3.1 | Baseline characteristics of participants according to selenium intake
Between 2001 and 2004, a total of 3821 participants were finally enrolled in our analysis. The characteristic distribution of participants was presented in Table 1 (according to selenium intake levels). Briefly, selenium intake was associated with age, race, income, hypertension, diabetes, smoking, CHD, and stroke (Table 1, p < 0.05). Besides, BMI, education, alcohol and high cholesterol weren’t associated with selenium intake.
Table 1
Baseline characteristics of participants according to Selenium
| Selenium Q1 (N = 954) | Selenium Q2 (N = 955) | Selenium Q3 (N = 956) | Selenium Q4 (N = 955) | P value | P for trend |
Age$ | 55.11 (19.43) | 50.90 (18.28) | 49.00 (17.67) | 43.36 (16.19) | < 0.001 | < 0.001 |
BMI (mean (SD))$ | 27.67 (5.44) | 28.02 (5.48) | 28.05 (5.42) | 28.32 (5.71) | 0.286 | 0.015 |
Race (%)* | | | | | 0.007 | 0.894Ns |
Mexican American | 172 (18.0) | 187 (19.6) | 217 (22.7) | 191 (20.0) | | |
Non-Hispanic Black | 224 (23.5) | 184 (19.2) | 148 (15.5) | 180 (18.8) | | |
Non-Hispanic White | 495 (51.9) | 529 (55.3) | 520 (54.4) | 529 (55.4) | | |
Other Hispanic | 33 ( 3.5) | 29 ( 3.0) | 42 ( 4.4) | 27 ( 2.8) | | |
Other Race - Including Multi-Racial | 30 ( 3.1) | 27 ( 2.8) | 29 ( 3.0) | 28 ( 2.9) | | |
Education = High School (%)* | 556 (58.3) | 486 (50.9) | 480 (50.2) | 488 (51.2) | 0.052 | 0.002 |
Income (mean (SD))$ | 2.55 (1.56) | 2.86 (1.60) | 2.95 (1.59) | 2.90 (1.62) | 0.028 | < 0.001 |
Alcohol = Yes (%)* | 121 (62.7) | 98 (62.4) | 77 (55.0) | 92 (61.3) | 0.572Ns | 0.491Ns |
Hypertension = Yes (%)* | 362 (38.4) | 323 (34.1) | 283 (30.0) | 213 (22.6) | < 0.001 | < 0.001 |
High_cholesterol = Yes (%)* | 266 (42.0) | 307 (45.5) | 266 (42.5) | 249 (40.4) | 0.837 | 0.382 |
Diabetes = Yes (%)* | 121 (12.7) | 98 (10.3) | 105 (11.0) | 68 ( 7.1) | 0.004 | < 0.001 |
Smoking = Yes (%)* | 617 (64.7) | 576 (60.4) | 568 (59.4) | 531 (55.6) | 0.002 | < 0.001 |
CHD = Yes (%)* | 75 ( 7.9) | 63 ( 6.6) | 69 ( 7.3) | 35 ( 3.7) | 0.011 | < 0.001 |
Stroke = Yes (%)* | 58 ( 6.1) | 35 ( 3.7) | 25 ( 2.6) | 16 ( 1.7) | < 0.001 | < 0.001 |
Note:
Data are presented as mean ± SD or n (%)
* represent χ2 test for categorical variables
$ Linear regression models for continuous variables
Selenium Q1 represent Selenium < 79.5 mcg
Selenium Q2 represent 79.5 ≤ Selenium < 110.4 mcg
Selenium Q3 represent 110.4 ≤ Selenium < 152.6 mcg
Selenium Q4 represent Selenium ≥ 152.6 mcg
CHD represent coronary heart disease
NS represent No significance
3.2 | The association of selenium intake with ED
Participants still were divided four groups according to the daily selenium intake. The ED sometimes (sometimes able + never able) and ED always (never able) were associated with selenium intake (P < 0.001). Neither testostosterone(P = 0.647) nor sex hormone binding globulin(P = 0.06) has any correlation with selenium intake. But sex hormone binding globulin (SHBG) demonstrates a correlated trend (P < 0.001). (Table 2)
Table 2
