Demographic characteristics of analysis population in survey 2020
All students submitted the questionnaire and the response rate was 100%. Excluding those invalid and inconsistency between the two inquiries, 1466 students in survey 2020 in total were included in the analysis sample of cross-sectional study. Sociodemographic information of participants, including gender, region, family structure, household monthly income per capita, father’s education level, mother’s education level and pocket money per month for students, were shown in Table 1.
The SSBs consumption frequency of students with different SSBs-related unhealthy attitudes or behaviors
Among 1466 respondents, 5.1 % consumed SSBs every day and 13.6 % consumed it 3-6 times/week. 680 (46.4%) junior school students would consume SSBs just because they want to drink, 121 (8.3%) would drink SSBs as water, 66 (4.5%) thought SSBs will not affect health, and 338 (23.1%) would purchase sugary drinks in advance. Students with above unhealthy attitude or behavior had higher SSBs consumption frequency (P < 0.05). See Table 2 for details.
Table 1 Sociodemographic characteristics of study subjects in survey 2020 (n=1466)
Sociodemographic Variable
|
n
|
%
|
Gender
|
Male
|
758
|
51.7
|
Female
|
708
|
48.3
|
Region
|
Urban
|
631
|
43.0
|
Rural
|
835
|
57.0
|
Family structure
|
Nuclear
|
1080
|
73.7
|
Non-nuclear
|
386
|
26.3
|
Household monthly income per capita
|
< ¥1000
|
94
|
6.4
|
¥1000-5000
|
982
|
67.0
|
> ¥5000
|
390
|
26.6
|
Father's education level
|
Primary school and illiteracy
|
77
|
5.3
|
Middle school
|
1099
|
75.0
|
College and above
|
290
|
19.8
|
Mother's education level
|
Primary school and illiteracy
|
117
|
8.0
|
Middle school
|
1040
|
70.9
|
College and above
|
309
|
21.1
|
Pocket money per month
|
None
|
296
|
20.2
|
<¥100
|
621
|
42.4
|
¥100-200
|
364
|
24.8
|
>¥200
|
185
|
12.6
|
Table 2 The SSBs Consumption frequency of junior school students with different beverage-related attitudes and behaviors, n (%)
|
n
|
≥1 times /day
|
3-6 times /week
|
1-2 times /week
|
1-3 times /month
|
<1 times /month
|
Z/χ²
|
P-value
|
Just want to drink SSBs
|
|
|
|
|
|
|
Yes
|
680
|
26 (3.8)
|
73 (10.7)
|
237 (34.9)
|
250 (36.8)
|
94 (13.8)
|
-6.192
|
<0.001
|
No
|
786
|
10 (1.3)
|
35 (4.5)
|
262 (33.3)
|
277 (35.2)
|
202 (25.7)
|
|
|
Drink SSBs as water
|
Yes
|
121
|
19 (15.7)
|
32 (26.4)
|
40 (33.1)
|
23 (19.0)
|
7 (5.8)
|
-9.343
|
<0.001
|
No
|
1345
|
17 (1.3)
|
76 (5.7)
|
459 (34.1)
|
504 (37.5)
|
289 (21.5)
|
|
|
Deem SSBs not affect health
|
Yes
|
66
|
5 (7.6)
|
17 (25.8)
|
19 (28.8)
|
19 (28.8)
|
6 (9.1)
|
31.653
|
<0.001
|
Not sure
|
135
|
10 (7.4)
|
16 (11.9)
|
48 (35.6)
|
392 (8.9)
|
22 (16.3)
|
|
|
No
|
1265
|
21 (1.7)
|
75 (5.9)
|
432 (34.2)
|
469 (37.1)
|
268 (21.2)
|
|
|
Purchase SSBs in advance
|
Yes
|
338
|
20 (2.5)
|
29 (7.4)
|
113 (34.0)
|
110 (35.9)
|
66 (20.2)
|
23.87
|
<0.001
|
No
|
1128
|
16 (1.4)
|
79 (7.0)
|
386 (34.2)
|
417 (37.0)
|
230 (20.4)
|
|
|
Binary Logistic Regression of the influence of unhealthy attitude or behavior factors on SSBs consumption frequency
In the multivariate analysis, frequency was taken as the dependent variable, and attitude or behavior factors, sociodemographic characteristic including gender, region, family structure, household monthly income per capita, father’s education level, mother’s education level and pocket money per month were taken as the independent variable. The code of the variable in model was shown in Table 3. OR means the multiple of the increased risk of high beverage consumption frequency for each grade of increase in independent variables. We fitted six multilevel logistic regression models. The first was the demographic characteristic model (Model 1). The second model included the family characteristic described above in addition to demographic characteristic (Model 2). The following four models included four attitudes or behaviors step by step (Model 3-6). The R2 of the models and odds ratios of variables were reported in Table 4.
