Carbon monoxide: Detectors save lives - except when they fail
This article was originally published on Oct. 19, 2008.
BULLFROG MARINA - "Can anybody hear me? We need help."
Glenn Howeth's faint voice hissed over the marine band radio. He had chest pains. The seven other people aboard the rented house boat were sick, and some were unconscious.
"There's something going on and we don't know what it is," the 62-year-old Howeth said.
The dispatcher at Lake Powell who heard his June distress call knew immediately. "This is a classic case of carbon monoxide poisoning," the dispatcher told him.
Howeth was pronounced dead hours later, after he suffered a heart attack on the boat.
Never heard during Howeth's call: a carbon monoxide alarm - although one with an 85-decibel siren was fastened inside the compartment where he was pleading for help. It's not known why it and others on the boat were silent.
Years of research has shown alarms can fail to warn people of carbon monoxide, the odorless and tasteless gas emitted by combustion. Research published in 2002 claimed six of 10 brands performed inconsistently. Only three of the brands performed to standards set by Underwriters Laboratories, the certification firm that ensures their reliability.
Research also has found factors such as low winter-time humidity - common in Utah - can impair their sensors. So can other natural gases in the air, or household chemicals.
The 2002 report, by the Gas Research Institute, a federally-chartered research center, also found alarms with digital displays often displayed inaccurate readings.
"They're inexpensive devices. Let's face it," said David Penney, a retired Wayne State University physiology professor who has done extensive research on carbon monoxide. "You're not going to get research-grade devices for $30. It's a compromise, isn't it?"
In Ogden, low-cost detectors were key to helping it become the first Utah city to require alarms in all residences, said Fire Chief Mike Mathieu.
"If these were 50 bucks," Mathieu said, holding a discounted alarm, "I couldn't get it to fly."
Real world vs. lab tests
Trace amounts of carbon monoxide exist almost everywhere. In small doses over short time spans, the gas is harmless, but prolonged or acute exposure can be deadly.
Alarm sensors must distinguish carbon monoxide from other gases and in temperatures that can vary every time the furnace blows or someone takes a shower and increases the humidity in a home.
Commercially available alarms are not designed or tested to detect the gas in real-world conditions, contends long-time critic Paul Clifford, a researcher who worked on the 2002 report and works for a private firm in Newark, Calif.
Clifford maintains UL tests the alarms in an environment that's too controlled - a small laboratory test chamber. He also contends UL "has no track record of actually enforcing compliance," and points to his 2002 research that found some alarms with the UL label did not actually meet UL's standards.
UL says it continually updates its alarm standards and maintains a committee, which includes Clifford, to review them.
Since the 2002 research, UL has mandated alarms must work in lower humidity levels, and required manufacturers to test alarms' effectiveness throughout their guaranteed life span. If an alarm has a digital display, its reading must be within 30 percent of the actual level.
John Drengenberg, UL consumer-affairs manager, acknowledged its testing may not mimic every "real world" condition, but said that's because UL needs a controlled environment to produce reliable data.
"We can't just put a CO alarm in a house and run a test," Drengenberg said.
Kidde and First-Alert, two major manufacturers, point out that their alarms meet UL standards.
Federal regulators also defend the alarms. The U.S. Consumer Product Safety Commission in 2004 published its own report in response to the critical 2002 study.
The government study found 18 of 40 alarms performed appropriately. The 22 others had at least a technical violation of the UL standards or some delay in sounding, but the safety commission emphasized it did not find a widespread risk of harm.
"The key message for us is: We have confidence in today's carbon monoxide alarms," said Scott Wolfson, a spokesman for the commission.
Detectors and dry, cold winters
UL requires alarms to sound when the amount of carbon monoxide in the air would cause the average person to begin feeling symptoms of poisoning.
Alarms do that by monitoring both the amount of the gas and how long it has been in the air. If 70 parts per million of carbon monoxide are present in a home, the typical alarm is supposed to sound within four hours. If 400 parts per million are present, the alarm is supposed to sound in 15 minutes or less.
But Clifford cited Salt Lake City, with its low humidity, as a place where the alarms are likely to fail.
Every alarm technology relies on liquid or water vapor. When humidity is low, components in an alarm can dry out or sensors can underestimate the amount of carbon monoxide present.
