Worsened renal function (WRF) during heart failure (HF) hospitalization is associated with in-hospital mortality, but there are limited data regarding its relation to long-term outcomes after discharge. The influence of WRF resolution is also unknown. This retrospective study analyzed patients who received care from a large health system and had a primary hospital discharge diagnosis of HF from January 2000 to June 2008. Renal function was estimated from creatinine levels during hospitalization. The first available value was considered baseline. WRF was defined a creatinine increase ≥0.3 mg/dl on any subsequent hospital day compared to baseline. Persistent WRF was defined as having WRF at discharge. Proportional hazards regression, adjusting for baseline renal function and potential confounding factors, was used to assess time to rehospitalization or death. Of 2,465 patients who survived to discharge, 887 (36%) developed WRF. Median follow-up was 2.1 years. In adjusted models, WRF was associated with higher rates of postdischarge death or rehospitalization (hazard ratio [HR] 1.12, 95% confidence interval [CI] 1.02 to 1.22). Of those with WRF, 528 (60%) had persistent WRF, whereas 359 (40%) recovered. Persistent WRF was significantly associated with higher postdischarge event rates (HR 1.14, 95% CI 1.02 to 1.27), whereas transient WRF showed only a nonsignificant trend toward risk (HR 1.09, 95% CI 0.96 to 1.24). In conclusion, in patients surviving hospitalization for HF, WRF was associated with increased long-term mortality and rehospitalization, particularly if renal function did not recover by the time of discharge.
A recent reanalysis of a clinical trial cohort demonstrated that worsened renal function (WRF) persisting at hospital discharge is associated with increased risk of death at 1 year (hazard ratio [HR] 1.5). This leaves unanswered whether WRF overall is associated with long-term outcomes and the relative contributions of WRF category (persistent vs transient) to postdischarge outcomes. To clarify these issues we performed a retrospective analysis of all patients hospitalized for heart failure (HF) at Henry Ford Hospital over an 8-year period, assessing the relation of WRF in the hospital to long-term death or rehospitalization.
Methods
The study was approved by the institutional review board at Henry Ford Hospital (Detroit, Michigan). Using automated data sources, we identified all patients ≥18 with a primary hospital discharge diagnosis of HF (see Appendix A for the International Classification of Diseases, Ninth Revision [ICD-9] codes used) from January 1, 2000 to June 30, 2008. The index hospitalization was the first inpatient admission during the period of observation. Patients also had to be members of a health plan for ≥1 year before the index hospitalization and to have received their care through health system physicians. Patients were followed until they reached an end point (death or rehospitalization) or were censored at the earlier of disenrollment or final follow-up on December 31, 2008.
Data for this study came from electronic administrative databases maintained by the health system, vital records from the Michigan Department of Community Health, and the Death Master File (DMF) from the Social Security Administration. Administrative data captured claims (i.e., coded diagnoses, procedures, and prescription fills) occurring within and outside the health system. A master patient index contained demographic data (i.e., date of birth, gender, and race). Laboratory data were available for all tests performed within the health system. The DMF, available through the National Technical Information Service, was supplemented with the Michigan State Division of Vital Records and Health Statistics, which were queried with patients’ social security numbers.
We obtained all serum creatinine laboratory measurements from the index hospitalization. The first creatinine value during hospitalization or in the emergency department (if that was the route of admission) was considered the baseline value. On each day of hospitalization, the last creatinine measurement was taken as the value for that day. Consistent with previous studies, WRF was defined as a ≥0.3-mg/dl increase in creatinine on any subsequent hospital day compared to baseline. Persistent WRF was defined as the patient meeting the WRF definition at the last available creatinine value during that hospitalization; this was required to be within 24 hours of discharge. If the patient had met the WRF definition on any hospital day but no longer met the definition at discharge, they were considered to have transient WRF (i.e., recovery of renal function by discharge).
Covariates examined included age, race, gender, baseline creatinine, and baseline co-morbidities (i.e., previous atrial fibrillation, diabetes, hypertension, vascular disease, stroke, heart failure, and coronary disease). These covariates were included in all multivariate models to assess the independent association between WRF and the outcome under evaluation. Except for diabetes and hypertension, baseline co-morbidities were defined as having a primary or secondary ICD-9 diagnosis in any setting and in the year before the index date (i.e., baseline year). Diabetes mellitus and hypertension required 2 claims with the relevant ICD-9 diagnostic codes from any clinical setting or ≥1 primary diagnosis from a hospitalization in the baseline year. Alternatively, a baseline history of diabetes mellitus could be defined as ≥1 prescription filled for a diabetic medication in the baseline year ( Appendix B lists medications). Procedure codes ( Appendix C ) were also used to ascertain the presence of some co-morbidities.
The primary end point was time to death or rehospitalization for any cause. Hospital readmissions were identified from claims data, which were available for health plan members enrolled in this study. Deaths were identified using data obtained from health system administrative data, vital records from the State of Michigan, and the Social Security Administration DMF, as described earlier.
