SEARCH

SEARCH BY CITATION

Keywords:

  • symptoms;
  • assessment tool;
  • validation;
  • factor analysis;
  • Chinese;
  • cancer

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

BACKGROUND

Symptom management is an essential component of cancer treatment for patients of every culture and nationality. Symptom assessment depends on subjective reporting, mandating linguistically equivalent versions of symptom assessment scales. Because disease-related and treatment-related symptoms most often occur in clusters, there is a global need for a standardized multiple-symptom assessment tool.

METHODS

The authors sought to validate the Chinese version of the M. D. Anderson Symptom Inventory (MDASI-C) by enrolling patients who had received various diagnoses of and different types of treatment for cancer (n = 249) in a cross-sectional symptom study conducted at an urban cancer center in China.

RESULTS

Factor analysis identified 2 underlying constructs, general symptoms and gastrointestinal symptoms, which had Cronbach alpha coefficients of 0.86 and 0.84, respectively. These results were consistent with English- and Japanese-language MDASI validation studies. Known-group validity was supported by the MDASI-C's ability to detect significant differences in symptom and interference levels according to Eastern Cooperative Oncology Group performance status (ECOG PS; P < 0.001) and chemotherapy status (P < 0.05). Fifty-five percent of the study cohort had ≥ 1 symptom that was considered severe (score ≥ 7 on a 0–10 scale). ECOG PS was strongly associated with symptom burden (total interference score: R2 = 0.26; P < 0.001). Fatigue, sadness, drowsiness, and lack of appetite accounted for most of the variability in the total interference score (R2 = 0.49; P < 0.05).

CONCLUSIONS

The authors demonstrated that the MDASI-C is a valid, reliable, and concise tool for measuring symptom severity and interference with functioning in Chinese cancer patients. Cancer 2004. © 2004 American Cancer Society.

Across all cultures and nationalities, symptom management for patients undergoing treatment or follow-up is an essential component of cancer care. The physical, cognitive, and affective symptoms caused by cancer and its treatments can be severe, and they rarely occur singly. These symptoms greatly influence a patient's functional status and may even cause the patient to change or abandon an active treatment plan.1–3 Oncology clinicians and researchers, therefore, have a critical need for a standardized tool to assess multiple symptoms. Although a multiple-symptom assessment tool may overlap with other measures (such as single-symptom pain and fatigue assessment tools and certain symptom subscales in health-related quality of life [QOL] measures), multisymptom scales nonetheless provide a rapid assessment method for monitoring and evaluating the effectiveness of treatments for cancer and its symptoms and for documenting treatment side effects observed in research and in clinical practice. Because symptom ratings depend on subjective reporting, instruments must be validated in the language most familiar to the patient.

Chinese is the most commonly spoken language in the world, and persons born in Chinese-speaking countries represent the second largest immigrant group in the United States.4 Nonetheless, no concise, standardized, and valid Chinese-language assessment tool has been available for multisymptom assessment of Chinese-speaking oncology patients in clinical settings, clinical trials, and epidemiologic studies. Furthermore, the crosscultural adaptability of an assessment tool is determined not only by the consistency of its psychometric properties but also by whether it successfully handles the linguistic variations that may exist among disparate groups. Because of the worldwide prevalence of the Chinese and English languages, a validated Chinese-language, patient-reported outcome measure that demonstrates a high degree of equivalency with English-based symptom-rating scales would have enormous global clinical and research utility.

When the M. D. Anderson Symptom Inventory (MDASI; Fig. 1) was introduced in 2000,1 several multisymptom assessment scales, including the Symptom Distress Scale,5 the Memorial Symptom Assessment Scale,6 the Rotterdam Symptom Checklist,7 and the Edmonton Symptom Assessment System,8 were already in use among patients with cancer in Western countries. The MDASI, however, has several advantages over these other tools. First, the MDASI assesses a concise but adequate set of symptoms that are applicable to most patients with cancer. Second, it asks patients to rate how severely their symptoms interfere with physical and affective functional domains, laying the groundwork for categorizing levels of symptom distress and judging symptom burden. Third, its 11-point rating system (a 0–10 numeric scale) is easily translated into other languages and is readily understood by patients with lower levels of education. These advantages make the MDASI an excellent candidate for foreign language translation and validation.

thumbnail image

Figure 1. English-language version of the M. D. Anderson Symptom Inventory.

Download figure to PowerPoint

Although cancer is a major cause of morbidity and mortality in China,9 little is known about the status of cancer-related symptom management in that country. Thus, the primary objective of the current study was to establish and validate a Chinese version of the M. D. Anderson Symptom Inventory (MDASI-C) for patients with cancer who reside in China. This would allow the measurement of symptom prevalence and symptom burden among Chinese patients with cancer and help to document the effectiveness of various treatments and educational programs. Once established in China, the MDASI-C can be used in conjunction with the already validated MDASI (U.S. version) and the native Japanese version of the MDASI (MDASI-J) to make crosscultural comparisons. The secondary goals of the current study were to examine the severity of symptoms caused by the most commonly occurring malignancies in China and to characterize symptom burden in the study cohort.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Patients and Data Collection

