The intended goal of the intervention is to increase masticatory activity and ultimately, QoL. Therefore, the basic conditions to enable mastication have to be met; one has to have a healthy mouth and must be given food that requires chewing. This results in an intervention aimed at improving oral health care and increasing food consistency.
Oral health care comprises brushing the teeth at least once a day, for at least one minute per jaw. This is a minimum requirement, brushing the teeth twice daily is recommended, in combination with the use of toothpicks and dental floss. Clinical lessons are offered to train the nursing staff in providing this oral health care. Instructions on how to care for a dental prosthesis is also topic of these lessons. The specific skills needed for providing oral health care are also practiced. This is in concordance with the 2007 ‘Guideline oral health care in (residential) care homes for elderly people’ .
Increasing food consistency is achieved by: a) evaluation of the need for pureed foods and b) offering food of tougher consistency. Due to apraxia (i.e. inability to perform tasks or movements) or dysphagia (i.e. difficulty in swallowing), some participants are not able to chew and swallow food of normal consistency; hence, their food needs to be pureed. However, some participants might be given pureed foods without medical need, for example out of convenience for the nursing staff. A qualified speech therapist can diagnose swallowing disorders and therefore the need for pureed food. Participants who are given pureed food without medical need will be reintroduced to more solid foods. All participants able to masticate normally are offered food of tougher consistency, such as apples, bread including its crust, crunchy cookies, raw vegetables and salads.
Quality of Life
The primary outcome variable is QoL, assessed with the Qualidem questionnaire . The Qualidem is considered the preferred instrument  for rating QoL for elderly persons with dementia and is appropriate for large and small-scale settings . A proxy is asked to rate observable behaviors on a 4-point Likert scale (i.e. ‘never’, ‘rarely’, ‘sometimes’, and ‘often’). Statements are for example: ‘is cheerful’, ‘refuses food’, ‘smiles’, or ‘wants to stay in bed’. For each statement, 0, 1, 2 or 3 points are awarded. The most positive outcome is given the highest point value (e.g., ‘smiles often’ is given 3 points, ‘refuses food often’ is given 0 points). A higher score suggests a higher QoL. The Qualidem score can be divided into several subscales. Not all subscales are appropriate for participants with severe dementia, so only the four subscales recommended for this group  will be included; ‘care relationship’ (0-21 points), ‘positive affect’ (0-18 points), ‘restless tense behavior’ (0-9 points), and ‘social isolation’ (0-9 points). The maximal score is 57 points.
As described earlier, QoL comprises many aspects of life. In order to assess this multicomponent aspect of QoL, the following secondary outcome variables are included.
Cognition, especially memory and executive function, is investigated using a pen-and-paper-based neuropsychological assessment. Trained examiners, blind for the intervention, will visit the participants at the care unit. First, the participant is screened with the MMSE. Based on the MMSE score, a set of neuropsychological tests will be conducted. If a participant is not able to give a single adequate response on any of the MMSE questions (i.e. MMSE score is zero) no further cognitive testing will take place. Participants scoring 1-4 on the MMSE will take the first four tests described below, and participants scoring 5 or higher on the MMSE will perform these tests and three additional tests (also described below). All these tasks are complementary as they assess different cognitive skills.
Category fluency is assessed in two separate instances, by asking the participant to name either as many animals or as many professions as possible in one minute [39, 40]. Time is measured with a stopwatch, and all responses are recorded. Identical responses are counted only once (e.g., horse, horse; doctor, doctor), responses assigning gender (e.g., lion, lioness; steward, stewardess) are counted as two correct responses. These rules are explained to the participant before starting. Faulty responses are ignored (i.e. not counted nor subtracted). If the participant starts a random conversation or remains quiet, a gentle reminder (‘can you name any other animals/professions?’) is given. The obtained score is the amount of correct responses.
Memory and attention
Memory and attention are assessed by verbally presenting sequences of numbers to the participant, who has to repeat them . The sequences start out with a length of two digits, and after three items, one extra digit is added. The task is cut off when two out of three responses are incorrect. Only correct responses are counted; faulty responses are ignored. The participant is allowed to make corrections. Maximal score is 21.
Working memory is assessed with a digit span backwards task, which is virtually the same as above; only this time the participant has to give the response in the reverse order . New sequences of digits are used. The task is cut off when two out of three responses are incorrect. Only correct responses are counted; faulty responses are ignored. The participant is allowed to make corrections. Maximal score is 21.
