Early detection of renal dysfunction in intensive care unit (ICU) patients is important. Indeed, an increase of the glomerular filtration rate (GFR) was demonstrated in some ICU patients populations by using urinary creatinine clearance (CL_CR) as a surrogate marker, and many of the drugs used in ICU patients need dose adjustment as a function of GFR 1 . Despite a normal serum creatinine measurement, a substantial number of burn patients demonstrated an increase in GFR with the need for increasing doses of renal elimination drugs to maintain therapeutic concentrations 1 2 . A study of our group showed that 42% of burn patients had a creatinine clearance greater than 120 mL minute^-1 1.73 m² 1 . Also, several studies performed in a general population of ICU patients suggested a poor correlation between serum creatinine concentration and GFR in polytrauma patients (PT) 3 4 5 . To the best of our knowledge, no study has specifically explored this population of PT patients.

The aim of this study, performed in a population of ICU patients with normal serum creatinine, was to estimate GFR, evaluated by measured CL_CR, in a population of PT patients through a comparison with a population of non-trauma patients (NPT).

Demographic and renal data are shown in Table 1. Two hundred, eighty-four patients were consecutively included in this observational study. The process of screening and inclusion in the study is shown in Figure 1 (flow chart).

The group of 144 PT patients was compared with the group of 140 NPT patients. No difference was found concerning hemodynamic data (Table 1). No difference was found concerning ventilation pressure. All the patients were ventilated with a tidal volume of 6 to 8 ml/kg, the PEP value was set at 5.8 ± 3 in PT vs 5.5 ± 3 in NPT patients (NS). FiO2 was 45 +/- 16% in PT vs 45 +/- 15 in NPT (NS), with Ph = 7.38 +/- 0.8 vs 7.39 +/- 0.8 (NS), PaO2 = 107 +/- 16 in PT vs 108 +/- 15 in NPT (NS), PaCO2 = 39 +/- 8 in PT vs 40 +/- 9, SaO2 = 97 +/- 3 vs 97 +/- 3 (NS). Glycemia was not different between groups (6.2 +/- 1.7 vs 6.1 +/- 1.8; NS). Twenty-three percent of PT vs 24% of NPT received norepinephrine (NS).

The overall results show that serum creatinine was normal (73 ± 22 μmol L^-1) and serum urea (8 ± 4 mmol L^-1) was slightly higher than the normal limits, but with no difference between groups. One hundred, six patients had a CL_CR above 120 mL minute^-1 1.73 m^-2, including 79 PT and 27 NPT (P < 0.0001). Only 63 patients had a CL_CR below 60 mL minute^-1 1.73 m^-2 with 15 PT and 48 NPT (P < 0.0001), whereas nine patients had a CL_CR below 30 mL minute^-1 1.73 m^-2, including two PT.

The overall urinary creatinine excretion was 929 ± 678 mg 24 h^-1 1.73 m^-2 for women and 1,369 ± 685 mg 24 h^-1 1.73 m^-2 for men. There was a significant difference between the urinary creatinine excretion of PT and NPT patients (1,489 ± 639 vs 969 ± 688 mg 24 h^-1 1.73 m^-2 respectively, P < 0.001). In the PT group, males had significantly higher urinary creatinine excretion than females (1,630 ± 644 vs 1,067 ± 392 mg 24 h^-1 1.73 m^-2, P < 0.001).

The overall measured CL_CR was 108 ± 57 mL minute^-1 1.73 m^-2. The CL_CR was higher in PT patients than in NPT patients when using measured CL_CR (131 ± 56 vs 85 ± 48 mL minute^-1 1.73 m^-2 respectively, P < 0.001).

Most patients with increased CL_CR (above 120 mL minute^-1 1.73 m^-2) were PT patients as shown in Table 2. On the opposite, most patients with decreased CL_CR (below 60 mL minute^-1 1.73 m^-2) were NPT patients.

