Transcultural adaptation and validation of the Chinese version of the cervical cancer awareness measure

Background Cervical cancer is the fourth most commonly diagnosed cancer in women globally. The creation of cervical cancer awareness are important determinants of participation in cervical cancer screening, prevention and control and contribute towards reducing delayed presentation and mortality from the disease. However, there is no reliable, valid and mature questionnaire in China. The aim of this study is to evaluate the content validity and psychometric properties of Cervical Cancer Awareness Measure (CAM-C) in Chinese female population. Methods This psychometric evaluation study enrolled 415 general women and 275 clinical/medical experts and was conducted in 2 stages: translation and evaluation of content validity and psychometric evaluation. Cognitive interviews (n = 10) were conducted using the Participant Interview Form. The psychometric evaluation (n = 690) included score distribution, homogeneity (interitem and item-total correlations), internal consistency (Cronbach’s α), construct validity, and test-retest reliability. Results The content validity supported by cognitive interview was excellent and psychometric properties of CAM-C were satisfactory. The internal consistency of subscale of CAM-C were acceptable, with a range between 0.86 and 0.92. The test-retest reliability was good, that ICC coecients for the total score of warning signs subscale, risk factors subscale, and knowledge score subscale varied between 0.78 and 0.84. There were signicant differences between the groups of general women and medical/clinical experts in subscales of knowledge, warning signs and risk factors (p<0.001), which supports construct validity. Conclusions The results provide evidence for the validity and reliability of CAM-C in a sample of female Chinese population. The Chinese version of the CAM-C was a valid instrument to assess and monitor awareness level in the various mode of collecting data (e.g., paper-pencil methods, computer/telephone-assisted interviewing, or self-administered methods), and could be used to evaluate the effects of interventional programme on awareness among Chinese female population.


Introduction
Cervical cancer is the fourth most commonly diagnosed cancer in women globally [1]. Approximately 270 000 people dead of cervical cancer globally, where nearly 90% death cases were reported in low-income and middle-income countries (LMIC) [2]. China is one of LMIC and there is about one fth of the world's population in China, so it is estimated that the health burden of cervical cancer will have a signi cant impact on global disease burden in the future [3]. Usually, the natural history of cervical cancer from mild dysplasia to malignant carcinoma will be 10 to 20 years, which make it possible to be a relatively early preventable disease. Over the past decades, the incidence and mortality of the disease have been reduced by 50% through carrying out formalized screening programmes in high-income countries, while the incidence of the disease have decreased in LMICs due to the introduction of opportunistic screening programme [4][5][6]. The avoidance of human papillomavirus infection plays an essential role in the primary prevention of cervical cancer [6]. Unfortunately, due to limited infrastructure, restricted introduction of vaccine programmes and absence of formalized screening programmes in LMIC, there is an enormous health disparity in incidence and survival rates of the disease globally, with approximately 90% of the disease occurring in LMIC and a wider range of estimates of 5-year survival rate from less than 50% to more than 70% [7,8].
It is advocated that prompt diagnosis and treatment are conducted to control the progression of cervical cancer [6]. Commonly, the delay in diagnosis and treatment of cervical cancer can be attributed to delay in patient, healthcare provider, referral and system [9]. The role of patient delay is much more important in developing countries [9,10]. The delay in patient is partly ascribed to lack of awareness, especially of early warning signs and symptoms and the risk factors of cancer [9]. The creation of cervical cancer awareness are important determinants of participation in cervical cancer screening, prevention and control and contribute towards reducing delayed presentation and mortality from the disease. Previous studies showed that raising awareness of cervical cancer and screening programme could encourage proper personal risk perceptions and prompt help-seeking and increase the acceptance and uptake of available screening services [11].
To address these issues, the Cancer Research UK developed the Cervical Cancer Awareness Measure (CAM-C) to provide a standardized and validated measure of awareness of cervical cancer [12]. It is a site-speci c and valid instrument for public use to measure awareness of cervical cancer. It consists of twelve items on warning signs awareness, eleven on risk factors, four about the NHS cervical screening and vaccination programmes, one each on peak age of incidence, anticipated time to help-seeking and con dence to detect symptom of disease [12]. The items of warning signs were scored as either 0 (no/don't know) or 1 (yes). The sum of 11 items were a total symptom knowledge score, with a range of 0 and 11. Each item of the risk factor was rated from 1 (strongly disagree) to 5 (strongly agree). The total score ranged from 11 and 55. The correct answer to peak age of incidence item is from 30 to 49 years old. For items of screening and vaccination programmes, a maximum score of 4 were given if participant have awareness of screening and vaccination programmes and answered 25 years for offering cervical screening and 12-18 years for offering vaccination [12]. The total score of knowledge was calculated from warning signs, risk factors, age of peak incidence, and awareness of the cervical screening and vaccination progammes, with a range of 11 and 71. Higher scores indicate greater knowledge [12]. The CAM-C has shown good psychometric properties in assessing cervical cancer awareness in general population [12]. However, there is no previous study to evaluate its content validity.
To the best of our knowledge, there is no reliable, valid and mature questionnaire of CAM-C in China.
Because effective health awareness programs or interventions are only possible to be achieved when a culturally sensitive measurement was designed according to the demographic characteristics and awareness needs of the target population, there is a need to evaluate new translation of CAM-C in China.
Therefore, the aim of this study was to evaluate the content validity and psychometric properties of CAM-C in Chinese population.

