Increasing adult CHD infective endocarditis hospitalisations

We identified a 125% increase in indexed infective endocarditis hospitalisations for the adult CHD population from 2001 to 2016, a trend similar to previously reported National Inpatient Sample data describing an increase in all-cause adult CHD hospitalisations over a similar time period. ^{Reference Agarwal, Sud and Menon24} It is important to note that this trend does not necessarily signify an increasing rate of infective endocarditis among adults with CHD, but rather may be more attributable to a growing adult CHD population that now comprises a larger relative percentage of the overall United States population. ^{Reference Marelli, Ionescu-Ittu, Mackie, Guo, Dendukuri and Kaouache1,Reference Gilboa, Devine and Kucik27} This finding highlights the importance of specialised providers and qualified medical centres familiar with the management of infective endocarditis among adult CHD populations. Notably, referral to specialised adult CHD centres has been associated with a significant reduction in mortality for this complex group of patients. ^{Reference Mylotte, Pilote and Ionescu-Ittu28}

Outcome differences and contributing factors

Comparatively lower in-hospital mortality related to infective endocarditis was identified in adults with CHD compared to the general population in the United States, offering an American perspective that demonstrates similar outcomes as recent studies from Europe. ^{Reference Maser, Freisinger and Bronstein29,Reference van Melle, Roos-Hesselink and Bansal30} We believe this mortality difference is likely multifactorial, though perhaps one key factor is the difference in microbiologic aetiology in patients with CHD. Streptococcus, which has been consistently identified as the leading causative pathogen of infective endocarditis in the CHD population, ^{Reference Moore, Cao, Kotchetkova and Celermajer4,Reference Dolgner, Arya, Kronman and Chan17,Reference Cahill, Jewell and Denne31} was also the most commonly identified cause of infective endocarditis in our adult CHD cohort. This is perhaps related to an increased burden of dysplastic and/or prosthetic valvular tissue in adults with CHD that increases the likelihood of transient bacteraemia from oral flora to adhere and proliferate. In contrast, Staphylococcus was the most commonly isolated organism in our cohort of adults without CHD, and it has also previously been demonstrated to predominate in patients with infective endocarditis and structurally normal hearts in the general population. ^{Reference Bor, Woolhandler, Nardin, Brusch and Himmelstein8,Reference Cabell, Jollis and Peterson10,Reference Pant, Patel and Deshmukh20,Reference Murdoch, Corey and Hoen32} Historically, infective endocarditis outcome parallels microbiologic aetiology. Streptococcus is associated with lower in-hospital mortality in patients with structurally normal hearts, ^{Reference Murdoch, Corey and Hoen32} and similar favourable outcomes have been identified when Streptococcus is isolated in CHD patients with infective endocarditis following Melody valve placement. ^{Reference Abdelghani, Nassif and Blom33} Conversely, Staphylococcus has been associated with increased infective endocarditis related complications and mortality in both populations. ^{Reference Cabell, Jollis and Peterson10,Reference Cahill, Jewell and Denne31,Reference Murdoch, Corey and Hoen32,Reference Mir, Uddin, Qureshi, Regmi, Tleyjeh and Saydain34–Reference Kim, Lee and Ku37} In our study, Staphylococcus was the most commonly isolated organism in patients who died in either group and was particularly prominent in patients without CHD who experienced mortality, where it was identified over three times as frequently as Streptococcus.

As has been previously argued, ^{Reference Tutarel, Alonso-Gonzalez and Montanaro5,Reference Maser, Freisinger and Bronstein29} early detection and intervention likely also play a significant role in the infective endocarditis adult CHD mortality difference. In a study of infective endocarditis in the general population, Thuny et al. ^{Reference Thuny, Di Salvo and Belliard9} postulated that high mortality rates may be due to delayed identification and late referral to institutions experienced in infective endocarditis management including capabilities of early surgery. In patients with CHD, infective endocarditis is often an early consideration with any indication of fever or with the onset of other systemic symptoms, particularly in the context of changing cardiac valve function.

Given risk factors of left-sided infective endocarditis such as thromboembolism and heart failure, we additionally considered the impact of left- versus right-sided infective endocarditis involvement on mortality. The large majority of infective endocarditis was left-sided in patients with and without CHD, which is similar to previously reported data. ^{Reference van Melle, Roos-Hesselink and Bansal30} Given our sidedness analysis was limited to patients undergoing surgical valve intervention, this data likely reflects a proclivity to surgically treat infective endocarditis that is left-sided. Infective endocarditis sidedness was similar between each group with only a modest difference in right-sided involvement driving statistical significance and is unlikely to meaningfully impact overall mortality. Compared to right-sided infective endocarditis, left-sided disease conferred an increase in mortality in the structurally normal group, likely related to age and potentially due to delayed detection resulting in development of complications associated with left-sided infective endocarditis as previously noted.

The United States opioid epidemic imparted considerable influence during the time period of our study, though its impact related to infective endocarditis affected both adults with and without CHD similarly given comparable rates in each group. In adults with CHD, drug abuse also did not convey significance in infective endocarditis outcome as no differences in mortality or surgery were demonstrated related to drug abuse.

