Trends for First-Time Hospitalization for AMI
In our study, a large (>60,000 patients) regional-wide population-based analysis, we observed a progressive implementation of evidence-based and guideline-recommended AMI treatments. In particular, the rate of in-hospital cardiac interventions such as coronary angiography and PCI increased considerably during the period of observation. The use of evidenced-based pharmacological therapies also increased in our population. Notably, these changes in practice were accompanied by a 3.3% absolute reduction of in-hospital mortality, by a 4.1% absolute reduction of in-hospital cardiovascular mortality, by a 2.4% absolute reduction of in-hospital HF.
The progressive decline in early mortality over time reported in our regional population-based analysis is consistent with other American and European observations. Obviously, this finding is largely explained by major improvements in the delivery of care for AMI, including the more frequent use of revascularization techniques and adjunctive evidenced-based therapies. However, it may also be related to changes in patient behavior such as faster calls for medical assistance after symptom onset and more frequent use of prehospital triage with direct referral for primary percutaneous transluminal coronary angioplasty as well as changes in regional organization of care with concentration of care provision in a smaller number of centers treating larger numbers of patients according to the Hub and Spoke model. However, it seems worth noting that also the increasing detection over time of n-STEMI less severe infarctions with troponin testing could contribute to an only apparent decline in AMI short-term mortality. Furthermore, as observed in our population, important changes in clinical characteristics potentially influencing the outcome of the patients can occur during long observation periods. In particular, in our first AMI population, in contrast to the increment of important comorbidities, such as diabetes, hypertension, renal failure, and so on, the median age of AMI presentation progressively decreased as well as the prevalence of peripheral vascular and chronic liver diseases during the 8-year period of observation. This finding is consistent with other recent European registries showing a progressive decrease in the age of patients hospitalized with AMI, secondary to an increased proportion of younger female patients.
The change in clinical characteristics observed over time obviously makes it difficult to interpret the observed trend of over-time mortality. Nevertheless, the fact that the mortality reduction remains significant even after adjustment for different confounders (intriguingly, this decrease in mortality became more apparent following AMI network implementation) highlights the favorable prognostic role played by the over-time increase of evidence-based treatments.
In our analysis concerning a quite old AMI population (median age 76 years) without a history of HF, in-hospital and 1-year postdischarge HF developed, respectively, in almost 22% and 8% of the patients. Quite surprisingly, our study showed a temporal paradox: in-hospital survival improved over the 8 years; but the incidence of early and late HF did not decrease consensually, offsetting the reductions observed in early mortality. Of note, similar results were recently reported in a wide Canadian population cohort of ≥65-year-old first-AMI patients.
A further disappointing finding of the current study consists in the observation that, in addition to the HF incidence, also the adjusted risk of postdischarge 1-year mortality and reinfarction did not show a significant temporal decline over the 8 years of observation. The reason for this apparent disconnect are not clear and are probably subordinate to several factors such as the following: (1) too short postdischarge period of observation; (2) progressive increment of very old fragile patients saved by the intensive in-hospital care, but discharged with short life expectancy; (3) detrimental role of the observed suboptimal postdischarge medical therapy adherence (at best, β-blockers, statins, and agents acting on the renin-angiotensin system were taken by about 80% of the patients); and (5) detrimental role of the increase over time of bleeding. Of note, this suggestive hypothesis was strongly supported by our sensitivity analysis that disclosed in-hospital bleeding as an independent predictor of cardiovascular and all-cause postdischarge mortality.
