Effect of Basal Insulin Glargine on First and Recurrent Episodes of Heart Failure Hospitalization

• heart failure
• hospitalization
• insulin glargine

Diabetes mellitus increases the incidence of recurrent heart failure through a variety of potential mechanisms, including ischemic myocardial damage, hyperglycemia, and the effects of the glucose-linked metabolic abnormalities on the myocardium.¹ To date, randomized controlled trials of more versus less glucose lowering with a menu of drugs, including insulin, have reported a neutral effect on incident heart failure. Nevertheless, epidemiological studies and the sodium-retaining properties of insulin have fueled concerns about a possible link.²

The ORIGIN trial (Outcome Reduction With Initial Glargine Intervention) allocated 12 537 people (mean age, 64 years; 35% women) with diabetes mellitus (88%), impaired glucose tolerance, or impaired fasting glucose to either basal insulin-mediated normoglycemia or therapy without insulin and reported a neutral effect on cardiovascular events during a median follow-up of 6.2 years. People with New York Heart Association class 3 or 4 heart failure were excluded. Heart failure requiring hospitalization was a prespecified component of the primary outcome and was defined as evidence of overnight hospitalization or attendance in an acute care setting for at least 2 of the following: signs or symptoms of heart failure; radiological evidence of congestive heart failure; treated with intravenous diuretics, or a first or increased dose of an oral diuretic, intravenous or oral vasodilator, or intravenous or oral inotrope.³ Both first and recurrent episodes were prospectively collected and blindly adjudicated; however, only analyses of the first heart failure hospitalization were prespecified. The trial was approved by local ethics committees at all sites, and all participants signed informed consent.

Cox regression was used to estimate the hazard of allocation to insulin glargine on the first heart failure hospitalization before and after accounting for the competing risk of death. It was also used to estimate the hazard of recurrent episodes in 2 ways. First, after adjustment for the competing risk of death, a joint frailty model was used that accounts for the multiplicity of events, the likelihood that a person who has had 1 event will have another, and the competing risk of death. Second, a proportional means model (an extension of the Cox model for the analyses of recurrent events) was used to estimate the hazard ratio.⁴ In addition, the effect of glargine allocation on recurrent events was estimated by calculating an event rate ratio with a Poisson model⁵ and by using the more conservative negative binomial model, which additionally accounts for the heterogeneity of each participant’s risk of recurrent events. P values were not adjusted for multiple comparisons. All statistical analyses were done with SAS version 9.4 (SAS Institute Inc, Cary, NC).

There was a high contrast in insulin use during the trial at both 2 years (89.7% and 3.5% of glargine and standard care participants, respectively) and 6 years (83.6% and 10.0%, respectively). During follow-up, the percentage of the 1916 deaths (2.6%/y) and 653 first heart failure hospitalizations (0.9%/y) did not differ by allocated treatment, whereas the 223 second heart failure hospitalizations (0.3%/y) occurred less frequently in participants allocated to insulin glargine compared with standard care (0.25%/y and 0.35%/y, respectively; P=0.01). Allocation to insulin glargine had a neutral effect (log-rank P=0.19) on the occurrence of at least 1 hospitalization for heart failure before (hazard ratio, 0.90; 95% confidence interval, 0.77–1.05)³ and after (hazard ratio, 0.90; 95% confidence interval, 0.77–1.05; P=0.19) accounting for the competing risk of death. Additional analyses with 4 different models (Figure) showed that allocation to insulin glargine generally had a neutral effect on recurrent events.

• Download figure
• Open in new tab
• Download powerpoint

Figure.

Risk of death and heart failure hospitalizations (HFHs). The effects of allocation to glargine on death, the composite of death or HFH, HFH accounting for the competing risk of death, and recurrent episodes of HFH using different methods are depicted as hazard ratios or, for the Poisson and negative binomial models, event rate ratios. The proportional means, Poisson, and negative binomial models do not account for the competing risk of death. CI indicates confidence interval.

These ORIGIN trial intention-to-treat analyses show that insulin glargine has a neutral effect on both initial and recurrent heart failure hospitalizations. They suggest that previous reports from observational studies linking insulin use to heart failure may have been affected by confounding of insulin use with other factors linked to a poor prognosis. The fact that ORIGIN participants were randomly allocated to either insulin or no insulin effectively deals with this possibility.

These findings are limited by the fact that information on history of New York Heart Association class 1 or 2 heart failure status was not collected. Nevertheless, the very large sample size provides reassurance that baseline history of heart failure was distributed equally across the treatment groups and that any postrandomization difference reflects the effect of the intervention. It is also limited by the inclusion of only those heart failure episodes serious enough to require hospitalization. Conversely, the high retention, adherence, and between-group differences in insulin use; adjustment for the competing risk of death; number of episodes; long follow-up; and blind adjudication of all heart failure hospitalizations are clear strengths. Overall, these findings are clearly reassuring for people using or considering the use of insulin and are relevant to future research into the effect of insulin on this serious and common cardiovascular outcome in people with diabetes mellitus.

• © 2018 American Heart Association, Inc.