OUP user menu

Temporal trends of the gaps in post-myocardial infarction secondary prevention strategies of co-morbid and elderly populations vs. younger counterparts: an analysis of three successive cohorts between 2003 and 2008

Alejandro Macchia, Marilena Romero, Antonio D'Ettorre, Javier Mariani, Gianni Tognoni
DOI: http://dx.doi.org/10.1093/eurheartj/ehr410 515-522 First published online: 16 November 2011


Aims Epidemiological studies reported two contrasting trends: on one hand, a significant improvement in the use of evidence-based treatments of patients discharged with a myocardial infarction (MI). On the other hand, the increasing number of elderly and co-morbid patients who are usually less treated. The aim of this study is to examine whether improvements in the treatment of MI are homogeneously distributed throughout all subgroups of patients.

Methods and results Based on record linkage of administrative registers, 21 423 patients discharged with MI in three different periods (2003, 2005, and 2007), were identified and followed up for major clinical events up to 1 year. Using as a reference temporal category those patients discharged in 2003 (odds ratios, 95% confidence intervals) and as a demographic category male patients aged ≤75 years (1.00), the study identified: in-hospital mortality significantly decreased in all periods and in all groups of patients; out-of-hospital mortality decreased only in younger patients and not in older patients; prescription of evidence-based treatments increased in all periods for all patients; however, the magnitude of improvement was mostly concentrated in younger patients.

Conclusion Although there was a mean improvement in the treatment and outcome of patients discharged from an MI, most of these benefits were strongly concentrated in younger, healthier patients. Old and co-morbid populations—although representing a substantial proportion of the burden of disease—received significant less attention and barely improved their survival.

  • Epidemiology
  • Cardiovascular prevention
  • Co-morbidities
  • Elderly


Large clinical trials have shown that several pharmacological and non-pharmacological treatments reduce mortality and morbidity in patients with myocardial infarction (MI),18 a success that must be considered a major achievement of modern medicine.

Published epidemiological surveys, however, document two contrasting trends beyond this overall positive picture. On one hand, many reports indicate that the appropriate use of evidence-based treatment strategies of MI has substantially improved over the last years.912 On the other hand, it is also noted that vulnerable and co-morbid populations often receive less or poorer treatment than their younger and ‘healthier’ counterparts.1315 This last observation is particularly disturbing, since increasing numbers of patients with MI tend to be more vulnerable because of older age and of co-morbid conditions.16,17 To the best of our knowledge, epidemiological reports have focused their attention on these two problems separately, without exploring specifically the interaction between temporal trends of improvement in the quality of treatments and under-treatment of old and co-morbid populations. The aim of this paper is to examine through successive cohorts of patients, representing the same background population, whether improvements in the treatment of MI were homogeneously distributed throughout all subgroups of patients.


Administrative databases covering: hospital discharge records, and prescription and personal data of a representative sample and of 12% of the Italian population distributed across 22 Local Health Areas were used in order to create three cohorts of patients hospitalized for MI during the Years 2003, 2005, and 2007. According to a validated methodology,18 patients were identified on the basis of a discharge diagnosis of acute MI (International Classification of Diseases, Ninth Revision, code 410) and followed up with their prescriptions of pharmacological treatments for 1 year. Hospital discharge records included information on primary diagnoses and up to five coexisting clinical diagnoses, anatomical localization of MI, procedures performed, dates of admission and discharge (index date), and in-hospital death.

Using encrypted affiliation numbers, data for this cohort were linked with the local health authority drug claims database. The latter data set provides prescription information (reimbursed by the National Health System); drugs are coded according to the Anatomical Therapeutic Chemical (ATC) Classification19 and qualified with respect to dosages, date of first prescription, and duration of exposure.

Definition of cohorts

In each index year, the cohort included all consecutive patients admitted for MI, qualified in terms of co-morbid conditions as resulting from their clinical (cardiovascular and non-cardiovascular) history over the 12 months preceding the index hospitalization; chronic exposure to pharmacological treatments was assumed as an identifier of underlying clinical conditions. In particular, cardiovascular co-morbidities included the presence of previous stroke, transient ischaemic attack (TIA), peripheral vascular disease, embolic episode, atrial fibrillation, hypertension, congestive heart failure (CHF), previous MI, and diabetes. Non-cardiovascular conditions included malignancy, hospitalization for major bleeding, chronic obstructive pulmonary disease, and depression.

Each cohort was first analysed for in-hospital and 1-year mortality; patients surviving up to 1 year were the cohort on which exposure to the recommended pharmacological treatments was evaluated.

