Professional Documents
Culture Documents
Introduction
The term ‘acute kidney injury’ (AKI) is currently recog Responses to a survey conducted in 2004 during an
nized as the preferred nomenclature for the complex international meeting on Critical Care Nephrology
clinical syndrome formerly known as acute renal failure revealed that clinicians tended to modify the AKI defi
(ARF). This transition in terminology also serves to nition according to the practices at their institution; more
emphasize that the spectrum of disease is much broader than 200 different definitions of AKI were provided.4 A
than the subset of patients who experience renal failure consensus definition was needed to bring order to the
requiring dialysis treatment. AKI occurs in a variety of literature on AKI. The success of this initiative can be
settings, and has clinical manifestations ranging from a gauged by the fact that in a similar survey conducted
minimal elevation in serum creatinine levels to anuric during an international meeting on Critical Care
renal failure.1 In fact, AKI exists along a continuum of Nephrology in 2007, the number of individualized AKI
disease: the acute decline in kidney function is often sec definitions had dropped to <20 and the vast majority of
ondary to an injury that causes functional or structural respondents were using the new consensus definition
changes in the kidneys. As the severity of the under of AKI to stratify the severity of renal dysfunction.5 In this
lying renal injury increases, the risk of an unfavorable Review, we briefly present the history, development and
outcome rises. clinical validation of the consensus definitions of AKI, Department of Pediatric
However, AKI is often overlooked, especially its milder which have been one of the most important develop Cardiac Surgery,
forms, even though all AKI severities can be associated ments in the management of this disorder. We also Bambino Gesù
Children’s Hospital,
with adverse outcomes. For example, in 2010 investiga describe possible modifications to these criteria based Piazza S. Onofrio 4,
tors reported that transient azotemia (defined as a rapid on new biomarkers and alternative measures of renal 00165 Rome, Italy
(Z. Ricci). International
recovery from AKI, within 72 h of its onset)2 was not a impairment, and consider their future role in studies Renal Research
benign condition. Transient azotemia represents about of AKI. Institute Vicenza, Viale
one-third of all cases of AKI in hospitalized patients, and Rodolfi 37, 36100
Vicenza, Italy
is independently correlated with a significantly increased The RIFLE criteria (D. N. Cruz).
risk of death (P <0.0001).2 An epidemiological study pub A consensus definition of AKI was published by the Department of
Nephrology, Dialysis
lished in 2004 demonstrated a wide variation in etiologies Acute Dialysis Quality Initiative (ADQI) in 2004.6 This and Transplantation, St
and risk factors for AKI and confirmed that this disease consensus definition is termed the RIFLE criteria, and Bortolo Hospital, Viale
is associated with increased mortality (particularly when uses the following categories: ‘at Risk’ is the least severe Rodolfi 37, 36100
Vicenza, Italy
dialysis is required).3 category of AKI, followed by ‘Injury’, ‘Failure’, ‘Loss’ and (C. Ronco).
‘End-stage renal disease’. This definition was intended
Correspondence to:
Competing interests to establish the presence or absence of clinical AKI in a Z. Ricci
The authors declare no competing interests. given patient or situation, and to describe the severity z.ricci@libero.it
those specified for the RIFLE class ‘at Risk’—are associ show any improvement in the sensitivity, robustness
ated with poor outcomes.13 Finally, the RIFLE criteria do and predictive ability of the definition and classifica
not consider the nature or site of the precipitating injury, tion of AKI.11,16 A secondary analysis of the Simplified
although the same could be said of consensus definitions Acute Physiology Score (SAPS) 3 database, published
used for other diseases.14 in 2009, evaluated and compared the performance of
the RIFLE and AKIN criteria. 17 The SAPS 3 database
AKIN criteria recorded serum creatinine levels and urine output for
In 2007, a modified version of the RIFLE criteria was 16,784 critically ill patients during the first 48 h after
published by the AKI Network (AKIN)—known as the ICU admission. In total, 1,504 patients (10.5%) were
AKIN criteria (Table 1).15 A number of national and classified as having AKI by the RIFLE criteria but not
international societies in the fields of nephrology by the AKIN criteria. The majority of these indivi
