you beter make your own operational definition which u can adopt and apply throghout ur study for all data sets without the need for any future change.

But make it feasible and practicable and reliable by incorporating all the feasible and analyzable components in ur datasets. Ensure that the components of ur definition are relevant to ur reserch question.

Good definition always depends upon ur study objectives and design. U have to frame it by yourself basing on ur requiremnts of achieving and supporting ur research hypothesis.

The diagnosis of AKI can be done by using the following parameters.

1) Serum creatinine > 1.3 mg% in females & > 1.5 mg% in males ( if baseline creatinine is not available ) * / or rise in serum creatinine of 0.3 mg% above baseline of patients creatinine , if available .

2) A recent precipitating event in the preceding 1 week ( eg. hypovolemia , sepsis , nephrotoxic drugs etc. ).

3) USG Kidneys : Normal sized kidneys with good cortico-medullary differentiation .

4) Reversible renal function : Decline in rising creatinine after management of AKI , if data is available .

* The normal serum creatinine in females is 0.8 - 1.0 mg% ( 0.8 mg% in elderly females & 1.0 mg% in young females ) & 1.0 - 1.3mg% in males ( 1.0 mg% in elderly males & 1.3 mg% in young males ). This is for non-black individuals.

The major problem with observing raised creatinine for the first time in a patient is to determine whether it is AKI or CKD . The parameters 2-4 might help in diagnosing AKI . AKI is a an acute , life threatening but usually reversible disease , While CKD is a gradual & irreversible renal disease . What I have suggested is from my personal experience in treating community acquired AKI in whom patients have been referred with raised creatinine for management , though recently many other criterias have been evolved .

If you want to publish, you must use a standard definition. The latest is the KDIGO definition - an increase of 0.3 mg/dl above baseline within 48hours or an increase of 50% within 7 days. See: KDIGO. Clinical Practice Guideline for Acute Kidney Injury Section 2: AKI Definition. Kidney Int Suppl 2012;2(1):19–36.

The biggest issue is deciding what is "baseline" creatinine. Often this is missing. The first consensus AKI definition (RIFLE) suggested "back-calculating" using the MDRD formula. We, and others, have shown this is erroneous. See: Pickering JW, Endre ZH. Back-calculating baseline creatinine with MDRD misclassifies acute kidney injury in the intensive care unit. Clin J Am Soc Nephro 2010;5(7):1165–73.

Your primary care data could be very valuable in this regard. The trick as Prof Shivakumar suggests, is to distinguish AKI from CKD. CKD is normally diagnosed over 90 days or more (ie the rise in creatinine takes >90d). Determination of a baseline creatinine for AKI diagnosis, therefore, needs to be carful to use the most recent, apparently stable, creatinine.

Finally, my experience is that at least with patients who are admitted to the ICU nearly all the AKI is community acquired in that by the time they are in the ICU the creatinine is already elevated (or it soon will be in the case of Cardiac Arrests, for example). One of the traps I've seen is for people to forget that creatinine changes are "after the fact" of a reduction of GFR. eg a Cardiac Arrest in the community is also AKI in the community even though creatinine may not be raised until day 2 in the ICU.

Happy to help more if you want - drop me a message.

Thanks to all of you for your answers and thank you John, for offering more help with this, which I would like to take you up on:

1) The background is that we want to use AKI as an outcome in a case control study nested within a cohort of users of antihypertensives to quantify the risk of a certain drug exposure in the community. We are therefore only interested in community acquired AKI. In order to address your last point, John, do you think it's reasonably to consider creatinine levels on the day or the following of admission to hospital?

2) The issue of baseline creatinine has certainly caused me some grief, becaue quite frequently, creatinine values are not available within three months of a potential AKI event (how can we identify this? - see point 3 below) I have seen an approach in the literature (Siew et al Kidney Int 2010;77(6):536-542), where they've used the latest creatinine within the year prior to hospital admission as a reference against which the performance of 3 commonly used ways of estimating the baseline were assessed. They are arguing that this approach closely matches the diagnostic workup in practice. I am thinking about using this approach and to restrict cases to AKI stage 2 or 3 (i.e. at least 2 fold increase from baseline, assuming that a two-fold increase over a period of 1 year or less is likely to be AKI rather than CKD). Do you think this is reasonable?

