1 Intensive Care Unit, 2 Renal Division and 3 Department of Clinical Chemistry, Ghent University Hospital, De Pintelaan 185, 9000 Gent, Belgium
Correspondence and offprint requests to: Eric Hoste, ICU, Ghent University Hospital, De Pintelaan 185, 9000 Gent, Belgium. Email: Erik.Hoste{at}UGent.be
![]() |
Abstract |
---|
![]() ![]() ![]() ![]() ![]() ![]() |
---|
Methods. This was a prospective observational study. A total of 28 adult patients with a serum creatinine <1.5 mg/dl, within the first week of ICU admission, were included in the study. Renal function was assessed with serum creatinine, timed 1 h urinary Ccr, and the CockcroftGault, MDRD and simplified MDRD equations.
Results. Serum creatinine was in the normal range in all patients. Despite this, measured urinary Ccr was <80 ml/min/1.73 m2 in 13 patients (46.4%), and <60 ml/min/1.73 m2 in seven patients (25%). Urinary creatinine levels were low, especially in patients with low Ccr, suggesting a depressed production of creatinine caused by pronounced muscle loss. Regression analysis and BlandAltman plots revealed that neither the CockcroftGault formula nor the MDRD equations were specific enough for assessment of renal function.
Conclusions. In recently admitted critically ill patients with normal serum creatinine, serum creatinine had a low sensitivity for detection of renal dysfunction. Furthermore, the CockcroftGault and MDRD equations were not adequate in assessing renal function.
Keywords: creatinine clearance; glomerular filtration rate; kidney failure; acute; kidney function; prediction equations; serum creatinine
![]() |
Introduction |
---|
![]() ![]() ![]() ![]() ![]() ![]() |
---|
Normal values obtained by a method for the assessment of renal function should correspond to a normal glomerular filtration rate (GFR). On the other hand, the method should also be able to detect a diminished GFR.
The most commonly used marker for the evaluation of renal function in ICU patients is serum creatinine, and the diagnosis of ARF is often defined on a single determination of serum creatinine [2,3]. However, serum creatinine may be not very suitable for this purpose. The serum creatinine level depends not only on renal elimination but also on creatinine generation, volume of distribution and renal elimination. Creatinine is metabolized from creatine released by the muscles, so that muscle mass and metabolic transformation of creatine have an impact on creatinine concentration. Many characteristics apart from renal function may influence creatinine concentration, such as age, gender and race: younger patients, males and blacks have higher serum creatinine levels for the same given GFR, compared with older patients, females and Caucasians [4,5]. In addition, serum creatinine and GFR are not linearly but hyperbolically related. Although an increase in serum creatinine from 4 to 8 mg/dl produces the same proportional decrease of GFR as an increase from 1 to 2 mg/dl, its clinical implication is completely different. Finally, critically ill patients are often in a non-steady state condition and it has been shown that changes of GFR are poorly reflected by daily changes in serum creatinine concentrations in patients with ARF [6].
Direct measurement of GFR, the golden standard for assessment of renal function, with exogenous substances such as inulin, non-radioactive contrast agents (iothalamate or iohexol) or radiolabelled compounds (e.g. [125I]iothalamate or technetium 99m-diethylenetriaminepenta-acetic acid) is not performed routinely in the ICU setting for practical reasons.
Urinary creatinine clearance (Ccr), however, can be easily measured in an ICU patient, and is the next best method for evaluation of renal function. However, clearance methods require a steady-state situation, a condition seldom fulfilled in critically ill patients. Changes in, for example, haemodynamic status can result in dramatic changes in renal function over a 24 h observation period. In order to circumvent this, we choose a short timed urine collection for calculation of urinary creatinine clearance in patients with stable renal function. According to previous studies, short time urinary creatinine clearances correlate well with longer urinary collection methods when patients are in a steady state [7,8].
Renal function can also be estimated by several equations based on serum creatinine and patient characteristics [4,5]. These equations can be easily calculated at the bedside with use of handheld computers. They are, however, not validated in critically ill patients.
Given these considerations, the objective of the study was the assessment of a single serum creatinine value for determination of renal function in a population of recently admitted (<1 week) critically ill patients with serum creatinine levels within the normal range, by comparing this with the measured urinary Ccr. In order to circumvent the impact of fluctuations in physiological conditions, Ccr was determined from a 1 h urine collection at the moment of creatinine determination. Additionaly, we evaluated three equations for determination of renal function.
