Living Alone With Cognitive Impairment

Abstract

Although most individuals experiencing cognitive impairment (CI) reside with a caregiver, an estimated 800,000 live alone. Such individuals may have an increased risk for injury to self or others through self-neglect as a result of the CI symptoms. While persons living alone with CI have been identified as an important area for needed research, few studies have been able to examine this population due to the challenges of identifying and recruiting study participants. By using the National Health & Aging Trends Study data set, the researchers explored the characteristics to describe this population. The results of this study indicated that the majority of persons living with CI were older, widowed females who were not diagnosed with Alzheimer’s or dementia but tested positive on cognitive screening measures. Further, the majority of persons living alone with CI relied on adult children and paid professionals as the primary care providers.

Introduction

Worldwide, there are an estimated 46.8 million individuals living with some form of cognitive impairment (CI). ¹ CI is a general term describing a group of disorders in which memory and thought processes become impaired. CI can be caused by many different dementias and disorders, but the majority of CI cases are caused by Alzheimer’s disease. ²

According to 2014 data from the Administration on Aging, just under 30% of Americans aged 65 or older (approximately 12.5 million people) live alone. The proportion of people who live alone increases with age and is higher among women than men. More than 56% of American women of at least 85 years old live alone. ³ While most individuals diagnosed with CI live with a caregiver, as many as 1 out of 7 persons with CI live alone. ⁴ Estimates have been as high as 800,000 individuals.

Over the years, researchers have explored the notion of “risk” for persons living alone with CI. Individuals living alone with CI are at an increased risk for self-neglect, ^{5 -15} delayed diagnosis, ^11,12 falling, ¹⁵ use of emergency services, ^8,14 undertreated psychiatric issues, ^{16 -19} and wandering. ⁸ Many individuals who live alone with CI lack support from community services that might help them with their activities of daily living (ADLs). ^6,7

We lack data, and especially national data, about the characteristics of persons living alone with CI. In this study, our intent is to identify the characteristics representing this subgroup and add to extent knowledge about the characteristics of persons with Alzheimer’s disease. A broader and more inclusive understanding of individuals with CI will increase service providers’ outreach to those with CI who need the most support. In addition, by enhancing our understanding of individuals with CI who live alone, we can identify those who are most vulnerable to social isolation and take into account their social networks and care providers and create programs to support these families. The following research questions guided this investigation: (1) What are the characteristics of persons living alone with CI? and (2) Who do persons living alone with CI identify as their primary care providers?

Sample

We used participants from the first wave of the National Health & Aging Trends Study (NHATS), which is a cross-sectional survey design supported by the National Institute on Aging. Trained professionals interviewed participants at their homes. In 2012, Johns Hopkins Bloomberg School of Public Health released the NHATS data set comprised of 8,245 participants, all of whom were drawn from Medicare enrollees in the United States and were aged 65 or older. A subsample of these data was used for this analysis based on the following criteria. First, we selected those who reported living alone based on the response to the question how many participants lived in their household. Those who responded telling 1 person resided in their household were selected. Second, we used the dementia classification constructed for this study, which we describe in more detail below, to identify participants who were having “possible” or “probable” dementia. This yielded a sample of 710 individuals who lived alone and had either possible or probable dementia. Forty-six percent (N = 327) of those who lived alone were classified as having probable dementia, while 54% (N = 383) were identified as having possible dementia and lived alone. In 16% of the cases, when the sample person agreed, proxy respondents were used to respond to items although the questions were modified to fit this context. For example, the question was rephrased as, “Now, I’d like to ask some questions about whether you have noticed any changes in the sample person in the last several years/since you have known [him/her] due to thinking of memory problems.” Another example was, “I’ll read some statements about him/her caused by [him/her] thinking or memory problems, please say Yes. Otherwise say No.”

Instrumentation

Although the data from NHATS include information about participants’ disabilities, lifestyle, values, and other biopsychosocial factors, for this investigation we used the demographic and contextual variables and CI measures, which we discuss in more detail below.

