Study design and subjects
The K-Stori is a nationwide survey designed to investigate broad health issues among Korean women according to five stages in the life cycle of women. Specific questionnaires were designed for each of the five stages (adolescence, childbearing, pregnancy & post-partum, menopause, and older adult), since life cycle approaches have been found to be effective in understanding and managing health problems and health promotion plans. In total, 15,000 women were recruited (3000 women per stage). Eligibility for inclusion and survey methods for each life cycle stage are shown in Table 1. The target population comprised women between 14 and 79 years of age for all households in Korea. Institutional households were excluded. Subjects who did not agree to participate in the survey and those with difficulties communicating were excluded.
Table 1 Eligibility and survey method for each life cycle stage
Considering the generalization of survey results, in-person interviews were carried out. However, for adolescents, the questionnaire was conducted via an online survey to minimize distortion of responses to sensitive health behavior items (Fig. 1).
Sampling
We randomly sampled 3000 women in each stage of the life cycle for a reliable and representative research design. Weights are presented to compare data among stages. For random sampling, a multi-level, stratified, probability-proportional statistics extraction method was used as a sampling framework using the 2010 Population and Housing Census. In order to generalize the survey results, the subjects were selected by random sampling for 16 cities and provinces (seven special and metropolitan cities, and nine provinces). The extraction method was used to stratify recruits by region and dong-eup/municipal district, after which 200 households were sampled. A sample of 15 households was extracted from each sampling area, in which the principle was to survey female household members aged 14–79 years in each sample household. Interviewers visited each of the 15 extracted households.
To recruit women in adolescence and pregnancy/post-partum period, interviewers planned to visit schools and obstetrics and gynecology or post-partum care centers. In order to select the same survey area as for the other life cycle stages, a system extraction method was used to identify local schools for adolescents and obstetrician and post-partum care centers for pregnant and post-partum women based on the same sample design area for the household survey subjects. If there were no schools or obstetrics and gynecology clinics in the sampling area, they were replaced by the nearest one.
Sample size calculations
Statistical power according to various odds ratios was calculated in order to examine whether the sample number of 3000 women for each life cycle stage was sufficient. Logistic regression analysis was used to calculate the statistical power that can be detected by various odds ratios when the risk group was assumed to be approximately 20% of the total subjects and when the outcome prevalence of the unexposed group was assumed to be about 10%. With these assumptions, the statistical power that can detect an OR = 1.6 was about 82% for surveying 3000 women in each life cycle stage. In addition, for subgroup analysis of a particular group of subjects (e.g., 1000 to 2000 persons), sufficient statistical power will be obtained in most analyses.
Assuming a 20% exposure to risk factors and a prevalence of 10% in the unexposed group is conservative. For general analysis, the analysis will be divided for 50% of the exposed and unexposed groups. As the prevalence of the outcome variable increases, the statistical power will be much higher. Therefore, the number of subjects to be surveyed in this study is expected to have sufficient power for statistical analysis of the survey data on women’s health issues.
Design of survey questionnaires
The survey questionnaires were structured differently for each life cycle stage to identify stage-specific health issues. Three strategies were implemented to develop a questionnaire. First, we analyzed national health data to understand the state of women’s health in Korea. We also reviewed both domestic and foreign health behavior-related surveys, especially those developed for women. Second, questionnaires were developed by gathering opinions from a number of experts: The Women’s Health Forum was held to diagnose women’s health problems, and multidisciplinary experts directly participated in the questionnaire development. Experts in each field identified issues related to women’s health from their viewpoint. The fields covered by these experts included medicine, public health, preventive medicine, statistics, nursing, geriatrics, communication health, social welfare, and sociology. Third, we analyzed health issues of interest to women through internet information analysis and focus group interviews. Using keywords, such as “women, health,” we searched the web and mapped blogs and internet cafes most used by women in Korea during January to October 2015. Focus group interviews were also conducted to identify women’s health issues and unmet needs in health care. Through these processes, we identified areas to be investigated for each life cycle stage. In questionnaire development, we tried to adopt questionnaire items that have been validated in previous studies.
