Risk of Cardiovascular Diseases and Stroke Events in Methamphetamine Users: A 10-Year Follow-Up Study

Risk of Cardiovascular Diseases and Stroke Events in Methamphetamine Users:

A 10-Year Follow-Up Study

Ming-Chyi Huang, MD, PhDa,c; Shu-Yu Yang, PhDb; Shih-Ku Lin, MDa,c; Kuan-Yu Chen, MD, PhDa,e; Ying-Yeh Chen, MD, ScDa,f; Chian-Jue Kuo, MD, PhDa,c,’ ¡; and Yen-Ni Hung, PhDd,’ ¡,*

Methamphetamine is highly addictive and may cause widespread organ toxicity because of its central and peripheral sympathomimetic effects.1 In addition to the extensively documented neuropsychiatric toxicity, cardiovascular and cerebrovascular complications are among the most adverse physical effects of the drug.1-4 The cardiovascular and cerebrovascular pathologies were reported to respectively account for 54% and 20% of the fatalities for which autopsy findings were available.3 The pathologies are more prevalent in methamphetamine users than in age-matched controls2,5 and occur at a significantly younger age than in the general population.2,6-8

Methamphetamine stimulates the release of catecholamines (dopamine and norepinephrine) and has an α- and β-adrenergic agonist effect, causing elevated heart rate and blood pressure, which is considered the primary mechanism underlying the cardiovascular and cerebrovascular harms.1,4-6 Excess catecholamine leads to a decreased cardiac oxygen supply caused by vasoconstriction and vasospasm and increases oxygen demand because of tachycardia and hypertension, resulting in an overall reduction in oxygen availability to the heart, which in turn causes cardiac muscle damage. These effects, together with the direct cardiotoxicity independent of catecholamine-mediated effects, predispose users to a high risk of cardiovascular complications.4,9-14 Similarly, the pharmacologic effect of methamphetamine on cerebral circulation has been associated with a higher risk of hemorrhagic and ischemic strokes.6,15

Literature supporting the association between methamphetamine exposure and cardiovascular and cerebrovascular events mainly contains case reports, case series, postmortem studies, and cross-sectional design-based research; however, long-term observation of a large number of methamphetamine users is lacking. Given that the risk of the aforementioned complications is unlikely to be limited to the period of methamphetamine use4 and may increase as the user ages, and that the risk can accumulate with each individual use episode over time,1 prospective follow-up may more accurately characterize the long-term adverse effects of methamphetamine use. Therefore, as an extension to previous research, we conducted a cohort study with observation time ranging from 10 to 14 years by analyzing a nationwide, population-based claims dataset to assess the long-term cardiovascular and cerebrovascular complications in methamphetamine users.

METHODS

Data Source

The single-payer National Health Insurance program was launched in Taiwan on March 1, 1995, and covers nearly 99% of the Taiwan population. Since 1996, the National Health Research Institute in Taiwan has maintained the National Health Insurance reimbursement claims data and established the National Health Insurance Research Database (NHIRD) (http://nhird.nhri.org.tw/en/), which provides data for research. The database includes standard claims information of beneficiaries, such as demographic data, diagnoses, prescriptions, and health care expenses. Information that can be used to identify beneficiaries and medical care providers were scrambled by the Bureau of National Health Insurance,16 and all investigators signed an agreement to ensure patient confidentiality before accessing the database.

Definition of Methamphetamine Cohort

We retrospectively selected the methamphetamine cohort from the Psychiatric Inpatient Medical Claims (PIMC) database, a subset of the NHIRD. This database was nested from nearly the entire Taiwan population (N = nearly 23 million) and comprised a cohort of patients hospitalized for any psychiatric disorder between 1996 and 2007 (n = 187,117). The database includes patients with at least 1 psychiatric inpatient record and 1 discharge diagnosis of mental illness coded according to the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM), codes 290-319. In Taiwan, each hospital that provides psychiatric inpatient service is accredited periodically by an independent nongovernment organization, the Taiwan Joint Commission on Hospital Accreditation. An accreditation requirement is that a board-certified psychiatrist diagnoses the patients. The inpatient database was used because the admission diagnoses based on complete clinical information were considered optimal and reliable.

