James C. Coyne Ph.D.

The Skeptical Sleuth

Therapy

Could Couples Therapy Prevent a Heart Attack?

Is there really a marital spat-heart attack connection?

Posted Jan 12, 2011

There is certainly evidence that acute stress such as a marital argument can precipitate a heart attack in men and women. But does that mean that going into couples therapy is an effective strategy for preventing a heart attack?

Considering this question will serve as the inaugural topic for a the Skeptical Sleuth, a blog that will discuss psychological research findings and claims about these findings in the media, with a critical look at the evidence that can be summoned for and against a particular claim. The goal will be to cultivate an open-minded skepticism about such claims, but, more importantly, to provide my readers with a set of tools to evaluate for themselves what they read.

To estimate the size of the link between marital conflict and cardiac events, we can look to a large study1 of almost 7700 British government workers who were followed for 12 years after filling out a questionnaire about positive and negative aspects of their close relationships. Persons in that study who reported negativity in their close relationships were one third more likely to have a coronary event in the 12-year time period, when coronary events were measured in terms of a fatal or nonfatal heart attack or evidence of objective confirmation of angina in their medical records. This sounds impressive, particularly when an increased risk of a quarter still held up after lots of other competing explanations or confounds were taken into account, like age, sex, marital status, and medical factors such as hypertension, high cholesterol, obesity and diabetes, as well as health behaviors such as smoking and drinking.

The hazards ratios of a third and quarter are ways of expressing the association between risk factors like a negative marriage, and health outcomes, like cardiac event. However, they can prove deceptive because they do not take into account absolute risks. So, we need to ask how likely overall was a cardiac event in this sample over 12 years of follow up and how did this risk vary with whether people reported a negative close relationship? The full sample of almost 9000 people, including the 7700 who completed the questionnaire, experienced 589 coronary events. Examining a table in the journal article, we find that 214 of the people in the worst top third of negativity in close relationships had a coronary event, versus 189 in the middle third and 186 in the lowest third. Only the difference between the highest and lowest third was significant, and the difference was no longer statistically significant when all the confounds were taken into account. So practically speaking, we are talking about a difference of 28 more cardiac events in 2757 persons with the most negative relationships compared to what occurred in the 2957 with the best close relationships. This sounds a lot less impressive than the figures of one third and one fourth more that were brought up earlier. How can this be? Answer: Most people in this sample did not have a coronary event in 12 years, and so, in absolute terms, there were not many events to be explained in terms of differences in the quality of relationships.

Should the 2757 persons with the most negative relationships have gone to therapy to prevent an excess of 28 coronary events among them? Perhaps for other reasons, but even if all got so much benefit from therapy that they moved from the highest to the lowest third in negativity, we probably could not expect to see much reduction in coronary events. And all of the 2757 people obtaining that much sustained benefit from couples therapy is rather unrealistic. So, there are lots of reasons for seeking couples therapy if a relationship is unhappy and full of conflict, but to do so simply with the aim of avoiding a coronary event is not realistic. Concentrating on lowering hypertension, stopping smoking, losing weight, or reducing cholesterol would be seem to be better strategies.

I am making a number of simplifying assumptions, but undoing them would not greatly affect the pattern of results. What is the lesson here? First, we need to look beyond the hazards ratios that are highlighted in journal articles and emphasized in media reports and ask about absolute differences in outcomes associated with particular risk factors. These absolute differences are typically downplayed and one has to go searching in the tables in the journal article to find them. Hazards ratios of a third or a quarter may not actually represent big differences. Second, in relatively healthy populations and with multiple risk factors at play, we cannot expect dramatic gains in targeting just one factor that does not explain much of the risk anyway.

1 De Vogli, R, Chandola, T; Marmot, MG. Negative Aspects of Close Relationships and Heart Disease, Arch Intern Med. 2007;167(18):1951-195 (available free at http://alturl.com/tadsv)