Is Weight Loss More Severe in Older People with Dementia?

Methods of dietary intake collection

The selection of methodology for dietary assessment of older people with dementia often present unique challenges. The impairment in memory and ability to recall invalidates the use of retrospective methods (such as 24 h dietary recall and food frequ-ency questionnaire) unless an appropriate surrogate is present [19]. The presence of cognitive decline, especially in severe cases of dementia, may also render prospective dietary assessment methods (such as food record and food diary) inaccurate due to the inability of older people with dementia to record food intake. Therefore, an appropriate surrogate is often necessary when studying the dietary intake of people with dementia. The use of a surrogate also presents problems as the surrogate may not be with the study participants at all times, may not know the exact composition of composite dishes (if food is prepared by others) and there may be multiple individuals involved in the care of the subject [20]. Additionally, surrogate reporters may also be subjected to incre-ased carer burden from having the additional task of reporting the participants’ food intake. This may create a reactivity bias where surrogates alter the dietary behaviors of the people under their care for ease of recording [19]. Therefore, it is important to recognize that dietary assessment per se is difficult in older people and that this difficulty is compounded in older people living with dementia due to cognitive decline affecting the accuracy and ability to recognize and estimate food intake as well as the complexity of invo-lving surrogate reporters.

Table 1 summarizes 24 studies involving more than 3,200 participants that investigated energy in-take in older people with dementia compared with non-dementia controls in both non-free living and free-living conditions [7, 21–43]. Most of these stu-dies employed prospective dietary assessment methods, except for three studies [24, 41] which used retrospective dietary assessment methods. One study [42] did not state the dietary assessment method at all. The three-day food record/diary (consisting of two-day weekdays and one-day weekend) dietary ass-essment method used by most studies is considered more representative and accurate than single observation 24 h recalls, especially for use in older people with cognitive decline [44–46]. However, as with all dietary assessment methods, the three-day food re-cord is still prone to underreporting due to many factors including mis-quantifying of amount of food consumed [45, 47–52]. The three-day plate waste method and three-day weighed food record methods used by some of the studies [7, 38] mitigate the issue of mis-quantifying food as everything is weighed. However, these methods impose a high burden on the caregivers or participants and therefore may not always be feasible in free-living individuals.

The use of new technologies to assist in the re-cording of intake events may improve accuracy and alleviate the participants’ burden in order to increase the quality of the data on food intake. Newer technologies utilizing computerized assessment coupled with image-assisted or image-based technology such as handheld devices, wearable video camera to augment established techniques have been well-validated against food diary and weighed food record among the older people [53–56]. The Novel Assessment of Nutrition and Aging (NANA) is one such tool which eliminates the limitation of retrospective assessment methods as dietary information as well as photographs of the food and drink are recorded as they are about to be consumed [55]. Furthermore, this tool reduces interviewer bias as the dietary intake collection is self-administered [55]. Another tool called the Dietary Intake Monitoring System (DIMS) integrates a weighing machine as well as a camera [56]. Utilizing this combination of weight and image data, the tool exerts its advantage over traditional dietary assessment and image-based assessment methods by more accurately estimating portion sizes [56].

Factors affecting energy intake in older people with dementia versus non-dementia controls

Increasing age had been independently associated with declining nutritional status in older people [57]. Therefore, age may be a confounding factor in terms of energy intake especially when non-dementia controls are not age-matched or correction for age was not made during analysis. The majority of studies in Table 1 were corrected for age and showed no significant differences between the two groups. However, four studies [21, 43] did not provide details of any statistical comparison of age between the group or any correction for age, and seven studies [25–27, 39] did not account for age differences between the participants. This lack of detail and consideration for age differences in the participants limit the validity of the results because the presence of age as a confounding factor was not considered when the comparison was made.

Most studies included participants of varying degree of dementia severity (from mild to severe) determined using the Mini-Mental State Examination (MMSE), with six studies using alternative or additional methods of classification such as Clinical Dementia Rating (CDR) and Global Deterioration Scale (GDS). However, three studies [21, 29] did not report on the severity of the dementia in the participants.

