FAQ is an acronym for Frequently Asked Question. This page is intended to provide answers to common questions, along with some useful tips and tricks that are presented as questions. Please visit the RxKinetics web site for the latest version of this document.
Q. Why are there two Vancomycin models in the program?
A. Vancomycin is a difficult drug to dose because of extreme inter-patient variability of pk parameters. No single model will fit all patient populations.
The two models roughly parallel the two divergent dosing methods:
Please bear in mind that vancomycin is not a very predictable drug, and these models will not fit all patient populations. It is important to save your data, and to periodically re-analyze your population parameters vs outcome. This is where the population analysis utility included with APK© will allow you to create models fitted to your patient population.
Q. How do I adjust a dose based only on a trough level?
A. You may analyze a single trough, a single peak, or a random level drawn at any point within the dosing interval by selecting the single point Bayesian method.
The serum level time is relative to the preceding dose. For example:
The timing of the trough level is therefore 10.5 hours after the end of the infusion.
One way to grasp this concept is to understand that the serum level is the result of a dose which has been given, not from a future dose. not from a future dose.
Q. Why does the creatinine clearance from the APK© program differ from what I calculate by hand?
A. The difference is most likely due to the weight that you are using in your hand calculation. Because creatinine is produced by muscle tissue, C&G recommend that you use lean body weight, except in morbidly obese patients. Because creatinine is produced by muscle tissue, not fat, additional weight in form of fat does not significantly alter the production of creatinine.
There is considerable difference of opinion among practitioners concerning the most appropriate creatinine clearance equation to use. APK© includes multiple creatinine clearance methods from which to choose. Use the method which is most appropriate for your patient.
Q. How do I dose IM gentamicin?
A. Enter an infusion length of 90 minutes, which is the average time for IM absorption of aminoglycosides. Be sure to draw your peaks at least 90 plus 15 minutes after the injection.
Q. What is my password for editing the drug models?
A. The program ships without a password.
Q. Must I always use steady-state analysis? For example, my patient received 2 doses of 1gm vanco q12hr, and for some reason the dose was decreased to 600mg around which levels were drawn.
A. You do not have to assume steady-state if you use the 3-point method. Draw a trough before the 600mg dose, a peak after, then a mid-point level after that. The program will use the pre-dose trough and the peak level to calculate the volume of distribution. The two post-dose levels are used to calculate the elimination rate.
Q. Why is the default dose for once-a-day gentamicin 5mg/kg instead of Hartford's 7mg/kg?
A. Please refer to the consensus document which appeared in the June 1997 issue of the International J. of Clinical Pharmacology and Therapeutics:
In many respects the approach of the Hartford nomogram is just as simplistic as giving 80mg Q 8 hours to all patients with a "normal" serum creatinine. Instead, all patients should be individually dosed in order to maximize therapeutic efficacy and minimize the risk of toxicity. The patient should receive a large enough dose to produce a peak level that is 8 to 10 times greater than the MIC. The dosing interval should then be extended to produce a trough less than 0.5 mcg/ml in order to provide a washout of aminoglycoside in the renal tubule thereby reducing toxicity.
Q. During prospective dosing of pediatric patients there seems to be no way to chose an initial target peak/trough value like there is for adults.
A. The initial pediatric calculations are weight based, as they are in every pediatric reference book you will find. Although you cannot target a specific peak and trough, the program will show estimated peak and trough levels from your selected dose.
The reason that initial doses are weight-based is because the creatinine clearance equations for peds are not accurate enough to be used for pk modeling.
Q. Do I enter serum level times relative to the beginning or the end of the infusion?
A. Times are relative the beginning and the end of the infusion.
Q. How does the program handle late doses?
A. Unfortunately, late doses can not be avoided, so how can you use these levels? There are at least two late dose scenarios, each would be handled differently, depending on the methodology you are using.
The first late dose scenario we'll describe occurs if the dose was late but the levels were correctly drawn relative to the actual hang time. For example, the nurse is supposed to be giving gent 80mg Q 8 hours, at 6-14-22, but the sample dose is given late, at 1600. The trough was drawn at 1530 and the peak at 1730. As you can see, the levels are correctly drawn relative to the late dose.
First, let's examine the Sawchuk and Zaske method. This method is the gold standard for evaluation of peak and trough levels in the one-compartment open model. Sawchuk and Zaske drew three levels: a trough before the dose, a peak after the dose, and another trough after the dose. However, many institutions take a shortcut and only draw two levels, a trough before the dose and a peak after. If the patient is at steady-state, then you can assume that the second trough after the dose will be the same as the trough before the dose. Under most circumstances, this is a valid assumption and will save laboratory expenses.
Next, look at Sawchuk and Zaske's equations for calculating pk parameters, you can find them in the program help file (Press the F1 key). Where is the dosing interval in these equations? ...... nowhere! The dosing interval is not used in any calculation.
If you are using the 3-point method, the interval is not used, and it doesn't matter if the dose was given late, as long as the levels are correctly drawn, relative to the time the dose was actually given.
If you are using the two point method, the dosing interval is only used to extrapolate the pre-dose trough to simulate the after-dose trough. Therefore, in the APK program, instead of entering the "ordered" interval, you would enter the time between the last dose and the sample dose. Using our same example, if the nurse is supposed to be giving gent 80mg Q 8 hours, at 6-14-20, but the sample dose is given late, at 1600. In the APK program you would enter an interval of 10 hours. Again, this is because the interval is only used to extrapolate the pre-dose trough.
