Validation and Application of the Arabic Version of the M. D. Anderson Symptom Inventory in Moroccan Patients With Cancer

Vol. 40 No. 1 July 2010 Journal of Pain and Symptom Management 75 Original Article Validation and Application of the Arabic Version of the M. D. Anderson Symptom Inventory in Moroccan Patients With Cancer Mati Nejmi, MD, PhD, Xin Shelley Wang, MD, MPH, Tito R. Mendoza, PhD, Ibrahima Gning, DDS, DrPH, and Charles S. Cleeland, PhD Départment d Anesthésiologie et de Traitment de la Douleur (M.N.), Institut National d Oncologie, Rabat, Morocco; and Department of Symptom Research (X.S.W., T.R.M., I.G., C.S.C.), Division of Internal Medicine, The University of Texas MD Anderson Cancer Center, Houston, Texas, USA Abstract Context. Little is known about symptom burdenea concept encompassing both symptom severity and the degree of symptom interference with daily livingein patients with cancer in Morocco or other countries with Arabic-speaking populations. Objectives. The goal of this study was to psychometrically validate the Arabic version of the M. D. Anderson Symptom Inventory (MDASI-A), a tool for measuring multiple symptoms in patients with cancer, and to test its utility in a sample of patients with late-stage cancer in Morocco. Methods. The MDASI-A was developed by standard forward-backward translation of the English MDASI. We used nonidiomatic Arabic in the MDASI-A to enhance its possible usefulness for all Arabic-speaking patients with cancer. One hundred sixty-five Arabic-speaking patients with various cancer types were recruited from a city hospital in Rabat, Morocco. The MDASI-A was administered by interview, as only 5% of the patients had a high school education. Results. Psychometric analysis demonstrated acceptable internal consistency, with Cronbach alpha values of 0.85 for all 19 items, 0.78 for symptom severity items, and 0.79 for interference items; known-group validity was demonstrated by significant differences in mean symptom severity and interference between patients with good vs. poor performance status. All patients had moderate to severe pain and were taking pain medications. Additional severe symptoms included fatigue, lack of appetite, and disturbed sleep. Patients with gastrointestinal or gynecological cancer reported relatively more symptom severity than patients with breast or lung cancer. Poor performance status, male gender, Funding source: This work was supported in part by The Hawn Foundation. Address correspondence to: Xin Shelley Wang, MD, MPH, Department of Symptom Research, The Ó 2010 U.S. Cancer Pain Relief Committee Published by Elsevier Inc. All rights reserved. University of Texas MD Anderson Cancer Center, 1515 Holcombe Boulevard, Unit 1450, Houston, TX 77030, USA. E-mail: xswang@mdanderson.org Accepted for publication: December 14, 2009. 0885-3924/$esee front matter doi:10.1016/j.jpainsymman.2009.12.007

76 Nejmi et al. Vol. 40 No. 1 July 2010 and current infection were significant predictors of high symptom interference (R 2 ¼ 0.48, P < 0.05). Conclusion. The MDASI-A is a valid and reliable patient-reported outcome instrument that can be used to assess Moroccan Arabic-speaking cancer patients multiple symptoms. Its utility for use in other Arab countries needs to be tested. J Pain Symptom Manage 2010;40:75e86. Ó 2010 U.S. Cancer Pain Relief Committee. Published by Elsevier Inc. All rights reserved. Key Words MDASI, symptom assessment, Arabic, validation, cancer, pain Introduction Patients with cancer suffer from a variety of symptoms caused by their disease or related treatment, especially when the cancer is in an advanced stage. To assess and manage these diverse symptoms, clinicians who provide oncology care need valid and reliable symptomreport measures. Although Arabic-language tools measuring the broad concept of healthrelated quality of life have been validated in Arabic-speaking patients, 1e5 a search of the medical literature in Medline, Scopus, and the Institut National de l Information Scientifique et Technique in France failed to retrieve any validated Arabic translations of established tools for measuring cancer-related symptom burden, a more focused concept that encompasses both the severity of symptoms and the patient s perception of the impact of the symptoms as the patient self-report closest to the biological processes of disease and the interactions between disease and treatment. 6 As a result, little is known about symptom burden in patients with cancer in Morocco or other countries with Arabic-speaking populations. The M. D. Anderson Symptom Inventory (MDASI), developed by Cleeland et al., 7 is a reliable, validated patient-reported outcome measure that can be used to assess symptom burden in patients with cancer. This comprehensive instrument assesses the most prevalent and common cancer-related symptoms, along with symptoms interference with daily life. It is easy to administer, is accessible to individuals with a low education level, and can easily be translated into other languages. Existing validated foreign-language versions include Japanese, 8 Chinese, 9 Filipino, 10 Greek, 11 Korean, 12 Russian, 13 and Taiwanese 14 translations. In a systematic review of 21 validated symptom assessment instruments, 15 the MDASI was judged to be the most appropriate in terms of flexibility, reliability and validity, ease of completion, and utility in symptom management. The purpose of this investigation was to create and validate an Arabic version of the MDASI (MDASI-A). We conducted the study among cancer patients receiving palliative care at the Institut National d Oncologie, Rabat, Morocco, examining