Maltese Translation and Adaptation of Champion’s Health Belief Model Scale and the Revised Illness Perception Questionnaire for Breast Screening Among Maltese Women

History and Development of the Champion’s Health Belief Model Scale

The health belief model (HBM), developed in the early 1950s, is a behavior prediction model, comprising six fundamental constructs: perceived susceptibility, perceived seriousness, perceived benefits, perceived barriers, cues to action, and self-efficacy (Jahanlou, Lotfizade, & Karami, 2013). Champion developed and validated a scale in 1984 (Champion’s Health Belief Model Scale [CHBMS]), consisting of 36 items to measure perceived susceptibility to breast cancer as well as perceived benefits and barriers to BS (Champion, 1984). In 1999, CHBMS-MS, excluding the breast self-examination used in the original studies, showing significant correlation between mammography compliance and high scores in the Susceptibility and Benefit subscales, whereas perceived barriers were associated with lower screening compliance (Huaman et al., 2011).

The scale was originally validated in Indiana, United States by Champion (Champion, 1999) in a cohort of 804 women aged 50 years and older in a population of Whites (68%) and African Americans (30%), accounting for 54% of the variance and showing adequate construct validity and reliability. Since then, Champion’s HBM scale has been tested for reliability and validity around the globe and translated for Iranian (Hashemian, Shokravi, Lamyian, Hassanpour, & Akaberi, 2013; Taymoori & Berry, 2009), Lithuanian (Zelviene & Bogusevicius, 2007), Malaysian (Parsa, Kandiah, Mohd Nasir, Hejar, & Nor Afiah, 2008), Arabic (Mikhail & Petro-Nustas, 2001), Korean (Lee, Kim, & Song, 2002), Chinese Australian (Kwok, Fethney, & White, 2010), Turkish (Secginli & Nahcivan, 2004; Norman & Brain, 2005; Lunt, Bowen, & Lee, 2005), African American (Champion et al., 2008), and Spanish-speaking American women (Medina-Shepherd & Kleier, 2010). Findings of these studies have provided support for the validity and reliability of these HBM-based scales, although poor construct validity was shown in a Peruvian-translated version (Champion et al., 2008) and in a Spanish version (Esteva et al., 2007).

Because HBM is widely cited (Noar & Zimmerman, 2005), we used CHBMS-MS (Champion, 1999) to translate, adapt, and test among Maltese women. HBM, however, only explains some of the variation in BS behavior such that it does not consider the impact of emotions (such as fear; Norman & Brain, 2005), nor does it accommodate social and environmental influences of past behavior (Lunt et al., 2005) which is why other models have been incorporated in studies to understand BS uptake (Cameron, 2008). In response to HBM’s limitations, an instrument associated with the common-sense model (CSM) of health and illness behavior (Cameron, 2008) was also translated, adapted, and tested.

History and Development of the Revised Illness Perception Questionnaire

In the late 1960s and early 1970s, Leventhal explored how fear messages in relatively acute situations might lead individuals to respond to the health threat communication by taking health-promoting actions (Broadbent et al., 2015), such as wearing seat belts or giving up smoking (Leventhal, Hudson, & Robitaille, 1997). Subsequent research by Leventhal and colleagues in 1980s led to the development of the CSM of self-regulation, which proposes that individuals develop two parallel, yet interrelated, representations of the stimulus (cognitive and emotional) in response to a perceived threat (Leventhal et al., 1997). Hence, CSM provides a framework for understanding how individual symptoms and emotions experienced during the health threat or diagnosis influence illness perceptions and guide subsequent coping behavior (Diefenbach & Leventhal, 1996). This model was later used to understand illness prevention and preventive behavior intentions (Figueiras & Alves, 2007).

The Illness Perception Questionnaire (IPQ; Weinman, Petrie, Moss-Morris, & Horne, 1996) was developed in light of self-regulation theory to provide a quantitative assessment of the five components of illness representation—identity, cause, timeline, consequences, and control/cure in Leventhal’s self-regulation model (Moss-Morris et al., 2002). These five dimensions have been studied in breast (Anagnostopoulos et al., 2012) and colorectal screening (Orbell et al., 2008).

