After three airplanes exploded over the last 11 years, the Federal Aviation Administration isolated the problem, and settled on a budding solution. It is believed the blasts were a consequence of a combustible mix of fuel vapors that can fill gas tanks on hot days. To the FAA’s critics, the agency was slow to uncover the culprit. To practical minds, the FAA was dealing with a remote problem and a rare effect.

The agency’s solution is retrofitting airplanes with onboard “inerting systems” that pump nitrogen into fuel tanks. It lowers the amount of oxygen in fuel vapors and reduces their flammability. The new systems, at a quarter of a million each, cost a lot of money and fall short of guaranteeing more airplanes won’t explode. But when it comes to air travel, this additional safety margin may be the best we can do, because we already made most of the easy fixes.

As the industry’s safety record grows more superlative, ending infrequent crashes becomes both more difficult, and expensive.

The same is true of prescription drugs, which like airlines, is one of the most regulated industries in the country. Over the last 50 years, we have added successive layers of testing and monitoring before new drugs are approved for sale to patients, to the point where the average development time for a new drug can span 10 years and cost almost $1 billion.

The result is that today we have the safest system in the world, but few glaring gaps to easily improve on. When it comes to making new drugs safer, most of the obvious solutions are already accounted for and we have reached the flat part of a curve that measures incremental safety against the additional cost. We can make our drug development system a little safer, but only at a very big cost.

This trade-off is at issue, after the pain medications Vioxx and Celebrex, known as Cox-2 inhibitors, were traced to small but higher risks of heart attacks among patients who use them. Drugs like Vioxx are called Cox-2 inhibitors because they selectively inhibit pathways that cause pain but, unlike similar painkillers like Advil, the Cox-2s do not block pathways in the body that also produce chemicals that protect the tummy lining.

Based on these concerns, the drug company Merck pulled its drug Vioxx from the market. The maker of Celebrex has taken a more cautious approach while it investigates the potential link, suspending promotion of its drug while investing in studies to separate fact from theory.

That approach may have been vindicated on Monday, when the FDA announced that the agency also discovered a link between heart attacks and the common over-the-counter drug Aleve, an older close cousin of the Cox 2s, raising the specter that the higher heart-attack risk may extend to this entire class of painkillers. The side effect may eventually emerge as a fact of life when it comes to these painkillers, and another trade-off that doctors and patients need to weigh.

Consumers are angry that these problems were not unearthed earlier. But the higher risk of heart attacks caused by Vioxx, for example, was on the order of about six or seven heart attacks for every 1,000 patients who took the drug. In an older-patient population that al ready suffered more heart attacks, such a risk could have been easily missed, even with a clinical trial that included 10,000 patients or more. With this additional testing, the benefits of an off chance of discovering a rare side effect before a new drug is approved is eventually outweighed by the cost of keeping promising drugs from patients.

Even delaying seemingly ordinary drugs can have dramatic consequence on the public health. The first non-sedating anti-allergy medicine, Claritin, took almost seven years to get approved, while sleepy drivers with sniffles continued to cause car accidents. Each of the popular anti-cholesterol drugs known as statins that today prevent 15% to 30% of heart attacks took several years to get approved. How many people died waiting? The math is straightforward.

Or consider this math: It’s estimated more than 20,000 people died between 1985 and 1987 waiting for streptokinase, the first drug that could be intravenously administered to reopen the blocked coronary arteries of heart attack victims. Between 1988 and 1992, about 3,500 kidney cancer patients died waiting for Interleukin-2, which was available in several European countries. In 1988 alone, it is estimated between 7,500 and 15,000 people died from gastric ulcers caused by aspirin and other non-steroidal anti-inflammatory drugs, waiting for the FDA to approve misoprostol, which was already available in 43 countries.

This is the price of too much caution, when drugs are held hostage to safety testing with lethal consequences. The poisonous compromise is becoming the prevailing orthodoxy inside the FDA right now, as the agency understandably withdraws under the withering attacks from a sound-bite-obsessed Congress and headline-hungry press.

The effects of this deadly atmosphere can already be seen in the FDA’s decision two weeks ago to delay approval of a promising cancer drug, Marqibo, for the treatment of aggressive non-Hodgkin’s lymphoma. This go-slow approach was also apparent in the lopsided vote to withhold approval of the testosterone patch for women, being touted as the “Viagra for women.” That decision was driven by theoretical fears of certain risks rather than problems already manifest.

Some have estimated that pumping nitrogen into gas tanks will save about one plane from crashing over the next 25 years. To many people, this is a fair price to pay, especially if you happen to be on that one plane when it is ready to blow. To others, it is a big price tag to stamp out a remote risk. When it comes to new drugs, the returns on our newfound caution may be even less striking.

Dr. Gottlieb is a practicing physician and resident fellow at the American Enterprise Institute. Between 2003 and 2004, he was a senior official at the Food and Drug Administration.