Citizen is 100% wrong
13 March 2013
Headline grossly misrepresents research findings to suggest that half of people living with HIV will die of heart failure.
Statistics are an extremely effective shorthand for conveying ratios and relationships with high impact. Their charisma is apparent in our daily references to well-known figures, like “90% of lung cancer cases are smoking-related” or “most accidents happen near the home”.
In fact, so effective are statistics that more than 85% of people make up their own percentages to prove a point. (See what I did there?)
That they so easily become part of the social currency makes it extremely tempting to misrepresent information for the sake of impact, or otherwise to over-simplify figures to get people’s attention.
Percentages are in effect highly condensed forms of data and need to be used with extreme care. The bigger picture is usually informed by substantial quantities of research, often gathered over extended periods of time from highly nuanced, individual cases.
As the purveyors of research findings (and many other statistics) to the public, journalists have a very serious responsibility to ensure that when they distil complex and complicated information, they do so correctly.
This is where a subeditor at the Citizen failed dismally last week in writing a headline for a snippet syndicated from the press agency Reuters.
In conveying the findings of an American study on the heart health of veterans living with HIV, the Citizen committed a classic error in its interpretation of research.
Whilst the flaw itself is truly worrying, this case offers a textbook example of how easily statistics fall prey to ignorance or laziness.
The misunderstood research, part of the Veterans Aging Cohort Study (VACS), involved over 82,000 US veterans of which one third are living with HIV.
Researchers found that nearly twice as many of the men who were living with HIV had suffered heart attacks than those who were HIV-negative.
The Citizen took this “twice as likely” risk to mean “50% of victims likely to have heart attack”.
Ignoring for a moment that the Citizen’s interpretation defies conventional math, with a bit more reading, this embarrassing error could have been avoided entirely.
In the study, the total number of heart attacks recorded was 871 (out of 82,458 participants). Among HIV-positive veterans aged 40 to 49, 2 heart attacks per 1000 were recorded, compared to 1.5 out of 1000 for HIV-negative veterans in the same age group. Among veterans aged 50 to 59, 3.9 out of 1000 HIV-positive men had heart attack events, compared to 3.3 out of 1000 among uninfected men of the same age. Once the researchers adjusted for variables, they were able to conclude that the HIV positive men were nearly twice as likely to suffer heart attacks.
From the research above it is quite apparent that the incidence of heart disease among men living with HIV is nowhere near half.
For the nearly 6 million South Africans living with HIV who may be reading this headline, or hearing it recounted by a friend, this information will no doubt incite a great deal of (largely unnecessary) panic.
It is a shame that South African readers cannot trust the press to bring them 100% accurate and reliable information.
To learn more about the ways in which statistics can be misinterpreted, go here: http://www.columbia.edu/cu/21stC/issue-3.3/ross.html.
For a tutorial on using statistics in the media, view this guide: http://nilesonline.com/stats/index.shtml
blog comments powered by Disqus