Patients and methods
The study was conducted in the 20 bed surgical ICU for adult patients in the Ghent University Hospital, a tertiary referral centre. Patients older than 18 years, with an indwelling arterial catheter, a urinary bladder catheter, a diuresis >400 ml/day, a serum creatinine <1.5 mg/dl, not treated by diuretics, and with an ICU stay of <1 week were included in the study. Patients were excluded when there was no information regarding body weight before admission, when they were haemodynamically unstable, when they were recovering from ARF or developing ARF, or when they were transferred from another ICU. In order to ensure that we evaluated patients in a steady-state condition, we also excluded patients in whom the difference between serum creatinine on the day of the experiment and 24 h later was 15%. Inclusion and exclusion criteria were defined by our concern to obtain reliable estimates of renal function in a population which, according to current standards, could be considered in steady state, and as having no renal failure.
An exactly timed 1 h urine collection was obtained for measuring the urinary creatinine concentration, and serum creatinine was measured at the end of the 1 h urine collection period. These data allowed us to calculate the timed, 1 h urinary Ccr (Table 1). Additionally, renal function was assessed by application of three well known, and widely used, equations that estimate renal function on the basis of demographic characteristics and biochemical values: the CockcroftGault formula [4], the original MDRD (Modification of Diet in Renal Disease) formula [5] and the simplified MDRD formula [9] (Table 1). Because body weight in most ICU patients is increased due to fluid accumulation, this may falsify the calculations based on equations developed for the general population. Therefore, the predicted body weight, based on gender and height, was calculated for each patient. The predicted body weight for male patients was calculated as 50 + 0.91 (cm of height 152.4), and for female patients as 45.5 + 0.91 (cm of height 152.4) [10]. The CockcroftGault equation was calculated once with the predicted body weight, and once with the actual body weight before ICU admission. The MDRD equations calculate the GFR corrected for body surface area (ml/min per 1.73 m2). To allow comparison, the CockcroftGault equation and measured Ccr were also normalized to 1.73 m2. The body surface area (m2) of the patients was calculated as: weight (kg)0.425 x height (cm)0.725 x 71.84/10 000. The urinary creatinine loss per day was calculated as urinary creatinine concentration (mg/dl) x urinary volume (ml) x 24 h x 1.73/100/body surface area (m2), expressed as (mg/24 h/1.73 m2), and as urinary creatinine concentration (mg/dl) x urinary volume (ml) x 24 h/100/ actual body weight (kg) (mg/kg/24 h), expressed as (mg/kg/24 h).
|
![]() |
Results |
---|
![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
Patients with a Ccr lower than 80 ml/min/1.73 m2 had a lower 24 h creatinine excretion (Table 3). A higher proportion of patients with low Ccr were treated with vasoactive therapy or mechanical ventilation. There were no differences in age, gender, APACHE II score or length of preceding ICU stay.
|
|
|
|
![]() |
Discussion |
---|
![]() ![]() ![]() ![]() ![]() ![]() |
---|
The most plausible explanation for the low sensitivity of serum creatinine for detection of renal insufficiency in critically ill patients is the depressed production of creatinine, as was suggested by the approximately one-third decrease in 24 h urinary creatinine excretion. Patients with a measured Ccr <80 ml/min/1.73 m2 had an even more pronounced muscle loss, as evidenced by the low urinary creatinine excretion. This muscle loss was probably caused by protracted illness preceding ICU admission, as there was a trend that more patients in the low GFR group were hospitalized prior to ICU admission.
Serum creatinine is the result of generation, distribution and excretion of creatinine; a lower generation will therefore result in a lower serum concentration for the same GFR, if distribution and excretion remain the same. Serum creatinine is produced by non-enzymatic hydrolysis of creatine, the major sources of this creatine being the release from endogenous muscles and exogenous nutritional intake of meat. Muscle wasting will initially result in an increase in the serum creatinine level but, once the reduction in decreased muscle mass becomes more pronounced, there will be a decreased release of creatine from the muscles, and subsequently a decreased serum creatinine. The low levels of urinary creatinine excretion in our patients suggest that an important degree of muscle loss had occurred already, possibly even caused by catabolism preceding their ICU admission. Two other factors may have contributed further to the relatively low serum creatinine levels in proportion to the assessed renal function. The dietary intake of creatine was negligible as all study patients received artificial enteral and/or parenteral nutrition. In addition, creatine is also produced by the liver, and disturbed liver metabolism is often present in critically ill patients.
