Does current smoking predict future frailty? The English longitudinal study of ageing Free

Abstract

Background

smoking is the single most preventable cause of morbidity and mortality. The evidence on independent associations between smoking in later life and incident frailty is scarce.

Objectives

to examine the effect of current smoking in older people on the risk of developing frailty, controlling for important confounders.

Methods

we used data of 2,542 community-dwelling older people aged ≥60 years in England. Participants were classified as current smokers or non-smokers. Frailty was defined using modified Fried criteria. Multivariable logistic regression models were used to examine risk of 4-year incident frailty in current smokers compared with non-smokers, adjusted for demographic, socioeconomic and health variables.

Results

of 2,542 participants, 261 and 2,281 were current smokers and non-smokers, respectively. The current smokers were significantly frailer, younger, with lower BMI, less educated, less wealthy and lonelier compared with non-smokers at baseline. In multivariable logistic regression models adjusting for age and gender, current smokers were twice as likely to develop frailty compared with non-smokers (odds ratio (OR) = 2.07, 95% confidence interval (CI) = 1.39–3.39, P = 0.001). The association is attenuated largely by controlling for socioeconomic status. Smoking remains significantly associated with incident frailty in fully adjusted models including age, gender, socioeconomic status, alcohol use, cognitive function and loneliness (OR = 1.60, 95% CI = 1.02–2.51, P = 0.04). The relationship is however attenuated when taking account of non-response bias through multiple imputation.

Conclusions

current smokers compared with non-smokers were significantly more likely to develop frailty over 4 years among community-dwelling older people. Given that smoking is a modifiable lifestyle factor, smoking cessation may potentially prevent or delay developing frailty, even in old age.

Introduction

When national surveys on smoking started in the UK in 1974, 41% of women and 51% of men were smokers [1]. The overall prevalence of smoking has been declining since then, down to 17% for women and 20% for men in 2014 [1]. Tobacco smoking is the single most preventable cause of morbidity and mortality in the UK [2]. National Health Service (NHS) spent £5.2 billion (approximately $7.5 billion) in treating smoking-related health conditions in 2005/06 [3].

Smoking also increases the risk of developing a number of other diseases, such as chronic obstructive pulmonary disease (COPD), coronary heart disease, stroke and peripheral vascular disease [4], all of which can potentially have negative effects on the physical, psychological and social health of smokers. Disability itself limits autonomy, increases the risk of dependence, reduces quality of life and contributes to mortality [5].

Frailty is considered a precursor to, but a distinct state from, disability [6]. Frailty has been described as a condition associated with decreased physiological reserve and increased vulnerability to adverse health outcomes with exposure to a stressor [7]. The outcomes include falls [8], fractures [9], disability [10], hospitalisation [11] and institutionalisation [12]. Frailty has also been shown to be linked to worse psychological or cognitive outcomes, such as poor quality of life [13] and dementia [14]. Due to the potential for reversibility of frailty [15], identifying potentially modifiable risk factors of frailty may help to develop strategies to prevent or slow progression of adverse health outcomes associated with both frailty and smoking. As maintaining independence is a key priority for older people, demonstrating links with smoking and frailty might provide additional motivation for older smokers to quit. A previous systematic review showed that only a few studies have examined longitudinal associations between smoking and risk of incident frailty [16]. Although most of these studies demonstrated that smokers were more likely to develop frailty, they provided effect measures that were unadjusted or adjusted for a limited number of confounders [16]. Therefore, the independent association of smoking with incident frailty has not been convincingly established. We thus aimed to examine the association of smoking with the risk of developing frailty, controlling for important confounding variables and using data from a nationally representative sample of community-dwelling older men and women living in England.

Methods

Study setting and population

The English Longitudinal Study of Ageing (ELSA) is a multi-centre longitudinal panel study of a nationally representative sample of community-dwelling men and women aged 50 years and older in England and its detail has been published elsewhere [17]. The initial participants (n = 11,391) at wave 1 in 2002 were recruited from households that participated in the Health Survey for England (HSE). The panel has been followed up with every 2 years. Ethical approval for all of the ELSA waves was obtained from the National Research and Ethics Committee and informed consent was obtained from all participants.

The current study used data of participants who were aged 60 years or older at wave 2 (baseline), since the gait speed was not measured for those aged less than 60 years, and who also participated at wave 4 (follow-up). Of 6,183 men and women aged 60 years or older who were interviewed at wave 2, those who missed any data regarding smoking status at wave 2 (n = 3) and frailty components at wave 2 (n = 1,688) were excluded. Those who were frail at wave 2 (n = 575) were also excluded in order to examine the risk of incident frailty. Among 3,918 participants left, 1,376 could not participate at the follow-up wave due to ill health (n = 44), death (n = 139), refusal (n = 547), being unable to contact (n = 132) or other reasons (n = 514). The final analytic sample for this study was 2,542 participants.

