May 26, 2011 (CIDRAP News) – It has long been suspected that influenza increases the risk of heart attacks, but the connection has been nebulous because it's difficult to distinguish the effects of flu from the effects of cold weather, which prevails in flu season in temperate regions.
Now, a British-Chinese research team says its analysis of 10 years' worth of flu, heart-attack, and weather data in the United Kingdom, a temperate country, and Hong Kong, a subtropical place, provides stronger evidence that flu contributes to myocardial infarction (MI).
Writing in the Journal of Infectious Diseases, the researchers conclude that up to 5.6% of MI-related deaths in Hong Kong and up to 3.4% of such deaths in England and Wales between 1999 and 2008 were attributable to flu.
"We found a consistent association between seasonal influenza circulation and acute MI-associated hospitalizations and deaths in 2 different settings characterized by differing populations, climates, and patterns of health-seeking behavior," says the report. The study was led by Charlotte Warren-Gash of Royal Free Hospital in London.
In an accompanying commentary, two flu specialists from the US Centers for Disease Control and Prevention (CDC) say the study strengthens the case for a link between flu and heart attacks and points up the need for flu vaccines that are more effective in elderly people.
Comparing MIs and flu activity
The authors say their study apparently is the first to examine the connection between flu activity and national rates of fatal and nonfatal MI in two different places and populations.
The researchers used government sources to gather data on MI-related hospitalizations and deaths in England and Wales and in Hong Kong for the period January 1999 through December 2008. They also relied on official sources for daily temperature and humidity data. For comparison, they also gathered data on two conditions not likely to be associated with flu: colon cancer and femoral neck fractures.
To track flu activity, the team examined general practitioner (GP) flu consultation rates for influenza-like illness (ILI) in the UK, while in Hong Kong they used the percentage of respiratory specimens that tested positive, since ILI data are deemed less specific for flu in subtropical areas. They compared the flu activity and MI data by week and used a regression analysis to control for the effects of temperature and humidity.
The analysis showed there were 1,219,150 MI-related hospitalizations and 410,204 MI-related deaths in England and Wales during the 10 years, for a median of 2,421 hospitalizations and 777 deaths per week, with a marked winter peak. ILI consultations ranged from 0.8 to 270.8 per 100,000 population per week and were highest in 1998-99 and 1999-2000.
For the same period, Hong Kong had 65,108 MI-linked hospitalizations and 18,780 deaths, for medians of 110 and 32 per week, respectively. The Hong Kong data showed a large winter peak and a small summer rise in MIs.
A strong association
After adjusting for temperature and humidity, the scientists found a "strong association" between GP consultations and MI-linked deaths in England and Wales. The association was clearest in models featuring "lag times" of 1 or 2 weeks between MI-related deaths and increases in flu indicators. For a 1-week lag, the incident rate ratio (IRR) was 1.051 (95% confidence interval [CI], 1.043 to 1.058; P<.01).
An association between flu activity and MI-related hospitalizations was also found for England and Wales, but it was smaller than the link between flu and MI deaths.
For Hong Kong the analysis showed a strong link between the proportion of specimens positive for flu and MI-related deaths occurring in the same week (no lag time) (IRR, 1.077; 95% CI, 1.013 to 1.145; P=.018), after adjustments for temperature and humidity. The analysis showed a slightly smaller association between flu diagnoses and MI hospitalizations in Hong Kong.
Depending on the seasonality model used, the researchers estimated that in England and Wales, 3.1% to 3.4% of MI-related deaths and 0.7% to 1.2% of MI-related hospitalizations were due to flu. For Hong Kong the estimates were 3.9% to 5.6% of deaths and 3.0% to 3.3% of hospitalizations. All of the estimates were found significant, with P values of .018 or lower. In both countries, the link between flu and MI was strongest in the oldest groups (80 years and older).
The findings were most striking for weeks of peak flu activity, according to the authors' models. At those times, 9.7% to 13.6% of MI deaths in Hong Kong and 10.7% to 11.8% of MI deaths in England and Wales were attributable to flu.
The analysis showed no association between flu activity and colon cancer or fractures of the femoral neck.
