Improving Clinical Prediction Rules In Acute Kidney

1 downloads 0 Views 965KB Size Report
Aug 6, 2013 - Department of Clinical & Experimental Medicine, Faculty of Health Sciences, ... Introduction: Early recognition of patients developing acute kidney injury is of ... This pilot study on general medical emergency admissions is the first to .... data (LH, RV, LF) were involved in the management of the patients.
Biomarkers

ISSN: 1354-750X (Print) 1366-5804 (Online) Journal homepage: http://www.tandfonline.com/loi/ibmk20

Improving Clinical Prediction Rules In Acute Kidney Injury With The Use of Biomarkers of Cell Cycle Arrest: A Pilot Study. Luke E Hodgson, Richard M Venn, Steve Short, Paul J Roderick, Duncan Hargreaves, Nicholas Selby & Lui G Forni To cite this article: Luke E Hodgson, Richard M Venn, Steve Short, Paul J Roderick, Duncan Hargreaves, Nicholas Selby & Lui G Forni (2018): Improving Clinical Prediction Rules In Acute Kidney Injury With The Use of Biomarkers of Cell Cycle Arrest: A Pilot Study., Biomarkers, DOI: 10.1080/1354750X.2018.1493617 To link to this article: https://doi.org/10.1080/1354750X.2018.1493617

Accepted author version posted online: 26 Jun 2018.

Submit your article to this journal

Article views: 9

View Crossmark data

Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=ibmk20

Title page Improving Clinical Prediction Rules In Acute Kidney Injury With The Use of Biomarkers of Cell Cycle Arrest: A Pilot Study. 1,2

2

2

1

Luke E Hodgson , Richard M Venn , Steve Short , Paul J Roderick , Duncan 2

3

4,5

Hargreaves , Nicholas Selby , Lui G Forni * Academic Unit of Primary Care and Population Sciences, Faculty of Medicine,

ip

University of Southampton, Southampton General Hospital, Tremona Road,

us

Western Sussex Hospitals NHS Foundation Trust, Anaesthetics Department, Worthing

Hospital, Lyndhurst Rd, Worthing, BN11 2DH.

Centre for Kidney Research and Innovation Division of Medical Sciences and

an

3

cr

Southampton, SO16 6YD 2

t

1

M

Graduate Entry Medicine, University of Nottingham & Department of Renal Medicine,

4

pt ed

Royal Derby Hospital, Uttoxeter Road, Derby, DE22 3NE

Intensive Care Department, The Royal Surrey County Hospital NHS Foundation Trust,

5

ce

Egerton Road, Guildford, GU2 7XX &

Department of Clinical & Experimental Medicine, Faculty of Health Sciences,

Ac

University of Surrey, Guildford, GU2 7XH.

* Corresponding Author email [email protected], Tel: 01483 571122 Fax: 01483 402724

Abstract Introduction: Early recognition of patients developing acute kidney injury is of considerable interest, we report the first use of a combination of a clinical prediction rule with a biomarker in emergent adult medical patients to improve AKI recognition. Methods: Single-centre prospective pilot study of medical admissions without AKI identified as high risk by a clinical prediction rule. Urine samples were obtained and tissue inhibitor of metalloproteinases-2 (TIMP-2) and insulin-like growth factor binding

cr

Outcome: Creatinine based KDIGO hospital-acquired AKI (HA-AKI).

ip

t

protein 7 (IGFBP7) - biomarkers associated with cell cycle arrest, were measured.

Results: Of 69 patients recruited, HA-AKI developed in 13% (n=9), in whom

us

biomarker values were higher (median 0.43 [interquartile range 0.21-1.25] vs.

0.07 [0.03-0.16] in cases without (P=0.008). Peak rise in creatinine was higher in

an

biomarker positive cases (median 30 μmol/l (7-72) vs 1 μmol/l (0-16), P=0.002). AUROC was 0.78 (95% CI 0.57-0.98). At the suggested cut-off (0.3) sensitivity

M

for predicting AKI was 78% (95% CI 40-97%), specificity 89% (78-95%), positive predictive value 50% (31-69%) and negative predictive value 96% (89-

pt ed

99%).

Discussion: Addition of a urinary biomarker allows exclusion of a significant number of patients identified to be at higher risk of AKI by a clinical prediction

ce

rule.

