Reducing loneliness and improving well-being among older adults with animatronic pets

Abstract

Background

Studies consistently demonstrate that older adults who are lonely have higher rates of depression and increased mortality risk. Pet ownership may be a solution for loneliness; however, challenges related to pet ownership exist for older adults. Therefore, researchers and practitioners are examining the use of animatronic pets to reduce loneliness.

Objective

To determine the feasibility of an animatronic pet program, and whether ownership of animatronic pets would decrease loneliness and improve well-being among lonely older adults.

Methods

Eligible individuals were identified as lonely through a prior survey. Participants were provided with the choice of an animatronic pet and completed T1/T2/T3 surveys.

Results

Attrition was high; 168 (63%) participants completed T1/T2 surveys, and 125 (48%) also completed a T3 survey. Post survey data indicated that loneliness decreased, while mental well-being, resilience, and purpose in life improved. Frequent interactions with the pets were associated with greater improvement in mental well-being and optimism.

Conclusions

Animatronic pets appear to provide benefits for the well-being of lonely older adults. Future studies should employ randomized controlled designs examining the impact of animatronic pets.

Keywords:

• Loneliness
• animatronic pets
• older adults
• quality of life

Introduction

Rapid growth of the older population is expected to continue through 2050 and beyond (Anderson & Perrin, 2017). Up to 90% of adults aged 65 and older have reported they prefer to remain living independently at home later in life, while maintaining a consistent routine in a familiar environment (Peek et al., 2014). The treatment and management of chronic health conditions is a significant part of the lives of many older adults, especially in their later stages of life. However, research has demonstrated that psychological health and social well-being have impacts just as significant as chronic conditions on older adults’ physical health and mortality (Luo, Hawkley, Waite, & Cacioppo, 2012; Uchino, 2006). Loneliness, in particular, has garnered increased research focus as a specific factor contributing to these physical health outcomes (Holt-Lunstad, Smith, Baker, Harris, & Stephenson, 2015). Interventions aimed at addressing the needs of lonely older adults that improve psychosocial well-being may provide great benefits. The primary purpose of this study was to test the feasibility and initial effectiveness of animatronic pets among community-dwelling lonely older adults and determine the psychological outcomes and level of satisfaction with the use of these pets.

Loneliness and older adults

Studies consistently demonstrate that older adults who are lonely or socially isolated have higher rates of depression, more health conditions, and greater mortality (Holt-Lunstad, Robles, & Sbarra, 2017). Loneliness is often defined as the gap between a person’s desired relationships and actual relationships; or a mismatch between those relationship standards (Luo et al., 2012). The research on social isolation primarily considers the absence of social relationships or the quantity of those relationships, while loneliness is conceptualized by the quality of the relationship or the self-perception of feeling alone (Holt-Lunstad et al., 2017).

Early work by Rook (1990) describes loneliness as a function of an individual’s inability to maintain the basic human need of social bonding. According to this research, this failure to maintain social bonding is most often involuntary and distressing to the individual. However, being physically alone may not always elicit loneliness for older adults as compared to younger cohorts (Larson, 1990). Loneliness is considered a multidimensional construct based on subjective evaluation of an individual’s own social network (De Jong-Gierveld, 1987). This evaluation can be broadly defined as how the individual considers 1) the absence of close relationships, 2) the perception as having an external versus internal cause and the ability to change the situation, and 3) different emotions associated with loneliness, such as shame and frustration (De Jong-Gierveld, 1998).

Meanwhile, in their review of the literature, Qualter et al. (2015) illustrate the reaffiliation motive model (RAM) as the process in which perceived social isolation or loneliness should motivate an individual to reconnect with others. However, in the RAM model, prolonged loneliness will produce the opposite result, leading to further negative affect. Further, risk factors for loneliness change over the lifespan (Qualter et al., 2015). Specific loneliness risk factors for older adults include being female, lower socioeconomic status, lower physical functioning, living alone and being unmarried or without a significant other (Luhmann & Hawkley, 2016).

Loneliness is common among adults aged 65 years and older, especially those age 80 and above, with prevalence ranging from about 30% to 60% (Aartsen & Jylhä, 2011; Musich, Wang, Hawkins, & Yeh, 2015). The prevalence of loneliness among older adults varies considerably depending on the assessment tool used and the type of loneliness being assessed. In one particular study, 55% of the age 65+ population reported they regularly experience moderate to severe loneliness (Musich et al., 2015). Research supporting the impact of loneliness is growing, suggesting that loneliness leads to depression, sleep problems, hypertension, functional decline, cognitive impairments, and mortality (Musich et al., 2015; Perissinotto, Cenzer, & Covinsky, 2012; Steptoe, Shankar, Demakakos, & Wardle, 2013).