and critical care endorsed this initiative. 6 One of the duals were classified as ‘at Risk’ by the RIFLE crite
tasks of the AKIN was further refinement of the defini ria, and none showed a rise in serum creatinine level
tion of AKI: the categories of Risk, Injury, and Failure ≥26.5 μmol/l (0.3 mg/dl) within 48 h of ICU admis
were replaced with Stages 1, 2 and 3, respectively, and sion. By contrast, 504 patients (3.5%) were classified as
the outcome categories Loss and ESRD were eliminated. having AKI by the AKIN criteria but not by the RIFLE
An absolute increase in serum creatinine levels of at least criteria. Over 90% of these patients were considered to
26.5 μmol/l (0.3 mg/dl) has been added to the minimum have Stage 1 AKI: of these, 54.5% had a serum creati
requirements for Stage 1. Patients starting renal replace nine level <70.7 μmol/l (0.8 mg/dl) at ICU admission
ment therapy (RRT) are automatically classified as having and the remainder showed a rise in serum creatinine
Stage 3 AKI, regardless of their serum creatinine levels levels of ≥26.5 μmol/l (0.3 mg/dl) within 48 h of ICU
and urine output. admission. Several conclusions can be drawn from
The AKIN criteria have partially addressed some of the these results, as highlighted by Bagshaw in an edito
limitations of the RIFLE criteria. However, some small rial:18 the diagnosis of any AKI stage correlates with
details might have been overlooked. Although the AKIN clinical outcome; the RIFLE criteria seem to be more
criteria still do not define the nature of the renal injury, sensitive for diagnosing AKI and to have greater preci
some attempt has been made to exclude easily reversible sion for predicting mortality than the AKIN criteria; at
causes of azotemia (such as volume depletion or urinary best, these data confirm that neither classification offers
obstruction) from being classified as AKI.15 Thus, the clear advantages over the other; and both systems still
AKIN criteria state that “diagnosis based on the urine have imperfections and limitations. For example, both
criterion alone will require exclusion of urinary tract criteria diagnose AKI using surrogate markers of renal
obstructions that reduce urine output or of other easily impairment that manifest at a relatively late stage after
reversible causes of decreased urine output” and that injury has occurred, both rely on knowing the patient’s
application of the diagnostic criteria “should be used baseline serum creatinine levels, and both provide equal
in the context of the clinical presentation and following weighting to urine output and serum creatinine level for
adequate resuscitation when applicable”.15 This approach classifying the severity of AKI.
attempts to address the inability of the RIFLE criteria to
exclude prerenal azotemia. However, since the presence The issue of baseline renal function
of these factors is virtually impossible to either verify or The conceptual model of AKI is one of a rapid decline in
exclude in retrospective studies, transient azotemia must renal function from baseline levels. Aside from requir
indeed be acknowledged as a form of AKI that, even if ing a time-frame within which this decline occurs (for
mild, can lead to adverse outcomes.2 Consequently, the example, 48 h or 1 week), this model assumes that the
negative prognostic value of small changes in serum patient’s baseline serum creatinine level (which is reflec
creatinine levels has been recognized. Another crucial tive of the patient’s premorbid kidney function) is known.
difference between the RIFLE and AKIN criteria is the Furthermore, the baseline serum creatinine level is
48 h time-frame within which the diagnosis of AKI required to ascertain the extent of recovery of renal func
must be made to fulfill the AKIN criteria (Table 1). For tion after the AKI event, which is a clinically important
patients whose serum creatinine levels are not measured end point. Unfortunately, the baseline serum creatinine
at least every other day, which includes many patients level might not be known in many patients, depending
in general hospital wards, a diagnosis within this time- on the population being studied. The baseline serum
frame would be difficult to achieve. The GFR criteria creatinine value has been estimated in various ways, such
have been eliminated, which helps to discourage the as the serum creatinine level on hospital admission,17,19
incorrect use of changes in estimated GFR for AKI diag the minimum serum creatinine level during the hospital
nosis, as described above. Finally, because of the lack of stay,19,20 the serum creatinine value back-estimated from
universal guidelines for RRT initiation, this decision is the MDRD equation,21,22 or the lowest value among these.
