Ment (MNA) . Body mass index (BMI = weight in kilograms/squared height in kilogram meters) was also calculated. Balance was assessed using the `one-leg balance’ test described by Vellas et al. . The test was performed by asking the subject to stand unassisted on one leg as long as possible (eyes open, barefoot or not, using whichever leg was spontaneously chosen by the subject). The `one-leg balance’ test was reported as abnormal when the subject was unable to stand on one leg for five seconds or more . The test was performed twice with the best result used for the analysis. Caregivers were asked about the patient’s current medication. Additionally, at assessment interview, subjects were asked to bring with them all their regular medication. Prescriptions of acetylcholinesterase (AChE) inhibitors (galantamine, donepezil or rivastigmine), psychotropic treatments (anxiolytics, neuroleptics, serotonin reuptake inhibitors and other antidepressants), anti-Parkinsonian treatment (dopamine agonist or dopaminergic agent) and vitamin D were collected and used as additional covariates. NMDA-receptor antagonist (memantine) was not available in France at that time, and, thus, not considered for the present analyses. Subjects and caregivers were asked about nursing-home admission and hospitalization. The reason for hospitalization was recorded. Subjects who did not attend follow-up appointments were contacted by telephone and post if necessary.Statistical analysis(disability, caregiver burden, cognitive and nutritional status, body mass index, balance, behavioral and psychiatric symptoms of dementia (BPSD), medication, hospitalization and institutionalization), survival analyses were based on the measure collected during the visit preceding the event, except for hospitalization and institutionalization which were considered regardless of the period when they occurred . The AChE inhibitor treatment (or the absence of treatment) was taken into account during at least the six-month period preceding the decline in walking ability. For time-independent variables, analyses were based on the measure collected at baseline. Cox proportional hazards models with discrete times (because decline in walking ability occurs between the follow-up visits) using a backwards selection procedure were performed with P 0.05 as PubMed ID:https://www.ncbi.nlm.nih.gov/pubmed/27241280 a removal criterion. Two models (with or without the ADL introduced in the initial model) were performed due to the fact that the ADL score is strongly correlated with walking ability and decline in walking ability may be both the cause and/or the consequence of the ability to perform basic motor tasks. Tests based on interaction with time were used to ascertain the proportional hazards assumption for timeconstant variables (age, gender, level of education, length of time from diagnosis of Ro 31-8220 (mesylate) References dementia, cardiovascular risk factors, cardiovascular disease or other comorbidities, CT). Statistical interactions were verified PubMed ID:https://www.ncbi.nlm.nih.gov/pubmed/24737349 in the final model. P-values were based on two-sided tests. All statistical analyses were performed using SAS software (version 9.3, SAS Institute Inc. Cary, NC, USA).Baseline characteristics were described using mean values ?standard deviation (SD) and proportions for quantitative and qualitative variables, respectively. Survival analyses (Cox models) were performed to identify independent predictive factors of decline in walking ability, using relative risk (RRs) and 95 confidence in.