Introduction

Globally, colorectal cancer (CRC) is among the highest in North America and Europe [1]. Early detection of CRC using the fecal occult blood test (FOBT) has been shown to reduce relative risk of mortality by 15% if performed biennially compared to no screening [2]. Population-based CRC screening using the FOBT or fecal immunochemical test has been implemented internationally [3]. Yet, CRC screening uptake is low among South Asians (SAs) settled in western countries [4, 5].

South Asian (SA) populations are growing in the United Kingdom (UK), United States of America (USA), and Canada due to increasing global migration. Prior studies report low CRC screening among SAs. For instance, CRC screening was low among SAs compared to non-Asian UK populations, 32.8% versus 61.3%, respectively [4]. In the USA, SAs were less likely to have obtained CRC screening compared to non-Latino Whites, 42.3% and 57.7%, respectively [5]. CRC risk is low among newcomer SAs; however, with time spent in settlement country, risk approaches similar incidence as native-born populations [6,7,8,9,10]. This increase has been linked to post migration factors such as the adoption of westernized lifestyle behaviors [10, 11].

Survey research is used to examine cancer screening behaviours; yet, less attention has been paid to survey development with inclusion of ethno-cultural relevance. Our team developed a survey to examine prevalence, beliefs and attitudes, facilitators and barriers to CRC screening among SAs in the UK, USA, and Canada. Data was drawn from phase one and phase two of a mixed method study [14, 15], and expert consultation [12].

Main text

Colon Cancer Screening Behaviours Survey

The survey was developed as part of an exploratory mixed method study conducted in Canada, and underpinned by critical social theory [13] that included a sco** study, focus group study, and survey development and cognitive pre-test study [12]. The sco** and focus group studies [14, 15] formed the basis of initial work to uncover concepts to examine CRC screening behaviours among SAs. The survey was cross-culturally translated and adapted into Urdu, and cognitively pre-tested (English and Urdu) with SAs in Canada [16]. This paper reports on the systematic and scientifically rigorous steps undertaken prior to reaching the final stage with the aim of encouraging comprehensive approaches in the field.

South Asian populations

SAs are individuals from India, Pakistan, Bangladesh, Sri Lanka, and the SA diaspora (i.e. SAs migrating from countries such as South Africa) [17, 18]. Rapidly growing in the west, SAs represent: the third largest Asian group in the USA [19]; the second largest minority group in the UK [20]; and, the first largest minority group in Canada [21]. Sco** and focus group studies [14, 15] elucidated on socio-cultural context of cancer screening among SAs.

Screening among South Asians

The sco** study reported on SA beliefs, attitudes, and barriers regarding breast, cervical, and CRC screening in the UK, USA, and Canada [14]. Common barriers included: (a) lack of knowledge [22,23,24,25]; (b) language barriers [22, 25]; (c) low literacy [23, 26]; (d) low self-perceived risk [22, 27]; and (e) cost and time [23]. Few studies examined SA beliefs and attitudes related to CRC screening, particularly in Canada [23, 26]. Given the uniqueness of CRC screening (i.e. procedures and gender preferences), a focus group study was required.

To examine CRC screening behaviours, focus groups with 42 SAs originating from India, Pakistan, Bangladesh, Mauritus, Uganda, and Kenya were conducted in Canada [15]. The Behavioural Reasoning Theory [28] incorporates behavioural and social context, and guided the interview protocol, which was later pilot tested with SAs. SA research assistants trained to conduct focus groups recruited participants from community settings in Ontario [15]. Findings revealed factors that influenced CRC screening, such as: benefits of early detection; screening was not believed to be necessary; lack of knowledge, and family physician support and access [15]. These collective findings informed our conceptual model.