Keywords: Acute Kidney Injury (AKI); tissue inhibitor of metalloproteinases-2

Ac

(TIMP-2); insulin-like growth factor binding protein 7 (IGFBP7);

Clinical significance (100 word limit) Early identification of AKI is of significant global importance to allow earlier institution of new and future therapies and to allow enrichment of future clinical trials. This pilot study on general medical emergency admissions is the first to report the employment of an electronically generated clinical prediction rule coupled with a urinary biomarker to predict adult patients who went on to develop AKI in hospital. Use of this approach could be further investigated to identify those at highest risk of AKI

Ac

ce

pt ed

M

an

us

cr

ip

t

which is associated with a high risk of mortality and considerable morbidity.

Introduction Acute kidney injury (AKI) complicates around 20% of hospital admissions and the development of AKI is associated with considerable morbidity and mortality risk (1, 2). This has led to calls to systematically highlight patients at risk of hospital-acquired AKI (HA-AKI); in the UK the National Institute for health and Care Excellence (NICE) guidance CG169 recommends that patients who are acutely ill (including all over 65s), undergoing general surgical procedures and receiving iodinated contrast be risk assessed for AKI (3-5). However, in practice this would result in excess of 80% of all adult acute medical admissions having risk assessment performed, although the

t

incidence in this group is significantly less than 10%. One approach to improving risk

ip

assessment is to implement clinical prediction rules which integrate multiple variables,

cr

such as medical history and diagnostic tests, to predict risk of an outcome (6-8).

However, a recent systematic review found relatively few available validated models

us

using clinical variables alone for AKI in general hospitalised populations, and all had limitations in predictive performance for use in AKI (9). This is due, in part, to the

an

heterogeneous nature of AKI reflecting the differing pathophysiological processes that may lead to acute renal injury. For example, some clinical prediction rules may perform

M

well under specific circumstances, such as after cardiac surgery, but less well when applied to a non-specific population such as acute medical emergency admissions. In

pt ed

tandem there has been significant interest in potential biomarkers to identify individuals with AKI in a timely fashion in order to intervene earlier and potentially improve clinical outcomes. This is driven in part by the well documented limitations of serum creatinine (SCr) and urine output for the detection of AKI with potential significant

ce

injury occurring before sufficient loss of excretory kidney function can be detected with

Ac

a change in SCr (10).

Therefore more sensitive and specific markers of renal injury and dysfunction could offer several advantages to the clinician. They could help identify AKI prior to a rise in SCr enabling early intervention. They could improve stratification of the severity of AKI and also potentially provide prognostic information (11, 12). Multiple candidate biomarkers have been described in the literature, with two biomarkers associated with cell cycle arrest, tissue inhibitor metalloproteinase-2 [TIMP-2] and insulin-like growth factor binding protein-7 [IGFBF-7], showing particular promise in a number of areas including intensive care and cardiac surgery, though not to date in general medical

emergency admissions (13, 14). Of relevance, the initial studies examined the ability of these biomarkers to predict worsening AKI in critically ill patients with stage 1 AKI and not in an unselected medical cohort. Clearly, the application of AKI biomarkers to an identified high risk cohort, if successful, would allow improved triage and hopefully translation into improved outcomes. We have previously described a validated AKI prediction model – the APS (AKI Prediction Score) (15, 16). This model uses a combination of physiological and patient demographic data to determine a score, derived from regression coefficients for each of the variables, which corresponds to a risk of developing HA-AKI over the next 7 days

ip

t

(Table 1). This model shows satisfactory calibration (Hosmer-Lemeshow p = 0.955)

and reasonable discrimination with an area under the receiver operating characteristic

cr

curve of 0.72 (95% CI 0.66–0.77) and was used to identify patients at an increased risk

us

of AKI. The main objective of this study was to assess the potential added value of TIMP2:IGFBP7 to risk assessment as per the APS tool. Although several studies have

an

investigated AKI biomarkers in an ICU environment this study, to the best of our knowledge, is the first report of the use of a AKI biomarker in conjunction with a

Ac

ce

pt ed

M

clinical prediction rule in emergency general medical patients.