Methamphetamine was the leading illicit drug in Taiwan before the 2000s, whereas other illicit drugs, such as “ecstasy,” ketamine, and marijuana, were uncommon.17,18 The current study cohort recruited eligible methamphetamine users from those who were diagnosed with amphetamine and other psychostimulant dependence (ICD-9-CM code 304.4x), amphetamine or related acting sympathomimetic abuse (ICD-9-CM code 305.7x), or drug psychosis (ICD-9-CM code 292.xx) between 1997 and 2000. Patients who had a diagnosis of drug psychosis (292.xx) and any primary or secondary diagnosis of other drug dependence (ICD-9-CM codes 304.0x, 304.1x, 304.2x, 304.3x, 304.5x, 304.6x, 304.7x, 304.8x, 304.9x, 305.0x, 305.1x, 305.2x, 305.3x, 305.4x, 305.5x, 305.6x, 305.8x, and 305.9x) within 1 year before or after receiving the 292.xx diagnosis were excluded from the best estimation analysis for diagnostic accuracy. The baseline time point (index date) for follow-up was defined as the date when each inpatient met the inclusion criteria. The study excluded patients younger than 18 years with any inpatient or outpatient diagnosis of cardiovascular disease or a stroke event before the index date or with any head injury diagnosis. Exclusion criteria were searched for in the time prior to the index date. The amount of time varied for each user and overall ranged from January 1, 1996, to the index date, which was between 1997 and 2000. For example, if the index date was December 31, 2000, the duration for which we searched the exclusion criteria was 5 years. Patient enrollment is illustrated in Supplementary eFigure 1.

• Existing evidence based on long-term observation of the association between methamphetamine exposure and cardiovascular and cerebrovascular events is scarce.
• This 10-year follow-up study showed that methamphetamine use is significantly associated with a risk of subsequent cardiovascular and cerebrovascular complications, particularly arrhythmia and hemorrhagic strokes.
• Age might be an effect modifier of the complications.

Selection of the Population Proxy Comparison Group

For the methamphetamine cohort, we selected an appropriate comparison group from the Longitudinal Health Insurance Database 2005 (LHID2005), which contains registration and claim data of 1,000,000 beneficiaries randomly sampled from the 2005 Registry for Beneficiaries of the NHIRD and has the same source population as that of PIMC (ie, nearly the entire population of Taiwan). No statistically significant differences were observed in age, sex, or medical expenditures between the patients in the LHID2005 and the complete NHIRD (http://nhird.nhri.org.tw/en/Data_Subsets.html#S3).

For comparison, we selected patients older than 18 years and who were admitted to hospitals between 1997 and 2000 (N = 191,678) from the LHID2005. The index date for the comparison group was defined as the date of admission to the hospital. We then excluded those who had been previously diagnosed with any cardiovascular or cerebrovascular diseases or had a previous diagnosis of methamphetamine use (ICD-9-CM codes 304.4, 305.7, and 292.xx) before the index date; the resulting group of 177,312 individuals served as the comparison candidates. Exclusion criteria were also searched from January 1, 1996, to the index date. The mean ± SD duration of searching period for exclusion criteria was comparable between the 2 groups (3.35 ± 1.02 and 3.34 ± 1.05 years for methamphetamine and comparison group, respectively).

We used the propensity score-based matching methods developed by Rosenbaum and Rubin19 to select patients for the comparison cohort. A multivariate logistic regression model for obtaining propensity scores was established according to the baseline characteristics of the population, including sex, age, Charlson Comorbidity Index (CCI),20,21 urbanization, and index year. The general medical condition was assessed using CCI, which is widely used in research and is the sum of weighted scores of 17 comorbid conditions.22,23 Records 1 year prior to the date of the index hospitalization were retrieved to identify comorbidities. The resulting predicted probabilities were used initially to apply the greedy match algorithm to generate the “best” matches followed by “next-best” matches hierarchically until no matches could be made.24 The cases were first matched to the controls to 8 digits of the propensity score; if no matches were found, they were matched to 7 digits. The algorithm proceeded sequentially to the lowest propensity score digit matched (1 digit). In this study, all cases were matched to 1 control at 8 digits and 3 controls at 7 digits, resulting in 4 comparison cohort subjects (n = 5,260) for each methamphetamine user, with similar baseline characteristics distributions as that of methamphetamine users (n = 1,315) (Table 1).

