OUP user menu

★ editor's choice ★

Long-term compliance with beta-blockers, angiotensin-converting enzyme inhibitors, and statins after acute myocardial infarction

Gunnar H. Gislason, Jeppe N. Rasmussen, Steen Z. Abildstrøm, Niels Gadsbøll, Pernille Buch, Jens Friberg, Søren Rasmussen, Lars Køber, Steen Stender, Mette Madsen, Christian Torp-Pedersen
DOI: http://dx.doi.org/10.1093/eurheartj/ehi705 1153-1158 First published online: 6 January 2006


Aims To study initiation, dosages, and compliance with beta-blockers, angiotensin-converting enzyme (ACE)-inhibitors, and statins in patients after acute myocardial infarction (AMI) and to identify likely targets for improvement.

Methods and results Patients admitted with first AMI between 1995 and 2002 were identified by linking nationwide administrative registers. A total of 55 315 patients survived 30 days after discharge and were included; 58.3% received beta-blockers, 29.1% ACE-inhibitors, and 33.5% statins. After 1, 3, and 5 years, 78, 64, and 58% of survivors who had started therapy were still receiving beta-blockers, 86, 78, and 74% were receiving ACE-inhibitors, and 85, 80, and 82% were receiving statins, respectively. Increased age and female sex were associated with improved compliance. The dosages prescribed were generally 50% or less of the dosages used in clinical trials, and dosages did not increase during the observation period. Patients who did not start treatment shortly after discharge had a low probability of starting treatment later.

Conclusion The main problem with underuse of recommended treatment after AMI is that treatment is not initiated at an appropriate dosage shortly after AMI. A focused effort in the immediate post-infarction period would appear to provide long-term benefit.

  • Acute myocardial infarction
  • Beta-blockers
  • ACE-inhibitors
  • Statins
  • Pharmacological treatment
  • Compliance


Beta-blockers, angiotensin-converting enzyme (ACE)-inhibitors, and statins are recommended for most patients following an acute myocardial infarction (AMI), but underuse is widely documented.1,2 Underuse may involve not starting therapy, poor compliance, and underdosing. Much attention has focused on the fact that many patients are never offered treatment,2 but compliance with treatment and dosing has received less attention. Comprehensive analyses of initiation, compliance, and dosing are necessary to learn the likely targets for improving long-term use. Since 1995, the Danish Registry of Medicinal Product Statistics has registered all prescriptions dispensed from pharmacies in Denmark according to each patient's unique civil registration number. In this study, we linked data from this Registry with data from the National Hospital Registry to study initiation of treatment, drug dosages, and long-term compliance with beta-blockers, ACE-inhibitors, and statins in 71 515 patients after their first AMI between 1995 and 2002.


The National Hospital Registry keeps records on all hospital admissions in Denmark since 1978, and each hospitalization is classified according to the International Classification of Diseases (ICD), until 1994 the ICD-8 and from 1994 the ICD-10. All patients are registered via a unique and permanent civil registration number. The Danish Registry of Medicinal Product Statistics includes information about all prescription medicines dispensed from pharmacies in Denmark since 1995. All prescriptions are registered at an individual level by using the civil registration number. The Registry classifies medicines according to the Anatomical Therapeutic Chemical (ATC) system, an international classification system of pharmaceuticals. As all residents in Denmark are covered by a national health security system and get the cost of drugs partly reimbursed, all pharmacies are required by law to register all prescription dispensed in this nationwide registry.


All patients aged 30 or older admitted with a diagnosis of first AMI (ICD-10: I21–I22) between 1995 and 2002 and alive 30 days after discharge (180 days for statins) were identified from the National Hospital Registry. The diagnosis of AMI in the National Hospital Registry has been validated and has a sensitivity of 91% and predictive value of 93%.3 The selection procedure and characteristics of the patients have been detailed previously.4,5 We registered whether patients had a prescription of beta-blockers (ATC code C07) and ACE-inhibitors (ATC code C09—also including angiotensin 2 receptor blockers) dispensed 0–30 days from discharge after first AMI and statins (ATC code C10AA) dispensed 0–180 days from discharge. The reason the interval for statins differed was that the reimbursement policies for statins changed during the period. In 1995, reimbursement required individual application, but after 1998, all AMI patients were reimbursed for the full cost of statins without need for individual application.

