Clinical Gerontologist

ISSN: 0731-7115 (Print) 1545-2301 (Online) Journal homepage: http://www.tandfonline.com/loi/wcli20

Alcohol Use and Frailty Risk among Older Adults over 12 Years: The Health and Retirement Study Mona Shah, Daniel Paulson & Vu Nguyen To cite this article: Mona Shah, Daniel Paulson & Vu Nguyen (2017): Alcohol Use and Frailty Risk among Older Adults over 12 Years: The Health and Retirement Study, Clinical Gerontologist, DOI: 10.1080/07317115.2017.1364681 To link to this article: http://dx.doi.org/10.1080/07317115.2017.1364681

Accepted author version posted online: 16 Aug 2017. Published online: 16 Aug 2017. Submit your article to this journal

Article views: 11

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=wcli20 Download by: [UNIVERSITY OF ADELAIDE LIBRARIES]

Date: 11 October 2017, At: 22:15

CLINICAL GERONTOLOGIST https://doi.org/10.1080/07317115.2017.1364681

Alcohol Use and Frailty Risk among Older Adults over 12 Years: The Health and Retirement Study Mona Shah, MS, MA, Daniel Paulson, PhD, and Vu Nguyen, MS

Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 22:15 11 October 2017

Department of Psychology, University of Central Florida, Orlando, Florida, USA ABSTRACT

KEYWORDS

Objectives: The primary goal was to examine the relationship between alcohol use and frailty, a variable characterizing late-life decline, in a national, longitudinal survey of older adults living in the United States. Methods: The sample drawn from the Health and Retirement Study included 9,499 stroke-free participants over age 65 in 2000. The sample was 59.1% female, and had a mean age of 74.25 years (SD = 6.99). Follow-up data was from 2004, 2008, and 2012. Frailty was defined phenotypically using the Paulson-Lichtenberg Frailty Index (PLFI). Alcohol use was measured via self-report. Control variables included age, race, education, socio-economic status (SES), depressive symptomatology, medical burden score, body mass index (BMI), and partner status. With abstinent participants as the reference group, logistic regressions were conducted to determine prevalent frailty at 2000, and Cox’s proportional hazard models were utilized to determine time to incident frailty over a 12-year period. Results: Results revealed that age, depressive symptomatology, and medical burden score were significant positive correlates of prevalent and incident frailty (p < .05) for both males and females. Logistic regressions revealed that consumption of 1–7 alcoholic drinks per week was associated with reduced prevalent frailty (OR = .49, p < .001) for females. Survival analysis results reveal that compared with nondrinkers, males and females who reportedly consumed 1–7 drinks per week had a decreased probability of incident frailty (HR = .78–081, p < .05). Conclusions: Findings suggest that moderate alcohol use confers reduced frailty risk for both older men and women. Future research should examine the mechanism(s) relating alcohol consumption and frailty. Clinical Implications: Findings support extant literature suggesting some healthcare benefits may be associated with moderate drinking.

Aging; late-life decline; risk reduction

Introduction Frailty is accompanied by high rates of comorbidity, disability, and risk for hospitalization (Fried et al., 2001), and thus, is a critical sign for healthcare need for older adults. An extensive literature has emerged on prognostic implications of frailty, though less is known about risk and protective factors for frailty. Meanwhile, an estimated 41% of adults 65 years and older report consuming alcohol (Substance Abuse and Mental Health Services Administration, 2014). Though 13% of older men and 8% of older women suffer ill consequences from abusing alcohol (Blazer & Wu, 2009), moderate alcohol use has been identified as a protective factor for some health outcomes, including frailty (Lang, Wallace, Huppert, & CONTACT Mona Shah

[email protected]

© 2017 Taylor & Francis Group, LLC

Melzer, 2007; Thun et al., 1997; Woods et al., 2005). Past work, however, underrepresents older men, and examines incident frailty over relatively brief periods of time. With the population of older Americans projected to double by the year 2050 (Shrestha & Heisler, 2011), identification of frailty risk and protective factors is imperative. The goal of this study is to identify how alcohol use, a highly prevalent behavior, relates to frailty, a disabling syndrome, among older adults. Alcohol use—A risk and protective factor

Excessive alcohol use confers health risks and can lead to cycles of abuse, withdrawal, physical and social impairment, and death (Ferreira & Weems,

4111 Pictor Lane, Orlando, FL 32816.

Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 22:15 11 October 2017

2

M. SHAH ET AL.

2008). Age-related changes place older adults at unique risk for these adverse consequences (Durfour, 1999; National Institute on Alcohol Abuse and Alcoholism, 1998; Sorocco & Ferrell, 2006). For example, age-related changes in body composition including increased fat and decreased water in the body can produce higher blood alcohol levels when drinking (Durfour, 1999), and polypharmacy escalates risk for adverse alcoholmedication interactions (Sorocco & Ferrell, 2006). Concurrently, numerous studies have demonstrated that moderate alcohol use is associated with health benefits (Lang et al., 2007; Thun et al., 1997). What constitutes “moderate” alcohol use among older adults, however, is inconsistently defined. The Dietary Guidelines for Americans 2010 characterize moderate alcohol use as no more than 1 drink per day for women, and no more than 2 drinks per day for men (U.S. Department of Agriculture & U.S. Department of Health and Human Services, 2010). The National Institute on Alcohol Abuse and Alcoholism (NIAAA) guidelines suggest no more than 7 drinks per week for older adults (National Institute on Alcohol Abuse and Alcoholism, 1998). Other recommendations are based on findings that two drinks per day does not escalate mortality risk (Lang, Guralnik, Wallace, & Melzer, 2007), and may actually reduce risk (Thun et al., 1997). For purposes of this study, and given the low prevalence of excessive alcohol use in the Health and Retirement Study data set, “moderate” alcohol use is defined by ‘1–7 drinks per week’; “substantial” alcohol use is defined as ‘8–14 drinks per week’; and “heavy” alcohol use is defined by ‘15 or more drinks per week.’ One of the most commonly cited benefits of moderate alcohol use is improved cardiovascular health (Abramson, Williams, Krumholz, & Vaccarino, 2001; Corrao, Rubbiati, Bagnardi, Zambon, & Poikolainen, 2000; Rimm, Klatsky, Grobbee, & Stampfer, 1996). A meta-analysis by Corrao and colleagues (2000) depicts the relationship between alcohol use and the risk of coronary heart disease as a J-shaped curve, suggesting moderate alcohol use decreases the risk of coronary heart disease compared with abstinence and heavy use. In addition to the quantity of consumption, alcohol’s effect on aging also appears to vary

