What is the effect of changing an treatment characteristic on patient preferences for that treatment? This is an important question of interest for many patient preference studies. In practice, it is often done by estimating the effect due to changing a characteristic of a profile while average on the distribution of remaining profile characteristics. Formally, it is known as the average marginal component effect or AMCE (see Hanmueller et al 2014). How do researchers do this?
… About 90% of the existing convenor analysis uses an equal distribution. The problem is that the resulting estimate of amce, which we call Uniform Amce (UAMCE), all convener gives equal weight to the profile, even when some of them are unrealistic from a significant perspective.
Consider the case of using a combination for various cancer remedies. Treatment characteristics may include overall existence of treatment, progress -free existence, adverse event rates and methods of administration. By using UAMCE, you are assuming a uniform distribution at all characteristic levels. In practice, however, it is likely that when the PFS characteristic level is high, the OS is high; Conversely, when PFS is low, the level of OS characteristic will also be low. UAMCE does not pay attention to that a PFS (high) OS (high) or PFS (low) S (low) profile More likely characteristic profile.
An alternative approach -sent by La Costa et al. (2021) -Using this population Amce (Pamce). This approach is average on the distribution of profile characteristics in a target population of interest rather than assuming a similar possibility for all characteristic combinations. Where will the data come from? Authors argue that it can come from real-world data- as the real-world market share- or a lectal distribution of various drugs that are of theoretical interest.
Author also proposes two new experimental design approaches using Pamce.
First approach, which we call Design-based confirmation analysisThe design phase includes the target profile distribution … in the most natural design, which we say joint population randomly, we propose random combination profiles according to their target profile distribution rather than uniform …
Our second approach, Model-based searching analysisAfter raising the profile and collecting data, the target profile at the analysis phase takes into account the distribution … This approach is useful in estimating Pamce when researchers randomly random profile It falls, such as uniform.
The author offers some useful applications of Pamce from the world of political science in full lesson.
A challenge to apply Pamce in the context of life science is that RWD may be rare or limited target characteristic profile to populate the profile. RWD may be limited that researchers may have to see patient preferences for new drug treatment already in the market. Since new treatment cannot be used in practice yet, it is not clear what weight it will receive. Additionally, a limited number of treatment can be available on the market. In this case, Pamce 2 or 3 may fall in comparison to direct remedies, rather than a fictional practice of patient preferences in all admirable combinations of treatment of different characteristic levels.
Authors also note that another challenge is that the joint distribution of all characteristics can be difficult to get in many research applications. It is proposed to use the author to solve it
Marginal population randomizing design, allowing researchers to freely remember each factor with its marginal distribution. However, this design requires a strong perception that there is no three-way or high-order interaction between the characteristics.
Nevertheless, it is useful to understand that when the characteristic distribution is considered as the same possibility (UAMCE), what is the perception of the vested and whether to consider the actual feature distribution in the goal population (PAMCE). Analysis studies.
You can read full study Here,
