Infections in Injection Drug Users

Immune Changes

Opiates have been demonstrated to reduce chemotaxis, phagocytosis, and production of cytokines and chemokines. Both heroin and morphine (a major metabolite of heroin) have been shown to decrease natural killer cell activity and decrease lymphocyte proliferation in response to phytohemagglutinin and other mitogens. Morphine has also been shown to impair antigen presentation by B lymphocytes. In addition to their direct effects on cells of the immune system (which have been demonstrated to have opiate receptors), opiates affect the immune system indirectly through the neuroendocrine system (mainly via the hypothalamic-pituitary-adrenal axis). Immunosuppression mediated through effects on the autonomic nervous system is also suggested. Despite the depressed cell-mediated immunity demonstrated in IDUs, opportunistic infections characteristic of T-cell deficiency were rarely reported before the HIV epidemic. In the lung, heroin appears to reduce the activity of inducible nitric oxide synthase, which may increase susceptibility to pulmonary infections. Methamphetamine increases the expression and production of metalloproteinase-2, leading to host collagen degradation and diminished wound healing, thereby at least partially explaining the prevalence of S. aureus in wounds of IDUs. It also impairs the function of phagocytic cells, further increasing susceptibility of users to infection.

Prolonged methadone maintenance has been shown to reverse some of the immunosuppressive effects of heroin. Buprenorphine, which is increasingly used as substitution therapy, also appears to allow a reversal of the immunosuppressive effects associated with heroin addiction.

Active IDUs have substantial humoral immunity dysregulation that is associated with increased systemic inflammation. IDUs who resist HCV infection despite repeated risky activity have significantly elevated levels of immunoglobulin G (IgG) and IgM against viral envelope glycoproteins as well as alterations of natural killer cells compared with healthy control subjects. Increased immunoglobulin levels tend to normalize after prolonged opiate withdrawal. Elevated immunoglobulin concentrations are accompanied by a high frequency of autoantibodies such as rheumatoid factor, as well as those directed against various microorganisms. The latter phenomenon often manifests as a biologic false-positive Venereal Disease Research Laboratory test result, which may create diagnostic confusion in IDUs who are at high risk for acquiring sexually transmitted diseases (STDs). Mixed cryoglobulinemia is also seen and increases with the duration of drug use. Hypergammaglobulinemia resulting from polyclonal B-cell activation may be the result of recurrent immunologic stimulation by injected foreign antigens as well as associated chronic liver disease and chronic infections with other pathogens. Regardless of HIV infection, injection drug use is associated with an increase in peripheral blood and mucosal cells. Interestingly, sharing injection paraphernalia also leads to the same increase in immune activation.

It has been shown that morphine may depress the monocyte functions essential for antiviral defense. These alterations could contribute to the high efficiency of transmission of certain viral pathogens in IDUs, including HBV, HCV, and HIV. In vitro, morphine has been shown to enhance HIV infection of human mononuclear cells through the downregulation of β-chemokine production and the upregulation of CCR5 receptor expression. More recent evidence shows that microRNAs involved in the processing of HCV are altered in IDUs and at least partially explain the innate anti-HCV immunity among drug users.

Infection Sites

The distribution of soft tissue lesions varies with the sites used for injection and reflects both the duration of drug use and the local practices among drug users. Typically, IDUs go through a progression of injection sites, starting with the upper extremity in the antecubital fossa, followed after approximately 2 years by the forearm. After about 4 years of use, users switch to injecting into the veins of the hand, and approximately 6 years after first injecting, they switch the veins of the neck, feet, and leg. By the 10th year of drug use, IDUs resort to the groin and peripheral digits. The groin becomes a favored site after other injection sites have been exhausted, although in some cases femoral injection is preferred for convenience and ease of use or as a means of ensuring immediate blood levels of the drug rather than facing the risk for the reduced euphoric effect associated with skin popping, that is, either intended or accidental injection of drugs into the skin and subcutaneous tissues.

