Counting Gun Traffic

S. Robert Lichter, Ph.D

“In Study of Gun Traffic, Va. Stands Out,” reads the front-page headline in the August 21 Washington Post. According to the lead paragraph,

“Law enforcement authorities traced more than 10,000 guns recovered in Virginia, Maryland and the District [of Columbia] last year – and nearly half came from Virginia, according to federal data released yesterday.”

Say what? It’s front-page news when you combine the number of guns from two states and a city, and one state accounts for nearly half the total of these three jurisdictions? In fact, Virginia’s share was actually 46 percent.

We know this because the Post kindly presented the actual figures in a table on an inside page: Virginia 7,571 guns recovered, Maryland 7,025 guns, and DC 1,831.That means Virginia outpaced Maryland by just under eight percentage points, and both far outstripped Washington DC, which has a far smaller population than either state. But that’s hardly front-page news.

To see just how misleading the Post’s lead was, we can make a simple statistical adjustment that reflects these population differences. According to July 2006 US Census estimates rounded to the nearest thousand, Virginia’s population is approximately 7,643,000, Maryland’s is 5,616,000, and the District of Columbia’s is only 582,000. That means Virginia, which accounted for eight percent more guns than Maryland, also accounts for 36 percent more people.

And on a per capita basis, the rate of guns recovered was about ten per 100,000 people in Virginia, thirteen per 100,000 in Maryland, and 31 per 100,000 in the District. So the article could just as easily have concluded that guns were recovered at a rate three times as high in DC as in Virginia, relative to population.

The article goes on to discuss other evidence, such as other east coast states where Virginia-traced guns have turned up. But the focus is on the DC-VA-MD comparison a local angle that backfires. In fact, we recommend that the Post switch to higher caliber information in its bulletins on firearms.

Update
The Washington Post is not the only paper to stumble over the data: In today’s New York Times, the editorial page notes,

“Virginia dealers have been a standout source for guns used in crimes up and down the seaboard, according to the federal study. They accounted for half of the 10,000 guns tracked by the study in the metropolitan Washington, D.C., area, and one in 11 in New York City.”

Originally published August 22, 2007

Are Schools Drug “Infested?”

Maia Szalavitz

Contrasting reports on a new “study” by Columbia’s National Center on Addiction and Substance Abuse: why some readers were better served by journalistic skepticism… more

Originally
published August 16, 2007

Is Tanning Addictive?

Maia Szalavitz

Do chronic tanners suffer withdrawal from the heroin-like effects of sunlight on skin? … more

Originally published August 15, 2007

Measuring the Toxicity of New Car Smell

Trevor Butterworth

Environmentalists have been warning about the poisonous smell of a new car for the past two years. But now that someone has actually tested the air, guess what stinks?… more

Originally published August 13, 2007

Look Who Isn’t Getting It On

Trevor Butterworth

Gina Kolata finds the math that makes men more sexually active than women doesn’t add up… more

Originally published August 13, 2007

The Cost of Media Scare Stories to Diabetics

Trevor Butterworth

The current issue of Biocentury, a trade publication that covers the pharmaceutical industry, should be essential and chastening reading for the journalists, editors and politicians who helped turn a limited, problematic study on the risks of the diabetes drug rosliglitazone (brand name Avandia) into a major international health scare.

In a comprehensive report on the FDA advisory panel meeting that analyzed the safety data on rosliglitazone, Biocentury editor Steve Usdin noted several key points that were absent from mainstream media accounts.

Drug safety advocates Steve Nissen (who kicked off the scare over Avandia with a meta-analysis in the New England Journal of Medicine purporting to show a 43 percent increased risk of heart attack for diabtetics taking Avandia) and Curt Furberg (who co-authored two editorials in the Journal critical of the FDA’s oversight of Avandia) both had publicly stated that Avandia was of no real benefit to diabetics given the risks. As Usdin noted,

“Nissen and the NEJM editorialists built their case on two assumptions. First, they stated that hemoglobin A1c, the efficacy endpoint in registration trials of Avandia and other marketed diabetes medicines, is not indicative of clinical benefit. The HbA1c value is the sum of both fasting and postprandial glucose excursions over a 2-3 month period.

Second, they said prevention of macrovascular events, those episodes involving large blood vessels such as myocardial infarction and stroke, is the primary reason to treat.”

But Robert Ratner, VP of scientific affairs at MedStar Research Institute countered that the 1993 National Institute of Health’s 10-year Diabetes Control and Complications Trial (DCCT) and the 1998 20-year United Kingdom Prospective Diabetes Study (UKPDS) both showed that glycemic control was not only effective in preventing and slowing the impact of microvascular complications associated with diabetes, such as eye, kidney and nerve diseases, it also reduced the incidence of fatal and non-fatal cardiovascular events.

If the FDA was to abandon glycemic control as a valid endpoint for approving diabetes drugs, and focus only on whether there was a risk of heart attack, there would be no drugs to treat diabetes, Ratner said.

This didn’t, of course, answer the critical question of whether there was clinical evidence that taking Avandia specifically prevented or slow these diabetes-related diseases. And David Graham, associate director for science and medicine in the FDA’s Office of Drug Safety called for Avandia to be withdrawn from the market because no such clinical evidence existed. However, he was challenged by panel member David Schade, director of the Core Laboratory and vice chairman for research at the University of New Mexico School of Medicine, who said,

“I agree that we haven’t seen rosiglitazone’s improvement of microvascular complications. I also argue I hope we never do. I don’t think anyone questions that if you lower hemoglobin A1c it will prevent or delay microvascular complications. No one would want to do that trial because nobody would want to be in the control group.”

In other words, given that we know glycemic control prevents microvascular complications, and given that rosiglitazone is proven at glycemic control, we can infer that it works at preventing such problems. To look for stronger causal evidence would be unethical as it would mean putting the diabetics in the control group at risk of disease because they wouldn’t be on any medication at all.

The important of maintaining glycemic control was underscored in comments to Biocentury by Chris Viehbacher, president of U.S. pharmaceuticals at GlaxoSmithKline.

“We surveyed around 200 physicians, primary care and endocinologists. Eighty percent said they had patients who just stopped taking Avandia. They didn’t switch to anything else, they stopped taking it on their own. Those same physicians said they have up to 40% of patients coming back with their glucose control out of whack.”

This is the real cost of the media choosing to run stories about Avandia’s risks without considering the statistical limitations of the study claiming those risks. The credibility of that study by Steve Nissen and Kathy Wolski was continually questioned by other statisticians and doctors – particularly those involved in the clinical management of diabetics – and yet the media coverage rarely acknowledged their criticism or weight of the evidence supporting their criticism.

And criticism of Nissen and Wolski keeps on coming: In the latest issue of the Annals of Internal Medicine, a team of researchers re-analyzed their data with more powerful meta-analytic tools to account for the studies which showed no adverse cardiovascular events and which Nissen and Wolski had left out of their analysis. When the results from these studies are factored in, there is no statistically significant increased risk of heart attack The researchers “conclude that the risk for myocardial infarction and death from cardiovascular disease for diabetic patients taking rosiglitazone is uncertain: Neither increased nor decreased risk is established.”

Originally published August 10, 2007

Healthiest Food Poses Greatest Risk When Eating Out

Trevor Butterworth

Healthinspections.com analyses CDC data and finds lettuce salads and raw veggies are the leading cause of food poisoning in restaurants… more

Originally published August 8, 2007