Ethics statement
The protocol in this study was approved by the Ethics Committee of the Institute of Clinical Pharmacology, Central South University, and signed informed consent was obtained from each interviewee. No minor or child was enrolled in this study.
Participants
This study was a subsequent follow-up study of a large community-based survey. The previous survey had covered 8 counties (Yueyang, Lingxiang, Huarong, Qianlianghu, Ziyang, Anxiang, Datonghu, and Longshan) that had been directly exposed to the 1998–1999 flood in Hunan province, China. All enrolled family members who were 16 years of age or over were asked to participate. Between January and May 2000, trained research assistants carried out face-to-face interviews (with on-site supervision by psychologists), using a pre-constructed questionnaire. The subjects were interviewed to ascertain PTSD and to collect demographic data. A total of 2336 (9.2 %) subjects were diagnosed as probable PTSD-positive individuals among a total of 25,478 study subjects according to the DSM-IV criteria [12].
The participants for the current study were enrolled from 3 counties: Huarong, Ziyang, and Anxiang, which have not had flood since 2000. In these three counties, 851 victims had been diagnosed as probable PTSD-positive individuals. All 851 victims were recruited for this follow-up study.
Data collection
439 victims were interviewed from house to house in Huarong in December, 2012; 412 victims were interviewed from house to house in Ziyang and Anxiang from May to June, 2014. Before survey, we wrote an investigator manual with the collaboration of psychologists and trained 8 interviewers, who worked at the local Centres for Disease Control and Prevention and had a bachelor’s degree or higher. With a unified understanding of all items, these 8 interviewers carried out face-to-face interviews using a questionnaire to ascertain PTSD and to obtain characteristics of each interviewee. The interviewers received onsite supervision from psychologists.
PTSD outcome
The diagnosis of PTSD was made according to PTSD Checklist-Civilian version (PCL-C) questionnaire developed from DSM-IV, which was also used in our former survey in 2000 [12]. The PCL-C was highly internally consistent (α = 0.94) and had good convergent and discriminant validities [15]. The questionnaire for PTSD had been previously tested in Chinese populations and had been proven to be valid and reproducible [10]. This survey includes 17 symptoms scored as 0 = none, 1 = slight, 2 = moderate, 3 = severe, and 4 = extreme. Subjects whose score was equal to or greater than 2 were defined as positive for that symptom. The 17 symptoms of PTSD were further divided into 3 groups, representing 3 sets of diagnostic criteria: B, C, and D. Subjects were given a diagnosis of probable PTSD if Criterion B, C, and D symptoms were all positive. All suspected cases were diagnosed as ‘probable PTSD’ because the diagnosis of PTSD may not be accurate although the interviewers received on-site supervision from psychologists.
Individual characteristics
The demographic characteristics, post-flood severe stressors, frequency of general collective action around 2006, and frequency of general collective action around 2013 were obtained during the interview. Demographic variables included gender, age, and education level. Post-flood severe stressors were measured by the question: “After 2000, have you experienced or witnessed an incident, which almost caused death or serious injury and caused you to be frightened?” Respondents could answer either “yes” or “no”. The question about post-flood severe stressors was put right after questions about demographic characteristics. If the answer of post-flood severe stressors was ‘yes’, the interview would be wrapped up.
Questions about “frequency of general collective action” were put together and shared a description: “General collective action means participating in activities organized by the following groups: political party, trade union, environmental group, parents’/school association, tenants’/residents’ group or neighborhood watch, church organization, voluntary service group, pensioners group/organization, social club/working men’s club, sports club, and the Women’s Institute [16].” Frequency of general collective action around 2006 was measured by the question: “Recalling the situation around 2006, to what extent did you take part in general collective action?” Frequency of general collective action around 2013 was measured by the question: “According the situation this year, to what extent do you take part in general collective action?” Respondents could answer 1 = “never”, 2 = “occasionally”, or 3 = “frequently”. Flood-related stressors and frequency of general collective action in 2000 were assessed in previous survey. Frequency of general collective action in 2000 was measured by the question: ‘To what extent did you take part in general collective action?’ Respondents could answer 1 = “never”, 2 = “occasionally”, or 3 = “frequently”. Flood-related stressors include: life-threatening experiences (e.g., narrow escape from flood waters requiring emergency rescue), death of a loved one, victimization after the flood (e.g., robbery–assault), physical illness/injury caused or exacerbated by the flood, extreme physical adversity (e.g., difficulty obtaining food or clothing), extreme psychological adversity (e.g., living in circumstances where the respondent had to use the toilet or change clothes without privacy), major property loss, income loss, and housing adversity (e.g., multiple moves). Victims answered each question with either a “yes” or “no”.
Statistical methods
Individual factors between interviewed victims and unreachable victims were first compared by χ2 test. Although the PCL-C had been used widely to diagnose PTSD, nine items on the questionnaire have no word about “stressor” in the stems. In order to avoiding the influence from other flood unrelated stressors, we excluded the victims who have experienced post-flood severe stressors in all following analyses.
Current probable PTSD positive rates for two interviewed time (2012 and 2014) in Huarong, Ziyang and Anxiang counties were described respectively and compared by χ2test, and then the date was put together and the positive case numbers and rates of each symptom cluter were described. To correct for possible confounding, the effect of the three counties was taken into consideration by including two dummy variables as predictors into the model.
In order to reveal the impact of general collective action on PTSD, a new indicators, fluctuation of frequency, was created. Score of 2006 minus score of 2000 is result1; score of 2013 minus score of 2006 is result2. If result1 and result2 are both 0, the result of “the fluctuation of frequency” is “no change”. If result1 and result2 are both positive number, or one of them is positive number, the other is 0, the result of “the fluctuation of frequency” is “becoming more frequent”. If result1 and result2 are both negative number, or one of them is negative number, the other is 0, the result of “the fluctuation of frequency” is “becoming less frequent”. If one of result1 and result2 is negative number, the other is positive number, the result of “the fluctuation of frequency” is “fluctuating”.
The rates of current probable PTSD among victims of different gender, age, education level, flood-related stressors, frequency of general collective action in 2000, around 2006, and around 2013, the fluctuation of frequency were compared respectively by univariate logistic regression models. Adjusted ORs for current probable PTSD were estimated with multiple logistic regression models. A total of 5 models were analyzed: frequency of general collective action in 2000, around 2006, around 2013, the fluctuation of frequency were included into 4 different models as independent variable, respectively; in “All FGCA model”, frequency of general collective action in 2000, around 2006, around 2013 and the fluctuation of frequency were all included. Gender, age, education level, and flood-related stressors were included as independent variables in every model. All analyses were performed with SPSS Version 18.0.