After the exclusions, 15,337 participants were left as the analytic sample. During the first follow-up, we identified 1516 cases of MAFLD, and the incidence of MAFLD in this study is 10.89%. As summarized in Table 1, the average age of all participants was 47.56 years, 54.92% (N = 8423) of participants were male, and the average BMI was 23.49 kg/m2. Participants were more likely to have a junior high school education, to be workers, to earn an average monthly income of ¥2,000-4,999, to never smoke, to never drink, and to perform frequent physical activity.
Table 1. Characteristics of the participants according to the presence of MAFLD.
Characteristics
|
Overall
(N=15,337)
|
MAFLD
(n=1,516)
|
Non-MAFLD
(n=13,821)
|
P value
|
Gender, n (%)
|
|
|
|
<0.001
|
Male
|
8423(54.92)
|
964(63.59)
|
7459(53.97)
|
|
Female
|
6914(45.08)
|
552(36.41)
|
6362(46.03)
|
|
Age, Mean±SD, years
|
47.56±7.87
|
49.72±8.14
|
47.32±7.80
|
<0.001
|
Educational attainment, n (%)
|
|
|
|
<0.001
|
Elementary school or below
|
1401(9.13)
|
180(11.87)
|
1221(8.83)
|
|
Junior high school
|
5082(33.14)
|
548(36.15)
|
4534(32.81)
|
|
Senior high school
|
4710(30.71)
|
464(30.61)
|
4246(30.72)
|
|
College or above
|
4144(27.02)
|
324(21.37)
|
3820(27.64)
|
|
Occupation, n (%)
|
|
|
|
<0.05
|
Worker
|
12318(80.32)
|
1217(80.28)
|
11101(80.32)
|
|
Servicer
|
516(3.36)
|
68(4.49)
|
448(3.25)
|
|
Technician
|
521(3.40)
|
56(3.70)
|
465(3.37)
|
|
Manager
|
1982(12.92)
|
175(11.55)
|
1807(13.08)
|
|
Average monthly income, n (%), Yuan
|
|
|
<0.001
|
<2000
|
5333(34.77)
|
455(30.01)
|
4878(35.29)
|
|
2000-4999
|
9575(62.43)
|
1010(66.62)
|
8565(61.97)
|
|
>4999
|
429(2.80)
|
51(3.36)
|
378(2.73)
|
|
Cigarette smoking, n (%)
|
|
|
|
<0.001
|
Never
|
8910(58.09)
|
788(51.98)
|
8122(58.77)
|
|
Current
|
5224(34.06)
|
568(37.47)
|
4656(33.69)
|
|
Former
|
1203(7.84)
|
160(10.55)
|
1043(7.55)
|
|
Alcohol drinking, n (%)
|
|
|
|
0.221
|
Never
|
11814(77.03)
|
1190(78.50)
|
10624(76.87)
|
|
Current
|
2893(18.86)
|
261(17.22)
|
2632(19.04)
|
|
Former
|
630(4.11)
|
65(4.29)
|
565(4.09)
|
|
Physical activity, n (%)
|
|
|
|
0.100
|
Seldom
|
1632(10.64)
|
148(9.76)
|
1484(10.74)
|
|
Moderate
|
5870(38.27)
|
617(40.70)
|
5253(38.01)
|
|
Frequent
|
7835(51.09)
|
751(49.54)
|
7084(51.26)
|
|
Fresh vegetable intake, n (%)
|
|
|
|
0.449
|
<2.5kg/week
|
3211(20.94)
|
306(20.18)
|
2905(21.02)
|
|
≥2.5kg/week
|
12126(79.06)
|
1210(79.82)
|
10916(78.98)
|
|
Fresh fruit intake, n (%)
|
|
|
|
<0.05
|
<2.5kg/week
|
5932(38.68)
|
627(41.36)
|
5305(38.38)
|
|
≥2.5kg/week
|
9405(61.32)
|
889(58.64)
|
8516(61.62)
|
|
Meat intake, n (%)
|
|
|
|
<0.05
|
≤350g/week
|
3648(23.79)
|
328(21.64)
|
3320(24.02)
|
|
>350g/week
|
11689(76.21)
|
1188(78.36)
|
10501(75.98)
|
|
Milk and dairy products, n (%)
|
|
|
|
<0.001
|
Never
|
1427(9.30)
|
199(13.13)
|
1228(8.89)
|
|
≤300ml/day
|
8401(54.78)
|
825(54.42)
|
7576(54.82)
|
|
>300ml/day
|
5509(35.92)
|
492(32.45)
|
5017(36.3)
|
|
Diabetes, n (%)
|
|
|
|
<0.001
|
No
|
840(5.48)
|
204(13.46)
|
636(4.61)
|
|
Yes
|
14497(94.52)
|
1312(86.55)
|
13185(95.4)
|
|
BMI, Mean ± SD, kg/m2
|
23.49±2.88
|
26.03±2.68
|
23.21±2.76
|
<0.001
|
Abbreviation: MAFLD, metabolic dysfunction-associated fatty liver disease
a Data presented as means (standard deviation) for continuous variables and numbers (percentages) for categorical variables. Examining differences between MAFLD and non-MAFLD based on t test (for continuous variables) and χ2 test (for categorical variables).