Conceptual model

Key concepts identified from our studies [14, 15] were charted (see Table 1); thereafter, a review of health behaviour theories was conducted to determine conceptual congruence. Behavioural concepts from the Health Belief Model (HBM) [29] and the Theory of Planned Behaviour (TPB) [30] aligned well to our key concepts (see Fig. 1, Box A & B). Perceived susceptibility aligned with SAs low perceived risk of cancer [14, 15]. Perceived severity reflected SA beliefs that cancer was scary, and had poor outcomes [14, 15]. Perceived benefits linked to reduced worry and improved survival with screening [14, 15]. Perceived barriers aligned to language and cost barriers [14, 15]. Perceived self-efficacy related to low confidence with completing the test (i.e. FOBT) [14]. Subjective norm reflected the influence of family and physician to have screening [14, 15]. Socio-contextual variables were also considered [31]. The emergent conceptual model is a product of primary research with SAs and existing theoretical literature (see Fig. 1).

Table 1 Key concepts identified
Full size table
Fig. 1
figure 1

Conceptual model of the Colon Cancer Screening Behaviors Survey. * Denotes an added item to cover missing content identified from the sco** and focus group studies [10, 11]

Full size image

The HBM [29] and TPB [30] were used to operationalize conceptual definitions (Table 1). The HBM [29] and the TPB [30] have been incorporated into measures that examined CRC screening among diverse populations [27, 32,33,34], and a few have combined concepts from both into a single survey [35, 36]. Thus, we were confident in our decision to utilize these behavioural concepts as the best fit to our key concepts.

As depicted in Fig. 1 (Box C), two behavioural outcomes include intention and adherence to CRC screening (Table 1). Intention is a precursor to CRC screening, while adherence is compliance with screening recommendations [37, 38]. According to the HBM [29] and TPB [30], perceived susceptibility, perceived severity, perceived benefits, perceived barriers, perceived self-efficacy, and subjective norm directly influence CRC screening intention and/or adherence [39, 40]. In our conceptual model, key concepts equally influence CRC intention or screening with no direct relationship between concepts specified.

According to the HBM [29], modifying factors indirectly influence behavioural outcomes (see Fig. 1). Screening history and socio-demographics represent socio-cultural context of screening and may directly influence outcomes [14, 15].

Comprehensive literature search

To identify articles that reported on candidate measures assessing cancer screening, a comprehensive literature search guided by DeVellis framework [41, 42], and librarian recommendation was conducted. Five databases were searched: Ovid Medline [1946 to March week 1 2015], EMBASE [1947 to 2015 March 09], PsychoINFO [1806 to March week 1 2015], CINAHL [1988 to 2015, March 9], and Health and Psychosocial Instruments [1985 to March 2015]. Grey literature search of the UK Bowel Screening Program and Cancer Research UK websites were completed. Reference lists were reviewed.

A combined total of 426 citations were returned. In selecting articles, inclusion and exclusion criteria were applied: (a) availability in English; (b) any cancer screening; (c) examination of beliefs, attitudes, facilitators or barriers using defined measures, and; (d) any population. Duplicates, dissertations, reviews, conference abstracts, and books were excluded. A total of 142 citations remained after applying inclusion and exclusion criteria. Of these, 78 were excluded because they were cross-sectional application studies that used previously developed or adapted measures. The remaining 64 articles reported on newly developed, previously created, and adapted measures; most were initially developed for breast cancer screening and later adapted to assess CRC screening [32, 43,44,45].

We decided to focus on measures that examined CRC screening because of unique procedures; 24 articles underwent full-text review. A further 19 articles were excluded because measures did not match key concepts or lacked conceptual definitions (Additional file 1: Literature search flow chart).

Five published surveys [45,46,47,48,49] were selected as the best match, and had the most promise because they were based on health behaviour theory [29, 30], had been previously validated, and provided sufficient detail to assess the conceptual basis [50,51,52,53,54].

