Sex differences in the diagnosis, management and outcomes of patients with infective endocarditis

Abstract

Background

Despite growing interest in male–female differences in cardiovascular disease, evidence in infective endocarditis (IE) is limited and contradictory.

Methods

This prospective study included all patients with definite or possible valvular IE discussed by the endocarditis team from 2016 to 2025. Baseline characteristics, diagnostics, treatment and outcomes were compared between sexes. A Cox model was conducted to assess survival adjusted for baseline characteristics, IE type and treatment.

Results

The cohort included 791 patients with definite or possible IE (72.8% males, 27.2% females). Age was not different (male: 67 (IQR 56–75), female: 71 (IQR 55–78) years, p=0.07). Females more often had hypertension (46.0% vs 35.2%, p=0.07) and mitral valve IE (45.6% vs 32.3%, p<0.001). Males had more predisposing conditions (71.5% vs 60.5%, p=0.004) and more aortic valve IE (71.0% vs 62.3%, p=0.02) and Cutibacterium species (5.7% vs 1.4%, p=0.007). There was no difference in indication for surgery. However, males were more often treated surgically (41.1% vs 29.3%, p=0.002), and females with an indication for surgery were more often treated conservatively (22.5% vs 37.0%, p=0.006). After a median follow-up of 2.1 (IQR 0.4–4.4) years, females had higher mortality (adjusted HR (aHR) 1.39 (95% CI 1.03 to 1.89), p=0.03). Surgery was associated with higher mortality in males (aHR 1.32 (95% CI 0.90 to 1.94) vs aHR 0.62 (95% CI 0.32 to 1.20), p value for interaction=0.03).

Conclusions

Differences were observed in baseline characteristics, IE type and survival. Females had worse overall adjusted survival, suggesting a less favourable overall prognosis. In terms of surgical decision-making, surgery was withheld more often in females despite surgical indication, and after excluding these patients, surgery appeared more protective in females than in males. These findings may reflect sex differences in surgical selection.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Infective endocarditis affects more males than females, with a ratio of approximately 2:1, but a clear explanation is missing. There are known sex differences in baseline characteristics and involvement of the aortic and mitral valve. However, evidence on sex differences concerning surgical indication, the effect of surgery and mortality is limited and contradictory.

WHAT THIS STUDY ADDS

• Although the indication for surgery is not significantly different between sexes, females are more often treated conservatively despite having an indication for surgery.

• After excluding patients who were managed conservatively despite having a surgical indication, females had higher overall adjusted mortality.

• Surgery was associated with higher mortality as opposed to conservative treatment in males compared with females.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Endocarditis is much more prevalent in males, but overall survival is worse in females, warranting more research to explain these findings. The observed higher operation rate and higher mortality after surgical treatment compared with conservative treatment in males suggests sex differences in surgical selection. The risk of mortality seems to be a complicated interplay of multiple factors, which justifies careful evaluation and advice by an experienced endocarditis team.

Introduction

Infective endocarditis (IE) is a rare disease with an incidence of approximately 14 per 100 000 persons per year globally.¹ Despite increasing interest in sex disparities in clinical characteristics, treatment and outcomes in cardiovascular research in the past years, sex disparities in patients with IE have not been widely studied yet. Although males more frequently develop IE, with a male-to-female ratio of approximately 2:1.²,⁴ Previous studies found that females were less often operated on than males, even after adjustment for baseline characteristics and comorbidity.⁴,⁶ Furthermore, some studies described higher short-term mortality in females and suggested that various factors contribute to these differences between sexes, including variations in age, comorbidities and treatment strategies.⁴ Indeed, some data regarding these male–female differences contradict. In this study, we further investigated male–female differences in management and outcomes in a real-world cohort of patients discussed by our endocarditis team (ET).⁷

Methods

Study design

All patients discussed in our ET from January 2016 to January 2025 with (suspected) IE were entered in our database. Patients with rejected IE, or infection involving only non-valvular structures, such as cardiac implantable electronic devices and vascular grafts, were excluded. Data were collected from the prospectively recorded and standardised ET documentation and electronic patient records.

