Multistage sampling was performed, initially based on calculating the sample size of the study nationally and for autonomous communities, adjusted to their population weights. Problems related to alcohol, smoking and other drugs were analyzed with an alpha error of 5%, a 2-tailed test, a statistical power of 90%, a control:case ratio of at least 4:1, and a more unfavorable proportion of cases in which cases were exposed and controls were not exposed, based on the available scientific evidence. The hospital selection was adjusted to the distribution of hospital groups in each autonomous community, which resulted in 87 hospitals being selected from all the autonomous communities in Spain.

Based on the written or digitalized information in the medical record, each patient's diagnoses, the external causes and the procedures applied were codified, following the ICD-9 (ninth revision of the International Classification of Diseases and Causes of Death). Specialized personnel, with solid training and experience in data recording, codified the data and introduced the information in the databases. These databases contained information on demographic characteristics, admission and discharge dates, type of admission and discharge, diagnostic codes for the main and secondary diagnoses, and external causes and procedures, classified using ICD-9 codes. These databases also included diagnosis-related groups and each hospital was classified into a group, depending on its size and complexity.9 The analysis was restricted to patients ≥ 18 years of age at the time of hospital discharge.

Cases of AMI, first episode, were defined as those in which the code appeared in the main diagnosis (codes 410.01-410.91). Cases were excluded if the code appeared in one of the secondary diagnoses, but not the main diagnosis. The ICD-9 codes were used to define cocaine use disorders as cocaine dependence (304.20-304.23) and cocaine abuse (305.60-305.63). Disorders due to cannabis, opioid, amphetamine, sedative or hypnotic, alcohol use and smoking were defined in a similar manner.

Age was quantified in years. The following comorbidities were identified: obesity, uncomplicated hypertension, complicated hypertension, arrhythmias, pulmonary circulation disorders, valvular disease, deficiency anemia, posthemorrhagic anemia, electrolyte disorders, weight loss, hypothyroidism, coagulopathy, previous AMI, congestive heart failure, peripheral vascular disease, cerebrovascular disease, dementia, chronic pulmonary disease, rheumatic disease, peptic ulcer, diabetes mellitus without chronic complications, diabetes mellitus with chronic complications, hemiplegia or paraplegia, kidney disease, moderate or severe liver disease, cancer, leukemia or lymphoma, metastatic cancer, AIDS, and depression. We used the ICD-9 codes proposed by Quan et al10 for these comorbidities. The Charlson comorbidity index was also calculated for each patient.11

The main aim was to calculate the association between the incidence of AMI and cocaine use disorders in patients admitted to hospital. The secondary aims were to determine mortality, length of hospital stay, and hospital costs in patients with and without cocaine use disorders who had experienced an AMI. We calculated costs by using the specific hospital costs for each diagnosis-related group stratified by hospital group and using the estimates published by the Ministry of health for 2008 to 2010.9

A univariate analysis was performed to examine the association between AMI and addictive disorders, age, sex, and comorbidities. Then, multivariate models were constructed, using unconditional logistic regression analysis to determine the association between cocaine disorders and other addictions with AMI and in-hospital death due to AMI, and controlling for the remaining variables. A multivariate analysis of covariance was performed to determine the effect of cocaine use disorders on length of hospital stay in days and on costs in patients with AMI. The data were adjusted by age, sex, addictions, hospital group, and the above-mentioned comorbidities. The analysis was performed with the STATA statistical program, version MP 12.1.

The characteristics of patients with and without cocaine use disorders are shown in Table 1. We identified 5 475 325 patients, of which 24 126 (0.44%) had cocaine use disorders and 79 076 (1.4%) were admitted for AMI. A total of 538 (2.2%) patients with cocaine use disorders were admitted for AMI, while 78 538 (1.4%) without these disorders were admitted.