Outcome Measures

We monitored the methamphetamine and comparison cohort patients for any incidence of cardiovascular disease or stroke event after the index date, which were the primary outcomes of interest. The ICD-9-CM codes for the cardiovascular disease and stroke event categories are presented in Table 2. The category of ischemic heart disease is subdivided in accordance with previous research4,10,12 into acute coronary syndrome and coronary artery disease. Acute coronary syndrome includes acute myocardial infarction and other acute and subacute forms of ischemic heart disease (eg, unstable angina), whereas coronary artery diseases are more chronic and include angina pectoris and other forms of chronic ischemic heart disease. The follow-up period began at the baseline date and progressed up to the date of the incidence of an event or date of censor (ie, date of death or the end of the study [December 31, 2010]).

Statistical Analyses

Incidence rates for cardiovascular diseases and stroke events were calculated as the ratio of incident cases and contributed person-years. Differences in incidences of diseases between the 2 groups were investigated using life table survival analysis. Cox proportional hazards analysis was used to calculate the hazard ratios (HRs) for each outcome of interest in the methamphetamine cohort relative to the comparison group. As recommended in a systematic review,25 matching variables were not included in the Cox modeling procedures for groups that had been balanced across covariates.

Because previous data suggest that methamphetamine users with cardiovascular complications were relatively young,10 we examined age as a potential effect modifier in our study. We defined the age variable according to different age cutoffs (< 25 y and ≥ 25 y, < 30 y and ≥ 30 y, and < 35 y and ≥ 35 y) and then examined the interaction effect between age and methamphetamine exposure on outcomes by using regression models. The results showed that the interaction effect was significant on the risk of cardiovascular diseases (P = .006) when the age cutoff was 30 years but not when it was 25 or 35 years. We subsequently determined whether the risk of complications of interest differed between young (aged < 30 y) and old (aged ≥ 30 y) patients. SAS 9.1.3 software (SAS Institute Inc, Cary, North Carolina) was used for data management and analysis. A P value of .05 or lower was considered significant in the regression analyses.

Sensitivity Analyses

Sensitivity analyses were used to test the robustness of the results. Because patients hospitalized for an appendectomy were relatively indistinguishable from the general population and no statistically significant differences were observed between these patients and the general population in Taiwan with regard to sex (P > .05) or age (P > .05),26,27 we selected patients hospitalized for appendectomy as a second comparison group from the LHID2005. These patients were selected at a 1:1 rather than 1:4 ratio because of a limited number of cases and then matched with the methamphetamine cohort (Supplementary eTable 1). Patients with a diagnosis of stroke before appendectomy were excluded. Furthermore, the appendectomy procedures have no known long-term impact on the increased risk of stroke or cardiovascular diseases.26,27

RESULTS

Characteristics of Methamphetamine Cohort and Comparison Groups

Table 1 summarizes the baseline characteristics of the 2 groups. In the methamphetamine cohort, the patients were mainly men (82.5%), and approximately 50% of them were younger than 30 years. We identified 142 and 124 patients in the methamphetamine and comparison groups, respectively, who died during the study period.

Incidence of Cardiovascular Diseases and Stroke Events

The total incidence of cardiovascular diseases and stroke events significantly differed between the 2 groups (87.5/10,000 person-years [methamphetamine cohort] vs 55.3/10,000 person-years [controls], P < .001) (Table 3). The incidence of arrhythmia (P = .011) and hemorrhagic stroke (P = .001) was significantly higher in the methamphetamine cohort.

Risk of Cardiovascular Diseases and Stroke Events

Figure 1 reveals a significantly increased risk of cardiovascular diseases and stroke events in the methamphetamine cohort compared with the comparison group (HR = 1.55, P < .001), particularly for arrhythmia (HR = 1.92, P = .014) and hemorrhagic stroke (HR = 2.09, P = .001).

