Relation Between Smoking Status and Subjective Cognitive Decline in Middle Age and Older Adults: A Cross-Sectional Analysis of 2019 Behavioral Risk Factor Surveillance System Data

Abstract

Background:

Smoking status may influence subjective cognitive decline (SCD); however, few studies have evaluated this association.

Objective:

To assess whether smoking status is associated with SCD among middle age and older adults, and to determine if this association is modified by sex at birth.

Methods:

A cross-sectional analysis was conducted using data from the 2019 Behavioral Risk Factor Surveillance System (BRFSS) survey to analyze the relationship between SCD and smoking status (current, recent former, and remote former). Eligible respondents included participants 45 years of age or older who responded to the SCD and tobacco questions of interest. Survey-weighted Poisson regression models were employed to estimate the crude and adjusted prevalence ratios (cPR/aPR) and corresponding 95% confidence intervals (CI) of the association between smoking status and SCD. A Wald test was computed to determine the significance of the interaction term between smoking status and sex (α= 0.05).

Results:

There were 136,018 eligible respondents, of which approximately 10% had SCD. There was a graded association between smoking and SCD, with the greatest prevalence of SCD among current smokers (aPR = 1.87; CI: 1.54, 2.28), followed by recent former smokers (aPR = 1.47; 95% CI: 1.02, 2.12), and remote former smokers (aPR = 1.11; 95% CI: 0.93, 1.33) each compared to never smokers. There was no evidence of effect modification by sex (p interaction = 0.73).

Conclusion:

The consistency of smoking as a risk factor for objective and subjective cognitive decline supports the need for future studies to further the evidence on whether changes to smoking status impacts cognition in middle age.

Keywords

Alzheimer’s disease cigarette smoking cognition cognitive aging dementia

INTRODUCTION

Cognitive decline is a public health priority as the older American population increases. By 2030, over 20% of the total U.S. population will be over the age of 64 years [1]. When considering the growing burden of neurocognitive disorders, it is important to understand how age-related changes in cognition impact memory, language, executive functioning, learning, and attention, which may be assessed by measuring cognitive decline subjectively [2]. Subjective cognitive decline (SCD) is the worsening of cognition attributed to the self-reported increase in the frequency of memory loss or confusion [3]. SCD may indicate early signs of mild cognitive impairment, Alzheimer’s disease, and other related dementias [4]. While objective measures of cognition include standardized neuropsychological tests such as cognitive batteries, normal performance on these standardized cognitive assessments is typically a criterion used to characterize those with SCD [5]. Although SCD does not indicate a formal diagnosis of cognitive decline by a health care professional, it has important implications for an individual’s ability to perform activities of daily living, manage chronic conditions, or care for themself [6]. Evaluating SCD may help to identify population level estimates for factors impacting cognition that are not often identified by formal assessments or diagnoses of cognitive decline [7].

SCD prevalence has been found to be higher among adults reporting modifiable risk factors for dementia including hypertension, physical inactivity, obesity, diabetes, depression, hearing loss, and current cigarette smoking, and all have been evaluated in association with SCD in the U.S. aside from cigarette smoking [6 , 8–12]. Smoking has been identified as a potential modifiable risk factor for dementia and cognitive decline due to its impact on lung function and risk of respiratory and cardiovascular disease [13]. The link between smoking and lung function influences the risk of cardiovascular risk factors, such as diabetes and hypertension, which in turn may exacerbate cognitive decline and dementia risk [13]. Reduced lung function alone has also been linked with worse performance on cognitive assessments and may impact future risk of cognitive decline and dementia through mechanisms such as the progression of a pro-inflammatory state [14]. The progression may be more stark in current smokers, as they may be at a greater risk of objective cognitive decline compared to former smokers [15 –18]. However, the association between former smoking status and cognitive decline is inconsistent [15 –18], as few studies have separated recent from remote former smokers, where recent former smokers last smoked a cigarette between 1 and less than 10 years ago and remote former smokers last smoked a cigarette 10 or more years ago. Nor has there been consistency in whether sex modifies the association between smoking status and cognitive decline. Across two previous cohort studies, no association between smoking and a five-year change in objective cognitive function existed by sex in one study [15], while the second study observed an association in 10-year cognitive decline among current smokers compared to never smokers in men but not in women [18]. While the differential association in the second study was attributed to the larger quantity of cigarettes smoked in men compared to women in the study population, a similar study with a larger sample size has not yet been conducted to further explore this potential difference. Furthermore, SCD represents a novel way to evaluate questions related to cognition as a potential self-reported marker of future cognitive decline [6, 7]. Thus, efforts need to be taken to understand the impact of modifiable dementia risk factors, such as smoking, on SCD [19, 20].