All subsequent prescriptions were identified until the end of year 2002 or the date of death. Among patients who did not receive treatment early after discharge, we identified those who started treatment between 30 days and 1 year to estimate whether a significant proportion initiated treatment later. As the registries do not include information of left ventricular function and that the ICD-10 cod for heart failure has a relatively low sensitivity, the dispensed prescription of loop diuretics (ATC code C03C) 90 days before to 30 days after discharge was used as a proxy for the diagnosis of heart failure. In the same way, dispensed prescription of antidiabetics (ATC code A10) 90 days before to 30 days after discharge was used as a proxy for the diagnosis of diabetes mellitus.


The Danish Registry of Medicinal Product Statistics includes information about the dispensing date of the prescription, the strength of the drug, and the total number of tablets dispensed but not the prescribed dose of the drug. The dose of each prescription was therefore calculated from the average dose given during up to three consecutive prescriptions. Excess tablets were allowed to be accumulated for up to three previous consecutive prescriptions at any time. On the basis of these assumptions, we calculated whether patients at any time had tablets available or not. We defined a patient as receiving treatment if tablets were available.

To determine long-term compliance, we calculated whether patients who did not receive treatment were in a break of 7, 30, 90, and 180 days or longer. We calculated the proportion of patients who restarted therapy after a break of a particular length and found that a substantial proportion of patients restarted treatment again after a break shorter than 90 days. We therefore used a break of at least 90 days as a proxy for poor compliance.

To validate the calculations, sensitivity analysis was performed with the average daily dose calculated also from single prescriptions, with the minimal dose equal to a previously calculated dose for the same patient and allowing each patient to keep residual tablets from one to five prescriptions. The manipulations resulted in 75–85% agreement between the calculated doses and 2–4% with changes in the number of patients stopping therapy.


We calculated the average dosages of the most frequently used medications within each group by multiplying the strength of the formulation by the number of tablets per day. The average dosage was calculated only for patients who had medication available for treatment. To estimate underdosing, we compared the average dosage with the dosages used in major randomized clinical trials.

Statistical analysis

For descriptive statistics, the results are given as mean values with standard deviation or as median values with range. To analyse differences in proportion in drug use and death rates between 1995 and 2002, we tested for linear trend in logistic regression models. The rates of first dispensing of treatment and break in treatment were estimated by the Kaplan–Meier method. To analyse the effect of available covariates (age, gender, year of AMI, and concomitant medical treatment) on long-term compliance, we used Cox multivariable proportional hazard models and censoring for death. Model assumptions—the linearity of continuous variables, the proportional hazard assumption, and lack of interaction—were tested and found valid unless otherwise indicated. Furthermore, we calculated the persistence of therapy, defined as number of patients having medicine available at any time (having filled at least one prescription early after the AMI) divided by the number of patients alive within particular treatment group. Patients with missing information or lost to follow-up (emigrated), n=22 (0.04%), were censored at the time of disappearance. All statistical calculations were performed using the SAS statistical software package, version 8.2 for UNIX servers (SAS Institute Inc., Cary, NC, USA).


The Danish Data Protection Agency approved this study, and data were made available to us such that individuals could not be identified.


During the 8-year study period, 71 515 patients had a first admission for AMI. The 30-day mortality was 22.7%, ranging from 26.0% in 1995 to 17.4% in 2002 (P<0.001). A total of 55 315 patients were alive 30 days after discharge and were included in the analysis of the use of beta-blockers and ACE-inhibitors after AMI. The 48 412 patients who were alive 180 days after discharge were included in the analysis of statins. The baseline characteristics of the study population are shown in Table 1.

View this table:
Table 1

Baseline characteristics of the study sample

Total patients (mean age±SD, years)55 315 (67.6±12.8)
Women (mean age±SD, years)20 361 (72.0±12.3)36.8
Men (mean age±SD, years)34 954 (65.1±12.5)63.2
Year of AMI
 19956 82712.3
 19966 48511.7
 19976 30011.4
 19986 51411.8
 19996 35111.5
 20007 15612.9
 20018 00814.5
 20027 67413.9
 Beta-blockersa32 25958.3
 ACE-inhibitorsb16 06829.1
 Statinsc16 43333.5
 Loop diuretics20 32336.7
 Antidiabetic drugs5 57110.1
  • SD, standard deviation.