Another problem: Some alarms reset themselves before all carbon monoxide leaves the air, Clifford said. Once the gas drops below a certain threshold, the alarm will erase all the information it has collected and begin again at zero parts per million, even though there could be up to 30 parts per million remaining.
If the level of carbon monoxide begins climbing again, the alarm will lag in its calculations, Clifford explained.
That can be important in cold weather, when furnaces turn on and off while keeping a home warm. The carbon monoxide level might decrease when the furnace is off but increase when the furnace is on.
"A person gets poisoned more and more and more because [a person's body] doesn't reset," Clifford said.
An unsafe standard?
Carbon monoxide exposure is measured by the amount of carboxyhemoglobin in a person's blood. The compound is formed when the gas bonds to hemoglobin, a protein in red blood cells that carries oxygen throughout the body.
Doctors agree an average adult starts feeling poisoning symptoms at 10 percent carboxyhemoglobin. But Penney and physician Lindell Weaver, director of hyperbaric medicine at LDS Hospital and Intermountain Medical Center, want the threshold lowered and alarms sounding sooner.
Current thresholds don't safeguard more vulnerable people, such as children, the elderly and pregnant women, Penney said.
Politicking plays a role in creating the standards. In 1994, a winter inversion settled over Chicago, boosted levels of carbon monoxide in the air and triggered thousands of alarms over 24 hours, even though the levels were not life-threatening.
The Consumer Product Safety Commission wanted alarms that were more accurate. But firefighters and the gas industry - both summoned when alarms sound - wanted a threshold that reduced the number of carbon monoxide-related calls.
At present, there is no readily available home or marine alarm designed to sound sooner than the 10-percent threshold. Consumers can purchase hand-held alarms used by firefighters or researchers, but they can run into thousands of dollars.
Weaver recommends buying a detector with a digital display, so even if the alarm does not sound, an owner can see how much carbon monoxide is present.
But possible inconsistencies in the alarms' performance are all the more reason to lower the threshold, Weaver added. "If you buy one as it is now, some people can be exposed to too much carbon monoxide without alarming."
Many manufactures guarantee the alarms for five years and recommend replacing them after seven years. But Trevor Perera, a manager with Gas Controls and Accessories, a certification service in Cleveland that has studied the alarms' performance, suggests replacing them sooner.
Household gases - from cleansers to hair spray to paint fumes - degrade their sensors, he said.
"In real life, those concentrations can be much heavier than what's in the standard," Perera said. "Most people in the industry know that after three years they should get another [alarm]."
For the cost, 'these have served us'
Separate research has shown the alarms can fail to alert boaters to carbon monoxide from engines and generators. It's not known why the hard-wired alarms aboard the Howeth family's boat did not sound the night Howeth died. The Park Service is considering spending thousands of dollars to have them tested.
One alarm had been periodically sounding earlier but the family could not find a problem and eventually someone removed its batteries. Park rangers found it on the roof.
Three other alarms, manufactured by Safe-T-Alert, were hard-wired in the cabin. In a preliminary report, the Park Service said it did not know how carbon monoxide entered the cabin. But once the gas was inside, the boat's recirculating air conditioning system likely spread the gas throughout the cabin.
A year earlier, Safe-T-Alert had posted a notice on its Web site warning that the model found on the boat might not sound at lower, but still dangerous, concentrations. It offered a free replacement.
David Buddingh, a spokesman for Safe-T-Alert's parent company, said the alarms "work fine but they don't meet the UL standard."
Despite challenges to the alarms' performance, most everyone agrees it is better to have one.
Wolfson pointed to Texas, where, in the wake of Hurricane Ike, people have died from the carbon monoxide produced by generators. On Tuesday night in Ogden, five workers were treated after they were poisoned at a cold storage business. It did not have carbon monoxide alarms.
"The No. 1 goal right now is to get people to install CO alarms in their house at the same rate of success we've had with smoke alarms," Wolfson said.
Mathieu said Ogden's requirement to have the detectors, adopted by the city council last year, has already helped prevent injuries or deaths. Utah building codes require alarms in new housing and some public buildings, but Mathieu worried for years about carbon monoxide levels in Ogden's older housing. "For the cost and the benefit, these have served us very well," Mathieu said.