In a new study, portable meters used to gauge blood sugar levels in pregnant women with diabetes gave readings that differed from lab tests by up to 16 percent.

That kind of inaccuracy is concerning, researchers say, because small differences in blood sugar during pregnancy can have potentially serious consequences for the mother and baby.

But that doesn’t mean women should stop using glucose meters. Instead, they should pay more attention to getting the readings right, Dr. David Sacks, who heads the clinical chemistry lab at the National Institutes of Health, told Reuters Health.

People with diabetes often rely on the devices as convenient indicators of when to take insulin and how much they need.

“They’re very small, you can put them in your pocket,” said Sacks, who was not involved in the new work. “You can prick your finger and get the result immediately. But the meters have never been as accurate as blood that’s drawn in central labs.”

To test just how accurate they really are, Dr. Nimalie Perera, of the Royal Prince Alfred Hospital in Sydney, Australia, compared readings from 102 women to a more precise lab test.

The women all had diabetes - most of them gestational diabetes, which begins during pregnancy - and came to a prenatal clinic for nurses to measure their blood sugar levels. For each patient, nurses drew blood with a finger prick using six different glucose meters.

The most accurate model, Stat-Strip, was off from the lab result by an average of 6 percent, according to findings published in Diabetes Care. The least accurate, Optium Xceed, was off by almost 16 percent. Both are available in the U.S.

The U.S. Food and Drug Administration has to approve all devices that are marketed for medical conditions in the country. Recently, Sacks says, the FDA has been trying to improve standards to increase the accuracy of blood glucose meters.