After repeated injections into a site, frequently without benefit of sterile technique, local ischemia or necrosis develops, and the tissues become susceptible to infection. In addition, the substances injected frequently contain materials added as diluents that commonly cause norepinephrine release and vasospasm or local damage to the vascular intima. This leads to thrombosis and further compromise of the soft tissues. Cocaine use may be associated with vascular thrombus at sites distant from injections and may cause muscle and skin infections even after inhalational use. Opiates also have immunosuppressive properties that may predispose to infection. In addition, among methamphetamine users, there are substantial differences based on the color of the drug they inject. Depending on the type of adulterant added to the drug the color may be clear, white, yellow, or pink. IDUs injecting a colored drug were more likely to develop an abscess compared with IDUs using the clear substance. HIV infection is now recognized as an important risk factor for skin abscesses. Women are at greater risk, presumably because of the difficulty they have accessing veins and the consequent injury to skin and subcutaneous tissues. In one British study, in addition to female sex, all of the following were associated with an injection site infection: being at least 25 years of age; having injected into the legs, groin, and hands over the last year; injecting on 14 or more days in the last 4 weeks; reusing needles and injecting crack cocaine; having HIV infection; and having previously received a prescribed substitute drug. The cost of such infections was estimated conservatively to be between £19.2 million and £30.5 million. It is also very common for IDUs to attempt self-management of their wounds by using a variety of manipulations, including applying heat or a rag soaked in salt water, squeezing out pus with their fingers or doing incision and drainage with a knife or syringe, or taking antibiotics orally or by applying crushed antibiotics to the wound. Although many of these behaviors may increase the risk for spread of the infection, in at least one study the infection healed in 89% of patients. Only 12% ended up going to a hospital, and most waited 5 days before doing so.

Abscess

Abscesses are the most common form of soft tissue infection, followed closely by cellulitis and skin ulcers. Cleaning the skin with alcohol before injection protects against abscess and offers a potential intervention to reduce disease and hospital admissions. Original optimism that needle exchange programs might reduce the incidence of abscesses has not been fully realized. In Vancouver, despite having a supervised injection facility program, 21.5% of IDUs had an abscess in the previous 6 months. However, other investigators demonstrated that for every 1000 needles exchanged, there was one less abscess; for every eight visits to the needle exchange program, one additional abscess was prevented. When the funding for the program was reduced, the prevalence of abscesses increased. With conflicting data, it may be that either complete cessation from injecting or scrupulous attention to sterile technique is the only effective measure to eliminate abscesses in IDUs.

The location of lesions depends on the sites injected. They are found commonly in both upper and lower extremities depending on where the user most frequently injects. In at least one comparison between IDUs and nonusers, IDUs were more likely to have abscesses in the groin, whereas the hand was a more common location in nonusers. Addicts also tended to present later and ended up having a longer hospital stay, with treatment costing almost 2.5 times more. Although abscesses occur with injection of heroin as a single drug, they are also found frequently in addicts who inject a mixture of heroin and cocaine and are more likely in IDUs with a long history of drug use and skin popping. Booting is also an independent risk factor for abscess formation. Additional risk factors for abscesses are female sex (odds ratio [OR], 1.7; P = .002), recent incarceration (OR, 1.7; P = .001), involvement in sex trade (OR, 1.4; P = .030), frequent cocaine use (OR, 1.5; P = .002), and being seropositive for HIV (OR, 1.5; P = .003).

Abscesses may spread to adjacent tissues, frequently with disastrous consequences. Mediastinitis may result from the extension of a cervical abscess, whereas lesions in the carotid triangle can erode into the carotid arteries resulting in massive hemorrhage. Thrombosis of the internal jugular vein has been reported as a complication of a deep neck abscess, as has acute vocal cord paralysis. This can lead to acute, severe airway obstruction and may necessitate immediate tracheostomy. Local venous thrombosis or extension to the retroperitoneal space may result from abscess in the femoral triangle.

Management of Soft Tissue Infections

The first decision in the management of patients with soft tissue infections is whether the patient should be admitted to a hospital. This can be a difficult decision, but it depends on the severity of the illness. Eron and colleagues recommend an approach in which patients are assigned to one of four classes. Class 1 patients have no signs of systemic toxicity and have no uncontrolled comorbid conditions. These patients usually respond to topical or oral therapy. Class 2 patients either have evidence of systemic illness but without any unstable, comorbid conditions or are systemically well but have one or more comorbid conditions that may complicate the outcome. Class 3 patients are toxic or are not toxic but have unstable comorbid conditions that may interfere with the response to therapy. Class 4 patients have sepsis syndrome or a serious life-threatening infection such as necrotizing fasciitis. Predictors of severe infection are hypotension, tachycardia, temperature lower than 35°C (95°F) or higher than 40°C (104°F), confusion, or a depressed level of consciousness. Patients with two or more of these findings have a blunted response to antibiotics and poor outcome. In terms of the decision whether to admit the patient to the hospital, class 1 patients most likely will do well if managed as outpatients. Class 2 patients may benefit from a period of observation; patients who respond quickly can be treated as outpatients, but others will require hospital admission. Patients in classes 3 and 4 should be treated as inpatients.

