Background

Improvements in healthcare, along with the development of medical treatments and vaccines, have increased life expectancy worldwide [1]. This has radically changed the global population demographic, with the proportion of older adults increasing, especially in developed countries [2]. Within Europe, 19.2% of the population were aged 65 years or over in 2016, with a projected increase to 29.1% by 2080 [3]. Considerable challenges arise with this increasing ageing population, among them promoting good health and well-being within this group so that they can live independently in the community for as long as possible [4].

One such challenge among older adults living in the community is risk of malnutrition and more specifically undernutrition (hereafter referred to as malnutrition) [5, 6]. Older adults are at increased risk of developing malnutrition due to natural age-related changes [7], namely, unfavourable changes in body composition, increased requirements for protein and certain micronutrients, alterations in appetite and declining sensory function. Left untreated, malnutrition can detrimentally affect cognitive and physical function, both of which can lead to loss of independence, increased risk of disease and poorer health outcomes [8,9,10,11]. Moreover, malnutrition is a complex multifactorial process, with many other components, such as, lifestyle, financial, social, psychological, presence of disease and medication use, known to contribute [12]. Within the published literature, there is little consistency between previously reported factors associated with malnutrition. In developed countries, malnutrition prevalence differs across community and healthcare settings depending on the individual’s characteristics, and the tools used to identify malnutrition. The greatest number of malnourished older adults in the UK is in the community setting (accounting for approximately 5% of the older population) [13, 14]. Community-dwelling older adults are a heterogeneous group who may experience remarkable differences in their ageing trajectory; namely, successful, usual or accelerated rates [15]. Successfully ageing older adults have few health conditions, are independent, rarely use healthcare services and their years of ill health are condensed into the end-of-life. Usually ageing older adults typically maintain their functional ability and independence but have health conditions and require frequent visits to their general practitioner (GP) to maintain their health status. Those experiencing an accelerated rate of ageing are frailer and more dependent than expected for their age, have multiple chronic diseases or experience rapid disease progression, and are frequent users of healthcare services [15].

With the global increase in life expectancy, more attention is being drawn to different rates of ageing. In particular, the concept of successful ageing is now acknowledged to be an important area of research. Nonetheless, whilst there is general agreement on the characteristics typical of a person ageing at a successful rate, to date, there is no consensus on how this concept should be defined. One of the most used definitions for successful ageing is someone who is ‘free of disease and disability, has a high physical and cognitive functioning ability and has an active engagement with life in general’ [16]. Rate of ageing can be influenced both positively and negatively by lifestyle, diet, psychological, psychosocial and disease related factors. Higher rates of physical activity throughout life are strongly linked to successful ageing [17, 18]. Older adults who self-report good health and no pain are more likely to age successfully than those that don’t [19]. Older adults experiencing different ageing trajectories may have different determinants of malnutrition which are specific to their rate of ageing.

Malnutrition in older adults is often under-recognised and poorly managed [20]. This can be attributed to the fact that it is a slow progressing condition and, therefore, its early signs and symptoms are not easily recognised either by affected individuals [21] or healthcare professionals (HCPs) [22]. Additionally, a universal definition and agreed diagnostic criteria have only recently emerged [20, 23, 24]. With the aim of achieving consensus on the definition of malnutrition, the European Society for Clinical Nutrition and Metabolism (ESPEN) stated (in 2015) that the following definition of malnutrition was generally accepted; “a state resulting from lack of uptake or intake of nutrition leading to altered body composition (decreased FFM) and body cell mass leading to diminished physical and mental function and impaired clinical outcome from disease” [25]. Furthermore, a global consensus for the diagnosis of malnutrition has recently (2019) been published based on a two-step approach; screening for risk of malnutrition using a validated tool, followed by assessment of the condition to provide a diagnosis of malnutrition and to grade its severity [26, 27].

