This is a multi-center study, involving three hospitals in Malaysia. The selected hospitals are University of Malaya Medical Centre (UMMC), Universiti Kebangsaan Malaysia Medical Centre (UKMMC) and Hospital Tuanku Jaafar, Seremban (HTJS).
Study population (inclusion or exclusion)
The study includes Malaysian citizens or permanent residents with histologically confirmed colon or rectal cancers. They should be receiving or planned to receive at least one form of treatment at the medical centers. Excluded patients are those aged less than 18
years, those with incomplete diagnosis and those with language problem or inability to understand any of the three languages of the instruments.
This study was approved by the ethics committees of the UMMC (MEC Ref.No:770.2), UKMMC (Project code: FF-274-2011) and the Ministry of Health Malaysia for using Hospital Tuanku Jaafar Seremban (NMRR-11-348-9245). The project was also under the guidance of the EORTC QOL office. We obtained written, informed consent from each participant as recommended by the ethics of medical research.
A data extraction form was developed for the purpose of gathering the relevant demographic and clinical data from the hospital records. Variables collected were: patient’s identification number, age, sex, race, marital status, nationality, educational status, employment status and cohabitation. Information on the index cancer was: site of primary cancer (according to the IARC and UICC cancer classification manual 8th edition), tumour stage (Dukes), and histopathological differentiation. Treatment planned or received such as; surgery, presence or absence of stoma, chemotherapy and radiotherapy.
The instruments selected for this study include a ‘core’ instrument EORTC QLQ-C30 (version 3.0) and two other ‘modules’, for colorectal cancer EORTC QLQ-CR29 and colorectal cancer liver metastasis (LMC 21) respectively. The choice of these instruments was guided by the availability, established psychometric properties, and non-superiority of other instruments [7
]. The QLQ-C30 is one of the most widely used instruments in cancer clinical research [8
]. EORTC QLQ-CR29 is the newest version of the colorectal cancer specific QoL questionnaire recently validated in Europe [9
]. The QLQ-LMC21 is the only liver metastasis specific instrument in use at present [10
]. A Karnofsky Performance Scale was used by the clinicians to rate the well-being of the patients [11
Sample size estimation
1. Pilot study:
according to the EORTC QOL group, each translated item of the questionnaire should be pilot-tested on 10 to 15 subjects before being field-tested on a larger sample [12
]. So for each of the two set of questionnaires: EORTC QLQ-CR29 and LMC21, we have included 30 subjects, ten each for Bahasa Malaysia, Chinese and Tamil respectively
1. 2. Validation:
For a multivariate analysis technique to gain reliable estimates, the number of subjects’ observations should be 10 times the number of variables in the model [13
]. Therefore, the sample size was estimated based on this recommendation as follows;
a) EORTC QLQ-CR29: There are 29 items in the CR29. Thus the minimum number of subject required is 290. In our study, we use 300 subjects to account for possible attrition.
b) EORTC QLQ-LMC 21: This questionnaire contains 21 items; therefore, we need a minimum of 210 subjects for its validation.
Three research assistants aided in data collection. Each assistant could read and speak a minimum of two languages from the three languages used in this study.
Prospective patients were identified using the eligibility criteria above. Eligible subjects received an invitation letter and/or telephone calls at least two weeks before their next visit to the clinic.
Baseline data and Karnofsky Performance Scale
Baseline data for all prospective patients were obtained from the medical records using a data extraction form mentioned above. Missing data was obtained from the patients during the interview. The clinician completed the Karnofsky Performance scale during patient’s visit on the same day as the day of the interview.
Personal characteristics such as sexual behavior and family life are better assessed through a self-administered method. It has the advantages of being reliable, preferred, in HRQoL studies and also cheap to undertake [5
]. The data collection method was self-administered and interviewer delivered the research administrators (research assistants and the researcher) presented the instruments, answered questions from respondents and were present throughout the sessions. This gave further motivation to the patients and encouraged them to answer every item. The reliability and quality in the answer were therefore ensured.
In preparation for the pilot study, we translated the chosen tools into three main Malaysian languages based on the recommendations contained in the EORTC QoL group translation procedure [13
The aim of pilot testing is to identify any potential problems in the translation of the instruments. Attention was on the following six areas; acceptability (face validity), clarity of the introduction and instructions, completeness, linguistic clarity, spontaneity of response and practicality of using the instruments. Interviews were conducted and it covered the following areas: difficulty in answering the questions, confusion, difficult words, upsetting nature, and an open comment on how the patient would ask a similar question if given opportunity. Responses were recorded and reported.