Interventions to address loneliness

Despite studies demonstrating the serious negative health outcomes due to experiencing loneliness later in life, interventions that have focused on reducing loneliness within this population have reported limited success (Cacioppo, Grippo, London, Goossens, & Cacioppo, 2015). In a review of the literature, Cacioppo et al. (2015) grouped interventions to reduce loneliness into four primary categories: 1) increase social contact; 2) improve social support; 3) enhance social skills; and 4) address maladaptive social cognition. In this context, a variety of initiatives and multidimensional intervention types may hold promise for addressing loneliness. One approach to consider for interventions is pet ownership, which encompasses multiple components, such as increased levels of social contact, higher levels of social support, and improved psychological well-being. (McConnell, Brown, Shoda, Stayton, & Martin, 2011; Stanley, Conwell, Bowen, & Van Orden, 2014). It is reasonable to infer that interventions focusing on pet ownership would be beneficial to negative outcomes such as feelings of loneliness.

Pet ownership among older adults

Older adults who own pets tend to be more physically active and have more social connections (Curl, Bibbo, & Johnson, 2016). In addition, pet ownership can decrease depression, anxiety, and loneliness later in life (Gee, Mueller, & Curl, 2017). However, many older adults may not be able to feasibly handle the ongoing physical and financial commitments of caring for pets. Challenges related to pet ownership for older adults include their cost, physical requirements to care for a pet, forgetting to take care of a pet, and tripping or falling over pets (Anderson, Lord, Hill, & McCune, 2015; Stevens, Teh, & Haileyesus, 2010). Further, some housing policies have limitations on pet ownership. Meanwhile, for older adults who experience loneliness, an additional concern of pet ownership may be the potential death of a longtime companion (Needell & Mehta-Naik, 2016). Therefore, the use of animatronic pets for older adults may provide the benefits of live pets without typical challenges associated with pet ownership.

Technology and animatronic pets as an intervention

Older adults have become more willing to adopt emerging technologies, with their use of the Internet, social media resources, and ‘smart’ technologies increasing throughout the past decade (Kuerbis, Mulliken, Muench, Moore, & Gardner, 2017; Vroman, Arthanat, & Lysack, 2015). However, compared to younger age groups, many older adults still remain digitally disconnected from emerging technological developments (Kuerbis et al., 2017). Many technological innovations are emerging as options to reduce loneliness, increase social connectedness, and thus improve quality of life through user-friendly technologies even if the options were not originally designed to address these issues or targeted for older populations. Digital options to support social connectedness and successful aging have expanded with technology including at-home interactive devices, animatronic pets, and other digital toys/games reaching the market. Digitally based devices and pets typically respond to voice commands, answer question prompts, and interact with users with audible responses and/or with movements (Crist, 2016). For technically savvy seniors, and those who are lonely or live alone, these options can potentially offer a sense of engagement with a social connection even through this type of digital connection (Reis, Paulino, Paredes, & Barroso, 2017).

Several studies have investigated the use of animatronic pets for older adults and the associated outcomes; however, these studies have been limited to older adults in nursing homes and assisted living facilities (Broadbent et al., 2012; Banks, Willoughby, & Banks, 2008; Sung, Chang, Chin, & Lee, 2015). These interventions have primarily focused on improving loneliness, social connections, interaction skills, and caregiving assistance. One small, randomized control study (N = 40) using a robotic seal (Paro) followed residents in a retirement home, combining a hospital and rest home facility, over the course of 12 weeks (Robinson, Kerse, Macdonald, & Broadbent, 2013). Participants in the experimental groups reported significantly lower levels of loneliness as compared to the control group at the end of the study although changes in depression and quality of life were not significant. Although this study had some success, the pet was considered high in cost and was not designed to be similar to familiar pets such as a cat or a dog.