Demographic and Contextual Variables

The demographic variables used for this study included participants’ age, gender (1 = “male” and 2 = “female”), marital status (1 = “married,” 2 = “living with a partner,” 3 = “separated,” 4 = “divorced,” 5 = “widowed,” and 6 = “never married”), race/ethnic background (asked as nonexclusive categories so respondents could select multiple races and/or ethnicities; 1 = “yes” and 2 = “no”), and self-reported health status (1 = “excellent,” 2 = “very good,” 3 = “good,” 4 = “fair,” and 5 = “poor”). In addition, questions regarding self-care, which often are referred to as ADLs, focused on whether or not a participant needed assistance with eating, getting cleaned up, using the toilet, and dressing (1 = “yes,” 2 = “no,” and 3 = “can’t do/doesn’t do”). Instrumental activities of daily living (IADLs; 1 = “yes” and 2 = “no”) assessed if participants needed help with laundry, shopping for groceries or other personal items, making hot meals, handling bills and banking, taking prescribed medications, and managing less common money matters. We computed the total scores, respectively, for ADLs and IADLs by adding participants’ responses, indicating that they needed help carrying out a particular activity.

Cognitive Impairment Measures

The CI assessment was designed to provide information about a participant’s memory (self-rated, whether memory interferes with daily activities, and immediate and delayed 10-word recall), orientation (date, month, year, day of week, and naming president and vice president), and executive function (clock drawing test). Scoring guidelines for the clock drawing test were based on those recommended by Schretlen, Testa and Pealson. ²⁰ Clocks were scored on a scale from 0, not recognizable as a clock, to 5, an accurate depiction of a clock. Extensive training was given for clock drawing test coders by neuropsychology specialists. The weighted kappa (κ) to assess interreliability between the 2 neuropsychology coders was 0.81. Two lay coders were also extensively trained, and their weighted κs ranged from 0.78 to 0.80.

In a technical report entitled, Classification of Persons by Dementia Status in the National Health and Aging Trends Study and published by the investigators of NHATS, 2 classifications—possible dementia and probable dementia—were constructed. ²¹ Previous researchers have identified a cutoff point of <1.5 standard deviations (SDs) below the mean on the cognitive measures to classify an individual as having CI. Following the criteria for diagnosing a dementia of Alzheimer’s disease, in order for a participant to meet the category of probable dementia, they would require impairment in at least 2 of the cognitive measures. The criteria used to classify possible dementia or mild cognitive impairment (MCI) were based on a cut point of <1.5 SDs below the mean in 1 (of any) domain (as opposed to multiple). ^22,23

Methods

This is a cross-sectional research design because we used the first wave of data from NHATS. We present descriptive frequencies and percentages for the demographic analyses as well as results from χ ² analyses used to test significant associations between variables. We used various strategies of imputation, specifically pairwise deletion, maximum likelihood, and full information maximum likelihood, to address missing data. However, most variables had less than 1% missing data. SPSS 20 software was used to analyze our data.

Results

The results from our first research question regarding the characteristics of participants who lived alone and had CI are shown in Table 1. Based on the criteria for CI used in NHATS and the question asking about the number of people in the household, we found that 710 participants (8.6%) lived alone and had CI. As shown in Table 1, these participants were mostly female and widowed, aged 80 or older, and from white backgrounds. About one-quarter of this subgroup reported African American backgrounds, and fewer self-identified as Hispanic. Less than 5% reported backgrounds that were American Indian, Asian American, and other while no one mentioned (0.0%) Alaskan Native, Native Hawaiian, and Pacific Island. Almost three-quarters (71.2%) of this subgroup was over the age of 80. Chi-square analysis revealed a significant difference between age and the likelihood of being classified as having possible or probable dementia.

Table 1.

