WLS Codebooks

Note: Codebooks contain frequency data from the private data release. For some variables, these differ from the frequencies in the public release. Variables that differ between the public and private versions include "(not available on public release)" as part of the variable description, or a note indicating the difference between the public and private versions.

Summary Variables

2020 ILIAD Summary Variables

OVERVIEW for all of ILIAD 2020

With funds from the National Institute on Aging the WLS partnered with UW
Madison's Alzheimer Disease Research Center (ADRC) to conduct a new round of
interviews starting in 2019.  ILIAD 2020 (Initial Lifetime's Impact on
Alzheimer's Disease and Related Dementias) includes either one or two
interviews for each participant for whom we had a prior measure of cognition or
another criterium. 

First, we conducted a "short" interview that took 30 to 40 minutes.  The focus
was on memory and thinking and we repeated activities used in prior rounds of
the study and for the first time, we administered the Telephone Interview for
Cognitive Status-modified (TICS-M). We also updated some family and health
measures.  Next, we recruited participants who scored below a cutoff on the
TICS-M to participate in a "long" interview with both a trained survey
interviewer (IV) and an Advanced Practive Provider (APP) also known as a nurse
practitioner. We planned for the long interview to be an in-person interview
with both the IV and APP present. Because of the Covid-19 pandemic we stopped
in-person interviews shortly after we started. We restarted the long interviews
after developing a comporable phone instrument. As vaccines and testing became
available we were able to return to in-person visits. By the end of our
fielding period for the long follow-up interview, participants who completed
the 2020 long instrument did so in one of four ways. See measure q1a942re for
the four different combination of modes for the long interview.

If we learned that the intended participant died or was too ill to be
interviewed, we recruited an informant to answer questions about the
participant's cognitive health.  These informant interviews used the Dementia
Questionnaire (DQ) which was scored for dementia. See measure stat20DQ for the
number and type of interviews we completed using the DQ

The information gathered by the IV and APP was presented to a group of
clinicians at a consensus conference.  Taking into account other medical
conditions and symptoms, the clinicians assigned each participant a level of
impairment (q1a951re) and accompanying primary, contributing and
non-contributing factors.  Along with level of impairment we include MCI
Subtype (q1a952re) and whether the consensus panel determined that Alzheimer's
Disease was primary, contributing or not present.  Along with Alzheimer's
disease the consensus committee also noted additional etiologies of impairment.
These other etiologies do not include enough cases to make available on the
public release of the data. Researchers needing these additional measures
should contact wls@ssc.wisc.edu.

The information collected on the DQ was first processed using an algorithim to
approximate a dementia diagnosis. Next a clinician looked at the outcome of the
algorithim as well as the detailed notes that the interviewers captured during
their conversations.  The clinician confirmed the diagnosis and also assigned a
level of confidence to the diagnosis based on the DQ.  See measures q1a954re
and q1a955re. 

Finally for the ease of researchers wishing to combine cases that completed the
long interview with cases for which we only have proxy data we create two
versions of a combined diagnosis measure (q1a956re and q1a957re)

Researchers wanting a precise measure that minimizes false positives (aka type
1 error) should consider q1a956re which we labeled "Positive Predictive Value
(PPV) Summary Score."  Conceptually PPV=(True Positives/(True Positives + False
Positives). For this measure we were more conservative in who we designated as
meeting the criteria for a research diagnosis of dementia, MCI, or other
impairment.  The cost of using this measure is that you might miss some cases
(i.e. have more false negatives).

Researchers wanting a broader measure that minimizes false negatives (aka type
2 error) should consider q1a957re which we labeled "Negative Predictive Value
(NPV) Summary Score." Conceptually NPV=(True Negatives)/(True Negatives + False
Negatives). For this measure any indication of poor cognitive performance at
any point (for any reason) was used to move cases from the "cognitively normal"
category into the "impaired" category. This is important from a population
health perspective because it's more likely to capture any signal of potential
cognitive problems. While it results in less precision for identifying cases,
it allows for the most comprehensive estimate of potential *needs* associated
with poorer cognitive performance.

Researchers wanting more details on how we created a resesarch diagnosis can
learn more by reading cor 1029 in Appendix X. 

We are currently repeating the same protocol with the same participants for
ILIAD 2023.


BRIEF VARIABLE DESCRIPTIONS
	2020 ILIAD Participation
		stat20short  2020 ILIAD Participation in Short Interview
		stat20long   2020 ILIAD Participation in Long Interview
		stat20DQ     2020 ILIAD Proxy Participation in Dementia Questionnaire

	Information about the 2020 ILIAD interviews 
		q1a003re     Age at time of Short interview
		q1a020re     State of Residence
		q1a934re     2020 FIPS code for home address
		q1a935re     2020 Census Tract code of home address
		q1a936re     2020 Census Block code of home address
		q1a937re     2020 Place FIPS code of home address
		q1a938re     2020 Minor Civil Divison code of home address
		q1a016rem    Month Completed ILIAD 2020 Short Interview
		q1a016rey    Year Completed ILIAD 2020 Short Interview
		q1a026rem    Month Informant Completed ILIAD 2020 DQ
		q1a026rey    Year Informant Completed ILIAD 2020 DQ
		q1a940re     Mode of Short interview
		q1a941re     Months between Short and Long interviews
		q1a942re     Mode of Long Interviews

	Outcome of 2020 ILIAD Long Interviews
		q1a951re     Level of cognitive impairment via Consensus
		q1a952re     MCI Subtype
		q1a953re     Consensus outcome for Alzheimer's Disease
			
	Outcome of 2020 ILIAD Dementia Questionnaire (DQ)
		q1a954re     Research Diagnosis Via Proxy
		q1a955re     Confidence of Proxy Diagnosis
		
	Combined Outcome of DQ and Long Interview
		q1a956re     Positive Predictive Value Summary Score
		q1a957re     Negative Predictive Value Summary Score
		
	Imputed Cognition
		q1a958re     Imputed Cognition for people who qualified for, 
		             but did not complete, the Time 1 long Interview

Menu