Increasing incidence of IV‐drug use associated endocarditis in southern West Virginia and potential economic impact

Background The opioid crisis has disproportionally affected Appalachia. One of the potentially lethal and costly complications associated with IV drug use is infective endocarditis (IE). The goal of this study was to assess the trend and costs of substance abuse associated IE admissions in Southern West Virginia. Methods This is a retrospective analysis of cost, incidence, and geographic patterns of all patients admitted over the last decade with concomitant drug abuse (cocaine, amphetamine, sedative, and other/mixed drug abuse) and IE in the largest tertiary care center for Southern West Virginia. A time series model was used to investigate the effect of drug use on the incidence of IE. Results A total of 462 patients were hospitalized with IE and concomitant illicit drug use. IE cases increased from 26 admissions in 2008 to 66 in 2015. Patterns of increases in mixed drug use (DRG most often associated with IV drug use in our center) mirrored increases in IE (P = 0.001). From 2008 to 2015, the total hospital charges were $17 306 464 on 462 cases of illicit drug associated IE. Only a fraction of the billed fees (22%) was collected ($3 829 701). Conclusions The number of patients hospitalized with IE has dramatically increased over the last decade in a pattern that mirrors the increase in mixed drug use. The majority of payers were from underfunded state programs or private pay and thus, only 22% of the hospital charges were paid, leaving a hospital deficit of over $13 476 763 during the study period.


| INTRODUCTION
Over the last decade, there has been a shift in the geographic patterns of intravenous drug use (IVDU) in the United States from focal urban locations to rural areas. 1,2 There are various socioeconomic factors that have influenced these patterns alongside more liberal opioid prescription practices. 3 The shifting burden of IVDU to more rural areas has created unique challenges for patient access (medical attention, addiction education, and rehabilitation), as well as an avalanche of additional costs for hospitals located within the endemic areas. Coinciding with the escalation of drug-related socioeconomic burden, the rural hospitals, and tertiary referral centers in Appalachia are caring for an increasing density of uninsured and underinsured patients. 4,5 In addition, poor sterile technique used by IVDUs and immune compromise from dietary deficiencies in drug abusers has caused an increase of hepatitis C and other related illnesses with dramatic loss of productivity, reduced long-term quality of life, and increased healthcare costs. 6,7 The Appalachian syndemic of the opioid crisis, worsening hepatitis C, recent human immunodeficiency virus (HIV) outbreak and economic challenges is punctuated by recent data suggesting West Virginia now leads the nation in mortality from drug overdoses (35.5 deaths per 100 000). 1 It is well known that poor sterile technique and repeated interruption of the skin/blood barrier have been implicated in IVDU associated infective endocarditis (IE) (IVDU-IE) that most frequently involves the tricuspid valve. 8 In addition, the use of unsterilized needles, bacteria in the injectate (drug, diluents, and fillers), can result in the introduction of bacteria or fungi directly into the bloodstream. These patients are also prone to developing immune compromise because of associated hepatitis, poor nutrition, and poor dentition. IVDUs also have higher rates of nasal and cutaneous colonization with Staphylococcus aureus than patients who use illicit drugs by the oral route exclusively. 9 S. aureus is the most common infectious agent in these patients and IE is on the rise within the United States. 10 Studies have analyzed differences in the effect of drug abuse on endocarditis, such as age, outcome, healthcare cost, and the patterns of geographical abuse. 11 It is clear that IVDUs are at an increased risk for IE, have a higher incidence of recurrent IE and higher expected mortality than those who develop IE without associated drug addiction. 12 These patients often require high-acuity hospital resources and prolonged intensive care. 9 There are also long recovery times from valve surgery, if needed. In addition, IVDUs-IE often seek delayed medical attention which poses significant challenges to emergency departments and front-line physicians. 11 At the same time that collateral healthcare costs from the drug epidemic have been increasing, the economic downturn, and contraction of revenue from the energy sector have negatively impacted the rural healthcare systems in Appalachia amplifying the economic impact on hospitals and healthcare providers. 13,14 The crisis created by the collision of rural economic instability and the drug epidemic in Appalachia has stimulated much discussion on needed policy reform. [13][14][15][16][17][18] Our center is the largest cardiovascular tertiary referral center in southern West Virginia and resides in the epicenter of the Appalachian opiate epidemic and energy sector decline associated economic collapse. The center has seen alarming increases in IE and we hypothesized that this was due to increases in IVDU-IE. Since IE is often very difficult to treat and consumes significant hospital resources, we set out to quantify changes in IE hospitalization patterns and associated costs in southern West Virginia. 15 The data variables obtained from the CAMC database included: Number of cases of patients hospitalized with IVDU and IE and patient variables including time of hospitalization, Zip codes, type of drug use (opiates, cocaine, amphetamines, sedatives, and mixed drug use), gender, age, race, and insurance status. Primary payers were identified by federal, state, and private payers and the study further subdivided all primary payers. We obtained detailed analysis of hospital charges to better understand the current annual economic burden for our center and federal/state funding agencies.