The current study was approved by the institutional review boards of the Tianjin Medical University–affiliated Tianjin Cancer Hospital (Tianjin, China) and by The University of Texas M. D. Anderson Cancer Center (Houston, TX). Eligible patients were 18 years of age or older and had a pathologic diagnosis of cancer; in addition, eligible patients were undergoing either active treatment or clinical observation on the day of enrollment. Patients were excluded if, in their physicians' estimation, they could not understand the intent of the study, if they did not elect to participate, or if they had an existing major psychiatric illness. Two hundred forty-nine patients with cancer were consecutively recruited from more than 10 clinical departments at Tianjin Cancer Hospital. All patients who were approached agreed to participate in the study. A study nurse asked patients to complete self-administered questionnaires, answered questions, and assisted with the completion of survey forms as necessary. Patients were able to complete the survey in approximately 20 minutes.

Measurements

The MDASI1 has been established as a valid and reliable tool for assessing cancer-related symptoms, regardless of therapy or specific diagnosis. It contains 13 core symptom severity items (fatigue, sleep disturbance, pain, drowsiness, poor appetite, nausea, vomiting, shortness of breath, numbness, difficulty remembering, dry mouth, distress, and sadness) and 6 symptom interference items (general activity, mood, work [both inside and outside the home], relations with other people, walking, and enjoyment of life). The underlying constructs include a general symptom severity factor and a gastrointestinal symptom severity factor. The internal consistency (reliability) of the symptom scales and the symptom interference scale is high (0.82–0.94). The known-group validity of these items was examined by testing groups stratified according to disease severity and treatment status.

The MDASI-C (Fig. 2) was developed using the standard translation/back-translation procedure10 that was used to create other validated versions of the MDASI. First, a physician who spoke both English and Chinese translated MDASI items as simply as possible into Chinese characters. A second translator who had not seen the original English items then back-translated the Chinese translation into English. Bilingual fluency was required of both translators. The items that had been back-translated into English were then compared with the original items. If the back-translated items and the original items did not agree, the first translator offered a second translation after comparing the original items and their back-translations, and a second back-translation was generated and compared with the original items once more. This translation/back-translation procedure was repeated until the translation was judged to be satisfactory.

thumbnail image

Figure 2. Chinese-language version of the M. D. Anderson Symptom Inventory.

Download figure to PowerPoint

The MOS 36-Item Short-Form Health Survey (SF-36) is a comprehensive, 36-question form that yields an 8-scale health profile along with summary measures of health-related QOL (physical and mental health component scores).11 Its Chinese version (SF-36-C) has been validated in other symptom studies in China.12, 13

The Questionnaire for Demographic Information was used to collect patient information. For each patient, the study coordinator completed a ‘Clinician Checklist’, which was used to record current disease information (including tumor site and stage and the presence of metastatic disease) and treatment information (e.g., whether the patient was currently receiving treatment or had received treatment within the previous month) based on the patient's medical chart. The patient's functional status was evaluated using the Eastern Cooperative Oncology Group performance status (ECOG PS) scale.14

Statistical Analysis

Following the method used in the original English-language MDASI validation study, we examined symptom severity and symptom interference items separately in the factor analysis. We used principal axis factoring with direct oblimin rotation to examine the construct validity of the MDASI-C15, 16 and the Harman rule17 to assess the fit of our model. The Harman rule states that model fit is acceptable if the standard deviation (SD) of the residuals derived from the fitted model is less than or equal to the reciprocal of the square root of the sample size. We considered scree plots, number of factors with eigenvalues ≥ 1.0, model fit, and clinical interpretability in deciding how many underlying constructs existed. After we identified the number of symptom severity subscales, we calculated separate Cronbach alpha coefficients for the symptom severity and symptom interference subscales.

We tested the known-group validity of the MDASI-C by categorizing patients according to their ECOG PS and chemotherapy status, using a global symptom score (mean of the 13 symptom items) and an interference component score (mean of the 6 interference items). We also assessed symptom severity scores for certain single items (pain, fatigue, nausea, vomiting, and lack of appetite) according to stratified ECOG PS and chemotherapy status to test known-group validity, because we expected to find higher levels of severe symptoms among patients who had more severe illness and patients who were undergoing chemotherapy. Mean differences, 95% confidence intervals (CI), and significance levels (evaluated using independent-sample t tests) are reported.

We tested convergent validity by calculating Spearman rank correlation coefficients for associations between composite interference scores or individual MDASI-C items and the eight domains in the SF-36-C. Based on our research on the impact of cancer-related pain in a four-country sample,18 we grouped interference items to create WAW (a composite of work, activity, and walking, representing the physical functioning domain) and REM (a composite of relations with other people, enjoyment of life, and mood, representing the mental health and social functioning domains) scores. Likewise, because of their conceptual similarity, we compared MDASI symptom items with SF-36 domain items as follows: pain with bodily pain subscale, fatigue with vitality subscale, distress with mental health subscale, and sadness with mental health subscale. Items were paired on the basis of previous knowledge so that the number of comparisons involving these pairs could be limited and Type I error rates could therefore be better controlled.

