Introduction

Once engaged and liked to care, access to anti-retroviral (ARV) therapy decreases morbidity and improves life expectancy among those with HIV-infection. HIV-providers are increasingly tasked to engage in risk modification strategies to prevent the mitigation of the benefits of ARV therapy.

Injection drug use (IDU) has a detrimental impact on progression and response to HIV therapy, including those aging with infection [1–3]. Providers attempting to provide care for HIV-infected patients with active substance abuse are additionally hampered by a general lack of resources to address treatment, a high prevalence of co-existing psychiatric diagnoses and often difficult to resolve psychosocial stressors faced by those with addiction [4–6]. As a result, IDU has been associated with an increase in use of health care services and in the need for hospitalization. Medial and non-opportunistic infectious complications of IDU place the patient at risk for poor outcomes and burden the health care system when reviewed in both the context of large public health databases and in cohorts of illicit drug abusers, the latter often found to be noncompliant with HIV therapy [2, 3, 7–9].

Shared drug-paraphernalia carries an inherent risk for infectious transmission, however the act of IDU also predisposes the user to damage and introduction of infectious agents directly into the skin and blood vessels. Clinically, the injection route may lead to local and systemic thromboembolism, skin and soft tissue infections, abscesses, bacteremia/fungemia and to the deep-seated consequences of blood stream infection, such as endocarditis. In the US, only a handful of studies have focused on the incidence of bacteremia in HIV-infected adults in the post-ARV era; these studies demonstrate blood stream infections tend to remain associated with IDU, low CD4 T-lymphocyte counts and with greater age [10, 11].

Similar to many urban centers, providers in Washington DC are challenged with managing a growing population of aging injection drug users receiving and seeking treatment for HIV/AIDS. We explore the impact of active abuse of injection drugs on the development of common medical and infectious complications of IDU among patients engaged in outpatient HIV care in the post-antiretroviral era.

Methods

Medical records were reviewed retrospectively at the Washington DC Veterans Affairs Medical Center (DCVAMC), utilizing our comprehensive electronic medical-record (EMR) and data management system (Vista_CPRS). The DCVAMC is a tertiary care, Complexity Level IB hospital with a full complement of medical and surgical referral services. The center offers individualized on-site outpatient substance abuse rehabilitation services that includes an opiate treatment program. Our Infectious Diseases outpatient clinic provides primary care services for HIV-infected veterans and is additionally staffed by a psychologist and two full-time social workers who assist providers in extensively documenting substance abuse issues and the psychosocial factors impacting continued abuse, where applicable. This study was approved by our Institutional Review Board and The Research and Development committee prior to study initiation.

HIV-infected patients were identified as having drug dependency through ICD-9 coding within our existing clinical case registry, between 1998 and 2009. The ICD-9 screening of the case-registry included searches for opioid dependence (304.00–304.03), opioid abuse (305.5–305.51) and drug withdrawal syndrome (292). Each subject was then confirmed as IDU through examination of the clinical EMR using notes documented by providers (both outpatient visits and inpatient hospitalization notes) and by further review of pharmacy records for opiate-replacement therapy with methadone, admission histories, discharge summaries, social service intervention notes and all the DCVAMC substance abuse rehabilitation visits during the observation period.

With the ability to review of available treatment records and the documentation of providers, each subject was further classified as either active-IDU (active use of injection drugs) or drug-users with no active IDU. The drug-users with no active IDU (non-active IDU) group included those with heroin or cocaine abuse through non-injection routes and/or those with remote IDU, where there was strong evidence for successful recovery and/or no evidence of documentation of any usage within the preceding 10 years. The subjects followed for <6 months in EMR were excluded from analysis.

Blood stream infections (BSI) were defined as a positive blood culture (confirmed by review of the microbiologic data) in conjunction with compatible clinical history and/or findings on examination as described by the treating physicians at the time and, if unclear in EMR, confirmed by the Infectious Diseases trained physicians reviewing the charts. Overall, all-cause mortality was noted and when applicable, attributable BSI mortality was defined as death within 30 days of the BSI episode.

Comparisons of the means for continuous variables employed the Student’s t test; proportions employed Fischer’s exact or Chi square testing and age-adjusted survival was assessed with Cox proportional hazard analysis over the observation period. All analyses were conducted with two-tailed tests, accepting a P value of <0.05 (SPSS v21, Chicago, IL).

Results

316 unique HIV-infected patients with a history of drug-abuse over an 11-year period were identified. We further classified patients as active-IDU (141/316, 45 %) and as drug-users, without active IDU (non-active IDU) (175/316, 55 %). The latter group was comprised of nearly equal proportions of patients with substance abuse but either no history of injection-abuse or documented remote IDU (51 and 49 %, respectively). In data not shown, our cohort was almost exclusively male (98 %) and predominately African-American (95 %).