Using 2019 Behavioral Risk Factor Surveillance System (BRFSS) data, we conducted a cross-sectional study to assess the association between smoking and SCD in middle age and older adults. Additionally, we sought to evaluate whether sex modifies the association between smoking status and SCD. We hypothesized that smoking status is associated with SCD, with current smokers having the highest rates of SCD compared to never smokers among all categories of smoking status. Further, we hypothesized that the association between smoking status and SCD may differ by sex, where the associations between current, recent, and remote former smoking status may be stronger in males compared to females.

MATERIALS AND METHODS

Study population

A secondary analysis was conducted using data from the BRFSS, an annual survey collected by states and territories in the United States (U.S.) since 1984 where the Centers for Disease Control and Prevention (CDC) aggregates and oversees the data collection system. BRFSS surveys noninstitutionalized adults residing in the U.S. to assess health-related risk behaviors, use of preventive services, and chronic health conditions [21]. A complex sampling design using both landline and cellular telephone-based survey samples is employed to ensure that participating states meet the BRFSS standards for telephone numbers selected as a probability sample of all households with a telephone in a given state. Further details on the BRFSS sample design and weighting methodology are published elsewhere [22].

A total of 418,268 adults from all 50 states, the District of Columbia, Guam, and Puerto Rico participated in the BRFSS during the year of 2019. The SCD question of interest was only asked to participants 45 years of age and older, thus, those under the age of 45 years were excluded from the analysis. Additionally, participants were excluded if they did not respond to both the question measuring SCD and the questions regarding tobacco use [23]. While all respondents were asked questions on tobacco use, only respondents residing in states that chose to incorporate the cognitive decline module within their 2019 BRFSS questionnaire were asked the SCD question of interest. Thus, the following 32 states were included in the final sample: Alabama, Connecticut, District of Columbus, Florida, Georgia, Indiana, Iowa, Kansas, Louisiana, Maryland, Michigan, Minnesota, Mississippi, Missouri, Nebraska, Nevada, New Mexico, New York, North Dakota, Ohio, Oklahoma, Oregon, Pennsylvania, Rhode Island, South Carolina, South Dakota, Tennessee, Texas, Utah, Virginia, West Virginia, and Wisconsin. There were 136,018 total eligible participants included in analysis. Publicly available and de-identified data were utilized in this analysis, therefore, this study is not deemed as human subjects research by The Ohio State University Institutional Review Board and Office of Responsible Research Practices.

Measures

The outcome of interest, SCD, was derived from the self-reported questionnaire item “During the past 12 months, have you experienced confusion or memory loss that is happening more often or is getting worse?” Those eligible participants who answered in the affirmative to this item were identified as participants with SCD. Eligible responses of “no” were coded as “no” for SCD, while responses of “don’t know/not sure” or refused were coded as missing.

The primary exposure of interest in this study was smoking status measured using four categories of smoking status: current smoker, recent former smoker (more than 1 year but less than 10 years since the last time a cigarette has been smoked), remote former smoker (10 years or more since the last time a cigarette has been smoked), and never smoker. Current smokers were participants who smoked at least 100 cigarettes in their life and smoked either every day or some days at the time of the questionnaire. Former smokers were participants who smoked at least 100 cigarettes in their life but were not currently smoking every day or some days. They were further categorized into two groups: those who reported smoking within the past 10 years or 10 years or more. Never smokers were identified as participants who had not smoked at least 100 cigarettes in their life. Participants who answered “don’t know/not sure” or refused to answer either the SCD or smoking status questions of interest were coded as missing.

