OUP user menu

Primary prevention of defibrillator implantation after myocardial infarction: clinical practice and compliance to guidelines

Johanna Sjöblom, Lina Ljung, Mats Frick, Mårten Rosenqvist, Viveka Frykman
DOI: http://dx.doi.org/10.1093/europace/eur354 490-495 First published online: 24 November 2011


Aims The purpose of this study is to investigate compliance to established guidelines regarding primary prevention with implantable cardioverter-defibrillator (ICD) in patients with left ventricular dysfunction after acute myocardial infarction (AMI) in a clinical setting.

Methods and results A retrospective study of medical records from patients admitted to two hospitals in Stockholm during 2008–2009 with a diagnosis of AMI and who at discharge had a left ventricular ejection fraction ≤35%. An evaluation of follow-up and echocardiographic recordings was performed 1–3 months after AMI. A total of 2023 patients <80 years, with AMI discharged from hospital, were screened for left ventricular dysfunction defined as ejection fraction (EF) ≤35%. Altogether 187 patients were identified and an ICD was implanted in 25 (13%)  patients. In 52 (28%) patients, there were contraindications for ICD therapy. An improved EF at follow-up, making ICD treatment redundant, was observed in 48 (41%)  patients who underwent an ECHO at follow-up or later. Seventeen (9%) patients without an ICD died during follow-up and out of these five patients died from sudden cardiac death (SCD), they had no contraindication to ICD therapy. An inadequate follow-up according to guidelines was found in 59 (32%) patients.

Conclusion The follow-up of post-myocardial infarction patients with left ventricular dysfunction according to guidelines was insufficient in this population and may have increased the risk for SCD. A significant proportion of patients experienced improved left ventricular function during short-term follow-up making preventive ICD treatment redundant.

  • Myocardial infarction
  • Sudden cardiac death
  • ICD therapy
  • Primary prevention


It is well known that reduced left ventricular function (LVED ≤35%) after acute myocardial infarction (AMI) increases the risk of sudden cardiac death (SCD).1,2 The MADIT II study3 has shown that prophylactic treatment with implantable cardioverter-defibrillators (ICD) provides an absolute risk reduction in mortality of 5.4% in patients with AMI and an ejection fraction (EF) of ≤30%. However, this is only valid for patients with AMI which occurred 1 month prior; in fact, there is no evidence of any mortality benefit of providing ICD immediately after an AMI in patients with EF ≤ 35%.47 One explanation for this may be that left ventricular (LV) function often improves within weeks tomonths after an AMI.8,9 Approximately 30–60% of the patients with impaired LV function improve in the first month after AMI.1012 The mortality benefit seems to remain even in patients with a prior AMI; in fact, there seems to be no difference in mortality benefit for ICD-implantation performed after 60 days, 6 months, or 1 year.1315 In contrast, the risk of SCD after AMI is the highest during the first 30 days as emphasized by data from several studies.2,16,17 At present, there are no established methods to identify post-AMI patients who will have a persistent LV dysfunction and thereby may benefit from early ICD implantation.18,19

International guidelines recommend ICD therapy in patients with ischaemic heart disease, reduced left ventricular function (EF ≤ 35%), and New York Heart Association (NYHA) classification I–III ≥40 days–3 months after AMI.20 It is not clear how these guidelines are followed in practice. Furthermore, it is not clear how many patients in a clinical setting have indication and no contraindications for primary prevention with an ICD and how often LV dysfunction at discharge improves over time with such a magnitude that treatment with ICD is no longer indicated. Therefore, the objectives of this study are as follows:

  1. To investigate whether post-AMI patients with LV dysfunction receive primary prevention with ICD according to established guidelines.

  2. To evaluate how many patients have an indication and no contraindications for ICD after the stipulated delay of at least 40 days after the AMI.

  3. To estimate the proportion of patients who experience significant improvements in LV function, in such way that there is no longer an indication for ICD treatment.

