Cardiovascular Outcomes Among Patients with Acute Coronary Syndromes and Diabetes: Results from ACS QUIK Trial in India

Study design and population

We examined patients presenting with AMI enrolled in the ACS-QUIK trial. The full details of the trial design were published elsewhere [8]. In brief, ACS QUIK was a large, multicenter, cluster-randomized, stepped-wedged pragmatic clinical trial that examined the role of a quality improvement tool kit on CV outcomes. A total of 63 hospitals located in Kerala, India, were randomized with a total of 21,374 patients enrolled between November 10, 2014, and November 9, 2016. Eligible patients had to be ≥18 years of age and present to one of the participating hospitals with either non-ST-segment myocardial infarction or ST-segment elevation myocardial infarction (STEMI) as defined by the Third Universal Definition of Myocardial infarction [9]. ACS QUIK was approved by the ethics boards of local, national, and international bodies and was approved by the Indian Health Ministry Screening Committee. All patients provided written informed consent prior to their participation.

In-hospital data collection

Patient-level in-hospital data were collected based on the 2013 ACCF/AHA key data elements and definitions for measuring the clinical management and outcomes of patients with acute coronary syndromes and coronary artery disease [10]. In-hospital data included the patient’s medical history, demographics, previous and concomitant medications, presenting vital signs and physical exam findings, and initial blood tests. Patients were recorded as having diabetes if they had a history of diabetes diagnosed and/or treated by a healthcare provider.

Therapies given on admission were recorded. This included the administration of aspirin, adjuvant antiplatelet therapy, beta-blockers, adjuvant anti-coagulation therapy, and primary reperfusion therapy. Primary reperfusion therapy for eligible patients with ST-segment elevation myocardial infarction included primary percutaneous coronary intervention or primary fibrinolytic therapy. Subsequent diagnostic angiography or PCI during admission were also recorded. Treatments received during hospitalization and discharge were also recorded.

Study outcomes

The primary outcome of this analysis was a composite endpoint of 30-day MACE including death, re-infarction, stroke, or major bleeding. Reinfarction was defined according to the Third Definition of Myocardial Infarction [9]. Major bleeding was defined using the Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Coronary Arteries (GUSTO) criteria [11]. We also analyzed 30-day death, 30-day CVD death, and in-hospital adverse events including in-hospital death, re-infarction, major bleeding, stroke, heart failure, cardiogenic shock, and cardiac arrest.

Statistical analysis

Categorical variables were summarized by their proportions and continuous variables were summarized by their means (SD) or medians (IQR) where appropriate. Proportions were compared between groups using a Chi² test. Normally distributed continuous variables were compared using Student t-test and skewed variables were compared using Mann-Whitney-U test.

Logistic regression was used to estimate the odds ratios and 95% confidence intervals for the associations between diabetes status and adverse outcomes. Standard errors of the estimated associations were adjusting for clustering by center. We presented unadjusted associations and performed hierarchical multivariable modeling whereby model 1 adjusted for clustering by center. Model 2 adjusted for clustering by center plus clinically important baseline factors including age, sex, smoking status, presence of STEMI, systolic blood pressure, heart rate, Killip class, initial creatinine, and troponin values. Model 3 adjusted for factors included in model 2 plus in-hospital treatments including administration of aspirin, anticoagulation, and PCI during admission. Model 3 was also adjusted for the quality improvement treatment arm that a given healthcare center was randomized to during the ACS-QUIK trial. Multiplicative interaction terms were calculated to test whether the association between diabetes and adverse outcomes differed by age, sex, history of hypertension, smoking status, presence of STEMI, receiving in-hospital aspirin or PCI during admission, and discharge medications. A 2-sided P-value < 0.05 was considered statistically significant. All analyses were performed using Stata Statistical Software version 16.1 (StataCorp LLC, TX).

Clinical characteristics on presentation

Among the 21,374 participants included in this analysis, the mean ± SD age was 60.1 ± 12.0 years and 24.3% were females. A total of 9,484 (44.4%) participants had diabetes. A total of 13,689 (64.0%) presented with a STEMI, and 13.6% presented with a Killip class of 2 or greater. Table 1 shows the demographic and clinical characteristics at baseline in the total population and stratified by diabetes status. At baseline, patients with diabetes were more likely to be older, female, non-smokers, and have a history of peripheral arterial disease. Patients with diabetes were more likely to have longer symptom-to-hospital arrival times (median time, 290 vs 225 minutes; P < 0.001), have elevated Killip class on presentation (proportion with Killip class ≥2, 20.1% vs. 11.1%; P < 0.001), but less likely to be diagnosed with a STEMI (57.7% vs. 69.1%; P < 0.001). The baseline demographic and clinical characteristics in patients with diabetes with and without STEMI are shown in supplemental Table 1.

