Habitual Sleep Duration and Health-Related Quality of Life in Family Caregivers: Findings from the Behavioral Risk Factor Surveillance System

ABSTRACT Background Insufficient sleep duration is associated with poor health-related quality of life (HRQoL). However, this relationship has not been studied in family caregivers, a group at high risk of insufficient sleep duration and poor HRQoL. Objective To examine the associations between habitual sleep duration and HRQoL measures in family caregivers. Methods This cross-sectional study used data from 23,321 caregivers in the 2016 Behavioral Risk Factor Surveillance System. The HRQoL measures were health status and poor mental and physical health days. A multivariable logistic regression model was used to assess the association between sleep duration (<7, 7–9, >9 hours) and health status (fair or poor versus good to excellent). Zero-inflated negative binomial models were used to analyze the association of sleep duration with poor mental and physical health days. Results Fair or poor health status was significantly higher in caregivers with short (odds ratio [OR], 1.40; 95% CI: 1.12, 1.74) and long (OR, 2.07; 95% CI: 1.34, 3.21) sleep duration. Short sleep duration was associated with a higher number of poor mental health days (IRR [incident rate ratio], 1.17; 95% CI: 1.04, 1.31) and poor physical health days (IRR, 1.26; 95% CI: 1.10, 1.45). Long sleep duration was associated with more poor mental health days (IRR, 1.31; 95% CI: 1.08, 1.60). Conclusion Extremes in sleep duration were associated with lower HRQoL. These findings point to the need for interventions that promote adequate sleep and address factors underlying extremes in sleep duration in the context of family caregiving.


Introduction
In the United States (US), 1 in 5 adults reports being a caregiver to an adult or child with special needs in the previous 12 months (AARP & National Alliance for Caregiving, 2020).Family caregivers, usually the care recipient's relatives, partners, friends, or neighbors (EmblemHealth & National Alliance for Caregiving, 2010), play a crucial role in long-term care and support in the US (Committee on Family Caregiving for Older Adults; Board on Health Care Services; Health and Medicine Division; National Academies of Sciences Engineering and Sciences, 2016).Even though caregiving can be rewarding to the caregiver (Spillman et al., 2014), the strain of caregiving is associated with negative health effects such as increased levels of anxiety (Song et al., 2018;Wharton & Zivin, 2017), depression (Keilty et al., 2018;Song et al., 2018), and low health-related quality of life (Boehmer et al., 2019;Thomas et al., 2015).Health-related quality of life (HRQoL), defined as one's perceived mental and physical health over time (Centers for Disease Control and Prevention, 2000), is a predictor of morbidity, hospitalization, and mortality (Brown et al., 2015;Dominick et al., 2002;Phyo et al., 2020).
Caregiving is also associated with poor sleep outcomes, with family caregivers more likely than non-caregivers to have sleep problems, including poor sleep quality and short sleep duration (Gao et al., 2019;Y. Liu et al., 2020;Song et al., 2021).Sleep plays a role in many of the body's physiologic processes, and the American Academy of Sleep Medicine and Sleep Research Society (AASM/SRS) recommends that adults sleep at least 7 hours daily to ensure optimal health (Watson et al., 2015).Short sleep duration is associated with health problems such as impaired cognitive performance (Curtis et al., 2018), metabolic syndrome (Hua et al., 2021;Titova et al., 2018), hypertension (Deng et al., 2017;Yadav et al., 2017), and depression (Li et al., 2017;Sun et al., 2018).Even though the relationship between sleep and HRQoL has been widely studied (Cho et al., 2020;W. Liao et al., 2022;Y. Liu et al., 2018;Lucena et al., 2020;Matsui et al., 2021;Stefan et al., 2017), only a few of these studies have focused on sleep duration (Cho et al., 2020;Y. Liu et al., 2018;Stefan et al., 2017) with none of them being in family caregivers.
The findings on sleep duration and HRQoL relationship are mixed.Some findings show a U-shaped relationship between sleep duration and HRQoL (Cho et al., 2020;Stefan et al., 2017), while other findings point to differences in this relationship based on the presence or absence of chronic illness (Y.Liu et al., 2018).Given the high risk of short sleep and poorer HRQOL in family caregivers (Boehmer et al., 2019;Gao et al., 2019;Keilty et al., 2018;Y. Liu et al., 2020), there is a need to examine the relationship between sleep duration and HRQoL in this population because habitual sleep as a health behavior can be modified to achieve better health outcomes (Grandner, 2019).We address this gap by using data from a population-based sample of family caregivers from the 2016 Behavioral Risk Factor Surveillance System (BRFSS) to analyze the relationship between habitual sleep duration and HRQoL and test whether having chronic health conditions modifies this relationship.