  • aRanging from 38.1% in 1995 to 67.9% in 2002 (P<0.001).

  • bRanging from 24.5% in 1995 to 35.5% in 2002 (P<0.001).

  • cRanging from 11.7% in 1995 to 63.0% in 2002 (P<0.001).

Figure 1 illustrates the timing of the first dispensing of each of the three drugs during the first year after AMI (censored for death). The probability of starting treatment was low if treatment was not started early for beta-blockers and ACE-inhibitors. Only 5621 (10.7%) patients started beta-blocker treatment and 6694 (12.1%) ACE-inhibitors between 31 days and 1 year after AMI. For statins, the start of treatment was slower, but between day 181 and 1 year, only 2868 (5.9%) patients started therapy. From 1995 to 2002, the number of patients initiating beta-blocker treatment later than 30 days after the infarction declined from 11.5 to 9.3% (P<0.001) and statins after 180 days declined from 6.7 to 1.7% (P<0.001), but the number of patients receiving ACE-inhibitors after 30 days increased from 10.5 to 12.0% (P<0.001). Hospital-based physicians issued the majority of the initial early prescriptions of beta-blockers, ACE-inhibitors, and statins (Table 2). The later prescriptions initiated were distributed almost evenly between hospital-based physicians and general practitioners.

Figure 1

Cumulative frequency of patients with first AMI who filled a first prescription of beta-blocker, ACE-inhibitor, or statin within 1 year after discharge (censored for death).

View this table:
Table 2

First prescription claim of beta-blockers, ACE-inhibitors, and statins dispensed 0–30 days and 31–365 days from discharge, according to the type of the physician issuing the prescription

0–30 days31–365 days0–30 days31–365 days0–180 days181–365 days
General practitioner (%)15.149.821.553.827.245.9
Specialist (private) (%)
Hospital physician (%)84.449.377.945.172.153.1

Figure 2 illustrates the long-term compliance for patients who filled a prescription for a beta-blocker or ACE-inhibitor within 30 days of AMI or a statin within 6 months. Short breaks in therapy were common, but many patients restarted therapy later (Table 3). We used a break of at least 90 days to analyse factors of importance for poor adherence with treatment.

Figure 2

Persistence with therapy in patients who claimed at least one prescription early after AMI. Each point represents number of patients with available medication divided by the number of patients alive at that time.

View this table:
Table 3

The proportion of AMI patients on beta-blocker, ACE-inhibitor, or statin therapy who experienced a break of at least 7, 30, 90, or 180 days within 5 years of treatment and the proportion who had re-initiated therapy within 1 year

Length of breakProportion experiencing a breakProportion re-initiating treatment within 1 year
Beta-blocker (%)ACE-inhibitor (%)Statin (%)Beta-blocker (%)ACE-inhibitor (%)Statin (%)
≥7 days96.594.797.089.092.996.7
≥30 days73.156.271.468.672.689.6
≥90 days51.734.434.442.245.169.3
≥180 days41.828.723.622.526.740.3

The Cox multivariable proportional hazard analysis (Table 4) revealed that although there were increasing number of patients initiating treatment from 1995 to 2002, this did not affect compliance. On the contrary, compliance improved with beta-blockers. Women had better compliance with beta-blockers and statins than men, and older patients had better compliance with beta-blockers but worse compliance with ACE-inhibitors and statins. Concomitant medical treatment did not worsen compliance, except for patients taking loop diuretics had slightly poorer compliance with beta-blockers.

View this table:
Table 4

Cox multivariable proportional hazard analysis of covariates predicting first break in therapy of more than 90 days (a proxy for low compliance) with beta-blockers, ACE-inhibitors, and statins in patients with first AMI, 1995–2002

Calendar yeara
 30–59 years1.001.001.00
 60–69 years0.850.81–0.90<0.0010.860.79–0.93<0.0010.880.82–0.94<0.001
 70–79 years0.860.81–0.90<0.0011.030.94–1.140.490.990.90–1.080.80
 ≥80 years0.880.82–0.95<0.0011.161.05–1.280.0031.401.16–1.68<0.001
Concomitant treatmentd
 Loop diuretics1.071.02–1.120.0060.910.85–0.980.0090.940.87–1.020.15
  • RR, relative risk; CI, confidence interval.

  • a1995–96 as reference.

  • bWomen as reference.

  • cAged 30–59 as reference.

  • dNon-users as reference.

Table 5 shows the average dosages of the most frequently used beta-blockers, ACE-inhibitors, and statins. We analysed whether dosages changed during the observation time by comparing the average dosages used during the first year of treatment with the average dosages used between 3 and 5 years of treatment. This revealed that the average dosages of beta-blockers and ACE-inhibitors changed only between 2 and 7%. The average dosages of simvastatin and pravastatin changed <3%, but the dosages for atorvastatin increased 21.5% from the first year when compared with 3–5 years of therapy.

View this table:
Table 5

Average daily dosage of the most frequently used beta-blockers, ACE-inhibitors, and statins after AMI

MedicationaProportionb (%)Average daily dosage [median (IQR); mg]Clinical trial dosages (mg)c
 Metoprolol76.875 (50–100)200
 Atenolol7.650 (25–50)100
 Bisoprolol6.55 (5–7.5)10
 Trandolapril26.32 (1.5–3)4
 Ramipril25.95 (3.75–10)10
 Captopril13.637.5 (25–62.5)150
 Enalapril13.410 (7.5–20)20
 Simvastatin63.820 (10–20)40
 Atorvastatin15.910 (10–20)80
 Pravastatin15.240 (20–40)40
  • IQR, interquartile range.

  • aThe most frequently used medications within each class.

  • bProportion of all medications used within each class.

  • cDosages used in randomized clinical trials with clinical endpoint reduction.


This study has four main findings. (i) If prophylactic treatment after a myocardial infarction is not initiated at the time of discharge, the likelihood of ever receiving treatment is small. (ii) If treatment is started early post-AMI, most patients adhere to treatment for many years. (iii) Initiation of treatment nearly always relies on the initiative of the hospital physicians. (iv) The prescribed doses of all three drugs were substantially lower than those recommended and seldom adjusted during long-term therapy.

Initiation of therapy

All three drugs were prescribed with increasing frequency during the study period, the increase being most pronounced for beta-blockers and statins. Beta-blockers and ACE-inhibitors were primarily initiated during the first 30 days after AMI, whereas statin therapy was more gradually initiated during the first 6 months (Figure 1). This was probably related to Denmark's reimbursement policies, which changed in 1998, and AMI patients no longer need to apply individually for reimbursement of the cost of statin therapy. Notably, the vast majority of early prescriptions were written by hospital-based physicians, whereas general practitioners initiated very few early prescriptions (Table 2). The relatively few patients who started late had a more even contribution by hospital-based physicians and general practitioners, but the general practitioners still only accounted for about half the initial prescriptions. Further, the number of patients initiating treatment later than 30 days for beta-blockers and 180 days for statins from 1995 to 2002 declined substantially. This emphasizes the importance of initiating treatment early, as the probability of starting treatment later is low.

It is interesting that although treatment was mostly initiated by hospital-based physicians, there still remains substantial proportion of patients who do not start treatment. Hospital-based physicians taking care of patients with AMI are most often specialists in internal medicine or cardiology, and thus should be updated in appropriate treatment of AMI patients. Most likely, some discharging physicians rely on the fact that the doctor taking care of the AMI patient after discharge initiates appropriate therapy not already started in the hospital, whereas doctors taking care of the patient after discharge depend on the judgement of the discharging physician. To prevent such misunderstanding, the discharging physician should initiate treatment in the hospital and if not started at discharge give clear message to the doctor taking care of the AMI patient after discharge regarding further medical treatment.

Long-term compliance with treatment

After 5 years of treatment, 58% of patients were still receiving beta-blockers, 74% ACE-inhibitors, and 82% statins (Figure 2). The literature is sparse regarding studies on long-term compliance with treatment post-AMI. Registries that included selected departments have shown 6-month compliance with ACE-inhibitors and beta-blockers exceeding 80%.2,6 Studies from North America among patients older than 65 years of age and patients participating in specific insurance systems showed that 60–80% continued treatment after 1 year.7 The most comprehensive long-term studies available are studies of lipid-lowering therapy. The long-term compliance in these studies was 50–85% after 1 year, 40–77% after 2–3 years, and 45–50% after 5 years.814

The difficulty in comparing these numbers with those in our study is that compliance has been defined in many different ways. Previous studies have not distinguished between breaks and discontinuation. Most of our patients restarted therapy after a break, especially after a brief break (Table 3). Even after a break of at least 90 days, almost half of the patients restarted therapy with beta-blockers and ACE-inhibitors within 1 year and almost 70% restarted statins.

