(C) PLOS One
This story was originally published by PLOS One and is unaltered.
. . . . . . . . . .



Longitudinal clustering of health behaviours and their association with multimorbidity in older adults in England: A latent class analysis [1]

['Alisha Suhag', 'Healthy Lifespan Institute', 'School Of Health', 'Related Research', 'University Of Sheffield', 'Sheffield', 'United Kingdom', 'Thomas L. Webb', 'Department Of Psychology', 'John Holmes']

Date: 2024-02

Abstract Background Health-risk behaviours such as smoking, unhealthy nutrition, alcohol consumption, and physical inactivity (termed SNAP behaviours) are leading risk factors for multimorbidity and tend to cluster (i.e. occur in specific combinations within distinct subpopulations). However, little is known about how these clusters change with age in older adults, and whether and how cluster membership is associated with multimorbidity. Methods Repeated measures latent class analysis using data from Waves 4–8 of the English Longitudinal Study of Ageing (ELSA; n = 4759) identified clusters of respondents with common patterns of SNAP behaviours over time. Disease status (from Wave 9) was used to assess disorders of eight body systems, multimorbidity, and complex multimorbidity. Multinomial and binomial logistic regressions were used to examine how clusters were associated with socio-demographic characteristics and disease status. Findings Seven clusters were identified: Low-risk (13.4%), Low-risk yet inactive (16.8%), Low-risk yet heavy drinkers (11.4%), Abstainer yet inactive (20.0%), Poor diet and inactive (12.9%), Inactive, heavy drinkers (14.5%), and High-risk smokers (10.9%). There was little evidence that these clusters changed with age. People in the clusters characterised by physical inactivity (in combination with other risky behaviours) had lower levels of education and wealth. People in the heavy drinking clusters were predominantly male. Compared to other clusters, people in the Low-risk and Low-risk yet heavy drinkers had a lower prevalence of all health conditions studied. In contrast, the Abstainer but inactive cluster comprised mostly women and had the highest prevalence of multimorbidity, complex multimorbidity, and endocrine disorders. High-risk smokers were most likely to have respiratory disorders. Conclusions Health-risk behaviours tend to be stable as people age and so ought to be addressed early. We identified seven clusters of older adults with distinct patterns of behaviour, socio-demographic characteristics and multimorbidity prevalence. Intervention developers could use this information to identify high-risk subpopulations and tailor interventions to their behaviour patterns and socio-demographic profiles.

Citation: Suhag A, Webb TL, Holmes J (2024) Longitudinal clustering of health behaviours and their association with multimorbidity in older adults in England: A latent class analysis. PLoS ONE 19(1): e0297422. https://doi.org/10.1371/journal.pone.0297422 Editor: Sandar Tin Tin, University of Auckland, NEW ZEALAND Received: August 20, 2023; Accepted: January 4, 2024; Published: January 25, 2024 Copyright: © 2024 Suhag et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability: The raw data involved in this analysis are available through the UK Data Service (https://www.elsa-project.ac.uk/accessing-elsa-data). Funding: For this analysis, AS was supported by a studentship awarded to the Healthy Lifespan Institute and funded by the University of Sheffield. The funders had no say in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. Competing interests: The authors have declared that no competing interests exist.

