Development and validation of an instrument to measure informational needs of breast cancer patients in Sri Lanka

Background: Breast cancer is the second most common cancer in the world and the commonest in Sri Lanka. The provision of relevant and accurate information in a social-culturally appropriate manner will lead to proper understanding of the disease and treatment options.  Objectives:The aim of this study was to develop and validate the Sri Lankans’ Informational Needs Assessment Questionnaire of Breast Cancer (SINAQ-BC). Methods: Item generation was through review of literature and using qualitative research techniques among stakeholders.  Thus developed and pretested 69 item SINAQ-BC was administered through interviews to 150 women at least 18 years of age who have been diagnosed as having breast cancer within the period of one year prior to the study and Exploratory Factor Analysis (EFA) was used for further item reduction. Internal consistency reliability and acceptability were also assessed. Results: Exploratory factor analysis determined that 58 items should be retained in the questionnaire and that they be best organized into five subscales lab elled as disease, diagnosis, treatment, physical care and psychosocial care. SINAQ-BC showed good internal consistency reliabilities with Cronbach’s alpha of 0.76 to 0.93.An average 30 minutes (+5 mins) was taken to complete the SINAQ-BC and it did not require any additional explanations demonstrating its acceptability. Conclusion: The SINAQ-BC demonstrated to be a valid and reliable tool to assess informational needs of breast cancer patients.It is recommended to be used to determine informational needs of women with breast cancer and as a tool to assess effectiveness of educational interventions.


Introduction
Breast cancer (BC) is the second most common cancer in the world and the commonest cancer among women. An estimated 1.67 million new cancer cases were diagnosed in 2012. Incidence rates vary nearly four-fold across the world regions, with rates ranging from 27 per 100,000 in Middle Africa and Eastern Asia to 92 in Northern America [1]. In Sri Lanka, breast cancer is the commonest cancer among women. The age standardized incidence rate (ASR) of BC in Sri Lanka was 23.0 per 100,000 population in 2010. The crude incidence rate (CR) of the BC has increased by approximately 1 per 100,000 population from the levels in 2009.One in every 40 females in Sri Lanka has a risk of developing BC during their lifetime (0-74 years) [2]. Upon diagnosis and during the life thereafter, BC causes a significant emotional impact upon the lives of patients as well as family and friends, work colleagues and the greater community. Many studies reveal that information plays an extremely important role in patients' ability in coping with the disease and their quality of life. Fredette in 1995 [3] stated that information on BC requires the afflicted woman to be well informed to understand her diagnosis, to develop a positive attitude towards the life amidst the threatening implications of the cancer, and to cope well with the condition. Provision of relevant and accurate information in a social-culturally appropriate manner will lead to proper understanding of their treatment options and outlook they need to adopt towards a healthy future [4]. It is therefore important to assess the information needs of women with BC since such an assessment has not been done in Sri Lanka.
Measuring of informational needs require a tool which should be culturally appropriate. A tool to measure informational needs of BC, named as Toronto Informational Needs Questionnaire (TINAQ), has been developed in the European region [5] but none have been developed for Asian Region. Therefore, development of an appropriate instrument to assess informational needs of BC survivors to be used in the Sri Lankan context was considered of paramount importance. It is imperative that validity of the survey instrument to measure informational needs of BC patients be also determined. This study aimed to develop a questionnaire to measure the informational needs of a BC patient and to determine its validity and reliability to assess the needs of information related to diagnosis, treatment modalities and follow up care.

Development of the tool
The development of the tool in the form of an interviewer administered questionnaire followed a step wise process.
First step was generation of items. Items were identified through review of literature and through focus group discussions. Focus group discussions were carried out with BC survivors, the lay people and the oncologist counselor attached to counselling unit in the National Cancer Institute, Maharagama, and other staff, to find out suitability of selected items and identify the new items. Sixty nine items were generated from a number of sources including findings in the literature and opinion from the multidisciplinary panel comprising experts as well as stakeholders i.e. oncologists, community physicians, radiologists, physiotherapists, nutritionist, medical counsellors, social counsellors, nurses and cancer patients.
In the second step, a consultative meeting was organized at which the same panel grouped the items into different subscales or domains that have been identified from literature survey [5]. Using collective professional experience, items that were thought to be repetitions and culturally inappropriate were removed by consensus. Items were then grouped into eight such subscales or domains of informational needs namely, diseases, diagnosis, investigative tests, treatments, physical care, psychological, care family and finance. The third step was to convert the items into questions and to develop a draft interviewer administered questionnaire to assess breast cancer patients' information needs. The researchers initially drafted the questionnaire in English. The questionnaire was designed to assess the information needs by inquiring the subjects to rate the importance of each item on a five point scale of 1 (not important), 2 (slightly important), 3 (moderately important), 4 (very important) and 5 (extremely important).