Demographic Characteristics.^a

Characteristics	Percentage of Total Sample	Percentage of Probable Sample	Percentage of Possible Sample
Gender of sample person
Female	69.7	72.5	67.4
Male	30.3	27.5	32.6
Race of sample person
White	69.9	65.4	66.5
African American	27.5	28.7	26.4
American Indian	1.7	3.0	2.3
Alaskan native	0.0	0.0	0.0
Asian American	1.9	1.2	2.6
Native Hawaiian	0.0	0.0	0.0
Pacific Islander	0.0	0.0	0.0
Other race	2.8	2.8	2.9
Ethnicity of sample person
Hispanic/Latino	6.8	6.1	7.3
Age of the sample person
65-69	4.4	2.8	5.7
70-74	9.9	6.4	12.8
75-79	14.5	14.1	14.9
80-84	22.1	21.4	22.7
85-89	24.9	24.5	25.3
90 or older	24.2	30.9	18.5
Marital status
Married (living separate)	3.5	4.6	2.6
Separated	4.1	1.8	6.0
Divorced	13.5	11.3	15.4
Widowed	69.9	70.9	68.9
Never married	8.5	10.4	6.8

^aN = 710.

Table 2 illustrates the various levels of education among the sample. Chi-square analyses between education and the dementia classification were statistically significant. In consideration of the highest level of education, 41.0% (probable dementia = 43.1%, possible dementia = 39.2%) of the sample did “not graduate from high school.” Again, χ ² analysis revealed a significant relationship between “severity of impairment” and “highest level of education,” suggesting that those with higher education resulted in less CI, χ ² (10, N = 710) = 30.603, P = .001.

Table 2.

Sample Person’s Highest Level of Education.^a

Characteristics	Percentage of Total Sample	Percentage of Probable Sample	Percentage of Possible Sample
Highest level of education
No school	2.1	3.7	<1.0
1st-8th grade	21.7	22.0	21.4
9th-12th grade	17.2	17.4	16.9
High school graduate	28.6	27.8	29.2
Vocational training	5.8	7.0	4.7
Some college	8.5	5.8	10.7
Associate’s degree	1.5	2.1	1.0
Bachelor’s degree	8.3	7.0	9.3
Master’s or Professional degree	3.1	1.8	4.2

^aN = 710.

In regard to respondents’ health, the majority of the sample felt they were in good health. Less than 1% of the sample reported as “don’t know” or “refused” when asked about health conditions. Of the total respondents, 26.2% had been diagnosed with diabetes and 20.8% had been diagnosed with heart disease. Interestingly, only 16.3% had been diagnosed with Alzheimer’s or dementia—all of whom met the criteria for probable dementia (36.1% of that sample). The results of the sample’s self-reported health conditions are listed in Table 3.

Table 3.

Summary of Sample Person’s Health Conditions.^a

Characteristics	Percentage of Total Sample	Percentage of Probable Sample	Percentage of Possible Sample
Self-reported health
Excellent	7.6	6.4	8.6
Very good	21.1	19.3	22.7
Good	32.3	31.5	32.9
Fair	26.3	25.4	27.2
Poor	12.4	17.1	8.4
Presence of health conditions
Heart attack	19.9	21.4	18.5
Heart disease	20.8	22.6	19.3
Blood pressure	64.8	62.1	67.1
Arthritis	59.6	61.5	58.0
Osteoporosis	20.4	24.5	17.0
Diabetes	26.2	23.5	28.5
Lung disease	14.9	15.3	14.6
Had stroke	16.6	20.8	13.1
Dementia or Alzheimer’s  disease	16.3	16.3	0.0
Cancer	24.8	23.5	25.8

^aN = 710.

The interview also included questions regarding ADLs and instrumental activities of daily living (IADLs). With regard to the items of ADLs items, 5.1% of the sample were not able to eat independently, 11.5% were not able to get out of bed without help, 15.6% were not able to get out of a chair without help, 17.7% were not able to walk independently, 30.3% were not able to go outside without assistance, 12.5% were not able to get dressed without assistance, 19.9% were not able to bathe independently, and 11% were not able to use the toilet independently. In summary, the individuals who were suspected of possible dementia were much less impaired when it came to ADLs compared to the probable dementia group. Among the probable dementia group, 30% had difficulty dressing without assistance, and just under 40% could not go outside without assistance. Table 4 presents this information.

Table 4.