Participants self-reported their age, sex at birth, race/ethnicity, highest education level completed, and whether they were ever told they had high blood pressure or diabetes. Supplementary Table 1 offers a breakdown of the categorization of all covariates according to their corresponding question(s) within the 2019 BRFSS. Missing data existed for approximately 2.14% of the self-reported responses of race/ethnicity, therefore, BRFSS imputed the missing responses for race/ethnicity with the most common race/ethnicity response for the region of the state in which the respondent resided [24]. Aside from race/ethnicity, responses of “don’t know/not sure” were coded as missing for the other covariates.

Statistical analysis

The 2019 BRFSS dataset was reweighted using clustering and stratification variables defined by BRFSS through complex sample weighting methodology [25]. Frequency distributions were reported for each covariate in the overall sample, and by the presence or absence of SCD. The reported observations for the overall sample included the nationally represented weighted percentages.

A univariate survey-weighted Poisson regression model was employed to first estimate the crude prevalence ratio (cPR) and 95% confidence interval (CI) of the unadjusted association between smoking status and SCD. A multivariable survey-weighted Poisson regression model was then fit to estimate the adjusted prevalence ratio (aPR) controlling for the following potential confounders of the association between smoking and cognitive decline, as presented in the literature: age, sex at birth, race/ethnicity, and highest education level completed [15 , 27]. Vascular risk factors, such as hypertension and diabetes, were not included in the adjusted Poisson regression model because of their identification as mediators of the association between smoking status and cognitive decline [28]. An interaction term between smoking status and sex at birth was then included and a Wald test was used to assess the presence of effect measure modification by sex (p < 0.05 considered statistically significant). All analyses were conducted using SAS 9.4 (SAS Institute Inc., Cary, NC) and Stata 17 (StataCorp, College Station, TX).

Secondary analysis

To understand nonresponse, as defined as participants with missing data on SCD or smoking status, we descriptively compared demographics of the overall eligible sample with the demographics of nonrespondents. Additionally, we descriptively assessed the eligible proportion of participants within each state, or the proportion of those who responded to the SCD and smoking questions of interest among all respondents 45 years of age and older across the states included in analysis.

To evaluate the robustness of our results, we evaluated the association between smoking and SCD within subgroups of middle-aged adults (i.e., between 45 to 59 years of age) and older adults (i.e., 60 years of age and older) to determine if the association varied by middle and older age groups.

RESULTS

A total of 136,018 respondents meeting the study eligibility criteria were included for analysis (Fig. 1). The proportion of the study population with SCD was 10.8%. Descriptive characteristics of the study population, both overall and according to SCD status, are presented in Table 1. Overall, more than half of the sample respondents were classified as never smokers (55.0%), were female (53.6%), and self-reported their race as white, non-Hispanic (74.2%). Similarly, over half of respondents without SCD were never smokers (56.6%), about half reported having hypertension (51.5%), and under a quarter reported having diabetes (17.3%). Among those with SCD, over half of respondents reported having hypertension (61.0%) and just over a quarter reported having diabetes (28.9%).

Fig. 1

Flowchart outlining the selection of eligible respondents residing in the United States from the 2019 Behavioral Risk Factor Surveillance System (BRFSS) included in analysis.

Table 1

Characteristics of a sample population (overall and according to SCD status) and weighted population estimates from a sample of BRFSS respondents of the 2019 landline and cellular telephone surveys (Sample N = 136,018)