It is also important to consider the stage of dementia as a possible confounding factor since later stages of dementia have been associated with eating problems such as slow eating and food refusal [14, 58]. However, despite grading the dementia severity, most studies did not account for this difference in dementia severity when comparing energy intake between the groups. Instead, all older people with dementia were considered as a single dementia group. One study [32] grouped dementia participants into early and late stage but did not find significant differences in energy intake between the two sub-groups. Significant differences were instead observed between the sub-group’s lean body mass and their energy balance (TEE –energy intake) [32]. Similarly, a recent study by Salminen et al. [59] found that there is no significant difference in energy intake between non-free living older people with severe dementia compared to the ones with mild dementia (classified according to CDR). Despite this observation, there is a significantly higher proportion of the former relying on oral nutritional supplements in order to match the energy intake of older people with mild dementia. Hence, older people with more severe dementia tend to require additional interventions in order to maintain their energy intake.

More studies (n = 13) were conducted in older people with dementia under free-living conditions than in non-free living older people with dementia (n = 8), with 3 studies conducted in mixed setting (non-free living and free-living) conditions. In the studies involving free-living participants, older people with dementia had caregivers reporting food intake while healthy controls self-reported. This dichotomy created the possibility of a reporting bias due to the presence of a proxy-reporter for older people with dementia when compared to controls who self-rep-orted their own intake. For example, in one study [7] it was reported that AD patients group ate more than healthy controls. However, it was unknown if the excessive eating was due to caregiver’s insistence of giving more food to the AD patients. It is also important to consider whether the caregivers reported what was served or what was actually eaten. This external influence might not exist in the non-dementia control group who ate ad libitum, hence creating a reporting bias.

The majority of the studies observed no significant difference in energy intake between older people living with or without dementia. Similarly, earlier reviews did not observe significant lowered energy intake in older people with dementia when compared to non-dementia controls [16, 60]. In the current review, five studies [21, 40] (male only in Thomas et al. [21]’s study and female only in Puranen et al. [40]’s study) observed significantly lower energy intakes in older people with dementia when compared to non-dementia controls, while five studies [7, 40] found significantly higher energy intake in older people with dementia compared to non-dementia controls. The observations that older people with dementia had significantly higher energy intake compared to healthy controls are in direct contrast to the frequently reported unintended weight loss commonly found in patients with dementia [60]. As mentioned above, the higher energy intake observed in older people with dementia compared to non-dementia controls may be attributed to caregiver persuading older people with dementia to consume more thereby increasing the latter’s intake [7] compared to healthy controls who might not have the same external motivation to increase their intake. It is also important to consider possible variations affecting the outcome such as differing study setting. For example, one of the studies [29] was conducted with the non-dementia controls living in the community compared to older people with dementia who lived in long-term care setting. This different environment means that the results are difficult to fully interpret.

One study [23] accounted for setting differences through comparing three environments, i.e., non-free living older people with dementia, free-living older people with dementia, and healthy free-living older people. The study observed that both older people with dementia living in a non-free living setting and free-living older people with dementia had significan-tly higher energy intake compared to healthy control. However, the dietary assessment was only conducted on a randomly selected subgroup (n not reported) of each of the main groups, hence reducing the sample size. The dietary assessment method was also different in that they used an investigator-led three-day weighed dietary record for six days in total for the healthy controls and free-living older people and 21 days for non-free living older people with dementia. There was no statement on whether the three-day weighed dietary record was performed consecutively on randomly selected days within the period of observation or whether the choice of days was left to the investigator’s discretion. This was a potential reporting bias which might have affected the results.

Most studies shown in Table 1 did not report whether participants with dementia lost weight or reported no active weight loss. Only four studies [7, 43] reported active weight loss among the participants. Wang et al. [7] observed that older people with dementia had significantly higher energy intake/kilogram (kg)/day compared with healthy control despite the dementia group having lower body mass index (BMI) and a higher percentage of participants living with dementia was actively undergoing weight loss. The reasons for this observation were unclear although the effect of focusing on food intake may draw the attention of both subjects and caregivers to food and therefore inadvertently encouraged food intake.