b Frequent physical activity is defined as exercising more than 3 times/week, and more than 30 minutes each time.
Table 2 displays the mean exposure to PM2.5 and its constituents over the course of the study period spanning from 2011 to 2015. The average concentrations of PM2.5, SO42-, NO3-, NH4+, OM, and BC over the five-year period were 35.57±11.47 μg/m3, 4.21±2.00μg/m3, 4.62±3.34 μg/m3, 3.37±2.00 μg/m3, 6.80±3.37 μg/m3 and 1.43±0.69 μg/m3, respectively.
Significant or moderate correlations were observed between various PM2.5 constituents and the total mass of PM2.5 (P < 0.001). (Pearson correlation coefficients ranging from 0.59 to 0.98, Fig. 1).
Table 2 Statistics of PM2.5 and its constituents during the study period from 2011 to 2015.
Pollutants
(μg/m3)
|
±s
|
Percentiles
|
IQR
|
|
Min
|
25th
|
50th
|
75th
|
Max
|
|
PM2.5
|
35.57±11.47
|
12.00
|
27.00
|
34.00
|
43.00
|
144.00
|
16.00
|
SO42-
|
4.21±2.00
|
0.35
|
2.75
|
4.02
|
5.40
|
17.83
|
2.65
|
NO3-
|
4.62±3.34
|
0.29
|
2.07
|
3.46
|
6.51
|
23.45
|
4.44
|
NH4+
|
3.37±2.00
|
0.29
|
1.84
|
2.92
|
4.43
|
18.17
|
2.59
|
OM
|
6.80±3.37
|
0.59
|
4.30
|
6.32
|
8.75
|
29.35
|
4.45
|
BC
|
1.43±0.69
|
0.13
|
0.93
|
1.34
|
1.82
|
7.31
|
0.89
|
Abbreviation: IQR, interquartile range; SD, standard deviation; PM2.5, particulate matter with an aerodynamic diameter less than or equal to 2.5 μm; OM, organic matter; BC, black carbon; SO42−, sulfate; NO3−, nitrate; NH4+, ammonium.
The associations between individual air pollutants and the risk of MAFLD are shown in Table 3. We found that all PM2.5 and its constituents were positively associated with the risk of MAFLD in the fully-adjusted model 3. Regarding the risk of incident MAFLD, the fully adjusted HRs for the highest quartiles compared with the lowest quartiles were 2.926 (95% CI: 1.223, 6.999), 2.856 (95% CI: 1.442, 5.659), 7.550 (95% CI: 3.385, 16.841), 4.832 (95% CI: 1.885, 12.383), 3.350 (95% CI: 1.348, 8.342), 1.894 (95% CI: 1.019, 1.624) for PM2.5, SO42-, NO3-, NH4+, OM, and BC, respectively. In the fully adjusted model 3, increased exposure levels to PM2.5 and its constituents were significantly associated with increased odds of MAFLD. The HRs of MAFLD were 1.490 (95% CI: 1.100, 2.017), 1.543 (95% CI: 1.216, 1.958), 1.949 (95% CI: 1.477, 2.572), 1.726 (95% CI: 1.239, 2.405), 1.528 (95% CI: 1.108, 2.105) and 1.311 (95% CI: 1.059, 1.624) per IQR increase in PM2.5, SO42-, NO3-, NH4+, OM, and BC, respectively.
We did not observe the significant nonlinear associations between the concentration of PM2.5 and its constituents and the incidence of MAFLD (both p-linear <0.001) (Fig. 2).