Subsequent measures include the Brief Illness Perception Questionnaire (B-IPQ; Broadbent et al., 2015), the Revised Illness Perception Questionnaire (IPQ-R; Moss-Morris et al., 2002), which examines illness beliefs and behaviors within specific groups of patients, or groups at risk from an illness, and an adapted version of the IPQ-R for “healthy” individuals (IPQ-RH) in recognition of the unique characteristics of asymptomatic populations (Figueiras & Alves, 2007). To remedy shortcomings in the original IPQ scale, the IPQ-R was developed by Moss-Morris et al. (2002) as a more comprehensive, psychometrically acceptable, quantitative measure to include measures of perceptions of illness duration (“acute/chronic timeline”), fluctuation in illness over time (“cyclical timeline”), perceptions of “treatment control” and “personal control” over illness, “illness coherence” (how clear and comprehensive an individual feels her illness to be), and “emotional representations” (feelings of depression, upset, anger, worry, and anxiety). Subsequently, the IPQ-R has been validated for use in diverse diseases or healthy populations (Chen, Tsai, & Lee, 2008), with language-specific validated IPQ measures, such as Italian (Giardini, Majani, Pierobon, Gremigni, & Catapano, 2007), Swedish (Brink, Alsén, & Cliffordson, 2011), Greek (Giannousi, Manaras, Georgoulias, & Samonis, 2010), Croatian and Lebanese (Petrak, Sherman, & Fitness, 2015), and Portuguese (Figueiras & Alves, 2007) versions. However, it has not yet been adapted and validated for Maltese asymptomatic and/or symptomatic women. Hence, we adapted the IPQ-R in this study to make it appropriate for both healthy women and those with cancer.

Data Sources and Study Design

The study was conducted during June 2015, as part of a larger cross-sectional study about BS in Malta. The parent study was approved by the School Research Ethics Committee at the School of Health Sciences, University of Stirling (SREC14/15-Paper No. 18v4), and by the Maltese Health Ethics Committee (HEC 02/2015). Permission to use the scales (CHBMS-MS and IPQ-R) was sought from the respective authors (Prof. Victoria Champion in 2013 for CHBMS-MS use and Prof. Rona Moss-Morris in 2014 for IPQ-R use). Permissions were also received from the chief medical officer, the chief executive officer (Primary Health Care Department), and the health data protection officers in primary and secondary care in Malta.

Sample and Procedures

Four translators were recruited for the translation pathway as follows: two translators (i.e., a European translator working in Brussels who was also a bilingual native speaker of both Maltese and English languages and a Maltese expert translator) translated the instrument from English to Maltese (Steps 2–3) and two different bilingual translators (i.e., a bilingual expert from the Health Ministry and an expert interpreter at the University of Malta) back-translated the instrument from Maltese to English (Step 4).¹ An expert panel (n = 12) was set up to ascertain content validity and to verify that it is clinically meaningful to experts in the clinical area (Anagnostopoulos, Dimitrakaki, Niakas, & Tountas, 2013). The 12 members comprised the lead researcher for this study, the 4 expert translators/interpreters, a statistician with 10 years’ experience in statistical research and analysis, 2 mammographers (Maltese and Scottish radiographers), a BS client, a breast cancer survivor, a consultant, and a clinician.

A focus group was conducted with a convenience sample of asymptomatic women (n = 6) to pilot test the adapted Maltese version of the instrument. Three of the women were housewives (53, 55, 58 years, respectively) who had attended BS, two were public employees (59, 60 years, respectively) who had not attended BS, and the other was a retired 62-year-old midwife who had also not attended BS when invited.

A convenience sample of 15 women (n = 15) participated in structured face-to-face interviews to assess comprehensibility and suitability of the research instrument and to ensure understanding of all scale items in both languages. Women were recruited from the BS center and were BS attendees, aged 50–60 years. The convenience sample was recruited because it was felt that such women would be interested in engaging with such a topic (Creswell & Plano Clark, 2011), thereby giving access to a range of women with different backgrounds (Kaltsa, Holloway, & Cox, 2013). Women with prior history of breast cancer or breast surgery, those who sought breast cancer treatment, as well as nonbilingual women were excluded.

Participants were assured that they had no obligation to participate, that their participation was voluntary, and that they could withdraw from the study at any time without the need to give any reason. The cover letter provided information to the women on how the researcher would protect their anonymity and confidentiality through coding. Following explanation on the nature of the research, informed consent was obtained from the participants.

Translation and Adaptation

Figure 1 illustrates the pathway in which the translation and adaptation of the mentioned scales was undertaken, based on published methods (Champion, 1984, 1999; Yilmaz & Sayin, 2014).

