This study is the first to examine associations between emerging measures of processes of care for patients hospitalized with heart failure with early and long-term postdischarge clinical outcomes. The principal findings are threefold. First, hospital-level performance measures for any beta-blocker, evidence-based beta-blocker, aldosterone antagonist, and ICD were significantly associated with 1-year patient-level mortality. Second, fewer hospital-level performance measures were associated with early outcomes or 1-year cardiovascular readmission. Third, in a secondary analysis, associations for some but not all of the emerging performance measures at the patient level were similar to those at the hospital level. Although existing Hospital Quality Alliance heart failure performance measures have been shown not to be associated with patient-level or hospital-level mortality, alternative measures have meaningful associations with patient-centered outcomes. These findings have important implications for quality-improvement efforts, public reporting of adherence to performance measures, and heart failure pay-for-performance programs.
The hospital discharge period has been a focus of heart failure guidelines and performance measures because of the ease of access to patients; patients’ receptivity to health care recommendations; and the opportunity to implement, manage, and measure intervention strategies.16
Existing criteria for development of performance measures include quantifying numerators and denominators and evaluating the interpretability, applicability, and feasibility of proposed measures so they accurately reflect quality of care.13
Performance measures are meant to measure structural aspects or processes of care for which evidence is so strong that failure to perform them reduces the likelihood of optimal patient outcomes.17
Although there are a number of evidence-based therapies for patients with heart failure with LVSD, current Joint Commission and CMS heart failure measures include 3 measures—use of discharge instructions, smoking cessation counseling, and assessment of left ventricular function—that no controlled trial has specifically addressed in patients hospitalized with heart failure. Guideline recommendations for these measure are based on expert opinion.13
In previous studies, none of the 4 current performance measures for heart failure in inpatient settings was associated with patient-level or hospital-level mortality, and only the measure for angiotensin-converting enzyme (ACE) inhibitor or angiotensin receptor blocker (ARB) in patients with LVSD was associated with mortality or readmission.
This study provides evidence to support expansion of heart failure performance measures to include 4 measures associated with clinical outcomes. After adjustment for case mix, an absolute 10% increase in hospital adherence to these measures was significantly associated with a 5% to 8% lower risk of 1-year mortality. Associations for these measures were stronger than for current Hospital Quality Alliance and ACC/AHA performance measures. Moreover, use of current heart failure performance measures for public reporting and CMS pay-for-performance programs may not be the most efficacious way to assess quality, given the lack of association between most current performance measures and early and long-term outcomes.
Attention has increasingly focused on improvements in short-term outcomes, given the high mortality and readmission rates observed. Hospitals are profiled on these early outcomes, but it is unclear what can be done to improve outcomes. We found that the hospital-level aldosterone antagonist and ICD measures were associated with lower risk of 60-day mortality, after adjustment for case mix and other therapies. Moreover, the measures for any beta-blocker and evidence-based beta-blocker were associated with a trend toward lower 60-day mortality. A clinical trial of aldosterone antagonists in patients hospitalized with LVSD and heart failure after myocardial infarction found significantly lower all-cause mortality in the first 30 days after randomization. Although clinical trials for ICD therapy have focused on long-term outcomes in ambulatory patients with chronic heart failure, there may be effects on early sudden cardiac death after hospitalization.18,19
Alternatively, there may be other unmeasured processes of care or clinical characteristics associated with ICDs and early outcomes. The significant associations of the beta-blockers, aldosterone antagonists, and ICD measures with mortality provide a rationale for incorporation into standard performance measures of quality of care for heart failure.
A more difficult challenge is the selection of process measures for readmission, now a focus of public reporting by CMS and national initiatives.20
Use of evidence-based beta-blockers was associated with lower short-term and long-term cardiovascular readmission. Coupled with the association with lower long-term mortality, evidence-based beta-blocker use could also be considered for a standard performance measure. Other processes of care, such as referral to disease management, were not associated with lower cardiovascular readmission. In previous studies, disease management has been associated with lower readmission rates.21
This discrepancy may be attributable to the fact that the processes of care studied here were at the point of discharge. Referral at time of discharge does not necessarily indicate that the patient enrolled or adhered to the program. Furthermore, heart failure disease management programs can be heterogeneous, and recent studies have suggested limited to no impact on clinical outcomes with certain programs.22
This study also reveals potential discrepancies between analysis of hospital-level performance measures and analysis of patient-level performance measures. For example, although hospital-level conformity to the ICD performance measure was associated with improved long-term outcomes, conformity at the patient level was not significantly associated with outcomes. The registry records whether a patient was discharged with an ICD but not when the ICD was placed. As a consequence, the ICD group may include patients who were healthy enough to receive an ICD during the registry hospitalization as well as sicker-than-average patients who received an ICD months or years earlier. When the relationship between exposure and outcome is measured at the patient level, this heterogeneity biases the results toward the null. Relationships between hospital-level adherence and patient-level outcomes are likely to be less affected by such heterogeneity.
Other studies have examined the link between performance measures evaluated at the hospital level and postdischarge outcomes. Werner et al6
found that the Hospital Compare performance measures for myocardial infarction, heart failure, and pneumonia predicted small or no differences in risk-adjusted in-hospital mortality at the hospital level. They concluded that efforts to develop performance measures “tightly linked” to patient outcomes would be worthwhile. In an previous OPTIMIZE-HF study, none of the Hospital Quality Alliance performance measures were significantly associated with lower 60-day to 90-day postdischarge mortality; only the ACE inhibitor/ARB measure was associated with lower mortality or readmission at the patient level.8
However, a beta-blocker measure was associated with lower mortality and a combined end point of mortality or rehospitalization at 60 to 90 days. Jha et al23
found a modest relationship between heart failure performance measures and inpatient outcomes.
This study has several limitations. First, eligibility for processes of care was based on medical record documentation and thus depended on the accuracy of the documentation. Patients may have had undocumented contraindications or intolerances, which may have led to overestimation of the number of patients eligible for each performance measure. Second, despite the size of the study, the analysis may not have had sufficient statistical power to detect small but clinically important differences in postdischarge outcomes. Third, we were not able to adjust for socioeconomic factors or adherence. Inclusion of these variables would have diminished our ability to detect process–outcome links, making the finding that some measures did have process–outcome links more remarkable. There may also be other measured or unmeasured confounding variables that, had they been included, would have strengthened or weakened the process–outcome link for some measures. We did not assess health-related quality of life, functional capacity, patient satisfaction, or other clinical outcomes that may be of interest. Fourth, the study population included Medicare fee-for-service beneficiaries enrolled in OPTIMIZE-HF and may not be representative of all patients hospitalized with heart failure, though recent data suggest that patients in OPTIMIZE-HF are reasonably representative of Medicare beneficiaries. Finally, we were not able to determine whether a therapy was initiated or discontinued after discharge, whether patients referred to disease management actually participated, or the intensity and procedures of the disease management program after referral, which limits the assessment of improved conformity to a process measure after discharge.
Adherence to heart failure process measures for any beta-blocker, evidence-based beta-blocker, aldosterone antagonist, and ICD are significantly associated with postdischarge clinical outcomes and can be used to effectively discriminate quality of care at the hospital level. These measures could be considered for inclusion in heart failure performance measure sets. Given the moderate associations between individual process measures and clinical outcomes, it may be appropriate to include multiple new measures in hospital profiling efforts and incentive programs aimed at improving the quality of care for patients with heart failure.