Predictors of the intention to use traditional Chinese medicine (TCM) using extended theory of planned behavior: a cross-sectional study among TCM users in Hong Kong

The Theory of Planned Behavior (TPB) is a well-known model in explaining health-related behavior. TPB has been used in many research, including the evaluation of smoking behavior interventions [9], adoption of social distancing [10], COVID-19 vaccination intention [11] and mobile health service adoption [12], and so TPB is a well-established and validated model. In TPB, behavioral intention, which is affected by subjective norms (perceived expectations from important people on performing the behavior), attitude (personal feelings and beliefs about the behavior) and perceived behavioral control (perceived ability of performing the behavior), is the predictor of actual behavior [13‐15].

Subjective norms relates to the likelihood that the important people will agree or disagree of certain behavior and personal motivation to meet their expectations [16]. In some studies, subjective norms has a positive effect to the attitude towards using TCM. Source of TCM information are mostly from family, friends and Chinese medicine practitioner [17]. More than 40% of the medical students in Hong Kong would like to use TCM because of the recommendation from family and friends [18]. Subjective norms significantly predicted attitude towards the use of TCM in China [19]. Additionally, recommendations and referrals from those who are trustworthy could motivate the adults to use acupuncture in Hong Kong [20]. Therefore, the following hypotheses are developed:

Perceived behavioral control is about the perceived ability to perform the behavior. The easier to perform the behavior leads to the increase in attitude and intention to perform the behavior. Attitude towards the use of TCM was found to be affected by perceived behavioral control [21]. Some studies have found that perceived behavioral control is one of the predictors of acquiring knowledge intention towards and the use of complementary and alternative medicine [22, 23]. Likewise, perceived behavioral control was found to significantly predict the intention to use TCM in China [24]. Therefore, this study proposes hypothesis 3 and 4 as below:

Positive attitude towards behaviors is correlated to the intention to the behaviors. The relationship between the attitude towards behaviors and intention to the behavior was found in many studies using TPB, including wearing PM2.5 mask [25], consumer purchasing behavior for organic food [26], and self-management of stroke [27]. Individual’s attitude towards herbal medicines consumption was correlated with intention of herbal medicine use among pregnant women [28]. Besides, attitude was found to significantly predict the intention to use TCM for health maintenance [24]. Positive attitude about traditional, complementary and alternative medicine had a positive relationship with the willingness to pay for the medicine [29]. The following hypotheses are thus proposed:

Individual’s knowledge about the behavior can be one of the predicting factors of attitude towards behavior and behavioral intention. Knowledge is important to provide meaningful and useful information for individuals. Knowledge can be viewed as a distal antecedent of behavior conveyed by attitude [30]. Knowledge is important to the attitudes towards health behaviour under the knowledge-attitude-behaviour model [31]. Patients’ attitude towards using acupuncture had a positive impact on the attitude towards acupuncture, affected by their knowledge [32]. Individual with sufficient knowledge can understand the safety and risk of performing behavior so the level of knowledge may affect the behavioral intention. Abamecha et al. had found that knowledge about cervical cancer was positively associated with intention to use cervical cancer screening [33]. Therefore, the following hypotheses are developed:

In some studies, customer satisfaction theory was used to predict the relationship of perceived service quality, satisfaction and behavioral intention [34, 35]. Perceived service quality refers to the comparison of prior expectations with perceived performance of a specific service [36, 37]. Satisfaction is interpreted as patient treatment satisfaction, clinical experience and service recommendation [38, 39]. The relationship between perceived service quality, patient satisfaction and behavioral intention was found in many studies [34‐37, 40‐42]. Therefore, the following hypotheses are developed:

A conceptual model which depicts the relationships between the constructs of TPB, knowledge, satisfaction and perceived service quality of TCM has been developed (Fig. 1).

This online cross-sectional survey by a questionnaire was conducted between 15 and 29 November 2021 to address the research question. This study was approved by the Research Committee of College of Professional and Continuing Education, The Hong Kong Polytechnic University. The methodology and results of this study were reported in accordance with the Checklist for Reporting Results of Internet E-Surveys (CHERRIES) [43].