Figure 1.

Translation, adaptation, face, and content validity (Maltese Breast Screening Questionnaire [MBSQ] pathway).

CHBMS-MS = Champion’s Health Belief Model Scale for Mammography Screening; IPQ-R = Revised Illness Perception Questionnaire; CHBMS-MS-M = Champion’s Health Belief Model Scale for Mammography Screening–Maltese version; IPQ-R-M = Illness Perception Questionnaire–Maltese version.

Steps 1–2: Identification of Scales and Forward Translation

Following the identification of validated scales by the researcher, initial translation of the questionnaire from English (original) to Maltese (target) languages was performed by two expert translators. This bilingual team first prepared their own translated versions; they then gave their versions to each other to verify each other’s work and finally came up with collaborative decisions about the translation.

Step 3: Reconciliation Session

The two experts met up with the researcher in a “reconciliation session” in Malta and reviewed the translation together for inconsistencies with the original English scale and to ensure that the language was kept simple to be understood by Maltese women.

Step 4: Back-Translation Into English

The adequacy of the Maltese translated instrument was evaluated using the back-translation technique. The Maltese version was back-translated into English (original language) by another team of experts (i.e., not the original translators in Steps 2–3).

Step 5: Adaptation Process

Both language versions were examined for conceptual equivalence by the expert panel (n = 12) which included the lead researcher and statistician for this study, the four translators (Steps 2–4), screening/medical professionals, and lay women. The back-translation and the original English instrument version were compared with attention given to grammar and the meaning conveyed by the words. In this “adaptation” process, the cultural and social characteristics of the translation are protected as much as possible (Kulis, Arnott, Greimel, Bottomley, & Koller, 2011).

Step 6: Face Validity Testing

A focus group was conducted with a convenience sample of asymptomatic women (n = 6) to pilot test the adapted Maltese instrument version. This cognitive debriefing procedure (Wild et al., 2005) was followed to ascertain face validity of the instrument, to ensure clarity and comprehensibility of the items, to highlight inappropriate items or response options, and to identify and test translation alternatives and modifications. This ensures that conceptual equivalence and cultural appropriateness are achieved (Anagnostopoulos et al., 2013). This group of screened/nonscreened women tested the instrument’s face validity and determined its cultural appropriateness and the accuracy of the translation, similar to the undertaken Turkish process (Yilmaz & Sayin, 2014). The researcher read the translated text aloud to the participants, following which each item was scored on a 5-point scale.

Step 7: Reconciliation Session

The scales were modified in a “reconciliation session” so that they could be administered by an interviewer, where two translators met up with the researcher in Malta to review the final version.

Step 8: Proofreading

Following proofreading, the final Maltese version was produced and entitled the Maltese Breast Screening Questionnaire (MBSQ). The following procedures were used to test the MBSQ:

Translation and Adaptation

Four queries of subcultural word comprehension were raised by the bilingual translators, which required consensus. The term breast lumps in the original instrument was translated to boċoċv f’sidrek. The second controversial term was mammogram, for which two panel members argued that some women in the target population may not be aware of early diagnostic breast tests. Although mammografija in the translated instrument was acceptable, the general known term was mammogram. Following this debate, the panel decided that both words were suitable and could be used interchangeably (i.e., mammogram, mammografija). A third controversial term was thickening of the breast. Following discussion, the panel decided on the phrase ħxuna tat-tessuti tas-sider. Another word discussed by all group members was nipple. Several controversies arose on whether to use the word nipple as is, nippla, or the pure technical phrase ras il-biżla. Most members argued that some women in the target population are not aware of the technical phrase but are familiar with the English term. This was then literally translated to nipil.

Because most women perceive breast cancer as a serious threat (Lagerlund, Sparén, Thurfjell, Ekbom, & Lambe, 2000), it was decided that the construct “perceived severity” would not be measured using HBM (Anagnostopoulos et al., 2012). Further removal of the item in the HBM-related scale would also avoid duplication because the seriousness of breast cancer was addressed in the IPQ-R scale. Moreover, because the use of both HBM and CSM often fails to address contextual constraints such as low income and education level that may influence women’s screening behavior, sociodemographic and socioeconomic factors as well as lifetime mammography use were added because of the acknowledgement of their contributions as BS determinants (Anagnostopoulos et al., 2012; Jepson et al., 2000; Lagerlund et al., 2000). The panel further added cues to action (such as physician recommendations and family history), which are often omitted from empirical studies through HBM use (Anagnostopoulos et al., 2012). Finally, based on these conclusions, the original version of the instrument consisted of 124 items and was presented to the focus group (n = 6) for testing.