Data source and study sample
This cross-sectional study used data from the 2016 BRFSS.In BRFSS, US state health departments use telephone surveys to collect annual data on behavioral risk factors and chronic disease in noninstitutionalized adults aged ≥18 years.In addition to the core questions, BRFSS has optional modules that states may include in their survey.In the 2016 BRFSS, 19 states (Arkansas, Arizona, California, Colorado, Connecticut, Georgia, Minnesota, Missouri, Montana, Nevada, New Jersey, New York, North Dakota, Ohio, Oregon, South Dakota, Tennessee, Texas, Utah), District of Columbia, and Puerto Rico included the caregiving module which 134,701 respondents completed (Centers for Disease Control and Prevention, 2019).Study approval was waived by the University of California Davis Institutional Review Board, because the BRFSS data used in this research is de-identified, publicly available.
For the present study, family caregivers (n = 24,152) were defined as adults aged ≥18 years who answered in the affirmative to the question, "People may provide regular care or assistance to a friend or family member who has a health problem, long-term illness, or disability.During the past month, did you provide any such care or assistance to a friend or family member?"Of these, 831 respondents were excluded due to missing data on sleep duration (n = 179) or HRQoL measures, poor mental health days (n = 353), or poor physical health days (n = 404), yielding a final study sample of n = 23,321.

Outcome variables: HRQoL measures
The outcome variables were the Centers for Disease Control and Prevention's self-rated HRQoL measures: health status, poor mental health days, and poor physical health days.Health status was assessed using the question, "Would you say that in general your health is: excellent, very good, good, fair, or poor" and dichotomized as good to excellent (good, very good, or excellent) and fair or poor (Y.Liu et al., 2018).Poor mental and poor physical health days were assessed using the following questions: "Now thinking about your mental health, which includes stress, depression, and problems with emotions, for how many days during the past 30 days was your mental health not good" and "Now thinking about your physical health, which includes physical illness and injury, for how many days during the past 30 days was your physical health not good".The validity and reliability of these HRQoL questions in population health assessment have been previously established (Andresen et al., 2003).

Predictor variable: habitual sleep duration
Habitual sleep duration was self-reported and derived from responses to the question, "On average, how many hours of sleep do you get in a 24-hour period".Based on sleep duration recommendations for adults by AASM/SRS (Watson et al., 2015) and previous studies (Alcantara et al., 2017;Ren et al., 2019), habitual sleep duration was categorized as short (<7 hours), recommended/average (7-9 hours), and long (>9 hours).
Chronic health conditions (coronary heart disease, stroke, diabetes, asthma, chronic obstructive pulmonary disease, arthritis, chronic kidney disease, and cancer) were assessed based on self-reported history of being told by a doctor or other health professional that one had the condition, and the number of chronic conditions the caregiver had categorized as 0, 1-2, ≥3.Depression was defined as a history of being told by a doctor or other health professional that one had a depressive disorder, including minor or major depression or dysthymia.Leisure time physical activity (LTPA) was defined as a history of physical activity or exercise unrelated to one's job in the previous month.Cigarette smoking status was categorized as never smoker (never smoked at least 100 cigarettes in a lifetime), current smoker (smokes every day or some days), and former smoker (smoked at least 100 cigarettes in a lifetime but not a current smoker).Alcohol intake was defined as drinking at least one alcoholic beverage in the previous 30 days.
Caregiving characteristics included the care recipient's primary health condition (dementia, mental illness, cancer, frailty/old age/infirmity, other condition), the care recipient's relationship to the caregiver (parent, spouse/partner, child, other family member, and non-relative), caregiving duration (<2 years and ≥2 years), and caregiving hours per week (<20 hours, 20 to 39 hours, and ≥40 hours).Caregivers responded to the question, "Of the following support services, which one do you most need that you are not currently getting": "classes about giving care, such as giving medications," "help in getting access to services," "support groups," "individual counseling to help cope with giving care," "respite care."Any caregiver who selected any support service needs was classified as having unmet support needs.

Missing data
In the final analytic sample (n = 23,321), 19% had missing data in the following covariates: caregiving duration (n = 365), caregiving hours per week (n = 1,383), education level (n = 40), marital status (n = 133), smoking status (n = 103), and alcohol intake (n = 206).The missing data were imputed using multiple imputation by chained equations (Royston & White, 2011).The imputation model comprised all variables in the analysis model, BRFSS weight and survey design variables, and auxiliary variables that showed significant associations with data missingness to improve the plausibility of data missing at random (White et al., 2011).The auxiliary variables included homeownership, annual household income levels, internet use in the previous 30 days, health insurance, inability to see a doctor because of medical costs, and owning a cell phone for personal use.The distribution of pre-and post-imputation data for the covariates with missing data was examined using survey-weighted percentages.The results showed minimal differences between the pre-imputation and post-imputation data (Supplementary Table 1).Consequently, all inferential analyses were completed using the imputed data.

Statistical analysis
Analyses were done using STATA 1C Version 15 (StataCorp, 2017).The BRFSS survey design variables and sample weights were applied in analyses to account for BRFSS's complex sampling procedure and increase the generalizability of results.In descriptive analyses, frequencies and percentages for categorical variables, means for continuous variables, and medians and interquartile ranges (IQR) for count variables whose data was skewed were calculated.In bivariate analyses, the distribution of the outcome variables and covariates across the three categories of sleep duration was assessed, and differences between groups were examined using the χ 2 test for categorical variables, one-way analysis of variance (ANOVA) for the continuous variable, age, and Kruskal Wallis test for the variables, poor mental and poor physical health days.
Logistic regression models were fit with health status (good to excellent [reference] and fair or poor) as a function of habitual sleep duration (<7 hours, 7-9 hours, and >9 hours).In the adjusted model, we controlled for caregiver sociodemographic characteristics (age, sex, race/ethnicity, marital status, education level, and employment status), caregiver health characteristics (number of chronic health conditions, history of depression, LTPA, cigarette smoking status, and alcohol intake), and caregiving characteristics ((relationship of care recipient to the caregiver, caregiving duration, caregiving hours per week, care recipient's primary health condition, and unmet support service needs).
A descriptive analysis of count outcomes, poor mental and physical health days, showed overdispersion and excess zero counts with over 55% zero days of poor mental (Figure 1) or physical health (Figure 2).We fitted zero-inflated negative binomial (ZINB) models since they fit better for zeroinflated count data (Neelon et al., 2016).The variables included in the zero-inflate part of the ZINB models were similar to those in the negative binomial part.The negative binomial part of the ZINB models was used to calculate the incident rate ratio (IRR) of poor mental and poor physical health days as a function of sleep duration adjusted for similar covariates as outlined in the logistic model above.
In separate analyses, we added an interaction term (sleep duration × the number of chronic health conditions) to each of the three adjusted models to assess for any effect modification of the association between sleep duration and the HRQoL outcomes by chronic health conditions.

Sample population characteristics
The caregivers' age ranged from 18 to 80 years (mean 48 ± 0.4).Most (81.6%) were below 65 years of age.More than a third (38.8%) had short sleep duration.Male caregivers constituted 41.6% of the sample.The majority (63.7%) had more than a high school education, and 55.7% were employed.Over half (58.9%) of caregivers reported having at least one chronic health condition, 21.9% had a history of depression, and 24.7% rated their health as fair or poor (Table 1).
In bivariate analyses, sleep duration differed by caregiver's HRQoL, age, marital status, race/ ethnicity, education level, employment status, number of chronic conditions, history of depression, LTPA, smoking status, alcohol intake, caregiving duration, caregiving hours per week (p < .05).Caregivers with short sleep duration were younger and more likely to be employed, while those with long sleep duration were older and more likely to have three or more chronic health conditions, have a history of depression, and not engage in LTPA (Table 1).

Association between habitual sleep duration and HRQoL measures
In the unadjusted logistic regression results, both short sleep duration (OR, 1.95; 95% CI: 1.60, 2.38) and long sleep duration (OR, 4.75; 95% CI: 3.13, 7.21) were associated with fair or poor health status   (Table 2).In the adjusted logistic model, the association between sleep duration and health status remained robust, with caregivers who had short sleep duration more likely to rate their health status as fair or poor compared to those with average sleep duration (OR, 1.40; 95% CI: 1.12, 1.74).Those with long sleep duration were twice more likely to rate their health status as fair or poor (OR, 2.07; 95% CI: 1.34, 3.21).
In other results from the adjusted logistic model, the other characteristics associated with fair or poor health status included being middle-aged (45-64 years old), Hispanic or other race/ethnicity, unmarried, having high school or lower education level, unemployed, history of depression, having at least one chronic health condition, not engaging in LTPA, caregiving for ≥40 hours per week, and caring for a spouse or partner (Table 2).
In the adjusted ZINB model examining the association of sleep duration and poor mental health days, caregivers with short sleep duration had more poor mental health days than caregivers with average sleep duration (IRR, 1.17; 95% CI: 1.04, 1.31).Those with long sleep duration had 31% more poor mental health days (IRR, 1.31; 95% CI: 1.08, 1.60).Other factors associated with an increased number of poor mental health days included some college or lower level of education, ≥40 caregiving hours per week, caring for a spouse or partner, having a chronic health condition, history of depression, not engaging in LTPA, and cigarette smoking (Table 3).
Results from the adjusted ZINB model on the association between sleep duration and poor physical health days showed that caregivers with short sleep duration had more poor physical health days than caregivers with average sleep duration (IRR, 1.26; 95% CI: 1.10, 1.45).There was no significant association between long sleep duration and the number of poor physical health days (IRR, 1.16; 95% CI: 0.95, 1.42).Other factors positively associated with poor physical health days included being middle-aged, unmarried, retired, or unemployed, caring for a parent, spouse, or partner, having chronic health conditions, and physical inactivity (Table 4).
Results from the adjusted models with the interaction term sleep duration × the number of chronic conditions yielded no significant interactions in the association between sleep duration and the HRQoL measures.

