Social Isolation Associated with Future Health Care Utilization

Abstract

Social isolation among individuals ages 65 years and older is associated with poor health outcomes. However, little is known about health care utilization patterns of socially isolated individuals. This retrospective, observational study evaluated associations between social isolation and hospital and emergency department (ED) utilization among Medicare patients ages 65 years and older. In a cohort of 18,557 Medicare members age 65 years and older at Kaiser Permanente Northwest, the authors compared rates of hospitalization and ED visits in the 12 months following a baseline survey between respondents who reported feeling lonely or socially isolated and those who did not, controlling for demographic and health variables and utilization in the 12 months prior to the survey. Statistical analysis was conducted in February 2020. In adjusted models, those who reported “sometimes” experiencing social isolation were more likely to have at least 1 hospital admission (odds ratio [OR_sometimes]: 1.17, 95% confidence interval [CI]: 1.01–1.35, P = 0.04), than those who “rarely” or “never” experienced social isolation. Those who experienced social isolation “sometimes” or “often/always” were more likely to have at least 1 ED visit (OR_sometimes: 1.28, 95% CI: 1.15–1.41, P < 0.0001, and OR_often/always: 1.51, 95% CI: 1.25–1.84, P < 0.0001, respectively) than those who “rarely” or “never” experienced social isolation. These findings suggest that self-reported social isolation may be predictive of future hospital admissions and ED utilization. Research is needed to determine how addressing social isolation needs within the health care system affects health care utilization and health outcomes.

Background

Social isolation, whether related to living alone or having few social network connections and infrequent social contact, is an important social risk factor for older adults.^1
–3 Social isolation is consistently associated with poor health outcomes^4

–8 and increased mortality.^9

–13 Social isolation is particularly prevalent in older populations, affecting an estimated 17%-24% of older adults.^14,15

Although previous research¹⁶ has shown that documented social and economic needs in the electronic health record are associated with future emergency department (ED) and hospital utilization, limited research has examined the specific effects of social isolation on health care utilization. Understanding the relationship between social isolation and health care utilization could lead to the development of cost-effective health system interventions that improve health outcomes. This study examined the relationship between social isolation and future hospital admissions and ED visits in patients ages 65 years and older with Medicare coverage.

Methods

Analytic cohort

This study included data from 18,557 Medicare Advantage members (ages 65 years or older) of Kaiser Permanente Northwest (KPNW), an integrated health system in Oregon and southwest Washington. Patients were included if they completed an operations-based survey, the Medicare Total Health Assessment (MTHA), between January 1, 2016, and June 30, 2018, and were enrolled at KPNW for 12 months prior to and following survey completion. MTHA has been implemented in several Kaiser Permanente regions (including KPNW) since 2012 and the domains of the survey have been described previously.¹⁷ The purpose of the survey is to assess required and optional domains as mandated by the Centers for Medicare & Medicaid Services for annual Medicare wellness visits. During these visits, Medicare beneficiaries and health providers develop a collaborative care plan, identify concerns about physical and mental health needs, and address other issues such as memory concerns and falls.^18
–20 Although CMS does not require assessment of social determinants of health (SDOH) during these visits, SDOH assessments are encouraged.²¹

Patients were eligible for the MTHA survey if they had a scheduled Medical Wellness Visit (MWV) during this period. Survey administration occurred either: (1) at the time of MWV or (2) prior to the visit using MWV (using electronic email administration or interactive voice recognitive technologies). The research team did not have specific information by mode of survey administration or about nonresponders to the survey. For patients who had completed >1 MTHA in the study time period, the first MTHA was used as the index survey for all analyses. This study was approved by the KPNW Institutional Review Board.

Outcome measures

The outcome measures of interest were inpatient hospital admissions and ED visits in the 12 months following the index MTHA (0 vs ≥1). Both variables were drawn from KPNW's electronic health record (EHR).

