Background Social support has been linked to positive health outcomes and one suggested mechanism for this effect is via health behaviours. Alcohol consumption is associated both with negative health and social outcomes and with positive social benefits, such as forming and maintaining friendships, relaxing and having fun.

A better understanding of which aspects of social support are associated with harmful alcohol consumption could help to improve public health messages and treatments for alcohol consumption; however, the existing evidence in this field is often contradictory and looks at the relationship at one time-point.

This study aimed to explore 1) the relationship between the quantity and quality of social support and harmful alcohol consumption, and 2) whether a change in the quantity and/or quality of social support over time is associated with a change in alcohol consumption.

Methods Data were from 6401 men/women aged 35–55 at Phase 1 (1985) and 3704 men/women aged 47–67 at Phase 5 (1997) from a prospective cohort study of UK civil servants (Whitehall II). Measures of quantity and quality of social support from the Close Persons Questionnaire and last week alcohol consumption regrouped into non-drinkers; lower-risk (≤14 units/week); increasing-risk (>14 to ≤35 units/week-women, >14 to ≤50 units/week-men) and higher-risk drinkers (>35 units/week-women, >50 units/week-men) were used to conduct multinomial logistic regression analyses.

Results 1) Amongst men, high levels of worsening support (RRR=1.95, 95%CI=1.13–3.37), seeing friends more often (RRR=2.20, 95%CI=1.10–4.42), seeing more friends/month and seeing relatives less frequently (RRR=1.89, 95%CI=1.12–3.19) increased the likelihood of being a higher-risk drinker.

Amongst women, seeing more friends/month (RRR=4.21, 95%CI=1.09–16.27) increased the likelihood, and being in the highest (RRR=0.25, 95%CI=0.06–0.95) and middle (RRR=0.21, 95%CI=0.06–0.79) tertiles for practical support reduced the likelihood of being a higher-risk drinker.

2) Amongst men, a reduction in confiding/emotional support (RRR=1.38, 95%CI=1.03–1.86) and an increase in practical support (RRR=1.54, 95%CI=1.01–2.37) were associated with a reduction in alcohol risk category. An increase in friends/month (RRR=1.29, 95%CI=1.03–1.63) and a reduction in practical support (RRR=1.23, 95%CI=1.00–1.52) were associated with an increase in alcohol risk.

Amongst women, only a reduction in the number of close persons was associated with a reduction in alcohol risk group (RRR=1.74, 95%CI=1.05–2.87).

Conclusion Good levels of practical support appear particularly protective of harmful alcohol consumption, whereas seeing more of friends is associated with harmful alcohol use. Interventions which target friendship groups may be more effective in addressing harmful consumption, whilst enabling the maintenance of social support.