Table 3.
Association between individual antidepressant drug classes and incident myocardial infarction
| Use of SSRIs | Use of TCAs | |||||
|---|---|---|---|---|---|---|
| Percentagea | Events | OR (95 % CI) | Percentagea | Events | OR (95 % CI) | |
| Analysis 1Ab | ||||||
| Nonuse | 92.2 | 595 | 1 (reference) | 90.6 | 587 | 1 (reference) |
| Current use | 1.8 | 7 | 0.70 (0.42–1.17) | 1.6 | 9 | 0.84 (0.53–1.33) |
| Past use | 6.0 | 48 | 1.33 (0.97–1.83) | 7.8 | 54 | 1.05 (0.79–1.41) |
| Analysis 1Bc | ||||||
| Nonuse | 92.4 | 681 | 1 (reference) | 90.7 | 672 | 1 (reference) |
| Current use | 1.8 | 8 | 0.65 (0.41–1.02) | 1.6 | 10 | 0.80 (0.52–1.24) |
| Past use | 5.8 | 55 | 1.42 (1.06–1.90) | 7.7 | 62 | 1.04 (0.79–1.38) |
| Analysis 2d | ||||||
| Past use | 5.8 | 50 | 1 (reference) | 7.7 | 54 | 1 (reference) |
| Current use | 1.8 | 7 | 0.58 (0.23–1.49) | 1.6 | 10 | 0.60 (0.26–1.41) |
Matched on age and sex and adjusted for: history of deep venous thrombosis, history of heart failure, systolic and diastolic blood pressure, maximum level of education, total cholesterol, smoking, blood-pressure-lowering agents, antithrombotic agents, antipsychotic agents, anxiolytics, hypnotics, depression and anxiety, and current use of the other antidepressant drug classes (including other antidepressants)
Abbreviations: 95 % CI 95 % confidence interval, OR odds ratio, SSRIs selective serotonin reuptake inhibitors, TCAs tricyclic antidepressants
aAs we studied the associations with time-varying exposure analysis, controls contributed more than once in the computation of the odds ratios before they were censored of became a case. For this reason, exposure is reported as a percentage
bAnalyses with never use of antidepressants as reference, using unimputed data
cAnalyses with never use of antidepressants as reference, using imputed data
dAnalyses with past use of antidepressants as reference, using imputed data. For the analysis on SSRIs, 1 case currently using had no matched controls and 5 cases who were past user had no matched controls. For the analysis on TCAs, 8 past users could not be matched. These were therefore not included in the analyses