Introduction A growing number of older adults are living with multimorbidity, defined as having two or more chronic diseases [1]. In England, for example, 67.8% of people aged 65 and older are expected to have multimorbidity by 2035 [1]. Multimorbidity is more common in recent cohorts and seems to be emerging earlier in the lifecourse, which places a significant strain on healthcare systems [2, 3]. However, chronic diseases (e.g., type-2 diabetes, coronary heart disease, chronic obstructive pulmonary disease) and some cancers that form a large proportion of the multimorbidity burden, have well-established modifiable risk factors, suggesting that there are opportunities for prevention [4]. Among such modifiable risk factors–health risk behaviours such as smoking, poor nutrition, alcohol consumption, and physical inactivity (collectively termed ‘SNAP’ behaviours) account for nearly one-third of disability-adjusted life years from chronic conditions [5]. Most preventative interventions for older adults focus on single behaviours [6]. However, research shows behavioural risk factors typically cluster in specific combinations within distinct populations. This suggests that interventions targeting these clusters may be more appropriate [7]. However, designing such interventions is challenging because individuals’ health behaviours not only cluster but may also change over time [8]. Additionally, engaging in multiple health-risk behaviours can have a negative impact on health that is greater than the sum of their individual effects [9]. While epidemiological studies have attempted to understand the combined impact of multiple behaviours on health, they often rely on simple indices that count the number of co-occurring behaviours without considering which behaviours, or combinations of behaviours, are driving the risk [10, 11]. Other studies have tackled this issue by examining the health risks associated with combinations of behavioural factors in dyads, triads, and tetrads [12]. However, they tend to overlook less common combinations involving multiple risk behaviours because of sparse data [12]. Clustering techniques, such as repeated measures latent class analysis (RMLCA), can better address these issues by grouping individuals with similar patterns of health-risk behaviours over time into clusters [10]. Identifying clusters of health-risk behaviours–and examining whether and how behaviours within these clusters change as people age–can inform interventions by: a) identifying high-risk populations (e.g. subgroups exhibiting combinations of behaviours that entail the greatest risk), b) informing the selection of target behaviours (e.g. those with the greatest health impact or highest reach), and c) detecting potential spillover effects (i.e. where targeting one health behaviour leads to compensatory changes in other behaviours) [13]. For instance, when some individuals reduce cigarette smoking, the reward value and consumption of ’treat foods’ increases, resulting in weight gain [13]. Previous studies that have used clustering techniques to analyse multiple behaviours in older adults have typically been cross-sectional, and so cannot test whether behaviour clusters change over time and if these changes affect long-term health [14–16]. While two studies have used longitudinal clustering to study a subset of SNAP behaviours in older adults [17, 18], none have examined the relationship between these clusters and multimorbidity. In addition, the aforementioned studies limit themselves to a basic definition of multimorbidity (i.e. a simple count of the number of diseases), which overlooks differences between diseases within one system and those spanning multiple systems [19]. This is crucial, as multimorbidity may have a larger impact on overall health if it arises out of chronic conditions in different body systems that are likely to compete for treatment, rather than closely related comorbidities that might have shared pathophysiology or shared approaches to management [20]. The construct of complex multimorbidity, defined as “the co-occurrence of three or more chronic conditions affecting three or more different body systems within one person without an index chronic condition”, addresses these issues by focusing on chronic conditions affecting multiple body systems [21, p1]. This definition also has the advantage of identifying the number and types of specialised health services involved in a patient’s care, thus identifying individuals with more complex needs [21]. Complex multimorbidity might also better reflect the biology of ageing as it involves a simultaneous breakdown or dysfunction of multiple, separate body systems, making it a more reflective measure to study in older adults [22]. Equally crucial is to recognise that multimorbidity is associated with social and economic determinants [23] that can add to (or interact with) behavioural determinants. For example, advanced age, female gender, low socioeconomic status, and education have all been identified as significant risk factors for the onset of multimorbidity [24]. These findings align with the Social Determinants of Health (SDoH) framework, which describes how broader societal structures–from economic policies to social norms–shape and segment populations hierarchically based on gender, race, education, occupation, and income [25, 26]. This stratification then directly and indirectly influences health outcomes. The present research therefore incorporates insights from the SDoH framework to examine whether socio-demographic factors predict membership within risk behaviour clusters. Given that socio-demographic determinants can not only shape individual health behaviours but also predict health outcomes through complicated multifactorial pathways, we will also adjust for them in examining the relationship between health behaviours and outcomes [24]. The present research The present research analyzes data from a longitudinal panel of older adults in England to: i) explore how the SNAP behaviours cluster over time in older adults, ii) investigate how membership in different behavioural clusters varies by socio-demographic characteristics, and iii) examine which, if any, behavioural clusters are prospectively associated with multimorbidity over time.

