Table 2.
Parameter estimates (with 95% confidence intervals) for the change in mean WC over time
| Model 1a | Model 2b | |
|---|---|---|
| Overall analysis (total sample) | ||
| 2009 | 2.97 (2.01 to 3.93) | 1.26 (0.35 to 2.17) |
| 2013 | 4.93 (3.87 to 6.00) | 1.88 (1.09 to 2.67) |
| Subgroup analyses | ||
| Sex | ||
| Men | ||
| 2009 | 3.00 (1.56 to 4.43) | 0.27 (-0.62 to 1.17) |
| 2013 | 4.77 (3.18 to 6.37) | 1.09 (0.39 to 1.80) |
| Women | ||
| 2009 | 3.01 (2.00 to 4.02) | 2.32 (1.25 to 3.39) |
| 2013 | 5.07 (4.00 to 6.14) | 2.68 (1.64 to 3.71) |
| Age group (years) | ||
| 18–24 | ||
| 2009 | 3.13 (1.85 to 4.42) | 1.39 (0.37 to 2.42) |
| 2013 | 3.74 (2.34 to 5.14) | 1.75 (0.91 to 2.58) |
| 25–34 | ||
| 2009 | 3.35 (1.98 to 4.73) | 1.53 (0.42 to 2.65) |
| 2013 | 4.95 (3.74 to 6.16) | 1.97 (1.09 to 2.84) |
| 35–44 | ||
| 2009 | 2.72 (1.32 to 4.13) | 1.25 (0.36 to 2.15) |
| 2013 | 4.90 (3.41 to 6.39) | 1.88 (0.86 to 2.89) |
| 45–54 | ||
| 2009 | 1.98 (0.22 to 3.73) | 0.65 (-0.49 to 1.79) |
| 2013 | 5.09 (3.43 to 6.74) | 1.63 (0.56 to 2.7) |
| 55–64 | ||
| 2009 | 3.49 (1.26 to 5.72) | 1.25 (-0.08 to 2.57) |
| 2013 | 4.82 (2.50 to 7.14) | 1.94 (0.54 to 3.34) |
| Area of residence | ||
| Urban | ||
| 2009 | 3.79 (2.25 to 5.33) | 2.12 (0.88 to 3.35) |
| 2013 | 5.83 (4.24 to 7.42) | 2.42 (1.21 to 3.63) |
| Rural | ||
| 2009 | 2.21 (0.95 to 3.47) | 0.43 (-0.95 to 1.81) |
| 2013 | 4.08 (2.59 to 5.57) | 1.37 (0.34 to 2.41) |
| Education | ||
| Tertiary educated | ||
| 2009 | 2.05 (0.54 to 3.55) | 0.80 (-0.59 to 2.19) |
| 2013 | 3.56 (2.19 to 4.94) | 1.86 (0.80 to 2.92) |
| Non-tertiary educated | ||
| 2009 | 3.18 (2.17 to 4.2) | 1.40 (0.49 to 2.32) |
| 2013 | 5.37 (4.19 to 6.54) | 1.91 (1.09 to 2.74) |
Abbreviations: WC waist circumference
aModel 1 regression model adjusted for age, sex, survey year and height
bModel 2 regression model adjusted for age, sex, survey year, height and weight
The coefficients shown in the table represent the estimated change in mean WC relative to 2005 (the reference category) Separate regression models were fit for each subgroup. The subgroup effects were estimated by including an interaction term between the relevant variable (sex, age group, education status or area of residence) and survey year. The model with the area of residence by survey year interaction effect also included the main effect for area of residence. Similarly, the model with the education status by survey year interaction effect also included the main effect for education status