Materials & Methods: A prospective pilot observational cohort study was performed on adult general medical patients at a UK single-centre acute hospital over a 12-month period (January 2016 January 2017). The primary outcome was hospital-acquired AKI within seven days of admission. KDIGO (Kidney disease: improving global outcomes) criteria for AKI were employed (SCr increase of ≥1.5 from the admission value or ≥26.5 μmol/L within a rolling 48 hours during the first 7 days of the patients first admission in the study period) (10). The hospital has an emergency department attendance of over 75000 per annum with approximately 30 acute medical admissions per 24-hour period. At

ip

t

admission, all inpatients routinely have physiological observations measured and

entered via handheld systems into a clinical data software system (Patientrack, Sydney,

cr

NSW, Australia). Inclusion criteria were an APS score ≥5 points at admission to

us

hospital, absence of community-acquired AKI as per KDIGO and age ≥18 years. The APS score of 5 was chosen as this corresponds to a risk of between 15-20% of

an

developing HA-AKI. Patients had to stay at least one night in the medical division over the study period and have at least two SCr assays performed. Exclusion criteria included patients with community AKI (defined using KDIGO change from baseline

M

SCr or absolute SCr value ≥354 μmol/dl), patients moved directly from the emergency department to the intensive care unit (as neither area uses the Patientrack data system),

pt ed

non-medical admissions (eg general surgery, obstetrics and gynaecology admissions) and those discharged without spending a night in hospital. Patients were detected through the Patientrack system which calculated the APS through electronic retrieval of

ce

previous International Statistical Classification of Diseases 10th revision (ICD-10) electronically coded history (heart failure, liver disease and diabetes mellitus). CKD was defined as an estimated glomerular filtration rate (eGFR) 0.3 indicating a positive test result. A total of 100 µL of urine is required for measurement, and the test takes about 21 minutes. The study was given ethical approval by NHS Research Ethics Committee London City Road and Hampstead. Patients were followed-up until discharge from hospital, or death during the inpatient spell. 1-year mortality was established through linked NHS data. A number of steps

ip

t

were employed to minimize bias. None of the researchers involved in analysis of the

data (LH, RV, LF) were involved in the management of the patients. The research team

cr

members responsible for data analysis had access only to the fully anonymised

us

individual-level data. A separate team member who was not involved in data analysis (SS) extracted SCr to make the diagnosis of AKI whilst another team member

an

performed biomarker measurement (DH). Finally an expert independent from the study group performed blinded adjudication of AKI diagnosis based on changes in SCr (NS). SPSS version 24 (IBM Corporation, Somers, NY 10589, USA) was used for data

M

analysis. Data are presented as percentages or median with interquartile ranges (IQR). Comparisons were performed by X2-tests for categorical variables and Mann–Whitney

pt ed

tests for continuous data. At the suggested biomarker cut-off sensitivity specificity, positive predictive value and negative predictive values were calculated for prediction of AKI development. Discrimination of the biomarker to predict the outcome was

Ac

ce

assessed by the area under the receiver operating characteristic curve (AUC).

Results Over the study period 103 patients with an APS≥5 were screened. As all patients had an APS calculated automatically by the Patientrack system before any outcome had occurred, there were no missing demographic or physiological data at admission. 69 patients were successfully recruited to the study with at least two SCr measurements and the urine sample collected. Reasons for non-recruitment included acute confusion (n=19), palliative care (n=9), plan for home with no repeat SCr (n=3), too unwell to consent (n=2) and unable to gain a urine sample (n=1). Of those recruited 13% (9/69) developed HA-AKI (Figure 1). Median age was 82 (interquartile range 75-86), 67% of

ip

t

the cohort were male and in-patient mortality was 10%.

cr

There were no significant differences between AKI and non-AKI cases regarding age, sex, baseline and admission SCr (table 2). Mortality was higher in those who developed

us

AKI as an in-patient (22% vs 8%) and on 1-year follow-up (56% vs 35%) but in this small sample this was not statistically significant (P=0.224 and P=0.282, respectively).

an

Seven of the nine patients who developed AKI were Nephrocheck positive (value >0.3). Biomarker values were significantly higher in those who developed HA-AKI (median

M

0.43 [interquartile range 0.21-1.25] vs 0.07 [0.03-0.16] in cases who did not, P=0.008) (Figure 2). At the suggested biomarker cut-off of >0.3, sensitivity for prediction of

pt ed

development of HA-AKI was 78% (95% CI 40-97%), specificity 89% (77-95%), positive predictive value 50% (31-69%) and negative predictive value 96% (89-99%). To discriminate patients who went on to develop AKI from those who did not, the biomarker test had an AUC of 0.78 (95% CI 0.57-0.98%) (figure 3). In biomarker

ce

positive cases median peak rise in SCr was 30 μmol/l (7-72) vs 1 μmol/l (0-16) in those with a negative biomarker result (P=0.002). 53 cases were identified by the biomarker

