“Smoking may be addictive but quitting is contagious, according to a provocative study of why people give up the weed”, reported The Times today. It says that the findings come from a 32-year study that collected data from more than 12,000 people. When people quit smoking it had a knock on effect on their families, friends and work colleagues. People whose spouses quit were 67% less likely to smoke, while friends of quitters were 36% less likely and siblings 25% less likely.
The research has used new methods to look at data from a previous study. The researchers assessed peoples’ smoking habits and looked at what effect quitting had on the chance that a husband, wife, brother, sister, friend or workmate would continue to smoke. This approach to looking at social influences on quitting provides reliable evidence and some measure of how groups of people can affect each other’s non-smoking habits. It sheds some light on what the researchers describe as the “the collective dynamics of smoking behaviour”.
Where did the story come from?
Dr Nicholas Christakis from Harvard Medical School and the Department of Sociology, Harvard University in Boston and James Fowler from the Department of Political Science, University of California in San Diego carried out the research.
The study was funded by grants from the National Institutes of Health and the Robert Wood Johnson Foundation, and by a contract from the National Heart, Lung, and Blood Institute to the Framingham Heart Study. The study was published in the peer-reviewed: New England Journal of Medicine.
What kind of scientific study was this?
This was a secondary analysis of data collected from a prospective cohort study. The researchers used computer models to perform a complex statistical analysis of the people who quit smoking and the probability that the people that they knew would later also quit.
The data came from a large, long-running study called the Framingham Heart Study, which has been following people and their social networks in the town of Framingham in the US for 32 years.
Since it began, 12,067 people have taken part and have had repeated assessments of their social networks and their smoking status. When it started in 1948, there were 5,209 subjects in the original group, or “cohort”. A second “offspring” group followed in 1971, which enrolled 5,124 of the original group’s children and their spouses. This was followed by another group of 508 people in 1994, and a “third generation” cohort in 2002, which consisted of 4,095 children of the offspring group.
The researchers concentrated on the “offspring cohort” of 5,124 subjects, and found 53,000 family ties with other people in the network, an average of 10.4 family ties per subject. Most of the subjects had wives and husbands, or at least one sibling who were also in the network. For example 83% of the spouses of subjects were also in the network. Less of the subjects, 45%, were connected by friendship with others in the network. Only people over 21 were included in the study (average age 38 years).
The previous studies had collected data on how many cigarettes the subjects smoked. However, the researchers decided to classify anyone who smoked more than one cigarette a day as a smoker.
They looked at this data over time by extracting it from the examinations and questionnaires completed at different time points. In this way, the researchers obtained smoking histories at seven time points, each spanning about three years of data collection, from 1973 to 1999. They also collected data on the level of education and geographical proximity of the subjects to their contacts. The statistical analysis was based on the first difference noted in the contacts’ smoking behaviour at the closest time point.
What were the results of the study?
The researchers found that smokers and non-smokers “clustered”, meaning that they grouped together in such a way so that that smokers were more likely to be linked to, or know other smokers and non-smokers were more likely to be linked to non-smokers. This clustering extended to three degrees of separation. The researchers say that “despite the decrease in smoking in the overall population, the size of the clusters of smokers remained the same across time, suggesting that whole groups of people were quitting in concert.
When a husband or wife quit the chance that their spouse would smoke, fell by 67%. When a brother or sister quit, the chance a person smoked decreased by 25%. Smoking cessation by a friend decreased the chances by 36% and among people working in small firms, smoking cessation by a co-worker decreased the chances by 34%. All these results were statistically significant. Friends with more education influenced one another more than those with less education. These effects were not seen among neighbours in the immediate geographic area.
What interpretations did the researchers draw from these results?
The researchers concluded that the person-to-person spread of smoking cessation appears to have been a factor in the decline in smoking seen in the population in recent decades. They say that smoking behaviour spreads through close and distant social ties in which “groups of interconnected people stop smoking in concert”.
What does the NHS Knowledge Service make of this study?
The large amount of data collected in these cohort studies has been reanalysed using computer simulations and mathematical modelling. There are some limitations common to these types of studies that should be considered.
- There may be additional factors that influence behaviour amongst groups of people that were not measured by these researchers. For example, the exposure to smoking cessation campaigns or cigarette taxes may affect all closely connected people together and could have had a small influence on the reliability of the results. However, the large size of the social effect shown and the fact that the researchers were able to demonstrate that one person quitting followed another suggests that these non-social factors may not be an important source of bias.
- The division of smokers into those who did not smoke and those who smoked more than one cigarette disguises a lot of variation in smoking behaviour. This point and the fact that questionnaires were used to collect the data, may have resulted in less accurate data on people who stopped and started multiple times or began to quit by cutting down on smoking. Although capturing these types of data would have added strength to the study, it is again unlikely to have overturned the main conclusions.
The results will not be surprising to social researchers and adds strength to the arguments for exploiting these sorts of social dynamics in encouraging the spread of healthy behaviours. The authors discuss how their findings suggest that collective interventions may be more effective than at first thought and particularly promote the notion that by targeting small groups, positive health behaviour changes might be spread to others.
Sir Muir Gray adds...
Smoking is an infectious disease.