A higher proportion of patients in the group with Ccr <80 ml/min/1.73 m2 were treated with vasoactive therapy and mechanical ventilation. Despite the fact that we selected patients who were haemodynamically stable, we cannot rule out that the low urinary creatinine excretion may also be a reflection of transient reductions in GFR associated with vasoactive therapy, blood pressure fluctuations and changes in rates of fluid administration.
There is a striking analogy of our findings with those obtained in patients with cirrhosis or quadriplegia [13,14]. In both patient groups, serum creatinine levels are lower for the same given GFR compared with a normal population, and reduction in muscle mass plays an important role in both conditions. Equations for assessment of renal function are also not valid for these patients.
It can be argued that we studied a relatively old patient population (median age 58 years), in whom a relatively lower muscle mass could already have been apparent. It is uncertain whether the same conclusions will hold for patients of a younger age. Nevertheless, it should be stressed that the age of the present population reflects the age of patients currently admitted to the ICU ward [15].
In addition, the CockcroftGault equation, calculated either on actual body weight before admission or on predicted body weight, and the MDRD formulae did not correspond very stringently with measured Ccr. The MDRD formulae were derived from patients with elevated serum creatinine; the applicability to patients with normal serum creatinine levels is therefore unclear. In addition to this, all formulae were validated initially on populations of non-ICU patients in whom the specific factors that may influence the serum creatinine values in ICU patients probably did not have the same impact or did not even play a role. The same line of reasoning holds true for serum albumin and serum urea, both used in the MDRD formula, but not in the CockcroftGault equation. Serum albumin levels can often be low in ICU patients, e.g. by dilution by fluid overload, by leakage of albumin into the extravascular compartment as a consequence of a capillary leak syndrome, or by decreased production in the course of an inflammatory response. Serum urea in this population is subject to changes due to intravascular volume status, gastrointestinal haemorrhage and catabolism. Hence, serum creatinine, albumin and urea values could have been influenced by many variables not present in the original reference populations.
It might be a concern that a 1 h urine sample is not representative of the urine production during a 24 h period. However, it needs to be emphasized that the 24 h urine volume was comparable with the urine volume obtained after extrapolation of the 1 h urine volume to a 24 h urine volume.
An important practical implication of our study is that early recognition of ARF in patients with normal creatinine values on the basis of one single creatinine measurement may be suboptimal. Evaluation of renal function on the basis of a timed urinary clearance is the most direct method, if urine collections are obtained in a proper way, based on correct timing, and after emptying of the urine bladder. Assessments of renal function with the CockcroftGault or the MDRD equations in our setting were not valuable alternatives.
Since the discussion for an appropriate definition for ARF is still ongoing [3,16,17], a definition of ARF should not be based only on a specific threshold value of serum creatinine. This is in agreement with the proposed definition of ARF by the Acute Dialysis Quality Initiative (ADQI) Working Group. This group proposes to use the change in serum creatinine and/or diminished urinary output as a threshold for the definition of ARF (http://www.adqi.net) [18]. Alternatively, one could also use other markers of glomerular filtration, e.g. cystatin C, a marker already available for routine use in some hospitals, or the kidney injury molecule-1 (KIM-1), a marker currently under investigation [19]. A meta-analysis has shown that cystatin C is superior to serum creatinine as a marker of renal function; however, information on the validity of cystatin C in critically ill patients is still scarce [20].
In conclusion, we found that in this cohort of critically ill patients studied within the first week of ICU admission, and with normal serum creatinine values, serum creatinine proved a very insensitive screening test for the early detection of renal dysfunction. The most probable explanation is that a decreased creatinine load resulted in a lower serum creatinine for a given GFR. The CockcroftGault and the MDRD equations proved not acceptable as alternatives. Therefore, a timed urinary creatinine clearance should be used to detect impaired renal function in critically ill patients with normal serum creatinine level.
Conflict of interest statement. None declared.
![]() |
References |
---|
![]() ![]() ![]() ![]() ![]() ![]() |
---|
|