Oops - missed dose

[pem]

Oh, and PEM, this model (used in a spreadsheet) will give you an approximation of what will be happening to the warfarin in your system.

Make C2 your warfarin dose, fill this down for the series of C

In Colum D
start with this series and continue
=C2
=D2+0.75*C2
=C4+(0.75*C3)+(0.5*C2)
=C5+(0.75*C4)+(0.5*C3)+(0.3*C2)
=C6+(0.75*C5)+(0.5*C4)+(0.3*C3)+(0.25*C2)
after here you can fill down

NOTE: this model does not take into account actual metabolism NOR the actual bioavailability of the dose (which varies from person to person depending on their metabolism).

However it does provide a simple working model for you to undertake a little what if analysis on "what if I miss a dose". Especially when you factor in the rate of drop of INR observed in my above chart to estimate the biological response to a missing dose you can get a feel for the reaction of the system (your bodys system) to the changes.

This is of course a SIMPLE model, however my experience in modelling suggests that more complex models while potentially producing more accurate results can vary more wildly when the input parameters are in error. Essentially the less parameters the lower the accuracy but the lower the tendency for wildy inaccurate results too.

That's great - thanks! I'll try it out and line it up with my actual INR response to the recent coumadin dosages. I agree with your opinion about more parameters potentially increasing the model's sensitivity to variance - I haven't examined that closely, but I get the same intuition.

Best,
pem

 

[pem]

Hi

having a bit of time on my hands after dinner this evening I thought (as I had "some" data) I'd throw together the data I had from my hospital stay into a spreadsheet alongside the simple mathematical model I have discussed. I applied a constant scaling factor (7.5) to bring the calculations to be close to that of the INR that I had to start with. I think the results are interesting

of course the mathematical model doesn't provide any buffering (as the body would), and so drops instantly that inputs are dropped, as well as raises tightly with no dampening (buffering) as soon as the inputs are raised. I find that the inter-reactions of the set are interesting.

I wish I had more INR data, but as my last reading was in the morning of going into hospital (and I didn't take my coaguchek) I just don't have it. I'd like to know how fast it drops! Also I wish I had opportunity to have INR measured on the day of surgery and in ICU.

Perhaps next time ;-)

PS: I don't expect that my INR went below 1.5 ... but then that'd be guessing

Interesting. I wonder why it over predicts for you. I think that "buffering" wouldn't be an issue on that time scale. I suppose at most it would be off by one day, so you could just shift the purple bars one spot to the left and see what the new alignment looks like.

If you search my old posts, you'll find detailed information on the model I developed. I think it is possible that there are individual differences in how home monitors respond to people and that it would be helpful to build a personal calibration mechanism into these monitors. I have ideas about how this could be done, but probably no interested parties.

Best,
pem

 

[pem]

Interesting findings, Pellicle.
One of the problems that I have with comparing meter results to lab results is that lab results aren't always very good. The quality of the lab results is also a product of how the sample is handled after the blood is drawn, and how accurate the value of the reagents is. I have my blood drawn at a hospital, and I trust the hospital lab. A few months ago - on the same day as the hospital blood draw - a clinic took my blood. The Hospital lab result put me right in range. The clinic's result was slightly below range. (Sorry, I'm being lazy and didn't look at my spreadsheet). The difference between the two labs was almost 1.0. The doctor at the clinic - apparently knowing nothing about anticoagulation management - prescribed a one time doubling up on my dose. This was crummy advice - but certainly fit the 'increase the weekly dose by 5%' model -- it would have made my INR spike a few days later, but drop me right back where I was in a week.

The point here -- you probably won't get the same result from different labs. I trust a hospital lab. I have less trust of a clinic or doctor office, where the blood may be mishandled, or improperly stored, and the actual lab may not be as careful about doing the test as a hospital lab would be.

It is interesting, protimenow, to consider pellicle's earlier chart comparing home monitor and lab results. That chart indicated very good consistency between the two.

Best,
pem

 

[pem]

The point here -- you probably won't get the same result from different labs. I trust a hospital lab. I have less trust of a clinic or doctor office, where the blood may be mishandled, or improperly stored, and the actual lab may not be as careful about doing the test as a hospital lab would be.

Also, I have obtained almost identical results across 10 readings from two different labs using reagents with wildly different ISI values. That doesn't necessarily negate your point, but perhaps there is an occasional lab or phlebotomist that is best avoided and hopefully the rest are pretty consistent. Agreed that hospital labs are probably trustworthy, but that's an empirical question too in the end, isn't it?

Best,
pem

 

[pellicle]

Hi

Interesting. I wonder why it over predicts for you. I think that "buffering" wouldn't be an issue on that time scale.

hard to know. Too many factors for my liking (to stick my neck out and suggest a reason). But then I only developed it to see if it would at all come close to reality. (developed some models in my masters thesis on water systems and sort of like doing that)

If you search my old posts, you'll find detailed information on the model I developed.

will do ...

