Van der Meer 1992a.

Study characteristics
Methods	Case‐control study
Participants	All 349 people who developed definite native‐valve endocarditis in the Netherlands over a 2‐year period (1 November 1986 to 1 November 1988) were collected. Cases (n = 48) were eligible if they had previously had congenital heart disease, coarctation of the aorta, rheumatic or other valvular dysfunction, or mitral valve prolapse with mitral regurgitation. Proxy responders (spouses or general practitioners) were used where cases were too ill to be interviewed or had died. Controls (n = 200) had not been diagnosed with endocarditis but had 1 of the cardiac conditions and were outpatients at a cardiology department of 1 of 5 hospitals. Controls were matched for age (within the same 5‐year age category). A random sample of potential controls was drawn, and, where there were at least 4 controls per case, all were contacted. Where there were fewer than 4 controls, a further random sample was drawn.
Interventions	Cases and controls had to have undergone a medical or dental procedure that required antibiotic prophylaxis within 180 days prior to the onset of symptoms of endocarditis (cases) or their interview (controls). Of the participants who underwent a dental procedure with definite indication for prophylaxis: 6 of 24 (25%) cases and 34 of 79 (43%) controls had removal of calculus plus polishing of teeth 9 of 24 (38%) cases and 15 of 79 (19%) controls had a dental extraction 1 case and 1 control had apical surgery 1 case and 1 control had dental extraction 1 (4%) case and 12 (15%) controls had dental avulsion 3 (13%) cases and 8 (10%) controls had removal of subgingival calculus 3 (13%) cases and 8 (10%) controls had root canal therapy. Median time from dental procedure to onset of endocarditis in cases was 10 days, range 0 to 175, and for the 7 who received antibiotics median time to onset was 18 days, range 7 to 60. Median time from dental procedure to interview in controls was 71 days, range 0 to 179 (12 missing values ignored), and for the controls who received antibiotics the median was 83 days, range 5 to 151 (1 missing value ignored). For both groups, all information about invasive procedures and use of prophylaxis was checked with medical or dental specialists and pharmacists.
Outcomes	Of the 349 people with definite native‐valve endocarditis, 197 had previous heart disease (10 proxy responders). Of these, 54 had undergone a medical or dental procedure with an indication for prophylaxis within the past 180 days. A causal relationship was ruled out in 6 of these 54 potential cases as the agent isolated from the blood was unlikely to have originated in the area of the procedure. Of the remaining 48 people with endocarditis, 44 had undergone a dental procedure with a definite (24) or possible (20) indication for prophylaxis (none of these cases had used a proxy responder). Of 889 potential controls who were sent an introductory letter, 689 were ineligible (53 had died, 29 had a prosthetic heart valve, 62 could not be located, 102 could not be contacted by phone and 418 had not undergone an invasive dental or medical procedure within the past 180 days) and the remaining 200 were interviewed by phone 2 to 5 days later. 181 of these controls had undergone a dental procedure with definite (79) or possible (102) indication for prophylaxis. The authors ensured that controls had not developed endocarditis, as defined by the diagnostic criteria of Von Reyn 1981. They also checked the appropriateness of antibiotic prophylaxis with medical, dental and pharmacy staff and against the Netherlands Heart Foundation recommendations, finding that 7 of 24 cases and 16 of 79 controls had had appropriate prophylaxis for a dental procedure requiring definite prophylaxis within the previous 180 days.
Notes	The published paper provided data on participants who had both medical and dental invasive procedures. The author kindly separated out those who had had invasive dental interventions.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Not undertaken (case‐control study). It is possible that, as dentists decide whether to give prophylaxis or not on the basis of the information about the patient in front of them, those patients appearing more frail may have been more likely to receive prophylaxis.
Allocation concealment (selection bias)	High risk	Not undertaken (case‐control study)
Blinding of outcome assessment (detection bias) All outcomes	High risk	Not undertaken
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Having died was an exclusion criterion for controls but not for cases, who could be included through a proxy.
Selective reporting (reporting bias)	Low risk	Expected outcomes and exposure reported
Confounding	Unclear risk	Participant sex and cardiac risk factor type was not described for the subgroup who had had a dental procedure, and the type of dental intervention appeared to be different in the cases and controls, although cases and controls were matched for age. Both groups were required to have undergone invasive dental techniques within 180 days prior to onset of symptoms/interview and data were split by time period for both groups.