Relevance of Polypharmacy for Clinical Outcome in Patients Receiving Vitamin K Antagonists

Lisa Eggebrecht, MSc; Markus Nagler, Dipl-Psych; Sebastian Göbel, MD; Heidrun Lamparter, Dipl-Psych; Karsten Keller, MD; Bianca Wagner, MPharm; Marina Panova-Noeva, MD, MSc, PhD; Vincent ten Cate, MSc; Christoph Bickel, MD; Michael Lauterbach, MD; Christine Espinola-Klein, MD; Roland Hardt, MD; Thomas Münzel, MD; Jürgen H. Prochaska, MD; Philipp S. Wild, MD, MSc


J Am Geriatr Soc. 2019;67(3):463-470. 

In This Article


Study Design

The rationale and design of the thrombEVAL study ( identifier NCT01809015) were previously described.[13] Briefly, the thrombEVAL study program comprises two prospective observational studies that investigated the quality of oral anticoagulation therapy in regular medical care and in a telemedicine-based coagulation service, both located in Rhineland-Palatinate, midwestern Germany. The study protocol conforms to the ethical guidelines of the Declaration of Helsinki as reflected in the a priori approvals by the local responsible Ethics Committee and the local data safety commissioner. All participants provided informed written consent to participate in this study.

Study Sample and Data Assessment

Between January 2011 and April 2, 2013, 2011 subjects were enrolled in the regular care cohort. The baseline examination comprised a standardized assessment of clinical status, sociodemographic variables, anthropometric measurements, medication, need for nursing care, and laboratory examinations. All study information underwent detailed quality control for completeness, plausibility, and validity according to prespecified criteria. Measurements of international normalized ratio (INR) values were collected retrospectively from study entry and during follow-up using inpatient laboratory data and the patients' anticoagulant card in regular medical care. The time in therapeutic range (TTR) was calculated according to the Rosendaal method.[14] Risk of stroke and major bleeding was determined according to the CHA2DS2-Vasc and HAS-BLED score, respectively.[15,16] Due to the nature of the study, design follow-up information was available for 1558 subjects.

Assessment of Polypharmacy

Medication use was defined as regular use (daily or weekly) and individually recorded from the prescription plan including active substances and daily doses of the medication. Drugs were coded according to the World Health Organization's Anatomical Therapeutic and Chemical (ATC) classification system.[17] The corresponding ATC codes are summarized in Supplementary Table 1. The analysis was based on the number of active substances and referred to as "number of drugs/medication" in this article. Drugs known to inhibit or potentiate the anticoagulant effect were studied and categorized according to the latest version of the summary of product characteristics form of phenprocoumon and warfarin.[18,19] Subjects were divided into the following three groups according to the number of concomitant medications (all participants taking at least one VKA): one to four, five to eight, or nine or more different medications.

Outcome Assessment

The clinical outcome comprised both anticoagulation-specific and non-anticoagulation-specific events: clinically relevant bleeding as a composite of major bleeding and clinically relevant nonmajor bleeding, and thromboembolic events defined as the occurrence of thrombosis, pulmonary embolism, systemic embolism, stroke, or myocardial infarction. Non-anticoagulation-specific events were defined as hospitalization and all-cause mortality.

Information on study end points was assessed in electronic case report forms via annual computer-assisted interviews with the study participants or from medical records. For all information on study end points, source data were obtained to evaluate validity. Subsequent adjudication of all reported end points was carried out by independent reviewers. In addition, electronic database systems of hospital records were screened for unreported events to reduce a potential recall bias. All-cause mortality was assessed via regular checks of the vital status at the registration offices.

Statistical Analysis

Baseline characteristics of the sample were expressed as absolute numbers and percentages for categorical variables, and medians with 25th and 75th percentile and means with corresponding standard deviation for continuous data. Disease burden was assessed using the age-adjusted Charlson Comorbidity Index (CCI).[20] Cumulative incidences were plotted graphically indicating the number of participants at risk in intervals of 4 months and for each outcome. Rate ratios (RRs; events per 100 patient-years) with corresponding 95% confidence interval (CIs) were calculated for outcome events according to categories of drug intake. Cox regression models with hazard ratios (HRs) with 95% CIs adjusted for potential confounders such as age, sex, traditional cardiovascular risk factors, and comorbidities were calculated to evaluate the effect of polypharmacy on the outcome. High-risk subgroups for bleeding per se were defined a priori and screened for potential interactions. All statistical analyses were conducted using the software program R, v.3.1.1 (