the psychometric properties of the MDASI-A and the prevalence and patterns of cancer-related symptoms in these Arabicspeaking patients. Our goal in creating the MDASI-A was to provide an instrument that will be useful in all Arab countries of the North African and Middle-Eastern regions as well as in immigrant Arab populations worldwide. Methods Subjects We recruited patients being treated for cancer in the Department of Anesthesiology, Institut National d Oncologie, Rabat, Morocco. Eligible patients had a diagnosis of cancer, were experiencing cancer pain and receiving pain medication, were at least 18 years old, and spoke Arabic. Patients were excluded if they were too ill or too cognitively impaired to respond to the questionnaire, as judged by the research staff. Most of the patients were receiving government-supported health care. The proposed study was approved by the General Secretary of the Ministry of Health in Morocco and the Institutional Review Board of The University of Texas MD Anderson Cancer Center in Houston, Texas. Participants

Vol. 40 No. 1 July 2010 Validity and Application of the Arabic MDASI 77 gave verbal consent after being fully informed about the purpose, survey process, and scientific goals of the study. Symptom Assessment The MDASI assesses 13 core cancer-related symptoms reported as highly frequent and severe: pain, fatigue, nausea, disturbed sleep, distress, shortness of breath, difficulty remembering, lack of appetite, drowsiness, dry mouth, sadness, vomiting, and numbness or tingling. Each symptom is rated on a 0e10 scale (0 ¼ not present and 10 ¼ as bad as you can imagine ) over the past 24 hours. Six interference items assessing symptom-related interference in general activity, mood, work (including work around the house), relations with other people, walking, and enjoyment of life also are included in the MDASI. Each interference item is rated on a 0e10 scale (0 ¼ did not interfere and 10 ¼ interfered completely ) over the past 24 hours. A more detailed description of this instrument and the methods used to validate the Englishlanguage version are published elsewhere. 7 The 13 symptom severity items of the MDA- SI can be averaged into a symptom subscale score and the six interference items can be averaged into an interference subscale score. Individual or subsets of symptom items also can be used as symptom outcomes, without summary scoring. Provisionally, the interference subscale can be subdivided into two component scores: WAW (mean of the physical items: walking, activity, and work) and REM (mean of the affective items: relations with other people, enjoyment of life, and mood ), as established in previous studies. 16 The MDASI-A (Fig. 1) was translated into Arabic from the original English-language MDASI using a multistep forward-backward translation method, which required collaboration between health care professionals and native Arab translators who were fluent in English. To avoid dialectical differences, items from the English version of the MDASI were translated using simple terms, and nonidiomatic expressions. Another bilingual professional translated the resulting Arabic version back into English. Comparison between the original and the back translation then followed. This process was repeated until agreement on each of the 19 items was reached. Study Procedures The study used a cross-sectional design. Patients completed the study questionnaires at the time of enrollment. Because of the high percentage of low literacy in the sample, the MDASI-A was administered by a trained researcher in an interview format. Patient demographic data were collected using a general survey questionnaire that included gender, age, marital status, education level, and employment status. Medical background information, such as cancer diagnosis, staging, the presence of metastases, and type of anticancer therapy, was obtained from hospital records. The physician-rated Eastern Cooperative Oncology Group Performance Status (ECOG PS) scale was used to evaluate a patient s performance status. Statistical Analysis Statistical analysis was conducted using Statistical Package of the Social Sciences (SPSS) version 15.0 (SPSS Inc., Chicago, IL). 17 Criterion for statistical significance was set using a twotailed alpha level of 0.05. When tests were not significant, effect sizes (ES) were computed as an indicator of the magnitude of the difference between known groups. ES values greater than or equal to 0.5 standard deviation (SD) are deemed moderate to large. 18,19 Instrument Validation. Construct validity of the instrument was assessed using principal axis factoring with direct oblimin rotation during confirmatory analysis. A factor solution with simple structure 20 and clinical interpretability was adopted. Model fit was tested according to Harman s criteria [N ( 1/2) ]. 21 Kaiser- Meyer-Olkin s test was used to determine whether the sample size was adequate; values of at least 0.50 are considered good. 22 Known-group validity was examined using independent-sample t-tests and stratification of patients by good ECOG PS (scores of 0e2) vs. poor ECOG PS (scores of 3e4). Internal consistency reliability of the MDASI-A was examined using Cronbach alpha values. Internal consistency reliability is considered good when Cronbach alpha values are 0.7 or higher. 20 Prevalence and Severity of Cancer Symptoms. In response to the paucity of symptom data in

78 Nejmi et al. Vol. 40 No. 1 July 2010 Fig. 1. Arabic version of the M. D. Anderson Symptom Inventory (MDASI-A). Copyright 1999 The University of Texas MD Anderson Cancer Center.