The endocarditis team

In our tertiary referral centre, a regional ET was installed at the end of 2015 following the implementation of the 2015 European Society of Cardiology (ESC) guidelines for the management of IE.⁸,¹⁰ The local members of the ET include a cardiologist, medical microbiologist or infectious diseases specialist, radiologist and/or nuclear medicine physician, and a cardiothoracic surgeon. Additionally, the presence of the referring physician and the microbiologist of the referring centre is preferred. Patients are discussed biweekly following referral by their treating physician, either local or from referring centres. The final diagnosis, that is, definite, possible or rejected IE, was defined as a consensus diagnosis after (re)discussion by the ET.

Type of IE

The type of IE was defined as native valve endocarditis, prosthetic valve endocarditis or transcatheter aortic valve endocarditis. Prosthetic valve endocarditis comprised infection of mechanical valves, tissue valves (including transcatheter valves in mitral, pulmonary or tricuspid position) and surgical valve repairs. Patients with transcatheter aortic valves are often older and more frail than patients with surgically implanted valves, and therefore, patients with transcatheter aortic valve endocarditis were categorised separately. Patients with combined prosthetic valve and native valve endocarditis were included in the prosthetic valve endocarditis group, and patients with combined transcatheter aortic valve endocarditis and native and/or other prosthetic valve endocarditis were categorised as transcatheter aortic valve endocarditis.

Treatment

Data on indicated and performed treatment were collected. The reason for the indication for surgery was defined as stated in the ESC guidelines for the management of IE, applicable at the time of discussion.^{8 11}

Follow-up

Data on relapse (<6 months) and reinfection (>6 months or different causative pathogen) were collected and scored in accordance with the ESC guidelines for the management of IE.^{8 11}

For all patients, data on all-cause mortality were obtained from the Dutch national population database on the 6th of January 2025. The follow-up time for survival was defined as the time between the first discussion by the ET and the date of death or the date the survival data was checked.

Statistics

Data were compared between the sexes. Categorical variables were reported as percentages and compared between groups using the χ² or Fisher’s exact test. Continuous data were reported using medians with 25th and 75th percentiles for non-parametric variables and compared using the Mann-Whitney U test. Missing data were handled using complete-case analyses. An overview of missing data is available in online supplemental table 1.

Survival analysis was visualised using Kaplan-Meier curves and compared across subgroups using a log-rank test. Follow-up completeness for survival was assessed using the Modified Clark C method.¹² A comparison of the survival rate in this cohort with that of the general Dutch population was performed using a patient-level matching strategy, matching patients based on age, sex, calendar year of diagnosis and country using data of the human mortality database.¹³ The survival of patients in our cohort relative to that of the general Dutch population was calculated ([survival IE cohort/survival general Dutch population ]*100). This allows for interpretation of mortality among males and females with IE, relative to their sex-specific background mortality.

A multivariable Cox proportional hazards model for overall survival was developed to adjust the hazard ratio (HR) for sex for confounding due to differences in baseline, IE type and treatment. Variables were selected based on clinical relevance and all significant univariable predictors were included in the multivariable model (p<0.05). We aimed to isolate the effect of undergoing surgery versus no surgical indication, and we believe that patients not operated on despite having an indication for surgery are a clinically important group with very high early mortality that should not be included in this analysis, as this would introduce confounding by (contra)indication. Therefore, patients who were treated conservatively despite having a surgical indication were excluded from the multivariable model. A time-varying covariate was added to the model to account for possible immortal time bias for the surgically treated patients. Correlations between variables were tested by Spearman’s correlation coefficient. The proportional hazard assumption was tested using Schoenfeld residuals. In cases of strong violation, the model was stratified by these variables. The final multivariable model included sex, age, medical history of diabetes, hypertension, heart failure, chronic obstructive pulmonary disease, treatment (surgery or conservative treatment), and was stratified for the type of IE.

Analyses were performed using IBM SPSS statistics (V.28.0.1.0) and R V.4.4.2. A p<0.05 was considered statistically significant.

Results

Within our cohort, 939 patients were referred with suspected IE, of which 791 patients had definite or possible IE (27.2% were female, 72.8% male) (figure 1). The diagnosis was significantly more often rejected in females (21.5% vs 13.4%, p=0.002). Baseline characteristics are described in table 1. The median age did not differ (male: 67 (56–75) vs female: 71 (55–78) years, p=0.07). Males had less hypertension (35.2% vs 46.0%, p=0.007) and more prior vascular prosthesis surgery (11.5% vs 6.0%, p=0.03).