Translation of the tool
Two translators, both with a high level of proficiency in English and Sinhala, independently translated the questionnaire into Sinhala. The two Sinhala translated versions were then reviewed by an independent expert proficient in both English and Sinhala languages. The items which were agreed upon were accepted for forward translation. Another two bilingual translators, both with a high level of proficiency in English and Sinhala, independently translated the provisional forward translation back into English without referring to the original English version. The two English translated versions were then reviewed by an independent expert, proficient in both English and Sinhala languages. Same way was used to translate the questionnaire into Tamil using two translators both with high level of proficiency in English and Tamil.
Then questionnaire thus developed was named as Sri Lankans' Information Needs Assessment Questionnaire of Breast Cancer (SINAQ-BC).

Pretesting of the SINAQ-BC
Pretesting the SINAQ-BC was done among ten BC patients at the Cancer Unit of Teaching Hospital in Galle, to assess the understanding and the clarity in the wording of questions. Administration of each questionnaire was followed by a structured interview conducted by the PI. Based on the interviews, a few modifications were done to the instructions stated for some items.

Validation of the SINAQ-BC
The SINAQ-BC together with its instructions was shown to a panel of experts, not involved in developing the study instrument, to assess the content validity. The experts agreed that the content assessed by the SINAQ-BC would be a good reflection to improve patients' knowledge on their disease and services provided for them. The experts also decided that content was subjectively appeared on the SINAQ-BC, confirming its content validity and face validity.
In appraising construct validity, Exploratory Factor Analysis (EFA) using principal component analysis (PCA) was used to assess whether SINAQ-BC had any redundant items and also to assess the suitability of its seven subscales structural. Internal consistency reliability of the SINAQ-BC was assessed using Cronbach's alpha. Bartletts test of sphericity and Kaiser-Meyer -Olkin measure was performed to assess sampling adequacy.

Study Population
The study population was women at least 18 years of age who have been diagnosed as having BC within the period of one year prior to the study.

Sample Size and sampling technique of the validation study
Nunnally [6] opines that an initial sample of at least 100 subjects is necessary to provide sufficient data for item analysis of a tool being validated. The researcher further says that domain's (subscales) internal consistency, reliability estimates can be made with five subjects per item. The largest subscale in the SINAQ-BC has 19 items with a minimum sample required being 95 (1:5).
Considering these references, 150 subjects were required since this instrument has 69 total items. Patient visiting outpatient clinics and / or admitted at National Cancer Institute, Maharagama were invited to participate in the study. The patients were not approached on the first day of any new treatment for ethical reasons. Written informed consent was obtained from each patient. A total of 170 BC patients were invited and 156 consented. Interviews were conducted at the convenience of the patient.

Data Collection
The SINAQ-BC developed above was administered by the PIafter the interview.

Data Analysis
Exploratory Factor Analysis (EFA) is essential to determine underlying constructs for a set of measured variables. Exploration for redundant items in the SINAQ-BC was based on the results of the exploratory Factor analysis. Following two criteria were used to determine whether an item would be retained. 1. Items with average inter-item correlation of less than 0.20 as the low correlation indicate the item may not be related to topic of interest. 2. When two items had inter-item correlations of 0.80 or higher, indicate that the two items measured the same informational need. Using the results of EFA, the subscale structure of the SINAQ-BC was also reviewed. Inter correlations of the subscales to total scale and the intercorrelations of the subscales were examined to confirm the suitability of the subscales that were decided upon. Internal consistency reliability for the SINAQ-BC and its subscales was assessed using Chronbach's alpha.
Acceptability was assessed using the average time taken to complete the interview and the response rates. All analysis was carried out using a statistical package for Social Science Version 21.0 (SPSS 21.0) by PI. The project was approved by the Ethics Committee in the Faculty of Medicine, University of Colombo, Sri Lanka [EC-10-035].

Validation of Sri Lankans' Informational Needs Assessment Questionnaire of Breast Cancer (SINAQ-BC)
The response rate was 96.2%. Age range of the study units were 27 -70years with mean age of 56.28 years (SD =± 12.87; 95% CI = 54.20 -58.36). The majority were married (70%) had completed either GCE (O/L) or GCE (A/L) (64%, n=96) and 9% (n= 9) possessed a degree or higher qualification. Approximately half (49%, n=74) reported an average household income of Rs. 20,000 -Rs. 30,000. Only 34% had only undergone surgery as the treatment modality for breast cancer. Thirty three percent had undergone chemotherapy and surgery while balance 33% had undergone radiation therapy and surgery.