Discussion
In this large population-based cross-sectional study, we find significant relationships between habitual sleep duration and three measures of HRQoL in family caregivers after controlling for various caregiver and caregiving factors.To the best of our knowledge, this is the first study to assess these relationships in family caregivers.Specifically, we find a U-shaped relationship between sleep duration and caregiver health status and poor mental health days, with short and long sleep duration adversely associated with these outcomes.The results also show significant associations between short sleep duration and poor physical health days.
Similar to our study, a U-shaped relationship between sleep duration and health status has been noted in previous population-based studies in adults living in the US (Y.Liu et al., 2018), Korea (Cho et al., 2020), and China (W.Liao et al., 2022).Similar findings were also reported among college students (aged 17-24 years) in Croatia, where <7 hours or >10 hours of sleep was associated with lower odds for good self-rated health compared to 7-8 hours of sleep (Stefan et al., 2017).
Our findings on sleep duration and mental and physical health days are partly consistent with a prior BRFSS study that assessed differences in the relationship between sleep duration and HRQOL in adults with and without chronic disease (Y.Liu et al., 2018).Specifically, in adults with chronic disease, short and long sleep duration was associated with frequent mental and physical distress, defined as ≥ 14 poor mental and physical health days, respectively (Y.Liu et al., 2018).However, in those without chronic disease, short sleep duration was associated with frequent mental and physical distress, while long sleep duration was associated with frequent mental distress but not physical distress (Y.Liu et al., 2018).Even though Liu et al. noted some differences in the two groups, our study findings in caregivers indicate that the number of chronic health conditions is not a modifier of the relationship between sleep duration and HRQoL measures but a strong predictor of unfavorable HRQoL.
Conversely, extreme sleep duration can indicate poor health, which is thought to be the mechanism underlying the association between long sleep duration and chronic conditions (Stewart et al., 2020).This premise is supported by evidence from longitudinal studies showing an association between long sleep duration and chronic conditions such as kidney disease (Bo et al., 2019), cancer, diabetes and stroke (Smagula et al., 2016), depression (Smagula et al., 2017), and dementia (Westwood et al., 2017).In chronic illness, the underlying disease processes, and in some cases, the treatments, may increase the homeostatic burden, fatigue, and sleep propensity (Gangwisch, 2014;Smagula et al., 2017), causing individuals who are ill to spend more time in bed resting or sleeping.Consequently, self-reported sleep duration may reflect time spent in bed rather than actual sleep duration (Watson et al., 2015).Therefore, the association between long sleep and HRQoL may indicate the severity of illness or other chronic conditions not controlled for in studies and may be an effective measure for identifying caregivers in poor health.Additional health-related, socioeconomic, and caregiving factors showed independent associations with caregivers' HRQoL.The link between HRQoL and various health indicators such as chronic illness and health behaviors like alcohol consumption, physical activity and cigarette smoking is well established (Megari, 2013;Orji et al., 2021), and our results collaborate these findings.Our findings align with prior studies in caregivers, which have identified significant links between diminished HRQoL and various forms of socioeconomic disadvantage, such as low education level and unemployment (Boyer et al., 2012;Lima-Rodríguez et al., 2022).Similarly, in line with our findings, previous research has demonstrated positive associations between caring for a spouse, being unmarried, caregiving hours, and decreased HRQoL (Cohen et al., 2017;Leng et al., 2019;Lima-Rodríguez et al., 2022;Yihedego et al., 2020).Our results on age and HRQoL align with other findings that showed a positive association between caregiver age and mental HRQoL, but a negative association with physical HRQoL (Lima-Rodríguez et al., 2022).In addition, our study found that middle-aged caregivers, often part of the sandwich generation that is responsible for providing care to both their children and aging parents (Lei et al., 2023), were more likely to rate their general health as poor or fair than did their younger counterparts.
We observed slight differences between our findings and previous research in the association between caregiving for individuals with dementia and caregiver HRQoL.In a prior BRFSS study, no differences were noted in any of the three HRQOL measures between caregivers of individuals with dementia and those caring for individuals with other chronic conditions (Secinti et al., 2021).While our study did not find significant differences in the number of poor mental or physical health days between caregivers of people with dementia and caregivers of individuals with other health conditions, it did reveal that caregivers of persons with dementia were less likely to rate their general health as poor compared to caregivers of individuals with other health conditions.These differences in the relationship between dementia and caregivers' HRQOL across studies may be partly explained by variations in study design, including sample characteristics and analysis procedures.For instance, our study included caregivers managing various health conditions, and compared dementia to a broader category named in BRFSS and in the current study as other health conditions.This differs from the previous BRFSS study that restricted their sample to caregivers of individuals with a few specific health conditions, specifically, dementia, diabetes, COPD, and cancer (Secinti et al., 2021).
There are limitations to consider when interpreting this study's findings.Since this is a crosssectional study, we can only infer associations, and no causal relationships, between sleep duration and HRQoL.Second, all the BRFSS data used in the study, including sleep duration and chronic health conditions, was self-reported.The sole reliance on self-reported data increases the risk of measurement error which may have impacted our study findings (Althubaiti, 2016).Third, the 2016 BRFSS did not have data for other sleep parameters, such as sleep quality and use of sleep aids, which are potential confounders of the relationship between sleep duration and HRQoL.Additionally, only nine chronic health conditions are assessed in BRFSS, which can misclassify some caregivers as having no or fewer chronic conditions.

Figure 1 .
Figure 1.Distribution of poor mental health days reported by family caregivers in 2016 BRFSS.

Figure 2 .
Figure 2. Distribution of poor physical health days reported by family caregivers in 2016 BRFSS.

Table 1 .
Characteristics of the 2016 BRFSS sample of family caregivers by habitual sleep duration status.

Table 2 .
Logistic regression results on the association of habitual sleep duration and general health status among family caregivers in the 2016 BRFSS.

Table 2 .
(Continued).Each variable adjusted for all other variables in the table. a

Table 3 .
Zero-inflated negative binomial regression results on the association of habitual sleep duration with poor mental health days among family caregivers in the 2016 BRFSS.

Table 3 .
(Continued).Each variable adjusted for all other variables in the table. a

Table 4 .
Zero-inflated negative binomial (ZINB) model of the association of habitual sleep duration with poor physical health days among family caregivers a .

Table 4 .
(Continued).Each variable adjusted for all other variables in the table. a