Independent variable

The independent variable was social isolation in response to the MTHA question, “How often do you feel lonely or isolated from those around you?” Responses were reported using a 5-point Likert scale: always, often, sometimes, rarely, or never. To obtain sufficient sample size and to analyze social isolation in the direction from most isolated vs. least isolated, results were consolidated into always/often, sometimes, or rarely/never. The social isolation question was adapted from the Patient-Reported Outcome Measurement Information System v 1.0 item bank and has been shown to have good psychometric properties.^22,23

Covariate measures

Covariate measures were all drawn from the EHR at the time of the index MTHA. Demographic variables included age, sex, race (white vs. nonwhite), and Area Deprivation Index (ADI).²⁴ The research team calculated the Charlson comorbidity index (CCI) for each patient,^25,26 categorizing them as having 0, 1, or ≥2 comorbidities documented in the 12 months prior to the index MTHA. Finally, prior health care utilization was assessed by measuring whether individuals had any hospital admissions, ED visits, or primary care visits in the 12 months prior to the index MTHA. All utilization variables were categorized as 0 vs ≥1.

Statistical analyses

Statistical analysis was conducted in February 2020. Multivariable logistic regression models were used to assess the association between social isolation with (1) future hospitalization and (2) ED utilization, separately, adjusting for all covariate measures. Odds ratios (ORs) and 95% confidence intervals (CIs) are reported for regression estimates. Statistical significance was defined as alpha <.05.

Results

Population characteristics

Table 1 shows population characteristics. Cohort members were predominantly white and averaged 73.4 years of age. They represented the full range of ADI values (range 1–10) and 42% had a CCI of ≥1. Across the full population, 6.2% (N = 1148) had at least 1 hospital admission in the year prior to survey completion, and 14.2% (N = 2642) had at least 1 ED visit. On the index MTHA, approximately 3.2% (n = 602) reported that they always or often feel socially isolated, 14.7% (n = 2725) reported feeling socially isolated sometimes, and 82.1% (n = 15,230) reported rarely or never feeling socially isolated.

Table 1.

Population Characteristics by Social Isolation Measure

	Social isolation			All	P
	Never/rarely	Sometimes	Often/always	All
	N = 15,230	N = 2725	N = 602	N = 18,557
	Mean +/-SD or %	Mean +/-SD or %	Mean +/-SD or %	Mean +/- SD %
Age (mean, standard deviation)	73.3 +/-6.5	73.9 +/- 7.1	73.5 +/- 7.0	73.4 +/- 6.6	<.0001
Female	54.7	64.9	65.5	56.5	<.0001
White	94.5	93.4	93.9	94.3	0.0741
Area Deprivation Index)					<.0001
1–3 (least disadvantaged)	40.1	35.2	33.2	39.1
4–6	33.8	33.8	33.1	33.8
7–10 (most disadvantaged)	24.3	29.1	31.9	25.3
Unknown	1.8	1.9	1.8	1.8
Charlson comorbidity index					<.0001
0	58.4	51.3	47.0	57.0
1	16.9	17.8	16.3	17.0
≥2	24.6	30.9	36.7	25.9
≥1 hospital admissions	5.8	7.7	9.5	6.2	<.0001
≥1 ED visits	13.0	18.8	25.9	14.2	<.0001
≥1 primary care visits	71.5	76.1	79.1	72.5	<.0001

Results based on 18,557 Medicare Advantage members (ages 65 years or older) of KPNW who completed an operations-based survey, the Medicare Total Health Assessment, between January 1, 2016, and June 30, 2018, and were enrolled at KPNW for 12 months prior to and following survey completion.

ED, emergency department; KPNW, Kaiser Permanente Northwest; SD, standard deviation.

Relationship between social isolation and hospitalizations

Social isolation was significantly associated with hospital admissions (P = .03) in a fully adjusted multivariate regression model. Patients who reported sometimes feeling socially isolated had significantly greater odds of having a hospital admission than those who reported social isolation rarely/never (OR_sometimes = 1.17, 95% CI = 1.01–1.35, P = .04); those who reported always/often feeling socially isolated also had a greater odds estimate of hospital admission compared to those who reported rarely/never feeling isolated, though this difference was not significant (P = .06) (Table 2). Seven of the 8 covariate measures in this model were significantly associated with hospital admissions, with the exception of ADI, which was not significant. Those with an ADI score between 7–10 were significantly more likely to have hospital admissions compared to the those with an ADI score between 1–3. There was no significant difference between 2 ADI categories and ADI score between 1–3: (1) ADI score between 4–6 and (2) unknown ADI values.

