16542 elderly people were joining the health screening in MJ Health Screening Centers between 2011 and 2012. After excluding people not fulfilling the inclusion criteria, 11198 non-diabetic and non-obese participants, including 6170 men and 5028 women, were included. The mean age was 70.3 ± 4.8 years old. Their basic demographic data were shown in Table 1. A significant difference in PC, age, gender, all metabolic parameters (including BMI, waist circumference, body fat, blood pressure, FPG, cholesterol, triglyceride, HDL, LDL, IR, GE, FPIS, and SPIS) and CRP was noted between the MetS (+) and MetS (-) subgroups (Table 1).
Table 1
Demographic data and metabolic parameters of the study participants
Variable | All | MetS (+) | MetS (-) | p value |
Number of participant | 11,198 | 3,050 | 8,148 | |
Platelet (µL×103) | 218.9 ± 58.1 | 226.2 ± 59.4 | 216.2 ± 57.4 | < 0.001 |
Age (year) | 70.3 ± 4.8 | 70.7 ± 5.0 | 70.1 ± 4.7 | < 0.001 |
Male | 6,170 (55.1) | 1,345 (44.1) | 4,825 (59.2) | < 0.001 |
Body mass index (kg/m2) | 22.7 ± 2.5 | 24.1 ± 2.0 | 22.2 ± 2.5 | < 0.001 |
Waist circumference (cm) | 79.8 ± 8.0 | 84.5 ± 7.3 | 78.1 ± 7.6 | < 0.001 |
Body fat (%) | 24.0 ± 7.4 | 27.9 ± 6.8 | 22.5 ± 7.1 | < 0.001 |
Systolic blood pressure (mmHg) | 134.7 ± 20.2 | 143.4 ± 18.6 | 131.5 ± 19.8 | < 0.001 |
Diastolic blood pressure (mmHg) | 75.4 ± 11.5 | 79.2 ± 11.2 | 73.9 ± 11.3 | < 0.001 |
Fasting plasma glucose (mg/dL) | 103.5 ± 23.5 | 112.9 ± 32.7 | 99.9 ± 17.6 | < 0.001 |
Total cholesterol (mg/dL) | 206.2 ± 38.0 | 210.7 ± 40.3 | 204.5 ± 36.9 | < 0.001 |
Triglyceride (mg/dL) | 120.8 ± 62.4 | 173.6 ± 72.7 | 101.1 ± 44.0 | < 0.001 |
HDL (mg/dL) | 54.9 ± 15.6 | 46.0 ± 12.5 | 58.3 ± 15.4 | < 0.001 |
LDL (mg/dL) | 127.1 ± 33.4 | 130.1 ± 35.1 | 126.0 ± 32.7 | < 0.001 |
IR (10− 4×min− 1×pmol− 1×L− 1) | 3.67 ± 0.02 | 3.69 ± 0.02 | 3.66 ± 0.02 | < 0.001 |
GE (10− 2×dL×min− 1×kg− 1) | 0.015 ± 0.002 | 0.013 ± 0.002 | 0.016 ± 0.002 | < 0.001 |
FPIS (µU/min) | 90.6 ± 62.2 | 125.4 ± 70.2 | 77.6 ± 53.4 | < 0.001 |
SPIS (pmol/mmol) | 0.059 ± 0.024 | 0.067 ± 0.024 | 0.056 ± 0.023 | < 0.001 |
C-reactive protein (mg/L) | 0.35 ± 1.05 | 0.39 ± 1.19 | 0.33 ± 1.00 | 0.009 |
Abbreviations: MetS, metabolic syndrome; HDL, high-density lipoprotein; LDL, low-density lipoprotein; IR, insulin resistance; GE, glucose effectiveness; FPIS, First phase insulin secretion, SPIS, Second phase insulin secretion; |
Data are shown as frequency (percentage) or mean ± standard deviation. |
The linear trend analysis showed that the higher level of PC was correlated to younger age, a higher proportion of female, higher prevalence of MetS, lower waist circumstance but higher body fat (due to the distribution of sex), higher levels of cholesterol, triglyceride, LDL, IR and CRP and lower level of GE. However, no correlation between PC levels and BMI, blood pressure, FPG, HDL, FPIS, and SPIS was observed (Table 2).
