A randomized, 6-month, prospective study was conducted at the clinic of the Behavioral Neurology Sector, in the Neurology and Neurosurgery Department of Universidade Federal de São Paulo - Escola Paulista de Medicina (UNIFESP/EPM).
The sample consisted of 90 subjects, of both genders, aged at least 65 years old and with probable AD, according to Diagnostic Statistical Manual 4th Edition (DSM IV) criteria and clinical dementia rating (CDR) of 1, 2 or 3.
For exclusion criteria were considered other forms of dementia, alternative feeding requirement (tube feeding), type 1 and 2 diabetes mellitus, and renal diseases.
Twelve subjects were included in the study but not in the statistical analys is: 3 subjects from CG and 4 from EG had difficulty in being transported to the hospital; 3 subjects from SG and 1 from CG died; 1 subject from SG needed tube feeding.
The remaining subjects were randomized into 3 groups: control group (CG), (N = 27), education group (EG), (N = 25), and supplementation group (SG), (N = 26).
All subjects were assessed at baseline and then at monthly intervals during the 6-month study period, including an orientation of health nutrition. The subjects nutritional status was assessed using anthropometric and biochemical data. The anthropometric data included: height (m), current weight (kg), Body Mass Index (BMI) (kg/m2), arm circumference (AC) (cm), arm muscle circumference (AMC) (cm), and triceps skinfold (TSF) (mm). For weight were used Welmy® mechanical scales for adults with a 150 kg capacity and for height was utilized stadiometer graduated in centimeters, to for subjects that were able to maintain erect posture. For the others with presented posture problems as kyphosis or lordosis and were usable to stand, the stature and knee equation proposed by Chumlea (1985)  was used, to avoid biases in measures of statures. For the compartmental muscle and fat mass assessment, Lange® Skinfold calipers were used, which expresses results in millimeters (mm). The biochemistry data included: total protein (TP), serum albumin and total lymphocyte count (TLC). Biochemical data were collected following a 12-hour fast and evaluated by the central laboratory of Hospital São Paulo.
For cognitive tracking, all subjects were assessed using the Mini-Mental State Examination (MMSE) assessing temporal and spatial orientation, memory, attention, calculation, language and praxis [12, 13] and Clinical Dementia Rating (CDR) [14, 15] administered by a duly qualified neuropsychologist.
The following demographic data were also collected for all subjects: education level (years of schooling), time of disease evolution calculated from the date of diagnosis and dependence during meals.
Subjects in the CG were monitored by monthly nutritional assessments and did not receive any form of intervention. The 25 Subjects caregivers and patients, in the EG participated in the educational program which consisted of 10 classes. Each class was taught to a maximum of 10 participants, with the aim greater interaction between the professional and caregivers. Each expositive class was supported by slides, with themes were of data proposed in accordance with the Brazilian Association of Alzheimer's (ABRAZ) Caovilla & Canineu (2002)  a nonprofit association that assists caregivers and family members of patients with Alzheimer's disease. Also it took into account the information received from the caregivers based on the main nutrition deficits in AD patients. The classes were developed with relevant topics to the needs of nutritional intervention, such as: the importance of nutrition in disease, behavioral changes during meals, attractive meals, constipation, hydration, administration of drugs, swallowing, food supplementation, lack of appetite, clarification of doubts. This order of exposition follows the progression of AD and the onset of symptoms that may be related to nutrition. Subjects in the SG received oral nutritional supplementation twice daily for 6 months in addition to their usual diet (Ensure with FOS®, Abbott Nutrition) and were measured monthly by means of anthropometry and biochemical parameters, as described before. Two servings provide 680 kcal and 25.6 g protein/day.
All subjects or their representatives signed the Informed Consent Form. This research was approved by the Ethics Committee of Universidade Federal de São Paulo (protocol 0552/06).
Statistical analysis was undertook using Kruskal-Wallis variance analysis to compare the three groups in relation to anthropometric, biochemistry, demographic and disease stage variables, and Siegel's chi-square test to study possible associations between the groups and the variables studied. The rejection level of the null hypothesis was fixed at α = 0.05.