- Split View
-
Views
-
Cite
Cite
Kali S. Thomas, Ucheoma Akobundu, David Dosa, More Than A Meal? A Randomized Control Trial Comparing the Effects of Home-Delivered Meals Programs on Participants’ Feelings of Loneliness, The Journals of Gerontology: Series B, Volume 71, Issue 6, November 2016, Pages 1049–1058, https://doi.org/10.1093/geronb/gbv111
- Share Icon Share
Abstract
Nutrition service providers are seeking alternative delivery models to control costs and meet the growing need for home-delivered meals. The objective of this study was to evaluate the extent to which the home-delivered meals program, and the type of delivery model, reduces homebound older adults’ feelings of loneliness.
This project utilizes data from a three-arm, fixed randomized control study conducted with 626 seniors on waiting lists at eight Meals on Wheels programs across the United States. Seniors were randomly assigned to either (i) receive daily meal delivery; (ii) receive once-weekly meal delivery; or (iii) remain on the waiting list. Participants were surveyed at baseline and again at 15 weeks. Analysis of covariance was used to test for differences in loneliness between groups, over time and logistic regression was used to assess differences in self-rated improvement in loneliness.
Participants receiving meals had lower adjusted loneliness scores at follow-up compared with the control group. Individuals who received daily-delivered meals were more likely to self-report that home-delivered meals improved their loneliness than the group receiving once-weekly delivered meals.
This article includes important implications for organizations that provide home-delivered meals in terms of cost, delivery modality, and potential recipient benefits.
The Older Americans Act of 1965 (OAA; Pub. L. 89-73, 79 Stat. 218, July 14, 1965) was among the first U.S. federal initiatives aimed at providing comprehensive services to help older adults age in place in their homes and communities. The home-delivered meals program offered through Title III-C2 of the Older American’s Act (OAA) is the largest program to be funded by the act. In 2012, the home-delivered meals program served more than 135 million meals to more than 841,000 participants in communities all across the United States ( Administration on Aging, 2015 ). The purpose of this OAA Nutrition Program is to “(i) reduce hunger and food insecurity among older individuals, (ii) promote socialization of older individuals, (iii) promote the health and well-being of older individuals, and (iv) delay adverse health conditions for older individuals.” The nation’s home-delivered meals program, more commonly known as “Meals on Wheels,” is funded (in part) by federal dollars from the OAA, state and local governments, private foundations, direct payment for services, fundraising, program participants, voluntary contributions (of time and/or money), and a variety of other sources (Administration for Community Living, 2015).
Meals on Wheels has become a significant part of national service strategies intended to support older adults in their own homes. Programs are generally popular with consumers and seen as beneficial in helping them to meet their basic food needs and remain in their homes ( Frongillo, Isaacman, Horan, Wethington, & Pillemer, 2010 ; Moore, Marceaux, Friedman, & Crixell, 2015 ; Sahyoun & Vaudin, 2014 ). Previous research has demonstrated a relationship between state spending on home-delivered meals and the ability of states to keep older adults with low-care needs out of institutions such as nursing homes ( Thomas & Mor, 2012 ). Research has also suggested that states that have increased their capacity in providing home-delivered meals also have recognized increased Medicaid savings by decreasing the proportion of low-care nursing home residents dually eligible for Medicaid and Medicare ( Thomas & Mor, 2013 ). Beyond providing savings to states, home-delivered meals are believed to improve the quality of life of older adults: the meals may help increase older adults’ independence, encourage autonomy, and thereby improve recipients’ quality of life. In addition to the nutrition received from this program, the socialization and “safety check” that is provided by the meal-delivery person (either a staff member or volunteer) is believed to be a contributor to these beneficial effects found in the literature ( Thomas & Mor, 2012 ).
Federal, state, and local funding cuts, increased transportation and food costs, and the lingering effects of the economic downturn have had significant impacts on Meals on Wheels programs ( Kamp, Wellman, & Russell, 2010 ). Over the past several years, these compounding factors have resulted in hundreds of thousands of fewer seniors served and an increase in the waiting lists for these programs ( Meals on Wheels America, 2013 ). Because of the challenge of meeting the increasing demand and need for home-delivered meals, decision makers at all levels are seeking lower cost solutions to serve homebound seniors. Once-weekly delivery of frozen meals has emerged as one such solution to reduce cost yet still meet the nutritional needs of at-risk older adults. In this model, participants are provided the full week’s meals in one bulk delivery. Proponents of the traditional, daily-delivery model believe that over the long term, these lower cost solutions—drop-shipped meals, less frequent meal deliveries with multiple meals, and limited personal contact—will have a negative impact on the health, independence, and well-being of homebound seniors receiving services. The objective of this study was to evaluate the extent to which home-delivered meals, and the type of delivery model, reduces homebound older adults’ feelings of loneliness.
Theoretical Framework
Social scientists have long believed that social connection is an inherent need that all human beings have, regardless of age, and was popularized by Bowlby’s Attachment Theory ( Bretherton, 1992 ). According to Weiss’ later version of the Attachment Theory, loneliness is posited to occur in the absence of these meaningful social connections ( Weiss, 1973 ). Important to his theory is that loneliness is not a reflection of distortions in social perception or unrealistically high need for companionship; rather, the result of ordinary human needs for connection to others not being met. According to Weiss, two types of loneliness exist: social and emotional loneliness ( Weiss, 1973 ). In the former, loneliness is caused by a lack of social integration, or social isolation. In the latter, loneliness is caused by the absence of a reliable attachment figure, such as a spouse. The difference between the two is most clearly understood by their remedies: absence or loss of an attachment figure can only be substituted by another close and intimate bond while social loneliness can best be resolved by acquiring new contacts and the sense of belonging to a group or social network.
We posit that individuals on waiting lists for home-delivered meals, characterized as being homebound with limited social support, are likely to experience social loneliness. Therefore, by accommodating new social contacts and “group membership” in the form of meal deliveries, (H1) we hypothesize that individuals receiving home-delivered meals will experience lower rates of loneliness as compared with a similar group that does not receive home-delivered meals, and thereby does not receive the additional socialization. Furthermore, (H2) we hypothesize that individuals who receive daily-delivered meals will be more likely to exhibit lower rates of loneliness and self-reported improvements in loneliness compared with individuals who receive once-weekly deliveries of meals given the frequency of their contact with another individual. In order to test these hypotheses, we analyzed data from a 15-week study comparing the benefits of a traditional daily contact Meals on Wheels program with those from less frequent weekly deliveries of frozen meals.
