Regular health check-ups are crucial in a secondary or tertiary prevention of diseases such as diabetes, hypertension, and cancer.1–3 These diseases could be detected through health check-ups at an early stage, which would be associated with better health outcomes such as fewer symptoms, delayed development of complications, and improved quality of life.4,5 Health check-ups are especially crucial for older adults because of high prevalence of co-occurring health issues and function decline that happen along with the aging process.6
Health beliefs and attitude towards health would have an impact on the performance of health check-ups.7,8 In particular, the deep-rooted Eastern culture belief plays a crucial role in the formation of health beliefs among Chinese older adults.9 The culture based health belief would contain attitudes like fatalism that could be a more salient barrier to performing regular check-ups than accessibility to healthcare facilities and socioeconomic factors such as income and education.9,10 Previous studies indicated that this cultural interpretation of health could impact Chinese women’s acceptance of cancer screening. 11–13Several studies have examined health check-ups among Chinese older adults; however, little is known about their attitudes and expectations towards regular check-ups that could be major obstacles in health promotion practices.14–16
Instruments that measure health beliefs have been developed and tested in China, most of which, however, measure health beliefs regarding a specific disease.17,18 Only does the Chinese version of the 48-item Nursing Outcome Classification (NOC) include a subscale that captures health beliefs regarding general health issues.19 The reliability and validity of the scale have been tested in hospital settings,19 but this health beliefs scale has not been validated in community settings; and no evidence showed its validity in measuring health beliefs related to health check-ups. Additionally, given that this measure was designed based on the Health Belief Model that has been questioned about its ability to capture cultural and social norms,20 the health beliefs scale may fail to measure the Chinese culture belief that is related to individual’s health belief.
A promising instrument to measure health belief towards health check-ups is the Attitudinal Index (AI).21 AI measures Asian women’s attitudes towards breast cancer screening, with a particular focus on cultural perceptions. 21Based on the Health Belief model as well as qualitative interviews, AI emphasizes 3 principles that allow participants to self-evaluate their cultural interpretation of health beliefs regarding breast cancer screening: barriers, fatalism, and detects on a 8-point Likert scale.21 Previous study reported good psychometric properties in a Chinese female population in Singapore, making it a strong candidate for using in Chinese older adults21. In this study, we aimed to explore psychometric properties of AI by testing its reliability and validity.
Methods:
The data in this present psychometric study were from the 2013 Survey of the Shanghai Elderly Life and Opinion, a study that followed a method of multistage cluster sampling procedure that yielded a total of 35 survey sites. A total of 3,418 individuals aged 50 or older were included.
Minor modification was made (e.g., changing the term “cancer” to “disease”) to better fit the targeted study population. All the items in this modified AI were rated on a 5-point Likert type scale: strongly agree, agree, disagree, strongly disagree, and don’t know.
The frequencies of missing values for the study variables ranged from 0.44% to 0.73%. Also, the proportions of respondents answered “don’t know” were relatively low, ranging from 2.11% to 6.96%. In this study, we treated the response “don’t know” as missing values and conducted Little's MCAR test22 to assess the missing pattern. The results revealed that the data were not completely missing at random (Chi-square = 3060.407, df = 2154, p < .001). Thus, multiple imputation23 was performed in this study using variables including demographics, health status, and regular check-ups to accommodate the problem of nonrandom missing data.24 Given the large sample size we have, we randomly split the entire dataset into two that each contained half of the sample (n = 1709). We conducted exploratory factor analysis using one half of the dataset and validated the results with confirmatory factor analysis using the other half.
We applied principal components to identify the key components that explain common and unique variance in the 16 items of the measure. Promax factor rotation was employed, and factors were extracted if their eigenvalues were greater than 1. The internal consistency of the measure was tested by calculating Cronbach’s alpha. We conducted confirmatory factor analysis to assess the fitness of our factor structural model using a number of different indices. In this study, the model was considered adequate if the p-value of chi-square test was greater than .05; the Root Mean Square Error of Approximation (RMSEA) was less than .08, preferably less than .05; Standardized Root Mean Square Residual (SRMR) was less than .05; and Comparative Fit Index (CFI) was greater than .90.
Results: While looking at the descriptive statistics of all 16 items, mean item values on this 4-point Likert scale ranged from 1.60 to 2.35. We did not observe any ceiling or floor effect across all 16 items and the total score, which indicated a relatively normal distribution.
The results from principal components analysis on all 16 items extracted four factors. All factor loadings revealed practical significance, within a range from 0.642 to 0.944. The first factor, labeled as Barrier, contained 3 items, all of which were related to individual’s objective barriers to preforming health check-ups. Fatalism, the second factor, contained 3 statements about individual’s perception that health and illness are predetermined. Unnecessary was the third factor that included 5 items that emphasized on one’s perceived necessity on health check-ups. The last factor was detect, which consisted of 5 items about individual’s awareness of the benefits of obtaining health check-ups. After rotation, the four factors were distinct from each other, and the correlations between the factors were considered low to moderate, with a range from .07 to .45.
Descriptive statistics showed the mean score for barriers was 1.89 (SD = 0.66), fatalism was 2.35 (SD = 0.86), unnecessary was 1.73 (SD = 0.61), and detects was 1.91 (SD = 0.53). The overall reliability was 0.835 and all the four subscales revealed good intra-item correlations among items. The Cronbach’s alpha for barriers, fatalism, unnecessary, and detects were 0.856, 0.908, 0.815, and 0.844, respectively.
The results of our confirmatory factor analysis indicated acceptable construct validity. All 16 items showed significantly factor loadings and the 4-dimensional structure of this modified AI was confirmed. In this model, CFI = .913 and SRMR = .048, achieving desired values; and RMSEA was reasonable but less than ideal (RMSEA = .051; 90% confidence interval: .047-.056). However, we failed to achieve a non-significant chi-square value (Chi-square = 516.806, df = 96, p < .001) in this analysis due to the large sample size (n = 1709).
Conclusion: To our knowledge, this is the first study to provide evidence on the validity and reliability of a measure of Chinese health beliefs in health check-ups. The extremely low proportions of missing values indicate its practicability and acceptability for Chinese older adults. Our findings suggest good psychometric properties of this modified AI, which could be implemented in community setting to assess Chinese older adults’ health beliefs in health check-ups.
The modified AI has strengthened the psychometric properties of the original scale by improving the reliability of the subscales. By rearranging items in the subscales and forming a new dimension, all the 4 subscales now show good internal reliability. Additionally, the modified AI indicated a new factor—unnecessary, which tapped mainly on individual’s perception of the necessity of receiving health check-ups. Results of our factor analysis indicated good reliability of this new subscale.
Our confirmatory factor analysis supported the 4-dimensional structure of the modified AI. All 16 items showed significant loadings and 3 out of 4 fit indices provide evidence of good model fit (RMSEA, CFI, and SRMR). The chi-square statistics was less than ideal because of the relatively large sample size. Other indices were less sensitive to sample sizes showed excellent model fit for this measure.
Understanding the factors that influence Chinese older adults health check-ups behaviors is critical for nurses to promote preventive care. This requires further studied using measures that are not only reliable and valid but also culturally relevant. AI can be a promising instrument that could provide nurses with insights into the provision of culturally sensitive health check-ups education and also for meeting the growing need for preventive care for Chinese older adults.
Our findings indicate that the modified AI is appropriate for assessing Chinese older adults’ health beliefs in health check-ups. Further studies using the modified AI are needed to assess its criterion validity. It would be important to examine the relationship between health knowledge, health beliefs, and check-ups behaviors.
See more of: Research Sessions: Oral Paper & Posters