Ac

to be at low risk of AKI whilst there were 7 ‘false positive’ results (biomarker >0.3 but no subsequent rise in SCr). Of the patients who developed AKI the aetiology of the AKI was sepsis (Table 3). Two of the patients with HA-AKI were biomarker negative (false negatives). One developed HA-AKI on day 7 after surgical intervention for a septic arthritis (Patient 8) the other developed AKI stage 1 complicating the management of congestive cardiac failure with

and the rise in serum creatinine may have reflected a change in volume status rather than renal injury particularly in the setting of CKD. Indeed, in patients with a baseline serum creatinine greater than 133µmol/dl a false positive diagnosis of AKI is made in roughly 30% of the patients where AKI is defined by an increase in serum creatinine > 26.5µmol/dl over 48 hours(21). Table 4 shows the patient characteristics in those who were biomarker positive but AKI negative (false positives). Of note 3 cases probably had community acquired AKI and one succumbed before a further serum creatinine test

ip

t

was obtained although the previous test had showed a rise in serum creatinine without

cr

reaching the definitive criteria for AKI. Therefore, the presence of a postive biomarker

us

in this cohort may have been associated with AKI in approaching 60%.

an

In part, our results reflect some of the practical problems encountered when translating new technologies to the bedside but also may reflect the observation that biomarker

M

positivity was associated with early identification of subclinical AKI. The

pt ed

identification of biomarker positivity in high risk patients would provide additional

Ac

ce

information which still may trigger intervention.

Discussion To our knowledge, this is the first report of an AKI biomarker study in general medical patients admitted non-electively to hospital in combination with a clinical risk prediction score. In a cohort identified as high risk for developing AKI by the electronically generated clinical prediction rule, addition of the biomarker allowed exclusion of patients who did not develop AKI given the high negative prediction value. This is of particular relevance when one considers current guidance regarding risk

ip

t

assessment in those admitted acutely to hospital. If the patient types described in the

cr

NICE guidelines are all assessed for AKI this would account for almost all acute

us

medical admissions resulting in a considerable clinical workload. A clinical risk score in isolation may help give some indication of those at highest risk, but currently available

an

models all have an appreciable false positive rate of approaching 80% that limits utility in a busy acute medical unit. The addition of the biomarker with a high negative

M

predictor value for the development of AKI would allow identification of the lower risk

pt ed

group thereby preventing unnecessary and costly interventions in this cohort. The use of a biomarker with a higher positive predictor ability would enable enrichment of the higher risk population identifying those who will require further input in a more

ce

targeted fashion towards the possible prevention or early treatment of AKI. Our study has focussed on the use of a clinical prediction rule to highlight those at highest risk of

Ac

AKI followed by biomarker estimation. Previously, Kimmel et al demonstrated, in a heterogenous cohort of emergency department patients that the addition of [TIMP2].[IGFBP-7] is complementary to serum creatinine as well as a clinical variable model concluding that the use of the biomarkers added valuable information for risk assessment of developing AKI.(22) This is in keeping with our study findings.

As highlighted, AKI prediction is a challenge as it is not a single disease but a complex syndrome with multiple underlying aetiologies (10). Our results suggest a promising two step approach could be employed where an electronically generated clinical prediction rule is followed in those at higher risk by the addition of a biomarker to further stratify risk. This approach has been described previously, although in very a different setting in paediatric ICU patients (22, 23).

ip

t

Of interest within these data is the meaning of the false positive and false negative

biomarker results. Firstly, Nephrocheck has not been validated under such non-specific

cr

conditions and yet still performs well in terms of discrimination. Moreover our

us

diagnosis of AKI was only determined by serum creatinine estimation and not urine output. Recent studies in surgical patients have shown that biomarker positivity is often

an

seen in patients diagnosed by changes in urine output alone, rather than changes in serum creatinine, which may explain our findings. Secondly, biomarker positivity in the absence of a serum creatinine rise may reflect renal injury which is not manifest through

