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

Disclosures

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

In This Article

Results

Baseline Characteristics and Comorbidities of Study Sample

Table 1 includes details of the demographic and clinical characteristics of the study population, categorized by the individual's number of drugs. The prevalence of polypharmacy in the study sample was 84.1%, with a prevalence of 44.7% for moderate polypharmacy (intake of five to eight drugs) and 39.4% for marked polypharmacy (intake of nine or more drugs), respectively. The median age of the study sample was 73.0 years (interquartile range [IQR] = 66.0-79.0), and 62.2% were male. The prevalence of most comorbidities and traditional cardiovascular risk factors increased significantly with the number of drugs. This trend was also reflected by the CCI. Also, the scores of risk measures such as CHA2DS2-Vasc and HAS-BLED increased with polypharmacy.

Concomitant Medication use

The mean number of medications used in addition to oral anticoagulation at baseline was 7.76 ± 3.28, with a range from zero to 24 (Supplementary Figure 1). Overall, 98.3% of study participants received phenprocoumon as an oral anticoagulant. The most commonly taken co-medications in the whole study population were diuretics and β-blocking agents (Supplementary Table 2). The frequency of concomitant medications known to increase or decrease the INR values (as listed in the recent summary of product characteristics for phenprocoumon and warfarin) and used by more than 5% of the study sample are presented in Table 1. The total number of co-medications was also related to increasing use of drugs that potentially influence the INR level stating that anticoagulated subjects with five drugs or more were using those drugs six times more frequently than participants on one to four drugs only. The most frequently used drugs that potentially influence the INR level in the group of subjects with five drugs or more were diuretics (72.9%), statins (46.6%), and low-dose acetylsalicylic acid (19.9%); in the medication classes not known to interact with VKA, it was β-blocking agents (67.4%), angiotensin-converting enzyme inhibitors (43.6%), and proton pump inhibitors (37.1%).

Supplemental Figure 1.

Distribution of number of medication in addition to vitamin K antagonist at baseline among 2,011 participants

Polypharmacy and Quality of Anticoagulation Control

Figure 1 outlines the quality of anticoagulation treatment stratified by the number of concomitant medications. Participants on nine or more drugs spend a significantly lower proportion of TTR compared with individuals on five to eight and one to four drugs, pattern A. In addition, the relative difference in time below the target range between medication groups is smaller than the relative difference in time above the target range (63% vs 88%), indicating a bigger shift toward higher INR values, pattern B.

Figure 1.

Quality of anticoagulation therapy stratified by the presence of polypharmacy.
The time in therapeutic range was calculated according to the Rosendaal method.14 Median values with corresponding interquartile ranges are displayed; hence percentages of totals do not add up to 100%.

Clinical Outcome According to the Number of Concomitant Drugs

During a mean follow-up period of 2.3 ± 1.0 years (maximum of 3 y), the overall incidence rates of thromboembolism, bleeding, hospitalization, and all-cause mortality were 3.4, 9.5, 62.8, and 9.6 events per 100 patients-years, respectively. The occurrence of thromboembolic events was not relevantly different with the extent of polypharmacy (RR5–8 drugs vs 1–4 drugs = 0.92; 95% CI = 0.53-1.66; p = .86; and RR≥9 drugs vs 1–4 drugs = 1.27; 95% CI = 0.74-2.26; p = .45). However, individuals on nine or more drugs had a higher incidence of clinically relevant bleeding events than subjects on one to four drugs (12.8 events per 100 patient-years vs 5.0 events per 100 patient-years; RR≥9 drugs vs 1–4 drugs = 2.58; 95% CI = 1.74-3.94; p < .001), and even subjects on five to eight drugs (12.8 events per 100 patient-years vs 8.5 events per 100 patient-years; RR≥9 drugs vs 5–8 drugs = 1.50; 95% CI = 1.20-1.89; p < .001; Supplementary Table 3). The 3-year cumulative incidence of bleeding increased across groups of medications (12.8% for 1–4 drugs; 19.7% for 5–8 drugs; 24.1% for ≥9 drugs). Similar relationships were observed for hospitalizations and all-cause mortality but not for thromboembolic events and the subtype major bleeding (Figure 2 and Supplementary Figure 2).

Figure 2.

Clinical outcome stratified by the presence of polypharmacy.

Supplemental Figure 2.

Cumulative incidence of major bleeding stratified by number of drugs

To evaluate specific effects of concomitant drugs on clinical outcome independent of the clinical profile, adjusted HRs were determined for thromboembolism, clinically relevant bleeding, hospitalizations, and all-cause mortality according to the categories of polypharmacy (Table 2). Marked polypharmacy was found to be independently associated with an increased risk of clinically relevant bleeding, even after adjusting for individuals' cardiovascular risk factors and comorbidities (HR = 1.62; 95% CI, 1.04-2.52; p = .033). Similarly, after adjustment for covariates, participants receiving nine or more drugs had a higher all-cause mortality compared with those taking between one to four drugs (HR = 2.16; 95% CI = 1.43-3.27; p < .001). The effects were smaller for the comparison of five to eight drugs vs one to four drugs than for nine or more drugs vs one to four drugs for the outcomes clinically relevant bleeding, hospitalization, and all-cause mortality, thus supporting a kind of dose-response relationship. However, effects were again not observed for the health outcome thromboembolic events (HR = 1.13; 95% CI = 0.60-2.13; p = .72) and the subtype major bleeding (HR = 1.28; 95% CI = 0.74-2.22; p = .37; Supplementary Table 4). When analyzing the number of drugs as the continuous variable, the risk of clinically relevant bleeding was increased by 4% (p = .05) per additional drug. Supplementary analyses regarding the effect of a combination of possible potentiating and/or inhibiting medication did not reveal any relevant associations in relation to thromboembolic events and clinically relevant bleeding (Supplementary Table 5). In Figure 3, sensitivity analyses are displayed for the outcome clinically relevant bleeding for predefined subgroups that reveal a high risk of bleeding per se. No significant interactions, however, were detected between the groups.

Figure 3.

Risk for clinically relevant bleeding in defined high-risk subgroups. CI, confidence interval; HR, hazard ratio.

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....