Baseline characteristics of participants with ED according to Selenium
| Selenium Q1 (N = 954) | Selenium Q2 (N = 955) | Selenium Q3 (N = 956) | Selenium Q4 (N = 955) | P value | P for trend |
Testostosterone (mean (SD))$ | 4.86 (2.30) | 4.94 (2.06) | 5.17 (2.09) | 5.43 (4.29) | 0.647Ns | 0.066Ns |
Sex_Hormone_Binding_Globulin$ | 48.27 (28.20) | 40.23 (25.83) | 40.49 (21.57) | 36.18 (24.19) | 0.06 | < 0.001 |
ED sometimes = Yes (%)* | 356 (37.3) | 282 (29.5) | 260 (27.2) | 162 (17.0) | < 0.001 | < 0.001 |
ED always = Yes (%)* | 150 (15.7) | 115 (12.0) | 89 ( 9.3) | 48 ( 5.0) | < 0.001 | < 0.001 |
Note:
Data are presented as mean ± SD or n (%)
* represent χ2 test for categorical variables
$ Linear regression models for continuous variables
Selenium Q1 represent Selenium < 79.5 mcg
Selenium Q2 represent 79.5 ≤ Selenium < 110.4 mcg
Selenium Q3 represent 110.4 ≤ Selenium < 152.6 mcg
Selenium Q4 represent Selenium ≥ 152.6 mcg
ED sometimes represent Massachusetts Male Aging Study (MMAS) response options were “Sometimes able” + “Never able”
ED always represent Massachusetts Male Aging Study (MMAS) response options were “Never able”
NS represent No significance
3.3 | Baseline characteristics of participants according to Erectile Function
There were 2017 normal participants, the other was ED. Mild, Moderate and Severe were 744, 658 and 402, respectively (Table 3). ED was associated with age, BMI, race, education, income, hypertension, high-cholesterol, diabetes, smoking, CHD, stroke, testostosterone, SHBG, as well as selenium intake. There is no relationship between alcohol and ED. Both as a continuous variable and as a categorical variable, selenium intake was significantly correlated with ED.
Table 3
Baseline characteristics of participants according to Erectile Function
| Normal (N = 2017) | Mild (N = 744) | Moderate (N = 658) | Severe (N = 402) | P value | P for trend |
Age (mean (SD)) | 40.44 (13.84) | 51.53 (16.96) | 62.21 (14.69) | 71.25 (14.83) | < 0.001 | < 0.001 |
BMI (mean (SD)) | 27.75 (5.38) | 28.24 (5.54) | 28.61 (5.84) | 28.01 (5.52) | 0.004 | < 0.001 |
Race (%) | | | | | < 0.001 | 0.876Ns |
Mexican American | 385 (19.1) | 165 (22.2) | 138 (21.0) | 79 (19.7) | | |
Non-Hispanic Black | 411 (20.4) | 160 (21.5) | 126 (19.1) | 39 ( 9.7) | | |
Non-Hispanic White | 1079 (53.5) | 374 (50.3) | 347 (52.7) | 273 (67.9) | | |
Other Hispanic | 72 ( 3.6) | 24 ( 3.2) | 27 ( 4.1) | 8 ( 2.0) | | |
Other Race - Including Multi-Racial | 70 ( 3.5) | 21 ( 2.8) | 20 ( 3.0) | 3 ( 0.7) | | |
Education = High School (%) | 939 (46.6) | 431 (57.9) | 389 (59.1) | 251 (62.4) | < 0.001 | < 0.001 |
Income (mean (SD)) | 2.98 (1.61) | 2.75 (1.61) | 2.59 (1.57) | 2.44 (1.46) | < 0.001 | < 0.001 |
Alcohol = Yes (%) | 178 (57.4) | 82 (63.6) | 83 (66.9) | 45 (58.4) | 0.259Ns | 0.279Ns |
Hypertension = Yes (%) | 387 (19.4) | 236 (32.2) | 333 (51.3) | 225 (56.5) | < 0.001 | < 0.001 |
High_cholesterol = Yes (%) | 438 (37.0) | 216 (43.5) | 278 (51.8) | 156 (46.8) | < 0.001 | < 0.001 |
