Construct Validity and Reliability of the European Organization for Research and Treatment of Cancer Quality of Life Tool Among Colorectal Cancer Patients in Ethiopia

Background: Colorectal cancer and its treatments can have a detrimental effect on patients’ quality of life. The European Organization for Research and Treatment of Cancer Quality of Life questionnaire (EORTC-QLQ) was developed to assess quality of life among colorectal cancer patients and is used worldwide. However, the tool has not been translated into a local language or validated in the Ethiopian context. Therefore, this study aimed to examine the psychometric properties of the EORTC-QLQ in Ethiopia among colorectal cancer patients. Methods: A cross-sectional study was conducted in a major referral hospital in Addis Ababa, Ethiopia from March to May, 2020. A total of 158 colorectal cancer patients were consecutively included. The construct validity was assessed using Multitrait Scale Analysis, convergent validity, and relationship with functional outcomes. The internal consistency was examined using Cronbach’s alpha. Result: Among the participants, 52.2% were men, with a median age of 46 years (IQR = 17.7 years). The time needed to complete the EORTC-QLQ was less than 20 minutes. The item-total correlation alpha values ranged from 0.47–0.91. All item correlations within their scales were greater than 0.4 except for the Blood and Mucus in Stool scale. The value of correlation coecients between all items and their own domain were higher than other domains, except for the Blood and Mucus in Stool scale. The correlation between the core questionnaire and the colorectal tool ranged from -0.453–0.581. The tool showed a signicant difference between stoma and non-stoma patients, and between patients who had good physical function and those who did not. Conclusion: Except for the Blood and Mucus in Stool scale, the construct validity and reliability of all scales were supported. Therefore, the Amharic version of the tool can be used to assess health related quality of life in Ethiopian colorectal cancer patients. However, due to the low correlation between items in the Blood and Mucus in Stool scale, we recommend that these items are treated as a single item.


Introduction
Colorectal cancer is the third leading cause of cancer cases worldwide. In 2018, there were more than 1.8 million new cases and 881,000 deaths from colorectal cancer [1]. The incidence of colorectal cancer is higher in high-income countries, but the mortality is higher in low-and middle-income countries. In Eastern Africa, colorectal cancer was the fourth most frequently diagnosed cancer type [2]. In Ethiopia, the second most populous country in sub-Saharan Africa, colorectal is the most common cancer diagnosis in males and it ranks fourth in females [3]. It is also responsible for 11.2 percent of cancer-related deaths in males and 4.8 percent of cancer-related deaths in females in Ethiopia.
Cancer symptoms and cancer treatments have several adverse effects and toxicity, which affect quality of life [4]. For instance, gastrointestinal cancer patients have been reported to suffer from problems that emerged as a consequence of both the disease and the treatments, which negatively affect their quality of life [5]. Colorectal cancer patients have signi cantly lower physical, role, cognitive and social function compared to the general public. Moreover, they have a higher frequency of constipation, diarrhoea and nancial di culties [6][7][8]. Thus, it is recommended to collect information on health-related quality of life and use this to guide decision making in treatment choice [9].
The World Health Organization (WHO) de nes quality of life as a broad concept that depends on a person's physical health, psychological state, level of freedom, social relationships, and personal beliefs [10]. Cancer treatments have started incorporating quality of life as an end point [11]. Thus, different sensitive, reliable and validated tools are available to assess quality of life among cancer patients.
Frequently used tools are the core EORTC-QLQ, the Functional Assessment of Chronic Illness Therapy (FACIT) Measurement System, the Rotterdam Symptom Checklist (RSCL), and the Symptom Distress Scale (SDS) [12]. From them, the core EORTC-QLQ has been recommended to be used when the aim of the study is to assess disease-speci c symptoms [13,14]. To complement the core EORTC-QLQ, different disease-speci c modules have been developed, and the EORTC-QLQ for colorectal cancer patients (EORTC-QLQ-CR29) is one of these tools. This tool has been developed to be used alongside the core tool in colorectal cancer patients. This disease-speci c tool has been validated in different countries such as Spain, Taiwan, Korea and China [15][16][17][18]. Even though the Ethiopian Federal Ministry of Health has developed a strategy to assess and treat distressing symptoms in cancer patients [19], there are no validated tools to assess quality of life among colorectal cancer patients. Therefore, this study intended to examine the reliability and construct validity of EORTC-QLQ-CR29 tool in Ethiopian colorectal cancer patients.