Influence of Age on the Risk of Cardiovascular Diseases and Stroke Events

We stratified all patients into younger (< 30 y) and older (≥ 30 y) age groups (Table 4). The younger methamphetamine group had a high risk of cardiovascular diseases (HR = 2.22, P = .001). By contrast, the older methamphetamine group was not associated with any type of cardiovascular disease, but did have a higher risk of stroke events (HR = 1.86, P = .001), particularly hemorrhagic stroke (HR = 2.41, P = .001).

The proportional hazards assumption was confirmed for all estimated variables, and the assumption was not violated when we stratified the sample by age of 30 years (P > .05).

Sensitivity Analyses

Consistent with the aforementioned results, methamphetamine users also carried higher risks of cardiovascular disease (HR = 1.89, P = .004) and stroke events (HR = 2.32, P = .002), particularly hemorrhagic strokes (HR = 2.53, P = .014), when compared with the second comparison group of patients (Supplementary eTable 1).

DISCUSSION

We found that methamphetamine users carried a long-standing risk of cardiovascular and cerebrovascular consequences, particularly arrhythmia and hemorrhagic stroke. In addition, the association with cardiovascular sequelae was more significant in the younger methamphetamine users than in the older users, who exhibited a higher risk of stroke. These findings expand previous studies by estimating real-world risks by using a large sample and nationwide retrospective cohort.

Risk of Cardiovascular Diseases

The results show a clear relationship between methamphetamine use and cardiovascular complications. Although the exact mechanisms are unclear, previous studies have suggested that in addition to indirect actions through excessive catecholamine release, methamphetamine causes direct cardiotoxicity, including cardiomyocyte hypertrophy or fibrosis and myofiber disarrangement.28 Cardiotoxicity can occur acutely2,4,10,12 or persist for long durations with subclinical findings even after methamphetamine discontinuation.29 A common pattern of methamphetamine abuse is characterized by a short period of repeated (several times a day) drug administration (“run” or “binge” use) followed by a drug-free period. In animal studies, this cyclic-use pattern led to sensitization of cardiovascular responses to methamphetamine,30,31 resulting in a greater increase in the heart rate and blood pressure when methamphetamine was administered intermittently over long intervals than over short intervals.4,30-32 Therefore, a complex interplay between the acute and chronic effects of methamphetamine use should be considered during estimation of the actual risk of cardiovascular complications. Thus, the follow-up data helped determine the long-term risk.

A higher incidence of electrocardiographic abnormalities33 and dysregulated heart rate variability34 has been associated with a high risk of arrhythmia in methamphetamine users. Repeated or intermittent use of methamphetamine may sensitize users to the tachycardia effects of the drug.14,30-32 Cardiac arrhythmia or palpitations are among the physical symptoms most commonly presented by methamphetamine users who visit medical or psychiatric emergency departments.10,35,36 Consistent with these findings, our study showed that methamphetamine use was associated with an elevated risk of arrhythmia. However, the enhanced risk of acute coronary syndrome, coronary artery disease, hypertension, or congestive heart failure suggested in previous studies was not demonstrated.1,2,4,5,9,13 Possible explanations for this include the small sample size, a lack of controls for existing cardiac diseases, the inclusion of participants with known cardiovascular diseases, and the cross-sectional design of previous studies.37

Risk of Stroke Events

Consistent with previous reports,3,15 we found that methamphetamine users had an increased risk of stroke events, particularly hemorrhagic strokes. The occurrence of stroke in adults younger than 50 years is uncommon and is associated with atypical etiologies such as amphetamine misuse.7,8 Most of the patients (97%) in our study were younger than 50 years. Previous studies have estimated a 4-fold increase in the risk of ischemic or hemorrhagic stroke among cocaine or amphetamine users.38 In further attempting to differentiate the effects of amphetamine on strokes from those of cocaine, Westover et al15 found that amphetamine use was typically associated with hemorrhagic events but not with ischemic strokes. Kaye et al3 noted that cerebral hemorrhage accounted for the majority of methamphetamine-related deaths associated with cerebrovascular complications. Our findings also support the relationship between methamphetamine use and the risk of hemorrhagic rather than ischemic stroke. It is postulated that transient hypertensive surges caused by the sympathomimetic effects of methamphetamine use lead to blood vessel injury (eg, cerebral vasculitis), which in turn causes intracerebral hemorrhage.6 The lack of association between ischemic stroke and methamphetamine use might be because the methamphetamine users we enrolled were young and thus less likely to experience ischemic stroke. In addition, we cannot eliminate the possibility that an extended duration (> 10 y) might be necessary to monitor the risk because the incidence of ischemic stroke development may require a long latent period.39

