Tools for the assessment of breast cancer screening beliefs in women: a literature review

CHBMS, which was the first tool in the literature to measure women’s breast cancer screening beliefs, was developed by Champion [11] based on the Health Belief Model (HBM). The CHBMS-1984 contained five domains. The item pool was generated from previous concept measurement studies, with 20–24 items in each domain. To test the content validity of the CHBMS, the list of items was made available to eight professional teachers and doctoral students who studied the HBM. Items accepted by more than six of the experts were included in the CHBMS-1984, resulting in 10–12 items in each domain. Through convenience sampling, 301 women, who were mostly white and Protestant, were eventually selected. As a result, 39 items were included in the CHBMS-1984.

After exploring the assessment contents of the CHBMS, the developers repeatedly revised it to better measure women’s breast screening beliefs and formulated several versions [12–15]. Each version had different characteristics, such as different breast cancer screening methods, with different CHBMS versions including breast self-examination (BSE), mammography or both of these methods (Table 1).

The Cancer Awareness Measure (CAM) was established to assess cancer screening awareness, but this instrument did not specifically target a specific type of cancer [16]. Using the CAM, Cancer Research UK, King’s College London and University College London developed the Breast Module (BCAM) to evaluate women’s breast cancer screening beliefs. The BCAM consists of the following seven domains: knowledge of symptoms; confidence, skills and behavior in relation to detecting a breast change; anticipated delay in contacting the doctor; barriers to seeking medical help; knowledge of age-related and lifetime risk; knowledge of the National Health Service (NHS) Breast Screening Program; and knowledge of risk factors. The reliability and validity of the scale were demonstrated by Linsell et al. [17]. The item pool was generated from a review of existing breast cancer screening awareness assessment studies and gray literature as well as previous research experience with breast cancer patients. Phrases or daily terms that did not apply to British women were revised, and items not used in previous questionnaires were added. After agreeing on the items listed, the initial version of BCAM was allocated to an expert panel comprising academic researchers, representatives of cancer charities, oncologists and experts in the design of evaluation questionnaires. Additionally, the researchers conducted cognitive interviews with 20 older women they knew and improved the wording of BCAM items.

This study conducted a BCAM pilot study on 82 women aged between 67 and 70 years who participated in the NHS Breast Screening Program and validated their applicability with different item formats. The scale was modified depending on the results of the pilot survey, and the revised version of BCAM was used for the national survey. The readability, acceptability and retest reliability of the scale were assessed by analyzing data from 712 women aged from 67 to 73 years. The finalized version of BCAM included three items with a total of 18 questions.

CBCSBQ, which was the initial version of BCSBQ, was developed by Kwok et al. [18] and was used to assess the beliefs, knowledge and attitudes of Chinese–Australian women toward breast cancer and screening practices. The CBCSBQ development process was divided into two phases: item generation, evaluation of content validity and cultural appropriateness, and translation; and evaluation of psychometric properties. The original items were generated through a comprehensive review of existing tools and in-depth interviews with the target population, and the original questionnaire contained 32 items. Then, a panel of four health experts (including a psychology professor, nursing professor, director of the screening program and research scholar) and four lay-people representing different social backgrounds assessed the content and face validity of the initial questionnaire and reviewed the phrasing, response formats, relevance and fitness of items. Eight experts also provided feedback on how well the items captured information about the measured variables. The content equivalence of the CBCSBQ was assessed by a literature review, and a consensus was finally reached. In addition to general items, the questionnaire contained special items for breast cancer, such as ‘I don't want to have a mammogram because I would need to take off my clothes and expose my breasts.’ Then, the questionnaire was translated using the back-translation technique to verify semantic equivalence.

In the second stage, participants were recruited from churches, social and community service centers and other Chinese community organizations. Through convenience sampling, 292 Chinese–Australian women aged 18 years and above were eventually included. The results of the multitrait-scaling analysis resulted in deletion of three items with a correlation coefficient of less than 0.3. Then, a factor analysis was conducted on the remaining 29 items, and 11 items with a factor load less than 0.4 were deleted. The factor analysis was conducted a second time for the remaining 18 items, among which five items failed to meet the standard of factor loading higher than 0.4 and were eliminated. Therefore, the final questionnaire contained 13 items.