Prescriptions of aspirin, clopidogrel, statins, β-blockers, angiotensin-converting enzyme inhibitors (ACE-Is)/angiotensin receptor blockers (ARBs), and n-3 polyunsaturated fatty acids (PUFA) defined the treated cohort if these drugs were prescribed at any time between discharge and the subsequent 365 days. Patients not receiving any of the above were classified as non-treated.

Statistical analysis

Continuous data are presented as means and standard deviation, and categorical data as counts and percentages. To assess temporal trends, patients were grouped for biennial periods according to the date of hospital discharge and of the follow-up period (2003–04, 2005–06, and 2007–08). Demographic, clinical, in-/out-of-hospital mortality and drug prescription patterns among these three groups were compared using the Cochrane–Armitage test for linear trends.

To assess vulnerable populations, patients were classified into four groups according to age (≥75 vs. <75 years) and sex. Using these criteria, Group 1 (reference category for analyses) were male patients with <75 years, Group 2 were females with <75 years, Group 3 were males with ≥75 years, and Group 4 were females with ≥75 years.

The rate of death between (reference category: males patients <75 years of age) and within groups (reference category: patients hospitalized in 2003 and followed-up in 2004) was examined and compared using adjusted logistic regression models.

To evaluate the probability of receiving evidence-based therapies, similar analyses were conducted, in this case limited to patients who were discharged alive and remained alive for at least 1 year. These analyses were conducted for each of the agents under study.

All comparisons and regression models were adjusted for the following variables: age, sex, previous CHF, diabetes, stroke, TIA, atrial fibrillation, peripheral vascular disease and type (Q-wave vs. non-Q-wave), and location of index MI. Additionally, correction models included the presence of non-cardiovascular co-morbidities: chronic obstructive pulmonary disease, depression and malignancy, and in-hospital revascularization (i.e. percutaneous coronary intervention and coronary artery bypass grafting).

All comparisons are reported using fully adjusted odds ratios (OR) with their corresponding 95% confidence intervals (CI). Analyses were performed with the use of SAS software, version 9.1 (SAS Institute).


A total of 26 367 patients were admitted for MI (8178 in 2003, 8837 in 2005, and 9352 in 2007). Of these, 11 864 were males ≤75 years (Group 1), 3689 females ≤75 years (Group 2), 5193 males >75 years (Group 3), and 5621 females >75 years (Group 4).

Fatality rates

A total of 2107 (8.0%) of the 26 367 patients died during the index hospitalization. The rate of in-hospital mortality decreased significantly from 8.7 to 7.3% from 2003 to 2007 (OR = 0.76; 95% CI 0.68–0.86, P < 0.001) (Figure 1A). Although the magnitude of this reduction was higher among younger patients (OR = 0.60; 95% CI 0.46–0.78) compared with older subjects (OR = 0.82; 95% CI 0.68–0.98), all groups experienced a significant decline in mortality over time (Figure 1B). At the same time, out-of-hospital mortality decreased from 11.3% in 2003 to 10.4% in 2007 (OR = 0.86; 95% CI 0.78–0.95, P = 0.004), albeit this reduction was significant only among younger patients (Figure 1A and B).

Figure 1

Rates for in-hospital (lower part of boxes) and out-of-hospital (upper part of boxes) mortality for different periods (A) and different groups (B). Lower notation refers to odds ratio for in-hospital mortality and upper notation for out-of hospital mortality.

Drug exposure in the surviving cohort

Overall, 21 423 patients surviving at least 1 year after discharge were identified and constitute the three biennial cohorts: 6552 (2003–04), 7182 (2005–06), and 7689 (2007–08). Significant changes occurred in both demographic and treatment characteristics of patients (Table 1). The mean age of patients increased from 68.5 years in 2003/04 to 69.8 years in 2007/08. The proportion between males and females remained stable at 2:1. As expected, the proportion of patients with NQWMI increased from 25.7 to 38.2%.