Predictor variable—smoking

Participants were classified as ‘current smoker’ or ‘non-smoker’ based on answers to the question ‘Do you smoke cigarettes at all nowadays?’ during the interview at wave 2. To examine effects of smoking cessation on frailty, the non-smokers were divided, based on data of when they quit smoking available from wave 3 (2 years after wave 2), into two groups: past smokers and never smoker. The past smokers were further divided into another two groups: those who quit within the last 10 years and those who quit more than 10 years ago [18].

Outcome variable—incident frailty

Frailty was defined using the frailty phenotype criteria that Fried et al. [6] described in the Cardiovascular Health Study (CHS). In CHS, frailty is defined using a combination of five physical frailty components: (i) unintentional weight loss, (ii) self-reported exhaustion, (iii) weakness, (iv) slow walking speed and (v) low physical activity. Frailty is classified as having three or more of the five criteria. An individual who meets one or two criteria is classified as prefrail, and an individual with no criterion is classified as robust. Please see Supplementary data, Appendix 1, are available at Age and Ageing online. for detail of definitions of the CHS criteria components, covariates and statistical analysis.

Results

Table 1 and Supplementary data, Appendix 2, are available at Age and Ageing online. present the baseline characteristics of the final analytic sample of 2,542 participants according to smoking status as well as 1,376 who were excluded at follow-up according to reasons for lost follow-up. Among the analytic sample at baseline, 2,281 participants were non-smokers (1,168 never smokers and 1,113 past smokers) and 261 were current smokers. Current smokers were significantly frailer, younger, with lower BMI, less educated, less wealthy and lonelier compared with non-smokers. There were no significant differences in gender, alcohol use and cognitive function between these two groups.

In the univariate logistic regression models, various factors were significantly associated with a higher risk of incident frailty over 4 years. Current smoking was associated with an approximately 50% increased risk of developing frailty (odds ratio (OR) = 1.56, 95% confidence interval (CI) = 1.06–2.29, P = 0.02). Other factors associated with an increased risk of incident frailty were belonging to the older age group, being a female, having a higher BMI, consuming alcohol less frequently, having completed a lower level of education, having a lower level of wealth, having a lower cognitive function and having more loneliness (Table 2).

Table 3 shows the results of the multivariable logistic regression models. In Model 1 adjusting for age and gender, current smokers were twice as likely to develop frailty at the time of follow-up compared with non-smokers (OR = 2.07, 95% CI = 1.34–3.19, P = 0.001). Further adjusting for alcohol use did not change the odds ratio drastically (OR = 2.17, 95% CI = 1.39–3.39, P = 0.001). Although adding education and wealth for adjustment in Model 3 decreased the odds ratio, current smoking remained a significant predictor of incident frailty (OR = 1.62, 95% CI = 1.05–2.52, P = 0.03). In Model 4, cognitive function and loneliness were further adjusted for, which made little change in the association (OR = 1.60, 95% CI = 1.02–2.51, P = 0.04). We repeated the final model (Model 4) with multiple imputation by chained equations, and this attenuated the association (OR = 1.48, 95% CI = 0.97–2.28, P = 0.07).

When COPD was added to the Model 4, current smoking was no longer a significant predictor of incident frailty and the OR decreased by 14.4% (OR = 1.37, P = 0.19). In this model, COPD was strongly associated with incident frailty (OR = 2.58, 95% CI = 1.59–4.20, P < 0.001). These findings suggest that current smokers are more likely to develop frailty due to COPD, rather than smoking itself. Adding CVD or cancers to Model 4 made little changes in the results, which suggest that CVD and cancers are not a modulator in the associations between current smoking and development of frailty.

In supplementary analyses, incident frailty risk for current and past smokers compared with never smokers was calculated. Compared with never smokers, current smokers were significantly more likely to develop frailty in Models 1 and 2, which became non-significant in Models 3 and 4. There was no significant association between past smoking and incident frailty in any models. (Supplementary data, Appendix 3, are available at Age and Ageing online.) Among 1,113 past smokers, 157 quit smoking within the last 10 years and 956 quit smoking for more than 10 years ago. Incident frailty risks of these two groups were not significantly different to that of non-smokers in all models (Supplementary data, Appendix 4, are available at Age and Ageing online.).

DISCUSSION

This prospective panel study of 2,542 British community-dwelling men and women aged 60 years or older who were free of frailty at baseline showed that current older smokers were 60% more likely to develop frailty than non-smokers over 4 years, controlling for a wide range of potential confounders including age, gender, alcohol use, education, wealth, cognitive function and loneliness.

Our findings are consistent with the limited previous longitudinal research, which has shown in the majority of studies that smoking worsened subsequent frailty status [19–23], except for one study [15] .