highly subjective: therefore, using RRT initiation as a The choice of estimation technique used to obtain the
criterion for AKI staging could result in an unforeseen baseline serum creatinine value has a marked effect on
confounding effect. the prevalence of AKI, the severity (or stage) of disease,
Generally speaking, studies that have used the AKIN and on the mortality risk associated with various stages
criteria rather than the RIFLE criteria did not seem to of AKI.11,12
Table 2 | Surrogate measures for baseline serum creatinine used to establish AKI diagnosis and severity 24
Surrogate measures for AKI diagnosis (%) Misclassification of AKI severity by AKIN Stage (%)
baseline serum creatinine
Frequency* Sensitivity Specificity Any AKIN stage AKIN Stage 1 and 2 only
level
MDRD back-estimation 38.3 84.2 77.4 29.5 11.6
Nadir level in hospital 35.9 81.7 79.8 24.0 5.2
Level at hospital admission 13.7 38.9 94.9 18.0 4.9
*For comparison purposes, the frequency of AKI in patients for whom baseline serum creatinine levels were available was 25.5%. 24 Values <0.1 were imputed
at 0.05 to compose this table.24 Abbreviations: AKI, acute kidney injury; AKIN, AKI Network; MDRD; Modification of Diet in Renal Disease.
The serum creatinine level at hospital admission could value overestimated the incidence of AKI at 38.3%; the
have been modified by the acute illness that prompted sensitivity of this approach was 84.2% and its specificity
hospitalization and is unlikely to be representative of was 77.4% (Table 2).
the true baseline state. The lowest serum creatinine level A post hoc analysis of data from the Beginning and
measured during the hospital stay also has a number of Ending Supportive Therapy for the Kidney (BEST) study
disadvantages. First, this measurement is, by definition, a also compared observed baseline serum creatinine levels
retrospective baseline (the patient’s hospitalization must with MDRD back-estimated values for determination of
have ended to identify the nadir value and, therefore, this RIFLE class.23 Use of the MDRD back-estimated values
measurement cannot be used in daily clinical practice). resulted in a false-positive rate of 18.8% and a false-
Second, the nadir serum creatinine value is likely to be negative rate of 7.3% for detecting AKI. Misclassification
lower than the true baseline level, thereby overestimating rates were lower when patients with chronic kidney
the incidence of AKI. When no information on pre disease (CKD) were excluded. The authors concluded
morbid renal function is available, for adults, the Acute that the MDRD back-estimated serum creatinine value
Dialysis Quality Initiative (ADQI) has recommended performed reasonably well for the determination of the
back-estimation of the baseline serum creatinine value RIFLE categories when premorbid renal function was
using the MDRD formula, assuming an estimated GFR of near-normal, but cautioned against the use of MDRD
75 ml/min/1.73 m2.6 Although convenient, the validity back-estimated serum creatinine values in patients with
of this approximation, as well as that of other surrogate suspected CKD. This warning is clearly logical, since the
measures of baseline renal function, has been the subject MDRD estimation method assumes that the patient has
of studies published since 2009.23–25 A study addressing near-normal premorbid renal function, an assumption
the issue of baseline serum creatinine levels in pediatric that is obviously invalid in individuals with suspected
patients with AKI has also been published.26 CKD—for example, patients with proteinuria on routine
In a study published in 2010, involving 4,863 hospital admission urinalysis, or patients with multiple risk
ized patients, investigators evaluated the performance of factors for CKD.
these three potential surrogates for baseline serum creati In a study published in 2010, the researchers compared
nine level (the serum creatinine value back-estimated the bias and accuracy of three methods that estimate
from the MDRD equation [as per ADQI recommenda baseline serum creatinine values and the incidence of AKI
tions], the serum creatinine level on hospital admission, in the ICU setting using the RIFLE criteria:25 the MDRD
and the nadir inpatient serum creatinine level).24 The back-estimated serum creatinine value, serum creatinine
authors concluded that all three surrogates resulted in values back-estimated by a newly developed equation, and
bidirectional misclassification of AKI. Although the a sex-specific, fixed baseline serum creatinine value of
