2009 HSR&D National Meeting Abstract
1078 — Estimating Dementia Prevalence Using Administrative versus Prospective Data
Williams JW (Durham COE), Holsinger TC
(Durham COE), Stechuchek KM
(Durham COE), Coffman C
To estimate an overall prevalence of dementia in primary care (PC) patients using administrative and prospectively collected data at Durham VAMC.
Prospective study (GEMS):
Subjects were identified from upcoming PC appointments based on eligibility criteria of age > = 65, at least 2 visits to primary care within the past 18 months, living within 90 miles of the Durham VAMC in a non-institutional setting with no ICD-9 diagnoses of dementia or psychosis and an AUDIT C < 7. Consenting subjects were evaluated by a research nurse using the CERAD neuropsychological battery, a semi-structured questionnaire and medical history review. Informants completed an interview about the subject’s cognitive and functional ability. An experienced panel of clinicians applied DSM-IV criteria to achieve a consensus diagnosis of normal, cognitive impairment no dementia, or dementia. We estimated unrecognized dementia using the percentage diagnosed with dementia.
Subjects were identified based on having a PC visit between May 2006-May 2008, age > = 65, at least 2 visits to PC within the past 18 months, living within 90 miles of the Durham VAMC, with no ICD-9 codes for psychosis. Excluded were veterans enrolled in the GEMS study. We estimated recognized dementia using the percentage having a dementia ICD-9 code.
Overall dementia prevalence:
We applied the unrecognized dementia rate to those who did not have an ICD-9 dementia code and combined them with those with recognized dementia to get the overall dementia prevalence rate.
Demographic characteristics for the 6,663 identified in the administrative data pull were: mean age 76.1 ± 6.7 years; 99% male; 62.5% white, 21.5% black, 14.8% unknown, and 1.2% other. GEMS study subjects were similar demographically.
In the GEMS study, 21 of 478 (4.4%) were given a consensus diagnosis of dementia. In administrative data, 463 veterans had a dementia ICD-9 code (6.9%). We estimate that 272 veterans have unrecognized dementia. The combined estimate of recognized and unrecognized dementia is 735/6,663 (11.0%).
Administrative data may significantly underestimate the prevalence of dementia in veterans.
As the U.S. population ages, accurate estimates of dementia prevalence are needed to inform planning for expenditures in all healthcare settings. Current VA projections for dementia range 3.6-5.1% and may significantly underestimate coming needs.
If these data are confirmed in other VA outpatient settings, policy planners should be better prepared for increased related expenditures as the Vietnam-era cohort ages.