Discussion
Our study found that there was an independent graded association between eGFR levels and risks of both 1-month and 1-year mortalities in patients with intracerebral hemorrhage after adjusting age, stroke severity and comorbidities. The risk of mortality increased as the eGFR declined.
Renal dysfunction is a known independent risk factor for mortality after stroke. A Scotland study followed 2042 stroke patients for 7-year and found that reduced creatinine clearance (<51.27 ml/min) and raised serum creatinine (≥1.35 mg/dL) could predict deaths in patients with acute stroke8. But, the study failed to show renal function as a significant predictor of mortality in hemorrhagic stroke because of small size of fatal events.8 A recent study from Taiwan found renal function could play a highly significant role in predicting mortality among ischemic stroke patients.11 An Israel study assessed risk factors associated with deaths for patients with acute stroke during one-year follow-up.7 Results demonstrated that eGFR was a strong predictor of mortality and poor functional outcomes, such as nursing home dwelling and Barthel index ≦75. Hao et al. reported that an eGFR of < 60 ml/min/1.73 m2 was a strong predictor of mortality and disability for hemorrhagic stroke but not for ischemic stroke.18 Molshatzki revealed that presence of moderate to severe reduction of eGFR (< 45 ml/min/1.73 m2) was associated larger, lobar hematomas and hence higher one-year mortality in patients with intracerebral hemorrhage.16 Data from the GET WITH The Guidelines-Stroke (GWTG-Stroke) program showed that, after intracerebral hemorrhage, patients had an increasing risk of in-hospital mortality with declining eGFR.17 On the contrary, a study analyzing China National Stroke Registry found that low eGFR (< 45 ml/min/1.73 m2) was independently associated with an increased risk of mortality in diabetic patients with ischemic events, but not in those with hemorrhagic stroke19. Our study focused on patients with intracerebral hemorrhage and found that eGFR levels could predict the risk of mortality in a graded relationship. This finding both confirms the results of previous studies and provides new information on the mortality prediction for intracerebral hemorrhage.
Decline in renal function is associated with anemia, increased oxidative stress, abnormal apolipoprotein levels, inflammation, calcium-phosphate derangement, elevated uremic toxins, hypercoagulability, and impaired immunity.4,22-24 All these factors may contribute to the increased risks of adverse outcomes such as cardiovascular events, infection episodes, and deaths. These mechanisms may explain the graded association between impaired renal function and the risk of death in patients with hemorrhagic stroke. Our data also showed that most of the baseline comorbidities were reversely associated with eGFR levels, indicating that patients with low eGFR levels were more critically ill. In addition, patients with a lower eGFR also had a higher NIHSS score, indicating that impaired renal function could be associated with stroke severity.
The strength of this study is using a large sample size from stroke registry data with a representative group of stroke patients in Taiwan. Thus, we could estimate the real world prognosis of patients with intracerebral hemorrhage in Taiwan. In addition, we used the CKD-EPI equation to estimate the eGFR as it is better than the MDRD equation for Asian people.25 However, there are several limitations in this study. First, there is no data regarding to the quantity of proteinuria in the TSR database. We were unable to evaluate eGFR and proteinuria simultaneously. Proteinuria is an important and independent risk factor for cardiovascular disease.5,26 Second, serum creatinine was measured and recorded once at the time of admission and it was confounded by acute illness. Therefore, it is difficult to determine the chronicity of renal dysfunction. Furthermore, some patients were excluded from analysis because of missing information. The precision of measurement might be affected. However, the demographic characteristics, the prevalence of comorbidities, the values of laboratory data, and the percentage of people using medication prior to admission were similar between the selected and excluded patients (Table S2). Moreover, although data on hematoma volume, location, and the type of intracerebral hemorrhage were unavailable in this database, the NIHSS score as stroke severity was adjusted in the multivariate analysis. In addition, the use of direct oral anticoagulant was not recorded in this database. This is a limitation for external validity.