use of serum creatinine levels at admission resulted in 70 μmol/l (0.8 mg/dl) for women and 88 μmol/l (1 mg/dl)
the least amount of misclassification in this study, this for men. Neither of the alternative methods offered a
parameter also had the lowest sensitivity (38.9%) for consistent improvement in the accuracy of AKI diagnosis
diagnosis of AKI (Table 2). When admission serum compared with MDRD-based estimates. Moreover, use
creatinine levels were used, nearly half of the ‘true’ AKI of the MDRD equation to back-estimate baseline serum
cases were missed: 25.5% of the cohort had AKI when creatinine values overestimated or underestimated cases
true baseline serum creatinine levels were used, versus of ‘Risk’, but rarely misclassified patients in ‘Injury’ and
13.7% with use of admission serum creatinine levels. ‘Failure’ categories. The same trend was seen in the study
The missed cases included community-acquired AKI, of Siew et al. (Table 2).24
which might improve or stabilize during hospitaliza Back-estimation of baseline serum creatinine values
tion. Increased mortality rates were associated with is thought to be more reliable in pediatric populations
each AKIN stage when admission serum creatinine than in adults, owing to the reduced prevalence of CKD
levels were used, but this finding reflects the bias inher among children.26 Zappitelli and colleagues26 compared
ent in this method. Only AKI that continues to worsen the performance of true baseline serum creatinine levels
during hospitalization can be detected by this method, with that of five methods for estimating baseline serum
and such a condition is associated with higher mortality creatinine values in critically ill and non-critically-ill
than transient AKI. In the same study, use of the MDRD pediatric populations: serum creatinine levels at ICU
equation to back-estimate the baseline serum creatinine admission (although none of the non-critically-ill
patients had serum creatinine levels measured at admis The issue of how to approach the lack of a true
sion); serum creatinine values back-calculated from baseline serum creatinine level in a patient with
the Schwartz formula27 assuming an estimated creati community-acquired AKI also remains unresolved. For
nine clearance of either 100 ml/min/1.73 m2 or 120 ml/ example, if a patient presents to the emergency room
min/1.73 m2; and use of either the upper limit or lower with an elevated serum creatinine level, the absence of a
limit of normal serum creatinine levels adjusted for known baseline serum creatinine level means that only
age and sex. Different methods of estimating baseline a retrospective diagnosis of community-acquired AKI
serum creatinine values led to marked differences in is possible, inferred from the downward trajectory of
the calculated incidence of AKI in the pediatric ICU, serum creatinine levels during the patient’s hospitaliza
from 12% using the admission serum creatinine value tion. The studies discussed above have pointed out the
to 87.8% using the lower limit of normal serum creati drawbacks associated with using back-estimated serum
nine level adjusted for age and sex. A similar trend was creatinine values (using either the MDRD or Schwartz
seen in the non-critically-ill group. When the research equations); however, none has been able to suggest a
ers compared the performance of estimated creatinine suitable alternative. As expected, the estimated values
values back-calculated using a creatinine clearance rate seem to result in more bias in a CKD population (com
of 120 ml/min/1.73 m2 with the performance of using the pared with a non-CKD population). Moving forward
lower limit of normal serum creatinine level adjusted for from this finding, a potential solution to the dilemma
age and sex, the median absolute percentage error was would be to use a different method of estimating the
smallest using the adjusted lower limit of normal serum baseline serum creatinine value (for example, back-
creatinine level estimation method, but the 5–95th per estimation using a GFR of 60 ml/min/1.73 m2 or 45 ml/
centile range of this error was substantially narrower min/1.73 m 2) in patients with community-acquired
when estimated creatinine values were back-calculated AKI who have multiple risk factors for CKD but lack
using a creatinine clearance rate of 120 ml/min/1.73 m2. a measurement of their premorbid serum creatinine
The authors of the study proposed that the latter method level. Such an approach would need to be tested in
should be used to estimate baseline serum creatinine rigorous study.
levels when the true value is unknown.
