Tonsillectomy remains a common surgical procedure with over half a million cases in the US per year, the most common surgical procedure in children. The indications and effects of tonsillectomy remain a matter of research and debate, as is appropriate. It is also a subject of popular misinformation and alarmism.

A recent article by Seth Roberts raises many of the issues with tonsillectomy, but also reveals the pitfalls of non-experts trying to understand the clinical literature and the effects of bias on evaluating a complex medical question. Throughout the article Roberts displays a persistent bias toward downplaying the benefits and exaggerating the risks of tonsillectomy, while accusing the medical establishment of doing the exact opposite.  The purpose of this post is not to defend the practice of tonsillectomy but to review some of the relevant issues and explore how bias can affect an assessment of the evidence.

Indications for Tonsillectomy

Roberts tells the story of Rachael who was offered tonsillectomy for her son and so did some research on her own. She looked on Pubmed (a good place to start) and found a Cochrane review from 2009.

The Cochrane Review that Rachael found (“Tonsillectomy or adeno-tonsillectomy versus non-surgical treatment for chronic/recurrent acute tonsillitis”) was published in 2009. It describes four experiments that compared tonsillectomy to the care a sick child would otherwise receive. All four involved children like Rachael’s son, and all four had similar results: Tonsillectomies had only a small benefit. (Contrary to what Rachael was told.) During the year after random assignment to treatment — the point at which some children had their tonsils removed, other children did not — children whose tonsils were removed had one less sore throat than children who were not operated on (two instead of three for children like Rachael’s son).

Roberts fails to mention that the benefit described above – one fewer sore throat – was for the mild group only. For those who had a more severe sore throat history, the review found greater benefit:

For more severely affected children adeno-/tonsillectomy will avoid three unpredictable episodes of any type of sore throat, including one episode of moderate or severe sore throat in the next year.

In addition there were fewer sore throat days, 17 with tonsillectomy compared to 22 without, even including the sore throat days from the surgery. One of the main points of Roberts’ article is to criticize the reviewers for omitting information from their review, but while discussing that very point he is omitting information stated plainly in the article abstract.

I find this to be a common source of bias in evaluating medical interventions – conflating the risk/benefit in low risk or severity groups with higher severity groups. Also, Roberts commits other common non-expert fallacies in evaluating the potential risk vs benefit of a therapy, narrowing the field of indications and potential benefits. In his article Roberts assumes that the only indication as a preventive measure for tonsillectomy is for recurrent sore throats. This is certainly a common indication, but not the only one. Tonsillectomy is also performed for acute abscess in the tonsils.

Further, there are complications from tonsillitis that go beyond merely have frequent sore throats, including dysphagia (difficulty swallowing) and sleep apnea (airway obstruction during sleep). In fact these are major considerations in deciding who should get a tonsillectomy.

Published guidelines reflect these additional factors that Roberts ignored – only doing tonsillectomy on severe cases and considering other issues like dysphagia and sleep apnea, and recommending medical management and watchful waiting in less severe cases. Roberts accuses physicians of ignoring relevant information about tonsillectomy, but he ignores published practice guidelines that look much more deeply into the issue than is reflected in his analysis.

Risks of Tonsillectomy

After downplaying the benefits and oversimplifying the indications for tonsillectomy, Roberts then goes on to exaggerate the risks. He seems to take the approach of listing any possible hypothesized risk as if it is established. The links he uses to defend each risk he cites does not support the claims he is making. Once again he is led to the conclusion that doctors are ignoring the risks and morbidity from tonsillectomy, while those alleged risks have not been established.

For example, he lists Hodgkins disease with links to evidence for an association with tonsillectomy. He does link to one article from 1972 and disputes the association, but did not link to a 1987 review that found no association between Hodgkins disease and tonsillectomy. As far as I can see this was the last word on the issue. Roberts still gets to list Hodgkins disease as a scary increased risk from tonsillectomy without fairly representing the state of the evidence.

He also lists variant Creutzfeld-Jacob disease, and links to two articles which do not establish a higher risk of developing the disease from tonsillectomy. Rather, the issue is about whether prion disease has an affinity for the tonsils, and whether tonsillar biopsy can be used for diagnosis. The correlation is about the natural history of prion disease, not a cause and effect from having a tonsillectomy. A deeper reading into the literature is needed in order to see this. Roberts, however, simply searched for tonsillectomy and complications and listed everything he found. In none of the cases is a cause and effect established. He wonders why doctors do not list all of the complications he found – that’s why.

