Junkfood Science: Suffer the little children: null link between tylenol and asthma

September 22, 2008

Suffer the little children: null link between tylenol and asthma

When little ones hurt, have a high fever and are crying and too miserable to rest or eat, their parents have been able to help them feel better with a few drops of tylenol or panadol. It’s the most-used analgesic around the world and used by parents for more than 50 years. This weekend, parents were frightened by 442 worldwide news reports of a study on more than 205,000 children just published in Lancet, claiming to find a link between acetaminophen use during infancy and asthma in childhood.

All the information was there to recognize that this study actually found no cause for concern, but how many parents understood that? How many babies will be left without pain relief, or worse, how many parents may go back to baby aspirin (putting their baby at risk for Reye’s syndrome) because, sadly, they got caught up in fear?

Per readers’ requests, let’s take a brief look at this study.

See those baloney words: linked to, associated with, correlated to...

Bottom line, this was a data dredge through a massive database of the International Study of Asthma and Allergies in Childhood (ISAAC) looking for correlations. In fact, the huge numbers of children included may have sounded impressive, but was our first clue that this study was most likely done in a computer rather than to have been a clinical trial on real children. Our second clue was the title of the paper, which reported on the association between paracetamol (known as acetaminophen in the United States) use in infancy and the risk of asthma...

These are our quickest and easiest baloney alerts that this report doesn’t warrant any reaction on our part. As we come to understand correlations and what risk factors mean, this study needs no more than a glance before going to line the bird cage.

These epidemiological studies were never meant to be used for anything more than the most preliminary-type of exploration for researchers to search for strong links among data that can be used to begin to form hypotheses that can be later tested in the laboratory and eventually in human clinical interventional trials. As we’ve seen again and again, no matter how “significant” any correlation, doesn’t mean it means anything or will ever hold up to later actual research that’s a fair test of a hypothesis. The only purpose served by publicizing the bazillions of — often meaningless, nonsensical and contradictory — correlations these data dredges scour up, is to provide daily fodder for media sensationalism and special interest groups. Sadly, the public is left to suffer from epidemiological whiplash, as one day’s scare is likely to be reported as good for us the next day, and bad for us again the day after.

Imagine how much less stressed and healthier we’d all feel if the media only reported actual clinical research that had scientific merit and was news we could really use. Since that’s not likely to happen, let’s look more closely at this study because its weaknesses offer additional examples of why this wasn’t a fair test of anything.

Overview of methodology

ISAAC is a database created in 1991, when Phase One began by sending out questionnaires to the parents (or primary caregivers) of 721,601 children in 56 countries. Questionnaires were translated into the local language with back-translated into English. This latest study published in Lancet was led by Richard Beasley, DSc, at the Medical Research Institute of New Zealand, Wellington, on behalf of the ISAAC Phase Three Study Group. It used data from Phase Three of the ISAAC project, which had sent out additional questionnaires asking parents (or primary caregivers) of 6-7 year olds and 13-14 year olds to think back to their child’s first year of life and symptoms of asthma, rhinoconjunctivitis (rhinitis with watery, itchy eyes) and eczema.

The questionnaires also asked the parents about 28 select environmental factors that the ISAAC authors wanted to examine as possible risk factors for asthma and allergies. As the authors explained: “Questions were about age, sex, family size, birth order, antibiotic use in the first year of life, breastfeeding, birthweight, diet, heating and cooking fuels, exercise, pets, socioeconomic status, immigration status, parental tobacco smoke, traffic pollution, and paracetamol use in the first year of life and in the past 12 months of children aged 6–7 years.”

Professor Beasley and colleagues said that they “used parent-reported symptoms [of asthma] rather than doctors’ diagnose [sic] to avoid major diagnostic differences related to access to medical care, language, and medical practice in populations worldwide.”

Already, the quality of the data going in is problematic. Not only was it restricted just to the information the ISAAC authors chose to examine, but it was retrospective and self-reported, subjective information. There was no attempt to validate the information with clinical examinations or examinations of the children’s medical records. Nor were any dosages of the medication recorded or included in the analysis. This is the very weakest of information and is at the greatest risk for recall bias.