Table 2
Difference in demographic data and metabolic parameters in groups with different level of platelet counts
Variable | Quartile by platelet count | p for linear trend |
Quartile 1 | Quartile 2 | Quartile 3 | Quartile 4 |
Number of participant | 2,782 | 2,811 | 2,804 | 2,800 | |
Platelet count (µL×103) | 153.5 ± 24.2 | 197.5 ± 9.4a | 230.9 ± 10.4ab | 293.4 ± 48.2abc | < 0.001 |
Age (year) | 71.2 ± 5.2 | 70.2 ± 4.7a | 69.9 ± 4.5ab | 69.8 ± 4.6ab | < 0.001 |
Male | 1,866 (67.1) | 1,651 (58.7)a | 1,406 (50.1)ab | 1,246 (44.5)abc | < 0.001 |
Prevalence of MetS | 633 (22.8) | 724 (25.8) | 786 (28.0)a | 906 (32.4)abc | < 0.001 |
Body mass index (kg/m2) | 22.6 ± 2.6 | 22.8 ± 2.5a | 22.7 ± 2.5 | 22.7 ± 2.5 | 0.209 |
Waist circumstance (cm) | 80.3 ± 8.3 | 80.0 ± 8.2 | 79.5 ± 8.0a | 79.5 ± 7.6a | < 0.001 |
Body fat (%) | 22.1 ± 6.9 | 23.6 ± 7.2a | 24.6 ± 7.3ab | 25.6 ± 7.6abc | < 0.001 |
Systolic blood pressure (mmHg) | 134.6 ± 20.8 | 134.5 ± 20.0 | 134.8 ± 19.7 | 135.0 ± 20.1 | 0.388 |
Diastolic blood pressure (mmHg) | 75.4 ± 11.9 | 75.3 ± 11.4 | 75.4 ± 11.4 | 75.5 ± 11.4 | 0.703 |
Fasting plasma glucose (mg/dL) | 103.2 ± 22.7 | 103.5 ± 23.4 | 103.1 ± 22.2 | 104.0 ± 25.4 | 0.301 |
Cholesterol (mg/dL) | 196.0 ± 36.0 | 206.3 ± 37.8a | 209.8 ± 36.1ab | 212.7 ± 39.8abc | < 0.001 |
Triglyceride (mg/dL) | 107.9 ± 55.9 | 118.0 ± 60.3a | 125.9 ± 63.2ab | 131.3 ± 67.0abc | < 0.001 |
HDL (mg/dL) | 54.3 ± 15.2 | 55.1 ± 15.8 | 55.5 ± 15.8a | 54.9 ± 15.7 | 0.070 |
LDL (mg/dL) | 120.1 ± 31.8 | 127.7 ± 33.3a | 129.2 ± 32.9a | 131.5 ± 34.6ab | < 0.001 |
FPIS (µU/min) | 91.1 ± 62.3 | 91.3 ± 61.9 | 90.1 ± 62.9 | 90.0 ± 61.8 | 0.412 |
SPIS (pmol/mmol) | 0.058 ± 0.024 | 0.060 ± 0.024 | 0.059 ± 0.024 | 0.059 ± 0.024 | 0.343 |
IR (10− 4×min− 1×pmol− 1×L− 1) | 3.665 ± 0.024 | 3.668 ± 0.024a | 3.669 ± 0.025a | 3.670 ± 0.025ab | < 0.001 |
GE (10− 2×dL×min− 1×kg− 1) | 0.0152 ± 0.0020 | 0.0149 ± 0.0022a | 0.0147 ± 0.0022ab | 0.0146 ± 0.0022ab | < 0.001 |
C-reactive protein (mg/L) | 0.29 ± 0.82 | 0.30 ± 0.98 | 0.31 ± 0.79 | 0.51 ± 1.46abc | < 0.001 |
Abbreviations: MetS, metabolic syndrome; HDL, high density lipoprotein; LDL, low density lipoprotein; FPIS, first phase insulin secretion; SPIS, second phase insulin secretion; IR, insulin resistance; GE, glucose effectiveness; |
Data are shown as frequency (percentage) or mean ± standard deviation; |
“a”, “b”, and “c” indicate significant difference (p < 0.05) versus quartile 1, quartile 2, and quartile 3, respectively. |
With adjustment of age and sex, the multivariable regression analyses showed that a greater PC level was significantly associated with a higher level of body fat, waist circumference, total cholesterol, LDL, triglyceride, CRP, and IR (P values < 0.05). In contrast, a higher PC level was significantly associated with a lower level of HDL (regression coefficient − 0.10, 95% confidence interval [CI] -0.15 to -0.05) and GE (regression coefficient − 0.40, 95% confidence interval [CI] -0.46 to -0.33). In addition, the multivariable logistic model suggested that a greater PC level was significantly associated with a higher risk of MetS (odds ratio 1.025, 95% CI 1.017 to 1.032) (Table 3).