Related to this issue: The KDIGO guidelines recommend that you could also use the nadir creatinine after the potential AKI event to confirm AKI. However, since the true baseline prior to the event is not known, patients who do not recover may be misclassified as non-cases. This misclassification may be differential if the rate of recovery differs between exposed and unexposed. Could you comment on the proportion of non-recovery from community acquired AKI and whether the recovery differs between drug-induced versus non-drug induced AKI?

3) From the above, it's clear that in order to identify AKI, you need to first identified a creatinine that looks sufficiently suspicious of AKI (an 'index creatinine'), which can then be varified using a bseline of some sort. People have used various thresholds for such an index creatinine (e.g. in the Grampian cohort that is described in the KDIGO guidelines they use a threshold of 130 for women and 150 for men). I do not know how they would have derived this threshold. I am thinking about using a threshold for index creatinine that suggests eGFR of

Sounds great Tobias. I'll be in contact via email

For observational studies we tend to take a hierarchical approach.

1. Within 1 year of an event. If unavailable:

2. The creatinine post event nearest to, but not greater than, 90 day's post event. (if >90 days they may have CKD on AKI). Also this should be post discharge. If unavailable:

3. Lowest of first hospital or final hospital.

Nadir in hospital is a problem I think because of fluid dilution of creatinine concentration and, so I recently discovered (published in Crit Care this year) the possible reduction in creatinine production in some cases. There will be many false positive AKI's if nadir is used.

You may want to exclude patients that don't have 1 or 2 as 3 may include patients with hospital acquired AKI.

If you have 1 and 2 then if they differ a lot there will be a need for adjudication as to which should be the baseline.

Restricting to stage 2 is a possibility for the reasons you suggest. However, I am a little cautious at "throwing away data" (ie stage I), and whether there is bias introduced by doing so. I'll need to think about this, as I will for the other good questions.

Dr Tobias Dreischulte has stated that this study was done to evaluate AKI in a cohort group of hypertensive patients to quantify the risk of a certain drug exposure in the community . I shall briefly share my experience on AKI in elderly hypertensive patients . Elderly patients admitted for fractures are assessed by me for fitness for surgery .If a female patient has a serum creatinine of 1.3 mgs% or more & a male patient with serum creatinine of 1.5 mgs% , the surgery is postponed by 48-72 hrs . Most of these patients are hypertensives on ACEI / ARB , who are also taking NSAIDS for osteoarthritis & also pain of fractures . I have to immediately assess whether the rise in creatinine is due AKI or CKD . Since these patients are hypertensives , they would have some record of baseline creatinine done as a routine about 1-2 years ago . The ACEI / ARB are withdrawn & replaced with Calcium Channel Blockers . The NSAIDS are also withdrawn & replaced with Paracetamol . They are adequately hydrated with IV fluids ( 0.9% Saline ) . The serum creatinine is checked daily & a drop of serum creatinine is misleading as John Pickering highlighted , but reassuring ! If there are no complications of AKI , they would undergo surgery & at discharge , the creatinine done would be considered as baseline . The patients are very anxious & it is my duty to explain to them whether it is AKI or CKD . The delay in surgery improves renal function & prevents AKI from progressing stage 1 & 2 to stage 3 .

As you are in the UK you could do worse than adopting the nationally mandated AKI alert algorithm ( https://www.england.nhs.uk/patientsafety/akiprogramme/aki-algorithm/ ).  This should already be being reported by your laboratories on secondary care creatinine requests. Not yet mandated for reporting to primary care but that is the next step of the roll out.

I do not know of the progress in Tayside but we started reporting the scores to primary care from our laboratories at the beginning of this month.  Score is calculated on all creatinine requests using the laboratory's database. This means the baseline includes inpatient & outpatient creatinine results as well as primary care values.  If there is a level calculated then it comes down the result messaging links to primary care systems as a number 1, 2 or 3 using a dedicated read code so it will drop into the system automatically.

Your lab may have been calculating values before they started reporting live or you could possibly implement the algorithm and calculate scores for historical data in your database.