curative treatment, and 35 patientswere follow up after curative therapy. The group under palliative treatment and curative treatment were statistically significant if they were classified as P-ICD 10-positive and P-ICD 10-negative (p < 0.05). In addition, regarding the staging of the disease, the group under no signs and symptoms and tumor progression were statistically significant if they were classified as P-ICD 10-positive and P-ICD 10-negative (p < 0.005).

Occurrences of each P-ICD-10 fatigue symptom in all patients, and patients classified as P-ICD 10-positive (N = 129) and P-ICD 10-negative (N = 380) are listed in Table 6.

According to the design of the questionnaire, the 1^st question, which corresponds to the A1 criteria of the P-ICD-10 criteria, should be answered "yes" in order for the questionnaire to proceed, thereby resulting in the non-discriminating property of this general symptom. Of the 10 other complaints, the most frequently reported by both P-ICD 10-positive and P-ICD 10-negative patients were "general weakness", "insomnia or hypersomnia", "experience of sleep as unrefreshing or non-restorative", and "perceived problems with short-term memory". The most discriminative complaints were related to physical fatigue, and they are listed according to the order of their discriminative properties: "Difficulty completing daily tasks attributed to feeling fatigued" (P-ICD 10-positive 76%, negative 19.9%), "Perceived need to struggle to overcome inactivity" (P-ICD 10-positive 70.5%, negative 16.2%), and "Post-exertional malaise lasting several hours" (P-ICD 10-positive 72.9%, negative 21.3%). The least discriminating of the complaints was about sleep quality, which is the 5^th question on the questionnaire: "insomnia or hypersonmia".

Internal consistency was established by calculating Cronbach alpha coefficients, which were 0.843, indicating good internal consistency for the Chinese-language version of the P-ICD 10 criteria. Removing non-reliable items would increase internal consistency. The alpha coefficient did not increase after removing any individual item, removing any item decreased the internal consistency, and the most reliable items, aside from the 1^st symptom, were the 7^th, 9^th, and 2^nd symptom, respectively.

To guarantee cultural equivalence and to establish expert validity, 7 authoritative specialists in the field of nursing, oncology, radio-oncology, psychiatry, and linguistics were formed to examine the Chinese-language version of the P-ICD 10 criteria. The experts were asked to rate the relevance of the content independently using a content validity index (CVI). They were asked to rate each item on the Chinese-language versions of the diagnostic interview guide based on relevance and semantic equivalence using the following 4-point Likert scale: 1 = not relevant (not appropriate), 2 = somewhat relevant (somewhat appropriate), 3 = relevant (quite appropriate), 4 = very relevant (very appropriate). The CVI was then computed based on the percentage of total items rated by the experts as either 3 or 4. According the Norwood(2000), a CVI rating exceeding .80 can be considered to show good content validity. All items in the Chinese-language versions of the diagnostic interview guide were rated by the panel experts as having a CVI greater than .90. For the final version, the CVIs for relevance and semantic equivalence both reached .99.