I think it is possible that there are individual differences in how home monitors respond to people and that it would be helpful to build a personal calibration mechanism into these monitors. I have ideas about how this could be done, but probably no interested parties.

sounds interesting. I would wonder at the benefits, thinking pareto principle here and that 80% of the benefits come from this 20% solution. Meaning that the Roche XS is near enough for me. Greater accuracy may be lost if it requires greater consistency in patient sampling. I know that in my first year biochem labs I had dreadful difficulty in calibrating oxygen sensors in solutions and it took me some weeks to get it right.

Love to hear about that though...

 

[Protimenow]

A few things:

I think that there can be substantial differences from one lab to another, if the blood is mishandled before it reaches the lab. This is what I think happened with the clinic.
The fact that reagents may have different values from lab to lab should be irrelevant - that's what INR is all about -- it is a ratio between prothrombin time and reagent value and is designed to yield results based on the reagent value.

As far as the papers I mentioned - I've posted one or more here over the past few months -- I googled INR (or something) and got the paper from the Duke Clinics, which discussed how they respond to low INRs. It probably wouldn't take a lot of looking to find it on the Internet.

 

[pem]

sounds interesting. I would wonder at the benefits, thinking pareto principle here and that 80% of the benefits come from this 20% solution. Meaning that the Roche XS is near enough for me. Greater accuracy may be lost if it requires greater consistency in patient sampling. I know that in my first year biochem labs I had dreadful difficulty in calibrating oxygen sensors in solutions and it took me some weeks to get it right.

Love to hear about that though...

Agreed. The calibration mechanism would have to rely on little or no change in user behavior. The idea is that some calibration is better than no calibration. In this sense, Pareto suggests that a 20% calibration may address 80% of discrepancy. Perhaps?

I'm glad to find others on here who also like to geek-out on this stuff.

Best,
pem

 

[pem]

A few things:

I think that there can be substantial differences from one lab to another, if the blood is mishandled before it reaches the lab. This is what I think happened with the clinic.
The fact that reagents may have different values from lab to lab should be irrelevant - that's what INR is all about -- it is a ratio between prothrombin time and reagent value and is designed to yield results based on the reagent value.

As far as the papers I mentioned - I've posted one or more here over the past few months -- I googled INR (or something) and got the paper from the Duke Clinics, which discussed how they respond to low INRs. It probably wouldn't take a lot of looking to find it on the Internet.

The argument I've heard is that reagents with higher ISI values can amplify error (if you look at the formula, this makes sense). So, in principle, INR should address differences in reagent sensitivity, but if your blood is mishandled en route to the lab, resultant errors may be magnified more by a lab that uses a reagent with a higher ISI than a lab that uses a low ISI reagent. Does that make sense?

One of my favorite quotes: "The difference between theory and practice is that, in theory, there is no difference between theory and practice. In practice, there is." I wish I knew who to attribute this to.

Best,
pem

 

[Protimenow]

Yes, that makes sense. But mishandled blood is still mishandled blood - and whether or not the error is larger than it would have been, it's still an error. So, although Alere told me recently to 'trust the lab' (versus the InRatio result), I don't think that all labs can, necessarily, be trusted to always be accurate.


That's a great quote. It's the first time I've seen it.

 

[pellicle]

One of my favorite quotes: "The difference between theory and practice is that, in theory, there is no difference between theory and practice. In practice, there is." I wish I knew who to attribute this to.

My favourite is similar, but different

Theory: you understand why it should work, but it doesn't
Practice: it works, but you have no idea why
Here Theory and Practice meet: it doesn't work and I don't know why

 

[Superman]

Probably late to the party here, but just for future reference, I use a bit of a different technique when missing a dose. This works in part because I've always taken my dose in the morning as it allows for more consistency for me. That being said, I missed a dose a couple of weeks ago and realized it while at work. Rather than messing with my dosing, I just took my regular dose that evening. Then, over the next couple of days - I tapered the timing of my dose back to a morning dosing.

I self test weekly. My results over the last five weeks have been 3, 3.1, 3, 3.1, and 3. I'm not always that consistent - but recently I've been on a roll even with a missed dose.

I just hate starting the roller coaster if I can avoid it by dropping significantly higher doses here and there.

 

[tom in MO]

I take 6 meds a day and split my meds into morning and night. This is because for me, BP meds are best taken before work and antihistamines better before bed.

I always set out the next dose's pills. So if I forget my 6 am meds, I see them sitting waiting when I go for my 6 pm meds. However, with two times a day dosing, if I forget, I am never more than 12 hours behind when I find out that I forgot, thus I catch up by taking the full dose. I've forgotten a few doses, but taking warfarin 12 hours late has never affected by INR with weekly and then biweekly testing.