View this table:
Table 1

Characteristics of patients discharged with diagnosis of acute myocardial infarction and survived at least 1 year after discharge by different periods

2003 (n = 6552)2005 (n = 7182)2007 (n = 7689)P-value for trend
 Mean (all)67.6 (12.7)68 (12.9)68.5 (12.9)<0.0001
 Mean (males)64.8 (12.4)65 (12.4)65.6 (12.6)
 Mean (females)73.4 (11.2)74.3 (11.5)74.4 (11.5)
Age >80, n (%)1158 (17.7)1376 (19.2)1637 (21.3)<0.0001
Males4485 (68.5)4581 (67.5)5159 (67.1)0.218
Females2067 (31.5)2331 (32.5)2530 (32.9)
MI location
 Anterior2163 (33.0)2073 (28.9)2072 (26.9)<0.0001
 Inferior2003 (30.6)1981 (27.6)1907 (24.8)
 Lateral154 (2.4)137 (1.9)129 (11.7)
 Posterior27 (0.4)23 (0.3)14 (0.2)
 Non-Q-wave MI1684 (25.7)2417 (33.7)2935 (38.2)
 Other521 (8)551 (7.7)632 (8.2)
In-hospital procedures
 Primary angioplasty1284 (19.6)2086 (29)2811 (36.3)<0.0001
 CABG16 (0.2)27 (0.4)12 (0.2)0.251
Previous co-morbidities
 Congestive heart failure673 (10.3)906 (12.6)1110 (14.4)<0.0001
 Hypertension3001 (45.8)3892 (54.2)4490 (58.4)<0.0001
 Diabetes1293 (19.7)1686 (23.5)1890 (24.6)<0.0001
 Chronic obstructive pulmonary disease476 (7.3)598 (8.3)731 (9.5)<0.0001
 Depression158 (2.4)334 (4.7)454 (5.9)<0.0001
 Treated dyslipidaemia781 (11.9)1201 (16.7)1595 (20.7)<0.0001
 Myocardial infarction155 (2.4)199 (2.8)191 (2.5)0.706
 Atrial fibrillation124 (1.9)157 (2.2)184 (2.4)0.042
 Neoplasm115 (1.8)140 (1.9)150 (2)0.405
In-hospital complications
 Heart failure894 (13.6)1003 (14.0)1074 (14.0)0.588
 Atrial fibrillation523 (8)586 (8.2)663 (8.6)0.161
Treatments within 1 year
 Aspirin5004 (76.4)6123 (85.3)6593 (85.7)<0.0001
 Statins4387 (67.0)5694 (79.3)6206 (80.6)<0.0001
 β-Blockers3883 (59.3)4983 (69.4)5475 (71.2)<0.0001
 ACE-inhibitors or ARBs4791 (73.1)5874 (81.8)6311 (82.1)<0.0001
 n-3 PUFA995 (15.2)1999 (27.8)2207 (28.7)<0.0001
 Clopidogrel893 (13.6)3383 (47.1)4461 (58.0)<0.0001

Baseline co-morbidities rose significantly during the 3 periods of observation. Congestive heart failure rose from 10 to 14%, hypertension from 46 to 58%, diabetes from 20 to 25%, chronic obstructive pulmonary disease from 7.3 to 9.5%, and depression from 2.4 to 5.9% (all P < 0.0001). The use of primary angioplasty increased from 19.6 to 36.3% (P < 0.0001) (Table 1).

Overall, prescription rates of all drugs under study rose during the three periods considered: aspirin prescription from 76.4% in 2003/04 to 85.7% in 2007/08; clopidogrel from 13.6 to 58.0%; statins from 67 to 80.6%; β-blockers from 59.3 to 71.2%. Treatment with either ACE-I or ARB increased from 73.1 to 82.1 and n-3 PUFA from 15.2 to 28.7% (Table 1).

The massive increase in clopidogrel is not further considered as it reflects simply the effect of its admission to reimbursement for the short post-MI period, but not for long-term secondary prevention.

Subgroups of patients

Of the 21 423 surviving patients, 10 980 were males ≤75 years; 3266 females ≤75 years; 3515 males >75 years and 3662 females >75 years. When considering the specific patient subgroups, it was clear that there was a significantly lower adjusted probability that older, female patients would be treated with evidence-based therapies (Figure 2).

Figure 2

Prescription patterns of recommended drugs within and between groups in three different periods.

When compared with men aged ≤75 years (OR = 1.00), the adjusted probabilities for being prescribed aspirin were lower for women aged >75 years (OR = 0.45; 95% CI 0.40–0.49; P < 0.0001). This difference was roughly the same in 2003–04, with an OR = 0.47 (95% CI 0.40–0.55) of receiving aspirin for women >75 years vs. men ≤75 years, and in 2007–08, OR = 0.40 (95% CI 0.34–0.48). The same trend can be seen for statins, β-blockers, ACE-Is/ARBs, and n-3 PUFA (Table 2).