Figure 1. Number of males and females in the total cohort, among patients with definite/possible infective endocarditis (IE), and the number of patients in each treatment group.

Table 1. Baseline characteristics, diagnostic criteria and final diagnosis in the total cohort, males and females.

	Total cohort N=791	Male N=576	Female N=215	P value
Baseline characteristics
Age (years)	68 (56–76)	67 (56–75)	71 (55–78)	0.07
Prior episode of endocarditis	48 (6.1)	38 (6.6)	10 (4.7)	0.39
Pre-existent valve disease	403 (50.9)	299 (51.9)	104 (48.4)	0.42
Congenital heart disease	115 (14.5)	90 (15.6)	25 (11.6)	0.19
Prior heart valve surgery	352 (44.5)	268 (46.5)	84 (39.1)	0.07
Prior CIED implantation	89 (11.3)	72 (12.5)	17 (7.9)	0.09
Type of CIED implanted	89 (11.3)	72 (12.5)	17 (7.9)	0.30
Pacemaker*	62 (69.7)	52 (72.2)	10 (58.8)
ICD*	17 (19.1)	12 (16.7)	5 (29.4)
CRTP*	2 (2.2)	1 (1.4)	1 (5.9)
CRTD*	8 (9.0)	7 (9.7)	1 (5.9)
Prior vascular prosthesis	79 (10.0)	66 (11.5)	13 (6.0)	0.03
Type of vascular prosthesis	219 (73.5)	153 (69.9)	66 (83.5)	0.02
Ascending aorta*	7 (2.3)	7 (3.2)	0 (0.0)
Ascending aorta+other*	54 (18.1)	47 (21.5)	7 (8.9)
Bentall*	9 (3.0)	7 (3.2)	2 (2.5)
Bentall+other*	6 (2.0)	4 (1.8)	2 (2.5)
Other*	3 (1.0)	1 (0.5)	2 (2.5)
Hypertension	302 (38.2)	203 (35.2)	99 (46.0)	0.007
Heart failure	97 (12.3)	76 (13.2)	21 (9.8)	0.24
Chronic renal failure	76 (9.6)	55 (9.5)	21 (9.8)	1.00
CVA/TIA	119 (15.0)	90 (15.6)	29 (13.5)	0.53
COPD	57 (7.2)	35 (6.1)	22 (10.2)	0.06
Diabetes mellitus	140 (17.7)	95 (16.5)	45 (20.9)	0.17
Intravenous drug use	6 (7.7)	5 (8.2)	1 (5.9)	1.00
Major diagnostic criteria
Positive blood cultures	647 (81.8)	468 (81.2)	179 (83.3)	0.58
Positive imaging	717 (90.6)	523 (90.8)	194 (90.2)	0.92
Minor diagnostic criteria
Predisposition	542 (68.5)	412 (71.5)	130 (60.5)	0.004
Fever	564 (71.4)	415 (72.2)	149 (69.3)	0.48
Embolic complications	321 (40.6)	233 (40.5)	88 (40.9)	0.97
Other microbiological evidence†	108 (13.7)	81 (14.1)	27 (12.6)	0.66
Immunological phenomena	17 (2.1)	13 (2.3)	4 (1.9)	0.95
Final diagnosis
Definite	691 (87.4)	502 (87.2)	189 (87.9)	0.87
Possible	100 (12.6)	74 (12.8)	26 (12.1)	0.87

Data are presented as n (%) or median with 25th–75th percentile.Examples of ‘other’ vascular prostheses are hemi aortic arch replacements, aortic arch replacements and transcatheter endovascular graft replacements.

^*Percentage within the subgroup.

^†Other microbiological evidence than major criterium.

^‡Within the subgroup of patients with right-sided or cardiac implantable electronic device infection.

CIED, cardiac implantable electronic device; COPD, chronic obstructive pulmonary disease; CRTD, cardiac resynchronisation therapy with a defibrillator; CRTP, cardiac resynchronisation therapy with a pacemaker; CVA, cerebrovascular accident; ICD, implantable cardioverter defibrillator; TIA, transient ischaemic attack.

Diagnostic criteria

There were no differences in the presence of major diagnostic criteria between sexes (table 1). Among the minor criteria, the presence of predisposing conditions (eg, previous IE, medical history of heart valve disease, cardiac implantable electronic devices, prosthetic valves, congenital heart disease) was observed more frequently in males (71.5% vs 60.5%, p=0.004).¹¹ The presence of other minor criteria was not different between sexes.