Diabetes = Yes (%) | 75 ( 3.7) | 69 ( 9.3) | 148 (22.5) | 100 (24.9) | < 0.001 | < 0.001 |
Smoking = Yes (%) | 1071 (53.1) | 472 (63.4) | 467 (71.1) | 282 (70.3) | < 0.001 | < 0.001 |
CHD = Yes (%) | 43 ( 2.1) | 45 ( 6.1) | 74 (11.5) | 80 (20.1) | < 0.001 | < 0.001 |
Stroke = Yes (%) | 21 ( 1.0) | 15 ( 2.0) | 49 ( 7.5) | 49 (12.2) | < 0.001 | < 0.001 |
Testostosterone (mean (SD)) | 5.50 (3.31) | 5.23 (2.21) | 4.60 (2.14) | 3.67 (1.87) | < 0.001 | < 0.001 |
SHBG (mean (SD)) | 34.88 (19.59) | 43.06 (30.69) | 47.47 (24.61) | 58.39 (29.38) | < 0.001 | < 0.001 |
Selenium (mean (SD)) | 133.07(73.38) | 123.89 (74.11) | 111.21 (59.59) | 100.61(52.67) | < 0.001 | 0.0000 |
Quartile (Selenium) | | | | | | |
Selenium Q1 | 404 (20.0%) | 194 (26.1%) | 206 (31.3%) | 150 (37.3%) | < 0.001 | 0.0000 |
Selenium Q2 | 490 (24.3%) | 184 (24.7%) | 167 (25.4%) | 115 (28.6%) | | |
Selenium Q3 | 516 (25.6%) | 180 (24.2%) | 171 (26%) | 89 (22.1%) | | |
Selenium Q4 | 607(30.1%) | 186 (25.0%) | 114 (17.3%) | 48 (11.9%) | | |
Note: |
Data are presented as mean ± SD or n (%)
* represent χ2 test for categorical variables
$ Linear regression models for continuous variables
SHBG represent Sex Hormone Binding Globulin
Selenium Q1 represent Selenium < 79.5 mcg
Selenium Q2 represent 79.5 ≤ Selenium < 110.4 mcg
Selenium Q3 represent 110.4 ≤ Selenium < 152.6 mcg
Selenium Q4 represent Selenium ≥ 152.6 mcg
CHD represent coronary heart disease
Massachusetts Male Aging Study (MMAS) response options answered “almost or almost always able” considered as “Normal”; Answered “usually able” considered as “Mild”; Answered “sometimes able” considered as “Moderate”.
Answered “Never able” considered as “Severe”.
NS represent No significance
3.4 | Daily food selenium intake and ED in the unadjusted and adjusted Logistic regression models
In our study, we used univariate and multivariate linear regression and constructed two main models and to explore the independent impact of selenium intake on ED. As a continuous variable, selenium intake was significantly correlated with ED in both unadjusted model (odds ratio [OR]: 0.99, 95%CI: 0.993–0.995, P < 0.01) and adjusted model (model01, [OR]: 0.99, 95%CI: 0.992–0.998, P < 0.01. model02, [OR]: 0.99, 95%CI: 0.992–0.999, P < 0.05. model03, [OR]: 0.99, 95%CI: 0.995–0.998, P < 0.01.). As a categorical variable, we found that the relationship between the selenium intake and ED was non-linear. Compared with group Q1, we found that Q2 group had 30% lower ED risk ([OR]: 0.70, 95%CI: 0.580–0.581, P < 0.01) and Q3 group had 37% lower ED risk ([OR]: 0.63, 95%CI: 0.517–0.761, P < 0.01), as well as Q4 group had 66% lower ED risk ([OR]: 0.34, 95%CI: 0.277–0.424, P < 0.01) in the unadjusted model. Similar trends are also shown in the adjusted model, that compared with Q1group, the higher selenium intake level groups had lower ED risk, but not significant (Table 4).
The above analysis revealed a significant relationship between ED and the intake of selenium intake. Does a higher intake of selenium mean that the risk of ED is lower?