Use of evidence-based therapies

Table 2 shows data on the use of evidence-based therapies in patients with and without diabetes. On admission, 97.9% of patients received aspirin and 98.2% received a second antiplatelet agent with no significant difference between those with and without diabetes. Adjuvant anticoagulation therapy was administered to 85.6% of participants but was more likely administered to those with diabetes than those without (84.9% vs. 86.1%; P = 0.012). The utilization of beta-blockers was low in the overall sample (40.1%) and did not differ by diabetes status.

Among those presenting with STEMI, 72.1% received reperfusion therapy. An equal proportion of STEMI patients with and without diabetes received primary PCI (49.1% vs 49.0%; P = 0.91). However, those with diabetes had significantly longer door-to-needle (median, 45 vs 43 minutes; P = 0.002) and door-to-balloon times (median, 91 vs 75 minutes; P < 0.001). In all comers (STEMI and non-STEMI patients), 12,681 (59.3%) patients had angiography during their hospital stay and 10,553 (49.4%) received PCI. Receiving a PCI during admission was less likely in those with diabetes compared to those without (47.5% vs 50.9%; P < 0.001).

At discharge, most patients (>95%) were prescribed aspirin, an adjuvant antiplatelet agent, statin therapy, and received smoking cessation counseling (for smokers) with no clinically meaningful difference in prescription patterns between those with and without diabetes except for prescribing aspirin, which was lower in patients with diabetes (97.6% vs 98.0; P = 0.048). Those with diabetes were more likely to be discharged on a beta-blocker but less likely to receive a referral to cardiac rehabilitation.

Adverse in-hospital and 30-day cardiac outcomes

The prevalence of in-hospital and 30-day adverse cardiac outcomes in those with and without diabetes is shown in Figure 1 (supplemental Table 2). Among those with and without diabetes, the incidence of heart failure was the most common in-hospital adverse cardiac event (8.4% vs 5.6%; P < 0.001). In those with diabetes, the second most frequent in-hospital adverse cardiac event was death (3.7%) followed by cardiac arrest (3.6%) and cardiogenic shock (3.4%). By contrast, among patients without diabetes, the second most common adverse in-hospital event was cardiogenic shock (3.2%) followed by cardiac arrest (2.6%) and death (2.5%).

Figure 1

At 30-day follow-up, MACE occurred in a total of 1,247 (5.9%) patients and death occurred in a total of 954 (4.5%) patients. CVD death was the most common type of death at 30-days comprising 97.3% of all deaths. MACE, death, CVD death, re-infarction, and stroke were all more prevalent in those with diabetes than those without diabetes. Supplemental Table 3 shows the results stratified by the presence of STEMI in patients with and without diabetes.

Table 3 shows univariable and multivariable adjusted odds ratios for the association between and in-hospital or 30-day adverse events. In an unadjusted model, compared to those without diabetes, individuals with diabetes had a statistically significant 50% higher odds of in-hospital death (OR, 1.5; 05% CI, 1.2 to 1.8), 47% higher odds of in-hospital re-infarction, (OR, 1.47; 95% CI 1.1 to 1.9), 53% higher odds of in-hospital heart failure (OR, 1.5; 95% CI, 1.3 to 1.8), and 39% higher odds for in-hospital cardiac arrest (OR, 1.4; 95% CI, 1.1 to 1.7). These associations remained statistically significant after adjusting for clinical presentation factors and therapies received during admission. Diabetes was not statistically associated with in-hospital stroke, major bleeding, or cardiogenic shock in any of the models.