The location and type of infected valve among patients with possible or definite IE is shown in online supplemental figures 1 and 2. The distribution of IE type trended toward male-female differences (p=0.053), mainly due to a higher proportion of prosthetic valve endocarditis in males (38.4% vs 29.3%: p=0.02). The position of the infected valve was different between sexes (p=0.02). The aortic valve was more often infected in males (71.0% vs 62.3%, p=0.02; online supplemental figure 1), but the mitral valve was less often involved (32.3% vs 45.6%, p<0.001) compared with females. There was no difference in the involvement of right-sided valves (pulmonary: p=0.19, tricuspid: p=0.24), and 91 (11.5%) patients had more than one valve involved.

Causative pathogen

The most common causative pathogens found in blood cultures were viridans group streptococci, Staphylococcus aureus and Enterococcus faecalis, and not different between sexes (table 2). Cutibacterium species were more frequently found in males (5.7% vs 1.4%, p=0.007).

Table 2. Causative pathogen among males and females with definite or possible infective endocarditis.

	Total N=791	Male N=576	Female N=215	P value
Viridans group streptococci	195 (24.7)	138 (24.0)	57 (26.5)	0.46
Staphylococcus aureus *	188 (23.8)	130 (22.6)	58 (27.0)	0.22
Enterococcus faecalis	99 (12.5)	69 (12.0)	30 (14.0)	0.47
Streptococcus bovis	56 (7.1)	45 (7.8)	11 (5.1)	0.22
Other streptococci	56 (7.1)	45 (7.8)	11 (5.1)	0.22
Coagulase negative staphylococci	42 (5.3)	35 (6.1)	7 (3.3)	0.15
Cutibacterium species	36 (4.6)	33 (5.7)	3 (1.4)	0.007
HACEK organisms	13 (1.6)	8 (1.4)	5 (2.3)	0.36
Streptococcus pneumoniae	8 (1.0)	6 (1.0)	2 (0.9)	1.00
Enterobacterales	8 (1.0)	8 (1.4)	0 (0.0)	0.12
Other enterococci	6 (0.8)	2 (0.3)	4 (1.9)	0.050
Staphylococcus lugdunensis	5 (0.6)	4 (0.7)	1 (0.5)	1.00
Tropheryma whipplei	2 (0.3)	1 (0.2)	1 (0.5)	0.47
Bartonella species	1 (0.1)	1 (0.2)	0 (0.0)	1.00
Other pathogens	29 (3.7)	16 (2.8)	13 (6.0)	0.04
Lactobacillus species	4 (0.5)	4 (0.7)	0 (0.0)
Aerococcus urinae	4 (0.5)	3 (0.5)	1 (0.5)
Corynebacterium species	4 (0.5)	3 (0.5)	0 (0.0)
Candida species	3 (0.4)	2 (0.9)	2 (0.9)
Neisseria species	3 (0.4)	1 (0.2)	2 (0.9)
Pseudomonas aeruginosa	3 (0.4)	0 (0.0)	3 (1.4)
Moraxella catarrhalis	2 (0.3)	0 (0.0)	2 (0.9)
Listeria monocytogenes	1 (0.1)	1 (0.2)	0 (0.0)
Campylobacter fetus	1 (0.1)	0 (0.0)	1 (0.5)
Rothia mucilaginosa	1 (0.1)	1 (0.2)	0 (0.0)
Fusobacterium necrophorum	1 (0.1)	0 (0.0)	1 (0.5)
Bifidobacterium species	1 (0.1)	1 (0.2)	0 (0.0)
Achromobacter xylosoxidans/dentrificans	1 (0.1)	0 (0.0)	1 (0.5)
No pathogen cultured	47 (5.9)	35 (6.1)	12 (5.6)	0.87

Data are presented as n (%). Causative pathogens are listed in descending order of frequency.

^*186 methicillin susceptible S. aureus, 2 methicillin-resistant S. aureus.