Clearly, some degree of misclassification of AKI occurs The issue of urine output criteria
when using any method of estimating the baseline Urine output is one of the most commonly used AKI
serum creatinine value in both adults and children, an markers: it can be measured in real time at the hospi
observation that highlights the need for investigators to tal bedside, it is easy and inexpensive to determine and
make every effort to find and use a true baseline serum urine flow variations generally trigger first attempts of
creatinine level before resorting to the use of estimated therapy. However, urine output is affected by the patient’s
values.11,12 A baseline serum creatinine level that is appro blood volume status and by diuretic use. Furthermore,
priate for use in diagnosis of AKI should be representa 6 h and 12 h urine output can only be accurately assessed
tive of the patient’s normal renal function, such as an in patients with a urinary catheter. The AKIN and RIFLE
outpatient serum creatinine level measured within the criteria both use urine output to define the severity of
previous year (preferably within the past 3 months).28 AKI, but the accuracy of this parameter has been less
When an outpatient serum creatinine level is simply studied than have changes in serum creatinine levels,
not available, the serum creatinine measurement taken since exact data on urine output are not always avail
at either hospital or ICU admission seems to result in less able in retrospective studies.8 In the aforementioned
misclassification of AKI cases than does back-estimation systematic review,7 the relative risk of death in studies
using the MDRD equation. However, use of admission that used both changes in serum creatinine level and
serum creatinine levels is specific but insensitive, and urine output criteria to assess the severity of AKI was
is likely to underestimate the incidence of AKI. 24,26 lower than in studies that used the creatinine criteria
Furthermore, this approach can only detect hospital- only. A possible explanation for this finding is that the
acquired AKI. Nevertheless, as remarked by the Program urine output criteria overestimate the severity of AKI;
to Improve Care in Acute Renal Disease (PICARD) study thus, a sudden decrease in urine output (which would
investigators,29 serum creatinine levels throughout ICU eventually be corrected by fluid infusion or diuretic
admission might be affected by hemodilution, which fre administration) might not correspond to an actual
quently occurs in critically ill patients. This factor could increase in serum creatinine levels but would still be
lead to underestimation of the severity of AKI in these ranked as a high-severity AKI class. Finally, urine output
patients; if this factor is not taken into account, hemo is probably reduced early in the pathogenesis of prerenal
dilution could mask the increase in serum creatinine azotemia but late in the course of established AKI and,
levels associated with AKI. In the PICARD study, this consequently, the therapy prescribed might correspond
effect was revealed by daily evaluations and adjustments to markedly different clinical situations even if such
of serum creatinine levels according to the patient’s fluid patients exhibit similar AKI stages.
balance. Use of nonadjusted serum creatinine levels
would have led to a median underestimation of the true Consensus definitions using novel biomarkers
serum levels of 2.1% on day 1, with a progressive increase In view of the limitations of using either urine output or
to a median of 14.3% on day 6.29 changes in serum creatinine levels to assess the severity
Table 3 | Relationship between biomarkers and RIFLE class or AKIN stage in adults
Study Biomarker Setting AKI criteria AKI severity by level of biomarkers
Siew et al. (2009) 32
Urine NGAL* ICU AKIN Non-AKI: 48.1 (12.7–179.3) ng/ml‡
Stage 1: 158 (27–430) ng/ml‡
Stage 2: 390 (39–4,317) ng/ml‡§
Stage 3: 390 (39–4,317) ng/ml‡§
Cruz et al. (2010)31 Plasma NGAL|| ICU RIFLE Non-AKI: 102.4 (67.4–147.5) ng/ml‡
Risk: 171.4 (101.9–281.7) ng/ml‡
Injury 2: 214.9 (137.5–419.5) ng/ml‡
Failure: 620.3 (406.4–1,300.0) ng/ml‡
Haase-Fielitz et al. Plasma NGAL (ng/ml)¶ Undergoing RIFLE and AKIN Non-AKI: ≤150 ng/ml
(2009)40 cardiac surgery Risk or Stage 1: >150 ng/ml
Injury or Stage 2: >150 ng/ml
Failure or Stage 3: >240 ng/ml
Herget-Rosenthal Serum cystatin C (% ICU and high risk RIFLE Non-AKI: 48.1 <50% increase
et al. (2004)35 increase from baseline)# of AKI Risk: ≥50% increase
Injury: ≥100% increase
Failure: ≥200% increase
*At enrollment. ‡Values presented as median (interquartile range). §Values for pooled group of Stage 2 and Stage 3 patients. ||Peak levels during the first 4 days
in the ICU. ¶Measured 6 h after the start of cardiopulmonary bypass. #Levels at enrollment were used to define AKI. Abbreviations: AKI, acute kidney injury; AKIN,
AKI Network; ICU, intensive care unit; NGAL, neutrophil gelatinase-associated lipocalin; RIFLE, Risk, Injury, Failure, Loss, End-stage renal disease.