by sex. Moderate alcohol use has been associated with cognitive (McGuire, Ajani, & Ford, 2007) and other health benefits, including lower hospitalization rates, (Balsa, Homer, Fleming, & French, 2008) for older women, but not for men. These findings underscore the importance of examining sex differences in the relationship between alcohol use and late-life decline. Frailty

Frailty, broadly conceptualized as the multisystemic dysregulation of homeostatic mechanisms (Fried et al., 2001), is a well-established indicator of late-life decline. Frailty is marked by loss of independence, high rates of health care utilization, and increased mortality risk (Clegg & Young, 2011). Frailty has been associated with diverse aspects of late-life decline, including vascular disease (Alonso-Bouzon et al., 2014), falls (Fried et al., 2001), and inflammation and clotting (Walston et al., 2002). Fried and colleagues (2001) operationalized frailty as a phenotype, defined as having three or more of the following conditions: unintentional weight loss, weakness, exhaustion, slow walking speed, and low physical activity. In using a phenotypic conceptualization, Fried and colleagues (2001) distinguish frailty from high comorbidity and disability both conceptually and empirically. This strategy has been replicated across several frailty indices, including the PaulsonLichtenberg Frailty Index (PFLI; Paulson & Lichtenberg, 2014), which was used in the this study. Frailty is positively associated with aging, being female, being African-American, and lower education and income (Fried et al., 2001; Paulson & Lichtenberg, 2014; Woods et al., 2005). In addition, heavy drinking (characterized in one study as more than 45 drinks per month), physical inactivity, depression, social isolation, fair or poor perceived health, and prevalence of chronic conditions have been identified as predictors of frailty in older adults (Strawbridge, Shema, Balfour, Higby, & Kaplan, 1998; Woods et al., 2005). Further, Woods and colleagues (2005) found a small but significant reduction in frailty risk (OR = .87) for older women who reported consuming < 1 drink per week, and a larger frailty risk reduction (OR = .69) for older women who reported consuming between 1 and 14 drinks per week. Those who reported consuming more than 14 drinks per week, however, were found

Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 22:15 11 October 2017

CLINICAL GERONTOLOGIST

to have similar frailty risk to those who reported abstaining from alcohol (Woods et al., 2005). Despite the study’s female-only sample and 3-year follow-up, these findings suggest that alcohol use at moderate levels may reduce frailty risk among older adults. Similarly, Ortola and colleagues (2015) examined the influence of “Mediterranean style” drinking, defined as moderate intake of alcohol (less than 4 standard drinks for men, less than 2.4 standard drinks for women, and no binge drinking), with wine preference (at least 80% of alcohol proceeds from wine), and alcohol consumption only with meals, in a sample of Spanish adults over age 60. They reported a significant frailty risk reduction effect from Mediterranean style drinking over a mean follow-up of 3.3 years. A comparable study reported similar findings over a 3-year follow-up; however, their Swiss sample included those aged 65 to 70 years, and thus was restricted to the youngold who experience frailty at dramatically lower rates than the older-old (Seematter-Bagnoud, Spagnoli, Bula, & Santos-Eggimann, 2014). Study objectives The primary goal of this study is to examine the relationship between level of alcohol use and frailty in a nationally representative, longitudinal survey of Americans over the age of 65. Hypothesis 1 is that, by comparison to older adults who are either abstinent or heavy consumers of alcohol, those older adults who report moderate alcohol use will experience lower frailty prevalence cross-sectionally. Hypothesis 2 is that, by comparison to abstinent older adults or heavy consumers of alcohol, older adults who drink moderately will have reduced incidence of frailty over 12 years. Hypothesis 3 is that the anticipated frailty risk reduction effect of moderate alcohol use will be more robust for females than for males. An auxiliary hypothesis was added to examine the possibility that support for Hypothesis 1 is not better explained by variable rates of attrition. Method Participants

The present study utilized data collected through the Health and Retirement Study (HRS). The HRS is a

3

longitudinal, cohort study on health, retirement, and aging conducted by the University of Michigan with support from the National Institute of Aging. The first wave of data collection occurred in 1992 with adults over the age of 50 years living in the United States. Data are collected biennially, and currently, the sample consists of over 30,000 adults. The method of data collection includes interviews, surveys, and links to personal records, including Social Security earnings and benefits records, National Death Index data, Medicare claims record data, and employer pension data. Further information on HRS survey design and data collection methods can be found in previously published reports (Hauser & Willis, 2004; Heeringa & Conner, 1995). This study used HRS data from years 2000 (baseline), 2004, 2008, and 2012, as these are the years when complete frailty data for participants are reported and available. There were a total of 19,579 participants with data in 2000. Participants had to be at least 65 years old at the study’s baseline to be included in the study, and thus 8,844 participants were excluded. Another 1,230 participants with a history of stroke and 6 with missing stroke data were excluded, as individuals who have experienced a stroke are more likely to experience subsequent depression, and cognitive and motor impairment (Mukherjee, Levin, & Heller, 2006). Of the remaining 9,499 participants, 8,074 had adequate data to complete examine frailty prevalence (i.e., 1,425 cases were deleted listwise due to missing data). The survival analyses of frailty incidence retained 6,073 participants of the original 9,499 given that participants were excluded if they did not have baseline data (499), were frail in the 2000 wave (1,492), and missing data for all years after 2000 (1,435). Measures Alcohol use Alcohol use was measured via self-report in 2000. Participants who reported drinking alcohol were asked to provide the average number of drinking days per week and average number of drinks per day in the past 3 months. The typical number of drinks per week was calculated by multiplying the number of drinks per day by the number of drinking days per week.

Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 22:15 11 October 2017

4

M. SHAH ET AL.

Outcome variables Frailty. Frailty was measured via self-report using the Paulson-Lichtenberg Frailty Index (PFLI; Paulson & Lichtenberg, 2014), which is based on Fried and colleagues’ (2001) conceptualization of frailty as a phenotype, in 2000, 2004, 2008, and 2012. The frailty index includes five symptom criteria: wasting, weakness, slowness, fatigue or exhaustion, and falls. The wasting criterion was met if the participant reported a loss of 10% body weight over a 2-year period. The weakness criterion was met if the participant responded yes to “Because of health problems, do you have difficulty with lifting or carrying weights over 10 pounds, like a heavy bag of groceries?” The slowness criterion was met if participants responded affirmatively to “Because of a health problem, do you have any difficulty with getting up from a chair after sitting for long periods?” The fatigue or exhaustion criterion was met if the participant endorsed yes to “Since we last talked with you [in the last wave], have you had any of the following persistent or troublesome problems: [. . .] severe fatigue or exhaustion?” Lastly, the falls criterion was met if the participant answered yes to “Have you fallen in the past 2 years?” Each affirmative response received a score of 1, and a score of 3 or more indicated that the participant is frail. No PLFI indicators were drawn from the depression measure utilized below. Control variables Demographic variables. Data on the demographic variables—age, sex, race, education, and socio-economic status (SES)—were collected through selfreport via telephone or in-person interviews. Socio-economic status was calculated by summing total wealth (excluding secondary home) and total income. Total wealth was calculated by the sum of all wealth components (i.e., combined value of: primary home; real estate, excluding primary home; vehicles; businesses; individual retirement account (IRA) and Keogh accounts; stocks and mutual funds; checking, savings and money market accounts; certificates of deposit (CDs), government and saving bonds, and treasury bills; bonds; and all other savings) less the sum of all debt (i.e., sum of first and second mortgages, home loans,

and debt). Total income for the last calendar year was calculated by summing the participant and spouse’s earnings, pensions and annuities, Supplemental Security Income and Social Security Disability, Social Security retirement, unemployment and workers compensation, other government transfers, household capital income and other income. Due to the large-scale values in the SES variable as compared with the other variables, SES values were transformed by dividing them by 100,000. Given that the number of participants not identified as White/Caucasian remained small comparatively, all races other than White/Caucasian were collapsed into one category. Medical burden. Medical burden was assessed by summing the number of endorsed comorbidities: hypertension, diabetes, cardiac disease, arthritis, pulmonary disorder, and cancer. Body mass index (BMI). The HRS data set provides a calculated BMI (in kg/m2) for each participant based on weight collected at each wave and the original height measurement. Depression (CES-D score). Depression was assessed using participants’ scores on an abridged, 8-item version of the Center for Epidemiological Studies Depression (CES-D) measure (Radloff, 1977). Participants answered “yes” or “no” to each itemstatement with respect to how they were feeling “much of the time” in the past week. Scores range on a scale from 0–8, with higher scores suggesting higher levels of depression. Partner status. Partner status was assessed by asking the participant their marital status (i.e., “married,” “married, spouse absent,” “partnered,” “separated,” “divorced,” “separated/divorced,” “widowed,” and “never married”). Participants who reported being “married” or “partnered” were identified as being “partnered,” indicating that they are currently cohabitating with a partner regardless of their marital status (i.e., “partnered”). All other categories were collapsed into one category—“not partnered.”

CLINICAL GERONTOLOGIST

Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 22:15 11 October 2017

Statistical methods

The association between alcohol use and prevalent frailty was determined using stepwise logistic regression models for each sex. The influence of alcohol use was assessed using three dummy coded variables: “Drinks 1–7” represents participants who reported consuming an average of 1–7 drinks per week; “Drinks 8–14” represents participants who reported consuming an average of 8–14 drinks per week; and “Drinks ≥ 15” represents participants who reported consuming an average of greater than or equal to 15 drinks per week. All three groups were then compared with those participants who reported consuming 0 drinks per week (see Table 1). HRS participants who met the age criteria, denied history of stroke, and had complete data in 2000 were used to identify the prevalence of frailty at baseline. Given past findings suggesting that health benefits of moderate alcohol use may be sex dimorphic (Balsa et al., 2008; McGuire et al., 2007), the relationship between alcohol use and frailty status was examined for males and females using separate logistic regression analyses. Using the incidence subsample described above, time to incident frailty over 12 years was examined using Kaplan-Meier survival curves and Cox regressions. To determine that the proportional hazard assumption was met, for each Cox proportional hazard model, we performed a chi-square test of whether Schoenfeld’s residuals are correlated with time as derived in Grambsch and Therneau (1994). The test results indicated the proportional hazard assumption was not violated for all variables except medical burden in the female sample. To safeguard against this violation, another set of survival analyses were run excluding the medical burden and CES-D variable for women as this eliminated the violation in assumption. Results from these analyses did not significantly differ from the original analyses, and thus, results from the original analyses have been reported (see Appendix: Supplementary Table 1). Two Cox regressions examined time to incident frailty for males and females, respectively. Data on participants who were lost to follow-up due to mortality, non-responsiveness or being dropped from the sample were censored at the first attrited wave. As in the prior logistic regression analyses,