The goal of the CHBMS-1984 was to measure breast cancer beliefs of American women. It included five subscales (perceived susceptibility’, ‘perceived seriousness’, ‘perceived benefits’, ‘perceived barriers’ and ‘health motivation’) with a total of 39 items. All items were based on Likert’s 5-point scale, with a range from one = ‘strongly disagreed’ to five = ‘strongly agreed’; the higher the score of each subscale, the higher the intensity of the measurement in that domain. Notably, the CHBMS-1984 rated the perceived benefits and barriers of BSE, which was consistent with the CHBMS-1993 and different from the other versions of the CHBMS.

Compared with use of the other versions, the CHBMS-1993 and CHBMS-1999 were widely utilized. The CHBMS-1993 included ‘perceived susceptibility’ (perception of personal risk of breast cancer), ‘perceived seriousness’ (perception of the level of personal breast cancer-related threats), ‘perceived benefits’ (perception of the benefits of BSE for the individual), ‘perceived barriers’ (perception of negative components of BSE for the individual), ‘health motivation’ (health-related beliefs and behaviors) and ‘perceived confidence’ (perception of the procedural ability to perform a BSE with the perception of competence to detect abnormal lumps). Different from the CHBMS-1984, this scale added the ‘perceived confidence’ dimension and contained 42 items in six subscales. All items were rated on a 5-Likert scale; the higher the score, the greater the intensity of the content. The CHBMS-1999 included only three subscales regarding the perceived susceptibility, benefits and barriers related to mammography. The researchers removed items with redundant or low factor loading from the original version, adjusted and added the susceptibility and barrier subscale content, and finalized 19 items (Table 1).

The BCAM was intended to measure British women’s breast cancer-related attitudes and behaviors and included three items (‘knowledge of breast cancer symptoms’, ‘age-related risk’ and ‘reported frequency of breast checking’) with18 questions. The BCAM scoring principle was as follows: the knowledge of breast cancer symptoms item included nine options, if participants chose five or more nonlump symptom options, the women were considered to have knowledge about breast cancer. The correct answer to the ‘age-related risk’ item was that 70-year-old women were most likely to develop breast cancer in the following year than younger people. For the last item, a self-reported breast examination at least once a week or month was an appropriate response. The correct answer to an item received one point, and thus the total score of BCAM was between zero and three points. The BCAM focused on specific early signs of breast cancer (such as breast and nipple changes), which was different from the other instruments. The study showed that BCAM performed well among young and old adult women and could be used to assess screening awareness for additional cancers.

CBCSBQ included: ‘attitude toward health checkups’, ‘knowledge and perception about breast cancer’, and ‘mammography screening practice’, for a total of 13 items. The items were rated on a 5-point Likert scale ranging from one = ‘strongly agree’ to five = ‘strongly disagree.’ The higher the score, the higher the understanding of breast cancer and more positive the attitude toward health examinations and breast cancer screening. Notably, the breast cancer screening method evaluated by the CBCSBQ was mammography, which was different from that of the CHBMS-1984.

Champion [11] developed the CHBMS-1984 to evaluate women’s beliefs about breast cancer screening. Verification of internal consistency was based on the removal of less relevant items until the correlation coefficient began to drop after deletion of an item. At this time, the Cronbach’s α was recognized as the largest value, and thus 39 items were finalized. The results showed that the correlation coefficients of the other three subscales were greater than 0.7 except for that of the ‘perceived benefit’ subscale (Cronbach’s α was 0.61) and 'health motivation subscale’ (Cronbach’s α was 0.62). Retesting was performed 2 weeks after the initial measurement. The results reported that the test–retest reliability coefficients of the other four subscales were significant and greater than 0.7 except for that of the 'perceived benefit’ subscale (r = 0.47). The structure validity of the scale was tested by factor analysis, and seven factors were extracted by principal component analysis. The analyses suggested that the 'perceived seriousness’ subscale contained three factors, whereas the remaining subscales contained only one factor. The structure validity of the CHBMS-1984 was also confirmed by mutual exclusion of five subscales. The multiple regression analysis results indicated that the multiple R was 0.51 (p ≤ 0.01), which explained 26% of the variance, and proved again that the scale had good structural validity. Although the variance explained by the 'perceived barrier’ and 'health motivation’ subscales was 23 and 4%, respectively, the variance explained by all subscales was not significant. Subsequently, the reliability and validity of the CHBMS were verified in Arabia, Korea, Turkey, Lithuania, Malaysia, Iran and Spain with consistent results [19–26].