  4. To investigate the number of SCD in patients waiting for an ICD implant as an effect of the recommended guidelines.


Medical records of all patients admitted, with a diagnosis of AMI, to two hospitals in Stockholm (Södersjukhuset and Danderyds Sjukhus) between January 2008 and December 2009 were retrospectively studied. Patients with a discharge EF of ≤35% and aged ≤80 years at the time of admission were included in the study. We reviewed how many patients had an indication and no contraindication to primary prevention of ICD following a new ECHO between 1 and 3 months after discharge and, if indicated, whether the responsible physician made a decision for treatment with ICD.

Contraindications for ICD treatment were assessed by two independent reviewers from different hospitals who were unaware of the outcome during follow-up. In case of conflicting opinions consensus was reached by a mutual decision. Contraindications to ICD were defined as severe mental illness, dementia, current alcohol or drug abuse, NYHA Class IV, expected survival <1 year due to underlying diseases, or patient refusing ICD implantation. In patients with repeated ECHO, we evaluated the proportion of patients with improved LV function to such a degree that an indication for ICD no longer was present. Echocardiographic recordings at discharge and at follow-up were assessed with respect to left ventricular EF (LVEF). We considered the follow-up as inadequate in patients with impaired LV function, no contraindications to ICD, if follow-up ECHO was not performed in 1–3 months, and/or the physician did not discuss ICD treatment.

Statistical analysis

Data are presented as mean ± SD. Proportions are presented as per cent. As this is a retrospective and descriptive study, no further statistical analysis was performed.


Acute myocardial infarction was diagnosed in 3372 patients during the pre-defined period. Of these, 2023 (60%) patients were of age ≤80 years, of which 187 (9.2%) had an EF of ≤35% and were discharged alive from hospital. Clinical characteristics of these patients are shown in Table 1.

View this table:
Table 1

Characteristics of patients with ejection fraction ≤35%, and age <80 years

n= 187 (%)
Age, years (±SD)67.5 ± 9
Male/female137 (73)/50 (27)
Diabetes mellitus62 (33)
Hypertension91 (49)
History of congestive heart failure55 (29)
Prior coronary artery bypass graft surgery43 (23)
Prior myocardial infarction71 (38)
Angina pectoris58 (31)
Atrial fibrillation52 (28)
Baseline EF, % (±SD)29 ± 6


Most patients with AMI and LV dysfunction received a follow-up visit and performed a new ECHO to evaluate the LVEF 1–3 months after discharge. An ICD for primary prevention was implanted in 25 patients (13%). The medium time to implantation was 6.6 months. At discharge from the AMI 52 (28%) patients had at least one contraindication for ICD implantation and in our study 59 (32%) patients did not receive an adequate follow-up. Fifty-three (90%) because they did not execute a follow-up ECHO in time, and six (10%) because the doctor did not consider ICD treatment (Figure 1).

Figure 1

Acute myocardial infarction 2008–2009 Södersjukhuset and Danderyds sjukhus.

Left ventricular recovery

Of the 187 patients included in the study, 117 (63%) underwent ECHO at follow-up or later. Of those, 48 (41%) patients experienced improvements in LV function to such a degree that there was no longer an indication for ICD, although 20 of them performed ECHO much later than 3 months. The mean baseline EF of 29% (10–35% ± 6 SD) increased to 36% (10–60% ± 10 SD) with an average follow-up of 5 months, (Figure 2A and B). Clinical characteristics of patients with LV function improvements compared with those without improvements are shown in Table 2. In patients with no recovery of LV function there was a higher frequency of cardiovascular co-morbidity and more often non-STEMI. The presence of treatment with angiotensin-converting enzyme-inhibitors/angiotensin receptor blocker and beta-blockers was pronounced in both groups.