Understanding and identifying factors that lead to malnutrition is critical for developing interventions aimed at preventing or delaying disability in older adults. This is particularly important in the community, where although prevalence is low, the greatest number of at-risk individuals reside [14, 28]. Community-dwelling older adults are a heterogeneous group; thus, the factors related to, or determinants of, malnutrition may vary according to individual differences in the rate of ageing. Potential differences in determinants of, and factors related to malnutrition according to differences in ageing rates may contribute to the heterogeneity between currently published studies. The aim of this review, therefore, is to summarise the current evidence relating to the sociodemographic factors associated with, and determinants of, malnutrition and malnutrition risk in community-dwelling older populations and, to explore potential differences according to different rates of ageing [15].

Methods

Search strategy

Two independent systematic searches (Search 1, LAB; Search 2, KL, ML, MGB) of PubMed Central and Embase databases were conducted in April 2019 to identify relevant papers on ageing and poor nutritional status. Duplicates were excluded (LAB and KL), and titles examined to assess suitability for inclusion (LAB and ML). Studies examining the sociodemographic factors associated with, or determinants of, malnutrition were included. The key search terms were as follows: the primary outcome (protein-energy malnutrition, malnutrition, undernutrition, weight loss, nutritional status); the population sub-group (elderly, older adults, ageing, aging); and the exposure (determinants, predictors, risk factors). Figure 1 shows the exact search terms used.

Fig. 1
figure 1

Search terms

Full size image

Inclusion criteria

Studies with populations with mean age > 65 years, majority community-dwelling (> 80%), and conducted in Western populations (specifically European, North American, Canadian, Australian and New-Zealanders) were selected for consideration. Studies containing populations from multiple countries were only included if the majority of the population came from the specified Western countries. As the standardised criteria for diagnosing malnutrition were only published in 2019 [26, 27], papers using any definition of malnutrition arising from use of screening tools, specific BMI cut-offs or weight loss percentages were considered for inclusion. Only papers which were published since 2000, peer-reviewed, available in full-text, written in English, conducted on humans and in which the study authors completed multivariate statistical analysis were considered for inclusion. As the main aim of this review was to assess the sociodemographic factors associated with malnutrition according to a population’s rate of ageing, studies examining a combination of biochemical or nutritional factors in addition to sociodemographic factors were excluded [29,30,31].

Study selection

Abstracts were screened for inclusion by two authors independently (LAB and ML). If a study appeared to meet the inclusion criteria, full text articles were read and analysed for inclusion by two authors working independently (LAB and MGB). Final inclusion was decided by consensus discussion with a senior researcher working on the topic of community malnutrition (PDC).

Data synthesis

Selected full-text articles were read in full and the investigated factors categorised into domains. Factors suggested as being associated with malnutrition or as determinants of malnutrition were categorised under nine known domains: demographic, food intake, oral, lifestyle, social, economic, physical functioning, psychological and disease-related [32]. For the purposes of this review, poverty was included in the social domain and both edentulousness and chewing difficulties included in the food intake domain. Factors reported within each domain are summarised in Table 1. Where possible, study populations were categorised into successful, usual or accelerated rate of ageing groups according to the criteria suggested by Keller et al. (2007), as summarised below [15] (Fig. 2).

  • Successful ageing: predominantly functionally independent (> 60%), not frail (< 40%), low prevalence of polypharmacy (< 40%), and low prevalence of multi-morbidity (< 40%).

  • Usual ageing: predominantly functionally independent (> 60%), not frail (< 40%), a high proportion regularly attending a GP (> 50%), high prevalence of multi-morbidity (> 50%) and polypharmacy (> 50%).

  • Accelerated ageing: predominantly frail (> 60%), functionally dependent (> 60%), users of home-care services (> 40%), and a high proportion was recently hospitalised (> 50%).

Table 1 Reported associated factors, and determinants, of malnutrition in community-dwelling older adults by domain
Full size table
Fig. 2
figure 2

Categorisation of studies included in review by rate of ageing

Full size image

For each of the parameters listed above, any measure or tool or definition used by a particular study was deemed acceptable. In order for study populations to be categorised, information had to be available for at least two of the above criteria. Study populations were placed between two categories if there was insufficient information to differentiate which specific category the population should be placed in.

