We recruited outpatients and inpatients who visited STI clinics on Wednesdays and Saturdays at two branch institutes of Shanghai Dermatology Hospital in the Jingan district (located on Qiujiang Road and Baode Road, respectively) from November 2017 to May 2018.
The inclusion criteria were according to the latest revised STI prevention and management measures in China, and included patients aged over 16 years old with syphilis, gonorrhea, condyloma acuminatum, and genital herpes. All diagnoses (including clinical and laboratory diagnoses) were conducted by doctors in the hospital.
The exclusion criteria were as follows: (1) infected with HIV (2) serious psychological or cognitive diseases; (3) unconscious status; (4) unwilling to cooperate with researchers; (5) vision or hearing loss or poor reading ability, leading to little understanding of the purpose and content of this research.
Our research team recruited all physicians working at STI clinics in two cooperative Shanghai Dermatology Hospitals. The researchers were in charge of training all doctors in how to select patients who met the inclusion and the exclusion criteria of this study. After training, doctors were responsible for informing eligible patients about the purpose and content of this research and determining their willingness to participate. The interviewer team comprised senior undergraduate or graduate students recruited from Shanghai Jiao Tong university school of medicine who had experience with administering questionnaire surveys. The interviewer team was responsible for conducting face-to-face interviews with patients. All interviews were conducted twice a week over nearly 7 months.
In each interview, one member of the interviewer team invited patients one by one to a quiet room, so as to keep patients focused on completing the questionnaire. The interview comprised three phases; during the first phase, patients were provided with an informed consent form to sign. Interviewers answered all patients’ concerns in principle. During the second phase, patients were required to complete the questionnaire on their own; no private information was collected during this process. In the third phase, interviewers verified the completed questionnaire to ensure no items were missed or ignored. At the end of the interview, patients receive a cash stipend (80 RMB, nearly 12 USD) in consideration of their cooperation in this study.
The questionnaire included self-reported demographic characteristics, free oral PrEP adherence questionnaire, and PrEP uptake behavior items, to evaluate patients’ adherence to PrEP14. Free oral PrEP adherence questionnaire used in this research is original, and eight items were included, based on factors found to be associated with the acceptability of free oral PrEP in our previous team work.14 Those eight items were used to measure participants’ attitudes, subjective norms, and perceived behavioral control. Behavioral intention was measured using an additional question separate from the questionnaire, to predict the actual performance of each individual. Considering the conciseness and clarity of the questionnaire items, four subscales were established: benefits, barriers, peer support, and self-efficacy. Eight items were included, with two items for each subscale. Each item on the questionnaire was scored using a three-point scale: 0 points for "disagree", 1 point for "unclear", and 2 points for "agree"; the total score on the questionnaire was 16 points. Higher scores indicated higher adherence to PrEP. Both positive and negative questions were included, to reduce participants’ response bias to a certain extent. (full questionnaire was shown in additional file 1)
Participants were asked the following question apart from the adherence questionnaire: Would you be willing to take PrEP on a daily basis during the following 6 months? There were five responses, corresponding to different degrees of PrEP uptake willingness or behavioral intention (answers: 1 = definitely not, 2 = unlikely, 3 = neutral, 4 = likely, 5 = definitely will). Given that not all behavioral intentions translate into action,26 we selected “definitely will” as the standard predicting positive behavior, in a conservative manner.
For item analysis, we used the critical ratio and Spearman correlation coefficient to test and evaluate the relevance and reliability of all items. In the calculation of critical ratio, participants were divided into two groups based on their total adherence scores: a high-adherence group (participants ranked among the 27% with the highest scores) and a low-adherence group (participants ranked among the 27% with the lowest scores). We conducted a t-test for scores of each item in the high- and low-adherence groups. A Spearman correlation test was conducted between each item score and the total score.
For reliability analysis, Cronbach’s alpha coefficient was used to evaluate the internal consistency of the questionnaire, and the Spearman–Brown split-half coefficient was used to evaluate reliability of the questionnaire.
A panel of eight experts specializing in population health and health behavior was invited to conduct the content validity analysis, scoring the relevance of each item to the corresponding content concept, the simplicity and clarity of each item, and the ambiguity of each item. In addition, the experts gave their opinions, based on a four-point scale, in evaluating the questionnaire using the content validity index (CVI), including the item-level CVI (I-CVI) and scale-level CVI (S-CVI);
For construct validity, factor analysis was performed. Samples were randomly divided into two groups. Group 1 (406 patients) was used for exploratory factor analysis (EFA) and Group 2 (410 patients) for confirmatory factor analysis (CFA). In the EFA, Bartlett's test of sphericity and the Kaiser–Meyer–Olkin (KMO) value were applied for eligibility; principal component analysis was adopted to extract common factors. In the CFA, the maximum likelihood method was used to verify the fitness of the initial model. Goodness-of-fit indices such as c2, degrees of freedom (df), p-value, root mean square error of approximation (RMSEA), goodness-of-fit index (GFI), adjusted goodness-of-fit index (AGFI), and comparative fit index (CFI) were adopted.
The value of average variance extracted (AVE) and construct reliability (CR) were used to evaluate convergent validity. We used the square root value of AVE and related analysis results to evaluate discriminant validity.
For predictive value and score translation, univariate logistic regression, hierarchical logistic regression, and receiver operating characteristic curve (ROC) analyses were carried out to assess the predictive value of adherence questionnaire scores in discriminating patients with positive behaviors (high-adherence group) and negative behaviors (low-adherence group), and to determine the cutoff score.
The data were analyzed using SPSS 25.0, AMOS 22.0, and MedCalc 19.0.