Descriptive statistics (mean values, SDs, and 95% CIs) were used to determine the prevalence of severe symptoms and their impact on functionality in Chinese patients with cancer. Chi-square and Mann–Whitney U tests were performed to compare percentages of symptoms considered to be severe among patients with the three most common types of cancer. Based on previous pain-related research,19, 20 we used a provisional criterion of score ≥ 7 on a 0–10 scale to define a symptom as being severe.

Symptom burden21 was examined by assessing the predictive value of the mean MDASI-C interference component score on linear regression analysis. We conducted exploratory univariate analyses to obtain candidate variables for multivariate regression models and used one-way analysis of variance to identify possible predictors among categoric variables. Correlation coefficients were used for continuous, ordinal, and binary variables. A predictor was considered to be a candidate if it exhibited at least a marginal association (P ≤ 0.25 and/or r ≥ 0.25) or if it possessed clinical significance. We decreased the cutoff P value from 0.25 to 0.05 in the multivariate regression model so that strong predictors would not be inadvertently excluded. We addressed concerns regarding collinearity by examining the tolerance of each independent variable (i.e., 1 − the squared multiple correlation of that variable with all other variables in the model), which was expected to be far from the lower limit of 0.02.22 We then conducted multivariate regression analyses (using forward and stepwise methods) of clinical models and a symptom model and performed residual diagnostic tests to evaluate the appropriateness of the selected linear regression models.

All statistical procedures were performed using SPSS software (SPSS Inc., Chicaco, IL).23 All reported P values are two tailed.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Patient Characteristics

Table 1 presents selected demographic and disease-related characteristics for the study cohort (N = 249). The majority of patients were married (86%) and had a high school education or less (59%). Thirty-nine percent of patients were retired, and 36% were working full-time or part-time. Although 48% of patients had metastatic disease, 58% had a favorable performance status (ECOG PS 0 or 1). Opioid analgesics had been prescribed to 19% of patients for pain management. Twenty-two percent experienced infection, 42% used antibiotics or antifungal medication, and 30% used antiemetic agents. One case of missing data regarding worst level of fatigue and 1 case of missing data regarding worst level of lack of appetite yielded a missing data rate of 0.04% out of a total of 4731 data points (19 items completed by 249 patients).

Table 1. Sample Demographic and Disease Characteristics (N = 249)
Patient characteristicNo. of patients (%)
  • a

    Treatment information was unavailable for one patient.

Median age in yrs (range)51 (18–77)
Gender 
 Female134 (54)
 Male115 (46)
Diagnosis 
 Gastrointestinal62 (25)
 Breast60 (24)
 Lung51 (21)
 Head and neck15 (6)
 Non-Hodgkin lymphoma10 (4)
 Sarcoma10 (4)
 Gynecologic9 (4)
 Genitourinary9 (4)
 Thyroid9 (4)
 Other14 (6)
Stage 
 I53 (21)
 II73 (29)
 III71 (29)
 IV52 (21)
Treatmenta 
 No treatment, but underwent follow-up25 (10)
 Chemotherapy87 (35)
 Surgery53 (21)
 Radiotherapy29 (12)
 Chemotherapy and surgery38 (15)
 Radiotherapy, chemotherapy, and surgery9 (4)
 Biotherapy and immunotherapy7 (3)

Factor Analysis and Internal Consistency

Principal axis factor analysis generated a two-factor solution (general and gastrointestinal symptoms) for symptom severity items in the study cohort. The SD of the residuals of the two-factor solution was 0.061 (which was < 0.063, the reciprocal of the square root of the cohort size [N = 249]), indicating good fit. This two-factor solution is consistent with the psychometric properties of the original MDASI and the MDASI-J1, 24 (Table 2). Although 3 factors had eigenvalues ≥ 1.0, accounting for 57.5% of the variation in the current analysis (initial eigenvalues were 5.1, 1.3, and 1.1 for the first 3 factors, accounting for 39.4%, 9.8%, and 8.2% of the total variance, respectively), we did not use a 3-factor solution, because it did not satisfy the Harman rule and was not clinically interpretable.

Table 2. Comparison of Factor Analyses for the Chinese, English, and Japanese Versions of the M. D. Anderson Symptom Inventory
Symptom itemChinese version (n = 249)English version (n = 527)Japanese version (n = 252)
Factor 1Factor 2Factor 1Factor 2Factor 1Factor 2
Distress0.83−0.140.78−0.070.820.05
Fatigue0.81−0.130.710.070.850.06
Sadness0.67−0.030.69−0.080.67−0.04
Sleep disturbance0.64−0.020.560.020.770.03
Drowsiness0.60−0.040.650.070.65−0.03
Difficulty remembering0.52−0.040.66−0.110.660.02
Poor appetite0.480.210.360.390.47−0.36
Numbness0.460.150.490.030.650.06
Pain0.450.120.520.120.69−0.03
Dry mouth0.450.120.450.120.56−0.05
Shortness of breath0.420.050.53−0.030.56−0.10
Nausea0.020.910.090.810.03−0.88
Vomiting0.080.74−0.070.85−0.02−0.96

A high degree of internal consistency among the symptom severity factors and the symptom interference factors was demonstrated in the Chinese study sample (Table 2). The Cronbach alpha coefficient was 0.86 for general symptom factors, 0.84 for gastrointestinal factors, and approximately 0.90 for symptom interference factors. When we deleted each symptom item individually and recalculated the alpha coefficient, this coefficient remained relatively unchanged, indicating that each symptom made a contribution and should remain in the group. The Cronbach alpha coefficients for the MDASI-C were comparable to those calculated for other versions of the MDASI (Table 3).