Age as an Effect Modifier for Risk

Age acted as an effect modifier for risk estimation. We found an increased risk of cardiovascular diseases among the patients aged < 30 years, whereas those aged ≥ 30 years were more vulnerable to hemorrhagic strokes. The reasons underlying the age-dependent differential risk is unclear. Prior studies have indicated that young patients are more likely to be associated with rapid progression of substance abuse and engage in a binge-like methamphetamine use pattern that can be harmful and potentially fatal.40,41 Animals treated with several high doses of methamphetamine mimicking that of humans during bingeing behavior showed age-dependent differences in methamphetamine-induced monoaminergic responses.42 We speculate that an age-dependent impact of methamphetamine on catecholamine toxicity could likewise have existed for the patients in our study. In addition, the increased risk of cardiovascular diseases in the group aged < 30 years might be because of the rarity of cardiovascular diseases, which potentially exacerbates the difference between groups. Additional studies are warranted to clarify the mechanisms that contribute to the age-different toxic effects of methamphetamine use on vascular systems.

Study Limitations

This study is limited in several ways. First, the diagnosis in the claims data is code based on the ICD-9-CM system, which does not have codes (304.4x) specific to methamphetamine use. To overcome this problem, we developed a best-estimate strategy for identifying subjects with methamphetamine use. However, the methamphetamine cohort may not be completely representative because the group might have included cases that involved the use of only amphetamine or of other psychostimulants but not methamphetamine. Furthermore, methamphetamine users who concomitantly had a diagnosis of other drug abuse would have been excluded. To examine the potential impact from misclassification of cases, we reanalyzed the data by not excluding subjects who used other drugs of abuse (n = 35) and compared them with an independent comparison group. The results were similar to what we had observed (data not shown). Second, relative to the comparison group, the methamphetamine users may have sought medical advice less often.43 This study enrolled only the methamphetamine users who had used the National Health Insurance program for inpatient psychiatric service and did not include those who had never been admitted or those who did not request medical attention. In addition, methamphetamine users who were hospitalized might have been lost to follow-up. Therefore, our findings cannot be generalized to all individuals with methamphetamine use. Third, most methamphetamine users might have been using methamphetamine long before their inpatient diagnosis. Although we excluded patients with preexisting cardiovascular diseases or stroke events in the claims database before the baseline date, this study might still have underestimated the effects of prior methamphetamine exposure before 1996, which would have already predisposed the patients to high risks of complications, resulting in a bias away from the null. Fourth, in some individuals, the vascular events may not have been diagnosed accurately. Similarly, the accuracy of stroke-related codes in the NHIRD might be debatable. A recent study44 evaluating the validity of the database reported that approximately 98% of ischemic stroke diagnoses were confirmed by imaging examinations. Another study45 using the same database showed that nearly 90% of cases with hemorrhagic strokes had undergone imaging examinations during hospitalization. Therefore, the accuracy of stroke diagnoses in the database is sufficient. Fifth, because of the limited number of incident cases, we determined not to adjust our results by using strategies such as Bonferroni correction. If a P value < .005 is reset as significant, all of the results except that for arrhythmia remained the same. Further research is required to validate the associations observed in this study.

The claims database has several inherent limitations that might affect our results. It does not record possible confounders, such as smoking, blood pressure, nonprescribed aspirin use, physical inactivity, and other types of unhealthy behaviors. Therefore, we could not evaluate whether these variables confounded the association between methamphetamine use and cardiovascular disease or stroke events. Not a single factor but various factors conjointly account for the increased incidence of complications. Furthermore, case files with respect to duration, severity, dosage, and route of methamphetamine administration were not available in the database. Thus, we were unable to investigate any possible dose-response relationship between methamphetamine exposure and outcome measures or the relative risk associated with different routes of methamphetamine administration. In addition, the database did not include the relevant clinical information that allowed us to specify effects of acute use (eg, vasospasm and tissue ischemia) compared with those of long-term methamphetamine use (eg, accelerated atherosclerosis and endothelial dysfunction), or information on the cause of mortality, which would have enabled us to clarify cardiovascular disease- or stroke-specific death.

