Socio-demographic and lifestyle determinants of multimorbidity ten years later in older people: a prospective cohort study

Problem

Multimorbidity affects a significant percentage of patients seen in Primary Care. Increasing age and deprivation have consistently been shown to be positively associated to multimorbidity. However, most investigations have been cross-sectional, and evidence regarding lifestyle determinants is scarce.

Approach

We investigated the socio-demographic and lifestyle characteristics that predicted having multi-morbidity ten years later in a nationally representative cohort (n=4,564) of people aged 50 years or older, who participated continuously in the English Longitudinal Study of Ageing (ELSA) from 2002/3 (baseline) to 2012/13. Participants self-reported whether a doctor had diagnosed them with any of 17 common conditions. Presence of two or more conditions was classified as multimorbidity. We constructed two logistic regression models to predict having multimorbidity in 2012/13 using baseline covariates; the first involving participants reporting no conditions at baseline (n=1,370) and the second involving participants reporting any one condition at baseline (n=1,474). A third logistic regression was constructed using the full sample to predict having an increase in the number of reported conditions between baseline and 2012/13, whilst controlling for the number of conditions reported at baseline. Covariates were age, gender, wealth, educational attainment, whether they lived alone, BMI category, smoking behaviour, physical activity, alcohol consumption, diet, social detachment, locus of control, and the presence of individual health conditions.

Findings

Two thirds of participants (3,047/4,564; 66.8%) had multimorbidity in 2012/13; 429/1,370 (31.3%) of participants with no baseline conditions, and 978/1,474 (66.4%) of participants with a single initial condition developed multimorbidity over the ten year study period. Across all models, odds of having increased morbidity ten years later were positively associated with age, negatively associated with wealth, and greater for females and participants who were obese. Additionally, participants with no baseline conditions were at increased odds of multimorbidity if they had any smoking history (p=0.038). For participants with a single initial condition, ischaemic heart disease (odds ratio=4.02, p=0.001), diabetes (OR=3.51, p=0.010) and osteoporosis (OR=3.31, p=0.017) were most predictive of future multimorbidity, whereas lifestyle factors were not associated. Using the full sample, any increase in number of conditions over the ten year period was predicted by increasing BMI category (trend: p<0.001), lower physical activity (trend: p=0.008), being a current smoker (OR=1.39, p=0.003), and believing that life events are outside of one’s control (OR=1.20, p=0.017), in addition to associations with age (trend: p<0.001), female gender (OR=1.29, p=0.003) and wealth (trend: p<0.001).

Consequences

Age, gender, wealth and BMI are important socio-demographic determinants of subsequent multimorbidity, yet increased morbidity is also predicted by key lifestyle factors of smoking behaviour and physical activity. Strategies for preventing greater morbidity should be informed by these findings, and may further benefit from empowering patients to believe they can alter their health trajectories.