On-line supplementary data file. Binary mediation analysis: independent variable =score calculated using age and gender; dependent variable =presence of detectable HS-CTnT; mediator =coronary artery calcification (4 ordered categories). binary_mediation, dv(HSCTnT) mv(CAC4) iv(Age&Sex) OLS regression: CAC4 on iv (a1 path) ------------------------------------------------------------------------------ CAC4 | Coef. Std. Err. t P>|t| Beta -------------+---------------------------------------------------------------- Age&Sex | .2830554 .0338098 8.37 0.000 .3751243 _cons | 1.401448 .0779021 17.99 0.000 . ------------------------------------------------------------------------------ Logit: dv on iv (c path) Logistic regression Number of obs = 430 LR chi2(1) = 83.76 Prob > chi2 = 0.0000 Log likelihood = -160.20912 Pseudo R2 = 0.2072 ------------------------------------------------------------------------------ HSCTnT | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- Age&Sex | 1 .1247678 8.01 0.000 .7554596 1.24454 _cons | 2.06e-08 .2084196 0.00 1.000 -.4084949 .4084949 ------------------------------------------------------------------------------ Logit: dv on mv & iv (b & c' paths) Logistic regression Number of obs = 430 LR chi2(2) = 86.41 Prob > chi2 = 0.0000 Log likelihood = -158.88814 Pseudo R2 = 0.2138 ------------------------------------------------------------------------------ HSCTnT | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- CAC4 | .2424155 .1483464 1.63 0.102 -.0483381 .5331691 Age&Sex | .9379524 .129924 7.22 0.000 .6833061 1.192599 _cons | -.3408045 .2964171 -1.15 0.250 -.9217714 .2401623 ------------------------------------------------------------------------------ Indirect effects with binary response variable HSCTnT indir_1 = .03843131 (CAC4, continuous) total indirect = .03843131 direct effect = .52533238 total effect = .5637637 c_path = .5638047 proportion of total effect mediated = .06816919 ratio of indirect to direct effect = .07315619 ratio of total to direct effect = 1.0731562 Binary models use logit regression Binary mediation analysis: independent variable = qrisk2 score; dependent variable =presence of detectable HS-CTnT; mediator =coronary artery calcification (4 ordered categories). binary_mediation, dv(HSCTnT) mv(CAC4) iv(QRisk2) OLS regression: CAC4 on iv (a1 path) ------------------------------------------------------------------------------ CAC4 | Coef. Std. Err. t P>|t| Beta -------------+---------------------------------------------------------------- QRisk2 | .3063965 .0410989 7.46 0.000 .3400726 _cons | 1.402749 .0860251 16.31 0.000 . ------------------------------------------------------------------------------ Logit: dv on iv (c path) Logistic regression Number of obs = 427 LR chi2(1) = 68.43 Prob > chi2 = 0.0000 Log likelihood = -167.28439 Pseudo R2 = 0.1698 ------------------------------------------------------------------------------ HSCTnT | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- QRisk2 | 1 .1321601 7.57 0.000 .7409709 1.259029 _cons | -8.23e-09 .2191248 -0.00 1.000 -.4294768 .4294768 ------------------------------------------------------------------------------ Logit: dv on mv & iv (b & c' paths) Logistic regression Number of obs = 427 LR chi2(2) = 73.07 Prob > chi2 = 0.0000 Log likelihood = -164.96326 Pseudo R2 = 0.1813 ------------------------------------------------------------------------------ HSCTnT | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- CAC4 | .3187812 .1467563 2.17 0.030 .0311441 .6064183 QRisk2 | .9118533 .1381598 6.60 0.000 .6410651 1.182642 _cons | -.4478226 .304573 -1.47 0.141 -1.044775 .1491296 ------------------------------------------------------------------------------ Indirect effects with binary response variable HSCTnT indir_1 = .04793459 (CAC4, continuous) total indirect = .04793459 direct effect = .44750458 total effect = .49543917 c_path = .49629888 proportion of total effect mediated = .09675171 ratio of indirect to direct effect = .1071153 ratio of total to direct effect = 1.1071153 Binary models use logit regression Binary mediation analysis: independent variable = Framingham score; dependent variable =presence of detectable HS-CTnT; mediator =coronary artery calcification (4 ordered categories). binary_mediation, dv(HSCTnT) mv(CAC4) iv(Framingham) OLS regression: CAC4 on iv (a1 path) ------------------------------------------------------------------------------ CAC4 | Coef. Std. Err. t P>|t| Beta -------------+---------------------------------------------------------------- Framingham | .3673421 .0580639 6.33 0.000 .2965837 _cons | 1.484098 .1092086 13.59 0.000 . ------------------------------------------------------------------------------ Logit: dv on iv (c path) Logistic regression Number of obs = 417 LR chi2(1) = 38.19 Prob > chi2 = 0.0000 Log likelihood = -174.31537 Pseudo R2 = 0.0987 ------------------------------------------------------------------------------ HSCTnT | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- Framingham | 1 .1678172 5.96 0.000 .6710843 1.328916 _cons | 5.13e-09 .2728095 0.00 1.000 -.5346968 .5346968 ------------------------------------------------------------------------------ Logit: dv on mv & iv (b & c' paths) Logistic regression Number of obs = 417 LR chi2(2) = 49.59 Prob > chi2 = 0.0000 Log likelihood = -168.61578 Pseudo R2 = 0.1282 ------------------------------------------------------------------------------ HSCTnT | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- CAC4 | .4789045 .1412116 3.39 0.001 .2021349 .7556741 Framingham | .8651384 .1746936 4.95 0.000 .5227452 1.207532 _cons | -.7045175 .3498948 -2.01 0.044 -1.390299 -.0187362 ------------------------------------------------------------------------------ Indirect effects with binary response variable HSCTnT indir_1 = .06574401 (CAC4, continuous) total indirect = .06574401 direct effect = .32331221 total effect = .38905622 c_path = .38474254 proportion of total effect mediated = .16898331 ratio of indirect to direct effect = .20334527 ratio of total to direct effect = 1.2033453 Binary models use logit regression Binary mediation analysis: independent variable = Assign score; dependent variable =presence of detectable HS-CTnT; mediator =coronary artery calcification (4 ordered categories). binary_mediation, dv(HSCTnT) mv(CAC4) iv(Assign) OLS regression: CAC4 on iv (a1 path) ------------------------------------------------------------------------------ CAC4 | Coef. Std. Err. t P>|t| Beta -------------+---------------------------------------------------------------- Assign | .3708425 .0710465 5.22 0.000 .2446379 _cons | 1.456901 .1240167 11.75 0.000 . ------------------------------------------------------------------------------ Logit: dv on iv (c path) Logistic regression Number of obs = 430 LR chi2(1) = 27.65 Prob > chi2 = 0.0000 Log likelihood = -188.26701 Pseudo R2 = 0.0684 ------------------------------------------------------------------------------ HSCTnT | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- Assign | 1 .1952919 5.12 0.000 .6172349 1.382765 _cons | -1.51e-08 .3092846 -0.00 1.000 -.6061866 .6061866 ------------------------------------------------------------------------------ Logit: dv on mv & iv (b & c' paths) Logistic regression Number of obs = 430 LR chi2(2) = 40.71 Prob > chi2 = 0.0000 Log likelihood = -181.73803 Pseudo R2 = 0.1007 ------------------------------------------------------------------------------ HSCTnT | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- CAC4 | .4896939 .1349686 3.63 0.000 .2251602 .7542275 Assign | .8436903 .201221 4.19 0.000 .4493044 1.238076 _cons | -.7345824 .375891 -1.95 0.051 -1.471315 .0021505 ------------------------------------------------------------------------------ Indirect effects with binary response variable HSCTnT indir_1 = .056778 (CAC4, continuous) total indirect = .056778 direct effect = .26378436 total effect = .32056237 c_path = .32173567 proportion of total effect mediated = .17711999 ratio of indirect to direct effect = .21524401 ratio of total to direct effect = 1.215244 Binary models use logit regression Binary mediation analysis: independent variable = JBS/BNF score; dependent variable =presence of detectable HS-CTnT; mediator =coronary artery calcification (4 ordered categories). binary_mediation, dv(HSCTnT) mv(CAC4) iv(JBS/BNF) OLS regression: CAC4 on iv (a1 path) ------------------------------------------------------------------------------ CAC4 | Coef. Std. Err. t P>|t| Beta -------------+---------------------------------------------------------------- JBS/BNF | .3795559 .061306 6.19 0.000 .2907804 _cons | 1.498545 .1133945 13.22 0.000 . ------------------------------------------------------------------------------ Logit: dv on iv (c path) Logistic regression Number of obs = 417 LR chi2(1) = 35.30 Prob > chi2 = 0.0000 Log likelihood = -175.76196 Pseudo R2 = 0.0913 ------------------------------------------------------------------------------ HSCTnT | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- JBS/BNF | 1 .1735019 5.76 0.000 .6599424 1.340058 _cons | -2.31e-09 .2817927 -0.00 1.000 -.5523035 .5523035 ------------------------------------------------------------------------------ Logit: dv on mv & iv (b & c' paths) Logistic regression Number of obs = 417 LR chi2(2) = 47.23 Prob > chi2 = 0.0000 Log likelihood = -169.79962 Pseudo R2 = 0.1221 ------------------------------------------------------------------------------ HSCTnT | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- CAC4 | .4885612 .1407825 3.47 0.001 .2126325 .7644899 JBS/BNF | .8566292 .1808351 4.74 0.000 .5021989 1.211059 _cons | -.7266356 .3585542 -2.03 0.043 -1.429389 -.0238823 ------------------------------------------------------------------------------ Indirect effects with binary response variable HSCTnT indir_1 = .06618277 (CAC4, continuous) total indirect = .06618277 direct effect = .30573353 total effect = .3719163 c_path = .36779769 proportion of total effect mediated = .17795072 ratio of indirect to direct effect = .21647208 ratio of total to direct effect = 1.2164721 Binary models use logit regression