In order to assess visuospatial function, incomplete line drawings are shown, while the participant has to indicate what the images represent . The drawings are of increasing difficulty, showing animals or everyday items or situations (e.g., a fish, a book, or a man carrying something heavy). After five incorrect responses, the task is cut off. In case of incomplete responses (e.g., ‘a man’ instead of ‘a man carrying something heavy’), the examiner asks the participant to elaborate: ‘please describe everything you see’. Only correct responses are counted; faulty responses are ignored. The participant is allowed to make corrections. Maximal score is 20. If a participant scores 5 points or higher on the MMSE, three tests are added:
Verbal long term memory
To assess verbal long term memory, the examiner reads out loud a list of eight everyday words (such as ‘pencil’ or ‘bird’), which the participant must repeat from memory after each presentation; this is repeated five times . Points are awarded for correct responses; the maximal score is 40. After approximately 10 minutes, a delayed free recall and recognition condition will be administered. During the recognition condition, the participant has to indicate whether a word does or does not belong to the original list. Sixteen words are now read out loud, the eight original words and eight new words. Maximal score for free recall is 8, maximal score for recognition is 16.
Nonverbal episodic memory
A visual memory task is used to measure nonverbal episodic memory . A card with eight red squares printed on it is placed between the examiner and the participant. The examiner taps the squares in a certain order which the participant is asked to repeat. The sequences start with a length of two squares and after two trials the sequences are lengthened with square. The task is cut off when both responses of a certain length are incorrect. Only correct responses are counted; faulty responses are ignored. The participant is allowed to make corrections. Maximal score is 14.
Nonverbal working memory
By using virtually the same task as above, nonverbal working memory is assessed . This time the participant is asked to give the response in the reverse order, the printed squares are colored green and new sequences are used. The task is cut off when both responses of a certain length are incorrect. Only correct responses are counted; faulty responses are ignored. The participant is allowed to make corrections. Maximal score is 14.
Assessment of mood
Mood is assessed with two questionnaires regarding observable behaviors, measuring either depression or agitation. Both questionnaires will be filled out by a proxy, typically a member of the permanent nursing staff.
The presence or absence of depression is qualified using the Cornell Scale for Depression in Dementia (CSDD;  Dutch version (CSDD-D; . Nineteen statements about the participant are scored on a Likert scale: a = not able to judge; 0 = absent; 1 = slightly or variably present; and 2 = severe. Maximal score is 38 points; a higher score indicates the presence of more depressed behaviors. Statements refer to behaviors such as suicidal tendencies and facial expressions of sadness or fear, but also to weight loss or sleep disturbances. A score below 8 is considered to be within the normal range and a score of 8 and higher is indicative of depression [43, 45].
The amount of agitated behaviors is scored using the Dutch version of the Cohen-Mansfield Agitation Inventory (CMAI) . A total of 29 observable behaviors are scored on a 7-point Likert scale. Items are behaviors such as spitting, (verbal) aggression, biting, screaming, complaining, and others. The behaviors can be observed never (1), less than once a week (2), once or twice a month (3), several times per week (4), once or twice per day (5), several times per day (6), or several times per hour (7). Minimal score is 29; maximal score is 203. A higher score is indicative of more agitated behaviors.
Assessment of independence
The ability to perform activities of daily living is assessed with the Katz index of Independence in Activities of Daily Living . This index rates five activities and the ability to perform them unaided. They are: dressing, using the toilet, eating, moving around, and taking a shower or bath. A sixth question regards whether the participant is incontinent. A score of 1 indicates independence, score 2 stands for ‘needs some help’, and a score of 3 is given when someone is completely dependent (or, completely incontinent). Minimal score is 6; maximal score is 18.
Assessment of the rest-activity rhythm
The rest-activity rhythm is a circadian rhythm that reflects the sleep-wake rhythm in an indirect way. The rest-activity rhythm is measured during a week (7*24 hours), using an Actiwatch activity monitor (Cambridge Neurotechnology Ltd., Cambridge, United Kingdom). The Actiwatch is a small device that is worn on the wrist. It is placed on the dominant arm, unless this leads to agitation (e.g., due to presence of a wristwatch or bracelet). The Actiwatch is taken off only during showering since it is not waterproof. The Actiwatch records the motions of the arm, which are a proxy for overall activity . The recorded data are analyzed for the following variables:
Interdaily stability (IS)
The interdaily stability is a measure for the degree of similarity of activity patterns within the measured period. A stable rhythm is characterized by a higher IS score; scores are between zero and 1.
Intradaily variability (IV)
The intradaily variability is the difference in activity levels in periods throughout the day. A normal activity pattern shows low IV; a sinusoid pattern results in a score of zero, a score of 2 is obtained for noise, and higher scores can arise obtained due to peaks in activity.
Relative amplitude (RA)
The relative amplitude is calculated by dividing the difference between the means of the ten most active hours (M10) and the five least active hours (L5) by the sum of these means within a day/night cycle.
A larger RA is associated with a more pronounced wake/sleep cycle. A more detailed description of these variables and their analysis is available elsewhere .
Assessment of blood pressure
Blood pressure is measured in millimeters of mercury (mmHg) with a blood pressure monitoring device (SpaceLabs Medical Inc., Redmond, Washington, United States of America). The participant is sitting down quietly, unless they are bedridden in which case they are lying down. Systolic blood pressure and diastolic blood pressure are noted, as well as heart rate. Hypertension is defined as systolic blood pressure higher than 160 mmHg and/or diastolic blood pressure over 95 mmHg .