Table 2.

Logistic Regression Results: Association of Social Isolation with Hospital Admissions and Emergency Department Utilization

Social isolation measure and population characteristics	≥1 Hospital admissions				≥1 ED visits
Social isolation measure and population characteristics	OR	95% CI		P	OR	95% CI		P
Social Isolation				0.03				<.0001
Never/Rarely	reference				reference
Sometimes	1.17	1.01	1.35	0.04	1.28	1.15	1.41	<.0001
Often/Always	1.30	0.99	1.69	0.06	1.51	1.25	1.84	<.0001
Age (continuous)	1.04	1.03	1.05	<.0001	1.05	1.04	1.05	<.0001
Female (reference = Male)	0.85	0.77	0.95	0.00	1.00	0.93	1.08	0.96
White (reference = Nonwhite)	1.55	1.18	2.04	0.00	1.10	0.93	1.30	0.28
Area Deprivation Index				0.04				0.95
1–3 (least disadvantaged)	reference				reference
4–6	1.03	0.90	1.17	0.70	0.98	0.89	1.07	0.59
7–10 (most disadvantaged)	1.20	1.05	1.37	0.01	0.98	0.89	1.08	0.63
Unknown	1.03	0.68	1.56	0.89	0.97	0.72	1.30	0.83
Charlson comorbidity index				<.0001				<.0001
0	reference				reference
1	1.37	1.17	1.60	0.00	1.30	1.17	1.45	<.0001
≥2	1.95	1.71	2.23	<.0001	1.78	1.62	1.96	<.0001
≥1 hospital admissions	1.63	1.36	1.96	<.0001	1.09	0.94	1.27	0.24
≥1 ED visits	1.62	1.40	1.86	<.0001	2.30	2.08	2.55	<.0001
≥1 primary care visits	1.18	1.03	1.36	0.02	1.21	1.10	1.34	0.0001

Results based on 18,557 Medicare Advantage members (ages 65 years or older) of KPNW who completed an operations-based survey, the MTHA, between January 1, 2016, and June 30, 2018, and were enrolled at KPNW for 12 months prior to and following survey completion. Multivariable logistic regression models were used to assess the association between social isolation with (1) future hospitalization and (2) ED utilization, separately, adjusting for prior health care utilization by measuring whether individuals had any hospital admissions, ED visits, or primary care visits in the 12 months prior to the index MTHA. All utilization variables were categorized as 0 vs ≥1.

CI, confidence interval; ED, emergency department; KPNW, Kaiser Permanente Northwest; MTHA, Medicare Total Health Assessment; OR, odds ratio.

Relationship between social isolation and ED visits

Social isolation was significantly associated with future ED visits (P < .0001) (Table 2). Patients who responded sometimes and those who responded always/often to the social isolation question had significantly greater odds of having at least 1 ED visit in the following year than those who responded rarely/never (Table 2). Age, CCI, and visits to the ED or to primary care in the prior year also were significantly associated with ED visits.

Discussion

This study found that self-reported social isolation was associated with future hospital admissions and future ED visits among a cohort of Medicare members. These findings support a robust literature showing that social isolation can have profound impacts on health outcomes^6,8 and mortality for older adults.⁹ These results are novel for 2 reasons. It is the first study to find that social isolation was predictive of future hospital admission and ED utilization among a population of Medicare Advantage members ages 65 and older. Second, the relationship between social isolation and utilization remained significant, even after adjusting for a comprehensive set of demographic and clinical measures, including age, sex, race, ADI, CCI score, and prior health care utilization.

