3.1 Sample characteristics
Table 1 shows the association between demographic characteristics and sleep problem in preschoolers. In our survey, approximately 15.3% of the preschoolers reported CSHQ total score > 54, and children with low age group, poor economic status of family, low age group of parents, highly educated parents, night noise, only child, extensive screen time have more sleep problems (P < 0.001 or 0.05). In addition, emotional and behavioral symptoms (SDQ is abnormal) and parents’ higher level hostile/coercive are also more likely to have sleep problems (P < 0.001).
Variables
|
Total ( n = 27200 )
|
|
CSHQ total score ≦ 54 (n = 23045)
|
|
CSHQ total score
> 54(n = 4155)
|
X2
|
P-value
|
Table 1
Characteristics of participants by sleep problems, data shown as n (%)
Gender
|
|
|
|
|
|
0.45
|
0.511
|
Boy
|
13975(51.4)
|
|
11860(84.9)
|
|
2115(15.1)
|
|
|
Girl
|
13225(48.6)
|
|
11185(84.6)
|
|
2040(15.4)
|
|
|
Age(years)
|
|
|
|
|
|
58.90
|
< 0.001
|
3-
|
5949(21.9)
|
|
4900(82.4)
|
|
1049(17.6)
|
|
|
4-
|
8724(32.1)
|
|
7342(84.2)
|
|
1382(15.8)
|
|
|
5-
|
9128(33.6)
|
|
7818(85.6)
|
|
1310(14.4)
|
|
|
6-
|
3399(12.5)
|
|
2985(87.8)
|
|
414(12.2)
|
|
|
Urban/rural
|
|
|
|
|
|
1.76
|
0.195
|
Rural
|
2566(9.4)
|
|
2197(85.6)
|
|
369(14.4)
|
|
|
Urban
|
24634(90.6)
|
|
20848(84.6)
|
|
3786(15.4)
|
|
|
Economic status of family
|
|
|
|
|
|
9.91
|
0.007
|
Good
|
4033(14.8)
|
|
3469(86.0)
|
|
564(14.0)
|
|
|
Moderate
|
21825(80.2)
|
|
18467(84.6)
|
|
3358(15.4)
|
|
|
Poor
|
1342(4.9)
|
|
1109(82.6)
|
|
233(17.4)
|
|
|
Mother’s age
|
|
|
|
|
|
47.48
|
< 0.001
|
< 29
|
6795(25.0)
|
|
5629(82.8)
|
|
1166(17.2)
|
|
|
29–34
|
13372(49.2)
|
|
11297(84.5)
|
|
2075(15.5)
|
|
|
> 34
|
7033(25.9)
|
|
6119(87.0)
|
|
914(13.0)
|
|
|
Father’s age
|
|
|
|
|
|
36.03
|
< 0.001
|
< 31
|
5855(21.5)
|
|
4824(82.4)
|
|
1031(17.6)
|
|
|
31–36
|
13932(51.2)
|
|
11839(85.0)
|
|
2093(15.0)
|
|
|
> 36
|
7413(27.3)
|
|
6382(86.1)
|
|
1031(13.9)
|
|
|
Mother's education level
|
|
|
|
|
|
28.41
|
< 0.001
|
Junior high school and below
|
3405(12.5)
|
|
2955(86.8)
|
|
450(13.2)
|
|
|
Senior middle school
|
7672(28.2)
|
|
6572(85.7)
|
|
1100(14.3)
|
|
|
Junior college
|
5354(19.7)
|
|
4522(84.5)
|
|
832(15.5)
|
|
|
Undergraduate or more
|
10769(39.6)
|
|
8996(83.5)
|
|
1773(16.5)
|
|
|
Father's education level
|
|
|
|
|
|
16.44
|
0.001
|
Junior high school and below
|
2775(10.2)
|
|
2347(84.6)
|
|
428(15.4)
|
|
|
Senior middle school
|
7445(27.4)
|
|
6379(85.7)
|
|
1066(14.3)
|
|
|
Junior college
|
5364(19.7)
|
|
4589(85.6)
|
|
775(14.4)
|