The underlying issue is the effect of tonsillectomy on the immune system. Roberts engages in very simplistic reasoning – the tonsils are part of the immune system, removing them therefore compromises immune function and is a bad idea. He ignores the fact that those tonsils that are removed are unhealthy, and perhaps they have become counterproductive to immune function. From the practice guidelines linked to above here is a summary of the current thinking:

With chronic or recurrent tonsillitis, the controlled process of antigen transport and presentation is altered due to shedding of the M cells from the tonsil epithelium. The direct influx of antigens disproportionately expands the population of mature B-cell clones and, as a result, fewer early memory B cells go on to become J-chain–positive IgA immunocytes. In addition, the tonsillar lymphocytes can become so overwhelmed with persistent antigenic stimulation that they may be unable to respond to other antigens. Once this immunological impairment occurs, the tonsil is no longer able to function adequately in local protection, nor can it appropriately reinforce the secretory immune system of the upper respiratory tract. There would therefore appear to be a therapeutic advantage to removing recurrently or chronically diseased tonsils. On the other hand, some studies demonstrate minor alterations of Ig concentrations in the serum and adjacent tissues following tonsillectomy. Nevertheless, there are no studies to date that demonstrate a significant clinical impact of tonsillectomy on the immune system.

What are the alternatives?

The other display of flagrant bias in the article, reflecting, in my opinion, the successful marketing of “alternative” medical thinking, is Roberts’ endorsement of the naturopathic approach to chronic sore throats. While decrying the lack of evidence-based practice on the part of medical doctors (while not understanding or properly reflecting the evidence), he then offers as an alternative that is blatantly not evidence-based. The naturopath, of course, offered nutritional and herbal treatments. Roberts is correct only in that such interventions are lower risk and less permanent than a surgical procedure, so the threshold of evidence efficacy can be lower, but that does not justify making non-evidence-based recommendations.

He mentions vitamin D. Here the evidence is preliminary and mixed, looking for a correlation with vitamin D levels and chronic sore throats. One study found:

There is no difference between the serum vitamin D level and receptor gene polymorphism among children with recurrent tonsillitis and healthy children. But vitamin D insufficiency is more prevalent in children with recurrent tonsillitis group (18%).

That’s pretty thin. Vitamin C was also raised. Roberts blames Linus Pauling for giving vitamin C a stigma among scientists. This may be true, but it’s quite irrelevant. There have been many studies of vitamin C and infections, and the bottom line is that there is no proven benefit. The same is true for multivitamins.

The naturopath also offered powdered larch bark. Here is a summary of the current evidence there:

Larch arabinogalactan is approved by the US Food and Drug Administration (FDA) as a food additive and fiber supplement. However, available scientific evidence does not support claims that larch bark is effective in treating cancer or any other disease in humans. Early laboratory evidence suggested that larch arabinogalactan may stimulate the immune system. However, a more recent study in mice contradicts this finding. Further studies are needed to identify other uses for larch in humans.


The point of this post is not to thoroughly review the research and make recommendations regarding tonsillectomy. That is not my area of expertise, and panels of experts have already done that. The point, rather, was to use this question, and Roberts’ article, as an example of how bias and lack of expertise affects how we view the evidence. Several common errors in looking at clinical questions were made here, and I find frequently – taking a simplistic approach to disease and treatment mechanism, failing to consider severity as an indication for a treatment, failing to consider all relevant outcomes when assessing potential benefit, and using different criteria of evidence when assessing alternatives.

I do think that health consumers should avail themselves of information that is available to the public, but it is not easy to do so. Pubmed is a good place to start, but for the non-expert I would also recommend finding practice guidelines or summaries of the evidence prepared for the non-expert. It is also a good idea to use the information found as a basis for a conversation with a real expert, rather than as a means of replacing expert opinion with a “Google University” opinion.


Posted by Steven Novella

Founder and currently Executive Editor of Science-Based Medicine Steven Novella, MD is an academic clinical neurologist at the Yale University School of Medicine. He is also the president and co-founder of the New England Skeptical Society, the host and producer of the popular weekly science podcast, The Skeptics’ Guide to the Universe, and the author of the NeuroLogicaBlog, a daily blog that covers news and issues in neuroscience, but also general science, scientific skepticism, philosophy of science, critical thinking, and the intersection of science with the media and society. Dr. Novella also contributes every Sunday to The Rogues Gallery, the official blog of the SGU.