M

a rise in serum creatinine termed sub-clinical AKI which has been shown to also be associated with poor outcomes. Thirdly, we have little knowledge as to the kinetics of

pt ed

these biomarkers outside specific environments such as the post-surgical patient where the AKI insult is well recognised in the patient pathway. In our cohort, it is possible that the biomarker could have been increasing or decreasing at time of sampling. This may well be the case in the false positives where several of the patients presented with

ce

community acquired AKI and as such were not identified by a rise in serum creatinine

Ac

although the TIMP-2/IGFBP-7 was positive. Limitations

Caution must be employed drawing any robust conclusions, including generalisability, from a small single-centre cohort however, our data does provide a proof of concept of design and the results suggest a high negative predictive value in a population at higher risk of HA-AKI that should be further explored in larger multi-centre studies. Our results also support the feasibility of measuring a urinary biomarker in the environment of a busy admissions ward, where historically obtaining urine samples can be

challenging. This highlights a further limitation of our study in that we have defined AKI only by changes in serum creatinine which may have missed patients with AKI defined by urine output criteria alone. Moreover, a significant number of screened patients were not recruited due to acute confusion precluding written consent – such patients represent a high risk group, as evidence by the high number of patient in this non-recruited cohort who did develop AKI and future researchers could look at strategies to overcome barriers to recruitment in this group. If the addition of the Nephrocheck to a clinical prediction rule is shown to enhance risk prediction the next steps will include its testing in combination with a risk prevention strategy, to determine

t

whether this leads to an improvement in outcome. Unfortunately where such measures,

ip

such as AKI care bundles, have been introduced the compliance rate is poor. This may

cr

reflect “alert fatigue” and thus a more focused approach may enhance compliance and

Ac

ce

pt ed

M

an

us

translate into improved outcomes.

Conclusions In conclusion we have shown that the combination of an electronic clinical prediction rule coupled with biomarker analysis is feasible and may provide a more specific way of identifying patients at the highest risk of AKI within a general medical cohort. The high negative predictor value observed in this cohort is of practical importance, as this would prevent further unnecessary risk assessment in most cases allowing for a more focused approach. Given clinical prediction rules are increasingly employed at the bedside then this design should be confirmed in larger studies which may inform targeted interventions which hopefully will impact on the incidence of hospital acquired

ip

t

AKI.

cr

Declaration of Interests

LGF has received honoraria and research support from Astute medical and Orthoclinical

us

diagnostics.

an

Acknowledgements

We are grateful to the research and outreach nurses at Western Sussex Hospitals NHS

M

Trust for their help in identifying the patients and collecting urinary samples in order to

Funding details

pt ed

perform the study. We are also grateful to Dr R Rivero for his encouragement and help.

A grant from the Small Business Research Initiative (National Institute for Health Research Devices for Dignity HTC partnered with the Department of Health) helped

ce

fund integration and implementation of the technology for the clinical prediction rule

Ac

from the Hospital Trust’s electronic servers.

Table 1 Acute Kidney Injury Prediction Score (APS). AVPU – level of responsiveness: A – alert, V – responds to voice, P – pain, U – unresponsive. CKD – chronic kidney disease. Score 0

1

Age

< 60

Respiratory Rate

80

≥20

cr

ip

t

Other

Liver Disease

No

Yes

an

No

Yes

M

Diabetes Mellitus

us

failure

Age, median (IQR) Sex

HA-AKI (n=9)

No AKI (n=60)

P value, OR (95% CI)

85 (80-88)

81 (72-85)

3M, 6F

43M, 17F

137 (95-167) 118 (90-186)

109 (81-139) 116 (83-146)

0.121 0.051, 0.20 (0.050.88) 0.284 0.708

0.43 (0.21-1.25)

0.07 (0.03-0.16)

0.008

78%

10%

0.3

pt ed

Table 2. Demographics and results of those with and without hospital-acquired AKI.

