LAUSR

Question In patients with chronic obstructive pulmonary disease (COPD), do long-acting 2-agonists (LABAs) or long-acting muscarinic antagonists (LAMAs) increase risk for cardiovascular (CV) events? Methods Design Nested casecontrol study using data from the Taiwan National Health Insurance Research Database, which includes medical and pharmacy claims from all medical settings. Setting Taiwan. Patients The base cohort included 284220 patients 40 years of age (mean age 71 y, 69% men) who had International Classification of Diseases, Ninth Revision, (ICD-9) codes for 1 inpatient visit or 2 outpatient visits for COPD in a single year between 2008 and Jun 2011, and 1 prescription for COPD medication at each visit (cohort entry date = date of first outpatient visit or hospital discharge for COPD). Exclusion criteria were lack of continuous insurance coverage or use of LABAs or LAMAS in the year before the cohort entry date. From the base cohort, 37719 cases had a CV event at a mean follow-up of 2 years (index date = first CV event date) and were individually matched with up to 4 randomly selected controls (n =146139) using risk-set sampling. Matching was based on cohort entry date and a regression-based disease risk score, estimated for each patient using multiple covariates. An additional 763 cases without a corresponding control were excluded. Risk factors Use of LABAs or LAMAs 1 year before the index date (current use = 30 d, recent use =31 to 90 d, past use =91 to 180 d, and remote use = >180 d; new use = current use plus no use at 31 to 365 d; prevalent use = current use plus any use at 31 to 365 d). Outcomes CV events (coronary heart disease, cardiac arrhythmia, heart failure, or ischemic stroke) based on ICD-9-Clinical Modification primary diagnosis codes for inpatient or emergency department visits. Main results 16% of cases and 15% of controls used any LABAs or LAMAs in the year before the index date. Associations between current LABA or LAMA use and CV events are shown in the Table. Recent, past, or remote use of LABAs or LAMAs were not linked to increased risk for CV events (data not shown). Conclusions In chronic obstructive pulmonary disease, new use of long-acting 2-agonists or long-acting muscarinic antagonists was linked to increased risk for cardiovascular events 30 days after initiation. The drugs were not linked to events occurring >30 days after initiation. Associations between current use vs nonuse of LABAs or LAMAs and CV events in chronic obstructive pulmonary disease* LABA or LAMA exposure Drugs Adjusted odds ratio (95% CI) for CV events Current use LABA 1.06 (0.99 to 1.12) LAMA 1.00 (0.92 to 1.10) LABA + LAMA 1.16 (1.05 to 1.28) Current new use LABA 1.50 (1.35 to 1.67) LAMA 1.52 (1.28 to 1.80) LABA + LAMA 2.03 (1.42 to 2.91) Current prevalent use LABA 0.91 (0.85 to 0.98) LAMA 0.88 (0.79 to 0.98) LABA + LAMA 1.11 (1.00 to 1.23) *CV = cardiovascular; LABA = long-acting 2-agonist; LAMA = long-acting muscarinic antagonist; other abbreviations defined in Glossary. Current use = 30 d before CV event date (index date) in cases or matched index date in controls. New use = current use and nonuse at 31 to 365 d before index date; prevalent use = current use and any use at 31 to 365 d. Adjusted for covariates with prespecified baseline differences. Odds ratio >1 indicates greater risk for CV events with the drugs; odds ratio <1 indicates less risk for CV events with the drugs. Commentary Because LABAs and LAMAs by definition act on the autonomic nervous system, it is plausible that they could increase CV events, yet most randomized controlled trials (RCTs) of these agents have not shown such increases (1). These RCTs are prone to healthy volunteer bias, and many of them excluded patients with known CV disease; therefore, the large, population-based, observational study by Wang and colleagues provides a valuable real-world perspective on this question. Unfortunately, the main findings are probably an artifact of residual confounding by indication, despite rigorous methodologic efforts and sensitivity analyses to try to address this problem. Patients with COPD typically start a LABA or LAMA when their disease is unstable, based on symptoms, exacerbations, or a decrease in lung functionall of which are accepted indications for therapy (2). Wang and colleagues had no data on symptoms or lung function and adjusted for frequency, but not recency, of exacerbations. Therefore, COPD instability, rather than the LABA or LAMA given in response, was probably the direct cause of the CV events observed. Moreover, the reversal in direction of the association over time seems biologically implausible; the increased odds of CV events with new use, but decreased odds with prevalent use, further suggest confounding. Notwithstanding the above, the association of dual LABALAMA use with an increased odds of CV events warrants greater attention; dual use showed a stronger association with CV events than either agent alone and was seen with both prevalent and new use in current users. This finding should be considered in the context of COPD guidelines that support dual LABALAMA therapy (2). COPD is a leading cause of morbidity and mortality, so the findings of Wang and colleagues should not dissuade clinicians from starting LABAs or LAMAs, the most effective COPD medications to date (2), when indicated. However, they should be vigilant for CV events when starting LABA or LAMA therapy, regardless of whether the drug or the disease is responsible.