	Sample		SCD		No SCD
	N^*	% ^†	N^*	% ^†	N^*	% ^†
Smoking Status
Never Smoker	76,099	55.0	6,637	42.8	69,462	56.6
Current Smoker	18,593	15.3	3,138	25.1	15,455	14.1
Recent Former Smoker	8,176	6.5	1,125	7.8	7,051	6.3
Remote Former Smoker	33,150	23.2	4,099	24.3	29,051	23.0
Missing	0		0		0
Sex at Birth
Male	58,511	46.4	6,609	44.6	51,902	46.6
Female	77,507	53.6	8,390	55.4	69,117	53.4
Missing	0		0		0
Age
45–49	11,734	12.3	1,083	10.6	10,651	12.5
50–54	14,780	15.9	1,553	16.2	13,227	15.8
55–59	17,685	14.5	1,899	14.3	15,786	14.5
60–64	20,465	16.3	2,096	15.6	18,369	16.4
65–69	21,393	12.9	2,007	10.9	19,386	13.1
70–74	19,515	11.4	1,974	10.4	17,541	11.6
75–79	13,629	7.8	1,715	9.5	11,914	7.6
≥80	16,817	8.9	2,672	12.5	14,145	8.5
Missing	0		0		0
Race/Ethnicity
White, Non-Hispanic	112,966	74.2	12,115	71.9	100,851	74.4
Black, Non-Hispanic	11,669	12.2	1,385	13.1	10,284	12.1
Other race, Non-Hispanic	5,630	4.8	785	5.5	4,845	4.7
Hispanic	5,753	8.8	714	9.5	5,039	8.8
Missing	0		0		0
Education Level^‡
Did not graduate high school	9,748	12.6	1,842	21.2	7,906	11.5
Graduated high school	37,528	28.7	4,666	30.9	32,862	28.4
Attended college/technical school	37,288	29.7	4,290	28.7	32,998	29.8
Graduated college/technical school	51,031	29.0	4,166	19.2	46,865	30.3
Missing	423		35		388
Hypertension^§
Yes	70,851	49.9	9,310	61.0	61,541	51.5
No	64,727	50.1	5,641	39.0	59,086	48.5
Missing	440		48		392
Diabetes^#
Yes	25,661	18.6	4,148	28.9	21,513	17.3
No	110,133	81.4	10,814	71.1	99,319	82.7
Missing	224		37		187

^*Unweighted sample size (overall and according to presence or absence of SCD). ^†Percentages were weighted to account for the complex weighting methodology employed. ^‡Information on education level was missing for 423 respondents. ^§Information on hypertension was missing for 440 respondents. ^#Information on diabetes was missing for 224 respondents.

According to the unadjusted univariate model, all categories of smoking status were associated with SCD (Table 2). A dose-response association between smoking status and SCD was present; current smokers had 2.12 times the prevalence of reporting SCD compared to never smokers (95% CI: 1.96, 2.29). Recent former smoking status and remote former smoking status were also both positively associated with SCD when compared to never smoking status (cPR = 1.55; 95% CI: 1.39, 1.73; cPR = 1.35; 95% CI: 1.26, 1.45, respectively). The dose-response association between all categories of smoking status and SCD was consistent after adjusting for age, sex at birth, race/ethnicity, and highest education level completed in a multivariable survey-weighted Poisson model. The adjusted association between current smoking status and SCD was slightly attenuated (aPR = 1.87; CI: 1.54, 2.38), as were the associations between recent former smoking status and SCD (aPR = 1.47; 95% CI: 1.02, 2.12) and remote former smoking status and SCD (aPR = 1.11; 95% CI: 0.93, 1.33). The difference in the prevalence of SCD between remote former smokers and current smokers was not significant with adjustment. There was no evidence of effect measure modification by sex at birth (Wald p interaction = 0.73).

Table 2

Univariate and multivariable survey-weighted Poisson regression models of the association between smoking status and SCD: BRFSS 2019

	Unadjusted Model^*		Adjusted Model^†
Characteristic	cPR	95% CI	aPR	95% CI
Smoking Status
Never Smoker	1.0 (referent)	–	1.0 (referent)	–
Current Smoker	2.12	1.96, 2.29	1.87	1.54, 2.28
Recent Former Smoker	1.55	1.39, 1.73	1.47	1.02, 2.12
Remote Former Smoker	1.35	1.26, 1.45	1.11	0.93, 1.33

^*Crude prevalence ratio (cPR) and 95% confidence interval (CI) for the outcome of subjective cognitive decline (SCD). ^†Adjusted prevalence ratio (aPR) and 95% CI for the outcome of subjective cognitive decline (SCD) and adjusting for age (45–49, 50–54, 55–59, 60–64, 65–69, 70–74, 75–79, and ≥80 years of age), sex at birth (male/female), race/ethnicity (White/non-Hispanic, Black/non-Hispanic, Other race/non-Hispanic, and Hispanic), and highest education level completed (did not graduate high school, graduated high school, attended college/technical school, and graduated college/technical school).