 

[Protimenow]

This sounds like a good idea, Tom. For me, I take mine before bed -- usually roughly the same time each night. Getting into that habit has worked for me.

Oddly, my INR has dropped in the past few weeks - I don't know why - and I've slightly upped my dose. I'll check again today.

There's no way to completely avoid missing a dose -- perhaps having a tiny vial that links to the keychain where my car keys are, with my warfarin in it would be of value if some kind of emergency keeps me from getting home to take my nightly dose.

 

[marc_kowal]

This sounds like a good idea, Tom. For me, I take mine before bed -- usually roughly the same time each night. Getting into that habit has worked for me.

Oddly, my INR has dropped in the past few weeks - I don't know why - and I've slightly upped my dose. I'll check again today.

There's no way to completely avoid missing a dose -- perhaps having a tiny vial that links to the keychain where my car keys are, with my warfarin in it would be of value if some kind of emergency keeps me from getting home to take my nightly dose.

If you're worried about not getting home because of some type of emergency, you could always carry a small pill box in your pocket. I have a plastic one that's about 1.5" X 1.5". Even if we're just going to the in-laws for the day (they live one state over), I always take my weekly pill box with me, "just in case".

 

[pem]

Rather than messing with my dosing, I just took my regular dose that evening. Then, over the next couple of days - I tapered the timing of my dose back to a morning dosing.

I'll have to remember that. In my case, I was late by a full 24 hours, but in retrospect, I realize I may have overcompensated by increasing my dose from 9m to 14mg. I think I just freaked out because my INR was 1.7. In the future, I can imagine maintaining my dose, or increasing by one or two mg over several days while bridging with Lovenox (I still have leftover that hasn't expired).

Thanks,
pem

 

[pem]

I take 6 meds a day and split my meds into morning and night. This is because for me, BP meds are best taken before work and antihistamines better before bed.

I always set out the next dose's pills. So if I forget my 6 am meds, I see them sitting waiting when I go for my 6 pm meds. However, with two times a day dosing, if I forget, I am never more than 12 hours behind when I find out that I forgot, thus I catch up by taking the full dose. I've forgotten a few doses, but taking warfarin 12 hours late has never affected by INR with weekly and then biweekly testing.

This is very interesting! I wonder if I should split my Coumadin dosing into two equal parts, taking half of the full dose in the morning and half in the evening. That would seem to be a buffer against being 24 hours late. On the other hand, there would be twice as many opportunities to miss a dose

Thanks,
pem

 

[pem]

perhaps having a tiny vial that links to the keychain where my car keys are, with my warfarin in it would be of value if some kind of emergency keeps me from getting home to take my nightly dose.

I do exactly that. And this is a good reminder to refresh my vial (since it is in my pocket all the time, it stays pretty warm and probably loses its "freshness").

If you want a link to vials that work for this, let me know. I think the one I have is waterproof.

Best,
pem

 

[pellicle]

On the subject of containers I keep a film container with a dose in my drawer at work. I also keep a small bottle (left over from the Marevan that has been consumed) with a small mixture of 1, 3, and 5mg in the car for coping with dinner invitations that turn into over night stays.

Takes the stress out of it

 

[pellicle]

Hi

thought I'd run two scenarios through the model I have on a "missed dose" scenario. This model run starts with INR 2.8 and dose of 7mg daily.

As may be obvious, the top graph is simply missing a dose and going on with it, the bottom graph is the scenario of adding about 1/2 of the missed dose to your next dose and then returning to normal.

In both cases the INR returns to 2.8 in the same period. I added in the moving 3 period average trendline as I felt that it fitted what I observed of data taken from my most recent hospital visit. Considering that each data sample is based on dose + half life of the remanent calculated for 4 periods back, I have no sound basis for it other than that perhaps it may represent the buffering I seem to see on actual results. I say this because in particular one would not expect to see a sudden drop in INR the next day from a missed dose as the model gives. (shrug)

The model is that spreadsheet formula set I gave earlier with a constant coefficient to bring the INR values to what I personally get with my metabolism. Of course it does not take into account compounding metabolic changes such as other drug interactions.

I find it interesting to see the results and will perhaps use myself as a guinea pig should I miss a dose any time soon. Data for each graph plot represents about $6 to me so I'll scratch my head over that one.

 

[pem]

Thanks for running the simulation. Now you've got me curious. I might play around with this too!

There is still the question of whether a quick return to nominal INR is better or worse (in terms of health outcomes) than a gradual return, even without overshooting. Some studies suggest that rapid fluctuations are more dangerous than being out of range. On the other hand, it may depend upon *how* out of range one is. Your thoughts?

Best,
pem