Vol. 40 No. 1 July 2010 Validity and Application of the Arabic MDASI 79 Fig. 1. (Continued). the literature on patients with cancer in Morocco, we assessed symptom severity and prevalence for this Moroccan sample with pain and advanced cancer. Means, SDs, 95% confidence limits, and percentage of patients with moderate ($5 on the 0e10 scale) or severe ($7 on the 0e10 scale) MDASI-A scores were computed for symptoms and subscales. Predictors of Symptom Interference. Regression analyses were conducted to detect potential predictors of symptom burden in the study sample, represented by the average interference subscale score. Factors included were symptoms, disease status, staging, whether receiving radiation or not, whether receiving chemotherapy or not, evidence of infection, ECOG PS, age, and gender. Collinearity diagnostics were used to account for close intercorrelations among symptoms, and outliers beyond three SDs were identified.

80 Nejmi et al. Vol. 40 No. 1 July 2010 Results Patient Characteristics Sociodemographic and disease characteristics for the sample are presented in Table 1. Of the 165 patients enrolled, 68% had Stage IV cancer, and 49% had multiple Table 1 Demographic and Disease Characteristics of the Patient Sample (n ¼ 165) Characteristic % n Age Mean (SD) 48.7 (13.7) Median (range) 49.0 (18e84) Younger than 60 years 76.4 126 Gender Female 56.4 93 Male 43.6 72 Education level Preschool 63.6 105 Primary school 14.5 24 Middle school 17.0 28 High school 4.2 7 Special technical school 0.6 1 Race Arab 82.4 136 Sub-Saharan African 17.6 29 Disease type Hematological malignancies Lymphoma 2.4 4 Solid tumors Breast 19.4 32 Gastrointestinal 21.2 35 Gynecological 15.2 25 Genitourinary 7.3 12 Head and neck 7.3 12 Lung 12.1 20 Sarcoma 3.6 6 Unknown 7.9 13 Other 3.6 6 ECOG PS 0 (fully active) 6.1 10 1 (restricted but ambulatory) 6.7 11 2 (ambulatory, capable of 25.5 42 self-care) 3 (capable of only limited 22.4 37 self-care) 4 (completely disabled) 39.4 65 Cancer stage IIeIII 32.1 53 IV 67.9 112 Treatment within past month or currently a Radiotherapy 14.0 23 Chemotherapy 7.9 13 Surgery 3.6 6 Opioids 99.4 164 Antidepressants 79.4 131 Antibiotics, antifungals 13.4 22 Antiemetics 31.5 52 Nutritional support 26.7 44 a Patients may have received more than one treatment. cancer metastases; 25% of the patients were receiving cancer therapy at the time of the study. Validation of the Arabic Version of the M. D. Anderson Symptom Inventory Internal Consistency Reliability. The reliability of the MDASI-A is supported by Cronbach alpha values of 0.85 for the entire instrument (19 items), 0.78 for the 13-item symptom subscale, and 0.79 for the six-item interference subscale. Alpha values remained higher than 0.7 if any one item was deleted (Table 2). Construct Validity. Factor analysis for the 13- item symptom severity subscale showed a twofactor solutionda general symptom factor and a gastrointestinal symptom factor (nausea and vomiting). This solution parallels the factor structures of previous MDASI module and language validations, and it also satisfies clinical interpretability criteria and Harman s criterion for model fit: the SD of the residuals (0.061) for this two-factor model is less than the reciprocal of the square root of the sample size (0.078). However, the items pain and numbness did not load with either factor. Diagnostic tests evidenced that the sampling adequacy was acceptable (Kaiser-Meyer-Olkin s value was 0.81, where values of at least 0.50 are considered good). Table 2 Reliability of the MDASI-A (n ¼ 165) MDASI-A a If Item Deleted All 13 symptom items: a ¼ 0.78 Pain 0.79 Fatigue 0.76 Nausea 0.76 Disturbed sleep 0.77 Distress 0.76 Shortness of breath 0.77 Difficulty remembering 0.78 Lack of appetite 0.76 Drowsiness 0.75 Dry mouth 0.76 Sadness 0.76 Vomiting 0.76 Numbness 0.81 All six interference items: a ¼ 0.79 Activity 0.71 Mood 0.79 Work 0.74 Relations with others 0.81 Walking 0.74 Enjoyment of life 0.71

