Premium
Does an advanced statistical technique perform better than a simple technique for dementia prediction?
Author(s) -
Goerdten Jantje,
Carriere Isabelle,
Terrera Graciela Muniz
Publication year - 2020
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1002/alz.040837
Subject(s) - covariate , proportional hazards model , dementia , regression , statistics , regression analysis , econometrics , demographics , linear regression , mathematics , medicine , demography , disease , sociology
Background Cox regression models (CM) employ two strong assumptions: the proportional hazards (PH) and the log‐linearity (LL) of covariates. These assumptions might not hold for all predictor variables, which in turn can lead to biased results. (Abrahamowicz et al., 1997; Ritchie et al., 2016). Generalised Cox regression models (GCM) enable to independently test and relax the assumptions of CM (Abrahamowicz & MacKenzie, 2007). Method Data is used from the Survey of Health, Ageing and Retirement in Europe (SHARE), which includes around 140,000 individuals aged 50 or older followed over seven waves. Data from The Aging, Demographics, and Memory Study (ADAMS) is used for external validation. CM and GCM are used to estimate dementia risk. Result None of the predictor included in the analyses fulfils the assumptions of Cox regression. The pseudo R 2 is 0.06 (95% CI: 0.047, 0.062; CM) and 0.493 (95% CI: 0.460, 0.506; GCM). The likelihood ratio test results in a p‐value of <0.001. Both models predict dementia moderately well (10‐year risk: 0.737; 95% CI: 0.695, 0.783; CM and 0.746; 95% CI: 0.711, 0.790; GCM). Conclusion The GCM performs significantly better than the CM in modelling dementia risk. However, this approach has also disadvantages: larger sample needed, only visual examination of effects, long computational time possible. Yet, the generalised Cox regression is an interesting extension of the Cox regression. The option to relax and independently test the assumptions is especially appealing when including continuous variables. References: (1) Abrahamowicz, M., du Berger, R., & Grover, S. A. (1997). Flexible modeling of the effects of serum cholesterol on coronary heart disease mortality. Am J Epidemiol, 145 (8), 714‐729. (2) Abrahamowicz, M., & MacKenzie, T. A. (2007). Joint estimation of time‐dependent and non‐linear effects of continuous covariates on survival. Stat Med, 26 (2), 392‐408. doi:10.1002/sim.2519. (3) Ritchie, K., Carrière, I., Berr, C., Amieva, H., Dartigues, J.‐F., Ancelin, M.‐L., & Ritchie, C. W. (2016). The clinical picture of Alzheimer's disease in the decade before diagnosis: clinical and biomarker trajectories. The Journal of clinical psychiatry, 77 (3), e305‐311. doi:10.4088/jcp.15m09989.