Seminal research suggests several reasons for the relationship between social isolation and poor health outcomes among older adults, including that social support acts as a buffer to stress²⁷ and that social support helps patients identify the need for medical intervention and for seeking care.^27,28 In the absence of this social support, older adults may have poorer health and put off care until illness becomes critical, leading to more ED visits and hospital care. These findings also are consistent with a recent systematic review that found weaker social relationships were associated with higher hospital readmissions.²⁹

Present study results showed a stronger association between social isolation and future ED visits than with future hospital utilization. This is consistent with prior research showing that elderly people without family in the area were 7 to 30 times more likely to use the ED,³⁰ and that loneliness is associated with emergency hospitalizations but not planned hospitalizations.³¹ This may be because of the role of social and material support in referring patients to care. Without others to help them seek care and transport them to appointments, socially isolated older adults may be more likely to wait until their care requires emergency attention, or they may rely on EDs to serve as referral systems³⁰ for nonemergent care.

Limitations

Study findings should be considered in light of certain limitations. First, the study population of Medicare members in an integrated health care system may not be generalizable to Medicare beneficiaries in fee-for-service settings. Second, no information was available regarding nonrespondents to the MTHA and, thus, the research team was not able to account for potential nonresponse bias. By pooling multiple years of MTHA data across years, the team aimed to gather responses from a large population of Medicare beneficiaries who reflect a representative sample the KPNW membership. Third, data on social isolation were collected via self-report and may be subject to recall and social desirability biases. Finally, the team treated hospital utilization and ED visits as binary variables; factors such as length of hospital stay may be more strongly influenced by “objective” measures of isolation, while admissions may be higher for patients who experience more “subjective” isolation.³²

Conclusion

These results highlight the potential of social isolation to influence future health care utilization in older patients. Programs that address social isolation and loneliness, such as fitness programs for older adults,³³ may be cost-effective interventions as they are likely to reduce health care expenditures. Future research is needed to assess whether such interventions can reduce health care utilization and improve health outcomes for older adults.

Footnotes

Acknowledgment

Results of this analysis were presented at the Annual Health Care System Research Network (HCSRN) Conference: April 8–10, 2020.

Author Disclosure Statement

The authors declare that there are no conflicts of interest.

Funding Information

Funding to complete this work was provided internally by Kaiser Permanente Northwest.

References

Merchant

, Liu

, Lim

, Fu

, Chan

. Factors associated with social isolation in community-dwelling older adults: a cross-sectional study. Qual Life Res 2020. DOI: 10.1007/s11136-020-02493-7

Arons

, DeSilvey

, Fichtenberg

, Gottlieb

. Documenting social determinants of health-related clinical activities using standardized medical vocabularies. JAMIA Open, 2019; 2:81–88.

National Academies of Sciences, Engineering, and Medicine. Social isolation and loneliness in older adults: opportunities for the health care system. Washington, DC: The National Academies Press, 2020.

Cacioppo

, Hawkley

. Social isolation and health, with an emphasis on underlying mechanisms. Perspect Biol Med, 2003; 46(3 suppl):S39–S52.

Cornwell

, Waite

. Social disconnectedness, perceived isolation, and health among older adults. J Health Soc Behav, 2009; 50:31–48.

Coyle

, Dugan

. Social isolation, loneliness and health among older adults. J Aging Health, 2012; 24:1346–1363.

World Health Ofganization. Social determinants of health: The solid facts. Copenhagen: WHO Regional Office for Europe, 2003.

Hammig

. Health risks associated with social isolation in general and in young, middle and old age. PLoS One, 2019; 14:e0219663.

Holt-Lunstad

, Smith

, Baker

, Harris

, Stephenson

. Loneliness and social isolation as risk factors for mortality: a meta-analytic review. Perspect Psychol Sci, 2015; 10:227–237.

10.

Holt-Lunstad

, Smith

, Layton

. Social relationships and mortality risk: a meta-analytic review. PLoS Med, 2010; 7:e1000316.

11.

Tanskanen

, Anttila

. A prospective study of social isolation, loneliness, and mortality in Finland. Am J Public Health, 2016; 106:2042–2048.

12.

Laugesen

, Baggesen

, Schmidt

SAJ

, et al. Social isolation and all-cause mortality: a population-based cohort study in Denmark. Sci Rep, 2018; 8:4731.

13.

Alcaraz

, Eddens

, Blase

, et al. Social isolation and mortality in US black and white men and women. Am J Epidemiol, 2019; 188:102–109.

14.

Ortiz

. Crossing new frontiers: benefits access among isolated seniors. 2011. https://www.ncoa.org/resources/crossing-new-frontiers/ Accessed April 16, 2020 .