Secondary analysis

Nonrespondents with missing SCD or smoking status data included 25,923 participants that were not incorporated into the final analytical sample. Overall, nonrespondents appeared to be younger than respondents, with a greater proportion of nonrespondents falling between the ages of 45 and 54 years (34.5%) compared to the proportion of respondents with SCD (26.8%) and without SCD (28.3%) within the same age range (Supplementary Table 2). The eligible proportions of respondents were common across all states, aside from Florida, Georgia, Utah, and North Dakota (Supplementary Table 3). Florida and Georgia had lower eligible proportions compared to all other states (76.7% and 77.6%, respectively), while Utah and North Dakota had the highest eligible proportions (91.7% and 92.0%, respectively). The median eligible proportion was 84.3%.

When stratifying by age, the association between current and recent former smoking status and SCD was stronger among middle-aged adults, whereas the prevalence of SCD did not significantly vary by smoking status among older adults (Supplementary Table 4).

DISCUSSION

Overall, we observed a positive, dose-response association between smoking status and subjective cognitive decline among adults 45 years of age and older. Among this graded association, current smokers were associated with the greatest rates of SCD, followed by those who had last smoked within the past 1 to <10 years (recent former smokers), followed by those who had last smoked 10 or more years ago (remote former smokers), all relative to those who had never smoked. The association between smoking and SCD was not modified by sex at birth.

Our findings are consistent with the previous knowledge on the association between smoking status and cognitive decline assessed longitudinally and according to a cognitive battery rather than a self-reported assessment [15–18 , 29]. Our results indicate that the prevalence of reporting SCD among current smokers is 1.87 times that of never smokers. Using a neuropsychological test battery, a previous cohort study found significant declines in memory function, cognitive flexibility, and global cognitive function that were 1.9, 2.4, and 1.7 times greater in current smokers compared to never smokers, respectively [15]. A meta-analysis of prospective studies with at least 12 months of follow-up assessing the association between smoking and objective cognitive decline also found a significant association between smoking status and cognitive decline, with current smokers experiencing a larger decline in cognition over the course of follow-up compared to never smokers [16]. Although direct comparisons are difficult due to the nature of the differing cognitive assessment mechanisms, our findings support the existing knowledge on the association between smoking status and cognitive decline with evidence from a large sample representative of 32 U.S. states using a population-based measure of cognition.

Few studies have distinguished between recent and remote former smokers when analyzing the association between smoking and cognitive decline. However, a previous cohort study found that remote former smokers did not experience greater cognitive decline compared to never smokers across study follow-up, possibly due to the reduction of the residual effects of smoking over time [18]. This aligns with the graded association implied by our results, where the prevalence of SCD was greater among recent former smokers compared to never smokers, with no significant difference between the prevalence of SCD in remote former smokers compared to never smokers. While the designs, methods used to assess cognitive decline (i.e., SCD versus use of a cognitive battery), and sample size of remote former smokers (fewer remote smokers in the cohort study by over 10-fold) are different, the results both reinforce a possible attenuation in the smoking and cognitive decline association with time since smoking cessation.

There was no evidence that sex at birth is an effect measure modifier of the association between smoking status and SCD. These results are consistent with one of the two previous cohort studies mentioned that observed no interaction effect between sex and smoking in association with cognitive decline assessed objectively [15]. While the similar cohort study observed an association between smoking history and 10-year objective cognitive decline in men but not in women, the findings were attributed to the larger quantity of cigarettes smoked and increased pack-years of smoking among the men compared to the women within their study population. With support from our large sample population, it may be implied that the association between smoking and SCD is similar for males and females. However, we were unable to account for the intensity of smoking, rather only the duration.

Evaluating SCD as a midstream marker of cognitive decline raises opportunities to consider the possible impacts of interventions at younger ages on future risk of objective cognitive function [7]. In our study, the dose-response relation between levels of smoking status and SCD implies that time since cessation may be linked to cognitive outcomes. Thus, our findings raise the question as to whether smoking cessation interventions in middle adulthood could have measurable impacts on future cognitive status. Future longitudinal studies are necessary to address this question, as temporality cannot be assessed using measures of a cross-sectional nature, such as SCD. However, our findings underline the role that population-based measures like SCD play in the early detection of cognitive decline to inform further research and interventions that could mitigate the future population-level burden of cognitive disorders. Targeting middle-aged current and recent former smokers for SCD screening efforts may be considered, as the associations between these two levels of smoking status in the overall sample were likely driven by respondents in this younger agegroup.

