Oral streptococcal infective endocarditis among individuals at high risk following dental treatment: a nested case-crossover and case-control study

BackgroundIt is not clear whether Viridans Group Streptococcal Infective Endocarditis (VGS-IE) among individuals at high risk is more frequent following bacteraemia caused by invasive dental procedures (IDPs) than after daily bacteraemia caused by chewing and tooth brushing. The aim of this nested study was to assess if VGS-IE was temporally associated with IDPs in a national cohort of individuals at high risk. MethodsThis nested case-control and case-crossover study was based on a Swedish national cohort study of 76,762 individuals at high risk of IE due to complex congenital heart disease, prosthetic heart valve or previous IE. Participants were living in Sweden between July 1st, 2008 and January 1st, 2018. The frequency of IDPs during the 3 months before VGS-IE was calculated and compared to controls (sampled 1:10). A case-crossover study was conducted to account for residual confounders. Participants were identified using the national patient register, and IDPs were identified using the national dental health register. Findings98,247 IDPs were carried out in the cohort during the study period: 624 occasions of oral surgery, 44,190 extractions and 53,433 sessions of subgingival scaling. The study could not confirm that IDPs were more common among cases (4.6%) than controls (4.1%), OR = 1.22 [95% Confidence Interval (CI) 0.64-2.3], or during case- (3.3%) than reference periods (3.8%), OR = 0.89 [95% CI: 0.68-1.17]. Restricting the analysis to the period when cessation of antibiotic prophylaxis for the prevention of IE in Swedish dentistry was recommended, from the 1st of October 2012 to the 1st of January 2018, did not alter the results of the case-control study: OR 0.64, 95% CI: 0.20-2.09, or the case-crossover study: OR 0.58, 95% CI: 0.15-2.19. InterpretationThe study could not confirm that VGS-IE is associated with IDPs among individuals at high risk. A study with larger sample size could clarify whether there is a lack of association. The finding of a small (