| STATISTICAL ANALYSIS
We created a time series model in STATA 11.4 to investigate the effect of drug use on charged cases of endocarditis. A time series model was created for hospital visits involving opiate, cocaine, amphetamine, sedative, and mixed drug abuse. A multiple variable time series explored the relationship between the one data set and assessed the potential relationship between certain drug types including opiates, cocaine, amphetamine, sedative, and mixed drug abuse with the incidence of endocarditis. To create differing maps and understanding the effect of cases on regions of the state of West Virginia, Quantum GIS was used to examine these regions and to create a hot spot analysis. A hot spot analysis was constructed to understand potential areas of concern that may benefit from further investigation.

Equation #1
Endocarditis (patients with drug-related case The exposure variables contain opiate, cocaine, amphetamine, sedative, and mixed drug abuse. The subscript "t" refers to the year sequence starting 2008 to 2015 for t − 1. Next, the time series model assumptions were verified with standard diagnostic tests which included (normal distribution, autocorrelation and partial autocorrelation, temporal autocorrelation, and Akaike's and Bayesian information criterion). A histogram was created to confirm that residuals were normally distributed of the five exposure variables. Opiates were normally distributed as well as cocaine and amphetamines. The sedative variable had a left-skewed distribution as well as the mixed drug use exposure variable. To normalize the skew, a Shapiro-Wilk test was used to transform the variables into a normal distribution. Autocorrelation and partial autocorrelation function did verify that five models and exposures did not contain significant temporal autocorrelation.
The final test was an Akaike's and Bayesian information criterion to ensure the best fitting exposure metric and corresponding time series model. Only patients with a drug-related case were used within this analysis this attempts to screen out potential non-cases.

| RESULTS
The demographics of the study population are shown in Table 1.
Through the years there was a significant increase in admissions with any type of drug use as well as associated endocarditis as illustrated in Table 2. Increases in mixed drug use strongly correlated with increased cases of IE during the study period (P = 0.001). This suggests that for every 100 cases of mixed drug use there is a 0.06% increase in relative risk of endocarditis. This strong relationship can further be investigated The overall state health metric variable of household income was not significantly associated with the increase in endocarditis (P = 0.176). The payer mix for the study cohort is detailed in Table 3 alongside all the hospital charges vs the actual payment received. Pri- 14% of all IE patients to 56% from 2009 to 2014. 16,22 In addition, other studies have shown that serious infections associated with IVDU have also increased significantly with associated costs in 2012 of $700 million in the United States. 23,24 It should be noted that the increase in IVDU-IE is not isolated to Appalachia with less dramatic trends also reported in differing regions of the country. [22][23][24][25][26][27][28] The average charges for our IVDU-IE patients in the last year of the study (approximately $70 000/patient) are well below the national average of $120 000 and may reflect our team's attempts to avoid open heart surgery since IVDU-IE patients are 10 times more likely to die or need a second surgery. 16,28 This study has many limitations. First, in our center, the diagnosis of "mixed drug use" strongly correlates with IV drug use but it is not definitive. We elected to use this DRG since often these patients deny IVDU and clear objective clinical signs are often lacking. Also, the use of hospital ICD9/10 codes to establish a diagnosis is associated with several known limitations. Thus, the findings must be considered in that context. As a countermeasure for these limitations, patient data and drug use data were combined to filter out other cases of endocarditis that did not appear to have a drug-related cause. This approach allows for multiple filters to view the relationship that is occurring in the region; unfortunately, this method may under report certain cases.
We acknowledge these limitations and accordingly describe the findings as associations without concluding causation. Also, the review of  contracting reimbursements and regional economic challenges. 17 The Southern West Virginia energy sector contraction and associated unemployment persists, in part, because of concerns from new industries regarding relocation to a region with such a high incidence of opiate dependence in a potential workforce. This economic downturn cycle along-side increases in prescription opioid medications and influx of inexpensive heroin, have all likely participated in the regional opioid crisis. 19,29 Further research is needed to understand the overall impact of illicit drug use in the region and the multiple socioeconomic issues or other causation variables that may be participating in the epidemic.
However, based on the findings reported herein, investment in resources for high-risk areas, defined by geo-map driven hot spot analysis, for needle exchanges and rehabilitation centers may be