In terms of macronutrient intake, older people have higher requirement for protein than younger adults, at 1.0–1.2 g/kg/day to compensate for age-related changes in protein metabolism [61]. Table 2 summarizes the protein, fat, and carbohydrate intake of older people with dementia and non-dementia controls. There were 15 studies [7, 38–41] which reported the protein intake of the participants, though only five studies [7, 40] reported the protein intake as g/kg/day. Of the 5 studies, only families with male caregivers in Puranen et al. [40]’s study did not meet the higher protein requirement of older people. It was reported, however, that this observation may potentially be caused by male caregivers having more comorbidities, being older, and less skilled in food preparation compared to families with female caregivers [40].

Comparing the protein intake of older people with dementia and older people without dementia, only six studies reported significant differences with three studies reporting a higher protein intake in older people without dementia [21, 40] (female only in Puranen et al. [40]’s study) while the other studies reported higher protein intake in older people with dementia [23, 40] (male only in Puranen et al. [40]’s study). Hence, similar to the lack of evidence for a lower calorie intake in older people with dementia, there is also a lack of evidence for a lower protein intake in older people with dementia compared to older people without dementia.

Other than protein, fats and carbohydrates intake have been shown to have impacts on the development of dementia. Both fats and protein intake have been theorized to be crucial in the maintenance of neuronal membranes and myelin sheath in the brain [62]. A diet high in carbohydrates and refined sugars, on the other hand, may have deleterious effects on cognitive functions as high insulin level has been associated to a decline in cognitive function and hyperglycemia may increase oxidative stress and the formation of advanced glycation endproducts (AGE) [63–65].

There were 11 [7, 41] and 10 [7, 41] studies, respectively, which reported participants’ fat and carbohydrate intake. However, only two studies [34, 36] observed significantly lower fat consumption in older people with dementia and two other studies [22, 34] observing significantly lower carbohydrate consumption in older people with dementia compared to older people without. Wang et al. [7]’s study was the only one which observed a significantly higher carbohydrate consumption in older people with dementia compared to older people without. Interestingly, one study observed a significantly higher sugar intake in older people with dementia [39, 39] and one study observed a significantly higher frequency of intake of cookies and snacks in older people with dementia [41]. As discussed above, diet high in caloric intake from carbohydrate and low in caloric intake from fat and protein was shown to increase the risk of mild cognitive impairment or dementia in older people [62] and refined carbohydrate-rich diet [66] have been associated with increased risk of dementia. Whether or not the participants’ intake in the studies of Ahmed et al. [39] and Kim et al. [41] reflected their intake prior to developing dementia was uncertain due to the nature of the cross-sectional study design; however, this observation is crucial to avoid further exacerbation and deterioration of the mental status of the older people. It is hence important to not only ensure that older people with dementia consume enough calories but also consume enough from the right sources of calories.

In summary, current available literature does not demonstrate a lower energy and macronutrient intake in older people living with dementia compared to non-dementia controls in non-free living and free-living setting. However, those older people living with dementia are at high risk of weight loss and malnutrition. In order to further elucidate the reasons behind this, it is useful to consider whether an incr-ease in energy expenditure is the cause of the imbalance in energy equilibrium and therefore leading to weight loss.

Measurement of resting energy expenditure (REE)

Table 3 summarizes eight studies [26, 68] that compared resting energy expenditure in older people with dementia to non-dementia con-trols in both non-free living and free-living conditions. All included studies measuring REE, except for two [42, 68], utilized the gold standard REE measurement of indirect calorimetry. Among the six studies utilizing indirect calorimetry, five of the studies [26, 30–32] ensured a fasting period of at least five hours before measurement and three of these studies [27, 31] ensured that the participants abstained from vigorous physical activity for at least 14 hours before REE measurements or moderate physical activity for at least two hours before measurements. However, most of the six studies failed to mention if the measurements were done in thermal-neutral conditions and if a brief period of rest and acclimatization was accorded to the participants, especially if the latter traveled in the morning to reach the research center where REE were measured. These are important factors which might affect the REE readings if not controlled [69].

The main challenges of REE measurement in older people using indirect calorimetry are to ensure that participants remain in a restful state and are able to breathe smoothly especially throughout the steady state measurement period [70]. Difficulties such as agitation, fear, restlessness, and easily falling asleep had been reported in this population group [16]. None of the studies quoted in Table 3, however, mentioned such challenges during REE measurement in this group.