Treatment

There were no differences in the duration of antimicrobial treatment (p=0.87, table 3) or indicated treatment (p=0.10). The reason for the indication for urgent surgery was also not different between sexes (table 3). However, males were more often operated on (41.1% vs 29.3%, p=0.002). The median time from first discussion to surgery was 9 (2–27) days for males and 7 (2–20) days for females (p=0.45). Among patients with an indication for surgery, females were moreoften treated conservatively than males (22.5% vs 37.0%, p=0.006). No differences were observed in baseline characteristics, type of IE and indication for surgery between males and females treated conservatively despite having an indication for surgery (table 4).

Table 3. Duration of antimicrobial treatment, indicated and performed treatment for males and females with definite or possible infective endocarditis.

	Total N=791	Male N=576	Female N=215	P value
Antimicrobial treatment
6 weeks	631 (84.9)	461 (85.1)	170 (84.6)	0.87
4 weeks	112 (15.1)	81 (14.9)	31 (15.4)
Chronic suppressive therapy	15 (1.9)	10 (1.7)	5 (2.3)	0.57
Indicated treatment
Conservative treatment	385 (48.7)	270 (46.9)	115 (53.5)	0.10
Indication for surgery	406 (51.3)	306 (53.1)	100 (46.5)	0.10
Elective surgery*	77 (9.7)	58 (19.0)	19 (19.0)	0.99
Urgent surgery*	305 (38.6)	228 (74.5)	77 (77.0)	0.62
Device extraction*	13 (1.6)	10 (3.3)	3 (3.0)	1.00
Urgent surgery+device extraction*	11 (1.4)	10 (3.3)	1 (1.0)	0.31
Indication for urgent surgery	n=316	n=238	n=78
Congestive heart failure*	92 (11.6)	65 (27.3)	27 (34.6)	0.22
Uncontrolled infection*	97 (12.3)	75 (31.5)	22 (28.2)	0.58
Embolisation*	67 (8.5)	49 (20.6)	18 (23.1)	0.64
Perivalvular abscess/mycotic aneurysm*	130 (16.4)	104 (43.7)	26 (33.3)	0.11
Other*	18 (2.3)	12 (5.0)	6 (7.7)	0.40
Performed treatment
Conservative treatment	385 (48.7)	270 (46.8)	115 (53.2)	0.11
Surgery performed	300 (37.9)	237 (41.1)	63 (29.3)	0.002
Elective surgery*	65 (8.2)	47 (19.8)	18 (28.6)	0.13
Urgent surgery*	219 (27.7)	176 (74.3)	43 (68.3)	0.34
Device extraction*	10 (1.3)	8 (3.4)	2 (3.2)	1.00
Urgent surgery+device extraction*	6 (0.8)	6 (2.5)	0 (0.0)	0.35
Conservative despite indication for surgery	106 (26.1)	69 (22.5)	37 (37.0)	0.006
Comorbidity/high risk*	92 (86.8)	59 (85.5)	33 (89.2)	0.77
Patients’ wish*	14 (13.2)	10 (14.5)	4 (10.8)

Data are presented as n (%).

^*Percentage within group.

Table 4. Baseline characteristics, type of IE, indication for surgery and reason surgery was not performed among males and females treated conservatively despite having a surgical indication.

