Poor long-term outcomes of intravenous drug users with infectious endocarditis

Objectives The optimal management of active endocarditis in intravenous (IV) drug users is still lacking. Methods From the years 1997 to 2017, 536 patients with active infectious endocarditis were surgically treated, including 83 (15%) with IV drug use (IVDU) and 453 (85%) without IV drug use (non-IVDU). Initial data were obtained from the Society of Thoracic Surgeons database and supplemented with chart review and national death index data. Results The IVDU group was significantly younger (43 vs 56 years old) than the non-IVDU group and had greater rates of psychiatric disorders, drug use, and tricuspid valve endocarditis (28% vs 8.6%). Hypertension, dyslipidemia, and diabetes mellitus were significantly more common in the non-IVDU group. Perioperative complications and operative mortality (7.2% vs 7.9%) were similar. IVDU was not a significant risk factor for operative mortality. Kaplan–Meier survival was significantly lower in the IVDU group (5-year survival, 46% vs 67%). Significant risk factors for long-time mortality included IV drug use (hazard ratio [HR], 1.92), age ≥65 years (HR, 1.78), congestive heart failure (HR, 1.87), and enterococcus endocarditis (HR, 1.54). The 5-year rate of reoperation was similar between IVDU and non-IVDU groups (2.4% vs 2.7%). Conclusions IVDU is a significant risk factor for long-term mortality. A multidisciplinary approach was preferred for IVDU patients to treat both endocarditis and substance use disorder and improve long-term survival.

Conclusions: IVDU is a significant risk factor for long-term mortality. A multidisciplinary approach was preferred for IVDU patients to treat both endocarditis and substance use disorder and improve long-term survival. No-IVDU IVDU 44 26 22 13 Survival of IVDU and non-IVDU patients after surgery.

CENTRAL MESSAGE
Endocarditis in IVDU requires multidisciplinary treatment. Surgery alone is not sufficient.

PERSPECTIVE
Although IV drug usage as the etiology of endocarditis did not increase the incidence of shortterm complications in patients with active infectious endocarditis compared with non-IVDU, the long-term survival was very poor due to addiction as a second disease. Surgery alone may not be adequate for this patient population.
Video clip is available online.
Active infective endocarditis is a disease process associated with increased morbidity and mortality. Of patients with acute endocarditis, 25% to 30% require surgical intervention, with a noted surgical mortality of 8% to 16%. 1,2 The incidence of infectious endocarditis has been on the rise, increasing from 11 to 15 cases per 100,000 people between 2000 and 2011. [3][4][5] This trend has continued despite the decline in rheumatic fever cases, a predominant cause of infectious endocarditis, as well as improvements in health care. An increase in intravenous drug use (IVDU) in the setting of the opioid epidemic is a potential contributing factor to the increase in infectious endocarditis.
The sale of opioid pain relievers increased by 4 times between the years 1999 and 2010, as noted by the Center of Disease Control and Prevention. 6 Kadri and colleagues 7 recently reported a significantly increased incidence of infectious endocarditis due to IVDU between 2002 and 2016 with an annual percent chance of 1.8%. Despite this alarming increase, the current literature regarding long-term outcomes for IVDU compared with non-IVDU is conflicting. Furthermore, studies have demonstrated similar survival between the 2 groups whereas others have demonstrated poor long-term outcomes in IVDU. 5,8,9 As such, we examined the short-and long-term outcomes of IVDU and non-IVDU patients who underwent surgical intervention for infectious endocarditis at our institution. We hypothesized that intravenous (IV) drug users have similar short-and long-term outcomes.

METHODS
This study was approved by the institutional review board (HUM0014297; March 31, 2018), and a waiver of informed consent was obtained.

Patient Selection
Between the years 1997 and 2017, 536 patients with active endocarditis underwent surgical management at our institution. Patients who were treated medically or turned down for surgery were excluded. Endocarditis involvement for these patients included the aortic valve, mitral valve, tricuspid valve, or multiple valves. These patients were further divided based on their history of IVDU. Of the 536 patients, 83 (15%) had a history of IVDU and 453 (85%) were non-IV drug users. Patients in the IVDU group were all IV drug users, but it was difficult to ascertain which patients were actively using IV drugs at the time of their surgery. IVDU was not limited by heroin use-some patients injected cocaine, methamphetamine, or were unspecified drug users.