Subject grouping and assessment plan
Eight groups of patients were examined. These were adopted and modified from the EORTC QLQ-CR29 validation study in Europe [9
a) Group 1
: patients with colon cancers, had surgery with no stoma, not receiving any form of chemotherapy. This group was being assessed once and the questionnaires were completed within 12
months of surgery.
b) Group 2
: patients with colon cancers, had surgery with no stoma, receiving any form of chemotherapy. This group was being assessed once and the questionnaires were completed within 12
months of surgery but within two weeks of receiving chemotherapy.
c) Group 3: Rectal cancer patients with preoperative radiotherapy, the questionnaires were completed within two weeks of the radiotherapy.
d) Group 4: Patients with a permanent stoma; irrespective of the cancer site. This group was being assessed within five years post-surgery.
e) Group 5: Temporary stoma patients; irrespective of the cancer site. This group was being assessed twice, within one month after surgery with stoma, and within one to three months after closure of the stoma.
f) Group 6: Palliative care group: these were patients who were being treated with palliative intention. They were being assessed twice, first within two weeks of receiving chemotherapy or radiotherapy and second assessment was performed three months after the first assessment.
g) Group 7:
Test-retest group; these patients were selected randomly from the six groups above and they were requested to complete the questionnaires within 7–14
days after the first assessment.
h) Group 8: Liver metastasis group; these patients were required to complete the questionnaires once.
Questionnaires were checked and data entered into a database in Microsoft Excel and later transferred to SPSS version 20.0 for Windows for analysis. Data coded based on the guidelines as contained in the EORTC scoring manual [14
]. Two-sided tests were used, and p-values of
0.05 were considered statistically significant.
Descriptive statistical analysis was performed for all variables. Continuous variables were reported using means and standard deviations or median and inter-quartile range. For dichotomous variables, absolute numbers and percentages were presented.
Completion rate and time taken to complete the questionnaires were assessed [15
]. Confirmatory factor analysis was performed.
Multitrait scaling analysis
Multitrait scaling analysis was employed to examine item convergent validity. Each items scale’s Pearson’s product moment correlation should exceed 0.4 for convergent validity on all scales. Inter scale correlations were used to measure discreminant validity which was a measure of item own scale correlation in relation to other scales. It is hypothesized item own correlation should be higher than with the other scales.
Internal consistency reliability
The internal consistency was assessed using the cronbach’s alpha coefficient. Coefficients of above 0.70 were considered acceptable for group comparisons.
Reproducibility (test-retest reliability)
Intraclass Correlation Coefficient (ICC) was used to assess the test-retest reliability. A score of one indicates perfectly reliable, zero perfectly unreliable test.
Subjects were compared based on the treatment groups such as subjects receiving chemotherapy, versus surgery alone, patients with and without stoma and performance status (KPS score of
81%). Non parametric test Wilcoxon rank sum test were used for the comparison of the groups.
Responsiveness was measured by comparing changes over time of the instruments in subgroup of patients undergoing palliative chemotherapy (Group 6) and after closure of a temporary stoma (Group 4).
Instrument acquisition and pilot testing
This study was designed to validate three EORTC questionnaires on a sample of colorectal cancer patients in Malaysia. The questionnaires included were; EORTC QLQ-C30 (version 3.0), EORTC QLQ CR29 and EORTC QLQ LMC21.
Questionnaires were acquired from the relevant developers and EORTC Quality of life group. Translation was performed according to the group guidelines; pilot testing was conducted and reported below. The validation testing is ongoing.
Pilot testing was conducted from January 02, 2012 to January 31, 2012. Questionnaires were administered to 10 patients for Bahasa Malaysia and 10 patients for Chinese-Malaysia versions. Due to inadequate number of patients suitable for QLQ LMC21 questionnaire and the Tamil version of all three questionnaires, we decided to remove them from our study.
Details of the Socio demographic and clinical characteristics of the patients were presented in Table .0 for Bahasa Malaysia and Chinese-Malaysia.
Socio demographic and clinical characteristics of patients included in the pilot testing of the EORTC QLQ-CR29 (Bahasa-Malaysia and Chinese-Malaysia)
Mean age was 58
years, Male: Female ratio 1:1, 30% had attained tertiary education, 60% had rectal cancer followed by sigmoid 30% and recto sigmoid 10% respectively. 50% had stoma, 70% chemotherapy and 60% had radiotherapy. Karnofsky performance status was ≥80 in 90% of patients.
Mean age was 67
years, Male: Female ratio 1:1, 70% had attained tertiary education, rectal cancer 30%, recto sigmoid, ascending and transverse colon each 20% respectively. 50% had stoma, 60% chemotherapy and 40% had radiotherapy. Karnofsky performance status was ≥80 in 50% of patients.
There was no difficulty in understanding the introduction as well as the instructions for completion of the questionnaire. Mean duration for completion of a set was found to be 8
Patients consider the time of administration which was immediately after the consultation with the doctors to be inappropriate. Questionnaire items 1–28, 31–54 were not associated with any difficulty in answering, nor were they confusing, difficult to understand or offensive. Questions 29–30 and 56–59 were associated with some problems. Questions 29 &30 are questions about the overall quality of life and general health status. Three patients answering Bahasa-Malaysia versions indicated their difficulty in differentiating between question 29 & 30. They considered the duo to mean the same and suggested the questions to be merged. Questions 56 to 59 are questions about the sexual activities. (Questions 56–57 are for male and 58–59 are for female patients respectively). Three patients (Bahasa-Malaysia version) and five patients (Chinese-Malaysia version) felt the questions were not necessary because they were no more sexually active.