The difference in rates of surgery during an infective endocarditis hospitalisation between adults with and without CHD is notable and may suggest a predilection for surgical management of infective endocarditis in CHD patients. Surgery for infective endocarditis in adults with CHD in our study occurred in nearly one third of patients, which is similar to previously reported surgical rates for adult CHD infective endocarditis. ^{Reference Tutarel, Alonso-Gonzalez and Montanaro5,Reference Maser, Freisinger and Bronstein29} Those with transcatheter Melody valves with infective endocarditis have even higher reported rates of surgical intervention (44%). ^{Reference Davtyan, Guyon and El-Sabrout38} One possible explanation for this management paradigm is the necessity for a more aggressive approach in instances of haemodynamically compromising infective endocarditis complications such as right ventricular outflow tract obstruction within conduits or transcatheter pulmonary valves. ^{Reference Abdelghani, Nassif and Blom33,Reference Davtyan, Guyon and El-Sabrout38} Another possibility is the influence of surgical familiarity for the management of patients with CHD. Institutions accustomed to frequent application of surgical or transcatheter intervention for patients with CHD may also be more apt to consider a surgical approach for complications such as infective endocarditis.

Adult CHD infective endocarditis sociodemographic outcome disparities

Significant sex and insurance differences were identified within the adult CHD infective endocarditis group, adding to the growing number of outcome disparities among vulnerable patient populations. In our study, women with adult CHD and infective endocarditis were less likely to undergo surgery. Our data is consistent with the literature, ^{Reference Bansal, Cremer, Jaber, Rampersad and Menon15,Reference Sambola, Fernandez-Hidalgo and Almirante16} which has identified lower rates of valve surgery in women with structurally normal hearts and infective endocarditis despite improved in-hospital and 1 year mortality for those who are surgically managed. One possible explanation for these findings might be an avoidance of surgery in women who are pregnant when diagnosed with infective endocarditis.

In our adult CHD cohort, those with governmental insurance had significantly higher mortality rates. Similar outcomes in insurance disparities have been identified in the paediatric population following congenital heart surgery. ^{Reference Chan, Pinto and Bratton39} As the adult CHD population continues to grow, ensuring access to care is paramount. National health policies such as the Affordable Care Act lower uninsured rates in adults with CHD, ^{Reference Salciccioli, Salemi, Broda and Lopez40} demonstrating one strategy of reducing underinsurance as a barrier to care. Additional system-level mechanisms of reducing disparity within heart centres include social determinants of health screening to assist in understanding access to medications and adult CHD physicians, broad implementation of implicit bias training to ensure equal recognition and treatment of all patients with infective endocarditis regardless of sociodemographic characteristics, centre-based quality improvement initiatives including timely and adequate imaging studies, as well as increasing awareness and education of the signs and symptoms of infective endocarditis, particularly to lower socio-economic status communities.

Limitations

Despite the large size of our overall population derived from utilisation of a national database from 2001 to 2016, this study has several limitations. Inherent limitations exist with retrospective data collection which can only provide association as opposed to causality. The deidentified nature of the National Inpatient Sample database creates the possibility that some infective endocarditis hospitalisations are readmissions as opposed to newly diagnosed cases, potentially overestimating the true incidence of infective endocarditis in the United States.

Generalisability within this dataset is dependent on reliable International Classification of Diseases coding, which may not be consistent across institutions. Coding errors have been identified in CHD, ^{Reference Rodriguez, Ephrem, Gerardin, Raskind-Hood, Hogue and Book26} and the current dataset may include inaccuracies in regard to presence of CHD and congenial heart disease complexity. For instance, the code for “non-rheumatic mitral valve insufficiency” may be utilised if the valve pathology is due to acquired valve degeneration or from a congenital aetiology, and non-trivial overlap likely exists. This limits our ability to identify patients, particularly in the non-adult CHD group, that may be at higher risk for mortality or surgery related to infective endocarditis such as those with acquired degenerative valve disease. Furthermore, International Classification of Diseases codes for history of prosthetic valve disease are likely under-coded in patients with adult CHD, limiting the comparative utility of this variable between the non-adult CHD and adult CHD groups.

An additional limitation of International Classification of Diseases coding includes an inability to identify less common microbiologic organisms such as the Haemophilus, Aggregatibacter, Cardiobacterium, Eikenella, and Kingella group. Such a limitation has been acknowledged previously ^{Reference Dolgner, Arya, Kronman and Chan17} and may account for a larger proportion of patients with unknown microbiologic aetiologies in our cohort. Given the heterogeneity of adult CHD anatomy, valve specificity does not always accurately denote a left-sided mitral and aortic valve and right-sided tricuspid and pulmonary valve, limiting our sidedness comparison and its impact on mortality. Inclusion of analyses to determine the impact of genetic syndromes such as 22q11 deletion was considered; however, we believe under-coding and/or inconsistent coding of such inherited disorders during inpatient admissions likely limits the reliability of this data. Institutional variability in the management of infective endocarditis such as empiric antimicrobial selection and varying thresholds for surgical intervention may affect outcome. Finally, complexity of surgical intervention may vary from simple valve debridement to more complex valve or conduit resections/replacements and could impact surgical outcome.