Our analysis included a large sample representative of 32 U.S. states, however, selection bias may be present due to the optional nature of the cognitive decline module impacting the number of states included in analysis, as well as the telephone-based survey design. States choose to include an optional module within their annual BRFSS questionnaire based on need [30]. Thus, it is plausible that increasing trends in the prevalence of Alzheimer’s disease and related dementias, increased long-term care needs, or similar reasons may explain why the states chose to opt into the cognitive decline module. A bias away from the null may exist, as the prevalence of SCD among states included may be higher than those states not included. However, all 50 states had included this module within their questionnaire at least once by 2018 [7]. Differential smoking policies across states, such as tobacco taxes, may also influence differences in the association between smoking and SCD among the states in the sample compared to those excluded. Additionally, despite the expansion of the sampling methods to include the cellular telephone survey in addition to the landline telephone survey, only non-institutionalized individuals are reached through the BRFSS design. However, the comparison study populations also consisted of only non-institutionalized adults. Further, the self-reported nature of the questionnaire poses an important tradeoff, as it supports the analysis of population-based estimates without requiring a clinical or objective assessment, such as SCD. Missing data due to item nonresponse to the SCD and smoking status questions of interest were identified across all states, which includes those participants who responded don’t know/not sure, refused, or were missing a response to the questions of interest. While a potential source of bias, item nonresponse may be explained by the nature of closed-ended dichotomous questions assessing SCD, which may lead to participants feeling forced into making a decision regarding the presence or absence of SCD and motivate responses of “don’t know/not sure” [31]. As exemplified in the descriptive sensitivity analysis, nonrespondents appeared to be younger in comparison to the respondents. Therefore, it is possible that these individuals were not able or willing to confidently respond to the SCD question of interest due to feeling forced or uncertain as to whether the question applies to an individual within their age group. The single question used to evaluate SCD in the BRFSS also does not confirm normal performance on a standardized cognitive assessment, nor does it address the concerns of the respondents regarding their feelings of worsening confusion or memory loss. While there is currently no gold standard measure to distinguish SCD from normal cognitive function, the standardization of SCD assessments is evolving to formalize a definition of the term to facilitate comparability across research settings and beyond [32]. For example, a set of criteria associated with SCD and preclinical Alzheimer’s disease is also being developed to evaluate several symptoms as an indication for preclinical dementia, which is referred to as SCD-plus [32].

Residual confounding of the estimated associations between smoking status and SCD may exist due to the inability to adjust for potential unmeasured risk factors of both SCD and smoking status. However, BRFSS collects data on a wide range of health-risk behaviors and socio-demographics, thus, important confounders identified a priori were accounted for in adjustment, and chronic conditions identified as prominent mediators, such as hypertension and diabetes, were able to be considered. Lack of adjustment for history of a depressive disorder may induce confounding, however, depression may also serve as a mediator or collider on the causal pathway between smoking and SCD, which is why it was not included in the analysis. While the potential impact of lifestyle factors, such as physical activity and obesity, and other vascular factors, such as history of stroke and myocardial infarction, are not directly addressed in this study, adults with SCD have been found to have a higher prevalence of reporting these factors [8]. Exposure to smoking in addition to these factors may exacerbate the observed associations in this study and should be further explored. Inability to account for quantity of tobacco use or number of cigarettes smoked across all categories of smoking status may also contribute to the presence of residual confounding. Moreover, although over- or under-reporting of the self-reported measures may have occurred, BRFSS prevalence estimates of SCD, tobacco use, and chronic conditions, such as diabetes, have been found to be comparable to corresponding estimates within the National Health and Nutrition Examination Survey (NHANES) and National Health Interview Survey (NHIS) [7, 33]. These findings support the validity of the SCD question as a population-based measure, as well as a history of reliable and valid measures within BRFSS questionnaires.

In summary, the findings of our cross-sectional study representative of 32 U.S. states support a possible dose-response trend in the relation between smoking status and SCD. While supporting the consistency in the relations between objective and subjective cognitive decline with smoking status, as well as a graded association by level of smoking status driven by middle-aged adults, our results raise the question as to whether smoking cessation intervention efforts in middle adulthood could indicate long-term cognitive benefits in later years, as well as the possible value of targeting middle-aged adult current and recent former smokers for SCD screening. Future epidemiological studies are necessary to further the evidence as to whether changes to smoking status in middle age could have measurable impacts on cognition. Understanding the mechanism of this relation is imperative to guide future prevention efforts using this population-based measure of SCD.

