Social Connectedness Among Medicare Beneficiaries Following the Onset of the COVID-19 Pandemic

Abstract

This cross-sectional study examines social connectedness among Medicare beneficiaries during the COVID-19 pandemic.

Social distancing is an effective strategy to limit contagion and mortality from the COVID-19 pandemic. However, these measures may also decrease perceived social connectedness and conversely increase social isolation, states which are associated with psychologic and physiologic morbidity.^1,2,3,4,5 Patients who are elderly or have disabilities are particularly encouraged to practice social distancing given their higher risk of severe COVID-19 infection, but they also have a higher baseline risk for reporting social isolation.^5,6 We hypothesized that the pandemic and social distancing measures have negatively affected perceptions of social connectedness among these high-risk patients.

Methods

The Medicare Current Beneficiary Survey is an in-person, nationally representative survey of Medicare beneficiaries that is sponsored by the Centers for Medicare & Medicaid Services. We used the COVID-19 Summer 2020 Supplement data collected from June 10, 2020, to July 15, 2020. The institutional review board of Yale University approved the study and granted a waiver of informed consent because of the use of publicly available, deidentified data. The study cohort consisted of 9634 respondents who represented a weighted 50 851 437 community-dwelling beneficiaries who reported on changes in social connectivity during the pandemic (eMethods in the Supplement). The outcome variable, reporting decreased social connectedness, was defined as a beneficiary stating that that they felt “less socially connected to family and friends” since the onset of the pandemic.

We selected variables that we hypothesized were potentially associated with decreased social connectedness (eMethods in the Supplement). These variables included demographic and socioeconomic characteristics, adherence to 6 specific COVID-related social distancing precautions, ownership and use of technologies that allow for remote social engagement, and medical comorbidities and care receipt that may make beneficiaries more dependent on social contacts for health reasons. We used univariable and multivariable logistic regression to identify factors associated with reported decreased social connectedness since the start of the pandemic. Predicted probabilities (PPs) were estimated using population-averaged estimates, with 95% CIs calculated using the δ method. Statistical analyses were conducted using Stata, version 13.1 (StataCorp), and statistical significance was determined from contrast estimates using the χ² statistic. The multivariable model included variables reaching a significance level of P < .10 on univariable analysis.

Results

Since the start of the COVID-19 outbreak, a weighted 36.7% of enrollees reported feeling less socially connected with friends and family. Decreased social connection was reported by 41.2% of female, 31.2% of male, 24.6% of Black non-Hispanic, and 38.9% of White non-Hispanic beneficiaries (Table 1). On multivariable analysis (Table 2), demographic characteristics that were independently associated with feeling less socially connected included female (PP, 40.4%; 95% CI, 38.6%-42.2%) vs male sex (PP, 33.2%; 95% CI, 31.2%-35.1%; P < .001), White non-Hispanic (PP, 38.1%; 95% CI, 36.6%-39.7%) vs Black non-Hispanic race/ethnicity (PP, 30.4%; 95% CI, 25.9%-34.9%; P < .001), income of $25 000 or greater (PP, 38.5%; 95% CI, 36.8%-40.1%) vs less than $25 000 (PP, 34.2%; 95% CI, 31.4%-37.1%; P = .02), no Medicaid eligibility (PP, 38.1%; 95% CI, 36.5%-39.6%) vs full Medicaid dual eligibility (PP, 28.6%; 95% CI, 23.5%-33.7%; P = .002), history of non–skin cancer (PP, 40.8%; 95% CI, 38.0%-43.6%) vs none (PP, 36.3%; 95% CI, 34.9%-37.7%; P = .005), and depression (PP, 41.6%; 95% CI, 39.0%-44.3%) vs none (PP, 35.5%; 95% CI, 33.9%-37.1%; P < .001). Behaviors that were associated with feeling less socially connected included performing all 6 assessed anti–COVID-19 social distancing measures (PP, 38.8%; 95% CI, 37.4%-40.2%) vs 4 or fewer (PP, 30.5%; 95% CI, 26.2%-34.8%; P < .001), computer ownership (PP, 38.4%; 95% CI, 36.6%-40.2%) vs not (PP, 34.4%; 95% CI, 32.2%-36.6%; P = .01), and use of internet teleconferencing software (PP, 40.2%; 95% CI, 38.5%-42.0%) vs not (PP, 33.3%; 95% CI, 31.1%-35.5%; P < .001). Among other variables, metropolitan (PP, 37.6%; 95% CI, 36.1%-39.1%) vs nonmetropolitan residence (PP, 35.9%; 95% CI, 33.3%-38.5%; P = .23) and geographic region were not independently associated with reported social connectivity.

