Table 1.
Characteristics of Participants in the Surveys in 1997, 2000, 2004, and 2006
| Characteristics* | 1997 (n = 1,668) | 2000 (n = 1,671) | 2004 (n = 2,154) | 2006 (n = 2,352) | p trend † |
|---|---|---|---|---|---|
| Age (years), mean (SD) | 69.1 (6.9) | 69.2 (6.7) | 69.6 (6.8) | 69.7 (7.0) | .005 |
| Age (years), n (%) | |||||
| 60–69 | 1,033 (61.9) | 1,003 (60.0) | 1,242 (57.7) | 1,341 (57.0) | |
| 70–79 | 488 (29.3) | 536 (32.1) | 730 (33.9) | 799 (34.0) | |
| ≥80 | 147 (8.8) | 132 (7.9) | 182 (8.4) | 212 (9.0) | .004 |
| Women, n (%) | 906 (54.3) | 904 (54.1) | 1,144 (53.1) | 1,249 (53.1) | .287 |
| Rural residents, n (%) | 995 (59.7) | 1,004 (60.1) | 1,335 (62.0) | 1,488 (63.3) | .002 |
| Education, n (%) | |||||
| No formal school | 1,098 (71.0) | 961 (62.4) | 1,102 (51.4) | 1,279 (54.6) | |
| Primary school | 229 (14.8) | 284 (18.4) | 519 (24.2) | 442 (18.9) | |
| Middle school or above | 220 (14.2) | 296 (19.2) | 525 (24.5) | 623 (26.6) | <.001 |
| Ever smoking, n (%) | 478 (28.8) | 478 (28.7) | 711 (33.1) | 766 (32.6) | <.001 |
| Alcohol intake, n (%) | 256 (15.7) | 295 (18.2) | 359 (16.7) | 373 (15.9) | .931 |
| Obesity, n (%) | 127 (8.6) | 148 (9.1) | 208 (10.2) | 214 (9.7) | .356 |
| Hypertension, n (%) | 735 (49.1) | 789 (48.1) | 995 (48.0) | 1,023 (45.3) | .020 |
| Diabetes, n (%) | 53 (3.2) | 68 (4.3) | 79 (3.7) | 90 (3.9) | .979 |
| Myocardial infarction, n (%) | 19 (1.2) | 27 (1.6) | 25 (1.2) | 43 (1.8) | .171 |
| Stroke, n (%) | 45 (2.8) | 51 (3.2) | 93 (4.3) | 76 (3.3) | .246 |
| Number of cardiometabolic diseases, n (%) | |||||
| 0 | 892 (53.5) | 843 (50.4) | 1,104 (51.3) | 1,261 (53.6) | |
| 1 | 707 (42.4) | 732 (43.8) | 923 (42.9) | 960 (40.8) | |
| ≥2 (multiple) | 69 (4.1) | 96 (5.7) | 127 (5.9) | 131 (5.6) | .810 |
Notes: *The number of subjects with missing values was 267 for education, 20 for smoking, 90 for alcohol intake, 516 for body mass index, 377 for hypertension, 155 for diabetes, 73 for myocardial infarction, and 137 for stroke. When these factors were considered as covariates in subsequent analyses, a dummy variable for each of these factors was created to represent the group of subjects with the missing value.
†The linear trend was tested using generalized estimation equations regression models, if applicable, controlling for age, sex, education, and living region.