The data used in the study were from the Social Environment and Biomarkers of Aging Study (SEBAS) in Taiwan, collected from a representative subsample randomly selected from the 1999 Taiwan Longitudinal Study of Aging (TLSA). The Bureau of Health Promotion of the Department of Health in Taiwan granted approvals for the protection of human subjects for SEBAS. Data were collected through home visits by interviewers. A written consent was obtained from each of all participants with the rare exceptions when a participant who could not read or write. In that case, a consent form was read by the interviewer and signed by a witness.
The 2000 wave of the SEBAS (hereafter the SEBAS 2000) was conducted from July through December 2000 using 27 original primary survey units (PSU) from the TLSA and 10 new townships . All respondents residing in a given PSU were selected for interviews. Data were collected through face-to-face interviews using a structured questionnaire, physical examination, and bio-specimen collection. This human subjects research was approved by the institutional review boards at Princeton University, RAND, Georgetown University, and the Bureau of Health Promotion in Taiwan.
The SEBAS 2000 in-home interview was conducted by a public health nurse who was well-known and highly-respected locally. The questionnaire covered chronic conditions, physical functioning, psychological well-being, cognitive capacity, utilization of health services, and social networks/support. The interviewer then evaluated each respondent’s health. Among the 1,497 respondents, 1,023 received physical exams at a nearby hospital several weeks later (111 were not eligible for health examination and 363 refused to participate). After list-wise deletion, of the 1,023 individuals who completed the survey, 1,005 subjects without any missing data were included in statistical analyses. The age range of participants was 54–91, with 628 persons aged 65 or older and 421 women.
Biospecimens were collected by survey staff during the in-home interview and related the hospital visit. Survey staff collected the 12-hour urine specimen at the participant’s home and accompanied the participant to the hospital on the morning of the scheduled appointment. During the hospital visit, participants were asked about their health history, family disease history, health-related behaviors, and current long-term medications. Blood pressure and anthropometric measurements (i.e., respondent’s height, weight, waist and hip circumference) were performed and a blood specimen was taken to measure biomarkers.
Union Clinical Laboratories (UCL) took responsibility for immediate shipment of the specimens to their headquarters in Taipei, followed standard laboratory protocols for conducting assays, and provided the results to the Bureau of Health Promotion (BHP, in the Department of Health of Taiwan) within two weeks. One genetic marker, the APOE gene, was also obtained by blood specimen using the polymerase chain reaction amplification refractory mutation system (PCR-ARMS) and polymerase chain reaction restriction fragment length polymorphism (PCR-RFLP) analysis. Data quality evaluations conducted during and after the fieldwork by BHP indicate that the SEBAS 2000 rendered reliable data .
The measure of self-rated health status was based on a simple question: “Regarding your current state of health, do you feel it is excellent, good, average, not so good, or poor?” A binary measure of poor SRH was coded for logistic regression analyses (1 = not so good or poor SRH, 0 = excellent/good/average SRH). Other ordinal categorizations were also tested and the conclusions were very similar.
The SEBAS 2000 measured fourteen diseases or conditions: high blood pressure, diabetes, heart diseases, stroke, cancer/malignant tumor, bronchitis/emphysema/pneumonia/lung disease/asthma and other lower respiratory tract diseases, arthritis/rheumatism, gastric ulcer/stomach ailment, liver/gallbladder disease, hip fracture, cataract, kidney disease, gout, and spinal/vertebrae spur. All the chronic diseases were self-reported in response to the question “Have you ever had this disease?” Each condition was coded 1 if an individual said yes and 0 otherwise. For this analysis we only included chronic diseases with a prevalence rate of 10 % or higher and with a significant bivariate association with SRH. While hypertension was frequent, it was found to be underreported through evaluation of the SEBAS 2000 with other survey datasets and validation with laboratory results . Eventually, only four diseases, diabetes, heart diseases, gastric ulcer, and chronic obstructive pulmonary disease (COPD) were used to measure physical health in this study.