|
|
Undergraduate or more
|
11616(42.7)
|
|
9730(83.8)
|
|
1886(16.2)
|
|
|
Night noise
|
|
|
|
|
|
34.95
|
< 0.001
|
Yes
|
9759(35.9)
|
|
8100(83.0)
|
|
1659(17.0)
|
|
|
No
|
17441(64.1)
|
|
14945(85.7)
|
|
2496(14.3)
|
|
|
BMI
|
|
|
|
|
|
0.59
|
0.745
|
Normal
|
22799(83.8)
|
|
19304(84.7)
|
|
3495(15.3)
|
|
|
Overweight
|
3049(11.2)
|
|
2586(84.8)
|
|
463(15.2)
|
|
|
Obesity
|
1352(5.0)
|
|
1155(85.4)
|
|
197(14.6)
|
|
|
Only child
|
|
|
|
|
|
21.09
|
< 0.001
|
Yes
|
17916(65.9)
|
|
15050(84.0)
|
|
2866(16.0)
|
|
|
No
|
9284(34.1)
|
|
7995(86.1)
|
|
1289(13.9)
|
|
|
Screen time
|
|
|
|
|
|
88.49
|
< 0.001
|
< 2h
|
18059(66.4)
|
|
15564(86.2)
|
|
2495(13.8)
|
|
|
≥ 2h
|
9141(33.6)
|
|
7481(81.8)
|
|
1660(18.2)
|
|
|
Outdoor activity
|
|
|
|
|
|
0.68
|
0.414
|
< 2h
|
17873(65.7)
|
|
15166(84.9)
|
|
2707(15.1)
|
|
|
≥ 2h
|
9327(34.3)
|
|
7879(84.5)
|
|
1448(15.5)
|
|
|
Exclusive breastfeeding
|
|
|
|
|
|
1.44
|
0.488
|
0
|
10830(39.8)
|
|
9141(39.7)
|
|
1689(15.6)
|
|
|
1–6 months
|
14781(54.4)
|
|
12553(84.9)
|
|
2228(15.1)
|
|
|
> 6 months
|
1583(5.8)
|
|
1346(85.0)
|
|
237(15.0)
|
|
|
SDQ total score
|
|
|
|
|
|
1066.21
|
< 0.001
|
Normal
|
20692(76.1)
|
|
18320(88.5)
|
|
2372(11.5)
|
|
|
Suspicious
|
3450(12.7)
|
|
2646(76.7)
|
|
804(23.3)
|
|
|
Abnormal
|
3058(11.2)
|
|
2079(68.0)
|
|
979(32.0)
|
|
|
Maternal supportive/engaged
|
|
|
|
|
|
5.88
|
0.053
|
≤ P25
|
6491(23.9)
|
|
5440(83.8)
|
|
1051(16.2)
|
|
|
P25-P75
|
13633(50.1)
|
|
11575(84.9)
|
|
2058(15.1)
|
|
|
> P75
|
7076(26.0)
|
|
6030(85.2)
|
|
1046(14.8)
|
|
|
Maternal hostile/coercive
|
|
|
|
|
|
493.39
|
< 0.001
|
≤ P25
|
5591(20.6)
|
|
5094(91.1)
|
|
497(8.9)
|
|
|
P25-P75
|
14456(53.1)
|
|
12419(85.9)
|
|
2037(14.1)
|
|
|
> P75
|
7153(26.3)
|
|
5532(77.3)
|
|
1621(22.7)
|
|
|
Paternal supportive/engaged
|
|
|
|
|
|
6.47
|
0.039
|
≤ P25
|
6245(23.0)
|
|
5230(83.7)
|
|
1015(16.3)
|
|
|
P25-P75
|
13665(50.2)
|
|
11600(84.9)
|
|
2065(15.1)
|
|
|
> P75
|
7290(26.8)
|
|
6215(85.3)
|
|
1075(14.7)
|
|
|
Paternal hostile/coercive
|
|
|
|
|
|
452.23
|
< 0.001
|
≤ P25
|
5792(21.3)
|
|
5214(22.6)
|
|
578(10.0)
|
|
|
P25-P75
|
13868(51.0)
|
|
11981(86.4)
|
|
1887(13.6)
|
|
|
> P75
|
7540(27.7)
|
|
5850(77.6)
|
|
1690(22.4)
|
|
|
3.2 Association of parenting behaviors with sleep problems and emotional and behavior symptoms in preschool children
Parenting behaviors takes the low-level group of each dimension as the reference group. In all model, the ORs for the sleep problem increased across the maternal hostile/coercive score and paternal hostile/coercive score.