View this table:
Table 2

Patients treated with recommended drugs in different groups and different periods

Men ≤75 years (n = 10 980)Women ≤75 years (n = 3266)Men >75 years (n = 3515)Women >75 years (n = 3662)
n%OR (95% CI)an%OR (95% CI)an%OR (95% CI)an%OR (95% CI)a
 All periods955687.01.00271883.20.72 (0.65–0.81)275878.50.60 (0.54–0.67)268873.40.45 (0.40–0.49)
 2003–04278180.41.0082278.30.83 (0.70–0.99)73371.50.65 (0.55–0.71)66865.70.47 (0.40–0.55)
 2005–06331589.61.0093486.20.73 (0.60–0.90)92780.60.55 (0.45–0.67)94775.90.43 (0.36–0.51)
 2007–08346090.61.0096284.90.60 (0.49–0.73)109881.90.57 (0.47–0.69)107376.80.40 (0.34–0.48)
 All periods939685.61.00265081.10.70 (0.63–0.78)222363.20.31 (0.28–0.34)201855.10.22 (0.20–0.24)
 2003–04269878.01.0077073.30.72 (0.61–0.85)49948.70.28 (0.24–0.32)42041.30.20 (0.17–0.23)
 2005–06330289.21.0092685.50.71 (0.58–0.87)76266.30.28 (0.23–0.33)70456.40.19 (0.16–0.22)
 2007–08339688.91.0095484.20.66 (0.54–0.81)96271.80.36 (0.30–0.43)89464.00.24 (0.20–0.28)
 All periods801073.11.00241273.90.99 (0.90–1.08)191854.60.46 (0.42–0.50)200154.60.43 (0.39–0.46)
 2003–04230666.61.0069866.50.93 (0.80–1.08)45043.90.40 (0.34–0.46)42942.20.34 (0.29–0.40)
 2005–06278075.11.0083777.31.07 (0.91–1.26)66758–00.50 (0.43–0.58)69956.00.43 (0.37–0.50)
 2007–08292476.61.0087777.41.02 (0.87–1.20)80159.80.47 (0.41–0.55)87362.50.49 (0.42–0.57)
 All periods871279.31.00265581.30.94 (0.84–1.04)277979.10.75 (0.67–0.83)283077.30.59 (0.53–0.65)
 2003–04252873.11.0079475.60.83 (0.70–0.99)74172.30.75 (0.64–0.89)728 71.60.63 (0.53–0.74)
 2005–06303582.01.0090983.90.95 (0.79–1.15)94882.40.82 (0.68–0.99)982 78.70.59 (0.49–0.70)
 2007–08314982.51.0095284.00.92 (0.76–1.11)109081.30.64 (0.54–0.77)1120 80.20.53 (0.45–0.64)
n-3 PUFA
 All periods337030.71.0075423.10.69 (0.63–0.75)58416.60.50 (0.45–0.55)49313.50.39 (0.35–0.44)
 2003–0464118.51.0015915.10.80 (0.66–0.97)10510.20.59 (0.47–0.74)90 8.80.50 (0.39–0.63)
 2005–06132635.81.0028326.10.66 (0.57–0.77)22119.20.49 (0.41–0.58)169 13.50.33 (0.27–0.39)
 2007–08140336.71.0031227.50.66 (0.57–0.76)25819.30.44 (0.38–0.52)23416.80.37 (0.31–0.44)
  • aAll OR (95% CI) are fully adjusted for confounders.

Temporal trends by subgroups

From 2003 to 2007, prescriptions of all the drugs under study increased in all groups of patients (Table 3). However, the increase was not constant across groups. While prescription of aspirin increased by 230% among younger males, it increased only by 52% among older females (Table 3), suggesting a disparity of treatment between younger, healthier patients and more co-morbid and older patients.

View this table:
Table 3

Temporal trends (2003–07) in treatment with aspirin, statins, β-blockers, angiotensin-converting enzyme inhibitors or angiotensin receptor blockers and n-3 polyunsaturated fatty acids

Men ≤75 years (n = 10 980)Women ≤75 years (n = 3266)Men >75 years (n = 3515)Women >75 years (n = 3662)
 2003 (ref.)
 20052.09 (1.82–2.39)1.63 (1.29–2.06)1.61 (1.31–1.98)1.57 (1.30–1.90)
 20072.30 (1.99–2.65)1.43 (1.14–1.80)1.69 (1.38–2.08)1.52 (1.26–1.84)
 2003 (ref.)
 20052.28 (1.99–2.61)2.02 (1.61–2.53)1.92 (1.60–2.30)1.64 (1.38–1.97)
 20072.15 (1.88–2.46)1.74 (1.39–2.17)2.28 (1.89–2.74)1.94 (1.62–2.32)
 2003 (ref.)
 20051.52 (1.36–1.68)1.65 (1.36–2.01)1.65 (1.38–2.00)1.60 (1.35–1.91)
 20071.62 (1.45–1.80)1.66 (1.36–2.02)1.73 (1.45–2.06)1.94 (1.63–2.31)
 2003 (ref.)
 20051.67 (1.48–1.88)1.54 (1.22–1.93)1.66 (1.34–2.05)1.28 (1.05–1.57)
 20071.69 (1.50–1.90)1.48 (1.18–1.86)1.37 (1.11–1.69)1.26 (1.03–1.55)
n-3 PUFA
 2003 (ref.)
 20052.51 (2.25–2.81)1.97 (1.58–2.45)2.02 (1.57–2.60)1.52 (1.15–2.00)
 20072.62 (2.34–2.92)2.09 (1.67–2.59)2.00 (1.56–2.58)1.85 (1.42–2.43)