A relevant finding of our study is that the increased aggressiveness of care was associated with an increase of in-hospital and postdischarge bleeding. In particular, the incidence of in-hospital bleeding almost doubled (from 3.7% to 5.4%, relative increase of 45.9%) during the 8 years of observation. Because of the design of our study, we could only detect the incidence of clinical (requiring hospitalization) bleedings (such as gastrointestinal, intracranial, urinary tract, airway, intraocular or intraarticular bleeding, hemopericardium) or the need for blood transfusions; but we could not classify the severity of such bleedings. In contrast with our data, in an 8-year (2000–2007) analysis, the GRACE registry recently reported a temporal reduction of major bleedings (from 2.6% to 1.8%), despite an increasing use of low–molecular weight heparin and thienopyridine therapies and PCIs. This discrepancy probably lies in differences existing between the 2 studies concerning the design of the analyses (administrative data vs voluntary participation registry data), classification of bleeding episodes (clinically significant bleeding vs major bleeding), and clinical characteristics of the patients. Of note, a significant temporal increment, over a period of 12 years, of the risk of severe bleeding was recently documented in STEMI patients enrolled in the Swedish RISK-HIA registry.
Overall, during the 8-year period of observation (2002–2009), the standardized annual incidence of AMI in the Emilia-Romagna region progressively increased up to 2004 and decreased afterwards. On the other hand, the incidence of STEMI progressively decreased up to 2009 and was partially counterbalanced by the specular increase of n-STEMI. Obviously, this finding could reflect a true epidemiological change in acute coronary syndromes due to a greater benefit of primary prevention measures for patients at risk of STEMI or changes in cardiovascular risk factors that may have differentially promoted the development of n-STEMI. However, the most likely explanation for our finding is that the increased n-STEMI hospitalization rate is mainly artificially and consequent to the expanded use of cardiac biomarkers like troponin I and T.
Few previous studies have separately analyzed trends in the outcomes of STEMI and n-STEMI patients. Although similar in-hospital mortality rates have been described in n-STEMI versus STEMI patients, other studies and our findings have shown higher in-hospital case-fatality rates among STEMI patients. Notably, both STEMI and n-STEMI populations presented significant reductions in early mortality over time, although n-STEMI showed an in-hospital case fatality reduction over time practically double that of STEMI. Although this result appears worthy of note, it should however be interpreted with caution because it may be biased by the progressive increase in troponin testing–related detection of n-STEMI less severe infarctions.
Notably, n-STEMI patients showed worse (30% higher) 1-year postdischarge unadjusted clinical outcomes. Although we did not specifically analyze baseline clinical characteristics of STEMI and n-STEMI populations, as already reported, the more unfavorable baseline clinical profile (older age, higher proportion of hypertensive and diabetic patients, etc) of n-STEMI patients can explain this finding to a large degree.
Several issues should be considered in interpreting our results. The population-based design in a large (>60,000 patients) population of an Italian region with universal health care greatly reduced the risk of selection biases. Moreover, this study reflects the variations over time in acute coronary care in unselected AMI patients evaluated for a quite long (8 years) period of observation. Furthermore, as reported above, all patients were followed until death or end of follow-up; and hence, no one had incomplete data. The main potential limitation of the study is the reliance on the accuracy and validity of routine hospital discharge data. Unfortunately, we do not have specific review data regarding random samples carried out in our population. However, previously reported international experiences suggested that the administrative diagnosis of AMI can be considered sufficiently accurate (accuracy rate >95%).
During the study period, the Emilia-Romagna Regional Health Agency had access to computerized prescription records regarding only medications dispensed by territorial pharmacies after hospital discharge. We therefore had no information on medical treatments (including thrombolysis) administered during the in-hospital stay or at discharge, or on the clopidogrel postdischarge administration rate (in our region, clopidogrel was dispensed for hospitalized and discharged patients only by nosocomial pharmacies).
As reported above, the baseline clinical and demographic characteristics of the study population changed significantly over time. To overcome this bias when analyzing time trends for major clinical outcomes, we determined crude and adjusted (using multivariate weighted Cox regression analysis) data. Nevertheless, considering that no type of statistical adjustment (including the inability to fully adjust for known and unknown confounders) can completely overcome all pitfalls of nonrandomized comparisons, our results should be interpreted with caution. Finally, this study focused on AMI patients living in an Italian region with universal health care coverage including a territorial network for the treatment of AMI. It should be underlined that the results of the present study may not be automatically replicated in different settings.