Face Validity

From the original 50-item IPQ-R, two items were removed from the cancer timeline domain (“Breast cancer will last for a long time”; “I expect to have breast cancer for the rest of my life”) because they were found to confuse the women and cause consistent heightened anxiety in responders, resulting in a 48-item Maltese (M) version (entitled IPQ-R-M). Participants were asked to report their personal views about breast cancer rather than their perceptions of an illness personally affecting them. For example, “My illness has serious economic and financial consequences” was replaced with “Breast cancer has serious economic and financial consequences"; “My illness will last for a long time” was replaced with “Breast cancer will last for a long time” following which reverse scoring was eliminated for this item to read “Breast cancer will last for a short time” because of the misunderstanding, confusion, and anxiety experienced by all women. The IPQ-R risk factors domain title were also amended to read “Risk/Lifestyle Factors,” whereas the sections “personal” and “treatment” control were categorized under the heading “Curability/Controllability.” For the lifetime mammography use domain, 1 item was deleted to avoid overlap (“a mammogram prior to breast screening” yes/no). Hence, the final Maltese instrument (MBSQ), comprising the Maltese (M) scales CHBMS-MS-M and IPQ-R-M, consisted of 121 items that were clustered into 11 subscales for sociodemographic and health status (20 items), 4 subscales for lifetime mammography use (17 items), 5 subscales for health beliefs (36 items), and 7 subscales for illness perceptions (48 items).

The earlier-mentioned method found the instrument to be acceptable and ready for use in psychometric testing among the target population. Of the convenience sample of 15 women (n = 15), the mean age was 54.5 years ± 3.2 years (SD); 6 women were from below-average-income families (lower than €16,113), 11 women were housewives, and 12 women had up to a secondary education level.

Instrument Scoring

For the scope of preliminary mean instrument scoring, the mean values at Time 1 in Maltese were analyzed (refer to mean Maltese T1 in Table 2). Subscale scores were retrieved as the mean of items (following reverse scoring [r] for only one item “There is no possibility of getting breast cancer” in the Perceived Susceptibility subscale). Higher scores for Health Belief subscales, for instance, indicate more susceptibility, benefits, barriers, cues to action and self-efficacy (Champion, 1999). Maltese women scored highest for perceived benefits and lowest for perceived barriers, and highest for cyclical timeline and lowest for acute/chronic timeline.

Internal Consistency and Correlation Analysis: Psychometric Estimates of Reliability

Table 1 presents measures of central tendency (mean), variability (standard deviation), and alpha coefficients for the scales. In terms of reliability, Cronbach’s alpha value was .93 for CHBMS-MS and .92 for IPQ-R (Table 1). Such a result in excess of .80 shows high internal consistency (reliability; Huaman et al., 2011). Cronbach’s alpha estimations of each subscale were as follows: Health Beliefs-Susceptibility (α = .91), Benefits (α = .75), Barriers (α = .88), Cues to Action (α = .86), Self-Efficacy (α = .90), whereas for Illness Perceptions-Breast Cancer Identity (α = .92), Causes of Breast Cancer (α = .90), Timeline Acute/Chronic (α = .88), Timeline Cyclical (α = .86), Consequences (α = .93), Personal Control (α = .90), Treatment Control (α = .90), Illness Coherence (α = .86), Emotional Representations (α = .96). These values showed that the scale items measured similar features with high reliability because each dimension was expected to have an alpha of at least .7 (Huaman et al., 2011). Hence, preliminary high Cronbach’s alpha values indicated that the Maltese instrument had internal consistency.

Reliability Over Time

The CHBMS-MS and IPQ-R subscales demonstrated acceptable stability over a 2-week period for all measures. Responses were compared between Time 1 (T1) and Time 2 (T2) after 2 weeks for both Maltese and English versions, respectively. Test–retest scores for all dimensions showed Pearson correlation coefficients higher than .6 for both languages. For test–retest reliability (Maltese; Table 2), Pearson’s correlation coefficients for CHBMS-MS-M and IPQ-R-M were .79 and .75, respectively. For test–retest reliability (English), Pearson’s correlation coefficients for CHBMS-MS and IPQ-R were .83 and .74, respectively (Table 3). Hence, all of the subscale items met the criteria of reliability and were retained.