Those with diabetes had a statistically significant 40% higher odds of 30-day MACE (OR, 1.4; 95% CI, 1.2 to 1.6), 46% higher odds of 30-day death (OR, 1.5; 95% CI, 1.3 to 1.7), 46% higher odds of 30-day CVD death (OR, 1.5; 95% CI, 1.3 to 1.7) and 41% higher odds of 30-day re-infarction (OR, 1.4; 95% CI, 1.1 to 1.8). These associations remained statistically significant after adjusting for baseline clinical factors (Model 2) and important therapies received during admission (Model 3). Moreover, diabetes was significantly associated with higher 30-day MACE after additional adjustment for therapies on discharge (Model 4) (adjusted OR, 1.28; 95% CI, 1.0 to 1.6).

Figure 2 shows the multivariable adjusted odds ratios for 30-day MACE and diabetes stratified by demographic and treatment factors. Two significant interactions were noted. First, diabetes was associated with higher 30-day MACE in those without hypertension but not in those with hypertension (P-interaction = 0.04). Second, the association between diabetes and higher MACE was stronger for those who were current smokers than for those who were not current smokers (P-interaction < 0.001). Supplemental Table 4 and supplementary Table 5 show the number of events according to diabetes status and stratified by history of hypertension and current smoking, respectively. Supplemental Figure 1 shows the multivariable adjusted probability of 30-day MACE across age in patients with and without diabetes stratified by history of hypertension and current smoking.

Figure 2

Differences in presentation and initial treatments

Consistent with prior studies, we show that patients with diabetes were more likely to be older and female [16]. Prior studies have shown that these factors together with diabetes are associated with more frequent atypical ischemic symptoms, which may explain why patients with diabetes in our study had significant delays in hospital presentation [18]. We also observe that patients with diabetes have a higher Killip class despite lower proportions of STEMI [17]. This has been observed in western ACS registries and may be due to the higher burden of ischemia in patients with diabetes or delayed hospital presentation in the face of ongoing myocardial necrosis due to myocardial infarction.

In patients with and without diabetes, the morbidity and mortality associated with STEMI is significantly reduced with prompt reperfusion [19]. Current guidelines advocate for achieving door-to-balloon times ≤90 minutes and, for non-PCI capable hospitals, a door-to-needle times ≤30 minutes when the anticipated time to mechanical reperfusion is >120 minutes. In our study, STEMI patients with and without diabetes were equally likely to receive primary PCI. However, STEMI patients with diabetes had significantly longer door-to-balloon times with more than half of those with diabetes having door-to-balloon times >90 min or door-to-needle times >45 min. These delays occurring after hospital presentation compound earlier delays from symptom onset-to-hospital presentation. Other studies have reported similar delays in reperfusion in those with diabetes [20, 21].

Among the various factors delaying reperfusion, the struggle to afford therapy costs is perhaps a critical issue in Kerala. In a substudy of ACS-QUIK addressing the microeconomics of AMI, 54% of patients experienced catastrophic health spending that exceeded 40% of the patient’s annual household expenditure, excluding food costs; 8% had to either sell or mortgage assets or borrow money to buffer the cost of AMI treatment [22]. As expected, these findings were more pronounced in those without baseline health insurance. In our study, those with diabetes were less likely to have health insurance and, therefore, more likely to face greater financial challenges, perhaps contributing to delays in reperfusion. Door-to-reperfusion (whether mechanical or pharmacologic) is an actionable measure of cardiovascular care. Our findings should catalyze institutional quality improvement aiming at achieving timely reperfusion in patients with diabetes. Enhancing access to healthcare insurance to facilitate the timely delivery of AMI therapies should be considered, especially among higher-risk populations such as those with diabetes.

Differences in outcomes between those with and without diabetes

The current study confirms the significant morbidity and mortality associated with diabetes reported in prior studies and extends these findings to patients in LMIC [20, 23, 24]. Our findings show that diabetes is independently associated with higher adverse outcomes and contradicts the findings of other studies suggesting diabetes was not associated with increased risk after adjusting for the lower utilization of evidence-based treatments and PCI in patients with diabetes [25, 26]. Diabetes has been associated with higher coronary artery disease burden and extreme serum glucose values at the time of AMI, which were associated with worse prognosis [26]. Diabetes is also associated with a higher incidence of heart failure, and cardiac arrest after AMI [27, 28, 29]. These associations were confirmed in the current study. Our findings highlight the importance of emphasizing primary and secondary CVD prevention in patients with diabetes. Future studies should examine the efficacy of different strategies aiming at enhancing primary and secondary prevention among patients with diabetes in LMIC.