Discussion
In our study, a large (>60,000 patients) regional-wide population-based analysis, we observed a progressive implementation of evidence-based and guideline-recommended AMI treatments. In particular, the rate of in-hospital cardiac interventions such as coronary angiography and PCI increased considerably during the period of observation. The use of evidenced-based pharmacological therapies also increased in our population. Notably, these changes in practice were accompanied by a 3.3% absolute reduction of in-hospital mortality, by a 4.1% absolute reduction of in-hospital cardiovascular mortality, by a 2.4% absolute reduction of in-hospital HF.
The progressive decline in early mortality over time reported in our regional population-based analysis is consistent with other American and European observations. Obviously, this finding is largely explained by major improvements in the delivery of care for AMI, including the more frequent use of revascularization techniques and adjunctive evidenced-based therapies. However, it may also be related to changes in patient behavior such as faster calls for medical assistance after symptom onset and more frequent use of prehospital triage with direct referral for primary percutaneous transluminal coronary angioplasty as well as changes in regional organization of care with concentration of care provision in a smaller number of centers treating larger numbers of patients according to the Hub and Spoke model. However, it seems worth noting that also the increasing detection over time of n-STEMI less severe infarctions with troponin testing could contribute to an only apparent decline in AMI short-term mortality. Furthermore, as observed in our population, important changes in clinical characteristics potentially influencing the outcome of the patients can occur during long observation periods. In particular, in our first AMI population, in contrast to the increment of important comorbidities, such as diabetes, hypertension, renal failure, and so on, the median age of AMI presentation progressively decreased as well as the prevalence of peripheral vascular and chronic liver diseases during the 8-year period of observation. This finding is consistent with other recent European registries showing a progressive decrease in the age of patients hospitalized with AMI, secondary to an increased proportion of younger female patients.
The change in clinical characteristics observed over time obviously makes it difficult to interpret the observed trend of over-time mortality. Nevertheless, the fact that the mortality reduction remains significant even after adjustment for different confounders (intriguingly, this decrease in mortality became more apparent following AMI network implementation) highlights the favorable prognostic role played by the over-time increase of evidence-based treatments.
HF Paradox and Late Events
In our analysis concerning a quite old AMI population (median age 76 years) without a history of HF, in-hospital and 1-year postdischarge HF developed, respectively, in almost 22% and 8% of the patients. Quite surprisingly, our study showed a temporal paradox: in-hospital survival improved over the 8 years; but the incidence of early and late HF did not decrease consensually, offsetting the reductions observed in early mortality. Of note, similar results were recently reported in a wide Canadian population cohort of ≥65-year-old first-AMI patients.
A further disappointing finding of the current study consists in the observation that, in addition to the HF incidence, also the adjusted risk of postdischarge 1-year mortality and reinfarction did not show a significant temporal decline over the 8 years of observation. The reason for this apparent disconnect are not clear and are probably subordinate to several factors such as the following: (1) too short postdischarge period of observation; (2) progressive increment of very old fragile patients saved by the intensive in-hospital care, but discharged with short life expectancy; (3) detrimental role of the observed suboptimal postdischarge medical therapy adherence (at best, β-blockers, statins, and agents acting on the renin-angiotensin system were taken by about 80% of the patients); and (5) detrimental role of the increase over time of bleeding. Of note, this suggestive hypothesis was strongly supported by our sensitivity analysis that disclosed in-hospital bleeding as an independent predictor of cardiovascular and all-cause postdischarge mortality.