Vol. 40 No. 1 July 2010 Validity and Application of the Arabic MDASI 81 Known-Group Validity. Sample t-tests showed that patients with poor ECOG PS had significantly higher symptom scores (mean ¼ 5.0, SD ¼ 1.53) than the patients with good ECOG PS (mean ¼ 4.2, SD ¼ 1.54)dP < 0.01. Significant differences in interference were also seen between the patients with poor ECOG PS (mean ¼ 7.1, SD ¼ 1.12) and good ECOG PS (mean ¼ 4.7, SD ¼ 2.27)dP < 0.001. Average physical (WAW) and affective (REM) components of the interference subscale showed similar group differences (Table 3). Severity and Prevalence of Multiple Cancer- Related Symptoms Descriptive statistics on symptom severity and prevalence are displayed in Table 3. Only seven of the 3,135 possible data points (19 MDASI-A items answered by 165 subjects) were left blank, resulting in a 0.2% missingdata rate. The patients in our sample had come to the treatment center because of severe pain, with 96% rating their pain at 7 or higher on the MDASI-A. In addition, the patients reported high prevalence of multiple severe symptoms (rated $7 on the MDASI-A s 0e10 scale), including fatigue (65%), lack of appetite (48%), sadness (31%), dry mouth (30%), shortness of breath (29%), and distress (28%). Pain was rated as the most severe symptom, followed by fatigue, lack of appetite, disturbed sleep, dry mouth, sadness, and distress. Difficulty remembering, nausea, and vomiting were the least severe symptoms reported by this sample. We also examined symptom profiles by type of cancer, the most prevalent of which were breast, lung, gastrointestinal, and gynecological cancer (approximately 68% of the subjects). Overall, symptom severity scores were higher for patients with breast or lung cancer than for patients with gastrointestinal or gynecological cancer (difference [diff] $ 0.8, ES $ 0.5, P < 0.05) (Table 4 and Fig. 2). The severity of specific symptoms varied by cancer type. As expected, patients with lung Table 3 Symptom Severity and PrevalencedGroup Mean Comparisons Based on ECOG PS (n ¼ 165) Mean (SD) MDASI-A Item Total PS 0e2 c PS 3e4 c Range LCL UCL % $5 a % $7 b Pain 9.62 (1.06) 9.59 9.64 5e10 9.46 9.78 100 96.4 Fatigue 7.05 (2.79) 6.05 7.68 0e10 6.63 7.48 85.5 64.8 Lack of appetite 6.07 (3.07) 5.37 6.51 0e10 5.60 6.55 78.8 47.9 Disturbed sleep 5.35 (2.34) 5.13 5.48 0e10 4.99 5.71 69.1 24.2 Dry mouth 5.07 (2.62) 4.37 5.51 0e10 4.67 5.47 67.9 29.7 Sadness 4.85 (3.41) 4.43 5.11 0e10 4.32 5.37 61.8 30.9 Distress 4.33 (3.25) 3.93 4.57 0e10 3.83 4.84 54.3 27.8 Numbness 3.76 (3.39) 3.51 3.91 0e10 3.24 4.28 45.5 21.8 Shortness of breath 3.89 (3.59) 3.78 3.96 0e10 3.34 4.44 47.3 29.1 Drowsiness 3.28 (3.07) 2.45 3.79 0e10 2.81 3.76 42.9 16.0 Nausea 3.03 (3.03) 2.45 3.38 0e10 2.56 3.50 35.4 15.9 Difficulty remembering 2.85 (3.10) 2.16 3.26 0e10 2.37 3.33 34.8 15.9 Vomiting 2.43 (2.93) 1.84 2.79 0e10 1.98 2.88 24.8 14.5 Interference with General activity 8.72 (2.67) 6.89 9.84 5e10 8.30 9.13 91.5 81.2 Work 8.68 (2.74) 7.00 9.73 0e10 8.26 9.11 89.7 83.0 Enjoyment of life 7.67 (3.06) 5.35 9.10 0e10 7.20 8.14 86.1 67.9 Walking 6.75 (3.50) 3.67 8.65 0e10 6.21 7.28 78.8 55.8 Mood 4.27 (3.09) 3.95 4.47 0e10 3.80 4.75 50.9 25.5 Relations with others 1.15 (2.35) 1.33 1.03 0e10 0.78 1.51 13.3 4.2 Summary scores Symptom severity 4.74 (1.58) 4.25 5.05 0.6e9.2 4.50 4.99 52.7 13.3 Symptom interference 6.21 (2.03) 4.70 7.14 0.0e9.8 5.89 6.52 80.6 60.0 WAW d 8.05 (2.56) 5.85 9.41 0e10 7.66 8.44 88.5 78.8 REM e 4.36 (2.04) 3.54 4.87 0.0e9.7 4.05 4.68 47.3 12.7 LCL ¼ lower 95% confidence limits; UCL ¼ upper 95% confidence limits. a Percent moderate to severe (rated $5 on the MDASI-A s 0e10 scale). b Percent severe (rated $7 on the MDASI-A s 0e10 scale). c Performance status was considered good for ECOG PS scores of 0e2 and poor for ECOG PS scores of 3e4. d WAW is the mean component score of the interference items walking, activity, and work. e REM is the mean component score of the interference items relations with other people, enjoyment of life, and mood.

