Socio-demographic and lifestyle determinants of multimorbidity in the English Longitudinal Study of Ageing (ELSA)

The problem

Long-term conditions are the main challenge facing healthcare systems worldwide. There is a substantial lack of knowledge as to the longitudinal determinants of multi-morbidity, particularly with regard to the impact of health behaviours.

The approach

We investigated the socio-demographic and lifestyle characteristics that predicted having multi-morbidity eight years later in a cohort (n=5,114) of people aged 50 years or older participating continuously in the English Longitudinal Study of Ageing (ELSA) from 2002/3 (baseline) to 2010/11. Participants reported in 2002/3 and 2010/11 whether or not a doctor had diagnosed them with any of 19 conditions; including cardiovascular, geriatric and psychological conditions. We constructed two logistic regression models to predict having two or more conditions in 2010/11; the first involving participants reporting no conditions at baseline (n=1,438) and the second involving participants reporting any one condition at baseline (n=1,590). A third logistic regression was constructed using the full sample to predict having an increase in the number of reported conditions between baseline and 2010/11, whilst controlling for the number of conditions reported at baseline. For each model, the baseline covariates entered as predictors were age band (50-64, 65-74, 75+ years old), gender, highest educational attainment (no qualification, intermediate, degree/higher education) as a proxy for socio-economic status, whether or not the participant was currently married/in a civil partnership, their level of physical activity (sedentary, low, moderate, vigorous) and smoking behaviour (never smoked, smoked in the past, current smoker). All models corrected for differential non-response using sample weighting.

Findings

408/1438 (28.4%) of participants reporting no conditions in 2002/3 had multi-morbidity in 2010/11. Increasing age band at baseline raised the odds of having multimorbidity by 90% (odds ratio=1.90, p<0.001), whereas higher education attainment lowered the odds by 21% (OR=0.79, p=0.008). There was also a marginal trend for higher levels of physical activity to decrease the odds of developing multi-morbidity (OR=0.85, p=0.064). Of participants reporting any one condition in 2002/3, 972 (61.1%) had multi-morbidity by 2010/11. A similar pattern of predictors was evident, with physical activity (OR=0.74, p<0.001) and having smoked previously (OR=1.38, p=0.008) additionally being significant. 3,006/5,114 (58.8%) of participants had an increase in reported conditions over the eight years. Again, an increase was predicted by older age (OR=1.25, p<0.001), lower educational attainment (OR=0.83, p<0.001), lower physical activity (OR=0.77, p<0.001), having previously smoked (OR=1.23, p=0.003) or being a current smoker (OR=1.23, p=0.027) at baseline. Gender was non-significant in all models.

Consequences

A significant proportion of people aged 50 and older will develop multimorbidity over an eight years period, with more than 1 in 4 of those not having a condition developing multimorbidity. Physical activity and tobacco use should be at the centre of lifestyle modifications for reducing the burden of multimorbidity.

Credits

• Luke Mounce
• Jose M. Valderas