In summary, we found that methamphetamine users were susceptible to long-term risks of cardiovascular complications and stroke events. The identification of the adverse consequences of methamphetamine use is of utmost importance in increasing the involvement of methamphetamine users in motivational interviews, educating potential methamphetamine abusers about avoidable risks, and reducing the overall burden produced by complications.

Submitted: February 9, 2015; accepted November 24, 2015.

Online first: August 30, 2016.

Drug names: ketamine (Ketalar and others).

Potential conflicts of interest: The authors have no conflict of interest to declare in this study.

Funding/support: This research was supported by grants from the National Science Council (NSC102-2628-B-532-001-MY3; NSC101-2314-B-532-002 -MY2) and Ministry of Science and Technology (MOST103-2628-B-532 -001-MY3) of Taiwan, and Taipei City Hospital (9900-62-049, 10101-62-008; 10101-62-055; 10101-62-032; 10301-62-042), Taipei, Taiwan.

Role of the sponsor: The funding organizations had no role in designing or conducting the study; the collection, management, analysis, or interpretation of data; or the preparation, review, or approval of the manuscript.

Supplementary material: Available at PSYCHIATRIST.COM.

REFERENCES

1. Darke S, Kaye S, McKetin R, et al. Major physical and psychological harms of methamphetamine use. Drug Alcohol Rev. 2008;27(3):253-262. PubMed doi:10.1080/09595230801923702

2. Karch SB, Stephens BG, Ho CH. Methamphetamine-related deaths in San Francisco: demographic, pathologic, and toxicologic profiles. J Forensic Sci. 1999;44(2):359-368. PubMed doi:10.1520/JFS14464J

3. Kaye S, Darke S, Duflou J, et al. Methamphetamine-related fatalities in Australia: demographics, circumstances, toxicology and major organ pathology. Addiction. 2008;103(8):1353-1360. PubMed doi:10.1111/j.1360-0443.2008.02231.x

4. Kaye S, McKetin R, Duflou J, et al. Methamphetamine and cardiovascular pathology: a review of the evidence. Addiction. 2007;102(8):1204-1211. PubMed doi:10.1111/j.1360-0443.2007.01874.x

5. Yeo KK, Wijetunga M, Ito H, et al. The association of methamphetamine use and cardiomyopathy in young patients. Am J Med. 2007;120(2):165-171. PubMed doi:10.1016/j.amjmed.2006.01.024

6. Ho EL, Josephson SA, Lee HS, et al. Cerebrovascular complications of methamphetamine abuse. Neurocrit Care. 2009;10(3):295-305. PubMed doi:10.1007/s12028-008-9177-5

7. Yeung M, Bhalla A, Birns J. Recreational drug misuse and stroke. Curr Drug Abuse Rev. 2011;4(4):286-291. PubMed doi:10.2174/1874473711104040286

8. de los R×­os F, Kleindorfer DO, Khoury J, et al. Trends in substance abuse preceding stroke among young adults: a population-based study. Stroke. 2012;43(12):3179-3183. PubMed doi:10.1161/STROKEAHA.112.667808

9. Ito H, Yeo KK, Wijetunga M, et al. A comparison of echocardiographic findings in young adults with cardiomyopathy: with and without a history of methamphetamine abuse. Clin Cardiol. 2009;32(6):E18-E22. PubMed doi:10.1002/clc.20367

10. Turnipseed SD, Richards JR, Kirk JD, et al. Frequency of acute coronary syndrome in patients presenting to the emergency department with chest pain after methamphetamine use. J Emerg Med. 2003;24(4):369-373. PubMed doi:10.1016/S0736-4679(03)00031-3

11. Waksman J, Taylor RN Jr, Bodor GS, et al. Acute myocardial infarction associated with amphetamine use. Mayo Clin Proc. 2001;76(3):323-326. PubMed doi:10.4065/76.3.323