Construct Validity

When applying correlation analysis between the English and the Maltese versions (Table 1), the Pearson correlation values for CHBMS-MS and IPQ-R were .87 and .89, respectively. All correlation values exceeded .6 and showed a significant correlation between the items of both versions (p < .001). The Pearson correlation values were tested at the .05 level of significance.

When applying a Pearson correlation between the two time points, the Pearson correlation value was .778, showing a strong positive correlation between the two time points. Such an association was found to be significantly different (p < .001).

Implications

The Maltese and English versions of the CHBMS-MS and IPQ-R can be used by nurses and other health care professionals as measures to assess Maltese women’s health beliefs and illness perceptions concerning breast cancer and screening. Nurses have frequent patient contact in various health care settings and are known to be valuable change agents and patient advocates (Arabi, Rafii, Cheraghi, & Ghiyasvandian, 2014). An important breast health promotion opportunity for public health nurses is raising public awareness on breast cancer by educating women about the importance of practicing screening. Likewise, nurses and health care professionals can structure patient education and counseling sessions guided by the conceptual theoretical framework proposed in this study to ensure comprehensiveness of approach and content. For instance, information on breast cancer risks, susceptibility to breast cancer, signs and symptoms of breast cancer, and its consequences, as operationalized by different HBM and CSM constructs, can increase patients’ knowledge to improve screening use (Noar & Zimmerman, 2005). Moreover, health care providers can use the HBM and CSM to understand patients’ needs, employing constructs of the models to guide patient interviewing. For instance, a BS invitation may be based on factors that influence BS behavior such as existing perceptions of benefits and barriers and on psychological and social factors (Kaltsa et al., 2013). Nurses can therefore assess women’s level of perceived risk and target their teaching about health-promoting behaviors to reduce risk perception by educating women about the risk factors for breast cancer. If women are aware they may be at risk for developing breast cancer, they may perceive themselves at risk and participate in screening. Counseling may be required to increase the likelihood that a woman attends for screening by increasing women’s confidence. Particular focus on the appointment related to screening could provide an opportunity for targeted interventions to increase BS uptake, such as assisting women with scheduling an appointment, ensuring that guidelines and information is provided about the recommended intervals between mammograms and addressing the importance of regular screening. This will ultimately affect the quality of an individual’s life and reduce the allocation of resources needed to treat those who develop breast cancer.

Because individuals possess multifaceted cognitive representations of various diseases (Lykins et al., 2008), nurses can support patients to explore beliefs and perceptions by helping them to relate personal accounts about their families, their culture, and their illness perceptions, including causal attributions for the disease (Richer & Ezer, 2000). Achieving this may be more attainable if nurses and health care providers are knowledgeable, competent, and feel supported in providing education and counseling in the clinical setting. This presents a challenge for all health care disciplines considering that health implications span the entire health care continuum. Furthermore, the gap in competency includes lack of recognition of the relevancy of screening to nursing practice which may impact the uptake of continuing education in this area. To overcome these challenges, robust interventions are needed with reliable measures that can adequately assess the outcomes of these strategies. Validated instruments for nursing and patient assessment should be made available in clinical settings as a priority. With reliable measures to inform the required interventions and outcomes associated with their implementation, nurse-led interventions make it possible to design cost-effective strategies focused on reducing disparities across diverse populations and increasing quality within health care systems.

Limitations

Although our preliminary internal consistency and test–retest reliability correlation scores were relatively similar or higher to those reported in prior validation studies of the CHBMS and IPQ-R research, we recognize this study’s limitations. First, the reported findings cannot be generalized because these are limited to a convenience sample. Our goal was not to obtain a representative sample but rather to obtain an indication of the instrument’s reliability and validity among women with varied backgrounds and diverse perspectives. For greater applicability, it is recommended that this instrument be tested among a larger sample. Second, recruitment of these women may have led to a biased sample of women with no socioeconomic inequalities. We acknowledge that those who participated may have been more interested in and knowledgeable about screening as compared with those who would not attend for screening. Third, for those who participated in this study, the formal consent to participate sets them apart from those who would refuse such an invitation. Moreover, although women were asked to express their true feelings, they may have responded in a way that is considered socially acceptable. Despite these limitations, our rigorous approach to translating and adapting the instrument gives us confidence in the instrument’s acceptability and readiness for use to collect data from the target population.