Table 3
Univariate and multivariable regression analyses for the association between platelet count (µL×104) and metabolic parameters
| Unadjusted | | Age and sex adjusted |
Outcome | β/OR 95% (CI) | P | | β/OR 95% (CI) | P |
Body fat (%) | 0.20 (0.18 to 0.22) | < 0.001 | | 0.06 (0.04 to 0.08) | < 0.001 |
Waist circumference (cm) | -0.04 (-0.07 to -0.02) | 0.001 | | 0.03 (0.01 to 0.06) | 0.009 |
Body mass index (kg/m2) | 0.001 (-0.007 to 0.009) | 0.878 | | -0.006 (-0.014 to 0.003) | 0.179 |
Systolic blood pressure (mmHg) | 0.008 (-0.056 to 0.073) | 0.800 | | -0.002 (-0.067 to 0.062) | 0.940 |
Diastolic blood pressure (mmHg) | -0.002 (-0.038 to 0.035) | 0.933 | | 0.002 (-0.035 to 0.039) | 0.922 |
Cholesterol (mg/dL) | 0.91 (0.79 to 1.03) | < 0.001 | | 0.70 (0.58 to 0.82) | < 0.001 |
HDL (mg/dL) | 0.002 (-0.05, 0.05) | 0.937 | | -0.10 (-0.15, -0.05) | < 0.001 |
LDL (mg/dL) | 0.64 (0.53 to 0.74) | < 0.001 | | 0.54 (0.44 to 0.65) | < 0.001 |
Triglyceride (mg/dL) | 1.39 (1.19 to 1.59) | < 0.001 | | 1.27 (1.07 to 1.47) | < 0.001 |
C-reactive protein (mg/dL) | 0.023 (0.020 to 0.026) | < 0.001 | | 0.026 (0.022 to 0.029) | < 0.001 |
Fasting plasma glucose (mg/dL) | 0.05 (-0.02 to 0.12) | 0.191 | | 0.06 (-0.02 to 0.13) | 0.132 |
FPIS (µU/min) | -0.03 (-0.22 to 0.17) | 0.796 | | 0.15 (-0.05 to 0.35) | 0.141 |
SPIS× (104×pmol/mmol) | 0.01 (-0.76 to 0.78) | 0.974 | | -0.60 (-1.38 to 0.17) | 0.128 |
IR (10− 1×min− 1×pmol− 1×L− 1) | 0.24 (0.16 to 0.31) | < 0.001 | | 0.12 (0.05 to 0.20) | 0.002 |
GE (102×dL×min− 1×kg− 1) | -0.35 (-0.42 to -0.28) | < 0.001 | | -0.40 (-0.46 to -0.33) | < 0.001 |
Metabolic syndrome (binary outcome) | 1.029 (1.022 to 1.037) | < 0.001 | | 1.025 (1.017 to 1.032) | < 0.001 |
Abbreviations: β, regression coefficient; OR, odds ratio; CI, confidence interval; HDL, high density lipoprotein; LDL, low density lipoprotein; FPIS, first phase insulin secretion; SPIS, second phase insulin secretion; IR, insulin resistance; GE, glucose effectiveness. |
RCS analysis (4 knots) rejected the null hypothesis that there is a linear relationship between PC with all the examined variables (P for non-linearity < 0.05), except FPG, systolic blood pressure, diastolic blood pressure, FPIS, and MetS. Thus, PC had significant associations with body fat, waist circumference, cholesterol, HDL, LDL, triglyceride, CRP, GE, and IR in both linear and non-linear models. However, our analysis showed that the non-linear model had a better fit than the linear model for body fat, cholesterol, LDL, triglyceride, CRP, GE, and IR based on a lower Akaike information criterion (AIC) and Bayesian information criterion (BIC). The results were inconsistent between AIC and BIC for waist circumference and HDL, suggesting that either linear or non-linear models are applicable for these two variables.
Alternatively, PC was significantly associated with BMI and SPIS in the non-linear model only. In contrast, there were neither linear nor non-linear associations between PC and FPG, systolic blood pressure, diastolic blood pressure, and FPIS.
Figure 1 depicts the graphic results of both the linear and RCS model for anthropometric measurements. The results demonstrated a non-linearity on body fat, waist circumference, and BMI. The PC levels were associated with greater body fat, waist circumference, and BMI values before approximately 250000/µL at a linear scale, while it turned to be a negative correlation afterward (inverted U shape, Fig. 1A-1C).
Figure 2 illustrates the graphic results of both linear and RCS models for laboratory data. The results demonstrated an apparent non-linearity on cholesterol, HDL, LDL, triglyceride, and CRP. Higher PC level was associated with greater levels of cholesterol, LDL, and triglyceride before approximately 250000/µL at a linear scale, whereas it reveals a slightly negative (inverted U shape, Fig. 2A and 2C) or null correlation afterward (upside down L shape, Fig. 2D). The PC level was slightly positively associated with HDL before 220000/µL and turned to be negatively associated with HDL afterward (Fig. 2B). The PC level was not associated with CRP level until approximately 250000/µL and it demonstrated that the association was linear afterwards (reverse L shape, Fig. 2E).
Figure 3 displays the results of both linear and RCS models for glucose metabolism factors. These results showed that a higher PC level was associated with a greater SPIS level before approximately 200000/µL, however, it reversed to be associated with a lower SPIS level after 200000/µL (inverted U shape, Fig. 3A). Higher PC level was associated with a lower GE value before approximately 250000/µL, but the association tended to be null afterward (shallow U, Fig. 3B). Similar to the result of SPIS, a higher PC level was associated with a greater IR value before approximately 250000/µL and turned to be associated with a lower IR value afterward (inverted U shape, Fig. 3C). At last, the association between PC level and the risk of MetS seemed to be linear in nature (Fig. 4).
The graphic associations between PC and several non-significant variables, including systolic blood pressure, diastolic blood pressure, FPG, and FPIS, were shown in online supplementary Figures S1A-S1D.