A recent review of the literature examined 58 studies that used robotic pets to address the health needs of older adults (Shishehgar, Kerr, & Blake, 2017). The review demonstrated support for robotic pets as companions, particularly for individuals experiencing loneliness. However, limited studies exist regarding the use of these pets among community-dwelling, independent living older adults. Therefore, the primary purpose of this study was to 1) test the feasibility and usability of animatronic pets among community-dwelling, independent living, and cognitively intact older adults; and 2) determine the impact these pets would have on loneliness and other psychological outcomes including quality of life, purpose, resilience, and optimism.

Methods

Recruitment and study population

Participants were recruited from a survey pool and screened for loneliness. The surveys were utilized as part of a larger research investigation within the UnitedHealthcare organization to improve customer experience. In 2016, approximately 5 million Medicare insureds were covered by an AARP® Medicare Supplement plan insured by UnitedHealthcare Insurance Company. These plans are offered in all 50 states, Washington DC, and various US territories. Eligibility criteria for this study included individuals covered by an AARP^® Medicare Supplement plan insured by UnitedHealthcare (for New York residents, UnitedHealthcare of New York), who were aged 65 and older, screened positive for loneliness on an initial survey, agreed to accept the animatronic pet, and to complete pre- and post-surveys. Exclusion criteria for participation included those who were already participating in UnitedHealthcare (UHC) clinical support, and current ownership of a live pet. A total of 3,937 individuals were contacted to participate in this study, and 277 (7%) agreed to participate. This recruitment rate was expected as our exclusion criteria included pet ownership (n = 1,660). Furthermore, qualitative data collected with other similar studies at UHC has found that many individuals have expressed the concern that participating in research with their healthcare providers may impact their health insurance rates. Thus, this impacted the ability of the research team to recruit participants effectively. In addition, no compensation was provided for participation in this study or the completion of the surveys. The final sample was reduced to 271 participants as few individuals were later identified as ineligible. Of those, 216 (80%) completed a T1 assessment (Table 1). This study was approved by the New England Institutional Review Board.

Table 1. Characteristics of respondents.

Characteristic	T1 N = 216	T1 and T2 N = 168	T1, T2 and T3 N = 125
	%	%	%
Age
65–74	51.4	48.2	49.6
75–84	30.6	31.5	29.6
> = 85	18.1	20.2	20.8
Female	52.8	55.4	56.8
HCC Score
< 0.8	43.1	44.0	44.0
0.8–2.0	39.8	40.5	41.6
> 2.0	17.1	15.5	14.4
Pet
Cat	29.6	30.4	31.2
Dog	70.4	69.6	69.4

Intervention

Psychological outcome metrics (purpose, resilience, optimism) were chosen for this study as part of this greater research effort discussed earlier. Participants received the animatronic pet of their choice (cat or dog) in the mail and were instructed to treat it as a pet. There were no other instructions given to the participants. They received a T1 survey at the same time as receipt of the pet, with a T2 survey delivered approximately 30 days after they had received the pet and a T3 survey approximately 60 days after receipt of the pet. In addition, twice a week for four weeks, participants received an interactive voice reminder (IVR) phone call encouraging interactions with the pet. The IVR phone call also asked participants to record whether they had been interacting with their pet (0-no, 1-yes) and, if so, how much time on an average day they were interacting with the pet (0, 1 (1–2 h), 2 (3–5 h), or 3 (6+ hours)). Participants were told in advance they would receive the calls on Tuesdays and Fridays between 10am and 6 pm in their time zone. If they were unavailable for the phone call, that data collection time point was missed.

Measures

Demographic information for all participants was collected either during the initial survey or obtained through administrative claims data. Demographics collected included age (65–74, 75–84, 85+), gender, caregiver status, and socioeconomic status.

Hierarchal Condition Category (HCC) risk scores were used to assess clinical health status (McCall & Cromwell, 2011). HCC scores are derived from the Centers for Medicare & Medicaid Services (CMS) guidelines and based on medical claims.

Quality of life

Quality of life (QoL) was assessed with the 12-item Veteran’s RAND (VR-12). The VR-12 is a validated general QoL measure that asks participants about their QoL in the previous four weeks. Two subscale scores are then derived from this measure: Physical component (PCS) and the mental component (MCS) scores. The algorithm is scored on a scale of 0–100, with higher scores indicating better physical and mental QoL (Selim et al., 2009). Cronbach’s α for the VR-12 was .92.

