One of the most common questions I get asked by parents is, “Don’t the tonsils and adenoids help to fight infections?” My usual response is that yes, they are part of the immune system, but only a small part. There are thousands of lymphoid tissues and glands spread throughout the body. However, if large tonsils are causing you to stop breathing 10 to 15 times every hour with your oxygen levels dropping to 75%, then taking them out is much more beneficial than any theoretical risk of future infections.
To better answer this question, researchers at the University of Melbourne combed through almost 1.2 million charts of children who were born in Denmark between 1970 and 1999. About 17,000 underwent adenoidectomy, 12,000 underwent tonsillectomy, and 31,000 underwent both procedures. What they found was that undergoing tonsillectomy was associated with almost 3 times the risk of upper respiratory tract diseases, compared with controls. This included respiratory, infectious and allergic disease. Adenoidectomy raised the risk to about 2 times normal. Infections were found to be 17% higher overall.
- Confounding. This occurs when a variable is related to both the predictor (surgery) and the outcome (respiratory problems), causing what seems to be an association. For example, they did not control for smoking or use of antibiotics, both of which can potentially increase rates of respiratory problem or infections.
- Reverse causation. This occurs when B follows A, so it’s assumed that A causes B, when in in fact, B causes A. The example he gives is that children who undergo adenoidectomy are more likely to undergo ear tube placement, but other conditions, such as allergies or infections can lead to more adenoidectomies and ear tube placement.
- Selection bias. This occurs when the group that receives therapy differs from the control patients. For example, there was no mention of why tonsil or adenoid surgery was performed. There was also no mention of how they chose the control group other than that their “health did not differ.”
- Measurement bias. How you measure diseases or outcomes can lead to errors, especially if you rely on diagnostic codes from a database.
My biggest problem with this study is related to #3, selection bias. In their methods section, they stated that each of the 3 surgeries was compared with controls (no surgery), after ensuring they were otherwise of comparable health. They did not go any more in detail regarding what “comparable health” means. Ideally, if they could pull out subjects with sleep apnea only, they could have followed those who underwent surgery vs. those that did not. Unfortunately, this is not possible with a database like this (#4: Measurement bias). Based on what they describe about controls, it’s likely that they were controls because they were not prone to get tonsil infections or develop sleep apnea. This is related to #2, reverse causation: An original unknown condition may lead to both tonsillectomy and respiratory conditions and infections.
This unknown condition may be obstructive sleep apnea, which has known associations with increased risk of respiratory infections, cancer and heart disease. Although tonsillectomy does a good job of treating OSA in children, it’s only fully effective in about 2/3 of the time.
Although this study may completely contradict what I’ve said in the past, upon further inspection, you have to take the results with a grain of salt. This brings to mind what every scientific authority says about any landmark research study that’s publicized in the media: “association does not imply causation.”