5

alcohol use was dummy coded and nondrinkers were utilized as the comparison group. Predictors of incident frailty at 2004, 2008, and 2012 were measured at the 2000 data collection. A sensitivity analysis was completed using another series of Cox regressions, but those only those respondents who reported persistent patterns of drinking over time. This analysis was completed to ensure that the observed results are not an artifact of classification bias. In this sensitivity analysis, alcohol use classification remained consistent for 12 years or the duration of a participant’s inclusion in the study. Finally, a post-hoc logistic regression was completed to examine the possibility that prior results (see below) were an artifact of a survival effect.

Results Participant characteristics are provided in Table 1. The changes in the demographic characteristics of the sample over time are a result of participants who attrited being older and less educated, having more depressive symptomatology, a higher medical burden score, and lower SES. Logistic regression results are depicted in Table 2. At baseline (2000), frailty among males significantly associated with age (OR = 1.04, p < .001, 95% CI = 1.02–1.06), SES (OR = .96, p < .05, 95% CI = .93–1.00), CES-D score (OR = 1.59, p < .001, 95% CI = 1.49–1.70), BMI (OR = .96, p < .01, 95% CI = .93–.99), and medical burden score (OR = 1.78, p < .001, 95% CI = 1.50– 2.00). Race, partner status, years of education, and number of drinks per week consumed did not significantly associate with prevalent frailty for males (p > .05). For females, frailty at baseline (2000) was significant associated with age (OR = 1.06 p < .001, 95% CI = 1.05–1.07), CESD score (OR = 1.36, p < .001, 95% CI = 1.31–1.41), BMI (OR = 1.02, p < .05, 95% CI = 1.00–1.03), medical burden score (OR = 1.66, p < .001, 95% CI = 1.54–1.79), and Drinks 1–7 (OR = .49, p < .001,95% CI = .35–.66). Race, partner status, years of education, and SES did not significantly relate to prevalent frailty for females (p > .05). Post-hoc logistic regression models were run excluding BMI as a control variable to eliminate effects of the potential confound BMI has with the

6

M. SHAH ET AL.

Table 1. Sample characteristics.

Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 22:15 11 October 2017

Variable Age Education (years) 2000 CES-D 2000 Medical Burden 2000 BMI 2000 SES* 2000 Drinks/Week Gender Male Female Race White Other Partnered Frail in 2000

Total Sample

Abstinent (0 drinks)

Moderate (1–7 drinks/week)

Substantial (8–14 drinks/week)

Heavy (>14 drinks/week)

n = 8,079

n = 6,037

n = 1,483

n = 398

n = 161

M(SD) M(SD) M(SD) M(SD) 74.25 (6.99) 74.62 (7.14) 73.45 (6.59) 72.57 (5.82) 11.90 (3.35) 11.46 (3.39) 13.15 (2.89) 13.37 (2.79) 1.54 (1.86) 1.68 (1.91) 1.10 (1.57) 1.12 (1.64) 1.84 (1.17) 1.91 (1.19) 1.64 (1.05) 1.68 (1.08) 26.38 (4.91) 26.59 (5.18) 25.73 (3.93) 25.82 (3.93) 3.88 (7.06) 3.11 (5.45) 6.14 (10.40) 6.57 (9.52) 1.78 (5.07) 0 (0) 3.61 (2.27) 12.18 (2.32) Percentage of Sample Percentage of Sample Percentage of Sample Percentage of Sample

M(SD) 71.66 (5.59) 13.12 (2.93) 1.40 (1.92) 1.62 (1.14) 26.04 (3.89) 5.31 (9.00) 26.16 (14.71) Percentage of Sample

40.90 59.10

35.40 64.60

51.70 48.30

68.10 31.90

80.10 19.90

85.60 14.40 58.10 14.40

83.10 16.90 54.40 16.90

93.30 6.70 68.00 6.70

93.00 7.00 72.40 7.80

94.40 5.60 71.40 6.80

Sample characteristics of the 2000 logistic regression prevalence sample and by the various drinking categories (abstinent, moderate, substantial, and heavy).

Table 2. Logistic regression results for prevalent frailty. Males Variable Age Education (years) Race Partnered 2000 CES-D 2000 SES 2000 BMI 2000 Medical Burden Drinks 1–7 Drinks 8–14 Drinks ≥ 15 Constant

Wald 12.76*** 2.37 .50 2.70 184.09*** 4.23* 6.81** 95.67*** .28 .07 .68 28.83

OR 1.04 .97 1.15 1.31 1.59 .96 .96 1.78 .90 .93 .70

Females 95% CI 1.02–1.06 .93–1.00 .78–1.68 .95–1.79 1.49–1.70 .93–1.00 .93–.99 1.59–2.00 .62–1.31 .54–1.60 .30–1.63

Wald 87.36*** .62 .00 .06 239.81*** 1.36 3.95* 182.21*** 21.54*** .77 .07 153.86

OR 1.06 1.01 1.00 1.02 1.36 .99 1.02 1.66 .49 .76 .86

95% CI 1.05–1.07 .98–1.04 .80–1.25 .85–1.23 1.31–1.41 .97–1.01 1.00–1.03 1.54–1.79 .35–.66 .41–1.41 .28–2.65

***p ≤ .001. **p ≤ .01. *p ≤ .05. OR = Odds Ratio. All analyses controlled for age, years of education, minority status, partnered status, baseline CES-D score, baseline socio-economic status (SES), baseline body mass index (BMI), and baseline medical burden score and the reference group includes abstinent participants (0 drinks per week).

