Almost a third of car booster seats tested do a poor job of fitting children to their seat belts and should be redesigned, according to a review by the U.S. insurance industry and researchers.

The Virginia-based Insurance Institute for Highway Safety and the University of Michigan Transportation Research Institute found Wednesday that 13 of 41 booster seats left children vulnerable to soft tissue injury because the seat belts crossed the abdomen instead of the pelvic bone.

Those boosters "may increase restraint use by making children more comfortable, but they don't position belts for optimal protection," said IIHS President Adrian Lund.

The institute rated 41 seats and found they could not recommend 13. Ten were rated as "best bets," and five were rated as "good bets."

Higher price did not correlate to better protection.

"Boosters that provide better belt fit aren't necessarily the priciest," said Anne McCartt, Institute senior vice president for research. "Parents don't have to spend a lot of money for a best bet or good bet booster."

The biggest issue for booster seats is where the lap belt crosses. A correctly positioned lap belt loads pelvic bones during a crash, not the abdomen. All the top rated boosters located this belt on children's upper thighs.

A good booster also positions the shoulder belt at mid-shoulder, keeping the webbing away from the neck so it won't chafe. That reduces the likelihood that kids will endanger themselves by putting the belt behind their back or under an arm, said lead author, Matt Reed, of the University of Michigan Transportation Research Institute.

"Our data show it's possible to design a booster with good lap and shoulder belt fit," Reed said in a statement. "Boosters that can't do that should be redesigned."

Companies reacted by defending their products, saying they do extensive testing that meet or exceeds safety standards.

Montreal-based Dorel Juvenile Group said it "welcomes the opportunity to review the evaluation conducted by the IIHS."

However, Evenflo, based in Ohio, called the study "misleading as it fails to consider the real world use and performance of the seats tested."

Evenflo has sold more than four million units of the seats cited by the Institute and "we are not aware of a single incident in which a child was injured as a result of an improper fit," said a spokesman.

This was the first time the Institute issued evaluations for booster seats.

Booster seats are recommended for children over 40 pounds until they are eight years old or four feet nine inches tall. The seat belt should go across a child's lower hips and mid-shoulders instead of the abdomen because the liver and spleen are more vulnerable to injuries.

With files from the Associated Press