Table 1.
Main baseline data of patients in the trials included
Fig 1.
Main analysis: net effect of Ezetimibe on Cancer stratified by comparator (E+lipid-lowering drug against the same lipid-lowering drug at the same dosage; E+simvastatin against simvastatin at the same dosage).
The medium degree of heterogeneity (I2 = 60.1%) should encourage the use of the random effect method of pooling. No test for interaction was done (only two trials). The overall result, although not significant, shows a definite trend toward damage (212% increment of Cancer Risk, p = 0.167), due to an impressive trend toward damage in the UK-HARP-II trial, where simvastatin was the Not-E drug.
Fig 2.
Main analysis: net effect of Ezetimibe on all-cause death stratified by comparator (E+lipid-lowering drug against the same lipid-lowering drug at the same dosage; E+simvastatin against simvastatin at the same dosage).
The low degree of heterogeneity (I2 = 10.3%) should encourage use of the fixed effect method of pooling. The test for interaction does not show the formal presence of a ‘subgroup effect’ (p = 0.2696); however, the direction of the result is toward damage only in the stratum where simvastatin was the no-E drug, showing a clinically important increment of death risk. The non-significant overall result shows a small trend toward damage (3% increment of death risk, p = 0.947).
Fig 3.
Main analysis: net effect of Ezetimibe on CV death stratified by comparator (E+lipid- lowering drug against the same lipid-lowering drug at the same dosage; E+simvastatin against simvastatin at the same dosage).
The low degree of heterogeneity (I2 = 33.7%) should encourage the use of the fixed effect method of pooling. The test for interaction does not show the formal presence of a ‘subgroup effect’ (p = 0.2378); however, the direction of the result is toward damage only in the stratum where simvastatin was the no-E drug, showing a clinically important increment of CV death risk. The non-significant overall result shows a trend toward benefit (10% decrement of CV death risk, p = 0.844).
Fig 4.
Main analysis: net effect of Ezetimibe on not-CV death stratified by comparator (E+lipid- lowering drug against the same lipid-lowering drug at the same dosage; E+simvastatin against simvastatin at the same dosage).
The low degree of heterogeneity (I2 = 0.0%) should encourage the use of the fixed effect method of pooling. The test for interaction does not show the formal presence of a ‘subgroup effect’ (p = 0.422); however, the direction of the result is toward damage only in the stratum where simvastatin was the no-E drug, showing a clinically important increment of not-CV death risk. The non-significant overall result shows a small trend toward damage (4% increment of not-CV death risk, p = 0.966). We calculated the number of not-CV deaths by subtracting the number of CV deaths from the number of all-cause deaths reported by each single trial.
Fig 5.
Main analysis: net effect of Ezetimibe on MI stratified by comparator (E+lipid lowering drug against the same lipid-lowering drug at the same dosage; E+simvastatin against simvastatin at the same dosage).
The low degree of heterogeneity (I2 = 0.0%) should encourage the use of the fixed effect method of pooling. The test for interaction does not show the formal presence of a ‘subgroup effect’ (p = 0.4901); however, the direction of the result is toward damage only in the stratum where simvastatin was the no-E drug, showing a clinically important increment of MI risk. The non-significant overall result shows a trend toward damage (37% increment of MI risk, p = 0.636).
Fig 6.
Main analysis: net effect of Ezetimibe on stroke stratified by comparator (E+lipid- lowering drug against the same lipid-lowering drug at the same dosage; E+simvastatin against simvastatin at the same dosage).
The low degree of heterogeneity (I2 = 0.0%) should encourage the use of the fixed effect method of pooling. The test for interaction does not show the formal presence of a ‘subgroup effect’ (p = 0.4385); however, the direction of the result is toward damage only in the stratum where simvastatin was the no-E drug, showing a clinically important increment of stroke risk. The non-significant overall result shows a trend toward damage (45% increment of stroke risk, p = 0.546).
Fig 7.
