Single and dual antiplatelet therapy in elderly patients of medically managed myocardial infarction

The National Health Insurance (NHI) program in Taiwan started in March 1995 and enrolled 22.60 million beneficiaries (more than 98% of inhabitants) at the end of 2007. NHI has reimbursed DAPT in all ACS patients including ST-segment elevation myocardial infarction (STEMI) and NSTEMI since October 1, 2007. Thus we included all patients who were admitted for AMI during the period through October 1, 2007 to December 31, 2010 from the NHI claims database. We extracted all of the records of the study subjects from the NHI claims database and linked to the National Death Registry for mortality outcomes using the identification number of each patient. To comply with data privacy regulations, personal identities were encrypted and all data were analyzed in a de-identified manner. The protocol for this study was approved by the Institutional Review Board of National Taiwan University Hospital.

AMI was identified using International Classification of Disease- Clinical Modification, 9th version (ICD-9-CM) codes for STEMI (410.0~ 410.6, 410.8) and NSTEMI (410.7, 410.9). To identify true incident cases of AMI, patients had records of AMI within January 1, 2002 ~ September 30, 2007 were excluded. In addition, patients with unknown discharge date and gender were excluded. We also excluded patients who were younger than 85 years at index hospitalization for AMI, who died in hospital or within 7 days of discharge, and who had ever undergone PCI within 1 year prior to the index hospitalization. Patients with prescriptions of aspirin or clopidogrel after discharge were retained in the study cohort and were grouped into three groups including DAPT, aspirin only, and clopidogrel only, according to their first prescription at outpatient clinic after their index AMI hospitalization. The antiplatelet exposure duration should be 7 days or more and the date of the first prescription of antiplatelet agents was operationally defined as the index date. Those who did not have any prescription of anti-platelet agents within 28 days after discharge were excluded. A further 1381 patients were excluded due to receiving PCI during the index hospitalization. The remaining 1469 individuals comprised the final study cohort (Fig. 1).

Potential confounders included age, sex, individual comorbidities such as diabetes mellitus, hypertension, congestive heart failure (CHF), peripheral artery disease (PAD), chronic pulmonary disease, renal disease, peptic ulcer disease and malignancy within the last 12 months prior to the index date in accordance with Charlson comorbidity measurements [21]. Use of statins, angiotensin-converting-enzyme inhibitors (ACEIs), angiotensin receptor blockers (ARBs), beta-blockers, diuretics and proton pump inhibitors (PPIs) within 60 days prior to the index date and at discharge of the index AMI hospitalization were also included as confounders. The clinical outcomes were all-cause death, cardiovascular death (ICD-9-CM codes 401–449 as the main cause of death) and gastrointestinal bleeding (ICD-9-CM codes 531.0, 531.2, 531.4, 531.6, 532.0, 532.2, 532.4, 532.6, 533.0, 533.2, 533.4, 533.6, 578.0, 578.1, 578.9). In Taiwan, the prescriptions of clopidogrel in ACS patients is reimbursed by NHI for only 9 months, rather than 12 months according to current international guidelines. [1, 2] Therefore, all the patients were followed until 9 months of treatment, occurrence of clinical outcomes, or December 31. 2010, whichever occurred first.

For comparison of the baseline characteristics between the three groups, the 2-sample t-test was used for continuous variables and the chi-square test was employed for categorical variables. Pair-wise comparison approach using Cox proportional hazards regression model with adjustment for age, gender, AMI presentation, comorbidities and medications exposed was used to compare the relative risks of different clinical outcomes among the three groups. The test for proportional hazards assumption in Cox model was performed by Schoenfeld’s global test. If the proportional hazards assumption was violated, we would partition the relative risks as separate hazard ratios (HR's) within different follow-up time periods according to inspection of survival curves. DAPT, the standard treatment for AMI patients, was chosen a priori as the reference category. The survival curves of the three groups were illustrated by using the Kaplan-Meier method.

Even though professional medical associations in Taiwan have developed clinical guidelines concerning AMI management [22], variations in management of AMI among different levels of medical facilities could not be ignored. We utilized the shared frailty model to account for variability at hospital level under the assumption that patients clustered within the same medical facility shared the same level of frailty [23, 24]. Two shared frailty models were employed as sensitivity analyses. The first one controlled for 174 individual hospitals and the second one controlled for 11 different levels of hospitals where our study subjects were cared during their index AMI.

After application of selection criteria in the NHI claims database, we identified 1469 patients. Among them, 390 (27%) patients were prescribed with DAPT, 549 (37%) with aspirin only, and 530 (36%) with clopidogrel only (Fig. 1). The mean age was 88 years and the AMI presentations were comparable among three groups, with NSTEMI predominant but with less male patients in clopidogrel only group. Patients prescribed with aspirin only or clopidogrel only were more likely to have hypertension and chronic pulmonary disease but less likely to have diabetes and prescriptions of statins, ACEIs/ARBs, and beta-blockers at discharge in comparison with DAPT group. Clopidogrel only group were more likely to have history of peptic ulcer disease and had more prescriptions of proton pump inhibitors either at discharge or within the past 60 days prior to admission in comparison with DAPT group. Compared with DAPT group, aspirin only group were less likely to have prescriptions of statins within the past 60 days prior to admission (Table 1).

