Population data suggest that there is a steady decline in estimated glomerular filtration rate (eGFR) as patients get older. However this takes little account of individual patient fluctuation; with many patients exceeding levels of change which should lead to referral to specialist care. In this study, we develop an algorithm to estimate the rate of change in renal function with age and eGFR as covariates. The produced chart enables clinicians to look up the probability that this change was within the normal range. To achieve this, we used a routine clinical database – the Quality Improvement in Chronic Kidney Disease (QICKD) trial database – in which 18,476 eligible patients have been identified. The key innovation here is the use of a regression model to smooth the underlying data, which enforces the dependency of the eGFR measurement on both the patient and time. A significant advantage of our approach is that daily fluctuation in the biological measurement is taken into account. This allows a clinician to reveal trends previously dismissed or ignored. As a possible application, we produced a plot of distribution of the rate of change in eGFR as a function of age and each level of eGFR. These data are also presented as a video online at: http://goo.gl/mtWfV. One the key findings is the presence of considerable variance of the rate of change in eGFR across patients at all ages and levels of renal function. Consequently, collating the data at the population level can adversely impact on the trend at the individual patient level. This points to the need for modeling the rate change at the patient level. Despite this, our timeand patient-enforced estimate is still sensitive enough to detect the minute change at the population level. This represents a significant improvement considering that the conventional approach that bins the patient by age group and eGFR value, as currently practiced, may not be able to detect such a minute change.
[1]
Adam Wright,et al.
Leveraging electronic health records to support chronic disease management : the need for temporal data views
,
2012
.
[2]
Bill Ravens,et al.
An Introduction to Copulas
,
2000,
Technometrics.
[3]
Marc Berg,et al.
Viewpoint Paper: Some Unintended Consequences of Information Technology in Health Care: The Nature of Patient Care Information System-related Errors
,
2003,
J. Am. Medical Informatics Assoc..
[4]
E. Luciano,et al.
Copula methods in finance
,
2004
.
[5]
Radford M. Neal.
Pattern Recognition and Machine Learning
,
2007,
Technometrics.
[6]
Jiaquan Xu,et al.
Deaths: preliminary data for 2009.
,
2011,
National vital statistics reports : from the Centers for Disease Control and Prevention, National Center for Health Statistics, National Vital Statistics System.
[7]
S. de Lusignan,et al.
The QICKD study protocol: a cluster randomised trial to compare quality improvement interventions to lower systolic BP in chronic kidney disease (CKD) in primary care
,
2009,
Implementation science : IS.