How to interpret CEA results with different cycle length
Linh Quan
Hello,
I am conducting a cost-utility analysis, I have built a Markov model where the cycle length is 4 months. In the CEA results text report, my incremental effectiveness is 3.56 and my ICER is ~$31,000. How would you interpret this to present in a paper? Should I convert it to years?
The intervention provides an additional 3.56 quarter life-years? I assume to convert to years the calculation would be 3.56 x 4/12 = 0.89, so an additional 0.89 life-years? And then to get cost by year: $31000/0.89 = ~$35,000 per LY?
Your help and recommendations are appreciated!
Comments
You can report effectiveness values and incremental effectiveness in any unit. However, people will generally expect results to be reported in QALYs. I recommend you use QALYs, so the values and the ICER are more recognizable to your audience.
If your cycle is 4 months, it is simple to report model results in QALYs.
For each utility value accumulated at a health state, adjust it by multiplying it by 4/12, which is explicitly changing your 4-month value to an annual value.
For example, let's say you have a state called Severe with utility uSevere, change the values entered at that state to uSevere*4/12.
Then you will accumulate effectiveness with QALYs units for each 4-month cycle.
I think this is an excellent answer, but what about the cost?
If costs are per cycle (that is, costs per 4 months) do you also multiply cSevere (the cost of being in the severe stae for 4 months) by 4/12? Why or why not?
Thank you!
Jonathon
The example provided just references the utilities in terms of providing a demonstration of how to adjust the units. But you are correct. If your model had a 4-month cycle, then you need to ensure the values you are accumulating each cycle are "per 4 months". So, if your model had annual costs of $9,000, then for the health states you should have costs of: (4/12)*9,000.
Thanks for asking the question.
Consistent with Belinda's comment above...
If your costs are already includes inputs that represent 4-month costs, then you would not adjust them by 4/12.
Thank you both.
To make sure I am fully clear on this.
If the cycle is 4 months and the utility per cycle (utility for 4 months) is 0.8 you multiply this by 4/12 to report model results as QALYs?
So, 0.33333333333*0.8 = 0.26666666666 or 0.27 QALYS?
Whereas costs per cycle in the model (costs per 4 months) do not need to be multiplied by 4/12?
It depends on your inputs values.
I think utility/QALY is the clearest example. Utilities are values that should be entered in the appropriate units - something like uStage1 = 0.75. That is the most natural form for the input. In this case, you would want to enter uStage * 4/12 into the Markov View to apply the utility over a 4-month period.
Cost is not as clear because the natural/recognizable input could be a 4-month cost or it could be an annual cost.
I would recommend specifically entering "4/12" rather than the fraction just to be super explicit about what you are doing. The results will be the same.
Andrew
The best way to think of this is the units of your utility. (And then we can consider costs too).
If your Utility per Annum (NOT per cycle) is 0.8, then (4/12)*0.8 will provide you the utility per (4-month) cycle, and return the outcome of the model as Utility over years (QALYs).
For costs, in your model if you already have the cost per 4-month cycle, you can use this directly in the model. If you had Annual costs, then the same process as above (4/12)*Annual_cost will give you cost per 4-month cycle to use in the model. Both costs and utilities used in your health states have to be of the same time unit as such.
Hi Belinda,
Should the following be will provide you the cost or will provide you the utility?
Just wondering if it's a typo?
That was a typo. I corrected in that note above.
Please sign in to leave a comment.