Additional file 1: eTable 1 also listed the distribution of study population in the 11 different levels of medical facilities among the 174 hospitals involved in this study.

Overall, the accumulated incidence was 13.8% for all-cause death, 7.9% for cardiovascular death and 4.2% for gastrointestinal bleeding (Table 2). Among three groups, patients treated with DAPT had numerically the lowest incidences of all-cause death (9.7%), cardiovascular death (6.1%), and gastrointestinal bleeding (3.1%) in comparison with aspirin only group (all-cause death, 12.9%; cardiovascular death, 8.7%; gastrointestinal bleeding, 4.9%) and clopidogrel only group (all-cause death, 17.9%; cardiovascular death, 8.5%; gastrointestinal bleeding, 4.3%). It is worth noting that the incidence of all-cause death in clopidogrel only group was nearly two-fold of that in DAPT group.

After adjusting for age, gender, AMI presentation, comorbidities and medications usage prior to and after index AMI, we found no difference in the risks of all-cause death (adjusted HR 1.20, 95% confidence interval [CI] 0.77–1.88, p = 0.42), cardiovascular death (adjusted HR 1.16, 95% CI 0.66–2.04, p = 0.60) and gastrointestinal bleeding (adjusted HR 1.66, 95% CI 0.77–3.61, p = 0.20) between aspirin only group and DAPT group. Nevertheless, patients treated with clopidogrel only possessed a higher risk of all-cause death (adjusted HR 1.50, 95% CI 1.00–2.25, p = 0.049) with comparable risks of cardiovascular death (adjusted HR 1.13, 95% CI 0.66–1.92, p = 0.65) and gastrointestinal bleeding (adjusted HR 0.96, 95% CI 0.44–2.09, p = 0.92) in comparison with patients treated with DAPT (Table 3). The Kaplan-Meier survival curves are illustrated in Fig. 2. The results remained unchanged in the two sensitivity analyses using the shared frailty model to control for characteristics of different levels of hospitals (Additional file 1: eTables 2 and 3).

Since the test for proportional hazards assumption was not satisfied concerning comparison of risk of all-cause death between clopidogrel only group and DAPT group (Table 3, p = 0.009), we chose the 7th day after the index date as cut-off time point for partition of separate HR’s based on inspection of the survival curves (Figs. 2 and 3). Of note, during the first 7 days, use of clopidogrel only was not associated with an increased risk of all-cause death in comparison with DAPT group (adjusted HR 1.08, 95% CI 0.41–2.80, p = 0.88). However, the risk of all-cause death in clopidogrel only group became statistically significant after the 7th day post-index date as compared with DAPT (adjusted HR 1.61, 95% CI 1.04–2.49, p = 0.035) (Fig. 3).

Our results and abstract were submitted to the 21st Cardiovascular Summit Transcatheter Cardiovascular Therapeutics Asia Pacific (TCTAP) and published in the supplement of Journal of the American College of Cardiology [25].

This study also has important limitations that should be acknowledged. Firstly, our results were limited by its non-randomized comparisons, selection bias and an uneven distribution of risk factors. Although we performed regression adjustment with all the covariates listed in Table 1, the results might be still confounded by other underlying conditions which were not included in analysis. Secondly, we estimated HR's of three different outcomes with six pairwise comparisons, which raised the question of statistical significance in multiple testing. While the most common way to control the type I error is using the Bonferroni correction to reduce the risk of false significance, p-value adjustments by the Bonferroni correction is also considered to be overly conservative and leads to a very high rate of false negatives [36, 37]. In addition, the eligible subjects in our analysis was quite small though there was a large number of patients in our database. It might result in wide confidence intervals of estimates. Therefore, we provided conventional statistical inference instead of adjusting the p-value to point out an interesting observation which needed further validation. In other words, the aim of our study was to generate a hypothesis or idea to inspire further clinical research. Thirdly, the NHI claims database does not include self-purchased drugs over the counter. Therefore, the possibility of misclassification could not be excluded particularly for self-purchases of aspirin over the counter in clopidogrel only group. However, any resulting misclassification was likely to result in a dilution of the true effect. Accordingly, our conclusion that clopidogrel users had a statistically significant higher risk of all-cause death compared with DAPT users seemed robust under this presumption. Fourthly, guidelines also recommend the use of prasugrel and ticagrelor combined with aspirin in ACS patients. Because prasugrel is not approved for use in Taiwan and ticagrelor is reimbursed by Taiwan NHI since 2013, these two medications were not examined in our study. Finally, the incidence of intracranial hemorrhage was too small to be included in our analysis.