The trend was less evident in the treatments with statins, ACE-Is/ARBs, and n-3 PUFA. β-Blockers were the only agents to show a proportionally greater increment in use among older groups of patients (Table 3).


The data of in-hospital and out-of-hospital mortality rates for MI document that the overall improvement in the survival of MI patients is a robust reality also in Italy. The clearly less favourable improvement of outcomes in the more vulnerable and co-morbid populations at 1 year is an indicator of the temporal increased gap for the transferability of recommended secondary prevention strategies to the more older and co-morbid populations.13,20

The detailed profiles of the pharmacological strategies applied to the four groups identified in the successive cohorts provide a very clear picture of the situation. While a substantial improvement of compliance to guidelines could be seen in all the subgroups, the evidence of increasing (not decreasing) gaps is well documented when the four groups are compared.

The overall, ‘mean’ positive results were achieved mostly by improvements attained only in the less vulnerable fraction of the population.

Age appears to play a primary role, and its interaction with sex defines the least exposed population (females >75 years). Aspirin seems to be the most evident case of comparative disadvantage, though a lower rate of prescription could reasonably be expected because of the age-related probability increase in bleeding complications (also due to the expected higher use of non-steroidal anti-inflammatory drugs).

The apparent exception of exposure to β-blockers could well be a marker of the concomitant increasing prevalence of heart failure and of the acceptance of the safety profile of this drug class among the elderly.

It is indeed encouraging, however, to witness less divergent OR in the third (2007–08) cohort.

A further contribution of our data to the assessment of the reasons underlying the persistence of what could be seen as a discriminatory attitude could be derived by comparing the trends of in-hospital mortality (where the improvements appear more equally distributed) and 1-year mortality (which reflects more closely long-term drug exposures). Although remains speculative, the absence of long-term (one year) survival benefits among elderly groups could be related to the persistent inequality in prescribing effective treatments.

Socio-economic factors, which are often quoted as leading determinants of discrimination,13 could hardly play a role in the Italian context, where the NHS provide full coverage. It is possibly more pertinent to interpret our data not principally as indicators of non-compliance by the prescribers, but as an expression of a real cultural problem, for which there cannot be a clear-cut answer.21 In fact, this attitude towards the underuse of beneficial medications based on age had been documented by others and termed ‘ageism’.22

An alternative explanation for our findings could be that the paucity of evidence-based data on the effectiveness of studied treatments in the elderly population guided the physician's decisions. However, available evidence indeed suggests that relative benefits of studied therapies are similar among older and younger patients. Specifically, for aspirin use, the European and ACC/AHA guidelines recommend the use of aspirin daily after an acute coronary syndrome in the absence of contraindications and without modification based on age.23,24

Also β-blockers are known to be safe and effective in elderly patients post-MI. Although no study included patients ≥75 years of age, several smaller trials as well as observational studies provide strong evidence that long-term β-blocker therapy improves outcomes after MI in patients up to 90 years of age.2527 Similar figures occur with long-term treatment with ACE-Is.2830

Finally, statins are the drugs that have the best direct evidence of benefit among elderly for the prevention of subsequent MI and death.31,32


Since our data are from administrative databases we do not have detailed information about some patient's characteristics that could eventually guide decisions on treatments and influenced outcomes (unmeasured confounders). However, our results were obtained using fully adjusted analyses for co-morbidities, cardiovascular history, in-hospital procedures, and MI characteristics.

We selected for our analyses a cohort that survived at least 1 year after the index event. Although this could result in a ‘healthier’ selection bias, this strategy was used to avoid comparing severely ill patients in the elderly groups with healthier younger groups. Our results are indeed strengthened since the selection approach could result in under-stimulation of actual gaps in treatment (i.e. bias results to the null).

Conflict of interest: none declared.


View Abstract