All respondents who reported having one of these conditions at the time of the interview indicated that a physician delivered the diagnosis (94.7 % for heart diseases, 97.3 % for diabetes, 88.5 % for COPD, and 87.7 % for ulcer). The prevalence rates of these four diseases in SEBAS 2000 were similar to those found in the National Health Interview Survey in Taiwan in 2001 . Moreover, the evaluation of the SEBAS 2000 revealed that, with the exception of hypertension, the accuracy of self-reported chronic disease information in the Taiwan study was similar to that in the United States .
The information of the APOE gene, which was obtained from blood specimens analyzed for allele variant, had three alleles: E2, E3 and E4. A binary measure of the APOE4 allele was coded as 1 if the individual carries one or two copies of the E4 allele (carrier), and 0 otherwise (non-carrier).
Stress was measured by traumatic events and chronic life stress. Experiencing housing damage during the 1999 earthquake, which occurred a year before the survey was conducted, was used to measure life stress due to a traumatic event. Environmental stress such as financial difficulty was a major element of psychological stress and was used to measure chronic stress.
The 1999 earthquake was the greatest disaster in late 21st century in Taiwan. It occurred in Jiji, Nantou County, Taiwan on Sept. 21, 1999. Some 2,415 people were killed and 11,305 were injured. The “Quake of the Century” had a profound effect on the whole island, and even on some mainland provinces. The Richter magnitude scale of the 1999 earthquake ranged from 4 in the south (Kaohsiung) to 5 in the north (Taibei) and east (Hualian) and 7 in the west (Yunlin and Jiayiin). Survey participants were asked “Was there any damage or loss to the house in which you usually lived prior to the earthquake?” Response categories were yes (1) and no (0).
Financial condition was measured by the question “Do you (and your spouse) have enough money or any difficulty meeting monthly living expenses or other expenditures?” Possible responses were: “1 = enough money, with some left over; 2 = just enough money, no difficulty; 3 = some difficulty; and 4 = much difficulty.” Financial difficulty was coded 1 if the individual selected the third or fourth categories, and 0 otherwise.
Health behaviors included smoking, drinking alcohol, physical exercise, and diet. Four health behaviors were recoded into binary variables (1 = yes, 0 = no) measured by the following questions: “In the past six months, did you smoke?” “In the past six months, did you drink alcohol?” “Do you drink milk every day?” and “Do you eat at least three servings of vegetables and two servings of fruit every day?” For physical exercise, a three-category option of frequency (<=1, 2–5, and 6+) was designed in the questionnaire, and we directly used its categorization without any modification.
To obtain robust results, we controlled for socio-demographic factors in the statistical analyses. Socio-demographic variables included gender, age, marital status, ethnicity, urban or rural residence, education, and occupation. We also included obesity as a measure of physical condition. In accordance with WHO’s criterion of body-mass index (BMI) for Asian populations, we defined obesity as a BMI greater than 23, weight/height (kg/m) .
Due to a documented relationship with SRH, disability in instrumental activities of daily livings (IADL) was also included in the analysis . IADL disability involved limitations on buying personal items, managing money/paying bills, riding bus or train by oneself, doing physical work at home, doing light tasks at home, and making phone calls. Respondents were coded as IADL disabled (1) if the individual had at least some difficulty with one or more items, and 0 otherwise.
After examining descriptive statistics (Table 1), we estimated a logistic regression model of the associations of stress factors and health behaviors with odds of poor SRH while controlling for socio-demographic factors (Model I in Table 2). We then incorporated chronic diseases and the APOE4 allele to examine how measures of physical health and genetic information may alter the associations in Model I (Model II, Table 2). Finally, because of significant gender differences in the relative importance of factors associated with SRH , we examined all interactions between gender and chronic diseases, the APOE4 allele, stress factors, and health behaviors. However, only one interaction (between gender and ulcer) was significant. We thus analyzed a model that included this interaction in addition to all factors in Model II (Model III, Table 2).