A reported “3.421-fold” association between methamphetamine-related disorders and cardiomyopathy is not a statement that every exposed person has a 3.421-fold probability of developing heart disease. Reading the underlying Taiwanese claims study requires separating diagnostic codes, observed events, person-time rates and an adjusted hazard model.
What population did the study follow?
Yu and colleagues analyzed historical Taiwanese insurance records from 2000–2015, matching 17,071 adults with methamphetamine-related disorder diagnoses to a comparison group at a 1:4 ratio by age, sex and inclusion year. The abstract and main text report 68,264 controls. Some table headings instead print 68,284; the stated total of 85,355 and the 1:4 design correspond to 68,264. That discrepancy should remain visible rather than silently becoming a new denominator.
The exposure was an administrative diagnosis involving dependence, abuse or psychosis categories. It was not a standardized administered dose, a measured blood concentration or verified consumption by a defined route. Absence of a recorded disorder diagnosis also does not prove a person had never used methamphetamine.
What do the events and rates show?
The paper reports 153 cardiomyopathy events in the diagnosed group and 274 in the comparison group. Table 4 gives 198,273.14 and 793,065.24 person-years, respectively. Dividing events by these denominators produces approximately 77.17 and 34.55 events per 100,000 person-years, as that table reports.
A later paragraph instead describes rates of 0.77 and 0.34 per 100,000 person-years. Those figures are inconsistent with both the table and the event/person-time arithmetic by roughly a factor of 100. They should not be repeated as interchangeable estimates. The event counts are not percentages, and person-year rates are not fixed-period individual probabilities.
What does the adjusted hazard ratio add?
The Cox analysis reported an adjusted hazard ratio of 3.421, with a 95% confidence interval of 1.589–5.014. The model incorporated recorded demographic, comorbidity and health-system variables. A hazard ratio compares modeled event occurrence over follow-up; it is not the crude ratio of the two observed rates and should not be multiplied directly by someone’s presumed lifetime risk.
The association is relevant evidence, but adjustment only addresses measured variables under the model’s assumptions. The study lacked detailed alcohol consumption, lifestyle, genetic, psychosocial and exposure-severity information. It also relied on people coming into medical care. Differences in diagnosis and surveillance may matter alongside biological effects.
Do subgroup labels establish a dose–response relationship?
No. The reported hazard ratios differed across dependence, abuse and psychosis codes, but those codes are not measured dose levels. The study did not supply comparable milligrams, frequency, route or cumulative exposure for each person. Treating the categories as a calibrated exposure ladder would add information the database did not contain.
Likewise, a sex association in this human claims model is a different question from sex differences in a controlled mouse experiment. Neither can be used to declare universal male or female susceptibility without attention to the population, outcome and comparison.
How should this paper inform a case review?
It can support a carefully qualified statement that recorded methamphetamine-related disorders were associated with subsequent recorded cardiomyopathy in this historical cohort. It cannot assign the cause of one person’s cardiomyopathy, reconstruct their dose or demonstrate impairment from a positive toxicology result. The reporting inconsistencies also make checking the original tables essential before quoting absolute rates.
Source
Yu PC et al. Increased risk of cardiomyopathy in individuals with methamphetamine related disorders in Taiwan. Scientific Reports. 2025;15:11449. doi:10.1038/s41598-025-94591-0. Observational human claims evidence using 2000–2015 records.
Leave a Reply