Table 1. Weighted Rates of Decreased Social Connectedness Since the Start of the COVID-19 Pandemic Among Selected Subpopulations.

Variable name	Variable categories	%
		Percentage of overall cohort in specified category	Percentage of beneficiaries within specified category who reported feeling less socially connected
Age group, y	<65	13.7	35.4
	65-74	55.8	39.1
	>74	30.5	33.4
Sex	Men	44.0	31.3
	Women	56.0	41.2
Race/ethnicity	White non-Hispanic	76.7	38.9
	Black non-Hispanic	9.9	24.6
	Hispanic	7.3	34.4
	Asian, American Indian or Alaska Native, Native Hawaiian or other Pacific Islander, or unknown	6.0	33.9
Metropolitan residence	No	20.0	33.3
	Yes	80.0	37.7
Geographic region	West	22.0	39.4
	Northeast	17.7	37.7
	Midwest	22.2	40.6
	South	38.1	32.7
Annual income, $	<25 000	28.9	29.5
	≥25 000	69.0	40.2
	Do not know	2.1	27.9
Non-English language spoken at home	No	90.2	37.1
	Yes	9.8	34.0
Medicaid dual eligible in 2019	No	86.8	38.1
	Fully dual eligible	7.9	26.6
	Partially dual eligible	3.0	31.2
	QMB eligible only	2.3	30.7
Primary location care received	Physician’s office/medical clinic	82.4	37.7
	Managed care plan center/HMO	3.7	44.5
	Other clinic (eg, rural, company)	1.6	38.2
	Urgent care/emergency department	4.3	29.5
	Veterans Affairs facility	2.8	23.4
	Other type of location	0.4	25.2
	No regular care at a location	4.9	31.4
Own computer	No	31.7	28.8
	Yes	68.3	40.6
Own smartphone	No	29.0	29.5
	Yes	71.0	39.8
Access to internet	No	15.7	26.7
	Yes	84.3	38.7
Used video or voice calls over the internet	No	46.3	29.2
	Yes	53.7	43.4
Missed any health appointment because of COVID-19	No	78.9	34.4
	Yes	21.2	45.8
Total COVID-19 precautions taken	≤4	9.5	29.0
	5	17.2	32.1
	6	73.3	39.9
Wore a face mask because of COVID-19	No	4.4	20.8
	Yes	95.6	37.7
Avoided contact with people with illness because of COVID-19	No	4.5	27.1
	Yes	95.5	37.5
Kept a 6-foot distance between self and others because of COVID-19	No	5.1	27.5
	Yes	94.9	37.5
Avoided large gatherings of people because of COVID-19	No	8.0	24.4
	Yes	92.0	38.2
Left home for essential purposes only because of COVID-19	No	13.9	32.0
	Yes	86.1	37.8
Avoided other people as much as possible	No	10.0	27.9
	Yes	90.0	37.9
Traditional news sources used for information on COVID-19	No	8.4	32.3
	Yes	91.7	37.2
Social media used for information on COVID-19	No	68.4	36.5
	Yes	31.6	37.6
Comments or guidance from government officials used for information on COVID-19	No	33.8	31.1
	Yes	66.2	39.7
Other webpages/internet used for information on COVID-19	No	52.3	31.9
	Yes	47.8	42.2
Friends or family used for information on COVID-19	No	46.9	37.1
	Yes	53.1	36.5
Health care clinicians used for information on COVID-19	No	50.7	35.0
	Yes	49.3	38.7
History of any immunodeficiency	No	81.3	35.7
	Yes	18.7	41.6
History of myocardial infarction	No	90.6	37.2
	Yes	9.5	33.0
History of congestive heart failure	No	94.5	36.9
	Yes	5.5	35.2
History of stroke	No	91.3	36.8
	Yes	8.7	36.9
History of dementia	No	98.5	37.0
	Yes	1.5	27.5
History of depression	No	73.5	35.1
	Yes	26.6	41.6
History of chronic obstructive pulmonary disease/asthma	No	80.9	36.8
	Yes	19.1	36.9
History of diabetes	No	66.3	37.6
	Yes	33.7	35.3
History of cancer	No	80.3	35.7
	Yes	19.7	41.4

Abbreviations: HMO, health maintenance organization; QMB, Qualified Medicare Beneficiary.

