Study setting
This study was conducted in tertiary general hospitals in China. Although the regulation policy for the clinical application of antimicrobials covers all levels of medical institutions, for the weakness of the primary medical services system in China, the provision of vast medical services is heavily dependent on hospitals, especially tertiary hospitals. In 2019, tertiary hospitals received 1.77 billion medical visits [19]. As one of the major consumers of antimicrobials, the irrational use of antimicrobials in tertiary hospitals is quite prominent. Thus, it’s necessary to regulate the use of antimicrobials of physicians in tertiary hospitals for reducing AMR.
Theoretical framework
The theoretical framework (Fig. 1) was adapted from the integration of TPB, TAM and TOE to illuminate the determinants of physicians’ intentions to use CPGs on antimicrobial from three levels, namely individual level (physicians), technical level (CPGs on antimicrobial), and organizational level (hospitals).
Individual-level factors
Proposed by TPB, behavioral intention is a function of three factors, including attitude, subjective norms and perceived behavioral control [20]. Attitude is defined as a positive or negative evaluation of a particular behavior [21], many studies showed a strong correlation between attitude and intention [22, 23]. Subjective norms are kinds of perceived criteria and social pressure to engage or not to engage in behavior [24], which may also significantly affect physicians’ intentions. Also, perceived behavioral control reflects the person’s belief that an action is under his or her control, such as perceived risk. Risk perception is associated with low intentions.
Technical-level factors
Proposed by TAM, the relative advantage is a degree to which new technology or product is more advantageous than the original or competing ones [25], while ease of use is a degree to which the potential user expects the product can perform with ease [26, 27]. Regarding the intentions to use CPGs, the physicians and managers are more inclined to adopt the guidelines having better outcomes and efficiency with no additional effort and time to learn how to implement.
Organizational-level factors
TOE suggests top management support “can foster innovation by creating an organizational context that welcomes change and is supportive of innovations” [28]. In hospitals, top management’s involvement in the use of CPGs on antimicrobials through formal measures (e.g. funding, training, and system building), can ensure the accomplishment of intended outcomes to a great extent [29]. Organizational implementation refers to the whole implementation process of CPGs on antimicrobials, including providing relevant information, supervision, and inspection, corrective feedback, respectively [30, 31].
Measurements
Based on the theoretical model, as well as the literature review of previous studies, a questionnaire with 30 items was developed for this study (Additional file 1). Three items in Part 1 were used to measure the intentions to use CPGs on antimicrobials of physicians. There were 21 items in Part 2, covering seven potential factors: attitude, subjective norms, perceived risk, relative advantage, ease of use, top management support and organizational implementation. Each item in Part 1&2 corresponding to the constructs was measured using a five-point Likert scale, where 1 = Strongly disagree, 2 = Disagree, 3 = Neutral, 4 = Agree, and 5 = Strongly agree. And Part 3 was a personal information card consisted of 6 items, including several basic characteristics of participants like gender, age, education, professional degree, department, and years in practice.
Sampling
Considering the diverse level of socio-economic development in different regions of China, a cross-sectional survey was conducted using a multistage sampling strategy. Firstly, Fujian, Hubei, Yunnan & Sichuan provinces were randomly selected respectively on behalf of eastern, central and western regions of China. Secondly, 4 tertiary general hospitals were selected from each of the selected regions. Lastly, in each selected hospital, 16–20 physicians were randomly sampled from major departments of internal medicine and surgery, respectively. And 3–5 physicians were randomly sampled from departments of gynecology and obstetrics, ophthalmology and otorhinolaryngology, orthopedics, and others, respectively. Thus, 50–60 physicians from each hospital were invited to participate in the survey.
Data collection
A cross-sectional questionnaire survey was performed to investigate the determinants of physicians’ intentions to use CPGs on antimicrobials in China. With the support of sampled hospitals, each round for filling out the questionnaire was accompanied by trained facilitators. The purpose of the study and the use of data will be explained in detail to ensure the participants understand what they needed to do and how to do it. All responses were anonymous, filled out by the participants at their convenience and returned directly to the researchers.
Data collection started from April 2018 and lasted for nearly one year. Overall, a total of 676 questionnaires were returned. After excluding responses that (1) provided the same response for all items, (2) incomplete questionnaires, we obtained 644 valid questionnaires with a valid response rate of 95.27%.
Data analysis
This study used SPSS 21.0 and AMOS 17.1 software programs as the two main statistical tools to analyze the data. To analyze the descriptive data and investigated variables clearly, several steps were followed. Firstly, descriptive statistics were performed for the analysis of participants’ distribution characteristics. Secondly, we conducted the assessment of reliability and validity via Cronbach’s α Coefficient and factor analysis to tell whether the questionnaire was acceptable. Finally, structural equation modeling (SEM) was used to analyze the mechanism and the relationship between the factors via path analysis and mediating effect test. The path coefficients calculated by path analysis are equivalent to the standardized regression coefficients and direct effects. The mediating effects (indirect effects) and total effects were obtained by mediating effect test. The indirect effects refer to the influence of one variable on another through a third variable, and its value was calculated through the Bootstrap method. If the value does not contain zero in its 95% confidence interval, the mediating effect is considered significant [32].