View this table:
Table 2

Demographic differences in patients according to left ventricular function

 Improved EF > 35%Non-improved EF ≤ 35%
n= 48 (%)n= 69 (%)
Male32 (67)56 (81)
Age, years (±SD)67 (±7)64 (±10)
Ejection fraction at baseline, % (±SD)32 (±5)28 (±6)
Diabetes mellitus14 (29)27 (39)
Hypertension21 (43)39 (57)
History of heart failure6 (12)15 (22)
Prior coronary artery bypass graft surgery4 (8)17 (25)
Atrial fibrillation10 (21)23 (33)
Valvular disorder1 (2)8 (12)
Left ventricular aneurysm2 (4)7 (10)
Prior myocardial infarction11 (23)25 (36)
ST elevation myocardial infarction26 (54)33 (48)
Non-ST elevation myocardial infarction22 (46)36 (52)
Percutaneous coronary intervention24 (50)38 (55)
Coronary artery bypass graft surgery9 (19)8 (12)
Troponin I > 520 (42)29 (42)
Multivessel disease9 (19)14 (20)
Beta-blockers47 (98)68 (99)
Aldosterone antagonist7 (15)16 (23)
ACE inhibitor or ARB47 (98)66 (96)
Figure 2

(A) Correlation of ejection fraction at baseline and at follow-up among responders. (B) Correlation of ejection fraction at baseline and at follow-up among non-responders.


Seventeen (9%) patients died during follow-up and prior to ICD decision but eight (47%) of them were considered to have contraindications to ICD already at discharge. Sudden cardiac death accounted for eight (47%) of all deaths (Figure 3). Patients who died had generally higher cardiovascular co-morbidity. Of those who suffered SCD, three patients (37%) were considered to have contraindications but five patients may have been rescued by early ICD implantation (Table 3). The causes of death are mainly based on clinical judgement since autopsy was performed rarely. Whether this finding can be generalized has to be confirmed in larger studies.

View this table:
Table 3

Demographic characteristic of the patients who died compared with survivors

n = 17 (%)n = 170 (%)
Male12 (71)121 (71)
Age, years (±SD)71 (±9)67 (±9)
Ejection fraction at baseline, % (±SD)27 (±5)29 (±6)
Diabetes mellitus9 (53)55 (32)
Hypertension9 (53)82 (48)
History of heart failure9 (53)49 (29)
Prior coronary artery bypass graft surgery6 (35)37 (22)
Atrial fibrillation6 (35)47 (28)
Valvular disorder3 (18)20 (12)
Left ventricular aneurysm1 (6)7 (4)
Prior myocardial infarction7 (41)62 (36)
ST elevation myocardial infarction5 (29)61 (36)
Multivessel disease7 (41)26 (15)
Percutaneous coronary intervention8 (47)77 (45)
Coronary artery bypass graft surgery1 (6)20 (12)
Troponin I > 58 (47)59 (35)
ACE inhibitor or ARB14 (82)164 (96)
Beta-blockers15 (88)165 (97)
Aldosterone antagonist5 (29)36 (21)


The main results of this study demonstrate that the number of patients who develop a low LVEF (≤35%) after AMI is small, and ICD implantation for primary prevention is suggested according to guidelines in ∼1–3% of all patients with AMI. In the AMI population with LV dysfunction a little more than one-fourth of the patients had contraindications to ICD implantation and in almost one-third, follow-up was insufficient and did not adhere to guidelines, this may increase the risk for sudden death. Furthermore, a significant proportion of patients experience improvements in LV function during short-term follow-up making preventive ICD treatment in this group redundant.

Patient demographics

The demography of our study group is in concordance with earlier studies.21 We excluded patients older than 80 years since treatment of these with primary prophylactic ICD rarely is indicated because of a low expected survival rate. Furthermore, the possibilities to retrospectively make an accurate judgement of the patients vital status becomes more difficult with increasing age.

The proportion of patients who did not recover their LV function after AMI is slightly lower than previous studies because we did not exclude patients with earlier AMI, known heart failure or previous coronary artery bypass graft surgery (CABG). Another difference is that in many previous studies most AMI patients have undergone percutaneous coronary intervention or CABG compared with 57% in our study population.

