Remaining Socially Connected at 100 and Beyond Reduces Impact of Loneliness on Nutritional Status

Abstract Background Understanding factors influencing centenarians’ nutritional status can offer insight into effective nutrition interventions to improve quality of life among this population. Objective This cross-sectional study was conducted to evaluate the role of social support and loneliness on nutritional status among Oklahoma centenarians (N = 151). Methods Nutritional status was assessed with the Short Form Mini Nutrition Assessment (MNA-SF). Perceived social support was assessed with the 24-item Social Provisions Scale. Loneliness was examined with the 10-item UCLA loneliness scale. Results Ordinal logistic regression revealed that those who lacked social support were at increased risk of poor nutritional status (OR = 2.28, p < .05). Further, the model revealed that centenarians who reported lack of social support and loneliness had almost 2.8 times higher likelihood of being at risk of poor nutritional status compared to their socially connected counterparts (p < .01). Conclusions and implications Findings have implications for geriatric dietitians, social workers, and clinical counselors seeking to implement services and programs aimed at helping long-lived adults feel socially connected and maintain proper nutritional well-being.


Introduction
The oldest old, people 85 years of age and older, are among the fastest growing segment of the population. In addition, the number of centenarians is also increasing. There were about 3,000 centenarians in the United States in 1950. By 2050, the number of persons living to age 100 worldwide is projected to reach more than 3.6 million. 1 Good nutrition is important for promoting quality of life in old and very old age. Age is an identified risk factor for poor nutritional status due to physiological changes such as sensory loss and decreased mobility, which are necessary for shopping, cooking, and eating. 2 This can contribute to increased malnourishment that not only affects an older adult's physical health, but their social and mental well-being. 3 The oldest old who commonly outlive family, friends, and familiar supports, are often at-risk for living the remaining years of life in relative solitude, therefore resulting in an increased likelihood of poor nutritional status. 4,5 However, Buettner reported than many people living to 100 years or older tend to eat with family or with groups of friends. 6 This phenomenon may be in accordance with the Socioemotional Selectivity Theory (SST), which posits that the importance of an individual's goals is reflective of their stage of life and changes in accordance with age. 7,8 As centenarians near the end of their lives, maintaining nutritional status may lose importance in comparison with other goals and desires such as eating for pleasure or maintaining social connectedness. If maintaining nutritional status becomes a lower priority, the risk of poor nutrition increases.
Loneliness has been identified as a risk factor for poor nutritional status among older adults. [9][10][11][12] Loneliness can affect older adults' appetite resulting in decreased amount and variety of foods consumed, fewer meals consumed, and increased use of convenience foods. 9,10 For many of those impacted by loneliness, declines in health related quality of life, including poor nutritional status, occur in near-isolation of others and remain unobserved or undetected. 3 Social support also shares an association with nutritional well-being in old age. Older adults with higher social support have been reported to consume a more balanced diet that includes fruits and vegetables and contributes to better health status; [13][14][15][16] whereas, older adults with lower social support have been reported to be at increased nutritional risk. 11,17 Family, friends and neighbors can support older adults' nutritional wellbeing through a variety of instrumental activities such as assisting with grocery shopping or preparing meals. 14 There are two main types of social support: emotional, which refers to the expression of positive feelings such as love and trust, and instrumental, which refers to the provision of practical aid. 16 Although there has been research investigating centenarian dietary patterns and nutritional status, 18,19 research regarding factors influencing centenarians' nutritional status is limited. Ji et al. did report age, poor cognitive function, gastrointestinal disease, poor self-rated health, and lower serum albumin levels were associated with poor nutritional status among the oldest-old. 2 However, no research has investigated the role of social support and loneliness with centenarians' nutritional status. Understanding factors influencing centenarians' nutritional status can offer insight into effective nutrition interventions to improve quality of life among this population. This cross-sectional study was conducted to evaluate the role of social support and loneliness on nutritional status among Oklahoma centenarians.