On average, patients had been living with their HIV diagnosis 10.7 ± 5.3 years and 76 % (240/316) had been prescribed ART (Table 1). Over 90 % of the patients were actively engaged in clinic visits during the observation period. Durability of response and compliance was not directly assessed, however the group not active in IDU was more likely to be on ARVs (82.3 vs. 68.1 %, P = 0.003) and were more successful in achieving undetectable viral load in the observation period (43.1 vs. 30.7 %, P = 0.02, Table 1).

Table 1 Clinical history and demographic data for HIV-infected patients with behavior of either active injection drug abuse (IDU) or non-active IDU, defined as either remote IDU or active non-injection route of abuse
Full size table

Concurrent tobacco-use, alcoholism and homelessness was frequent and rates were significantly higher among those with active IDU (P < 0.01, Table 1). Viral hepatitis co-infection rates were also predictably high for the cohort; however there was a strong correlation of active IDU to serologic evidence of exposure to Hepatitis B (HBV) and Hepatitis C (HCV), with rates of 88 and 95 % respectively (Table 1). There was a modest distribution of underlying hypertension, diabetes and congestive heart failure; however prevalence was not significantly different among the groups. Psychiatric illnesses such as schizophrenia, depression and post-traumatic stress disorder were also equally represented (Table 1).

Among the active IDU group, there was a higher prevalence of DVT/PE and a trend towards a greater number of cerebrovascular events (12.1 vs. 2.3 %, P = 0.001 and 5.7 vs. 1.1 %, P = 0.05). There were 0.97 hospitalizations/year in the cohort, over the observation period. The rate of hospitalizations was higher for active IDUs (Table 1).

In the observation period, a total of 120 clinically significant BSI episodes were identified and 60 % (72/120) were among active IDUs. There were 63 (19.9 %) individuals with at least one clinically relevant BSI (mean of 0.09 BSI/year) and 54 % (34/63) BSIs occurred in those with active IDU. More than one BSI occurred in 9.2 % of patients. There was no difference in the proportion of individuals with BSI, with regard to IDU activity, however the occurrence of multiple BSI was significantly higher in those with active IDU (13.5 vs. 5.7 % P = 0.02, Table 1).

There were 107 deaths in the 11-year observation period. All-cause mortality was significantly higher in active IDU compared to non-active group (42.6 vs. 26.9 %, P = 0.003, Table 1). Patients were followed for a mean of 7.0 ± 3.5 years, with 78 % of those surviving to the end of the observation period reaching ages above 50 years. The mean age was significantly greater in the active IDU group (46.2 ± 5.1 vs. 44.7 ± 6.9 years, P = 0.03, Table 1), however no significant difference in age was noted among those with and without BSI (data not shown). Over the observation period, age-adjusted survival was 63 % shorter for the active IDU group (HR 1.7, CI 1.16–2.51, P = 0.006). BSIs contributed to significant all-cause mortality (62.1 vs. 26.9 %, P < 0.001). Among those with BSIs, active IDU did not appear to be a secondary risk factor for mortality.

Discussion

Our comprehensive EMR combined with the presence of the on-site substance abuse services and a well-staffed comprehensive HIV-care clinic allowed us to capture retrospective data over a decade in what is usually considered a potentially unreliable cohort and, where historically, a variety of healthcare venues might be used with acute illness. While we were limited in our ability to categorize the frequency of active IDU, our control group was comprised of non-injection route substance abusers and those with remote IDU in an attempt to minimize bias resulting from complex addiction-related sociodynamics, particularly difficult to detect or measure in a retrospective approach.

Across the US, HIV-providers routinely perform clinical risk-reduction counseling, with increasing emphasis on age-related management of medical comorbid conditions and modifiable risks, such as tobacco usage. We found that the practice of active IDU, compared to addiction not coupled with injection as a route of abuse or remote IDU, placed patients engaged in HIV-care at greater risk for morbidity and for mortality. When offered routine preventative vaccination or when referred for treatment for hepatitis co-infection, HIV-infected patients with IDU should also be counseled on the avoidable risks incurred by injection substance abuse. These risks include thrombosis, BSIs and subsequent hospitalizations. Over time, advancing age may place HIV-infected patients at additional risk for bacterial complications.

Conclusions

In many urban centers, HIV-providers are increasingly responsible for the care for aging patients with drug abuse issues. Addressing addiction and active-abuse has the potential to impact avoidable medical complications and contribute to the continued, overall health of patients linked to HIV-care.