Methods

Consecutive patients (n = 448) implanted with a first-time ICD at the Erasmus Medical Center (Rotterdam, The Netherlands) from August 2003 to February 2010, were enrolled in the Mood and personality as precipitants of arrhythmia in patients with an Implantable cardioverter Defibrillator: A prospective Study (MIDAS). The exclusion criteria were a life-expectancy of <1 year, being on the waiting list for heart transplantation, a history of psychiatric illness other than affective/anxiety disorders, or insufficient knowledge of the Dutch language.

The medical ethics committee of the Erasmus Medical Center approved the study protocol, and the study was conducted according to the Helsinki Declaration. An ICD nurse provided written and oral information on the study before ICD implantation to all patients, and all patients provided written informed consent. The aim of the present study was a part of the broader objective to create a more complete picture of the interrelation between ICD patients’ psychological functioning and clinical risk profile.

Patients’ medical records and purpose-designed questions in the questionnaires were used to obtain the baseline demographic and clinical information. The demographic variables included age, gender, marital status, and educational level. The clinical variables included indication for ICD therapy (primary vs secondary), treatment with cardiac resynchronization therapy, left ventricular ejection fraction ≤35%, QRS duration, mean heart rate, the presence of coronary artery disease, symptomatic heart failure (defined as New York Heart Association functional class III-IV), atrial fibrillation, peripheral artery disease, previous percutaneous coronary intervention or coronary artery bypass grafting, smoking, and the use of cardiac (i.e., β-blockers, amiodarone, diuretics, angiotensin-converting enzyme inhibitors, and digoxin) and psychotropic medication. Information with respect to statin use, including the type and dosage, was also collected at baseline. Because statin use was stable during the 12 months of follow-up in almost all patients, we used the baseline information on statin use for the analyses during all follow-up visits.

In our cohort, 5 types of statins were prescribed: rosuvastatin, atorvastatin, simvastatin, pravastatin, and fluvastatin. Because of differences in pharmacologic efficacy and potency, we assigned relative weights to the different types and calculated a relative dose for each patient. According to the published data, the following relative potencies were allocated: fluvastatin, 1; pravastatin, 2; simvastatin, 4; atorvastatin, 8; and rosuvastatin, 16. Thus, rosuvastatin is 16 times more potent than fluvastatin at the same dosage. Subsequently, the original statin dosage was multiplied by the relative potency to obtain a relative dosage for each patient, enabling comparisons among the different statin types. Furthermore, a distinction was made between the lipophilic (atorvastatin, simvastatin, and fluvastatin) and hydrophilic (rosuvastatin and pravastatin) statins according to their capacity to penetrate the blood–brain barrier to compare the effects of statin types on patients’ psychological functioning.

The Hospital Anxiety and Depression Scale, a 14-item self-report questionnaire, with 7 items measuring anxiety and 7 items measuring depression, was administered at baseline and at 3, 6, and 12 months after implantation. All items are rated on a 4-point Likert scale, with scores ranging from 0 to 3 (total score range 0 to 21), and higher scores reflecting more symptoms. The psychometric properties of the Hospital Anxiety and Depression Scale are good.

Patients’ health status at baseline and at 3, 6, and 12 months after implantation was assessed using the validated Dutch language version of the Medical Outcomes Study Short Form 36-item Health Survey. The questionnaire consists of 36 items that contribute to 8 subscales: physical functioning, role limitations–physical, bodily pain, social functioning, mental health, role limitations–emotional, vitality, and general health. Each subscale has a score range of 0 to 100, with higher scores indicating better health status. The psychometric properties of the Medical Outcomes Study Short Form 36-item Health Survey are adequate.

The baseline demographic and clinical variables for patients with versus without statin therapy were compared using the chi-square test for nominal variables and Student’s t test for continuous variables, respectively. To assess the longitudinal association between statin therapy and psychological functioning, generalized linear mixed modeling was used. The major advantage of this technique is that missing data from 1 measurement point do not lead to exclusion of that patient from the analyses. Thus, the available data were used optimally. The described effects in the “Results” section are the relation of statin use at any measurement point with the level of psychological functioning over time, including all measurement occasions. We adjusted for variables that have been associated with impaired psychological functioning in the published arrhythmia data, including atrial fibrillation, symptomatic heart failure, coronary artery disease, diabetes mellitus, appropriate and inappropriate shocks during follow-up, the use of amiodarone and psychotropic medication, and smoking. In addition, we adjusted for variables that were expected to be related to psychological functioning or functioning of the cardiovascular system, including age, gender, educational level, peripheral artery disease, and the use of β blockers. Statin use, including type and dosage, was set as a fixed variable (i.e., not varying over time) after ascertaining the stability of statin use in our data set during the 12-month follow-up period. All covariates were also set as fixed variables. The results of the generalized linear mixed modeling analyses are presented as estimates with accompanying t and p values and 95% confidence intervals. In a secondary analysis, the association between statin type (lipophilic vs hydrophilic statins) and psychological functioning was longitudinally assessed with generalized linear mixed modeling, adjusting for the same covariates. For all tests, p <0.05 (2 sided) was considered significant. Analyses were performed using PASW Statistics, version 19, statistical software (IBM, Armonk, New York).