Heart failure (HF) contributes to a high rate of hospitalizations. Acute kidney injury (AKI), with or without chronic kidney disease (CKD), is a common complication in patients with HF. The link between AKI and the risk for readmission for repeat episodes of HF is not well studied. In this study, 6,535 patients discharged with primary diagnoses of HF derived from a statewide inpatient database were examined. The association between AKI, with and without CKD, and risk for 30-day readmission with HF was assessed. Logistic regression was used to test the relations between predictor variables and outcomes. The mean age was 73.8 ± 14.6 years, and 51% of patients (n = 3,351) were women. AKI occurred in 6.5% of patients during the index hospitalization, whereas 16% had CKD. Nine hundred seventy-seven patients (15%) required readmission within 30 days for HF. Index hospital mortality was 1.7% in those without AKI or CKD compared to 11% and 13% in those with AKI without and with CKD, respectively (p <0.0001). Patients with AKI had a 30-day readmission rate of 21%, compared to 14% in those without AKI (p <0.0001). On multivariate analysis, AKI without CKD was associated with the highest risk for readmission (odds ratio 1.81, 95% confidence interval 1.35 to 2.39) compared to those with neither of the 2 diagnoses. In conclusion, patients with HF who have AKI experience a high rate of 30-day readmission for repeat episodes of HF. Reducing the risk for AKI, and follow-up monitoring after AKI, may improve care and reduce health care costs in patients with HF.
Heart failure (HF) is a growing public health problem in the United States, affecting >5 million patients, and is one of the most common reasons for hospitalization in the United States. The readmission rate after hospitalization with HF is reflective of the quality of care and health care costs, and it is nearly 23% at 30 days and 60% during 9-month follow-up. Acute kidney injury (AKI) is a common complication in patients admitted for HF. Several studies have examined the predictors and outcomes of AKI occurring in patients admitted for HF. There is limited information regarding the association between AKI and readmissions for HF, thus making it difficult to assess the potential biologic link between kidney dysfunction and progression of cardiovascular morbidity. In this study, we examined all patients discharged with primary diagnoses of HF from a statewide inpatient database (Washington State) to assess the relation between the diagnosis of AKI, with or without chronic kidney disease (CKD), and the risk for readmission within 30 days for a repeat episode of HF. We also sought to examine other indicators of resource utilization, such as length of stay during index hospitalization and total hospital days within 30 days of index discharge.
Methods
We examined the State Inpatient Database for Washington for 2006. These databases are developed as part of the Healthcare Cost and Utilization Project, a federal-state-industry partnership sponsored by the Agency for Healthcare Research and Quality. Of the 587,185 available discharges from January 1 to December 31, 2006, we included only those patients who had primary discharge diagnoses of HF, as identified by validated International Classification of Disease, Ninth Revision, diagnosis codes ( Appendix ). To allow sufficient time to examine readmissions, we only considered discharges occurring from January 1 to September 30, 2006, leaving 6,535 patients for analysis.
The primary end point was readmission to a hospital with either primary or secondary diagnosis of HF within 30 days of the index discharge. We also examined the secondary outcome of length of stay, assessed as hospital days during the index hospitalization as well as total number of days spent in the hospital during the index and 1 subsequent hospitalization occurring within a 30-day period. Major predictors of interest were the disease states of AKI and CKD (identified by International Classification of Disease, Ninth Revision, diagnosis codes, as listed in the Appendix ), associated with the index hospitalization for HF. On the basis of conditions, we classified patients into 4 groups: AKI present with CKD, AKI present without CKD, CKD present without AKI, and both AKI and CKD absent. From the State Inpatient Database file, we considered the following variables for the analysis: demographic information (age and gender), primary payer (Medicare, Medicaid, private insurance, self-pay, or other), resident location (urban or rural), median income (reported in 4 quartiles), number of procedures during index hospitalization (0, 1, 2, or ≥3), number of chronic conditions, and 27 of the 29 available co-morbid disease groups (HF and renal failure were not considered). For the analysis, we excluded patients aged <21 or >100 years at their index hospitalizations. We also removed patients with transfer visits.
Distributions of categorical variables are expressed as frequencies and continuous variables as mean ± SD. The overall rate of readmission and mortality were compared across patient groups using chi-square tests, while length of stay and total hospital days were compared using F tests. Analysis comparing mortality included all discharges as well as those patients who died during the index hospitalization. Using multivariate logistic regression, we assessed the relation between 4 disease groups of interest and the primary outcome of 30-day readmission due to HF. Absence of both AKI and CKD was chosen as a reference group, with the other 3 groups as comparators. We used the method of stepwise selection, allowing variables with bivariate p values ≤0.25 to be entered into the model. Certain variables (age and gender) were forced into the model regardless of their bivariate p values. The risk estimates were expressed as odds ratios and 95% confidence intervals. Two-tailed p values <0.05 were considered significant. All statistical analysis were conducted using JMP version 9.0 (SAS Institute Inc., Cary, North Carolina).
