Multi-study and multivariate evaluation of healthy life expectancy (HLE): An IALSA workshop on multistate modeling using R . edit me
- shared dropbox folder with manuscript production
- Follow
./utility/reproducibility-instructions.mdto create dynamic documents linked below: ./manipulation/map/0-ellis-island-map.R./reports/review-variables.R./manipulation/map/1-encode-multistate.R
- letter of invitation
- venue: Amsterdam, April 14-16, 2016
- instructional and academic resources
- overview of the studies and variables
This is a hands-on workshop for the purpose of active data analysis on a variety of datasets during the two-day workshop. We expect this to be a productive workshop with the aim of publishing outcomes of analyses that are developed during the workshop. Although this is also a learning opportunity, we encourage participants to participate in a multi-study papers before publishing independent results.
The objectives of this workshop are:
-
- Evaluate sensitivity of using alternative variables/definitions of HLE and Cognitive Impairment/Dementia-Free HLE and replication across studies.
-
- Evaluate utility of multistate model for making individual predictions of future transitions and confidence in making individual predictions.
- Candidate Studies : HRS, ELSA , LASA , WLS , MAP , OCTO-Twin, H-70, LBSC
- Outcomes : ADL/IADL, MMSE < 24, self-rated health, diagnosis of dementia, other cognitive variables (decisions about thresholds and alternative models required)
- Major covariates : age at baseline, sex , education, SES/SEP
- Minor covariates : BMI, sedentary behavior (physical activity) , smoking (various), alcohol use (various), chronic conditions (comorbidity index)
- closer look
The workshop lectures will be delivered by Dr. Ardo van den Hout (UCL, Lecturer at the Department of Statistics), developer of the R package ELECT that estimates these models. Participants are encouraged to review at the ELECT web site to prepare for this workshop.
Here are some example operationalizations: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2966893/ and http://www.ncbi.nlm.nih.gov/pubmed/25150976. For example, the BRFSS question used to assess health status was "Would you say that in general your health is excellent, very good, good, fair, or poor?" For this study, participant responses of "fair or poor" were categorized as "unhealthy" and "excellent, very good, or good" as "healthy."
encode-multistate-map]: