LitCog Caregiver Cohort: a prospective, observational cohort study investigating US caregivers’ health literacy, self-care skills and cognitive function

Abstract

Introduction

Many older adults receive assistance in managing chronic conditions. Yet complicating the utility of caregiver support is whether caregivers have sufficient skills to aid in a patient’s self-care. Health literacy and cognition are important determinants of older adults’ health outcomes, but few studies have examined caregiver health literacy, cognition and self-care skills and their relations to patient outcomes.

Methods and analysis

We will expand an ongoing cognitive ageing cohort study (LitCog) to enroll a parallel caregiver cohort. Caregivers are eligible if they are (1) ≥18 years of age, (2) provided care for ≥6 months and (3) assisted with at least one activity of daily living, instrumental activity of daily living or health management task. Caregivers will complete interviews at time points corresponding with the LitCog participant interviews. Caregivers will complete assessments of health literacy, self-care skills, cognitive function, caregiver healthcare task difficulty, caregiver burden, caregiver self-efficacy, activation, technology use, busyness and routine and relationship quality. Caregivers will self-report the nature and intensity of care provided, and their own health status. Associations between caregiver presence and caregiver capacity with patient outcomes will be examined in a series of regression models, and mediating and moderating factors will be tested.

Ethics and dissemination

The Institutional Review Board at Northwestern University has approved the study protocol (STU00026255). Results will be published in peer-reviewed journals and summaries will be provided to the funders of the study as well as patients and caregivers.

Strengths and limitations of this study.

• This study will enroll caregivers to older adults participating in an ongoing cognitive ageing cohort study.

• Recruitment of caregivers requires a two-step consent process (first participants, followed by caregivers), and bias may exist between older adults whose caregivers do or do not participate in the study.

• Caregivers will complete measures of caregiver support, health literacy, cognition and health outcomes.

• As with many longitudinal studies, a primary limitation will be loss to follow-up and missing data that may introduce attrition bias.

• This study is only recruiting individuals from the USA, and findings may not be generalisable to older adults and caregivers living in other countries.

Introduction

Estimates suggest that between 62% and 80% of older adults are managing multiple (≥2) chronic conditions (MCC),^{1 2} resulting in significant healthcare and self-management responsibilities. Individuals living with MCC contend with considerable disease and treatment burden,^{3 4} requiring sufficient health literacy (HL) and self-management skills (SM).^5–7 Studies have consistently documented the challenges older adults disproportionately face when navigating health systems, engaging healthcare providers, managing and adhering to prescribed regimens and self-monitoring conditions.^8–10 In addition to greater healthcare demands and increasing comorbidities, older adults also contend with age-related cognitive decline,¹¹ which impacts self-management behaviours and health outcomes.^12–15 Few cognitive ageing studies have been able to map baseline cognitive performance and changes over time to the development of disease in later life and health outcomes, and fewer have detailed assessments of HL, SM skills, cognitive abilities and health outcomes. Beginning in 2008, the Health Literacy and Cognitive Function Study, ‘LitCog’ was designed to complement existing cognitive ageing studies by taking a pragmatic, health services perspective on cognition-health associations.¹⁶ The LitCog study investigates changes in cognition, HL and SM skills and their relation to health outcomes, purposefully examining less-studied, potentially modifiable pathways that could inform how health systems can address age-related cognitive decline.

To date, the LitCog study has examined older adults’ ability to engage in self-care independent of their social context; however, health behaviours and health management occurs in the contexts of people’s everyday lives, and many older adults manage their health with other people, such as a caregiver. Given the central role caregivers may play in supporting the self-management of adults living with MCC and preventing declines in health, their presence and assistance cannot be ignored.^17–22 A caregiver’s assistance may mitigate SM challenges experienced by patients with cognitive decline or limited HL or SM skills. In particular, caregivers may provide tangible help, such as assistance with self-care tasks, as well as emotional support and companionship, which may indirectly impact health outcomes.²³ Despite an expansive body of literature related to caregiving, this research has historically focused on caregivers of elderly adults with physical disabilities or dementia, rather than those managing other chronic conditions.^24–27 Given the high prevalence of multimorbidity, this gap in the literature is significant.

Complicating the utility of available caregiver support is the complexity of chronic disease self-management, as those providing assistance must also have considerable skills and competencies. Literature regarding informal caregiving of chronically ill adults has found that those who assume these roles often receive little to no training.^{28 29} Furthermore, as older adult’s needs grow and their dependence on caregivers increase, the caregiver’s ability to maintain a high level of care may be affected.²⁴ Yet evidence is both limited and inconsistent with regards to caregiver skills. A prior study among paid caregivers found high rates of limited HL (36%) and inadequate SM skills; nearly two-thirds made significant errors when dosing out multidrug regimens.³⁰ This is not surprising, given the minimal training and supervising practices of home health agencies.^{31 32} Research on family caregivers commonly focuses on activities of daily living (ADL), instrumental activities of daily living (IADL), and psychosocial concerns, but does not explore their interactions with the healthcare system or their training and support for medical tasks.³³ Thus, it is possible a significant determinant of LitCog patients’ health status is not only the presence of a caregiver involved in their SM, but one with sufficient cognitive, HL and SM skills to compensate for a patient’s self-care deficits. Very little is known about the associations between caregiver HL and SM skills and care recipients’ outcomes; however this could be a modifiable target for health systems.

In response, we are expanding our LitCog cohort to enrol patients’ caregivers to examine the presence, skills and roles of caregivers over time and assess how their involvement optimally benefits patients. Understanding the availability and adequacy of caregiver support over time may be critical to patients’ health. Our specific aims and hypotheses are presented in table 1.

Table 1.

LitCog Caregiver Cohort aims and hypotheses

Aims and corresponding hypotheses
Aim 1	Investigate associations between the presence of an involved caregiver with treatment adherence, chronic disease outcomes and functional health status among older adults.
Hypothesis 1	Older adults with caregivers will maintain better health outcomes compared with those without a caregiver.
Hypothesis 2	Those with caregivers possessing adequate cognitive, health literacy and self-management skills will have better health outcomes and slower decline in functional status compared with those with no caregiver or one with more limited skills.
Aim 2	Identify factors influencing associations between caregiver involvement and patient outcomes.
Hypothesis 3	Patients’ treatment burden and self-care independence will moderate associations.
Hypothesis 4	Caregivers’ own health status and treatment burden over time will moderate associations.
Hypothesis 5	Caregivers’ healthcare activation and health-related technology use will mediate associations.

Methods and analysis

Sample

We will enroll a parallel cohort of caregivers to participants enrolled in the LitCog study. Beginning in 2008, 900 adults were recruited from one academic general internal medicine practice and six federally qualified health centres in Chicago, Illinois, USA. English speaking adults who sought regular care (defined as two clinical visits within the past 2 years) from study sites were identified through practice records, and research coordinators contacted patients by telephone to screen for eligibility. Patients were eligible if they (1) were between the ages 55 and 74, (2) spoke English (3) and had adequate cognitive capacity, as defined by ≤2 errors on the six-item screener.³⁴ Enrolled participants are invited to complete follow-up interviews every 3 years, with a total of six time points currently planned.

Beginning at the fourth time point (T4), we invited LitCog participants’ caregivers to enroll into a parallel cohort and complete separate corresponding interviews. Caregivers are eligible if they (1) are ≥18 years of age, (2) provided care for ≥6 months and (3) assisted with at least one activity of daily living, instrumental activity of daily living or health management task. Caregivers will complete a modified battery to the patient interview (table 2); data is collected and recorded using Research Electronic Data Capture (REDCap) survey software.³⁵ Caregivers complete interviews in-person at the first interview (corresponding to patient T4 interview), and by telephone at the second (corresponding to patient T5 interview) and third interview (corresponding to patient T6 interview). Caregivers enter the cohort at any point at which they become eligible.

Table 2.

LitCog patient and caregiver measures

Variable	Instrument(s) or measure(s)	Data source
		Patient		Caregiver
		Interview	EHR	Interview
Health literacy and self-management skills
Health literacy	Test of Function Health Literacy in Adults, Rapid Estimate of Adult Literacy in Medicine, Newest Vital Sign, Brief Health Literacy Screen	■		■
Self-management skills	Comprehensive Health Activities Scale	■		■
Cognitive function
Processing speed	Digit Comparison, Pattern Comparison, Symbol Digit	■		■
Working memory	Size Judgement Span, Spatial Working Memory, Spatial Span	■		■
Inductive reasoning	Ravens Progressive Matrices, Educational Testing Service Letter Sets, Stockings of Cambridge	■		■
Long-term memory	New York Paragraph, Verbal Recognition – Immediate and Delayed	■		■
Verbal abilities	American National Adult Reading Test, Shipley Institute of Living Scale, Graded Naming Test	■		■
Global cognitive function	Mini-Mental State Examination, Telephone Interview for Cognitive Status	■		■
Health outcomes: health status, clinical outcomes
Physical health	Patient Reported Outcomes Measurement Information Service (PROMIS) Physical Function	■		■
Mental health	PROMIS Anxiety and Depression short forms	■		■
Chronic conditions	Self-reported chronic conditions, ICD-10 codes	■	■	■
Chronic disease outcomes	Haemoglobin A1c, blood pressure, LDL cholesterol, glomerular filtration rate		■
Mortality	National Death Index		■
Self-care regimen complexity
Regimen complexity	Number of Medications, Healthcare Task Difficulty, Treatment Burden Questionnaire; Medication Regimen Complexity Index	■	■	■
Psychological factors
Patient activation	Consumer Health Activation Index	■		■
Personality	NEO Five Factor Inventory-3	■
Behavioural factors
Medication adherence	Adherence Stars With Knowledge	■
Clinical preventive services	Colon and breast cancer screening, vaccinations	■
Protective health behaviours	Berkeley fat and fruit/vegetable intake	■
Health risk behaviours	BRFSS smoking and drinking	■
Technological factors
Technology use	Health Information Trends Survey Pew Research Center Survey, Portal Use	■		■
Social support and caregiver factors
Social support	Lubben Social Network Scale, Tangible Support	■
Caregiver burden	Caregiver Burden Inventory			■
Self-efficacy	Caregiver Inventory			■
Nature and intensity of care provided	Relationship to patient, hours of care provided, duration of caregiving responsibilities, nature of assistance provided (disability-related, health system logistics, health management)			■
Socio-demographics
Socio-demographics	Age, sex, race/ethnicity, education, income, housing, employment, insurance	■		■

BRFSS, Behavioral Risk Factor Surveillance System; EHR, Electronic Health Record; ICD-10, International Classification of Diseases, Tenth Revision (ICD-10); LDL, Low-density lipoprotein; NEO, Neuroticism, Extraversion, Openness.

Patient and public involvement

Patients and the public were not involved in the design or conduct of this research study.

Measures

HL and SM skills

HL is assessed using the Newest Vital Sign (NVS) and Brief Health Literacy Screener (BHLS). The NVS is a screening tool used to determine the risk of limited HL. Patients are given a copy of a nutrition label and asked six questions about how they would interpret and act on the information contained on the label.³⁶ The BHLS includes three single-item screening questions asking participants to self-report (1) the frequency that someone else helps them read hospital materials; (2) their difficulty understanding written information; and (3) their confidence filling out forms independently.³⁷ SM skills are assessed using the Short Comprehensive Health Activities Scale, which measures skills necessary to navigate the healthcare system. Participants are presented five scenarios depicting common health-related tasks in a variety of formats, including print documents, prescription medication bottles, spoken health communication and multimedia video and asked a series of questions to assess comprehension.³⁸

Cognitive abilities

Caregivers will complete a modified version of the cognitive battery that LitCog participants complete, which will include one test per cognitive domain. Domains that are assessed and the specific tests include: working memory (Size Judgement Span),³⁹ executive function (Educational Testing Service Letter Sets),⁴⁰ attention/processing speed (Symbol Digit),⁴¹ episodic memory (New York University Paragraph).⁴² Global cognitive ability will be assessed using the Mini Mental State Examination⁴³ and Telephone Interview for Cognitive Status.⁴⁴

Nature and intensity of care provided

To characterise the nature of assistance provided, questions will be modelled on the National Study of Caregiving.²¹ We will collect information on whether caregivers provide assistance related to disability related activities (eg, shopping, transportation, housework), health system logistics (eg, making appointments, ordering medications, communicating with care recipient’s medical provider) and health management (eg, administering medications, diet, skin care, exercise). To characterise the level of care provided, caregivers will report information about the relationship of the caregiver to the participant, number of hours of care provided/week, duration of caregiving responsibilities and travel time to care recipients’ residence.

Caregiver-specific measures

Participants will complete several caregiver-specific assessments including caregiver healthcare task difficulty, caregiver burden, caregiver self-efficacy, activation, technology use, busyness and routine and relationship quality. Caregiver healthcare task difficulty will be assessed with the Caregiver Healthcare Task Difficulty (HCTD) scale.^{3 45} The caregiver HCTD scale was designed to inquire whether caregivers assisted the care recipient with a series of tasks, and if so, how much difficulty they had performing the task. Caregiver burden is assessed using the Caregiver Burden Inventory, a multidimensional measure of caregiving burden which includes 24 items that address time-dependence, developmental, physical, social and emotional burden.⁴⁶ Caregiver self-efficacy will be measured with the Caregiver Inventory, a non-disease-specific measure of self-efficacy for caregiving that reflects the comprehensive relationship.⁴⁷ Caregiver activation will be assessed using the Consumer Health Activation Index,⁴⁸ which measures health activation and level of motivation to engage in healthcare decisions/actions. Caregiver access to and use of various forms of technology use will be measured using items from Pew Research Center survey on technology use. Self-reported busyness and routine will be measured using the Martin and Park Environmental Demands Scale,⁴⁹ and relationship quality will be assessed using a single item in which caregivers self-report the degree of tension or strain in their relationship with the care recipient.

Caregiver health status and urgent healthcare usage

Caregiver’s physical and mental health will be assessed using the Patient Reported Outcomes Measurement Information Service (PROMIS) short-form instruments of physical function, depression and anxiety.⁵⁰ Caregivers will also self-report the number of chronic conditions and prescription medications. Urgent healthcare usage was assessed by self-report of emergency department visits and unplanned inpatient hospitalisation in the past 12 months.

Socio-demographics

We will collect self-reported socio-demographic information including, age, gender, educational attainment, race and ethnicity, living situation (own/rent), marital status, household composition, household income, working status and health insurance.

Analysis plan

Caregiver data will be combined with their corresponding participant data, and analyses, per aim, are described below.

Aim 1: investigate associations between the presence of an involved caregiver with treatment adherence, chronic disease outcomes and functional health status among older adults

Hypothesis 1: Older adults with caregivers will maintain better health outcomes compared with those without a caregiver. Hypothesis 2: Those with caregivers possessing adequate cognitive, HL and SM skills will have better health outcomes and slower decline in functional status compared with those with no caregiver or one with more limited skills. The association between caregiver involvement/skills with health outcomes over time (T4–T6) will first be investigated using Generalized linear mixed models (GLMM), with caregiver involvement (yes/no) being the main variable of interest for Hypothesis 1. For Hypothesis 2, a three category variable will be created indicating no caregiver, caregiver with limited or caregiver with adequate cognitive, HL and SM skills. Relevant patient characteristics (eg, socio-demographic, HL skills and treatment burden) as well as caregiver characteristics including type (informal, paid), healthcare task difficulty, caregiver burden and the nature of care provided will be examined as covariates.

Aim 2: identify factors influencing associations between caregiver involvement and patient outcomes

Hypothesis 3: Patients’ treatment burden and self-care independence will moderate associations. Hypothesis 4: Caregivers’ own health status and treatment burden over time will moderate associations. Hypothesis 5: Caregivers’ healthcare activation and health-related technology use will mediate associations. To test for moderation in Hypothesis 3, interaction terms between caregiver involvement and skills and patient health outcomes with each of the potential moderators (patient treatment burden, self-care independence) will be added separately to each model described in Aim 1.^{51 52} To test Hypothesis 4 and Hypothesis 5, analyses will be limited to participants with caregivers. For Hypothesis 4, the caregiver’s health status and treatment burden included as moderators as described for Hypothesis 3. To test for mediation for Hypothesis 5, we must show: (1) the independent variable (caregiver HL, SM skills) predicts the potential mediator (caregiver activation, technology use), (2) the independent variable predicts the dependent variable (patient health outcomes) and (3) the addition of the potential mediator eliminates or decreases the strength of the relationship between the independent and dependent variables.^51–53 Models similar to those described in Aim 1 but with activation or technology use as the outcome will be run to satisfy #1. If this model holds for a particular mediator, we will proceed by adding it to respective models in H₅, where significant relationships between caregiver HL or SM skills and health outcomes are found (#2). Caregiver HL or SM coefficients will then be examined to determine whether the addition of mediators eliminates or decreases the strength of the relationship (#3). Bootstrap resampling techniques will obtain an empirical 95% CI for the effect.

Exploratory analyses

We recognise it is possible the primary caregiver may change over time, although we estimate most caregivers will be familial and less likely to change over 6 years. If this occurs, we will examine whether a change (from unpaid to paid or variable paid caregivers) affects relationships.

Power calculation

Power for this study was based on comparisons of the patient PROMIS physical function scores (standardised mean=50, SD=10) between those with and without a caregiver for Aim 1. We expect that 400 patients will complete T₅ interviews and 350 will complete T₆ interviews; 60% of each sample will have an involved caregiver. At T₅, a sample of 400 patients (240 with a caregiver, 160 without) will enable us to detect a minimum difference of 2.9 between groups assuming a type I error of 5%, with 80% power. With similar assumptions at T₆, and n=350 patients (210 with a caregiver, 140 without) we can detect a difference of 3.1.

Ethics and dissemination

The study was approved by the Northwestern University Institutional Review Board (STU00026255). All participants will provide consent prior to participation. Results will be published in peer-reviewed journals and summaries will be provided to the funders of the study.

Ethics statements

Patient consent for publication

Not applicable.