Determining the appropriate antibiotic is the next decision. Class 1 patients usually respond to oral antibiotics, as do some in class 2. Initial therapy with parenteral agents with rapid switch to oral agents may be preferable for class 2 patients. Patients with more serious illness should be treated with parenteral therapy, at least until they respond satisfactorily. Data from randomized trials to guide the duration of therapy are insufficient. In most cases, 1 to 2 weeks of treatment is adequate, but infection may recur in up to 20%; thus prevention of future infections cannot be ignored. Empirical therapy aimed at the usual pathogens is usually adequate, although in at least one medical center, patients with hand abscesses were found to be 11 times more likely to be infected with a clindamycin-resistant MRSA than other patients. Thus it may be important to be aware of local resistance patterns before initiating empirical treatment.

As with other patients, incision and drainage remains the mainstay of treatment of abscesses in IDUs. Infection in the deltoid region is a consequence of skin popping and subsequent necrosis of the underlying fascia. In these cases, drainage may be facilitated by ultrasound-guided aspiration, even in the emergency department setting. Regardless of location, cure rates for drainage alone may approach 80%; however, the addition of trimethoprim-sulfamethoxazole for 7 days is associated with a significantly higher cure rate and is effective at preventing many secondary complications such as need for repeat drainage or metastatic spread of infection.

Antibiotic therapy is directed at the organisms recovered from the blood or purulent material. In uncomplicated cellulitis, cultures are seldom helpful. In such cases, therapy is empirical and is based on the pathogens most commonly encountered in that geographic location. Prolonged antibiotic treatment is frequently required. As noted previously, early surgical drainage of abscesses is essential; because of the tendency of these lesions to spread to adjacent or distant regions, multiple drainage procedures may be required. Deep infections of the hand are far more common in IDUs than in non-IDUs, and they mandate a unique approach. The microbiology of such infections varies depending on the injected substances. Patients who primarily inject cocaine have a high frequency of mixed anaerobic infection, whereas heroin users are more likely to harbor streptococci and staphylococci. In either case, surgical débridement is far more likely to be required in IDUs than in non-IDUs. Some caution is indicated before incising a lesion in the vicinity of blood vessels because a mycotic pseudoaneurysm can easily be misdiagnosed as an abscess. Inadvertent entry into such a lesion can have disastrous consequences. Inadvertent intraarterial injection of crushed tablets leading to hand swelling mimicking infection has also been reported. Failure to make the correct diagnosis may lead to loss of digits due to extreme ischemia.

Skin Ulcers

Skin ulcers are extremely common in IDUs. They are found at every conceivable site but are particularly common below the knee, close to the ankle. They arise from tissue damage caused by repeated nonsterile injection into the same site with associated thrombosis and infection. Synergy between streptococcal infection and cocaine-induced tissue ischemia may lead to large necrotic ulcerations and extensive tissue loss. Alternatively, skin ulcerations may result from necrosis induced by the illicit substance injected. Skin ulcers may persist for years and are a frequent reason for hospitalization. Typically, they have ragged edges and seropurulent drainage. Patients experience severe pain, and it is often the pain, rather than the ulcer itself, that brings the patient to medical attention. The microbiology of these lesions is similar to that of other soft tissue infections in addicts, although they more frequently contain more than one organism. S. aureus and β-hemolytic streptococci remain the most common isolates, with gram-negative bacilli—most often Klebsiella, Pseudomonas, Escherichia coli, and Proteus —playing an important role. Skin ulcers present particularly difficult management problems when they involve the hands and feet, and they may ultimately lead to loss of function.

Treatment of skin ulcers requires administration of systemic antibiotics and prolonged local wound care, including gentle washing, wet-to-dry dressings, and application of topical antibacterial creams. Elevation of the leg to reduce edema is an important component of the therapy and plays a role in pain management. Generally, parenteral antibiotics are continued until the wound is covered by granulation tissue. Very large lesions may require skin grafting or muscle flaps, but these are effective only after all necrotic tissue has been removed and the wound is clean and granulating. An important adjuvant treatment is the application of compression dressings, such as Unna boots, which, when properly applied, serve to reduce edema as well as to promote wound healing. With time, most skin ulcers heal completely, leaving circular, punched-out scars.

The most important complication is contiguous osteomyelitis, which may be difficult to diagnose because radiologic evidence frequently indicates periosteal reaction in bones immediately beneath large ulcers. When there is still a question of osteomyelitis, a triple-phase bone scan or MRI may be helpful. Ultimately a diagnosis of osteomyelitis may be impossible without a bone biopsy, which may be difficult to obtain without traversing infected superficial tissues. In such cases, prolonged parenteral antibiotic therapy directed at the organism cultured from the ulcer and careful radiographic follow-up may be the best approach. A combination of codeine tablets cooked with a solvent (gasoline, lighter fluid, or paint thinner) creates a drug called desomorphine, known as krokodil because of the scaly green patches that develop at the injection sites. Ultimately these lesions may ulcerate, and underlying osteomyelitis is known to occur. As with similar ulcers, a team approach and local care are required for resolution.

Paradoxically, IDUs may wish to prevent healing of these ulcers because once well-vascularized granulation tissue forms, the ulcers become an excellent location in which to inject. These lesions are referred to as “shooter's patches.” They may be successfully managed by use of a well-vascularized muscle flap, providing even better access to the circulation. The competing goals of the clinicians and the IDU are a source of continued frustration for the health care team.

Necrotizing Fasciitis

Necrotizing fasciitis, without or with myositis, is the single infection in IDUs that is most likely to need immediate and appropriate treatment. In one institution the prevalence increased significantly in the period 2009–2010 solely due to an increase in admissions of IDUs. However, the clinical picture is subtle and rarely elicits the emergency response required. At San Francisco General Hospital, 1% of IDUs in need of incision and drainage for soft tissue infection were found to have necrotizing infection requiring extensive débridement. The classic findings of high fever, bullae, crepitance, and skin necrosis are usually absent initially, and the impression may be that of mild cellulitis. In some cases, the true nature of the disease may be so subtle as to be missed during a procedure to débride an abscess or cellulitis. Alternatively, infection may spread after an apparently effective incision and drainage. The major indication of the true nature of the infection is the fact that signs and symptoms, such as pain and hemodynamic instability, are disproportionate to the apparent extent of the local process. However, this can be misleading because the clinical presentation may be no different from routine cellulitis, with no more than erythema of the involved area, providing no clue to the serious underlying pathologic process. Also, because addicts are frequently viewed as drug-seeking complainers, patients with excessive complaints for what appears to be a minor disease may be interpreted as exhibiting narcotic-seeking behavior. This misperception can further delay recognition of the need for aggressive and rapid action. In one study the correct diagnosis was made in only 59% of patients who presented to an emergency department, and many were initially admitted to a nonsurgical service. Others argue that the best results are achieved by admission directly to a vascular surgery service. Regardless of the admitting service, it is clear that a high index of suspicion is required so as not to miss the correct diagnosis. Additional clues to the serious nature of the problem are hemodynamic instability, increased creatinine level, local anesthesia, rapid progression of inflammation, or the presence of blue or hemorrhagic bullae. Crepitance is an important clue when present. Although a rapid test is not yet available, the finding of the proinflammatory cytokine interleukin-1 receptor antagonist with markedly elevated white blood cell count has been associated with fatal outcome in patients with necrotizing fasciitis, especially in the absence of fever. Finally, a slow response to appropriate antibiotic treatment suggests a deeper underlying problem. MRI and CT may be useful diagnostic tools. Characteristic findings include asymmetrical fascial thickening and fat stranding, followed by gas tracking along fascial planes. Abscesses may also be seen. CT scans may be misleading because both false-positive and false-negative results have been reported, and contrast enhancement contributes no additional information. The only definitive test is surgical exploration, which is both diagnostic and therapeutic. The finding of necrosis is characteristic; however, it may be necessary to explore more than one area. A negative biopsy result from one location does not preclude the diagnosis in adjacent tissues.

As with most infections in IDUs, gram-positive organisms are usually found. However, β-hemolytic streptococci predominate in approximately 50% of cases, followed by S. aureus, α-hemolytic streptococci, and coagulase-negative staphylococci. Gram-negative organisms are infrequent and are usually represented by enteric pathogens, especially E. coli, Klebsiella, Proteus mirabilis, Pseudomonas, and Enterobacter. Anaerobes are recovered in 12% of cases, including Clostridium sordellii in one series of black tar heroin users ; yeasts ( Candida spp.) are uncommon. Polymicrobial infection is common. Cultures are imperative, as occasionally unusual organisms are detected, as in the case of IDUs with C. novyi as the sole cause of necrotizing fasciitis.

Management of necrotizing fasciitis by antibiotics alone leads to progression of the infection in 75% of patients. Parenteral antibiotics and aggressive surgery coupled with reexploration at 24 hours and as often as necessary afterward to ensure complete removal of all necrotic tissue offer the best prognosis. In one study, IDUs required an average of 3.4 débridements for necrotizing fasciitis. Aggressive nutritional support and early coverage of the soft tissue defect have been shown to improve the outcome. Even with aggressive treatment, the mortality rate is high, ranging from 10% to 23%, and amputation is required in up to 10% of patients. Patients with group A streptococcal infection and streptococcal sepsis appear to have the worst prognosis. The best results are achieved with the use of vacuum-assisted therapy in addition to extensive débridement. Addicts, who tend to be young and relatively healthy, have the best outcome of any patient group with this disease.

Pyomyositis

Pyomyositis, a less serious infection involving the musculature, occurs frequently in IDUs. Direct inoculation of bacteria into the musculature has been implicated. Hematogenous spread also occurs, occasionally as a complication of endocarditis. Most patients who have pyomyositis present with pain and swelling of the affected area. Lesions have been reported in the deltoid, psoas, biceps, gastrocnemius, gluteal, and quadriceps muscles. Ultrasound, CT, or MRI shows the underlying defect within the muscle. S. aureus is the most common pathogen; infection with viridans streptococci, infection with aerobic gram-negative bacilli, and mixed infection with anaerobes have also been reported. Patients respond well to drainage and antibiotic therapy. A rare but related condition, uterine pyomyoma, has also been reported. The cause appears to be hematogenous dissemination to an infarcted leiomyoma.

Just as needle exchange programs reduce HIV infection among IDUs, combining needle exchange programs with a wound and abscess clinic may substantially reduce the cost of care (to as low as $5.00 per patient) and the number of visits to emergency departments. Widespread implementation of such clinics could have a major impact on the management of skin and soft tissue infections among IDUs.

Injection Anthrax

Over the past several years a cluster of cases of injection anthrax have been identified among IDUs in the United Kingdom and Germany. The first recognized case occurred in a skin popper who was diagnosed in 2000. After a lull, additional cases were seen between December 2009 and December 2010. There were 54 confirmed cases, and 18 deaths occurred. Sporadic cases still occur. In contrast to classic cutaneous anthrax, cases do not manifest as an eschar. These patients typically present with severe tissue swelling and evidence of soft tissue infection occurring 1 to 10 days after injecting heroin. The injection was frequently in the subcutaneous tissues. Initially the lesions were often similar to typical soft tissue infections seen in IDUs; however, blood cultures were often positive for Bacillus anthracis. Other diagnoses were also confirmed via polymerase chain reaction (PCR) assay of tissue from wounds. Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry was also used successfully in several cases. Cultures of B. anthracis can be easily misidentified as Bacillus cereus. In suspected anthrax cases, special techniques should be used to verify the identification of the isolate.

Patients in this outbreak were treated with a variety of different antibiotic regimens. Empirical therapy, including ciprofloxacin and clindamycin with additional antibiotics combined into a five-drug regimen, is recommended. In addition, aggressive surgical débridement with or without fasciotomy is considered necessary, and repeat débridement may be required.

The epidemiology of this outbreak is currently unclear, but it is most likely related to the transportation of heroin from Afghanistan, the site of origin, to Western Europe and the United Kingdom hidden in goat skins. It appears these goat skins were likely contaminated with anthrax spores that germinated once injected into the warm tissues of the patients.