baseline wasting frailty indicator given that the same weight variable was used in calculating each variable. No substantive variation occurred in the results and thus, the variable was included. Results from the survival analysis can be found in Tables 3 and 4. Findings for males suggest that older age (HR = 1.07, p < .001, 95% CI = 1.06– 1.09), greater depressive symptomatology (HR = 1.15, p < .001, 95% CI = 1.10–1.21), lower SES (HR = .98, p < .05, 95% CI = .97–1.00), and higher medical burden scores (HR = 1.34, p < .001,

95% CI = 1.25–1.44) were associated with increased probability of incident frailty. In addition to these covariates, compared with nondrinkers, participants who consumed 1–7 drinks had a 22% reduced probability of incident frailty (HR = .78, p < .05, 95% CI = .63–.96). Results from the survival analysis for females suggest that older age (HR = 1.06, p < .001, 95% CI = 1.05–1.07), greater depressive symptomatology (HR = 1.11, p < .001, 95% CI = 1.08–1.14), higher medical burden scores (HR = 1.27, p < .001, 95%

CLINICAL GERONTOLOGIST

Table 3. Kaplan-Meier non-parametric results. Simplistic Approach

Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 22:15 11 October 2017

Year 2004 2008 2012

N. Risk 6,073 4,226 2,667

N. Event Survival 878 .86 667 .72 485 .59 By Drinking Groups

Non-drinkers 2004 4,359 2008 2,938 2012 1,803 Drinks 1–7 2004 1,245 2008 933 2012 630 Drinks 8–14 2004 324 2008 251 2012 166 Drinks 15+ 2004 135 2008 96 2012 65

SE .00 .01 .01

95% CI .85–.86 .71–.73 .58–.60

712 508 356

.84 .69 .56

.01 .01 .01

.83–.85 .68–.71 .54–.57

125 111 82

.90 .79 .69

.01 .01 .01

.88–.92 .77–.82 .66–.72

23 35 29

.93 .80 .66

.01 .02 .03

.90–.96 .75–.85 .60–.72

17 12 17

.87 .77 .57

.03 .04 .05

.82–9.93 .69–.84 .47–.67

By Sex Males 2004 2008 2012 Females 2004 2008 2012

2,670 1,864 1,207

282 233 184

.89 .78 .66

.01 .01 .01

.88–.91 .77–.80 .64–.69

3,403 2,362 1,460

596 434 301

.83 .67 .53

.01 .01 .01

.81–.84 .66–.69 .52–.55

Table 4. Cox proportional hazard model results. Males Variable Age Education Race Partnered CES-D Score SES BMI Medical Burden Score Drinks 1–7 Drinks 8–14 Drinks 15+

Females

HR Z 95% CI HR Z 1.07 10.63*** 1.06–1.09 1.06 13.2*** .99 −1.15 .96–1.01 .98 −2.05* .88 −.98 .68–1.14 .92 −1.06 .88 −.41 .47–1.64 1.62 2.01* 1.15 5.72*** 1.10–1.21 1.11 6.92*** .98 2.22* .97–1.00 1.00 .71 1.01 1.00 .99–1.03 1.03 5.37*** 1.34 8.01*** 1.25–1.44 1.27 9.36*** .78 −2.37* .95 −.33 1.30 1.40

.63–.96 .81 −2.49* .70–1.28 1.08 .46 .90–1.88 1.65 1.71

95% CI 1.05–1.07 .96–1.00 .78–1.08 1.01–2.60 1.08–1.14 .99–1.01 1.02–1.04 1.21–1.34

.69–.96 .77–1.52 .93–2.94

***p ≤ .001. **p ≤ .01. *p ≤ .05. HR = Hazard Ratio. The reference group includes abstinent participants (0 drinks per week).

CI = 1.21–1.34), higher BMI (HR = 1.03, p < .001, 95% CI = 1.02–1.04), higher education (HR = .98, p < .05, 95% CI = .96–1.00), and being partnered (HR = 1.62, p < .05, 95% CI = 1.01–2.60) was associated with increased probability of incident frailty.

7

In addition to these covariates, compared with nondrinkers, participants who consumed 1–7 drinks had a 19% reduced probability of incident frailty (HR = .81, p < .01, 95% CI = .69–.96). Given that changes in alcohol use over the 12year time period could lead to a misclassification bias, a post-hoc sensitivity analysis was done with the subset of the sample that had their alcohol consumption level remain the same throughout the 12-year span of the study. Results (see Appendix: Supplementary Table 2) obtained from this analysis were similar to that from the original analysis and suggested that male and female participants who consumed 1–7 drinks had a 37–44% reduced probability of incident frailty (for males, HR = .56, p < .01, 95% CI = .40–.80; for females, HR = .63 p < .01, 95% CI = .48–.85). These results indicate that our original findings were less susceptible to misclassification bias. One possible explanation for these findings may be that moderate alcohol use is associated with higher mortality rates in this sample, thus producing the appearance of a protective effect with respect to frailty. This auxiliary hypothesis was tested using post-hoc logistic regression analyses with alcohol use predicting attrition. Results were that participants who reported consuming any alcohol (greater than 1 drink per week) were 29% to 31% less likely to attrite compared with those who reported abstaining (OR = .69–.71, p < .05). These results suggest that the reported relationship between alcohol use and frailty is not an artifact of heightened mortality rates among alcohol users in this sample. Lastly, post-hoc logistic regressions were run excluding participants who met criteria for binge drinking according to the NIAAA guidelines (4 or more drinks for women, 5 or more drinks for men), and results remained similar to those reported above.

