Development and validation of a 5‐year mortality prediction model using regularized regression and Medicare data

Abstract

Purpose De‐implementation of low‐value services among patients with limited life expectancy is challenging. Robust mortality prediction models using routinely collected health care data can enhance health care stakeholders' ability to identify populations with limited life expectancy. We developed and validated a claims‐based prediction model for 5‐year mortality using regularized regression methods. Methods Medicare beneficiaries age 66 or older with an office visit and at least 12 months of pre‐visit continuous Medicare A/B enrollment were identified in 2008. Five‐year mortality was assessed through 2013. Secondary outcomes included 30‐, 90‐, and 180‐day and 1‐year mortality. Claims‐based predictors, including comorbidities and indicators of disability, frailty, and functional impairment, were selected using regularized logistic regression, applying the least absolute shrinkage and selection operator (LASSO) in a random 80% training sample. Model performance was assessed and compared with the Gagne comorbidity score in the 20% validation sample. Results Overall, 183 204 (24%) individuals died. In addition to demographics, 161 indicators of comorbidity and function were included in the final model. In the validation sample, the c‐statistic was 0.825 (0.823‐0.828). Median‐predicted probability of 5‐year mortality was 14%; almost 4% of the cohort had a predicted probability greater than 80%. Compared with the Gagne score, the LASSO model led to improved 5‐year mortality classification (net reclassification index = 9.9%; integrated discrimination index = 5.2%). Conclusions Our claims‐based model predicting 5‐year mortality showed excellent discrimination and calibration, similar to the Gagne score model, but resulted in improved mortality classification. Regularized regression is a feasible approach for developing prediction tools that could enhance health care research and evaluation of care quality.

Validation of a 5‐Year Mortality Prediction Model amongU.S.Medicare Beneficiaries

“…In this external validation study using a nationwide sample of Medicare beneficiaries, the Lund-Lewis claims-based prediction model for 5-year mortality, which includes indicators of comorbidity and functional impairment, showed strong discrimination and no decline in performance compared with the NC Medicare population where it was developed. 5 Further, the model showed consistent performance over time and by census region, sex, age group, and race, indicating strong generalizability. The model also had high discrimination at 30, 90, 180, and 365 days.…”

“…See Lund et al for full details of the development of the predictive model. 5 We estimated predicted 5-year mortality for each beneficiary in the U.S. Medicare sample by applying the model parameter estimates derived in the original NC training cohort. We compared performance with the comorbidity score model, 13 also obtaining mortality predictions using the model parameter estimates from a logistic model including the comorbidity score, age polynomials (linear, quadratic, and cubic), and sex fit in the original NC training cohort.…”

“…1,2 In quality measurement (eg, Centers for Medicare & Medicaid Services [CMS] Five-Star Quality Rating System and the National Committee for Quality Assurance Healthcare Effectiveness Data and Information Set [HEDIS] measures), mortality prediction models can be used for case-mix adjustment to compare and benchmark the performance of different providers, institutions, or plans more accurately, particularly when sample size constraints do not allow for adjustment of several variables. 3,4 Lund and colleagues developed a Medicare claims-based risk prediction model for 5-year mortality 5 that considered demographics, several comorbid conditions, [6][7][8][9] as well as indicators of frailty, 10 disability, 11 and functional impairment 12 (hereafter referred to as the Lund-Lewis model). The model was developed in a 2008 training cohort of North Carolina (NC) Medicare beneficiaries following an office visit by patients aged 66 and older.…”

“…The Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis (TRIPOD) statement describes the importance of external validation to identify and address potential overfitting. 15 Because the Lund-Lewis model was developed using Medicare data from a single state, the internal validation results 5 may not be applicable to the broader U.S. Medicare population. To that end, we aimed to validate the Lund-Lewis model externally in a nationwide sample of U.S. Medicare beneficiaries.…”

Impact of Rubrics, Addie and Gagne Model on the Performance of Students in Programming Subject

“…In short Impact of Addie model has already been proven in versatile domain. In this study author has experimented the delivery steps of Addie model for the performance enhancement of programming subject along with the coordination of well design rubrics.in [8][9][10] usage of gagne model in the various domain has been explained. In [11][12][13][14][15] importance of rubrics in the assessment and its designed parameters are explained.…”