Results

Search results

The initial database search yielded 21,326 papers once duplicates were deleted. All papers were considered for inclusion; reasons for exclusion are outlined in Fig. 3. The most common reasons for exclusion were studies conducted in non-Western populations, younger populations (mean < 65 years), populations with a specific disease or condition (e.g., Parkinson’s disease), studies whose primary focus was not malnutrition and studies completed in hospital, residential care, or rehabilitation settings. A total of 68 papers met the final inclusion criteria (Fig. 3).

Fig. 3
figure 3

Flow chart of selection criteria for inclusion in review

Full size image

The articles included were heterogeneous in study design (Table 2). Studies were predominantly of cross-sectional (N = 54) or longitudinal design (N = 11). There were two systematic reviews of observational studies and one meta-analysis of longitudinal studies. Sample size of the studies ranged from 49 to 15,669 participants. The majority of included studies were conducted within European countries.

Table 2 Factors associated with, and, determinants of malnutrition
Full size table

Categorisation of studies according to rate of ageing

Nine studies were classified as ageing at a usual rate [35, 36, 39, 49, 63, 66, 70, 84, 99]. Three studies were classified as ageing successfully and five studies were categorised as ageing at an accelerated rate. Six studies were placed between the successful and usual ageing groups [34, 41, 42, 45, 68, 100] and five studies were placed between the usual and accelerated ageing categories [38, 44, 53, 77, 93]. In order to include as many studies as possible in our results, studies classed within the usual to successful ageing category were collapsed into the successful ageing category [21, 85, 87] whilst studies within the usual to accelerated category were collapsed into the accelerated ageing category [40, 46, 94,95,96] (Fig. 2). Forty studies remained uncategorised so were omitted from the synthesis of studies by ageing rate; however, the details of each of these studies are described in Table 2. Primary reasons for not categorising studies included lack of information on presence of chronic diseases, polypharmacy, functionality, frailty or use of social or medical services not being provided or that the study included multiple cohorts (details of all studies included in this review are within Table 2).

Factors associated with, and determinants of, malnutrition

Factors in the demographic and disease-related domains were most-commonly examined (63 and 54 studies respectively), followed by the social (50 studies), psychological and physical functioning domains (46 studies each) (Table 2). Factors under the food intake and lifestyle domains were the least well studied (32 and 20 studies respectively). The factors most-commonly reported to be associated with malnutrition were within the demographic (41 studies), disease-related (34 studies), physical functioning (30 studies) and psychological (30 studies) domains. Domains less commonly reported as associated with malnutrition were the social (27 studies), food intake (23 studies) and lifestyle (7 studies). The evidence for individual factors within each domain is critically considered.

The frequency of factors reported as associated with malnutrition according to the rate of ageing category is presented in Table 3. In this review, demographic factors such as being female (successful, N = 2; usual, N = 1; accelerated, N = 1) and increasing age (successful, N = 2; usual, N = 3; accelerated, N = 1) were commonly reported as associated with malnutrition/malnutrition risk across all ageing rate categories. Other demographic (unmarried status (N = 4) [42, 45, 85, 100] and a low education level (N = 2) [34, 68]) and physical functioning factors were more commonly reported within the successful ageing category compared to the other ageing rate categories. Factors within the food intake and disease-related domains were most-commonly reported in older adults who are ageing at an accelerated rate.

Table 3 Factors associated with malnutrition in community-dwelling older adults stratified by ageing rate
Full size table

This review found that factors reported to be associated with malnutrition from the food intake domain increased in frequency and severity across the three ageing categories (successful, usual, accelerated). Food insecurity was reported as a risk factor in the successfully ageing category [42], choosing foods that were easy to chew was a risk factor in the usual ageing category [39], whilst difficulties eating and eating dependency were associated with malnutrition risk in the accelerated ageing category [77]. Having a poor or reduced appetite is reported as being associated with malnutrition or malnutrition risk across all categories of ageing rate [39, 44, 77, 87].

