Monday August 15 2011
The study examined over 80,000 women
“Depressed women have increased risk of stroke,” BBC News has today reported, saying that depression increases risk of stroke by 29% in women. The Daily Mail also presents this figure but claimed that antidepressants such as Prozac increase the risk by nearly 40%.
The news is based on a large US study that followed more than 80,000 female nurses between 2000 and 2006. They found that women with a history of depression – defined by meeting a symptom score on a mental health test, having a doctor’s diagnosis or antidepressant use – had a 29% greater risk of stroke during the follow-up period compared with women with no history of depression.
Further analysis found that meeting a symptom score or having a doctor’s diagnosis was not in itself associated with stroke risk if the person had never taken antidepressant medications. Taking antidepressant medications was associated with increased risk, even if they did not meet depression criteria. While this might initially suggest that antidepressants were behind the increased stroke risk, it should not be assumed that this is the case, and the risk may be related to the underlying condition that is being treated, rather than the drugs itself. For example, people who have required antidepressants to treat their depression may have had more severe depression than those who have not required medication. Also, ‘antidepressants’ can be prescribed for a number of other conditions aside from depression, such as anxiety and chronic physical pain.
The observed association between depression, antidepressant medication use and risk of stroke is complex and deserves further study. Nevertheless it is important to note that the benefits of effective treatment for depression are likely to outweigh any small risk of stroke – if there is indeed one at all. As said, the reasons for the observations are unclear and not definitely known. People should not stop taking any medications that they have been prescribed on the basis of this study.
Where did the story come from?
The study was carried out by researchers from the University of Harvard and the University of Bari in Italy. It was funded by the Institute of Health, the US National Heart, Lung, and Blood Institute, and the US National Alliance for Research on Schizophrenia & Depression. The study was published in the peer-reviewed medical journal, Stroke.
BBC News has accurately reported this research, but the Daily Mail’s focus on one part of this observed complex relationship, is misleading. In particular, it is completely incorrect to highlight Prozac as a risk factor, as this study has not examined any individual drug.
What kind of research was this?
This was a large prospective cohort study, which aimed to investigate whether women with depression are at increased risk of stroke. The authors say that although depression has been associated with an increased risk of coronary heart disease, prospective data for the specific association with stroke are limited.
The approach used in this study, i.e. following women with the exposure (depression) before they have experienced the outcome (stroke), is a good way to examine the issue as it would ensure that depression definitely preceded stroke. However, to be most accurate, the study would also need to ensure that women were free of any cardiovascular disease prior to onset of depression.
What did the research involve?
This research included women in the large Nurses’ Health Study, a research project established in 1976 to look at various aspects of health. The study enrolled 121,700 female nurses aged 30-55 at the start of the study from across the US. Lifestyle and medical health were assessed by mailed questionnaire every two years. Through to 1996 the study had retained over 94% of the entire cohort for follow-up.
This study on stroke risk specifically used the questionnaire from the year 2000 as the starting point for its analysis because it was the first year to explicitly record physician-diagnosed depression. After exclusion of women who had no history of depression symptoms, depression diagnosis or antidepressant drug use, and exclusion of women who had already experienced a stroke, the researchers were left with 80,574 women for their study, ranging from 54 to 79 years old.
In 2000 (and also in 1992 and 1996) a Mental Health Index (MHI-5) score was used to assess women on the presence of depressive symptoms. Clinically-significant depressive symptoms were defined as a score of 52 or less. Women with a score above 52 were defined as not having depression. Questions on use of antidepressant medication were given every two years since 1996. Reporting of physician-diagnosed depression began in 2000. This study defined depression as currently reporting or having a history of any of these three conditions (a MHI5 score of 52 or less, a physician’s diagnosis of depression, or taking antidepressant medications).
The researchers followed these women for six years up to 2006, and they looked for the outcome of stroke using the National Death Index and postal authorities. They attempted to verify all reports of stroke through medical records, autopsy reports and death certificates. Strokes were classed as confirmed strokes if a medical record or death certificate was available, and it met the National Survey of Stroke criteria (requiring neurological deficit of rapid or sudden onset lasting more than 24 hours, or until death). Those which were self-reported by the person or their next of kin but could not be verified were designated as probable strokes.
What were the basic results?