In-patient mortality 1-yr mortality

2

HA-AKI – hospital-acquired AKI, IQR – Interquartile range, Mann Whitney U tests or χ NC – nephrocheck biomarker results, OR – odds ratio, SCr – serum creatinine

Table 3: Details of Patients who developed hospital acquired AKI AKI Stage

CKD

BM

1

Chest sepsis

85

F

1

Y

+

2

Bowel perforation

77

F

2

N

+

3

Urosepsis, NSTEMI, UGI bleed

86

F

1

Y

+

4

AECOPD

82

M

1

N

+

5

CAP

90

M

1

Y

6

Neutropenic sepsis

75

F

1

7

Sepsis unknown source

90

M

8

Septic Arthritis

84

M

9

CCF, IV diuretics

+

1

Y

+

2

N

-

M

1

Y

-

cr

+

an

N

M

86

t

Age Sex

ip

AKI Cause(s)

us

Patient

pt ed

NSTEMI – Non ST Elevation Myocardial Infarction, AECOPD – Acute Exacerbation of Chronic Obstructive Airways Disease, CAP – Community acquired pneumonia, CCF

Ac

ce

– Congestive cardiac failure. BM – Biomarker status.

Table 4 : Details of Patients who were biomarker positive and did not develop hospital acquired AKI. AKI Diagnosis

Age Sex

Notes CKD

Stage CCF, LRTI

88

M

0

Y

Possible community injury

LRTI

77

M

0

Y

Possible community injury

80

M

0

N

Metastatic Prostate

Only single repeat Creatinine on

Cancer

t

day of discharge

M

0

N

Clinicians diagnosed AKI on

cr

85

ip

Possible community injury;

CAP 82

F

0

N

SBO, Cellulitis

66

M

0

N

91

M

0

M

Died during admission, bowel

Y

perforation with rising creatinine

pt ed

Bowel perforation

Rhabdomyolysis

an

Rhabdomyolysis, CAP

us

discharge

CCF – Congestive cardiac failure, LRTI – lower respiratory tract infection, CAP –

ce

Community acquired pneumonia, SBO- small bowel obstruction.

Ac

Legends to Figures Figure 1 Study flow chart Figure 2. Area under the receiver operating characteristic curve for Nephrocheck urine biomarker to predict hospital-acquired AKI 0.78 (95% CI 0.57-0.98). Figure 3 Median interquartile box plots with range for Nephrocheck urine biomarker at admission for those who went on to develop AKI (n=9) and those who did not (n=60). Biomarker result was significantly higher in those who developed AKI (using the Mann Whitney U test, P=0.008). Dashed line represents 0.3 recommended cut-off by manufacturer for high risk of AKI.

References

Ac

ce

pt ed

M

an

us

cr

ip

t

1. Uchino S, Bellomo R, Bagshaw SM, Goldsmith D. Transient azotaemia is associated with a high risk of death in hospitalized patients. Nephrology, dialysis, transplantation : official publication of the European Dialysis and Transplant Association - European Renal Association. 2010;25(6):1833-9. 2. Coca SG, Peixoto AJ, Garg AX, Krumholz HM, Parikh CR. The prognostic importance of a small acute decrement in kidney function in hospitalized patients: a systematic review and meta-analysis. American journal of kidney diseases : the official journal of the National Kidney Foundation. 2007;50(5):712-20. 3. Mehta RL, Cerda J, Burdmann EA, Tonelli M, Garcia-Garcia G, Jha V, et al. International Society of Nephrology's 0by25 initiative for acute kidney injury (zero preventable deaths by 2025): a human rights case for nephrology. Lancet. 2015;385(9987):2616-43. 4. National Confidential Enquiry into Patient Outcome and Death. Adding insult to injury: a review of the care of patients who died in hospital with a primary diagnosis of acute kidney injury (acute renal failure) http://www.ncepod.org.uk/2009report1/Downloads/AKI_report.pdf. NCEPOD; 2009. 5. National Institute for Health and Care Excellence. Acute Kidney Injury: Prevention, Detection and Management Up to the Point of Renal Replacement Therapy. (Clinical guideline CG169). London: National Clinical Guideline Centre.; 2013 Aug. 6. James MT, Hobson CE, Darmon M, Mohan S, Hudson D, Goldstein SL, et al. Applications for detection of acute kidney injury using electronic medical records and clinical information systems: workgroup statements from the 15(th) ADQI Consensus Conference. Can J Kidney Health Dis. 2016;3:9. 7. Hoste EA, Kashani K, Gibney N, Wilson FP, Ronco C, Goldstein SL, et al. Impact of electronic-alerting of acute kidney injury: workgroup statements from the 15(th) ADQI Consensus Conference. Can J Kidney Health Dis. 2016;3:10. 8. Royston P, Moons KG, Altman DG, Vergouwe Y. Prognosis and prognostic research: Developing a prognostic model. Bmj. 2009;338:b604. 9. Hodgson LE, Sarnowski A, Roderick PJ, Dimitrov BD, Venn RM, Forni LG. Systematic review of prognostic prediction models for acute kidney injury (AKI) in general hospital populations. BMJ Open. 2017;7(9):e016591. 10. KDIGO Clinical Practice Guideline for Acute Kidney Injury. Kidney Int 2012;Suppl 2(1):1-136. 11. Ostermann M, Philips BJ, Forni LG. Clinical review: Biomarkers of acute kidney injury: where are we now? Crit Care. 2012;16(5):233. 12. Murray PT, Mehta RL, Shaw A, Ronco C, Endre Z, Kellum JA, et al. Potential use of biomarkers in acute kidney injury: report and summary of recommendations from the 10th Acute Dialysis Quality Initiative consensus conference. Kidney Int. 2014;85(3):513-21. 13. Chawla LS, Kellum JA. Acute kidney injury in 2011: Biomarkers are transforming our understanding of AKI. Nat Rev Nephrol. 2012;8(2):68-70.