	Total cohort N=106	Male N=69	Female N=37	P value
Baseline characteristics
Age (years)	73 (63–78)	72 (63–77)	75 (65–81)	0.15
Prior episode of endocarditis	8 (7.5)	6 (8.7)	2 (5.4)	0.71
Pre-existing valve disease	60 (56.6)	38 (55.1)	22 (59.5)	0.69
Congenital heart disease	8 (7.5)	5 (7.2)	3 (8.1)	1.00
Prior heart valve surgery	48 (45.3)	34 (49.3)	14 (37.8)	0.31
Prior CIED implantation	18 (17.0)	13 (18.8)	5 (13.5)	0.59
Prior CIED type				0.63
Pacemaker*	10 (55.6)	7 (53.8)	3 (60.0)
ICD*	5 (27.8)	3 (23.1)	2 (40.0)
CRTP*	0 (0.0)	0 (0.0)	0 (0.0)
CRTD*	3 (16.7)	3 (23.1)	0 (0.0)
Prior vascular prosthesis	7 (6.6)	7 (10.1)	0 (0.0)	0.09
Prior vascular prosthesis type				1.00
Ascending aorta*	0 (0.0)	0 (0.0)	0 (0.0)
Ascending aorta+other*	5 (71.4)	5 (71.4)	0 (0.0)
Bentall*	2 (28.6)	2 (28.6)	0 (0.0)
Bentall+other*	0 (0.0)	0 (0.0)	0 (0.0)
Other*	0 (0.0)	0 (0.0)	0 (0.0)
Hypertension	43 (40.6)	27 (39.1)	16 (43.2)	0.68
Heart failure	19 (17.9)	14 (20.3)	5 (13.5)	0.44
Chronic renal failure	13 (12.3)	9 (13.0)	4 (10.8)	1.00
CVA/TIA	19 (17.9)	12 (17.4)	7 (18.9)	1.00
COPD	10 (9.4)	4 (5.8)	6 (16.2)	0.09
Diabetes mellitus	28 (26.4)	18 (26.1)	10 (27.0)	1.00
Type of IE
NVE	61 (57.5)	38 (55.1)	23 (62.2)	0.81
PVE	29 (27.4)	20 (29.0)	9 (24.3)
TAVI-IE	16 (15.1)	11 (15.9)	5 (13.5)
Indication urgent surgery
Congestive heart failure	25 (23.6)	13 (18.8)	12 (32.4)	0.15
Uncontrolled infection	17 (16.0)	13 (18.8)	4 (10.8)	0.41
Embolisation	18 (17.0)	10 (14.5)	8 (21.6)	0.42
Abscess/mycotic aneurysm	34 (32.1)	25 (36.2)	9 (24.3)	0.28
Other	5 (4.7)	3 (4.3)	2 (5.4)	1.00
Reason surgery was not performed
Patients wish	14 (13.2)	10 (14.5)	4 (10.8)	0.77
High risk/comorbidity	92 (86.8)	59 (85.5)	33 (89.2)	0.77

Data are presented as n (%) or median with 25th–75th percentile. Examples of ‘other’ vascular prostheses are hemi aortic arch replacements, aortic arch replacements and transcatheter endovascular graft replacements.

^*Percentage within group.

CIED, cardiac implantable electronic device; COPD, chronic obstructive pulmonary disease; CRTD, cardiac resynchronisation therapy with a defibrillator; CRTP, cardiac resynchronisation therapy with a pacemaker; CVA, cerebrovascular accident; ICD, implantable cardioverter defibrillator; IE, infective endocarditis; NVE, native valve endocarditis; PVE, prosthetic valve endocarditis; TAVI, transcatheter aortic valve implantation; TIA, transient ischaemic attack.

Outcomes

Follow-up completeness for survival was 99.0% and the median follow-up time was 2.1 (0.4–4.4) years. Survival was significantly different between the sexes (p=0.01; figure 2), with a 30-day, 1-year and 5-year survival of 90.4%, 83.6%, and 81.6% for males, and 70.4%, 63.8% and 56.8% for females. Survival for patients relative to that of the general Dutch population was 73.4% for males and 64.5% for females 5 years after the first discussion by the ET (figure 2). Figure 2 also shows that, in males, survival was initially worse in the conservative group (p=0.03), but as follow-up progressed, the survival curves intersected. In females, surgically treated patients had better survival compared with conservatively treated patients (p<0.001). For both males and females, patients with an indication for surgery that did not undergo surgical intervention had worse survival compared with patients treated conservatively (male: p<0.001, female: p=0.002) or surgically (both p<0.001). Survival was not significantly different between different types of valves affected by IE (p=0.39, online supplemental figure 3).

Figure 2. The upper Kaplan-Meier curve shows survival of males and females with definite or possible infective endocarditis (solid lines) compared with the general Dutch population (dashed lines). The lower Kaplan-Meier curves show survival of patients with definite or possible infective endocarditis stratified for treatment. The Kaplan-Meier curve on the left shows survival for males and the Kaplan-Meier curve on the right shows survival for females.

The outcomes of the multivariable Cox proportional hazards model showed that overall mortality was significantly higher in females (adjusted hazard ratio (aHR) 1.39 (95% CI 1.03 to 1.89), p=0.03; figure 3, online supplemental tables 3 and 4). Surgery was not significantly associated with the hazard for mortality compared to conservative treatment in the subgroup of males or females; however, the effect of surgery on mortality was significantly different between sexes (figure 3). Surgery was associated with significantly higher mortality compared to conservative treatment in males compared with females (male: aHR 1.32 (95% CI 0.90 to 1.94), female: aHR 0.62 (95% CI 0.32 to 1.20), p value for interaction=0.03). In the subgroup of patients operated on, no differences in 30-day mortality for males (n=21/237, 8.9%) and females (n=3/63, 4.8%) after surgery (p=0.43) were observed.