Loneliness

Loneliness was measured using the 10-item short version of the UCLA Loneliness Scale (Russell, 1996). Participants are asked on a four-point scale about their general feelings related to being lonely, left out, or isolated. Responses range from 1 (never) to 4 (often) and are summed to create a score ranging from 10 to 40. Higher scores indicate higher levels of loneliness. A score of 24+ is considered lonely. Cronbach’s α for loneliness was .88.

Psychological well-being

Measures of resilience, purpose in life, and optimism were used to assess psychological well-being. Resilience was measured with the six-item Brief Resilience Scale (BRS) (Smith et al., 2008). In this measure, participants are asked about their ability to bounce back from life experiences in the previous month on a five-point scale from 1 (strongly disagree) to 5 (strongly agree). Responses to the items are averaged so that total scores from 1 to 5, with higher scores indicating higher levels of resilience. Cronbach’s α for resilience was .87.

Purpose in Life was measured with seven items adapted from the National Institutes of Health (NIH) Tuberculosis Meaning and Purpose Scale Age 18+ (Hedberg, Gustafson, & Brulin, 2010). Responses, ranging from 1 (strongly disagree) to 5 (strongly agree), were averaged across the questions such that totals range from 1 to 5. Higher scores indicate higher levels of purpose in life. Cronbach’s α for purpose was .93.

Optimism was measured using the six-item Life Orientation Test-Revised (LOT-R) (Herzberg, Glaesmer, & Hoyer, 2006). One item was inadvertently deleted from the survey (In uncertain times I usually expect the best). Therefore, only five items were used in this study. Participants are asked how much they agree with statements about expecting the best or being optimistic. Responses range from 0 (disagree a lot) to 4 (agree a lot) and are summed to create a total score ranging from 0 to 20 with higher scores indicating greater optimism. Cronbach’s α for the optimism was .84.

Statistical testing

All data were imported into SAS. Matched pairs t-test analyses and repeated measures ANOVA were conducted to determine if there were significant changes on the study variables of QoL, loneliness, and psychological well-being. Regression analysis was used to determine if the interaction with the pet predicted the outcome variables.

Results

Response bias analysis, using claims data, was conducted for those who agreed to participate (n = 277) versus those who declined (n = 3,660) and for respondents (n = 216) versus non-respondents (n = 55). Analysis included age, gender, emergency room (ER) visits, inpatient (IP) admissions, clinical and drug adherence, prevalence of chronic conditions, and overall medical and drug costs. Overall, those who agreed to participate were very similar to those who declined. However, those who had agreed to participate in the study had more frequent ER visits in the last 12 months, higher levels of depression, and overall higher medical costs (but not drug costs) (p < .05). There were no differences for respondents versus non respondents.

Demographics of study participants for T1 (n = 216), those who completed T1 and T2 (168), and those who completed T1, T2, and T3 (n = 125) are shown in Table 1. At baseline, about half the respondents were between 65 and 74 and female, and most participants chose the animatronic dog (70%). In addition, 86% of participants reported previously owning a pet. Finally, baseline HCC scores indicated that participants were considered sicker than a typical Medicare Supplement plan population. Claims analysis at baseline indicated that most participants had diagnoses of hypertension (70%), almost half had hyperlipidemia (49%), about a third were obese (30%), and about a quarter had diabetes (27%).

Matched pairs t-tests were conducted for respondents who completed T1 and T2 (Table 2). As shown in Table 2, there was significant improvement from T1 to T2 in most of the study variables including mental well-being, purpose in life, resilience, and optimism. Further, loneliness decreased from T1 to T2. Repeated measures ANOVA was conducted to detect change over time, T1, T2, to T3 (Table 3). The sample size decreased significantly from T1 to T3 (from 168 to 125); however, after controlling for gender, characteristics of mental well-being, purpose in life, resilience, and loneliness remained significant. In addition, physical well-being declined over time.

Table 2. Matched paired t-test of survey variables pre and post (N = 168).

Variables	Mean T1	Mean T2	Difference	p-value|
Physical Health	37.86	37.18	−0.68	0.24
Mental Health*	44.26	47.17	2.91	0.01
Purpose*	26.08	27.17	1.09	0.01
Resilience*	3.21	3.30	0.09	0.05
Loneliness*	26.49	25.92	−0.57	0.05
Optimism*	11.10	11.66	0.56	0.05

Means of the score variables and their change from Pre-survey to Post-survey.

* Measurements with significant improvement from Pre-survey to Post-survey at p<.05.

Table 3. Repeated measures ANOVA (N = 125).