Study design, area and participants
A cross-sectional study was carried out at Tikur Anbessa Specialized Hospital (TASH) from March to May, 2020. TASH is the largest referral and teaching hospital in Ethiopia, serving as the only radiotherapy centre in the country. In this oncology centre, there are two radiotherapy machines, 36 inpatient beds and 12 outpatient chemotherapy beds. There are also six clinical oncologists who provide the oncology services [20].
The sample size was calculated based on the recommendations given for scaling analysis [21,22]. Since Amharic is the o cial working language of Ethiopia and is spoken by millions of Ethiopians as a second language, 158 pathologically-con rmed colorectal cancer patients who were ≥ 18 years old and who could speak Amharic were consecutively included while attending the TASH oncology department. Instrument EORTC-QLQ-CR29 is a 29-item module developed to complement the core EORTC-QLQ. It comprises 19 single items and four scales to assess urinary frequency, blood or mucus in the stool, stool frequency, body image and other problems faced by patients. The tool scales are generally classi ed as functional and symptom scales [23]. Along with the colorectal tool, the core EORTC-QLQ and an intervieweradministered structured questionnaire were used to assess participants' sociodemographic and clinical characteristics.
The EORTC-QLQ-CR29 tool was translated into Amharic according to the EORTC translation guidelines [24]. A pilot study was conducted to identify di cult, confusing and upsetting questions. None of the questions were found to be di cult, confusing or upsetting for the participants. Based on the recommendations for pilot studies [24], ten colorectal cancer patients were included in the pilot and these participants were excluded from the actual study.
After the completion of the pilot, the data collection commenced. The data was collected by two nurses who have bachelor's degrees and work in the oncology centre. Training on how to collect the data was given to the data collectors by the principal investigator for three days, focusing on the purpose of the study, contents of the questionnaire, and how to approach and get consent from patients. The principal investigator supervised the data collection process every other day to monitor the data collection procedure. Patients were approached in the waiting room while waiting for their turns.

Data analysis
Descriptive statistics were used to assess the frequency distributions of socio-demographic characteristics and clinical data. Numbers and percentages were used for categorical variables and median and inter-quartile range were used for continuous variables. The acceptability of the EORTC-QLQ-CR29 was assessed in term of response rate, missing data and time needed to complete the questionnaire. The reliability was assessed using Cronbach's alpha. A value of Cronbach's alpha of 0.70 or greater was considered to be adequate [22]. Convergent validity, the extent to which two measures are related to the same construct, was determined by employing Multitrait scaling analysis. Multitrait scaling analysis focuses on items as the unit of analysis [22].
Known-groups validity was checked using the Mann-Whitney test to see whether the tool is able to detect differences between groups. The known groups that were used for comparison were treatment intent, physical function, and presence of stoma [23]. Based on the median value of the physical scale of the core EORTC-QLQ, patients were classi ed as having better or worse physical function. Those patients with a median score of ≥ 43.3 for the physical scale were considered to have better physical function, whereas those below 43.3 were considered to have worse physical function [25].
The correlations between the items of the colorectal cancer and core tools scales were determined using Spearman's correlation coe cient. All scales and items were transformed into a 0-100 score as per the EORTC-QLQ scoring manual [26]. Statistical analyses were performed using SPSS version 21.

Page 5/17
Socio-demographic characteristics of respondents A total of 158 participants were included in this study, with a median age of 46 (IQR = 17.7). About 50 (32.3%) study participants had no formal education and were employed (32.3%), while 45 (28.5%) respondents had attended college. The majority of the study participants were men (52.5%), married (63.3%), and residents of Addis (64.4%). Clinical characteristics of respondents Rectal cancer accounted for 85 (53.8%) respondents. Seventy-eight (49.4%) of respondents were being treated with chemotherapy alone. About 120 (75.9%) of respondents were treated with palliative intent.
The majority of (72.2%) respondents were stage four cancer patients (Table 2). Acceptability of EORTC-QLQ-CR29 The acceptability of the tool was assessed in terms of response rate, missing data and time needed to complete the questionnaire. The EORTC-QLQ-CR29 along with the core EORTC-QLQ tool took an average of 20 minutes to complete. There were no missing responses to the questionnaire. In general, the tool was found to be acceptable.

