SP303QUANTIFICATION OF THE ASSOCIATION BETWEEN CHRONIC KIDNEY DISEASE AND CAUSE-SPECIFIC HOSPITALISATION: A UK POPULATION-BASED COHORT STUDY

Abstract

INTRODUCTION AND AIMS: Although chronic kidney disease (CKD) status or stage is associated with increased risk of all-cause hospitalisation, it remains unclear which causes of hospitalisation are specific to patients with CKD. Therefore, we aimed to quantify the association between CKD status and cause-specific hospitalisation in absolute and relative terms, and to identify causes of hospitalisation specific to patients with CKD, as compared with those without known CKD in the UK general population. METHOD(S): In Clinical Practice Research Datalink linked to Hospital Episode Statistics, we identified adult patients with CKD (estimated glomerular filtration rate <60 mL/min/1.73m2 twice for >=3 month) not on renal replacement therapy (RRT) between 2004 and 2014. As a control group, we selected patients without known CKD randomly from the database, matched for age, sex, general practice, and calendar time. Outcomes were hospitalisations due to 9 common conditions as the primary diagnosis after cohort entry: heart failure; myocardial infarction; cerebral infarction; pneumonia; urinary tract infection; gastrointestinal bleeding; intracranial bleeding; venous thromboembolism; and hip fracture. For each outcome, follow-up was until the first incidence of that outcome, death, RRT initiation, or last data collection. We calculated an incidence rate difference for each outcome between matched patients with and without CKD. Cox regression was used to estimate a relative risk (patients with vs. without CKD) for each outcome, stratifying by matched pair and adjusting further for ethnicity, socio-economic and smoking status, BMI and diagnoses of 17 chronic diseases in a UK pay for performance system (Quality Outcome Framework). Subgroup analysis was done by baseline CKD stage (GFR category 3a, 3b, and 4 or 5). Sensitivity analysis was conducted by regarding death and RRT initiation as competing risk. RESULT(S): We created a cohort of 242,349 matched pairs (median age 76 [IQR 70 - 82], male 39.3%) with and without CKD. The largest incidence rate difference was seen for heart failure with 6.6/1000 person-years (PY) (due to the incidence rate of 9.7 vs. 3.1/1000 PY in patients with and without CKD), followed by urinary tract infection with 5.2/1000 PY (13.1 vs. 7.9/1000 PY) and pneumonia with 4.4/1000 PY (12.6 vs. 8.2/ 1000 PY). Relative risk was largest for heart failure with adjusted hazard ratio of 1.66 (95% CI, 1.59 - 1.75), followed by venous thromboembolism with 1.55 (1.46 - 1.64) and myocardial infarction with 1.40 (1.34 - 1.46). In subgroup analysis by CKD stage, the effect size was also largest for heart failure (see Figure). Competing risk analysis did not change the rank of effect size: heart failure marked the largest sub-hazard ratio. CONCLUSION(S): Among the cause-specific hospitalisations studied, heart failure marked the largest absolute risk difference and relative risk between patients with and without CKD. Although CKD status and stage can be a marker of multimorbidity, priority should be given to prevention and outpatient-based management of heart failure for narrowing the gap of hospitalisation rate between patients with and without CKD.