Data Collection
Data elements were obtained from the Society of Thoracic Surgeons from our institution's Cardiac Surgery Data Warehouse to determine pre-, intra-, and postoperative characteristics. Investigators supplemented data collection through a review of medical records and operative reports using traditional chart review as well as the Electronic Medical Record Search Engine data-retrieval tool. 10 Survival data were obtained through the National Death Index database through December 12, 2021, and supplemented with medical record review and questionnaire response via mailed survey or phone call through February 2019. 11 Reoperation events included reoperation of prosthesis for any reason and for reinfection.

Statistical Analysis
Data are presented as median (interquartile range) for continuous data and n (%) for categorical data. Univariate comparisons between IVDU and non-IVDU were performed using c 2 tests for categorical data. The Kolmogorov-Smirnov test was used to test the normality of the data. Univariate comparisons between groups were performed using the Mann-Whitney U test for continuous data. Multivariable logistic regression was used to calculate the odds ratios of risk factors for operative mortality. The risk factors included group, age, sex, cirrhosis, congestive heart failure (CHF), cardiogenic shock, and preoperative sepsis. Cox proportional hazard regression was performed to calculate the hazard ratio (HR) for late mortality including group, age 65 years, sex, CHF, enterococcus endocarditis, cirrhosis, and preoperative sepsis. The C-statistic for the whole population model was 0.64. All variables passed the proportional hazards assumption for this model. For the proportional hazards regression model within IV drug uses, the C-statistic was 0.66. The Enterococcus variable was used as a strata variable as it did not pass the proportional hazards assumption. A Kaplan-Meier survival curve was used to assess long-term survival using time to death since valve operation due to endocarditis. The log-rank test was used to compare survival between groups (IVDU and non-IVDU). Cumulative incidence function curves for reoperation were generated adjusting for the competing risk of death. The Gray test was used to assess the difference in the cumulative incidence function curves between groups. Statistical calculations were performed using SAS (SAS Institute), SPSS, Version 28 (IBM Corp), and Quickcalcs (GraphPad).

Operative Data
The IVDU group had significantly greater rates of tricuspid valve repair (25% vs 15%; P ¼ .02) and

Abbreviations and Acronyms
CHF ¼ congestive heart failure HR ¼ hazard ratio IVDU ¼ intravenous drug use MET ¼ multidisciplinary endocarditis team replacement (16% vs 3.1%, P < .001) but similar rates of other valve procedures compared with the non-IVDU group. The IVDU group had significantly fewer ascending aorta and aortic root procedures but similar rates of coronary artery bypass grafting, hemiarch, and total arch procedures. Cardiopulmonary bypass and crossclamp times were also similar between groups (Table 3).

Perioperative Outcomes
IVDU patients had a lower rate of postoperative atrial fibrillation (17% vs 27%; P ¼ .04) and a greater rate of postoperative pacemakers (12% vs 5.5%; P ¼ .03) compared with non-IVDU patients. Nevertheless, there were no significant differences in other perioperative complications or mortality (operative mortality: 7.2% vs 7.9%) between groups. ICU stays, ventilation hours, and red blood cell units given were also similar between IVDU and non-IVDU groups (Table 4). IVDU was not a significant risk factor for operative mortality with an odds ratio of 1.02, P ¼ .97 (Table 5).