Figure 3. Forest plot visualising the HR (dots) with 95%CI (whiskers) of the terms included in the multivariable Cox proportional hazards models adjusted for sex, age, diabetes, hypertension, heart failure, COPD, type of infective endocarditis and surgical treatment (excluding those not operated despite having a surgical indication). The HR and 95% CI is presented for the overall model (black), males (blue) and females (red). The aHR for female sex in the overall model and the p-values for the interaction of the different terms with sex are presented at the right of the figure. The HR for age is presented per 10-year increase in age. Proportional hazard assumption was violated for the type of infection (native valve endocarditis, prosthetic valve endocarditis, transcatheter aortic valve endocarditis) and the Cox model is stratified for these variables, as such no covariates of this variable are presented. aHR, adjusted HR; CI: Confidence interfval; COPD, chronic obstructive pulmonary disease.

The proportional hazards assumption was violated for the term surgery in males, indicating that the HR is not constant over time (p=0.004), but the assumption for the global test was not violated (p=0.12, online supplemental figure 4).

Discussion

In this study, assessing differences between males and females with IE, several differences in baseline, treatment and outcomes were found. First, there were more males in our cohort, consistent with prior studies, although the underlying reasons remain unclear. The indication for surgery was not different, but males were more often actually operated on, while females were more often treated conservatively despite having an indication for surgery. Finally, in the total cohort, mortality was higher in females, but surgery appeared more protective in females than in males.

Proportion of males and females

The male predominance in our cohort might partially be explained by predisposing conditions for IE in males. The difference in IE location is probably caused by a difference in prior valve pathology, because it is well-known that bicuspid aortic valve is more prevalent in males but rheumatic heart valve disease is more prevalent in women. The higher prevalence of predisposing conditions will at least partly be explained by the lower surgical rate in females with heart valve disease.¹⁴,¹⁸

Microbiological evidence

We found a higher percentage of males with Cutibacterium species as causative pathogen, all of which were cases of prosthetic valve endocarditis. This is in line with another recent study and could be related to the proliferation of Cutibacterium species in sebaceous glands under stimulation of androgens, suggesting that extra attention for skin health in males may be valuable.¹⁹

Treatment

Current literature on sex differences in surgical treatment for IE show controversial results. The results of most studies are in line with our results and show that male patients were more likely to undergo surgery, even after propensity score matching or adjusting for age, comorbidities (including Charlson comorbidity index) and IE-related complications (eg, embolic complications, severe valvular regurgitation, periannular complications, heart failure).^{25 15 20},²² We hypothesised that this difference might be related to greater frailty in females and/or a consequent tendency to refrain from surgery.¹⁵ Frailty was not corrected for in this study, nor in the current literature. Some studies adjusted for selected comorbidities or the Charlson comorbidity index; however, frailty encompasses more than comorbidity alone and is therefore only partially accounted for in these studies. Another suggested explanation for the higher surgical rate in males is the higher proportion of aortic valve involvement and consequent complications requiring surgical treatment in males.^{2 5 21 23} For example, Jordal et al found a lower surgical rate for females in patients with aortic valve IE, but no difference between sexes in mitral valve IE.²⁴ The study performed by Bansal et al found that females were less likely to undergo surgery, and more specifically, females were less likely to undergo surgery for isolated aortic IE or combined aortic and mitral valve IE, but not isolated mitral valve IE.²¹ Finally, Curlier et al adjusted for aortic valve IE and combined aortic and mitral valve IE, and showed that not female sex, but younger age, aortic valve involvement, valve perforation, intracardiac abscess, severe regurgitation and congestive heart failure were associated with surgery.²⁵ Regarding the tendency to refrain from surgery, it is unclear whether this is clinician-derived (eg, risk averse behaviour in women) or patient-derived (eg, patient refusal). In conclusion, the differences could be related to other baseline differences or perhaps be more related to gender than to sex alone. More studies are needed to investigate whether there are indeed male-female differences.