The adjusted associations between sleep problems (CSHQ total score > 54) and parenting behaviors and emotional symptoms are shown in Table 2. The crude model ORs (95%CI) for sleep problems across the maternal hostile/coercive were 1.00, 1.46 (1.31, 1.63) and 2.21 (1.96, 2.50) (P for trend < 0.001), respectively. After adjusting for confounding factors, the ORs (95%CI) for sleep problems across the maternal hostile/coercive were 1.00, 1.38 (1.23, 1.54) and 1.86 (1.64, 2.10), respectively, suggesting a significant dose-response relationship (P for trend < 0.001). The crude model ORs (95%CI) for sleep problems across the paternal hostile/coercive were 1.00, 1.21 (1.09, 1.84) and 1.83 (1.64, 2.06) (P for trend < 0.001), respectively. After adjusting for multiple confounding factors, the ORs (95%CI) for sleep problems across the paternal hostile/coercive were 1.00, 1.17 (1.05, 1.30) and 1.59(1.42, 1.79), respectively, suggesting a significant dose-response relationship (P for trend < 0.001). The crude model ORs (95%CI) for sleep problems across the SDQ scores were 1.00, 2.35 (2.15, 2.57) and 3.64 (3.33, 3.97) (P for trend < 0.001), respectively. After adjusting for confounding factors, the ORs (95%CI) for sleep problems across the SDQ scores were 1.00, 2.30 (2.10, 2.52) and 3.57 (3.27, 3.90), respectively, suggesting a significant dose-response relationship (P for trend < 0.001).
Table 2. Odds ratio of sleep problems by level of parenting behaviors and emotional symptoms
Variables
|
Model 1a
|
|
Model 2b
|
Odds ratio
|
95%CI
|
P-value
|
|
Odds ratio
|
95%CI
|
P-value
|
Maternal Supportive/engaged
|
|
|
|
|
|
|
|
≤P25
|
Ref
|
|
|
|
Ref
|
|
|
P25-P75
|
0.94
|
0.86-1.02
|
0.141
|
|
1.01
|
0.93-1.11
|
0.779
|
>P75
|
0.97
|
0.87-1.07
|
0.513
|
|
1.06
|
0.95-1.19
|
0.281
|
P-value for trend
|
0.024
|
|
|
|
0.001
|
|
|
Maternal Hostile/coercive
|
|
|
|
|
|
|
|
≤P25
|
Ref
|
|
|
|
Ref
|
|
|
P25-P75
|
1.46
|
1.31-1.63
|
< 0.001
|
|
1.38
|
1.23-1.54
|
< 0.001
|
>P75
|
2.21
|
1.96-2.50
|
< 0.001
|
|
1.86
|
1.64-2.10
|
< 0.001
|
P-value for trend
|
< 0.001
|
|
|
|
< 0.001
|
|
|
Paternal Supportive/engaged
|
|
|
|
|
|
|
|
≤P25
|
Ref
|
|
|
|
Ref
|
|
|
P25-P75
|
0.90
|
0.82-0.98
|
0.012
|
|
0.93
|
0.85-1.02
|
0.101
|
>P75
|
0.90
|
0.81-1.00
|
0.055
|
|
0.97
|
0.87-1.08
|
0.564
|
P-value for trend
|
0.017
|
|
|
|
0.001
|
|
|
Paternal Hostile/coercive
|
|
|
|
|
|
|
|
≤P25
|
Ref
|
|
|
|
Ref
|
|
|
P25-P75
|
1.21
|
1.09-1.34
|
< 0.001
|
|
1.17
|
1.05-1.30
|
0.004
|
>P75
|
1.83
|
1.64-2.06
|
< 0.001
|
|
1.59
|
1.42-1.79
|
< 0.001
|
P-value for trend
|
< 0.001
|
|
|
|
< 0.001
|
|
|
SDQ total score
|
|
|
|
|
|
|
|
Normal
|
