Setting and participants
A cross-sectional study was conducted in Wuhan, Hubei, China, from 1 July, 2019 to 31 July, 2019. Trained investigators recruited respondents in public places such as bus stops and parks by convenient sampling method. The inclusion criteria of the subjects were: (1) adults aged 18 and older; (2) those who were able to read and write Chinese. During the data collection period, a total of 4832 questionnaires were distributed, and 3420 respondents completed the questionnaire, with a response rate of 70.78%. Among them, 129 questionnaires were deemed invalid because more than 25% of the items were missing, and 85 questionnaires were excluded because the respondents were younger than 18 years old. Finally, 3206 qualified questionnaires were included in the analysis.
Measurement
The main outcome of this study was the prevalence of SMA, measured by asking "Did you use antibiotics without seeing a doctor in the past 6 months?" In order to identify the relevant factors of SMA, this study also collected the social demographic characteristics (gender, age, residence, educational level, self-perceived economic status, self-perceived health status), antibiotic knowledge, health beliefs of SMA.
A self-designed scale was used to measure the health beliefs of SMA. The scale is based on health belief model, including four dimensions: perceived threats, perceived benefits, perceived barriers and self-efficacy. Perceived threats refer to the degree of perceived harm caused by SMA. Perceived benefits refer to the perceived benefit of taking antibiotics under the guidance of a doctor. Perceived barriers refer to the perception of difficulty in taking antibiotics under the guidance of a doctor. Self-efficacy is the determination to overcome obstacles to seeking medical care and to take antibiotics under the guidance of a doctor. The overall Cronbach's α coefficient of the scale was 0.80. The Cronbach's α coefficients of each dimension are as follows: perceived threats (0.89), perceived benefits (0.94), perceived barriers (0.80), self-efficacy (0.84). There are 4 items of perceived threats and 5 items of perceived benefits, perceived barriers and self-efficacy respectively. Each item used Likert’s 5-point scoring method, from "totally disagree" to "totally agree", scoring 1–5 points respectively. According to the lower quartile and the upper quartile of the score, the participants' health beliefs were divided into three groups: high, middle and low.
The participants' antibiotic knowledge was measured by an Antibiotic Knowledge Scale developed by our group. The scale included 10 items with the Cronbach's α coefficient of 0.72. Each knowledge item was given a score of one point if the participant answered correctly, and a score of zero if the answer was wrong or "I don’t know". So, the total score of this scale ranged from 0 to 10. The higher the score, the better the antibiotic knowledge of participants. According to the lower quartile and the upper quartile of the score, the participants' antibiotic knowledge level was divided into three groups: high, middle and low.
Statistical analysis
We used SPSS version 22.0 for Windows (IBM Corp., Armonk, NY, USA) for all analysis (see Additional file 1: Appendix S1 for the statistical analysis process). Descriptive statistical methods such as frequency, percentage, mean and standard deviation (SD) were used to present demographic characteristics, antibiotic knowledge, health belief and other variables. Pearson Chi-Square test was used to compare the prevalence of self-medication among participants with different characteristics. Binary Logistics regression model was used to explore the related factors of SMA. Adjusted odd ratios (ORs) and 95% confidence intervals (CIs) for each variable were given. Possible multicollinearity of variables in the model was assessed by calculating variance inflation factors (VIFs) and pairwise correlation coefficient (see Additional file 2: Appendix S2 for the multicollinearity text). Differences were tested by using two-tailed tests. If P < 0.05, the result was considered statistically significant.