Study design and participants
The population was sampled to represent the general population of Shandong Province, China (18–49 years of age). General population refers to males and females who had lived in their current residence for at least the past 6 months. The estimated sample size was 8074 individuals based on the national survey conducted in 1999–2000 [6]. (Additional file 1) The multi-stage sampling process included four steps, as follows: Shandong Province was divided into four strata based on geographic region (east, northwest, south, and middle); 2–4 urban districts or counties were randomly extracted from each stratum, producing 12 primary sampling units (PSUs; Fig. 1); 4–7 subunits (rural townships and urban street districts) were selected from each PSU using the probability proportionate to size sampling (PPS) method; a total of 59 subunits were selected; 3–12 rural villages or urban communities were selected from each subunit using the aforementioned sampling method, with more populated subunits having a higher probability of selection; villages and communities were selected with a proportion of 1:1; a total of 184 rural villages and 183 urban communities were included; residents, 18–49 years of age, who had lived in their current residence for at least the past 6 months were selected and arrayed in order by age and gender; systematic sampling was performed based on the gender and age distribution of the Shandong Province population to produce the sample; Thus, 22 persons were drawn from each village or community according to sample size and number of selected villages or communities.
The inclusion criteria were as follows: participants born between 1 January 1966 and 31 September 1997; continuous residence at the study site for at least the past 6 months; consent to provide urine specimens for testing; and willing to participate in the study and complete the questionnaires. The exclusion criteria were as follows: individuals denying sexual debut (including oral sex); and inability to provide correct information, such as individuals with a mental illness or alcohol abusers.
This study was approved by the Ethics Committee of Shandong Provincial Institute of Dermatology and Venereology (approval number: 2016–04). Oral informed consent was obtained from each participant by interviewers before interview.
Procedures
Before implementation of this survey, 3 days of training on fieldwork staff was provided by epidemiology and clinical laboratory professionals in each of the PSUs. One laboratory technician for urine collection and coding, one male and one female questionnaire interviewers, and one financial staff for providing rewards to participants were selected on the basis of course performance for each field team. A trained supervisor from the study group was also assigned to each field team for ethics and quality control of the interview.
Before sampling, individuals who continuously lived in their current residences for at least the past 6 months were identified by village/community committee staff or physicians by 3 steps. First, they referred to the official registers of household or health record of local residents for eligible individuals. Second, they inquired and recorded unregistered migrants who had lived in the local village/community for 6 months or more. Third, these two groups of people were combined in a roster for sampling. The sampled participants were informed to be interviewed in the village/community clinic or meeting room in private. For individuals who could not be reached via telephone, staff or physicians would visit them in their home. Individuals who could not be reached after repeated visits were excluded from the study. To retain privacy, questionnaire interviews were performed in a private room, and interviewers were the same gender as the respondent. Most of the selected respondents worked during the daytime, so the interviews were usually performed early in the morning or in the evening. Computer-based interviews were not performed because the pilot investigation showed that many participants could not use the computer to navigate the questionnaire, especially in villages. We adopted questionnaire-based interviews to collect information. Face-to-face interviews addressed demographic information, including gender, age, marital status (unmarried, married, divorced, or widowed), educational level (elementary school or below, middle school, senior high school or technical school, and university or above), and location of residence (rural or urban). Other information on the questionnaires was completed by the respondents themselves. The structured questionnaires included information regarding occupation, income, living status (living with family, living with roommates, or living alone), living with a spouse (yes or no), smoking (a smoker was defined as someone who had smoked > 100 cigarettes during their lifetime), number of episodes of intoxication in the past year (0, 1–3, 4–6, and > 6), STIs (including syphilis, gonorrhea, CT infection, condyloma acuminatum, genital herpes, and HIV infection) in the past 5 years (yes or no), quality of sex life (satisfied or not satisfied), sexual behaviors, such as age at first intercourse (≤20 years and > 20 years), number of sex partners (1 and ≥ 2), and number of new sex partners in the past year (≤ 1 and ≥ 2). After the interview, the supervisor checked the questionnaires for missing values and logical errors and wrote a unique alphanumeric code on the questionnaires instead of the individual’s name.
Every respondent was instructed not to urinate for at least 1 h before they participated in the interview. Approximately 30–50 ml of first-void urine was collected into a urine cup. Laboratory technicians promptly transferred the urine into Cobas PCR Media tubes (Roche Molecular Systems, Inc., Branchburg, New Jersey, USA) using a disposable pipette and the tube was inverted 5 times to mix. Each specimen was labeled with a unique alphanumeric code, which was the same as on the questionnaire. Urine was transported inside a Styrofoam cooler to the local laboratory within 3 h of collection and was stored 4 °C as soon as the staff received the specimen. Urine was transported inside the cooler to the laboratory of Shandong Provincial Institute of Dermatology and Venereology in Jinan within 2 weeks.
Chlamydia trachomatis and Neisseria gonorrhoeae (NG) DNA in urine specimens were detected with a Roche Cobas 4800 CT/NG Assay (Roche Molecular Systems, Inc.), which had a sensitivity of 96.6% and a specificity of 100% [16]. Only participants with positive results were notified, then referred to a local STI clinic or general hospital for treatment.
Statistical analysis
Questionnaire information, as well as testing results, was double-entered into Epidata 3.1 (EpiData association, Odense, Denmark); discrepancies were checked against raw data. We applied selection probability weights which were inversely proportional to the selection probabilities for the number of units at each sampling step to balance differences in the probabilities of selection. Non-response weighting was made by adjustment among those who participated in this survey with those who did not. Post-stratification weighting was made by adjusting data according to age and gender distribution of the Shandong census data. We analyzed data with surveyfreq and surveylogistic methods using SAS 9.3 (SAS Institute, Inc., Cary, NC, USA). The prevalence with 95% confidence interval (CI) of CT and NG was calculated based on total population weights and Taylor series linearization. We determined the association between CT infection and demographic and behavioral variables with logistic regression and reported crude odds ratio (ORs) in univariate analysis. Risk factors with a p value < 0.20 in univariate analysis were applied into the multivariate logistic model. Adjusted ORs and 95% CIs are presented. The variance inflation factor (VIF), condition index, and variance proportions were calculated to examine multicollinearity among risk factors. Maximal variance inflation factor values ≥10 or maximal condition index values ≥10 and a corresponding variance proportion value > 0.5 indicated multicollinearity among variables. All tests were two-sided, and a p value < 0.05 was considered statistically significant.
Proportion of symptomatic and treatment, incidences of various complications after CT infection were on the basis of our primary data as well as previous published studies [17, 18] (Figs. 2 and 3, Additional file 2). Medical costs for CT infections were based on diagnosis and treatment of the infection and complications; costs related to transportation, lodging, and time missed from work were not included into analysis. According to guidelines for STI treatment and control in China, we assumed that the diagnosis of CT infections was established with PCR testing and the treatment of CT infections involved a single 1 g dose of oral azithromycin [19]. Medical costs of cases with complications were estimated based on published research in Shandong Province [17]. The discounted rate of medical costs was assumed to be 3%. Sensitivity analysis was performed by adding uncertainty factors, such as the prevalence of CT infections, proportion of asymptomatic CT infections [20], and diagnostic method (antigen detection), into the computation.