Association between Opioid Use and Atrial Fibrillation: the REasons for Geographic And Racial Differences in Stroke (REGARDS) Study

TO THE EDITOR

It has been estimated that over 4.3 million United States adults are taking opioids regularly in any given week.¹ Opioid receptors are down-regulated in animal models of atrial fibrillation (AF).² However, the association between opioid use and AF has not been examined in population-based studies. We examined the cross sectional association between prescription opioid use and AF using data from the REasons for Geographic And Racial Differences in Stroke (REGARDS) Study.

METHODS

Details of REGARDS and its design have been published before.³ Briefly, between 2003 and 2007, 30,239 participants were recruited using postal mailings and telephone calls from across the United States. Demographic information, medical histories, blood tests and electrocardiograms were obtained using a computer-assisted telephone interview system and in-home study visits by trained staff. The study was approved by institutional review boards at all participating centers. AF was identified by electrocardiogram and self-reported history of a previous physician diagnosis.⁴ Opioid use was ascertained by pill-bottle review during the in-home visit. The association between opioid use and AF was examined in multivariable adjusted logistic regression models using SAS version 9.3(SAS Inc. Cary, NC). Subgroup analyses by age, sex, race, baseline CHD, hypertension, and diabetes were also conducted.

RESULTS

A total of 24,632 participants (mean age: 65 ± 9.4 years; 54% women; 40% blacks) were included in the analysis. A total of 1,887 (7.6%) participants reported opioid use and 2,086 (8.5%) had AF. The most commonly used opioid was hydrocodone (n=779, 41% of opioid users) followed by propoxyphene (n=470, 25% of opioid users), and tramadol (n=378, 20% of opioid users). Several differences were observed between opioid users and non-users. Opioid users were slightly younger, more likely to be female, black and have cardiovascular comorbidities (Table 1). The prevalence of AF was higher in opioid users than non-users (12% vs. 8%, p <0.001). As shown in Table 2, opioid use was associated with increased odds of AF (OR=1.35, 95% CI=1.16, 1.57) after adjustment for potential confounders, and the results were consistent in several subgroups of REGARDS participants. Since it is possible that this association could be confounded by substance abuse, we further adjusted for benzodiazepines use and alcohol use. The association remained statistically significant (OR=1.29, 95% CI=1.11, 1.51). Also, given the known cardiotoxicity of propoxyphene, we excluded 434 participants on this drug in a sensitivity analysis, and the association remained statistically significant (OR=1.33, 95%CI=1.11, 1.58).

Table 1.

Characteristics of study participants by opioid use

Characteristic Mean ± SD; n (%)	Opioid Use (n=1,887)	No Opioid Use (n=22,745)	P-value*
Demographic variables
Age (years)	64 ± 9.4	65 ± 9.4	0.048
Male Sex	656 (35)	10,673 (47)	<0.001
Black	830 (44)	9,069 (40)	<0.001
Region			<0.001
Stroke Belt	706 (37)	7,790 (34)
Stroke Buckle	463 (25)	4,724 (21)
Non-belt	718 (38)	10,231 (45)
Education, High school or less	927 (49)	8,374 (37)	<0.001
Annual income, <$20,000	540 (29)	3,696 (16)	<0.001
Clinical variables
Current or former smoker	1,171 (62)	12,233 (54)	<0.001
Alcohol use			<0.001
Heavy	53 (3)	914 (4)
Moderate	457 (24)	7742 (34)
None	1377 (73)	14089 (62)
Diabetes	543 (29)	4,551 (20)	<0.001
Prior CHD	428 (23)	3,861 (17)	<0.001
Prior Stroke	170 (9.0)	1,274 (5.6)	<0.001
Aspirin use	802 (43)	9,960 (44)	0.28
Antihypertensive medication use	1,237 (66)	11,808 (52)	<0.001
Lipid-lowering medication use	728 (39)	7,555 (33)	<0.001
Body mass index (kg/m2)	31 ± 7.1	29 ± 6.0	<0.001
Systolic Blood Pressure (mm Hg)	129 ± 18	127 ± 16	0.002
Laboratory variables
Total cholesterol (mg/dL)	190 ± 42	192 ± 40	0.10
HDL-cholesterol (mg/dL)	51 ± 16	52 ± 16	0.10
Prior PAD	62 (3.3)	455 (2.0)	<0.001
Log(hs-CRP) (mg/L)	1.2 ± 1.2	0.76 ± 1.2	<0.001
Serum creatinine (mg/L)	0.92 ± 0.52	0.91 ± 0.40	0.14
Log(ACR) (mg/g)	2.4 ± 1.3	2.3 ± 1.2	<0.001
Left ventricular hypertrophy	210 (11)	2,196 (9.7)	0.04