Ac

ce

pt ed

M

an

us

cr

ip

t

14. Kashani K, Al-Khafaji A, Ardiles T, Artigas A, Bagshaw SM, Bell M, et al. Discovery and validation of cell cycle arrest biomarkers in human acute kidney injury. Crit Care. 2013;17(1):R25. 15. Forni LG, Dawes T, Sinclair H, Cheek E, Bewick V, Dennis M, et al. Identifying the patient at risk of acute kidney injury: a predictive scoring system for the development of acute kidney injury in acute medical patients. Nephron Clin Pract. 2013;123(3-4):143-50. 16. Hodgson LE, Dimitrov BD, Roderick PJ, Venn R, Forni LG. Predicting AKI in emergency admissions: an external validation study of the acute kidney injury prediction score (APS). BMJ Open. 2017;7(3):e013511. 17. Sawhney S, Fluck N, Marks A, Prescott G, Simpson W, Tomlinson L, et al. Acute kidney injury-how does automated detection perform? Nephrology, dialysis, transplantation : official publication of the European Dialysis and Transplant Association - European Renal Association. 2015;30(11):1853-61. 18. Mori K, Lee HT, Rapoport D, Drexler IR, Foster K, Yang J, et al. Endocytic delivery of lipocalin-siderophore-iron complex rescues the kidney from ischemiareperfusion injury. J Clin Invest. 2005;115(3):610-21. 19. Mishra J, Dent C, Tarabishi R, Mitsnefes MM, Ma Q, Kelly C, et al. Neutrophil gelatinase-associated lipocalin (NGAL) as a biomarker for acute renal injury after cardiac surgery. Lancet. 2005;365(9466):1231-8. 20. Mayer T, Bolliger D, Scholz M, Reuthebuch O, Gregor M, Meier P, et al. Urine Biomarkers of Tubular Renal Cell Damage for the Prediction of Acute Kidney Injury After Cardiac Surgery-A Pilot Study. J Cardiothorac Vasc Anesth. 2017;31(6):20729. 21. Lin J, Fernandez H, Shashaty MGS, Negoianu D, Testani JM, Berns JS, et al. False-Positive Rate of AKI Using Consensus Creatinine–Based Criteria. Clinical journal of the American Society of Nephrology : CJASN. 2015;10(10):1723-31. 22. Kimmel M, Shi J, Latus J, Wasser C, Kitterer D, Braun N, et al. Association of Renal Stress/Damage and Filtration Biomarkers with Subsequent AKI during Hospitalization among Patients Presenting to the Emergency Department. Clinical journal of the American Society of Nephrology : CJASN. 2016;11(6):938-46. 23. Menon S, Goldstein SL, Mottes T, Fei L, Kaddourah A, Terrell T, et al. Urinary biomarker incorporation into the renal angina index early in intensive care unit admission optimizes acute kidney injury prediction in critically ill children: a prospective cohort study. Nephrol Dial Transplant. 2016;31(4):586-94.

pt ed

ce

Ac

cr

us

an

M t

ip

pt ed

ce

Ac

cr

us

an

M t

ip

pt ed

ce

Ac

cr

us

an

M t

ip