12. Wijetunga M, Bhan R, Lindsay J, et al. Acute coronary syndrome and crystal methamphetamine use: a case series. Hawaii Med J. 2004;63(1):8-13, 25. PubMed

13. Wijetunga M, Seto T, Lindsay J, et al. Crystal methamphetamine-associated cardiomyopathy: tip of the iceberg? J Toxicol Clin Toxicol. 2003;41(7):981-986. PubMed doi:10.1081/CLT-120026521

14. Yu Q, Larson DF, Watson RR. Heart disease, methamphetamine, and AIDS. Life Sci. 2003;73(2):129-140. PubMed doi:10.1016/S0024-3205(03)00260-1

15. Westover AN, McBride S, Haley RW. Stroke in young adults who abuse amphetamines or cocaine: a population-based study of hospitalized patients. Arch Gen Psychiatry. 2007;64(4):495-502. PubMed doi:10.1001/archpsyc.64.4.495

16. Gau SS, Chung CH, Gau CS. A pharmacoeconomic analysis of atypical antipsychotics and haloperidol in first-episode schizophrenic patients in Taiwan. J Clin Psychopharmacol. 2008;28(3):271-278. PubMed doi:10.1097/JCP.0b013e3181723713

17. Chen WJ, Fu TC, Ting TT, et al. Use of ecstasy and other psychoactive substances among school-attending adolescents in Taiwan: national surveys 2004-2006. BMC Public Health. 2009;9(1):27. PubMed doi:10.1186/1471-2458-9-27

18. Chiang SC, Chen CY, Chang YY, et al. Prevalence of heroin and methamphetamine male users in the northern Taiwan, 1999-2002: capture-recapture estimates. BMC Public Health. 2007;7(1):292. PubMed doi:10.1186/1471-2458-7-292

19. Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika. 1983;79:516-524.

20. Quan H, Sundararajan V, Halfon P, et al. Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care. 2005;43(11):1130-1139. PubMed doi:10.1097/01.mlr.0000182534.19832.83

21. Romano PS, Roos LL, Jollis JG. Adapting a clinical comorbidity index for use with ICD-9-CM administrative data: differing perspectives. J Clin Epidemiol. 1993;46(10):1075-1079, discussion 1081-1090. PubMed doi:10.1016/0895-4356(93)90103-8

22. Kuo CJ, Yang SY, Liao YT, et al. Second-generation antipsychotic medications and risk of pneumonia in schizophrenia. Schizophr Bull. 2013;39(3):648-657. PubMed doi:10.1093/schbul/sbr202

23. Quail JM, Lix LM, Osman BA, et al. Comparing comorbidity measures for predicting mortality and hospitalization in three population-based cohorts. BMC Health Serv Res. 2011;11(1):146. PubMed doi:10.1186/1472-6963-11-146

24. D’ Agostino RB Jr. Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat Med. 1998;17(19):2265-2281. PubMed doi:10.1002/(SICI)1097-0258(19981015)17:19<2265::AID-SIM918>3.0.CO;2-B

25. Shah BR, Laupacis A, Hux JE, et al. Propensity score methods gave similar results to traditional regression modeling in observational studies: a systematic review. J Clin Epidemiol. 2005;58(6):550-559. PubMed doi:10.1016/j.jclinepi.2004.10.016

26. Lin HC, Tsai SY, Lee HC. Increased risk of developing stroke among patients with bipolar disorder after an acute mood episode: a six-year follow-up study. J Affect Disord. 2007;100(1-3):49-54. PubMed doi:10.1016/j.jad.2006.09.016

27. Lin HC, Tsai SY, Lee HC. No higher risk of myocardial infarction among bipolar patients in a 6-year follow-up of acute mood episodes. Psychosom Med. 2008;70(1):73-76. PubMed doi:10.1097/PSY.0b013e31815c1e93

28. Maeno Y, Iwasa M, Inoue H, et al. Methamphetamine induces an increase in cell size and reorganization of myofibrils in cultured adult rat cardiomyocytes. Int J Legal Med. 2000;113(4):201-207. PubMed doi:10.1007/s004149900088