Materials And Methods
A two-stage study was designed to evaluate the content validity and psychometric properties of the Chinese version of CAM-C. The rst stage involved translation of the English version into Chinese and evaluation of content validity and second stage was psychometric evaluation.

Stage 1: Instrument translation and content validity
With the permission of the initial developer of CAM-C, the instrument was translated from English into Chinese (Mandarin) according to the "forward-backward" procedure proposed by Beaton et al [14]. Firstly, two professional translators performed forward translation (translated original English version into Chinese) independently. They were native Chinese speakers and uent in English. The two translators and research group (NJ and LJZ) compared, reviewed and amended translation to develop an acceptable Chinese version. Then two independent bilingual translator who were native English speakers but pro cient in Chinese translated the Chinese adapted version back into English. The back translators did not have medical background and know the initial questionnaire previously. The translated English version was compared with the original one by the researchers (NJ and LJZ) and minor changes were made based on consensus.
Cognitive interviews were conducted with ten Chinese women (aged 21-66 years). Respondents who were female, ≥18 years and able to read and speak Mandarin were invited to participant in cognitive interview.
All the respondents were interviewed in the circumstances of having informed consent and the right of con dentiality, anonymity, privacy, and voluntary participation. In the cognitive interview, the respondents read the instructions and questions of the CAM-C and ll out the questionnaire in a controlled environment where they could think aloud and narrate their experiences during processing and answering questions [14,15]. After completing the questionnaire, the researchers interviewed respondents based on the Participant Interview Form, where all the respondents were encouraged to express their thoughts, understanding, di culties, and challenges in the process of answering questions. The purpose of this interview is to evaluate whether the instrument could be self-administered and whether the instructions, questions, and response options of the Chinese version of the CAM-C were clear, easy, unambiguous, and relevant to answer. The duration for each cognitive interview ranged 25 and 30 minutes. The results of the cognitive interviews were discussed by researchers and incorporated in the revision of the questionnaire. The unprompted questions of warning signs and risk factors were excluded in the nal Chinese version of the CAM-C.
Seven medical, linguistic and cultural experts as a panel were invited to do evaluation of the content validity of the CAM-C according to the item level content validity index (I-CVI) with a value above 0.78, the scale-level content validity index/universal agreement (S-CVI/UA) above 0.70, and S-CVI/average agreement (S-CVI/Ave) above 0.9 [16]. After test, the Chinese version of the CAM-C had satisfactory content validity with I-CVI values of 0.80~1, S-CVI/UA value of 0,82, and S-CVI/Ave value of 0.96.

Participants
This study was conducted between May 2020 and October 2020. People who were ≥18 years old, and could read and speak Mandarin Chinese were invited. The participants included general women and clinical/medical experts. The general women were chosen by multi-stage cluster random sampling method from six districts of Tianjin and the medical and clinical experts were randomly selected from physicians and nurses at six hospitals a liated with Tianjin Medical University.
A total of 750 Chinese women were invited to participant in this study and 60 women refused. Thus, 690 participants (415 general women and 275 clinical/medical experts) were recruited into this study. 100 participants lled in the Chinese version of the CAM-C for twice (2 weeks apart) to assess the test-retest reliability. Questionnaires from 54 general women and 66 medical/clinical experts were used to evaluate construct validity.
All the participants were invited to complete a questionnaire package (by paper-pencil or telephone methods). The questionnaire package included demographic data and the Chinese version of the CAM-C.