Summary of Sample Person’s Activities of Daily Living (ADLs).^a

Characteristics	Percentage of Total Sample	Percentage of Probable Sample	Percentage of Possible Sample
Sample person cannot ___ without help of another person or special equipment
Eat	5.1	7.6	2.9
Get out of bed	11.5	18.3	5.7
Get out of chair	15.6	23.5	8.9
Walk around inside	17.7	25.4	11.2
Go outside	30.3	38.8	23.0
Dress	12.5	19.3	6.8
Bathe	19.9	30.0	11.2
Use toilet	11.4	19.0	5.0

^aN = 710.

With respect to the items of IADLs, 26.8% were not able to prepare meals independently, 31.1% were not able to do laundry without assistance, 23.1% were not able to do light housework without assistance, 45.1% were not able to shop for groceries independently, 35.2% were unable to manage money independently, 27.0% were not able to take medication without assistance, and 14.4% were unable to make telephone calls without assistance. Again, the individuals who were suspected of possible dementia were much less impaired when it came to instrumental activities of daily living (IADLs) compared to the probable dementia group. These data are summarized in Table 5.

Table 5.

Summary of Sample Person’s Activities of Daily Living (IADLs).^a

Characteristics	Percentage of Total Sample	Percentage of Probable Sample	Percentage of Possible Sample
Sample person cannot ___ without help of another person or special equipment
Prepare meals	26.8	41.3	14.4
Do laundry	31.1	46.5	18.0
Do light housework	23.1	35.8	12.3
Shop for groceries	45.1	34.3	30.5
Manage money	35.2	54.4	18.8
Take medicine	27.0	44.6	12.0
Make telephone calls	14.4	25.4	5.0

^aN = 710.

Respondents were also asked to rate their memory (16.3% respondents with proxies were not asked this question). Of the self-respondents, 7.7% reported their “memory as excellent,” 18.7% reported their “memory as very good,” 28.7% reported “memory as good,” 20.7% reported “memory as fair,” and 7.7% reported “memory as poor.” When these same respondents were asked to compare their memory to 1 year ago, the majority (60.0%) of respondents felt their memory was about the same from the previous year. Those who did not rate their self-rated memory as excellent or very good were also asked how much of their memory problem interfered with their daily functioning. Of the 66.8% of the sample who were asked this question, 4.8% stated they experienced memory problems “daily,” 4.1% “most days,” 15.8% “some days,” 16.6% “rarely,” and 15.5% “never.”

The majority (75.5%) of persons living alone with CI had visited their friends and family in the last month, and most felt that these informal interactions were important to them. However, they were less involved in clubs and other organized activities and fewer felt this was important. Finally, when asked about going out for entertainment or enjoyment, most participants (54.2%) had gone out in the last month. They differed in how important they felt going out was in their lives. For example, about one-third of the participants felt it was “very important” to go out for enjoyment, whereas almost another third felt it was “somewhat important.” Overall those classified with a probable dementia were less socially engaged than those suspected of a possible dementia. See Table 6 for a more detailed presentation of these results.

Table 6.

Summary of Sample Person’s Social Participation.^a

Characteristics	Percentage of Total Sample	Percentage of Probable Sample	Percentage of Possible Sample
Sample person has visited friends or family in last month
Yes	75.5	73.7	77.0
Sample person’s health or functioning impacted his or her ability to visit family or friends in the last month
Yes	18.2	24.2	13.1
Importance of visiting friends/family for the sample person
Very important	53.0	51.7	54.0
Somewhat important	25.1	23.2	26.6
Not so important	20.6	22.9	18.5
Sample person has participated in clubs or activities in last month
Yes	24.1	19.3	71.5
Sample person’s health or functioning impacted his or her ability to participate in clubs or activities in the last month
Yes	19.9	25.4	84.6
Importance of participating in clubs or activities for the sample person
Very important	18.0	14.4	21.1
Somewhat important	22.5	21.7	23.2
Not so important	58.7	62.7	55.4
Sample person went out for entertainment in the last month
Yes	54.2	50.2	57.7
Sample person’s health or functioning impacted his or her ability to go out for entertainment in the last month
Yes	21.8	29.7	15.1
Importance of going out for entertainment to the sample person
Very important	31.3	28.4	33.8
Somewhat important	32.7	30.6	34.5
Not so important	35.2	40.1	31.1

^aN = 710.