Materials and Methods
Participants
Participants’ data come from a three-arm, parallel, fixed, randomized controlled trial conducted by Meals on Wheels America in the winter of 2013 and spring of 2014 and designed (i) to compare receiving home-delivered meals (regardless of delivery method) with not receiving meals and (ii) to compare receiving daily-delivered meals to receiving frozen, once-weekly delivered meals. Participants were selected from a convenience sample of waiting lists at eight sites across the United States. The sites were selected based on survey responses to a Meals On Wheels America survey assessing the effects of the federal budget sequester on their programs and conducted in September to October of 2013. Specifically, the eight sites were chosen because they each reported having average waitlist times of 3 or more months. Their waiting lists at the time of survey was anywhere from 101 to 569 individuals and the program sizes varied and served anywhere from 230 clients to 6,000 annually. Three sites were located in Texas and the rest were in Florida, Georgia, North Carolina, New Jersey, and Rhode Island.
The study was designed to enroll 620 older adults across all of the sites. The target number of participants was chosen based on the funding available in the grant to pay for the intervention groups’ meals and to provide compensation to programs for conducting the assessments. Each site agreed to target a specific number of participants. This target number was discussed and agreed upon between Meals on Wheels America and the participating site and was based on the site’s willingness to take on the workload associated with this project and the ability to add the determined number of clients to their service. In total, 626 older adults from these sites agreed to participate in the study.
Protocol
Participants were surveyed by local Meals on Wheels staff, and in one site, by Meals on Wheels trained volunteers. A survey guide was developed to collect information about participants’ demographic information and baseline social support, mental health, self-rated health, and loneliness. The survey guide was pilot tested with clients at the local Meals on Wheels affiliate in Rhode Island. Feedback from these sessions was utilized to make final changes to the data collection tools. The baseline survey was conducted in person, in the participants’ homes following seniors’ written consent to participate. After the initial survey, sites randomized participants by alphabetizing participants last names and sequentially assigning them to one of three groups: (a) daily, traditional meal delivery ( n = 214), (b) frozen, once-weekly meal delivery ( n = 202), and (c) a control group who were to remain on the waiting list until service became available ( n = 210). Participants in the frozen meal group received once-weekly deliveries by Meals on Wheels staff/volunteers of 5 days of frozen meals and participants in the daily delivery group received daily delivery by Meals on Wheels staff/volunteers of hot/chilled meals during weekdays. Fifteen weeks after receiving the first meal—or for the control group, 15 weeks after the initial survey—participants were called and a follow-up telephone survey was scheduled. The follow-up telephone survey again queried participants about their social support, mental health, self-rated health, and loneliness. Among participants who received meals, they were also queried about the various effects that home-delivered meals have had on their daily lives.
After randomization, 18 participants (9 receiving daily and 9 receiving frozen, weekly) no longer wanted to receive meals, 17 died, 24 moved (either to a nursing home or to a different location), 11 were in the hospital during the follow-up period, 15 refused to participate in the follow-up telephone survey, and 45 could not be reached for follow-up. A final sample of 154 remained in the control group, 174 in the daily-delivery group, and 131 in the frozen, once-weekly delivery group leaving an analytic sample of 459. There were no statistically significant differences in rates of attrition by group, by site.
Outcomes
Loneliness
Loneliness was specified as a primary outcome for this pilot randomized control trial (RCT). The three-item University of California, Los Angeles (UCLA) Scale ( Hughes, Waite, Hawkley, & Cacioppo, 2004 ) was used to measure overall loneliness for individuals at baseline and follow-up. This scale is a subset of the items used in the Revised UCLA Loneliness Scale (D. Russell, Peplau, & Cutrona, 1980 ). Previous work conducted to develop the three-item scale suggested the alpha coefficient of reliability is .72 ( Hughes et al., 2004 ). The three questions that made up the scale included, “How often do you lack companionship? How often do you feel left out? How often do you feel isolated from others?” The response categories were coded 0 ( never ) 1 ( rarely ), 2 ( sometimes ), and 3 ( often ). Each person’s responses to the questions are summed (on a range from 0 to 9), with higher scores indicating greater loneliness. This scale has been validated and used in previous studies examining loneliness ( Cattan, White, Bond, & Learmouth, 2005 ). Although it is not a direct measure of social loneliness, it has been shown to be significantly associated with objective measures of social isolation among older adults (i.e., marital status, living arrangements, volunteering, providing help to others, and neighborhood safety; ( Hughes et al., 2004 ).
We also examined responses to the question, “Do services received from the home-delivered meals program help you feel less lonely?” that was asked of participants who received daily-delivered and weekly-delivered meals. This question was utilized in addition to the UCLA loneliness score because it is a self-reported measure of improvement in loneliness and reflects participants’ beliefs about the influence of the home-delivered meals programs on their lives.
Control Variables
We controlled for a number of baseline characteristics that the literature has suggested is related to loneliness. Specifically, we included measures of demographic characteristics, socioeconomic status, self-rated health, group membership, social support, and mental health.
Race and education have frequently been shown to be related to loneliness ( Barg et al., 2006 ; Perissinotto, Stijacic Cenzer, & Covinsky, 2012 ; Savikko, Routasalo, Tilvis, Strandberg, & Pitkälä, 2005 ) and therefore, we include dichotomous variables for education (i.e., education greater than a high-school degree = yes/no). To control for differences in race, we created dummy variables to indicate whether the participant was White (yes/no) or Black (yes/no) with all other races being the reference group. We also controlled for age given its positive relationship to loneliness ( Perissinotto et al., 2012 ; Steptoe, Shankar, Demakakos, & Wardle, 2013) and marital status (married = yes/no) because of its documented inverse relationship to loneliness ( Fokkema, De Jong Gierveld, & Dykstra, 2012 ; Theeke, 2009 ).