Table 3. Results of Descriptive and Reliability Analyses Performed in China, the United States, and Japan
Symptom itemCronbach alpha coefficient
ChinaUnited StatesJapan
  • a

    Cronbach alpha coefficient for subscale. All other coefficients are Cronbach alpha if symptom item is deleted.

13 symptoms0.87a0.87a0.92a
6 interference items0.90a0.91a0.93a
General symptom subscale0.86a0.85a0.91a
Gastrointestinal symptom subscale0.84a0.82a0.92a
Pain0.850.850.90
Fatigue0.830.840.89
Sleep disturbance0.840.850.90
Distress0.830.840.89
Shortness of breath0.850.850.90
Difficulty remembering0.850.850.90
Poor appetite0.840.850.90
Drowsiness0.840.840.90
Dry mouth0.850.860.90
Sadness0.840.850.90
Numbness0.850.860.90
Nausea
Vomiting

Known-Group Validity

The MDASI-C was sensitive enough to detect different levels of symptom severity and symptom interference with patient functioning. Patients with poorer ECOG PS (2–4) reported significantly more severe pain and fatigue on the MDASI-C and had poorer symptom and interference composite scores (P < 0.001; Table 4). Comparing patients who received chemotherapy with patients who were not treated, the mean composite scores for gastrointestinal symptom items (nausea and vomiting) differed significantly (2.0 vs. 0.9, respectively; P < 0.01). The mean lack of appetite score also was higher for patients receiving chemotherapy than for patients who were not treated (3.7 vs. 2.2, respectively; P < 0.05), and patients in the former group reported significantly greater symptom severity and interference according to their composite scores compared with patients in the latter group (P < 0.05; Table 5).

Table 4. Means and Standard Deviations of MDASI-C Scores According to ECOG Performance Status (N = 249)
MDASI-C measurePossible rangeMean score (standard deviation)Mean difference (95% CI)
Better ECOG performance status (0–1) (n = 145)Poorer ECOG performance status (2–4) (n = 103)
  • MDASI-C: Chinese version of the M. D. Anderson Symptom Inventory; ECOG: Eastern Cooperative Oncology Group; CI: confidence interval.

  • a

    P < 0.001.

Paina0–102.3 (2.4)4.4 (2.9)−2.0 (−2.7, −1.4)
Fatiguea0–103.6 (2.5)6.0 (2.7)−2.4 (−3.0, −1.7)
13 symptom itemsa0–102.2 (1.5)3.6 (1.7)−1.4 (−1.8, −1.0)
6 interference itemsa0–102.0 (2.0)4.7 (2.4)−2.8 (−3.3, −2.2)
Table 5. Means and Standard Deviations of MDASI-C Scores According to Chemotherapy Status (N = 249)
MDASI-C measurePossible rangeMean score (standard deviation)Mean difference (95% CI)
No treatment, but follow-up (n = 25)Chemotherapy (n = 87)
  • MDASI-C: Chinese version of the M. D. Anderson Symptom Inventory; CI: confidence interval.

  • a

    P < 0.01.

  • b

    P < 0.05.

Mean of gastrointestinal items (nausea and vomiting)a0–100.9 (1.4)2.0 (2.2)−1.14 (−2.1, −0.2)
Poor appetiteb0–102.2 (2.5)3.7 (3.0)−1.5 (−2.8, 0.2)
13 symptom itemsb0–102.1 (1.4)3.0 (1.7)−0.9 (−1.6, −0.1)
6 interference itemsb0–102.7 (2.6)3.3 (2.8)−0.6 (−1.8, 0.5)

Convergent Validity

MDASI-C WAW interference items were strongly correlated with the SF-36-C physical functioning subscale (r = −0.73; P < 0.001), and REM score was closely correlated with the SF-36-C mental health (r = −0.60; P < 0.001) and social functioning subscales (r = −0.60; P < 0.001). Similarly, correlation coefficients were moderately high for MDASI-C item/SF-36-C domain pairs, such as the pain/bodily pain subscale pair (r = −0.51; P < 0.001), the fatigue/vitality subscale pair (r = −0.51; P < 0.001), the distress/mental health subscale pair (r = −0.59; P < 0.001), and the sadness/mental health subscale pair (r = −0.59; P < 0.001). WAW score was more strongly correlated with the physical component score of the SF-36-C than was REM score (r = −0.69 vs. r = −0.44; P < 0.05), whereas REM score was more strongly correlated with the mental health component score than was WAW score (r = −0.59 vs. r = −0.38; P < 0.05).