Reliability of EORTC-QLQ-CR29
The reliability of the tool was examined based on the value of Cronbach's alpha coe cients (alpha ≥ 0.70). As presented in Table 3, all of the scales had an alpha value greater than or equal to 0.7, except for the Mucus and Blood in Stool scale (0.47). Urinary Frequency had the highest alpha value (0.91), followed by Stool Frequency (0.85). The correlation between the core EORTC-QLQ and CR29 scales The correlation between the core EORTC-QLQ and CR29 scales ranged from − 0.45-0.58. The highest correlation was seen between the Body Image scale from the colorectal tool and the Emotional scale from the core questionnaire. The highest negative correlation was observed between Flatulence and Cognitive Function (r = -0.45). Most of the correlations between the core EORTC-QLQ and CR29 were below 0.4 (Table 5).

Known-groups validity of EORTC-QLQ-CR29
To assess the known-groups validity of the tool, comparisons of scores of multi-item scales and single items of EORTC-QLQ-CR29 among three clinically-distinct groups were made. These three groups were treatment intent, presence of stoma and physical function. This study hypothesised that patients with a stoma had a higher symptom score for Embarrassment and lower functional score for the Anxiety and Body Image items and scale. Similarly, patients with worse physical health and palliative patients were expected to have higher scores for the symptom items and scales and lower scores for functional scales and items.
The Mann-Whitney test revealed that there was no signi cant difference between the curative and the palliative treatment groups in any of the scores of the colorectal tool scales or items. In patients with and without a stoma, the presence of a stoma led to a deterioration in quality of life by increasing Urinary Frequency, Flatulence, Embarrassment and Anxiety. Patients with better physical function scored higher for functional items and scales and scored lower for symptom items and scales. This suggested the known-groups validity of the tool is ful lled.  [15,18,23,27], where the tool was found to differentiate between the groups. However, the Body Image and Sore Skin items did not discriminate between patients with and without stoma. This is inconsistent with previous ndings where a difference was seen according to the presence of a stoma [15,18,23,27]. This difference might be due to the fact that, in our study, there was an age difference between stoma and non-stoma patients. However, the previous studies did not report the age of stoma and non-stoma patients. In the current study, patients in the stoma groups were older than those in the non-stoma groups. Based on a study conducted in Australia, elderly patients have a greater appreciation for their function and body image [29].
The correlation between the core EORTC-QLQ and CR29 ranged from weak to moderate. The highest correlation was observed between the Emotional and Body Image scales. This agrees with studies conducted in Korean and Spain [15,17]. This result suggests that the tools are designed to cover different dimensions of health-related quality of life. Therefore, the core questionnaire should be used along the speci c modules.

Strengths And Limitation Of The Study
During translation of the tool, every step of the EORTC translation process was followed, under the supervision of the EORTC translation group. Nonetheless, this study had its own limitations. Even though patients were included irrespective of their place of residence, treatment and disease stage, most of them were in the nal stages of cancer. This is because most cancer patients in Ethiopia come to hospital at the late stages of the disease. There is also a delay in diagnosis and a prolonged treatment waiting time.
Thus, the generalisability of the nding for early stage patients cannot be ascertained.

Conclusion
The current study supported the acceptability, reliability, and validity of the Amharic version of the EORTC-QLQ-CR29 in Ethiopian colorectal cancer patients. However, the construct validity of the Mucus and Blood in Stool scale was not supported. Based on the ndings, we recommend the Mucus and Blood in Stool scale is separated and treated as a single item. We also recommend that researchers and clinicians use the core questionnaire alongside the disease-speci c module while assessing health related quality of life Ethical clearance was obtained from the ethical clearance committee of Addis Ababa University (AAU) College of health science, school of public health ethical review committee. Permission to use the questionnaires was secured from the EORTC research group using an online form at https://qol.eortc.org/form. Informed written consent was given by all participants.