The questionnaire items and the assessment tools adapted in this survey are shown in Table 2. The main questionnaire items assess health status, perceived health, reproductive health, physique and body type, physical activity and exercise, eating habits, food intake, smoking, drinking, sleeping, medical service use, health communication, social support and relationships, violence, social-economic status, and demographic characteristics. Each of the five stage-specific questionnaires consists of 26 questions common among all questionnaires and more than 100 questions specified to each life cycle stage.
Table 2 Questionnaires items and assessment tools in K-Stori
A pilot study was conducted with 30 women per each life cycle stage (total of 150 women) to determine the feasibility and validity of the survey. Through the pilot study, we examined whether the questionnaire constructed by the experts was able to convey the correct meaning of each question to the general public. Based on the results from the pilot study, the wording of some of the expressions in the questionnaire was modified.
Data collection
A total of 15,000 women aged 14–79 years completed surveys between April 2016 and June 2016. Trained interviewers from a professional research agency conducted door-to-door interviews to assess study eligibility. In order to select the survey subjects, the interviewers checked whether there was an eligible person in the household. Interviewers also visited schools and obstetrics and postnatal care centers to recruit eligible adolescent and pregnancy and post-partum women. The interviewers contacted subjects daily from 10 am until the evening, including weekends. Public holidays were excluded. Up to three attempts to contact an individual were made, at different times of the day. If contact was not made upon the third attempt, neighboring, alternative survey households were selected according to the predetermined contact order, and contact was tried again. Once the interviewers found eligible women, they explained the survey to and obtained signed consent from subjects who agreed to participate in the survey. Of the 37,334 people who were contacted, 15,084 interviews were completed. The survey response rate was 40.4%. Of these, a total of 15,000 interviews were included in the final analysis, excluding those who did not answer the main questions.
Respondent characteristics according to stage in life cycle are shown in Table 3. The mean age of the study participants was 41.2 years. Almost 80% of women resided in urban areas. Nearly 50% of women were included in the middle household income group (between $1700 and $3499). About 30% of women, not including adolescent girls, had a high school level of education. General characteristics differed according to stages in the female life cycle.
Table 3 General characteristics of the study participants in K-Stori
Quality management
In order to ensure the reliability of the survey, step-by-step quality control was carried out in accordance with the survey preparation, survey process, and data entry and verification process. In order to enhance cooperation with the initial contact process, the purpose of and information about the K-Stori were posted on the homepage of The Korea Centers for Disease Control and Prevention, and National Cancer Center. To reduce non-sampling error, a standardized guideline was developed and disseminated to interviewers, and an education session was held for all interviewers prior to beginning the survey.
Regarding data entry and verification, a double check system was carried out by the researchers and assistant researchers to minimize errors when editing. Surveys were reviewed according to standardized guidelines upon completion of the entire survey. Inaccurate response items were confirmed by calling the subjects’ phone. Regional surveillance supervisors maintained a Computer Aided Telephone Interviewing system for 30% of randomly chosen surveys to check the accuracy of responses. In order to minimize errors in the coding process, experienced coding staff members were selected, and preliminary education on the coding guidelines was carried out. Punching error was minimized by operating a double punching system in which two people simultaneously punched the same questionnaire for all items.
Ethical considerations
The study was approved by the institutional review board of the National Cancer Center, Korea (NCC2016–0062). An approved study description was provided to all eligible participants. The study description covered the research purpose, subject, content, duration, voluntary participation, withdrawal of consent, expected risks and benefits from participating in the research, and confidentiality. As the health information questions in this study assessed information on health behaviors, as well as mental and psychological factors, information security was strengthened. If the subjects agreed to participate in the study after reading the study description, participants were asked to provide written informed consent. For adolescents, because they were under 18 years of age, parental consent was obtained at the time of recruitment, or a parental consent form was sent to the adolescents’ homes and the survey was carried out after confirming consent. Women who were pregnant or who had recently given birth were allowed to consult with their spouse or partner and discuss their participation in the survey.