Linsell et al. [17] validated the psychometric properties of BCAM by evaluating 1035 women who participated in the NHS Breast Screening Service. All participants were divided into three groups to test the scale’s acceptability, change sensitivity and test–retest reliability. The BCAM acceptability was assessed by 292 women aged between 67 and 73 years who received interventions to improve breast cancer awareness and early symptom prevention. After completing the 18 questions included in the three items, the participants were asked to answer some feedback questions about the scale, such as whether any items needed improvements, the time required to complete the scale and the evaluations of these items. A total of 87% of the respondents answered these questions, and more than 90% of them were content with BCAM. In terms of changing sensitivity, breast cancer awareness was compared between the two groups by evaluating 576 older women receiving two different interventions (10-min interaction with health professional supported by booklet VS using only booklet). The results indicated that both interventions had significant effects but that the interaction group had the best improvements. The test–retest reliability was measured by recruiting 167 women over the age of 50 years who participated in the mammography screening program in south-east London. Two evaluations were conducted during the day of the screening and at 2 weeks of follow-up, with 76% of the participants completing the reassessment; the results confirmed the retest reliability of BCAM. The Kappa statistics for all items ranged from moderate to good (0.42–0.70), although the Kappa value for one item (whether the breast lumps are the alarming sign of breast cancer) was low (0.28).

The scale was distributed to 18 female cancer experts and 33 nonmedical senior scholars, and the structural validity was verified using the ‘known group’ method and Fisher’s exact test. The results suggested significant differences in 'knowledge of breast cancer symptom’ and 'age-related risk’ items between the two groups, but no significant difference was found in the 'reported frequency of breast checking’ item. Based on the Flesch Reading Ease formula with a value range of 0–100, 60–70 is an acceptable standard. The higher the score, the higher the readability. The results showed that BCAM had a readability score of 87.9, which meant that it was easier to understand than standard adult books. Currently, BCAM has been translated into Kenyan and Arabic versions, and studies proved that they could accurately assess women’s breast cancer awareness [27,28].

Kwok et al. tested the multitrait scaling and verified the psychometric properties of the CBCSBQ [18]. The researchers recruited Chinese–Australian women over the age of 18 years who had literacy with Chinese and English and no history of breast cancer. The correlation coefficients between most items and their corresponding subscales exceeded the standard of r = 0.3. Specifically, in the 'attitude toward general health checkups’ subscale, the correlation coefficients between all items and their corresponding subscales were above 0.7; however, the coefficients of three items was lower than 0.3 in the subscales 'knowledge and perception about breast cancer’ and 'mammography screening practices’, which were excluded from further analysis. The construct validity of the CBCSBQ was verified through factor analysis of the KMO measure and Bartlett’s test of Sphericity results. According to the factor analysis results, after excluding items with a factor loading lower than 0.4, the questionnaire ultimately contained only 13 items. After eliminating items that did not meet the eligibility criteria, an internal consistency verification was conducted for the final revised questionnaire. The results suggested that the Cronbach’s α of the three subscales ranged from 0.70 to 0.79 and was 0.76 for the 13-item questionnaire, indicating that the questionnaire had sufficient internal consistency. Then, the CBCSBQ was translated into multiple versions to test the breast cancer screening beliefs of Chinese–Australian women (Hong Kong), Arabic–Australian women, African–Australian women, Australian–Indian community women and Australian women living in China and Korea, and their reliability and validity were verified [29–34].