Survival

Previous studies show conflicting results regarding both short-term and long-term mortality between sexes. We found that female sex was independently associated with higher all-cause mortality, while adjusting for baseline characteristics, type of IE, and treatment. Higher short-term adjusted mortality was also found in several other studies,^{15 20 22} and a possible explanation could lie in differences in the immune response in males and females due to differences in sex hormones.²⁶ Whether differences in presentation and perhaps delay in diagnosis play a role is more difficult to investigate.

We did not find an association between surgery and mortality in the overall model (both males and females), but several previous studies showed a protective effect of surgical treatment. However, in most studies, patients treated conservatively despite a surgical indication were assigned to the conservative arm, which may have introduced significant confounding by (contra)indication.^{5 24 27} By excluding these patients in our survival analyses, we may derive less biased estimates.

The observation that females with an indication for surgery are more often treated conservatively, combined with a higher HR for surgery compared with conservative treatment in males than in females, raises the hypothesis that differences in surgical risk acceptability between sexes exist. This interpretation is further supported by a higher absolute 30-day postoperative mortality in males, although this difference did not reach statistical significance, potentially due to limited power. Across healthcare more broadly, males have been shown to undergo more aggressive and invasive treatments than females, often with associated improvements in outcomes in males.²⁸,³¹ In the present setting, however, the opposite pattern seems to surface, with the surgical strategy in males potentially being associated with impaired survival.

In contrast, other studies showed higher mortality after surgical treatment in females compared with males.^{22 25 32 33}

Clinical implications

The proportion of females in IE studies and the global incidence of IE in females are lower, but a clear explanation is still missing.¹ Males more often have a predisposing condition, such as prosthetic heart valves or devices, which might explain the larger proportion of males, but this difference cannot completely explain the observed differences. It remains essential to discuss and highlight the importance of preventive measures for IE to patients with predisposing conditions. To overcome potential underdiagnosis of IE among females, it remains important to carefully assess imaging findings and consider additional cardiac imaging if findings are equivocal. Moreover, physicians should consider IE in females with an infective profile and obtain adequate blood cultures before initiating antibiotic therapy to find a possible causative pathogen.

For males, it is important to be aware that Cutibacterium species are more frequently found, which can be more difficult to diagnose. Those patients often have atypical symptoms, absence of fever, low inflammatory markers and it takes a long time for blood cultures to become positive. Therefore, it is essential to not only collect blood cultures adequately but also to culture them for a sufficient period to allow for the growth of this pathogen.^{19 34}

Additionally, we might have to be more selective for surgical treatment in males, since surgery is associated with higher mortality as opposed to conservative treatment in males compared with females. Thus, careful weighing of risk and benefits of surgery, preferably by discussion of patients in a multidisciplinary ET, remains important to hopefully abate the sex differences in association of surgery with mortality.

Finally, females had worse survival, and based on our findings, we do not have a clear explanation. Various other factors have previously been shown to be associated with mortality, indicating that the mortality risk is a complicated interplay of multiple factors, which justifies careful evaluation by an experienced ET.

Limitations

This study is limited by several factors. There are missing values, but fortunately, they are limited and comprise less than 1%, except for the intended duration of antibiotic therapy (6.1% missing). Our cohort is limited to patients referred to the ET for discussion by their treating physician and therefore does not include all IE patients within the region. Uncomplicated cases or patients with bad prognosis may not be referred, which could result in a difference between the representation of males and females in our cohort compared with the prevalence of IE in the population. Furthermore, the clinical parameters collected in this study did not allow for calculation of the time from symptom onset and admission to the first discussion by the ET, nor allowed comorbidity, frailty or surgical risk score calculation to add to the Cox model for adjustment.

Conclusions

In this large prospective cohort of patients with IE there were striking differences between sexes. First, only 30% of the patients were female. Males had less comorbidity and mitral valve IE, but more predisposing conditions and aortic valve IE. Cutibacterium species were more often the causative pathogen in males and only found in cases with prosthetic valve endocarditis. Females had worse overall adjusted survival, suggesting a less favourable overall prognosis. In terms of surgical decision-making, surgery was withheld more often in females despite indication, and after excluding patients treated conservatively despite a surgical indication, surgery appeared more protective in females than in males. Together, the latter two findings may reflect sex differences in surgical selection.

Data availability statement

Data are available on reasonable request.