Within this review, lifestyle factors were rarely reported as being associated with malnutrition or malnutrition risk in any of the ageing categories. Lack of physical activity was reported once in both the successfully [68] and accelerated [46] ageing categories. Alcohol use was reported as being associated with a lower risk of malnutrition once within the usual ageing category [49]. Smoking was reported to be associated with malnutrition in one study from the accelerated ageing category [46].

Cognitive impairment, a factor within the psychological domain was reported as being associated with malnutrition by one study in the successful ageing category [85], whilst dementia was reported as associated with malnutrition risk in both the usual (N = 1) [36] and accelerated (N = 2) [53, 94] ageing categories. Depressive symptoms were reported in the successful ageing (N = 2) [34, 45], usual ageing (N = 3) [35, 36, 49] and accelerated ageing (N = 2) [38, 53] categories.

Indicators of declining mobility (difficulty walking 100 m and difficulty climbing a flight of stairs) were reported in the successful ageing category only (N = 2) [85, 87]. Factors indicative of physical dependency (being unable to go outside) were reported in one study from the accelerated ageing category [46]. Falls were reported to be associated with malnutrition or malnutrition risk in the successful ageing (N = 1) [85] and accelerated ageing (N = 1) [96] categories.

Living with others was associated with reduced risk of developing malnutrition in the successful ageing category (N = 1) [45], whilst living alone was associated with increased risk of malnutrition risk in the usual ageing category (N = 2) [35, 39]. Social support was reported to be associated with malnutrition or malnutrition risk in both the successful (N = 2) [85, 100] and usual (N = 2) [35, 99] ageing categories.

This review found factors from the disease-related domain were commonly reported across all ageing rate categories but increased in severity as the ageing rate progressed into the accelerated ageing category. Recent hospitalisation was reported in the successful (N = 1) [85] and accelerated (N = 2) [94, 96] ageing categories. Factors such as multi-morbidity were more commonly reported in the successful ageing category (N = 2) (N = 0, usual ageing category, N = 1, accelerated ageing category) whilst individual diseases such as cancer and osteoporosis (N = 1) [46] and extended hospital stays (N = 1) [96] were reported in the accelerated ageing category.

Discussion

This review provides a summary of the factors associated with malnutrition and malnutrition risk reported in community-dwelling older adults with an emphasis on differences identified according to rate of ageing [15]. This novel approach has found that as the rate of ageing accelerates, an increasing number of factors are reported within the food intake, social and disease-related domains; and these factors increase in severity in the accelerated ageing category. Within the usual and accelerated ageing categories, dementia is reported to be associated whilst cognitive impairment appears in the successful ageing category. Indicators of declining mobility and function are associated with malnutrition and these indicators increase in severity across the ageing categories. Within the successful ageing category, demographic factors such as low education level and unmarried status appear to be most important. Factors such as hospitalisation and falls appear to be relevant regardless of rate of ageing.

The findings presented in this paper contribute to our understanding of the factors associated with, and determinants of, malnutrition in older adults and may explain differences in factors associated with, and determinants of, malnutrition reported in previously published studies. Standardised criteria for the diagnosis of malnutrition were only published as recently as 2019 [26, 27]. The majority of studies included in this review were published prior to this date; thus, many differing definitions of malnutrition were used. The lack of consistency between studies makes comparisons difficult; however, implementation of these 2019 criteria in future studies should help to reduce the heterogeneity.

Factors associated with, and determinants of, malnutrition

Demographic domain

Numerous cross-sectional studies included in this review reported no association between marital status and malnutrition [34, 37, 46, 47, 50, 51, 62, 70, 88, 93]. Conversely, other studies, including a recent meta-analysis of longitudinal studies, did report a relationship, whereby not being married was associated with an increased risk of developing malnutrition [33, 42, 45, 56, 73, 76, 95, 99]. This may be attributed to the fact that being married is linked to better health behaviours across life, with this effect being more pronounced in men [52]. In this review, unmarried status was frequently reported to be associated with malnutrition or malnutrition risk in the successful ageing category. Most of the evidence in this review suggested that level of education is not associated with malnutrition [32, 33, 35, 36, 38, 42, 45, 46, 49, 50, 54, 55, 61, 62, 66, 67, 70, 71, 73, 76, 85, 87, 88, 93, 99,100,101,102]. However, when stratified by rate of ageing, a low level of education appeared to be more commonly reported as being associated with malnutrition within the successful ageing category. These demographic factors could be playing a key role in the development of malnutrition within the successful ageing group as older adults in this category are not burdened with chronic diseases, mental or physical functional limitations to the same extent as older adults in the other ageing rate categories.