Table 2. Selected Characteristics and Behaviors Statistically Significantly Associated With Reporting Decreased Social Connectedness on Multivariable Analysis.

Variable name	Variable categories	Predicted probabilities, % (95% CI)	P value
Sex	Men	33.2 (31.2-35.1)	1 [Reference]
	Women	40.4 (38.6-42.1)	<.001
Race/ethnicity	White non-Hispanic	38.0 (36.5-39.5)	1 [Reference]
	Black non-Hispanic	30.3 (25.9-34.8)	.001
	Hispanic	38.9 (34.5-43.2)	.72
	Asian, American Indian or Alaska Native, Native Hawaiian or other Pacific Islander, or unknown	34.4 (28.2-40.5)	.27
Annual income, $	<25 000	34.2 (31.4-37.1)	1 [Reference]
	≥25 000	38.5 (36.8-40.1)	.02
	Do not know	31.7 (24.8-38.6)	.51
Medicaid dual eligible in 2019	No	38.0 (36.5-39.6)	1 [Reference]
	Fully dual eligible	28.6 (23.5-33.6)	.001
	Partially dual eligible	35.9 (27.3-44.5)	.65
	QMB eligible only	36.6 (26.6-46.5)	.78
Primary location of care receipt	Physician’s office/medical clinic	37.6 (36.2-38.9)	1 [Reference]
	Managed care plan center/HMO	43.8 (37.3-50.4)	.06
	Other clinic (eg, rural, company)	39.6 (29.0-50.2)	.71
	Urgent care/emergency department	27.1 (22.2-32.0)	<.001
	Veterans Affairs facility	28.9 (20.9-36.8)	.04
	Other type of location	29.4 (5.2-53.7)	.51
	No regular care at a location	52.5 (25.4-79.7)	.28
Own computer	No	34.4 (32.2-36.6)	1 [Reference]
	Yes	38.4 (36.6-40.2)	.01
Used video or voice calls over the internet	No	33.3 (31.1-35.4)	1 [Reference]
	Yes	40.3 (38.6-42.0)	<.001
Missed any health appointment because of COVID-19	No	35.7 (34.3-37.1)	1 [Reference]
	Yes	42.4 (39.4-45.4)	<.001
Total COVID-19 precautions taken	≤4	30.6 (26.3-34.9)	1 [Reference]
	5	33.6 (31.1-36.2)	.17
	6	38.7 (37.3-40.2)	.001
Other webpages/internet used for information on COVID-19	No	35.2 (33.2-37.3)	1 [Reference]
	Yes	39.1 (37.0-41.3)	.02
History of depression	No	35.5 (33.9-37.1)	1 [Reference]
	Yes	41.7 (39.0-44.3)	.001
History of cancer	No	36.3 (34.8-37.7)	1 [Reference]
	Yes	40.8 (38.0-43.6)	.004

Abbreviations: HMO, health maintenance organization; QMB, Qualified Medicare Beneficiary.

Discussion

More than one-third of Medicare beneficiaries reported feeling less socially connected to friends and family since the start of the COVID-19 pandemic. Medicare beneficiaries who were women, had higher incomes, were not of Black non-Hispanic race/ethnicity, and had a history of cancer or depression were more likely to report a negative association of the pandemic with perceptions of social connectedness. The likelihood of reporting decreased social connection was widespread nationally and was associated with practicing more social distancing measures. The limitations of this study include an inability to assess the magnitude of the decreased connection reported and survey exclusion of the subset of Medicare-beneficiaries who are not community dwelling. The public health benefits and psychosocial costs of prolonged social distancing measures should be balanced carefully in this doubly vulnerable population.

Supplement.

eMethods. Supplementary methodological description