Validation of Claims Algorithms to Identify Alzheimer’s Disease and Related Dementias

Abstract

BACKGROUND Using billing data generated through healthcare delivery to identify individuals with dementia has become important in research. To inform tradeoffs between approaches, we tested the validity of different Medicare claims-based algorithms. METHODS We included 5,784 Medicare-enrolled, Health and Retirement Study participants aged >65 years in 2012 clinically assessed for cognitive status over multiple waves and determined performance characteristics of different claims-based algorithms. RESULTS Positive predictive value (PPV) of claims ranged from 53.8-70.3% and was highest using a revised algorithm and 1-year of observation. The trade-off of greater PPV was lower sensitivity; sensitivity could be maximized using 3-years of observation. All algorithms had low sensitivity (31.3-56.8%) and high specificity (92.3-98.0%). Algorithm test performance varied by participant characteristics, including age and race. CONCLUSIONS Revised algorithms for dementia diagnosis using Medicare administrative data have reasonable accuracy for research purposes, but investigators should be cognizant of the trade-offs in accuracy among the approaches they consider.

Trends in US Medicare Decedents’ Diagnosis of Dementia From 2004 to 2017

“…Medicare claims have recently been shown to have a positive predictive value between 60% and 70%, which may make it challenging to disentangle upcoding, correction of historical underdiagnosis, and misdiagnosis. 36 Additionally, 2 recent articles 36 , 37 and the present analysis demonstrate that underdiagnosis may no longer warrant using 1 claim as opposed to the standard approach of using 2 in ADRD-diagnostic criteria. Thus, changes in ADRD diagnosis over time among Medicare decedents could affect payment and comparison of quality-of-care outcomes.…”

Psychotropic medication prescribing in assisted living and nursing home residents with dementia after the National Partnership

“…First, we limited our analysis to fee‐for‐service Medicare beneficiaries, and our results may not be generalizable to beneficiaries covered by Medicare Advantage for whom we did not have access to encounter data for the entire study period. We used a validated claims‐based algorithm with high specificity to identify beneficiaries with ADRD limiting our population to those with an ADRD diagnosis on an administrative claim 16 . In AL facility populations, there are no federally mandated sources of clinical data such as the Minimum Data Set for Medicare and Medicaid certified NHs to add to our ADRD identification.…”

Time-lagged relationships between a decade of air pollution exposure and first hospitalization with Alzheimer's disease and related dementias

“…Sixth, hospital billing codes classifying ADRD may be susceptible to misclassification ( Festa et al, 2022 ). A recent study showed the CCW algorithm applied to Medicare parts A and B data had a 92.3 % specificity and 56.8 % sensitivity in identifying patients with ADRD ( McCarthy et al, 2022 ). Because our dataset only includes part A claims we believe our outcome will have similar specificity but lower sensitivity, resulting in potential bias towards the null ( Taylor et al, 2002 ).…”