Prevalence of antidepressant prescription or use in patients with acute coronary syndrome: a systematic review

Abstract

Background and objectives: Depression is common among acute coronary syndrome (ACS) patients and is associated with poor prognosis. Cardiac side effects of older antidepressants were well-known, but newer antidepressants are generally thought of as safe to use in patients with heart disease. The objective was to assess rates of antidepressant use or prescription to patients within a year of an ACS.

Methods: PubMed, PsycINFO, and CINAHL databases searched through May 29, 2009; manual searching of 33 journals from May 2009 to September 2010. Articles in any language were included if they reported point or period prevalence of antidepressant use or prescription in the 12 months prior or subsequent to an ACS for ≥100 patients. Two investigators independently selected studies for inclusion/exclusion and extracted methodological characteristics and outcomes from included studies (study setting, inclusion/exclusion criteria, sample size, prevalence of antidepressant prescription/use, method of assessing antidepressant prescription/use, time period of assessment).

Results: A total of 24 articles were included. The majority were from North America and Europe, and most utilized chart review or self-report to assess antidepressant use or prescription. Although there was substantial heterogeneity in results, overall, rates of antidepressant use or prescription increased from less than 5% prior to 1995 to 10-15% after 2000. In general, studies from North America reported substantially higher rates than studies from Europe, approximately 5% higher among studies that used chart or self-report data.

Conclusions: Antidepressant use or prescription has increased considerably, and by 2005 approximately 10% to 15% of ACS patients were prescribed or using one of these drugs.

Figure 1. Flow Diagram of Study Selection Process.

Figure 2. Prevalence of Antidepressant Use in ACS Patients.

Studies are grouped by geographic region and sorted by midpoint of data collection period. Benazon et al. rates are denoted with grey markers. Error bars are +/−1.96 standard errors, and marker sizes are weighted according to the inverse of the squared standard error. *Medication information was abstracted from patient charts but only prescriptions actually filled were included. ^†Study reported, “only two patients were taking antidepressants prior to hospitalization,” which was coded as being assessed at hospital admission.

Figure 3. Heterogeneity of Unadjusted Prevalence Estimates.

The funnel plot shows individual study prevalence estimates (x-axis) plotted against estimated standard errors (y-axis). The white cone inside the gray region indicates a 95% probability region that points would be expected to fall into assuming no heterogeneity and no publication bias. The North American studies are indicated by the filled circles and the non-North American studies are indicated by the × symbols.

Figure 4. Heterogeneity of Adjusted Prevalence Estimates.

The funnel plot shows the residuals from the meta-regression on setting and time (x-axis) plotted against estimated standard errors (y-axis). The white cone inside the gray region indicates a 95% probability region that points would be expected to fall into assuming the assumptions of the meta-regression were valid. The North American studies are indicated by the filled circles and the other studies are indicated by the × symbols.