Results
Of the 6,535 patients discharged with HF, 51% (n = 3,351) were women. The mean age of the cohort was 73.8 ± 14.6 years. The distribution of patient characteristics for the overall cohort, and comparison by those who met the primary outcome, is listed in Table 1 . Patients discharged with HF also had chronic co-morbidities, with 81% of patients reported to have ≥5 chronic conditions. Among the most common co-morbid conditions associated with these patients were hypertension (50%), chronic pulmonary disease (31%), and diabetes mellitus (32%). Patients grouped by disease states of AKI and CKD by readmission status are also listed in Table 1 . As listed in Table 2 , the mean length of stay was longer in discharge groups of AKI with CKD (7.2 ± 21.9 days) and AKI without CKD (6.2 ± 6.0 days), compared to 3.6 ± 3.3 days in those with neither diagnosis ( Table 2 ). Index hospital mortality (analysis inclusive of deaths and discharges linked with index hospitalization; n = 6,714) was lowest in patients without AKI or CKD (1.73%) but was 11% in those with AKI without CKD and 13% in those with AKI in addition to CKD (p <0.0001).
| Patient characteristics | Readmission Within 30 Days | p Value | |
|---|---|---|---|
| No (n = 5,558 [85%]) | Yes (n = 977 [15%]) | ||
| Age (years) (mean 73.8 ± 14.6) | 73.87 (14.6) | 73.57 (14.58) | 0.5502 |
| Men (49.7%) | 2,706 (48.7%) | 480 (49.1%) | 0.836 |
| Women (51.3%) | 2,852 (51.3%) | 497 (50.9%) | |
| Primary payer | |||
| Medicare (70.8%) | 3,924 (70.6%) | 702 (71.8%) | 0.4261 |
| Medicaid (6.9%) | 362 (6.5%) | 87 (8.9%) | 0.0084 |
| Private insurance (18.4%) | 1,048 (25.3%) | 155 (15.9%) | 0.0238 |
| Self-pay (2.7%) | 157 (2.8%) | 23 (2.3%) | 0.3975 |
| Other (1.2%) | 67 (1.2%) | 10 (1.0%) | 0.6204 |
| Resident location | |||
| Urban (88.8%) | 4,937 (88.8%) | 869 (88.9%) | 0.9133 |
| Rural (11.2%) | 621 (11.2%) | 108 (11.0%) | 0.9134 |
| Median income | |||
| Low (32.3%) | 1,811 (32.6%) | 303 (31.0%) | 0.7078 |
| Medium (26.8%) | 1,466 (26.4%) | 283 (29%) | 0.0940 |
| High (22.2%) | 1,224 (22.0%) | 227 (23.2%) | 0.4026 |
| Very high (18.7%) | 1,057 (19.0%) | 164 (16.8%) | 0.0951 |
| Chronic conditions | 0.0093 | ||
| 0–4 (18.6%) | 1,062 (19.1%) | 153 (15.6%) | |
| ≥5 (81.4%) | 4,496 (80.9%) | 824 (84.3%) | |
| Disease group | |||
| AKI− CKD− (79.7%) | 4,486 (80.7%) | 725 (74.2%) | 0.0001 |
| AKI+ CKD− (4.6%) | 231 (4.1%) | 69 (7.0%) | 0.0002 |
| AKI+ CKD+ (1.9%) | 103 (1.8%) | 21 (2.1%) | 0.5383 |
| AKI− CKD+ (13.8%) | 738 (13.3%) | 162 (16.6%) | 0.0069 |
| AKI (6.48%) | 334 (6%) | 90 (9.2%) | 0.0003 |
| CKD (15.66%) | 841 (15.1%) | 183 (18.7%) | 0.0052 |
| Outcome Event | AKI− CKD− | AKI− CKD+ | AKI+ CKD− | AKI+ CKD+ | p Value |
|---|---|---|---|---|---|
| (n = 5,211) | (n = 900) | (n = 300) | (n = 124) | ||
| Length of stay, index hospitalization | 3.6 ± 3.3 | 4.1 ± 3.1 | 6.2 ± 6.0 | 7.2 ± 21.9 | <0.0001 |
| Total hospital days ⁎ | 5.2 ± 4.8 | 6.1 ± 4.7 | 8.5 ± 7.6 | 9.5 ± 7.1 | <0.0001 |
⁎ Total hospital days including index hospitalization and 1 subsequent readmission within 30 days.
Of the 6,535 patients discharged, 977 (15%) were readmitted within 30 days with diagnoses of HF. The 30-day readmission rate was 21% in those with AKI compared to 14.5% in those without AKI (p <0.0001). As shown in Figure 1 , AKI without CKD was associated with the highest rate of readmission (23%). Those without AKI or CKD experienced the lowest rate of readmission, at 14%. Table 3 indicates unadjusted and case mix–adjusted risk for 30-day readmission by patient groups. Other variables included in final model included age, gender, number of chronic conditions, primary payer, diabetes, valvular heart disease, drug abuse, and psychoses. AKI without CKD was associated with an increased risk for readmission (odds ratio 1.81, 95% confidence interval 1.35 to 2.39) compared to those patients with neither of the 2 diagnoses.