Epidemiology of Skeletal Infections

Skeletal infections are common in IDUs, most occurring via hematogenous seeding by bacteria or fungi. Target sites for infection are determined by the blood supply and predominantly involve the axial skeleton. The original source of these infections may not be apparent, or the infections may represent metastatic complications of endocarditis. Injection drug use has been identified as a risk factor for osteoarticular infection in patients with endocarditis. In addition, bone and joint infections frequently result from contiguous spread from adjacent, often neglected, areas of infection in skin and soft tissues. IDUs with hematogenous infection often have multiple sites involved simultaneously, and blood cultures frequently are negative at the time of presentation. IDUs with HIV infection do not appear to be at increased risk for osteoarticular infections, and when these infections do occur, they are most frequently caused by the usual bacterial pathogens, not opportunistic pathogens.

IDUs with skeletal infection tend to be young and otherwise healthy. Clinical findings include constitutional manifestations and local signs and symptoms depending on the site involved. Patients with osteomyelitis often have a paucity of findings, presenting with local pain and tenderness only. Lack of signs and symptoms frequently results in delay in diagnosis. Fever is absent in one-third of patients. Similarly, signs of sepsis and leukocytosis and radiologic signs may be absent in patients with osteomyelitis.

Pyogenic infections predominate, with almost 90% being bacterial in origin, although virtually any organism can cause skeletal infections. The predominant pathogen isolated is S. aureus, with one large series finding almost one-fifth of cases due to coagulase-negative staphylococci (half associated with hardware). Gram-negative bacilli are responsible for one-fifth of cases, with about half of these due to Pseudomonas aeruginosa . Polymicrobial infections are reported in 46% of osteomyelitis cases but only 15% of septic arthritis cases. Anaerobes are isolated from as many as one-fifth of bone infections but rarely cause septic arthritis. IDUs who lick their needles or the skin surface before injection may develop osteomyelitis or septic arthritis with E. corrodens (“needle licker's osteomyelitis”). Candida spp. have been increasingly recognized as a possible etiology of skeletal infections in IDUs, particularly spondylodiscitis and vertebral osteomyelitis. A characteristic form of systemic candidiasis has been reported among IDUs who inject heroin that includes folliculitis, usually of the scalp and beard; endophthalmitis; and bone and joint lesions, most often costochondritis. Lastly, Mycobacterium tuberculosis should always be considered among the possible causes of skeletal infection in IDUs, particularly vertebral osteomyelitis. As many as 20% of infections are culture negative. Because the microbiologic differential diagnosis is broad, nucleic acid amplification techniques should be used in these cases in an attempt to confirm the microbial etiology of the infection.

Site of Skeletal Infection

Joint infections usually involve the extremities, most commonly affecting the knee. The incidence of left-sided knee arthritis exceeds right-sided knee arthritis, possibly because of the tendency of right-handed IDUs to inject into the left groin veins; this suggests a relationship between site of injection and infection. IDUs are particularly susceptible to vertebral osteomyelitis. IDUs with vertebral osteomyelitis are more likely to present with symptoms of a shorter duration and have more profound neurologic deficits, likely owing to the higher incidence of cervical spine involvement compared with non-IDUs. Primary sternal osteomyelitis, often associated with an antecedent history of blunt trauma to the sternum, is reported in IDUs. This group is also prone to septic arthritis in unusual sites such as the sternoclavicular and costochondral joints and the pubic symphysis. Other sites frequently involved include wrist, shoulder, hip, and sacroiliac joints. Vertebral osteomyelitis may extend into the subdural or epidural spaces and may cause formation of an abscess, with consequent cord compression and paraplegia. In addition, lumbosacral vertebral osteomyelitis may be associated with psoas abscess.

Diagnosis and Management of Skeletal Infections

Because of the wide spectrum of organisms that may be involved, diagnostic needle aspiration for smear and culture is necessary in all cases. Even when blood cultures are positive, invasive diagnostic steps are advised because skeletal and bloodstream infection may represent two separate processes, and infections may be polymicrobial. Frequent arthrocentesis, arthroscopic or open drainage, and débridement of nonviable bone are also advised if clinically indicated. Antibiotic therapy is required for 4 to 6 weeks, with the selection of drug based on the identity of the responsible microorganisms and susceptibility data. Increasingly, oral antimicrobial agents with high bioavailability and good bone penetration are used for at least a portion of the therapeutic course.