Table 4
Relationship between daily food selenium intake and ED in the unadjusted model and adjusted Logistic regression models
Characteristic | Unadjusted model | Adjust model 01 | Adjust model 02 | Adjust model 03 |
Selenium (Continuous) | 0.99 (0.993,0.995) 0.000 | 0.99 (0.995,0.998) 0.000 | 0.99 (0.992, 0.999) 0.045 | 0.99 (0.995, 0.998) 0.000 |
Q1 | Reference | Reference | Reference | Reference |
Q2 | 0.70 (0.580,0.851) 0.000 | 0.87 (0.697,1.082) 0.207Ns | 0.941 (0.544,1.630) 0.827Ns | 0.89 (0.707,1.131) 0.354Ns |
Q3 | 0.63 (0.517,0.761) 0.000 | 0.75 (0.601,0.943) 0.014 | 0.976 (0.556,1.714) 0.932Ns | 0.77 (0.607, 0.984) 0.037 |
Q4 | 0.34 (0.277,0.424) 0.000 | 0.53 (0.419,0.681) 0.000 | 0.509 (0.273, 0.936) 0.031 | 0.61 (0.471,0.789) 0.000 |
Note: |
Data present as OR(95% confidence interval) P value
Selenium Q1 represent Selenium < 79.5 mcg
Selenium Q2 represent 79.5 ≤ Selenium < 110.4 mcg
Selenium Q3 represent 110.4 ≤ Selenium < 152.6 mcg
Selenium Q4 represent Selenium ≥ 152.6 mcg
Unadjusted model for selenium & Number for Unadjusted model = 3820
Adjust model 01 Race + Age + Income + Education + Selenium + BMI + Smoking + Drink & Number for Adjust model 01 = 3521
Adjust model 02 Selenium + Testostosterone + SHBG (Sex Hormone Binding Globulin) & Number for Adjust model02 = 587
Adjust model 03 Race + Age + Income + Education + Selenium + BMI + Smoking + Drink + Hypertension + Diabetes + coronary heart disease + Stroke & Number for Adjust model03 = 3460
NS represent No significance
3.5 | Threshold effect analysis of the relationship between the daily food selenium intake and ED in the unadjusted model and adjusted Logistic regression models
We therefore further analyze the relationship between the daily food selenium intake and Erectile Dysfunction (ED). The adjusted smooth curve fitting revealed that the association between selenium intake levels and ED was non-linear after adjusting for Race, Age, Income, Education, Selenium, BMI, Smoking, Drink, Hypertension, Diabetes, coronary heart disease and Stroke (Fig. 2). In order to present the relationship accurately, we adopted two-piecewise binary logistic regression. Using trial and error method, the inflection point was 180.5mcg. When the selenium intake level was lower than 180.5 mcg, it was negatively correlated with the likelihood of ED ([OR]: 0.99, 95%CI: 0.994–0.998, P < 0.01, model03). Whereas when the selenium intake level was higher than 180.5 mcg, the relationship was not significant([OR]: 0.99, 95%CI: 0.991-1.000, P = 0.076, model03) (Table 5).
Table 5
Threshold effect analysis of the relationship between the daily food selenium intake and ED in the unadjusted model and adjusted Logistic regression models
Characteristic | Unadjusted model | Adjust model 01 | Adjust model 02 | Adjust model 03 |
Selenium(Continuous) | 0.99 (0.993,0.995) 0.0000 | 0.99 (0.995,0.998) 0.0000 | 0.99 (0.992, 0.999) 0.045 | 0.99 (0.995, 0.998) 0.000 |
Inflection point | 180.5 | 180.5 | 180.5 | 180.5 |
Selenium < 180.5 | Reference | Reference | Reference | Reference |
Selenium ≥ 180.5 | 0.43 (0.336,0.545) 0.000 | 0.61 (0.470,0.801) 0.000 | 0.51 (0.242,0.976) 0.034 | 0.68 (0.516,0.906) 0.009 |
Selenium < 180.5 (Continuous) | 0.99 (0.991,0.995) 0.0000 | 0.99 (0.988,0.998) 0.014 | 0.99 (0.993,0.997) 0.001 | 0.99 (0.994,0.998) 0.004 |
Selenium ≥ 180.5 (Continuous) | 0.99 (0.991,1.000) 0.126Ns | 1.00 (0.987, 1.012) 0.942Ns | 0.99 (0.994,1.000) 0.170Ns | 0.99 (0.991,1.000) 0.076Ns |
Note: |
Data present as OR(95% confidence interval) P value;
Unadjusted model for selenium & Number for Unadjusted model = 3820;
Adjust model 01 Race + Age + Income + Education + Selenium + BMI + Smoking + Drink & Number for Adjust model 01 = 3521;
Adjust model 02 Selenium + Testostosterone + SHBG (Sex Hormone Binding Globulin) & Number for Adjust model02 = 587;
Adjust model 03 Race + Age + Income + Education + Selenium + BMI + Smoking + Drink + Hypertension + Diabetes + coronary heart disease + Stroke & Number for Adjust model03 = 3460;
Ns represent No significance