Prevalence of Severe Symptoms

Table 6 presents a comparison of mean values and SDs for all MDASI-C items across the three most common diagnosis groups and across five different treatment groups. Overall, fatigue, sleep disturbance, distress, and pain were reported to be the four most severe symptoms. Fatigue and sleep disturbance were the primary symptoms for patients with breast and lung cancer, whereas fatigue and lack of appetite were the worst symptoms for patients with gastrointestinal cancer. Pain was more severe among patients with gastrointestinal or lung cancer than among patients with breast cancer. Sadness was ranked among the four worst symptoms only by patients with breast cancer. Distress was ranked third by breast cancer patients and fourth by lung cancer patients.

Table 6. Means and Standard Deviations of Symptom and Interference Scores According to Diagnosis and Treatment
MDASI-C itemDisease typeTreatment typeTotal (N = 249)
Breast (n = 60)Gastrointestinal (n = 62)Lung (n = 51)No treatment (n = 25)Surgery (n = 53)Chemotherapy + surgery (n = 38)Chemotherapy (n = 87)Radiotherapy (n = 29)
  1. MDASI-C: Chinese version of the M. D. Anderson Symptom Inventory.

Fatigue4.1 (2.8)4.7 (2.7)4.8 (3.4)3.7 (2.9)5.4 (2.9)4.3 (2.7)4.7 (2.9)4.2 (2.7)4.6 (2.9)
Sleep disturbance4.2 (3.0)3.3 (2.8)4.3 (3.3)2.8 (2.6)4.2 (2.7)4.4 (3.0)3.9 (2.7)3.6 (3.4)3.9 (2.9)
Distress3.4 (2.8)2.8 (3.1)3.6 (3.0)3.1 (3.2)3.2 (3.1)2.8 (2.8)4.1 (3.1)3.2 (3.1)3.5 (3.1)
Pain2.4 (2.4)3.3 (2.9)3.9 (3.2)2.1 (2.6)4.3 (3.1)3.5 (2.9)3.0 (2.5)2.6 (2.7)3.2 (2.8)
Poor appetite2.2 (2.7)3.6 (2.7)3.6 (3.2)2.2 (2.5)3.0 (2.7)3.1 (3.0)3.7 (3.0)3.2 (2.8)3.1 (2.9)
Dry mouth2.6 (2.8)2.9 (2.6)3.3 (2.9)2.9 (2.9)2.8 (2.9)2.9 (2.7)3.4 (3.2)3.1 (2.7)3.1 (2.9)
Sadness3.2 (3.3)2.5 (2.8)2.8 (3.2)2.8 (2.9)2.7 (3.0)2.2 (2.8)3.1 (3.2)3.2 (3.0)2.9 (3.0)
Drowsiness2.0 (2.4)2.7 (2.7)2.7 (2.7)2.2 (2.6)2.8 (2.8)2.2 (2.4)2.8 (2.8)2.0 (2.6)2.5 (2.7)
Shortness of breath2.0 (2.2)2.0 (2.6)3.4 (3.2)1.3 (1.9)2.2 (2.3)2.6 (2.7)2.3 (3.0)2.4 (2.8)2.3 (2.7)
Nausea1.8 (2.5)2.0 (2.7)2.3 (2.6)1.0 (1.8)1.8 (2.6)2.8 (3.2)2.3 (2.5)1.7 (1.8)2.1 (2.5)
Difficulty remembering2.2 (2.2)1.7 (2.1)1.9 (2.3)1.6 (1.9)1.5 (2.0)2.0 (2.2)2.2 (2.5)1.9 (2.2)2.0 (2.3)
Numbness1.5 (2.4)1.3 (1.9)1.7 (2.6)1.1 (1.7)1.8 (2.4)1.9 (2.5)1.5 (2.5)0.9 (1.4)1.6 (2.3)
Vomiting1.3 (2.3)1.7 (2.6)1.6 (2.5)0.8 (1.3)1.5 (2.5)1.9 (2.8)1.8 (2.4)0.9 (1.7)1.5 (2.4)
Work3.9 (3.6)5.3 (3.6)4.8 (3.7)3.3 (3.7)5.4 (3.6)4.4 (3.6)4.7 (3.6)4.9 (3.7)4.7 (3.6)
Enjoyment of life2.6 (2.8)3.1 (3.1)3.5 (3.3)3.0 (3.3)2.8 (3.1)2.4 (2.6)3.5 (3.4)4.1 (3.3)3.1 (3.2)
Mood2.9 (2.7)2.5 (2.7)3.1 (3.0)3.6 (3.3)2.7 (2.5)2.6 (2.4)3.1 (3.1)2.9 (2.9)3.0 (2.8)
Walking2.1 (2.8)3.2 (3.0)3.6 (3.6)1.8 (2.3)3.4 (3.0)2.8 (3.0)3.3 (3.5)2.1 (2.9)2.9 (3.2)
General activity2.1 (2.8)3.0 (2.9)3.1 (3.0)2.5 (3.0)3.1 (2.8)2.8 (2.7)3.1 (3.4)2.2 (3.0)2.8 (3.1)
Relations with other people1.7 (2.4)2.6 (2.8)2.3 (2.7)2.0 (2.7)2.6 (3.0)1.7 (2.3)2.3 (2.9)2.6 (3.1)2.3 (2.8)