WRF categories were compared using chi-square tests for categorical variables or 2-sample Student’s t tests for continuous variables. Those variables that were not distributed normally were compared using a 2-sample Mann-Whitney test. Baseline creatinine values were natural-log transformed to normalize their distribution. Event rate estimates were generated using Kaplan-Meier curves and these curves were compared using log-rank tests. Proportional hazards regression models were used to assess the relation of WRF (WRF vs no WRF) and WRF category (no WRF vs transient WRF vs persistent WRF) with the composite end point of mortality or rehospitalization after discharge, with adjustment for all baseline covariates including renal function at admission (which was log transformed due to skewed distribution). Logistic regression was used to assess the relation of baseline patient and clinical factors to persistent WRF. All analyses were performed in SAS 9.1.3 (SAS Institute, Cary, North Carolina).
Results
In total 2,537 patients were initially identified. Of these, 72 died during the index hospitalization. In this group WRF was strongly associated with in-hospital death (HR 2.8, 95% confidence interval [CI] 1.73 to 4.52) and was associated with a near doubling in length of stay (3.8 vs 7.0 days, p <0.001). Excluding patients who did not survive to discharge left 2,465 meeting full inclusion criteria. Baseline characteristics of these 2,465 patients who made up the study cohort are listed in Table 1 . Overall, 887 patients (36%) developed WRF. This occurred at an average of 2.5 ± 3.73 days of hospitalization. Age, diabetes, loop diuretic use, peripheral vascular disease, and number of creatinine measurements were associated with WRF in bivariate analyses (all p values <0.05; Table 1 ).
| Variable | All Patients | WRF | No WRF | Persistent WRF | Transient WRF |
|---|---|---|---|---|---|
| (n = 2,465) | (n = 887) | (n = 1,578) | (n = 528) | (n = 359) | |
| Age (years), mean ± SD | 70.4 ± 13.4 | 71.2 ± 12.9 † | 69.9 ± 13.6 † | 72.2 ± 12.5 † | 69.8 ± 13.4 † |
| Women | 1,206 (48.9%) | 441 (49.7%) | 765 (48.5%) | 273 (51.7%) | 168 (46.8%) |
| Race | |||||
| Caucasian | 1,145 (47.7%) | 417 (48.7%) | 728 (47.2%) | 224 (44.0%) † | 193 (55.5%) † |
| African-American | 1,256 (51.0%) | 440 (49.6%) | 816 (51.7%) | 285 (54.0%) † | 155 (43.2%) † |
| Other | 64 (2.5%) | 30 (3.4%) | 34 (2.2%) | 19 (3.6%) † | 11 (3.1%) † |
| Baseline creatinine (mg/dl) | 1.31 ± 0.59 | 1.34 ± 0.63 | 1.29 ± 0.57 | 1.36 ± 0.69 | 1.31 ± 0.52 |
| Baseline hemoglobin (g/dl) | 12.0 ± 2.0 | 11.9 ± 2.0 | 12.0 ± 2.0 | 11.7 ± 2.0 † | 12.2 ± 2.0 † |
| Angiotensin-converting enzyme inhibitor/angiotensin receptor blocker use during hospitalization | 1,920 (77.9%) | 710 (80.1%) | 1,210 (76.7%) | 415 (78.6%) | 295 (82.2%) |
| Loop diuretic use during hospitalization | 2,336 (94.8%) | 870 (98.1%) † | 1,466 (92.9%) † | 518 (98.1%) | 352 (98.1%) |
| β-Blocker use during hospitalization | 1,224 (49.5%) | 428 (48.3%) | 796 (50.4%) | 251 (47.5%) | 177 (49.3%) |
| Previous atrial fibrillation | 722 (29.3%) | 266 (30.0%) | 456 (28.9%) | 152 (28.8%) | 114 (31.8%) |
| Previous diabetes | 1,058 (42.9%) | 406 (45.8%) † | 652 (41.3%) † | 244 (46.2%) | 162 (45.1%) |
| Previous hypertension | 1,648 (66.9%) | 612 (69.0%) | 1,036 (65.7%) | 378 (71.6%) † | 234 (65.2%) † |
| Previous peripheral vascular disease | 334 (13.6%) | 139 (15.7%) † | 195 (12.4%) † | 78 (14.8%) | 61 (17.0%) |
| Previous stroke | 390 (15.8%) | 156 (17.6%) | 234 (14.8%) | 89 (16.9%) | 67 (18.7%) |
| Previous heart failure | 1,229 (49.9%) | 439 (49.5%) | 790 (50.1%) | 252 (47.7%) | 187 (52.1%) |
| Previous coronary artery disease ⁎ | 749 (30.3%) | 284 (32.0%) | 464 (29.4%) | 172 (32.6%) | 112 (31.2%) |
| Days in hospital, median (quartiles 1–3) | 4 (2–6) | 5 (4–8) † | 3 (2–5) † | 4 (3–7) † | 6 (5–9) † |
| Proportion of days with creatinine values | 0.927 ± 0.12 | 0.914 ± 0.12 † | 0.934 ± 0.12 † | 0.920 ± 0.12 | 0.905 ± 0.11 |
| First day with worsened renal function, median (quartiles 1–3) | NA | 4 (3–6) | NA | 4 (3–7) † | 4 (3–6) † |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