Exploratory Factor Analysis
Exploratory Factor Analysis performed by principal component analysis using data. Sampling adequacy was confirmed by Barlett's test of Sphericity which was significant at p<0.01 (chi square = 1735.4, df = 105, p = 0.000 and Kaiser-Meyer-Olkin measure of sampling adequacy which was 0.697, being well above the accepted level of 0.6).
Nine items were deleted because their average interitem correlations were below 0.20. Finally the SINAQ-BC contained 58 items. All 58 items had item-total correlations of 0.25 or higher, levels considered acceptable by Nunnally [6]. In all retained items, the means fell above the midpoint of the Likert scale (1-5) with standard deviations in the range of 0.4 -1.7.
Analyzed 58 items in SINAQ-BC, indicated five subscales. The Eigen values ranged from 1.020 -5.074. The total variance explained was 68.1%. In the Exploratory Factor Analysis, five subscales were formed out of the seven original subscales in the SINAQ-BC where the two subscales diagnosis and investigative tests were combined as diagnosis and the two subscales on family and finances were combined into a subscale of psychological care. Thus, titles of the subscales after combining, the number of items of each subscale with the relevant factor loadings are shown in Table 1. Behaviour of newly diagnosed women .422 As shown in the Table 1 all the retained items contained a factor loading of 0.393 or more. Table 2 shows that the inter correlations of the subscale to total scale ranged from 0.77 -0.89 and the inter-correlations of the SINAQ-BC subscale scores ranged from 0.47-0.84. This confirmed that each subscales best correlated with the total tool compared to its correlation to other subscales.

Acceptability of SINAQ-BC
None of the women had difficulty in responding to the SINAQ-BC and the meantime taken to complete the interview was 30 minutes (±5 mins) and all were able to respond to the items without additional explanations.

Discussion
The SINAQ-BC is the only validated instrument developed to assess informational needs of BC patients in Sri Lankan context. It is expected to function as a valid measure to identify information needs which facilitate designing of the patient educational programmes.
The existing literature is not rich enough to provide a sound conceptual foundation for informational needs of BC patients. As Flick(1998) and Neuman (1997) have suggested that focus group is useful in exploratory research or in generating new ideas to develop hypotheses [7,8]. Hence, the present study used both these methods to generate items of the SINAQ-BC.
A hospital-based cross sectional study design was employed to assess the validity and reliability of the SINAQ-BC. Thus, purposive sampling technique was considered as the best technique to ensure representation of women with an initial diagnosis of breast cancer in the treatment phases of surgery, chemotherapy, radiation therapy.
To be considered as acceptable items in the questionnaire, item analysis was completed as done by Nunnally [6]. Items with item-total correlations of 0.25 or higher levels was considered as an acceptable item in the questionnaire, correlation below 0.20 considered as not interesting items and if two items have inter-item correlation of 0.80 or higher, it considered as same informational need, hence, retained only one item. Rarely both items were retained if they seem to be dealing with different informational needs. Content validity of the instrument was established as emphasized by Latwin in 1994 to assess the content completeness of the instrument [9]. Nunnally [6] opines that an initial sample of at least 100 subjects is necessary to provide sufficient data for item analysis to decide whether the research should be continued or abandoned. The researcher further says that domain's (subscales) internal consistency, reliability estimates can be made with five subjects per item. Accordingly, the largest subscale in the SINAQ-BC has 19 items, a minimum sample would be 95 (1:5). Ferguson & Cox [10] stated at their pre-analysis that minimum participants to variable ratio would be in between 2:1 -6:1. Considering these references, 150 subjects were used for the study since this instrument has 69 total items.
Factor analysis is a complex statistical procedure, which essentially reduces a large set of variables into much smaller set of variables with common characteristics [11]. As such, it is an empirical method of determining 'the underlying dimensionality' of a set of variables [12]. Factor analysis suggested by Galloway et al. (1997) [5] for the Toronto informational Needs Questionnaire-Breast Cancer (TINQ-BC) was used in the present research.
Reliability refers to the dependability and repeatability of an instrument. Wilkin et. al. [13] explained that if an instrument is more reliable, there is lower element of random error. Reliability also provides an estimate of homogeneity. Cronbach's α is used to estimate the internal consistency reliability of SINAQ-BC. The alpha coefficient for each of the subscales and the total scale was above 0.80, which is considered as a satisfactory estimate of reliability by Burns & Grove [14]. Nunnally also said that this is the level necessary for instruments used in basic research [6]. A tool (TINQ-BC) developed by Galloway et.al in 1997 [5] also showed that overall alpha coefficient was 0.97 while the corresponding figure for SINAQ-BC was 0.95. Domain α of TINQ-BC ranged from 0.73 to 0.93 while for SINAQ-BC it was 0.76 to 0.93.

Conclusion
The SINAQ-BC demonstrated to be a valid and reliable tool to assess informational needs of BC patients in relation to diagnosis, treatment modalities and follow up care. This is evident by high internal consistency, reliability, content validity and construct validity. Its validity to assess informational needs was confirmed irrespective of the treatment modality that the patient was undergone. It is recommended that the SINAQ-BC be used to determine informational needs of women with BC in Sri Lanka and as a tool to assess educational interventions to improve knowledge and coping of BC among patients.