Discussion The present research investigated the relationship between clusters of health-risk behaviours over time and multimorbidity in older adults. We identified seven distinct clusters of behaviour that resemble those found in previous studies from Germany [14], Australia [15], and Taiwan [17] including a cluster characterised by an overall low level of risk, a cluster characterised by physical inactivity, and a cluster characterised by heavy alcohol consumption, non-smoking and low physical activity. Similarly, with the exception of a study focusing on Taiwanese men [17], where the smokers were split across two clusters because of a relatively high prevalence of smoking, the smallest subgroup in each study comprised smokers who exhibited two or more risky behaviours, which parallels our finding that the High risk smokers represented the smallest cluster (~11% of the sample). However, our clusters diverged from the findings of a study focusing on six international ageing cohorts, likely because their study: excluded dietary data, included social activity as a behaviour, and used different measures for physical activity, alcohol consumption, and smoking [16]. The present research moved beyond existing research, however, by using longitudinal data to not only examine whether distinct clusters of health behaviours are found in older adults but also whether and how patterns of behaviour within each cluster change over time. We found that patterns of behaviour within the clusters were largely stable over time, with two exceptions: The proportion of current smokers steadily declined in the High-risk smokers cluster, while the proportion of alcohol abstainers gradually increased in clusters characterised by moderate or no alcohol consumption (i.e. the clusters labelled Low-risk and Abstainer yet inactive). Notwithstanding these exceptions, our findings support the idea that SNAP behaviours in older people are fairly stable and likely reflect lifelong habits [8], emphasising the importance of addressing risk behaviours early in the life course to prevent negative health outcomes [44]. Additionally, the finding that behavioural patterns are relatively stable over time suggests that clustering in older adults can be accurately captured by cross-sectional studies. The clusters also had different socio-demographic profiles. Consistent with alcohol consumption patterns in the UK [32], the two clusters of heavy drinkers were predominantly male. The clusters characterised by physical inactivity but no other risky behaviours (i.e. the Low-risk yet inactive and Abstainer yet inactive clusters) were primarily female, similar to findings in previous studies [14–17]. High-risk smokers were younger on average and, in contrast to previous research, we did not find evidence that high-risk smokers more likely to be men [45]. This may be due to survivorship bias, as smoking is the leading cause of lung cancer deaths, but lung cancer occurs less frequently and has a better prognosis in women [45]. We also found a marked consistency with previous studies looking at clusters of health behaviour among older adults [14, 15, 17], in that we found that clusters characterised by physical inactivity (in combination with other risky behaviours) were less likely to be wealthy or well-educated, suggesting a link between socio-demographic inequalities and health behaviour clustering. Importantly, identified clusters also differed in their disease status. Participants characterised as Abstainers but inactive and Low-risk yet inactive had a higher prevalence of complex multimorbidity and endocrine disorders than other low-risk clusters that engaged in health-promoting behaviours (i.e. Low-risk and Low-risk yet heavy drinkers), and they also had higher rates of multimorbidity compared to the Low-risk cluster. Notably, participants in the cluster characterised as Abstainers but inactive had a higher prevalence of endocrine disorders than participants in all clusters except Low-risk yet inactive. That the cluster characterised by physical inactivity (but no other risk behaviours) was associated with worse health outcomes than clusters characterised by multiple risk behaviours suggests that the relationship between behaviour and health outcomes is more complex than a linear dose–response relationship [46]. Indeed, it is important to recognise the possibility of a bidirectional relationship between physical activity and multimorbidity, since not only is physical activity a risk factor for multimorbidity, but multimorbidity, in turn, can reduce function and reduce adherence to recommended levels of physical activity [47]. Some associations were more straightforward and predictable. For example, we found that High-risk smokers had higher rates of respiratory disorders than every other cluster as might be expected. However, High-risk smokers also had a higher prevalence of complex multimorbidity and endocrine, nutritional and metabolic disorders compared to the Low-risk yet heavy drinkers cluster, a finding that is harder to explain using health behaviour patterns alone. Similarly hard to explain is the finding that Inactive, heavy drinkers had a higher prevalence of complex multimorbidity than Low-risk yet heavy drinkers. One explanation for the lower prevalence of complex multimorbidity and endocrine, nutritional