winter December – February was not significantly dif- ferent from those in other seasons, we did not take
season into account as a covariate.
3. Results
In our study population, the geometric mean of plasma fibrinogen was 2.48 gl 95 CI; 2.47 2.49,
and smoking rate was 57.5. Table 1 shows age-adjusted plasma fibrinogen levels by employment grade; the lower
employment grades had higher plasma fibrinogen level P B 0.001. The difference between the highest and the
lowest grade was about 0.23 units or about 10. Differ- ences in plasma fibrinogen levels by educational back-
ground were similar to those by employment grade Table 2; participants with lower educational back-
grounds had higher plasma fibrinogen levels P B 0.001.
Table 1 Age-adjusted plasma fibrinogen levels according to employment grade
a
Employment grade Total
1 High 2
3 4
5 6
7 Low 690
542 4375
N 416
206 833
904 784
2.56 2.58
2.48 2.41
2.52 Mean fibrinogen gl
2.55 2.35
2.55
a
Trend test: PB0.001. Table 2
Age-adjusted plasma fibrinogen levels according to educational background
a
Educational background Total
IV low III
II I high
240 2548
691 4374
N 895
Mean fibrinogen gl 2.41
2.51 2.48
2.49 2.50
a
Educational background: more than 15 years = I, 13–15 years = II, 11–12 years = III and less than 11 years = IV. Trend test: PB0.001.
Table 3 The relationship between explanatory variables and age-adjusted plasma fibrinogen level
a
Standardised linear Linear regression estimate
Standardised odds N
Variables Odds ratios
d
95 CI ratios
c
95 CI regression
b
95 CI estimate
c
95 CI 4375
1.13 1.071.18 Employment
1.15 0.751.56 grade
1.44 0.862.02 4374
1.15 1.061.23 Educational
background 1.48 0.952.02
0.52 0.330.70 4374
Body mass index 1.16 1.081.23
1.05 1.031.08 kgm
2
1.35 1.201.49 1.87 1.292.46
3.41 2.354.49 4191
1.18 1.101.26 Waist to hip
ratio×10 0.92 0.841.00
4374 Height cm
− 0.08 −0.180.01
0.99 0.971.00 −
0.52 −1.080.04 Smoking
4312 2.33 1.832.83
1.31 1.231.38 0.13 −0.660.41
Alcohol 1.00 0.951.05
4299 1.00 0.931.08
− 0.08 −0.440.28
gweek100 0.96 0.881.05
Exercise 4313
0.10 −0.490.69 1.07 1.001.14
1.01 1.001.01 0.84 0.311.37
0.06 0.020.10 Systolic blood
4374 pressure
mmHg White blood cell
2.90 2.613.20 5.19 4.665.72
1.33 1.291.37 1.66 1.581.73
4374 10
9
l 4374
6.27 5.217.35 Hemoglobin A1c
1.35 1.271.43 1.75 1.611.89
3.31 2.763.87
a
Smoking: no = 1, ex = 2, current520day = 3, current\20day = 4; exercise: no = 1, light = 2, briskB3w = 3, brisk\2w = 4.
b
Per cent change in plasma fibrinogen level per unit change in explanatory variable.
c
The standardised estimates show the effects on fibrinogen of changing the explanatory variables by one standard deviation.
d
Odds ratio of having plasma fibrinogen level above the fifth quintile of plasma fibrinogen 2.85 gl.
