Changes in the agreement between the Minimum Data Set and hospital Medicare claims measures of dementia

A comparison of dementia diagnoses and cognitive function measures in Medicare claims and the Minimum Data Set

BACKGROUND

Gold standard dementia assessments are rarely available in large real-world datasets, leaving researchers to choose among methods with imperfect but acceptable accuracy to identify nursing home (NH) residents with dementia. In healthcare claims, options include claims-based diagnosis algorithms, diagnosis indicators, and cognitive function measures in the Minimum Data Set (MDS), but few studies have compared these. We evaluated the proportion of NH residents identified with possible dementia and concordance of these three.

METHODS

Using a 20% random sample of 2018-2019 Medicare beneficiaries, we identified MDS admission assessments for non-skilled NH stays among individuals with continuous enrollment in Medicare Parts A, B, and D. Dementia was identified using: (1) Chronic Conditions Warehouse (CCW) claims-based algorithm for Alzheimer's disease and non-Alzheimer's dementia; (2) MDS active diagnosis indicators for Alzheimer's disease and non-Alzheimer's dementias; and (3) the MDS Cognitive Function Scale (CFS) (at least mild cognitive impairment). We compared the proportion of admissions with evidence of possible dementia using each criterion and calculated the sensitivity, specificity, and agreement of the CCW claims definition and MDS indicators for identifying any impairment on the CFS.

RESULTS

Among 346,013 non-SNF NH admissions between 2018 and 2019, 57.2% met criteria for at least one definition (44.7% CFS, 40.7% CCW algorithm, 26.0% MDS indicators). The MDS CFS uniquely identified the greatest proportion with evidence of dementia. The CCW claims algorithm had 63.7% sensitivity and 78.1% specificity for identifying any cognitive impairment on the CFS. Active diagnosis indicators from the MDS had lower sensitivity (47.0%), but higher specificity (91.0%).

CONCLUSIONS

Claims- and MDS-based methods for identifying NH residents with possible dementia have only partial overlap in the cohorts they identify, and neither is an obvious gold standard. Future studies should seek to determine whether additional functional assessments from the MDS or prescriptions can improve identification of possible dementia in this population.

Developing a Machine Learning Risk-adjustment Method for Hospitalizations and Emergency Department Visits of Nursing Home Residents With Dementia

BACKGROUND

Long-stay nursing home (NH) residents with Alzheimer disease and related dementias (ADRD) are at high risk of hospital transfers. Machine learning might improve risk-adjustment methods for NHs.

OBJECTIVES

The objective of this study was to develop and compare NH risk-adjusted rates of hospitalizations and emergency department (ED) visits among long-stay residents with ADRD using Extreme Gradient Boosting (XGBoost) and logistic regression.

RESEARCH DESIGN

Secondary analysis of national Medicare claims and NH assessment data in 2012 Q3. Data were equally split into the training and test sets. Both XGBoost and logistic regression predicted any hospitalization and ED visit using 58 predictors. NH-level risk-adjusted rates from XGBoost and logistic regression were constructed and compared. Multivariate regressions examined NH and market factors associated with rates of hospitalization and ED visits.

SUBJECTS

Long-stay Medicare residents with ADRD (N=413,557) from 14,057 NHs.

RESULTS

A total of 8.1% and 8.9% residents experienced any hospitalization and ED visit in a quarter, respectively. XGBoost slightly outperformed logistic regression in area under the curve (0.88 vs. 0.86 for hospitalization; 0.85 vs. 0.83 for ED visit). NH-level risk-adjusted rates from XGBoost were slightly lower than logistic regression (hospitalization=8.3% and 8.4%; ED=8.9% and 9.0%, respectively), but were highly correlated. Facility and market factors associated with the XGBoost and logistic regression-adjusted hospitalization and ED rates were similar. NHs serving more residents with ADRD and having a higher registered nurse-to-total nursing staff ratio had lower rates.

CONCLUSIONS

XGBoost and logistic regression provide comparable estimates of risk-adjusted hospitalization and ED rates.

Dementia diagnosis in the hospital and outcomes among patients with advanced dementia documented in the Minimum Data Set

BACKGROUND

Individuals with dementia do not always have a diagnosis of dementia noted on their hospital claims. Whether this lack of documentation is associated with patient outcomes is unknown. We examined the association between a dementia diagnosis listed on a hospital claim and patient outcomes among individuals with a Minimum Data Set (MDS) assessment.

METHODS

A retrospective cohort study was conducted using administrative claims data and nursing home MDS assessments. Hospitalized patients aged 66 and older with advanced dementia noted on an MDS assessment completed within 120 days prior to their first hospitalization in 2017 were included. Advanced dementia was defined based on an MDS diagnosis of dementia, dependency in four or more activities of daily living, and a Cognitive Function Scale score indicative of moderate to severe impairment. Multilevel regression with a random intercept at the hospital level was used to examine the relationship between documentation of dementia in inpatient hospital Medicare claims and the following patient outcomes after adjusting for patient and hospital characteristics: invasive mechanical ventilation (IMV) use, intensive care unit or coronary care unit (ICU/CCU) use, 30-day mortality, and hospital length of stay (LOS).

RESULTS

In 2017, among 120,989 patients with advanced dementia and a nursing home stay, 90.57% had a dementia diagnosis on their hospital claims. In adjusted models, documentation of a dementia diagnosis was associated with lower use of the ICU/CCU (adjusted odds ratio [AOR]: 0.78 [95% confidence interval 0.74, 0.81]), use of IMV (AOR: 0.50 [0.47, 0.54]), and 30-day mortality (AOR: 0.81 [0.77, 0.85]). Patients with a dementia diagnosis had a shorter LOS.

CONCLUSIONS

Among patients with advanced dementia, those whose dementia diagnosis was documented on their inpatient hospital Medicare claim experienced lower use of ICU/CCU, use of IMV, lower 30-day mortality, and shorter LOS than those whose diagnosis was not documented.