Table 3 Multivariable-adjusted HR (95% CI) for the associations of PM2.5 and its constituents in quintiles with risk of MAFLD.
|
Quartile 1
|
Quartile 2
|
Quartile 3
|
Quartile 4
|
Per IQR
|
P-trend
|
PM2.5
|
<35.01
|
35.01~36.37
|
36.37~38.91
|
>39.91
|
|
|
n (%)
|
446(11.63)
|
409(10.67)
|
318(8.29)
|
343(8.95)
|
|
|
Model 1
|
Ref.(1.00)
|
0.731(0.532,1.003)
|
1.125(0.625,2.028)
|
1.741(0.729,4.161)
|
1.194(0.873,1.634)
|
0.267
|
Model 2
|
Ref.(1.00)
|
0.810(0.590,1.114)
|
1.251(0.693,2.260)
|
2.151(0.898,5.154)
|
1.273(0.931,1.742)
|
0.131
|
Model 3
|
Ref.(1.00)
|
0.990(0.716,1.368)
|
1.842(1.004,3.381)
|
2.926(1.223,6.999)
|
1.490(1.100,2.017)
|
<0.05
|
SO42-
|
<4.15
|
4.15~4.31
|
4.31~4.61
|
>4.61
|
|
|
n (%)
|
447(11.65)
|
399(10.41)
|
328(8.55)
|
342(8.92)
|
|
|
Model 1
|
Ref.(1.00)
|
0.791(0.607,1.005)
|
1.137(0.716,1.806)
|
2.242(1.159,4.338)
|
1.311(1.032,1.667)
|
<0.05
|
Model 2
|
Ref.(1.00)
|
0.849(0.659,1.094)
|
1.213(0.763,1.930)
|
2.552(1.316,4.949)
|
1.369(1.077,1.739)
|
<0.05
|
Model 3
|
Ref.(1.00)
|
0.938(0.717,1.226)
|
1.370(0.830,2.262)
|
2.856(1.442,5.659)
|
1.543(1.216,1.958)
|
<0.001
|
NO3-
|
<4.55
|
4.55~4.82
|
4.82~5.25
|
>5.25
|
|
|
n (%)
|
385(10.03)
|
424(11.07)
|
381(9.94)
|
326(8.50)
|
|
|
Model 1
|
Ref.(1.00)
|
0.814(0.636,1.041)
|
1.273(0.811,1.998)
|
2.965(1.589,5.531)
|
1.529(1.219,1.917)
|
<0.001
|
Model 2
|
Ref.(1.00)
|
0.892(0.696,1.143)
|
1.430(0.908,2.253)
|
3.440(1.836,6.445)
|
1.609(1.283,2.017)
|
<0.001
|
Model 3
|
Ref.(1.00)
|
1.160(0.865,1.554)
|
2.424(1.389,4.231)
|
7.550(3.385,16.841)
|
1.949(1.477,2.572)
|
<0.001
|
NH4+
|
<3.31
|
3.31~3.47
|
3.47~3.73
|
>3.73
|
|
|
n (%)
|
430(11.21)
|
400(10.43)
|
340(8.87)
|
346(9.02)
|
|
|
Model 1
|
Ref.(1.00)
|
0.755(0.553,1.030)
|
1.338(0.741,2.415)
|
2.554(1.089,5.988)
|
1.373(1.008,1.871)
|
<0.05
|
Model 2
|
Ref.(1.00)
|
0.819(0.599,1.120)
|
1.473(0.813,2.667)
|
2.931(1.243,6.912)
|
1.439(1.055,1.962)
|
<0.05
|
Model 3
|
Ref.(1.00)
|
1.020(0.725,1.434)
|
2.085(1.096,3.964)
|
4.832(1.885,12.383)
|
1.726(1.239,2.405)
|
<0.001
|
OM
|
<6.71
|
6.71~7.09
|
7.09~7.54
|
>7.54
|
|
|
n (%)
|
424(11.05)
|
418(10.91)
|
316(8.23)
|
358(9.35)
|
|
|
Model 1
|
Ref.(1.00)
|
0.735(0.539,1.002)
|
1.532(0.848,2.768)
|
2.589(1.091,6.144)
|
1.293(0.845,1.769)
|
0.108
|
Model 2
|
Ref.(1.00)
|
0.807(0.590,1.103)
|
1.691(0.933,3.062)
|
3.033(1.272,7.232)
|
1.370(1.002,1.874)
|
<0.05
|
Model 3
|
Ref.(1.00)
|
0.888(0.637,1.237)
|
1.894(1.010,3.554)
|
3.350(1.348,8.324)
|
1.528(1.108,2.105)
|
<0.05
|
BC
|
<1.41
|
1.41~1.45
|
1.45~1.60
|
>1.60
|
|
|
n (%)
|
507(13.22)
|
309(8.06)
|
365(9.54)
|
335(8.72)
|
|
|
Model 1
|
Ref.(1.00)
|
0.965(0.752,1.238)
|
0.662(0.429,1.022)
|
1.256(0.680,2.323)
|
1.083(0.871,1.346)
|
0.472
|
Model 2
|
Ref.(1.00)
|
0.992(0.773,1.273)
|
0.702(0.454,1.084)
|
1.419(0.766,2.628)
|
1.127(0.906,1.402)
|
0.282
|
Model 3
|
Ref.(1.00)
|
1.204(0.934,1.552)
|
0.902(0.575,1.415)
|
1.894(1.019,3.523)
|
1.311(1.059,1.624)
|
<0.05
|
a Model 1 was adjusted for age and gender.