Discussion Primary findings suggest consuming approximately 1–7 alcoholic drinks per week confers reduced frailty risk for older women at baseline. Specifically, after controlling for age, race, education, SES, partner status, BMI, medical burden, and CES-D score, prevalent frailty was reduced

Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 22:15 11 October 2017

8

M. SHAH ET AL.

by 51% for women who drank 1–7 alcoholic beverages per week at baseline. Findings also suggest that compared with nondrinkers, consuming 1–7 alcoholic drinks per week reduced the probability of incident frailty by 19% to 22% (HR = .79–.81, p < .05). These results are consistent with past findings that moderate alcohol use reduces frailty risk (Ortola et al., 2015; Seematter-Bagnoud et al., 2014; Woods et al., 2005). Consistent with past findings (Fried et al., 2001; Strawbridge et al., 1998; Woods et al., 2005), older age, greater medical burden, and more depressive symptomatology was consistently associated with greater frailty risk for both older men and women. These findings are suggestive of the debate on depression and frailty as separate constructs in the extant literature. Though depression and frailty may share risk factors, symptoms, and consequences (Lohman, Dumenci, & Mezuk, 2015), conventional disease models have distinguished between the two constructs. Other work, also based on a subsample of women over age 80 drawn from the HRS (Paulson & Lichtenberg, 2013) found that depression longitudinally predicted frailty, and yet, frailty is not a longitudinal predictor of subsequent depression. The current finding that depression is a robust predictor of incident frailty for both older men and women partially supports a longitudinal relationship between these variables. Future research should further examine this interesting relationship. A secondary finding of this study is that higher BMI scores was consistently associated with greater incident frailty risk for older women, but not older men. This sex discrepant finding is consistent with previous findings that suggest frailty is associated with being overweight among females (Woods et al., 2005), but not among males (Cawthon et al., 2007). This suggests that older women may be more vulnerable to certain risks, frailty in particular, associated with obesity. It is interesting that our findings did not support prior research linking excessive alcohol use to worsened health outcomes (i.e., frailty). This null effect may be attributed to having relatively few participants who reported consuming more than 15 drinks per week and the consequent reduction of statistical power. It also may be attributed to a survival effect whereby only the most resilient

heavy users of alcohol would be represented in this sample of older adults. This null finding also may reflect measurement error inherent to subjective data. Future research should identify underlying mechanisms contributing to these relationships among older adults. The primary limitation of this study is the reliance on subjective data, including alcohol use, medical burden, and frailty data. While this method is subject to measurement error, it is consistent with the vast majority of both alcohol research due to the difficulties in obtaining objective measures (e.g., blood alcohol concentration), and epidemiological research. Previous research has demonstrated reliability and validity of selfreported alcohol use measures (Del Boca & Darkes, 2003; Embree & Whitehead, 1993; Williams, Aitken, & Malin, 1985). Additionally, reviews of the reliability and validity of self-report alcohol use measures suggest that the context of the situation and interaction between the respondent and interviewer may influence the response (Midanik, 1988). Given the context of participant confidentiality and a large-scale epidemiological research project as opposed to a medical visit and perceived pressures to comply with medical professionals’ alcohol use recommendations, there is reduced risk for response bias in providing the quantity and frequency of their alcohol use habits. Prior work has shown adequate concordance between self-report of medical data and objective medical data (Bush, Miller, Goldsen, & Hale, 1989). While the self-report components of the frailty measure are subject to errors inherently associated with self-reporting, the measure overall has demonstrated construct validity (Paulson & Lichtenberg, 2014). The findings of this study extend our knowledge of the relationship between alcohol use and frailty by examining that relationship using a larger, more representative sample over a 12-year period. With the rapidly growing population of older adults, and high prevalence of alcohol use among older adults (Sorocco & Ferrell, 2006), it is important to identify how alcohol use affects the aging process as these findings can provide further empirically supported drinking guidelines for older adults. In combination with research examining harmful drinking in older adults, these

Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 22:15 11 October 2017

CLINICAL GERONTOLOGIST

findings may also be useful in both the assessment of alcohol abuse among the older adults and the development of primary interventions targeting healthy aging. Knowing how alcohol consumption relates to critical indicators of late-life decline such as frailty allows clinical healthcare personnel to provide more informed guidelines and recommendations to their patients. Future research should seek to identify the mechanism by which alcohol consumption and frailty relate, specifically the protective effects of moderate use. Various mechanisms have been proposed thus far, ranging from biological basis including HDL cholesterol (Thornton, Symes, & Heaton, 1983) to lifestyle and behaviors associated with moderate drinking, such as regular socialization (Maraldi et al., 2009).

Clinical implications ● Findings are that moderate drinking, defined as one to seven drinks per week, does not escalate health risk with respect to frailty. ● Moderate alcohol use reduces risk with respect to frailty. ● Healthcare clinicians and researchers should consider sex when examining risk factors of late-life decline.

