We agree with Carrillo et al with regard to the importance of the information bias when dealing with cocaine use. Nevertheless, other factors also explain the differences between our results1 and the findings reported in their study,2 carried out in a coronary critical care unit (CCCU). In a study by Gupta et al3 involving 102 952 acute myocardial infarction patients in 364 hospitals participating in the American College of Cardiology Registry of the United States, only 924 patients (0.9%) of all ages were cocaine-positive, defined as use of the drug within the preceding 72hours or its presence in urine.
Extrapolation of the results of a study involving 87 hospitals (Spain) or 364 hospitals (United States)—in patients with confirmed acute myocardial infarction from different geographical areas, treated in different types of hospitals and departments, and of different ages—to the findings in a single CCCU in which acute coronary syndrome was studied in patients under 50 years of age is problematic. It would also be risky to extrapolate the results of a single CCCU to all the patients hospitalized for acute myocardial infarction.
We were surprised by the high mortality rate among cocaine-positive patients in the CCCU. There were no statistically significant differences between the study by Gupta et al3 and ours.1 Hollander et al4 reported a 0% mortality rate among cocaine-positive patients, which could be due to 2 different circumstances:
- 1.
In their study, Carrillo et al2 performed a simple analysis based on 2 deaths among 24 cocaine-positive patients and 3 deaths among 379 patients with a cocaine-negative urine test. If we calculate the odds ratio (OR) and use an exact method to determine its 95% confidence intervals (CI), we obtain an OR=11.4 (95% CI, 0.89-103.3). These authors only evaluated the statistical significance (P=.03), but overlooked the imprecision of the OR which, with a 95% CI, indicates that a cocaine-positive urine test can be associated with an increased risk of death, with no effect at all (OR=1) or, surprisingly, with a reduced risk of death (OR=0.89-0.99).
- 2.
Our study1 and that of Gupta et al3 (but not that of the above-mentioned authors) involved a multivariate analysis that included important prognostic variables (age, sex, other addictive disorders, comorbidities, complications, analytical findings, and treatments, depending on the study).
For all these reasons, we consider that, to calculate the risk of mortality attributable to cocaine use in patients with acute coronary syndrome admitted to a CCCU, the sample should be large enough to guarantee the statistical power of the study, the accuracy of the estimators of the effect size, and control of confounding bias by means of a multivariate analysis of the results.
FUNDINGThis study was financed by the Spanish Government Delegation for the National Plan on Drugs, Ministry of Health, Social Services and Equality (grant no. 2009I017, project reference no. G41825811).