Table 2.

Summary of articles reviewed

Author	Location	Population size	Study design	Percentage operated on	Intervention/ exposure	Endpoint(s)	Results	Main limitations
Kang et al. [2]	Seoul, South Korea	76	RCT	48.7%	Early surgery vs. conventional therapy to prevent emboli in IE patients with vegetations > 10 mm.	Death, embolic event, or recurrence of IE within 6 weeks and 7 years.	Primary endpoint reached in 3% vs. 23%, respectively. 7-year follow-up revealed event-free survival of 87% vs. 59%, respectively.	Patients of lower risk and mean age (47y) much lower than overall IE-patient population. Small population size.
Fosbøl et al. [19]	Multinational	1,006	Retrospective register study using ICE-Plus database, propensity matching used.	53.4%	Vegetation size and surgery’s effect on 6-month mortality.	In-hospital or 6-month mortality.	Larger vegetation size carries higher mortality. Mortality after surgery is not affected by vegetation size.	Lack of data regarding cause of death in cohort. All ICE centres are tertiary centres, possibility of selection and referral bias.
Kim et al. [20]	Seoul, South Korea	132	Prospective enrolment from 1998–2006, with 44 pairs propensity matched.	48.5%	Effect of early surgery in prevention of emboli, compared to antibiotic therapy alone in patients with IE.	Embolism, cardiovascular mortality, recurrence of IE.	Higher 5-year survival rate, and lower percentage reaching endpoint in early surgery group [HR 0.14 (95% CI 0.03–0.64)].	Early surgery patients tended to have larger vegetations. Propensity matching not sufficient to completely avoid confounding.
Wang et al. [17]	Multinational	148	Retrospective register study using ICE-MD database, propensity matching used.	42.0%	Mortality in patients with PVIE who underwent surgery, compared to similar patients who did not undergo surgery.	In-hospital mortality.	Lower in-hospital mortality and increased survival in surgically treated group [OR 0.56 (95% CI 0.23–1.36)].	Certain variables were incomplete. Propensity matching not sufficient to completely avoid confounding.
Young et al. [20]	London, United Kingdom	142	Retrospective study using electronic patient records. Propensity matching was not used.	69.0%	Association between vegetation size on valve destruction, embolism and mortality.	6-month mortality.	Vegetation area was associated with mortality in antibiotic therapy alone group (HR 1.01 [95% CI 1.00–1.02, p-value < 0.01], but not in surgery + antibiotic therapy group.	Selection and referral bias, due to retrospective design at tertiary centre.
Song et al. [21]	Seoul, South Korea	419	Retrospective study using electronic patient records. Propensity matching was not used.	65.2%	Association between vegetation size and survival and cerebral emboli.	Long-term mortality.	Vegetation size was only associated with mortality in patients treated with antibiotic therapy alone. Surgical treatment was strongly associated with better 1-, 3-, and 5- year survival (97.4% vs. 89.7% and 96.3% vs. 85.6% and 95.4% vs. 83.3%) and lower risk of embolism (OR 0.78 (0.5–1.14)).	Selection bias. Unknown if patients had already suffered cerebral embolism before echo.
Scheggi et al. [18]	Multinational	638	Retrospective study using electronic patient records. 49 surgical patients propensity matched with 98 patients treated with antibiotic therapy alone.	33.3%	All-cause mortality in patients with prevention of embolism as only surgical indication.	5-year survival.	Surgically managed patients have lower mortality rates, compared to patients treated with antibiotic therapy alone [HR 0.223 (95% CI 0.079–0.656)]. 5-year survival rate higher in surgical patients (82.4% vs. 66.5%)	Changes in management of IE during long study period. Tertiary centre with lower surgical mortality than smaller centres.
Desch et al. [22]	Luebeck, Germany	71	Observational study. All patients diagnosed with IE at single centre between 2000 and 2012 included. Propensity matching was not used.	83.1%	All-cause mortality in patients with prevention of embolism as only surgical indication.	Long-term mortality.	Surgery was independently associated with mortality, compared to antibiotic therapy alone [HR 3.9 (95% CI 0.9–16 − 6)]	Selection and scientific bias. Physicians possibly tended to operate on higher-risk patients. Small sample size.
Cabezón et al. [19]	Madrid, Spain	726	Retrospective study using electronic patient records. 70 surgically treated patients matched with 69 patients treated with antibiotic therapy alone.	50.4%	All-cause mortality in patients with prevention of embolism as only surgical indication.	90-day mortality.	Surgically treated with no other indications for surgery had lower survival probability than patients treated with antibiotic therapy alone (81.4% vs. 88.4%).	Selection and referral bias, due to retrospective design at tertiary centre.

RCT = Randomized controlled trial; IE = Infective endocarditis; ICE = International Collaboration on Endocarditis; HR = Hazard ratio; CI = Confidence interval