Variables	Mean T1	Mean T2	Mean T3	F	p-value
Physical Health*	37.73	37.29	36.06	4.26*	0.02
Mental Health*	44.75	48.13	48.93	12.84**	0.01
Purpose*	26.57	27.72	27.76	4.72*	0.01
Resilience*	3.26	3.37	3.28	2.98*	0.05
Loneliness*	26.22	25.41	25.80	3.87*	0.02
Optimism	11.19	11.86	11.64	2.17	0.11

Means of the score variables and their change from T1/T2/T3 after controlling for gender.

* Measurements with significant change at p < 0.05

** Measurements with significant change at p < .01

A frequency distribution depicting highest level of engagement in the IVR calls demonstrated that about 4% never responded to the calls (Table 4), while about 15% reported interacting with the pet six or more hours within the previous few days. This distribution demonstrates that participants likely provided data accurately. Self-reported interaction with the pet did not have a significant effect on any outcome variable at T2, although it did predict the likelihood of participants completing a T2. However, regression analysis determined that greater time spent interacting with the pets was associated with an improvement in participants’ mental health (Table 5). There was a marginal effect on optimism (p < .10) at T3.

Table 4. Distribution of greatest level of engagement with pet as reported during IVR call.

Q1	Q2	N	%
Interact with Pet	Hours of Interaction with Pets
No response	No response	6	3.6%
No	No response	2	1.2%
Yes	0	3	1.8%
Yes	1 (1–2 hours)	80	47.6%
Yes	2 (3–5 hours)	52	31.0%
Yes	3 (6+ hours)	25	14.9%
Total		168	100.0%

Each participant who completed a T1 and a T2 is represented by his/her greatest level of engagement out of 8 possible interactions (N = 168).

Table 5. Improvement of measurements associated with interaction hours with pets (N = 125).

Variables	Odds Ratio	95% confidence interval	p-value
Mental health	1.61	(1.036, 2.492)	0.05
Optimism	1.47	(0.934, 2.301)	0.10

Logistic regression models using interaction hours with pets as predictor and the improvement indicators of the variables as dependent.

Additional analyses were conducted for participants who chose an animatronic cat versus a dog. No significant association was found between the type of animatronic pet with any outcome variable.

Gender differences

Additional analysis was conducted to evaluate the effects of gender. Independent sample t tests were conducted to compare the means between males and females at T1. As shown in Table 6, male respondents were in overall better psychological health, reporting better mental well-being, purpose, resilience, and optimism. At T2, the gender differences between purpose, resilience, and optimism remained, while the gender difference for mental well-being disappeared. However, females were lonelier than males at T2. Finally, at T3, only the gender differences for resilience and optimism remained. Overall, the males in this study were healthier, although it does appear that the females did benefit from the animatronic pets.

Table 6. Gender difference on study measures (N = 125).

Variables	Survey	Female N = 71	Male N = 54	Difference	p-value
Physical Health	T1	36.88	37.54	−0.67	0.78
	T2†	35.59	39.50	−3.91	0.09
	T3	35.21	36.46	−1.26	0.57
Mental Health	T1*	43.30	46.31	−3.01	0.05
	T2	47.02	49.49	−2.47	0.22
	T3†	46.85	49.99	−3.14	0.08
Purpose	T1*	25.08	27.51	−2.43	0.05
	T2*	26.45	29.27	−2.82	0.01
	T3†	26.62	28.39	−1.78	0.08
Resilience	T1*	3.07	3.41	−0.34	0.05
	T2*	3.17	3.60	−0.43	0.01
	T3*	3.04	3.53	−0.49	0.01
Loneliness	T1	26.67	25.78	0.90	0.26
	T2^†	26.24	24.56	1.68	0.06
	T3	26.30	25.17	1.12	0.17
Optimism	T1*	10.39	12.19	−1.80	0.05
	T2*	10.95	13.01	−2.06	0.01
	T3*	10.54	12.94	−2.39	0.01

* Survey measurements with significant difference at p < 0.05.

† Survey measurements with significant difference at p< 0.10.