Ref
|
|
|
|
Ref
|
|
|
Suspicious
|
2.35
|
2.15-2.57
|
< 0.001
|
|
2.30
|
2.10-2.52
|
< 0.001
|
Abnormal
|
3.64
|
3.33-3.97
|
< 0.001
|
|
3.57
|
3.27-3.90
|
< 0.001
|
P-value for trend
|
< 0.001
|
|
|
|
< 0.001
|
|
|
Note.Sleep problems are equivalent to CSHQ total scores > 54.
aUnadjusted model.
bAdjusted for gender, age, residence, economic status of family, parents' age, parents' education level, night noise, BMI, only child, screen time, outdoor activity, exclusive breastfeeding.
|
3.3 Association of parenting behaviors with emotional and behaviors symptom in preschool children
Parenting behaviors takes the low-level group of each dimension as the reference group, SDQ scores takes the normal group of each dimension as the reference group. Table 3 shows parenting behavior and childrens’ emotional and behavior symptoms are significantly related (P < 0.001). The model ORs (95%CI) for SDQ scores across the maternal supportive/engaged were 1.00, 0.65 (0.45, 0.57) and 0.51 (0.59, 0.71) (P < 0.001), respectively. The model ORs (95%CI) for SDQ scores across the maternal hostile/coercive were 1.00, 1.64 (1.45, 1.86) and 2.68 (2.34, 3.06) (P < 0.001), respectively. Paternal result is the same as maternal.
Table 3
Odds ratio of SDQ score by level of parenting behaviors
|
Suspicious
|
|
Abnormal
|
Odds ratio
|
95%CI
|
P-value
|
|
Odds ratio
|
95%CI
|
P-value
|
Maternal Supportive/engaged
|
|
|
|
|
|
|
|
≤P25
|
1.00
|
|
|
|
1.00
|
|
|
P25-P75
|
0.65
|
0.45–0.57
|
< 0.001
|
|
0.54
|
0.49–0.59
|
< 0.001
|
>P75
|
0.51
|
0.59–0.71
|
< 0.001
|
|
0.46
|
0.41–0.53
|
< 0.001
|
Maternal hostile/coercive
|
|
|
|
|
|
|
|
≤P25
|
1.00
|
|
|
|
1.00
|
|
|
P25-P75
|
1.64
|
1.45–1.86
|
< 0.001
|
|
1.52
|
1.33–1.74
|
< 0.001
|
>P75
|
2.68
|
2.34–3.06
|
< 0.001
|
|
3.06
|
2.65–3.53
|
< 0.001
|
Paternal supportive/engaged
|
|
|
|
|
|
|
|
≤P25
|
1.00
|
|
|
|
1.00
|
|
|
P25-P75
|
0.79
|
0.72–0.86
|
< 0.001
|
|
0.67
|
0.61–0.73
|
< 0.001
|
>P75
|
0.56
|
0.50–0.64
|
< 0.001
|
|
0.50
|
0.44–0.57
|
< 0.001
|
Paternal hostile/coercive
|
|
|
|
|
|
|
|
≤P25
|
1.00
|
|
|
|
1.00
|
|
|
P25-P75
|
1.25
|
1.12–1.41
|
< 0.001
|
|
1.29
|
1.14–1.47
|
< 0.001
|
>P75
|
1.93
|
1.70–2.19
|
< 0.001
|
|
2.33
|
2.04–2.67
|
< 0.001
|
3.4 The relationship between parenting behaviors and children sleep mediated by emotional and behaviors symptoms
In order to further understand the role of children emotional behavior between the parenting behavior and children sleep problem, we performed a mediating effect analysis. In this model, SDQ scores were used as mediating variables, parenting behavior as independent variables and CSHQ score as dependent variables.
