Posttraumatic Stress Disorder and Electronically-Measured Medication Adherence after Suspected Acute Coronary Syndromes

The experience of a suspected acute coronary syndrome (ACS) can induce symptoms of posttraumatic stress disorder (PTSD), even among patients who rule out for ACS.(1) Patients who develop PTSD symptoms after ACS have twice the risk for recurrent cardiovascular events and mortality.(2) Yet, the reasons for this association between ACS-induced PTSD symptoms and adverse prognosis are poorly understood.

A hallmark of PTSD is avoidance of reminders of traumatic events. Patients with ACS-induced PTSD report skipping cardiovascular medications because they serve as traumatic reminders.(3) Self-reported adherence, however, is a poor indicator of true medication-taking behavior.(4) This study is the first to test whether suspected ACS-induced PTSD symptoms are associated with electronically-monitored medication nonadherence.

From 2014–2017, we enrolled a cohort of patients with suspected ACS from the emergency department (ED) of Columbia University Irving Medical Center (CUIMC).(5) Patients were included in this adherence substudy if they were prescribed a cardiovascular medication at discharge. Patients were excluded if they were non-English or non-Spanish speaking, cognitively impaired, in need of psychiatric intervention, terminally ill, or incapable of using an electronic pill bottle. The CUIMC institutional review board approved the protocol. All patients provided written informed consent.

PTSD was evaluated by telephone one month after discharge using the 17-item PTSD Checklist for a specific stressor (PCL-S), cued to the ACS symptoms that brought patients to the ED. A PCL-S score of 30 has ~80% sensitivity and specificity for PTSD diagnosis in primary care, and a score ≥44 is consistent with probable PTSD.

Adherence to a single cardiovascular medication (typically aspirin) was assessed for up to three months after discharge using eCAPs (Information Mediary Corp., Ottawa, Canada). eCAPs resemble ordinary pill bottle caps but record the date and time when pill bottles are opened. Substudy participants transferred their monitored medication into an eCAP mailed to their address after hospital discharge and mailed back the eCAP after a follow-up phone call. Generalized estimating equations were used to model percentage of adherent days per month (i.e., days adherent/days monitored) from PTSD symptoms. Month was a repeated-measures factor. Covariates included age, sex, ethnicity, discharge diagnosis (ACS confirmed or ruled-out), comorbidity (Charlson score), depression (8-item Patient Health Questionnaire), and dosing frequency of monitored medication.

Of 783 eligible patients, 662 (84.5%) agreed to participate, 474 (60.5%) returned eCAPs, and 400 (51.1%) had complete data on covariates and were included in these analyses. Patients with greater PTSD symptoms were less likely to return eCAPs (P=.03).

The mean (SD) age was 61.4 (11.8) years, 45.3% were women, 61.5% Hispanic, 12.3% white; 40.3% had prior cardiovascular disease, and 37.0% had ACS confirmed. PTSD symptoms were elevated (PCL-S≥30) in 18.5% of patients [10.7% possible PTSD (PCL-S 30-43); 7.8% probable PTSD (PCL-S≥44)]. Adherence was measured electronically for a mean (SD) of 52.8 (20.6) days. The most common electronically-monitored medication classes were aspirin (66.0%) and antihypertensives (19.8%).

The mean percent of days patients were adherent (i.e., took the correct number of doses) was 69.5% (SD 28.5%). Adherence declined from the first month (74.2%) to the third month after discharge (63.2%; P <0.001). There was a graded association between PTSD symptoms and medication adherence after discharge (Figure); this association did not vary by month (P-interaction=0.85). Patients with probable PTSD (PCL-S≥44) had more than twice the odds of medication nonadherence (i.e., odds of taking the incorrect number of doses on a given day; OR = 2.50, 95% CI 1.48, 4.22) compared to those with few/no PTSD symptoms (PCL-S<30). In covariate-adjusted analyses, probable PTSD remained associated with nonadherence (OR_Adjusted = 2.39, 95% CI 1.37, 4.18). The association between probable PTSD and nonadherence remained present in secondary analyses restricted to those with confirmed ACS (OR_Adjusted = 4.40, 95% CI 1.93, 10.03). Month since discharge (P<0.001) and Charlson score were also associated with nonadherence (P=.01).

Figure.

Electronically-Measured Adherence to Cardiovascular Medication in the 3 Months following Discharge in Patients with Suspected Acute Coronary Syndrome (N=400). Adherence is predicted as a function of month and PTSD group.

This study is the first to demonstrate that patients with ACS-induced PTSD symptoms are less adherent to risk-reducing cardiovascular medications. This association was present irrespective of whether the ACS was confirmed or ruled out and independent of depression. Results add to evidence that patients with PTSD symptoms are less likely to engage in risk-reducing health behaviors, and suggest medication nonadherence as a mechanism by which PTSD increases secondary cardiovascular risk. Strengths included electronic measurement of adherence and recruitment of diverse patients. Limitations included enrollment from a single urban ED; lack of assessment of reasons for nonadherence; and lack of adjustment for mental illness other than depression. There were also missing adherence data, especially in those with elevated PTSD symptoms. This may have led the association between PTSD and nonadherence to be underestimated. Overall, these findings suggest that screening and treatment for PTSD after acute cardiovascular events may be warranted. Interventions that address avoidance of traumatic reminders may be needed to optimize medication adherence in suspected ACS patients with PTSD symptoms.

Abbreviations:

PTSD

posttraumatic stress disorder

PCL

PTSD Checklist for DSM-IV-Specific Stressor (suspected acute coronary syndrome)