This study was approved by the Committee for Ethical Affairs of Zhongnan Hospital of Wuhan University.
After having obtained ethics approval, we designed and conducted a robust survey study between January 2019 and July 2019 which complies with recommended methods  and aims to maximize compliance with reporting guidelines .
A multitude of factors, including enablers and barriers of guideline adherence, clinician, organization, and system levels, may influence whether and how guidelines are used [7,8,9, 13,14,15,16]. In order to formally assess the determinants of guideline implementation, we used the following literature and resources for the conceptualization of our research framework: (a) ideas of implementability formalized by Gagliardi et al.  consisting of 22 elements within eight domains, including adaptability, usability, applicability, validity, accommodation, communicability, implementation, and evaluation; (b) Guideline Implementability Appraisal (GLIA and GLIA 2.0) tool to provide information about implementability to authoring groups enabling them to decide on content in anticipation of potential problems in implementation  and taking into account decidability, executability, validity, flexibility, measurability, effect on process of care, novelty/innovation, and computability ; (c) Qualitative approach to exploring the medical practitioners’ perceptions and experiences regarding guideline implementation with general, open-ended and non-leading questions having developed a basic understanding of the reaction of medical practitioners and system mechanism to the introduction of guidelines ; (d) Systematic reviews of guideline implementation [19,20,21] literature with five main areas identified: (1) the guideline, (2) the target health care professional user, (3) the patient characteristics, (4) the work environment, and (5) the implementation strategy; (e) Systematic examination of the content of guideline development manuals to identify implementation methodology of known organizations [22,23,24]; and (f) Behavior change and social-cognitive theory applied in implementation research for improving understanding of determinants of evidence-based medicine (EBM) practice and guideline use [25,26,27].
Items collected during the review of literature and existing instruments or frameworks formed the basis of an item pool. This was then further extended with items emerging from interviews with medical practitioner and discussed based on expert opinions and behavior change and social-cognitive theory. Parsimony was achieved by combining multiple items into one, and the number of items was reduced by several expert meetings. The first version was field tested in single interviews among clinicians from Zhongnan Hospital of Wuhan University. A series of draft versions were piloted.
Questionnaire construction, piloting and reliability testing
A bespoke questionnaire, consisting of four parts, was developed as a self-administered survey and its design was based on recommended methods . The aim of the survey was to investigate barriers and enablers related to guideline adherence. First, the survey instrument covered background information about the participants (qualifications, education level, clinical department, years of practice) and some specific questions related to guideline implementation (e.g., “Have you had EBM or EBM related education?”, “Do you agree that high-quality guidelines provide basic guidance for healthcare delivery?”, “Are you are willing to acquire and read high quality guidelines?”, and “To what extent do you think you are applying the guidelines in your clinical practice?”). A 4-point Likert-type scale was used to rate the extent of guideline adherence.
The second section captured knowledge of a broad and comprehensive range of the methods and processes for producing guidelines with 17 items. Questionnaire items were based on the manuals considered in our framework section (e) as mentioned above, with particular reference to NICE and WHO [22, 23].
Given that guideline noncompliance may come from difficulty in searching or downloading guidelines which is a separate issue to guideline implementation barriers, like guideline flaw, lack of atmosphere of EBP etc., we divided the third section of barriers into guideline acquisition barriers and guideline implementation barriers. So the third section had four multiple choice items relating to barriers to guideline acquisition and 15 multiple choice items relating to barriers to guideline implementation which were categorized into three areas: intrinsic flaw in guideline (eight items); deficient or incomplete system mechanism and external environment (four items); and awareness and ability of clinicians (three items). The fourth part consisted of questions which looked at methods for improving guideline implementation; this included seven multiple choice items which addressed external enablers and four multiple choice items which adopted a microcosmic perspective to focus on internal enablers relating to the guideline implementation. The full questionnaire is provided in Additional file 1.
The content validity and readability of the questionnaire were tested by five experts of guideline development, six clinical EBM experts and 20 clinical experts from different medical specialties. All experts commented on the clarity and relevance of each survey item. There was agreement among the experts on the clarity and relevance of most of the included items, and we revised some items to improve clarity. Before implementing the study survey, we tested for repeatability by administering the questionnaires to the same population of 40 participants twice, with a two-week interval in between the first and second survey. The test–retest reliability coefficient was excellent at 0.80 (1 = perfect repeatability).
Survey sampling, questionnaire administration and data collection
A cross-sectional survey was used which took into account the differences of geographical location and the number of medical institutions. We used a multi-stage stratified sampling strategy based on China's economic regions (the East with seven provinces and three municipalities; the Middle with six provinces; the West with 11 provinces and one municipality; and the Northeast with three provinces). Two to five provinces were selected for each region with two to three cities selected for every province, and each city included both secondary and tertiary hospitals. Sampling procedures for hospitals was decided by Medical Standards Bureau of Management Center of Medical Management Services, National Health Commission of the People's Republic of China Mainland based on Proportion Report of Hospital Institutions of Health Statistics Yearbook 2018 . We did not include Hong Kong, Macao or Taiwan in this survey. In total, 32 cities from three provinces and two municipalities in the East, three provinces in the Middle, three provinces and one municipality in the West, and two provinces in the Northeast were chosen. More provinces and municipalities were selected from the East because there is a higher concentration of medical institutions in that region. Doctors in each hospital were recruited using the hospitals’ directories which held a database of their ID numbers. Licensed doctors, pharmacists and nurses, regardless of specialty, with over 5 years of continuous working experience of providing direct or indirect clinical care optimizing health promotion, wellness, and disease prevention were invited to take part in the survey.
The survey was administered during the period from January 2019 to July 2019. Four researchers were each allocated to one of the regions and all used the agreed set of instructions included in the protocol. The researcher explained the nature and purpose of the study to the participants in a meeting room. Informed consent was obtained before the printed copies of questionnaires were distributed. The survey data were anonymized. Data were validated using Epidata (version 3.1, Odense Denmark, EpiData Association, 2010). Questionnaires with more than 10% of data missing were excluded from the analysis.
We hypothesized that the guidelines use is associated with demographic characteristics, attitudes, and knowledge. All included data were analyzed using SPSS [version 17.0 (SPSS, Chicago, IL, USA)]. Categorical variables from survey items were described using frequencies and percentages. Continuous variables were described as median with interquartile range (IQR 25–75% percentile) or mean with standard deviation (SD) as appropriate. We used Chi-squared test to explore if there were differences in the barriers to guideline implementation based on the different grade of hospitals.
Univariate and multivariate analyses were carried out using logistic regression. The dependent variable was the self-reported guideline adherence, and the independent variables were region, hospital grade, years of practice, professional title, EBM education in work unit, education background, EBM education in college, participation in guideline development, acknowledgment of guideline for clinical practice, knowledge score, and professional practice area. Factors of P < 0.1 in the univariate analysis were included into the multivariate analysis to identify the independent determinants of guideline adherence. The associations are reported as adjusted odds ratios (ORs) with 95% confidence interval (95% CI). Two-sided P < 0.05 was considered to be statistically significant.