DISCUSSION
In this study, we found that IVDU did not significantly increase short-term complications or operative mortality in patients with infective endocarditis. The greater rate of postoperative atrial fibrillation in the non-IVDU group was likely related to the older age of the patients and the greater rates of comorbidities in this group. The cumulative incidence of reoperation for endocarditis was also similar between groups. Nevertheless, there was significantly lower long-term survival associated with IVDU on Kapan Meier analysis ( Figure 1). IVDU (HR, 1.92) was a significant risk factor for long-term mortality. Additional risk factors for late mortality included age 65 years, CHF, and E endocarditis.
The IVDU group had significantly more psychiatric disorders and preoperative embolic events (Tables 1 and 2). IV drug users were more likely to have anxiety, bipolar disorder, as well as additional substance use including alcohol, heroin, cocaine, pain medications, methamphetamines, or other drugs. A meta-analysis performed by Conner and colleagues 12 demonstrated a positive association between intravenous drug use and depression, supporting the idea that IV drug abuse may be a complication of a patient's psychiatric conditions. Given the IVaccess for drug use in these patients, it was expected for IVDU patients to have more tricuspid valve endocarditis as shown in our study (Table  2) and other studies. 13 If a physician encounters a patient    valves. When patch repair of the valve or aortic root was required, autologous pericardium was preferred to minimize prosthetic material or bovine pericardium. If the valve required replacement, a bioprosthetic valve or a mechanical valve was used for patients. For tricuspid valve, we did not use tricuspid valvectomy for those patients. We either repaired or replaced the tricuspid valve.
With regards to perioperative outcomes, our study did not show any significant difference in operative mortality between the IVDU and non-IVDU groups. This is in concordance with a large meta-analysis, which did not show any significant difference in all-cause perioperative mortality between IVDU and non-IDVU groups. 14 A large retrospective study with more than 34,000 patients showed lower unadjusted operative mortality among IV drug users, with greater risk of major morbidity after adjusting for other factors. 15 Nevertheless, the adjusted odds ratio for operative mortality was not significant. The operative mortality in this study was 10.7% in the non-IVDU group compared with 6.7% in the IVDU group. Although our sample size was smaller, our operative mortality was similar (7.9% vs 7.2%). It is likely that the lack of difference in perioperative mortality between groups is because IV drug users were significantly younger and had fewer comorbidities than their non-IVDU counterparts.
Consistent with multiple studies, the long-term survival of our patients after surgical intervention was much worse in IVDU patients compared with non-IVDU patients. 5,9,16 One-half of the IVDU patients died 4 years after surgery ( Figure 1). Based on the entire cohort, risk factors for long-term mortality in our study were IVDU, age   67%; 95% CI, 63%-71%). therefore be cautious to offer surgery to older patients and patients with active infectious endocarditis and CHF. E endocarditis should also be treated aggressively, as it has been shown to have the greatest 1-year mortality along with Staphylococcus aureus according to a study by Shah and colleagues. 17 Among IV drug users with infective endocarditis, we should also be prepared to offer holistic care. A recent study found that the leading cause of death for IVDU patients with endocarditis after surgery was recidivism compared with malignancy, cardiopulmonary, and kidney disease for the non-IVDU. 5 As surgeons, we can treat endocarditis in IV drug users with fairly good perioperative results comparable with non-IV drug users. However, if we just operated on those patients and left the psychiatric disorder and substance use disorder untreated, we would not be treating these patients adequately.
Interestingly, among these IV drug users, long-term survivors were younger while long-term nonsurvivors of IVDU patients had significantly greater rates of CHF (Tables E1  and E2). There were more triscuspid valve replacements in survivors, but post-operative outcomes were similar between survivors and non-survivors (Tables E3 and E4). The significant risk factors for late death within the IVDU group included age, female sex, and CHF, suggesting that we should be cautious in offering surgery to older female IVDU patients, especially if they have CHF (Table 7).
In 2018, the Multidisciplinary Endocarditis Team (MET), a collaborative environment for the discussion and management of every endocarditis case, was established at our institution. A multidisciplinary approach was used to treat patients in the inpatient and outpatient setting with the help of social workers, pharmacists, and physicians from infectious disease, cardiology, cardiac surgery, and neurology. Indications for surgery included but were not limited to (1) valve dysfunction resulting in heart failure; (2) left-sided endocarditis caused by highly resistant microorganisms; (3) the presence of heart block, root abscess, or destructive lesions; (4) persistent or fever despite multiple days of appropriate antibiotics; (5) and large or persistent vegetations. 18,19 In terms of structure, patients were typically discussed before surgery, but also discussed again if any problems arose. Postoperatively, patients were in the cardiovascular intensive care unit until they were stabilized and then transferred to a step-down medicine service. The patients were then followed by cardiac surgery and any of the other medical specialties as needed. This change led to a substantial impact on the management (diagnosis, treatment, communication, mortality) of all endocarditis patients as evidenced by a significant decrease in in-hospital mortality from 29.4% to 7.1% between 2014 and 2019 at the University of Michigan, although patients in hospice were not included. 20,21 Furthermore, Chirillo and colleagues 22 reported a 2-fold decrease in in-hospital mortality after implementation of their multidisciplinary protocol.
Given the complexity of many of these patients, multidisciplinary teams are essential for successful management of the short-and long-term survival of these patients both in the in-patient and out-patient settings. This is particularly important in patients with IVDU, who have greater rates of readmission and disease recurrence. 23 Although psychiatry and addiction medicine are not part of our core The variables in bold are statistically significant. IV, Intravenous; CHF, congestive heart failure. The variables in bold are statistically significant. CHF, Congestive heart failure. *Enterococcus as a causative organism was set as a strata for model. multidisciplinary endocarditis conference attendance, our colleagues recommend attendance of these teams as part of treatment in patients with substance abuse disorder, complicating the diagnosis of endocarditis as in the case of many IVDU patients. 24 We have both of these teams at our institution, and it is the responsibility of the primary care team to make the appropriate consults if necessary.
As part of the multidisciplinary endocarditis team, it would be beneficial to make this part of the recommendations as indicated for these patients. There should also be a movement to enhance training in addiction within infectious disease as to provide an additional perspective in the multidisciplinary conference when a formal addiction team is not available. 25 The cumulative incidence of reoperation was overall similar between the groups (Figure 2). The relatively low reoperation rate supported our choice of bioprosthesis for patients who needed a valve replacement. Kaiser and colleagues 8 noted greater rates of reoperation in IV drug users as compared with non-IV drug users, likely from reinfection. However, the reoperation rates between the IVDU group and the non-IVDU group were comparable in our study, which was likely obscured by the high mortality of the IVDU group. By operating on patients with recurrent endocarditis due to relapse of IVDU, we are not targeting the underlying cause and contributing to additional morbidity with each reoperation for the same disease. Therefore, before operating on these patients, we emphasize that our group does not perform any additional surgeries related to relapse from IVDU. This could also be a contributor to the low reoperation rate. VIDEO 1. While IV drug use does not appear to affect perioperative mortality in patients with active infective endocarditis, it is a significant risk for long-term mortality. A multidisciplinary team is essential to mitigate this risk. Video available at: https://www.jtcvs.org/article/S2666-2736 (22) 00239-X/fulltext.