Patients with cocaine use disorders were younger (mean age, 37.3 years), were mainly male (78.2%) and showed a high prevalence of addiction to all drugs, mainly tobacco (58.9%) and alcohol (50.3%), but also cannabis (35.7%), opioids (33.1%), amphetamines (4.5%), and sedatives or hypnotics (9.8%).

The mean Charlson comorbidity index score was lower in patients with cocaine use disorders, which could be explained by their mean age; in addition, in most of these patients, their rates of specific comorbidities were significantly lower–or without statistically significant differences–than those of patients without these disorders. However, patients with cocaine use disorders had a higher prevalence of weight loss, coagulopathies, mild, moderate or severe liver disease, AIDS, and depression.

A more detailed analysis of the distribution of the prevalence of cocaine use disorders by age and sex is presented in Figure 1, which shows that, in both men and women, the group with the highest prevalence of these disorders was that aged 35 years to 44 years, followed by the group < 35 years, that aged 45 years to 54 years and, finally, that aged 55 years to 64 years.

Figure 1.

Distribution by groups and age and sex of the prevalence of cocaine use disorders among patients admitted to a sample of 87 Spanish hospitals from 2008 to 2010.

(0.07MB).

The characteristics of patients with and without AMI are shown in Table 2. Those admitted for AMI were older (mean age, 67.7 years), were mainly male (70.8%), and showed the two most prevalent addictions: tobacco (45.7%) and cocaine (0.7%).

The mean Charlson score was higher among patients with AMI, mainly due to the comorbidities associated with risk of AMI: obesity, hypertension, cardiac arrhythmias, pulmonary circulation disorders, valvular disease, prior AMI, congestive heart failure, cerebrovascular disease, diabetes mellitus without chronic complications, and diabetes mellitus with chronic complications.

A more detailed analysis of the distribution of cases of AMI by groups of age and sex is presented in Figure 2, which shows that the prevalence of AMI increased with age and reached a peak in the group aged 55 years to 64 years; however, the odds ratio was significantly higher among patients with cocaine use disorders, with a significant positive tendency: the prevalence of AMI doubled among the group aged 35 years to 44 years, tripled among that aged 45 years to 54 years and quadrupled among the group aged 55 years to 64 years when compared with the reference group aged < 35 years (P for tendency < 0.0001). These analyses were limited to the population aged 18 years to 64 years, because only 3 patients aged ≥ 65 years had a cocaine use disorder and AMI.

Figure 2.

Frequency of myocardial infarction in patients with and without cocaine use disorders by age groups. The odds ratio of each age group is shown in comparison with that in the group aged 18 years to 34 years and the P-value for tendency (chi-squared). 95%CI, 95% confidence interval; OR, odds ratio.

(0.1MB).

The results of the multivariate analysis are shown in Table 3, which reveals that the prevalence of AMI was 3-fold higher in patients with cocaine use disorders than in those without these disorders when the analysis was adjusted by age, sex, the remaining addictions, and the 30 comorbidities listed previously. The positive tendency between age and AMI was maintained in the multivariate analysis, but the differences between odds ratios were less marked than in the univariate analysis.

The multivariate analysis showed no association between having a cocaine use disorder and mortality among patients with AMI.

The multivariate analysis of covariance, which included age, sex, hospital group, and all the above-mentioned addictions and comorbidities, showed that, among patients with an AMI, those with cocaine use disorders had a mean excess length of hospital stay of 1.5 days (95% confidence interval, 1.4-1.6 days). Overexpendure due to hospital stay among patients with an AMI and cocaine use disorder was 382 (95% confidence interval, 298-464) euros.

Our study has several limitations. The data used were drawn exclusively from the Minimum Basic Data Set (MBSD) and were not completed by additional data from the patients. Throughout the study, we used the definitions of addictive disorders, AMI and comorbidities as assigned the physicians in each center, which were then codified and introduced in the databases by the codifiers. Another limitation is the potential under-registration of information due to the possibility that the medical record did not contain all the data required for the codifiers to assign codes or due to variation in the codifiers’ interpretations.