As even Dr. R. Graham Barr, M.D., DrPH, assistant professor of epidemiology at Columbia University Mailman School of Public Health in New York wrote in a commentary in the same issue of Lancet:

[A] cross-sectional survey with a retrospectively ascertained primary exposure is not a design on which we prefer to make therapeutic decisions. Recall bias (parents of children with asthma might better remember giving paracetamol in the first year of life) and reporting bias (parents more attuned to their children’s maladies might be more likely to give paracetamol and report the current wheeze) could account for the findings.

Nearly every child on the planet has received this analgesic. As the ISAAC authors also commented in their Discussion Section, fever is common in infants, so most babies would likely have received paracetamol for fever on more than one occasion during their first year of life. But since respiratory illnesses during early childhood (especially respiratory syncytial virus infection) is associated with increased problems with wheezing in children, these infants would also have been more likely to have received paracetamol for their accompanying fever and discomfort than typical little ones.

In other words, respiratory infections and other illnesses accompany the use of acetaminophen in babies. Had the researchers asked about other comfort measures, such as use of humidifers, favorite cuddly toys, or popsicles, the correlations to asthma might have been every bit as significant. Repeated respiratory problems, however, have considerably more biological plausibility for a potentially meaningful link. The tylenol connection just came along for the ride.

As Dr. Barr wrote, underlying respiratory disease, differences in hygiene and the use of other fever medications could also explain the findings.

Epidemiological studies have linked asthma to hundreds of things, but this type of data is not very reliable Dr. Richard Lockey, M.D., professor of Medicine, Pediatrics and Public Health and director of the Division of Allergy and Immunology at the University of South Florida College of Medicine, told ABC News. Dr. Lockey is a past President of the American Academy of Asthma, Allergy and Immunology and former director of the Board of Allergy and Immunology. He said he seriously doubted there is anything to this link.

When enough data is thrown into a computer, odds are it will pull out all sorts of meaningless correlations, just like a popsicle. But no one would try to claim a popsicle causes asthma. Yet, like all such databases, this one has been the source of countless other links with asthma, many of which contradict each other or make no sense at all. Authors from New Zealand, for example, dredged ISAAC and reported children who ate a hamburger more than once a week had a 65% odds ratio of having a history of wheeze and if it was from a fast food restaurant, the odds went up to 141%. But no one would really believe that a hamburger causes asthma. Turkish authors used the database to report a 154% higher odds ratio of asthma among those whose family incomes were below $300/month; whereas Brazilian authors used it to report higher risks for asthma associated with wealthier families.

This weekend’s Lancet study, also suffered from exceedingly high attrition rates. Of the original cohort, 89,514 babies (46%) who’d received tylenol for fever during 1st year of life were lost in follow-up and their parents didn’t complete the questionnaire. Only data on about half (105,041) of the original group of babies who’d received acetaminophen at least once during infancy were included in the multivariate analysis. Were the parents of children with asthma more likely to have completed the questionnaires, seeking answers to their children’s illnesses?

The ISAAC researchers then loaded all this questionnaire data into a computer and applied several different computer models to identify links, calculate odds ratios and adjust for specific confounders. They adjusted for these covariates: maternal education, antibiotic use in the first year of life, ever breastfed, parental smoking, current diet and siblings.

As they stated: “The primary outcome measure was the association between paracetamol use for fever in the first year of life and asthma symptoms at 6–7 years of age, expressed as odds ratios, as measured by the multivariate analysis.”


From their multivariate analysis, they reported a 46% odds ratio (OR=1.46) associated with acetaminophen use in the first year of life and risks of asthma symptoms at age 6-7 years.

Of course, this is an untenable correlation, made even moreso by the use of odds ratios rather than relative risks. This computer-derived correlation, like others under at least 200%, is no better than what might have surfaced for this type of study by a random error or chance, and most likely explained by co-factors. The relative risks derived from epidemiological studies that later prove out in clinical trials to be real, are considerably higher — by several hundred percent! Forty-six percent may sound like a lot, but it’s still a null finding. It’s even less than a hamburger.:-)

In other words, there was no valid link found between acetaminophen usage in infancy and childhood asthma. Without even a tenable link, it’s not a viable avenue to look for a cause, of course.