Study helps build the case
In the commentary, Lynn Finelli and Sandra S. Chaves of the CDC write, "This well-designed and -analyzed study adds to the pool of evidence supporting a relationship between influenza and AMI [acute MI] independent of temperature and humidity and is the first to our knowledge to propose influenza-attributable proportions for acute myocardial hospitalization and death."
The commentators note a couple of limitations in the study. For one, the authors did not look into the effects of viruses other than flu. Another problem is the "inconsistent lag times observed between countries and among outcomes for ILI visits/influenza circulation and hospitalizations and deaths."
The study does not address exactly how influenza might contribute to cardiovascular events. Finelli and Chaves say acute respiratory infections trigger responses such as an increase in white blood cells and release of cytokines, which may make atherosclerotic plaques more prone to rupture and block coronary arteries.
As for the overall case for flu as a contributor to MI, the commentators say, "There is consistent ecologic evidence that overall cardiovascular mortality is related to influenza virus circulation and ILI activity, and there is consistent ecologic and individual level evidence that occurrence of AMI can be temporally related to acute respiratory infection including ILI."
The data "are compelling so far," but the evidence is not yet conclusive, they add. Still needed are prospective studies of MI in individuals with laboratory evidence of flu infection.
Praise and questions
The study drew qualified praise from another veteran flu researcher, Lone Simonsen, PhD, research director at Georgetown University's Department of Global Health in Washington, DC.
"This is a carefully conducted time series analysis, making a good case for MI being associated with influenza. I especially like the consideration of non-influenza outcomes like colon cancer and hip fractures," she commented by e-mail.
She also hailed the use of two locations with different climates and flu seasons and the tracking of MI events in entire populations—"very different from longitudinal studies where the number of MI events is typically limited."
However, Simonsen said she was bothered by the finding that in the UK, the link between flu and MI was clearest in models showing increased MI events before, rather than after, increases in flu activity.
"Usually in this type of study one would expect to see a temporal pattern of influenza peaks followed by MI peaks," she said. "For example, it is understood that pneumonia hospitalizations and deaths typically peak a few weeks AFTER peak influenza activity. . . . For England and Wales, the MI peak for hospitalizations (and deaths) occurred 2, 3, and 4 weeks BEFORE the influenza peak.
"This may of course be explained by some delay in adult patients seeking care for influenza—2 to 3 weeks is a long time. But one has to wonder about the possibility that other winter-seasonal pathogens are playing a role; for example RSV [respiratory syncytial virus] activity often peaks earlier than influenza."
Simonsen suggested that further research on the issue should focus on flu epidemics with unusual timing, such as the early flu season of 2003-04 and the H1N1 pandemic in the summer and fall of 2009. If there were early peaks of MI activity in those seasons, it would strengthen the evidence that the association is real, she said.
The study authors address the issue of the UK lag times between MI and flu activity in their discussion, commenting that one British study showed that phone calls to the National Health Service for colds and flu preceded GP reports of the same symptoms by 1 to 3 weeks. "This may explain why we saw the best model fits, and greatest estimates of effect, when UK influenza data was lagged by -1 to -3 weeks (representing our assumptions that reported ILI consultations represent illness occurring in the community some time earlier)."
However, they acknowledge that "peaks in MIs might precede GP reports of ILI if triggered by other synchronous environmental events." But they add that they used sensitivity analyses to control for temperature.
Policy implications
Finelli and Chaves comment that the findings imply that increasing vaccination coverage in people at risk for heart attacks might reduce their risk of such events. However, they note that several retrospective studies of the effects of flu vaccination on the risk of cardiovascular events have yielded inconsistent results.
"Because influenza vaccine effectiveness is suboptimal, especially in older people, the opportunity to prevent influenza-related complications in this population will benefit from the development of more immunogenic and effective vaccines," they write.
Warren-Gash C, Bhaskaran K, Hayward A, et al. Circulating influenza virus, climatic factors, and acute myocardial infarction: a time series study in England and Wales and Hong Kong. J Infect Dis 2011 Jun 15;203(12):1710-7 [Abstract]
Finelli L, Chaves SS. Influenza and acute myocardial infarction. (Commentary) J Infect Dis 2011 Jun 15;203(12):1701-4 [Full text]
See also:
Sep 21, 2010, CIDRAP News story "Study on flu shots and heart-attack risk questioned"