β-blocker intensification (initiation or uptitration) occurred in approximately half of patients following index hospital discharge for HF. Using an adjusted matched case–control analysis to better accommodate the short-time window between exposure to β-blocker intensification and readmission, we found no significant increase in exposure to intensified β-blocker dose in the 30
days preceding hospital readmission; although, the low rates of intensification limited the ability to detect even moderate relative differences in readmission rates. In a separate analysis of the subgroup of patients initiated on β-blocker in the ambulatory setting, we found an association of borderline significance between higher starting doses β-blocker and subsequent short-term readmission events. Although the original rationale for the analysis was to test these associations, an equally important finding that emerged from the results was the low absolute rate of β-blocker intensification preceding readmission events (<3% in the prior 30
days) despite initiation and uptitration occurring in nearly half of patients at some point in follow up. This provides objective data that the short-term negative inotropic and chronotropic effects of β-blockers are unlikely to play a major role in the high readmission rates among patients with a history of HF hospitalization.
These data from contemporary clinical practice complement trials data. Randomized controlled trials have generally shown long-term decreases in hospitalization following the use of β-blockers in patients with HF and LVSD, effects which appear relatively early and continue over time [8
]. Additionally, initiation of β-blockers during hospitalization for worsening HF in a controlled trial (IMPACT-HF) was not associated with increased readmission at 60
]. The highly controlled, closely monitored setting of randomized trials among a select patient population may favor the beneficial effects of β-blockers while minimizing the potential adverse events. The ratio of benefit to risk in trials may be further optimized with rigid algorithms guiding β-blocker intensification. The applicability of trials to the general care setting is always a concern, especially if many of these safety measures are not routinely employed. Our data show relatively high starting doses for β-blockers and that higher doses were associated with a statistically significant but small absolute increase in readmission. These data thus support guideline recommendations to start β-blocker at low doses and uptitrate gradually.
Heart failure hospitalization and readmission are receiving increasing attention. However, the precipitants of heart failure hospitalizations remain unknown. While various interventions have been proposed to reduce readmission, specific targets for improvement remain elusive. The findings here are relevant to clinical practice and quality improvement initiatives as the low absolute attributable risk of β-blocker intensification for short-term readmission suggests that interventions to improve rates of β-blocker intensification among the LVSD population appear relatively safe. provided guideline recommended dosing is followed.
The overall utilization of β-blockers among patients with LVSD who would be expected to have a guideline indication for β-blocker therapy was 91%. The observed 9% rate of nonuse included nonadherence, as utilization data were derived from pharmacy fills rather than prescriptions. We were unable to perform a more detailed assessment of contraindications to β-blocker use, which might explain a portion of β-blocker non-use. Despite this high overall utilization rate in patients with LVSD, a more detailed assessment of intensification timing and dose maximization of β-blockers suggests potential areas for improvement. The median time from index hospital discharge to dose uptitration was nearly 4
months in patients with LVSD, far longer than is recommended by clinical practice guidelines. Furthermore, the maximum β-blocker doses achieved in this cohort was about half that seen in randomized trials. The reasons for these patterns are unclear. Clinicians may be reluctant to uptitrate β-blocker dose soon after hospitalization for HF decompensation, particularly in a patient population with relatively high levels of comorbidity. However, such reluctance was not seen in the median time to β-blocker initiation (1
week in patients with LVSD), making such an explanation less likely. An alternative explanation is that clinicians are responding to current quality measures which tend to assess whether patients with an indication for a therapy receive it or not, but at present do not typically capture performance on dose maximization.
Certain issues should be considered in the interpretation of this study. First, the retrospective cohort design precluded an assessment of clinical decisions around β-blocker. However, worsening symptoms should generally prompt clinicians to avoid rapid intensification of β-blockers, such that the direction of bias should be towards greater readmission in patients with less aggressive intensification. Because we found the opposite association, this should alleviate some concern for residual treatment selection bias. Second, KPCO is an integrated health organization, and the findings here may not be representative of the fee-for-service setting. Third, we were not able to characterized New York Heart Association functional classification, although we were able to adjust for LVEF and other clinical correlates of HF severity, including blood pressure, heart rate, kidney function, and ischemic heart disease, which are strongly associated with adverse outcomes in HF and rarely available in administrative datasets. Fourth, intensification of β-blockers was determined using pharmacy dispensing that may not capture β-blocker intensification (e.g.
doubling the dose) that occurred prior to a change in dispensing. Fortunately, such undocumented practices are discouraged in Kaiser’s integrated electronic health records. Fifth, much of the current attention to HF hospitalization has focused on 30-day readmission rates. Because of the need to create a 30-day window to assess pre-hospitalization exposure to β-blockers and because we found that multiple changes to therapies were occurring immediately following index hospital discharge, we had to exclude patients with readmission in these first 30
days. Sixth, our primary analysis looked at all patients with HF, whereas β-blockers have a HF indication only in HF patients with LVSD. However, the majority of patients with HF and preserved ejection fraction are prescribed β-blockers for a variety of reasons, and questions around short-term safety may apply to patients with a wide range of left ventricular ejection fraction. Finally, the study was underpowered to detect small relative differences, and our study is not adequately powered to look carefully at differences by important HF subgroups (e.g.
LVSD). However, this low power was due in part to a lower than expected intensification rate in the 30
days prior to readmission, which itself became one of the prevailing findings from the data.