b Model 2 was adjusted for model 1 plus educational attainment, occupation, average monthly income, fresh vegetable intake, fresh fruit intake, cigarette smoking status, alcohol drinking status, physical activity.
c Model 3 was adjusted for model 2 temperature and relative humidity.
Fig. 3 showed the adjusted HRs and 95%CIs of MAFLD associated with each IQR increase in PM2.5 and its constituents by demographic characteristics and lifestyle factors. Stratified analyses suggested that the associations between exposure to PM2.5 and its constituents and MAFLD were modified by sex, occupation, educational attainment, alcohol drinking status, and physical activity. For example, the HRs for the incidence of MAFLD associated with each IQR increase in PM2.5 were 0.894 (95% CI: 0.619, 1.296) and 3.268 (95% CI: 1.884, 5.671) for males and females; 1.458 (95% CI: 1.024, 2.077) and 2.214 (95% CI: 1.142, 4.293) for workers and other occupations; 1.651 (95% CI: 1.026, 2.657) and 1.419 (95% CI: 0.937, 2.150) for the lower and higher educational groups, respectively; 1.606 (95% CI: 0.481, 5.367), 0.982 (95% CI: 0.619, 1.557) and 2.064 (95% CI: 1.324, 3.218) for the seldom moderate and frequent physical activity groups, respectively; 1.941 (95% CI: 0.941, 4.114), 0.549 (95% CI: 0.103, 2.938), and 1.511 (95% CI: 1.072, 2.130) for the current, former and never drinkers, respectively. Similar effect modifications were also observed for SO42-, NO3-, NH4+, OM, and BC.
We used Quantile g-calculation to explore the relationship between mixed exposure of SO42-, NO3-, NH4+, OM, BC and MAFLD, and the Fig. 4 showed that the risk of MAFLD increased by 33.3% for each quartile increase of mixed exposure (95% CI: 1.031, 1.724, P < 0.05).
In the mixed exposure of SO42-, NO3-, NH4+, OM and BC, the effect weight of NO3- and NH4+ on MAFLD was positive, with a total positive weight coefficient of 0.574. The weight of NO3- was 75.1%, and the weight of NH4+ was 24.9%. The effect weights of SO42-, OM and BC on MAFLD was negative, with a total negative weight coefficient of -1.720. The weight of BC, SO42-, and OM were 79.5%, 12.6% and 7.9%, respectively (Table 4 and Fig. 5). The weight of BC is the highest in the mixed exposure (79.5%).
Table 4. HR (95% CI) for incidence of MAFLD associated with AP in multi-pollutant models
Pollutants
|
weight
|
Coefficient β
|
Effect of Mixtures ψ (log HR)
|
95%CI
|
HR(95%CI)
|
P
|
BC
|
0.795
|
-1.367
|
-0.945
|
(-1.196,-0.695)
|
0.389(0.302,0.499)
|
<0.001
|
SO42-
|
0.126
|
-0.217
|
OM
|
0.079
|
-0.136
|
NO3-
|
0.751
|
0.431
|
NH4+
|
0.249
|
0.143
|