Based on pain-related research,19, 20 we provisionally used a score range of 5–6 on the MDASI-C to define moderate symptoms, while scores ≥ 7 were considered indicative of severe symptoms. We found that 55% of the study cohort reported scores ≥ 7 for 1 or more symptoms, indicating that those symptoms were severe enough to interfere significantly with daily life. Of these patients, 14.5% had 1 symptom with a score ≥ 7, 10% had 2 symptoms, and another 10% had 3 symptoms. In addition, nearly 15% had 4–6 symptoms with score ≥ 7, and approximately 5% had 7–11 symptoms with score ≥ 7. Thus, approximately 40% of patients reported scores ≥ 7 for at least 2 symptoms.

At least 30% of lung cancer patients ranked 7 symptoms as being moderate to severe (score 5–10), whereas ≥ 30% of gastrointestinal or breast cancer patients reported only 4–5 symptoms as being moderate to severe. Figure 3 lists percentages of patients who rated symptoms as being severe (score 7–10) for the 3 most common types of cancer in the current study. Generally speaking, patients with lung cancer had decidedly greater symptom distress than did patients with gastrointestinal or breast cancer. Patients with breast cancer were more likely to experience sleep disturbance and to be sad. Patients with lung cancer reported significant levels (P < 0.05) of shortness of breath.

thumbnail image

Figure 3. Prevalence of severe symptoms (score ≥ 7 on the Chinese version of the M. D. Anderson Symptom Inventory) according to disease type for patients with the 3 most common types of cancer in the current study. GI: gastrointestinal.

Download figure to PowerPoint

Symptom Burden—Predictors of Symptom Interference

Symptom interference as assessed by the MDASI-C is a measurement of the impact of multiple symptoms on patient functioning. In the Chinese cohort, the mean score for the 6 interference items was 3.1, with an SD of 2.5. The highest level of interference was reported for work, followed by enjoyment of life, mood, walking ability, general activity, and relations with others (Table 6).

To verify the utility of the interference component score as one method of representing symptom burden21 in the current cohort, we performed linear regression analyses to examine the predictive value of clinical variables and cancer-related symptoms on total symptom-related interference (Table 7). The dependent variable was the interference component score (i.e., the mean of 6 interference items scored on a 0–10 scale). The 17 independent biomedical variables examined included age, gender, marital status, educational level, job status, diagnosis, disease stage (0-IV), disease status, ECOG PS, presence of infection, treatment (radiotherapy, chemotherapy, or surgery), use of opioids to treat pain, use of an antiemetic, hemoglobin level, and albumin level. Malignancy-related symptom severity data were provided by the 13 core items on the MDASI-C.

Table 7. Linear Regression Models: Predictors of Mean MDASI-C Interference Component Score
ModelIndependent variablesaPredictors of severe interferenceUnstandardized coefficients (95% CIs)Standardized coefficients
  • MDASI-C: Chinese Version of the M. D. Anderson Symptom Inventory; CIs: confidence intervals; ECOG PS: Eastern Cooperative Oncology Group performance status.

  • a

    Dependent variable was the mean score for the 6 interference items on the MDASI-C.

  • b

    0-IV.

Clinical model (demographic data and data from clinical checklist)
 Model 1 (R2 = 0.26; n = 86; P < 0.001)Age, gender, ECOG PS, disease stage,b disease status, use of opioids for pain, presence of infection, hemoglobin level, diagnosis (three most common types)1) ECOG PS (0–1 vs. 2–4)2.43 (1.53–3.33)0.51
 Model 2 (R2 = 0.05; n = 121; P < 0.05)Age, gender, disease stage,b disease status, use of opioids for pain, presence of infection, hemoglobin level, diagnosis (three most common types)1) Disease stage (I, II, III, IV)0.55 (0.13–0.97)0.23
Multiple symptom model (MDASI-C)
 Model 3 (R2 = 0.49; n = 247)Pain, fatigue, sleep disturbance, distress, difficulty remembering, poor appetite, drowsiness, dry mouth, sadness, numbness1) Fatigue score (0–10) (P < 0.001)0.28 (0.18–0.38)0.31
2) Sadness score (0–10) (P < 0.001)0.27 (0.18–0.36)0.32
  3) Drowsiness score (0–10) (P < 0.01)0.15 (0.05–0.25)0.16
  4) Poor appetite score (0–10) (P < 0.05)0.12 (0.02–0.21)0.13
Univariate analysis

After screening independent variables in a univariate analysis, we identified 9 of the 17 biomedical variables and 10 of the 13 symptom variables from the MDASI-C as candidates for multivariate regression analysis (Table 7). These candidates met 1 of 2 criteria; ECOG PS and diagnosis met the first criterion (P ≤ 0.25 and/or r ≥ 0.25), and 7 other variables met the second criterion (clinical importance). Ten symptom variables were significantly correlated with the MDASI-C interference component score (r = 0.34–0.59; P < 0.001).