Data analysis
The classical test theory was selected to evaluate the psychometric properties of the Chinese version of CAM-C. Descriptive statistics was used to evaluate data completeness, refusal rate, ceiling/ oor effects.
For data completeness, the standard of lower than 10% of responders was considered to be acceptable [17]; for ceiling and oor effects, the standard of less than 15% were considered acceptable [18]. Itemtotal correlations and inter-item correlations were calculated to evaluated the homogeneity of the CAM-C.
The item-total correlation coe cients should range from 0.3 to 0.7 [19], and the inter-item correlation coe cients should range from 0.2 to 0.4 [20].
Cronbach's α was used to measure internal consistency of sections of risk factors, warning signs and knowledge. Cronbach's α of 0.70 or above was considered to be acceptable internal consistency [18]. Test-retest reliability was evaluated by using intraclass correlation coe cient (ICC) with 2-way randomeffects model of average measure for sections of warning signs, risk factors and knowledge score and by the unweighted kappa statistic for each item. ICC above 0.70 was considered excellent test-retest reliability [18]; the unweighted kappa statistic with the value less than 0.20 was considered as poor stability, 0.21-0.40 as fair stability, 0.41-0.60 as moderate stability, 0.61-0.80 as good stability, and 0.81-1.00 as very good stability [21].
The 'known-group' method was conducted to assess construct validity. The construct validity is supported when signi cantly differences are found between the scores of two groups known to differ in cancer awareness levels [22]. Differences between two groups were tested by t test and chi-square test.
p<0.05 was considered to be statistically signi cant. All statistical analyses were conducted by using SPSS statistics 22.0 (IBM Corp, Armonk, New York).

Sample characteristics
A total of 690 women participated in the psychometric evaluation of the Chinese version of CAM-C. The mean age was 42.7 (SD, 10.6) years. More than 50% of the women were married/cohabiting (56.1%), and 46.8 % had college or higher education level. More than half (51.3%) women employed full-time. Most of women did not have close cancer experience (70.9%), while only 29.1% had previous experience of cancer with themselves, family, relatives or friends. Details of the sample were presented in Table 1.
Acceptability 98.6% of cases completed the Chinese version of the CAM-C. Problems with missing data for all answered items were low (2.4%), which indicated that the Chinese version of the CAM-C had good acceptability. The refusal rate is about 8.0%. The duration of lling questionnaire ranged from 7 minutes to 12 minutes.

Ceiling or oor effects
The ceiling or oor effects of the total score of knowledge and risk factors sections were less than 10%, with a range between 0.1 and 4.8; the ceiling or oor effect of the total score of warning signs section was less than 20%, with a range between 13.5 and 16.2. These results showed that the CAM-C had an acceptable level of variance (Table 2).

Homogeneity
The inter-item correlations except for two items (age when invited for screening and age invited for vaccination) were in an acceptable level, with a range from 0.20 and 0.37. The item-total correlations were satisfactory for all items in subscales of knowledge, warning signs and risk factors, with the lowest correlation as 0.51, 0.49 and 0.31 respectively ( Table 2).

Internal consistency
The internal consistency of warning signs subscale and risk factors subscale was excellent with a Cronbach's α of 0.90 and 0.92 respectively. The internal consistency of knowledge score subscale was good with a Cronbach's α of 0.86 (Table 2).

Test-retest reliability
The ICC coe cient for the total score of warning signs subscale, risk factors subscale, and knowledge score subscale varied between 0.78 and 0.84 (shown in table 3), which showed that all these three subscales of the CAM-C were well consistent between the two administrations (Table 3). The computed kappa coe cients of items about cervical screening and vaccination programme were very good (0.88 and 0.90); the kappa coe cients of items on anticipated time to help seeking, con dence noting a symptom, and age when invited for screening were good, with a range between 0.61 and 0.67; and the kappa coe cients of items about age invited for vaccination and peak age of incidence were fair (0.54 and 0.55) ( Table 3).
The consistency percentages of correct responses to the proposed items were 78%-92% in the testing and re-testing administrations, which indicated that the CAM-C had acceptable test-retest reliability.