Age and female sex are reported to be associated with malnutrition and malnutrition risk across all ageing rate categories. It has been reported that females have a 45% higher chance of developing malnutrition compared to their male counterparts [72]. This could be due to a multitude of factors including the fact that globally women have longer life expectancies than men [72, 82]. Women are also more likely to experience adverse social and economic circumstances in old age [72, 103,104,105], which are themselves independently associated with increased risk of malnutrition. Within the included studies, many reported an independent association between increasing age and deteriorating nutritional status [34, 42, 43, 53, 57, 62, 63, 76, 79, 81, 83, 88, 106, 107]; conversely, a systematic review concluded there was moderate strength evidence to suggest that older age and malnutrition are not associated [32]. Furthermore, a second systematic review concluded that it is likely that frailty is driving the association seen between malnutrition and advancing age [89]. Factors within the demographic domain are frequently reported to be associated with malnutrition; however, consideration should be given as to whether these are true determinants of the condition or whether the associations seen are false positives due to frequency of assessment.

Food intake domain

Factors affecting food intake, such as the amount of food eaten or the ability to eat/feed oneself, appear to be particularly associated with malnutrition within the accelerated ageing category, compared to the other categories. This may be in line with the fact that this group comprises a sicker, and more diseased population group. The escalation in the severity of these factors across the ageing categories (from food insecurity to factors affecting food choice to having difficulty or being unable to self-feed), highlights that as older adults deteriorate in health and function, they become more vulnerable to developing malnutrition.

In this review, a reduced/poor appetite appears to be associated with malnutrition across all ageing rate categories. Reduced appetite can be a consequence of many factors known or suggested to be associated with, or determinants of, malnutrition, including depression, cognitive decline, chewing or swallowing difficulties and sensory changes [90, 108, 109]. Two systematic reviews included in this review reported that reduced appetite is associated with malnutrition with one of these reviews reporting that poor appetite was the only factor that had strong evidence to support an association with malnutrition [32, 110]. Conversely however, a meta-analysis of longitudinal studies reported no association with incident malnutrition [73]. These differences may be related to the fact most studies included in the systematic reviews and categorised by rate of ageing in this review were cross-sectional in design, whilst the meta-analysis only included longitudinal studies. In addition, variances in the way the question on appetite was asked between studies may have contributed to these differences.

Evidence surrounding the association between dental status and presence of chewing problems and malnutrition is conflicting [39, 44, 55, 61, 63, 79, 93, 110, 111]. Having no/few teeth and difficulties chewing can be detrimental to diet quality and lead to malnutrition as nutrient-dense foods (e.g., meat, fruit and vegetables) may be avoided in favour of softer, higher calorie but less nutrient-dense foods which may be easier to eat [98]. However, difficulties chewing or swallowing may also be a consequence of malnutrition as a decline in physical function is a known outcome of malnutrition which may explain the conflicting findings found amongst the cross-sectional studies included in this review [41, 44, 63, 97].

Lifestyle domain

Lifestyle factors were seldom reported across all categories of ageing rate; thus, the evidence surrounding lifestyle factors, such as alcohol consumption, smoking and low physical activity and malnutrition is weak. Few associations have been reported for physical activity as a protective factor [51, 56, 68] and smoking as increasing risk [33, 46] of malnutrition in cross-sectional studies. One cross-sectional study reported alcohol intake as protective against malnutrition [49]. This study was conducted in The Netherlands which is one of the lowest alcohol-consuming countries in Europe; therefore, this finding may not be applicable in countries with higher consumption rates [112]. All other included studies, including a meta-analysis and two systematic reviews, failed to report associations between alcohol consumption and malnutrition [32, 37, 46,47,48,49, 70, 73, 85, 87, 96, 110, 111, 113]. As reported in a previous systematic review [110], our review reinforces the conclusion that factors within the lifestyle domain do not appear to be determinants of malnutrition in older adults.