Outcomes of Intravenous Drug use in Patients with Infectious Endocarditis
Methods Implications 1. IV drug use is a significant risk factor for long-term mortality. 2. Multidisciplinary team is essential in improving longterm survival. FIGURE 3. IV drug use is a significant risk factor for long-term mortality. A multidisciplinary approach is essential for patients with IV drug use to treat both endocarditis and substance use disorder and improve long-term survival. IV, Intravenous; IVDU, intravenous drug users.
Nevertheless, if patients developed prosthetic valve endocarditis without relapse of IVDU, we performed a second operation. If they developed prosthetic valve endocarditis due to relapse of IVDU, most surgeons did not perform a reoperation, especially in patients without good social support. We also discuss these cases with the MET. In such cases, it would be beneficial to include the medical ethics team in this discussion, which is available at our institution. Our MET's recommendations have led to surgery when an operation was not planned, indicating that these discussions can change management. Additionally, we now focus on providing patients with multidisciplinary care to prevent recidivism and therefore reinfection.
In summary, surgical intervention alone for IVDU patients is not enough. More needs to be done to combat the psychosocial comorbidities that exist in these patients. With surgery, we help alleviate the active infectious process and prevent poor short-term outcomes. However, in the long-run, proper treatment of underlying psychiatric disorders, and substance use is key for successfully decreasing morbidity and mortality in patients with IVDU. Furthermore, a standardized multidisciplinary approach may also be beneficial. 20,24,26 As such, guidelines highlighting surgical management in IV drug users for first-time and recurrent disease, including recommendations for substance use interventions and multidisciplinary team-based care, should continue to be developed (Video 1).
This study is a single-center experience and is therefore limited in breadth. It also has all the limitations of a retrospective study. We had a large sample of patients with endocarditis but had a relatively small sample size of patients with IVDU. Our survey response rate was not 100% and could have underestimated the cumulative incidence of reoperation. The study was also limited by the possibility of Type II error. Nevertheless, our Cox model was able to provide us with enough sensitivity to show the considerable effect of IVDU on long-term mortality in patients with endocarditis.