To analyse the factors influencing long-term compliance, we used the time to first break of 90 days, or longer, in multivariable proportional hazard analysis (Table 4). The increase in early initiation of treatment during the study period was not associated with decreased long-term compliance—instead, compliance improved moderately but statistically significantly. Notably, multiple drug therapy did not reduce compliance, and increased age and female sex were also associated with improved compliance. These results emphasize that an increased effort during the study period to initiate more medicines in more patients was not associated with worse compliance.


Analysis of the dosages of the most frequently used drugs within each class revealed substantial underdosing of almost all medicines, with dosages generally 50% or less than the dosages that randomized trials have proved to be effective (Table 5).1519 We also analysed whether dosages changed from the first prescription dispensed when compared with later on but found that patients largely stayed at their initial dosage. Defining a target dose for statins is more difficult because this may be related to the effect on serum cholesterol levels. Our data do not cover patients' cholesterol levels, and thus we cannot conclude about the dosages of statins. Although not as prominent as for beta-blockers and ACE-inhibitors, the dosages of statins were generally lower than those used in randomized trials.2022

Strengths and limitations of the study

The AMI diagnosis in the National Hospital Registry has been validated and is sensitive as well as specific.3 The registry includes all hospital admissions in Denmark from 1978 and therefore not affected by selection bias by only including selected hospitals, certain health insurance systems, or age groups. A comprehensive accumulation of nationwide admission data and linking to other registries is unique for Denmark. Although the study is based on registries from a single country, Denmark's health insurance system, which partly reimburses all patients for drug expenses, is typical of many western countries.23

A break in therapy of at least 90 days was arbitrarily selected to define long-term compliance, as one-third to half of the population experienced break of at least 90 days. Analyses of the effects of shorter or longer breaks in therapy did not change the results noticeably.

The diagnostic criteria for AMI changed in 1999, and during the study period more sensitive diagnostic markers have been introduced (i.e. troponins). Nevertheless, this has not resulted in dramatic change in the prognosis of AMI 24 and it is unlikely that this affects our study.

The National Hospital Registry does not include information about basal clinical variables such as location or size of infarction, smoking status, systolic function, and co-morbidity. However, the indication for treatment with beta-blocker or statin after AMI is independent of these factors. Further, ACE-inhibitors are primarily indicated for patients with reduced systolic function and diabetes. We therefore used the concomitant use of loop diuretics and antidiabetics as a proxy for, respectively, heart failure and diabetes, to define subgroups of patients where there was definite indication for treatment. The registries do not include information about contraindications to treatment, adverse reactions, or allergies that might have caused the treatment not to start or to be terminated early after start. Also, we do not know how many patients started treatment in hospital and did not continue treatment after discharge. Furthermore, factors such as race/ethnicity, history of depression or other psychiatric illness, and the specialty of the physician taking care of patient could also influence long-term compliance.

Clinical implications

This study demonstrates that long-term compliance can be expected if treatment with beta-blockers, ACE-inhibitors, and statins is started early after AMI, although dosages are seldom uptitrated during long-term therapy. In contrast, if therapy is not started in the early phase after AMI, the patient is unlikely to ever receive appropriate medical treatment. Most general practitioners and other physicians treating survivors of AMI seem to rely on the initiative and judgement of the hospital department from which the patient is discharged. This underscores the importance of hospital departments initiating appropriate long-term therapy in survivors of AMI. A more focused effort from hospitals treating AMI patients in starting these patients on relevant medicines and titrating them to an effective dose will result in long-term benefits.

The role of the funding source

The study sponsors had no involvement in study design; in the collection, analysis, and interpretation of data; in the writing of the report; and in the decision to submit the article for publication.

Conflict of interest: several of the authors have given industry-sponsored lectures and taken part in clinical trials sponsored by the industry, but there are no specific conflicts of interest related to the current article.


The study was supported by an unrestricted research grant from the Danish Pharmaceutical Association (Grant no. 31-03) and the Danish Heart Foundation (Grant no. 05-04-B46-A522-22207).


View Abstract