Overall, with early diagnosis, the immediate prognosis of bone and joint infection in IDUs is excellent, but long-term follow-up data are lacking. Many IDUs present late in the clinical course, and delays in diagnosis and institution of therapy are accompanied by a high likelihood of chronic osteomyelitis and late relapse of disease. Also contributing to this late but frequent complication is the problem of nonadherence. IDUs tend to leave the hospital against medical advice when confronted with prolonged intravenous antibiotic therapy in a skilled nursing facility, and many physicians are reluctant to consider outpatient parenteral antibiotic therapy in patients who are active IDUs. A tendency for medical personnel to underappreciate and undertreat the pain associated with skeletal infection may also lead to patients leaving the hospital before discharge planning can be completed. A retrospective review of discharge to a residential addiction treatment facility to complete parenteral antibiotics showed high rates of success and substantial cost savings for the health system.

Epidemiology

Bacteremia is common in IDUs, often with resultant infective endocarditis (IE). In the Detroit Medical Center, 74 of 180 addicts with bacteremia had endocarditis. In North Carolina, there was more than a 12-fold increase in the number of hospitalizations for injection drug use–related IE from 0.2 to 2.7 per 100,000 per year between 2010 and 2015. The associated hospital costs during the same period increased from $1.1 million in 2010 to $22.2 million in 2015. At the same time there was a remarkable shift in the prevalence of IE from urban to rural populations in North Carolina and elsewhere. The increase occurred while the use of heroin remained stable and was largely the result of the dramatic increase in the use of prescription opioid medications. This transformation from primarily urban to rural populations forced to handle this new influx of addiction-related endocarditis has major implications for the overall health care system. In one setting, a single patient who had multiple episodes of IE was responsible for the institutional cost of $380,000, not counting the cost of providers and eventual cardiac surgery that was performed elsewhere. In part, this was due to a system unwilling to pay for drug rehabilitation services that may have prevented subsequent admissions. In contrast to the United States and with some exceptions, in Western Europe the prevalence of IE among IDUs has decreased, leading to a corresponding decrease in hospital admissions for right-sided endocarditis; serious cases of left-sided IE are still seen. In Central and Eastern Europe, there has been an increase in injection drug use and, in contrast to the West where polymicrobial IE has decreased, there has been an apparent increase in IE. In Scotland the injection of a new psychoactive substance referred to a “burst” (variably butylone, methiopropamine, or ethylphenidate) led to a dramatic increase in IE and necrotizing pneumonia cases, all of which were associated with S. aureus infection. A Swedish referral center saw an increase of cases of addiction-related IE but experienced a lower mortality rate, perhaps due to the implementation of a multidisciplinary team to handle these infections.

HIV infection also has had an effect on the epidemiology of IE among IDUs. The rate in this population now appears to be similar to the rate of prosthetic valve infection, approximately 1% per year. The incidence of IE is higher among IDUs with HIV than among seronegative IDUs (24.8 vs. 3.9 cases/10,000 person-years). With the widespread adoption of antiretroviral therapy (ART), the overall incidence has decreased. An inverse relationship between IE and CD4 ⁺ lymphocyte count (OR for 200–499 cells/mm ³ , 2.01; OR for <200 cells/mm ³ , 3.61) had been reported and undoubtedly still exists among patients with advanced disease.

In most early studies, men were affected more often than women (5.4 : 1 in the Detroit Medical Center, 2 : 1 in Chicago), and men with IE were older than women and had significantly longer histories of drug use (10.2 years vs. 7.1 years). However, among addicts with HIV infection, women had an increased risk (OR, 3.26), a rate that was similar to addicts with increased injection frequency. The reason for this difference is unclear, but it may be due to women having smaller veins, making it more difficult to inject, which might lead to more local infections and subsequent bacteremia. Greater risk-taking behavior may also be a factor. HCV infection was an associated factor in 100% of IDUs with IE in Taiwan. Injection drug use is also a risk factor for relapsing (≤41%) and polymicrobial (8%) IE. Among patients with recurrent disease, the median interval between episodes is far shorter in addicts than in nonaddicts. In areas where buprenorphine lacks the inhibitor compound that is active only when injected, such as in Singapore, addicts have turned to injection, with the consequence of an increased rate of IE. Cocaine use is an additional risk factor. In contrast, alcohol consumption confers protection against endocarditis, perhaps by inducing an inhibitory effect on platelet function.