Patients with CL_CR greater than 120 mL minute^-1 1.73 m ^-2 were younger (40 ± 16 years vs 56 ± 18 years), had a lower SAPS II score (43 ± 14 vs 50 ± 15) and a higher male ratio as compared with patients presenting a CL_CR lower than 120 mL minute.

All factors presenting a statistical difference between hyperfiltration and hypofiltration subgroups (Table 2) were analyzed. Through a logistic regression analysis, including goodness of fit of the model, age and trauma were the only factors independently correlated to CL_CR (Table 3).

The present results comparing a population of hemodynamic stable PT patients to a population of hemodynamic stable NPT patients with steady state serum creatinine concentration with a normal creatinine serum value demonstrate that (i) PT patients exhibit dramatic variations of their CrCl; (ii) CrCl is higher in PT patients than in NPT patients; (iii) Age and trauma are independently correlated factors to CL_CR in our study and in these study conditions.

Considering serum creatinine values, no significant difference was found between PT and NPT groups despite the variations of CrCl. These data demonstrated that a wide range of measured CL_CR variations exists and, therefore, confirm Hoste data obtained in critically ill patients with serum creatinine within normal range 16 . These authors demonstrated that "serum creatinine has a low sensitivity for detection of renal dysfunction". Our results also revealed some opposite trends between CrCl and creatinine measurements, as some patients had significantly lower values of serum creatinine for CL_CR > 60 mL minute^-1 1.73 m^-2 than for CL_CR < 60 mL minute^-1 1.73 m^-2. These data underline the inaccuracy of serum creatinine values in estimating the renal function.

Fifty-five percent of PT patients, and only 19% of NPT patients presented a measured CL_CR above 120 mL minute^-1 1.73 m^-2. In addition, only 10% of PT patients vs 34% of NPT patients presented a measured CL_CR below 60 mL minute^-1 1.73 m^-2. In clinical practice, the diagnosis of increased CL_CR as a surrogate marker of GFR is important and has largely been demonstrated in burn patients in the setting of antibiotics monitoring: ceftazidime, cefepime, vancomycin and amikacin 4 5 17 18 . In critically ill patients, high CL_CR required high doses of drugs, which are eliminated by the kidneys to obtain therapeutic concentration. Recently we confirmed the need for CL_CR monitoring in order to accurately monitor renal function and, therefore, to optimize the doses of antibiotics 4 19 .

Our results demonstrate that age, gender, ideal body weight, severity index, trauma patients, and serum creatinine are factors for a CL_CR above normal (> 120 mL minute^-1 1.73 m^-2), and for a moderate renal impairment (CL_CR < 60 mL minute^-1 1.73 m^-2). Through a logistic regression analysis, only two factors (age and trauma patients) remained significantly correlated with a CL_CR above normal and for a moderate renal impairment. In the current results, 12% of elderly patients (over 65 years) have a CL_CR greater than 120 mL minute^-1 1.73 m^-2. The impact of age on CL_CR is well known and this parameter was, therefore, introduced in the formulas estimating CL_CR (Cockcroft-Gault, Robert and simplified MDRD) 13 14 15 . The decrease in glomerular filtration, the involution of nephronic units and the reduction of the renal blood flow explain the high frequency of renal impairement in elderly patients. However, it should be kept in mind that glomerular ageing is correlated to age in only two-thirds of the patients, and this phenomenon accounts for the inaccuracy of the CL_CR estimated by calculated formulae 20 .

