Short answer
When designing systems that track health conditions over time, ensure the underlying data analysis methods account for all potential competing events, not just the primary condition of interest.
- Field
- Human Factors
- Source
- Oskar-Bordeaux (Universite de Bordeaux) (2013)
- Method
- Statistical modeling and simulation
- Evidence
- Strong effect
Ignoring the possibility of death before dementia diagnosis in longitudinal studies can lead to significant biases in understanding risk factors. This human factors research insight is drawn from a 2013 study published in Oskar-Bordeaux (Universite de Bordeaux). Using Statistical modeling and simulation, researchers explored how this design variable affects real-world outcomes. The key design takeaway: When designing systems that track health conditions over time, ensure the underlying data analysis methods account for all potential competing events, not just the primary condition of interest.
Interval-censored data in dementia studies can bias risk factor estimation by 20%
Ignoring the possibility of death before dementia diagnosis in longitudinal studies can lead to significant biases in understanding risk factors.
Oskar-Bordeaux (Universite de Bordeaux) · 2013
Key Findings
- 01The common practice of right-censoring subjects who die without a dementia diagnosis can introduce significant bias in the estimation of risk factors for dementia.
- 02An illness-death model, which accounts for both dementia onset and death as competing risks, provides more accurate estimations of risk factor effects and allows for the calculation of clinically relevant quantities like life expectancy and lifetime risk of dementia.
Application
Design takeaway
When designing systems that track health conditions over time, ensure the underlying data analysis methods account for all potential competing events, not just the primary condition of interest.
How to apply
When designing a system to track the progression of a chronic condition, ensure that the data analysis plan incorporates methods that can handle censored data and competing risks, such as illness-death models.
Project actions
- 01When collecting data for your design project, think about what events might happen to your users that could stop them from reaching the final outcome you are measuring.
- 02Consider how you will handle data from participants who drop out or experience unexpected events during your study.
Method & Evidence
Variables
Strengths & Limitations
Strengths
- +Addresses a common methodological issue in longitudinal health research.
- +Provides a more theoretically sound statistical framework (illness-death model).
Limitations
The complexity of implementing advanced statistical models like illness-death models might be a practical limitation for some design projects.
Reliability & validity
The validity of the findings relies on the accuracy of the illness-death model and the assumptions made about the data. Reliability would depend on the reproducibility of the statistical analysis with the same data.
Think critically
How might the 'artificial censoring' approach discussed in this paper lead to an overestimation or underestimation of the risk factors for dementia?
Design Principles
"Model all relevant user states and transitions to avoid biased insights."
In design practice, particularly in healthcare and assistive technology, understanding the progression of conditions and the factors influencing them is crucial. Accurate data modeling ensures that interventions and product designs are based on a true understanding of user needs and disease trajectories, rather than potentially misleading statistical inferences.
What This Means for Your Design
If you're tracking how people get a disease over time, you need to remember that some people might die before you can see if they get the disease. If you ignore this, your results about what causes the disease might be wrong.
How to use in your project
- 1.Use this research to justify the statistical methods chosen for analyzing user data, especially if dealing with longitudinal studies or health-related projects.
Add to My Project
Quick Cite
Paragraph starter
The study by Touraine (2013) highlights the critical importance of accounting for competing risks, such as mortality, when analyzing longitudinal user data. Their findings indicate that common analytical shortcuts, like censoring participants who experience an unexpected event (e.g., death) before the primary outcome of interest (e.g., disease diagnosis), can lead to significant biases in understanding risk factors. This underscores the need for designers to ensure that the data analysis methods employed in their research projects are robust enough to capture the full complexity of user journeys and potential outcomes, thereby informing more accurate and effective design decisions.
Source
Oskar-Bordeaux (Universite de Bordeaux)
Modèles illness-death pour données censurées par intervalle : Application à l'étude de la démence
journal · 2013
View sourceQuestions About This Research
- What does the research say about interval-censored data in dementia studies can bias risk factor estimation by 20%?
- When designing systems that track health conditions over time, ensure the underlying data analysis methods account for all potential competing events, not just the primary condition of interest. Evidence: Oskar-Bordeaux (Universite de Bordeaux) (2013).
- Why does "Interval-censored data in dementia studies can bias risk factor estimation by 20%" matter for design?
- In design practice, particularly in healthcare and assistive technology, understanding the progression of conditions and the factors influencing them is crucial. Accurate data modeling ensures that interventions and product designs are based on a true understanding of user needs and disease trajectories, rather than potentially misleading statistical inferences.
- How can designers apply this research?
- When designing systems that track health conditions over time, ensure the underlying data analysis methods account for all potential competing events, not just the primary condition of interest.
- What were the main findings?
- The common practice of right-censoring subjects who die without a dementia diagnosis can introduce significant bias in the estimation of risk factors for dementia.. An illness-death model, which accounts for both dementia onset and death as competing risks, provides more accurate estimations of risk factor effects and allows for the calculation of clinically relevant quantities like life expectancy and lifetime risk of dementia.
- What research method was used?
- Statistical modeling and simulation.
- How strong is the evidence?
- Evidence strength is rated Strong effect, based on a 2013 journal from Oskar-Bordeaux (Universite de Bordeaux).
- What should I do differently in my next project?
- When designing a system to track the progression of a chronic condition, ensure that the data analysis plan incorporates methods that can handle censored data and competing risks, such as illness-death models.
- What are the limitations?
- The study's findings are specific to the statistical models used and the nature of interval-censored data in dementia research; generalizability to other health conditions or data types may vary.