Construct validity
There were signi cant differences between the group of general women and medical and clinical experts in subscales of knowledge, warning signs and risk factors (p<0.001). The medical and clinical experts scored higher than the general women on the above three subscales (Table 4).
Clear discriminations were found in items of peak age of incidence, cervical screening and vaccination programmes, and age when invited for screening and vaccination between two groups, which reached statistical signi cance (p<0.05). Compared with general women, the medical/clinical experts were more able to identify a 30-49-year-old woman as most likely to get cervical cancer, and reported a greater proportion of knowing programmes of cervical screening and vaccination, and answering correct age of inviting for screening and vaccination (Table 4).

Discussion
The increase of level of cervical cancer awareness makes great contribution to early detection, which then in turn, decreases the incidence and mortality of disease. However, the lack of an effective and valuable tool to measure awareness of disease has hampered the development of an early detection programme for cervical cancer in China. Therefore, this study aimed to evaluate the content analysis and the psychometric properties of the CAM-C in Chinese female population. The content validity from experts and general women was excellent; and the psychometric properties regarding acceptability, homogeneity, internal consistency, test-retest reliability, and construct validity were satisfactory. These results indicated that Chinese version of the CAM-C is a psychometrically valid and reliable instrument for measurement of cervical cancer awareness in Chinese population.
The acceptability of the CAM-C was very good due to low level of missing data and high level of completion rate. The responders in cognitive interview found the questions were easy to understand, and recognized minimal items to be confused, which further indicated the instrument with a good acceptability. Compared with previous studies of other site-speci c versions of CAM, the refusal rate in this study was obviously lower [23,24]. The reasons for high refusal rate were time constraints, lack of knowledge to answer the survey, not useful and fear, worry, or anxiety about the survey [23,24]. The lower refusal rate of Chinese version of the CAM-C might be because the unprompted questions about warning sign and risk factor were excluded in the survey.
No ceiling and oor effects were reported. In addition, homogeneity was acceptability with ideal coe cients of item-to-total and inter-item correlations. The excellent internal consistency and satisfactory test-retest reliability were demonstrated in the present study, which is consistent with the original English version of the CAM-C in UK population [12]. These results indicated that the Chinese version of the CAM-C might have the capacity to maintain its stability and yield the same results when being repeatedly used among other similar population.
Construct validity of the Chinese version of the CAM-C was supported by the "known-groups" methods that cervical cancer experts achieved signi cantly higher scores or correct rate of questions than educated non-experts. This was in line with the previous study [12] and showed that the CAM-C were able to differentiate between groups with established differences of awareness level.
As the validation studies of the original version of the CAM-C proved that limited value of data was generated by the item on lifetime risk and responders felt too di cult to answer it, this item was subsequently dropped from the instrument [12]. In this study, cervical cancer expertise and general women in the cognitive interview suggested to delete the unprompted items of warning signs and risk factor in Chinese version of the CAM-C. Chinses people disliked answering open-ended questions in questionnaire because they found it di cult to answer and easy to feel annoying to spend too much time and energy to write answers [25]. Most of people chose to skip over the unprompted questions [25].
When considering the response rate and completeness rate of the data which are important indicators for data quality [26], items about the unprompted warning sign and risk factor were excluded in the nal version of the CAM-C. We found the Chinese version of the CAM-C is more suitable to be used in population surveys in the various mode of collecting data (e.g., paper-pencil methods, computer/telephone-assisted interviewing, or self-administered methods), because the exclusion of unprompted items could avoid participants making changes on their previous response answers to the prompted format of subsequent items.
It is important to recognize the limitations. Firstly, it is considered that the results of this study were provisional because all the participants in this study were female, which might not be representative of the population and limit generalization to male populations. Second, the researchers administered questionnaire by using different modes across two stages of research (Bowling, 2005), which might cause potential impacts on the quality of data collection. However, the deleted unprompted questions tended to make these potential impacts to be minimum.

Conclusions
It is demonstrated that the Chinses version of the CAM-C questionnaire had good psychometric properties in the present study sample in terms of the internal consistency, test-retest reliability, content validity and construct validity. It is anticipated that the Chinese version of the CAM-C could be a valid instrument to assess and monitor awareness level, and evaluate the effects of interventional programme on awareness among Chinese population. Validation of the CAM-C in male population might be done in the future studies.

Declarations
Ethics approval and consent to participate The study was approved by Tianjin Medical University Ethics Committee (TMUHMEC 2016015). All methods were performed in accordance with the relevant guidelines and regulations (Declaration of Helsinki) and all participants signed the written informed consent.