3.1 Cross-Sectional Relationship between Physical Activity and Depression
Following the methodology of prior research [28], we categorized standardized levels of physical activity (MET-min/week) into five quintiles (Q1-Q5). Table 2 displays the association between quintiles of physical activity and depression based on 2020 cross-sectional data, adjusted for various covariates. After adjusting for demographic characteristics, lifestyle factors, and chronic health conditions, we observed a significant reduction in the risk of depression among the lower quintile group (Q2) compared to the lowest quintile group (Q1), by 29% ([OR] 0.71, 95% Confidence Interval [CI] 0.56-0.90, P=0.004). However, no significant differences were found in other quintile groups, suggesting a potential nonlinear relationship between physical activity and depression risk.
To further analyze the dose-response relationship between physical activity and the risk of depression, we utilized a restrictive cubic spline model on top of Model 3. Our findings revealed a U-shaped dose-response relationship (Figure 3), with the lowest risk of depression observed at approximately 3023 MET-min/week of physical activity.
Table 2. Association between physical activity quintile level and depression
Physical activity
|
n
|
event (%)
|
OR For Depression (95%CI)
|
Model 1
|
Model 2
|
Model 3
|
Q1
|
734
|
276(37.60)
|
1.00 (reference)
|
1.00 (reference)
|
1.00 (reference)
|
Q2
|
728
|
197(27.06)
|
0.69 (0.55-0.87)*
|
0.69 (0.55-0.87)*
|
0.71 (0.56-0.90)*
|
Q3
|
730
|
229(31.27)
|
0.80 (0.64-1.00)
|
0.82 (0.65-1.03)
|
0.86 (0.69-1.09)
|
Q4
|
731
|
262(35.84)
|
0.97 (0.77-1.21)
|
0.99 (0.79-1.24)
|
1.02 (0.81-1.28)
|
Q5
|
731
|
256(35.02)
|
0.95 (0.76-1.19)
|
1.00 (0.79-1.25)
|
0.99 (0.79-1.25)
|
Note:*p<0.05, Q1-Q5 represent quintiles of physical activity (MET-min/week) from low to high.
3.2 Longitudinal Relationship between Trajectories of Physical Activity and Depression
In the cross-sectional analysis, we identified a U-shaped relationship between levels of physical activity and the risk of depression. However, directly constructing a group-based trajectory model (GBTM) might overlook potentially beneficial long-term trajectories of physical activity. Therefore, we divided participants into three groups based on the change point (3023 MET-min/week) in the dose-response relationship between MET and the risk of depression and constructed group-based trajectory models accordingly. Subsequently, we identified five distinct trajectory groups (Figure 4).
These trajectory groups are as follows: low-stable group (A, 11.5%), moderate-stable group (B, 14.6%), higher-decreasing group(C, 21.4%), highest-decreasing group (D, 41.1%), highest-stable group (E, 11.3%).
Figure 5 depicts the results of multivariable logistic regression analyses, adjusted for covariates (based on model 3), illustrating the independent associations between longitudinal trajectories of physical activity, residential location, and depressive symptoms. Specifically, compared to the low-stable group (A), the moderate-stable group (B) exhibited the lowest risk of depression, with a risk reduction of 47% (Odds Ratio [OR] 0.53, 95% Confidence Interval [CI] 0.39-0.71, P<0.001). Conversely, the highest-stable group (E) demonstrated the highest risk of depression, with a risk increase of 37% (OR 1.37, 95% CI 1.02-1.82, P=0.033). However, the differences in depression risk between Groups C and D and Group A were not statistically significant. This observation suggests that early reductions in physical activity levels may be beneficial for mental health (from group E to group D).
Due to the non-significant differences in depression risk among Groups A, C, and D, we combined them into one group and conducted subgroup analyses to ascertain whether trajectories of physical activity exerted protective or detrimental effects across different subpopulations. Analysis results (see Table 3) revealed that compared to the combined group of A, C, and D, Groups B and E exhibited no interaction with any subgroups. However, a gender interaction was identified in the comparison between Group B (protective) and Group E (risky). Subsequently, we performed further analyses stratified by gender (see Figure 6), wherein among males, Groups B, C, and D all demonstrated significant protective effects, while the harmful effect of Group E was non-significant. Conversely, among females, moderate-maintenance (Group B) effectively reduced the risk of depression, whereas high-maintenance (Group E) emerged as a significant risk factor.