The results showed that the final binary logistic regression was significant (R2=31.1%, P=0.001) (Table 4). Variance in membership group (SSBs consumption frequency >3 times/week) explained by each group of variables was: demographic characteristics, 0.2%; family characteristics, additional 8.8%; “Just want to drink SSBs”, additional 3.7%; “Deem SSBs not affect health”, additional 6.0%; "Drink SSBs as water”, additional 10.8%; and "Purchase in advance", additional 1.6%. The attitude and behavior factors contributed to 22.1% of the 31.1% variation that could be explained. In the final model, the four most influential factors were "Drink SSBs as water"(OR=10.288), “Deem SSBs not affect health” (OR=2.735), “Just want to drink SSBs” (OR=2.302) and "Purchase SSBs in advance" (OR=2.245). In addition, gender, household income per capita, father's education level and pocket money per month were also significant in final model (P < 0.05).
Table 3 The code of variables in binary logistic regression model
Variables
|
Points
|
|
|
Gender
|
Male=1, Female=2
|
|
Region
|
Urban=1, Rural=2
|
|
Family structure
|
Nuclear=1, Non-nuclear=2
|
|
Household income per capita
|
<¥1000=1
¥1000-5000=2
>¥5000=3
|
|
Father's education level
|
Primary school or illiteracy=1
Middle school=2
College and above=3
|
|
Mother's education level
|
Primary school or illiteracy=1
Middle school=2
College and above=3
|
|
Pocket money per month
|
None=1
<¥100=2
¥100-200=3
>¥200=4
|
|
Just want to drink SSBs
|
No=0, Yes=1
|
|
Deem SSBs affect health
|
No =0, Not sure=1, Yes =2
|
|
Drink SSBs as water
|
No =0, Yes =1
|
|
Purchase in advance
|
No =0, Yes =1
|
|
Table 4 Results of binary logistic regression analysis: risk factors versus SSBs consumption frequency
Model
|
Nagelkerke’s R2
|
Variables
|
OR:95%CIa
|
P-valuea
|
Total
|
Change
|
Model 1
|
0.002 (n.s.)
|
0.002 (n.s.)
|
Gender
|
0.581 (0.388-0.871)
|
0.009
|
|
|
|
Region
|
n.s.
|
|
Model 2
|
0.09*
|
0.088*
|
Family structure
|
n.s.
|
|
|
|
|
Household income per capita
|
2.227 (1.528-3.245)
|
<0.001
|
|
|
|
Father's education level
|
0.539 (0.327-0.887)
|
0.015
|
|
|
|
Mother's education level
|
n.s.
|
|
|
|
|
Pocket money per month
|
1.621 (1.308-2.011)
|
<0.001
|
Model 3
|
0.127*
|
0.037*
|
Just want to drink SSBs
|
2.302 (1.524-3.478)
|
<0.001
|
Model 4
|
0.187*
|
0.06*
|
Deem SSBs not affect health
|
2.735 (2.032-3.681)
|
<0.001
|
Model 5
|
0.295*
|
0.108*
|
Drink SSBs as water
|
10.288 (6.392-16.558)
|
<0.001
|
Model 6
|
0.311*
|
0.016*
|
Purchase SSBs in advance
|
2.245 (1.454-3.465)
|
<0.001
|
n.s.: not significant
a: OR and P value in final model
*: P<0.005
The relationship between SSBs consumption frequency and the risk scores for unhealthy attitude and behavior combinations
In order to observe the comprehensive effect of unhealthy attitudes and behaviors to SSBs consumption, we created a risk scores variable (See the methods section for details). As shown in Figure 2, as this risk score increased, the proportion of students with high SSBs consumption frequency (e.g. “≥1 times/day”) increased, while the proportion of that with low frequency (e.g. “<1 times/month”) decreased. The consumption frequency of SSBs was positively correlated with the risk score (the χ²-trend value was 127.470, P<0.001). The higher the risk score, the higher the consumption frequency.