Microbiology

S. aureus remains the most common pathogen, affecting the tricuspid or pulmonary valve in approximately 90% of cases. Coagulase-negative staphylococci are an uncommon cause of endocarditis in IDUs. Streptococci, particularly groups A, B, and G, are the second most common pathogens. These two organisms account for up to 75% of cases. Ruppen and colleagues found Streptococcus dysgalactiae to be common over a 10-year period, 2006–15, with a predominance among female addicts. Enterococcus played a major role in the past, but its prevalence is decreasing. Gram-negative organisms are infrequent causes. Intermittent epidemics of P. aeruginosa endocarditis have occurred in Detroit and Chicago. The most recent epidemic occurred in Detroit during the years 2006–08. There were 10 cases, all in HIV-negative patients. Five were left-sided, three of which occurred on prosthetic valves. Four patients had tricuspid infection alone, and one had aortic and tricuspid disease. In one case, Pseudomonas was combined with Candida parapsilosis . Serratia marcescens was responsible for a sustained epidemic in the Oakland, California, area and occurs sporadically. Gram-negative endocarditis in IDUs is not common outside the HACEK group ( Haemophilus spp., Aggregatibacter spp., Cardiobacterium hominis, E. corrodens, and Kingella spp.). In one report, only 2 of 49 cases of gram-negative endocarditis occurred in IDUs.

Among the fungi, Candida endocarditis (typically non– C. albicans ) is most common. In one series of Candida endocarditis, 9 of 30 cases were in IDUs: 3 of C. parapsilosis; 3 of C. albicans; and 1 each of C. tropicalis, C. guilliermondii, and C. pelliculosa. All patients had left-sided infection, but the mortality risk was lower among IDUs (OR, 0.12; 95% confidence interval [CI], 0.02–0.7; P = .03). One death occurred at day 4 owing to a brain embolism, and the others occurred remotely: one at 3 months from an unknown cause and one at 1 year due to S. aureus IE. Aspergillus endocarditis in addicts has also been reported. Two of three patients were HIV infected. Duration of symptoms ranged from 2 weeks to 1 month before detection. Blood cultures were negative, but there were large vegetations and peripheral embolization. Aspergillus frequently contaminates illicit drugs, but whether the inoculum associated with injection is sufficient to cause endocarditis is unclear. That two of three patients were HIV positive is of interest and may indicate that altered immune status played a role.

Polymicrobial endocarditis is observed with increasing frequency among drug users. Usually only a few organisms are involved, but rarely there may be numerous pathogens. In such cases, standard laboratory techniques may be inadequate to isolate and identify the full microbial spectrum, placing a burden on the clinician to suspect polymicrobial endocarditis caused by salivary contamination of needles or injection sites whenever uncommon oropharyngeal organisms are cultured from the blood. E. corrodens is often found in polymicrobial IE, frequently in association with Streptococcus constellatus. These organisms may be a synergistic combination, owing to coinciding metabolic niches. They coaggregated during in vitro growth, and in animal models, E. corrodens produced endocarditis only if injected with streptococci. Among the fastidious organisms, in addition to Eikenella there are reports of endocarditis caused by anaerobes such as Fusobacterium spp. and Clostridium. Sporadic reports document the occurrence of B. cereus bacteremia in IDUs, who, in contrast to immunocompromised patients infected with this pathogen, tend to have uncomplicated infections that may even resolve without therapy. In addition, numerous reports describe endocarditis in IDUs caused by a variety of organisms that are frequently considered nonpathogens. These infections may be related to altered host immunity resulting from HIV infection or to unusual practices among addicts such as licking needles before use or “cleaning” the injection site with saliva. Given the proclivity of addicts to experiment with novel substances to inject, almost any microorganism can be found in the blood of IDUs, as exemplified by the isolation of Paenibacillus larvae, an organism that lives in honey and causes severe disease in honeybees, in the blood of addicts. These IDUs injected methadone that had been previously mixed with honey to make it too viscous to inject.