Page 82 - IJPS-8-1

International Journal of
Population Studies Urbanization and body weight

45.7% in 2015 (Table S2 in Supplementary File). During
the same period, the sample percentage of overweight men
more than doubled from 14.8% to 51.4%. Weight changes
are also seen in the measures related to abdominal obesity.
For example, the average WC increased by about 8% from
74.9 cm in 1993 to 80.6 cm in 2015 for women and by
about 12% from 76.4 cm to 85.4 cm for men during the
same period. Using WC ≥90 cm for men and WC ≥85 cm
for women as cutoffs, the sample percentage of abdominal
obesity grew by more than 150% from 14.8% in 1993 to
38.1% in 2015 for women and more than tripled from 8.3%
to 38.3% for men during the same period.
The secular trend of gender gap in weight gain varies
across different measures of body weight status. Among
the continuous measures depicted in Figure 1, men Figure 3. Trend of urbanicity index at the community level, stratified by
had on average significantly lower BMI and WHtR than community types.
women in the early 1990s, but the gender gaps narrowed
over time. In fact, the gender gap in average BMI lost its
statistical significance by 2000 and men’s average BMI pace of urbanization was uneven across different types
gradually surpassed that of women thereafter. Men had of communities – sustaining the between-community
on average significantly higher WC and WHpR than gaps in the level of urbanization over two decades, which
women in the early 1990s, with the gender gap growing might be attributable to public policies and investments
for WC but remaining relatively stable for WHpR in the in favor of urban development during China’s reform era
study period. In contrast, men had generally lower rates (Xie & Hannum, 1996). Therefore, the observed growth in
of abdominal obesity according to WC and WHpR cutoffs population-level body weight described above could have
between 1993 and 2015, but the gender gap in WC-based been driven by both within- and between-community
rate of abdominal obesity converged by 2006 (Figure 2). differences in urbanization.
The gendered patterns across measures of weight growth Table 1 shows that the average age of the respondents was
between the continuous and dichotomous measures of WC around 42 years; more than 80% were married; and most had
and WHpR may be attributed to gender-specific cutoffs of completed elementary or middle school education, with men
abdominal obesity. on average having a higher level of education than women.
The person-year observations in the analytical sample were
4.2. Descriptive statistics of independent variables
by and large evenly distributed across household income
Figure 3 shows the trends of average urbanicity index quartiles, provinces, and survey waves.
values and the associated 95% confidence intervals,
stratified by four community types used in the CHNS 4.3. Regression results for continuous outcomes
sampling stage. Rapid urbanization took place in the Table 2 reports multilevel regression disaggregation results
sampled communities, with the average urbanicity index for the longitudinal associations between urbanization and
score growing by almost 50% from 46.4 in 1991 to 71.5 continuous measures of body weight status. To preserve
in 2015 (Table S3 in Supplementary File). The pace of the space, only the main coefficient estimates of interest
urbanization was fastest in village communities, where were presented. Full regression estimates from selected
the average urbanicity index score rose by nearly 70% models can be found in Table S4 in Supplementary File and
from 34.8 to 59 over two decades. Nevertheless, Figure 3 full results from other models can be requested from the
indicates that the average level of urbanization in village author. In the female subsample, the urbanicity index score
communities remained significantly lower than in any was significantly associated with BMI and WC (Model 1),
other type of community throughout the period of 1991- but unrelated to WHpR or WHtR. Both associations
2015. In addition, the average level of urbanization in were attributed to between-community difference but
suburban communities was significantly lower than that not within-community difference in urbanicity index,
in city communities throughout the same period, and regardless of which disaggregation method was employed
significantly lower than that in town communities during (Models 2 and 3). After disaggregation, WHpR was not
most of the period. These findings suggest that despite a associated with either between- or within-community
growing urbanicity within each type of community, the difference in urbanicity index; neither was WHtR.

Volume 8 Issue 1 (2022) 76 https://doi.org/10.36922/ijps.v8i1.334