Mechanisms by which current smokers are more likely to develop frailty are unknown, but may be multifactorial given that tobacco smoke is a mixture of numerous kinds of toxic chemicals and compounds and can affect every organ in the body. Smoking has been shown to be associated with various physical and mental illnesses [4], any of which can contribute to the development of frailty. These health risks can be reduced substantially by smoking cessation, according mostly to findings from studies among middle aged adults [24]. Although scarce, the evidence supports that one is never too old to quit smoking and older smokers can still benefit from quitting [25]. One study showed that the risks of myocardial infarction and stroke were reduced by 40% within 5 years of smoking cessation in German older people aged 50 and over [26]. Smoking cessation can potentially be an effective strategy to prevent or delay developing frailty among older smokers. This possible benefit of smoking cessation is supported by our findings that past smokers did not have higher risk of incident frailty than never smokers. Evidence suggests that older people may be less motivated by preventing disease such as heart attacks than younger people [27]. However it is their priority to remain independent, able to look after themselves and engaged socially [28]. Therefore, knowledge that continued smoking in later life may increase the risk of frailty, which itself is strongly associated with increased dependency and increased risk of moving into care home settings, may provide additional motivation to encourage older smokers to quit.

In the multivariable logistic regression models, the odds ratio of developing frailty in current smokers compared with non-smokers decreased from 2.17 to 1.62 (−25.3%) when further adjusted for education and wealth, which suggests that the association between smoking and incident frailty can partially be explained by socioeconomic status. Lower socioeconomic status has been shown to be associated with a higher prevalence of smoking [3] and a higher level and faster progression of frailty [29]. Socioeconomically disadvantaged smokers typically are found to have developed their smoking habit earlier in their lives, and are likely to be more nicotine dependent, to have less social support for smoking cessation and to be less likely to succeed in smoking cessation attempts [30]. In order to reduce the smoking-related health inequalities, smoking cessation measures should be effective on these hard-core smokers with low socioeconomic status [30]. In our supplementary analysis using multiple imputation of covariates the relationship of smoking with frailty was attenuated further and becomes non-significant.

COPD, CVD and cancers were separately added to the final multivariable logistic regression model to see if these smoking-related diseases fully explained the association between current smoking and incident frailty, or if there appeared to be a further independent effect of smoking on frailty. Only COPD changed the results significantly and current smoking no longer predicted incident frailty in that model, which suggests that the association appears to be explained by COPD. Finally our supplementary analysis suggests that the harmful effects of smoking on frailty are largely restricted to those who were smoking at baseline. Those who had more recently quit (within the last 10 years) showed no increased risk of frailty compared to never-smokers.

There are some limitations and strengths of this study. First, due to the limited availability of data at the baseline wave, only current smokers and non-smokers were defined. We had to retrieve data from a later wave to create past and never smoker groups. We had no information on the extent of smoking exposure (quantity of cigarettes consumed or length of exposure) and we were therefore unable to explore a ‘dose–response’ relationship. It should be noted that the information on smoking status was self-reported and potentially subject to response bias. Second, our sample was restricted to those who had completed measurements of frailty status (e.g. gait speed, handgrip strength) in nurse interviews at two time points in ELSA. Those who were excluded due to missing data at follow-up were significantly frailer and more likely to be current smokers compared with those who were included, which suggests that those excluded were missing data that were not random. Therefore, this exclusion is likely to attenuate an association between smoking status and incident frailty. Whilst we attempted to account for attrition bias by using non-response weights, this differential loss to follow-up may have underestimated the associations between frailty and smoking. Third, the ELSA cohort only includes the English population and may not be generalisable to other populations. Fourth, as in other studies, components of CHS criteria were slightly modified according to availability of ELSA data, which may have affected the findings [31]. Fifth, we used only two time points 4 years apart to assess incident frailty risk according to smoking. Given that COPD may be an important mediator in the association between smoking and frailty, over many years, a study with a longer follow-up period and multiple data collection time points would be justified.

The major strengths of this analysis are a large sample size, a prospective study design and the use of a wide range of potential confounders for adjustment.

In conclusion, current smokers compared with non-smokers were significantly more likely to develop frailty over 4 years among British community-dwelling older people. This result is in line with findings of a recent systematic review [16]. Given that smoking is a modifiable lifestyle factor, smoking cessation may potentially prevent or delay developing frailty, even in old age.

Supplementary data

Supplementary data mentioned in the text are available to subscribers in Age and Ageing online.

Conflicts of interest

None.

Funding

G.K. is funded by a University College London (UCL) Overseas Research Scholarship, which did not have any influence on the study design, the collection, analysis, and interpretation of data, the writing of the article or the decision to submit it for publication.

ELSA has been funded by the National Institute of Aging in the United States and a consortium of UK government departments coordinated by the Office for the National Statistics, and the data are available through the UK Data Archive (http://data-archive.ac.uk).

References