29. Inouye DS, Navin JJ, Hardman JM. Fatal postoperative arrhythmia in a man with a remote history of methamphetamine and cocaine use: a case report. Hawaii Med J. 2004;63(3):82-86. PubMed

30. Varner KJ, Ogden BA, Delcarpio J, et al. Cardiovascular responses elicited by the “binge” administration of methamphetamine. J Pharmacol Exp Ther. 2002;301(1):152-159. PubMed doi:10.1124/jpet.301.1.152

31. Yoshida K, Morimoto A, Makisumi T, et al. Cardiovascular, thermal and behavioral sensitization to methamphetamine in freely moving rats. J Pharmacol Exp Ther. 1993;267(3):1538-1543. PubMed

32. Fukunaga T, Mizoi Y, Adachi J. Methamphetamine-induced changes of peripheral catecholamines: an animal experiment to elucidate the cause of sudden death after methamphetamine abuse. Nippon Hoigaku Zasshi. 1987;41(4):335-341. PubMed

33. Haning W, Goebert D. Electrocardiographic abnormalities in methamphetamine abusers. Addiction. 2007;102(suppl 1):70-75. PubMed doi:10.1111/j.1360-0443.2006.01776.x

34. Henry BL, Minassian A, Perry W. Effect of methamphetamine dependence on heart rate variability. Addict Biol. 2012;17(3):648-658. PubMed doi:10.1111/j.1369-1600.2010.00270.x

35. Pasic J, Russo JE, Ries RK, et al. Methamphetamine users in the psychiatric emergency services: a case-control study. Am J Drug Alcohol Abuse. 2007;33(5):675-686. PubMed doi:10.1080/00952990701522732

36. Richards JR, Bretz SW, Johnson EB, et al. Methamphetamine abuse and emergency department utilization. West J Med. 1999;170(4):198-202. PubMed

37. Maulik N, Wei Z, Liu X, et al. Improved postischemic ventricular functional recovery by amphetamine is linked with its ability to induce heat shock. Mol Cell Biochem. 1994;137(1):17-24. PubMed doi:10.1007/BF00926035

38. Petitti DB, Sidney S, Quesenberry C, et al. Stroke and cocaine or amphetamine use. Epidemiology. 1998;9(6):596-600. PubMed doi:10.1097/00001648-199811000-00005

39. Ohta K, Mori M, Yoritaka A, et al. Delayed ischemic stroke associated with methamphetamine use. J Emerg Med. 2005;28(2):165-167. PubMed doi:10.1016/j.jemermed.2004.06.015

40. Anker JJ, Baron TR, Zlebnik NE, et al. Escalation of methamphetamine self-administration in adolescent and adult rats. Drug Alcohol Depend. 2012;124(1-2):149-153. PubMed doi:10.1016/j.drugalcdep.2012.01.004

41. Baumeister SE, Tossmann P. Association between early onset of cigarette, alcohol and cannabis use and later drug use patterns: an analysis of a survey in European metropolises. Eur Addict Res. 2005;11(2):92-98. PubMed doi:10.1159/000083038

42. Kokoshka JM, Fleckenstein AE, Wilkins DG, et al. Age-dependent differential responses of monoaminergic systems to high doses of methamphetamine. J Neurochem. 2000;75(5):2095-2102. PubMed doi:10.1046/j.1471-4159.2000.0752095.x

43. Falck RS, Wang J, Carlson RG, et al. Perceived need for substance abuse treatment among illicit stimulant drug users in rural areas of Ohio, Arkansas, and Kentucky. Drug Alcohol Depend. 2007;91(2-3):107-114. PubMed doi:10.1016/j.drugalcdep.2007.05.015

44. Cheng CL, Kao YH, Lin SJ, et al. Validation of the National Health Insurance Research Database with ischemic stroke cases in Taiwan. Pharmacoepidemiol Drug Saf. 2011;20(3):236-242. PubMed doi:10.1002/pds.2087

45. Wu CS, Wang SC, Gau SS, et al. Association of stroke with the receptor-binding profiles of antipsychotics-a case-crossover study. Biol Psychiatry. 2013;73(5):414-421. PubMed doi:10.1016/j.biopsych.2012.07.006