Figure 2. Composition of SSBs consumption frequencies among junior school students with different risk scores
The unhealthy attitude or behavior factors and SSBs consumption in long term trend analysis 2012-2020
Using data from three surveys 2012, 2018 and 2020, we further analyzed the relationship between attitude or behavior factors and frequency of beverage consumption as well as long-term trends. Among junior school students, the consumption frequency decreased with the growth of the year (χ²-trend =144.153, P<0.001) under the same gender composition (χ²=1.096, P=0.578) (Table 5). The proportion of students who choose "Just want to drink SSBs" was different between 2012, 2018 and 2020 (χ²=24.513, P<0.001), but there was no linear trend (χ²=0.034, P=0.853). The proportion of students who “Drink SSBs as water” decreased with the increase of the year (χ²-trend=154.730, P<0.001). The proportion of students who "Deem SSBs not affect health” or “not sure” also reduced with the increase of the year (χ²-trend=145.847, P<0.001) (Table 6). We converted the frequency into times of monthly consumption to show the long term trend in the Figure 3. And the long term trend of the risk score combined three unhealthy attitude or behavior factors was also showed in Figure 3. Both SSBs consumption times (P<0.001) and risk score (P<0.001) decreased with the passage of time and there was significant positive correlation between them (r=0.314, P<0.001). (Fig. 3). The detailed description was shown in Table 7.
Table 5 Variation trend of SSBs consumption frequency among junior school students in 2012, 2018 and 2020, n (%)
|
n
|
≥1times /day
|
3-6times /week
|
1-2times /week
|
1-3times /month
|
<1times /month
|
χ²-trend
|
P-trend
|
2012
|
189
|
15 (7.9)
|
64 (33.9)
|
80 (42.3)
|
18 (9.5)
|
12 (6.3)
|
144.153
|
<0.001
|
2018
|
1338
|
86 (6.4)
|
193 (14.4)
|
458 (34.2)
|
357(26.7)
|
244 (18.2)
|
|
|
2020
|
1200
|
24 (2.0)
|
129 (10.8)
|
429 (35.8)
|
388 (32.3)
|
230 (19.2)
|
|
|
Table 6 Variation trend of three unhealthy attitudes and behaviors in 2012, 2018 and 2020, n (%)
|
|
|
2012
|
2018
|
2020
|
χ²
|
P-value
|
Just want to drink SSBs
|
Yes
|
111 (58.7)
|
548 (41.0)
|
560 (46.7)
|
24.51
|
0.003
|
|
No
|
78 (41.3)
|
790(59.0)
|
640 (53.3)
|
|
|
Drink SSBs as water
|
Yes
|
81 (42.9)
|
269 (20.1)
|
101 (8.4)
|
164.52
|
<0.001
|
|
No
|
108 (57.1)
|
1069 (79.9)
|
1099 (92.6)
|
|
|
Deem SSBs not affect health
|
No
|
30 (15.9)
|
77 (5.8)
|
51 (4.2)
|
409.98
|
<0.001
|
|
Not sure
|
115(60.8)
|
164(12.3)
|
126 (10.5)
|
|
|
|
Yes
|
44 (23.3)
|
1097 (82.0)
|
1023 (85.2)
|
|
|
Table 7 Variation trend of consumption times per month and risk scores in 2012, 2018 and 2020, Mean ± SD
|
|
2012
|
2018
|
2020
|
F
|
P-value
|
Consumption times per month
|
11.2±8.4
|
7.0±7.9*
|
4.7±5.5*#
|
85.163
|
<0.001
|
Risk scores
|
1.9±1.0
|
0.8±0.9*
|
0.7±0.8*#
|
179.335
|
<0.001
|
*: compared with 2012, P<0.05
#:compared with 2018, P<0.05