Sample recruitment
Cross-sectional data was collected from community-dwelling centenarians (Age: M ¼ 101; SD ¼ 1.74) residing in Oklahoma. Participants were purposefully sampled through the Oklahoma Centenarian Club, the Oklahoma State University Family and Consumer Sciences Cooperative Extension network, senior nutrition/activity center sites, and assisted-living/long-term care centers. A trained member of the research team employed the Short-Portable Mental Questionnaire (SPMSQ) to screen for cognitive capacity to consent. 20 Cognitive screening of centenarian research participants in social behavioral investigations is a recommended practice for purposes of protecting cognitively frail individuals from risk of confusion, mental fatigue, agitation, or stress that may arise during semi-structured interview procedures. 21 The SPMSQ is a brief 10-item interview that tests short-and longterm recall of information. 20 A normative score of 0-2 errors on the SPMSQ is indicative of no cognitive deficit. A total of 3-4 errors reflects mild cognitive limitation. In addition, a score of 5-7 errors represents moderate cognitive problems, whereas 8 or more questions answered incorrectly is suggestive of severe cognitive impairment. The SPMSQ can also be adjusted by education. Persons with a grade school education or less are allowed one additional error. For purposes of this study, the recommended standard cutoff score of 4 or less errors was used. 22 Forty-five centenarians or 29.2% of recruited participants did not meet the cutoff criteria and were therefore not included in the final study sample. The remaining 154 participants who met the pre-established cutoff criteria were asked to read and signed a university IRB approved informed consent document. Afterwards each participant completed a one-to-one semi-structured interview with a trained member of the research team. This semi-structured interview included standardized measurement assessments commonly used within gerontology-geriatric research including the Geriatric Depression Scale-10 (GDS-10). 23 A score of 3 or more out of 10 on the GDS-10 is used as a cutoff to indicate depression. 23

Nutritional status
The Short Form Mini Nutrition Assessment (MNA-SF) was used as the primary outcome measure of nutritional status. 24 The MNA-SF was adapted as a self-report assessment index of food intake over the past three months, weight loss over the last three months, mobility, psychological stress over the past three months, self-reported problems with dementia/ depression, and Body Mass Index based on self-reported height and weight. A high total score is indicative of normal nutritional status; whereas a low score reflects malnourishment. MNA-SF total score ranges from 0 to 14. Due to the sample size we converted SNA-SF scores into a binary variable (normal nutritional status, MNA-SF scores 12-14 versus at risk of poor nutritional status, MNA-SF scores < 12). Per guidelines in the literature, those with scores from 0 to 11 were considered at risk of low nutritional status. 25 Previous literature has confirmed the validity of the MNA-SF among the older adult population in numerous living situations, including both inpatient and community dwelling. 24,26 Social support The Social Provision Scale (SPS) was used as the primary measure of social support. 27 The SPS is a 24-item assessment of perceived social provisions. The SPS incorporates six sub-scale measures of social support including guidance (i.e. There is someone I could talk to about important decisions in my life), reassurance of worth (i.e. There are people who admire my talents and abilities), social integration (i.e. There are people who enjoy the same activities I do), attachment (i.e. I feel a strong emotional bond with at least one other person), nurturance (i.e. There are people who depend on me for help). Participants were asked to rate their level of agreement on a four-point Likert scale (1 ¼ Strongly disagree; 4 ¼ Strongly agree). A high score indicates high social support provisions; whereas a low score indicates low social support provisions. Social support score was calculated by averaging the responses to the 24 items. Original reliability of the SPS has been reported as being high (a ¼ 0.92). 27 Alpha reliability of the SPS in the current sample was 0.76.

Loneliness
A 10-item version of the UCLA Loneliness Scale-Version 3 was used to evaluate loneliness. 28 Sample items included "How often do you feel you lack companionship?" and "How often do you feel left out?" Participants were asked to respond to each statement using a four-point Likert scale (1 ¼ Never; 4 ¼ Always). A high score reflects frequent feelings of loneliness, whereas a low score reflects low occurrence of feeling lonely. Russell reported a high Cronbach alpha reliability of 0.94. 28 Loneliness score was calculated by averaging the responses to the 10 items. Alpha reliability of the 10-item version used within this study was 0.67.

