Why Science Reporting Stinks 0

Because it doesn’t tell enough and it misses the point.

Case in point missed:

The Daily Progress reports on a study by the University of Virginia in conjunction with the Dial Soap people. The study purports to prove that alcohol hand sanitizers–the kind the everyone set out in the waiting room last flu season–aren’t very effective against colds and flu.

Here’s the two crucial pieces from the Daily Progress story:

Study subjects who used the sanitizers had 42 rhinovirus infections per 100 volunteers, compared with 51 infections per 100 volunteers who didn’t take special precautions. Influenza infections hit 12 of 100 subjects who used the sanitizer, compared with 15 per 100 subjects who didn’t take special precautions.

(and the last paragraph)

Turner said his findings aren’t cause for panic. He said studies have shown that hand sanitizer is effective for gastrointestinal diseases, particularly in the developing world. According to a 2002 CDC study, sanitizers did a better job reducing bacteria on hands than did antibacterial soap.

The Daily Progress is the hometown paper for Charlottesville, Va., the site of the University of Virginia. The version that went out over the wires is much shorter, more like this.

To evaluate this, the reader needs to know how large the sample was. The results are given in pseudo-percentages, such as “42 our of 100.” A someone skimming the story could conclude that 100 persons were in the sample, but that’s neither stated nor implied under a careful reading.

I also find the conclusion questionable. Here’s the lead from the Daily Progress and most wire versions I’ve found:

Using alcohol-based hand sanitizer doesn’t significantly decrease how often someone is infected with a cold or flu, a University of Virginia study has found.

Look at the figures: the incidents of colds was reduced by 9 per 100, to use the story’s construct. That’s 18%. Given the pervalence of colds, that’s not insignificant.

The incidence of flu was reduced from 15 per 100 to 12 per 100. Let’s do the math:

3 case reduction divided by 15 cases in control group = 20%

Again, not an insignificant reduction.

I suspect that the stories are based on a UVa press release; I haven’t looked for it and it’s not the issue any way. The issue is that the numbers do not support the headlines. If the headlines are based on the wording of press release, reporters ought to have done the math and pressed (heh) for more information.

Frankly, an ad campaign that X may reduce your chance of catching the common cold by 20% would be a pretty good ad campaign, especially if it didn’t include fully dressed people in bath tubs.

Full Disclosure: I don’t commonly use alcohol hand sanitizers unless I’m visiting someone in the hospital, something I fortunately haven’t had to do lately.