^*Statistical significant for categorical variables tested using the chi-square method and for continuous variables the Wilcoxon-rank sum was used.

ACR=urine albumin-to-creatinine ratio; AF=atrial fibrillation, CHD=coronary heart disease; HDL=high-density lipoprotein; hs-CRP=high-sensitivity C-reactive protein; PAD=peripheral arterial disease; SD=standard deviation.

Alcohol use was classified by the number of drinks per week reported by study participants using the following criteria: none, moderate (1 to 2 drinks/day for men and 1 drink/day for women), and heavy (>2 drinks/day for men and >1 drink/day for women). Stroke belt includes states of Alabama, Arkansas, Indiana, Kentucky, Louisiana, Mississippi, Tennessee, Virginia; Stroke buckle include states of North Carolina, South Carolina and Georgia

Table 2.

Association of opioid use with atrial fibrillation, Overall and by subgroups of REGARDS participants

	*OR (95%CI)	P-value	Interaction P-value
All population	1.35 (1.16, 1.57)	<0.001	N/A
Age
<65 years	1.27 (1.01, 1.59)	0.04	0.95
≥65 years	1.41 (1.15, 1.73)	0.001
Sex
Female	1.35 (1.12, 1.63)	0.002	0.70
Male	1.36 (1.05, 1.76)	0.02
Race
Black	1.51 (1.20, 1.91)	<0.001	0.13
White	1.27 (1.04, 1.55)	0.02
Coronary Heart Disease
No	1.28 (1.06, 1.56)	0.01	0.54
Yes	1.44 (1.13, 1.83)	0.004
Hypertension
No	1.41 (1.03, 1.94)	0.03	0.86
Yes	1.33 (1.12, 1.58)	0.001
Diabetes
No	1.35 (1.13, 1.62)	0.001	0.94
Yes	1.32 (1.003, 1.73)	0.047

^*Adjusted for age, sex, race/ethnicity, region of residence, income, education, systolic blood pressure, HDL-cholesterol, total cholesterol, body mass index, smoking, diabetes, antihypertensive and lipid-lowering medications, aspirin, alcohol use, coronary heart disease, stroke, log (hs-CRP), serum creatinine, log(ACR), peripheral arterial disease, and left ventricular hypertrophy.

ACR=albumin-to-creatinine ratio; AF=atrial fibrillation; CI=confidence interval; hs-CRP=high-sensitivity C-reactive protein; OR=odds ratio.

DISCUSSION

In this analysis from the REGARDS study, a biracial community-based population study, opioid use was independently associated with increased prevalence of AF.

In the past, propoxyphene has been linked with fatal cardiac arrhythmia that led to eventual cessation of its sales in United States due to safety concerns.⁵ However, chronic arrhythmia like AF has not been linked with opioids. Endogenous opioid-peptides open mitochondrial K⁺_ATP channels, making mitochondria resistant to oxidative stress during episodes of ischemia. Loss of this protective mechanism against oxidative stress may render atrial-myocytes amenable to damage and thus lead to AF.⁶

Our results should be read in the context of certain limitations including the cross sectional design, possibility of residual confounding by unmeasured factors and lack of data on opioid dosage and length of therapy. Also, the use of self-reported history of a previous physician diagnosis as one of the methods to ascertain AF is subject to recall bias.

In conclusion, using data from the REGARDS study we showed that opioid use is associated with increased prevalence of AF. Over the past couple of decades, there have been significant increases in both opioid use and AF in United States. It is a thought provoking parallel temporal trend that needs further investigation.