Main analysis: net effect of Ezetimibe on SAEs stratified by comparator (E+lipid- lowering drug against the same lipid-lowering drug at the same dosage; E+simvastatin against simvastatin at the same dosage).
The low degree of heterogeneity (I2 = 0.0%) should encourage the use of the fixed effect method of pooling. The test for interaction does not show the formal presence of a ‘subgroup effect’ (p = 0.4803); however, the direction of the result is toward damage only in the stratum where simvastatin was the no-E drug, showing a clinically important 45% increment of SAEs risk. The non-significant overall result shows a trend toward damage (24% increment of SAEs risk, p = 0.216). The larger risk of major clinical endpoints was in fact noted in the E arms of trials where E plus simvastatin were compared with simvastatin alone (Figs 1–7).
Fig 8.
Complementary analysis: net effect of the Ezetimibe/Simvastatin combination (E+Simvastatin against placebo) on cancer.
The simvastatin dosage was different in the two trials (20 mg/day in SHARP, 40 mg/day in SEAS). The high degree of heterogeneity (I2 = 83.7%) should encourage use of the random effect method of pooling. The non-significant overall result shows a trend toward damage (18% increment of cancer risk, p = 0.395).
Fig 9.
Complementary analysis: net effect of the Ezetimibe/Simvastatin combination (E+simvastatin against placebo) on all-cause death.
The simvastatin dosage was different in the two trials (20 mg/day in SHARP, 40 mg/day in SEAS). The low degree of heterogeneity (I2 = 0.0%) should encourage the use of the fixed effect method of pooling. The nonsignificant overall result shows a small trend toward damage (2% increment of all-cause death risk, p = 0.596).
Fig 10.
Complementary analysis: net effect of the Ezetimibe/Simvastatin combination (E+Simvastatin against placebo) on CV death.
The simvastatin dosage was different in the two trials (20 mg/day in SHARP, 40 mg/day in SEAS). The low degree of heterogeneity (I2 = 0.0%) should encourage the use of the fixed effect method of pooling. The nonsignificant overall result shows a trend toward benefit (9% decrement of CV death risk, p = 0.162).
Fig 11.
Complementary analysis: net effect of the Ezetimibe/simvastatin combination (E+simvastatin against placebo) on not-CV death.
The simvastatin dosage was different in the two trials (20 mg/day in SHARP, 40 mg/day in SEAS). The low degree of heterogeneity (I2 = 0.0%) should encourage the use of the fixed effect method of pooling. The nonsignificant overall result shows a trend toward damage (8% increment of not-CV death risk, p = 0.091). We calculated the number of not-CV deaths by subtracting the number of CV deaths from the number of all-cause deaths reported by each single trial.
Fig 12.
Complementary analysis: net effect of the Ezetimibe/simvastatin combination (E+simvastatin against placebo) on MI.
The simvastatin dosage was different in the two trials (20 mg/day in SHARP, 40 mg/day in SEAS). The low degree of heterogeneity (I2 = 0.0%) should encourage the use of the fixed effect method of pooling. The non-significant overall result shows a trend toward benefit (19% decrement of MI risk, p = 0.051).
Fig 13.
Complementary analysis: net effect of the Ezetimibe/simvastatin combination (E+simvastatin against placebo) on Stroke.
The simvastatin dosage was different in the two trials (20 mg/day in SHARP, 40 mg/day in SEAS). The low degree of heterogeneity (I2 = 19.8%) should encourage the use of the fixed effect method of pooling. The non-significant overall result shows a trend toward benefit (14% decrement of Stroke risk, p = 0.115).
Fig 14.
Complementary analysis: net effect of the Ezetimibe/simvastatin combination (E+simvastatin against placebo) on SAEs.
The simvastatin dosage was different in the two trials (20 mg/day in SHARP, 40 mg/day in SEAS). The low degree of heterogeneity (I2 = 0.0%) should encourage the use of the fixed effect method of pooling. The non-significant overall result shows a small trend toward benefit (1% increment of SAEs risk, p = 0.837).