Underuse of implantable cardioverter-defibrillator for primary prevention

Treatment with ICD has increased continuously but their use varies significantly by sex, ethnicity, hospitals, and between industrialized countries.2224 Implant in the USA is more frequent than in Europe.25 The average implantation rate in western European countries was ∼155 per million population in 2007, but the implantation rate in Sweden during the same period was 64 per million.26 Although ICD implantation rate is high in Germany, the use of ICD for primary prevention after myocardial infarction is low, shown in a recent publication corresponding to 7% in patients with LVEF of 31–40% post-myocardial infarction.4 The same pattern is presented in an Italian multicentre study in which only 1.3% of patients with EF < 40% post-AMI received an ICD for primary prevention during 38 months follow-up.27

There are several explanations why ICD treatment is underused despite similar guidelines. Such reasons could be financial restraints, lack of knowledge, insufficient routines or referral patterns, individual physician's preference regarding ICD therapy, and safety of the devices. In Sweden, there are several small hospitals that do not perform ICD implantations. Previous studies have shown that implanting hospitals were larger and more likely to have an academic affiliation.28 Post-AMI patient follow-up visits are usually carried out by a cardiologist and not an electrophysiologist or device specialist and, compared with their electrophysiology colleagues, cardiologists were referring only about half of the patients who met ICD implantation indication criteria.25 The routine to postpone the decision to implant an ICD at least 40 days after the AMI may explain part of the results.

Improvement of left ventricular function

In the present study, many patients (four out of ten) experienced LV function improvements to such a degree that ICD treatment was no longer indicated at follow-up. To our knowledge, this is the first time the magnitude of this improvement, with regard to ICD indication, has been reported. Improvement of LV function might be an explanation why previous studies with early implantation after AMI have shown no benefit in survival.7,29

However, since patients are at the highest risk for SCD during the first months after AMI, it is important to find methods to detect which patients need early ICD treatment after AMI. This problem is highlighted by the finding that 4.3% (8 of 187) of the patients, in our population, died from a possible ventricular arrhythmia early after discharge.

Predicting left ventricular recovery

Numerous studies have tried to find predictors of LV functional recovery after an AMI. In the present study, improvement of LV function was associated with less cardiovascular co-morbidity and diabetes mellitus.12,30,31 However, the multiplicity of predictors of LV function recovery in all these studies indicates that these may be difficult to use in clinical practice.


Retrospective observational studies entail several limitations. Relevant details may be missed in medical records, in this study the absence of patient data may lead to an overestimation of the number of patients not given adequate follow-up. The echocardiographic assessments were often visual and not based on Simpsons method. However, this method is frequently used in clinical practice and has been shown to correlate to reference methods such as radionuclide imaging.32 In addition, we may have misjudged patients as not adequate follow-up because it might not be apparent from the medical record that they have a contraindication to ICD therapy. It is also possible that we erroneously concluded that ICD was contraindicated by misinterpreting content in the medical records.


The present study suggests an insufficient adherence to guidelines with respect to ICD treatment in patients with AMI and LV dysfunction eventually causing SCD in patients eligible for ICD treatment. In addition, a significant proportion of patients improve their LV function during follow-up with such a magnitude that an indication for ICD no longer was present. There is a need in prospective studies to develop methods to identify patients who could benefit from an early ICD treatment.


Follow-up of patients with AMI and LVEF ≤ 35% appears not to be adhering to guidelines with respect to identifying candidates for preventive ICD treatment. A significant proportion of patients with LV dysfunction at discharge seem to improve during follow-up in such a way that ICD treatment becomes redundant.


The study complies with the Declaration of Helsinki and the research protocol is approved by the locally appointed ethics committee.

Conflict of interest: none declared.


This work was supported by The Swedish Heart and Lung Foundation and Medtronic Inc.


View Abstract