Two measures of socioeconomic status were included given the inverse relationship found between wealth/income and loneliness in previous studies ( Cohen-Mansfield & Parpura-Gill, 2007 ; Savikko et al., 2005 ; Steptoe et al., 2013). Specifically, we included an indicator if participants were enrolled in Medicaid or reported there are “times when you don’t have enough money to buy the food you need?”
Self-rated health was included as a control variable given its inverse relationship to loneliness in previous studies ( Theeke, 2009 ). The variable consists of participants’ responses to the question “Would you say that in general your health is Excellent, Very Good, Good, Fair Poor,” in which we created dummy variables with the reference group being individuals who responded either “Excellent” or “Very Good.” To control for whether or not participants needed assistance with personal care needs, we included participants’ responses to the question “Because of a physical, mental, or emotional condition, do you sometimes need the help of another person with personal care needs, such as eating, bathing, dressing, or getting around inside the home?” Additionally, previous work has suggested that voluntary group and religious/church membership is protective against loneliness ( Cattan et al., 2005 ; Johnson & Mullins, 1989 ); therefore, we included a measure of group membership collected at baseline. The item, “Do you participate in any groups, such as a senior center, social or work group, religious-connected group, self-help group, or charity, public service, or community group?” was used to assess this construct.
Social support has been found to have a direct link to loneliness in that older adults with social support report less loneliness ( Schnittger, Wherton, Prendergast, & Lawlor, 2012 ) and therefore, we included the item, “Looking back over the last few months, how often do you have contact with friends or family?” Two dummy variables were created to indicate respondents had (i) contact once or twice a month or less and (ii) weekly contact with friends or family with the reference group being daily contact. Additionally, we included the dichotomous response to the item, “Is there a family member, friend, or neighbor that you feel you can call on for help if you need it?” Finally, we included a measure of whether or not the participants lived alone given its documented positive relationship to loneliness ( Perissinotto et al., 2012 ; Savikko et al., 2005 ; Theeke, 2009 ).
Previous work has confirmed a relationship between mental health, particularly depression and anxiety, and loneliness ( Bekhet & Zauszniewski, 2012 ). In this study, depression was measured using the Patient Health Questionnaire 2-item (PHQ-2) depression screener that has been validated and used in other studies (K. Kroenke, Spitzer, & Williams, 2003 ; Löwe, Kroenke, & Gräfe, 2005 ). A score greater than or equal to three was used to signal individuals that screened positive for depression. To evaluate anxiety, we used the Generalized Anxiety Disorder 2-item measure (GAD-2), developed as a screening test to detect anxiety disorders (K. Kroenke, Spitzer, Williams, Monahan, & Löwe, 2007 ). A score of greater than or equal to three denoted a screening cutpoint for clinically significant anxiety (K. Kroenke, Spitzer, Williams, Löwe, 2010).
Analyses
We removed one site from the analyses because they did not collect information on living arrangement, an important variable in this study ( n = 83). There were no statistically significant differences between responses from the site that we dropped and the seven that remained related to the items measured at baseline or their follow-up loneliness score and self-reported improvement in loneliness. Seven sites with 376 participants were included in these analyses. Equivalence between groups at baseline was assessed for all descriptive control variables and the loneliness score by using analysis of variance for continuous variables, and chi-square tests for categorical variables in SAS version 9.3.8.
The pilot RCT conducted by MOWA had two primary aims (i) to compare receiving meals (regardless of delivery method) with not receiving meals and (ii) to compare receiving hot meals with receiving frozen meals. To answer the question of whether home-delivered meals improve feelings of loneliness, we pooled participants who received both daily-delivered and frozen meal delivery. We evaluated whether the mean follow-up loneliness score was equal across the two groups (control group and meals group) while statistically controlling for the effects of the covariates using the analysis of covariance (ANCOVA) modeling approach (SAS version 9.3.8). We specified the following model:
where the unit of analysis is the individual participant, i , and LPost, is the post loneliness score for participant i ; LPre, is the presurvey loneliness score for participant i ; Group is a dummy variable indicating the intervention condition to which participant i was assigned (i.e., Group i = 1 if participant was assigned to one of the meals groups); X is a vector of the control variables including site dummy variables and the baseline characteristics described earlier; and e represents the unexplained variance. β2 is the key coefficient used to determine whether the adjusted follow-up loneliness scores for the meals group were significantly different from the adjusted follow-up loneliness scores for the control group. We repeated these analyses to compare the difference in the postsurvey loneliness scores between the two groups receiving meals.
Additionally, l L differences in self-reported improvement in loneliness were limited to individuals who received meals and completed the follow-up survey. Logistic regression controlling for site, baseline loneliness score, and the control variables described earlier was utilized to test for significant differences between the two groups receiving meals on their self-reported improvement in loneliness. Secondary analysis of these study data received expedited review from the Brown University Institutional Review Board.
Results
Characteristics of Sample
Descriptive characteristics of the final analytic sample are presented in Table 1 . There were no statistically significant differences between groups on their baseline loneliness scores. There was a statistically significant difference between the groups in the proportion of individuals who participated in groups, with the control group having a statistically larger proportion that participated in groups compared with the groups receiving meals. Additionally, there was a statistically significant difference in the proportion of members in the control group that were married compared with the groups receiving meals. Overall, study participants represent a vulnerable population at risk for loneliness: over half lived alone, 14% reported having no one to call on for help, only a quarter of participants reported participating in groups, 30% screened positive for depression and 26% for anxiety, 24% were married, 22% had poor self-rated health, and almost 20% reported having contact with friends or family less than once or twice a month. In addition, the majority of participants (54%) reported needing assistance with personal care activities such as bathing, eating, or dressing and reported sometimes not having enough money to buy the food they need.