82 Nejmi et al. Vol. 40 No. 1 July 2010 Table 4 Analysis of Variance ResultsdGroup Mean Comparison Based on Major Disease Groups (n ¼ 112) MDASI-A Summary Score Means Comparison Mean Difference ES 95% Confidence Limit Mean symptom severity Breast: 4.53 (1.60) GI 0.85 0.53 1.58 0.11 GYN 0.79 0.53 1.59 0.02 GI: 5.38 (1.58) GYN 0.06 0.04 0.73 0.85 GYN: 5.32 (1.34) Lung 0.82 0.58 0.09 1.73 Lung: 4.50 (1.49) Breast 0.03 0.02 0.89 0.83 GI 0.88 0.57 1.73 0.03 Mean symptom interference Breast: 5.87 (2.52) GI 0.67 0.30 1.71 0.37 GYN 0.72 0.34 1.86 0.41 GI: 6.54 (1.87) GYN 0.06 0.03 1.17 1.06 GYN: 6.59 (1.77) Lung 0.60 0.29 0.68 1.88 Lung: 5.99 (2.38) Breast 0.12 0.05 1.09 1.34 GI 0.55 0.26 1.74 0.65 Mean WAW a Breast: 7.78 (3.48) GI 0.69 0.25 1.99 0.60 GYN 0.35 0.12 1.77 1.06 GI: 8.48 (2.14) GYN 0.34 0.15 1.05 1.73 GYN: 8.13 (2.36) Lung 0.80 0.34 0.79 2.39 Lung: 7.33 (2.40) Breast 0.45 0.15 1.96 1.07 GI 1.14 0.50 2.63 0.34 Mean REM b Breast: 3.96 (1.90) GI 0.64 0.33 1.63 0.34 GYN 1.10 0.61 2.17 0.02 GI: 4.60 (1.98) GYN 0.45 0.25 1.51 0.60 GYN: 5.05 (1.69) Lung 0.40 0.19 0.80 1.61 Lung: 4.65 (2.61) Breast 0.69 0.31 0.45 1.84 GI 0.05 0.02 1.08 1.18 GI ¼ gastrointestinal cancer; GYN ¼ gynecological cancer. a WAW is the mean component score of the interference items walking, activity, and work. b REM is the mean component score of the interference items relations with other people, enjoyment of life, and mood. cancer had significantly more severe shortness of breath, whereas patients with gastrointestinal cancer experienced more severe nausea, vomiting, and lack of appetite (Table 4). Patients with gastrointestinal cancer reported significantly greater fatigue (mean ¼ 8.3; SD ¼ 2.2) than the patients with either breast cancer (diff ¼ 2.0, P < 0.05) or lung cancer (diff ¼ 1.7, P < 0.01) and greater disturbed sleep (mean ¼ 6.0, SD ¼ 2.3) and dry mouth (mean ¼ 5.9, SD ¼ 2.4) than the patients with breast cancer (for disturbed sleep, diff ¼ 1.2, P < 0.05; for dry mouth, diff ¼ 1.54, P < 0.05). For affective symptoms, patients in the gynecological cancer group reported significantly more severe sadness (mean ¼ 6.4, SD ¼ 2.4) than the patients with gastrointestinal cancer (diff ¼ 1.6, P < 0.5) and reported greater interference with mood (mean ¼ 5.6, SD ¼ 2.3) than the patients with breast cancer (diff ¼ 1.6; P < 0.5) (Fig. 2). Predictors of High Symptom Interference The interference of symptoms in daily life was primarily reflected in the ratings of general activity, work, enjoyment of life, and walking. Compared with patients with Stage II and III disease, patients with Stage IV disease (68% of the sample) reported more functional impairment, with significantly higher interference scores of general activity (diff ¼ 1.07, P < 0.05), work (diff ¼ 1.74, P < 0.001), and walking (diff ¼ 2.5, P < 0.001). We examined the patient characteristics relative to the severity of symptoms to identify which characteristics, if any, would predict the total symptom interference score. In a stepwise linear regression model that included only patient characteristics (Model 1), we found that ECOG PS (b ¼ 0.68, P < 0.001), male gender (b ¼ 0.15, P < 0.05), and evidence of current infection (b ¼ 0.13, P < 0.05) were predictive of mean symptom interference. A model with symptoms only (Model 2) indicated that four symptom items were predictive of mean total interference, including fatigue (b ¼ 0.41, P < 0.001), sadness (b ¼ 0.21, P < 0.01), numbness (b ¼ 0.22, P < 0.01), and dry mouth (b ¼ 0.20, P < 0.01). A model with both symptoms and patient characteristics (Model 3) showed that ECOG PS, gender, fatigue, sadness, numbness, and vomiting were significant predictors of symptom interference. This model explained 66% of the variability in symptom interference, whereas the first two models explained 48% and 42%,

Vol. 40 No. 1 July 2010 Validity and Application of the Arabic MDASI 83 Fig. 2. Proportion of patients with severe symptoms by four major disease groups (n ¼ 112). Patients with a diagnosis of breast, gastrointestinal, gynecological, or lung cancer were compared in terms of proportions of severe ($7 on a 0e10 scale) vs. nonsevere fatigue (a), disturbed sleep (b), lack of appetite (c), and sadness (d). Chi-square tests for proportions were computed to detect significant association between each of the four symptoms and the four disease groups. There was a significant association between disease category and fatigue (X 2 ¼ 10.7, P < 0.05) and between disease category and lack of appetite (X 2 ¼ 10.8, P < 0.05). Disturbed sleep and sadness showed no significant association. GI ¼ gastrointestinal cancer; GYN ¼ gynecological cancer. respectively, of the variation in interference subscales. Thus, at least 18% of the variability in symptom interference is attributable to symptom reports over and above clinical and patient characteristics (Table 5). Discussion This study demonstrated the satisfactory psychometric properties (i.e., reliability, construct validity, convergent validity) of the MDASI-A as a patient-reported symptom assessment instrument for use with Moroccan cancer patients. Although the sample of 165 patients was characterized overall as having low literacy, advanced-stage disease, and severe pain, they were able to respond quite well to the MDASI-A interview questions using numeric rating scales. The MDASI-A translation uses simple, nonidiomatic Arabic terms, which should make the instrument testable and useful in other Arabic-speaking countries of the North African and Middle-Eastern regions as well as in the immigrant Arab populations worldwide. Cronbach alpha coefficients greater than 0.7 evidenced the MDASI-A s good internal consistency reliability. 20 The extraction of gastrointestinal and general symptom factors was similar to results reported in the original MDA- SI validation; 7 however, for the MDASI-A, pain did not load with either factor, reflecting the fact that almost all patients in this sample rated their pain at a severe level and were receiving pain medications at the time of enrollment.

84 Nejmi et al. Vol. 40 No. 1 July 2010 Model Table 5 Regression Models: Predictors of Total Symptom Interference Independent Variables in the Model Predictors Correlation and Significance Model 1: Patient and clinical factors (n ¼ 163) Age, gender, education, race, ECOG PS, disease stage, presence of infection, currently receiving antidepressants, currently receiving cancer treatment Poorer ECOG PS Male gender Infection Model 2: Symptoms (n ¼ 158) 13 MDASI symptom items Fatigue Sadness Numbness Dry mouth Model 3: Symptoms and patient characteristics (n ¼ 156) 13 MDASI symptom items, age, gender, education, race, ECOG PS, disease stage, presence of infection, currently receiving antidepressants, currently receiving cancer treatment Poorer ECOG PS Male gender Fatigue Sadness Numbness Vomiting R 2 ¼ 0.48; P < 0.05 R 2 ¼ 0.42; P < 0.01 R 2 ¼ 0.66; P < 0.001 Numbness also did not load with either factor, reflecting the fact that only 8% of the sample was receiving chemotherapy and possibly experiencing treatment-related numbness; even so, it is possible that some patients experienced numbness because of low-level disease or comorbidities. Further investigation using a different sample of Arabic-speaking patients with more variation in symptom experience may more clearly identify whether the underlying constructs of the MDASI-A are, in fact, different from previous MDASI validations. Most of the patients in this sample had severe pain, which resulted in a ceiling effect for pain. The SD for pain was approximately 1.05, which was smaller than the SD of approximately 2.5 typically seen in other MDASI validation studies. 7,9 Another result of the ceiling effect was the underestimated correlation of pain with other symptoms because of the truncation of pain values. We questioned whether the item numbness was clearly understood by patients during the interview, but investigation with two native Arab translators confirmed that the words used in this item ( Your numbness or tingling at its worst ) were correct in their Arabic meaning. In addition, removing or introducing the item pain in the analysis did not significantly change the correlation coefficients, the number of factors, the reliability, or the regression results. The most severe symptoms observed in this MDASI-A validation study, including pain, fatigue, lack of appetite, and disturbed sleep, were consistent with other MDASI validation studies and longitudinal cancer symptom studies that use the MDASI. Moderate to severe symptom levels (rated as 5 or greater on the 0e10 scale) were reported for pain (100% of the sample), fatigue (85%), lack of appetite (79%), and disturbed sleep (69%). Also consistent with other validation studies were findings that symptoms interfered most with general activity and work, and that there was a lower prevalence of strictly treatment-related symptoms, such as nausea and vomiting. This evidence supports the MDASI-A as comparable to other linguistic versions of the MDASI for capturing the symptom profiles of the study sample. Validated symptom assessment tools, such as the MDASI-A, might be useful for promoting the clinical measurement and management of symptoms in cancer patients and for conducting multinational clinical trials in Arabic-speaking patients with cancer. A notable finding in this study was that, compared with most of the other language groups we have studied, 9,11,13,14 these Moroccan patients with cancer reported extremely low levels of symptom interference with their relations with other people relative to the other interference items on the MDASIeA(Table 3). This phenomenon also was found in the validation study of the MDASIeFilipino. 10 It is customary for Moroccans to be secretive about having cancer, hiding illness and treatment even from extended family, because of religious and cultural beliefs that often associate cancer with shame, futility, or the will of God. 23e28 Our analysis showed that the unusually low