of AKI discussed above, various research groups have in serum cystatin C level from a baseline value, using
proposed that novel AKI biomarkers could be used to cut-offs analogous to those for serum creatinine in the
define and stage AKI in conjunction with the RIFLE and RIFLE criteria.35
AKIN criteria. The Neutrophil Gelatinase-Associated Overall, these studies suggest that new biomarkers
Lipocalin (NGAL) Investigator Group has published the could be potentially useful in determining a patient’s
results of a meta-analysis of data from 19 studies con RIFLE or AKIN class. If confirmed by future studies, the
ducted in eight countries, involving 2,538 patients, of incorporation of new biomarkers into current consensus
whom 487 (19.2%) developed AKI.30 The group reported AKI definitions could represent an important and very
that NGAL levels in plasma, serum or urine seem to be useful improvement.11 These results are encouraging, but
of diagnostic and prognostic value for AKI, RRT and are far from definitive. Important challenges remain in
mortality, especially in patients who have undergone creating a novel biomarker-based definition of AKI for
cardiac surgery, and in children. NGAL allows AKI diag clinical use. For example, all biomarker studies have used
nosis a few hours after the onset of kidney damage with serum creatinine levels, with or without urine output, as
elevated specificity and sensitivity. At present, limited the reference standard to which the new biomarker is
data are available describing the associations between compared. Experts have, however, lamented the inad
levels of this biomarker and either RIFLE class or AKIN equacies of both serum creatinine changes and urine
stage, but a few studies have demonstrated a correla output as biomarkers of renal impairment. If we ‘validate’
tion between the severity of AKI and levels of NGAL new tests against flawed reference standards, then this
in blood and urine. In an adult ICU population, peak methodology should be considered in the interpretation
plasma levels of NGAL were increased among patients of results. Less than ideal specificities or positive predic
with high RIFLE class AKI (Table 3), and a significant tive values could potentially reflect not the inadequacy of
correlation existed between the two (r = 0.55, P <0.001).31 the new biomarker to detect renal tubular injury or stress,
The same relationship was seen when looking at either but rather the limitation of the serum-creatinine-based
peak or median plasma NGAL levels over the first 4 days reference standard.
spent in the ICU, or NGAL level during the first ICU day. Many studies of new biomarkers have also exam
Similarly, urine NGAL values were higher in the group ined their capability to predict ‘hard’ outcomes, such
of adult patients with AKIN stages 2 and 3 than in those as the need for RRT or death.30 In the absence of broad
with AKIN Stage 1 in the ICU (Table 3).32 consensus with regard to the time of initiation of
Similarly, urinary levels of both NGAL and inter RRT, however, this ‘hard’ end point is subjective, and
leukin (IL)‑18 have been studied in critically ill pediatric one that is strongly influenced not only by physician
patients requiring mechanical ventilation (Table 4). 33,34 practice, but also by serum creatinine levels and urine
A strong association exists between worse AKI as output. Renal biopsies would enable histopathologic
assessed by the pediatric RIFLE class, and increasing confirmation of tubular damage but are clearly imprac
mean and peak levels of NGAL in urine.33 Peak urine tical for large studies. Important steps forward for AKI
IL‑18 concentrat ions also demonstrated a stepwise biomarker research would include the identificat ion
increase with worsening AKI class as assessed by the of a better gold standard than either serum creati
pediatric RIFLE criteria.34 Finally, in a small study of 85 nine level or urine output, as well as evidence that a
pediatric patients in the ICU, researchers tried to define biomarker-directed therapeutic approach can improve
the severity of AKI according to the percentage change clinical outcomes.
It is important to remember that a fundamental Table 4 | Relationship between biomarkers and RIFLE class in children in the ICU
requirement for a good consensus definition is that Study Biomarker AKI severity by level of biomarker
it is easy to apply in a variety of clinical and research
Zappitelli et al. (2007)33 Urine NGAL* Non-AKI:‡ 24.6 ± 45.5 ng/ml
settings.11 However, the novel biomarkers described Risk:‡ 34.5 ± 47.4 ng/ml
above are not readily available in many countries, and Injury:‡ 82.9 ± 122.9 ng/ml
are likely to remain economically impractical for wide Failure:‡ 103.2 ± 107.3 ng/ml
use in many developing countries in the foreseeable Washburn et al. (2008)34 Urine IL‑18* Non-AKI:‡ 102.8 ± 175.1 pg/ml
future. If one of the objectives of having a consensus Risk:‡ 128.9 ± 259.7 pg/ml
definition is to have the medical community speaking Injury:‡ 179.9 ± 224.2 pg/ml
Failure:‡ 366.1 ± 631.0 pg/ml
the same language, these restrictions must be taken into
*Peak levels in the first 4 ICU days. ‡All values are presented as mean ± standard deviation. Abbreviations:
consideration. If the epidemiology of AKI in a devel AKI, acute kidney injury; ICU, intensive care unit; IL, interleukin; NGAL, neutrophil gelatinase-associated
oped country is described using a biomarker-based lipocalin; RIFLE, Risk, Injury, Failure, Loss, End-stage renal disease.