Figure 1 shows the model of the relationship between maternal hostile/coercive and sleep problems mediated by SDQ total score. The mediation model was significant (F = 1027.26, p < 0.001), accounting for 16% [0.024/(0.125 + 0.024)] of the mediating effect, where the total effect is equal to the direct effect (0.125) plus the indirect effect (0.024). Both the direct effect of maternal hostile/coercive on SDQ (a = 0.200, SE = 0.004, p < 0.001) and the SDQ on CSHQ (b = 0.118, SE = 0.006, p < 0.001) were significant. Moreover, the direct effect of maternal hostile/coercive on CSHQ (c’ = 0.125, SE = 0.005, p < 0.001) was also significant. Finally, the bootstrapping results proved the significant effect of maternal hostile/coercive on CSHQ score through SDQ score [B = 0.024, bias-corrected and accelerated 5,000 bootstrapping (BCa) 95%CI = 0.020 ~ 0.027] (Table 4). The full model explains 11% of the total variance of maternal hostile/coercive on CSHQ score. These results suggest that emotional and behavioral symptoms partially mediates the relationship between maternal hostile/coercive and sleep problems.
Table 4. Testing of bootstrap the mediation effect of maternal hostile/coercive on CSHQ scores
Effect type
|
B
|
Boot SE
|
Boot 95%CI
|
Relative effect
|
LLCI
|
ULCI
|
Total effect
|
0.1485
|
0.005
|
0.1394
|
0.1576
|
|
Direct effect
|
0.1250
|
0.005
|
0.1156
|
0.1343
|
84.11%
|
Indirect effect
|
0.0236
|
0.002
|
0.0204
|
0.0271
|
15.89%
|
Figure 2 shows the model of the relationship between paternal hostile/coercive and sleep problems mediated by SDQ score. The mediation model was significant (F = 827.87, p < 0.001), accounting for 17% [0.024/(0.118 + 0.024)] of the mediating effect, where the total effect is equal to the direct effect (0.118) plus the indirect effect (0.024). Both the direct effect of paternal hostile/coercive on SDQ (a = 0.187, SE = 0.005, p < 0.001) and the SDQ on CSHQ (b = 0.127, SE = 0.006, p < 0.001) were significant. Moreover, the direct effect of paternal hostile/coercive on CSHQ (c’ = 0.118, SE = 0.005, p < 0.001) was also significant. Finally, the bootstrapping results proved the significant effect of paternal hostile/coercive on CSHQ score through SDQ score [B = 0.024, bias-corrected and accelerated 5,000 bootstrapping (BCa) 95%CI = 0.021 ~ 0.027] ( Table 5 ). The full model explains 12% of the total variance of paternal hostile/coercive on CSHQ score. These results suggest that emotional and behavioral symptoms partially mediates the relationship between paternal hostile/coercive and sleep problems.
Table 5. Testing of bootstrap the mediation effect of paternal hostile/coercive on CSHQ score
Effect type
|
B
|
Boot SE
|
Boot 95%CI
|
Effect ratio
|
LLCI
|
ULCI
|
Total effect
|
0.1422
|
0.005
|
0.1325
|
0.1519
|
|
Direct effect
|
0.1184
|
0.005
|
0.1086
|
0.1283
|
83.10%
|
Indirect effect
|
0.0238
|
0.002
|
0.0206
|
0.0269
|
16.90%
|