Critical appraisal of selected measures

Critical appraisal examined the match between key concepts, selected candidate measures, and SAs because conceptual relevance and socio-cultural alignment were more important than statistical outcomes [55, 56]. Nine items from the Evaluating the Measurement of Patient-Reported Outcomes (EMPRO) [57] were used to assess selected measures for conceptual and measurement model (n = 7), content validity (n = 1), and response burden (n = 1). Two appraisers independently critiqued the first articles, met to discuss results, and reach consensus. Appraisers were selected based on expertise with survey measurement research, and cancer screening research with SAs. The remaining critique of articles was completed by one assessor.

Fifteen potential measures were assessed, and they all met EMPRO criteria for conceptual match (see Table 2). Overall assessment results of “strong recommendation” or “recommendation with potential alterations” were deemed acceptable for inclusion. A final step involved expert consultation of selected measures to retain in the survey.

Table 2 EMPRO tool assessment and scores
Full size table

Expert consultation

Public health practitioners (n = 3) with expertise in cancer screening research and program evaluation were consulted because they worked closely with SAs in Ontario. Ethics approval was obtained from the University of Toronto (#27857) and Brock University (#12-036) Research Ethics Boards. Verbal consent was approved for consultations. Nominal group methods was used [58] to obtain input and endorsement on selected measures to ensure cultural relevance and acceptability for SAs.

The process began with presenting the background and key concepts. Each candidate measure was independently reviewed to ascertain which aligned best to key concepts. Voting cards were used to log selections and provide additional comments. Afterwards, discussion occurred regarding selected measures and potential problems with some items.

Although measures provided good overall matches, selected measures did not fully cover concepts relevant to SAs uncovered in our prior studies [14, 15]. Consequently, three experts in survey measurement and cancer screening were consulted at a separate meeting to provide feedback and ensure complete conceptual coverage in the survey [41]. A total of 17 items [59,60,61,62,63,64,65] were added to key concepts for completeness (see Fig. 1). Modifications to items were also required. Informed by our conceptual model, the Colon Cancer Screening Behaviours Survey incorporated 84 items.

Cross-cultural translation and adaptation and cognitive testing

Cross-cultural translation and adaption into Urdu was conducted [16] following recommended procedures [66, 67]; two individual forward translations; a discussion meeting including a final synthesis report; and, expert committee review. This process resulted in the identification of key issues including missing terms, and difficult or incorrect translation of terms. Thereafter, the survey was cognitively pre-tested with 30 SA immigrants in Canada [16]. General design, culture, and gender related revisions were made, and the survey was further tested with no major problems.

Conclusions and recommendations

This study adds to prior CRC screening research conducted with SAs in the USA [27, 68] and the UK [25]. Our survey is unique because it was cross-culturally translated and adapted into Urdu, a language chosen because it is widely understood among diverse SAs in the spoken form. In other studies examining CRC screening, surveys targeted English [68] and Hindu and Gujarati speaking SAs [27]. Assessing CRC screening behaviours among SAs requires an adaptation to socio-cultural context. The purpose of our survey is to examine prevalence, beliefs, attitudes, facilitators and barriers to screening among SAs in Canada. Once psychometrically tested, it may be used with English and Urdu speaking SAs in other contexts.

Changes made to published measures were considered necessary to cover key concepts; however, changing survey measures altered measurement properties, which improved measures because of the relevancy to assess CRC screening among SAs; conversely, they could also have been weaken. We believe cognitive testing improved the survey, but it requires further assessment of psychometric properties.

Limitations

The directed literature review was successful in yielding validated measures; however, because we restricted it to psychometrically tested measures, some untested measures conceptually aligned may have been missed. Nevertheless, modified measures in the survey require psychometric testing. The sco** study findings [14] provided relevant concepts applicable to diverse SAs in the UK, USA and Canada where most studies emerged, while focus group study findings [15] reflected SAs in Canada and thus, may not be representative of those in other contexts. We believe incorporating findings from both studies [14, 15] expanded the breath of understanding CRC screening among SAs in multiple contexts. Consultation capitalized on expertise from individuals working directly with SAs promoting cancer screening and research, and survey measurement; however, only a few experts had international experience.