Health survey dilemma: Can you hear me now?

Washington -- As more Americans ditch their traditional home telephones in favor of cell phones, researchers who keep tabs on the nation's health via telephone are becoming increasingly concerned that important segments of the population are being missed.

For the first time, bias threatens surveillance systems used by state and local health departments to establish policies on an enormous range of health behaviors, from alcohol use and smoking to diabetes awareness and who's getting a flu shot. The fear is that population-wide interventions could be off-target, and even physicians' understanding of patients' health risks may be somewhat askew.

Many telephone surveillance systems -- including the world's largest, the Behavioral Risk Factor Surveillance System, which was established in 1984 by the Centers for Disease Control and Prevention and state health departments -- traditionally have collected data by landline phones. Now, sizeable chunks of the population have chosen to become entirely wireless, making it much more difficult -- and expensive -- to reach them.

Data from the 2006 National Center for Health Statistics' National Health Interview Survey, which is conducted in person, indicate that 25% of 18- to 29-year-olds and 32% of low-income young adults live in households with only wireless telephones.

The number of cell-only users rises to nearly a third among young adults ages 25 to 29, said Mike Battaglia, a vice president at Abt Associates, a research consulting firm in Cambridge, Mass.

Missing these people means that there could be underestimates of the prevalence of binge drinking, smoking and HIV testing, or that obesity may have been overestimated and physical activity underestimated, according to a report by NCHS researchers slated for publication in an upcoming Public Opinion Quarterly.

The NHIS has been tracking cell-phone use for several years on its survey instrument. When tracking began, the cell-phone-only group made up a tiny minority of the population. Since then, that number has exploded and shows no signs of slowing.

The quest for accuracy

"We are starting to see for the first time some bias that is meaningful," said Stephen Blumberg, PhD, senior scientist at the NCHS and an author of the Public Opinion Quarterly report. "Now we do need to think more carefully about what to do about it."

Ali H. Mokdad, PhD, branch chief of the CDC Behavior Surveillance Branch, is thinking a lot about the Behavioral Risk Factor Surveillance System and is experimenting with collecting data on cell phones and via mail surveys. "I can tell you for sure that BRFSS, which is the largest telephone survey and one of the most important for state and local health departments, will be -- maybe in two or three years -- not a telephone-only survey but a multiple-mode survey."

As soon as the growth in cell-phone-only households became apparent, the CDC began to look for ways to bring that population back under the surveillance umbrella, Dr. Mokdad said. But the challenges are many, and, for those rolling up their sleeves, it's new terrain. "The cell-phone-only population is a totally new game for us," Dr. Mokdad said.

One alternative being explored in several pilot studies is to call people on their cell phones -- 16 states will do so next year for the BRFSS.

But this approach is not simple. For one thing, it's more expensive, costing $75 to $100 per interview, versus $40 to $45 on landlines, Dr. Mokdad said.

The cost is higher because interviewers have to dial a lot more numbers to reach their party. By using random-digit-dialing programs on computers, a representative sample of landline users can be reached fairly easily. Such automatic dialing is illegal for cell phones, so surveyors must dial by hand.

They have to dial many more numbers to get a response, since cell-phone numbers had been changing frequently and lists became outdated. Another wrinkle is that they have to make sure the person who answers isn't driving or in some other unsafe place, Dr. Mokdad said.

Concerns have been raised about the quality of the data gathered. Since multitasking is common while talking on a cell phone, the answers provided may not be as thoughtful as they would be if obtained from someone who is sitting at home. Plus, are the individuals in a location where they can speak freely? No one wants to answer questions about their drinking or sexual practices while in the mall.

The question of who pays for the interview is also an issue. Since cell-phone users pay for their incoming calls, reimbursement is generally offered by the data seekers.

Meanwhile, another line of research is examining whether mail surveys might be a good way to reach the cell-phone users. "Surprisingly, the response rate from mail surveys doesn't differ that much from the response rates we were getting on the telephone," said Scott Keeter, director of survey research for the Pew Research Center for the People and the Press, an independent, Washington, D.C.-based opinion research group.

While mail surveys may be cost effective, they don't necessarily work well for people who are less educated and for some minority populations, particularly Hispanics who are not English-speaking, Keeter said.

In addition, statisticians are working on adjustments that can be made to account for the cell-phone population that might be missing from the landline surveys, Dr. Blumberg said.

Despite the challenges, the researchers remain optimistic. "I think we will successfully find ways to get the information that public health needs," Dr. Blumberg said. "Those ways might become more expensive. They might be more time consuming and they might need more statistical adjustment, but I have no doubt that we will be able to continue to collect the information. I just can't tell you exactly how we will be doing it five years from now."

Dr. Mokdad also is confident in the data he is gathering. "But that doesn't mean I'm resting," he said. "Today we have 10% to 12% of the population that uses only cell phones, but that may double. This is a moving target."