Factors affecting resting energy expenditure in older people with dementia versus non-dementia controls

The finding of increased metabolic rate in a triple transgenic mice model of AD [71] has contributed to the hypothesis that hypermetabolism may be the cause of weight loss in older people living with dementia. Furthermore, the fact that no significant difference between energy intake in older people living with dementia compared to non-dementia controls has been confirmed, it is then reasonable to suggest that a state of hypermetabolism exists which causes energy imbalance leading to weight loss. However, conflicting results have been seen in studies on energy expenditure in older people with dementia [26, 68]. Most of the studies comparing REE in older people living with dementia compared to non-dementia controls have not found significant differences between the groups [26, 68].

Wolf-Klein et al. [67] observed significantly higher REE levels in older people with dementia than non-dementia controls. They attributed this finding to higher fat-free mass (FFM) among older people with dementia [67] but did not discuss the possible reasons why FFM was higher in the dementia group. Unfortunately, the relatively small sample size of the study limits the generalizability of the findings.

Gender difference has been recognized as a contributing factor toward REE differences, with REE in males being reported as higher than in females [72, 73] of the same age and weight from puberty to adulthood. This difference is mainly due to differences in body composition between males and females namely FFM, fat mass and fat distribution [74]. However, this gender difference in REE may be less significant amongst those aged 60 and above due to age related decline in FFM. A review on energy requirements in older people, encompassing 2,450 participants, concluded that gender was not a determinant of REE for individuals in this age group [75]. Aside from study by Niskanen et al. [26] (whose participants were all women), the other seven studies identified did not perform subgroup analysis on REE differences between the gender. However, most of them either had matched participants for FFM or corrected for body composition differences in the analysis [27, 67]. Only two studies [42, 68] matched the participants by their BMI (surrogate measure of body fatness). Hence, the presence of confounding factors such as gender and body composition limit the validity of their results.

Ethnicity is another factor that may influence the REE of an individual. Individuals born in tropical regions have been observed with lower REE compared to Caucasians of similar age groups [76, 77]. Adzika Nsatimba et al. [78] also observed that there was a significant difference between the REE of sub-Saharan Africans compared to Australians of European descent. Furthermore, the various reports of over- and underestimation of REE when predictive equations were applied to different ethnic groups from which predictive equations were developed [79] suggested that the differences in REE between different ethnic groups may be independent of the individuals’ body composition. This may be explained by research on uncoupling protein genes (linked to an individual’s susceptibility to obesity) which suggested that these genes may be involved in the processes leading to lower REE of African American females compared to white females [80, 81]. Research investigating REE differences of older people based on ethnicity is limited. One which compared the REE of older people including Spanish, Amerindian, and African found no differences in REE (after adjustment for FFM) of older people [82]. However, the relatively small sample size and limited ethnicity involved in the study mean that the results may not be applicable in other study groups. Hence, ethnicity may still be an important factor to be considered in REE and TEE studies as it may be a confounding factor but this requires further elucidation. This is especially true if the participants of the study were of different ethnicity from each other. Unfortunately, none of the eight studies identified gave details on the ethnicity of their study populations and this may confound the conclusion that there was no difference in REE between dementia and control group.

In the above studies, variations also exist in the severity of the disease as well as the presence or absence of weight loss and its severity in the participants. These confounding factors may affect the REE values of participants directly or indirectly through the loss of FFM. Regardless, from the eight studies identified, it is apparent that there is not a significant elevation in the REE of older people living with dementia compared to non-dementia controls. Niskanen et al. [26] and Heshka et al. [83] hypothesized a state of hypermetabolism early in the disease with REE later decreasing to a level similar to the ones found in non-dementia older people as FFM was reduced as weight was lost. This concept of hypermetabolism early in the disease phase was further supported by recent findings from Doorduijn et al. [84] where they showed a higher REE (kcal/kg FFM) but similar energy intake in individuals with mild AD and mild cognitive impairment compared to individuals without. This may explain the early weight loss observed in individuals even before the signs of dementia became apparent [9]. A prospective cohort study, tracking the weight history, REE, and energy intake of the participants, would elucidate whether hypermetabolism may actually be the cause of elevated REE driving early weight loss in individuals before the onset of dementia starts. Although, the existence of various comorbidities are often present in older people, and the possible interplay from other factors, such as low food intake, may create difficulties in singling out hypermetabolism caused by dementia (if, at all) to be a significant cause of weight loss in an individual.

