Trends in Drug UseAssociated Infective Endocarditis and Heart Valve Surgery, 2007 to 2017

Abstract

Infective endocarditis (IE), an infection of 1 or more heart valves, is a severe complication of injection drug use that may occur through inoculation of bacteria or fungi into the blood or by hematogenous spread from localized infections. Concurrent with the growing number of overdose deaths and rising injection drug use, drug useassociated IE (DUA-IE) has increased markedly in the United States since 2000 (14). The rising incidence of DUA-IE aligns with the rapid growth of the opioid epidemic. During 2000 to 2013, Wurcel and colleagues (1) estimated a greater than 2-fold increase in DUA-IE hospital admissions nationwide, a trend generally seen in other national studies from recent years (2, 5, 6). However, a report from North Carolina for 2010 to 2015 described a 12-fold increase in hospitalizations for DUA-IE, with the sharpest escalation beginning in 2013 (7). The standard of care for IE is a prolonged course of intravenous antibiotics, often accompanied by surgical valve replacement. However, the use of valve surgery for DUA-IE has invoked controversy because of concern regarding postoperative injection drug use and the associated risk for prosthetic valve infection. Despite the lengthy and expensive course of DUA-IE treatment, drug use disorders and harm reduction often go unaddressed during hospitalization, and the nationwide outpatient infrastructure for treating drug use disorders is often inadequate (8, 9). As a result, clinicians have debated the practical and ethical considerations of valve surgery for DUA-IE (1013). No large-scale research has been done recently to examine the rate of valve surgery for DUA-IE or to characterize surgical patients with DUA-IE and their hospital stays. To quantify the impact of rising injection drug use on the application of valve surgery for IE treatment, we examined the case of North Carolina. As of 2016, the state was still seeing an upsurge in overdose deaths, which are increasingly driven by heroin and synthetic narcotics and are disproportionately distributed in certain regions of the state (14, 15). Although our study focuses on 1 state, findings in North Carolina may reflect nationwide trends. Past-year misuse of pain relievers or heroin (16) and the overdose death rate in the state align closely with U.S. averages (17). The health care and political landscape of North Carolina harbors many complexities that have shaped the opioid epidemic, including suboptimal but growing access to opioid treatment and syringe programs as well as a large uninsured population in the wake of Medicaid nonexpansion (14, 1820). In this study, we used statewide data to examine the annual trends, characteristics, and charges related to DUA-IE hospitalizations with valve surgery, and we updated overall DUA-IE hospitalization trends through mid-2017. Our findings may frame national discussions about the medical consequences of drug use disorders, contemporary management of IE, and long-term care after valve replacement in young persons. Methods Study Design and Data We conducted a retrospective study using the North Carolina Hospital Discharge Database, which includes demographic, diagnostic, procedural, and billing data from all short-term, nonfederal, acute care hospitals in the state, covering roughly 1 million hospitalizations yearly. The units of analysis were individual hospitalizations, not unique persons, because personal identifiers were not available for the entire study period. The study population consisted of all North Carolina residents hospitalized for IE from 1 July 2007 to 30 June 2017 who were aged 18 years or older at the time of admission. Operationalization of IE, DUA-IE, and Valve Surgery Hospitalizations for IE were defined by diagnosis codes from the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM), and 10th Revision, Clinical Modification (ICD-10-CM). Cases due to rare pathogens (such as Histoplasma species) were excluded. Among IE cases, DUA-IE was identified by a diagnosis code indicating recreational drug use, withdrawal, dependence, or poisoning, excluding codes for marijuana or unspecified drug use. Persons born after 1965 with a diagnosis of hepatitis C virus (HCV) infection also were coded for DUA-IE, given the high propensity for HCV to be transmitted through recreational injection and the common method of using HCV infection as a criterion in identifying DUA-IE (1, 21). All IE hospitalizations not meeting these criteria were classified as nonDUA-IE. Drugs were classified as opioids (including opiates), cocaine, amphetamines, or other drugs (benzodiazepines, hallucinogens, or unspecified drugs). For all hospitalizations, we used ICD procedure codes to determine whether valve surgery was performed. We generated the list of diagnosis and procedure codes used to operationalize IE, DUA-IE, and valve surgery by reviewing published literature and directly examining ICD-9-CM and