. | Control . | Daily delivered . | Frozen, weekly delivered . |
---|---|---|---|
% or Mean ( SD ) . | % or Mean ( SD ) . | % or Mean ( SD ) . | |
Loneliness | |||
Baseline Loneliness Score | 3.5 (2.9) | 3.5 (2.7) | 3.1 (2.6) |
Unadjusted Follow-up Loneliness Score a,b | 4.2 (2.7) | 3.7 (2.9) | 3.2 (2.6) |
Self-rated improvement in loneliness a,b | 71.5 | 56.3 | |
Social support | |||
Live alone | 58.8 | 56.1 | 50.0 |
Contact with friends or family daily | 56.7 | 58.7 | 64.4 |
Contact with friends of family once or twice a month or less | 19.7 | 20.3 | 10.6 |
Has someone to call for help | 86.6 | 87.2 | 84.8 |
Group membership | |||
Participates in groups a | 34.9 | 22.5 | 21.7 |
Self-rated health | |||
Excellent | 0.8 | 0.7 | 1.9 |
Very good | 4.0 | 10.6 | 4.8 |
Good | 24.6 | 19.7 | 22.9 |
Fair | 46.0 | 49.3 | 47.6 |
Poor | 24.6 | 19.7 | 22.9 |
Risk factors | |||
Anxiety | 20.8 | 21.3 | 22.9 |
Depression | 32.3 | 28.7 | 29.3 |
Enrolled in medicaid | 33.8 | 28.5 | 35.8 |
Not enough money to buy food | 52.0 | 53.2 | 53.3 |
Needs assistance with personal care needs | 47.2 | 54.6 | 62.9 |
Demographics | |||
Age | 75.7 (9.6) | 77.4 (10.3) | 76.2 (9.1) |
Married a | 31.5 | 18.9 | 21.9 |
White | 57.6 | 60.0 | 48.0 |
Black | 37.6 | 32.8 | 45.0 |
Education | |||
Less than high school | 34.9 | 32.5 | 26.0 |
High school diploma or GED | 34.1 | 35.0 | 36.5 |
Some college | 17.5 | 23.1 | 18.3 |
College degree or higher | 13.5 | 7.7 | 19.2 |
Site | |||
Site 1 | 3.2 | 3.5 | 3.8 |
Site 2 | 16.5 | 14.0 | 16.0 |
Site 3 | 16.5 | — | 21.7 |
Site 4 | 22.1 | 28.0 | — |
Site 5 | 27.6 | 34.3 | 34.9 |
Site 6 | — | 5.6 | 7.6 |
Site 7 | 14.2 | 14.7 | 16.0 |
. | Control . | Daily delivered . | Frozen, weekly delivered . |
---|---|---|---|
% or Mean ( SD ) . | % or Mean ( SD ) . | % or Mean ( SD ) . | |
Loneliness | |||
Baseline Loneliness Score | 3.5 (2.9) | 3.5 (2.7) | 3.1 (2.6) |
Unadjusted Follow-up Loneliness Score a,b | 4.2 (2.7) | 3.7 (2.9) | 3.2 (2.6) |
Self-rated improvement in loneliness a,b | 71.5 | 56.3 | |
Social support | |||
Live alone | 58.8 | 56.1 | 50.0 |
Contact with friends or family daily | 56.7 | 58.7 | 64.4 |
Contact with friends of family once or twice a month or less | 19.7 | 20.3 | 10.6 |
Has someone to call for help | 86.6 | 87.2 | 84.8 |
Group membership | |||
Participates in groups a | 34.9 | 22.5 | 21.7 |
Self-rated health | |||
Excellent | 0.8 | 0.7 | 1.9 |
Very good | 4.0 | 10.6 | 4.8 |
Good | 24.6 | 19.7 | 22.9 |
Fair | 46.0 | 49.3 | 47.6 |
Poor | 24.6 | 19.7 | 22.9 |
Risk factors | |||
Anxiety | 20.8 | 21.3 | 22.9 |
Depression | 32.3 | 28.7 | 29.3 |
Enrolled in medicaid | 33.8 | 28.5 | 35.8 |
Not enough money to buy food | 52.0 | 53.2 | 53.3 |
Needs assistance with personal care needs | 47.2 | 54.6 | 62.9 |
Demographics | |||
Age | 75.7 (9.6) | 77.4 (10.3) | 76.2 (9.1) |
Married a | 31.5 | 18.9 | 21.9 |
White | 57.6 | 60.0 | 48.0 |
Black | 37.6 | 32.8 | 45.0 |
Education | |||
Less than high school | 34.9 | 32.5 | 26.0 |
High school diploma or GED | 34.1 | 35.0 | 36.5 |
Some college | 17.5 | 23.1 | 18.3 |
College degree or higher | 13.5 | 7.7 | 19.2 |
Site | |||
Site 1 | 3.2 | 3.5 | 3.8 |
Site 2 | 16.5 | 14.0 | 16.0 |
Site 3 | 16.5 | — | 21.7 |
Site 4 | 22.1 | 28.0 | — |
Site 5 | 27.6 | 34.3 | 34.9 |
Site 6 | — | 5.6 | 7.6 |
Site 7 | 14.2 | 14.7 | 16.0 |
a Differences between groups statistically significant at the p < .05 level.
b Measured at follow-up.