Bleeding in AMI Patients
A relevant finding of our study is that the increased aggressiveness of care was associated with an increase of in-hospital and postdischarge bleeding. In particular, the incidence of in-hospital bleeding almost doubled (from 3.7% to 5.4%, relative increase of 45.9%) during the 8 years of observation. Because of the design of our study, we could only detect the incidence of clinical (requiring hospitalization) bleedings (such as gastrointestinal, intracranial, urinary tract, airway, intraocular or intraarticular bleeding, hemopericardium) or the need for blood transfusions; but we could not classify the severity of such bleedings. In contrast with our data, in an 8-year (2000–2007) analysis, the GRACE registry recently reported a temporal reduction of major bleedings (from 2.6% to 1.8%), despite an increasing use of low–molecular weight heparin and thienopyridine therapies and PCIs. This discrepancy probably lies in differences existing between the 2 studies concerning the design of the analyses (administrative data vs voluntary participation registry data), classification of bleeding episodes (clinically significant bleeding vs major bleeding), and clinical characteristics of the patients. Of note, a significant temporal increment, over a period of 12 years, of the risk of severe bleeding was recently documented in STEMI patients enrolled in the Swedish RISK-HIA registry.
STEMI versus n-STEMI Subanalysis
Overall, during the 8-year period of observation (2002–2009), the standardized annual incidence of AMI in the Emilia-Romagna region progressively increased up to 2004 and decreased afterwards. On the other hand, the incidence of STEMI progressively decreased up to 2009 and was partially counterbalanced by the specular increase of n-STEMI. Obviously, this finding could reflect a true epidemiological change in acute coronary syndromes due to a greater benefit of primary prevention measures for patients at risk of STEMI or changes in cardiovascular risk factors that may have differentially promoted the development of n-STEMI. However, the most likely explanation for our finding is that the increased n-STEMI hospitalization rate is mainly artificially and consequent to the expanded use of cardiac biomarkers like troponin I and T.
Few previous studies have separately analyzed trends in the outcomes of STEMI and n-STEMI patients. Although similar in-hospital mortality rates have been described in n-STEMI versus STEMI patients, other studies and our findings have shown higher in-hospital case-fatality rates among STEMI patients. Notably, both STEMI and n-STEMI populations presented significant reductions in early mortality over time, although n-STEMI showed an in-hospital case fatality reduction over time practically double that of STEMI. Although this result appears worthy of note, it should however be interpreted with caution because it may be biased by the progressive increase in troponin testing–related detection of n-STEMI less severe infarctions.
Notably, n-STEMI patients showed worse (30% higher) 1-year postdischarge unadjusted clinical outcomes. Although we did not specifically analyze baseline clinical characteristics of STEMI and n-STEMI populations, as already reported, the more unfavorable baseline clinical profile (older age, higher proportion of hypertensive and diabetic patients, etc) of n-STEMI patients can explain this finding to a large degree.
Study Strengths and Limitations
Several issues should be considered in interpreting our results. The population-based design in a large (>60,000 patients) population of an Italian region with universal health care greatly reduced the risk of selection biases. Moreover, this study reflects the variations over time in acute coronary care in unselected AMI patients evaluated for a quite long (8 years) period of observation. Furthermore, as reported above, all patients were followed until death or end of follow-up; and hence, no one had incomplete data. The main potential limitation of the study is the reliance on the accuracy and validity of routine hospital discharge data. Unfortunately, we do not have specific review data regarding random samples carried out in our population. However, previously reported international experiences suggested that the administrative diagnosis of AMI can be considered sufficiently accurate (accuracy rate >95%).
During the study period, the Emilia-Romagna Regional Health Agency had access to computerized prescription records regarding only medications dispensed by territorial pharmacies after hospital discharge. We therefore had no information on medical treatments (including thrombolysis) administered during the in-hospital stay or at discharge, or on the clopidogrel postdischarge administration rate (in our region, clopidogrel was dispensed for hospitalized and discharged patients only by nosocomial pharmacies).
As reported above, the baseline clinical and demographic characteristics of the study population changed significantly over time. To overcome this bias when analyzing time trends for major clinical outcomes, we determined crude and adjusted (using multivariate weighted Cox regression analysis) data. Nevertheless, considering that no type of statistical adjustment (including the inability to fully adjust for known and unknown confounders) can completely overcome all pitfalls of nonrandomized comparisons, our results should be interpreted with caution. Finally, this study focused on AMI patients living in an Italian region with universal health care coverage including a territorial network for the treatment of AMI. It should be underlined that the results of the present study may not be automatically replicated in different settings.
Source...