Social domain

Factors within the social domain, predominantly factors related to social support, were apparent within the successful and usual ageing categories, where social factors related to use of services were more prevalent amongst the accelerated ageing category, likely reflecting increased dependency among this group and subsequently, a higher need for these services. This finding is supported by two longitudinal studies which reported that meals-on-wheels use, which may be linked to reduced social (and physical) functioning, was associated with increased risk of malnutrition [40, 94]. Amongst fit, community-dwelling older adults, those with the highest levels of social vulnerability (defined using the social vulnerability index) have been reported to be more than twice as likely to die compared to their counterparts who had the lowest levels of social vulnerability [58]. In contrast, a meta-analysis has reported that living alone or receiving social support do not predict incident malnutrition [73]. These differences may be related to study design as our review is predominantly comprised of cross-sectional studies.

Physical functioning domain

Evidence surrounding a relationship between inability or difficulty completing activities of daily living (ADLs) and malnutrition is conflicting [34, 36, 43, 45, 46, 49, 50, 63, 71, 83, 88, 92,93,94,95, 97, 111]. A systematic review has stated there was inconclusive evidence to identify whether there was an association with malnutrition [32]. This conflicts with other studies which suggest that declining health and/or functionality can make cooking, personal transport and grocery shopping difficult; therefore, negatively affecting nutritional status [114]. Further work is required to fully understand this.

Low handgrip strength (HGS) did not appear to be associated with malnutrition across any of the rate of ageing categories. Furthermore, HGS, was reported to have no association with incident malnutrition following a meta-analysis of longitudinal cohorts [73]. Although this may seem surprising as HGS is often used as a marker for functionality and/or frailty, it may be explained by the fact that declines in physical function are a known outcome of malnutrition and, therefore, low HGS is likely a consequence as opposed to a determinant of the condition. As such, low HGS may be a useful indicator of those who are severely malnourished as opposed to those exhibiting early signs or risk of developing malnutrition.

Falls among older adults, can be an indicator of declining cognition or onset of frailty [91, 115] and can result in fractures and hospitalisation, known risk factors for nutritional decline [116, 117]. Increased risk of falling appears to be associated with malnutrition in both the successful ageing and accelerated ageing groups, suggesting a bidirectional relationship between falling and malnutrition, whereby it could be a determinant of malnutrition for an older person ageing at a successful rate, initiating a rapid deterioration in health. Equally, it could be a consequence of malnutrition in an older person ageing at an accelerated rate. Adding weight to this hypothesis, a recent meta-analysis of six longitudinal studies reported no association between falls and incident malnutrition. However, this study reported that difficulty walking 100 m and difficulty climbing a flight of stairs (indicators of mobility) were determinants of incident malnutrition [73]. Indicators of declining mobility associated with malnutrition appear to increase in severity across ageing rate categories. Difficulties walking 100 m or climbing a flight of stairs appeared as associated factors in the successful ageing group whilst being unable to go outside is an associated factor in the accelerated ageing category.

Psychological domain

The prevalence of malnutrition is significantly higher among people with dementia; however, this is more likely to impact on the determinants of malnutrition in long-term care settings where dependency is higher compared to the community setting [118, 119]. Difficulties assessing whether cognitive decline is a determinant of malnutrition are compounded by the under-representation of this cohort of older adults within studies. Cognitive decline appears to be associated with malnutrition in one study in the successful ageing category, while dementia is associated with malnutrition within the usual ageing and accelerated ageing categories. It is likely that this is signifying the progressive decline in health as older adults move from ageing at a successful rate into the other less successful ageing categories.