. | Control . | Daily delivered . | Frozen, weekly delivered . |
---|---|---|---|
% or Mean ( SD ) . | % or Mean ( SD ) . | % or Mean ( SD ) . | |
Loneliness | |||
Baseline Loneliness Score | 3.5 (2.9) | 3.5 (2.7) | 3.1 (2.6) |
Unadjusted Follow-up Loneliness Score a,b | 4.2 (2.7) | 3.7 (2.9) | 3.2 (2.6) |
Self-rated improvement in loneliness a,b | 71.5 | 56.3 | |
Social support | |||
Live alone | 58.8 | 56.1 | 50.0 |
Contact with friends or family daily | 56.7 | 58.7 | 64.4 |
Contact with friends of family once or twice a month or less | 19.7 | 20.3 | 10.6 |
Has someone to call for help | 86.6 | 87.2 | 84.8 |
Group membership | |||
Participates in groups a | 34.9 | 22.5 | 21.7 |
Self-rated health | |||
Excellent | 0.8 | 0.7 | 1.9 |
Very good | 4.0 | 10.6 | 4.8 |
Good | 24.6 | 19.7 | 22.9 |
Fair | 46.0 | 49.3 | 47.6 |
Poor | 24.6 | 19.7 | 22.9 |
Risk factors | |||
Anxiety | 20.8 | 21.3 | 22.9 |
Depression | 32.3 | 28.7 | 29.3 |
Enrolled in medicaid | 33.8 | 28.5 | 35.8 |
Not enough money to buy food | 52.0 | 53.2 | 53.3 |
Needs assistance with personal care needs | 47.2 | 54.6 | 62.9 |
Demographics | |||
Age | 75.7 (9.6) | 77.4 (10.3) | 76.2 (9.1) |
Married a | 31.5 | 18.9 | 21.9 |
White | 57.6 | 60.0 | 48.0 |
Black | 37.6 | 32.8 | 45.0 |
Education | |||
Less than high school | 34.9 | 32.5 | 26.0 |
High school diploma or GED | 34.1 | 35.0 | 36.5 |
Some college | 17.5 | 23.1 | 18.3 |
College degree or higher | 13.5 | 7.7 | 19.2 |
Site | |||
Site 1 | 3.2 | 3.5 | 3.8 |
Site 2 | 16.5 | 14.0 | 16.0 |
Site 3 | 16.5 | — | 21.7 |
Site 4 | 22.1 | 28.0 | — |
Site 5 | 27.6 | 34.3 | 34.9 |
Site 6 | — | 5.6 | 7.6 |
Site 7 | 14.2 | 14.7 | 16.0 |
. | Control . | Daily delivered . | Frozen, weekly delivered . |
---|---|---|---|
% or Mean ( SD ) . | % or Mean ( SD ) . | % or Mean ( SD ) . | |
Loneliness | |||
Baseline Loneliness Score | 3.5 (2.9) | 3.5 (2.7) | 3.1 (2.6) |
Unadjusted Follow-up Loneliness Score a,b | 4.2 (2.7) | 3.7 (2.9) | 3.2 (2.6) |
Self-rated improvement in loneliness a,b | 71.5 | 56.3 | |
Social support | |||
Live alone | 58.8 | 56.1 | 50.0 |
Contact with friends or family daily | 56.7 | 58.7 | 64.4 |
Contact with friends of family once or twice a month or less | 19.7 | 20.3 | 10.6 |
Has someone to call for help | 86.6 | 87.2 | 84.8 |
Group membership | |||
Participates in groups a | 34.9 | 22.5 | 21.7 |
Self-rated health | |||
Excellent | 0.8 | 0.7 | 1.9 |
Very good | 4.0 | 10.6 | 4.8 |
Good | 24.6 | 19.7 | 22.9 |
Fair | 46.0 | 49.3 | 47.6 |
Poor | 24.6 | 19.7 | 22.9 |
Risk factors | |||
Anxiety | 20.8 | 21.3 | 22.9 |
Depression | 32.3 | 28.7 | 29.3 |
Enrolled in medicaid | 33.8 | 28.5 | 35.8 |
Not enough money to buy food | 52.0 | 53.2 | 53.3 |
Needs assistance with personal care needs | 47.2 | 54.6 | 62.9 |
Demographics | |||
Age | 75.7 (9.6) | 77.4 (10.3) | 76.2 (9.1) |
Married a | 31.5 | 18.9 | 21.9 |
White | 57.6 | 60.0 | 48.0 |
Black | 37.6 | 32.8 | 45.0 |
Education | |||
Less than high school | 34.9 | 32.5 | 26.0 |
High school diploma or GED | 34.1 | 35.0 | 36.5 |
Some college | 17.5 | 23.1 | 18.3 |
College degree or higher | 13.5 | 7.7 | 19.2 |
Site | |||
Site 1 | 3.2 | 3.5 | 3.8 |
Site 2 | 16.5 | 14.0 | 16.0 |
Site 3 | 16.5 | — | 21.7 |
Site 4 | 22.1 | 28.0 | — |
Site 5 | 27.6 | 34.3 | 34.9 |
Site 6 | — | 5.6 | 7.6 |
Site 7 | 14.2 | 14.7 | 16.0 |
a Differences between groups statistically significant at the p < .05 level.
b Measured at follow-up.
As documented in Table 1 , all three groups had similar loneliness scores at baseline. However, the groups’ unadjusted loneliness scores at follow-up were significantly different from one another. Specifically, the control group had the highest loneliness score (indicating more loneliness) with an average score of 4.2, followed by the group receiving daily-delivered meals (3.7) and then the group receiving frozen, once-weekly delivered meals (3.2). In terms of self-rated improvement in loneliness, participants receiving daily-delivered meals were more likely to indicate that the home-delivered meals program helped to improve their feelings of loneliness as compared with the group receiving frozen, once-weekly delivered meals when not adjusting for any baseline characteristics.
Comparing Feelings of Loneliness at Follow-Up Between the Control Group and the Group Receiving Meals
Results from an ANCOVA controlling for baseline loneliness score and adjusting for covariates that have been shown to be related to loneliness suggest that participants who received home-delivered meals (either daily-delivered or weekly-delivered) had lower adjusted loneliness scores (3.4) than the control group (4.2; see Table 2 ) at the end of the study period. This difference was statistically significant (estimate = 0.73, p < .05).
. | . | Adjusted Post-Loneliness Score . | Estimate . | SE . | p Value . |
---|---|---|---|---|---|
Model 1 comparison | Control | 4.17 | .73 | 0.31 | .018 |
Meals | 3.44 | ||||
Model 2 comparison | Frozen, weekly delivered | 3.23 | −0.39 | 0.42 | .359 |
Daily delivered | 3.62 |
. | . | Adjusted Post-Loneliness Score . | Estimate . | SE . | p Value . |
---|---|---|---|---|---|
Model 1 comparison | Control | 4.17 | .73 | 0.31 | .018 |
Meals | 3.44 | ||||
Model 2 comparison | Frozen, weekly delivered | 3.23 | −0.39 | 0.42 | .359 |
Daily delivered | 3.62 |
Note: Models control for the baseline loneliness score, study site, education, age, self-rated health, race, marital status, their frequency of contact with friends or family, and whether or not the participant lives alone, has someone to call on for help, has enough money to buy the food they need, participates in groups, reports needing assistance with personal care needs, is enrolled in Medicaid, screens positive for anxiety, and screens positive for depression.