Current evidence suggests that PT (mainly, young patients without significant comorbidities) present with a CL_CR increase. However, this phenomenon as received little attention in the literature, and dose modification are therefore rarely considered. The present results clearly demonstrate for the first time that trauma is a major factor for CL_CR increase. Several factors may explain this increase in CL_CR in PT patients. First, urinary creatinine excretion may be involved in such a phenomenon. A higher creatinine urinary excretion was observed in PT compared with NPT patients whereas serum creatinine was similar in both groups. However, the higher creatinine urinary excretion observed in PT patients was within a normal range. Serum protein variations may impact our results. However, all the patients had a serum protein value between 50 and 55 gL^-1. It is, therefore, very unlikely that serum protein variations interfere with the present results. Also, regarding hemodynamics, CL_CR were studied and measured at a steady state in both groups (that is, distant from the admittance). It should be noted that our patients were hemodynamically stable at the time of data collection with no sign of dehydration. Although interference due to some cephalosporin has been described when the creatininemia was measured by using the Jaffe method 21 ; no changes in this parameter were observed during the overall period of the study. Sepsis can also reduce creatinine production as described in mice 22 . Moreover, in critically-ill patients, a positive fluid balance may lead to underestimation of the severity of AKI and delay the recognition of a 50% relative increase in sCr 23 . Finally, it should be hypothesized that humoral and inflammatory mechanisms encountered after severe trauma 24 or burn 1 are involved in the observed CL_CR increase.

The present study encountered some limitations. Increased CL_CR is related with enhanced renal elimination of circulating drugs. However, describing this phenomenon in terms of current available measures of the GFR at the bedside is still debated. In our study, the GFR was estimated by measured creatinine clearance on a 24-hour urine collection. However, the gold standard for GFR assessment is the measure of inuline clearance 25 but the cost and complexity of this tool limits its application in routine. Another limitation is the lack of consensus regarding the upper limit of normal GFR. However, increasing data support the concept of increased GFR in PT patients, and several reports demonstrated subtherapeutic concentrations of drugs in PT patients 5 26 . Also, a cross-sectional single 24-hr measure of CL_CR at 10 days in relatively stable patients was performed but fast modifications of kidney function may occur and there is a need for a continuous re-evaluation. Finally, some other factors may influence our results. In particular, fluid status, cardiac output may be significantly altered from baseline. However, whatever the causes of these alterations, the ATLS (Advanced Trauma Life support) principles are applied in our institution regarding resuscitation of PT patients. We, therefore, believe that the current results are broadly representative of the population of PT patients. It could be argued that the external validity of this single-center study may be limited. However, our findings may be relevant to the vast majority of level I trauma centers, provided that ATLS principles are applied in these institutions.

In hemodynamic ICU stable patients with steady state serum creatinine concentration, CL_CR, which is a surrogate marker of GFR, is higher in polytrauma patients than in other critically ill patients. In ICU patients, the drug monitoring must take into account the glomerular filtration rate. The measure of CL_CR should be routinely proposed for PT patients in order to adjust dose regimen, especially for drugs with renal elimination (betalactams, ceftazidime, cefepime, piperacillin, vancomycin, aminoglycosides, and so on).

• In ICU patients with normal serum creatinine, CL_CR, is higher in trauma than in non-trauma patients.

• The measure of CL_CR should be proposed in routine for ICU patients in order to adjust dose regimen, especially for drugs with renal elimination.

• Age and trauma were the only factors independently correlated to CL_CR.

• Glomerular filtration rate should be measured in ICU patient to detect renal filtration abnormalities.

• Serum creatinine is not a good marker for renal function estimation.

AKI: Acute Kidney Injury; CL_CR: creatinine clearance; GFR: glomerular filtration rate; ICU: intensive care unit; NPT: non polytrauma patients; PT: polytrauma patients; sMDRD, Modification of Diet in Renal Disease index; SOFA: score, Sequential Organ Failure Assessment score.

The authors declare that they have no competing interests.

AJ and IT carried out the serum creatinine measurement and calibration. SR, AB and TS carried out the patients' inclusions. KA helped to draft the manuscript and reviewed the intellectual content. BG, SS, OF and KS participated in the design of the study and helped to draft the manuscript. JMC and VM conceived of the study, and participated in its design and coordination and helped to draft the manuscript and performed the statistical analysis. All authors read and approved the final manuscript.