Multivariate model

Poor ECOG PS was a significant predictor of the interference component score in the clinical model (Model 1; Table 7). However, we were interested in identifying other significant predictors when ECOG PS was excluded from the pool of candidate variables because of its strong correlation with the interference component score and its possible confounding effects. Advanced-stage disease was found to be a second significant predictor of the interference component score (Model 2). Finally, increased fatigue, sadness, drowsiness, and lack of appetite were significant predictors of, and accounted for 49% of, the interference component score in the symptom model (Model 3).

Tolerances for all independent variables ranged from 0.65 to 0.77. Thus, multiple collinearity issues did not exist in the models that were used.22 Standardized residuals in the three models were normally distributed, or else there was no discernible pattern of correspondence between the residuals and predicted values; these features were consistent with the assumptions of the linear regression model.

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

As evidenced by its satisfactory construct validity, known-group validity, convergent validity, and reliability, the MDASI-C is a simple, accurate, valid, and reliable tool for measuring multisymptom severity and the related impact on patient functioning. Patients with cancer often are not well enough to complete long or complicated questionnaires. The negligible incidence of missing data (0.04%) in the current study indicates that the MDASI-C is not burdensome for Chinese cancer patients (including patients with severe symptoms) to use.

We also documented the most common severe symptoms and the interference caused by these symptoms in Chinese patients with the most commonly occurring types of cancer. More than one-half (55%) of all patients reported having more than 1 severe symptom. Led by fatigue, multiple physical and psychologic symptoms (such as sleep disturbance, distress, and pain) interact to create a further burden for patients who are already striving for cure and longer survival. This symptom burden, whether due to the cancer itself or to treatment, severely reduces a patient's ability to function in daily life (with the greatest impact seen in the areas of work and enjoyment of life).

Proper documentation of the research and clinical findings related to multiple cancer-related symptoms requires a valid and reliable instrument for measuring symptoms and their concurrent effects on patients. Lack of such an instrument impedes successful symptom research and improvements in routine symptom management during cancer care.2, 25 The MDASI-C, the first validated multisymptom assessment tool in China, can be applied in both clinical practice and clinical trials to evaluate the effectiveness of treatment interventions and in epidemiologic studies to estimate the prevalence of symptoms. With psychometric properties comparable to those of the MDASI-J and the original MDASI, the MDASI-C should also be widely applicable in the United States, where Chinese-speaking persons are the second largest foreign-born resident group.4

Current treatments and new therapies derived from clinical trials are often evaluated, in part, on the basis of how effectively they reduce the number of symptoms experienced by patients. Despite the brevity of the MDASI-C, its 13 items form a relatively comprehensive list of the symptoms commonly experienced by Chinese patients with cancer, regardless of disease site or stage. Its 0–10 numeric scale is simple, familiar, and, therefore, effective in describing symptom severity and symptom interference in real time. The advantages of such a scale have been demonstrated in previous validation studies of the Brief Pain Inventory10, 26 and in comparisons made across culturally and linguistically varied groups in several countries.18 Also, the 0–10 scale of the MDASI-C is easily used in conjunction with modern technology, such as computer-aided interactive voice telephone systems, electronic mail, and the Internet, and this feature gives both hospitals and patients the flexibility to communicate efficiently. Furthermore, preliminary multisymptom data obtained from the current cohort can serve as a baseline for future comparisons of cancer-related symptoms among patients in the same hospital or among other Chinese patients.

The current study had certain limitations. We did not address the test-retest reliability of the MDASI-C, leaving this issue for a future longitudinal study. We tested only 4 of 13 symptoms (pain, fatigue, distress, and sadness) for concurrent validity by analyzing correlations with SF-36-C subscales. There was no other gold-standard tool in China, however, with which to measure symptom severity and its impact on the functional ability of patients with cancer. In addition, the categories we used to delineate symptom severity require verification. The use of score ≥ 7 as the defining criterion for severe symptoms can be traced to our previous crosscultural studies on cancer-related pain; those studies suggested that on a 0–10 scoring scale, 5–6 was the optimum range for defining moderate symptoms and 7–10 was the optimum range for defining severe symptoms.19, 20 This categorization has proven clinically useful in assessing the varying levels of pain and fatigue that patients experience, and this assessment is critical for managing symptoms and reducing levels of interference with functioning. Nonetheless, further research is needed to define moderate and severe levels for other symptoms experienced by Chinese-speaking cancer patients. Finally, because we only enrolled Chinese-speaking patients living in mainland China, the psychometric properties of the MDASI-C will require validation in symptom studies involving Chinese-speaking patients with cancer who live elsewhere.

