The results need to be interpreted in light of some limitations. First, despite the large cohorts, power was still limited for several of the individual AD, particularly streptococcus-related diseases (such as rheumatic fever with or without chorea), which means that we could not calculate some of the specific risks or these resulted in less precise estimates (wide CI). Second, because the Swedish versions of ICD-8 and ICD-9 do not differentiate between type 1 and type 2 diabetes mellitus, we only included individuals diagnosed with type 1 diabetes mellitus according to ICD-10, resulting in the exclusion of older cases diagnosed before ICD-10 was introduced in Sweden (1997). Third, our cohorts do not represent the totality of all OCD and TD/CTD patients in Sweden; many sufferers do not seek help, the coverage of the register before 2001 is incomplete, and patients diagnosed by general practitioners and non-medical specialists are not included. In contrast, AD may be more likely to be seen by specialist physicians in Sweden, resulting in better coverage in the patient register.