During six years of follow-up 1,033 strokes were documented: 538 ischaemic strokes (caused by a blood clot), 124 haemorrhagic (caused by a bleed in the brain), and 371 unknown strokes. Of these 1,033 strokes, 648 were confirmed strokes and 385 were probable.
In a model that adjusted for multiple confounders (including age, cardiovascular risk factors, socio-demographics, lifestyle factors and other medical illness), women with any history of depression (meeting any of the three above criteria) had a 29% increased risk of any type of stroke compared to women with no history of depression (hazard ratio 1.29, 95% confidence interval 1.13 to 1.48).
The researchers then looked at the different combinations of defining criteria for depression separately. They found that:
- Women who met the MHI-5 symptom score or had depression diagnosed by a doctor but had no history of antidepressant use had no significantly increased risk.
- Women who met the MHI-5 symptom score or had depression diagnosed by a doctor and had a history of antidepressant medication use had a 39% increased risk (HR, 1.39, 95% CI 1.15–1.69).
- Women who had a history of antidepressant medication use, but did not have depressive symptoms on the MHI-5 symptom score and did not have depression diagnosed by a doctor had 31% increased risk of stroke (HR 1.31, 95% CI 1.03 to 1.67).
On further subanalysis they also found that, compared to women with no history of depression or use of antidepressants, women who reported current depression at questioning had an increased risk of stroke, whereas individuals who only had a history of depression did not have increased risk of stroke.
How did the researchers interpret the results?
The researchers conclude that depression is associated with a “moderately increased” risk of stroke.
This study has several benefits, including its large size and thorough follow-up which used clear criteria to define cases of depression, including use of Mental Health Index scores and clinical diagnoses to classify cases of depression. It also attempted to validate all reports of stroke using medical records, and also featured adjustments to account for the influence of multiple potential medical, lifestyle and sociodemographic confounders.
The researchers found that a history of depression (defined through doctor’s diagnosis, antidepressant use or MHI-5 score) was associated with a 29% increased risk of stroke. Interestingly, antidepressant medication use seemed to have a particular association: meeting a symptom score or having a doctor’s diagnosis was not associated with stroke risk if the person had never taken antidepressant medications. But taking antidepressant medication was associated with increased risk, even if they did not meet depression criteria.
Importantly though, the reasons for this and the mechanisms underlying the associations are unclear. It should not be assumed that antidepressants themselves carry an increased risk of stroke based on these results. It may be that people who have required antidepressant medications to treat their depression have had more severe depression than those who have not required medication. Also, although being named ‘antidepressants’ these types of drugs are not only used in the treatment of depression. They are prescribed for various other mental health conditions (e.g. anxiety) or physical conditions (e.g. treatment of chronic pain). Therefore it is difficult to unpick the complex relationship that may exist between stroke risk and the underlying conditions that ‘antidepressants’ are used to treat.
Further points to note on this study:
- Only 63% of strokes were validated by medical records and death certificates - the remainder were considered ‘probable’ strokes obtained by self-report only. The researchers have combined both confirmed and probable strokes in their analyses and do not appear to have carried out separate analysis using confirmed strokes only, which may have been more accurate.
- Although participants with a past history of stroke were excluded from the analyses, people with cardiovascular disease or cardiovascular risk factors such as high blood pressure or diabetes (which are associated with stroke risk), do not appear to have been excluded.
- Likewise, it is unclear whether people with a history of transient ischaemic stroke (mini-stroke lasting for <24 hours) were also excluded at baseline. If they were not excluded, the study would have included people who were already at high risk of stroke when their depression or medication use was assessed. Therefore it is difficult to definitely conclude a temporal relationship and assume that the depression or antidepressant medication use has preceded the development of the cardiovascular disease process that led to stroke.
- The study population was all women and all nurses. With such a distinct population there may be difficulty in generalising the results elsewhere as their health-related behaviours may not match that of the general population.
As the researchers say, their observed association between depression and antidepressant medication use, and risk of stroke deserves further study. Nevertheless it is important to note that the benefits of effective treatment for depression are likely to outweigh any small risk of stroke – if there is indeed one at all. In short, the reasons for the observations are unclear and not definitely known. Therefore people should not stop taking any medications that they have been prescribed on the basis of this study.