Table 1

Characteristics of the included studies

Study	Country	Study design	Intervention (N)	Type of valve	Duration of follow-up (days or months)	Outcomes	Main finding
Paparella 201917	Italy	Retrospective cohort	miAVR (386)	Mechanical (Bicarbon and CarboMedics); Biological (Hancock II and Mosaic)*	–	30-day ACM, stroke, repeat intervention, AF, AKI, hospitalisation duration	The miAVR arm required more blood transfusion, longer hospitalisation,and had a higher incidence of AKI. The TAVI arm had more permanent pacemakers. No differences in terms of 30-day mortality or stroke.
Paparella 201917	Italy	Retrospective cohort	TAVI (386)(98% TF; 2% TA)†	CoreValve (93.2%), Lotus (6.8%)†	–
Furukawa 201818	Germany	Retrospective cohort	miAVR (177)	miAVR: Perimount Magna (74%), Perimount Magna Ease (14.1%), Hancock II (2.8%), Trifecta (7.9%), Perceval (1.1%)	766 days	30-day ACM, bleeding, stroke, AMI, AF, AKI, paravalvular leakage, hospitalisation duration	Longer hospitalisation in the transapical arm. The transapical arm trended towards worse midterm survival. No differences in terms of 30-day mortality, stroke or myocardial infarction.
			TA–TAVI (177)	TA–TAVI: Sapien XT (41.8%), Sapien 3 (32.2%), Accurate TA (26%)
			TF–TAVI (177)	CoreValve (55.9%), Sapien XT 15.8%, Sapien 3 (14.7%), Accurate TF neo (2.3%), Direct Flow (11.3%)
Calle-Valda 201719	Spain	Retrospective cohort	miAVR (50)	NA	46.7 months	30-day ACM, repeat intervention (re-exploration for bleeding), bleeding (postoperative bleeding mL/24 hours), stroke (30 days), AF, 30-day readmission, hospitalisation duration (days)	The TAVI arm required more pacemakers and had higher rates of paravalvular leakage. The TAVI arm required shorter hospitalisation.
Calle-Valda 201719	Spain	Retrospective cohort	TF–TAVI (50)	CoreValve (100%)	46.7 months		No statistically significant differences in terms of survival.
Bruno 201722	Italy	Retrospective cohort	miAVR (19)	Intuity Valve (100%)	29.1 months 27.7 months	ACM, bleeding, stroke, AMI, AF, paravalvular leakage (early and midterm), hospitalisation duration	The TAVI arm required more pacemakers and had higher rates of paravalvular leakage. No significant differences in terms of mortality.
Bruno 201722	Italy	Retrospective cohort	TF–TAVI (30)	CoreValve (100%)	29.1 months 27.7 months
Hijri 201730	USA	Retrospective cohort	SAVR (722)	Bioprosthetic (92%), Mechanical (8%)	35 months	Operative mortality, AKI, hospitalisation duration	The TAVI (irrespective of approach), miAVR and conventional surgical arms had comparable rates of intraoperative and midterm mortality.
Hijri 201730	USA	Retrospective cohort	TAVI (306)	Sapien (28.4%), Sapien XT (23.9%), Sapien 3 (30.7%), CoreValve (15%)	35 months	Operative mortality, AKI, hospitalisation duration
Nguyen 20179	USA	Retrospective cohort	miAVR (699)	NA	–	30 day ACM, stroke, AF, dialysis, hospitalisation duration	The results showed an increasing rate of adoption for TF–TAVI and miAVR, but a decrease in TA–TAVI and conventional surgery. 30-day mortality was highest for TA–TAVR, followed by TF–TAVR, SAVR and miAVR.
			TF–TAVI (727)	NA
			TA–TAVI (303)	NA
Tokarek 2015/201623	Poland	Retrospective cohort	TF–TAVI (39)	Sapien XT (79%), CoreValve (21%)	583.5 days	ACM (30 days, 6 months, 1 year), bleeding, stroke, AMI, AF, paravalvular leakage QoL (EQ-5D-3L and MLHFQ 24 M)	The TAVI arm had a higher ejection fraction, but there were no differences in mortality (2015). The TAVI arm had better QoL for up to 1 year, but no differences persisted at 2 years (2016).
			MT (50)	NA
			MS (44)	NA
Miceli 201621	Italy	Retrospective cohort	RT (37)	Perceval S (100%)	13 months	Mortality, bleeding, stroke, paravalvular leakage, AKI, hospitalisation duration	The TAVI arm had a significantly higher rate of paravalvular leakage. No significant differences in terms of stroke, 1 year and 2 year survival.
Miceli 201621	Italy	Retrospective cohort	TAVI (37)(51.6% TF; 48.3% TA)**	Sapien (100%)	13 months
Santarpino 201416	Germany	Retrospective cohort	MIS (37)	Perceval (100%)	18.9 months	In-hospital mortality, stroke, paravalvular leakage, AKI	In high-risk patients, cumulative survival was higher in the miAVR arm compared with the TAVI arm. TAVI had significantly higher rates of paravalvular leakage, which was significantly associated with mortality.
Santarpino 201416	Germany	Retrospective cohort	TAVI (37) (59% TA; 40.2% TF; 0.8% transaortic)†	Sapien, Sapien XT*	18.9 months	In-hospital mortality, stroke, paravalvular leakage, AKI
Haldenwang 201429	Germany	Retrospective cohort	miAVR (77)	Perimount, Trifecta*	–	AKI	TA–TAVI carried a higher risk of AKI.
Haldenwang 201429	Germany	Retrospective cohort	TA–TAVI (56)	SAPIEN (100%)	–	AKI	TA–TAVI carried a higher risk of AKI.
Zierer 200920	Germany	Retrospective cohort	TA–TAVI (21)	Cribier-Edwards (100%)	–	ACM (30 days, 1 year), repeat intervention, stroke, AMI, AF, hospitalisation duration	TA–TAVI had shorter postoperative recovery. There were no significant differences in terms of morbidity or mortality.
Zierer 200920	Germany	Retrospective cohort	PUS-AVR (30)	Perimount (100%)	–

*Insufficient data provided to specify the percentages used for each valve type.
†Percentages given for overall cohort rather than the propensity-matched cohort used for analysis, as data for the latter were not available.
ACM, all-cause mortality; AF, atrial fibrillation; AKI, acute kidney injury; ; AMI, acute myocardial infarction; miAVR, minimally invasive aortic valve replacement; MIS, minimally invasive sutureless; MS, ministernotomy; MT, mini-thoracotomy; NA, not available; PUS-AVR, partial upper sternotomy-aortic valve replacement; RD-AVR, rapid-deployment aortic valve replacement; RT, right anterior mini-thoracotomy; SAVR, surgical aortic valve replacement; TA, transapical; TAVI, transcatheter aortic valve implantation; TAVR, transcatheter aortic valve replacement; TF, transfemoral.