Research Viewpoints Hospitalization Risk in CKD

Viewpoint On

Wong E, Ballew SH, Daya N, Ishigami J, Rebholz CM, Matsushita K, Grams ME, Coresh J: Hospitalization Risk among Older Adults with Chronic Kidney Disease. Am J Nephrol DOI 10.1159/000501539

Chronic kidney disease (CKD) as currently defined is believed to be associated with an increased risk for a variety of adverse events, including shortened life expectancy, end stage kidney disease (ESKD), cardiovascular disease (CVD), and acute kidney injury. All-cause risk for hospitalization is an understudied area of risk associated with CKD, particularly in the older age group which already has a heightened frequency of all-cause hospitalizations.

Wong and co-workers have remedied this deficiency by a very comprehensive observational analysis of hospitalization rates among 4706 older adults (average age 75.7 years) enrolled in the atherosclerosis risk in communities (ARIC) study, 29% of whom had CKD by estimated glomerular filtration rate (eGFR)-creatinine and albumin to creatinine ratio criteria according to the 2013 KDIGO schema. About 70% had an eGFR-creatinine ≥60 ml/min/1.73m2; 20% (G1/G2) had an eGFR creatinine of 45–59ml/min/1.73m2 (G3A) and about 9% had an eGFR creatinine <45ml/min/1.73m2 (G3B, G4/G5). Females exceeded males in all CKD categories, except those <30ml/min/1.73m2. African Americans constituted 21% of the cohort. When classified by eGFR cystatin C about 50% of the cohort were categorized as having CKD; with eGFR creatinine + cystatin C the frequency of CKD was intermediate at about 38%. Using a Reference Group of eGFR creatinine of 60-89ml/min/1.73m2 (G2) without albuminuria (A1), the relative of all-cause hospitalization was increased in a graded fashion from 1.2 (95% CI = 1.1-1.4) in category G3/A1 to 3.3 (95% CI= 2.0-5.5) in category G4/5. Hospitalization rates for CVD were increased in all categories of CKD.

Although CKD was diagnosed more frequently by eGFR cystatin C compared to eGFR creatinine, the pattern of increased relative risk for hospitalization according to CKD category was not appreciably altered. With eGFR creatinine values there was a tendency for an increase in all-cause hospitalization risk when the eGFR increased to >90ml/min/1.73m2 in the presence of abnormal albuminuria. This was less evident when eGFR cystatin C was used for CKD categorization. As expected, co-morbidity (diabetes, CVD, cancer, hypertension) was common in all groups and the prevalence was incrementally increased as eGFR declined and albuminuria increased (except for diabetes).

This study contains a wealth of information concerning the association of CKD category and hospitalization in a carefully characterized and longitudinally followed cohort of older adults with age-related co-morbidity. The effect of eGFR cystatin C on CKD categorization is not unexpected as the non-GFR determinants of serum cystatin C would likely be affected by the co-morbidity present in the cohort. The J-curve seen in eGFR creatinine categorization might have been due to frailty and sarcopenia. It is too bad that a “frailty score” was not included in the analysis. The reference group selected for the relative risk analysis had a rather broad range of eGFR (60-89 ml/min/1.73m2) and included values of GFR that are well above the “normal” values of eGFR for healthy subjects of 75 years of age. Whether this “reference group selection” effect confounded the analysis cannot be determined. About 77% of the C3A/A1 category had no abnormal albuminuria and most had some co-morbidity, including 35% with diabetes. One wonders if the modest risk of increased hospitalization in this category of CKD was due to co-morbidity rather than the CKD itself. It is very hard to disentangle CKD from co-morbidity in an observational analysis of this kind.

All things considered, this is a valuable addition to the literature associating risk of an adverse event (hospitalization in this case) and CKD. It points out that very high risk CKD (eGFR <45 ml/min/1.73m2; G3B/G4) and/or high grade albuminuria (A2) have very high hospitalization rates, often exceeding 500 per 1000 patient years of observation. This is a very tempting target for improved management. Case managers in medical care consortia need to pay attention to this data.