Table 3. Association between different physical activity trajectories and risk of depression in subgroups
|
Event
|
group A,C,D VS group B
|
group A,C,D VS group E
|
group B VS group E
|
OR (95%CI)
|
P-int
|
OR (95%CI)
|
P-int
|
OR (95%CI)
|
P-int
|
Overall
|
1220
|
0.61(0.49-0.77)
|
|
1.23(1.00-1.51)
|
|
2.00(1.51-2.67)
|
|
Gender
|
|
|
0.204
|
|
0.081
|
|
0.023
|
Male
|
414 (33.9)
|
0.73(0.50-1.05)
|
|
1.11(0.81-1.50)
|
|
1.52(0.98-2.38)
|
|
Female
|
806 (66.1)
|
0.54(0.40-0.72)
|
|
1.62(1.21-2.17)
|
|
3.01(2.05-4.43)
|
|
Age
|
|
|
0.070
|
|
0.718
|
|
0.084
|
54-59
|
314 (25.7)
|
0.38(0.21-0.65)
|
|
1.41(0.99-2.00)
|
|
3.68(2.02-7.01)
|
|
60-69
|
583 (47.8)
|
0.58(0.41-0.80)
|
|
1.19(0.89-1.59)
|
|
2.06(1.37-3.15)
|
|
≥70
|
323 (26.5)
|
0.82(0.56-1.18)
|
|
1.15(0.66-1.95)
|
|
1.41(0.75-2.60)
|
|
Marital status
|
|
0.316
|
|
0.468
|
|
0.195
|
Single
|
353 (28.9)
|
0.74(0.48-1.12)
|
|
1.09(0.67-1.77)
|
|
1.48(0.81-2.72)
|
|
Married
|
867 (71.1)
|
0.57(0.43-0.74)
|
|
1.33(1.06-1.67)
|
|
2.34(1.68-3.26)
|
|
Drink status
|
|
0.134
|
|
0.562
|
|
0.569
|
No
|
918 (75.2)
|
0.69(0.53-0.89)
|
|
1.44(1.11-1.87)
|
|
2.10(1.49-2.97)
|
|
A little
|
80 (6.6)
|
0.56(0.22-1.25)
|
|
1.16(0.56-2.31)
|
|
2.08(0.76-6.13)
|
|
Yes
|
222 (18.2)
|
0.36(0.18-0.64)
|
|
1.13(0.75-1.67)
|
|
3.16(1.61-6.60)
|
|
Smoking status
|
|
0.485
|
|
0.235
|
|
0.134
|
Former
|
153 (12.5)
|
0.81(0.45-1.41)
|
|
1.17(0.66-2.01)
|
|
1.43(0.69-2.99)
|
|
Never
|
828 (67.9)
|
0.56(0.42-0.74)
|
|
1.50(1.14-1.97)
|
|
2.68(1.86-3.89)
|
|
Now
|
239 (19.6)
|
0.67(0.40-1.09)
|
|
1.00(0.66-1.48)
|
|
1.48(0.83-2.72)
|
|
Chronic disease comorbidity
|
|
0.211
|
|
0.891
|
|
0.637
|
0
|
688 (56.4)
|
0.61(0.45-0.82)
|
|
1.24(0.95-1.60)
|
|
2.02(1.40-2.93)
|
|
1 or 2
|
450 (36.9)
|
0.52(0.35-0.76)
|
|
1.25(0.87-1.80)
|
|
2.41(1.47-3.98)
|
|
≥3
|
82 (6.7)
|
1.27(0.51-3.31)
|
|
1.61(0.57-4.97)
|
|
1.27(0.34-4.86)
|
|
Note: OR, odds ratio; CI, confidence interval. All ORs were adjusted for gender, age, marital status, income, sleep duration, Nap duration, Drink status, smoking status, Chronic disease comorbidity and Epidemic impact on medical treatment. P-int (p for interaction) has calculated by the likelihood ratio test.