. | . | Adjusted Post-Loneliness Score . | Estimate . | SE . | p Value . |
---|---|---|---|---|---|
Model 1 comparison | Control | 4.17 | .73 | 0.31 | .018 |
Meals | 3.44 | ||||
Model 2 comparison | Frozen, weekly delivered | 3.23 | −0.39 | 0.42 | .359 |
Daily delivered | 3.62 |
. | . | Adjusted Post-Loneliness Score . | Estimate . | SE . | p Value . |
---|---|---|---|---|---|
Model 1 comparison | Control | 4.17 | .73 | 0.31 | .018 |
Meals | 3.44 | ||||
Model 2 comparison | Frozen, weekly delivered | 3.23 | −0.39 | 0.42 | .359 |
Daily delivered | 3.62 |
Note: Models control for the baseline loneliness score, study site, education, age, self-rated health, race, marital status, their frequency of contact with friends or family, and whether or not the participant lives alone, has someone to call on for help, has enough money to buy the food they need, participates in groups, reports needing assistance with personal care needs, is enrolled in Medicaid, screens positive for anxiety, and screens positive for depression.
Comparing Feelings of Loneliness at Follow-Up Between the Two Groups Receiving Meals
When using ANCOVA and adjusting for baseline loneliness and other covariates, there was no statistically significant difference in the follow-up loneliness scores between the groups receiving daily-delivered and weekly-delivered meals ( Table 2 ). However, results from logistic regression analysis suggest that individuals receiving daily-delivered meals were three times more likely than individuals receiving frozen, once-weekly delivered meals to indicate that the services received from the home-delivered meals program helped them feel less lonely (odds ratio = 3.23; 95% confidence interval = 1.39–7.54; p < .01) when controlling for baseline loneliness scores and other characteristics that may influence feelings of loneliness ( Table 3 ).
. | Odds ratio . | 95% Confidence interval . | p Value . | |
---|---|---|---|---|
Daily-delivered meal | 3.23 | 1.39 | −7.54 | .007 |
Baseline Loneliness Score | 1.10 | 0.93 | −1.29 | .258 |
Social support | ||||
Live alone | 1.51 | 0.66 | −3.44 | .329 |
Contact with friends or family weekly (Ref. Daily) | 0.45 | 0.17 | −1.19 | .107 |
Contact with friends of family once or twice a month or less (Ref. Daily) | 0.29 | 0.09 | −0.93 | .038 |
Has someone to call for help | 0.23 | 0.06 | −0.98 | .047 |
Group membership | ||||
Participates in groups | 0.45 | 0.18 | −1.09 | .075 |
Self-rated health | ||||
Good (Ref. Excellent or Very Good) | 1.47 | 0.34 | −6.31 | .604 |
Fair (Ref. Excellent or Very Good) | 1.16 | 0.29 | −4.58 | .837 |
Poor (Ref. Excellent or Very Good) | 0.78 | 0.17 | −3.49 | .740 |
Risk factors | ||||
Anxiety | 1.09 | 0.42 | −2.85 | .862 |
Depression | 1.05 | 0.41 | −2.73 | .915 |
Enrolled in medicaid | 0.89 | 0.38 | −2.12 | .795 |
Not enough money to buy food | 1.23 | 0.53 | −2.85 | .637 |
Needs assistance with personal care needs | 1.20 | 0.53 | −2.74 | .660 |
Demographics | ||||
Age | 1.02 | 0.98 | −1.06 | .385 |
White | 1.96 | 0.36 | −10.84 | .439 |
Black | 1.00 | 0.21 | −4.89 | .999 |
Married | 1.05 | 0.39 | −2.85 | .925 |
Highest level of education greater than high school | 2.00 | 0.89 | −4.53 | .096 |
c -statistic = .772 |
. | Odds ratio . | 95% Confidence interval . | p Value . | |
---|---|---|---|---|
Daily-delivered meal | 3.23 | 1.39 | −7.54 | .007 |
Baseline Loneliness Score | 1.10 | 0.93 | −1.29 | .258 |
Social support | ||||
Live alone | 1.51 | 0.66 | −3.44 | .329 |
Contact with friends or family weekly (Ref. Daily) | 0.45 | 0.17 | −1.19 | .107 |
Contact with friends of family once or twice a month or less (Ref. Daily) | 0.29 | 0.09 | −0.93 | .038 |
Has someone to call for help | 0.23 | 0.06 | −0.98 | .047 |
Group membership | ||||
Participates in groups | 0.45 | 0.18 | −1.09 | .075 |
Self-rated health | ||||
Good (Ref. Excellent or Very Good) | 1.47 | 0.34 | −6.31 | .604 |
Fair (Ref. Excellent or Very Good) | 1.16 | 0.29 | −4.58 | .837 |
Poor (Ref. Excellent or Very Good) | 0.78 | 0.17 | −3.49 | .740 |
Risk factors | ||||
Anxiety | 1.09 | 0.42 | −2.85 | .862 |
Depression | 1.05 | 0.41 | −2.73 | .915 |
Enrolled in medicaid | 0.89 | 0.38 | −2.12 | .795 |
Not enough money to buy food | 1.23 | 0.53 | −2.85 | .637 |
Needs assistance with personal care needs | 1.20 | 0.53 | −2.74 | .660 |
Demographics | ||||
Age | 1.02 | 0.98 | −1.06 | .385 |
White | 1.96 | 0.36 | −10.84 | .439 |
Black | 1.00 | 0.21 | −4.89 | .999 |
Married | 1.05 | 0.39 | −2.85 | .925 |
Highest level of education greater than high school | 2.00 | 0.89 | −4.53 | .096 |
c -statistic = .772 |
N = 249. Model includes site dummy variables.
