On 2017 Aug 29, Andreas Lundh commented:

Comment on Association of streptococcal throat infection with mental disorders

A recent Danish register-based cohort study(1) concludes that “individuals with a streptococcal throat infection had elevated risks of mental disorders, particularly OCD and tic disorders.” However, some methodological issues need consideration.

Firstly, the choice of exposure has a risk of misclassification. Exposure (i.e. Group A Streptococcal (GAS) throat infection) was the combination of a rapid antigen test performed by the patient’s general practitioner and subsequent antibiotic prescription. In Denmark use of rapid antigen test and antibiotic prescription are guided by modified Centor criteria(2,3) where symptomatic patients at high risk of GAS throat infection are not tested, but instead receive empirical antibiotic treatment. This leads to patients being misclassified as unexposed since they are never tested.

Secondly, the choice of outcome has a similar risk of misclassification. Psychiatric diagnoses are identified from national databases that only contain hospital information. Psychiatric patients that are not treated in hospitals (e.g. treated by primary care psychiatrists) are misclassified as not having had the outcome. Misclassification seems likely as only 0.1% and 0.2%, respectively, had a diagnosis of OCD or tics in the study period.

Thirdly, the analytical strategy has a risk of bias. The authors compared patients that had received both a rapid antigen test and antibiotics with a group that was never tested. GAS throat infection will in most cases resolve spontaneously and many patients will never contact their general practitioner for testing. The group tested therefore likely differs from the group not being tested and represents a group with certain healthcare seeking behavior. This is substantiated by the findings that risk of mental disorders seems to increase with number of tests and regardless of whether the tests are negative or positive. A more reasonable analysis that avoids confounding by test indication would be to compare the group of tested patients prescribed antibiotics with the group of tested patients without prescribed antibiotics. This comparison weakens the association and it is no longer statistically significant for tics.

Instead of describing this as a possible source of bias the authors conclude that nonstreptococcal throat infection was also associated with increased risk of mental disorders, a theory that was not part of the original study hypothesis. Another interpretation is that these associations can be explained by a certain healthcare seeking behavior of patients and parents leading to an increased probability of receiving an antigen test, being prescribed an antibiotic and being treated in hospital.

References

1) Orlovska S, Vestergaard CH, Bech BH, Nordentoft M, Vestergaard M, Benros ME. Association of Streptococcal Throat Infection With Mental Disorders: Testing Key Aspects of the PANDAS Hypothesis in a Nationwide Study. JAMA Psychiatry 2017;74:740-6.

2) Bjerrum L, Gahrn-Hansen B, Hansen MP, Córdoba G, Aabenhus R, Monrad RN. [Airway infections – diagnosis and treatment. Clinical guideline for general practitioners]. Copenhagen: Danish College of General Practitioners; 2014.

3) Choby BA. Diagnosis and treatment of streptococcal pharyngitis. Am Fam Physician 2009; 79:383-90.

On 2017 Aug 29, Andreas Lundh commented:

Comment on Association of streptococcal throat infection with mental disorders

A recent Danish register-based cohort study(1) concludes that “individuals with a streptococcal throat infection had elevated risks of mental disorders, particularly OCD and tic disorders.” However, some methodological issues need consideration.

Firstly, the choice of exposure has a risk of misclassification. Exposure (i.e. Group A Streptococcal (GAS) throat infection) was the combination of a rapid antigen test performed by the patient’s general practitioner and subsequent antibiotic prescription. In Denmark use of rapid antigen test and antibiotic prescription are guided by modified Centor criteria(2,3) where symptomatic patients at high risk of GAS throat infection are not tested, but instead receive empirical antibiotic treatment. This leads to patients being misclassified as unexposed since they are never tested.

Secondly, the choice of outcome has a similar risk of misclassification. Psychiatric diagnoses are identified from national databases that only contain hospital information. Psychiatric patients that are not treated in hospitals (e.g. treated by primary care psychiatrists) are misclassified as not having had the outcome. Misclassification seems likely as only 0.1% and 0.2%, respectively, had a diagnosis of OCD or tics in the study period.

Thirdly, the analytical strategy has a risk of bias. The authors compared patients that had received both a rapid antigen test and antibiotics with a group that was never tested. GAS throat infection will in most cases resolve spontaneously and many patients will never contact their general practitioner for testing. The group tested therefore likely differs from the group not being tested and represents a group with certain healthcare seeking behavior. This is substantiated by the findings that risk of mental disorders seems to increase with number of tests and regardless of whether the tests are negative or positive. A more reasonable analysis that avoids confounding by test indication would be to compare the group of tested patients prescribed antibiotics with the group of tested patients without prescribed antibiotics. This comparison weakens the association and it is no longer statistically significant for tics.

Instead of describing this as a possible source of bias the authors conclude that nonstreptococcal throat infection was also associated with increased risk of mental disorders, a theory that was not part of the original study hypothesis. Another interpretation is that these associations can be explained by a certain healthcare seeking behavior of patients and parents leading to an increased probability of receiving an antigen test, being prescribed an antibiotic and being treated in hospital.

References

1) Orlovska S, Vestergaard CH, Bech BH, Nordentoft M, Vestergaard M, Benros ME. Association of Streptococcal Throat Infection With Mental Disorders: Testing Key Aspects of the PANDAS Hypothesis in a Nationwide Study. JAMA Psychiatry 2017;74:740-6.

2) Bjerrum L, Gahrn-Hansen B, Hansen MP, Córdoba G, Aabenhus R, Monrad RN. [Airway infections – diagnosis and treatment. Clinical guideline for general practitioners]. Copenhagen: Danish College of General Practitioners; 2014.

3) Choby BA. Diagnosis and treatment of streptococcal pharyngitis. Am Fam Physician 2009; 79:383-90.