Data analysis
The outcome of interest in the current study was nutritional status, assessed by the MNA-SF. In order to examine the association between MNA-SF as a binary variable, loneliness, and social support, multiple logistic regression analyses were conducted. A multiplicative term between social support and loneliness was introduced in subsequent models, along with relevant covariates including age, sex, race, marital status, depression (GDS-10), 23 and co-morbidity. Odds ratios (OR) and 95% confidence intervals (CI) were calculated for each variable in the model. The main effect model (model without a multiplicative term) was compared to the interaction model using the nested likelihood ratio test. A significant effect (p < .05) indicates the interaction model is superior to the main effect model. All data analyses were conducted using Stata 14.2.
Some supplementary analyses were conducted in order to test the robustness of the current model. In line with the literature and meta-analyses, 29 both social support and loneliness were tested in multiple specifications, including the sum of all scores, the average (sum of scores/number of valid answers), categories (low, medium, high), and binary variable (bottom 25% vs. everyone else). Other covariates, such as number of children, religious participation, living arrangement (i.e. living alone), perceived economic security, and income were considered; however, they were not included in the final analyses since they were not significant, and the substantive conclusions were unchanged. Table 1 presents the descriptive statistics of the core study variables. A total of 151 centenarians were included in the data analysis after the exclusion of 3 participants due to incomplete data for nutritional status and social support. In regard to living arrangements, approximately 45% of participants were community dwelling, 36% resided in assisted living, and 19% resided in nursing homes. Descriptive statistics showed that about 45% of participants were categorized as at risk for poor nutritional status, indicating that poor nutritional status is a significant concern among centenarians. On average, participants reported a high amount of support from various sources (M ¼ 3.33, SD ¼ 0.30) albeit a higher level of loneliness (M ¼ 3.34, SD ¼ 0.46). Demographically, the average age of participants was 101, 71% were women, 88% were white, and 64% completed college education. Approximately 85% of participants were widowed. On average, participants reported low levels of depressive symptoms (M ¼ 1.36, SD ¼ 1.60) and about one in three centenarians reported comorbid conditions. Table 2 represents the results from logistic regression. The findings revealed that a one unit increase in loneliness did not have a main effect on nutritional status (OR ¼ 1.08, p ¼ 0.76, 95% CI 0.49-2.37), adjusting for other relevant covariates. Social support showed a direct effect on nutritional status (OR ¼ 0.94, p ¼ 0.03, 95% CI 0.88-0.98). In particular, a one unit increase in social support was associated with a 6% lower likelihood of being at risk for poor nutritional status. These findings indicate that social support contributes to centenarians' nutritional status even after controlling for covariates. The subsequent multiplicative model revealed that there was a significant interaction between loneliness and social support (OR ¼ 2.33, p ¼ 0.03). In addition, for the ease of interpretation, the top and bottom quartiles for social support and loneliness were compared. Predicted probabilities showed that those who reported both loneliness (top 25th percentile) and  lack of social support (bottom 25th percentile) were at an 18% higher likelihood of being at risk for poor nutritional status (95% CI 0.11-0.29), compared to 9% (95% CI 0.04-0.15) of those who reported high social support (top 25th percentile) and lower feelings of loneliness (bottom 25th percentile). Centenarians who reported a combination of loneliness and lack of social support were at the highest risk of poor nutritional status. Other interaction terms (e.g. social support and depressive symptoms, loneliness and depressive symptoms) were not significant and therefore eliminated from the final analyses.