Our second research question asked the following question: Who do persons living alone with CI identify as their primary care provider? These results are shown in Table 7. The majority (66.4%) of care providers were adult children: 39.7% were respondents’ daughters, 26.7% were respondents’ sons, and 1.7% were daughters-in-law. Another common care provider for this population was paid assistance (6.9%) and other paid staff (5.2%). Additionally, siblings and their children were a common proxy among the sample. Extended family were also mentioned. For example, 4.3% of care providers were nieces and 2.6% were nephews. Other types of care providers (<4%) who were noted included grandchildren, cousins, friends, and spouses who did not live with them.

Table 7.

Care Provider.^a

Characteristics	Percentage
Relationship to the sample person
Daughter	39.7
Son	26.7
Paid assistant	6.9
Staff	5.2
Sister	4.3
Niece	4.3
Spouse	2.6
Nephew	2.6
Daughter-in-law	1.7
Brother	<1
Grandchild	<1
Cousin	<1
Friend	<1

^aN = 116.

Discussion

As stated in the results, 69.7% were female and 71.2% of the subsample was over the age of 80. As expected, among the subsample the older an individual became, the more likely they were to be classified as having CI. Many studies have found that age is the greatest risk factor for the onset of Alzheimer’s disease. ²⁴ Our findings also corroborated previous studies that have shown that more women than men have CI, which in part is due to women’s higher life expectancies compared to men’s. ²⁵ Given that widows outnumber widowers, we expected higher rates of CI among bereaved women than among bereaved men. Future cohorts are likely to have higher numbers of people living alone with CI who were divorced or never married. ²⁶

Education

In contrast to younger cohorts, these older adults included a substantial percentage of persons who had not completed high school. The significant association we found between education and CI is consistent with previous studies, and at least 2 explanations could account for this. First, educational biases (ie, Mini-Mental State Examination) on neuropsychological tests are possible. ^{24,27 -30} For example, the clock drawing test has been criticized for inaccuracy particularly with education bias. ³¹ The validity of these tests are especially questionable Among those who are illiterate, which mostly include those with a grade school education. The tests do not capture what they are intended to capture–CI. Rather they capture these individual’s lower level of education. ^32,33 Previous studies have, indeed, found that reading disorders can lead to misdiagnosis of MCI. ³⁴ Those with lower levels of school years also may not have the problem-solving skills, executive functioning ability, or appropriate knowledge prior to taking the exam. These factors inevitably will generate “false positives” on such cognitive assessments.

A second and more recent explanation for the significant association between education and CI lies with the cognitive reserve framework. ³⁵ According to Stern, individuals with substantial cognitive reserve less often succumb to CI, or if they do, they succumb later than individuals with limited cognitive reserve. ³⁵ By cognitive reserve, Stern and others ^35,36 are referring to the consequences of a lifetime of intellectual enrichment as a result of many years of education, stimulating social interactions, and engagement in complex and challenging cognitive activities. The accumulation of these lifelong intellectual pursuits actually modifies the brain in such a way that it acts as a cognitive reserve, that is, a buffer to cognitive pathology, such as what occurs with Alzheimer’s disease. ^35,36 After comparing patients matched for clinical severity in Alzheimer’s disease, Stern found an inverse relationship between resting cerebral blood flow and years of education. ³⁵ Education seemed to modify the association between Alzheimer’s disease pathology and levels of cognitive functioning. Almeida and colleagues found that older adults with at least 16 years of education had less evidence of neurodegeneration in their cerebrospinal fluid compared to people with fewer years of education. ³⁶ More recent studies suggest that cognitive reserved can be strengthened after appropriate cognitive training. ³⁷