and metabolic disorders could be that, compared to other clusters, the Low-risk yet heavy drinkers cluster had the largest proportion of individuals in the highest wealth tertile and in intermediate and professional jobs–indicators of elevated socioeconomic status. This higher socioeconomic status, a known protective factor, may influence health outcomes, as it has consistently been identified as an important determinant of multimorbidity [23]. Thus, examining the interaction between health behaviour clusters and socio-demographic variables on multimorbidity, could further help clarify the patterns of risk. Additionally, our focus on the adverse health effects of risky behaviours might have overshadowed the protective effects of engaging in some behaviours (i.e., adequate fruit and vegetable intake and being physically active) [46]. Recognizing the potential benefits of these behaviours and their associated factors is crucial, as they offer functional, social, and psychological resilience against multimorbidity [48]. Strengths and limitations Several strengths distinguish this study. It is the first to examine the association between longitudinal clusters of multiple health-risk behaviours and multimorbidity in older adults. Health behaviour experts helped to choose the most viable of the measures available in ELSA and how these might be used. It uses a robust, model-based, probabilistic approach (namely, RMLCA), demonstrates stable results in split-half replication, is reproducible (i.e. diagnostic criteria and programming codes are accessible) [49]. Furthermore, the results adjust for baseline disease and a range of socio-demographic variables that may confound the relationship between health behaviour and outcomes. Despite these strengths, the study has some limitations. ELSA relies on self-report data, which can be subject to recall limitations and social desirability bias. Having said this, longitudinal analyses are less susceptible to misclassification bias due to consistent measures across survey waves. It is also important to note that alcohol consumption was only measured for the past week, which may misestimate drinking behaviour for those with inconsistent drinking patterns. Relatedly, missing data are unavoidable in general population cohorts such as ELSA and we had to exclude participants with missing sociodemographic data at baseline. As a result, participants who were included were slightly older, better educated, and more likely to have more intermediate and professional level jobs than those who were excluded. This may limit the generalisability of our findings. Finally, as there are relatively few ethnic minority participants in ELSA, the findings may not generalise to non-white populations. Implications The present research offers new insights into the relationship between clusters of health behaviours and multimorbidity in older adults and has practical implications for interventions to improve health outcomes. For instance, by identifying distinct profiles of risk behaviour, our findings can help to identify high-risk subgroups and select behaviour(s) to target with interventions. For example, our data suggest that targeting physical inactivity, which characterised all five clusters associated with negative health outcomes and represented the majority of the population (70%), could have the greatest potential reach and health impact. The present findings also demonstrate how targeting different clusters may require tailored approaches. For instance, interventions targeting participants in the Abstainer yet inactive cluster, which comprised mostly women and had lower levels of education and wealth, may need to address barriers to physical activity that are specific to their socio-demographic profile. This aligns with existing evidence indicating that interventions tailored to specific target audiences are more effective in promoting changes in multiple behaviours in the general adult population [50, 51] and in patients with chronic conditions [52], than interventions that are not tailored.

Conclusions The present research identified seven clusters of older adults with distinct patterns of behaviour that were associated with socio-demographic characteristics and the prevalence of multimorbidity. Notably, we found that the number or combination of risk behaviours alone could not explain why some clusters had worse health outcomes than others. A closer examination of how behaviour clusters interact with socio-demographic characteristics could offer a more nuanced understanding of their combined effect on health outcomes. Integrating this additional layer of complexity into our current study would have made its breadth unmanageable, but it remains an important area for future investigations to explore. Additionally, our findings show that health-risk behaviours tend to be stable as people age, emphasising the importance of addressing them early. Future research should take a lifespan approach to investigate how risk behaviours cluster at earlier life stages.

Acknowledgments We would like to thank Nicola Buckland, Robert Copeland, and Matt Field for their contributions to helping identify which measures from ELSA to focus on.

[END]
---
[1] Url: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0297422

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.