. | Odds ratio . | 95% Confidence interval . | p Value . | |
---|---|---|---|---|
Daily-delivered meal | 3.23 | 1.39 | −7.54 | .007 |
Baseline Loneliness Score | 1.10 | 0.93 | −1.29 | .258 |
Social support | ||||
Live alone | 1.51 | 0.66 | −3.44 | .329 |
Contact with friends or family weekly (Ref. Daily) | 0.45 | 0.17 | −1.19 | .107 |
Contact with friends of family once or twice a month or less (Ref. Daily) | 0.29 | 0.09 | −0.93 | .038 |
Has someone to call for help | 0.23 | 0.06 | −0.98 | .047 |
Group membership | ||||
Participates in groups | 0.45 | 0.18 | −1.09 | .075 |
Self-rated health | ||||
Good (Ref. Excellent or Very Good) | 1.47 | 0.34 | −6.31 | .604 |
Fair (Ref. Excellent or Very Good) | 1.16 | 0.29 | −4.58 | .837 |
Poor (Ref. Excellent or Very Good) | 0.78 | 0.17 | −3.49 | .740 |
Risk factors | ||||
Anxiety | 1.09 | 0.42 | −2.85 | .862 |
Depression | 1.05 | 0.41 | −2.73 | .915 |
Enrolled in medicaid | 0.89 | 0.38 | −2.12 | .795 |
Not enough money to buy food | 1.23 | 0.53 | −2.85 | .637 |
Needs assistance with personal care needs | 1.20 | 0.53 | −2.74 | .660 |
Demographics | ||||
Age | 1.02 | 0.98 | −1.06 | .385 |
White | 1.96 | 0.36 | −10.84 | .439 |
Black | 1.00 | 0.21 | −4.89 | .999 |
Married | 1.05 | 0.39 | −2.85 | .925 |
Highest level of education greater than high school | 2.00 | 0.89 | −4.53 | .096 |
c -statistic = .772 |
. | Odds ratio . | 95% Confidence interval . | p Value . | |
---|---|---|---|---|
Daily-delivered meal | 3.23 | 1.39 | −7.54 | .007 |
Baseline Loneliness Score | 1.10 | 0.93 | −1.29 | .258 |
Social support | ||||
Live alone | 1.51 | 0.66 | −3.44 | .329 |
Contact with friends or family weekly (Ref. Daily) | 0.45 | 0.17 | −1.19 | .107 |
Contact with friends of family once or twice a month or less (Ref. Daily) | 0.29 | 0.09 | −0.93 | .038 |
Has someone to call for help | 0.23 | 0.06 | −0.98 | .047 |
Group membership | ||||
Participates in groups | 0.45 | 0.18 | −1.09 | .075 |
Self-rated health | ||||
Good (Ref. Excellent or Very Good) | 1.47 | 0.34 | −6.31 | .604 |
Fair (Ref. Excellent or Very Good) | 1.16 | 0.29 | −4.58 | .837 |
Poor (Ref. Excellent or Very Good) | 0.78 | 0.17 | −3.49 | .740 |
Risk factors | ||||
Anxiety | 1.09 | 0.42 | −2.85 | .862 |
Depression | 1.05 | 0.41 | −2.73 | .915 |
Enrolled in medicaid | 0.89 | 0.38 | −2.12 | .795 |
Not enough money to buy food | 1.23 | 0.53 | −2.85 | .637 |
Needs assistance with personal care needs | 1.20 | 0.53 | −2.74 | .660 |
Demographics | ||||
Age | 1.02 | 0.98 | −1.06 | .385 |
White | 1.96 | 0.36 | −10.84 | .439 |
Black | 1.00 | 0.21 | −4.89 | .999 |
Married | 1.05 | 0.39 | −2.85 | .925 |
Highest level of education greater than high school | 2.00 | 0.89 | −4.53 | .096 |
c -statistic = .772 |
N = 249. Model includes site dummy variables.
Discussion
The home-delivered meals program has remained one of the most popular programs for delivery of nutrition and supportive services to older Americans. Although advocates have long argued that the socialization and “safety check” provided by the meal delivery driver is as important as the meal itself; this study is the first of its kind to quantify the effect of home-delivered meals on participants’ feelings of loneliness. Our results confirmed our hypothesis that individuals who received home-delivered meals would have lower loneliness scores compared with the control group that did not receive meals. These results also partially supported our hypothesis that individuals who received daily-delivered meals, and therefore more frequent social contacts, would experience the greatest improvements in their self-reported feelings of loneliness.
Our findings are aligned with Weiss’ revision of the Attachment Theory, in that when new social contacts are added, social loneliness decreases. We speculate that this was likely to occur for three reasons. First, individuals who are on waiting lists for home-delivered meals are characterized as being homebound and therefore, the majority of their social contact takes place within the home. As evidenced by our sample, over half live alone and more than 40% do not have daily contact with friends or family. Therefore, it is reasonable to assume that the majority of this population has limited social contact. By becoming a Meals on Wheels recipient, these homebound individuals, many of whom are socially isolated, now have an additional person(s) present in their life. Although the level of interaction during the delivery process varies by client and driver, the premise is that this additional social contact provides them with uncompensated and nonobligatory socialization and care. We believe that this recognition may be one mechanism that decreases their feelings of social loneliness. Secondly, Meals on Wheels volunteers and staff report that friendships can and do develop between drivers who deliver meals and their clients. It is reasonable to assume that this new friendship decreases feelings of loneliness among clients. Finally, for individuals on waiting lists, being given the opportunity to be “part of the group” receiving meals may increase their feelings of belonging and therefore decrease their feelings of loneliness as posited by Weiss. One if not all of these assumptions are believed to explain the findings and justify our hypothesis that home-delivered meals decrease feelings of social loneliness.
The role of home-delivered meals in reducing loneliness is a laudable outcome in its own right, but these findings potentially have significant implications for our health care system as a whole. The health risks posed by loneliness may be particularly severe for older adults ( Cacioppo & Hawkley, 2003 ; Holt-Lunstad, Smith, Baker, Harris, & Stephenson, 2015 ), especially as they are likely to face stressful life course transitions, health problems, and disabilities. Previous research has suggested that loneliness is associated with an increased risk of functional decline and death ( Holt-Lunstad et al., 2015 ; Perissinotto et al., 2012 ), health-related behavioral and biological risk factors ( Hawkley, Thisted, Masi, & Cacioppo, 2010 ; Shankar, McMunn, Banks, & Steptoe, 2011 ), as well as an increased risk of ER visits ( Molloy, McGee, O’Neill, & Conroy, 2010 ) likelihood of having a coronary condition ( Sorkin, Rook, & Lu, 2002 ), and nursing home placement (D. W. Russell, Cutrona, de la Mora, & Wallace, 1997 ). Therefore, by identifying loneliness, as well as ensuring that isolated or lonely older adults receive more social contact through their service providers, we can improve the health and well-being of homebound older adults, particularly individuals who live alone, and potentially decrease the influence of these modifiable risk factors on our health care system.