Assessment of masticatory function
Mastication can be assessed in several ways. Most notably is the distinction between subjective, self-rated masticatory ability and the objective outcome masticatory efficiency (also described as masticatory performance) [50, 51]. In the present study, masticatory ability is not assessed, since the cognitively impaired participants are not likely to be able to reliably answer questions about their chewing capacity. Participants’ masticatory efficiency is assessed using several techniques. First of all, the dental status is recorded: is the participant dentate, is he/she wearing a partial of full prosthesis, etc. If possible, the dental records are used for obtaining this information, otherwise, nursing staff is interviewed, or a visual inspection is performed. Secondly, several assessments are taken:
The number of pairs of upper and lower teeth that touch each other when the mouth is closed (i.e. occluding units) are measured using dental modeling wax (Alminax; Müller & Weygandt, Büdingen, Germany). The wax is solid at room temperature and becomes soft when immersed in warm water. The participant bites down on a plate of softened wax and then it is allowed to harden again. Determination of the number of occluding units is done by visual inspection. In complete dentitions, 8 teeth are present in each quadrant (two incisors, one canine, two premolars, and three molars). Thus, the maximal score is 16 occluding units (i.e. 32 teeth).
Maximal mandibular excursions
Active, voluntary, mandibular mobility is assessed as the distance between the upper and lower incisal edges, in millimeters, and is measured with a flexible plastic ruler (Dental Union, Nieuwegein, The Netherlands) [30, 52]. First, the participant is asked to open his/her mouth as wide as possible without experiencing any pain. Next, the participant is encouraged to open his/her mouth as far as possible, even if this is painful, thus enabling the assessment of the maximal mouth opening. During the maximal opening, the participant is continuously encouraged. Maximal protrusion (i.e. outward forward movement of the mandible) is assessed by asking the participant to push the lower jaw forward as far as possible. Laterotrusion (i.e. outward movement of the jaw to the side) to the left and right are also measured. After every mandibular excursion, the participant is asked whether this is painful and if so, where the pain is located and what the intensity of the pain is. Pain location is recorded as joint area, pre-auricular, cheek area, floor of the mouth, temporal area, and other. Pain intensity is recorded on a 5-point Likert scale: 0 = no pain, 1 = tenderness, 2 = mild pain, 3 = moderate pain, and 4 = severe pain. Overjet (i.e. the antero-posterior distance between the upper and lower incisors), overbite (i.e. the vertical overlap of the upper and lower incisors), and midline deviation (i.e. the horizontal distance between the upper and lower dental arch midlines) are recorded while the participant rests in occlusion. The vertical maximal opening is corrected for overbite, the protrusion is corrected for the overjet, and the laterotrusions are corrected for the midline deviation.
Maximal voluntary bite force (MVBF)
The maximal voluntary bite force (MVBF) is measured with the VU University Bite Force Gauge (VU-BFG). The VU-BFG is a hand-held device which uses a load cell to measure maximal voluntary bite force in kilograms. The VU-BFG can be used centrally between the incisors or unilaterally between the (pre-) molars. For this study, the VU-BFG will be used between the incisors. The participant is instructed to bite as long and hard as possible and is encouraged continuously during the sampling. If the sampling fails (e.g., due to losing the prosthesis), a second attempt will be made after a rest period. Maximal sampling time is 20 seconds, and sampling takes place at a frequency of 50 Hz. All bite force samples are logged; the highest (i.e. peak) value is used as MVBF.
In order to quantify actual masticatory performance, a mixing ability test in which the participant has to orally knead two viscoelastic colored materials [53, 54] is used. A four-gram sample made of blue and pink chewing gum (Bubblicious® Ultimate Original and Twisted Tornado; Cadbury, London, United Kingdom) is given to the participant, with the instruction to chew naturally. The sample resembles a piece candy due to its general appearance (bicolor capsule-shaped sample in a cellophane wrapper) and smell (sugary and sweet) which makes it easier for the participant to accept it as test food. A casting mold ensures consistent production of samples. Several protocols use a fixed number of chewing cycles, e.g. . However, participants suffering from dementia may find it hard to count and chew at the same time. An observer cannot accurately distinguish individual chewing cycles either, due to for example swallowing in between chewing motions, movements of the head while chewing (e.g., looking around), presence of a tremor, or obscuring of the jaw due to facial features such as sagging skin or beard (pilot data, not shown). Using a fixed amount of chewing cycles is therefore not possible in this population. Since in healthy people the average chewing frequency is stable at approximately 1.4 hertz (1.45 Hz for women and 1.30 Hz for men; , the assumption is made that this is also the case in persons suffering from dementia. This was confirmed in a pilot study (data not shown) and therefore a standard chewing time of 20 seconds will be used. A stopwatch is used to measure time. After digital optical analysis, a mixing ability score is obtained [53, 54].