Adherence to guidelines for the prescription of secondary prevention medication at hospital discharge after acute coronary syndrome: a multicentre study

The study was conducted in a cross-sectional design.

In 2012, 91 hospitals provided ACS care in the Netherlands. From this pool, 13 hospitals were selected by means of a multistage random sampling procedure to participate in the evaluation of the national quality improvement program.

Potentially eligible study charts were selected from the hospitals’ financial system codes for ST-segment elevation myocardial infarction (STEMI), non-ST-segment elevation myocardial infarction (NSTEMI) and unstable angina (UA). All patients discharged in 2012 with a diagnosis of ACS (as confirmed in the discharge letter) were considered for inclusion. Charts of patients transferred to another hospital or department for further evaluation or treatment, patients who died during hospital admission, who received palliative care, who left the hospital against medical advice or who had no information about the prescribed medication at discharge in their chart were excluded.

Data were collected by means of retrospective chart review. The chart reviewers visited the participating hospitals monthly. When the number of charts exceeded the screening capacity, charts were selected per month in chronological order of discharge until the screening capacity limit was reached.

In this study, guideline adherence was defined as the prescription of acetylsalicylic acid, P2Y₁₂ receptor inhibitor, statin, beta-blocker and ACE inhibitor at discharge, or a documented contraindication or other motivation for not prescribing these medicines. From the charts, information related to the prescription of these five medicines was abstracted. In case one or more medicines were not prescribed, documented contraindications as reported in an annually updated Dutch database of pharmacotherapy [12], or as motivated by the treating physician were retrieved (e.g. the prescription of anticoagulants instead of acetylsalicylic acid). The full list of contraindications was reported previously [11]. Additionally, patient, care and hospital characteristics (n = 40), e.g. age, sex, cardiac medical history, risk factors, resuscitation and discharge diagnosis were recorded. Hospitals were characterised by type (academic, tertiary teaching or general) and presence of percutaneous coronary intervention (PCI) and/or coronary artery bypass grafting (CABG) facilities (yes/no).

In total, 0.82 % of the data were missing, ranging per variable from 0.04 % (date of discharge) to 2.5 % (heart failure or arrival). Little’s test [13] was non-significant (p = 0.57) and missing value analyses showed no relationship between the missing data and the complete data, indicating the missing data were missing completely at random [14]. Therefore missing data were imputed by means of full conditional specification using the imputation procedure in IBM SPSS (Version 20 for Windows). As a sensitivity analysis, the results of the final analysis were compared with the results of a full case analysis to determine the accuracy of the imputation procedure.

Characteristics of the study population, participating hospitals and guideline adherence were determined by means of descriptive statistics. Associations of predictor variables with guideline adherence (complete adherence vs incomplete adherence) were studied by means of generalized linear mixed models for binary outcomes. To correct for clustering of patients within hospitals, hospital was entered as random effect in the analyses.

The associations of the predictor variables with guideline adherence were tested in univariate models (Table 1). All predictor variables with a significant association (p ≤ 0.05) were added to a multivariable model. To account for collinearity, all predictor variables without a significant association in the univariate analyses were added to the multivariable model one by one. In case of a significant improvement of the model fit (p ≤ 0.05), they were added to the multivariable model. Additionally, several potential interactions between variables were tested in the multivariable model: age with treatment, discharge diagnosis and sex; and discharge diagnosis with treatment and sex. In case of a significant improvement of the model fit (p ≤ 0.05), the interactions were added to the multivariable model. From this model, the fixed effects were presented as odds ratios (OR) with 95 % confidence intervals (CI).

The study protocol was approved by the medical ethics review committee of the VU University medical center.

In total, 3427 charts of patients with a confirmed discharge diagnosis of ACS in 2012 were screened. Of these, 876 patients (26.6 %) were transferred to another hospital or department for further evaluation or treatment, 56 (1.6 %) died during admission, information concerning discharge medication was missing for 14 patients (0.4 %), 6 (0.2 %) left the hospital against medical advice and 4 (0.1 %) received palliative care. After exclusion of these charts, 2471 patients were eligible for further analyses. Their mean age was 66.9 years and the majority were male (67.6 %) (Table 1).

Overall, 49.1 % of the patients were prescribed all five guideline-recommended medicines at discharge from the hospital, while an additional 20.0 % had contraindications documented for the medicines that were not prescribed. Consequently, the complete guideline adherence for the combination of the five medicines was 69.1 %. Guideline adherence for the individual medicines ranged between 99.6 % for acetylsalicylic acid and 76.8 % for the ACE inhibitor. Complete guideline adherence for the combination of the five medicines ranged from 42.1 to 87.0 % between hospitals. Prescription rates and guideline adherence are presented in Table 2 and Fig. 1.

In univariate generalized linear mixed model analyses, discharge diagnosis, type of treatment, age, sex, resuscitation, transport from another hospital and length of stay were significantly associated with the probability of guideline adherence (Table 1). Additionally, the risk factors diabetes mellitus, kidney failure, a prior detected coronary stenosis, current smoking, and a medical history of angina pectoris, coronary artery disease, prior PCI or prior CABG were associated with the probability of guideline adherence. These were entered in a multivariable model. The variable ‘recent PCI, CABG or myocardial infarction (< 6 months before admission)’ significantly improved the multivariable model fit and was therefore subsequently added to the model secondarily. In addition, an interaction between age and sex was added to the multivariable model. No significant associations between hospital characteristics and guideline adherence were found.

In the final model, patients with NSTEMI or UA compared with patients with STEMI, and patients who had a prior CABG were less likely to receive the guideline-recommended medication at discharge (Table 3). Further, adherence was higher for patients who were treated with PCI compared with patients who received pharmacological or CABG treatment, who had diabetes mellitus or kidney failure, who were resuscitated on admission, who had longer lengths of hospital stay, or who had a recent PCI, CABG or myocardial infarction (< 6 months before admission). Additionally, the effect of age differed between men and women, i.e. the medication was less likely completely according to the guidelines for older women compared with older men.