Professor Beasley and colleagues admitted that their odds ratio “may be overestimates of the risk” because of confounding factors likely to have been present. The authors didn’t report the odds ratios from any of the other 28 questions in the questionnaire, nor did they provide any of the actual (absolute) numbers of children with asthma, to enable us to put these odds into any credible perspective. [Providing only odds ratios, instead of actual numbers, is not valid medical reporting. If 1 child among 1,000 who'd not been given tylenol developed asthma compared to 2 children among 1,000 who'd received tylenol, the risk is double, but clinically meaningless.] By comparison, authors at the Isra University Hospital in Hyderabad reported in the July issue of the Journal of the College of Physicians and Surgeons Pakistan an association between asthmatic children and a parental history of asthma, with an odds ratio of 26.8 — 1,800% the risks associated with acetaminophen in this week's study.

BTW: The only clinical research the ISAAC authors referenced as support for a potential medical link between tylenol use and asthma was found: “in one randomized controlled trial, paracetamol use for fever in childhood was associated with an increased risk of hospital outpatient attendance for asthma when compared with ibuprofen.” This study had been conducted by doctors from the Slone Epidemiological Unit School of Public Health, Boston University School of Medicine and published in the February 2002 issue of Pediatrics. It was a randomized, double-blind trial where 1,879 asthmatic children were prescribed either low doses of acetaminophen or two different doses of ibuprofen for fever control as needed. The children’s outpatient medical visits and hospitalizations over the next four weeks were obtained from parental questionnaires. A total of 18 children had been hospitalized for asthma during those four weeks, two more in the acetaminophen group compared to the higher ibuprophen dose group, but “there were too few hospitalizations to permit computation of stable dose specific risk estimates.”

The ISAAC authors said “causality cannot be established from a study with this design” and the “evidence is insufficient to advise parents and healthcare workers of the risk-benefit of taking paracetamol in childhood.”

Aubrey Grayson, Medical writer for ABC News, reported on doctors issuing much stronger statements about this ISAAC study, telling parents to not worry or react to it. Dr. Anita Gewurz, M.D., professor of allergy and immunology at Rush University Medical Center, said the study contains serious methodological flaws and the findings should be taken with a grain of salt. Several doctors urged parents not to deny their children relief because of scary news reports:

"Don't panic or go back to aspirin," said Dr. N. Franklin Adkinson, Jr., professor in the division of allergy and clinical immunology at the Johns Hopkins Asthma and Allergy Center in Baltimore. "Continue to use acetaminophen until further research is done."

Parents should also remember that other painkillers are not without their risks. In 1982, the government issued a warning to avoid giving young children aspirin to relieve cold and flu pain. Aspirin use in young children has been linked to the development of Reye's syndrome — a rare but serious children's disease that can lead to brain damage, liver failure and death. Acetaminophen has never been linked with the development of Reye's syndrome, and doctors have since urged parents to choose acetaminophen-based pain relievers to give to their feverish children.

But some doctors worry that many parents may read about the new findings and begin denying their sick children any form of pain relievers. Even worse, parents might begin to choose aspirin over acetaminophen once again -- thus possibly placing their children at risk for Reye's.

"Misrepresenting this will cause unnecessary panic," said Dr. Peter Catalano, chairman of the department of otolaryngology at the Lahey Clinic in Burlington, Mass. "The science is absent."

© 2008 Sandy Szwarc

More information

Tylenol has long been the standard remedy for fever and pain in children and to be safe and effective when used as directed. However, like everything, overdoses can be harmful and tylenol overdoses can result in serious liver damage. More information on how much tylenol to give your child and other precautions is available at the Tylenol Dosage Calculator.

More information on paracetamol is available here, and on acetaminophen here.

Study Disclosures:

Funding The BUPA Foundation, the Health Research Council of New Zealand, the Asthma and Respiratory Foundation of New Zealand, the Hawke’s Bay Medical Research Foundation, the Waikato Medical Research Foundation, Glaxo Wellcome New Zealand, the New Zealand Lottery Board, Astra Zeneca New Zealand, and Glaxo Wellcome International Medical Affairs. Richard Beasley received honoraria for lectures and participation in advisory boards, and grant support from GlaxoSmithKline, the manufacturer of paracetamol. All other authors declare that they have no conflict of interest.

Bookmark and Share