Discussion

The main objectives of this study were to 1) test the usability and acceptability of animatronic pets among older adults, and 2) determine if these animatronic pets could have an impact on the psychological well-being of lonely older adults, including the outcomes of level of loneliness, QoL, purpose, resilience, and optimism. The results provide evidence that study participants regularly interacted with their animatronic pets, which provided benefits including reduced loneliness, improved QoL, and psychological well-being. In addition, greater hours of interaction predicted improved mental well-being and optimism. These are important findings as there were minimal instructions given to the participants in this study regarding the interaction with pet. Meanwhile, attrition rates were high; for those who completed the surveys, loneliness decreased while purpose, resilience, and mental well-being improved over a 60-day period.

Previous interventions to address loneliness have reported limited success. However, as previously discussed, Cacioppo et al. (2015) has described four loneliness intervention categories that may support the success of these initiatives, including: increasing social contact, improving social support, enhancing social skills, and addressing maladaptive social cognition. In this context, it is likely the success of the current intervention focusing on animatronic pets was primarily achieved by increasing social contact, enhancing social skills, or addressing maladaptive social cognition. For instance, qualitative data on this sample found that some participants used their pets in social settings, which helped them to improve social interactions (Hudson et al., under review). In addition, for some participants, the pets provided comfort and helped improve their confidence, perhaps an explanation for the continued improvement in purpose and mental well-being, as well.

Meanwhile, the decline in loneliness was only detected at T2 and remained stable at T3, perhaps due to the loss of about one quarter of the sample at T3. In addition, while loneliness decreased, it was not at the level of what is typically considered no longer lonely. Meta-analysis by Masi, Chen, Hawkley, and Cacioppo (2011) reported similar trends of small and significant reductions that illustrates that individuals ‘improve but do not recover’ from loneliness (Masi et al., 2011).

Improved purpose and mental well-being were encouraging findings as researchers try to address QoL concerns for lonely older adults. Many of these older adults had previously owned pets (86%); thus, it is possible these animatronic pets provided a gap that was missing in their lives. Finally, regarding the choice of pet, it was initially surprising that most participants chose the animatronic dog rather than the cat. However, in 2018 pet ownership was included on the General Social Survey (GSS), with data indicating that dog ownership is the more prevalent than cat ownership. Therefore, the more popular choice may reflect participants’ previous pet ownership patterns.

To our knowledge, the current study is the first to examine the use of these pets within a generally healthy community-dwelling sample of older adults with no reported cognitive impairments. Importantly, although there was no control group for this study, the investigation did include a larger sample than many of the other studies and, perhaps as a consequence, found more significant relationships than other studies (Shishehgar et al., 2017). This provides some evidence that older adults are open to engaging in digital options to support social connectedness and successful aging. Furthermore, the context of this study within a healthcare environment demonstrates that low-tech digital interventions may have a strong impact in many areas. This has implications for healthcare providers considering intervention opportunities. Animatronic pets are relatively low in cost, ranging from $55 to $110 and available online. Although the individuals in this study were provided these pets free of charge and the pets were automatically mailed to them upon enrollment in the study, the pets may also be accessible to older adults regardless of their mobility and some level of financial constraints.

Study limitations include the lack of a control group; therefore certain findings may need further investigation. However, qualitative data collected following the study substantiated survey results, as participants described the benefits they derived from these pets (Hudson et al., under review). Additional limitations include the short-term follow-up and the study population (e.g.only lonely older adults). However, the strengths of high utilization and positive health outcomes on multiple constructs demonstrate that animatronic pets may be a unique, low cost, and simple solution to address loneliness among older adults. Future research should include a control group and expand the use of these animatronic pets to other older adult populations, including those who are not lonely as well ascaregivers. In addition, larger samples could identify whether the animatronic cat or dog is more effective or if having a pet prior to this study is more meaningful than owning an animatronic pet.

Conclusions

Loneliness and social isolation are common among older adults, impacting their overall health and quality of life; these issues have become important determinants of health later in life. Animatronic pets offer potential opportunities to support successful aging and reduce loneliness, primarily among those who choose to remain in their homes and who experience loneliness. Although the technology is still developing, robotic pets could provide enhanced social support, feelings of connectedness, and reduced loneliness while avoiding the higher costs of many types of interventions.

Disclosure statement

The authors declare no conflicts of interest with respect to the research, authorship, and/or publication of this article. The authors received no financial support for the research, authorship, and/or publication of this article.

Funding

This work was funded by the Supplemental Health Insurance Program. All authors are employed or contracted by UnitedHealth Group, ASI Services, Inc. However, their compensation was not dependent upon the results obtained in this research, and the investigators retained full independence in the conduct of this research.