Dementia Classifications

Another important factor that could lead to inaccurate results lies with the sensitivity and specificity of the assessment tools. Interestingly, 16.6% (all from the probable dementia sample, which was 36.1%) had been diagnosed with Alzheimer’s or dementia. Possible explanations for this may be due to CI assessment tests not being sensitive enough to pick up MCI, which is a common issue discussed in the neurological literature. ^38,39 Another possible explanation is that some clinical assessments may be too sensitive—again resulting in false positive. At this time, it remains unclear as to how accurate the possible and probable dementia classifications are at truly reflecting the incident of CI in this national sample. Longitudinal analyses may help to expand our understanding of the effectiveness of neuropsychological tests as cognitive screenings, particularly among the population of those living alone.

Functional Impairment

As previously mentioned, ADLs and instrumental activities of daily living (IADLs) differed substantially between the 2 groups of probable and possible dementia—the individuals who were suspected of possible dementia were much less impaired when it came to ADLs compared to the probable dementia sample. This also suggests there may be a relationship between functional impairment and CI, but it would be important to consider age and other characteristics such as frailty as confounding factors.

According to the practice literature, general practitioners (GPs) are using the IADL scales to help them with CI disease diagnoses. In one paper, De Lepeleire and colleagues described the effectiveness of using it as a tool in addition to other cognitive assessments to help GPs more confidently diagnose CI in patients. ⁴⁰ Additionally, the IADL scale has been used to help determine the level of impairment of individuals by taking into account the hierarchy of “difficulty,” with “shopping” being the most difficult item and “telephone use” being the least difficult item. Shopping was the most discriminatory item differentiating between patients with different levels of ability from both a cognitive and functional perspective. ⁴¹ While it was originally assumed functional impairment might predict CI, new research suggests that CI precedes and predicts subsequent functional impairment in patients with mild Alzheimer’s disease. ⁴⁰ This could explain the disparate functional capacities we observed between the possible and probable dementia persons. ⁴²

Social Interactions

The majority (75.5%) of respondents “visited with friends and family in the last month.” However, a fifth (20.6%) of the sample felt such interactions were “not important at all.” Consistently a fourth-to-fifth of the sample reported a lack of social engagement. Some individuals (primarily those in the probable dementia sample) reported wanting social interactions but were hindered by their health or functional limitations. Finally some people might have less of a need for social interaction and enjoy more solitary activities, such as painting or playing a musical instrument. However, when a person changes and becomes less interested in previously enjoyed activities, some experts might view such a withdrawal as a symptom or precursor to CI or dementia. ⁴³

Conclusion

In this study, we examined a subsample of the NHATS data to explore the characteristics of older adults with CI who were living alone in their communities. Although the majority of these older adults had not been formally diagnosed with a CI, most of them tested positive, with 1 or 2 deficits on the cognitive assessment tools used in this study. With only a very small percentage officially diagnosed with Alzheimer’s disease or dementia (16.3%) within this subsample, researchers and practitioners need to explore more innovative ways to access and identify persons living alone who might be at risk for CI. While it may not always be appropriate when only a few deficits are present to provide a clinical diagnosis (such as Alzheimer’s disease), more GPs and geriatricians need more knowledge and training on when to assess CI in their patients and how to intervene when they identify people with CI. ⁴⁴

Researchers must investigate how these older persons meet their daily needs, given that more older persons with CI are expected to be living alone in the future. Many questions remain such as how do the older persons manage their shopping? How do they compensate for reductions in social interactions? How often do these adults end up in hospitals? How are communities supporting these older persons? We also need more studies examining the safety of these older persons to live alone. Finally, we lack data on evidence-based interventions specifically for this subgroup. If we create and implement programs designed to support these older persons’ independence, we will more effectively avoid premature institutionalizations, which, in turn, will lower costs. Most older adults prefer living out their later years in their homes, that is, they prefer to “age in place.” However, until we have more information about older persons with CI who live alone, about their characteristics, and the factors that might influence these older adults’ functioning, we will fail to meet these older persons’ needs and preferences. We will also continue to overspend on residential care that might be unnecessary or at least delayed.