Table 4 The relationship between employment gradeeducational background and plasma fibrinogen level while adjusting for age and other explanatory
variables Adjusted for variables
a
N Linear regression estimate of employment grade
Odds ratios of employment grade 95 CI
95 CI A
a
Age
c
4375 1.15 0.751.56
1.13 1.071.18 0.85 0.351.35
4374 1.11 1.041.17
Educational background 4374
Body mass index 1.24 0.831.65
1.14 1.081.19 4191
Waist to hip ratio 1.35 0.931.77
1.16 1.101.21 1.12 0.711.53
4374 1.12 1.061.17
Height 4312
Smoking 1.24 0.831.64
1.15 1.091.20 1.24 0.821.65
4299 1.14 1.091.19
Alcohol 4313
Exercise 1.21 0.801.63
1.14 1.081.19 4374
Systolic blood pressure 1.11 0.701.52
1.12 1.071.18 0.78 0.391.17
4374 1.09 1.031.14
White blood cell 1.03 0.631.43
1.12 1.061.17 Hemoglobin A1c
4374 4178
0.96 0.551.37 1.13 1.071.19
All variables except for educational background
4177 All variables
0.85 0.351.34 1.12 1.051.19
B
b
1.44 0.862.02 4374
1.15 1.061.23 Age
c
0.73 0.031.44 Employment grade
1.05 0.951.15 4374
1.49 0.912.07 4373
1.15 1.071.24 Body mass index
4190 Waist to hip ratio
1.66 1.062.25 1.20 1.111.29
1.40 0.821.98 Height
1.14 1.051.22 4373
1.25 0.671.84 4311
1.14 1.051.23 Smoking
4298 Alcohol
1.61 1.022.20 1.18 1.091.27
4312 Exercise
1.56 0.972.15 1.18 1.091.26
1.39 0.811.97 4373
1.14 1.061.23 Systolic blood pressure
4373 White blood cell
0.87 0.311.42 1.08 1.001.17
4373 Hemoglobin A1c
1.28 0.711.86 1.13 1.051.22
4177 0.95 0.381.53
1.13 1.031.22 All variables except for employment
grade All variables
0.27 −0.420.97 4177
1.02 0.911.13
a
The relationship between employment grade and plasma fibrinogen level while adjusting for age and other explanatory variables.
b
The relationship between educational background and plasma fibrinogen level while adjusting for age and other explanatory variables.
c
The first row of the table shows the age-adjusted effects of educational background on plasma fibrinogen level this result is also shown in Table 3. The following rows show the same effects while adjusting for age and the particular explanatory variable of interest.
In Table 3, we show the relationship between plasma fibrinogen level and its determinants from linear regres-
sion models. The result again demonstrates that em- ployment grade and educational background were
related to plasma fibrinogen level P B 0.001. Each step down the employment grade ladder was associated
with a 1.15 increase in plasma fibrinogen level, and the corresponding figure was 1.44 for educational
background. Body mass index BMI; kgm
2
, waist to hip ratio, smoking, systolic blood pressure, WBC and
HbA1c were positively and significantly associated with plasma fibrinogen level P B 0.001. Height was nega-
tively associated with plasma fibrinogen level, but the effect was small and bordering on significant P = 0.07.
There were no significant associations between plasma fibrinogen level and alcohol consumption P = 0.65,
and plasma fibrinogen level and physical activity at leisure P = 0.75. In logistic regression analyses, the
relationships between all the variables and plasma fibrinogen level were similar to those obtained in the
linear regression analyses. The odds ratio was 1.13 per unit of employment grade and 1.15 per unit of educa-
tional background, respectively.
Table 4A shows the relationship between employ- ment grade and plasma fibrinogen level while adjusting
for age, educational background, BMI, waist to hip ratio, height, smoking, alcohol, exercise, systolic blood
pressure, WBC, HbA1c, all these variables except for educational background, and all variables together.
Employment grade was independently associated with plasma fibrinogen level after adjusting for each variable
individually, and remained significant after adjusting for all the variables together P B 0.001. However, the
regression coefficient declined on adjusting for educa- tion background, WBC and HbA1c. Similar results
were shown in logistic regression analyses.
The relationship between educational background and plasma fibrinogen level while adjusting for age and
all other variables individually, as well as all variables together is shown Table 4B. Although educational back-
ground was independently associated with plasma fibrinogen level after adjusting for each variable, the
regression coefficients were reduced after adjusting for employment grade and WBC. Finally, the association no
longer remained significant when all variables were added to the model P = 0.44. In logistic regression
analyses, the odds ratio for educational background was not significant after adjusting for employment grade
P = 0.37, and was borderline significant after adjusting for WBC P = 0.07.
4. Discussion