As current available literature suggests that there is no significant difference between REE of older people living with dementia compared to healthy controls, TEE should now be considered. Higher TEE compared to energy intake would lead to weight loss. Hence, TEE measurement will enable direct comparison with energy intake to analyze if weight loss observed in older people living with dementia is truly caused by imbalance between energy intake and TEE.

Measurement of total energy expenditure (TEE)

Studies comparing TEE between older people living with dementia and healthy controls, summarized in Table 4 [30–32, 68], are limited. Traditionally, two methods commonly used to measure TEE are direct calorimetry (using whole-room calorimeter) and doubly-labeled water (DLW). The latter is the preferred method and is often considered the gold standard to measure TEE. Direct calorimetry, on the other hand, requires a whole-room calorimeter with complex engineering which limits its accessibility [85]. Furthermore, whole-room calorimeter requires the participants to remain within the physically confined space for an extended period of time [85] and hence may not be suitable for some participants, especially older people with dementia. Two of the included studies measuring TEE [30, 31] used DLW, while two other studies [42, 68] used a physical activity monitor (Actiheart, CamNtech), which calculated activity energy expenditure (AEE) using branched equations on electrocardiogram (ECG) reading and activity level [86]. Estimated REE (using the Schofield equation [68, 86], diet-induced thermogenesis (assumed to be a constant 10% of TEE)) and AEE were then used to estimate the TEE of the wearer [68, 86]. Actiheart, however, has not been successfully validated against DLW for use on either healthy older people, nor people with dementia.

Another study [32] utilized a modified version of two physical activity questionnaires, namely the Yale Physical Activity Questionnaire and the Baecke Questionnaire, in the derivation of TEE values. Neither of which have been well-validated against DLW for use in older people with dementia [87, 88].

Based on the quality statements developed by Porter et al. [89], the DLW protocol employed by Poehlman et al. [30] and Dvorak and Poehlman [31] were well-designed with both studies stating the DLW dosage as well as their methods for converting carbon dioxide (CO₂) production to energy expenditure. Poehlman et al. [30]’s isotope ratio analysis also met the 10% precision requirements for TEE measurement. However, both studies did not clearly report if the study subjects were fasted for 5–12 hours before administration of DLW.

Factors affecting total energy expenditure in older people with dementia versus non-dementia controls

The current review of the literature has not found sufficient evidence to suggest that TEE in older people living with dementia is higher than in healthy controls. In fact, Poehlman et al. [30] and Kwan et al. [32] observed a lower TEE in older people with dementia compared to healthy controls. Poehlman et al. [30] attributed the differences in TEE to lower REE and when adjusted for different body composition, they found no significant differences between the REE of the dementia and control group. However, Kwan et al. [32] attributed the lower TEE in the dementia group to lower AEE found in the group as caregivers may have restricted the movements of the participants for safety reasons.

An aspect which often needs to be considered in older people living with dementia is AEE as fidgeting and wandering behavior had been reported in people living with dementia [90]. These characteristics may contribute to the increase in AEE and hence increase the overall TEE, a possible contributor to the weight loss experienced in this group.

Prentice et al. [91] commented that TEE was much higher in participants who showed the characteristic wandering and rocking often seen in people with dementia compared to participants who were mostly sedentary. In addition, Moyle et al. [92] also observed this heightened agitation levels among older people with dementia especially between 3 pm and 6:59 pm, which may be a characteristic of ‘sundowning’. Dvorak and Poehlman [31] reported on non-purposeful movements such as fidgeting and shivering that their participants exhibited. However, they found no differences in TEE between the participants with and without dementia [31]. Wang et al. [7] also observed higher incidence of wandering behavior in older people with dementia but found that the group still had lower metabolic equivalent of task (MET) per day compared to non-dementia controls, thereby expending less energy. Evidence seems to be conflicting on whether or not wandering and fidgeting behaviors increase TEE. As weight loss is also often seen in older people with dementia who do not display the characteristics of wandering and fidgeting, the latter may not be a significant contributor towards weight loss in older people with dementia. Noninvasive activity measurement tools such as Actiheart and SenseWear may be useful to integrate into future studies to determine the extent of wandering and fidgeting behaviors. Further validation of these tools in this population is needed.