Chronic kidney disease (CKD) is an important and common medical problem that disproportionately affects older adults. Estimates from a national study suggest that 28% of adults aged 60 and older have at least stage III CKD (with estimated kidney function of <60 ml/min/1.73m2). CKD carries increased risk of mortality;indeed, particularly among older adults, the risk of mortality far exceeds their risk of ultimately requiring dialysis. The risk of mortality for older adults increases both with advancing age and with decreasing eGFR. Prognostic risk scores can be used in the clinical setting to elucidate individual patients'risk for adverse outcomes, and can subsequently help to provide more patient-centric care. At present, few predictive models have been developed to predict mortality in patients with CKD and none have been developed in older adults with CKD. Research has shown that predicting death in older adults with chronic illness is quite challenging;this is due at least in part to the fact that many patient characteristics which predict death in younger adults, such as specific co-morbid illnesses, do not have the same prognostic value in older adults. Instead, risk of death in older patients has been found to be related to functional measures (e.g., functional status, frailty, and depression), which may be harder to assess than more traditional health characteristics. The primary aim of this proposal is to develop risk scores which will predict short term (6 month) and long term (24 month) risk of death in adults aged 65 years and older with moderate to severe CKD (estimated kidney function of 15-30 ml/min/1.73m2). We will utilize Kaiser Permanente Northwest's patient database to examine those characteristics predictive of mortality in this population. We will delineate patients into two age cohorts: ages 65-74 and 75 and older. This is necessary because these two age cohorts will have different baseline risk of mortality (because age is such a powerful predictor of mortality);creating two different risk scores based on these age groups will allow us to examine other risk prediction characteristics in the context of these cohorts'different baseline mortality hazard. Using two Cox regression analyses, we will evaluate patient characteristics that might predict mortality within six months and within 24 months of patients'first estimated kidney function <30 mL/min/1.73m2. We will translate the Cox regression coefficients into a points-based risk score;the higher number of points will indicate a higher risk of mortality. We will assess the accuracy of each risk score by comparing predicted to observed risks within quintiles of patients. We will validate the model statistically using a bootstrapping re-sampling technique described by Harrell. We anticipate that development of this risk score will be a positive step towards helping care providers to predict mortality in older CKD patients, allowing more individualized and patient-centered care for this unique population.
Chronic kidney disease is estimated to affect 30% of older adults and is associated with an increased risk of death. The ability to define which older adults with chronic kidney disease will experience death in the short term (6 months) and longer term (24 months) could help clinicians provide more individualized and patient-centric care for this population. This project describes development of a risk score to predict the risk of death among older adults (ages 65 and older) with moderate to severe chronic kidney disease at 6 months and 24 months.
|Weiss, Jessica W; Platt, Robert W; Thorp, Micah L et al. (2015) Predicting mortality in older adults with kidney disease: a pragmatic prediction model. J Am Geriatr Soc 63:508-15|