Discussion
The results of this study revealed that loneliness did not independently increase the likelihood of being at risk for poor nutritional status among centenarians. This finding contradicts previous research which has identified loneliness as a risk factor for poor nutritional status among older adults. [10][11][12] This may be due to the particular sample of centenarians used for this study, as participants were recruited from various clubs, senior nutrition/activity center sites, and assisted-living/long-term care centers, indicating that this sample of centenarians may have been more socially connected than most and therefore less likely to be living or eating alone. Loneliness has been reported to affect eating choices, one of the primary determinants of nutritional status among older adults. 30, 31 Whitelock & Enaff reported that older adults who admit to feeling lonely have decreased motivation to cook meals or even eat at all, as many prefer not to grocery shop and cook for just one person. 31 Other research has similarly indicated that loneliness can affect older adults' appetite, resulting in decreased amount and variety of foods consumed, fewer number of meals consumed, and increased use of convenience foods, all of which can negatively impact nutritional status. 9,10 Interestingly, the current study results do not align with any of these previous findings. However, it is important to note that the previous studies were not conducted with centenarians and many did not evaluate levels of social support. Thus, among centenarians, loneliness by itself may not be as strong of a predictor of poor nutritional status. Additionally, in a previous study by the current research team, with 171 rural older adults ranging in age form 65-101 years, loneliness was not significantly associated with nutritional status; however, loneliness was significantly associated with depressive effect, which had a direct negative association with nutritional status. 32 Results from this study further revealed that centenarians who reported lack of social support had a significantly higher likelihood of being at risk of poor nutritional status. This is consistent with previous research reporting that older adults with higher social support have increased diet quality, diet variety, and health status; [13][14][15][16] whereas, older adults with lower social support tend to be at increased nutritional risk. 11,17 Other researchers have similarly acknowledged that very old adults who are socially supported financially, emotionally, or physically also experience positive health effects, 32 including prolonged survivorship. 33 Martin et al. reported approximately 90% of centenarians in reported centenarian studies needed help with transportation and another 30% suffered from appetite loss, digestive issues, or chewing and swallowing problems. Meanwhile many more had difficulties performing functional activities, including cooking, cleaning, and money management, illustrating that a lack of social support can be truly detrimental to this population. 33 Furthermore, the results of this study revealed that centenarians who reported both a lack of support and loneliness were almost 2.8 times more likely to be at risk for poor nutritional status. Several studies have reported associations between social support and loneliness among older adults. Winningham and Pike reported that compared to older adults who participated in a social support intervention, those who did not participate reported a decrease in perceived social support and an increase in loneliness. 34 Both receiving and providing social support has been reported to affect loneliness among older adults. Chalise, Kai, and Saito reported older adults who either received or provided social support had significantly lower loneliness. 35 Similarly, Rodriguez, Gierveld, and Buz reported that receiving emotional support and receiving or providing instrumental support were protective against loneliness among older adults. 36 In addition, the pathway of social support may vary based on the type of loneliness affect. Schnittger, Wherton, Lawlor, and Lawlor reported social support influenced emotional loneliness indirectly through other factors, such as depression, whereas social support had a direct effect on social loneliness. 37 Thus, social support can have an additional impact on nutritional status through its impact on loneliness. Social support is a determinant of isolation and loneliness, 38 and in centenarians in particular, perceived social support has been reported to lower reported loneliness. 39 In a study evaluating an older adult meal sharing program, 53% of older adults who participated in meal sharing lived alone, but only 2% reported being very lonely; whereas among older adults who did not participate in meal sharing, 16% reported being very lonely, even though only 34% lived alone. 40 This suggests increased social support may decrease loneliness and subsequently increase nutritional status.
This study is not without limitations. First, this study involved a crosssectional study conducted with Oklahoma centenarians and thus the results cannot be extrapolated to all centenarians as a whole. Second, the cross-sectional design of the study is not appropriate for exploring causal relationship between the variables. Results only reflect associations between loneliness, support, and nutritional status and not causation. Third, results are based on standardized self-reported assessments. Inclusion of qualitative methods such as recording and journaling of daily dietary and food preparation practices, as well as social engagement experiences of centenarians could have provided expanded insight into other underlying conditions that might further explain nutritional health among long-lived adults. In addition, differences in living arrangements have the potential to impact loneliness, social support, and nutritional status. While analyses considering living arrangements as a covariate did not show any significant changes to outcome variables, there is a possibility that this may be due to small sample size and not lack of relationship.

Conclusion
Despite the limitations, the findings from this study highlight the importance of social support in moderating the association between loneliness and nutritional status among Oklahoma centenarians. The findings suggest interventions aimed at improving centenarians' nutritional status should include efforts to improve social support and offer opportunities to feel socially connected, especially if there are indications of loneliness.

Research implications
Future research into the interrelationship between social support and loneliness on nutritional status among centenarians is needed. In regard to social support, additional research is needed investigating if the effect of social support differs based on if it is received or provided and whether the type of social support is instrumental or emotional in nature. As aforementioned, living arrangements may play a larger role in loneliness, social support, and nutritional status than the results of the current study suggest. Therefore, further research is warranted exploring availability, distribution, and quality of senior meal services among old-old adults who may reside alone in private homes, as well as those residing in senior housing sites, assisted-living facilities, and skilled nursing facilities.

Takeaway points
While centenarians who reported lack of social support were at a significantly higher likelihood of being at risk of poor nutritional status, loneliness did not independently increase this likelihood. However, centenarians who reported both a lack of support and loneliness had almost 2.8 times higher likelihood of being at risk of poor nutritional status. As social support was found to significantly impact the nutritional status of this population, providing centenarians with more opportunities for social interaction may have a positive impact on their nutritional status. Organizing congregated meals may decrease loneliness, increase social support, and provide easier access to food, all of which may result in improved nutritional status among centenarians.

Funding
The author(s) reported there is no funding associated with the work featured in this article.