Using DALYs in TreeAge Markov Models
Occasionallly (and recently), I received a question about how to use DALYs in a Markov model within TreeAge Pro. People nearly always use QALYs rather than DALYs in their models, so I have limited experience using DALYs.
I understand DALYs to be a measurement of mortality and morbidity on patients, representing the overall disease burden. This is a little trickier to use in TreeAge Pro than QALYs.
With QALYs, you can simply enter the utility for each state and that amount will be accumulated as the cohort passes among the health states in the model.
With DALYs, it is unclear exactly what to do. I believe it would make sense to enter a kind of disutility related to DALYs at all health states other than dead and fully healthy. Please comment if you have thoughts on this.
This brings us to the dead state. We need to add DALYs for the dead state to account for a portion of the cohort being dead from this point in time into the future. However, I'm unclear on the details.
- Should we accumulate 1 DALY for every cycle until the end of the time horizon?
- Should we accumulate 1 DALY for every cycle until the end of expected end of life?
- Is there a better approach?
I welcome all comments on this post. Thank you for your input.
Andrew Munzer
TreeAge Software
Comments
Here are my thoughts, in the "for what it's worth" category. I would love to see other approaches.
Since DALY = YLL + YLD (years of life lost + years of life with disability), a Markov model could include a tracker that accumulates:
I think this would give every simulated subject a DALY value. Exporting the simulation results to Excel would allow you to calculate the mean, SD and other statistics on the DALY in each therapy strategy.
I think this is what you suggested, Andy, but I supplied a little more detail that I hope is helpful to others.
Melissa,
I think I would take the same approach of setting the Dead state up not as an absorbing state/terminal node, but as a label node returning the cohort back to the same Dead state for the remainder of the time horizon. I also agree that the DALY/disutility would be 1 for the Dead state. I'm not totally sure about how long to run people in the Dead state. It kind of makes sense to run it all the way to the end of the time horizon, but I'm not totally sure. Maybe it would be more appropriate to run it to the average age of death. I just don't know.
With respect to sick/disabled states, it feels like the DALY/disutility value should not be 1 (like death). I would think that healthier states would have small fractional values, while very sick/more disabled states would have larger fractional values. What do you think?
Andrew
Andrew,
You are right; please accept my apologies. According to the paper (link below), disability years are not weighted at 1, but at a fraction depending on the severity of the disability. Then, of course, each year of life lost or disabled has to be discounted appropriately.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3776440/
I would like to use DALYs and am wondering if any body is using DALYs in Treeage
Hi Caroline,
I will be using DALYs in Treeage and wondering how you went with your analysis if you did end up using DALYs.
Hi Melissa,
I tried accessing the link you had posted but its showing up as invalid. just wondering if you were able to post the article link or name here?
Please refer to the comments in this article:
https://treeage.zendesk.com/hc/en-us/community/posts/1500000044161-DALYs-in-Markov-Model-Cost-Effectiveness-Analysis
The referenced example model is included with your TreeAge Pro software.
Please sign in to leave a comment.