Disease-related domain

Disease-related factors appear across all ageing categories; however, specific diseases such as cancer and osteoporosis only appear within the accelerated ageing category. Malnutrition is common among older adults with cancer, with the prevalence ranging from 30 to 85% depending on the cancer type [120]. Recent hospitalisation is the factor most likely to impact negatively on an older person’s nutritional status within the disease-related domain [32, 73, 110]. Hospitalisation appears as an important factor within the successful and accelerated ageing categories in this review. However, prolonged hospital stay (> 4 weeks) only appears as a factor within the accelerated ageing category. Similar to falls, hospitalisation is likely to have a bidirectional relationship with malnutrition, being a determinant of the condition for those ageing at a successful rate and a consequence for those within the accelerated ageing category.

Numerous studies have reported associations between poor self-rated (SR) health and malnutrition [32, 37, 42, 47, 81, 84, 88, 94, 106, 110, 121, 122] with SR health being a prevalent factor across all ageing rate categories in this review. This contrasts with a recent meta-analysis (of longitudinal studies) that reported no association between SR health and incident malnutrition [73]. Contradictory results have been reported surrounding the relationship between polypharmacy and the number of chronic diseases with malnutrition and malnutrition risk. Two systematic reviews have concluded that the evidence for polypharmacy as a factor associated with malnutrition was inconclusive and that there was moderate evidence to support no association with number of comorbidities [32, 110]. The conflicting results reported for these factors is likely due to the differing numbers of medications/diseases being used to define polypharmacy/multimorbidity between studies.

Strengths and limitations

This review used a novel approach of categorising community-dwelling older adults according to their rate of ageing (successful, usual or accelerated) and assessed whether differences occurred in the factors associated with malnutrition for each category. To the best of our knowledge, no other study has taken this approach previously. This approach may contribute to reducing the heterogeneity in factors reported to be associated with malnutrition among older adults in the community setting.

There are a number of limitations associated with the published literature on the determinants of malnutrition in older adults. Whilst 68 studies were initially identified as relevant for inclusion in this review, 40 could not be categorised according to rate of ageing due to the lack of detailed information on the characteristics of the study population provided within the published manuscripts. These studies were, therefore, omitted from the synthesis of factors associated with, and determinants of, malnutrition by rate of ageing. Had these manuscripts contained sufficient information to permit categorisation by rate of ageing, our results would have been strengthened or potentially different. Where possible, the current review sub-categorised the study populations from the included studies into successful, usual or accelerated ageing. It is likely that there was heterogeneity between the participants included in individual studies; however, group means were used to categorise the populations from individual papers as a whole into rate of ageing categories. Furthermore, there was variation in the parameters used in different studies, for example, number of medications to define polypharmacy and method of measuring functional independence.

The majority of studies included in this review used convenience samples, often with small sample sizes. This limits the use of the results as they cannot be extrapolated to represent the general population of community-dwelling older adults. A number of studies only investigated factors from one or two domains, thus, failing to acknowledge the multifactorial aetiology of malnutrition. Furthermore, a factor or determinant could not be identified as associated with malnutrition if it had not been included in the original study. The majority of current published literature is cross-sectional in design. Studies of longitudinal design are superior to definitively determine the factors which predict malnutrition as cross-sectional studies cannot distinguish between the causes and consequences of malnutrition. This review included only studies published in the English language and from the year 2000 onwards. The timeframe chosen was to ensure a more standardised approach to the identification of malnutrition and malnutrition risk and allowed for the identification of potential factors and determinants of malnutrition relevant to the health of older adults in the past 20 years. Nonetheless, these factors may have introduced selection bias into our results.

Conclusions

Numerous changes occur with ageing, increasing the vulnerability of older adults to developing malnutrition. Older adults are a heterogeneous group; thus, assessment of individuals’ rate of ageing could aid in identifying specific determinants for different cohorts of community-dwelling older adults. In the future, categorising community-dwelling older adults according to their rate of ageing could also be incorporated into malnutrition screening methods; this would allow for a more personalised approach to identifying malnutrition in older adults as different domains and different individual factors appear to be important depending on the ageing category. Further longitudinal studies and meta-analyses, segregating elderly by ageing rate, are warranted to clearly distinguish which factors are true determinants of malnutrition and not simply the consequences of the condition.