References Abramson, J. L., Williams, S. A., Krumholz, H. M., & Vaccarino, V. (2001). Moderate alcohol consumption and risk of heart failure among older persons. Journal of American Medical Association, 285(15), 1971–1977. doi:10.1001/jama.285.15.1971 Alonso-Bouzon, C., Carcaillon, L., Garcia-Garcia, F. J., Amor-Andres, M. S., El Assar, M., & Rodriguez-Manas, L. (2014). Association between endothelial dysfunction and frailty: The Toledo study for healthy aging. Age, 36 (1), 495–505. doi:10.1007/s11357-013-9576-1 Balsa, A. I., Homer, J. F., Fleming, M. F., & French, M. T. (2008). Alcohol consumption and health among elders. The Gerontologist, 48(5), 622–636. doi:10.1093/geront/48.5.622 Blazer, D. G., & Wu, L. T. (2009). The Epidemiology of atrisk and binge drinking among middle-aged and elderly community adults: National Survey on Drug Use and

9

Health. The American Journal of Psychiatry, 166(10), 1162–1169. Bush, T. L., Miller, S. R., Goldsen, A. L., & Hale, W. E. (1989). Self-report and medical record report agreement of selected medical conditions in the elderly. American Journal of Public Health, 79(11), 1554–1556. Cawthon, P. M., Marshall, L. M., Dam, Y. M., Ensrud, T.-T., Barrett-Connor, K. E., & Orwoll, E. S. (2007). Frailty in older men: Prevalence, progression, and relationship with mortality. Journal of American Geriatrics Society, 55(8), 1216–1223. doi:10.1111/j.1532-5415.2007.01259.x Clegg, A., & Young, J. (2011). The frailty syndrome. Clinical Medicine, 11(1), 72–75. Corrao, G., Rubbiati, L., Bagnardi, V., Zambon, A., & Poikolainen, K. (2000). Alcohol and coronary heart disease: A meta-analysis. Addiction, 95(10), 1505–1523. Del Boca, F. K., & Darkes, J. (2003). The validity of selfreports of alcohol consumption: State of the science and challenges for research. Addiction, 98(s2), 1–12. Durfour, M. C. (1999). What is moderate drinking? Defining “drinks” and drinking levels. Annual Review of Medicine, 23(1), 5–14. Embree, B. G., & Whitehead, P. C. (1993). Validity and reliability of self-reported drinking behavior: Dealing with the problem of response bias. Journal of Studies on Alcohol, 54(3), 334–344. Ferreira, M. P., & Weems, M. K. (2008). Alcohol consumption by aging adults in the United States: Health benefits and detriments. Journal of American Dietetic Association, 108(10), 1668–1676. doi:10.1016/j.jada.2008.07.011 Fried, L. P., Tangen, C. M., Walston, J., Newman, A. B., Hirsch, C., Gottdiener, J., . . . McBurnie, M. A. (2001). Frailty in older adults: Evidence for a phenotype. Journal of Gerontology, 56A(3), M146–M156. Grambsch, P. M., & Therneau, T. M. (1994). Proportional hazards tests and diagnostics based on weighted residuals. Biometrika, 81(3), 515–526. doi:10.1093/biomet/81.3.515 Hauser, R. M., & Willis, R. J. (2004). Survey design and methodology in the Health and Retirement Study and the Wisconsin Longitudinal Study. Population and Development Review, 30, 209–235. Heeringa, S. G., & Conner, J. (1995). Technical description of the Health and Retirement Study sample design: HRS/ AHEAD Documentation Report DR-002. Ann Arbor, MI: Survey Research Center, Institute for Social Research, University of Michigan. Lang, I., Guralnik, J., Wallace, R. B., & Melzer, D. (2007). What level of alcohol consumption is hazardous for older people? Functioning and mortality in U.S. and English national cohorts. Journal of American Geriatrics Society, 55(1), 49–57. doi:10.1111/j.1532-5415.2006.01007.x Lang, I., Wallace, R. B., Huppert, F. A., & Melzer, D. (2007). Moderate alcohol consumption in older adults is associated with better cognition and well-being than abstinence. Age and Ageing, 36(3), 256–261. doi:10.1093/ ageing/afm001

Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 22:15 11 October 2017

10

M. SHAH ET AL.

Lohman, M., Dumenci, L., & Mezuk, B. (2015). Depression and frailty in late life: Evidence for a common vulnerability. Journals of Gerontology Series B: Psychological Sciences and Social Sciences. doi:10.1093/geronb/gbu180 Maraldi, C., Harris, T. B., Newman, A. B., Kritchevsky, S. B., Pahor, M., Koster, A., . . . Health, A. B. C. S. (2009). Moderate alcohol intake and risk of functional decline: The Health, Aging, and Body Composition study. Journal of the American Geriatric Society, 57(10), 1767–1775. doi:10.1111/j.1532-5415.2009.02479.x McGuire, L. C., Ajani, U. A., & Ford, E. S. (2007). Cognitive functioning in late life: The impact of moderate alcohol consumption. Annals of Epidemiology, 17(2), 93–99. doi:10.1016/j.annepidem.2006.06.005 Midanik, L. T. (1988). Validity of self-reported alcohol use: A literature review and assessment. Addiction, 83(9), 1019– 1029. doi:10.1111/j.1360-0443.1988.tb00526.x Mukherjee, D., Levin, R. L., & Heller, W. (2006). The cognitive, emotional, and social sequelae of stroke: Psychological and ethical concerns in post-stroke adaptation. Topics in Stroke Rehabilitation, 13(4), 26–35. National Institute on Alcohol Abuse and Alcoholism. (1998). Alcohol and aging. Retrieved from http://pubs.niaaa.nih. gov/publications/aa40.htm Ortola, R., Garcia-Esquinas, E., Leon-Munoz, L. M., GuallarCastillon, P., Valencia-Martin, J. L., Galan, I., & Rodriguez-Artalejo, F. (2015). Patterns of alcohol consumption and risk of frailty in community-dwelling older adults. The Journals of Gerontology. Series A, Biological Sciences and Medical Sciences. doi:10.1093/gerona/glv125 Paulson, D., & Lichtenberg, P. A. (2013). Vascular depression and frailty: A compound threat to longevity among olderold women. Aging and Mental Health, 17, 901–910. doi:10.1080/13607863.2013.799115 Paulson, D., & Lichtenberg, P. A. (2014). The PaulsonLichtenberg Frailty Index: Evidence for a self-report measure of frailty. Aging and Mental Health, 19(10), 892–901. doi:10.1080/13607863.2014.986645 Radloff, L. S. (1977). The CES-D scale: A self-report depression scale for research in the general population. Applied Psychological Measurement, 1(3), 385–401. Rimm, E. B., Klatsky, A., Grobbee, D., & Stampfer, M. J. (1996). Review of moderate alcohol consumption and reduced risk of coronary heart disease: Is the effect due to beer, wine, or spirits? Bmj, 312(7033), 731–736. doi:10.1136/bmj.312.7033.731