In conclusion, the current study demonstrated that the MDASI-C is a valid and reliable multisymptom assessment tool that is highly suitable for Chinese patients with cancer. The standard methods of translation/back-translation and testing for reliability and validity that we described are useful for researchers wishing to create alternate-language versions of existing assessment tools. In addition, the results of the current study support the crosscultural and crosslinguistic validity of the MDASI itself. As a standard international measurement tool for symptom assessment, the MDASI should enable investigators to examine cultural and regional differences in symptom management and to use the data obtained by this tool as an outcome measure in international clinical trials.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

The authors thank Jeanie F. Woodruff for her editorial assistance.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES
  • 1
    Cleeland CS, Mendoza TR, Wang XS, et al. Assessing symptom distress in cancer patients: the M.D. Anderson Symptom Inventory. Cancer. 2000; 89: 16341646.
  • 2
    Cleeland CS. Cancer-related symptoms. Semin Radiat Oncol. 2000; 10: 175190.
  • 3
    Borden EC, Parkinson D. A perspective on the clinical effectiveness and tolerance of interferon-alpha. Semin Oncol. 1998; 25(1 Suppl 1 ): 38.
  • 4
    U.S. Census Bureau. Census 2000 brief: the foreign-born population. Washington, DC: U.S. Census Bureau, 2003.
  • 5
    McCorkle R, Young K. Development of a symptom distress scale. Cancer Nurs. 1978; 1: 373378.
  • 6
    Portenoy RK, Thaler HT, Kornblith AB, et al. The Memorial Symptom Assessment Scale: an instrument for the evaluation of symptom prevalence, characteristics and distress. Eur J Cancer. 1994; 30A: 13261336.
  • 7
    de Haes JC, van Knippenberg FC, Neijt JP. Measuring psychological and physical distress in cancer patients: structure and application of the Rotterdam Symptom Checklist. Br J Cancer. 1990; 62: 10341038.
  • 8
    Bruera E, Kuehn N, Miller MJ, Selmser P, Macmillan K. The Edmonton Symptom Assessment System (ESAS): a simple method for the assessment of palliative care patients. J Palliat Care. 1991; 7: 69.
  • 9
    Li L, Zhang S, Lu F. [ Research on characteristics of mortality spectrum and type composition of malignant tumors in China]. Zhonghua Zhong Liu Za Zhi. 1997; 19: 323328.
  • 10
    Wang XS, Mendoza TR, Gao SZ, Cleeland CS. The Chinese version of the Brief Pain Inventory (BPI-C): its development and use in a study of cancer pain. Pain. 1996; 67: 407416.
  • 11
    Ware JE Jr., Sherbourne CD. The MOS 36-Item Short-Form Health Survey (SF-36). I. Conceptual framework and item selection. Med Care. 1992; 30: 473483.
  • 12
    Ren XS, Wang XS, Liu S, Hao X. Psychometric and clinical evaluation of the Chinese version of the SF-36 health survey among cancer patients in China. Qual Life Res Newsletter. 2003; 30: 57.
  • 13
    Wang XS, Cleeland CS, Mendoza TR, et al. The effects of pain severity on health-related quality of life: a study of Chinese cancer patients. Cancer. 1999; 86: 18481855.
  • 14
    Oken MM, Creech RH, Tormey DC, et al. Toxicity and response criteria of the Eastern Cooperative Oncology Group. Am J Clin Oncol. 1982; 5: 649655.
  • 15
    Gorsuch RL. Factor analysis (2nd edition). Hillsdale, NJ: Lawrence Erlbaum Associates, 1983.
  • 16
    Mulaik SA. The foundations of factor analysis. New York: McGraw-Hill, 1971.
  • 17
    Harman HH. Modern factor analysis (2nd revised edition). Chicago: University of Chicago Press, 1967.
  • 18
    Cleeland CS, Nakamura Y, Mendoza TR, Edwards KR, Douglas J, Serlin RC. Dimensions of the impact of cancer pain in a four country sample: new information from multidimensional scaling. Pain. 1996; 67: 267273.
  • 19
    Serlin RC, Mendoza TR, Nakamura Y, Edwards KR, Cleeland CS. When is cancer pain mild, moderate or severe? Grading pain severity by its interference with function. Pain. 1995; 61: 277284.
  • 20
    Mendoza TR, Anderson KO, Wang XS, Easley M, Mobley G, Cleeland CS. Phone it in or fill it in? Comparing symptom assessment using the interactive voice response system with the traditional paper and pencil method [abstract]. Presented at the Eighth Annual Meeting of the International Society for Quality of Life Research, Amsterdam, The Netherlands, November 7–10, 2001.
  • 21
    Cleeland CS, Reyes-Gibby CC. When is it justified to treat symptoms? Measuring symptom burden. Oncology (Huntingt). 2002; 16: 6470.
  • 22
    Kleinbaum DG, Kupper LL, Muller KE. Applied regression analysis and other multivariable methods (2nd edition). Boston: PWS-Kent Publishing Co., 1988.
  • 23
    SPSS, Inc. SPSS statistical software program for Windows (Version 12.0). Chicago: SPSS, Inc., 2003.
  • 24
    Okuyama T, Wang XS, Akechi T, et al. Japansese version of the M. D. Anderson Symptom Inventory: a validation study. J Pain Symptom Manage. 2003; 26: 10931104.
  • 25
    Ward SE, Goldberg N, Miller-McCauley V, et al. Patient-related barriers to management of cancer pain. Pain. 1993; 52: 319324.
  • 26
    Caraceni A, Mendoza TR, Mencaglia E, et al. A validation study of an Italian version of the Brief Pain Inventory (Breve Questionario per la Valutazione del Dolore). Pain. 1996; 65: 8792.