The second late dose scenario occurs when the dose is give late and the levels are correctly drawn relative to the scheduled hang time, not the actual hang time. For example, if the nurse is supposed to be giving gent 80mg Q 8 hours, at 6-14-20, the sample dose is given late, at 1600, but the trough was drawn at 1330 and the peak at 1530. As you can see the levels are correctly drawn relative to the scheduled dose, not the actual hang time, the "peak" is not a peak at all, but a trough drawn before the dose was given.
This scenario is the result of a lack of communication between lab and nursing. In this case, throw out the peak and just use the trough. With Bayesian you only need a single trough level.
The APK program is deceptively simple in this area. When you hit a snag like this, you have to stop and think, and that's the downside to having such a simple interface for entering levels. The other alternative, and the one taken by most other PK programs, is to have you enter each and every dose and the time each dose was administered. This approach punishes you each time you have to enter a simple set of levels. The APK program only makes you work hard in a problem situation.
Q. How do I enter my own drug model?
A. Please read the step-by-step tutorial on the RxKinetics web site.
Q. Where I can find linear regression equations for beta-lactam antibiotics? I'd like to be able to get a good set of Kel values to plug in for cephalosporins and penicillins.
A. Unfortunately, this information is not found in one single reference. Bennett's Drug Prescribing in Renal Failure and Chernow's Critical Care Pharmacotherapy have tables of pk data for common drugs. The FDA package insert of newer drugs usually has an excellent pharmacokinetics section.
Q. We sometimes see an "outlier" message when analyzing serum levels with Bayesian, what does this mean?
A. Bayesian places significant weight on the population model. Because of this reliance on the population model, it is very important with Bayesian that you pick the right model for your patient. This is a particular problem with vancomycin because of the extreme inter-patient variability in pk parameters.
When the program flags an "outlier", it means that, for whatever reason, the serum levels can't be statistically reconciled with the population model. If, after 100 iterations, the Bayesian object (sum of least squares) is not reduced, it gets kicked out of the loop and the patient is labeled an outlier.
Outliers are especially challenging when we are analyzing a single trough level. If the patient's trough is out of range, then you need to get a truer picture of the patient's pk parameters by measuring a peak and trough series instead of a single trough.
If a significant percentage of your patients are flagged as outliers, then there is a good possibility that the model you are using is not appropriate for your patient population. This is a situation where the population analysis tool would be helpful. Each time you print a consult based on serum level data, the calculated pk parameters are saved. This data is a virtual gold mine of information about your patient population. With this tool you can derive a model better fitted to your patients.
Q. My patient is a double amputee, the program won't let me enter his height (36"), what should I do?
A. The program does not make any specific adjustments for amputees, so you'll need to do some calculations the old-fashioned way. Enter the patient's pre-amputation height. The program will calculate a baseline lean body weight.
Because creatinine is produced by lean tissue, to get a more accurate estimate of CrCl, you'll need to estimate post-amputation muscle mass. Start with the LBW calculated from the patient's pre-amputation height. Then deduct a percentage depending on the extent of the amputation:
In this case, since your patient is a double amputee, you would decrease the LBW by 32%. Let's call this the Amputee-Adjusted LBW (AALBW).
If the patient's actual weight is less than the AALBW, then the patient is underweight and you should use the patient's current weight to calculate CrCl.
If the patient weighs more than the AALBW, then you should hand calculate the CrCl using the AALBW.
Q. My patient is bedridden and I cannot obtain his/her height, what should I do?
A. A patient may also have scoliosis or severe contractions that prevent a true height measurement. There are (at least) three methods that may be used to estimate height when actual measurement is not possible: knee height, forearm length and demi-span. Please search the RxKinetics web site for the topic "Estimating height in bedridden patients".
Q. Does APK interface with our Pharmacy Information System?
A. APK© may be interfaced with a PIS, using our open interface specification. Please contact your PIS vendor to request implementation.
Q. What does the message "Y-intercept less than zero" mean when I try to perform population analysis?
A. "Y-intercept less than zero" is displayed when the regression line intercepts the Y axis below zero. "Regression analysis yielded a negative slope" is displayed when the regression line is inverted.
The regression line is a plot of Y vs X where X is creatinine clearance (the independent variable) and Y is the elimination rate (the dependent variable). The Y intercept is the Nonrenal Kel, the slope of the regression line is the Renal Kel for the following equation:
Kel = NonRenalK + [CrCl x RenalK]
It would be impossible (of course) for there to be a negative slope or a negative NonRenal Kel, that would mean you are somehow putting drug back into the system instead of excreting it.
Both of these "errors" are caused by a widely scattered data set for which a regression line cannot be accurately calculated. This would be expected with a drug like Vancomycin which has a wide variability in pk parameters. Usually it takes a higher N in your data set before the data becomes analyzable. Keep saving your consults and eventually you will acquire a data set that will give usable results.
Desktop problem solutions
Q. We had a power outage while running APK© and now we can't access the database!
A. Because of the sudden loss of power, the database locks were not released. Also, the patient data indexes are usually corrupted when this happens.
To fix the problem delete the patient data indexes and the lock files in your data directory. Use the Windows "Find" utility to find and delete these files:
The indexes will automatically be rebuilt when you restart APK©.
Q. We need to reinstall APK© on our new computer, how do we transfer our current patient from our old computer?
A. Follow these steps:
Q. We get an Error #33 (file lock violation) dialog and when we click OK the program stops running.
A. Error #33 (file lock violation) indicates one or more of the lock files are corrupt or inaccessible. Lock files have the "@S@" extension and are used to keep track of which records are open and which are locked for writing. Lock files are the gate-keepers of the database, allowing multiple users to simultaneously access records without corrupting the underlying data.