This is the first study in 30 years to examine the outcomes associated with home-delivered meals and different meal delivery methods ( Osteraas et al., 1983 ). Our findings using the ANCOVA model adjusting for baseline loneliness scores suggest that once-weekly delivery (a cost-savings option) may have equally beneficial effects as daily-delivered meals in terms of reducing participants’ loneliness. However, when relying on participants’ self-reported beliefs that meals helped improve their loneliness, we do see a significant difference between meal delivery type. In a time of increased resistance to supporting social programs, finding ways to reduce costs and staff and volunteer time that could additionally serve more people on waiting lists is key. Therefore, future work is needed to understand the discrepancy in these findings and ensure that providing frozen, once-weekly delivered meals by trained Meals on Wheels staff or volunteers is indeed as beneficial as the daily contact afforded by the traditional program.
Although these findings help inform our current understanding of the benefits of this program, it also sets the stage for additional work in the area of loneliness among this population. Future research should examine the pathways by which the home-delivery process may impact feelings of loneliness and other quality of life outcomes. In addition, it would be worthwhile to go beyond measuring just the frequency of interaction, as we did in this study, but to also quantify the level and quality of the interaction that is afforded through the home-delivered meals program to determine whether there is a “dosage” response by which participants begin to experience improvements in subjective feelings like loneliness.
Limitations
It is important to note this study’s limitations. Although there were no differential rates in attrition and our retention rate was quite high for a community-based study, because of attrition, our findings may underestimate the true effect of the intervention on outcomes. For example, the loneliest individuals may have attrited from the study, thereby biasing our results. However, it is important to note that we found no differences in loneliness measured at baseline among individuals who dropped out of the study suggesting that the effect of attrition may have been minimal. Future work should include larger sample sizes and should assure closer follow-up in order to determine the differential effects of the treatment on participants.
Our sample consisted of individuals on waiting lists for home-delivered meals. It is important to note that all but one site reported that they prioritized their waiting lists so that those most in need would receive services first. In addition, our sample is younger, less likely to live alone, and less likely to live in poverty (as indicated by receiving Medicaid assistance) than the national population of older adults receiving home-delivered meals in 2013 ( Administration on Aging, 2015 ). Therefore, we believe that the effects seen may have been modest in comparison with those we would have witnessed were to examine the effectiveness of home-delivered meals in the population currently receiving meals and therefore, might not be generalizable to that population. Future research should include individuals who are determined to be most at need and already enrolled in the program in order to more fully understand the impact that the different meal delivery methods can have on clients’ health, health care utilization, and quality of life, in addition to feelings of loneliness.
It is important to note that the weekly delivery of frozen meals in this study consistently maintained some degree of social contact in the form of weekly driver visits. Therefore, the follow-up loneliness scores among individuals who received once-weekly delivery of frozen meals may not be generalizable to older adults who receive frozen meals via mail courier. With the growth in new and different meal delivery modalities, such as more affordable drop-shipped meals delivered via postal courier, it is important that decisions made regarding which model is most beneficial to older adults’ overall quality of life is informed by rigorous research. Additional research is needed to understand older adults’ experience receiving meals from these newer delivery models with limited to no person contact.
Finally, it is important to highlight some limitations of our measures. Although we hypothesized that home-delivered meals impact participants’ feelings of social loneliness, we did not have a question that directly measured social loneliness. Therefore, we relied on an overall measure of loneliness, the UCLA 3-Item Loneliness Scale. This measure has been shown to be strongly associated with commonly used measures of objective social isolation among older adults ( Hughes et al., 2004 ). For this reason, we believe that it is a suitable measure to understand feelings of loneliness, particularly social loneliness, among this population. Nevertheless, we recognize that there are other scales of loneliness that might have added different context to this analysis ( de Jong Gierveld & Van Tilburg, 2006 ; D. Russell, Cutrona, Rose, & Yurko, 1984 ; D. Russell et al., 1980 ; Wittenberg & Reis, 1986 ).
Additionally, use of a subjective self-reported single-item measure of improvement in loneliness is not ideal given that there are some potential sources of recall and response bias (e.g., inaccurate memory and social desirability) and because it lacks the depth, breadth and accuracy of a multi-item scale. However, we would expect that should bias occur, it would be evenly distributed across the two groups receiving meals and therefore would not account for the difference in responses that we observed. In addition, a single-item measure of loneliness has been suggested to be better adapted for use with the oldest old age groups ( Holmén, Ericsson, Andersson, & Winblad, 1992 ), and a meta-analysis of loneliness studies of middle-age to older adults found that 49% used a single-item measure of loneliness ( Pinquart & Sorensen, 2001 ). Therefore, we believe that this measure has value despite its limitations.
Finally, it is important to also emphasize that these findings are based on self-report to Meals on Wheels staff and therefore, there is the potential for response variation resulting from the use of different data collectors at each site. However, all interviewers were trained in the data collection procedures and we saw no statistically significant differences in responses, between sites.
Conclusions
Our study suggests that home-delivered meals improve the well-being of older adults, specifically by reducing feelings of loneliness. In terms of participant’s self-rated improvements in loneliness, our results signal that the benefits of the social contact afforded by daily-delivered meals exceed those of once-weekly-delivered, frozen meals. As the nation looks to serve more older adults who will be needing these services, it is important that research findings from rigorously conducted studies guide the decisions about how to best provide these services. This study confirms that the home-delivered meals program provides more than just nutrition to clients; it improves their quality of life. Furthermore, this work is a first look at the differences in outcomes associated with meal delivery modalities and sets the stage for additional work in this area.
Funding
This research project has been sponsored by Meals on Wheels America and has been made possible by a grant from AARP Foundation.
Conflict of Interest
The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of the funding sources or the organizations with which they are affiliated.
Acknowledgments
K. S. Thomas and D. Dosa were responsible for conception and study design, analysis and interpretation of data, and drafting and revising the manuscript. U. Akobundu coordinated data collection, managed the More than a Meal project, and helped revise the manuscript.
References
Author notes
Decision Editor: Deborah Carr, PhD