The prevalence of cocaine use disorders in the hospitalized population was markedly higher in men than in women, and the group with the highest rate (3.5%) consisted of men aged 35 years to 44 years. The real prevalence is probably underestimated in hospitalized patients, given the results of several studies comparing the reported addiction rate with other objective indicators such as the prevalence of cocaine metabolites in urine. In a study by Bosch et al,26 19% of the patients did not admit to having used cocaine, even though cocaine metabolites were detected in their urine. In other studies, the proportion of patients denying cocaine use varied: 28% of the patients in a study by Hollander et al27 and 48.2% of those in a study by Lee et al.28 All these factors indicate that the present study probably underestimated the number of cases of AMI concomitant to cocaine use and that the impact of cocaine is probably considerably greater.

Databases such as the MBSD also have considerable advantages.29 Registered data is usually complete in most hospital discharges and, because these databases include practically all cases, they allow fairly accurate estimates on incidence, prevalence, comorbidities, and mortality from diseases attended in the hospital setting.30 These data can be retrospectively analyzed, unlike other designs that require prospective data collection. Data from long periods and a large number of patients, as in the present study, can be collected relatively rapidly and easily. Because the data are systematically gathered, the cost reduction is considerable. Studies based on these databases may show fewer selection biases, such as those biases that lead patients or their legal representatives to refuse to sign informed consent and participate in the study.

The results of this study are supported by several factors. We used the 410.x1 codes of the ICD-9 for AMI, first episode as the main diagnosis (excluding those that appeared as a secondary diagnosis), following the inclusion and exclusion criteria of the United States’ AHRQ (Agency for Healthcare Research and Quality) in its health care quality indicator “in-hospital mortality from acute myocardial infarction”31 to obtain maximal validity of AMI diagnosis. A study carried out and published in Spain that used the MBSD and the AHRQ criteria found that mortality from AMI was highly similar to that of patients who were followed up in observational studies such as PRIMHO, PRIMHO II, RISCI, and ARIM.32 The latter study also found that other characteristics, such as distribution by sex, mean age, and the frequency of cardiovascular risk factors, were similar, indicating that the use of the MBSD led to adequate representation of the epidemiological profile of AMI.32 The results of another study on the validity of the comorbidities of AMI in Spain, calculated from MBSD data, which also used the AHRQ inclusion and exclusion criteria, suggest that the information obtained from the MBSD is reliable and valid.33 The latter study, which analyzed in-hospital mortality from AMI from 2003 to 2009, observed that mortality was decreasing, a finding that also coincides with the tendency noted in other observational studies carried out in Spain.34,35

The overexpenditure incurred by patients with cocaine use disorders was due to 2 factors. First, the higher number of diagnostic and therapeutic procedures and the greater number of complications in these patients, which led to their inclusion in more expensive diagnosis-related groups. Second, the excess cost was attributable to longer hospital stay.

Because of the sample size and the diversity of hospitals, our data can be generalized and are not limited to patients admitted to one or a few hospital centers. Calculation of overexpenditure was facilitated by the availability of costs for each diagnosis-related group stratified by hospital groups and by each year. To our knowledge, this is the first study that calculates the excess hospital stay and overexpenditure attributable to cocaine use disorders in patients with AMI.

One of the main therapeutic objectives after hospital discharge is cessation of cocaine use. The incidence of chest pain, AMI, and death decrease in patients who cease using cocaine.36,37 The remaining cardiac risk factors should be modified, especially cigarette smoking,12 as well as any other addictions. Activities with demonstrated effectiveness that could prevent AMI recurrence in these patients include a brief intervention on the risks of these substances and referral to specialized rehabilitation services. Given the current economic restrictions, decreasing the number of admissions and readmissions due to cocaine would help to reduce hospital costs and increase the availability of hospital beds. Each case of AMI and of other diseases avoided or of prevented admission would also reduce the burden of the multiple problems experienced by these patients in a particularly difficult and stressful era.