Vol. 40 No. 1 July 2010 Validity and Application of the Arabic MDASI 85 ratings on the item relations with other people resulted in a slightly lower overall Cronbach alpha (0.79) for the interference items than in previous validation studies. However, the alpha value is still greater than the acceptable value of 0.7 promoted by generally accepted statistical standards 20 and by the U.S. Food and Drug Administration s guidelines for developing patient-reported outcome tools. 29 We also investigated the usefulness of this multiple symptom assessment tool for capturing symptoms in major cancers, such as in breast, gastrointestinal, gynecological, and lung cancers. The component score for all symptoms indicated the highest severity from gastrointestinal cancer, followed in order by gynecological, breast, and lung cancer. More patients with gastrointestinal or gynecological cancer reported severe fatigue (rated 7 or greater). There was no difference between men and women on reported interference with relations with other people. Poor ECOG PS, male gender, and presence of infection at the time of the survey were the patient-related and clinical factors that significantly contributed to total interference, explaining nearly half of the variance. The study had several limitations. Most notable is the possibility of selection bias in this sample of largely illiterate palliative care patients, 90% of whom reported their pain severity as 10 on the 0e10 scale and most of whom were taking opioids. Consequently, this patient sample presented with a higher level of pain severity overall compared with the patient samples in other MDASI language validation studies. Second, the study did not include another Arabiclanguage patient-reported outcome measurement tool to fully support the MDASI-A s construct validity. Although there is no other word for pain in Arabic, including a second measure of pain might have confirmed whether pain actually was that prevalent or whether the assessment tool itself affected the rating. Furthermore, because most of this patient sample had a primary school education or less, all MDASI-A assessments were conducted through face-to-face interview instead of the typical paper-and-pencil format, which may have introduced method variance because of the mode of administration. Finally, as almost the entire sample had end-stage disease, the symptom profile model is overwhelmed by disease-driven symptoms. Further investigation with a different Arabic-speaking study sample is needed to confirm the generalizability of MDASI-A among Arabic-speaking patients and in other Arab countries. In conclusion, the MDASI-A is a simple, concise symptom assessment tool that may be useful for assessing the symptom status of Arabicspeaking cancer patients worldwide. Clinicians who treat patients with cancer and agencies that develop much needed cancer control programs and strategies in Arab countries will benefit from the availability of a multisymptom assessment tool in their native language. Acknowledgments The authors express special thanks to Jeanie F. Woodruff, ELS, who provided English editing. They also thank the Hawn Foundation for its financial support and the patients from the Institut National d Oncologie, Rabat, Morocco, who participated in this project. References 1. Abourazzak FE, Allali F, Rostom S, et al. Factors influencing quality of life in Moroccan postmenopausal women with osteoporotic vertebral fracture assessed by ECOS 16 questionnaire. Health Qual Life Outcomes 2009;7:23. 2. Awad MA, Denic S, El Taji H. Validation of the European Organization for Research and Treatment of Cancer Quality of Life questionnaires for Arabic-speaking populations. Ann N Y Acad Sci 2008;1138:146e154. 3. McKenna SP, Doward LC, Kohlmann T, et al. International development of the Quality of Life in Depression Scale (QLDS). J Affect Disord 2001; 63(1e3):189e199. 4. Ohaeri JU, Awadalla AW. The reliability and validity of the short version of the WHO Quality of Life Instrument in an Arab general population. Ann Saudi Med 2009;29(2):98e104. 5. Sabbah I, Drouby N, Sabbah S, Retel-Rude N, Mercier M. Quality of life in rural and urban populations in Lebanon using SF-36 Health Survey. Health Qual Life Outcomes 2003;1(1):30. 6. Cleeland CS. Symptom burden: multiple symptoms and their impact as patient-reported outcomes. J Natl Cancer Inst Monogr 2007;37:16e21. 7. Cleeland CS, Mendoza TR, Wang XS, et al. Assessing symptom distress in cancer patients: the M.D. Anderson Symptom Inventory. Cancer 2000; 89(7):1634e1646.