An interesting observation of the current analysis was the significant interaction between diabetes and a history of hypertension or smoking with regards to 30-day MACE. In those without hypertension, diabetes was associated with higher 30-day MACE. However, in those with hypertension, events were not different in those with and without diabetes. Both hypertension and diabetes are strong CVD risk factors and are associated with a pro-thrombotic state and impaired left ventricular function [30, 31, 32]. Data from observational studies suggest a synergistic interaction between hypertension and diabetes with regards to stroke [33], kidney function worsening, and heart failure [34]. By contrast, a pooled analysis of four randomized clinical trials including 28,771 patients with MI and evidence of heart failure or left ventricular dysfunction found no significant interaction between history of diabetes and hypertension with regards to adverse cardiac outcomes [35]. Our finding of a sub-additive interaction between hypertension and diabetes was unexpected. Both hypertension and diabetes may be manifestations of an underlying adverse cardiometabolic state that is prevalent in the study population. As a result, our findings may be due to collinearity among these variables contributing to the risk of adverse events.

The presence of both diabetes and current smoking was adversely synergistic on 30-day MACE. Data from Nurses’ Health Study suggested that the association between diabetes and CHD was weaker in female smokers than non-smokers, while other studies including both males and females have shown no interaction [36, 37, 38]. Our finding is important, especially that tobacco use is growing the fastest in LMICs and was present in 22% of those with diabetes in our sample [39]. This interaction highlights the importance of counseling patients with diabetes on tobacco cessation. Reassuringly, in our study, receiving tobacco cessation advice on discharge was equally high (>95%) in those with and without diabetes. However, the percentage of those achieving tobacco cessation at 30-day follow-up is unclear. Future studies should investigate the effectiveness of implementing post-MI smoking cessation interventions in low- or middle-income countries.

Non-STEMI patients with diabetes are at high risk of cardiac events

Among non-STEMI patients, we observed higher in-hospital and 30-day adverse outcomes in those with diabetes compared to those without diabetes. These findings are consistent with prior studies of patients with non-STEMI [16, 40, 41]. Furthermore, non-STEMI patients with diabetes have as high in-hospital and 30-day adverse outcomes as STEMI patients without diabetes. In the acute phase of non-STEMI management, the current AHA/ACC guideline recommends that clinical decisions for medical therapy and revascularization should be similar in those with and without diabetes [42].

Clinical risk scores to risk stratify non-STEMI patients such as the TIMI risk score or GRACE risk calculator patients do not account for diabetes status [43, 44]. Our findings that diabetes was independently associated with the risk of adverse outcomes suggests the importance of accounting for diabetes status at the time of risk stratification. Moreover, our findings that the proportion of in-hospital heart failure was the highest in non-STEMI patients with diabetes highlights the significant morbidity associated with diabetes that is not captured by risk scores.

Strengths and Limitations

A major strength of our study is that it leverages data from the largest trial in India to date. Furthermore, the high prevalence of diabetes, smoking, and hypertension allow for exploring their interaction with regards to 30-day MACE. Our study has some notable limitations. This was a post-hoc analysis of data from ACS QUIK. Although we used multivariable regression to adjust for potential confounding, our analysis is outside of randomization making unmeasured or residual confounding possible. Second, we analyzed the difference in tobacco cessation advice and cardiac rehabilitation referral between those with and without diabetes. However, data on the percent of patients who were successful in achieving tobacco cessation or receiving cardiac rehabilitation are missing; thus, the association between tobacco cessation or cardiac rehabilitation and 30-day MACE cannot be assessed. Cardiac rehabilitation in India remains grossly underutilized and the number of cardiac rehabilitation centers remains insufficient; as such, efforts to improve the uptake of cardiac rehabilitation in India are needed [45]. Furthermore, we have no data on adherence to discharge medication or the patient’s ability to schedule post-discharge follow-up, both important predictors of lower adverse events after myocardial infarction.

Third, while we report on the percentage of patients receiving PCI, data on the angiographic difference between those with and without diabetes were not collected. Lastly, although the ACS QUIK was a cluster-randomized clinical trial at the hospital center level, informed consent was required for 30-day follow-up. This might have resulted in some recruitment bias. However, Huffman et al. showed that baseline characteristics were balanced in those with and without 30-day follow, except for troponin values which were higher in those missing 30-day follow-up (median, 1.3 vs 4.6 ng/ml) [46].