ICD-10-CM code lists. We used the General Equivalence Mappings from the Centers for Medicare & Medicaid Services, mapping forward from ICD-9-CM to ICD-10-CM codes (2224). This strategy ensured that we identified all ICD-10-CM codes that correspond (map) to included ICD-9-CM codes. We assessed the appropriateness of including each code (Supplement Tables 1 to 3). Supplement. Supplement Tables Data Analysis For all IE hospitalizations and for those with surgery, we examined annual trends in the aggregate and stratified by drug use. Yearly rates were calculated for the number of IE hospitalizations per 100000 North Carolina residents aged 18 years or older by using state population data obtained from the U.S. Census Bureau. We described the distribution of demographic characteristics (age, race/ethnicity, and sex), insurance payer, valves targeted for surgery, length of stay, disposition location, and hospital charges. Distributions among DUA-IE and nonDUA-IE groups were compared by using Wilcoxon rank-sum tests for continuous variables and 2 tests for categorical variables, with statistical significance defined as a P value less than 0.010. To evaluate the influence of our assumption that HCV-infected patients born after 1965 acquired IE through injection drug use despite having no specific drug-related diagnosis, we conducted a sensitivity analysis restricting our DUA-IE definition to include only hospitalizations with a diagnosis code specific to drug use. Analyses were performed with SAS, version 9.4 (SAS Institute). The study was approved by the Institutional Review Board at the University of North Carolina at Chapel Hill. Role of the Funding Source No agency funding this work had any role in design, data collection, analysis, interpretation, authorship, review, or the decision to submit the findings for publication. Results Annual Trends of All IE Hospitalizations and IE Hospitalizations With Valve Surgery During the 10-year period, 22825 IE hospitalizations occurred in North Carolina. Of these, 2602 (11%) were for DUA-IE and 20223 (89%) for nonDUA-IE (Table 1). Valve surgery was performed in 1655 (7%) of all IE hospitalizations. Of hospitalizations with surgery, 285 (17%) were for DUA-IE and 1370 (83%) for nonDUA-IE (Table 2). Table 1. Characteristics of Hospitalizations for IE in North Carolina, by Drug Use Status, 2007 to 2017* Table 2. Characteristics of Hospitalizations for IE Treated With Valve Surgery in North Carolina, by Drug Use Status, 2007 to 2017* Annual admissions for IE increased during the study period, from 1936 to 2992, corresponding to 27.74 per 100000 North Carolina residents in the first year and 37.80 per 100000 in the final year (Figure 1). Annual hospitalizations for DUA-IE rose from 64 (0.92 per 100000) in 2007 to 2008 to 867 (10.95 per 100000) in 2016 to 2017, an increase of approximately 12-fold. The number of admissions for nonDUA-IE was generally stable across the period (26.84 per 100000 in the final year), although an upsurge was observed in 2015 to 2016. Figure 1. Rates of hospitalization for IE in North Carolina, by drug use status, 2007 to 2017. Rates are hospitalizations per 100000 North Carolina residents aged 18 y or older. North Carolina population is based on U.S. Census Bureau estimates of residents aged 18 y or older on 1 July. Because annual estimates in this study incorporated 2 calendar years (1 July to 30 June), the average of the 2 y was used as the population denominator. DUA-IE = drug useassociated infective endocarditis; IE = infective endocarditis; nonDUA-IE = infective endocarditis not associated with drug use. Annual hospitalizations with valve surgery for all IE cases increased from 1.62 per 100000 residents in 2007 to 2008 to 3.26 per 100000 in 2016 to 2017 (Figure 2). With regard to DUA-IE hospitalizations with surgery, fewer than 10 occurred per year until mid-2013 (representing 0.01 to 0.10 per 100000). Annual hospitalizations then rose to 109 (1.38 per 100000) by 2016 to 2017, an overall increase of 13-fold. NonDUA-IE hospitalizations with surgery increased from 106 (1.52 per 100000) in the first year to 149 (1.88 per 100000) in the last year. In the final year, DUA-IE accounted for 42% of all IE hospitalizations in which surgery was performed (109 of 258 hospitalizations) (Supplement Tables 4 and 5). Figure 2. Rates of hospitalization for IE in North Carolina treated with valve surgery, by drug use status, 2007 to 2017. Rates are hospitalizations per 100000 North Carolina residents aged 18 y or older. North Carolina population is based on U.S. Census Bureau estimates of residents aged 18 y or older on 1 July. Because annual estimates in this study incorporated 2 calendar years (1 July to 30 June), the average of the 2 y was used as the population denominator. DUA-IE = drug useassociated infective endocarditis; IE = infective endocarditis; nonDUA-IE = infective endocarditis not associated with drug use. Hospitalizations for DUA-IE increased across the study period for all drug cate