86 Nejmi et al. Vol. 40 No. 1 July 2010 8. Okuyama T, Wang XS, Akechi T, et al. Japanese version of the M.D. Anderson Symptom Inventory: a validation study. J Pain Symptom Manage 2003; 26(6):1093e1104. 9. Wang XS, Wang Y, Guo H, et al. Chinese version of the M.D. Anderson Symptom Inventory: validation and application of symptom measurement in cancer patients. Cancer 2004;101(8):1890e1901. 10. Wang XS, Laudico AV, Guo H, et al. Filipino version of the M.D. Anderson Symptom Inventory: validation and multisymptom measurement in cancer patients. J Pain Symptom Manage 2006;31(6): 542e552. 11. Mystakidou K, Cleeland C, Tsilika E, et al. Greek M.D. Anderson Symptom Inventory: validation and utility in cancer patients. Oncology 2004;67(3e4): 203e210. 12. Yun YH, Mendoza TR, Kang IO, et al. Validation study of the Korean version of the M.D. Anderson Symptom Inventory. J Pain Symptom Manage 2006;31(4):345e352. 13. Ivanova MO, Ionova TI, Kalyadina SA, et al. Cancer-related symptom assessment in Russia: validation and utility of the Russian M.D. Anderson Symptom Inventory. J Pain Symptom Manage 2005;30(5):443e453. 14. Lin CC, Chang AP, Cleeland CS, Mendoza TR, Wang XS. Taiwanese Version of the M.D. Anderson Symptom Inventory: symptom assessment in cancer patients. J Pain Symptom Manage 2007;33(2): 180e188. 15. Kirkova J, Davis MP, Walsh D, et al. Cancer symptom assessment instruments: a systematic review. J Clin Oncol 2006;24(9):1459e1473. 16. Cleeland CS, Nakamura Y, Mendoza TR, et al. Dimensions of the impact of cancer pain in a four country sample: new information from multidimensional scaling. Pain 1996;67(2e3):267e273. 17. Statistical Package for the Social Sciences, v. 15.0. Chicago, IL: SPSS Inc., 2006. 18. Norman GR, Sloan JA, Wyrwich KW. Interpretation of changes in health-related quality of life: the remarkable universality of half a standard deviation. Med Care 2003;41(5):582e592. 19. Cohen J. Statistical power analysis for the behavioral sciences, 2nd ed. Hillsdale, NJ: Lawrence Earlbaum Associates, 1988. 20. Nunnally J, Bernstein I. Psychometric theory, 3rd ed. New York: McGraw-Hill, 1994. 21. Harman H. Modern factor analysis, 2nd ed. Chicago, IL: University of Chicago Press, 1976. 22. Kaiser HF. A second generation little jiffy. Psychometrika 1970;35:401e415. 23. Koopman C, Angell K, Turner-Cobb JM, et al. Distress, coping, and social support among rural women recently diagnosed with primary breast cancer. Breast J 2001;7(1):25e33. 24. McCorkle R, Quint-Benoliel J. Symptom distress, current concerns and mood disturbance after diagnosis of life-threatening disease. Soc Sci Med 1983;17(7):431e438. 25. Whelan TJ, Mohide EA, Willan AR, et al. The supportive care needs of newly diagnosed cancer patients attending a regional cancer center. Cancer 1997;80(8):1518e1524. 26. Lalla Salma Association Against Cancer. Cancer in Morocco. Available from http://www. contrelecancer.ma/en/cancer-in-morocco. Accessed July 2, 2009. 27. Nejmi M, Wang XS, Mendoza TR, Gning I, Cleeland CS. [Preliminary validation of the Arabic version of the Brief Pain Inventory (BPI-A) for cancer pain patients]. [in French]. Le Courrier de l Algologie 2006;5(1):27e35. 28. Lalla Salma Association Against Cancer. Connaissances, perceptions & attitudes a l egard du cancer. [in French]. Available from http:// www.contrelecancer.ma/fr/etude%20perception.pdf. Accessed July 6, 2009. 29. US Food and Drug Administration, Center for Drug Evaluation and Research, Center for Biologics Evaluation and Research, Center for Devices and Radiological Health. Guidance for industry. Patient-reported outcome measures: use in medical product development to support labeling claims. Available from http://www.fda.gov/downloads/ Drugs/GuidanceComplianceRegulatoryInformation/ Guidances/UCM071975.pdf. Accessed June 9, 2009.