The online questionnaire was initially and purposely sent to participants with varying age and educational level. The link of the survey was posted on social media platforms hosted by the academic institution. Participants were encouraged to forward the questionnaire link to relatives and friends whom they considered suitable. Participants who were literate in Chinese and able to understand the questionnaire; Hong Kong residents aged 18 or above; could access to the internet via smart-phone or computer; consented to participating before they could start the survey; and had used TCM before were included in this study. Interested subjects would read a statement describing the background of the survey, anonymity assurance and participation agreement at the beginning of the survey. Checking the consent box represents that they gave their agreement for study participation. Participation was optional and anonymous. To avoid multiple submissions, the prefix and first three digits of Hong Kong Identity Card (HKID) were collected in the questionnaire. Participants could leave their email address if they were interested to be contacted for survey clarification or join future follow up studies. Identifiers, such as HKID, were removed before data analysis. All decisions were made at subjects’ own discretion. No financial incentives were given after the completion of the questionnaire. The calculation of target sample size should be 10 times the maximum number of measuring items pointing at the construct in the PLS path model [44]. In the present study, there are 33 items measuring seven constructs so the target sample size is 330 (10 × 33).

The questionnaire consisted of four parts to collect the data regarding the attitude, subjective norms, perceived behavioral control, intention, knowledge, perceived service quality and satisfaction of TCM of respondents. All items in the TPB framework and satisfaction of the use of TCM were measured using the 7-point Likert scale (1 = strongly disagree / Never to 7 = strongly agree / Always). For the perceived service quality, a validated short questionnaire to assess perceived service quality was used [45]. Five items regarding perceived service quality was asked using the 7-point Likert scale (1 = strongly disagree to 7 = strongly agree). The knowledge construct was measured by four questions regarding their self-report knowledge of basic theory, medicine, diagnosis and diet using the 7-point Likert scale (1 = strongly disagree to 7 = strongly agree). The questionnaire items of the seven constructs in this study is shown in Supplementary file 1. Several demographic questions (i.e., age, gender, level of education, economic status, religion, and self-report health status) were also asked in the survey. Respondents were allowed to select more than one religion in the questionnaire.

To examine the conceptual framework in this study, the partial least squares structural equation model (PLS-SEM) was adopted. The reflective measurement model and structural model were examined with the SmartPLS 3.0 statistical software. The reliability and validity of the measurement instrument were verified. To evaluate the reliability of this study, value of Cronbach’s alpha and composite reliability greater than 0.7 is considered to be reliable [44]. The convergent validity is considered good if the outer loadings of the measurement items are greater than 0.6 and the Average Variance Extracted (AVE) of each construct greater than 0.5 [44, 46]. To evaluate discriminant validity, both Fornell Larcker criterion and heterotrait-monotrait (HTMT) ratio were assessed.

The reliability of the model is presented in Table 2. The variance inflation factor (VIF) level of each item was lower than 5, indicating no critical levels of collinearity [44]. All items had outer loadings of greater than 0.6, except one item of attitude construct and one item of perceived service quality. Hence, these two items were deleted. The composite reliability and Cronbach’s alpha were higher than the threshold value of 0.7, indicating a satisfactory internal consistency reliability. The AVE of all constructs were above 0.5, suggesting that the constructs had good convergent validity [44]. The results of the assessment of discriminant validity are presented in Table 3. The discriminant validity in this study was supported. With adequate reliability, convergent validity and sufficient discriminant validity, it was possible to move forward to assess the structure model.

The results of the direct effects of structural model are shown in Table 4. These results supported the hypotheses H1, H3 and H8, indicating that the subjective norms (β = 0.408; p <  0.01), perceived behavioral control (β = 0.313; p <  0.01) and knowledge of TCM (β = 0.117; p <  0.05) are positively related to the attitude towards using TCM but the result exhibited weak effect sizes on knowledge. The results showed that subjective norms (β = 0.175; p <  0.01), perceived behavioral control (β = 0.168; p <  0.01), attitude (β = 0.398; p <  0.01) and satisfaction (β = 0.201; p <  0.01) have significantly positive effects on the intention to use TCM. However, there was a non-significant relationship between knowledge and behavioral intention of using TCM (β = 0.000; p > 0.05) and perceived service quality and behavioral intention of using TCM (β = 0.068; p > 0.05). H12 stated that the perceived service quality was positively related to the satisfaction and the results provided statistical support for this hypothesis (β = 0.487; p <  0.01). The results of structural model are shown in Fig. 2. The mediation model results are shown in Table 5. All meditation hypotheses were supported. Attitude had a partial mediating effect on the relationship between subjective norms, perceived behavioral control and intention to use TCM. Attitude had a full mediating effect on the relationship between knowledge and intention to use TCM. Satisfaction had a full mediating effect on the relationship between perceived service quality and intention to use TCM.