One paradox of modern health care is as disturbing as it is baffling: Even though almost every 20th-century law aimed at improving drug safety was enacted in response to a pediatric drug disaster, the systems we put in place to improve safety ultimately proved problematic.

Cynthia A. Connolly, who has spent decades as a pediatric nurse and is an associate professor at the University of Pennsylvania School of Nursing, has been trying to understand why.

Recently, she received a fellowship from the National Endowment for the Humanities to finish her book, Children, Drug Therapy, and Pharmaceuticals in the United States, 1906-1979.

Connolly, one of four faculty directors at Penn's Field Center for Children's Policy, Practice, and Research, and fellow at Penn Nursing's Barbara Bates Center for the Study of the History of Nursing, spoke to us recently about her work.

What prompted your interest in this topic?
I was a legislative fellow on Capitol Hill in 2002. At that point, I had been a pediatric nurse for more than 20 years. I had recently finished my Ph.D. I thought I knew something about children. I thought I knew something about history.

I was asked to sit in on a hearing for what became the Best Pharmaceuticals for Children Act. It was introduced because many of the drugs that were used for children had not been tested for safety and efficacy in children, only adults. I was shocked. I had given thousands of medications to thousands of children. I wondered: How did this happen?

Tell us about some of the disasters involving children that led to new laws.
The FDA was founded in 1906. A major argument behind its creation was to regulate the sale of dangerous products. One, in particular, was opium-laced "soothing syrups" that had been on the market for more than a century. They were advertised for everything from teething to colic, even for behavioral issues. If you were a poor woman and needed to leave your child in the care of her 6-year-old sister so you could go to work in one of Philadelphia's factories, the opium would keep the baby calm. But many babies died.

The next big change came in 1938, in response to one of the first sulfonamide drugs. It was laced with sweet-smelling and -tasting diethylene glycol, a form of antifreeze, so it would appeal to children's palates. Again, many died.

The third law came after thalidomide. Widely prescribed to pregnant women in other countries because of its antinausea properties, the drug caused hideous bone and limb deformities in their babies. Few American women took the drug because a physician at the FDA refused to approve it, despite enormous pressure from her superiors. But some women did get it. Incredibly, until the 1962 law enacted in the wake of thalidomide was passed, a drug company could test a drug without going to the FDA for permission.

In a nutshell, the 1906 law said you had to have accurate labeling; anything could be in a drug so long as you labeled it accurately. The 1938 law stipulated that the drug also had to be safe, and the 1962 legislation said that a product also needed to be efficacious, meaning that it did what its manufacturer claimed, based on rigorous testing.

Why didn't this solve the problems?
Multiple reasons. First, in the wake of the 1962 law, drug development got a lot more expensive. And there was no specific mandate that a new drug be tested in every population in which it might be used. So companies had an economic incentive to test drugs only in adults. Pediatricians would then use their clinical knowledge to adapt dosing for children.

This occurred as scientific understanding of how children differ from adults - cognitively, emotionally, physiologically - was increasing. Pediatric clinicians had always known children were not just small adults. But now they had the evidence to better understand that, for example, toddlers' incredibly high rates of metabolism meant that for some drugs, they needed a higher dosage per pound than did babies, older children, or adults. Drug dosing in pediatrics was not just as simple as cutting an adult dosage in half or quarters.

At the same time, new federal rules mandating informed consent were more complicated for children. An adult can volunteer himself or herself. But in the 1960s and 1970s, there were major debates about who should decide for children. Parents? Pediatricians? Finally, even if parents were allowed to volunteer their children, who should pay for the more complicated and expensive testing? Insurers? Drug companies? Hospitals? Parents? Medical schools? Within a few years, children became what pediatrician and pharmacist Harry Shirkey called "therapeutic orphans." Drugs had received FDA approval, but doses in children were worked out in hospitals, private practices, and clinics.

Didn't anyone care?
It wasn't that people weren't thinking about children or didn't care about them. Everyone wanted better drug safety, but those important and positive changes to American drug law did not have child-specific language. This is one example of how we need more careful attention to children's needs across the board in terms of laws and policies. We have that in some places - juvenile justice, for example. But we need it everywhere.

Give us a teaser - one interesting moment in history with regard to children's drugs and safety.
Here is one that really sticks with me: During World War II, researchers were trying to gather data on penicillin - which bacteria it killed, in what doses, etc. The drug was critically important to the war effort. Thousands of men were dying from wound-related infections.

But the drug was incredibly scarce; every drop was precious. So it was not generally available to civilians. When a civilian did receive it, it was because he or she contracted an infection similar to those in soldiers in the field.

A lot of those civilians were infants. When I first read old newspaper articles about planes rushing penicillin to sick children, I didn't bat an eye. Everyone is worried about a baby with an infection, and you'd do what you could.

But that wasn't the rationale. It was because you needed much less of the drug for a baby. You could test it in 10 babies with pneumonia, vs. one adult. Of course, that reason, if we used it today, sounds horrible.