15.

Cudjoe

TKM

, Roth

, Szanton

, Wolff

, Boyd

, Thorpe

. The epidemiology of social isolation: national health and aging trends study. J Gerontol B Psychol Sci Soc Sci, 2020; 75:107–113.

16.

Mosen

, Banegas

, Benuzillo

, Hu

, Brooks

, Ertz-Berger

. Association between social and economic needs with future healthcare utilization. Am J Prev Med, 2020; 58:457–460.

17.

Steiner

, Stenmark

, Sterrett

, et al. Food insecurity in older adults in an integrated health care system. J Am Geriatr Soc, 2018; 66:1017–1024.

18.

Chung

, Lesser

, Lauderdale

, Johns

, Palaniappan

, Luft

. Medicare annual preventive care visits: use increased among fee-for-service patients, but many do not participate. Health Aff (Millwood), 2015; 34:11–20.

19.

, Jensen

, Nerenz

, Tarraf

. Medicare's annual wellness visit in a large health care organization: who is using it?. Ann Intern Med, 2015; 163:567–568.

20.

Ganguli

, Souza

, McWilliams

, Mehrotra

. Trends in use of the US Medicare annual wellness visit, 2011–2014. JAMA, 2017; 317:2233–2235.

21.

Partnership for

Prevention

, Thomson Reuters (Healthcare) Inc. Guidance on development of health risk assessment as part of the annual wellness visit for Medicare beneficiaries—(Section 4103 of the Patient Protection and Affordable Care Act). General Proceedings from a Public

Forum

, Expert

Input

, and the Research Literature for the Design of Patient-Centered Health Assessments: Final

Report

. 2011. https://www.cdc.gov/policy/hst/hra/hraawvguidancereportfinal.pdf Accessed June 4, 2020 .

22.

Hahn

, DeWalt

, Bode

, et al. New English and Spanish social health measures will facilitate evaluating health determinants. Health Psychol, 2014; 33:490–499.

23.

Cella

, Riley

, Stone

, et al. The Patient-Reported Outcomes Measurement Information System (PROMIS) developed and tested its first wave of adult self-reported health outcome item banks: 2005–2008. J Clin Epidemiol, 2010; 63:1179–1194.

24.

Tuliani

, Shenoy

, Parikh

, Jutzy

, Hilliard

. Impact of Area Deprivation Index on coronary stent utilization in a Medicare nationwide cohort. Popul Health Manag, 2017; 20:329–334.

25.

Kieszak

, Flanders

, Kosinski

, Shipp

, Karp

. A comparison of the Charlson comorbidity index derived from medical record data and administrative billing data. J Clin Epidemiol, 1999; 52:137–142.

26.

Deyo

, Cherkin

, Ciol

. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol, 1992; 45:613–619.

27.

Gourash

. Help-seeking: a review of the literature. Am J Community Psychol, 1978; 6:413–423.

28.

Bass

, Noelker

. The influence of family caregivers on elder's use of in-home services: an expanded conceptual framework. J Health Soc Behav, 1987; 28:184–196.

29.

Valtorta

, Moore

, Barron

, Stow

, Hanratty

. Older adults' social relationships and health care utilization: a systematic review. Am J Public Health, 2018; 108:e1–e10.

30.

Coe

, Wolinsky

, Miller

, Prendergast

. Elderly persons without family support networks and use of health services. A follow-up report on social network relationships. Res Aging, 1985; 7:617–622.

31.

Molloy

, McGee

, O'Neill

, Conroy

. Loneliness and emergency and planned hospitalizations in a community sample of older adults. J Am Geriatr Soc, 2010; 58:1538–1541.

32.

Shaw

, Farid

, Noel-Miller

, et al. Social isolation and Medicare spending: among older adults, objective social isolation increases expenditures while loneliness does not. J Aging Health, 2017; 29:1119–1143.

33.

Brady

, D'Ambrosio

, Felts

, Rula

, Kell

, Coughlin

. Reducing isolation and loneliness through membership in a fitness program for older adults: implications for health. J Appl Gerontol, 2020; 39:301–310.