Seematter-Bagnoud, L., Spagnoli, J., Bula, C., & SantosEggimann, B. (2014). Alcohol use and frailty in community-dwelling older adults persons aged 65 to 70 years. Journal of Frailty and Aging, 3(1), 9–14. Shrestha, L. B., & Heisler, E. J. (2011). The changing demographic profile of the United States. CRS Report for Congress. Sorocco, K. H., & Ferrell, S. W. (2006). Alcohol use among older adults. The Journal of General Psychology, 133(4), 453–467. Strawbridge, W. J., Shema, S. J., Balfour, J. L., Higby, H. R., & Kaplan, G. A. (1998). Antecedents of frailty over three decades in an older cohort. Journal of Gerontology, 53B(1), 9–16. Substance Abuse and Mental Health Services Administration. (2014). Results from the 2013 national survey on drug use and health: Summary of national findings (NSDUH Series H-48, HHS Publication No. (SMA) 14-4863). Rockville, MD: Author. Thornton, J., Symes, C., & Heaton, K. (1983). Moderate alcohol intake reduces bile cholesterol saturation and raises HDL cholesterol. The Lancet, 322(8354), 819–822. doi:10.1016/S0140-6736(83)90738-9 Thun, M. J., Peto, R., Lopez, A. D., Monaco, J. H., Henley, S. J., Heath, C. W., & Doll, R. (1997). Alcohol consumption and mortality among middle-aged and elderly U.S. adults. The New Engalnd Journal of Medicine, 337(24), 1705– 1714. U.S. Department of Agriculture, & U.S. Department of Health and Human Services. (2010). Dietary guildelines for Americans 2010. Washington, DC: US Government Printing Office. Walston, J., McBurnie, M. A., Newman, A., Tracy, R. P., Kop, W. J., Hirsch, C. H., . . . Fried, L. P. (2002). Frailty and activation of the inflammation and coagulation systems with and without clinical comorbidities. Archives of Internal Medicine, 162(20), 2333–2341. doi:10.1001/ archinte.162.20.2333 Williams, G. D., Aitken, S. S., & Malin, H. (1985). Reliability of self-reported alcohol consumption in a general population survey. Journal of Studies on Alcohol, 46(3), 223– 227. Woods, N. F., LaCroix, A. Z., Gray, S. L., Aragaki, A., Cochrane, B. B., Brunner, R. L., . . . Women’s Health, I. (2005). Frailty: Emergence and consequences in women aged 65 and older in the Women’s Health Initiative Observational Study. Journal of American Geriatrics Society, 53(8), 1321–1330. doi:10.1111/ j.1532-5415.2005.53405.x

CLINICAL GERONTOLOGIST

11

Appendix: Supplementary tables Supplementary Table 1. Cox proportional hazard model results without medical burden and CES-D. Males

Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 22:15 11 October 2017

Characteristic Age Education Race Partnered SES BMI Drinks 1–7 Drinks 8–14 Drinks 15+

HR 1.08 .98 .90 .97 .98 1.04 .78 .88 1.24

Z 11.47*** −2.06* −1.14 −.29 −3.03** 3.55*** −2.57** −.86 1.18

Females 95% CI 1.06–1.09 .96–1.00 .75–1.08 .80–1.12 .96–.99 1.02–1.06 .64–.94 .66–1.18 .87–1.76

HR 1.06 .96 .94 .95 1.00 1.04 .79 1.06 1.57

Z 13.71*** −3.98*** −1.01 −.78 −.23 7.69*** −2.88** .36 1.54

95% CI 1.05–1.07 .95–.98 .82–1.06 .85–1.07 .99–1.01 1.03–1.05 .67–.93 .76–1.49 .88–2.77

***p ≤ .001. **p ≤ .01. *p ≤ .05. The reference group includes abstinent participants (0 drinks per week). These survival analyses results are without the medical burden and CES-D variable in order to meet all of the assumptions.

Supplementary Table 2. Complete sensitivity analysis results. Males Characteristic Age Education Race Partnered CES-D SES BMI Medical Burden Drinks 1–7 Drinks 8–14 Drinks 15+

HR 1.06 1.00 .89 1.04 1.16 1.00 1.01 1.30 .56 .61 .88

Z 7.29*** .04 −.90 .28 4.69*** −.54 .67 5.81*** −3.24** −.97 −.29

Females 95% CI 1.05–1.08 .97–1.03 .70–1.14 .80–1.35 1.09–1.23 .98–1.01 .98–1.04 1.19–1.42 .40–.80 .23–1.65 .36–2.13

HR 1.06 .98 .93 .92 1.11 1.00 1.03 1.28 .63 1.25 2.71

Z 10.48*** −1.82 −.98 −1.23 5.92*** .41 4.20*** 8.49*** −3.09** .50 1.98*

95% CI 1.05–1.07 .96–1.00 .81–1.07 .80–1.05 1.07–1.15 .99–1.01 1.01–1.04 1.21–1.35 .48–.85 .52–3.03 1.01–7.27

***p ≤ .001. **p ≤ .01. *p ≤ .05. The reference group includes abstinent participants (0 drinks per week). This post-hoc sensitivity analysis was done with the subset of the sample that had their alcohol consumption level remain the same throughout the 12-year span of the study.