DISCLOSURE STATEMENT

Authors’ disclosures available online (https://www.j-alz.com/manuscript-disclosures/22-0501r2).

Footnotes

The supplementary material is available in the electronic version of this article: .

References

Alzheimer’s Association (2018) 2018 Alzheimer’s disease facts and figures. Alzheimers Dement 14, 367–429.

Wagster

, King

, Resnick

, Rapp

(2012) The 87% guest editorial. J Gerontol 67, 739–740.

Alzheimer’s Association and Centers for Disease Control and Prevention (2013) The Healthy Brain Initiative: The Public Health Road Map for State and National Partnerships, 2013–2018.

Hebert

, Weuve

, Scherr

, Evans

(2013) Alzheimer disease in the United States (2010–2050) estimated using the 2010 census. Neurology 80, 1778–1783.

Studart

, Nitrini

(2016) Subjective cognitive decline: The first clinical manifestation of Alzheimer’s disease? Dement Neuropsychol 10, 170–177.

Jessen

, Amariglio

, Buckley

, van der Flier

, Han

, Molinuevo

, Rabin

, Rentz

, Rodriguez-Gomez

, Saykin

, Sikkes

SAM

, Smart

, Wolfsgruber

, Wagner

(2020) The characterisation of subjective cognitive decline. Lancet Neurol 19, 271–278.

Olivari

, Baumgart

, Taylor

, McGuire

(2021) Population measures of subjective cognitive decline: A means of advancing public health policy to address cognitive health. Alzheimers Dement (N Y) 7, e12142.

Omura

, McGuire

, Patel

, Baumgart

, Lamb

, Jeffers

, Olivari

, Croft

, Thomas

, Hacker

(2022) Modifiable risk factors for Alzheimer disease and related dementias among adults aged ≥45 years — United States, 2019. MMWR Morb Mortal Wkly Rep 71, 680–685.

Curhan

, Willett

, Grodstein

, Curhan

(2019) Longitudinal study of hearing loss and subjective cognitive function decline in men. Alzheimers Dement 15, 525–533.

10.

Curhan

, Willett

, Grodstein

, Curhan

(2019) Longitudinal study of self-reported hearing loss and subjective cognitive function decline in women. Alzheimers Dement 16, 610–620.

11.

Omura

, Brown

, McGuire

, Taylor

, Fulton

, Carlson

(2020) Cross-sectional association between physical activity level and subjective cognitive decline among US adults aged ≥45 years, 2015. Prev Med 141, 106279.

12.

Zlatar

, Muniz

, Galasko

, Salmon

(2018) Subjective cognitive decline correlates with depression symptoms and not with concurrent objective cognition in a clinic-based sample of older adults. J Gerontol B Psychol Sci Soc Sci 73, 1198–1202.

13.

Pathan

, Gottesman

, Mosley

, Knopman

, Sharrett

, Alonso

(2011) Association of lung function with cognitive decline and dementia: The Atherosclerosis Risk in Communities (ARIC) Study. Eur J Neurol 18, 888–898.

14.

Engström

, Lind

, Hedblad

, Wollmer

, Stavenow

, Janzon

, Lindgärde

(2002) Lung function and cardiovascular risk. Circulation 106, 2555–2560.

15.

Nooyens

ACJ

, van Gelder

, Verschuren

WMM

(2008) Smoking and cognitive decline among middle-aged men and women: The Doetinchem Cohort Study. Am J Public Health 98, 2244–2250.

16.

Anstey

, von Sanden

, Salim

, O’Kearney

(2007) Smoking as a risk factor for dementia and cognitive decline: A meta-analysis of prospective studies. Am J Epidemiol 166, 367–378.

17.

Deal

, Power

, Palta

, Alonso

, Schneider

ALC

, Perryman

, Bandeen-Roche

, Sharrett

(2020) Relationship of cigarette smoking and time of quitting with incident dementia and cognitive decline. J Am Geriatr Soc 68, 337–345.

18.

Sabia

, Elbaz

, Dugravot

, Head

, Shipley

, Hagger-Johnson

, Kivimaki

, Singh-Manoux

(2012) Impact of smoking on cognitive decline in early old age: The Whitehall II Cohort Study. Arch Gen Psychiatry 69, 627–635.

19.

Molinuevo

, Rabin

, Amariglio

, Buckley

, Dubois

, Ellis

, Ewers

, Hampel

, Klöppel

, Rami

, Reisberg

, Saykin

, Sikkes

, Smart

, Snitz

, Sperling

, van der Flier

, Wagner

, Jessen

(2017) Implementation of subjective cognitive decline criteria in research studies. Alzheimers Dement 13, 296–311.

20.

Rabin

, Smart

, Crane

, Amariglio

, Berman

, Boada

, Buckley

, Chételat

, Dubois

, Ellis

, Gifford

, Jefferson

, Jessen

, Katz

, Lipton

, Luck

, Maruff

, Mielke

, Molinuevo

, Naeem

, Perrotin

, Petersen

, Rami

, Reisberg

, Rentz

, Riedel-Heller

, Risacher

, Rodriguez

, Sachdev

, Saykin

, Slavin

, Snitz

, Sperling

, Tandetnik

, van der Flier

, Wagner

, Wolfsgruber

, Sikkes

SAM

(2015) Subjective cognitive decline in older adults: An overview of self-report measures used across 19 international research studies, J Alzheimers Dis 48, S63–S86.

21.

(2019) Behavioral Risk Factor Surveillance System Overview: 2019. https://www.cdc.gov/brfss/annual_data/2019/pdf/overview-2019-508.pdf.

22.

(2019) BRFSS Combined Landline and Cell Phone Weighted Response Rates by State, 2019. https://www.cdc.gov/brfss/annual_data/2019/pdf/2019-response-rates-table-508.pdf.

23.

Centers for Disease Control and Prevention (CDC) (2019) Behavioral Risk Factor Surveillance System Survey Questionnaire.

24.

LLCP 2019: Codebook Report. https://www-cdc-gov.proxy.lib.ohio-state.edu/brfss/annual_data/2019/pdf/codebook19_llcp-v2-508.HTML.

25.

(2020) Behavioral Risk Factor Surveillance System (BRFSS) Complex Sampling Weights and Preparing 2019 BRFSS Module Data for Analysis. https://www.cdc.gov/brfss/annual_data/2019/pdf/Complex-Smple-Weights-Prep-Module-Data-Analysis-2019-508.pdf.

26.

Yaffe

, Fiocco

, Lindquist

, Vittinghoff

, Simonsick

, Newman

, Satterfield

, Rosano

, Rubin

, Ayonayon

, Harris

(2009) Predictors of maintaining cognitive function in older adults: The Health ABC Study. Neurology 72, 2029–2035.

27.

Kalmijn

, van Boxtel

MPJ

, Verschuren

MWM

, Jolles

, Launer

(2002) Cigarette smoking and alcohol consumption in relation to cognitive performance in middle age. Am J Epidemiol 156, 936–944.

28.

Launer

, Feskens

, Kalmijn

, Kromhout

(1996) Smoking, drinking, and thinking. The Zutphen Elderly Study. Am J Epidemiol 143, 219–227.

29.

Weuve

, Tchetgen Tchetgen

, Glymour

, Beck

, Aggarwal

, Wilson

, Evans

, Mendes de Leon

(2012) Accounting for bias due to selective attrition: The example of smoking and cognitive decline. Epidemiology 23, 119–128.

30.

CDC - BRFSS - BRFSS Frequently Asked Questions (FAQs), https://www.cdc.gov/brfss/about/brfss_faq.htm.

31.

Tandetnik

, Farrell

, Cary

, Cines

, Emrani

, Karlawish

, Cosentino

(2015) Ascertaining subjective cognitive decline: A comparison of approaches and evidence for using an age-anchored reference group. J Alzheimers Dis 48, S43–S55.

32.

Rabin

, Wang

, Mogle

, Lipton

, Derby

, Katz

(2020) An approach to classifying subjective cognitive decline in community-dwelling elders, Alzheimers Dement (Amst) 12, e12103.

33.

Pierannunzi

, Hu

, Balluz

(2013) A systematic review of publications assessing reliability and validity of the Behavioral Risk Factor Surveillance System (BRFSS), 2004–2011. BMC Med Res Methodol 13, 49.